Final Report Summary - POPJIM (Plug and Produce Joint Interface Modules)
Executive Summary:
PoPJIM is a four year collaborative project (funded by Seventh Framework Program, call Factory of Future, area FoF.NMP.2010-1 Plug-and-Produce components for adaptive control). The PoPJIM goal was to intro-duce advanced concepts beyond the state of the art in the domain of intelligent machine-tools and to enable high precision mass customization and machining of complex components with zero defects. Increased ac-curacy of machines and processes offers substantial benefits to a wide range of applications that demand high quality and reliability of machining process.
The precision of machining systems is ever increasing in order to keep up with components’ accuracy re-quirements. At the same time product variants are increasing and order quantities are decreasing, which in-troduces high demands on the capability of machining systems. The machining system is an interaction be-tween the machine tool structure, the process and the control system and is defined in terms of capability by the positional, static, dynamic and thermal accuracy. So far, the control of the machining system, in terms of static and dynamic stability is process based which is often translated into sub-optimum process parame-ters and therefore low productivity.
Today's machine tool design criteria focus mainly on optimized lightweight and dynamically flexible struc-tures. This results in more efficient and fast machines, but unfortunately, it often also means less chatter re-sistant machine. Chatter is a dynamic instability of the cutting process which brings catastrophic conse-quences to machining process, such as poor surface quality, tool breakage or machine component failures. In order to succeed in high accuracy and high productivity machine building, dynamic instabilities during ma-chining must be kept under control.
For that purpose PoPJIM project brought a new approach: plug and produce Joint Interface Modules (JIMs). These modules consists of self-optimising mechatronic elements located in machine tool joints whose func-tion is to control the dynamic behaviour of the machine in order to assure process stability, avoid chatter and therefore improve machining accuracy.
JIMs will have plug and produce capability, with an underlying modular and flexible machine building con-cept and they will be communicated and controlled through a reliable Distributed Wireless Control Network (DWCN). The DWCN is responsible for the control of JIMs so that the overall dynamic stiffness of the ma-chining system (machine tool structure and machining process) remains optimized at the system level and it facilitates swift and reliable reconfiguration of JIMs and machine modules for plug and produce capabilities.
The project consortium has implemented an incremental approach in the realization of the full capability JIM. First, a mechanical JIM prototype has been built in order to show its dynamic capability. This first pro-totype has been equipped with piezoelectric tuning and capability and has been used to demonstrate the via-bility of the JIM concept at the very early stage in the project.
Project Context and Objectives:
The Strategic Research Agenda of “Manufuture” stipulates that European manufacturing industry needs to reinvent itself by shifting from cost-based global competition towards a dynamic creation of knowledge based added value [1]1. Such a shift necessarily requires introduction of advanced concepts beyond the state of the art which address the technological hurdles that limit their performance.
Complex components machined with zero defects are an essential condition in high precision mass customization and it becomes a new challenge for the new generation of intelligent machine-tools. Increasing the precision and accuracy of machines, products and processes offers substantial benefits to a wide range of applications from ultra-precision to mass customization with higher quality and better reliability.
Within this context, PoPJIM identifies critical performance limiting problems in machine tool design & development and plans to address them by coordinating various expertises with novel solutions. The outcomes of the envisaged effort will transform machine tools design and performance to new levels which in turn greatly contribute to realisation of low-cost self-optimising reconfigurable machine tools with modular plug-and-produce capability.
The PoPJIM advances two crucial innovations for the future machine tools. The first one refers to a concept of portable Joint Interface Module (JIM) as a self-optimising mechatronic element that controls the dynamic behaviour of a machine-tool during its interaction with the cutting process. Traditionally dynamic instability in a machining process is controlled by tuning the process parameters to match with the inherent dynamic characteristics of the machine tool structure which often results in lowering the rates of production. Novelty of the JIM concept is to control the machining process dynamics by adjusting dynamic stiffness of the machine tool at structural joints. In other words, instead of changing the process parameters, dynamic stiffness of the machine tool is tuned to maintain the process stability. With this approach, the process efficiency is not sacrificed as the process parameters remain unchanged while the overall dynamic stiffness of the machine tool structure follows the demands of voluntary or involuntary changing machining conditions.
Carefully designed damping and stiffness characteristics i.e. the dynamic stiffness in the JIMs allows the dynamic behaviour of the machine tool to be predictable with a higher degree of accuracy. The mechantronic design of the JIM includes an integrated control system which enables it to be self-adaptive for optimising the dynamic stiffness within its design range during a machining operation. A demand beyond the design range of a specific JIM is met by reconfiguring the machine tool with one designed for the required level of performance. The adaptive characteristic for maintaining optimum dynamic stiffness is achieved by exploiting both passive and active multifunctional materials in the JIM design. For the first time, this innovation will allow that a real representation of the joint behaviour can be introduced as a virtual model at an early design stage for fully model based design application.
Second facet of this innovation refers to a Distributed Wireless Configuration and Control Network (DWCN) which allows for plug and produce capability and decentralised control of JIMs through a wireless communication network. This development is essential for achieving operational flexibility and efficiency required for the modularity characteristics inherent in the JIM-based machine tools. The DWCN is responsible for configuration and communication among the network components, JIMs, network coordinator, DSP proxy etc. for establishing two major tasks:
1. Collaborative control of JIMs so that the overall dynamic stiffness of the machining system (machine tool structure and machining process) remains optimised at the system level
2. Facilitate swift and reliable reconfiguration of JIMs and machine modules for plug and produce capabilities.
Due to highly critical nature of the information flow and the engineering noise in production environment, communication reliability and efficiency are the two essential and challenging demands on design and development of the DWCN. The innovative concepts of the mechantronic JIM, application of novel multifunctional materials and the DWCN are essential pre-requisites in realising a self-adaptive, plug and produce machine tools with globally optimised dynamic stiffness. Furthermore, implementation of JIMs will enable model based analysis of machining systems2 at early stage of the design phase. Ultimate objective of these innovations is to keep the machining processes always stable while maintaining production rates and/or the quality of parts in terms of tolerance and surface finish demands as per specifications.
Joints are known to contribute significantly to the dynamic stiffness of the machine too l during its interaction with machining processes. Research in this area suggests that damping from mechanical joints accounts for up to 90% of the damping and 60% of the dynamic stiffness6 in a typical machine tool structure. Although structural joints are an excellent source of damping, the modern design practices fail to exploit this potential in developing machine tools with enhanced dynamic stiffness. This is mainly due to lack of:
• Controllability of joint characteristics
• Suitable functional materials which can compensate for loss of stiffness while increasing damping capacity
In addition to this, lack of theoretical models also prevents accurate prediction of the behaviour of joints under dynamic loading at the design stage. All the theories of elasticity available at present deal only with the problem of the monolithic elastic body, i.e. elastic body with infinitely rigid joints. Dynamic properties of joints can be altered by introducing functional materials of known properties. Research shows that addition of viscoelastic composites at the interface of structural joints results in considerable increase in damping. In addition, pre -stress on joint interface has a significant impact on the magnitude of damping in such joint8. Significant amount of research in this area has been conducted at the Royal Institute of Technology (KTH) in Stockholm. Findings of this research show that addition of high damping interfaces or re -designing critical interfaces in machine tool structures using high damping materials such as viscoelastic (metal-polymers) composites improves their dynamic stiffness by orders of magnitude. And this improvement has remarkable impact on the production rate, cutting tool life and surface quality of the machined surfaces9. Research at KTH in collaboration with industrial partners such as system 3R Intl. AB, Spirex Tools AB and Mircona AB also shows that design and operational variables such as interface design, configuration of the viscoelastic composites and interface pre-stress, pjoint, can be exploited in achieving dynamic stiffness to compensate for changes in operation conditions of the machine tools. Concerning the development of new multifunctional materials for improving dynamic stiffness in machine tool structures, KTH is collaborating closely with an industrial partner Plasmatrix Materials AB. This cooperation has been successful in developing a unique patented structural material named EDS (enhanced dynamic stiffness) material. The EDS material belongs to a family of nanocomposite materials consisting of a metal matrix with embedded carbon nanoparticles. By varying the proportion of nanoparticles in the metal matrix, the stiffness-damping properties can be varied from very stiff to highly damped.
Results of various research projects driven by Swedish industries show that increase in dynamic stiffness of different machine tool components and or working holding fixtures have a very positive effect on productivity and quality.
Introducing functional materials of known characteristics at the joint interfaces offer an opportunity to design joints of predictable characteristics. The core design principle is that the damping effect of the rigid joints between a JIM and machine tool modules is insignificant compared to that of JIM itself. On the other hand stiffness in the JIM will be good enough to direct all vibratory energy to its own interface for a localized dissipation due to high damping.
Since the joint characteristics in case of JIM will be dictated by the properties of the functional materials, few necessary changes in the models of these materials would suffice for quantifying the dynamic behaviour of these joints.
The objectives of the PoPJIM proposal as envisaged by the concept of JIMs are:
1. Develop portable Joint Interface Modules (JIMs) where an appropriate configuration of one or more functional materials of known characteristics act as an interface between two metallic elements and has dominant influence on the behaviour of the JIMs
2. Exploit the properties of both the interface geometric design and functional materials in developing the JIMs with varying and adaptable damping capacity and thus the dynamic stiffness as well
3. Develop active control mechanism in the JIMs so that they can optimise their dynamic stiffness as per demands of their operational environment
4. Develop a reliable communication network based on distributed wireless control technology which is inevitable for exploiting the flexibility that JIMs offer in realising modular machine tools with plug and produce capability
5. Demonstrate the feasibility of these developments by implementation in a milling machine
Project Results:
Development of the JIM concept
As mentioned in the preceding paragraphs, a typical JIM as visualized in the project offers both predictability and adaptability of its dynamic stiffness. These two characteristics are obviously valid within the design range of the JIM. When the operational demands on dynamic stiffness surpass the design capacity of the JIM, it has to be replaced with another JIM of higher capacity. This methodology thus defines three different levels of adaptability:
• Adaptability based on active control
• Adaptability based on interface design
• Adaptability based on dimensional design
It is worth mentioning that in the first two adaptability levels, nominal dimensions of the JIMs remain unchanged.
Adaptability based on active control
Voluntary and involuntary changes in operational conditions require corresponding adjustments in dynamic stiffness of the machine tool structure in order to keep the process parameters at optimal values. The adaptability in the dynamic stiffness is achieved by varying the joint interface pre-stress through active control during machining operations. The interface of a typical JIM is realised with a given design of selected functional materials placed into two rigid metallic components of a given geometric design of the interfacing surfaces. In order to pre-stress the joint limits of the control based adaptability in the dynamic stiffness lie between the lowest and highest allowable pre-stress for a given JIM and the controllable range then depends on factors such as:
• Design (mainly thickness) of the viscoelastic composite
• Pre-stressing capacity of the piezo actuator
• Load bearing capacity of the metallic pre-stress plates
The upper and lower pre-stress plates in this construction also provide for mechanical coupling interfaces for rigid connections with the adjacent machine tool modules. In order to maintain envisaged operational efficiency of the JIM, it is imperative that the stiffness of the JIM never exceeds the stiffness of rigid connections with the machine tool modules on the two ends of the JIM. The goal here is to divert maximum vibration energy into the damping interface of the JIM so that it is dissipated effectively through the viscoelastic composites. When a machining process requires dynamic stiffness that exceeds the design limits of adaptability through active control, then a JIM with higher stiffness capacity will replace it.
Adaptability based on interface design
Dynamic stiffness in JIMs can be increased or decreased by changing the geometric design at their interface. Design features such as conical or spherical surfaces, vgroves, key slots etc. can be used to enhance the JIM capability. In combination with different configurations of the multifunctional materials these design features at the interface provide for another level of dynamic stiffness in the JIMs.
Adaptability based on dimensional design
The third level of adaptability caters for the needs of changing dimensions of both JIMs and machine tool modules. The extend level of plug & produce capability in PoPJIM concept is the ability to reconfigure the machine tool from one type to completely another, for example from a conventional milling to a high speed milling machine. This reconfiguration may require that nominal dimensions of both the JIMs and the machine modules such as spindles may vary considerably. The system should be able to detect these changes and then adapt to the requirements in terms of maintaining the zero reference of the machine; this implying a channel for communication with numerical control of the machine tool as well.
Vital consideration in implementation of the JIM concept
Online and offline compensation for dimensional variation in the machine tool structure: The dimensional variations, either online due to actuation for prestress, temperature variations or offline due to use of JIMs and machine modules of different nominal dimensions, will be given due consideration in the overall design of the system. Any dimensional variation will be compensated with the precision commensurate to the requirements of the product quality.
Flexibility, speed and precision in mechanical coupling mechanism: The connections of the JIMs with machine tool structure/modules have to be more rigid than the highest levels of stiffness in the JIMs. In addition to this basic design requirement, the mechanical coupling for this purpose has to provide flexibility for easy changeovers, swiftness to minimise down times during the changeovers and obviously the necessary positional precision. Designs and solutions in this regard are available commercially; the best fitting to the requirements will be investigated and used for this application. One of the examples is System 3R AB which develops high precision mechanical coupling systems with full automation capability for changeovers.
The JIM design as described here assumes piezo-actuators as primary means of actuation for control of the dynamic stiffness. This however does not rule out the possibility of using other smart materials like magnetoristrictive, shape memory pneumatically or hydraulically operated mechanical forcing systems, etc. These materials and solutions may be considered in situations where the piezo-actuation is not a suitable approach.
Design methodology for JIM based modular machine tools
In line with the objectives of plug & produce approach, JIMs will ensure flexibility and efficiency both in the design & development and operational phases of the machine tools. In the design phase digital models of JIMs and machine modules will facilitate virtual analysis of the envisaged machine tool and assist in achieving a design with optimal dynamic performance. Including the influence of joint interfaces with known and controllable characteristics at the design stage will minimise the requirement of prototype manufacturing. The JIM based machine tools will be realised more efficiently with higher levels of reliability.
The self-adaptive control of dynamic stiffness in the JIMs and ultimately of the machining system comprised of two essential features namely monitoring and control.
Monitoring: While the machine tool is in operation, the dynamic behaviour of the machining process will be constantly monitored. An online vibration monitoring system, supported by a model-based identification of the machining process, will be implemented for extracting information regarding the machining system’s dynamic stability. The monitoring system will continuously transmit this information to a network coordinator for data analysis.
State-of-the-Art monitoring systems periodically record measurement data in chunks and perform offline analysis on the same data extracting characteristics of the signals based on time and frequency domain analysis. Since this procedure leads to significant time delays, the response behaviour of an underlying control system dramatically suffers. In fact, instabilities in machine tools such as chatter can occur in the fraction of a second.
Since the control system does not directly regulate mechanical states such as displacements, velocities, accelerations (which can be directly measured or dynamically estimated by state observers), but mechanical parameters of the system, real-time availability of accurate parameter values, especially global stiffness and damping parameters of the machine tool, are pivotal to quickly and reliably counteract instabilities. The proposed plug and produce approach doesn’t require mathematical models to be derived from underlying complex numerical models of the machine tool, but parameterized by intrinsic mechanical properties such as modal gains, eigenfrequencies and damping ratios. Estimation of these parameters will be carried out online, i.e. parameters appear as states of a dynamical system and are driven to their actual values by measurement data.
Due to the non-linearity of the underlying adaptive laws for parameter estimation and the use of IIR (infinite impulse response) models described by modal parameters, stability cannot be guaranteed a priori for all configurations. Hence, implementation and appropriate management of stability criteria for parameter estimation will be at the centre of investigations with respect to the monitoring system.
Self-adaptive control:
The control of the JIM interaction with the machining system is performed in two stage monitoring and control strategy:
• Modelling of the interaction between machine tool elastic structure and the machining process
• Self-adaptive control of the JIM units
In the first step the machining system’s operational damping and stiffness are extracted from the identified models. The input to the adaptive controller is the operational parameters from the first step. The output from the adaptive controller is the parameter that controls the JIM dynamic state, i.e. JIM’s damping and stiffness. It is important to point out between the difference between individual JIMs dynamic behavior and machining system dynamic parameters. By changing the dynamics of JIMs, a control of machining system dynamics will be achieved.
Model-based identification
Generally, a model is used to approximate data and it has a specific structure, i.e. a mathematical formulation which includes a number of parameters. These parameters can have a physical interpretation, in which case the identified model is a structural model, or retain only the mathematical interpretation in which case and regarded as a synthetic model. An advantage of the model based vibration analysis is that it can be used for obtaining both synthetic and structural models of machining systems.
The model based identification is carried out in two steps
1. Evaluation of ARMA model’s parameters – synthetic model
2. Formulation of the structural model, thereby the operational damping and stiffness can be extracted.
Parametric models
Parametric models is special class of models, driven by white noise processes and possessing rational system functions, includes autoregressive (AR) (Burg, least square, Yule Walker, geometric lattice, instrumental variable), ARX (arx, iv4), moving average (MA), autoregressive-moving average (ARMA), Box Jenkins, Output Error models.
The output processes of this class of models have power spectral densities that are totally described in terms of model parameters and the variance of the white noise process. The motivation for parametric models of random processes like machining is the ability to achieve better power spectrum densities estimators based upon the model than produced by classical spectral estimators like FFT. The modeling of a stationary time series as the output of a dynamic system whose input is white noise n(t), can be carried out in several ways. One way is to use the parsimonious parameterization which is employing ARMA(p,q) representation. Given a time series of data Xt, the ARMA model is a tool for understanding and, predicting future values in this series. The model consists of two parts, an autoregressive (AR) part and a moving average (MA) part. The model is usually then referred to as the ARMA(p,q) model were p is the order of the autoregressive part and q is the order of the moving average part. In its original representation ARMA parameter estimate is based on iteration schemes by help of a non-linear algorithm.
One of the main benefits of the model based identification is that it can be performed in-process and thereby be used for monitoring and control of machining system stability which requires fast response in order to detect chatter before full development. One important tool used to compare in model performance is spectral analysis. In spectral analysis the power spectral density (PSD) is defined as the discrete-time Fourier transform of an infinite autocorrelation sequence (ACS).
This transform (relationship) between the PSD and ACS may be considered as a nonparametric description of the second-order statistics of a random process. However a parametric description of the second-order statistics may also be devised by assuming a time-series model of the random process.
Self-adaptive control of JIMs
The controller is designed to work into two states:
(i) Control state, when the machining system approach instability threshold or in some situations, by deliberate actions to bring the system in a desired stability state.
The JIM is designed based on two calibration curves: pressure–damping and pressure–stiffness. These calibration curves are estimated for individual JIMs at the JIM design stage and then when integrated in the machine tools. As an initial pre-stress is necessary in the JIM, a self-learning technique will be implemented.
(ii) Monitoring state to follow the system behaviour
As long as the machining system is in stable region, or in the desired range, the control is disabled. For this reason the selected control at this stage will be the self-tuning type. At the start of the project a review of other control system types will be appropriate.
The heart of the adaptive controller as for the model-based identification is the recursive estimation. In the initial phase of the project, various recursive estimation schemes will be analyzed and tested. The self-tuning controller is based on the Pole Assignment Control (PAC). The aim is to modify the dynamic response of the JIMs in order to keep the machining system in the stable state.
Information from the monitoring system will drive the control system to self-adapt the dynamic characteristics of the machining system by varying damping and stiffness parameters of the JIMs. This will be achieved by changing the pre-stress on damping interface using smart materials as actuators in the JIMs. The control system will change the damping and stiffness characteristics of the critical joints in a way that energy directed to each interface is proportionate to its dissipation capacity so that the system does not experience any localized overload. The process monitoring and JIM control systems are linked to each other through a network coordinator. The network coordinator has a very vital role in the realization of the zero vibration machining system concept in PoPJIM. The machining process identification algorithms based on the estimate of overall system damping reside in the network coordinator. On the other hand the information on operational condition of each JIM is also processed simultaneously, but still in the network coordinator. The network coordinator will keep track of stiffness and damping parameters in each of the JIMs using mapping functions, also available in it (see Figure 6). As the system detects the machining process approaching its instability threshold, a coordinator function will invoke commands for JIM actuators to act in a manner that global dynamic stiffness of the machining system is altered to avoid the impending process instability.
It is important to note that the coordinator function and the underlying control algorithms do not work iteratively, i.e. offline analysis of a stream of vibration data followed by decision making with respect to pre-stress based on static mappings, but dynamically drive the system to a stable operating point based on adaptive control laws.
Such control laws will be derived from energy-like functions describing the quantity to be minimized (vibrational energy) or even maximized (dissipated energy) along the trajectories of the system. Performance of these control laws is, of course, governed by accurate models and proper estimates of states and parameters of the system. Candidates of update laws will be benchmarked with respect to stability and response to transient changes in the excitation spectrum. Since the machine tool will be equipped with multiple JIMs, cross-coupling of the JIMs and, hence, the multivariable nature of the control architecture has to be appropriately considered. A robust transducer system will be integral part of the JIMs to transmit information regarding their instantaneous dynamic state to the network coordinator. The network coordinator, by keeping track of both the process condition and the state of the JIMs, is providing control directions to each of the JIM for activating the pre-stressing mechanism whenever necessary.
It’s important to reiterate the fact that in conventional control systems the dynamic instability is controlled by changing the machining process parameters, often reducing the metal removal rate and ultimately productivity of the machine tool system. In this breakthrough concept the process instability will be controlled by changing structural dynamics of the machine tool system instead.
The result of this control strategy is that the system productivity will not be compromised at all; actually there will be a possibility to increase the removal rate as the combination of new dynamic characteristics may allow for more aggressive process parameters, obviously within limits of overall design capability.
Distributed Wireless Configuration and Control Network
PoPJIM will implement a Distribute Wireless Configuration and Control Network, DWCN, based on a reliable wireless multi-hop-networking facility to support plug and produce configuration and the JIM distributed control.
Machining environments pose specific challenges when deploying wireless networks.
In many cases, different signals interfere with the wireless channel and metal objects are in places that may sub optimize the network connectivity. However, to connect systems wirelessly no existing network infrastructure is needed and no additional network cables need to be installed which can reduce the time and cost of module replacement significantly. Also, sensitive equipment (like cable connectors that would be exposed to damaging environment factors such as vibration) is not needed on the systems. If carefully planned and carried through, wireless networks are an interesting alternative to traditional cable-bound industrial environments. Wireless network is a prerequisite for efficient plug and produce environment.
The usage of wireless communication allows plug-and -produce not only with respect to multiple JIMs, but also for the interaction of the machine tool components with a configuration support environment. In order to bring such a networking technology to a production system level, PoPJIM will do the extension and tailoring of an industry standard radio technology – IEEE 802.15.4.
The novel aspect of the plug-and-produce scenario supports configuration following mechanical mounting of the modules. This demands the wireless connectivity layer to be both self-configuring and self-organizing, but still under the given time demands of a scalable and reliable network. Administrative intervention is unwanted and contradicts the Plug and Produce principle. This particular demand is denoted (with respect to similar IP network capabilities) as zero-conf networking. This particular feature forms a novel extension for state-of-the-art in IEEE 802.15.4 based wireless networks.
The research in this aspect will be driven by a set of key objectives, in alignment to the overall project goals and the goals from the cross-thematic NMP / ICT call: support environment for ICT-based production systems, self-adaptive and portable plug-and-produce components with an open architecture, and high degree of autonomy and adaptability. From the development goals stated in the FoF.NMP.2010-1 call text, the work contributes to the “components and methods for intelligent, self-sufficient plug-and-produce systems”. PoPJIM will focus on the software part of “an open architecture to facilitate any additions of new modules as needed for implementation in a new process environment”.
The resulting networking facility has to consider a specific set of non-functional demands:
• Wireless communication to support easier mechanical integration and reconfiguration of JIMs
o True plug-and-produce support for JIMs and external control units
o Application of industry standard RF hardware for cost optimization
o Dynamic adaption with modules being installed, removed, or moved
• Automated establishment of a wireless network topology
• Research on dependable multi-hop networking strategies and algorithms
• Immediate setup of networking without manual intervention
• Research on application of zero-conf approaches in machining environments
• Automated scaling of communication
• Tailoring for harsh operation environments (e.g. RF interferences)
Configuration support environment
Apart from the mechanical aspect of plugging (mounting) of JIMs in the joints,
PoPJIM considers the Plug and Produce configuration of JIMs as composed of two phases:
1. Communication phase: a JIM needs to establish communication and send/receive data to/from the network coordinator, process proxy (the DSP proxy) and other JIMs.
2. Setup phase: the JIM self-optimizes to initialize and set its pre-stress value.
Plugging a JIM into the machine tool amounts to reconfiguration of a machine tool and hence it requires the selection of the right module through configuration analysis. To support this configuration task a support environment is required that is composed of two major components:
• Module library
• Configuration knowledgebase
Module Library Service
The module library service is meant to facilitate the configuration/reconfiguration process by providing digital models of the modules: the JIM modules as well as machine tool modules such as spindles, tables, slides, tool holders, chucks etc.
As mentioned in the preceding paragraphs, JIMs are to be designed with different operational range and geometrical features to accommodate various operation conditions and joint interface types. The description essentially contains the module’s basic attributes such as modules id, feature parameters, high-low values, etc and should be able to fulfil all the information requirement and format to conduct a digital configuration analysis. The existing device specification languages will be investigated for application in the project. In order to address a wider configuration scenario involving several module suppliers using different specification, a semantic module specification scheme will be developed and applied.
Configuration Knowledgebase Service
The knowledge base service is intended to offer the storage of data about machine tools and their modules, such as models, reference values, relevant standards, model templates or partial models, and connectivity and interfacing restrictions and possibilities. The knowledge base content will continuously be updated as new knowledge elements are acquired through experience and analysis of run time data. In sum, the DWCN deals with three major building blocks for the work can be identified: (i) the software part of wireless networking, (ii) the hardware part of wireless networking, and iii) the configuration support service.
Overall strategy and general description:
The objective is to develop and demonstrate the JIM concept for the machining centers. The project has duration of 48 months and is structured into 6 work packages.
As described in section 1.1 the main innovative technological advances envisaged by PoPJIM include the following functional areas:
1. Design and realization of a mechatronic Joint Interface Module (JIM)
2. Distributed Control for the JIMs
3. A multi-hop wireless network for the realization of DWCN
For such a highly innovative and multi-disciplinary research activity a well defined work strategy is essential to define clearly the scope of the project and the interdependence of the activities to achieve the objective with the resources available.
In addition to the common work packages WP1 Project Management and WP6
Dissemination and Exploitation, the project activities are organized in the following four work packages.
Work package 2 – Mechatronic Design of JIM with Functional Materials
The mechanical design of Joint Interface Modules (JIMs) with novel functional materials, Modelling and analysis of JIM integrated machining center, integration of the mechatronic components and prototyping of the JIMs.
Work package 3 – JIM Distributed Control
The JIM distributed architecture, the sensing and actuation system in the JIM and the embedded control. The output of these activities will be integrated to the module in WP2 integration task.
Work package 4 – Multi-hop Wireless Network
The Multi-hop topology for communication among the JIMs and other network components, the wireless infrastructure and the configuration support service to enable and support virtual configurations by providing module library of JIMs and other machine tool modules as well as a configuration knowledgebase.
Work package 5 – Demonstration
The JIMs developed in the RTD work packages will be demonstrated using a machining centre in an industrial context. The interdependencies and the flow of information and results among these work packages are summarized in subsection 1.3-d. The work plan will be monitored and updated, if necessary, by the project management to ensure the achievement of the objectives within the time and budgetary limits. These amendments are primarily meant to avert or mitigate risks that may arise due to anticipated/monitored or unforeseen events or make use of relevant and promising technology developments available outside the project.
Work Package 2: JIM Mechatronic Design and Realization
Task 2.1 JIM Requirement Specification
This task comprised mainly of two subtasks dealing with the dynamic performance analysis of legacy ma-chine tool and selection of the critical joint interface locations. Major objective of this task was to analyse the machine tools selected by the machine tool manufacturers in the project consortium i.e. AFM and Soraluce. AFM and Soraluce dedicated for the project a turning machine (TAE-35N) and a heavy duty floor type milling machine (FL6000) respectively. The two machines were analysed for their dynamic behavior both through experimental modal analysis and machining tests. Results from these analyses and tests laid the foundation for preparing the requirement specification.
The requirement specification is now being used as reference document for all developments as envisaged in the project plan including the JIM mechatronic design, JIM control and the multi-hop wireless network. The structural dynamic parameters such as the natural vibration modes, excitation amplitudes and modal damping of the two machining systems provide the guidelines for the developments related to the JIM design, control and wireless communication strategies.
Following analysis of the selected machine tools it’s found that the damping design, control and communication solutions should be able to handle the structural dynamics up to 2500 Hz. For the proof of concept, the critical control and communication range can be limited up to 1500 Hz as the higher frequency excitations were observed with rather extreme machining parameters.
Task 2.2 JIM Mechanical Design with Functional Materials
Main objective of this task is to develop model based design of JIMs and methodology for development of joint interfaces with known and controllable characteristics. Since the principle of the JIM based design is that the JIMs assume dominating role in defining the dynamic behaviour of the machining systems, the interfaces have to be designed with special materials with exceptionally high damping and acceptable stiffness. In this case two different types of interfacial materials are used. One is a metal-polymer compo-site with exceptionally high damping capacity and the other is plasma coated carbide interface with high damping and stiffness. Since the high damping materials, especially the polymer based, have usually low stiffness special emphasis on the interface design is necessary to cater for resulting lower stiffness. In this context three design parameters including the geometric design, material design and the boundary conditions become critical drivers for the JIM design.
Sub task 2.2.1 Joint interface property characterization
Main focus of this sub task is to characterize the dynamic properties of the interface materials. The viscoelastic polymers exhibit a highly nonlinear behaviour where the material modulus and loss factor (a measure of material damping) change nonlinearly with frequency or temperature. Both experimental and computational approaches have been used in characterizing the material behaviour. Following what has been developed in the 1st period, the method and calculation algorithm was further improved to measure and identify the joint interface properties with only assembled structures. In the developed calculation algorithm, the structure component’s properties are not necessary known for the identification process. As the joint interface’s static stiffness and damping values are defined as frequency dependent, the joint interface material’s properties are compared with what has been calculated by the algorithm depending on frequency. Following the scheme, the method is applied to measure the joint interface properties with pure VEMs without screws and it was found out that the VEMs Young’s modulus and damping property does not change under the pressure range the bolt screws can supply. The change of joint interface’s stiffness and damping is coming from the bolt screws instead. In the virtual simulations, five different designs have been analyzed through finite element analysis for the flexible fixture:
• Case I (Without VEM): Original system, no VEM layer added.
• Case II (1mm damping layer): 2 viscoelastic layers (3M™ Damping Foil 2552) with aluminium foil embedded between the base plate and the flexible fixture.
• Case III (1mm damping layer in base and screws): 2 viscoelastic layers (3M™ Damping Foil 2552) with aluminium foil embedded between the base plate and the flexible fixture, and between the screws and the flexible fixture (see Fig.3).
• Case IV (3mm damping layer): 6 viscoelastic layers (3M™ Damping Foil 2552) with aluminium foil embedded between the base plate and the flexible fixture.
• Case V (6mm Sylomer): 1 Sylomer layer (Sylomer ® SR850-6) of 6mm between the base plate and the flexible fixture.
It is clearly seen that after VEM layers have been applied in the interface, embed VEM layer in the joint screws can improve the flexible fixture’s performance even further.
Subtask 2.2.3 JIM design update with Added Interface Module concept
It was concluded that JIM approach was suitable for low damping and high frequency applications, such as the flexible fixture tested. This can be the case of numerous slender and thin walled structure machining applications, such as aeronautical parts. AFM case is in this group, with a slender workpiece to machine and a high frequency and low damping mode limiting cutting stability. However, Soraluce’s case is different, with low frequency (high mass) and high damping modes involved in process instability.
Therefore, a different approach will be considered for Soraluce’s case. An added mass supported by a tunable viscoelastic damping layer was used. The purpose of the damping layer is to add damping and to tune the frequency of the mass in order to make it work as a vibration absorber, through changing the preload of the viscoelastic layer.
Figure 16 shows a scheme of the operating concept of the tunable viscoelastic damping layer, which will be hereafter called Added Interface Module (AIM).
Sub task 2.2.4 EDS material embedded in JIM
Parallel to the mechanical design of the first JIM, use of plasma coatings as Enhanced Damping Surfaces, or EDS, to form damping layers at the interfaces is being investigated and a batch of experiments has been conducted and the work is still on progress. The experimental evaluation work is carried out in parallel with the Viscoelastic material.
One JIM prototype has been applied with the HiDS coating and the cone of JIM with coating is shown in Figure 18 (see the attached document). The thickness of the coating is 500 μm.
While the coated cone is embedded in JIM prototype with a workpiece clamped on the fixture (as shown in Figure 19), direct machining was performed to test the function of the damping coating.
Task 2.3 Mechatronic Design and Prototyping
Following the last report, the JIM mechatronic design and prototyping is updated during this reporting period. The main activities that are included in this reporting period include:
• The development of JIM with appropriate actuation and sensing elements integrated
• Performance analysis of the integrated actuation and sensing elements
• Performance analysis of the mechatronics JIM with variable stiffness
•
Since it was found out that the main obstacle for the 1st JIM mechatronic prototype to function is the piezo electronic device which forms a low impedance path for vibration energy, the second JIM mechatronic de-sign is updated with piezo electronic devices which are integrated with HDI. Apart from understanding that VEM in JIMs can enhance the mechanical structure’s vibration damping capacity, the effect of pre-stress on the JIMs performance against vibration is investigated in this period. When a workpiece is clamped on the JIM prototype with integrated mechatronic devices, the model analysis is done to characterize the mechanical systems properties with low pre-stress pressure and high pre-stress pressure.
Task 2.4 JIM Based Machining System Modelling and Experimental Validation
Main objective of this task is to develop model based design methodology for machine tool design although applicable in general but especially for JIM based machine tools. The idea is to provide designers a reliable and efficient design tool for assessing the dynamic behaviour of the machine tools so that they can design the machines with required dynamic performance. This approach will move the expensive retro-fitting of vibration control solution at the user facilities to inexpensive and efficient integration at the de-sign stage.
The JIM interface characterization is an important requirement for this task and is being met through the characterization work being done in Task 2.2. The interface behaviour models will be implemented in the design and analysis of JIMs in virtual environment. In the work done so far, representative models of the subject machines have been analysed through FEM. The dynamic behaviour of these models is being tuned to simulate behaviour closest to the experimental results. Figure 22 & 23 show examples of FEM analysis of the AFM turning machine and the vertical plate for the Soraluce machine. Similar analysis has been carried out for the Soraluce case. In the next phase of analysis these HDIs will be integrated in the models and the dynamic performance of the machines will be analysed for different boundary conditions. In parallel with the FE modelling of subject machines and model based design methodology is being developed. This activity resulted in deliverable D2.3 which was completed in month 20 of the project.
The tasks in the work package WP2 were completed in due time and the mechanical JIM concept was suc-cessfully proved. Some changes in the distribution of tasks between partners were necessary since an im-portant part of the project was focusing on implementation of PoPJIM concept in an industrial turning machine.
Static and dynamic capability of the machining system is determined by the interaction between machine tool and cutting process. In this respect, the control of machining system stability can be achieved by changing dynamic parameters of the machine tool, the process or both. The traditional way to control machining system dynamic behaviour is by controlling the process parameters, i.e. depth of cut, rotational speed, speed rate as well cutting tools’ micro-geometry and material. By this, static stiffness in the direction of cutting force and overall damping of machining system are improved.
Although being aware about the significant contribution of joints’ stiffness and damping to the overall capability of the machining systems, the classical theory of the machining system lacks a unified concept for consciously designing structural interfaces with controllable characteristics. Apparently, there is a tendency today both among scholars and manufacturers to develop and implement complex schemes for on-line monitoring and control of machining systems. The simple explanation is the existing knowledge gap between machine tool manufacturers and machine tool users regarding static and dynamic capability. Due to unlimited combinations of tools’ geometries and materials, workpieces’ shapes, dimensions, and materials, fixtures and toolholders it is impossible to predict the behavior of a machine tool and by this the accuracy of resulted parts. The consequence is that the machine tool users are forced to add advanced sensor systems for monitoring and controlling the machining system. In an industrial environment these solutions are costly and complex and not reliable due to adverse conditions. PoPJIm concept has proven the fundamental relationship between the conditions of machine tool joint inter-faces and the process static and dynamic stability.
Work package WP3: JIM Distributed Control
Adaptive control architecture was formulated at concept level, implemented at the integration level and tested on the AFM TAE 35N machine at the machine level. The system identification, optimization and control algorithms applied at various stages of the control architecture were then validated with respect to the machining tests on AFM TAE 35N machine. The sensing and actuating systems were developed according to the technical requirements requested by WP2 in such a way that the control architecture can be realized on the AFM TAE 35N machine at the functional level. In Work Package WP3, the following objective was to be achieved in the last reporting period i.e. from 22.10.2013 till 31.08.2014
• JIM Embedded Control: The control architecture for the reduction of unwanted vibrations which was developed earlier with respect to one active JIM has to be adapted for 2 active JIMs such that it will enable them to converge to a global optimum. The following tasks were to be performed in the last reporting period to achieve this particular objective in practice:
• Integration of wireless hardware (developed by WP4 for process monitoring stage) with the JIM global control (Master Controller Node) hardware
• Machining tests on AFM TAE 35N machine with active JIM to validate the process monitoring stage, JIM global control, JIM local control stages of the developed control architecture
• Integration of the master controller node (MCN) hardware with the communication support platform (CSP) via general purpose wireless network node setup (developed by WP4 for communication be-tween MCN and CSP)
• Re-design and development tests of amplified pre-stress actuators (APA) and the improvement of displacement signals obtained from eddy current sensor (ECS)
Task 3.1 Control architecture defines the features, functional and non-functional properties of control
Task 3.1 focuses on developing the control architecture for the self-adaptive control of mechanical JIM’s when integrated within a machining system. It defines the software as well as the hardware interfaces to WP2 and WP4 and also the kind of sensors and actuators which were considered at the initial stages. This Task 3.1 is of interest to all project partners in particular partners from WP2 and WP4 and within WP3.
The control architecture of Task 3.1 consists of primary and secondary stages of control. In the primary stage of control, a global master controller will acquire the incoming vibration signals with respect to wire-less communication medium and estimates the JIM parameters like operational damping, Eigen frequency and optimal pre-stress. Whereas in the secondary stage of control, a local controller acquires the incoming optimal pre-stress from the global master controller and provides the appropriate actuating signals to the Sensing and Actuation systems. So that, the actual pre-stress measured by the sensing and actuation systems tracks the optimal pre-stress in the presence of external disturbances. Here, the computed optimal pre-stress is regarded as a set point for local controller.
Task 3.2 Sensing and actuating systems develops an actuation system which can provide sufficient force capabilities to cover JIM pre-stress bandwidth
Task 3.2 focuses on developing the sensing and actuation systems to provide the mechanical input to pre-stress the JIM’s. This Task 3.2 is of interest to all project partners in particular partners from WP2 and with-in WP3. Actuators of Task 3.2 are Amplified Pre-Stress Actuators (APA). The APA was designed to introduce damping capabilities and sealed ceramics to protect the piezo-ceramic from the critical machining environment. The APA was designed with 5 layers of 10μm of thick VEM to prevent the force shunting of its damping interface.
In the previous integration attempts, the APA was tested on the AFM TAE 35N turning machine included with a single JIM. In practice, the force generated by the APA was not sufficient enough to apply a mechanical effort to pre-stress the JIM on the AFM TAE 35N turning machine. The sensor signal was relatively low to be applied as a feedback signal for the local controller. It was decided to perform the re-design and complete testing of the APA along with the improvement of ECS. Now, the re-design and the complete testing of the APA along with the improvement of ECS are under progress. According to Task 3.2 Deliverable D3.2 “Sensors and Actuators for JIM” was prepared. At present, the deliverable D3.2 is being checked by the project consortium.
Work package WP 4: Multi-hop wireless network
WP4 provides networking solutions for the PoPJIM. The developed software need to be provide reliable networking in presence of high amount of metal (machines, walls, machined components, etc.) that adversely influence the radio signal propagation due to increased electromagnetic noise. These effects reduce quality of wireless communication by increasing packet losses and end-to-end communication de-lay, but may also affect overall PoPJIM system configuration developed in Task 4.3.
The research and development in WP4 are performed in three tasks. Task 4.1 developed network that connects machines among themselves, and machines to the configuration support platform. We have developed a low-cost, flexible and maintainable network that allows remote management of any number of machines that may be deployed anywhere in factory floor. Task 4.2 has developed a real time, high speed vibration monitoring of the machine that allows control of the JIMs within the machine.
The research and development within tasks has progressed as planned and all R&D tasks were completed in time. The main remaining effort in the last year of project will be integration of solutions developed in WP4 with work of other work packages. In project deliverable D4.3 we have presented detailed integration approach that will allow us timely complete the demonstrator. It is planned to perform the integration pro-cess in a set of successive integration workshops that will be used for detection of issues and incremental adding of new functionalities. The approach offers the possibility to the project partners to integrate and test their solutions in real factory environment, on the demonstrator machine. If issues are identified, they will be resolved by the next integration workshop, which will include testing of fixed features, and addition of new ones. We first present network architecture, and then proceed with detailed reports on activities and results of each of the tasks. Each tasks first summarizes the main achievements, and then proceeds with the detailed description of activities and achievements.
Network Architecture
The main architectural and technical challenge of the DWCN is that it has to accommodate two very distinct communication cases:
• Continuous, synchronized, high data rate, short range, low delay communication between the vibration monitors and the MCN, and MCN and JIMs. For this purpose we implemented a dedicated wireless sensor network (WSN) and associated Medium Access Control (MAC) protocol that is utilized to control the machine. The network supports sampling rates of the vibration sensors is in range of 5kHz and nodes are synchronized with accuracy of approximately 10 μs.
• Non-real time, long range, reliable communication between the Master Controller Node (MCN) and the Configuration Support Platform (CSP) through the general- purpose wire-less network (GWN). To support this, GWN provides stable and reliable routing in multi-hop wireless network, low maintenance and ease of configuration, and low wireless signal interference with the WSN.
The network deployment diagram is shown in Figure 44. The diagram shows components that comprise the DWCN and their interconnections:
• WSN nodes and software operate on ARM Cortex M3 based hardware running custom embedded software (Fast-MAC) developed by Inertia.
• MCN consists of two physical nodes: xPC-Target performs real-time machine control, xPC-Host provides GUI and communicates with the CSP.
• GWN hardware nodes are wireless routers running embedded Linux on Broadcom MIPS processors. They are also four-port Ethernet switches. They execute customized multi-hop routing extensions and configuration solutions software.
• CSP runs on x86-based servers running Microsoft Windows and specialized software (OPC Matricon, Matlab, MySQL).
Interfaces between subsystems in the project are realized by various technologies, depending on the capabilities of devices and implementation preferences:
• In-GWN: Proprietary network routing and management protocols on top of
• WLAN (WiFi), visible as fully-fledged IP network by users of GWN
• In-WSN: Inertia proprietary FastMAC
• Vibration sensor – wireless node: ICP (Integrated Circuit - Piezoelectric)
• MCN ↔ WSN Gateway: Inertia Serial Framing Protocol over Ethernet
• WSN Gateway ↔ JIM: Inertia Serial Framing Protocol over UART
• MCN – CSP: OPC through GWN.
Task 4.1: Multi-hop Networking and Network Topology (leader: UBER)
This task has developed reliable routing (networking) protocol and the network configuration and management services. The key developments in this task include:
• We have procured wireless hardware nodes and installed embedded version of Linux on them for further work. The nodes prepared in this manner were used as deployment platform for the routing and configuration software that we devel-oped in course of project.
• We have modified the existing OLSR 1routing protocol, so that it provides more reliable and efficient traffic routing in wireless multi-hop networks.
• We have provided a complete network management and configuration solution. The solution is based on standard protocols, with extensions that allow efficient network reconfiguration in the harsh industrial environment.
• The solutions were tested in factory of of one of partners. The routing pro-tocols showed performance improvements in comparison to its standard version. The network configuration operated as specified and was extensively used during the performance and reliability tests of the routing protocol.
Detailed description of the task
General Purpose Wireless Network provides connectivity between MCNs and the CSP. The main requirement is that GWN has to support use of OPC for communication between MCN and CSP. The GWN communication stack is executed on GWN nodes. The layers of communication stack that are addressed in PoPJIM are presented in light blue in figure that describes the architecture of the system. Networking layer is responsible for the address resolution of nodes and forwarding of packets to destination nodes. In case of GWN its focus in on reliable and efficient management of routes in multi-hop wireless communication between all communicating parties attached to the GWN. The network exploitation and maintenance functionalities need to be provided as well in order to allow the developed technology to establish foothold in practice. Due to core nature of PoPJIM as a Plug’n’Produce system, the components of PoPJIM system need to have capability to be dynamically reconfigured. This includes the GWN, so we develop reliable solution for the (re)configuration and management of GWN.
The distributed wireless control and configuration network will be deployed in a production environment of varying size (with regard to number of nodes and machines involved as well as to the physical area covered by the factory floor). This large area that needs to be covered is used for machining that affect the wireless signal propagation and reduce the communication range of wireless nodes: numerous walls of reinforced concrete, large metal machines and workpieces, electromagnetic noise created by working machines.
For these reasons, it may not be possible to provide direct (single-hop) wireless communica-tion between MCN and CSP. To overcome this issue, we develop a multi-hop GWN. A pair of nodes that want to communicate and that are not in direct wireless communication range needs assistance of other nodes in the network to forward their data, hop by hop, until packet reaches the desired destination. In order to provide sufficient reliability and performance of network, it is important to identify the best route (ordered list of nodes that will perform packet forwarding) between the communicating pair of nodes.
The network is run by programmable embedded systems: the LinkSys series of Linux WLAN access points. Nodes have Broadcom BCM5352 200MHz processor, WLAN and integrated 4- port 100Mbit Ethernet switch. They have 16 MB of RAM and 4 MB of Flash memory. They provide a stable Linux kernel, open drivers for the WLAN hardware and a customization op-tions for the networking layer.
We base our routing solution on the OLSR routing protocol, and we improve it to fit the needs of the PoPJIM project and industrial environment. It is a proactive routing protocol, where nodes exchange the locally obtained link information in order to build the global topology knowledge at each node in network. This topology is used by network nodes to determine the routes.
The OLSR protocol provides a satisfying solution with regard to its usual operation, but in course of project we have identified one weakness of it – its link quality estimation is inflexible and may not operate adequately in industrial environment (details and impact of it can be found in project deliverable D4.32). Inaccurate link quality estimation leads to sub- optimal route selection, unnecessary packet losses, and to issues of applications that use the network.
There does not exist a consensus on the best link detection algorithm and its parameters in re-search community. An example of behaviour of different link estimators on a dynamic link such as those in industrial environment is shown in Figure 4.3. Simple moving average SMA(10) estimator has inherent large error and is rather chaotic. SMA(50) and Exponential Moving Average - EMA(0.4 10) are rather accurate. EMA(0.1,30) performs very well when the link quality is changing slowly, but due to its slow reaction time, its error increases greatly towards end, when the link quality promptly changes. Obviously, their usability will depend on the use case scenario, as we have also shown in previous related work3 through rigorous stochastic modelling of a class of link detectors.
This implies that if an inadequate link detector is used in protocol, the protocol will not operate correctly – the errors in individual link estimation will accumulate and the protocol will choose sub-par routes. To correct this inflexibility, we develop an adaptive approach to link detection:
• We perform online evaluation of effectiveness of multiple link detection approaches and select the best one. This removes the issue of pre-selection of approach without feedback on its performance.
• The link detector algorithms are selected on link basis (each link has its custom detector), not on network basis (all links use the same detector and parameters). This further reduces the
• Errors in link state estimation process and improves accuracy of routing.
We have tested this approach on the set of data collected in Berlin’s large wireless multi-hop network that contains more than 300 nodes4 and on data collected in GWN testbed that we operate in the Institute for Computer Science at UBER. Out of the available data, we use 20% of lifetime of a link for the learning phase, to select the best link-state estimator. The remaining 80% of data of a link is analyzed as production phase where we apply the selected estimator and calculate its MSE on the data that is not encountered before. Our approach is consistently better than all pre-determined, network-based approaches. Its MSE is very close to the MSE of optimal approach (we define optimal as a single type of detector is used over whole link lifetime so that the MSE is minimal. Optimality comes from a priori knowledge of link behaviour - for this reason optimality is not reachable in practice) in all analysed cases. It is some 10 to 20% more accurate than the use-case specific best estimator, and within 10% of the MSE of the optimal case (Table 2 in project deliverable D4.3).
However, such learning approach that uses only the initial set of measurements (20% of avail-able measurements) for the decision making process is also static in its nature and cannot adapt to later changes in link and network behaviour: if conditions change, network will remain stuck with the type of detectors that was selected in the first (and only) learning phase, leading to inflexibility and issues.
To offset this issue, we have implemented continuous learning approach for link-state estimation. A node starts with an arbitrary link estimator. It collects data that is delivered by the detector for 10minutes (free parameter that can be configured) and then executes the de-scribed algorithm for estimator selection on the measured data. Link estimator is set to new type and collects its data for the next 10 minutes, after which the whole adaptation process re-peats.
The results of the continuous approach are very good – we were able to considerably reduce the MSE to approximately half of the previous learning MSE. It is even considerably better than the static optimal approach.
The approach is fully implemented in GWN and has two components. The described learning and decision-making component runs on the GWN configuration server. The learning component is implemented in Java, and is thus platform independent. Once it reaches decision, it sends commands to GWN nodes via UDP. We have developed an OLSR plug-in soft-ware component that replaces the standard OLSR link-quality estimator plug-in. It accepts commands and can switch between different estimators.
Network Monitoring and Management
GWN, as essential part of the PoPJIM concept, will be deployed in manufacturing companies where dedicated (wireless) network administrators will not be permanently available. Once PoPJIM enters deployments in production facilities, it will be used on multiple machines, for prolonged periods of time when system configuration and management becomes the major ac-tivity from the network perspective and one of detrimental factors for overall usability of the system. We improve system usability by providing monitoring and management facilities in network that are suited for the developed GWN and PoPJIM concept.
During system’s use, any of networked components may need replacement, it may be moved to another machine (i.e. attached to a different GWN), or machine operator may switch Ethernet ports to which a networked subsystem is connected to. For instance, a failed node is replaced by operational one. These common, operational changes in PoPJIM network Each GWN node has its own unique IP address assigned to its WLAN interface. This is the only fixed and necessary parameter that has to be set in GWN prior to its deployment (other-wise the GWN routing will not function). All other GWN node and network configuration parameters can be set during network lifetime.
Each GWN node has its own IP /24 sub-network for assignment of addresses to Ethernet hosts attached to its internal Ethernet switch. Each GWN node runs its own DHCP server that as-signs unique IP addresses to Ethernet hosts that are connected to one of the four available Ethernet ports of the node. GWN configuration server remotely manages and configures the Local DHCP parameters, such as the IP sub-network to be assigned to Ethernet ports, default gateway, default DNS, DHCP lease time.
We have developed configuration solution based on SNMP SET commands and the “extend” options in the SNMP daemon that runs on GWN node. The needed SNMP extensions are de-veloped in C language net-snmp-API for AgentX5 . They can be easily recompiled and deployed to other nodes than those that we currently selected for GWN implementation. The configuration scripts are implemented so that they can run on every Linux system, further ensuring the cross-platform portability of the developed solution. The configuration changes are saved by the GWN node in its configuration files, so even if GWN node is restarted, its configuration will be valid.
For purpose of name resolution, we have developed a SNMP TRAP extension that monitors the local DHCP configuration changes. The SNMP TRAP changes both the GWN configura-tion server and the local SNMP information base, so any authorized SNMP client can ac-cess the correct information. GWN Configuration Server accepts the SNMP TRAP (SNMP Master Component in Figure 4.4) parses the information from it, creates and executes (re)configuration scripts for its DNS server. As the result, the new IP address of the host that has just been connected to the GWN network has its entry in DNS server, which allows the resolution of its name to its IP address. Opposite is also valid – if an Ethernet host has been disconnected from the GWN node, its entry in DNS is removed.
must not impact the system functionality and this requirement renders any sort of fixed, pre-determined network and host configuration infeasible.
When a node (such as xPC-Host) is connected to the GWN, it must obtain a unique IP address to be able to communicate within it. An IP address is not permanently associated to a node. A node may obtain another (again unique at the time being) IP address at any time during its lifetime in system. Since it does not have a fixed IP address, a node should be addressable by its symbolic name. For this purpose, there must be a system that dynamically resolves names to IP addresses. Since IP addresses of nodes change, they must be accessible by their symbolic name (host name). There must be a system that resolves names to IP addresses.
These requirements are not new, or specific only to the PoPJIM system. The Internet relies on the same principles of configurability and provides tools that provide for these requirements. However, the PoPJIM system has some specifics that are different than in a typical enterprise Ethernet LAN. These specifics need to be taken into account while developing the network configuration subsystem:
• The (re)configuration traffic in GWN should be minimized. GWN, same as every system, is particularly sensitive to loss of reconfiguration data, so the less recon-figuration data we sent through unreliable GWN, the higher overall reliability is achieved.
• The machine should be able to continue operation even if GWN experiences an outage (for instance a GWN gets partitioned due to transient node failures).
• GWN nodes have available resources that are by orders of magnitude lesser than in case of their server counterparts used in enterprise LANs. This affects both available tools and design decisions that we make in the development.
To address these requirements, we have developed a hierarchical, loosely coupled configura-tion system (Figure 4.4). A centralized GWN configuration server is used for overall network management and configuration of GWN nodes. It prevents local configuration errors through remote configuration and monitoring of GWN nodes. Host configuration is performed directly by GWN nodes.
Our solution is based on well-established technologies and standards and guarantees interoper-ability with existing systems. This considerably reduces the network maintenance costs. We have customized existing software to fit to PoPJIM needs, and to capabilities of GWN nodes almost instantaneous updates in the network configuration, and at the same time it generates minimal traffic in GWN – the traps are activated only when it is necessary.
Experimental Evaluation of GWN in AFM Workshop
We have performed a GWN tests in AFM factory in Andrychov, Poland. The factory consists of three joined workshops (Figure 4.6) with numerous machines of different sizes. The factory size is 124 x 76m. We placed nodes throughout the factory to create a connected GWN (Figure 4.5). GWN network consisted of six nodes. Two types of measurements were per-formed, one for TCP traffic performance, to emulate behaviour of OPC RPC calls, and one for tests of the network control and configuration mechanisms. We analyse two principal cases: one that uses the default link detectors from OLSR and one that uses our adaptive approach. We first execute the default OLSR approach, and then use the collected topology data to execute adaptive link detection. In order to eliminate effects of transient wireless noise on measurements, we used a measurement tactics where we interleave standard OLSR link detectors and our adaptive approach. The order of execution and marks of measurements are: A: SMA(32), B: adaptive, C: SMA(32), D: adaptive.
Each measurement set consists of 50 samples for each of the measured categories (i.e. 50 samples of 5KB message size, 50 of 10KB message size, 50 of 100KB message size and 50
SNMP SET tests). For evaluation purposes we combine then measurement results from sets A
and C, and from sets B and D.
The measurements are heavily influenced by the surrounding levels of signal noise, as well as the internal TCP states, so there exists a large dispersion of measured values (a sample can be seen in Figure 4.7). In order to remedy large dispersion, we use median instead of the mean, as it is common for distributions with numerous outliers. Median is much more robust metric and intended for use in such scenarios.
The medians of all experiments are shown in Figure 8, and it can be seen that our adaptive ap-proach results in a slight but constant advantage over the default link detection in TCP transfer evaluation cases, and a considerable advantage for the SNMP-SET tests. It must be noted that although the overall results of the adaptive approach are better, in some experiments we have observed cases where the default link detection approach delivered better TCP transfer times. Obviously and as expected, there is a strong overall dependence between measurement results and current background noise on wireless, compounded with the long-term effects of internal TCP states on the flow performance. In all experiments is our approach provides a clear and notable advantage for the stateless forms of communications, such as the SNMP-SET commands where our approach results in consistently better reaction times than the default link detection approach.
Summary
We have developed routing protocol extensions and network configuration toolchain. Unlike many approaches from research, our solutions are fully operational and deployable. We have tested them in experiments in AFM facilities. The experiments have demonstrated that the de-veloped solution is reliable and robust (there were no node or software glitches in two days of experiments) and that our additions to the existing stack result in performance improvements.
Our initial notions about networking in factory environments were confirmed:
• There is no readily available Ethernet/WLAN connectivity at factory floor. The only Intra/Internet connected part of the factory are offices, which are separated from the workshop floor.
The wireless coverage from the office space is far from sufficient to cover the factory: our WLAN stations were running on full power but due to numerous obstacles in form of machines, walls and massive metal workpieces, we needed
six nodes to cover approximately 30 to 40% of the factory, with moderate to very bad connectivity with the other factory workshops.
Wireless multi-hop networks remains as the only viable and financially acceptable solution for Intranetworking in such environments. Any attempt at installation of Ethernet cables to the workshops would induce downtime of machines creating production breaks of prohibitive cost. With only six nodes we were able to cover 30 to 40% of the factory floor and provide Intranet for devices and machines that otherwise would not be able to access it. With a modest in crease in node number it is possible to provide full network coverage at a very low cost: the price of as single node is below 50€, so for less than 1000€ allows full Internet access.
Task 4.2: Wireless communication component (leader: Inertia)
The key developments in this task include:
• A high-speed wireless solution was developed, that fits the PoPJIM use case re-quirements (high-speed 5KHz data rate, synchronized among multiple points).
• Inertia's initial hardware platform was completely redesigned. New hardware is based on Cortex A3 processor, it has IP67 casing.
• Ethernet gateway was developed from scratch, to interface with the MCN developed in WP3.
• Inertia's real-time operating system ported on the new platform, Inertia's wireless FastMAC protocol ported to the new high-speed radio, integration with the Inertia ProMove GUI software for data acquisition. It is the fastest wireless sensor network platform currently on the market.
• A set of experiments was done in an industrial workshop in order to assess the wireless performance. The experiments showed a good overall performance, with the warning that instances of burst packet loss may occur.
• The integrated hardware platform was presented at several exhibitions (Hannover
Messe 2013, IDTechEx 2013, High-Tech Systems 2013, Precision Fair 2013.)
Detailed description of the task
The WSN hardware incorporates the necessary electronics, including microcontrollers, memory, radio, antennas, data interfaces, and casing. Software implements the Inertia Fast-MAC protocol that regulates the access to the wireless medium and data communication within the WSN as well as interfacing to systems with which the WSN interacts.
The extremely high sampling rates required controlling the AFM TAE-35N turning machine required introduction and use of a proprietary high speed 2.4GHz radio transceiver. that can communicate up to 4Mbps data rate in FSK or GFSK modulation. The maximum range/coverage for line-of-sight conditions is approximately 30 meters. The radios operate in the license-free 2.4GHz band. The main hardware subcomponents of the wireless nodes are:
• ARM Cortex M3 microcontroller implementing the application logic (sampling vibrations sensor, communicating with the MCN or JIM Local Controller) and the WSN MAC protocol. Memory/storage support.
• Interface electronics for the ICP accelerometer sensors.
• Radio transceiver and antenna
• Due to harsh environment in which nodes operate, the node casing was chosen to be according to IP67 standard.
Figure 52 shows the prototype of the WSN node with a modular design: the motherboard contains the microcontroller, radio part and storage support, while the daughterboard provides the electronic interface and powering circuitry for the ICP accelerometer sensors.
• The MAC layer is based on the proprietary Inertia Fast-MAC. This is a Time Division Multiple Access (TDMA) MAC protocol, optimized for accurate sensor sampling synchronization and timely transfer of sampled data. TDMA prevents packet collisions among communicating parties (each sender knows when to send and others are silent in this time period) and it introduces a known and fixed delay in communication (each sender has a fixed slot in a frame, so delay between two successive transmissions equals to the duration of the frame). Both of these features make it a perfect choice for real-time control application, such as in the PoPJIM project. Finally, nodes are tightly synchronized and they are able to perform accurate time-stamping of the data exchanged in the network, which is needed by the MCN control algorithm.
• Developed MAC protocol is used for the process monitoring (vibration sensing). Process monitoring comprises of two principal steps. The first is vibration sensing which is performed by wireless nodes placed on machine structure over the accelerometers that are attached to them. The sensors provide analog signal over ICP connector to the wireless node and this signal is converted to a digital value, suitable for transmission through WSN. According to requirement specification delivered by WP3, the vibration data resolution is set to 12 bits. The sensor and the A/D converter capable of sampling at 5kHz rates, in order to capture the full span of vibration signal and its later reconstruction at the MCN. The accelerometer is external to node, and connected to it by a cable. Such configuration al-lows that sensor is placed closer to the machining process, without possibility of wireless node damage by the machining process, cooling liquid, and flying debris.
• A wireless node assembles the vibration samples in a packet. As soon as data payload of pack-et is filled, the packet is queued for transmission. Once the time slot of the wireless node is due, the packet is immediately sent to the WSN gateway, attached to MCN, using the Fast- MAC. The gateway then provides the vibration data to the MCN by Inertia Serial Framing Protocol.
Inertia High-Speed Wireless Platform Experiments in an Industrial Workshop
• The objective of the experiments is to assess the overall performance of the 4 Mbit high-speed radio used for the Inertia WSN. The experiments are performed in an industrial workshop environment (machine tools, metallic structures, people working with the tools and machines).
• Three different situations are investigated: line-of-sight, small obstructions (e.g. tools trolley between gateway and node) and large obstructions (e.g. machine tools or walls between gate-way and node). Each setup is tested over 3 different distances: 2, 5 and 10 m. The wireless node is sampling and communicating data samples at 9.6kHz. The experiments lasted approximately 1200s per scenario.
• Two additional tests are made with the wireless nodes placed on machine tools:
• The node is placed on top of an upright drill. The drill is not running during the test.
• The node is placed on top of a large milling machine running at 1200 rpm.
• For each test, the average packet loss percentage is computed. The packet loss is also translated into time gaps between two consecutive samples. Ideally, without packet loss, the time gap between two consecutive samples should be 1/9600 = 0,0001042s. The distribution of packet loss / time gaps over time was determined for each test. From it, we derive the mean packet gap and the standard deviation of it. Figure 4.11 shows the results of the experiments for nodes placed on various places in the workshop.
Additional tests were made with nodes placed on the machine structure, a case that is more relevant for the PoPJIM project. The experiments were run for approximately 60minutes, to observe long-time behaviour of the developed WSN MAC protocol.
The experiment setup can be seen in Figure 4.12. The nodes are approximately 3m distant. The packet loss remained low (0.434%), with mean packet gap of 0.0001046s. Figure 4.13 shows the histogram of the time gaps. It is important to note that the histogram is represented at logarithmic scale. Ideally, all values in the histogram should be grouped at the default value of 0,0001042s. The same figure shows the packet loss per second, as the test progresses over time. In the second experiment, the nodes were placed on operating milling machine, also on distance of 3m. The results remained similar, with very low packet loss of 0.217% and mean gap between packets of 0.0001044s.
.
Task Summary
The experiments indicate a relatively good overall performance, with less than 3.5% packet loss on average even in settings with large obstructions between the gateway and node. The behaviour is stable over larger periods of time. The operating conditions of the machine tool (with the wireless node placed on the working machine tool) do not seem to affect the wireless performance. However, there are a number of important remarks and findings that have to be carefully considered with respect to the possible adverse impact of the wireless over the control performance.
Firstly, the packet loss appears to vary with distance in a counter-intuitive manner when obstructions (both small and large) are between the gateway and node: the performance is better for longer distances. This is not the case in the clear line of sight case, where the packet loss increases with distance. A likely explanation is that obstructions placed close to either the gateway or node tends to limit the possibilities for reflected propagation of the radio signals. Secondly, it occurred in our tests that placing the gateway on a higher, non-metallic support improves the overall wireless performance.
Finally, it is important to analyze the statistics (histograms and distribution over time) of the packet loss. In some tests, even if the average packet loss is low, there are instances where a burst packet loss occurred. This means that time gaps of additional 100ms or possibly more may appear, where data is not received from the sensors. The controller needs to cope with such unpredictable situations.
Task 4.3 Configuration Support Platform
Configuration Support Platform, CSP is an enabler for the Plug and Produce property of a modular machine tool with joint interface modules. This document describes the use cases, architectural features and the implementation of a CSP. The document defines also the functions and features of the interactions to the DWCN to serve the control system and for configuration/reconfiguration tasks in WP2 and WP3. The functional architecture of CSP is composed of the Module library, CSP host and the communication interface.
The module library is the repository of the relevant module specification and configuration knowledge. The module specification should include the
• Basic module data such as JIM components both mechanical and electronic, the interface geometry (flat, tapered, etc) and assembly of the JIM, CAD models, etc.
• BOM, supplier of the module or component information need to be captured. The specifics of these are also depending on the business model used to acquire or use the JIM and its components.
• The JIM characteristics, i.e. the profile of the JIM displacement or force obtained from the JIM as the function of the applied pre-stress value.
• Considering future extension of the application to support reconfiguration of a reconfigurable machine tool, additional JIM characteristics such as module trans-formation matrix, stiffness matrix, compliance matrix etc need to be included.
Configuration knowledge including the dynamic behaviour of the configuration and selection rules to identify the best possible JIM. The dynamic behaviour of a given configuration where the particular JIM has been installed in the machine structure. These behaviours include the updated profile of the operational damping and the Eigen-frequencies as the function of time. The module Library is implemented in MYSQL database with the following data model.
The CSP host is the hub of CSP and provides the following front and back end services to the user:
• Front end services: via a user interface, the user can input the queries as specified in the use case descriptions above, get visual the required information to enable a selection decision or analysis.
• Establish communication with the repository for a reliable and secured communication with the repository
• Establish communication with MCN via the wireless network DWCN. Invoke the configuration rules and specific query functions (as described in section 3.4.2 to execute them.
The CSP Host is implemented in Matlab that provides not only a secured communication with the Module Library but also provides utilities and toolboxes for communication with the control system and the DWCN.
The communications interface with MCN is via the DWCN as stated above. CSP and DWCN are connected over Ethernet, same as MCN and DWCN. The communication interface of CSP should support reliable and secured communication between CSP and MSN. For data exchange, MCN and CSP use OPC communication middle-ware.
A user interface is also developed on the host for an offline access and analysis. The figure below shows the graphical interface for this purpose.
Following the architecture and the specifics of the chosen implementation application tools and architectural components, the Configuration Support Platform is implemented.
The communication integration between the controller (MCN) via the wireless net-work is achieved and for this the OPC protocol
Populating the module library, the knowledge model as well as the machining sys-tem configuration has been done. Three main features of the CSP architecture can be summarized as follows.
1. Open architecture: The CSP architecture is inherently open, i.e. any one of the modules can be upgraded or changed whenever better ones are available for use.
2. Extensible: The implemented architecture is extensible in its functionality, i.e. analysis can be added either by making use of the toolboxes in Matlab or integrating the analysis applications.
3. Capable of accommodating different business models. Acquisitions of the modules can be effected in different ways depending on the business model between the module supplier and the user. The model library, as implemented here, can be accessible to the suppliers over the internet so that they can add and update module’s specification.
Work Package 5 Demonstration
Contributions made by the partners with respect to the above mentioned objectives are the following:
• KTH >> Technical expertise for performing machining tests in order to acquire acceleration measurements of AFM TAE 35N machine included with active JIMs and to obtain experimental results with respect to real-time system identification and optimization algorithm tests. Also provided invaluable feedback on deliverables “D3.1 JIM Control Architecture Specification” and “D3.3 JIM Embedded Control”.
• INERTIA and HUB >> Technical expertise for the integration of wireless com-ponents of process monitoring stage with the control hardware. Also provided invaluable feedback on deliverables “D3.1 JIM Control Architecture Specification” and “D3.3 JIM Embedded Control”.
• HUB >> Technical expertise for the integration of wireless nodes (developed specifically for the communication aspect between JIM global control and communication support platform) with the control hardware. Also provided in-valuable feedback on deliverables “D3.1 JIM Control Architecture Specification” and “D3.3 JIM Embedded Control” and technical expertise for the integration of Inertia’s wireless components (developed for process monitoring stage) with JIM global control.
• AFM >> Technical expertise and invaluable time for performing machining tests on AFM TAE 35N machine.
• CEDRAT Tech. >> Technical expertise for the integration of Amplified Pre-Stress Actuators (APA) and Eddy Current Sensors (ECS) with the control hard-ware and AFM TAE 35N machine. Also provided invaluable feedback on deliverables “D3.1 JIM Control Architecture Specification” and “D3.3 JIM Embedded Control”.
• SORALUCE, S. Coop., >> Provided the basic hardware requirements for the setup of control hardware with AFM TAE 35N machine regarding the preparation of demo for Final Review meeting on 12.09.2014.
Task 5.1 Mechanical JIM Demonstrator
Damped Tooling System
Modal testing of the tools clamped in HSK100 tool holder indicates that the EDS coating material can improve mode damping of the structure without sacrificing its static stiffness. Machining tests were performed to determine the experimental stability lobes diagram for both the coated tool and conventional tool.
The machining results indicated that the damping coating can improve the critical depth of cut, from 0.2mm to 0.3mm when slotting in X direction and from 0.15mmto 0.3mm when slotting in Y direction. In total, the coating almost doubled the cutting capability of the process.
Important Note: The experimental results of each stage of control architecture correspond to the AFM use case (i.e. AFM TAE 35N machine included with standard turret and active JIM located at its tailstock side). Active JIM is a mechatronic component actuated using the Amplified Pre-Stress Actuators mounted along the walls of tailstock holder.
Demonstrator IDEKO/Soraluce
Mechanical realization
From previous investigations, it was concluded that the JIM approach was not appropriate for the low frequency structural modes damping and, for that reason, the AIM approach was posed instead. The AIM approach keeps the same concept as the JIM inter-face, as a stand-alone added module on the machining system that increases dynamic stiffness, is wireless and has a self-tuning capability, but the principle for dynamic stiff-ness maximization is based on the dynamic absorber effect.
As it was explained in the previous periodic report, two different designs have been carried out. The first one is a preload type damper, in which the stiffness is adjusted by a preload increase of a viscoelastic material. The preload is achieved by turning a two-direction screw, which in turns acts on a wedge that moves an especially shaped preload plate back and forth. This plate pushes or releases the viscoelastic material which surrounds the moving mass. In this case the damping is provided by the viscoelastic material itself.
The second design is a spring type damper, in which the stiffness is adjusted by turning a variable stiffness spring at a determined position. As the moving mass is connected through bearings to this spring, the natural frequency of the system is changed by turning the variable stiffness spring. The damping is provided through eddy current effect in this case. The eddy current effect is achieved by adding specifically located magnets to the moving mass. When the moving mass starts vibrating, eddy currents are induced in several fixed copper plates, which in turns create a force opposing to the mass movement, adding damping to the system.
The spring type AIM is unidirectional and its movement is guided through linear bearings, whereas the preload type AIM is bidirectional and its movement is guided by sup-porting balls. The approximate mass of the moving mass of both AIMs is around 180kg,
First, these two prototypes were tested in their passive form, that is, with no self-tuning capability. The cutting tests were performed after tuning the system manually. Once it was observed that the system was working correctly and that the cutting capability could be increased to a large extent, the self-tuning capability was added.
Self-tuning capability
Control algorithm
In order to provide the system with the self-tuning capability, a control algorithm was developed. An accelerometer was used to monitor online the cutting process and wifi ports were used to enable the communication with a control algorithm in a external PC, which sends back a proportional command in order to tune the AIM accordingly and maximize the dynamic stiffness.
The control algorithm consists of a chatter detection method, followed by main frequency identification and a AIM tuning in order to match its frequency to the identified chatter frequency. The AIM is tuned through the viscoelastic material pre-stress in case of the preload type damper and the spring orientation in case of the spring type damper. The software user interface shows the tuning position and an online time and frequency monitoring of the acceleration signal.
Physical implementation
The control scheme described above was physically carried out by adding an upper floor to the passive prototype in order to enclose the sensors, actuators and communication ports.
The tool is a special tool that allows machining at low depth of cut and high feeds. Due to the lead angle of its inserts, the force is oriented in the axial direction. This is a good characteristic for the milling machine itself, since the axial direction is the stiffest, but it is a problem when machining on a flexible fixture in the axial direction. In this case, as the workpiece holding system is a vertical plate, the dynamic flexibility of this in the axial direction will determine the cutting capability of the process.
Cutting tests layout
As described in previous reports, the bending mode of the vertical plate in Z direction is the main mode limiting cutting stability. A 75% improvement of the dynamic stiffness magnitude is achieved at the critical mode. When the AIM is activated, the cutting tests are turned from stable to unstable, reducing the vibration level and improving the surface finish. Cutting capability is increased to a large extent. Without the AIM on the vertical plate, chatter problems started from a depth of cut of 0.5mm whereas when the AIM is integrated, a depth of cut of 2,5mm is machined in a smooth way, This means that the machine’s cutting capability is increased in more than 5 times.
2.5.2 Self -Tuning JIM demonstrator
Process Monitoring - Wireless acquisition of acceleration signal & system identi-fication algorithm
The acceleration signal of AFM machine is measured with respect to wire-less communication medium and the resultant operational damping and Eigen frequency characteristics are shown. The graphs denoted in red correspond to the unstable machining cut with maximum voltage applied to the Amplified Pre-Stress Actuators (APA) and the graphs denoted in blue correspond to the stable cut with minimum voltage applied to the Amplified Pre-Stress Actuators (APA).
In the unstable machining cut, we have three vibration modes ω - 1 at 1600Hz, ω - 2 at 900 Hz and ω - 3 at 403.3Hz. Out of these three vibration modes, 4ω - 403.3 Hz is the dominant vibration mode. The operational damping which corresponds to the vibration mode ω - 3 at 403.3 Hz is negligible which indicates the existence of chatter phenomenon. In contrast to the vibration mode ω - 3 , the operational damping which corresponds to vibration modes ω - 1 and ω - 2 are 15% and 19% respectively.
Whereas in the case of stable machining cut, we have only one vibration mode ω - 1 at 1600Hz. The operational damping which corresponds to the vibration mode ω - 1 at 1600 Hz is 15% on average throughout the operating time window of the machining process which is being performed. According to the operational damping characteristics which correspond to the vibration mode ω - 1 at 1600Hz, the chatter phenomenon didn’t exist.
Eigen frequency characteristics of unstable and stable machining cuts in a specific time window where the effect of vibration mode ω - 3 at 403.3 Hz is clearly visible on the resultant surface finish of the bars shown in Figure 16. From Figure 3.3 the resultant surface finish of the bars with respect to unstable and stable machining cuts is shown. In Figure 3.3 the top bar and bottom bar represents the surface finishes which corresponds to unstable and stable machining cuts respectively. Based on the light reflections on the bars, we can tell whether the resultant surface finishes of the bars are good or worse. It can be clearly seen that, the dominant vibration mode ω - 3 at 403.3 Hz has a deteriorating effect (where chatter phenomenon is significant in between time instants ω - 1=26 ω - and ω - 2=34ω - ) on the surface finish of top bar which corresponds to the unstable machining cut. Whereas, the surface finish of bottom bar is rather good which corresponds to the stable machining cut.
According to Figure 1, Figure 2 and Figure 3 the conclusion drawn from the experimental results is that the process monitoring stage is able to provide the acceleration signals with respect to wireless communication medium and compute its resultant operational damping and Eigen frequency characteristics in an accurate manner.
From Figure 17, the experimental result of 1D optimization tested on AFM TAE 35N turning machine is shown. For setting a global pre-stress optimum on the active JIM, Gradient Descent based optimization algorithm is adopted. On the left side of Figure 17, the graphs denoted in green represent the measured acceleration signal with respect to Wireless communication medium, estimated operational damping and Eigen frequency. On the right side of Figure 17, the graphs denoted in blue are the gradient of the estimated operational damping and the direction signal computed from the gradient of estimated operational damping and finally the graph denoted in green is the ‘pre-stress’ of the active JIM computed in terms of the voltage applied to the Amplified Pre-Stress Actuators.
Since the pre-stress of the active JIM is computed in terms of the voltage applied to the Amplified Pre-Stress Actuators, there is no need to enable the JIM local control. Ac-cording to Figure 17, the pre-stress parameter is directly dependent upon the direction signal. This direction signal is computed based on the gradient of the estimated operational damping. The direction signal tells the pre-stress parameter in which direction it should start to update. 1D optimization algorithm is initialized after 1s. The active JIM parameter ‘pre-stress’ starts its update initially from higher voltage value of 42V and converges towards optimal value of 0V within 7s. Based on the earlier unstable and stable machining cuts, the conclusion which can be drawn from the JIM global control stage is that the gradient descent based optimization algorithm is able to set the global pre-stress optimum on a single active JIM.
In Figure 18, the graphs denoted in red and green the acceleration signals, estimated operational damping and Eigen frequency characteristics with respect to unstable machining cut and 1D optimization algorithm enabled machining cut respectively.
From Figure 18, it can be seen that the Eigen frequency characteristics were slightly shifted which correspond to the 1D optimization algorithm enabled machining cut. This gives an indication that the pre-stress which is being applied on the active JIM is trying to avoid the structural modes of the machine tool-workpiece system while it is excited by cutting forces. This is clearly evident from the Eigen frequency characteristics computed for 1D optimization algorithm enabled machining cut compared to the Eigen frequency characteristics computed for unstable machining cut. Also the operational damping computed for the 1D optimization algorithm enabled machining cut which corresponds to the vibration mode ω - 1 at 1600 Hz is slightly higher compared to the operational damping computed for unstable machining cut which corresponds to the vibration mode ω - 3 at 403.3 Hz.
From Figure 20, the resultant surface finish of the unstable and 1D optimization algorithm enabled machining cuts is shown. In Figure 20, the top bar and bottom bar represents the surface finishes which corresponds to unstable and 1D optimization algorithm enabled machining cuts respectively. If we have a look at the top bar, the surface finish.
Master Controller Node (MCN) – CSP Integration via GWN (General Purpose Wireless Network) Nodes
The integration of MCN (which is responsible for JIM global control stage) with CSP using GWN nodes was performed such a way that the communication between MCN and CSP exists in a wireless network setup. In general, MCN is located near the AFM TAE 35N turning machine. On the other hand, CSP is located far away from the AFM TAE 35N turning machine. The following Figure 13 describes the experimental result which describes how well the communication aspect between MCN (JIM global control) and communication support platform (CSP) works.
In Figure 26, the acceleration signal of AFM TAE 35N turning machine (included with active JIM at the tailstock side), estimated operational damping and Eigen frequency characteristics with respect to 1D optimization enabled cut are shown. Here the opera-tional damping, Eigen frequency and optimal pre-stress (computed in terms of voltage applied to the APA) parameters are being sent from MCN to CSP via GWN nodes using OPC Write functionality. On the other side, CSP is acquiring these parameters using OPC read functionality and holds these parameter values in its respective module li-brary. From Figure 13, as we can clearly see that there is no loss of information between MCN and CSP during the machining process.
In addition to these parameters, the following specifications which are static during the machining process are also provided to CSP:
Run Number
Workpiece Specifications
• Diameter in mm
• Length in mm
Process Specifications
• Depth of cut in mm
• Feed rate in mm/rev
• Spindle speed in rpm
JIM Interface Specification
• Pre-load applied on tailstock JIM in N-m
Using the communication aspect between MCN and CSP, the possibility of adding or removing the existing JIM modules without having a negative impact to the overall control architecture in order to meet the high damping requirement requested by an end user (in our case end user is the machinist) before the start or after the end of a machining process is realized is rather rough (chatter phenomenon is significant between time instants ω - 1=26 ω - and ω - 2=32ω - ) coinciding with the dominant vibration mode ω - 3 at 403.3 Hz.
To a certain extent the surface finish of the bottom bar is smooth with respect to the 1D optimization algorithm enabled machining cut compared to the surface finish of top bar which corresponds to the unstable machining cut. During the 1D optimization algorithm enabled machining cut, the force window which is available from the Amplified Pre-Stress Actuators is very short. Due to this particular limitation, the surface finish of the bottom bar doesn’t really match with the earlier surface finish of the bar which corresponds to the stable machining cut.
Potential Impact:
Strategic impact
The machine tool (MT) sector contributes to the productivity of virtually all industrial sectors, in that few products can be manufactured without the help of machine tools.
The industry of machine tools has been a fast growing market in the last years, led by technological innovation and precision. In 2008, the global production grew by 8% to 24.6 billion €21. Despite the current difficult economic environment, European machine tool industry is still leading the global market for the production of machine tools with a 44% market share. The competitive advantages of European Machine Tools industry resides today in a high innovative, diversified and precise offer. However, the European manufacturing industry is in need of modernization and upgrade following several years of underinvestment. To maintain for the coming years a secure position as sustainably world-wide leadership and competitiveness machine tools and manufacturing industry, European industry needs to apply unique and innovative technologies and to develop sustainable strategies and excellent trained workers.
Performance impact
By endowing future machine tools with new innovative intelligent and integrated manufacturing features well beyond the state of the art and requiring collaboration of among several scientific and technological areas, PoPJIM anticipates a strong impact on following industrial applications.
• Design and realization of new machine tools
• Users of machine tools for component manufacturing
• Improving the performance of existing machine tools
In this context, PoPJIM identifies three main industrial stakeholders: i) module providers ii) machine tool builders and iii) end users. By focusing on JIM (component) level, manufacturing unit level and manufacturing system level, PoPJIM defines intelligent and integrated manufacturing features as the application of advances in design and technological dimensions to achieve highly productive and reliable machine tools and manufacturing systems that are easily adaptable and reconfigurable in response to changing conditions and new requirements.
Improving quality
Component Level
• Introducing JIM component with known dynamic characteristics will drastically reduce the unknown random variations at joints generated by tribological conditions at interfaces thus ensuring a robust design with reliable functions.
• Simple and highly efficient self-adaptive control at JIM level will essentially improve the control performance and reliability (80%) in comparison to the by far complex conventional control at system level.
• JIM mechatronic self-adaptive control improve significantly the response time and reduces or eliminates the risk for human errors
• Multi-functional passive materials as primary constituents of JIM mechanical structure contribute to a strong improvement of dynamic performance
• Higher portability through JIM mechatronic component standardization Machine Tool (Unit) Level
• Dynamic model-based design methodology to model and simulate the machining systems leads to robust and predictable design of machine tool structures. This contributes to improving quality and notably shortened ramp-up time of at least 30% and in consequence to a reduction of the time-to market of JIM based MT.
• Modularity offered by JIM based MT will allow for high reconfigurability and adaptability to a wide range of products of complex geometry and various materials.
• Wireless distributive technology leads to non-expensive and secure industrial applications
End user level
PoPJIM enables higher quality of parts produced by JIM based Machining Systems
• Consistently running at their optimal capacity
• Receiving real-time information about JIM status
• Monitors actual cutting conditions in real-time and automatically self-adapt the JIM’s structural parameters to the level corresponding to the global optimum operation conditions.
Due to a more stable operation condition, up to 100% improvement in product non-conformance rejection and 75-90 % reduction in machine tool breakdowns and tool breakage can be achieved.
Increasing productivity
Increasing productivity in machining process demands high material removal rate
(MRR) in stable cutting conditions and depends strongly on dynamic properties of machine tool structure. New and wide business opportunities will be opened for knowledge driven European SMEs in the machine tool sector.
Machine tool level
PoPJIM creates the conditions for predictive models that are capable of determining the dynamics of different configurations without the need for vibration measurement repetitions. Model based is based on the predictability of joint interfaces behaviour and therefore it can be stated that up to 50-75% reduction in machine tool lead time due to concurrent design and development process is possible. PoPJIM will drastically reduce the non value-added activities at the end user facility by integrating vibration control solution during machine configuration thereby shorter time to customer can be achieved
End user level
Eliminate the needs for stability process analysis which becomes expensive and time consuming when experimental measurements are needed for each individual possible
Combination of machine tool configurations will result in significantly reducing the lifecycle costs of machine tools (20-30 times)
Since the damping in these JIM based machine tools due to added damping at the joint interfaces, will be considerably higher than the conventional machine tools, more aggressive machining parameters can be used to produce quality products at a much faster rate (1.5-17 times) and with maximum tool life increase (up to 200%) –with shorter lead times (up to 32%) at a lower production cost.
The enhanced capability of JIM based machine tools will minimize the number of operations necessary for producing parts as per specifications. System 3R has reported the reduction of number of milling operation in an automotive part production from seven to three through integration of a high damping interface in a workholding fixture. These high damping interfaces used in these cases are precursors to JIM.
Reduce the need for part inspection due to higher predictability of the JIMbased machining systems operating in safe conditions.
Through its self-diagnosis functions, PoPJIM will eliminate the time for unforeseen failures at machine tool structure level, while virtually eliminating the risk for tool breakage by reducing the dynamic loads at tool-work piece interface. The availability times of JIM-based machine tools will be increased in a proportion of 50-80%.
• Increase of the material removal rate while reducing the specific power consumption through reduced cutting forces and eliminating post operations.
• The energy saving can further be improved up to 30 to 60% by making use of lighter structural modules in combination with JIMs.
Industrial relevance and impact
Many manufacturers are being forced to become more innovative on how best to process parts and to improve quality and reduce total manufacturing costs. Additionally, changes in part requirements have also brought new challenges. High heat-resistant materials, improved strength-to-weight ratios, tighter tolerances, and finer, more controlled surface finishes have challenged traditional methods of machining. PoPJIM creates new opportunities for European manufacturing industry to bring itself to a higher level by offering total quality solutions and new approaches both at the machine tool development level and at the parts produced.
Example of industrial relevance
Impact in Internal Grinding Application: A typical example of industrial application is a supplier of high precision and high volume grinding machines. The extreme conditions with long and small bores and deep located seats with limited room for access, combined with very high demands for accuracy and a premium for short cycle time, require machine tool components with particular emphasis on dimensional stability, low vibration and high overall stiffness. Bore, seat and face on two diesel injection nozzles produced with company’s grinding machines are simultaneously finish ground to tolerances that should eliminate the need for super finishing operations. Grinding fuel injectors (Figure 15) are characterized by very high production requirements, small wheel diameters (2 to 6mm), limited coolant access, and extremely tight finishes and tolerances. Bore roundness is especially stringent and must be held to less than 0.5 μm. CBN wheels dominate and quills are carbide. As new manufacturing processes evolve, grinding quills become increasingly critical for high precision operations. They must satisfy unconditional criteria:
• Rigidity to withstand high pressure
• Lightweight to counteract critical eigen frequencies
• Perfect rotation to achieve speeds up to 200 000 rpm
There are some important aspects related to the grinding process:
1) The quill (Figure 16) bends during rough grinding of long and small bores. As a result only the inner part of the grinding wheel is acting. This causes uneven grinding wheel wear, which equals poor bore straightness, cylindricity and taper
2) There are numerous ways of mounting the wheels. Gluing direct to the carbide is problematic. Wheels are often supplied on wheel screws. Producing quills in carbide with high accuracy electrodischarge machine to cut internal threads for screws is very complex and expensive. Only few companies such as Håmex Hårdmetallverktyg Linköping can produce to the necessary tolerance. The drawback with gluing is that due to higher rotational speed the wheel is jerked out from the wheel. Screws cause eccentricities. A quill rotation of 2 to 3 microns can achieve a finish smoothness of 6 to 8 microns.
3) For new applications, especially grinding at the end of long bores, FEA analysis of the dynamic stiffness and resonant frequencies of the quill is required.
4) Feedrates are limited by chatter onset. With taper compensation, faster federates may be used for larger wheels and the control algorithm applied to gradually reduce rates to those of fixed infeed grinder as the wheel gets smaller.
5) Rotational speeds are limited to about 70% of the maximum values due to the resonance on the critical eigenfrequencies. By integrating a JIM module in the spindle elastic bearing support, another JIM at the interface quill-spindle nose as well by coating diamond layers on the quill by help of Pulse Plasma Deposition process of Plasmatrix Materials, PoPJIM concept claims the following benefits:
• An increase of straightness and cylindricity by up to 40% when grinding small fuel-injection components
• Meantime, the cycle time can be reduced by 30% through maintaining constants feed rates at higher spindle speeds.
• Increasing eigenfrequencies while reducing the resonance amplitude
• Increasing life time of quill by up to 100%
• Reduced number of experiment while designing robust quills and other
machine components
• Spindle holder controlled by JIM mechatronic system allows faster overall cycles times for weak systems susceptible to chatter
Impact on Milling and Turning Application: The application cases reported in the table below and Figure 17 clearly indicate the huge potential benefits of the increase in machining system dynamic stiffness by integrating High Damping Interfaces (HDI) in workholding fixtures and at selected locations in machine tools. Research results show that the use of HDI has resulted in better product surface finish, shorter cycle time and increased metal removal rates. The design solution with self-adaptive control for dynamic stiffness as envisaged in PoPJIM is far more comprehensive than simply using a workholding device with high dynamic stiffness. And therefore, this also offers much higher potential benefits such as higher metal removal rate, better surface finish, longer cutting tool life and less scrap of nonconforming components: overall this results in higher productivity, better quality also energy savings due to shorter cycle times.
Engineering relevance and impact
PoPJIM will significantly change the way the engineering activities particularly design of machine tools and process planning are performed in companies. The JIM concept can be extended to other production equipment, such as material handling where the dynamic phenomena affect performance.
Design of new machine tools will radically be changed from a trial and error activity to a model-based design due to the use of corresponding standardized JIM mechatronic modules with prescribed dynamic characteristics. Mechanical design time and therefore the led times will be dramatically reduced (3-4 times) due to reducing the number of cycles in the process sequence: design –prototype-testing.
Control design: A new control feature based distributed architecture will be integrated to control the JIMs via a wireless multihop network and DWCN. Wireless communication network is for the first being implemented for machine tools outside research labs. This is a completely new feature that ensures the performances as claimed by the PoPJIM concept and opens a potentially huge impact on the way machine tools are design and used. The application of smart materials will be introduced for actuation purpose.
Process planning for component manufacturing
The process planning is the most laborious engineering activity in a manufacturing company. As the majority of its phases are manually performed this results in a time exhausting activity with need for high qualified personnel. JIM based machine tool concept enables for the first time to plan the dynamic performance at process planning phase. As JIM library contains standardized JIM modules with specified range of dynamic load – stiffness/damping characteristic, the process planner has to select for each operation the optimal cutting conditions and the corresponding JIM modules.
Furthermore, supporting the process planning with knowledgebase and module libraries ensures reuse of configuration patterns and facilitates the process while minimising the human errors. In cases where the manufacturing system is composed of multiple alternatives JIM based machine tools for a given set of operations the impact can extend to optimise at the production planning and scheduling level.
Supporting EU policies and European dimension
The implementation of portable plug-and-produce for a new generation of production systems and generally accepted European standards for self-optimizing mechatronic components will be of vital importance in helping the European industry:
• For European machine tools will strengthen the competitive position in the world market.
• For the European manufacturing sector as a whole will translate into competitive advantages ranging from shorter product development cycles to new value-added products and services.
The PoPJIM project is part of the core of the European manufacturing strategy enabling at Machine-level technologies and systems that improve manufacturing productivity, quality, flexibility and safety and at Systems-level technologies for innovation in the manufacturing enterprise, including controls, sensors, radio frequency (RF) wireless networks, and information technologies; methods and approaches that improve design and decision making and integrated and collaborative product and process development
The “Vision 2020” document of ManuFuture has indicate among the drivers of manufacturing innovation, advances in science and technology, specifically in the fields of nanotechnologies, materials science, electronics, mechatronics, information and communication technologies (ICT) and biotechnology. New production processes based on new research, and the integration of hitherto separate technologies exploiting the convergence of scientific and technological disciplines, may radically change the scope and the scale of manufacturing.
The contribution of PoPJIM project to Manufuture refers to what Manufuture considers key technologies for the future:
Mechatronic modules will become essential parts of production machines.
Mechatronic production systems, where all the signals from mechatronic components are integrated for complete manufacturing control, are seen as greatly aiding the rapid reconfiguration demanded for flexible and high-speed manufacturing.
Mechatronics and process control are two of the key research areas for the future.
With major breakthroughs in three technological areas, (i) JIM mechatronic component and EDS multifunctional material, (ii) self-adaptive active control and self-diagnoses and (iii) distributive wireless network, PoPJIM will contribute to generating high added-value products. PoPJIM will stimulate the creation of a European knowledge platform with great potential at both machine tool development level and at manufacturing industry level and consequently contributing to transformation of European industry from a resource-intensive to a knowledge intensive industry.
Contribution to European societal objectives
Quality of Life of the European citizens
PoPJIM implemented in machine tool industry and in manufacturing technologies have the potential to improve quality of life by means of new and enhanced products and services, generating wealth and employment. Research activities in the field proposed will provide economic growth and the consequent wealth level needed for social and environmental improvements at the basis of a sustainable development. It will not only provide innovative solutions to existing problems but will also offer new opportunities through the development of innovative materials (EDS multifunctional material) and innovative mechatronic devices (JIM), European designed products and processes with new functionality with the potential to improve the quality of life for everybody. Furthermore, the PoPJIM will allow module suppliers to customise their products and services to the OEM manufacturer requirements.
Customisation means that the OEMs will not only be capable of bringing flexibility into their production processes, they will also be able to switch to niche markets for JIM based machine tools swiftly and efficiently, leading to high quality products with significant added value. In particular the project will lead to high quality, superior products in the field of high precision industrial components. Eventually it will lead to a strong and economically sound industry and to the creation of additional jobs. Research activities in PoPJIM could lead to successful mass production of many consumer product that would otherwise be manufactured in Asia. European employment can be safeguarded, in the sector of manufacturing that has been under pressure of low wage country competition for years. Contribution to the European Skills Base
This project will contribute to the preservation of economically important machine tool industry for Europe, by offering high added value in products, processes and services. Moreover, this project will lead to the enhancement of the competitiveness of the manufacturing industry, leading to a continuous growth in employment. A very important benefit of this project will be that it will increase the attractiveness of the industry for young professionals, as jobs will become more challenging and more rewarding. The daily tasks of workers in the manufacturing industry will become more flexible, more attractive and more complex, as more engineering skills will be required. Therefore education and training of workers will be an integral part of the future implementation and exploitation.
The PoPJIM entire project concept brings together two key research themes within FP7 namely the NMP and ICT themes, together and creates a venue to exploit the synergy between them and to advance the knowledge gain in their respective domains.
European scientists are recognised worldwide for their leading-edge research in the field of process control and optimization, plug and produce techniques. This is a crucial area for European level research. Thus, excellent technical solutions will give the EU as a whole a large competitive advantage to develop new leading industries in this field with vast economic potential. This will result in an increased magnetism of knowledge intensive SMEs on young technical professionals, because of high quality jobs. Should consortium partners not undertake this project, it is likely that the solutions envisaged will be developed in the USA and/or Asia and that the industry will shift towards Asia because of lower production costs and the steady fading of the present added value differential. This would not only result in larger unemployment figures in the industry, but will also imply that European industry will buy products developed and manufactured outside of the EU. Train both post docs and PhD students in multiple academic and analytical environments, by providing for exchange among the different institutions and labs brought to the project by the participants. Such exchange and exposure will inevitably provide a small well-trained body of young scientists who could lead technical efforts at research or industrial areas.
Task 5.1 Dissemination of project activities and results
Introduction
The overall objective of PoPJIM Project is to develop Plug and Produce Joint interface Modules. These modules are mechatronic elements located in machine tool joints in or-der to improve dynamic behaviour and therefore enhance cutting performance.
The objective of the Dissemination Plan is to identify and arrange the activities to be performed in order to spread the knowledge and achievements generated during the pro-ject and promote the commercial exploitation of the project’s results. It is essential that the dissemination of the results reach the maximum possible audience. For that purpose, the active participation of all project partners is needed in order to spread project results to different sectors, countries and fields.
Dissemination means and contents must be restricted to preserve project confidentiality and protect project results. Therefore, every public exposure of project results must be previously approved by the PoPJIM Consortium. The content of this document is subject to updated along the project life.
Standard dissemination material
The first task will be to prepare effective dissemination standard material that can be used for project overview exhibition. The following physical dissemination material will be created and, if needed, periodically updated as the project advances:
• PoPJIm Logo
• PoPJIm Poster
• PoPJIM Flyer
• PoPJIM Brochure
This material will be uploaded onto the Project Management private section on PoPJIM’s web-site, under WP6: Dissemination and Exploitation folder.
Dissemination channels
The periodical release of news, reports and multimedia material communicating project progress and achievements is of capital importance for project dissemination purposes.
PoPJIM website
PoPJIM website will be the main communication channel for PoPJIM project. This website will consist of a public and a private section. The private section will be used as a project monitoring and file and report administration site. It will be a shared domain for the communication among project partners. The public part will include project gen-eral information, Consortium details and a living section of project news related to pro-ject progress. All project partner’s own website will contain a link to PoPJIM’s website.
LinkedIn Group
A LinkedIn project will be created in order to create an open discussion forum for pro-ject technical and commercial issues. Invitations to project partner’s partners will be done in order to have a feedback and an external view from other technicians and stake-holders familiar with the project related activities.
PoPJIM - Plug and Produce Joint Interface Modules
Youtube channel
A Youtube channel will be opened in order to create and show project related multime-dia content, particularly for the demonstration works.
POPJIM Group
Scientific and conference papers
PoPJIM’s project development will generate scientific paper publication opportunities related to project’s findings and advances. These papers must be related to project’s side technology and in no case will disclose technical details about PoPJIM’s main objec-tive.
Both scientific journal and congresses will be considered.
Technical article
Other non-scientific but technical publications, closer to the industrial environment will be also produced. This will mainly include informational and pedagogical literature in order to raise PoPJIM project’s awareness in technical or sectorial journals.
Company’s newsletter
It is not the aim of the Consortium to create a periodic Newsletter for PoPJIM, but the inclusion of project’s main milestones in each partner’s particular newsletter (in the case it exists) will be promoted. The reason for this is that already settled newspapers from Consortium companies and research entities have a wider scope than a newly created project specific Newsletter would have.
Project results exhibition
Apart from the standard project dissemination channels, other dissemination events will be used for project result dissemination and promotion. In these dissemination events, standard dissemination material (poster, flyer and brochure) as well as the material re-leased through the foreseen dissemination channels (Youtube Channel, Technical Re-ports…) will be used. Also physical demonstrators will be shown.
Fairs and commercial events
General purpose fairs or company’s commercial events will be used for PoPJIM project result exhibition and promotion.
Courses to stakeholders
RTD entities will be in charge of organizing courses to stakeholders to inform about currently existing problems in machining and train about how to solve these problems through PoPJIM’s solution. Also individual project results will also presented in these courses.
Coordinated Seminars
PoPJIM project will interact with other ongoing European Research projects that are dealing with chatter avoidance solutions in order to prepare a Seminar in which different approaches for the solution of similar target problems will be demonstrated. The objective is to enhance the dissemination impact, increase the number of potential audience and take advantage of senergies among the involved projects to explore new variants in project’s objectives. The venue for this Seminar must be an important machine tool and manufacturing fair or event.
Dissemination procedure
Every dissemination procedure will follow the steps described below:
When a significant result or event is achieved in the project, the partner responsible for the realization of that result or event will create a piece of news to include in the PoP-JIM website. This piece of news will be sent to Ideko, who, as the responsible for dissemination, will decide if the information discloses project’s critical foreground. If this could be the case, it will be circulated among all partners for study and modification if needed. Once the note is edited, it will be forwarded to Humboldt for publication in the website. These news could be also published through different dissemination means foreseen in the dissemination plan: newsletters, papers, technical articles…
On the other hand, all the other dissemination events will be also tracked and published in PoPJIM’s news section. Therefore, the responsible partner of every dissemination event will be in charge of creating a short note for publication on project’s website.
It will be very important to try to document both project achievements and dissemination events through pictures and videos for the enhancement of PoPJIM’s Youtube Channel.
All Scientific Papers and Technical Articles must include the following sentence for PoPJIM project addressing:
The research leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 260048.
Exploitation of project results Introduction
1. Introduction
Realization of the project concepts will evidently be through a concerted innovation process that will bring multiple exploitable outcomes. The intent of this exploitation plan is to
• identify the potential exploitable results,
• assign rapporteurs to each identified exploitable items to further develop exploitation venues as the project progresses
• to identify and detail the background knowledge brought by the consortium partners in relation to each exploitable item
• identify the mode of appropriate exploitations, i.e. license, produce and sell, internal use, consultancy, etc, .
• perform preliminary multi aspect risk assessment, i.e. technical, market, legal, etc
For the first task, to identify possible results and characterise them, an initial estimation of development costs, market situation and result protection means is done. Corresponding to each identified exploitable item actors and rapporteurs were assigned.
Then IPRs are analysed with respect to each items. Every partner claims background and foreground IPRs. This will be helpful to distinguish between knowledge brought to the project by each partner and knowledge acquired during the project. Exploitation claims are also made, this way the list of partners involved in the exploitation of each result and exploitation means will be clearly stated. All this information will be helpful for future exploitation and co-ownership agreements among partners.
Finally a risk assessment of each result is performed. This tool enables the identification of the critical aspects that jeopardize successful exploitation of each result. The risks are evaluated as a combination of impact and likelihood and, in order to minimise or eliminate these risks, solution and action points are planned. Each action point will have a responsible for its fulfilment
Therefore all the information contained in this document will pave the way for the corresponding business plan implementation and future cooperation agreements. This document will be an evolving document and will be updated as the project progresses under consensus among all partners.
During the preparation of this document, a one day Seminar was conducted on the 18th of October of 2011 in Steyr (Austria) in collaboration with the European Commission in which a more structured approach was used to outline the plan on which the content of this deliverable is based. In this Seminar the basics for the definition of a successful exploitation plan were addressed and the specific work on exploitation issues for PoPJIM project was carried out.
Consequently, the tables and their content in this document are largely shaped as a result of the seminar and the subsequent discussion conducted within the consortium.
List of Websites:
www.popjim.com
PoPJIM is a four year collaborative project (funded by Seventh Framework Program, call Factory of Future, area FoF.NMP.2010-1 Plug-and-Produce components for adaptive control). The PoPJIM goal was to intro-duce advanced concepts beyond the state of the art in the domain of intelligent machine-tools and to enable high precision mass customization and machining of complex components with zero defects. Increased ac-curacy of machines and processes offers substantial benefits to a wide range of applications that demand high quality and reliability of machining process.
The precision of machining systems is ever increasing in order to keep up with components’ accuracy re-quirements. At the same time product variants are increasing and order quantities are decreasing, which in-troduces high demands on the capability of machining systems. The machining system is an interaction be-tween the machine tool structure, the process and the control system and is defined in terms of capability by the positional, static, dynamic and thermal accuracy. So far, the control of the machining system, in terms of static and dynamic stability is process based which is often translated into sub-optimum process parame-ters and therefore low productivity.
Today's machine tool design criteria focus mainly on optimized lightweight and dynamically flexible struc-tures. This results in more efficient and fast machines, but unfortunately, it often also means less chatter re-sistant machine. Chatter is a dynamic instability of the cutting process which brings catastrophic conse-quences to machining process, such as poor surface quality, tool breakage or machine component failures. In order to succeed in high accuracy and high productivity machine building, dynamic instabilities during ma-chining must be kept under control.
For that purpose PoPJIM project brought a new approach: plug and produce Joint Interface Modules (JIMs). These modules consists of self-optimising mechatronic elements located in machine tool joints whose func-tion is to control the dynamic behaviour of the machine in order to assure process stability, avoid chatter and therefore improve machining accuracy.
JIMs will have plug and produce capability, with an underlying modular and flexible machine building con-cept and they will be communicated and controlled through a reliable Distributed Wireless Control Network (DWCN). The DWCN is responsible for the control of JIMs so that the overall dynamic stiffness of the ma-chining system (machine tool structure and machining process) remains optimized at the system level and it facilitates swift and reliable reconfiguration of JIMs and machine modules for plug and produce capabilities.
The project consortium has implemented an incremental approach in the realization of the full capability JIM. First, a mechanical JIM prototype has been built in order to show its dynamic capability. This first pro-totype has been equipped with piezoelectric tuning and capability and has been used to demonstrate the via-bility of the JIM concept at the very early stage in the project.
Project Context and Objectives:
The Strategic Research Agenda of “Manufuture” stipulates that European manufacturing industry needs to reinvent itself by shifting from cost-based global competition towards a dynamic creation of knowledge based added value [1]1. Such a shift necessarily requires introduction of advanced concepts beyond the state of the art which address the technological hurdles that limit their performance.
Complex components machined with zero defects are an essential condition in high precision mass customization and it becomes a new challenge for the new generation of intelligent machine-tools. Increasing the precision and accuracy of machines, products and processes offers substantial benefits to a wide range of applications from ultra-precision to mass customization with higher quality and better reliability.
Within this context, PoPJIM identifies critical performance limiting problems in machine tool design & development and plans to address them by coordinating various expertises with novel solutions. The outcomes of the envisaged effort will transform machine tools design and performance to new levels which in turn greatly contribute to realisation of low-cost self-optimising reconfigurable machine tools with modular plug-and-produce capability.
The PoPJIM advances two crucial innovations for the future machine tools. The first one refers to a concept of portable Joint Interface Module (JIM) as a self-optimising mechatronic element that controls the dynamic behaviour of a machine-tool during its interaction with the cutting process. Traditionally dynamic instability in a machining process is controlled by tuning the process parameters to match with the inherent dynamic characteristics of the machine tool structure which often results in lowering the rates of production. Novelty of the JIM concept is to control the machining process dynamics by adjusting dynamic stiffness of the machine tool at structural joints. In other words, instead of changing the process parameters, dynamic stiffness of the machine tool is tuned to maintain the process stability. With this approach, the process efficiency is not sacrificed as the process parameters remain unchanged while the overall dynamic stiffness of the machine tool structure follows the demands of voluntary or involuntary changing machining conditions.
Carefully designed damping and stiffness characteristics i.e. the dynamic stiffness in the JIMs allows the dynamic behaviour of the machine tool to be predictable with a higher degree of accuracy. The mechantronic design of the JIM includes an integrated control system which enables it to be self-adaptive for optimising the dynamic stiffness within its design range during a machining operation. A demand beyond the design range of a specific JIM is met by reconfiguring the machine tool with one designed for the required level of performance. The adaptive characteristic for maintaining optimum dynamic stiffness is achieved by exploiting both passive and active multifunctional materials in the JIM design. For the first time, this innovation will allow that a real representation of the joint behaviour can be introduced as a virtual model at an early design stage for fully model based design application.
Second facet of this innovation refers to a Distributed Wireless Configuration and Control Network (DWCN) which allows for plug and produce capability and decentralised control of JIMs through a wireless communication network. This development is essential for achieving operational flexibility and efficiency required for the modularity characteristics inherent in the JIM-based machine tools. The DWCN is responsible for configuration and communication among the network components, JIMs, network coordinator, DSP proxy etc. for establishing two major tasks:
1. Collaborative control of JIMs so that the overall dynamic stiffness of the machining system (machine tool structure and machining process) remains optimised at the system level
2. Facilitate swift and reliable reconfiguration of JIMs and machine modules for plug and produce capabilities.
Due to highly critical nature of the information flow and the engineering noise in production environment, communication reliability and efficiency are the two essential and challenging demands on design and development of the DWCN. The innovative concepts of the mechantronic JIM, application of novel multifunctional materials and the DWCN are essential pre-requisites in realising a self-adaptive, plug and produce machine tools with globally optimised dynamic stiffness. Furthermore, implementation of JIMs will enable model based analysis of machining systems2 at early stage of the design phase. Ultimate objective of these innovations is to keep the machining processes always stable while maintaining production rates and/or the quality of parts in terms of tolerance and surface finish demands as per specifications.
Joints are known to contribute significantly to the dynamic stiffness of the machine too l during its interaction with machining processes. Research in this area suggests that damping from mechanical joints accounts for up to 90% of the damping and 60% of the dynamic stiffness6 in a typical machine tool structure. Although structural joints are an excellent source of damping, the modern design practices fail to exploit this potential in developing machine tools with enhanced dynamic stiffness. This is mainly due to lack of:
• Controllability of joint characteristics
• Suitable functional materials which can compensate for loss of stiffness while increasing damping capacity
In addition to this, lack of theoretical models also prevents accurate prediction of the behaviour of joints under dynamic loading at the design stage. All the theories of elasticity available at present deal only with the problem of the monolithic elastic body, i.e. elastic body with infinitely rigid joints. Dynamic properties of joints can be altered by introducing functional materials of known properties. Research shows that addition of viscoelastic composites at the interface of structural joints results in considerable increase in damping. In addition, pre -stress on joint interface has a significant impact on the magnitude of damping in such joint8. Significant amount of research in this area has been conducted at the Royal Institute of Technology (KTH) in Stockholm. Findings of this research show that addition of high damping interfaces or re -designing critical interfaces in machine tool structures using high damping materials such as viscoelastic (metal-polymers) composites improves their dynamic stiffness by orders of magnitude. And this improvement has remarkable impact on the production rate, cutting tool life and surface quality of the machined surfaces9. Research at KTH in collaboration with industrial partners such as system 3R Intl. AB, Spirex Tools AB and Mircona AB also shows that design and operational variables such as interface design, configuration of the viscoelastic composites and interface pre-stress, pjoint, can be exploited in achieving dynamic stiffness to compensate for changes in operation conditions of the machine tools. Concerning the development of new multifunctional materials for improving dynamic stiffness in machine tool structures, KTH is collaborating closely with an industrial partner Plasmatrix Materials AB. This cooperation has been successful in developing a unique patented structural material named EDS (enhanced dynamic stiffness) material. The EDS material belongs to a family of nanocomposite materials consisting of a metal matrix with embedded carbon nanoparticles. By varying the proportion of nanoparticles in the metal matrix, the stiffness-damping properties can be varied from very stiff to highly damped.
Results of various research projects driven by Swedish industries show that increase in dynamic stiffness of different machine tool components and or working holding fixtures have a very positive effect on productivity and quality.
Introducing functional materials of known characteristics at the joint interfaces offer an opportunity to design joints of predictable characteristics. The core design principle is that the damping effect of the rigid joints between a JIM and machine tool modules is insignificant compared to that of JIM itself. On the other hand stiffness in the JIM will be good enough to direct all vibratory energy to its own interface for a localized dissipation due to high damping.
Since the joint characteristics in case of JIM will be dictated by the properties of the functional materials, few necessary changes in the models of these materials would suffice for quantifying the dynamic behaviour of these joints.
The objectives of the PoPJIM proposal as envisaged by the concept of JIMs are:
1. Develop portable Joint Interface Modules (JIMs) where an appropriate configuration of one or more functional materials of known characteristics act as an interface between two metallic elements and has dominant influence on the behaviour of the JIMs
2. Exploit the properties of both the interface geometric design and functional materials in developing the JIMs with varying and adaptable damping capacity and thus the dynamic stiffness as well
3. Develop active control mechanism in the JIMs so that they can optimise their dynamic stiffness as per demands of their operational environment
4. Develop a reliable communication network based on distributed wireless control technology which is inevitable for exploiting the flexibility that JIMs offer in realising modular machine tools with plug and produce capability
5. Demonstrate the feasibility of these developments by implementation in a milling machine
Project Results:
Development of the JIM concept
As mentioned in the preceding paragraphs, a typical JIM as visualized in the project offers both predictability and adaptability of its dynamic stiffness. These two characteristics are obviously valid within the design range of the JIM. When the operational demands on dynamic stiffness surpass the design capacity of the JIM, it has to be replaced with another JIM of higher capacity. This methodology thus defines three different levels of adaptability:
• Adaptability based on active control
• Adaptability based on interface design
• Adaptability based on dimensional design
It is worth mentioning that in the first two adaptability levels, nominal dimensions of the JIMs remain unchanged.
Adaptability based on active control
Voluntary and involuntary changes in operational conditions require corresponding adjustments in dynamic stiffness of the machine tool structure in order to keep the process parameters at optimal values. The adaptability in the dynamic stiffness is achieved by varying the joint interface pre-stress through active control during machining operations. The interface of a typical JIM is realised with a given design of selected functional materials placed into two rigid metallic components of a given geometric design of the interfacing surfaces. In order to pre-stress the joint limits of the control based adaptability in the dynamic stiffness lie between the lowest and highest allowable pre-stress for a given JIM and the controllable range then depends on factors such as:
• Design (mainly thickness) of the viscoelastic composite
• Pre-stressing capacity of the piezo actuator
• Load bearing capacity of the metallic pre-stress plates
The upper and lower pre-stress plates in this construction also provide for mechanical coupling interfaces for rigid connections with the adjacent machine tool modules. In order to maintain envisaged operational efficiency of the JIM, it is imperative that the stiffness of the JIM never exceeds the stiffness of rigid connections with the machine tool modules on the two ends of the JIM. The goal here is to divert maximum vibration energy into the damping interface of the JIM so that it is dissipated effectively through the viscoelastic composites. When a machining process requires dynamic stiffness that exceeds the design limits of adaptability through active control, then a JIM with higher stiffness capacity will replace it.
Adaptability based on interface design
Dynamic stiffness in JIMs can be increased or decreased by changing the geometric design at their interface. Design features such as conical or spherical surfaces, vgroves, key slots etc. can be used to enhance the JIM capability. In combination with different configurations of the multifunctional materials these design features at the interface provide for another level of dynamic stiffness in the JIMs.
Adaptability based on dimensional design
The third level of adaptability caters for the needs of changing dimensions of both JIMs and machine tool modules. The extend level of plug & produce capability in PoPJIM concept is the ability to reconfigure the machine tool from one type to completely another, for example from a conventional milling to a high speed milling machine. This reconfiguration may require that nominal dimensions of both the JIMs and the machine modules such as spindles may vary considerably. The system should be able to detect these changes and then adapt to the requirements in terms of maintaining the zero reference of the machine; this implying a channel for communication with numerical control of the machine tool as well.
Vital consideration in implementation of the JIM concept
Online and offline compensation for dimensional variation in the machine tool structure: The dimensional variations, either online due to actuation for prestress, temperature variations or offline due to use of JIMs and machine modules of different nominal dimensions, will be given due consideration in the overall design of the system. Any dimensional variation will be compensated with the precision commensurate to the requirements of the product quality.
Flexibility, speed and precision in mechanical coupling mechanism: The connections of the JIMs with machine tool structure/modules have to be more rigid than the highest levels of stiffness in the JIMs. In addition to this basic design requirement, the mechanical coupling for this purpose has to provide flexibility for easy changeovers, swiftness to minimise down times during the changeovers and obviously the necessary positional precision. Designs and solutions in this regard are available commercially; the best fitting to the requirements will be investigated and used for this application. One of the examples is System 3R AB which develops high precision mechanical coupling systems with full automation capability for changeovers.
The JIM design as described here assumes piezo-actuators as primary means of actuation for control of the dynamic stiffness. This however does not rule out the possibility of using other smart materials like magnetoristrictive, shape memory pneumatically or hydraulically operated mechanical forcing systems, etc. These materials and solutions may be considered in situations where the piezo-actuation is not a suitable approach.
Design methodology for JIM based modular machine tools
In line with the objectives of plug & produce approach, JIMs will ensure flexibility and efficiency both in the design & development and operational phases of the machine tools. In the design phase digital models of JIMs and machine modules will facilitate virtual analysis of the envisaged machine tool and assist in achieving a design with optimal dynamic performance. Including the influence of joint interfaces with known and controllable characteristics at the design stage will minimise the requirement of prototype manufacturing. The JIM based machine tools will be realised more efficiently with higher levels of reliability.
The self-adaptive control of dynamic stiffness in the JIMs and ultimately of the machining system comprised of two essential features namely monitoring and control.
Monitoring: While the machine tool is in operation, the dynamic behaviour of the machining process will be constantly monitored. An online vibration monitoring system, supported by a model-based identification of the machining process, will be implemented for extracting information regarding the machining system’s dynamic stability. The monitoring system will continuously transmit this information to a network coordinator for data analysis.
State-of-the-Art monitoring systems periodically record measurement data in chunks and perform offline analysis on the same data extracting characteristics of the signals based on time and frequency domain analysis. Since this procedure leads to significant time delays, the response behaviour of an underlying control system dramatically suffers. In fact, instabilities in machine tools such as chatter can occur in the fraction of a second.
Since the control system does not directly regulate mechanical states such as displacements, velocities, accelerations (which can be directly measured or dynamically estimated by state observers), but mechanical parameters of the system, real-time availability of accurate parameter values, especially global stiffness and damping parameters of the machine tool, are pivotal to quickly and reliably counteract instabilities. The proposed plug and produce approach doesn’t require mathematical models to be derived from underlying complex numerical models of the machine tool, but parameterized by intrinsic mechanical properties such as modal gains, eigenfrequencies and damping ratios. Estimation of these parameters will be carried out online, i.e. parameters appear as states of a dynamical system and are driven to their actual values by measurement data.
Due to the non-linearity of the underlying adaptive laws for parameter estimation and the use of IIR (infinite impulse response) models described by modal parameters, stability cannot be guaranteed a priori for all configurations. Hence, implementation and appropriate management of stability criteria for parameter estimation will be at the centre of investigations with respect to the monitoring system.
Self-adaptive control:
The control of the JIM interaction with the machining system is performed in two stage monitoring and control strategy:
• Modelling of the interaction between machine tool elastic structure and the machining process
• Self-adaptive control of the JIM units
In the first step the machining system’s operational damping and stiffness are extracted from the identified models. The input to the adaptive controller is the operational parameters from the first step. The output from the adaptive controller is the parameter that controls the JIM dynamic state, i.e. JIM’s damping and stiffness. It is important to point out between the difference between individual JIMs dynamic behavior and machining system dynamic parameters. By changing the dynamics of JIMs, a control of machining system dynamics will be achieved.
Model-based identification
Generally, a model is used to approximate data and it has a specific structure, i.e. a mathematical formulation which includes a number of parameters. These parameters can have a physical interpretation, in which case the identified model is a structural model, or retain only the mathematical interpretation in which case and regarded as a synthetic model. An advantage of the model based vibration analysis is that it can be used for obtaining both synthetic and structural models of machining systems.
The model based identification is carried out in two steps
1. Evaluation of ARMA model’s parameters – synthetic model
2. Formulation of the structural model, thereby the operational damping and stiffness can be extracted.
Parametric models
Parametric models is special class of models, driven by white noise processes and possessing rational system functions, includes autoregressive (AR) (Burg, least square, Yule Walker, geometric lattice, instrumental variable), ARX (arx, iv4), moving average (MA), autoregressive-moving average (ARMA), Box Jenkins, Output Error models.
The output processes of this class of models have power spectral densities that are totally described in terms of model parameters and the variance of the white noise process. The motivation for parametric models of random processes like machining is the ability to achieve better power spectrum densities estimators based upon the model than produced by classical spectral estimators like FFT. The modeling of a stationary time series as the output of a dynamic system whose input is white noise n(t), can be carried out in several ways. One way is to use the parsimonious parameterization which is employing ARMA(p,q) representation. Given a time series of data Xt, the ARMA model is a tool for understanding and, predicting future values in this series. The model consists of two parts, an autoregressive (AR) part and a moving average (MA) part. The model is usually then referred to as the ARMA(p,q) model were p is the order of the autoregressive part and q is the order of the moving average part. In its original representation ARMA parameter estimate is based on iteration schemes by help of a non-linear algorithm.
One of the main benefits of the model based identification is that it can be performed in-process and thereby be used for monitoring and control of machining system stability which requires fast response in order to detect chatter before full development. One important tool used to compare in model performance is spectral analysis. In spectral analysis the power spectral density (PSD) is defined as the discrete-time Fourier transform of an infinite autocorrelation sequence (ACS).
This transform (relationship) between the PSD and ACS may be considered as a nonparametric description of the second-order statistics of a random process. However a parametric description of the second-order statistics may also be devised by assuming a time-series model of the random process.
Self-adaptive control of JIMs
The controller is designed to work into two states:
(i) Control state, when the machining system approach instability threshold or in some situations, by deliberate actions to bring the system in a desired stability state.
The JIM is designed based on two calibration curves: pressure–damping and pressure–stiffness. These calibration curves are estimated for individual JIMs at the JIM design stage and then when integrated in the machine tools. As an initial pre-stress is necessary in the JIM, a self-learning technique will be implemented.
(ii) Monitoring state to follow the system behaviour
As long as the machining system is in stable region, or in the desired range, the control is disabled. For this reason the selected control at this stage will be the self-tuning type. At the start of the project a review of other control system types will be appropriate.
The heart of the adaptive controller as for the model-based identification is the recursive estimation. In the initial phase of the project, various recursive estimation schemes will be analyzed and tested. The self-tuning controller is based on the Pole Assignment Control (PAC). The aim is to modify the dynamic response of the JIMs in order to keep the machining system in the stable state.
Information from the monitoring system will drive the control system to self-adapt the dynamic characteristics of the machining system by varying damping and stiffness parameters of the JIMs. This will be achieved by changing the pre-stress on damping interface using smart materials as actuators in the JIMs. The control system will change the damping and stiffness characteristics of the critical joints in a way that energy directed to each interface is proportionate to its dissipation capacity so that the system does not experience any localized overload. The process monitoring and JIM control systems are linked to each other through a network coordinator. The network coordinator has a very vital role in the realization of the zero vibration machining system concept in PoPJIM. The machining process identification algorithms based on the estimate of overall system damping reside in the network coordinator. On the other hand the information on operational condition of each JIM is also processed simultaneously, but still in the network coordinator. The network coordinator will keep track of stiffness and damping parameters in each of the JIMs using mapping functions, also available in it (see Figure 6). As the system detects the machining process approaching its instability threshold, a coordinator function will invoke commands for JIM actuators to act in a manner that global dynamic stiffness of the machining system is altered to avoid the impending process instability.
It is important to note that the coordinator function and the underlying control algorithms do not work iteratively, i.e. offline analysis of a stream of vibration data followed by decision making with respect to pre-stress based on static mappings, but dynamically drive the system to a stable operating point based on adaptive control laws.
Such control laws will be derived from energy-like functions describing the quantity to be minimized (vibrational energy) or even maximized (dissipated energy) along the trajectories of the system. Performance of these control laws is, of course, governed by accurate models and proper estimates of states and parameters of the system. Candidates of update laws will be benchmarked with respect to stability and response to transient changes in the excitation spectrum. Since the machine tool will be equipped with multiple JIMs, cross-coupling of the JIMs and, hence, the multivariable nature of the control architecture has to be appropriately considered. A robust transducer system will be integral part of the JIMs to transmit information regarding their instantaneous dynamic state to the network coordinator. The network coordinator, by keeping track of both the process condition and the state of the JIMs, is providing control directions to each of the JIM for activating the pre-stressing mechanism whenever necessary.
It’s important to reiterate the fact that in conventional control systems the dynamic instability is controlled by changing the machining process parameters, often reducing the metal removal rate and ultimately productivity of the machine tool system. In this breakthrough concept the process instability will be controlled by changing structural dynamics of the machine tool system instead.
The result of this control strategy is that the system productivity will not be compromised at all; actually there will be a possibility to increase the removal rate as the combination of new dynamic characteristics may allow for more aggressive process parameters, obviously within limits of overall design capability.
Distributed Wireless Configuration and Control Network
PoPJIM will implement a Distribute Wireless Configuration and Control Network, DWCN, based on a reliable wireless multi-hop-networking facility to support plug and produce configuration and the JIM distributed control.
Machining environments pose specific challenges when deploying wireless networks.
In many cases, different signals interfere with the wireless channel and metal objects are in places that may sub optimize the network connectivity. However, to connect systems wirelessly no existing network infrastructure is needed and no additional network cables need to be installed which can reduce the time and cost of module replacement significantly. Also, sensitive equipment (like cable connectors that would be exposed to damaging environment factors such as vibration) is not needed on the systems. If carefully planned and carried through, wireless networks are an interesting alternative to traditional cable-bound industrial environments. Wireless network is a prerequisite for efficient plug and produce environment.
The usage of wireless communication allows plug-and -produce not only with respect to multiple JIMs, but also for the interaction of the machine tool components with a configuration support environment. In order to bring such a networking technology to a production system level, PoPJIM will do the extension and tailoring of an industry standard radio technology – IEEE 802.15.4.
The novel aspect of the plug-and-produce scenario supports configuration following mechanical mounting of the modules. This demands the wireless connectivity layer to be both self-configuring and self-organizing, but still under the given time demands of a scalable and reliable network. Administrative intervention is unwanted and contradicts the Plug and Produce principle. This particular demand is denoted (with respect to similar IP network capabilities) as zero-conf networking. This particular feature forms a novel extension for state-of-the-art in IEEE 802.15.4 based wireless networks.
The research in this aspect will be driven by a set of key objectives, in alignment to the overall project goals and the goals from the cross-thematic NMP / ICT call: support environment for ICT-based production systems, self-adaptive and portable plug-and-produce components with an open architecture, and high degree of autonomy and adaptability. From the development goals stated in the FoF.NMP.2010-1 call text, the work contributes to the “components and methods for intelligent, self-sufficient plug-and-produce systems”. PoPJIM will focus on the software part of “an open architecture to facilitate any additions of new modules as needed for implementation in a new process environment”.
The resulting networking facility has to consider a specific set of non-functional demands:
• Wireless communication to support easier mechanical integration and reconfiguration of JIMs
o True plug-and-produce support for JIMs and external control units
o Application of industry standard RF hardware for cost optimization
o Dynamic adaption with modules being installed, removed, or moved
• Automated establishment of a wireless network topology
• Research on dependable multi-hop networking strategies and algorithms
• Immediate setup of networking without manual intervention
• Research on application of zero-conf approaches in machining environments
• Automated scaling of communication
• Tailoring for harsh operation environments (e.g. RF interferences)
Configuration support environment
Apart from the mechanical aspect of plugging (mounting) of JIMs in the joints,
PoPJIM considers the Plug and Produce configuration of JIMs as composed of two phases:
1. Communication phase: a JIM needs to establish communication and send/receive data to/from the network coordinator, process proxy (the DSP proxy) and other JIMs.
2. Setup phase: the JIM self-optimizes to initialize and set its pre-stress value.
Plugging a JIM into the machine tool amounts to reconfiguration of a machine tool and hence it requires the selection of the right module through configuration analysis. To support this configuration task a support environment is required that is composed of two major components:
• Module library
• Configuration knowledgebase
Module Library Service
The module library service is meant to facilitate the configuration/reconfiguration process by providing digital models of the modules: the JIM modules as well as machine tool modules such as spindles, tables, slides, tool holders, chucks etc.
As mentioned in the preceding paragraphs, JIMs are to be designed with different operational range and geometrical features to accommodate various operation conditions and joint interface types. The description essentially contains the module’s basic attributes such as modules id, feature parameters, high-low values, etc and should be able to fulfil all the information requirement and format to conduct a digital configuration analysis. The existing device specification languages will be investigated for application in the project. In order to address a wider configuration scenario involving several module suppliers using different specification, a semantic module specification scheme will be developed and applied.
Configuration Knowledgebase Service
The knowledge base service is intended to offer the storage of data about machine tools and their modules, such as models, reference values, relevant standards, model templates or partial models, and connectivity and interfacing restrictions and possibilities. The knowledge base content will continuously be updated as new knowledge elements are acquired through experience and analysis of run time data. In sum, the DWCN deals with three major building blocks for the work can be identified: (i) the software part of wireless networking, (ii) the hardware part of wireless networking, and iii) the configuration support service.
Overall strategy and general description:
The objective is to develop and demonstrate the JIM concept for the machining centers. The project has duration of 48 months and is structured into 6 work packages.
As described in section 1.1 the main innovative technological advances envisaged by PoPJIM include the following functional areas:
1. Design and realization of a mechatronic Joint Interface Module (JIM)
2. Distributed Control for the JIMs
3. A multi-hop wireless network for the realization of DWCN
For such a highly innovative and multi-disciplinary research activity a well defined work strategy is essential to define clearly the scope of the project and the interdependence of the activities to achieve the objective with the resources available.
In addition to the common work packages WP1 Project Management and WP6
Dissemination and Exploitation, the project activities are organized in the following four work packages.
Work package 2 – Mechatronic Design of JIM with Functional Materials
The mechanical design of Joint Interface Modules (JIMs) with novel functional materials, Modelling and analysis of JIM integrated machining center, integration of the mechatronic components and prototyping of the JIMs.
Work package 3 – JIM Distributed Control
The JIM distributed architecture, the sensing and actuation system in the JIM and the embedded control. The output of these activities will be integrated to the module in WP2 integration task.
Work package 4 – Multi-hop Wireless Network
The Multi-hop topology for communication among the JIMs and other network components, the wireless infrastructure and the configuration support service to enable and support virtual configurations by providing module library of JIMs and other machine tool modules as well as a configuration knowledgebase.
Work package 5 – Demonstration
The JIMs developed in the RTD work packages will be demonstrated using a machining centre in an industrial context. The interdependencies and the flow of information and results among these work packages are summarized in subsection 1.3-d. The work plan will be monitored and updated, if necessary, by the project management to ensure the achievement of the objectives within the time and budgetary limits. These amendments are primarily meant to avert or mitigate risks that may arise due to anticipated/monitored or unforeseen events or make use of relevant and promising technology developments available outside the project.
Work Package 2: JIM Mechatronic Design and Realization
Task 2.1 JIM Requirement Specification
This task comprised mainly of two subtasks dealing with the dynamic performance analysis of legacy ma-chine tool and selection of the critical joint interface locations. Major objective of this task was to analyse the machine tools selected by the machine tool manufacturers in the project consortium i.e. AFM and Soraluce. AFM and Soraluce dedicated for the project a turning machine (TAE-35N) and a heavy duty floor type milling machine (FL6000) respectively. The two machines were analysed for their dynamic behavior both through experimental modal analysis and machining tests. Results from these analyses and tests laid the foundation for preparing the requirement specification.
The requirement specification is now being used as reference document for all developments as envisaged in the project plan including the JIM mechatronic design, JIM control and the multi-hop wireless network. The structural dynamic parameters such as the natural vibration modes, excitation amplitudes and modal damping of the two machining systems provide the guidelines for the developments related to the JIM design, control and wireless communication strategies.
Following analysis of the selected machine tools it’s found that the damping design, control and communication solutions should be able to handle the structural dynamics up to 2500 Hz. For the proof of concept, the critical control and communication range can be limited up to 1500 Hz as the higher frequency excitations were observed with rather extreme machining parameters.
Task 2.2 JIM Mechanical Design with Functional Materials
Main objective of this task is to develop model based design of JIMs and methodology for development of joint interfaces with known and controllable characteristics. Since the principle of the JIM based design is that the JIMs assume dominating role in defining the dynamic behaviour of the machining systems, the interfaces have to be designed with special materials with exceptionally high damping and acceptable stiffness. In this case two different types of interfacial materials are used. One is a metal-polymer compo-site with exceptionally high damping capacity and the other is plasma coated carbide interface with high damping and stiffness. Since the high damping materials, especially the polymer based, have usually low stiffness special emphasis on the interface design is necessary to cater for resulting lower stiffness. In this context three design parameters including the geometric design, material design and the boundary conditions become critical drivers for the JIM design.
Sub task 2.2.1 Joint interface property characterization
Main focus of this sub task is to characterize the dynamic properties of the interface materials. The viscoelastic polymers exhibit a highly nonlinear behaviour where the material modulus and loss factor (a measure of material damping) change nonlinearly with frequency or temperature. Both experimental and computational approaches have been used in characterizing the material behaviour. Following what has been developed in the 1st period, the method and calculation algorithm was further improved to measure and identify the joint interface properties with only assembled structures. In the developed calculation algorithm, the structure component’s properties are not necessary known for the identification process. As the joint interface’s static stiffness and damping values are defined as frequency dependent, the joint interface material’s properties are compared with what has been calculated by the algorithm depending on frequency. Following the scheme, the method is applied to measure the joint interface properties with pure VEMs without screws and it was found out that the VEMs Young’s modulus and damping property does not change under the pressure range the bolt screws can supply. The change of joint interface’s stiffness and damping is coming from the bolt screws instead. In the virtual simulations, five different designs have been analyzed through finite element analysis for the flexible fixture:
• Case I (Without VEM): Original system, no VEM layer added.
• Case II (1mm damping layer): 2 viscoelastic layers (3M™ Damping Foil 2552) with aluminium foil embedded between the base plate and the flexible fixture.
• Case III (1mm damping layer in base and screws): 2 viscoelastic layers (3M™ Damping Foil 2552) with aluminium foil embedded between the base plate and the flexible fixture, and between the screws and the flexible fixture (see Fig.3).
• Case IV (3mm damping layer): 6 viscoelastic layers (3M™ Damping Foil 2552) with aluminium foil embedded between the base plate and the flexible fixture.
• Case V (6mm Sylomer): 1 Sylomer layer (Sylomer ® SR850-6) of 6mm between the base plate and the flexible fixture.
It is clearly seen that after VEM layers have been applied in the interface, embed VEM layer in the joint screws can improve the flexible fixture’s performance even further.
Subtask 2.2.3 JIM design update with Added Interface Module concept
It was concluded that JIM approach was suitable for low damping and high frequency applications, such as the flexible fixture tested. This can be the case of numerous slender and thin walled structure machining applications, such as aeronautical parts. AFM case is in this group, with a slender workpiece to machine and a high frequency and low damping mode limiting cutting stability. However, Soraluce’s case is different, with low frequency (high mass) and high damping modes involved in process instability.
Therefore, a different approach will be considered for Soraluce’s case. An added mass supported by a tunable viscoelastic damping layer was used. The purpose of the damping layer is to add damping and to tune the frequency of the mass in order to make it work as a vibration absorber, through changing the preload of the viscoelastic layer.
Figure 16 shows a scheme of the operating concept of the tunable viscoelastic damping layer, which will be hereafter called Added Interface Module (AIM).
Sub task 2.2.4 EDS material embedded in JIM
Parallel to the mechanical design of the first JIM, use of plasma coatings as Enhanced Damping Surfaces, or EDS, to form damping layers at the interfaces is being investigated and a batch of experiments has been conducted and the work is still on progress. The experimental evaluation work is carried out in parallel with the Viscoelastic material.
One JIM prototype has been applied with the HiDS coating and the cone of JIM with coating is shown in Figure 18 (see the attached document). The thickness of the coating is 500 μm.
While the coated cone is embedded in JIM prototype with a workpiece clamped on the fixture (as shown in Figure 19), direct machining was performed to test the function of the damping coating.
Task 2.3 Mechatronic Design and Prototyping
Following the last report, the JIM mechatronic design and prototyping is updated during this reporting period. The main activities that are included in this reporting period include:
• The development of JIM with appropriate actuation and sensing elements integrated
• Performance analysis of the integrated actuation and sensing elements
• Performance analysis of the mechatronics JIM with variable stiffness
•
Since it was found out that the main obstacle for the 1st JIM mechatronic prototype to function is the piezo electronic device which forms a low impedance path for vibration energy, the second JIM mechatronic de-sign is updated with piezo electronic devices which are integrated with HDI. Apart from understanding that VEM in JIMs can enhance the mechanical structure’s vibration damping capacity, the effect of pre-stress on the JIMs performance against vibration is investigated in this period. When a workpiece is clamped on the JIM prototype with integrated mechatronic devices, the model analysis is done to characterize the mechanical systems properties with low pre-stress pressure and high pre-stress pressure.
Task 2.4 JIM Based Machining System Modelling and Experimental Validation
Main objective of this task is to develop model based design methodology for machine tool design although applicable in general but especially for JIM based machine tools. The idea is to provide designers a reliable and efficient design tool for assessing the dynamic behaviour of the machine tools so that they can design the machines with required dynamic performance. This approach will move the expensive retro-fitting of vibration control solution at the user facilities to inexpensive and efficient integration at the de-sign stage.
The JIM interface characterization is an important requirement for this task and is being met through the characterization work being done in Task 2.2. The interface behaviour models will be implemented in the design and analysis of JIMs in virtual environment. In the work done so far, representative models of the subject machines have been analysed through FEM. The dynamic behaviour of these models is being tuned to simulate behaviour closest to the experimental results. Figure 22 & 23 show examples of FEM analysis of the AFM turning machine and the vertical plate for the Soraluce machine. Similar analysis has been carried out for the Soraluce case. In the next phase of analysis these HDIs will be integrated in the models and the dynamic performance of the machines will be analysed for different boundary conditions. In parallel with the FE modelling of subject machines and model based design methodology is being developed. This activity resulted in deliverable D2.3 which was completed in month 20 of the project.
The tasks in the work package WP2 were completed in due time and the mechanical JIM concept was suc-cessfully proved. Some changes in the distribution of tasks between partners were necessary since an im-portant part of the project was focusing on implementation of PoPJIM concept in an industrial turning machine.
Static and dynamic capability of the machining system is determined by the interaction between machine tool and cutting process. In this respect, the control of machining system stability can be achieved by changing dynamic parameters of the machine tool, the process or both. The traditional way to control machining system dynamic behaviour is by controlling the process parameters, i.e. depth of cut, rotational speed, speed rate as well cutting tools’ micro-geometry and material. By this, static stiffness in the direction of cutting force and overall damping of machining system are improved.
Although being aware about the significant contribution of joints’ stiffness and damping to the overall capability of the machining systems, the classical theory of the machining system lacks a unified concept for consciously designing structural interfaces with controllable characteristics. Apparently, there is a tendency today both among scholars and manufacturers to develop and implement complex schemes for on-line monitoring and control of machining systems. The simple explanation is the existing knowledge gap between machine tool manufacturers and machine tool users regarding static and dynamic capability. Due to unlimited combinations of tools’ geometries and materials, workpieces’ shapes, dimensions, and materials, fixtures and toolholders it is impossible to predict the behavior of a machine tool and by this the accuracy of resulted parts. The consequence is that the machine tool users are forced to add advanced sensor systems for monitoring and controlling the machining system. In an industrial environment these solutions are costly and complex and not reliable due to adverse conditions. PoPJIm concept has proven the fundamental relationship between the conditions of machine tool joint inter-faces and the process static and dynamic stability.
Work package WP3: JIM Distributed Control
Adaptive control architecture was formulated at concept level, implemented at the integration level and tested on the AFM TAE 35N machine at the machine level. The system identification, optimization and control algorithms applied at various stages of the control architecture were then validated with respect to the machining tests on AFM TAE 35N machine. The sensing and actuating systems were developed according to the technical requirements requested by WP2 in such a way that the control architecture can be realized on the AFM TAE 35N machine at the functional level. In Work Package WP3, the following objective was to be achieved in the last reporting period i.e. from 22.10.2013 till 31.08.2014
• JIM Embedded Control: The control architecture for the reduction of unwanted vibrations which was developed earlier with respect to one active JIM has to be adapted for 2 active JIMs such that it will enable them to converge to a global optimum. The following tasks were to be performed in the last reporting period to achieve this particular objective in practice:
• Integration of wireless hardware (developed by WP4 for process monitoring stage) with the JIM global control (Master Controller Node) hardware
• Machining tests on AFM TAE 35N machine with active JIM to validate the process monitoring stage, JIM global control, JIM local control stages of the developed control architecture
• Integration of the master controller node (MCN) hardware with the communication support platform (CSP) via general purpose wireless network node setup (developed by WP4 for communication be-tween MCN and CSP)
• Re-design and development tests of amplified pre-stress actuators (APA) and the improvement of displacement signals obtained from eddy current sensor (ECS)
Task 3.1 Control architecture defines the features, functional and non-functional properties of control
Task 3.1 focuses on developing the control architecture for the self-adaptive control of mechanical JIM’s when integrated within a machining system. It defines the software as well as the hardware interfaces to WP2 and WP4 and also the kind of sensors and actuators which were considered at the initial stages. This Task 3.1 is of interest to all project partners in particular partners from WP2 and WP4 and within WP3.
The control architecture of Task 3.1 consists of primary and secondary stages of control. In the primary stage of control, a global master controller will acquire the incoming vibration signals with respect to wire-less communication medium and estimates the JIM parameters like operational damping, Eigen frequency and optimal pre-stress. Whereas in the secondary stage of control, a local controller acquires the incoming optimal pre-stress from the global master controller and provides the appropriate actuating signals to the Sensing and Actuation systems. So that, the actual pre-stress measured by the sensing and actuation systems tracks the optimal pre-stress in the presence of external disturbances. Here, the computed optimal pre-stress is regarded as a set point for local controller.
Task 3.2 Sensing and actuating systems develops an actuation system which can provide sufficient force capabilities to cover JIM pre-stress bandwidth
Task 3.2 focuses on developing the sensing and actuation systems to provide the mechanical input to pre-stress the JIM’s. This Task 3.2 is of interest to all project partners in particular partners from WP2 and with-in WP3. Actuators of Task 3.2 are Amplified Pre-Stress Actuators (APA). The APA was designed to introduce damping capabilities and sealed ceramics to protect the piezo-ceramic from the critical machining environment. The APA was designed with 5 layers of 10μm of thick VEM to prevent the force shunting of its damping interface.
In the previous integration attempts, the APA was tested on the AFM TAE 35N turning machine included with a single JIM. In practice, the force generated by the APA was not sufficient enough to apply a mechanical effort to pre-stress the JIM on the AFM TAE 35N turning machine. The sensor signal was relatively low to be applied as a feedback signal for the local controller. It was decided to perform the re-design and complete testing of the APA along with the improvement of ECS. Now, the re-design and the complete testing of the APA along with the improvement of ECS are under progress. According to Task 3.2 Deliverable D3.2 “Sensors and Actuators for JIM” was prepared. At present, the deliverable D3.2 is being checked by the project consortium.
Work package WP 4: Multi-hop wireless network
WP4 provides networking solutions for the PoPJIM. The developed software need to be provide reliable networking in presence of high amount of metal (machines, walls, machined components, etc.) that adversely influence the radio signal propagation due to increased electromagnetic noise. These effects reduce quality of wireless communication by increasing packet losses and end-to-end communication de-lay, but may also affect overall PoPJIM system configuration developed in Task 4.3.
The research and development in WP4 are performed in three tasks. Task 4.1 developed network that connects machines among themselves, and machines to the configuration support platform. We have developed a low-cost, flexible and maintainable network that allows remote management of any number of machines that may be deployed anywhere in factory floor. Task 4.2 has developed a real time, high speed vibration monitoring of the machine that allows control of the JIMs within the machine.
The research and development within tasks has progressed as planned and all R&D tasks were completed in time. The main remaining effort in the last year of project will be integration of solutions developed in WP4 with work of other work packages. In project deliverable D4.3 we have presented detailed integration approach that will allow us timely complete the demonstrator. It is planned to perform the integration pro-cess in a set of successive integration workshops that will be used for detection of issues and incremental adding of new functionalities. The approach offers the possibility to the project partners to integrate and test their solutions in real factory environment, on the demonstrator machine. If issues are identified, they will be resolved by the next integration workshop, which will include testing of fixed features, and addition of new ones. We first present network architecture, and then proceed with detailed reports on activities and results of each of the tasks. Each tasks first summarizes the main achievements, and then proceeds with the detailed description of activities and achievements.
Network Architecture
The main architectural and technical challenge of the DWCN is that it has to accommodate two very distinct communication cases:
• Continuous, synchronized, high data rate, short range, low delay communication between the vibration monitors and the MCN, and MCN and JIMs. For this purpose we implemented a dedicated wireless sensor network (WSN) and associated Medium Access Control (MAC) protocol that is utilized to control the machine. The network supports sampling rates of the vibration sensors is in range of 5kHz and nodes are synchronized with accuracy of approximately 10 μs.
• Non-real time, long range, reliable communication between the Master Controller Node (MCN) and the Configuration Support Platform (CSP) through the general- purpose wire-less network (GWN). To support this, GWN provides stable and reliable routing in multi-hop wireless network, low maintenance and ease of configuration, and low wireless signal interference with the WSN.
The network deployment diagram is shown in Figure 44. The diagram shows components that comprise the DWCN and their interconnections:
• WSN nodes and software operate on ARM Cortex M3 based hardware running custom embedded software (Fast-MAC) developed by Inertia.
• MCN consists of two physical nodes: xPC-Target performs real-time machine control, xPC-Host provides GUI and communicates with the CSP.
• GWN hardware nodes are wireless routers running embedded Linux on Broadcom MIPS processors. They are also four-port Ethernet switches. They execute customized multi-hop routing extensions and configuration solutions software.
• CSP runs on x86-based servers running Microsoft Windows and specialized software (OPC Matricon, Matlab, MySQL).
Interfaces between subsystems in the project are realized by various technologies, depending on the capabilities of devices and implementation preferences:
• In-GWN: Proprietary network routing and management protocols on top of
• WLAN (WiFi), visible as fully-fledged IP network by users of GWN
• In-WSN: Inertia proprietary FastMAC
• Vibration sensor – wireless node: ICP (Integrated Circuit - Piezoelectric)
• MCN ↔ WSN Gateway: Inertia Serial Framing Protocol over Ethernet
• WSN Gateway ↔ JIM: Inertia Serial Framing Protocol over UART
• MCN – CSP: OPC through GWN.
Task 4.1: Multi-hop Networking and Network Topology (leader: UBER)
This task has developed reliable routing (networking) protocol and the network configuration and management services. The key developments in this task include:
• We have procured wireless hardware nodes and installed embedded version of Linux on them for further work. The nodes prepared in this manner were used as deployment platform for the routing and configuration software that we devel-oped in course of project.
• We have modified the existing OLSR 1routing protocol, so that it provides more reliable and efficient traffic routing in wireless multi-hop networks.
• We have provided a complete network management and configuration solution. The solution is based on standard protocols, with extensions that allow efficient network reconfiguration in the harsh industrial environment.
• The solutions were tested in factory of of one of partners. The routing pro-tocols showed performance improvements in comparison to its standard version. The network configuration operated as specified and was extensively used during the performance and reliability tests of the routing protocol.
Detailed description of the task
General Purpose Wireless Network provides connectivity between MCNs and the CSP. The main requirement is that GWN has to support use of OPC for communication between MCN and CSP. The GWN communication stack is executed on GWN nodes. The layers of communication stack that are addressed in PoPJIM are presented in light blue in figure that describes the architecture of the system. Networking layer is responsible for the address resolution of nodes and forwarding of packets to destination nodes. In case of GWN its focus in on reliable and efficient management of routes in multi-hop wireless communication between all communicating parties attached to the GWN. The network exploitation and maintenance functionalities need to be provided as well in order to allow the developed technology to establish foothold in practice. Due to core nature of PoPJIM as a Plug’n’Produce system, the components of PoPJIM system need to have capability to be dynamically reconfigured. This includes the GWN, so we develop reliable solution for the (re)configuration and management of GWN.
The distributed wireless control and configuration network will be deployed in a production environment of varying size (with regard to number of nodes and machines involved as well as to the physical area covered by the factory floor). This large area that needs to be covered is used for machining that affect the wireless signal propagation and reduce the communication range of wireless nodes: numerous walls of reinforced concrete, large metal machines and workpieces, electromagnetic noise created by working machines.
For these reasons, it may not be possible to provide direct (single-hop) wireless communica-tion between MCN and CSP. To overcome this issue, we develop a multi-hop GWN. A pair of nodes that want to communicate and that are not in direct wireless communication range needs assistance of other nodes in the network to forward their data, hop by hop, until packet reaches the desired destination. In order to provide sufficient reliability and performance of network, it is important to identify the best route (ordered list of nodes that will perform packet forwarding) between the communicating pair of nodes.
The network is run by programmable embedded systems: the LinkSys series of Linux WLAN access points. Nodes have Broadcom BCM5352 200MHz processor, WLAN and integrated 4- port 100Mbit Ethernet switch. They have 16 MB of RAM and 4 MB of Flash memory. They provide a stable Linux kernel, open drivers for the WLAN hardware and a customization op-tions for the networking layer.
We base our routing solution on the OLSR routing protocol, and we improve it to fit the needs of the PoPJIM project and industrial environment. It is a proactive routing protocol, where nodes exchange the locally obtained link information in order to build the global topology knowledge at each node in network. This topology is used by network nodes to determine the routes.
The OLSR protocol provides a satisfying solution with regard to its usual operation, but in course of project we have identified one weakness of it – its link quality estimation is inflexible and may not operate adequately in industrial environment (details and impact of it can be found in project deliverable D4.32). Inaccurate link quality estimation leads to sub- optimal route selection, unnecessary packet losses, and to issues of applications that use the network.
There does not exist a consensus on the best link detection algorithm and its parameters in re-search community. An example of behaviour of different link estimators on a dynamic link such as those in industrial environment is shown in Figure 4.3. Simple moving average SMA(10) estimator has inherent large error and is rather chaotic. SMA(50) and Exponential Moving Average - EMA(0.4 10) are rather accurate. EMA(0.1,30) performs very well when the link quality is changing slowly, but due to its slow reaction time, its error increases greatly towards end, when the link quality promptly changes. Obviously, their usability will depend on the use case scenario, as we have also shown in previous related work3 through rigorous stochastic modelling of a class of link detectors.
This implies that if an inadequate link detector is used in protocol, the protocol will not operate correctly – the errors in individual link estimation will accumulate and the protocol will choose sub-par routes. To correct this inflexibility, we develop an adaptive approach to link detection:
• We perform online evaluation of effectiveness of multiple link detection approaches and select the best one. This removes the issue of pre-selection of approach without feedback on its performance.
• The link detector algorithms are selected on link basis (each link has its custom detector), not on network basis (all links use the same detector and parameters). This further reduces the
• Errors in link state estimation process and improves accuracy of routing.
We have tested this approach on the set of data collected in Berlin’s large wireless multi-hop network that contains more than 300 nodes4 and on data collected in GWN testbed that we operate in the Institute for Computer Science at UBER. Out of the available data, we use 20% of lifetime of a link for the learning phase, to select the best link-state estimator. The remaining 80% of data of a link is analyzed as production phase where we apply the selected estimator and calculate its MSE on the data that is not encountered before. Our approach is consistently better than all pre-determined, network-based approaches. Its MSE is very close to the MSE of optimal approach (we define optimal as a single type of detector is used over whole link lifetime so that the MSE is minimal. Optimality comes from a priori knowledge of link behaviour - for this reason optimality is not reachable in practice) in all analysed cases. It is some 10 to 20% more accurate than the use-case specific best estimator, and within 10% of the MSE of the optimal case (Table 2 in project deliverable D4.3).
However, such learning approach that uses only the initial set of measurements (20% of avail-able measurements) for the decision making process is also static in its nature and cannot adapt to later changes in link and network behaviour: if conditions change, network will remain stuck with the type of detectors that was selected in the first (and only) learning phase, leading to inflexibility and issues.
To offset this issue, we have implemented continuous learning approach for link-state estimation. A node starts with an arbitrary link estimator. It collects data that is delivered by the detector for 10minutes (free parameter that can be configured) and then executes the de-scribed algorithm for estimator selection on the measured data. Link estimator is set to new type and collects its data for the next 10 minutes, after which the whole adaptation process re-peats.
The results of the continuous approach are very good – we were able to considerably reduce the MSE to approximately half of the previous learning MSE. It is even considerably better than the static optimal approach.
The approach is fully implemented in GWN and has two components. The described learning and decision-making component runs on the GWN configuration server. The learning component is implemented in Java, and is thus platform independent. Once it reaches decision, it sends commands to GWN nodes via UDP. We have developed an OLSR plug-in soft-ware component that replaces the standard OLSR link-quality estimator plug-in. It accepts commands and can switch between different estimators.
Network Monitoring and Management
GWN, as essential part of the PoPJIM concept, will be deployed in manufacturing companies where dedicated (wireless) network administrators will not be permanently available. Once PoPJIM enters deployments in production facilities, it will be used on multiple machines, for prolonged periods of time when system configuration and management becomes the major ac-tivity from the network perspective and one of detrimental factors for overall usability of the system. We improve system usability by providing monitoring and management facilities in network that are suited for the developed GWN and PoPJIM concept.
During system’s use, any of networked components may need replacement, it may be moved to another machine (i.e. attached to a different GWN), or machine operator may switch Ethernet ports to which a networked subsystem is connected to. For instance, a failed node is replaced by operational one. These common, operational changes in PoPJIM network Each GWN node has its own unique IP address assigned to its WLAN interface. This is the only fixed and necessary parameter that has to be set in GWN prior to its deployment (other-wise the GWN routing will not function). All other GWN node and network configuration parameters can be set during network lifetime.
Each GWN node has its own IP /24 sub-network for assignment of addresses to Ethernet hosts attached to its internal Ethernet switch. Each GWN node runs its own DHCP server that as-signs unique IP addresses to Ethernet hosts that are connected to one of the four available Ethernet ports of the node. GWN configuration server remotely manages and configures the Local DHCP parameters, such as the IP sub-network to be assigned to Ethernet ports, default gateway, default DNS, DHCP lease time.
We have developed configuration solution based on SNMP SET commands and the “extend” options in the SNMP daemon that runs on GWN node. The needed SNMP extensions are de-veloped in C language net-snmp-API for AgentX5 . They can be easily recompiled and deployed to other nodes than those that we currently selected for GWN implementation. The configuration scripts are implemented so that they can run on every Linux system, further ensuring the cross-platform portability of the developed solution. The configuration changes are saved by the GWN node in its configuration files, so even if GWN node is restarted, its configuration will be valid.
For purpose of name resolution, we have developed a SNMP TRAP extension that monitors the local DHCP configuration changes. The SNMP TRAP changes both the GWN configura-tion server and the local SNMP information base, so any authorized SNMP client can ac-cess the correct information. GWN Configuration Server accepts the SNMP TRAP (SNMP Master Component in Figure 4.4) parses the information from it, creates and executes (re)configuration scripts for its DNS server. As the result, the new IP address of the host that has just been connected to the GWN network has its entry in DNS server, which allows the resolution of its name to its IP address. Opposite is also valid – if an Ethernet host has been disconnected from the GWN node, its entry in DNS is removed.
must not impact the system functionality and this requirement renders any sort of fixed, pre-determined network and host configuration infeasible.
When a node (such as xPC-Host) is connected to the GWN, it must obtain a unique IP address to be able to communicate within it. An IP address is not permanently associated to a node. A node may obtain another (again unique at the time being) IP address at any time during its lifetime in system. Since it does not have a fixed IP address, a node should be addressable by its symbolic name. For this purpose, there must be a system that dynamically resolves names to IP addresses. Since IP addresses of nodes change, they must be accessible by their symbolic name (host name). There must be a system that resolves names to IP addresses.
These requirements are not new, or specific only to the PoPJIM system. The Internet relies on the same principles of configurability and provides tools that provide for these requirements. However, the PoPJIM system has some specifics that are different than in a typical enterprise Ethernet LAN. These specifics need to be taken into account while developing the network configuration subsystem:
• The (re)configuration traffic in GWN should be minimized. GWN, same as every system, is particularly sensitive to loss of reconfiguration data, so the less recon-figuration data we sent through unreliable GWN, the higher overall reliability is achieved.
• The machine should be able to continue operation even if GWN experiences an outage (for instance a GWN gets partitioned due to transient node failures).
• GWN nodes have available resources that are by orders of magnitude lesser than in case of their server counterparts used in enterprise LANs. This affects both available tools and design decisions that we make in the development.
To address these requirements, we have developed a hierarchical, loosely coupled configura-tion system (Figure 4.4). A centralized GWN configuration server is used for overall network management and configuration of GWN nodes. It prevents local configuration errors through remote configuration and monitoring of GWN nodes. Host configuration is performed directly by GWN nodes.
Our solution is based on well-established technologies and standards and guarantees interoper-ability with existing systems. This considerably reduces the network maintenance costs. We have customized existing software to fit to PoPJIM needs, and to capabilities of GWN nodes almost instantaneous updates in the network configuration, and at the same time it generates minimal traffic in GWN – the traps are activated only when it is necessary.
Experimental Evaluation of GWN in AFM Workshop
We have performed a GWN tests in AFM factory in Andrychov, Poland. The factory consists of three joined workshops (Figure 4.6) with numerous machines of different sizes. The factory size is 124 x 76m. We placed nodes throughout the factory to create a connected GWN (Figure 4.5). GWN network consisted of six nodes. Two types of measurements were per-formed, one for TCP traffic performance, to emulate behaviour of OPC RPC calls, and one for tests of the network control and configuration mechanisms. We analyse two principal cases: one that uses the default link detectors from OLSR and one that uses our adaptive approach. We first execute the default OLSR approach, and then use the collected topology data to execute adaptive link detection. In order to eliminate effects of transient wireless noise on measurements, we used a measurement tactics where we interleave standard OLSR link detectors and our adaptive approach. The order of execution and marks of measurements are: A: SMA(32), B: adaptive, C: SMA(32), D: adaptive.
Each measurement set consists of 50 samples for each of the measured categories (i.e. 50 samples of 5KB message size, 50 of 10KB message size, 50 of 100KB message size and 50
SNMP SET tests). For evaluation purposes we combine then measurement results from sets A
and C, and from sets B and D.
The measurements are heavily influenced by the surrounding levels of signal noise, as well as the internal TCP states, so there exists a large dispersion of measured values (a sample can be seen in Figure 4.7). In order to remedy large dispersion, we use median instead of the mean, as it is common for distributions with numerous outliers. Median is much more robust metric and intended for use in such scenarios.
The medians of all experiments are shown in Figure 8, and it can be seen that our adaptive ap-proach results in a slight but constant advantage over the default link detection in TCP transfer evaluation cases, and a considerable advantage for the SNMP-SET tests. It must be noted that although the overall results of the adaptive approach are better, in some experiments we have observed cases where the default link detection approach delivered better TCP transfer times. Obviously and as expected, there is a strong overall dependence between measurement results and current background noise on wireless, compounded with the long-term effects of internal TCP states on the flow performance. In all experiments is our approach provides a clear and notable advantage for the stateless forms of communications, such as the SNMP-SET commands where our approach results in consistently better reaction times than the default link detection approach.
Summary
We have developed routing protocol extensions and network configuration toolchain. Unlike many approaches from research, our solutions are fully operational and deployable. We have tested them in experiments in AFM facilities. The experiments have demonstrated that the de-veloped solution is reliable and robust (there were no node or software glitches in two days of experiments) and that our additions to the existing stack result in performance improvements.
Our initial notions about networking in factory environments were confirmed:
• There is no readily available Ethernet/WLAN connectivity at factory floor. The only Intra/Internet connected part of the factory are offices, which are separated from the workshop floor.
The wireless coverage from the office space is far from sufficient to cover the factory: our WLAN stations were running on full power but due to numerous obstacles in form of machines, walls and massive metal workpieces, we needed
six nodes to cover approximately 30 to 40% of the factory, with moderate to very bad connectivity with the other factory workshops.
Wireless multi-hop networks remains as the only viable and financially acceptable solution for Intranetworking in such environments. Any attempt at installation of Ethernet cables to the workshops would induce downtime of machines creating production breaks of prohibitive cost. With only six nodes we were able to cover 30 to 40% of the factory floor and provide Intranet for devices and machines that otherwise would not be able to access it. With a modest in crease in node number it is possible to provide full network coverage at a very low cost: the price of as single node is below 50€, so for less than 1000€ allows full Internet access.
Task 4.2: Wireless communication component (leader: Inertia)
The key developments in this task include:
• A high-speed wireless solution was developed, that fits the PoPJIM use case re-quirements (high-speed 5KHz data rate, synchronized among multiple points).
• Inertia's initial hardware platform was completely redesigned. New hardware is based on Cortex A3 processor, it has IP67 casing.
• Ethernet gateway was developed from scratch, to interface with the MCN developed in WP3.
• Inertia's real-time operating system ported on the new platform, Inertia's wireless FastMAC protocol ported to the new high-speed radio, integration with the Inertia ProMove GUI software for data acquisition. It is the fastest wireless sensor network platform currently on the market.
• A set of experiments was done in an industrial workshop in order to assess the wireless performance. The experiments showed a good overall performance, with the warning that instances of burst packet loss may occur.
• The integrated hardware platform was presented at several exhibitions (Hannover
Messe 2013, IDTechEx 2013, High-Tech Systems 2013, Precision Fair 2013.)
Detailed description of the task
The WSN hardware incorporates the necessary electronics, including microcontrollers, memory, radio, antennas, data interfaces, and casing. Software implements the Inertia Fast-MAC protocol that regulates the access to the wireless medium and data communication within the WSN as well as interfacing to systems with which the WSN interacts.
The extremely high sampling rates required controlling the AFM TAE-35N turning machine required introduction and use of a proprietary high speed 2.4GHz radio transceiver. that can communicate up to 4Mbps data rate in FSK or GFSK modulation. The maximum range/coverage for line-of-sight conditions is approximately 30 meters. The radios operate in the license-free 2.4GHz band. The main hardware subcomponents of the wireless nodes are:
• ARM Cortex M3 microcontroller implementing the application logic (sampling vibrations sensor, communicating with the MCN or JIM Local Controller) and the WSN MAC protocol. Memory/storage support.
• Interface electronics for the ICP accelerometer sensors.
• Radio transceiver and antenna
• Due to harsh environment in which nodes operate, the node casing was chosen to be according to IP67 standard.
Figure 52 shows the prototype of the WSN node with a modular design: the motherboard contains the microcontroller, radio part and storage support, while the daughterboard provides the electronic interface and powering circuitry for the ICP accelerometer sensors.
• The MAC layer is based on the proprietary Inertia Fast-MAC. This is a Time Division Multiple Access (TDMA) MAC protocol, optimized for accurate sensor sampling synchronization and timely transfer of sampled data. TDMA prevents packet collisions among communicating parties (each sender knows when to send and others are silent in this time period) and it introduces a known and fixed delay in communication (each sender has a fixed slot in a frame, so delay between two successive transmissions equals to the duration of the frame). Both of these features make it a perfect choice for real-time control application, such as in the PoPJIM project. Finally, nodes are tightly synchronized and they are able to perform accurate time-stamping of the data exchanged in the network, which is needed by the MCN control algorithm.
• Developed MAC protocol is used for the process monitoring (vibration sensing). Process monitoring comprises of two principal steps. The first is vibration sensing which is performed by wireless nodes placed on machine structure over the accelerometers that are attached to them. The sensors provide analog signal over ICP connector to the wireless node and this signal is converted to a digital value, suitable for transmission through WSN. According to requirement specification delivered by WP3, the vibration data resolution is set to 12 bits. The sensor and the A/D converter capable of sampling at 5kHz rates, in order to capture the full span of vibration signal and its later reconstruction at the MCN. The accelerometer is external to node, and connected to it by a cable. Such configuration al-lows that sensor is placed closer to the machining process, without possibility of wireless node damage by the machining process, cooling liquid, and flying debris.
• A wireless node assembles the vibration samples in a packet. As soon as data payload of pack-et is filled, the packet is queued for transmission. Once the time slot of the wireless node is due, the packet is immediately sent to the WSN gateway, attached to MCN, using the Fast- MAC. The gateway then provides the vibration data to the MCN by Inertia Serial Framing Protocol.
Inertia High-Speed Wireless Platform Experiments in an Industrial Workshop
• The objective of the experiments is to assess the overall performance of the 4 Mbit high-speed radio used for the Inertia WSN. The experiments are performed in an industrial workshop environment (machine tools, metallic structures, people working with the tools and machines).
• Three different situations are investigated: line-of-sight, small obstructions (e.g. tools trolley between gateway and node) and large obstructions (e.g. machine tools or walls between gate-way and node). Each setup is tested over 3 different distances: 2, 5 and 10 m. The wireless node is sampling and communicating data samples at 9.6kHz. The experiments lasted approximately 1200s per scenario.
• Two additional tests are made with the wireless nodes placed on machine tools:
• The node is placed on top of an upright drill. The drill is not running during the test.
• The node is placed on top of a large milling machine running at 1200 rpm.
• For each test, the average packet loss percentage is computed. The packet loss is also translated into time gaps between two consecutive samples. Ideally, without packet loss, the time gap between two consecutive samples should be 1/9600 = 0,0001042s. The distribution of packet loss / time gaps over time was determined for each test. From it, we derive the mean packet gap and the standard deviation of it. Figure 4.11 shows the results of the experiments for nodes placed on various places in the workshop.
Additional tests were made with nodes placed on the machine structure, a case that is more relevant for the PoPJIM project. The experiments were run for approximately 60minutes, to observe long-time behaviour of the developed WSN MAC protocol.
The experiment setup can be seen in Figure 4.12. The nodes are approximately 3m distant. The packet loss remained low (0.434%), with mean packet gap of 0.0001046s. Figure 4.13 shows the histogram of the time gaps. It is important to note that the histogram is represented at logarithmic scale. Ideally, all values in the histogram should be grouped at the default value of 0,0001042s. The same figure shows the packet loss per second, as the test progresses over time. In the second experiment, the nodes were placed on operating milling machine, also on distance of 3m. The results remained similar, with very low packet loss of 0.217% and mean gap between packets of 0.0001044s.
.
Task Summary
The experiments indicate a relatively good overall performance, with less than 3.5% packet loss on average even in settings with large obstructions between the gateway and node. The behaviour is stable over larger periods of time. The operating conditions of the machine tool (with the wireless node placed on the working machine tool) do not seem to affect the wireless performance. However, there are a number of important remarks and findings that have to be carefully considered with respect to the possible adverse impact of the wireless over the control performance.
Firstly, the packet loss appears to vary with distance in a counter-intuitive manner when obstructions (both small and large) are between the gateway and node: the performance is better for longer distances. This is not the case in the clear line of sight case, where the packet loss increases with distance. A likely explanation is that obstructions placed close to either the gateway or node tends to limit the possibilities for reflected propagation of the radio signals. Secondly, it occurred in our tests that placing the gateway on a higher, non-metallic support improves the overall wireless performance.
Finally, it is important to analyze the statistics (histograms and distribution over time) of the packet loss. In some tests, even if the average packet loss is low, there are instances where a burst packet loss occurred. This means that time gaps of additional 100ms or possibly more may appear, where data is not received from the sensors. The controller needs to cope with such unpredictable situations.
Task 4.3 Configuration Support Platform
Configuration Support Platform, CSP is an enabler for the Plug and Produce property of a modular machine tool with joint interface modules. This document describes the use cases, architectural features and the implementation of a CSP. The document defines also the functions and features of the interactions to the DWCN to serve the control system and for configuration/reconfiguration tasks in WP2 and WP3. The functional architecture of CSP is composed of the Module library, CSP host and the communication interface.
The module library is the repository of the relevant module specification and configuration knowledge. The module specification should include the
• Basic module data such as JIM components both mechanical and electronic, the interface geometry (flat, tapered, etc) and assembly of the JIM, CAD models, etc.
• BOM, supplier of the module or component information need to be captured. The specifics of these are also depending on the business model used to acquire or use the JIM and its components.
• The JIM characteristics, i.e. the profile of the JIM displacement or force obtained from the JIM as the function of the applied pre-stress value.
• Considering future extension of the application to support reconfiguration of a reconfigurable machine tool, additional JIM characteristics such as module trans-formation matrix, stiffness matrix, compliance matrix etc need to be included.
Configuration knowledge including the dynamic behaviour of the configuration and selection rules to identify the best possible JIM. The dynamic behaviour of a given configuration where the particular JIM has been installed in the machine structure. These behaviours include the updated profile of the operational damping and the Eigen-frequencies as the function of time. The module Library is implemented in MYSQL database with the following data model.
The CSP host is the hub of CSP and provides the following front and back end services to the user:
• Front end services: via a user interface, the user can input the queries as specified in the use case descriptions above, get visual the required information to enable a selection decision or analysis.
• Establish communication with the repository for a reliable and secured communication with the repository
• Establish communication with MCN via the wireless network DWCN. Invoke the configuration rules and specific query functions (as described in section 3.4.2 to execute them.
The CSP Host is implemented in Matlab that provides not only a secured communication with the Module Library but also provides utilities and toolboxes for communication with the control system and the DWCN.
The communications interface with MCN is via the DWCN as stated above. CSP and DWCN are connected over Ethernet, same as MCN and DWCN. The communication interface of CSP should support reliable and secured communication between CSP and MSN. For data exchange, MCN and CSP use OPC communication middle-ware.
A user interface is also developed on the host for an offline access and analysis. The figure below shows the graphical interface for this purpose.
Following the architecture and the specifics of the chosen implementation application tools and architectural components, the Configuration Support Platform is implemented.
The communication integration between the controller (MCN) via the wireless net-work is achieved and for this the OPC protocol
Populating the module library, the knowledge model as well as the machining sys-tem configuration has been done. Three main features of the CSP architecture can be summarized as follows.
1. Open architecture: The CSP architecture is inherently open, i.e. any one of the modules can be upgraded or changed whenever better ones are available for use.
2. Extensible: The implemented architecture is extensible in its functionality, i.e. analysis can be added either by making use of the toolboxes in Matlab or integrating the analysis applications.
3. Capable of accommodating different business models. Acquisitions of the modules can be effected in different ways depending on the business model between the module supplier and the user. The model library, as implemented here, can be accessible to the suppliers over the internet so that they can add and update module’s specification.
Work Package 5 Demonstration
Contributions made by the partners with respect to the above mentioned objectives are the following:
• KTH >> Technical expertise for performing machining tests in order to acquire acceleration measurements of AFM TAE 35N machine included with active JIMs and to obtain experimental results with respect to real-time system identification and optimization algorithm tests. Also provided invaluable feedback on deliverables “D3.1 JIM Control Architecture Specification” and “D3.3 JIM Embedded Control”.
• INERTIA and HUB >> Technical expertise for the integration of wireless com-ponents of process monitoring stage with the control hardware. Also provided invaluable feedback on deliverables “D3.1 JIM Control Architecture Specification” and “D3.3 JIM Embedded Control”.
• HUB >> Technical expertise for the integration of wireless nodes (developed specifically for the communication aspect between JIM global control and communication support platform) with the control hardware. Also provided in-valuable feedback on deliverables “D3.1 JIM Control Architecture Specification” and “D3.3 JIM Embedded Control” and technical expertise for the integration of Inertia’s wireless components (developed for process monitoring stage) with JIM global control.
• AFM >> Technical expertise and invaluable time for performing machining tests on AFM TAE 35N machine.
• CEDRAT Tech. >> Technical expertise for the integration of Amplified Pre-Stress Actuators (APA) and Eddy Current Sensors (ECS) with the control hard-ware and AFM TAE 35N machine. Also provided invaluable feedback on deliverables “D3.1 JIM Control Architecture Specification” and “D3.3 JIM Embedded Control”.
• SORALUCE, S. Coop., >> Provided the basic hardware requirements for the setup of control hardware with AFM TAE 35N machine regarding the preparation of demo for Final Review meeting on 12.09.2014.
Task 5.1 Mechanical JIM Demonstrator
Damped Tooling System
Modal testing of the tools clamped in HSK100 tool holder indicates that the EDS coating material can improve mode damping of the structure without sacrificing its static stiffness. Machining tests were performed to determine the experimental stability lobes diagram for both the coated tool and conventional tool.
The machining results indicated that the damping coating can improve the critical depth of cut, from 0.2mm to 0.3mm when slotting in X direction and from 0.15mmto 0.3mm when slotting in Y direction. In total, the coating almost doubled the cutting capability of the process.
Important Note: The experimental results of each stage of control architecture correspond to the AFM use case (i.e. AFM TAE 35N machine included with standard turret and active JIM located at its tailstock side). Active JIM is a mechatronic component actuated using the Amplified Pre-Stress Actuators mounted along the walls of tailstock holder.
Demonstrator IDEKO/Soraluce
Mechanical realization
From previous investigations, it was concluded that the JIM approach was not appropriate for the low frequency structural modes damping and, for that reason, the AIM approach was posed instead. The AIM approach keeps the same concept as the JIM inter-face, as a stand-alone added module on the machining system that increases dynamic stiffness, is wireless and has a self-tuning capability, but the principle for dynamic stiff-ness maximization is based on the dynamic absorber effect.
As it was explained in the previous periodic report, two different designs have been carried out. The first one is a preload type damper, in which the stiffness is adjusted by a preload increase of a viscoelastic material. The preload is achieved by turning a two-direction screw, which in turns acts on a wedge that moves an especially shaped preload plate back and forth. This plate pushes or releases the viscoelastic material which surrounds the moving mass. In this case the damping is provided by the viscoelastic material itself.
The second design is a spring type damper, in which the stiffness is adjusted by turning a variable stiffness spring at a determined position. As the moving mass is connected through bearings to this spring, the natural frequency of the system is changed by turning the variable stiffness spring. The damping is provided through eddy current effect in this case. The eddy current effect is achieved by adding specifically located magnets to the moving mass. When the moving mass starts vibrating, eddy currents are induced in several fixed copper plates, which in turns create a force opposing to the mass movement, adding damping to the system.
The spring type AIM is unidirectional and its movement is guided through linear bearings, whereas the preload type AIM is bidirectional and its movement is guided by sup-porting balls. The approximate mass of the moving mass of both AIMs is around 180kg,
First, these two prototypes were tested in their passive form, that is, with no self-tuning capability. The cutting tests were performed after tuning the system manually. Once it was observed that the system was working correctly and that the cutting capability could be increased to a large extent, the self-tuning capability was added.
Self-tuning capability
Control algorithm
In order to provide the system with the self-tuning capability, a control algorithm was developed. An accelerometer was used to monitor online the cutting process and wifi ports were used to enable the communication with a control algorithm in a external PC, which sends back a proportional command in order to tune the AIM accordingly and maximize the dynamic stiffness.
The control algorithm consists of a chatter detection method, followed by main frequency identification and a AIM tuning in order to match its frequency to the identified chatter frequency. The AIM is tuned through the viscoelastic material pre-stress in case of the preload type damper and the spring orientation in case of the spring type damper. The software user interface shows the tuning position and an online time and frequency monitoring of the acceleration signal.
Physical implementation
The control scheme described above was physically carried out by adding an upper floor to the passive prototype in order to enclose the sensors, actuators and communication ports.
The tool is a special tool that allows machining at low depth of cut and high feeds. Due to the lead angle of its inserts, the force is oriented in the axial direction. This is a good characteristic for the milling machine itself, since the axial direction is the stiffest, but it is a problem when machining on a flexible fixture in the axial direction. In this case, as the workpiece holding system is a vertical plate, the dynamic flexibility of this in the axial direction will determine the cutting capability of the process.
Cutting tests layout
As described in previous reports, the bending mode of the vertical plate in Z direction is the main mode limiting cutting stability. A 75% improvement of the dynamic stiffness magnitude is achieved at the critical mode. When the AIM is activated, the cutting tests are turned from stable to unstable, reducing the vibration level and improving the surface finish. Cutting capability is increased to a large extent. Without the AIM on the vertical plate, chatter problems started from a depth of cut of 0.5mm whereas when the AIM is integrated, a depth of cut of 2,5mm is machined in a smooth way, This means that the machine’s cutting capability is increased in more than 5 times.
2.5.2 Self -Tuning JIM demonstrator
Process Monitoring - Wireless acquisition of acceleration signal & system identi-fication algorithm
The acceleration signal of AFM machine is measured with respect to wire-less communication medium and the resultant operational damping and Eigen frequency characteristics are shown. The graphs denoted in red correspond to the unstable machining cut with maximum voltage applied to the Amplified Pre-Stress Actuators (APA) and the graphs denoted in blue correspond to the stable cut with minimum voltage applied to the Amplified Pre-Stress Actuators (APA).
In the unstable machining cut, we have three vibration modes ω - 1 at 1600Hz, ω - 2 at 900 Hz and ω - 3 at 403.3Hz. Out of these three vibration modes, 4ω - 403.3 Hz is the dominant vibration mode. The operational damping which corresponds to the vibration mode ω - 3 at 403.3 Hz is negligible which indicates the existence of chatter phenomenon. In contrast to the vibration mode ω - 3 , the operational damping which corresponds to vibration modes ω - 1 and ω - 2 are 15% and 19% respectively.
Whereas in the case of stable machining cut, we have only one vibration mode ω - 1 at 1600Hz. The operational damping which corresponds to the vibration mode ω - 1 at 1600 Hz is 15% on average throughout the operating time window of the machining process which is being performed. According to the operational damping characteristics which correspond to the vibration mode ω - 1 at 1600Hz, the chatter phenomenon didn’t exist.
Eigen frequency characteristics of unstable and stable machining cuts in a specific time window where the effect of vibration mode ω - 3 at 403.3 Hz is clearly visible on the resultant surface finish of the bars shown in Figure 16. From Figure 3.3 the resultant surface finish of the bars with respect to unstable and stable machining cuts is shown. In Figure 3.3 the top bar and bottom bar represents the surface finishes which corresponds to unstable and stable machining cuts respectively. Based on the light reflections on the bars, we can tell whether the resultant surface finishes of the bars are good or worse. It can be clearly seen that, the dominant vibration mode ω - 3 at 403.3 Hz has a deteriorating effect (where chatter phenomenon is significant in between time instants ω - 1=26 ω - and ω - 2=34ω - ) on the surface finish of top bar which corresponds to the unstable machining cut. Whereas, the surface finish of bottom bar is rather good which corresponds to the stable machining cut.
According to Figure 1, Figure 2 and Figure 3 the conclusion drawn from the experimental results is that the process monitoring stage is able to provide the acceleration signals with respect to wireless communication medium and compute its resultant operational damping and Eigen frequency characteristics in an accurate manner.
From Figure 17, the experimental result of 1D optimization tested on AFM TAE 35N turning machine is shown. For setting a global pre-stress optimum on the active JIM, Gradient Descent based optimization algorithm is adopted. On the left side of Figure 17, the graphs denoted in green represent the measured acceleration signal with respect to Wireless communication medium, estimated operational damping and Eigen frequency. On the right side of Figure 17, the graphs denoted in blue are the gradient of the estimated operational damping and the direction signal computed from the gradient of estimated operational damping and finally the graph denoted in green is the ‘pre-stress’ of the active JIM computed in terms of the voltage applied to the Amplified Pre-Stress Actuators.
Since the pre-stress of the active JIM is computed in terms of the voltage applied to the Amplified Pre-Stress Actuators, there is no need to enable the JIM local control. Ac-cording to Figure 17, the pre-stress parameter is directly dependent upon the direction signal. This direction signal is computed based on the gradient of the estimated operational damping. The direction signal tells the pre-stress parameter in which direction it should start to update. 1D optimization algorithm is initialized after 1s. The active JIM parameter ‘pre-stress’ starts its update initially from higher voltage value of 42V and converges towards optimal value of 0V within 7s. Based on the earlier unstable and stable machining cuts, the conclusion which can be drawn from the JIM global control stage is that the gradient descent based optimization algorithm is able to set the global pre-stress optimum on a single active JIM.
In Figure 18, the graphs denoted in red and green the acceleration signals, estimated operational damping and Eigen frequency characteristics with respect to unstable machining cut and 1D optimization algorithm enabled machining cut respectively.
From Figure 18, it can be seen that the Eigen frequency characteristics were slightly shifted which correspond to the 1D optimization algorithm enabled machining cut. This gives an indication that the pre-stress which is being applied on the active JIM is trying to avoid the structural modes of the machine tool-workpiece system while it is excited by cutting forces. This is clearly evident from the Eigen frequency characteristics computed for 1D optimization algorithm enabled machining cut compared to the Eigen frequency characteristics computed for unstable machining cut. Also the operational damping computed for the 1D optimization algorithm enabled machining cut which corresponds to the vibration mode ω - 1 at 1600 Hz is slightly higher compared to the operational damping computed for unstable machining cut which corresponds to the vibration mode ω - 3 at 403.3 Hz.
From Figure 20, the resultant surface finish of the unstable and 1D optimization algorithm enabled machining cuts is shown. In Figure 20, the top bar and bottom bar represents the surface finishes which corresponds to unstable and 1D optimization algorithm enabled machining cuts respectively. If we have a look at the top bar, the surface finish.
Master Controller Node (MCN) – CSP Integration via GWN (General Purpose Wireless Network) Nodes
The integration of MCN (which is responsible for JIM global control stage) with CSP using GWN nodes was performed such a way that the communication between MCN and CSP exists in a wireless network setup. In general, MCN is located near the AFM TAE 35N turning machine. On the other hand, CSP is located far away from the AFM TAE 35N turning machine. The following Figure 13 describes the experimental result which describes how well the communication aspect between MCN (JIM global control) and communication support platform (CSP) works.
In Figure 26, the acceleration signal of AFM TAE 35N turning machine (included with active JIM at the tailstock side), estimated operational damping and Eigen frequency characteristics with respect to 1D optimization enabled cut are shown. Here the opera-tional damping, Eigen frequency and optimal pre-stress (computed in terms of voltage applied to the APA) parameters are being sent from MCN to CSP via GWN nodes using OPC Write functionality. On the other side, CSP is acquiring these parameters using OPC read functionality and holds these parameter values in its respective module li-brary. From Figure 13, as we can clearly see that there is no loss of information between MCN and CSP during the machining process.
In addition to these parameters, the following specifications which are static during the machining process are also provided to CSP:
Run Number
Workpiece Specifications
• Diameter in mm
• Length in mm
Process Specifications
• Depth of cut in mm
• Feed rate in mm/rev
• Spindle speed in rpm
JIM Interface Specification
• Pre-load applied on tailstock JIM in N-m
Using the communication aspect between MCN and CSP, the possibility of adding or removing the existing JIM modules without having a negative impact to the overall control architecture in order to meet the high damping requirement requested by an end user (in our case end user is the machinist) before the start or after the end of a machining process is realized is rather rough (chatter phenomenon is significant between time instants ω - 1=26 ω - and ω - 2=32ω - ) coinciding with the dominant vibration mode ω - 3 at 403.3 Hz.
To a certain extent the surface finish of the bottom bar is smooth with respect to the 1D optimization algorithm enabled machining cut compared to the surface finish of top bar which corresponds to the unstable machining cut. During the 1D optimization algorithm enabled machining cut, the force window which is available from the Amplified Pre-Stress Actuators is very short. Due to this particular limitation, the surface finish of the bottom bar doesn’t really match with the earlier surface finish of the bar which corresponds to the stable machining cut.
Potential Impact:
Strategic impact
The machine tool (MT) sector contributes to the productivity of virtually all industrial sectors, in that few products can be manufactured without the help of machine tools.
The industry of machine tools has been a fast growing market in the last years, led by technological innovation and precision. In 2008, the global production grew by 8% to 24.6 billion €21. Despite the current difficult economic environment, European machine tool industry is still leading the global market for the production of machine tools with a 44% market share. The competitive advantages of European Machine Tools industry resides today in a high innovative, diversified and precise offer. However, the European manufacturing industry is in need of modernization and upgrade following several years of underinvestment. To maintain for the coming years a secure position as sustainably world-wide leadership and competitiveness machine tools and manufacturing industry, European industry needs to apply unique and innovative technologies and to develop sustainable strategies and excellent trained workers.
Performance impact
By endowing future machine tools with new innovative intelligent and integrated manufacturing features well beyond the state of the art and requiring collaboration of among several scientific and technological areas, PoPJIM anticipates a strong impact on following industrial applications.
• Design and realization of new machine tools
• Users of machine tools for component manufacturing
• Improving the performance of existing machine tools
In this context, PoPJIM identifies three main industrial stakeholders: i) module providers ii) machine tool builders and iii) end users. By focusing on JIM (component) level, manufacturing unit level and manufacturing system level, PoPJIM defines intelligent and integrated manufacturing features as the application of advances in design and technological dimensions to achieve highly productive and reliable machine tools and manufacturing systems that are easily adaptable and reconfigurable in response to changing conditions and new requirements.
Improving quality
Component Level
• Introducing JIM component with known dynamic characteristics will drastically reduce the unknown random variations at joints generated by tribological conditions at interfaces thus ensuring a robust design with reliable functions.
• Simple and highly efficient self-adaptive control at JIM level will essentially improve the control performance and reliability (80%) in comparison to the by far complex conventional control at system level.
• JIM mechatronic self-adaptive control improve significantly the response time and reduces or eliminates the risk for human errors
• Multi-functional passive materials as primary constituents of JIM mechanical structure contribute to a strong improvement of dynamic performance
• Higher portability through JIM mechatronic component standardization Machine Tool (Unit) Level
• Dynamic model-based design methodology to model and simulate the machining systems leads to robust and predictable design of machine tool structures. This contributes to improving quality and notably shortened ramp-up time of at least 30% and in consequence to a reduction of the time-to market of JIM based MT.
• Modularity offered by JIM based MT will allow for high reconfigurability and adaptability to a wide range of products of complex geometry and various materials.
• Wireless distributive technology leads to non-expensive and secure industrial applications
End user level
PoPJIM enables higher quality of parts produced by JIM based Machining Systems
• Consistently running at their optimal capacity
• Receiving real-time information about JIM status
• Monitors actual cutting conditions in real-time and automatically self-adapt the JIM’s structural parameters to the level corresponding to the global optimum operation conditions.
Due to a more stable operation condition, up to 100% improvement in product non-conformance rejection and 75-90 % reduction in machine tool breakdowns and tool breakage can be achieved.
Increasing productivity
Increasing productivity in machining process demands high material removal rate
(MRR) in stable cutting conditions and depends strongly on dynamic properties of machine tool structure. New and wide business opportunities will be opened for knowledge driven European SMEs in the machine tool sector.
Machine tool level
PoPJIM creates the conditions for predictive models that are capable of determining the dynamics of different configurations without the need for vibration measurement repetitions. Model based is based on the predictability of joint interfaces behaviour and therefore it can be stated that up to 50-75% reduction in machine tool lead time due to concurrent design and development process is possible. PoPJIM will drastically reduce the non value-added activities at the end user facility by integrating vibration control solution during machine configuration thereby shorter time to customer can be achieved
End user level
Eliminate the needs for stability process analysis which becomes expensive and time consuming when experimental measurements are needed for each individual possible
Combination of machine tool configurations will result in significantly reducing the lifecycle costs of machine tools (20-30 times)
Since the damping in these JIM based machine tools due to added damping at the joint interfaces, will be considerably higher than the conventional machine tools, more aggressive machining parameters can be used to produce quality products at a much faster rate (1.5-17 times) and with maximum tool life increase (up to 200%) –with shorter lead times (up to 32%) at a lower production cost.
The enhanced capability of JIM based machine tools will minimize the number of operations necessary for producing parts as per specifications. System 3R has reported the reduction of number of milling operation in an automotive part production from seven to three through integration of a high damping interface in a workholding fixture. These high damping interfaces used in these cases are precursors to JIM.
Reduce the need for part inspection due to higher predictability of the JIMbased machining systems operating in safe conditions.
Through its self-diagnosis functions, PoPJIM will eliminate the time for unforeseen failures at machine tool structure level, while virtually eliminating the risk for tool breakage by reducing the dynamic loads at tool-work piece interface. The availability times of JIM-based machine tools will be increased in a proportion of 50-80%.
• Increase of the material removal rate while reducing the specific power consumption through reduced cutting forces and eliminating post operations.
• The energy saving can further be improved up to 30 to 60% by making use of lighter structural modules in combination with JIMs.
Industrial relevance and impact
Many manufacturers are being forced to become more innovative on how best to process parts and to improve quality and reduce total manufacturing costs. Additionally, changes in part requirements have also brought new challenges. High heat-resistant materials, improved strength-to-weight ratios, tighter tolerances, and finer, more controlled surface finishes have challenged traditional methods of machining. PoPJIM creates new opportunities for European manufacturing industry to bring itself to a higher level by offering total quality solutions and new approaches both at the machine tool development level and at the parts produced.
Example of industrial relevance
Impact in Internal Grinding Application: A typical example of industrial application is a supplier of high precision and high volume grinding machines. The extreme conditions with long and small bores and deep located seats with limited room for access, combined with very high demands for accuracy and a premium for short cycle time, require machine tool components with particular emphasis on dimensional stability, low vibration and high overall stiffness. Bore, seat and face on two diesel injection nozzles produced with company’s grinding machines are simultaneously finish ground to tolerances that should eliminate the need for super finishing operations. Grinding fuel injectors (Figure 15) are characterized by very high production requirements, small wheel diameters (2 to 6mm), limited coolant access, and extremely tight finishes and tolerances. Bore roundness is especially stringent and must be held to less than 0.5 μm. CBN wheels dominate and quills are carbide. As new manufacturing processes evolve, grinding quills become increasingly critical for high precision operations. They must satisfy unconditional criteria:
• Rigidity to withstand high pressure
• Lightweight to counteract critical eigen frequencies
• Perfect rotation to achieve speeds up to 200 000 rpm
There are some important aspects related to the grinding process:
1) The quill (Figure 16) bends during rough grinding of long and small bores. As a result only the inner part of the grinding wheel is acting. This causes uneven grinding wheel wear, which equals poor bore straightness, cylindricity and taper
2) There are numerous ways of mounting the wheels. Gluing direct to the carbide is problematic. Wheels are often supplied on wheel screws. Producing quills in carbide with high accuracy electrodischarge machine to cut internal threads for screws is very complex and expensive. Only few companies such as Håmex Hårdmetallverktyg Linköping can produce to the necessary tolerance. The drawback with gluing is that due to higher rotational speed the wheel is jerked out from the wheel. Screws cause eccentricities. A quill rotation of 2 to 3 microns can achieve a finish smoothness of 6 to 8 microns.
3) For new applications, especially grinding at the end of long bores, FEA analysis of the dynamic stiffness and resonant frequencies of the quill is required.
4) Feedrates are limited by chatter onset. With taper compensation, faster federates may be used for larger wheels and the control algorithm applied to gradually reduce rates to those of fixed infeed grinder as the wheel gets smaller.
5) Rotational speeds are limited to about 70% of the maximum values due to the resonance on the critical eigenfrequencies. By integrating a JIM module in the spindle elastic bearing support, another JIM at the interface quill-spindle nose as well by coating diamond layers on the quill by help of Pulse Plasma Deposition process of Plasmatrix Materials, PoPJIM concept claims the following benefits:
• An increase of straightness and cylindricity by up to 40% when grinding small fuel-injection components
• Meantime, the cycle time can be reduced by 30% through maintaining constants feed rates at higher spindle speeds.
• Increasing eigenfrequencies while reducing the resonance amplitude
• Increasing life time of quill by up to 100%
• Reduced number of experiment while designing robust quills and other
machine components
• Spindle holder controlled by JIM mechatronic system allows faster overall cycles times for weak systems susceptible to chatter
Impact on Milling and Turning Application: The application cases reported in the table below and Figure 17 clearly indicate the huge potential benefits of the increase in machining system dynamic stiffness by integrating High Damping Interfaces (HDI) in workholding fixtures and at selected locations in machine tools. Research results show that the use of HDI has resulted in better product surface finish, shorter cycle time and increased metal removal rates. The design solution with self-adaptive control for dynamic stiffness as envisaged in PoPJIM is far more comprehensive than simply using a workholding device with high dynamic stiffness. And therefore, this also offers much higher potential benefits such as higher metal removal rate, better surface finish, longer cutting tool life and less scrap of nonconforming components: overall this results in higher productivity, better quality also energy savings due to shorter cycle times.
Engineering relevance and impact
PoPJIM will significantly change the way the engineering activities particularly design of machine tools and process planning are performed in companies. The JIM concept can be extended to other production equipment, such as material handling where the dynamic phenomena affect performance.
Design of new machine tools will radically be changed from a trial and error activity to a model-based design due to the use of corresponding standardized JIM mechatronic modules with prescribed dynamic characteristics. Mechanical design time and therefore the led times will be dramatically reduced (3-4 times) due to reducing the number of cycles in the process sequence: design –prototype-testing.
Control design: A new control feature based distributed architecture will be integrated to control the JIMs via a wireless multihop network and DWCN. Wireless communication network is for the first being implemented for machine tools outside research labs. This is a completely new feature that ensures the performances as claimed by the PoPJIM concept and opens a potentially huge impact on the way machine tools are design and used. The application of smart materials will be introduced for actuation purpose.
Process planning for component manufacturing
The process planning is the most laborious engineering activity in a manufacturing company. As the majority of its phases are manually performed this results in a time exhausting activity with need for high qualified personnel. JIM based machine tool concept enables for the first time to plan the dynamic performance at process planning phase. As JIM library contains standardized JIM modules with specified range of dynamic load – stiffness/damping characteristic, the process planner has to select for each operation the optimal cutting conditions and the corresponding JIM modules.
Furthermore, supporting the process planning with knowledgebase and module libraries ensures reuse of configuration patterns and facilitates the process while minimising the human errors. In cases where the manufacturing system is composed of multiple alternatives JIM based machine tools for a given set of operations the impact can extend to optimise at the production planning and scheduling level.
Supporting EU policies and European dimension
The implementation of portable plug-and-produce for a new generation of production systems and generally accepted European standards for self-optimizing mechatronic components will be of vital importance in helping the European industry:
• For European machine tools will strengthen the competitive position in the world market.
• For the European manufacturing sector as a whole will translate into competitive advantages ranging from shorter product development cycles to new value-added products and services.
The PoPJIM project is part of the core of the European manufacturing strategy enabling at Machine-level technologies and systems that improve manufacturing productivity, quality, flexibility and safety and at Systems-level technologies for innovation in the manufacturing enterprise, including controls, sensors, radio frequency (RF) wireless networks, and information technologies; methods and approaches that improve design and decision making and integrated and collaborative product and process development
The “Vision 2020” document of ManuFuture has indicate among the drivers of manufacturing innovation, advances in science and technology, specifically in the fields of nanotechnologies, materials science, electronics, mechatronics, information and communication technologies (ICT) and biotechnology. New production processes based on new research, and the integration of hitherto separate technologies exploiting the convergence of scientific and technological disciplines, may radically change the scope and the scale of manufacturing.
The contribution of PoPJIM project to Manufuture refers to what Manufuture considers key technologies for the future:
Mechatronic modules will become essential parts of production machines.
Mechatronic production systems, where all the signals from mechatronic components are integrated for complete manufacturing control, are seen as greatly aiding the rapid reconfiguration demanded for flexible and high-speed manufacturing.
Mechatronics and process control are two of the key research areas for the future.
With major breakthroughs in three technological areas, (i) JIM mechatronic component and EDS multifunctional material, (ii) self-adaptive active control and self-diagnoses and (iii) distributive wireless network, PoPJIM will contribute to generating high added-value products. PoPJIM will stimulate the creation of a European knowledge platform with great potential at both machine tool development level and at manufacturing industry level and consequently contributing to transformation of European industry from a resource-intensive to a knowledge intensive industry.
Contribution to European societal objectives
Quality of Life of the European citizens
PoPJIM implemented in machine tool industry and in manufacturing technologies have the potential to improve quality of life by means of new and enhanced products and services, generating wealth and employment. Research activities in the field proposed will provide economic growth and the consequent wealth level needed for social and environmental improvements at the basis of a sustainable development. It will not only provide innovative solutions to existing problems but will also offer new opportunities through the development of innovative materials (EDS multifunctional material) and innovative mechatronic devices (JIM), European designed products and processes with new functionality with the potential to improve the quality of life for everybody. Furthermore, the PoPJIM will allow module suppliers to customise their products and services to the OEM manufacturer requirements.
Customisation means that the OEMs will not only be capable of bringing flexibility into their production processes, they will also be able to switch to niche markets for JIM based machine tools swiftly and efficiently, leading to high quality products with significant added value. In particular the project will lead to high quality, superior products in the field of high precision industrial components. Eventually it will lead to a strong and economically sound industry and to the creation of additional jobs. Research activities in PoPJIM could lead to successful mass production of many consumer product that would otherwise be manufactured in Asia. European employment can be safeguarded, in the sector of manufacturing that has been under pressure of low wage country competition for years. Contribution to the European Skills Base
This project will contribute to the preservation of economically important machine tool industry for Europe, by offering high added value in products, processes and services. Moreover, this project will lead to the enhancement of the competitiveness of the manufacturing industry, leading to a continuous growth in employment. A very important benefit of this project will be that it will increase the attractiveness of the industry for young professionals, as jobs will become more challenging and more rewarding. The daily tasks of workers in the manufacturing industry will become more flexible, more attractive and more complex, as more engineering skills will be required. Therefore education and training of workers will be an integral part of the future implementation and exploitation.
The PoPJIM entire project concept brings together two key research themes within FP7 namely the NMP and ICT themes, together and creates a venue to exploit the synergy between them and to advance the knowledge gain in their respective domains.
European scientists are recognised worldwide for their leading-edge research in the field of process control and optimization, plug and produce techniques. This is a crucial area for European level research. Thus, excellent technical solutions will give the EU as a whole a large competitive advantage to develop new leading industries in this field with vast economic potential. This will result in an increased magnetism of knowledge intensive SMEs on young technical professionals, because of high quality jobs. Should consortium partners not undertake this project, it is likely that the solutions envisaged will be developed in the USA and/or Asia and that the industry will shift towards Asia because of lower production costs and the steady fading of the present added value differential. This would not only result in larger unemployment figures in the industry, but will also imply that European industry will buy products developed and manufactured outside of the EU. Train both post docs and PhD students in multiple academic and analytical environments, by providing for exchange among the different institutions and labs brought to the project by the participants. Such exchange and exposure will inevitably provide a small well-trained body of young scientists who could lead technical efforts at research or industrial areas.
Task 5.1 Dissemination of project activities and results
Introduction
The overall objective of PoPJIM Project is to develop Plug and Produce Joint interface Modules. These modules are mechatronic elements located in machine tool joints in or-der to improve dynamic behaviour and therefore enhance cutting performance.
The objective of the Dissemination Plan is to identify and arrange the activities to be performed in order to spread the knowledge and achievements generated during the pro-ject and promote the commercial exploitation of the project’s results. It is essential that the dissemination of the results reach the maximum possible audience. For that purpose, the active participation of all project partners is needed in order to spread project results to different sectors, countries and fields.
Dissemination means and contents must be restricted to preserve project confidentiality and protect project results. Therefore, every public exposure of project results must be previously approved by the PoPJIM Consortium. The content of this document is subject to updated along the project life.
Standard dissemination material
The first task will be to prepare effective dissemination standard material that can be used for project overview exhibition. The following physical dissemination material will be created and, if needed, periodically updated as the project advances:
• PoPJIm Logo
• PoPJIm Poster
• PoPJIM Flyer
• PoPJIM Brochure
This material will be uploaded onto the Project Management private section on PoPJIM’s web-site, under WP6: Dissemination and Exploitation folder.
Dissemination channels
The periodical release of news, reports and multimedia material communicating project progress and achievements is of capital importance for project dissemination purposes.
PoPJIM website
PoPJIM website will be the main communication channel for PoPJIM project. This website will consist of a public and a private section. The private section will be used as a project monitoring and file and report administration site. It will be a shared domain for the communication among project partners. The public part will include project gen-eral information, Consortium details and a living section of project news related to pro-ject progress. All project partner’s own website will contain a link to PoPJIM’s website.
LinkedIn Group
A LinkedIn project will be created in order to create an open discussion forum for pro-ject technical and commercial issues. Invitations to project partner’s partners will be done in order to have a feedback and an external view from other technicians and stake-holders familiar with the project related activities.
PoPJIM - Plug and Produce Joint Interface Modules
Youtube channel
A Youtube channel will be opened in order to create and show project related multime-dia content, particularly for the demonstration works.
POPJIM Group
Scientific and conference papers
PoPJIM’s project development will generate scientific paper publication opportunities related to project’s findings and advances. These papers must be related to project’s side technology and in no case will disclose technical details about PoPJIM’s main objec-tive.
Both scientific journal and congresses will be considered.
Technical article
Other non-scientific but technical publications, closer to the industrial environment will be also produced. This will mainly include informational and pedagogical literature in order to raise PoPJIM project’s awareness in technical or sectorial journals.
Company’s newsletter
It is not the aim of the Consortium to create a periodic Newsletter for PoPJIM, but the inclusion of project’s main milestones in each partner’s particular newsletter (in the case it exists) will be promoted. The reason for this is that already settled newspapers from Consortium companies and research entities have a wider scope than a newly created project specific Newsletter would have.
Project results exhibition
Apart from the standard project dissemination channels, other dissemination events will be used for project result dissemination and promotion. In these dissemination events, standard dissemination material (poster, flyer and brochure) as well as the material re-leased through the foreseen dissemination channels (Youtube Channel, Technical Re-ports…) will be used. Also physical demonstrators will be shown.
Fairs and commercial events
General purpose fairs or company’s commercial events will be used for PoPJIM project result exhibition and promotion.
Courses to stakeholders
RTD entities will be in charge of organizing courses to stakeholders to inform about currently existing problems in machining and train about how to solve these problems through PoPJIM’s solution. Also individual project results will also presented in these courses.
Coordinated Seminars
PoPJIM project will interact with other ongoing European Research projects that are dealing with chatter avoidance solutions in order to prepare a Seminar in which different approaches for the solution of similar target problems will be demonstrated. The objective is to enhance the dissemination impact, increase the number of potential audience and take advantage of senergies among the involved projects to explore new variants in project’s objectives. The venue for this Seminar must be an important machine tool and manufacturing fair or event.
Dissemination procedure
Every dissemination procedure will follow the steps described below:
When a significant result or event is achieved in the project, the partner responsible for the realization of that result or event will create a piece of news to include in the PoP-JIM website. This piece of news will be sent to Ideko, who, as the responsible for dissemination, will decide if the information discloses project’s critical foreground. If this could be the case, it will be circulated among all partners for study and modification if needed. Once the note is edited, it will be forwarded to Humboldt for publication in the website. These news could be also published through different dissemination means foreseen in the dissemination plan: newsletters, papers, technical articles…
On the other hand, all the other dissemination events will be also tracked and published in PoPJIM’s news section. Therefore, the responsible partner of every dissemination event will be in charge of creating a short note for publication on project’s website.
It will be very important to try to document both project achievements and dissemination events through pictures and videos for the enhancement of PoPJIM’s Youtube Channel.
All Scientific Papers and Technical Articles must include the following sentence for PoPJIM project addressing:
The research leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 260048.
Exploitation of project results Introduction
1. Introduction
Realization of the project concepts will evidently be through a concerted innovation process that will bring multiple exploitable outcomes. The intent of this exploitation plan is to
• identify the potential exploitable results,
• assign rapporteurs to each identified exploitable items to further develop exploitation venues as the project progresses
• to identify and detail the background knowledge brought by the consortium partners in relation to each exploitable item
• identify the mode of appropriate exploitations, i.e. license, produce and sell, internal use, consultancy, etc, .
• perform preliminary multi aspect risk assessment, i.e. technical, market, legal, etc
For the first task, to identify possible results and characterise them, an initial estimation of development costs, market situation and result protection means is done. Corresponding to each identified exploitable item actors and rapporteurs were assigned.
Then IPRs are analysed with respect to each items. Every partner claims background and foreground IPRs. This will be helpful to distinguish between knowledge brought to the project by each partner and knowledge acquired during the project. Exploitation claims are also made, this way the list of partners involved in the exploitation of each result and exploitation means will be clearly stated. All this information will be helpful for future exploitation and co-ownership agreements among partners.
Finally a risk assessment of each result is performed. This tool enables the identification of the critical aspects that jeopardize successful exploitation of each result. The risks are evaluated as a combination of impact and likelihood and, in order to minimise or eliminate these risks, solution and action points are planned. Each action point will have a responsible for its fulfilment
Therefore all the information contained in this document will pave the way for the corresponding business plan implementation and future cooperation agreements. This document will be an evolving document and will be updated as the project progresses under consensus among all partners.
During the preparation of this document, a one day Seminar was conducted on the 18th of October of 2011 in Steyr (Austria) in collaboration with the European Commission in which a more structured approach was used to outline the plan on which the content of this deliverable is based. In this Seminar the basics for the definition of a successful exploitation plan were addressed and the specific work on exploitation issues for PoPJIM project was carried out.
Consequently, the tables and their content in this document are largely shaped as a result of the seminar and the subsequent discussion conducted within the consortium.
List of Websites:
www.popjim.com
