Final Report Summary - IREMO (Intelligent Reactive polymer composites Moulding)
Executive summary:
The IREMO project (intelligent Reactive polymer composites Moulding) involves 10 industrial, academic and Research and Developement partners from France, Spain, Greece, Germany, Cyprus and the UK. The target is to develop process monitoring and intelligent control technologies for the intelligent processing of liquid composite moulding and pultrusion aiming at the acceleration of the curing stage, reduction of development time and trial-and-error solutions and ensure product quality. The overall objective of the project is to develop all the supporting science and technology for the optimal production of polymer composites in closed moulding.
The scientific and technological achievements during the three years of the project are:
1. The development of optimal and self-learning process control tools that ensure the appropriate impregnation of fibres and the optimal curing of the composite parts resulting in more than 25% cycle time reduction at all industrial conditions considering production constraints and targets.
2. A process monitoring system capable of monitoring the full range of all major types of reactive polymers processing such as epoxies and polyesters with adequate accuracy and repeatability while providing in-situ control quality capabilities by relating the sensing signal to the materials' properties using reliable modelling tools.
3. A process modelling tool able to provide detailed process information and predictions for a wide range of polymers through an in-situ self-learning procedure.
4. Wireless communications and networking to minimise the wiring in the shop floor and development of a user-friendly interface tool.
5. Retrofitting of the control tool to an existing 2-components injection machine to become fully sensorised to measure material-state and control temperature automatically.
6. Integration of the monitoring and control system in a vacuum infusion process for large and thick composite parts and in a light RTM process for fast production of low-cost composite parts.
Project Context and Objectives:
Reactive polymer (thermosets and few thermoplastics) liquid composite moulding is used more and more into aircrafts, cars, boats, sports, construction, health, energy and almost in all industrial sectors because of numerous advantages. However, one large disadvantage is their complicated, expensive, risky and slow production hindering its wider industrial acceptance. The most common 'problems' in the production of polymer composite parts are the fibre impregnation with the polymer and the curing of this polymer until its consolidation. While fibre impregnation is very important and curing dominates the production cycle there are several conventional processes such as the hand layup, pultrusion and filament winding that the problems are easily detectable and solvable by the operators. Whereas in modern processes, such as the reactive polymer liquid composite moulding (Resin Transfer Moulding and its variants, Vacuum infusion but also Sheet Moulding Compounding (SMC) and Structural Reaction Injection Moulding (SRIM), the problems are significant, the existing knowledge is limited while there is no possibility to realise the problems and act on them as they occur. In a typical reactive polymer composite moulding production cycle after the fibre placement and the mould clamping and the cavity completely filled with resin, the polymerisation phase follows with or without external heating capturing from 65 up to 80% of the cycle time which also may vary depending on the application from a few minutes (automotive) to days (large boats). This a problem of all reactive polymer manufacturing but in the case of liquid composite moulding the problem is worse as the material and process deviations are significant due to the in-situ fibre impregnation (pre-pregs have a standard and much higher quality) and the cheaper production facilities.
In order to solve these problems turning reactive polymer liquid composite moulding to an agile industrial process it is important to have:
- automatic process control tools ensuring the optimal moulding of the materials
-self-learning process control tools that can handle efficiently production and material deviations without requiring extensive lab trials and scientific models
-production databases and self-learning material models that can provide the essential information to cope automatically with different matrices, batch-to-batch variations and small scale productions
-Robust and reliable sensors that can sense the state of the moulded materials inside the mould cavity, providing reliable and real-time information on the status of the process.
IREMO concept
The concept of our approach is to address all of the above issues in a holistic and systematic way in order to make reactive closed composite moulding an automatic, flexible and efficient industrial process. The proposed strategy to reach this target is based on the experience gained with the advanced control concepts developed for batch-to-batch processes in the chemical processing industry and a Receding Horizon Model Based Control developed and applied for closed composite moulding. The new elements will be the introduction of self-learning control schemes in conjunction with optimisation tools while advancing and integrating all the required supporting technologies i.e. self-calibrating process monitoring, self-learning process modelling and data analysis and communication tools.
IREMO Scientific objectives
The general scientific objective of the project is to develop all the supporting science behind the successful production of reactive polymer composites in closed moulding. More specifically the main scientific objectives of the project are:
- Development of an optimal self-learning process control that would ensure the appropriate impregnation of fibres and the optimal curing of the composite parts.
-Real-time optimisation of the polymerisation of fibre-reinforced polymer matrix in real production conditions considering all production constrains and targets.
-Relating the evolution of the sensing signal to the materials' properties using fast self-learning Neural Network-based tools
- Development of self-learning material models capable in providing the most valuable information and prediction based on in-situ process data rather than on extensive lab analyses.
IREMO Technological objectives
IREMO aims to develop an innovative moulding system for the efficient and intelligent mass production of reactive polymer composite parts.
The project has developed all the supporting technologies for the efficient and intelligent processing towards fast and scrap-free production with the following main technological objectives:
- A fully automatic, self-learning production process ensuring an optimal production cycle (20 to 25% cycle reduction) and low scrap ratio (below 1%) even during start-up.
- A process monitoring system capable of monitoring the full range of all major types of reactive polymers processing such as epoxies and polyesters with adequate accuracy and repeatability while providing in-situ control quality capabilities.
- A process modelling tool able to provide detailed process information and predictions for a wide range of polymers through an in-situ self-learning procedure.
- Wireless communications and networking to minimise the wiring burden in the shop floor.
- A retrofitting to an existing 2-components injection machine to become fully sensorised to measure material-state and control temperature and mixing ratio of at least 3% accuracy automatically.
- Integration of the monitoring and control system in a vacuum infusion process for large and thick composite parts.
IREMO baseline and performance
As IREMO is a performance-oriented project it is also very important to define the performance baseline with which the new systems would be compared to. However, this baseline is subjective to the technology of each industrial partner and to the specific application so this baseline will be clearly defined in the first months of the project based on the difference between the expected and the achieved values regarding set-up time, cycle time, start-up time, resources, scrap ratio and costs for each of the selected platform-cases.
Project Results:
WP1: System's specifications, requirements and targets, initial installations and data collection
The main objectives of Work Package 1 were:
i) to provide a detailed definition of processes,
ii) to install the process monitoring systems and
iii) to start the data collection.
Task 1.1 definition of processes, targets, requirements and specifications
Pultrusion case : ACCIONA
The product is a superstructure used in Building and Construction. Among the various advantages associated with it are light weight, ease of installation and high strength.
Production Set-up
The mould consists of three heating zones with one temperature sensor in each zone. The cavity size inside the mold is uniform except at entry and exit (a slight radius is provided on each side for avoiding damage to the raw materials).
Production machine
Caterpillar type Pultrusion machine of 5T Pulling force has been used. Temperature is controlled automatically.Pulling speed is controlled manually.
Infusion (ACCIONA)
The product is a superstructure used in Construction. Among the various advantages associated with are light weight, ease of installation and high strength.
Currently existing control capabilities: mould opening time and change catalyst ratio for faster production
High Speed RTM (SOTIRA)
The products are an outer door and fenders for automotive vehicles.
Production Set-up: The mould is equipped with both injection and vacuum heads, an injection channel and a seal.
A variety of requirements have to be met by order to be adopted by industrial practice:
- Functional requirements: The basic functionality of wireless sensor networks in IREMO is the permanent acquisition and transmission of process data during the production of reactive polymer composite parts. The sensor data will ensure an optimal monitoring of the moulding process in terms of efficiency as well as product quality. Concerning the storing of these data, a storage strategy must be developed to hold, share and protect valuable data about the moulding process.
- Requirements regarding embedding: The solution must be a retrofit of existing plants. This means that all required sensors can be integrated in the existing moulds. In addition, an ICT-infrastructure has to be installed to assure the acquisition, transmission and storing of sensor data.
- Performance requirements: Requirements concerning the performance of the sensor solution primarily arise from the complexity and size of the products or moulds and the placement of sensors, the resulting number of sensors, the sensor network topology, the data collection strategy and particularly the sensor hardware performances regarding transmission rate, band width and power consumption. In order to derive and initiate necessary measures in real-time, data collection, transmission, and processing must be extremely efficient and reliable, especially considering the amount of data expected for big parts.
- Handling requirements: Human-machine interfaces must be created that allow a comprehensible display and documentation of gathered information about the current status of the manufacturing process. This requires the design of an user-friendly use model, which visualizes the process sequence and initiates the measures of the control system. The system must furthermore recognize internal errors such as malfunctioning sensors and respond to them if necessary.
Neural network modelling of nonlinear processes UNEW
UNEW has extensive expertise in neural network modelling of nonlinear processes, optimisation control of batch processes, batch to batch iterative learning control of batch processes with applications to batch polymerisation reactors and bio-production processes. The expertise can be directly applied to the modelling, optimisation, and control of polymer composite moulding processes, which are batch processes in nature. UNEW will cooperate with other project partners in control objective specification and control strategy determination.
BENCHMARKING (ATOU)
A benchmarking of the existing competitive technologies has been realized. The objective of this work was to identify the main competitive systems already on the market, or in development, and to collect data concerning their principle, their characteristics and the suppliers which develop these systems.
The competitive systems have been divided in four sections according their characteristics:
- IREMO competitive sensors for real-time cure monitoring of FRP parts.
- Real-time monitoring systems of industrial processes.
- Real-time control systems of industrial processes.
- Self-learning control systems of industrial processes.
T1.2 Definition of pilot case plant
ACCIONA (infusion)
In order to monitor:
i) the arrival of the resin at two locations (i.e. the beginning and the end of the filling)
ii) the cure inside the mould and the lines,
A minimum of 2 mould sensors and 1 in line sensor is necessary
SORA (RTM)
In order to monitor
I) the arrival of the resin at two locations (inlet and outlet)
ii) the cure inside the mould and the lines ( as close as possible to the injection gate)
A minimum of 2 mould sensors and 1 in line sensors is necessary
Karnic (Light RTM)
In order to monitor:
i) the arrival of the resin at two locations (inlet and outlet)
ii) the cure inside the mould.
A minimum of 2 mould sensors is necessary
ACCIONA (Pultrusion)
In order to monitor:
i) In the die
ii) At the die exit
iii) In the resin bath
iv) the cure inside the die and the resin bath
A minimum of 3 mould sensors is necessary
T1.3 Preparation and installation of monitoring units
Monitoring systems have been delivered by SYNT to ACCI, NTUA, CEMC, SORA,TECN, KARN and replacement sensors also.
T1.4 Data collection, requirements and specifications
The number of acquisitions carried out with the monitoring units installed during the first period of the project is about 5 to 10 for pultrusion, 10 for infusion, 75 for light RTM, 150-200 for RTM and more than 150 for lab scale trials.
WP2: Development of self-learning optimal process control tools
The main objective of this WP is to develop the generic self-learning and optimal control tools and control platform. The basic platform for the control tools has been based on Labview (National Instruments) as well as algorithms coded in Matlab and C. The control platform is using simulation tools based on partial differential equations (PDE) and Neural networks for self-learning in conjunction with the fuzzy logic environment to maximise the applicability and the robustness of the developed tools.
?2.1 Control Strategies
Numerous important parameters that affect the production were identified: Resin, mould or die temperature, resin, flow rate, speed process, curing time and related demoulding time, pressure, vacuum, and cleaning/rinsing of the feeding lines. One of the main targets is the development of a friendly user interface i.e. a computer screen indicating to the operator all necessary information: experimental parameters, degree of cure, alarms etc. if some automatic actions can be envisaged in the control process, it is preferable that the final decision remains in the hands of operators (ex: opening of the mould, important change of the temperature .), and this for safety reasons.
The targets that were identified for the control system is the determination and the display of:
- degree of curing
- glass transition temperature
- gelation time
- viscosity
These properties have to be extracted for all production applications (batch-to-batch variations, conditions, fibres etc) and are of primary importance to optimize the following parameters:
- Mould opening (SORA/KARN/ACCI)
- Mould temperature (SORA/KARN/ACCI infusion)
- Partially or fully rinsing of the feeding lines (SORA/KARN)
- Time of demoulding, with or without post-curing (ACCI infusion)
- Pulling speed (ACCI Pultrusion)
- Mixing ratio check (SORA/KARN/ACCI infusion)
The quantifiable targets were set with respect to the characteristics of the initial process:
- SORA/KARN: time of curing reduced by at least 40%
- SORA/KARN: Wasted resin reduced by at least 70%
- ACCI infusion: time of curing reduced by at least 40%
- ACCI Pultrusion: increase of performance by at least 40%
?2.2 Development of Neural Networks and Optimisation tools
Neural Networks Tools
Neural network modeling of reactive polymer composite moulding processes has been carried out using both simulated data and experimental data from the simulation programme supplied by NTUA, 7 batches were simulated and the sampling time is 4 minutes.
The Robust Evolution Strategies
The Evolution Strategies emerged from the Genetic Algorithms and are better suited to optimisation problems. Their fundamental advantage over the conventional optimisation methods is that they do not use any gradient information so it is more difficult to get trapped in local minima. In this task an alternative Evolution Strategy, namely the Robust Evolution Strategy (RES) was developed to deal with the inherent variability of the basic parameters of the models that are used for the model-based optimization and control. This new algorithm provides the optimal solution of a mathematical problem within a realistic variation of its major parameters. In this way the provided solution is not optimal just for a single combination of model parameters for within a realistic range of these parameters. In the case of composites manufacturing we have the model parameters which for example may be some or all of the parameters of the kinetic model of the resin and the control parameters which may be the temperature profile. Seeking for the optimal temperature profile we can include the model parameters variation in order to ensure that even for the expected variations in the model parameters we can still have the optimal profile.
T2.3: Batch-to-batch control tools
Mould opening prediction (controlling mould opening)
The static model can predict the time to open the mould for any given moulding temperature in advance (using the predictions of degree of cure and Tg at different times). The mould open time can be obtained by solving an off-line optimisation problem, for example, minimise curing time so that the degree of cure reach 97% of the maximum degree of cure. Using the neural network model for the SOTIRA data, mould opening time (curing time) is obtained in relation to curing temperature.
Batch to batch iterative learning control
In this study, the batch time (moulding time) is fixed as 28 minutes and the sampling time is 4 minutes. Thus the batch is divided into 28/4=7 stages. The final degree of cure is the controlled variable and the applied temperature is the control action which is a vector with 28/4=7 elements. Note that it is also possible to control the degree of cure trajectory during the moulding process. In this study, two cases with different control actions are considered. In Case I, the applied temperature is kept constant during each stage and U contains the 7 piecewise constant temperatures. In Case II, the applied temperature is linearly rising during each stage at a particular rate and U contains the 7 temperature increasing rates.
Control Optimisation tools
Process robust optimisation is necessary in order to ensure that the production is optimised in a robust way always taking into consideration materials deviations and process constraints and limitations. This optimisation is important that it is done as continuously as possible to ensure that the process parameters are always optimised. At the past this process optimisation could be done only offline due to the requirements of increased computational effort so the batch-to-batch or at most the run-to-run cycle optimisation was possible. However as computational effort is significantly improving an online version will be also tested to check the capability for real-time cycle optimisation. In any case the concept of cycle optimisation is the same: the temperature profile is shot in order to achieve the best solution taking into account process constraints and limitations.
Optimizing the curing process may have various meanings so it is important to integrate into the process models simulation tools or at least criteria that will provide the quantification of the optimality criteria with respect to certain characteristics of the process:
1. Targeted degree of cure
2. Minimum processing time
3. Minimum exothermic peak temperature
4. Minimum curing generated internal stresses
5. Minimum cooling generated internal stresses
6. Homogeneous degree of cure
7. Minimum temperature gradients
T2.4: Development of the control platform and data manipulation tools
The control platform is a real-time tool for the control of the manufacturing process. It includes all the algorithms for the real-time calculation; it is responsible for the communication between the other control modules and also has a user graphical interface for handling also the input/ output dataflow. Although initially the Fuzzy Logic was researched to deal with the several data and user specs finally due to its complex structure it was decided to use a rule-based approach with specific thresholds for each variable depending on each process.
The Production Control Tool
The Production Control Tool has the basic user interface and graphical results and controls all the primary production equipment such as the injection machine, the Press or the heating/cooling devices. In particular the production control tool has the following tasks:
1) To collect the data from the process control tool (resin arrival, degree of cure, control decisions)
2) Sent several information to the process control tool (state of the press, state of the injection, target Tg, resin type etc)
3) Sent to the trigger box the following information (start of acquisition, decision of opening press)
4) Sent data for the processing data storage (Sensor data, time stamp, part name, part number, name of operator, resin type)
T2.5: Integration of the control tools for low-volume production
Following the work in T2.4 the main characteristics of the low-volume production such as new resin models, temperature effects and targets have been incorporated successfully for three sensing points wand can handle flexible but also durable sensors (different parameters are necessary for each sensor type). Tests at lab-scale conditions but also simulated trials in Acciona showed initially some problems which were successfully overcome and the control system is working very satisfactorily. It should be highlighted that this is a generic platform although based on the specific resin it can handle a large variety of temperatures and processing conditions e.g. carbon fibres, flexible/durable sensors etc.
The Production Control Tool
The production control tool has the following roles:
1) To collect the data from the process control tool (resin arrival, degree of cure, control decisions)
2) Sent several information to the process control tool (state of the infusion, target Tg, resin type, etc)
3) Alarm and decision of stop heatingSent data for the processing data storage (Sensor data, time stamp, part name, part number, name of operator, resin type)
Neural Networks Tools
The neural network model is of the following form: y(t) = f[t, y(t-1), T] where y is resistance (log transformed), T is the average middle temperature during the first 2.25 hours, and t is the discrete curing time. A bootstrap aggregated neural network containing 5 neural networks were developed from bootstrap re-sampling of the original training and testing data set. Each neural network is a single hidden layer feedforward neural network with 12 hidden neurons.
T2.6: Integration of the control tools for mid-volume production
Following T2.4 the main characteristics of the mid-volume production such as resin, temperature effects and targets have been incorporated. Lab-scale implementation and tests but also trials at real conditions at Karnic have shown the good performance of the algorithms to predict demoulding time. The self-learning tools have really helped for the fast modelling and implementation to batch-to-batch variations that already faced successfully during the trials. After a long series of trials performed by Karnic it was evident that, in contrast to the Sotira case, temperature variations would play a major role to the control design. Another challenge is the lack of resources for Karnic (and consequently of this industry) to realize an expensive and sophisticated control tool.
T2.7: Integration of the control tools for high-volume production
Based on the platform of T2.4 the main characteristics of the high-volume production case such as resin and targets as well as automatic opening of the press and series production have been incorporated. The material models include the classic models, the neural network models and the parametric-based models for providing directly a real-time estimation of the Tg. In order to maintain the flexibility of the system, 5 different modules run simultaneously: Java tool and database tool from BIBA, Optiview+ (Synthesites), Control tool (NTUA/UNEW/Synthesites) and Production control tool (CEMC/NTUA).
Real-time control of the feeding line
Currently, after a RTM injection, the lines who conducted the reactive resin to the mold through an injection head, are systematically cleaned with acetone and air. Indeed, without these two steps, a gelation will appear in the lines (after few minutes) with two consequences: Stop of the production and degradation of the materials. However, the use of solvents like acetone presents two majors disadvantages: COV emissions and important cost.
Deviations from plan
After the midterm review meeting there was major delay in WP2. In order to save time and to be more efficient, integration and lab-scale trials have been merged: tasks 4.4 4.5 4.6 and 4.7 have been merged into tasks 2.4 to 2.7 respectively.
WP3: Self-learning process modeling and monitoring
The objective of this WP was the development of a self-learning process modelling tool. An in-situ self-learning process modelling and monitoring tool will overcome the necessity of rigorous process models avoiding long, expensive and, sometimes, ineffective lab analyses.
T3.1: Process modeling and monitoring requirements and specifications.
The objective of Task 3.1 was to determine the most significant issues that are important for the successful process modeling and monitoring
T3.2: Adaptation of self-learning techniques to process modeling
The objective of Task 3.2 was to adapt the tools developed in T2.2 to the process (viscosity and cure) modeling of the composite materials.
T3.3: Development of the self-calibrating measuring concept.
The objective of Task 3.3 was to study the influence of electrical properties measurements at industrial level (with reinforcements and possibly conductive carbon fibres and various impurities mixed with the resin) because the conditions could be different providing uncontrolled variability.
T3.4: Development of soft sensors
The objective of Task 3.4 was the development of soft sensors. Soft sensing, is necessary as it is practically impossible to have sensors everywhere in the mould cavity where we need process information. So it is important to define the needs for sensing based on the size of the part and the restrictions for hard sensors placement
T3.5: Integration of the self-calibrating measuring to the self-learning process modeling
The aim of task 3.5 was the integration of the self-learning process modeling with the self-calibrating measuring system.
T3.6: Laboratory trials, comparison and validation
The aim of this task was to check the overall performance of the process modeling and monitoring tools developed in the previous tasks for the three pilot cases.
T3.7: Development of secondary solutions for special applications
The purpose of this task 3.7 added to the original Dow, was to implement the necessary additional developments for improving the measuring accuracy of Optimold for the vacuum infusion and room temperature curing (Acciona) and viscosity and mixing ratio sensing (Acciona VI and in-line in Sotira and Karnic).
The WP3 is completed.
WP4: System integration, wireless technology, lab-scale implementation and evaluation
The main objective was to develop the supporting technology and tools for wireless communication, storage and data processing for the control system.
T4.1: Fast data acquisition, analysis and storage tools
Data Acquisition:
Data collection:
For the collection of sensor data two different approaches may be applied for IREMO. The simplest approach (pushing method) is the collection of a continuous output from the sensor node. The advanced und complex method is that of a polled sensor.
The duration of data acquisition depends as well on the collection method as on other many factors like wireless technology, network topology and routing.
Data processing:
The second central and essential task in data acquisition is how and where to process the data. For data processing many technique (signal conditioning, data conversion, data cleaning, data aggregation and data compression) can be used to ensure clean and quality data and produce reliable and accurate measurements.
Data Storage:
Storage approaches can be classified principally into two categories: the first approach is to push the data to a centralized storage server (external storage) and the second, that called In-network storage, is to store the data on the same sensor node that produced the data or on another sensor in the network.
Experiments and prototype:
To facilitate the selection of relevant methods for data acquisition and storage for IREMO scenarios a lot of experiments have been implemented to build a solution prototype that ensures fast acquisition and storage of measurements.
After the set up of the database base, a suite of programs are written:
- The first is a microcontroller program for the sensor node written in Wiring/Arduino. It waits for a command from the acquisition server to read the sensor data and to send the result.
- The second program is for the acquisition sever and written in Processing. It requests continuous data via its serial port, converts the data to temperature (°C) and transfers the result to the database. This program requests data every 4 seconds avoiding to overwhelm the notebook (Acquisition server).
The experiments for acquisition and storage of data were very successful.
T4.2: Wireless data transmission from mould to control station
The final selection of the wireless data transmission solution could be only realized after more experiments which were done in Fibre Institute at the university Bremen (November 2010) and in CEMC (December 2010). Fibre Institute has a similarly environment as in IREMO. SYNT has provided a monitoring system in September 2010 (Meeting in Bremen) at our disposal. Furthermore SYNT is expected to provide an update of its software to cope with the specific requirements of the wireless communications issues. After the experiments and based on the common requirements for wireless technologies three communication protocols (Bluetooth, Wi-Fi and ZigBee) were selected. For the evaluation and selection of the wireless transmission solution it was important to identify requirements and restrictions towards using wireless technology which are relevant for data transfer in IREMO project.
The main requirements and restrictions for IREMO are categorized as follows:
Mobility:
The wireless network should offer connectivity to both static (e.g. fixed equipment) and mobile devices in the plant. Mobility aspect has to be guaranteed because cabling connects equipment to fixed locations but reduces the flexibility in equipment placement and reorganization. It can also be very expensive to install and maintain in terms of material and lab costs.
Power Consumption:
It is essential that wireless devices such as sensor nodes operate on batteries to guaranty the mobility aspect and energy must be saved. If battery operated, power consumption should be kept low; rechargeable battery is preferred to reduce costs. Suitable power consumption conservation techniques have to be used.
Reliability:
The connectivity should be provided to ensure communication ability between all network devices despite the many obstacles that the industrial space presents. High reliability is a key requirement in IREMO. A minimum aggregate rate of information must be delivered from the transmitter to the receiver node. Some critical scenarios require high or even total end-to-end reliability. Link reliability mechanisms such as MAC layer automatic repeat request - ARQ can significantly reduce the end-to-end packet loss ratio.
Availability:
The availability of the wireless network is the main requirement and the most decisive factor on the connectivity and data acquisition because real-time measurements are necessary as the only input for the control station and the neural network for process control. Due to the problem with electromagnetic emissions created by large motors (e.g CEMC), heavy equipment, and typical industrial machinery (moulds) a degradation of reliability by real-time measurements and a frequently data loss can be provoked. These emissions can create extremely high levels of noise that interfered with wireless equipment.
Selection of wireless Technology:
In order to make a decision, which technology suits best to the IREMO application, it is necessary to compare them based on essential requirements listed above and essentially from a typical industrial environment perspective such as in SORA.
T4.3: Organisation of control system's nodes
Based on the identified requirements BIBA is working on the implementation of the overall concept using open source tools for data acquisition and Storage. The solution will provide ACCI, KARN and SORA the possibility to use two different wireless technologies (WiFi and ZigBee) at the same time.
For the placement and organisation of sensor nodes BIBA has considered each case (SORA, KARN, and ACCI) apart. For ease case BIBA will work out a solution with different requirements:
SORA's case:
Based on the map of SORA plant and our observation during the visit of SORA (March 2010) we found out some criteria which are very important and decisive for the topology of the sensor network and the organisation of the network components to guarantee the most of requirements listed above.
1- The number of moulds
2- The distance between moulds
3- The obstacles in the pant
The number of router in the network depends on
1- the distance between the gateway and the mould (press).
2-the transmission range that is linked to the signal strength of the antenna of the wireless device
Wi-Fi
Concerning Wi-Fi, the technology is also tested in BIBA and it should be adequate for the IREMO application. The only disadvantages are power consumption and the need of more equipment (router with bridging function) for the connectivity of network components. Routers and/or access points cannot be operated on batteries because of the very high-power consumption of Wi-Fi devices. To minimise cabling (between OptiMold and Wi-Fi) and to enhance mobility degree BIBA proposed for all cases two solutions: the Wi-Fi device (Netgear WNCE2001 adapter) or the combination of open source platforms (Arduino microcontroller, Wi-Fi and Ethernet module).
ACCI's case:
The case of ACCINA is similar as in SOTIRA. The pultrusion machines are located in an environment influenced by many metallic objects which are the major problem with radio frequency based networks. They can lead to unreliable network connectivity. Compared to SOTIRA plant, ACCIONA plant has no walls between monitoring system and gateway which facilitate the communication and avoid the installation of relays (routers) for data forwarding.
KARNIC's case:
In the KARNIC case (low-volume production) with only one mould in the plant and due to the distance between the production area and the server room, it is not obligatory to use relay stations between OptiMold and the gateway which is connected to the server (in server room 2).
Deviation and corrective actions
Following the revised DOW after the midterm review, some tasks of WP4 have been merged in WP2.
The WP4 is completed.
Lot of modifications (Adjustments) regarding the architecture, the communication modules (acquisition and Trigger) and the processing tools have been provided after the finishing of WP4. This work has been done in WP5.2.
WP5: Systems' integration, trials and evaluation
The main objectives of this WP were
i) the development different prototypes, i.e 3 control production systems,
ii) trials and
iii) evaluations of the systems developed in the others WPs.
T5.1: Man-machine interface, standardization and inter-operability
In perfect agreement with original plans, involved partners have worked on the user man/machine user interface. Accordingly to the description of work, the high volume RTM case has been first treated. The others cases was also finalized.
Knowledge of the user's need:
The first approach has consisted to associate the user to the definition and specifications of the final tool, in other words: how do they want the system works and for who ?
Regarding now the infusion case, it has been described how the system of the control could look like. Some of the values will always be shown, such as:
- vacuum level (at the pump or at the resin trap); and
- elapsed time.
There are three different issues that should be monitored:
- the resin pot or the mixer (during the infusion process/stage);
- the resin arrival (during the infusion process/stage); and
- the laminate (while it is curing).
The main features that have to be monitored are:
- mixing ratio of the resin in the pot. It has to be between the acceptable upper and lower limits;
- temperature of the resin in the pot. The value has to be between the acceptable upper and lower limits;
- viscosity of the resin in the pot. This value has to be between the acceptable upper and lower limits. It is linked to the temperature of the resin.
- Counter showing the stage of gelation. The time shown in this field shows the time left during which it is possible to continue the infusion with the resin that is left in the resin pot. When this time is lower than 20 minutes the colour of the numbers will change to red to show that a new resin mix must be prepared.
- Quantity of resin within the laminate. A digital caudal meter or a digital counter will be placed in the pipe that feeds the laminate. When this measurement tends to zero, the infusion can be stopped.
- mixing ratio of the resin being injected. It has to be between the acceptable upper and lower thresholds;
- Temperature of the resin being injected. The value has to be between the acceptable upper and lower thresholds;
- viscosity of the resin being injected. This value has to be between the acceptable upper and lower thresholds. It is linked to the temperature of the resin.
- Quantity of resin within the laminate. A digital caudal meter or a digital counter will be placed in the pipe that feeds the laminate. When this measurement tends to zero, the infusion can be stopped.
- degree of conversion. The desired degree to be reached will be highlighted in the graphs as a threshold.
- Tg. The desired Tg to be reached will be highlited in the graphs as a threshold. When all sensors (real and soft) achieve the desired Tg, the oven can be stopped. An alarm should let the user know that the oven can be stopped manually.
- Would it be useful to monitor the viscosity during this phase?
- Temperature at the sensor locations (real and soft). An upper threshold will be highlighted in the graph. If any of the sensors reach this limit, the temperature of the oven will automatically decrease 200C until the temperature at that sensor location gets 100C lower. At that moment, the oven will recover its original temperature (200C higher).
- Temperature inside the oven. Several classical temperature sensors (thermocouples or RTDs) will be placed at different locations along the beam (especially in resin rich areas). The same upper threshold than in the previous point will be highlighted in the graph. If any of the sensors reach this limit, the temperature of the oven will automatically decrease 200C until the temperature at that sensor location gets 100C lower. At that moment, the oven will recover its original temperature (200C higher).
Interfaces-software part:
The results in term of interface is presented hereafter
The idea was to limit, as much as possible, the number of entries. These last ones have to be short and do not need any particular formation. The messages have to be clear with no superfluous information.
A particular attention has been paid to:
i) Choice of color
ii) The structure of the document
iii) The localization of the information (alert, signal etc) and their identification
Problems encountered and corrective actions:
No significant problems were found in the realization of this task.
T5.2: Software and hardware Systems integration and adaption (full scale)
Several moduli were created and compose the final system: a communication box, a trigger box, a java tool for acquisition, processing and data storage, a control tool (including off and on line models), and a production tool. First all these moduli was successfully tested and individually validated. Then the final prototype was constructed i.e the development of a unique tool integrating all the elements.
Regarding the modifications done after the finishing of WP4, as mentioned before, the final solution (Software and Hardware) has been improved.
- Software:
For the data processing tool (Java-based for acquisition, processing and storage) lot of modifications (based on the results of the trials) have been realized in order to optimize the final solution and to facilitate the identification of problems (connection of sensors, checking the actually state of the process, etc.), especially during the integration and testing phases. After running up the tool and by entering the command (function) HELP (or LS), relevant functions (GO, STOP, STATE, SENSORS) will appear on the display and can be used also in background to:
- start the process: Informing workers to close the mold and starting data acquisition, processing and storage.
- stop the process: Opening the mold (automatically) and stopping data acquisition, processing and storage.
- show the current state of the process and to
- list the connected sensors to the tool.
T5.3: Installation at low-volume case and full-scale trial.
ACCIONA Infraestructuras has already manufactured two composite beams by vacuum infusion process using the control tools developed within the IREMO project. The materials, curing cycle, control and software have been checked. Tg values obtained in laboratory are consistent with those obtained by the models, and consequently as proposed by the control system.
Problems encountered and corrective actions
Communication problems were faced during the first demo but they have been solved for the second infusion. Furthermore, improvements have been adapted in the control tools (both based on Neural Network and on Evolution Strategy) from the first to the second infusion in order to get more accurate and fast predictions on the physical-chemical parameters, mainly Tg.
T5.4: Installation at Mid-volume case and full-scale trial
The production is now using the system and operators supervise it. One of the most significant results is that the control system is able to highlight the fact that the first part produced in a more than one batch requires longer curing time (due to the lower temperature of the mold).
T5.5: Installation at High-volume case and full-scale trial
Several solutions have been studied for the packaging of the control system. Several points are requested:
i) Robust (use in an industrial environment, high level of IP (6X)
ii) Touch screen
iii) Easily Transportable
iv) Adaptable and compatible with existing system and equipments
Installation of the high volume production prototype was done. In this case, existing tools (injection machine/press) needed some modification in order to integrate the IREMO system.
Modifications of existing tools (injection machine/press) for the integration of the control system.
Characterisations of the thermal properties are in progress by measuring the effective Tg of the parts. These last one appears conform with the target, i.e around 115 °C, but also parts with artificial problems such as mixing ratio deviations.
T5.7: Control system evaluation and standardization
The objective of this task is to evaluate the performance of the whole control system (hardware and software)
Although this task is not fully achieved, the first trials with control are very promising. As described in WP2, the control system is tested in production conditions in different plants.
Particular attention is paid to the following points: the collect of the data such as: resin arrival, degree of cure, start of acquisitions and control decisions.
In the 3 cases, the systems are working satisfactorily for all different processing conditions: temperature, types of resins, carbon/glass fibers etc.
Acciona's case:
Two beams were manufactured by ACCIONA Infraestructuras using the control tools and the first has already been evaluated with the performance of DSC tests at ACCIONA Infraestructuras' laboratory with good agreement (+/- 5°C)
Karnic's case:
Data were collected during trials, using single and pendulum.
During the trials the following observations were made:
- During production using the pendulum arrangement it was difficult for the operator to observe visually on the computer screen the control indicator. For this reason, the control software was modified to give an audio signal as well.
- Data from the pendulum setup were collected using an additional Optimould box and a second computer to avoid confusion and conflict of the information.
- The system setup for the first mould (Mould-A) performed throughout with reasonable accuracy. However, the system for the second mould (Mould-B), even though is identical, was not fully reliable and control signal was found a few times to be taking longer time than expected. The problem was identified to be on the hardware and relevant changes were made.
- As expected, the first part produced in a more than one batch, required longer time to cure due to the lower temperature of the mould. The control appeared to identify successfully this instant.
The results collected during the control trials period have assisted in refining the system to provide reasonably accurate curing time predictions. The inconsistency of some readings however, suggests that further testing and refinements are necessary (in progress during July and demonstration activities)
Sotira's case:
Sotira Automotive manufactures for Audi RS3 the two front fenders in epoxy /carbon composite.
We have chosen those parts, with the agreement of AUDI, for IREMO project.
WP6: Industrial scale demonstration
For WP6 the main objective is to demonstrate the new production system for the three main applications
T6.1: Demonstration of low-volume case [ACCI]
ACCIONA Infraestructuras has already manufactured two composite beams by vacuum infusion process using the control tools developed within the IREMO project as per the requirement of the DoW.
The set up in the demo was to monitor in real time:
- Curing of resins and temperature at six different locations using flexible curing and temperature sensors;
- Resin in the resin bath using a pot sensor;
- temperature in the oven during curing of the part.
Thanks to these sensors, that the following parameters could be predicted using the IREMO system during the trial:
- viscosity of the resin;
- degree of cure of the resin;
- Tg of the resin.
Preliminary trial and demo #1.
One preliminary trial with the same monitoring set up was performed prior to the demo#1 trial itself in order to:
- avoid any errors in the manufacturing of the product that has been chosen for the demo;
- check the wireless communication;
- check the control tools;
- confirm that the IREMO system has been well understood within the ACCIONA team.
T6.2: Demonstration of mid-volume case [KARN]
The production of certain type parts produced by Karnic using the RTM method is a repetitive process where operations are repeated in every part production cycle with target to achieve maximum possible output during a specified batch period. For one complete cycle, that is the completion of one single part, the required production time can be divided in two categories depending on the occupancy of the operator.
These are:
(a) Time-A: the time at which the operator is fully occupied to perform value adding work in the process. This consists of all tasks from de-moulding the part already in the mould to preparation, mould closing and resin injection of the next part.
(b) Time-B: the time required for the part to cure during which the operator is unable to do value adding work and can only be treated as idle time.
The actual production time of the part is therefore the sum of (Time A + Time B).
As all value adding operations can be controlled by careful process planning and the adoption of specialised tools, Time-A can be of a fixed known value for every part. On the contrary, Time-B is dependent upon the chosen resin type, the environment temperature, the use of post curing and the percentage of catalyst used therefore, it may have significant variation from part to part and the overall efficiency of the process.
T6.3: Demonstration of high-volume case [SORA]
For the demonstration case, SOTIRA uses the IREMO system for moulding AUDI RS3 Fender. SOTIRAs target is to use the system with a complete automation to optimize the cycle time. SOTIRA wants to make a minimum of 18 parts per shift with the monitoring system and order directly the cycle of the couple 'injection machine - press' according to the degree of cure.
Each operation as monitoring or automation of the press with the IREMO system has been tested successfully, especially with prototype system. That was the work of the WP5 stage.
WP7
T7.1: Market analysis and perspectives
The following documents have been prepared to understand how the new technology could be used in potential applications, which could be the potential customers and their needs, and what are the competitor's strategies and the competing systems or solutions already available on the market:
- The functional schemes of the IREMO system (one scheme by pilot case) to clarify the objectives of the project and to define the 'IREMO system'.
- The benchmarking of the existing competitive technologies (cf. Task 1.1) whose the objective is to identify the main competitive systems already on the market, or in development, and to collect data concerning their principle, their characteristics and the suppliers which develop these systems.
- The SWOT analysis to understand the strengths, weaknesses of the consortium/project, and to identify opportunities and threats for the end results.
- The market analysis with data on the global composite market and focus by applications sectors (automotive, wind turbine, transportation, building and construction, aerospace, boat building) to identify opportunities and limitations to market the results of the project and to define a marketing strategy.
T7.2: Dissemination of results
Communication tools have been realized in order to disseminate the results of the project:
- Creation of the project logo.
- Creation of the public website (see http://www.IREMO.eu online) with inscription to Google Analytics to follow the traffic.
- Creation of the project flyer for public diffusion.
- Creation of a Powerpoint presentation of the project that any partner can use to present the project to non-project members.
- Creation of posters and banners to promote the project during exhibitions.
- Creation of a list of relevant seminars, exhibitions, conferences or workshops where there is potential opportunity to present the results of the project.
- Creation of a list of relevant magazines where the results of the project could be presented.
Many actions of dissemination have been undertaken since the beginning of the project.
- Project members have made technical and scientific presentations in twenty conferences and symposiums all over the world.
- Project members have published almost fifteen articles and papers in specialized press like scientific journals, technical magazine and conference proceedings.
- Four press releases have been diffused in order to announce and promote dissemination actions
- Around ten articles and news have been published by the specialized press about IREMO project, consecutively to the publication of IREMO press release and news by the consortium.
Moreover, in order to disseminate as better as possible the results of the project, the consortium has decided to attend two composites trade shows as exhibitor:
- Composites Europe 2011 - Stuttgart, Germany - 27/29 September 2011
- JEC Europe 2012 - Paris, France - 27/29 March 2012
The objectives were to have direct contact with industrial and scientific communities and also to communicate about the two workshops organized by IREMO in parallel of these events.
T7.3: Exploitation of results
Atoutveille had prepared and organized the Exploitation Strategy Seminar (ESS) which held the 7th of June 2011 in close relation with Alberto Belle, the expert consultant who animated the seminar.
Following the seminar, the following tasks have been achieved:
- ESS seminar synthesis
- Animation of the thinking about the definition of exploitable results
- Collect of information for the PUDF document following the definition of new exploitable results
- Improvement of the marketing strategy: modification and updating of the IREMO Market Strategy Table
T7.4: Workshop organization
According to the Document of Work (DoW), three public workshop sessions were planned in order to disseminate the result of the project in effective manner.
The first workshop has been held in September 2011 at Stuttgart in Germany, in parallel of Composites Europe 2011 exhibition (27-29 September 2011).
Potential Impact:
The global composites industry is now producing an annual 8m tonnes of product, worth an estimated 60bn EUROS in 2010, and with an average value of 7.2 EUROS per kg. However, the EUROS per kg value of that product currently differs widely from region to region. With a 36% market share by value and 35% by volume, the value of North American composites production is estimated at 22bn EUROS, equating to an average unit price of 8.2 EUROS per kg. Overall production volume in 2010 was 2.7m tonnes. In the EMEA (Europe, Russia, the Middle East and Africa), 2m tonnes were produced, 33% of the market by value and 22% by volume, with a value of 20bn EUROS and an average unit price of 10.0 EUROS per kg. In Asia-Pacific and the rest of the world (including Australia and South America), the average unit price of composites is much lower, at 5.5 EUROS per kg. The 3.3m tonnes produced in these regions in 2010 had a value of 18bn EUROS, representing 43% of the market by volume, but just 31% by value. This is explained by the increasing use of primarily carbon composites in higher-end applications in North America and EMEA - most notably so far in aerospace and wind energy - with consequently higher prices. Growth similarly varies greatly by sector, with the smaller markets - by volume, though certainly not added value - continuing to grow rapidly.
Market Perspectives
The overall market for advanced composites - which comprises composites based on carbon fibres, carbon nanotubes and graphene - will more than triple to USD 25.8 billion by 2020, according to the Lux Research report Carbon Fibre and Beyond: The USD 26 Billion World of Advanced Composites. Lux forecasts that the use of advanced composites by wind turbines manufacturers will increase from USD 2.5 billion in 2011 to USD 15.4 billion in 2020. Meanwhile, growth in aerospace will lag, despite the introduction of new aircraft that use large quantities of carbon fibre reinforced plastic (CFRP), such as Boeing's 787 Dreamliner. In 2020, wind energy will account for nearly 60% of the market for composites, compared with the current 35%.
Industrial impact
The main impact of the IREMO project is the significant increase of production rates, maintaining quality and improving the robustness of liquid composite moulding, one of the most advanced and upcoming manufacturing processes. This impact will be mainly achieved by the integration of machines, tooling, process monitoring and simulations under the same platform of process optimisation and control.
In particular IREMO is expected to contribute significantly to the:
-Reduction of the number of rejected components or products by 50%
-Reduction of power consumption by at least 15%
-Increased throughput and capability of processes and productivity maintaining repeatability and accuracy by a factor of 30% with respect to specific Tg and reduced porosity.
-Minimisation (or even elimination) of the use of services, e.g. acetone by a factor of more than 60% in some cases
As well as to:
-the reduction of down times during product exchange and conflict situations;
-the improvement of product quality while reducing the need for inspection;
-the increase in machine availability and reduction in maintenance; and
-improved efficiency of complex production systems
So besides the positive contributions of the project to the specific targeted impacts of this call further contribution to wider-range impacts are:
-the decrease in the usage of environmental unfriendly solvents
-easier advancement of the composites manufacturers towards closed moulding (zero emissions).
Societal impact
In a wider perspective IREMO responds to the societal needs via the improvement of working conditions by introduction of more automated processes minimising contact and exposure with potential irritating agents such as resin, and via the implementation of technological progresses in materials and manufacturing techniques increasing the skilfulness of the European workforce and securing the employment level in Europe.
Health and safety impact by the industrial production of components in Constructions
IREMO will contribute to the improvement of the occupational health and safety conditions of the workers at the construction of infrastructure structural elements by the improvements associated at the manufacturing process (automated control), transport (less weight to transport) and assembly (less weight to handle, less time, better guidance system) compared to conventional concrete elements. In particular, this aspect is of relevant importance, as construction sector is one of the most dangerous based on the fact that many more accidents occur in construction per every 100.000 workers than overall in the workforce : in 2001, there were 7.200 non-fatal accidents at work per 100.000 construction workers, as compared to 3.800 accidents per 100 000 workers for the total of the nine economy sectors for which comparable statistics exist. For fatal accidents at work the difference was even greater.
Health and safety impact at industrial production level
IREMO will improve significantly the occupational health and safety conditions of the employees at the production level. As confirmed by SORA and KARN their target to introduce closed liquid composite moulding is driven also by their motivation to eliminate VOC (Volatile of Carbon) emissions in the production site completely. This is extremely important as on one hand the styrene emissions from the polyester and vinylester use and on the other hand the unhealthy emissions of the epoxy resins when cured create extremely unhealthy working conditions.
Main dissemination activities
It has been defined that project results will be presented in workshops, conferences and symposia by the academic participants (NTUA, UNEW) and the research institutions (CEMC, TECN and BIBA). The industrial partners (SORA, ACCI and KARN) will disseminate giving publicity to the project and its results in technical magazines and well-known organizations, also industrial partners will be in charge of demonstration activities.
IREMO website
-Atoutveille has created and published the official website of the project (with the technical support of a web design company which hosts the site): http://www.IREMO.eu.
-The website can be entirely managed by Atoutveille, it will be maintained during the whole duration of the project and the 2 years following the end of the project.
-This public website is associated with an internal website platform used by the members of the consortium as a dynamic tool for data exchange and report depositary. It is also a means for European officers in charge of IREMO to be informed about the advancement of the project. Furthermore, a mailist containing all consortium member e-mail address has been established to facilitate exchange between IREMO partners
Press Releases
Four press releases have been sent to specialized press in the field of composite materials, like technical magazine, online technical website. They have generated news published in several support.
- 1st Press Release has been disseminated the 2nd of April 2010 in order to announce the beginning of the project and to detail the consortium, the context and the technical objectives of the project.
- The 2nd Press Release has been disseminated the 06th of January 2011 in order to announce the participation of IREMO consortium as speaker to the Symposium PPE RTM & Infusion in St Avold, France (9th and 10th February 2011).
- The 3rd Press Release has been disseminated the 15th of February 2012 in order to announce the organisation of the 2nd Technical Workshop of the IREMO project during a session at SEICO 12 International Conference in Paris, France, the 27th of March 2012. (Session 4B: IREMO from 08:30 to 09:50).
- The 4th Press Release has been disseminated the 11th of May 2012 with the objective to announce the organization of the 3rd and Final Workshop and Demonstration Sessions of the IREMO project at SOTIRA plant in Saint-Méloir-des-Ondes, France, the 18th of July 2012.
List of Websites:
http://www.iremo.eu
The IREMO project (intelligent Reactive polymer composites Moulding) involves 10 industrial, academic and Research and Developement partners from France, Spain, Greece, Germany, Cyprus and the UK. The target is to develop process monitoring and intelligent control technologies for the intelligent processing of liquid composite moulding and pultrusion aiming at the acceleration of the curing stage, reduction of development time and trial-and-error solutions and ensure product quality. The overall objective of the project is to develop all the supporting science and technology for the optimal production of polymer composites in closed moulding.
The scientific and technological achievements during the three years of the project are:
1. The development of optimal and self-learning process control tools that ensure the appropriate impregnation of fibres and the optimal curing of the composite parts resulting in more than 25% cycle time reduction at all industrial conditions considering production constraints and targets.
2. A process monitoring system capable of monitoring the full range of all major types of reactive polymers processing such as epoxies and polyesters with adequate accuracy and repeatability while providing in-situ control quality capabilities by relating the sensing signal to the materials' properties using reliable modelling tools.
3. A process modelling tool able to provide detailed process information and predictions for a wide range of polymers through an in-situ self-learning procedure.
4. Wireless communications and networking to minimise the wiring in the shop floor and development of a user-friendly interface tool.
5. Retrofitting of the control tool to an existing 2-components injection machine to become fully sensorised to measure material-state and control temperature automatically.
6. Integration of the monitoring and control system in a vacuum infusion process for large and thick composite parts and in a light RTM process for fast production of low-cost composite parts.
Project Context and Objectives:
Reactive polymer (thermosets and few thermoplastics) liquid composite moulding is used more and more into aircrafts, cars, boats, sports, construction, health, energy and almost in all industrial sectors because of numerous advantages. However, one large disadvantage is their complicated, expensive, risky and slow production hindering its wider industrial acceptance. The most common 'problems' in the production of polymer composite parts are the fibre impregnation with the polymer and the curing of this polymer until its consolidation. While fibre impregnation is very important and curing dominates the production cycle there are several conventional processes such as the hand layup, pultrusion and filament winding that the problems are easily detectable and solvable by the operators. Whereas in modern processes, such as the reactive polymer liquid composite moulding (Resin Transfer Moulding and its variants, Vacuum infusion but also Sheet Moulding Compounding (SMC) and Structural Reaction Injection Moulding (SRIM), the problems are significant, the existing knowledge is limited while there is no possibility to realise the problems and act on them as they occur. In a typical reactive polymer composite moulding production cycle after the fibre placement and the mould clamping and the cavity completely filled with resin, the polymerisation phase follows with or without external heating capturing from 65 up to 80% of the cycle time which also may vary depending on the application from a few minutes (automotive) to days (large boats). This a problem of all reactive polymer manufacturing but in the case of liquid composite moulding the problem is worse as the material and process deviations are significant due to the in-situ fibre impregnation (pre-pregs have a standard and much higher quality) and the cheaper production facilities.
In order to solve these problems turning reactive polymer liquid composite moulding to an agile industrial process it is important to have:
- automatic process control tools ensuring the optimal moulding of the materials
-self-learning process control tools that can handle efficiently production and material deviations without requiring extensive lab trials and scientific models
-production databases and self-learning material models that can provide the essential information to cope automatically with different matrices, batch-to-batch variations and small scale productions
-Robust and reliable sensors that can sense the state of the moulded materials inside the mould cavity, providing reliable and real-time information on the status of the process.
IREMO concept
The concept of our approach is to address all of the above issues in a holistic and systematic way in order to make reactive closed composite moulding an automatic, flexible and efficient industrial process. The proposed strategy to reach this target is based on the experience gained with the advanced control concepts developed for batch-to-batch processes in the chemical processing industry and a Receding Horizon Model Based Control developed and applied for closed composite moulding. The new elements will be the introduction of self-learning control schemes in conjunction with optimisation tools while advancing and integrating all the required supporting technologies i.e. self-calibrating process monitoring, self-learning process modelling and data analysis and communication tools.
IREMO Scientific objectives
The general scientific objective of the project is to develop all the supporting science behind the successful production of reactive polymer composites in closed moulding. More specifically the main scientific objectives of the project are:
- Development of an optimal self-learning process control that would ensure the appropriate impregnation of fibres and the optimal curing of the composite parts.
-Real-time optimisation of the polymerisation of fibre-reinforced polymer matrix in real production conditions considering all production constrains and targets.
-Relating the evolution of the sensing signal to the materials' properties using fast self-learning Neural Network-based tools
- Development of self-learning material models capable in providing the most valuable information and prediction based on in-situ process data rather than on extensive lab analyses.
IREMO Technological objectives
IREMO aims to develop an innovative moulding system for the efficient and intelligent mass production of reactive polymer composite parts.
The project has developed all the supporting technologies for the efficient and intelligent processing towards fast and scrap-free production with the following main technological objectives:
- A fully automatic, self-learning production process ensuring an optimal production cycle (20 to 25% cycle reduction) and low scrap ratio (below 1%) even during start-up.
- A process monitoring system capable of monitoring the full range of all major types of reactive polymers processing such as epoxies and polyesters with adequate accuracy and repeatability while providing in-situ control quality capabilities.
- A process modelling tool able to provide detailed process information and predictions for a wide range of polymers through an in-situ self-learning procedure.
- Wireless communications and networking to minimise the wiring burden in the shop floor.
- A retrofitting to an existing 2-components injection machine to become fully sensorised to measure material-state and control temperature and mixing ratio of at least 3% accuracy automatically.
- Integration of the monitoring and control system in a vacuum infusion process for large and thick composite parts.
IREMO baseline and performance
As IREMO is a performance-oriented project it is also very important to define the performance baseline with which the new systems would be compared to. However, this baseline is subjective to the technology of each industrial partner and to the specific application so this baseline will be clearly defined in the first months of the project based on the difference between the expected and the achieved values regarding set-up time, cycle time, start-up time, resources, scrap ratio and costs for each of the selected platform-cases.
Project Results:
WP1: System's specifications, requirements and targets, initial installations and data collection
The main objectives of Work Package 1 were:
i) to provide a detailed definition of processes,
ii) to install the process monitoring systems and
iii) to start the data collection.
Task 1.1 definition of processes, targets, requirements and specifications
Pultrusion case : ACCIONA
The product is a superstructure used in Building and Construction. Among the various advantages associated with it are light weight, ease of installation and high strength.
Production Set-up
The mould consists of three heating zones with one temperature sensor in each zone. The cavity size inside the mold is uniform except at entry and exit (a slight radius is provided on each side for avoiding damage to the raw materials).
Production machine
Caterpillar type Pultrusion machine of 5T Pulling force has been used. Temperature is controlled automatically.Pulling speed is controlled manually.
Infusion (ACCIONA)
The product is a superstructure used in Construction. Among the various advantages associated with are light weight, ease of installation and high strength.
Currently existing control capabilities: mould opening time and change catalyst ratio for faster production
High Speed RTM (SOTIRA)
The products are an outer door and fenders for automotive vehicles.
Production Set-up: The mould is equipped with both injection and vacuum heads, an injection channel and a seal.
A variety of requirements have to be met by order to be adopted by industrial practice:
- Functional requirements: The basic functionality of wireless sensor networks in IREMO is the permanent acquisition and transmission of process data during the production of reactive polymer composite parts. The sensor data will ensure an optimal monitoring of the moulding process in terms of efficiency as well as product quality. Concerning the storing of these data, a storage strategy must be developed to hold, share and protect valuable data about the moulding process.
- Requirements regarding embedding: The solution must be a retrofit of existing plants. This means that all required sensors can be integrated in the existing moulds. In addition, an ICT-infrastructure has to be installed to assure the acquisition, transmission and storing of sensor data.
- Performance requirements: Requirements concerning the performance of the sensor solution primarily arise from the complexity and size of the products or moulds and the placement of sensors, the resulting number of sensors, the sensor network topology, the data collection strategy and particularly the sensor hardware performances regarding transmission rate, band width and power consumption. In order to derive and initiate necessary measures in real-time, data collection, transmission, and processing must be extremely efficient and reliable, especially considering the amount of data expected for big parts.
- Handling requirements: Human-machine interfaces must be created that allow a comprehensible display and documentation of gathered information about the current status of the manufacturing process. This requires the design of an user-friendly use model, which visualizes the process sequence and initiates the measures of the control system. The system must furthermore recognize internal errors such as malfunctioning sensors and respond to them if necessary.
Neural network modelling of nonlinear processes UNEW
UNEW has extensive expertise in neural network modelling of nonlinear processes, optimisation control of batch processes, batch to batch iterative learning control of batch processes with applications to batch polymerisation reactors and bio-production processes. The expertise can be directly applied to the modelling, optimisation, and control of polymer composite moulding processes, which are batch processes in nature. UNEW will cooperate with other project partners in control objective specification and control strategy determination.
BENCHMARKING (ATOU)
A benchmarking of the existing competitive technologies has been realized. The objective of this work was to identify the main competitive systems already on the market, or in development, and to collect data concerning their principle, their characteristics and the suppliers which develop these systems.
The competitive systems have been divided in four sections according their characteristics:
- IREMO competitive sensors for real-time cure monitoring of FRP parts.
- Real-time monitoring systems of industrial processes.
- Real-time control systems of industrial processes.
- Self-learning control systems of industrial processes.
T1.2 Definition of pilot case plant
ACCIONA (infusion)
In order to monitor:
i) the arrival of the resin at two locations (i.e. the beginning and the end of the filling)
ii) the cure inside the mould and the lines,
A minimum of 2 mould sensors and 1 in line sensor is necessary
SORA (RTM)
In order to monitor
I) the arrival of the resin at two locations (inlet and outlet)
ii) the cure inside the mould and the lines ( as close as possible to the injection gate)
A minimum of 2 mould sensors and 1 in line sensors is necessary
Karnic (Light RTM)
In order to monitor:
i) the arrival of the resin at two locations (inlet and outlet)
ii) the cure inside the mould.
A minimum of 2 mould sensors is necessary
ACCIONA (Pultrusion)
In order to monitor:
i) In the die
ii) At the die exit
iii) In the resin bath
iv) the cure inside the die and the resin bath
A minimum of 3 mould sensors is necessary
T1.3 Preparation and installation of monitoring units
Monitoring systems have been delivered by SYNT to ACCI, NTUA, CEMC, SORA,TECN, KARN and replacement sensors also.
T1.4 Data collection, requirements and specifications
The number of acquisitions carried out with the monitoring units installed during the first period of the project is about 5 to 10 for pultrusion, 10 for infusion, 75 for light RTM, 150-200 for RTM and more than 150 for lab scale trials.
WP2: Development of self-learning optimal process control tools
The main objective of this WP is to develop the generic self-learning and optimal control tools and control platform. The basic platform for the control tools has been based on Labview (National Instruments) as well as algorithms coded in Matlab and C. The control platform is using simulation tools based on partial differential equations (PDE) and Neural networks for self-learning in conjunction with the fuzzy logic environment to maximise the applicability and the robustness of the developed tools.
?2.1 Control Strategies
Numerous important parameters that affect the production were identified: Resin, mould or die temperature, resin, flow rate, speed process, curing time and related demoulding time, pressure, vacuum, and cleaning/rinsing of the feeding lines. One of the main targets is the development of a friendly user interface i.e. a computer screen indicating to the operator all necessary information: experimental parameters, degree of cure, alarms etc. if some automatic actions can be envisaged in the control process, it is preferable that the final decision remains in the hands of operators (ex: opening of the mould, important change of the temperature .), and this for safety reasons.
The targets that were identified for the control system is the determination and the display of:
- degree of curing
- glass transition temperature
- gelation time
- viscosity
These properties have to be extracted for all production applications (batch-to-batch variations, conditions, fibres etc) and are of primary importance to optimize the following parameters:
- Mould opening (SORA/KARN/ACCI)
- Mould temperature (SORA/KARN/ACCI infusion)
- Partially or fully rinsing of the feeding lines (SORA/KARN)
- Time of demoulding, with or without post-curing (ACCI infusion)
- Pulling speed (ACCI Pultrusion)
- Mixing ratio check (SORA/KARN/ACCI infusion)
The quantifiable targets were set with respect to the characteristics of the initial process:
- SORA/KARN: time of curing reduced by at least 40%
- SORA/KARN: Wasted resin reduced by at least 70%
- ACCI infusion: time of curing reduced by at least 40%
- ACCI Pultrusion: increase of performance by at least 40%
?2.2 Development of Neural Networks and Optimisation tools
Neural Networks Tools
Neural network modeling of reactive polymer composite moulding processes has been carried out using both simulated data and experimental data from the simulation programme supplied by NTUA, 7 batches were simulated and the sampling time is 4 minutes.
The Robust Evolution Strategies
The Evolution Strategies emerged from the Genetic Algorithms and are better suited to optimisation problems. Their fundamental advantage over the conventional optimisation methods is that they do not use any gradient information so it is more difficult to get trapped in local minima. In this task an alternative Evolution Strategy, namely the Robust Evolution Strategy (RES) was developed to deal with the inherent variability of the basic parameters of the models that are used for the model-based optimization and control. This new algorithm provides the optimal solution of a mathematical problem within a realistic variation of its major parameters. In this way the provided solution is not optimal just for a single combination of model parameters for within a realistic range of these parameters. In the case of composites manufacturing we have the model parameters which for example may be some or all of the parameters of the kinetic model of the resin and the control parameters which may be the temperature profile. Seeking for the optimal temperature profile we can include the model parameters variation in order to ensure that even for the expected variations in the model parameters we can still have the optimal profile.
T2.3: Batch-to-batch control tools
Mould opening prediction (controlling mould opening)
The static model can predict the time to open the mould for any given moulding temperature in advance (using the predictions of degree of cure and Tg at different times). The mould open time can be obtained by solving an off-line optimisation problem, for example, minimise curing time so that the degree of cure reach 97% of the maximum degree of cure. Using the neural network model for the SOTIRA data, mould opening time (curing time) is obtained in relation to curing temperature.
Batch to batch iterative learning control
In this study, the batch time (moulding time) is fixed as 28 minutes and the sampling time is 4 minutes. Thus the batch is divided into 28/4=7 stages. The final degree of cure is the controlled variable and the applied temperature is the control action which is a vector with 28/4=7 elements. Note that it is also possible to control the degree of cure trajectory during the moulding process. In this study, two cases with different control actions are considered. In Case I, the applied temperature is kept constant during each stage and U contains the 7 piecewise constant temperatures. In Case II, the applied temperature is linearly rising during each stage at a particular rate and U contains the 7 temperature increasing rates.
Control Optimisation tools
Process robust optimisation is necessary in order to ensure that the production is optimised in a robust way always taking into consideration materials deviations and process constraints and limitations. This optimisation is important that it is done as continuously as possible to ensure that the process parameters are always optimised. At the past this process optimisation could be done only offline due to the requirements of increased computational effort so the batch-to-batch or at most the run-to-run cycle optimisation was possible. However as computational effort is significantly improving an online version will be also tested to check the capability for real-time cycle optimisation. In any case the concept of cycle optimisation is the same: the temperature profile is shot in order to achieve the best solution taking into account process constraints and limitations.
Optimizing the curing process may have various meanings so it is important to integrate into the process models simulation tools or at least criteria that will provide the quantification of the optimality criteria with respect to certain characteristics of the process:
1. Targeted degree of cure
2. Minimum processing time
3. Minimum exothermic peak temperature
4. Minimum curing generated internal stresses
5. Minimum cooling generated internal stresses
6. Homogeneous degree of cure
7. Minimum temperature gradients
T2.4: Development of the control platform and data manipulation tools
The control platform is a real-time tool for the control of the manufacturing process. It includes all the algorithms for the real-time calculation; it is responsible for the communication between the other control modules and also has a user graphical interface for handling also the input/ output dataflow. Although initially the Fuzzy Logic was researched to deal with the several data and user specs finally due to its complex structure it was decided to use a rule-based approach with specific thresholds for each variable depending on each process.
The Production Control Tool
The Production Control Tool has the basic user interface and graphical results and controls all the primary production equipment such as the injection machine, the Press or the heating/cooling devices. In particular the production control tool has the following tasks:
1) To collect the data from the process control tool (resin arrival, degree of cure, control decisions)
2) Sent several information to the process control tool (state of the press, state of the injection, target Tg, resin type etc)
3) Sent to the trigger box the following information (start of acquisition, decision of opening press)
4) Sent data for the processing data storage (Sensor data, time stamp, part name, part number, name of operator, resin type)
T2.5: Integration of the control tools for low-volume production
Following the work in T2.4 the main characteristics of the low-volume production such as new resin models, temperature effects and targets have been incorporated successfully for three sensing points wand can handle flexible but also durable sensors (different parameters are necessary for each sensor type). Tests at lab-scale conditions but also simulated trials in Acciona showed initially some problems which were successfully overcome and the control system is working very satisfactorily. It should be highlighted that this is a generic platform although based on the specific resin it can handle a large variety of temperatures and processing conditions e.g. carbon fibres, flexible/durable sensors etc.
The Production Control Tool
The production control tool has the following roles:
1) To collect the data from the process control tool (resin arrival, degree of cure, control decisions)
2) Sent several information to the process control tool (state of the infusion, target Tg, resin type, etc)
3) Alarm and decision of stop heatingSent data for the processing data storage (Sensor data, time stamp, part name, part number, name of operator, resin type)
Neural Networks Tools
The neural network model is of the following form: y(t) = f[t, y(t-1), T] where y is resistance (log transformed), T is the average middle temperature during the first 2.25 hours, and t is the discrete curing time. A bootstrap aggregated neural network containing 5 neural networks were developed from bootstrap re-sampling of the original training and testing data set. Each neural network is a single hidden layer feedforward neural network with 12 hidden neurons.
T2.6: Integration of the control tools for mid-volume production
Following T2.4 the main characteristics of the mid-volume production such as resin, temperature effects and targets have been incorporated. Lab-scale implementation and tests but also trials at real conditions at Karnic have shown the good performance of the algorithms to predict demoulding time. The self-learning tools have really helped for the fast modelling and implementation to batch-to-batch variations that already faced successfully during the trials. After a long series of trials performed by Karnic it was evident that, in contrast to the Sotira case, temperature variations would play a major role to the control design. Another challenge is the lack of resources for Karnic (and consequently of this industry) to realize an expensive and sophisticated control tool.
T2.7: Integration of the control tools for high-volume production
Based on the platform of T2.4 the main characteristics of the high-volume production case such as resin and targets as well as automatic opening of the press and series production have been incorporated. The material models include the classic models, the neural network models and the parametric-based models for providing directly a real-time estimation of the Tg. In order to maintain the flexibility of the system, 5 different modules run simultaneously: Java tool and database tool from BIBA, Optiview+ (Synthesites), Control tool (NTUA/UNEW/Synthesites) and Production control tool (CEMC/NTUA).
Real-time control of the feeding line
Currently, after a RTM injection, the lines who conducted the reactive resin to the mold through an injection head, are systematically cleaned with acetone and air. Indeed, without these two steps, a gelation will appear in the lines (after few minutes) with two consequences: Stop of the production and degradation of the materials. However, the use of solvents like acetone presents two majors disadvantages: COV emissions and important cost.
Deviations from plan
After the midterm review meeting there was major delay in WP2. In order to save time and to be more efficient, integration and lab-scale trials have been merged: tasks 4.4 4.5 4.6 and 4.7 have been merged into tasks 2.4 to 2.7 respectively.
WP3: Self-learning process modeling and monitoring
The objective of this WP was the development of a self-learning process modelling tool. An in-situ self-learning process modelling and monitoring tool will overcome the necessity of rigorous process models avoiding long, expensive and, sometimes, ineffective lab analyses.
T3.1: Process modeling and monitoring requirements and specifications.
The objective of Task 3.1 was to determine the most significant issues that are important for the successful process modeling and monitoring
T3.2: Adaptation of self-learning techniques to process modeling
The objective of Task 3.2 was to adapt the tools developed in T2.2 to the process (viscosity and cure) modeling of the composite materials.
T3.3: Development of the self-calibrating measuring concept.
The objective of Task 3.3 was to study the influence of electrical properties measurements at industrial level (with reinforcements and possibly conductive carbon fibres and various impurities mixed with the resin) because the conditions could be different providing uncontrolled variability.
T3.4: Development of soft sensors
The objective of Task 3.4 was the development of soft sensors. Soft sensing, is necessary as it is practically impossible to have sensors everywhere in the mould cavity where we need process information. So it is important to define the needs for sensing based on the size of the part and the restrictions for hard sensors placement
T3.5: Integration of the self-calibrating measuring to the self-learning process modeling
The aim of task 3.5 was the integration of the self-learning process modeling with the self-calibrating measuring system.
T3.6: Laboratory trials, comparison and validation
The aim of this task was to check the overall performance of the process modeling and monitoring tools developed in the previous tasks for the three pilot cases.
T3.7: Development of secondary solutions for special applications
The purpose of this task 3.7 added to the original Dow, was to implement the necessary additional developments for improving the measuring accuracy of Optimold for the vacuum infusion and room temperature curing (Acciona) and viscosity and mixing ratio sensing (Acciona VI and in-line in Sotira and Karnic).
The WP3 is completed.
WP4: System integration, wireless technology, lab-scale implementation and evaluation
The main objective was to develop the supporting technology and tools for wireless communication, storage and data processing for the control system.
T4.1: Fast data acquisition, analysis and storage tools
Data Acquisition:
Data collection:
For the collection of sensor data two different approaches may be applied for IREMO. The simplest approach (pushing method) is the collection of a continuous output from the sensor node. The advanced und complex method is that of a polled sensor.
The duration of data acquisition depends as well on the collection method as on other many factors like wireless technology, network topology and routing.
Data processing:
The second central and essential task in data acquisition is how and where to process the data. For data processing many technique (signal conditioning, data conversion, data cleaning, data aggregation and data compression) can be used to ensure clean and quality data and produce reliable and accurate measurements.
Data Storage:
Storage approaches can be classified principally into two categories: the first approach is to push the data to a centralized storage server (external storage) and the second, that called In-network storage, is to store the data on the same sensor node that produced the data or on another sensor in the network.
Experiments and prototype:
To facilitate the selection of relevant methods for data acquisition and storage for IREMO scenarios a lot of experiments have been implemented to build a solution prototype that ensures fast acquisition and storage of measurements.
After the set up of the database base, a suite of programs are written:
- The first is a microcontroller program for the sensor node written in Wiring/Arduino. It waits for a command from the acquisition server to read the sensor data and to send the result.
- The second program is for the acquisition sever and written in Processing. It requests continuous data via its serial port, converts the data to temperature (°C) and transfers the result to the database. This program requests data every 4 seconds avoiding to overwhelm the notebook (Acquisition server).
The experiments for acquisition and storage of data were very successful.
T4.2: Wireless data transmission from mould to control station
The final selection of the wireless data transmission solution could be only realized after more experiments which were done in Fibre Institute at the university Bremen (November 2010) and in CEMC (December 2010). Fibre Institute has a similarly environment as in IREMO. SYNT has provided a monitoring system in September 2010 (Meeting in Bremen) at our disposal. Furthermore SYNT is expected to provide an update of its software to cope with the specific requirements of the wireless communications issues. After the experiments and based on the common requirements for wireless technologies three communication protocols (Bluetooth, Wi-Fi and ZigBee) were selected. For the evaluation and selection of the wireless transmission solution it was important to identify requirements and restrictions towards using wireless technology which are relevant for data transfer in IREMO project.
The main requirements and restrictions for IREMO are categorized as follows:
Mobility:
The wireless network should offer connectivity to both static (e.g. fixed equipment) and mobile devices in the plant. Mobility aspect has to be guaranteed because cabling connects equipment to fixed locations but reduces the flexibility in equipment placement and reorganization. It can also be very expensive to install and maintain in terms of material and lab costs.
Power Consumption:
It is essential that wireless devices such as sensor nodes operate on batteries to guaranty the mobility aspect and energy must be saved. If battery operated, power consumption should be kept low; rechargeable battery is preferred to reduce costs. Suitable power consumption conservation techniques have to be used.
Reliability:
The connectivity should be provided to ensure communication ability between all network devices despite the many obstacles that the industrial space presents. High reliability is a key requirement in IREMO. A minimum aggregate rate of information must be delivered from the transmitter to the receiver node. Some critical scenarios require high or even total end-to-end reliability. Link reliability mechanisms such as MAC layer automatic repeat request - ARQ can significantly reduce the end-to-end packet loss ratio.
Availability:
The availability of the wireless network is the main requirement and the most decisive factor on the connectivity and data acquisition because real-time measurements are necessary as the only input for the control station and the neural network for process control. Due to the problem with electromagnetic emissions created by large motors (e.g CEMC), heavy equipment, and typical industrial machinery (moulds) a degradation of reliability by real-time measurements and a frequently data loss can be provoked. These emissions can create extremely high levels of noise that interfered with wireless equipment.
Selection of wireless Technology:
In order to make a decision, which technology suits best to the IREMO application, it is necessary to compare them based on essential requirements listed above and essentially from a typical industrial environment perspective such as in SORA.
T4.3: Organisation of control system's nodes
Based on the identified requirements BIBA is working on the implementation of the overall concept using open source tools for data acquisition and Storage. The solution will provide ACCI, KARN and SORA the possibility to use two different wireless technologies (WiFi and ZigBee) at the same time.
For the placement and organisation of sensor nodes BIBA has considered each case (SORA, KARN, and ACCI) apart. For ease case BIBA will work out a solution with different requirements:
SORA's case:
Based on the map of SORA plant and our observation during the visit of SORA (March 2010) we found out some criteria which are very important and decisive for the topology of the sensor network and the organisation of the network components to guarantee the most of requirements listed above.
1- The number of moulds
2- The distance between moulds
3- The obstacles in the pant
The number of router in the network depends on
1- the distance between the gateway and the mould (press).
2-the transmission range that is linked to the signal strength of the antenna of the wireless device
Wi-Fi
Concerning Wi-Fi, the technology is also tested in BIBA and it should be adequate for the IREMO application. The only disadvantages are power consumption and the need of more equipment (router with bridging function) for the connectivity of network components. Routers and/or access points cannot be operated on batteries because of the very high-power consumption of Wi-Fi devices. To minimise cabling (between OptiMold and Wi-Fi) and to enhance mobility degree BIBA proposed for all cases two solutions: the Wi-Fi device (Netgear WNCE2001 adapter) or the combination of open source platforms (Arduino microcontroller, Wi-Fi and Ethernet module).
ACCI's case:
The case of ACCINA is similar as in SOTIRA. The pultrusion machines are located in an environment influenced by many metallic objects which are the major problem with radio frequency based networks. They can lead to unreliable network connectivity. Compared to SOTIRA plant, ACCIONA plant has no walls between monitoring system and gateway which facilitate the communication and avoid the installation of relays (routers) for data forwarding.
KARNIC's case:
In the KARNIC case (low-volume production) with only one mould in the plant and due to the distance between the production area and the server room, it is not obligatory to use relay stations between OptiMold and the gateway which is connected to the server (in server room 2).
Deviation and corrective actions
Following the revised DOW after the midterm review, some tasks of WP4 have been merged in WP2.
The WP4 is completed.
Lot of modifications (Adjustments) regarding the architecture, the communication modules (acquisition and Trigger) and the processing tools have been provided after the finishing of WP4. This work has been done in WP5.2.
WP5: Systems' integration, trials and evaluation
The main objectives of this WP were
i) the development different prototypes, i.e 3 control production systems,
ii) trials and
iii) evaluations of the systems developed in the others WPs.
T5.1: Man-machine interface, standardization and inter-operability
In perfect agreement with original plans, involved partners have worked on the user man/machine user interface. Accordingly to the description of work, the high volume RTM case has been first treated. The others cases was also finalized.
Knowledge of the user's need:
The first approach has consisted to associate the user to the definition and specifications of the final tool, in other words: how do they want the system works and for who ?
Regarding now the infusion case, it has been described how the system of the control could look like. Some of the values will always be shown, such as:
- vacuum level (at the pump or at the resin trap); and
- elapsed time.
There are three different issues that should be monitored:
- the resin pot or the mixer (during the infusion process/stage);
- the resin arrival (during the infusion process/stage); and
- the laminate (while it is curing).
The main features that have to be monitored are:
- mixing ratio of the resin in the pot. It has to be between the acceptable upper and lower limits;
- temperature of the resin in the pot. The value has to be between the acceptable upper and lower limits;
- viscosity of the resin in the pot. This value has to be between the acceptable upper and lower limits. It is linked to the temperature of the resin.
- Counter showing the stage of gelation. The time shown in this field shows the time left during which it is possible to continue the infusion with the resin that is left in the resin pot. When this time is lower than 20 minutes the colour of the numbers will change to red to show that a new resin mix must be prepared.
- Quantity of resin within the laminate. A digital caudal meter or a digital counter will be placed in the pipe that feeds the laminate. When this measurement tends to zero, the infusion can be stopped.
- mixing ratio of the resin being injected. It has to be between the acceptable upper and lower thresholds;
- Temperature of the resin being injected. The value has to be between the acceptable upper and lower thresholds;
- viscosity of the resin being injected. This value has to be between the acceptable upper and lower thresholds. It is linked to the temperature of the resin.
- Quantity of resin within the laminate. A digital caudal meter or a digital counter will be placed in the pipe that feeds the laminate. When this measurement tends to zero, the infusion can be stopped.
- degree of conversion. The desired degree to be reached will be highlighted in the graphs as a threshold.
- Tg. The desired Tg to be reached will be highlited in the graphs as a threshold. When all sensors (real and soft) achieve the desired Tg, the oven can be stopped. An alarm should let the user know that the oven can be stopped manually.
- Would it be useful to monitor the viscosity during this phase?
- Temperature at the sensor locations (real and soft). An upper threshold will be highlighted in the graph. If any of the sensors reach this limit, the temperature of the oven will automatically decrease 200C until the temperature at that sensor location gets 100C lower. At that moment, the oven will recover its original temperature (200C higher).
- Temperature inside the oven. Several classical temperature sensors (thermocouples or RTDs) will be placed at different locations along the beam (especially in resin rich areas). The same upper threshold than in the previous point will be highlighted in the graph. If any of the sensors reach this limit, the temperature of the oven will automatically decrease 200C until the temperature at that sensor location gets 100C lower. At that moment, the oven will recover its original temperature (200C higher).
Interfaces-software part:
The results in term of interface is presented hereafter
The idea was to limit, as much as possible, the number of entries. These last ones have to be short and do not need any particular formation. The messages have to be clear with no superfluous information.
A particular attention has been paid to:
i) Choice of color
ii) The structure of the document
iii) The localization of the information (alert, signal etc) and their identification
Problems encountered and corrective actions:
No significant problems were found in the realization of this task.
T5.2: Software and hardware Systems integration and adaption (full scale)
Several moduli were created and compose the final system: a communication box, a trigger box, a java tool for acquisition, processing and data storage, a control tool (including off and on line models), and a production tool. First all these moduli was successfully tested and individually validated. Then the final prototype was constructed i.e the development of a unique tool integrating all the elements.
Regarding the modifications done after the finishing of WP4, as mentioned before, the final solution (Software and Hardware) has been improved.
- Software:
For the data processing tool (Java-based for acquisition, processing and storage) lot of modifications (based on the results of the trials) have been realized in order to optimize the final solution and to facilitate the identification of problems (connection of sensors, checking the actually state of the process, etc.), especially during the integration and testing phases. After running up the tool and by entering the command (function) HELP (or LS), relevant functions (GO, STOP, STATE, SENSORS) will appear on the display and can be used also in background to:
- start the process: Informing workers to close the mold and starting data acquisition, processing and storage.
- stop the process: Opening the mold (automatically) and stopping data acquisition, processing and storage.
- show the current state of the process and to
- list the connected sensors to the tool.
T5.3: Installation at low-volume case and full-scale trial.
ACCIONA Infraestructuras has already manufactured two composite beams by vacuum infusion process using the control tools developed within the IREMO project. The materials, curing cycle, control and software have been checked. Tg values obtained in laboratory are consistent with those obtained by the models, and consequently as proposed by the control system.
Problems encountered and corrective actions
Communication problems were faced during the first demo but they have been solved for the second infusion. Furthermore, improvements have been adapted in the control tools (both based on Neural Network and on Evolution Strategy) from the first to the second infusion in order to get more accurate and fast predictions on the physical-chemical parameters, mainly Tg.
T5.4: Installation at Mid-volume case and full-scale trial
The production is now using the system and operators supervise it. One of the most significant results is that the control system is able to highlight the fact that the first part produced in a more than one batch requires longer curing time (due to the lower temperature of the mold).
T5.5: Installation at High-volume case and full-scale trial
Several solutions have been studied for the packaging of the control system. Several points are requested:
i) Robust (use in an industrial environment, high level of IP (6X)
ii) Touch screen
iii) Easily Transportable
iv) Adaptable and compatible with existing system and equipments
Installation of the high volume production prototype was done. In this case, existing tools (injection machine/press) needed some modification in order to integrate the IREMO system.
Modifications of existing tools (injection machine/press) for the integration of the control system.
Characterisations of the thermal properties are in progress by measuring the effective Tg of the parts. These last one appears conform with the target, i.e around 115 °C, but also parts with artificial problems such as mixing ratio deviations.
T5.7: Control system evaluation and standardization
The objective of this task is to evaluate the performance of the whole control system (hardware and software)
Although this task is not fully achieved, the first trials with control are very promising. As described in WP2, the control system is tested in production conditions in different plants.
Particular attention is paid to the following points: the collect of the data such as: resin arrival, degree of cure, start of acquisitions and control decisions.
In the 3 cases, the systems are working satisfactorily for all different processing conditions: temperature, types of resins, carbon/glass fibers etc.
Acciona's case:
Two beams were manufactured by ACCIONA Infraestructuras using the control tools and the first has already been evaluated with the performance of DSC tests at ACCIONA Infraestructuras' laboratory with good agreement (+/- 5°C)
Karnic's case:
Data were collected during trials, using single and pendulum.
During the trials the following observations were made:
- During production using the pendulum arrangement it was difficult for the operator to observe visually on the computer screen the control indicator. For this reason, the control software was modified to give an audio signal as well.
- Data from the pendulum setup were collected using an additional Optimould box and a second computer to avoid confusion and conflict of the information.
- The system setup for the first mould (Mould-A) performed throughout with reasonable accuracy. However, the system for the second mould (Mould-B), even though is identical, was not fully reliable and control signal was found a few times to be taking longer time than expected. The problem was identified to be on the hardware and relevant changes were made.
- As expected, the first part produced in a more than one batch, required longer time to cure due to the lower temperature of the mould. The control appeared to identify successfully this instant.
The results collected during the control trials period have assisted in refining the system to provide reasonably accurate curing time predictions. The inconsistency of some readings however, suggests that further testing and refinements are necessary (in progress during July and demonstration activities)
Sotira's case:
Sotira Automotive manufactures for Audi RS3 the two front fenders in epoxy /carbon composite.
We have chosen those parts, with the agreement of AUDI, for IREMO project.
WP6: Industrial scale demonstration
For WP6 the main objective is to demonstrate the new production system for the three main applications
T6.1: Demonstration of low-volume case [ACCI]
ACCIONA Infraestructuras has already manufactured two composite beams by vacuum infusion process using the control tools developed within the IREMO project as per the requirement of the DoW.
The set up in the demo was to monitor in real time:
- Curing of resins and temperature at six different locations using flexible curing and temperature sensors;
- Resin in the resin bath using a pot sensor;
- temperature in the oven during curing of the part.
Thanks to these sensors, that the following parameters could be predicted using the IREMO system during the trial:
- viscosity of the resin;
- degree of cure of the resin;
- Tg of the resin.
Preliminary trial and demo #1.
One preliminary trial with the same monitoring set up was performed prior to the demo#1 trial itself in order to:
- avoid any errors in the manufacturing of the product that has been chosen for the demo;
- check the wireless communication;
- check the control tools;
- confirm that the IREMO system has been well understood within the ACCIONA team.
T6.2: Demonstration of mid-volume case [KARN]
The production of certain type parts produced by Karnic using the RTM method is a repetitive process where operations are repeated in every part production cycle with target to achieve maximum possible output during a specified batch period. For one complete cycle, that is the completion of one single part, the required production time can be divided in two categories depending on the occupancy of the operator.
These are:
(a) Time-A: the time at which the operator is fully occupied to perform value adding work in the process. This consists of all tasks from de-moulding the part already in the mould to preparation, mould closing and resin injection of the next part.
(b) Time-B: the time required for the part to cure during which the operator is unable to do value adding work and can only be treated as idle time.
The actual production time of the part is therefore the sum of (Time A + Time B).
As all value adding operations can be controlled by careful process planning and the adoption of specialised tools, Time-A can be of a fixed known value for every part. On the contrary, Time-B is dependent upon the chosen resin type, the environment temperature, the use of post curing and the percentage of catalyst used therefore, it may have significant variation from part to part and the overall efficiency of the process.
T6.3: Demonstration of high-volume case [SORA]
For the demonstration case, SOTIRA uses the IREMO system for moulding AUDI RS3 Fender. SOTIRAs target is to use the system with a complete automation to optimize the cycle time. SOTIRA wants to make a minimum of 18 parts per shift with the monitoring system and order directly the cycle of the couple 'injection machine - press' according to the degree of cure.
Each operation as monitoring or automation of the press with the IREMO system has been tested successfully, especially with prototype system. That was the work of the WP5 stage.
WP7
T7.1: Market analysis and perspectives
The following documents have been prepared to understand how the new technology could be used in potential applications, which could be the potential customers and their needs, and what are the competitor's strategies and the competing systems or solutions already available on the market:
- The functional schemes of the IREMO system (one scheme by pilot case) to clarify the objectives of the project and to define the 'IREMO system'.
- The benchmarking of the existing competitive technologies (cf. Task 1.1) whose the objective is to identify the main competitive systems already on the market, or in development, and to collect data concerning their principle, their characteristics and the suppliers which develop these systems.
- The SWOT analysis to understand the strengths, weaknesses of the consortium/project, and to identify opportunities and threats for the end results.
- The market analysis with data on the global composite market and focus by applications sectors (automotive, wind turbine, transportation, building and construction, aerospace, boat building) to identify opportunities and limitations to market the results of the project and to define a marketing strategy.
T7.2: Dissemination of results
Communication tools have been realized in order to disseminate the results of the project:
- Creation of the project logo.
- Creation of the public website (see http://www.IREMO.eu online) with inscription to Google Analytics to follow the traffic.
- Creation of the project flyer for public diffusion.
- Creation of a Powerpoint presentation of the project that any partner can use to present the project to non-project members.
- Creation of posters and banners to promote the project during exhibitions.
- Creation of a list of relevant seminars, exhibitions, conferences or workshops where there is potential opportunity to present the results of the project.
- Creation of a list of relevant magazines where the results of the project could be presented.
Many actions of dissemination have been undertaken since the beginning of the project.
- Project members have made technical and scientific presentations in twenty conferences and symposiums all over the world.
- Project members have published almost fifteen articles and papers in specialized press like scientific journals, technical magazine and conference proceedings.
- Four press releases have been diffused in order to announce and promote dissemination actions
- Around ten articles and news have been published by the specialized press about IREMO project, consecutively to the publication of IREMO press release and news by the consortium.
Moreover, in order to disseminate as better as possible the results of the project, the consortium has decided to attend two composites trade shows as exhibitor:
- Composites Europe 2011 - Stuttgart, Germany - 27/29 September 2011
- JEC Europe 2012 - Paris, France - 27/29 March 2012
The objectives were to have direct contact with industrial and scientific communities and also to communicate about the two workshops organized by IREMO in parallel of these events.
T7.3: Exploitation of results
Atoutveille had prepared and organized the Exploitation Strategy Seminar (ESS) which held the 7th of June 2011 in close relation with Alberto Belle, the expert consultant who animated the seminar.
Following the seminar, the following tasks have been achieved:
- ESS seminar synthesis
- Animation of the thinking about the definition of exploitable results
- Collect of information for the PUDF document following the definition of new exploitable results
- Improvement of the marketing strategy: modification and updating of the IREMO Market Strategy Table
T7.4: Workshop organization
According to the Document of Work (DoW), three public workshop sessions were planned in order to disseminate the result of the project in effective manner.
The first workshop has been held in September 2011 at Stuttgart in Germany, in parallel of Composites Europe 2011 exhibition (27-29 September 2011).
Potential Impact:
The global composites industry is now producing an annual 8m tonnes of product, worth an estimated 60bn EUROS in 2010, and with an average value of 7.2 EUROS per kg. However, the EUROS per kg value of that product currently differs widely from region to region. With a 36% market share by value and 35% by volume, the value of North American composites production is estimated at 22bn EUROS, equating to an average unit price of 8.2 EUROS per kg. Overall production volume in 2010 was 2.7m tonnes. In the EMEA (Europe, Russia, the Middle East and Africa), 2m tonnes were produced, 33% of the market by value and 22% by volume, with a value of 20bn EUROS and an average unit price of 10.0 EUROS per kg. In Asia-Pacific and the rest of the world (including Australia and South America), the average unit price of composites is much lower, at 5.5 EUROS per kg. The 3.3m tonnes produced in these regions in 2010 had a value of 18bn EUROS, representing 43% of the market by volume, but just 31% by value. This is explained by the increasing use of primarily carbon composites in higher-end applications in North America and EMEA - most notably so far in aerospace and wind energy - with consequently higher prices. Growth similarly varies greatly by sector, with the smaller markets - by volume, though certainly not added value - continuing to grow rapidly.
Market Perspectives
The overall market for advanced composites - which comprises composites based on carbon fibres, carbon nanotubes and graphene - will more than triple to USD 25.8 billion by 2020, according to the Lux Research report Carbon Fibre and Beyond: The USD 26 Billion World of Advanced Composites. Lux forecasts that the use of advanced composites by wind turbines manufacturers will increase from USD 2.5 billion in 2011 to USD 15.4 billion in 2020. Meanwhile, growth in aerospace will lag, despite the introduction of new aircraft that use large quantities of carbon fibre reinforced plastic (CFRP), such as Boeing's 787 Dreamliner. In 2020, wind energy will account for nearly 60% of the market for composites, compared with the current 35%.
Industrial impact
The main impact of the IREMO project is the significant increase of production rates, maintaining quality and improving the robustness of liquid composite moulding, one of the most advanced and upcoming manufacturing processes. This impact will be mainly achieved by the integration of machines, tooling, process monitoring and simulations under the same platform of process optimisation and control.
In particular IREMO is expected to contribute significantly to the:
-Reduction of the number of rejected components or products by 50%
-Reduction of power consumption by at least 15%
-Increased throughput and capability of processes and productivity maintaining repeatability and accuracy by a factor of 30% with respect to specific Tg and reduced porosity.
-Minimisation (or even elimination) of the use of services, e.g. acetone by a factor of more than 60% in some cases
As well as to:
-the reduction of down times during product exchange and conflict situations;
-the improvement of product quality while reducing the need for inspection;
-the increase in machine availability and reduction in maintenance; and
-improved efficiency of complex production systems
So besides the positive contributions of the project to the specific targeted impacts of this call further contribution to wider-range impacts are:
-the decrease in the usage of environmental unfriendly solvents
-easier advancement of the composites manufacturers towards closed moulding (zero emissions).
Societal impact
In a wider perspective IREMO responds to the societal needs via the improvement of working conditions by introduction of more automated processes minimising contact and exposure with potential irritating agents such as resin, and via the implementation of technological progresses in materials and manufacturing techniques increasing the skilfulness of the European workforce and securing the employment level in Europe.
Health and safety impact by the industrial production of components in Constructions
IREMO will contribute to the improvement of the occupational health and safety conditions of the workers at the construction of infrastructure structural elements by the improvements associated at the manufacturing process (automated control), transport (less weight to transport) and assembly (less weight to handle, less time, better guidance system) compared to conventional concrete elements. In particular, this aspect is of relevant importance, as construction sector is one of the most dangerous based on the fact that many more accidents occur in construction per every 100.000 workers than overall in the workforce : in 2001, there were 7.200 non-fatal accidents at work per 100.000 construction workers, as compared to 3.800 accidents per 100 000 workers for the total of the nine economy sectors for which comparable statistics exist. For fatal accidents at work the difference was even greater.
Health and safety impact at industrial production level
IREMO will improve significantly the occupational health and safety conditions of the employees at the production level. As confirmed by SORA and KARN their target to introduce closed liquid composite moulding is driven also by their motivation to eliminate VOC (Volatile of Carbon) emissions in the production site completely. This is extremely important as on one hand the styrene emissions from the polyester and vinylester use and on the other hand the unhealthy emissions of the epoxy resins when cured create extremely unhealthy working conditions.
Main dissemination activities
It has been defined that project results will be presented in workshops, conferences and symposia by the academic participants (NTUA, UNEW) and the research institutions (CEMC, TECN and BIBA). The industrial partners (SORA, ACCI and KARN) will disseminate giving publicity to the project and its results in technical magazines and well-known organizations, also industrial partners will be in charge of demonstration activities.
IREMO website
-Atoutveille has created and published the official website of the project (with the technical support of a web design company which hosts the site): http://www.IREMO.eu.
-The website can be entirely managed by Atoutveille, it will be maintained during the whole duration of the project and the 2 years following the end of the project.
-This public website is associated with an internal website platform used by the members of the consortium as a dynamic tool for data exchange and report depositary. It is also a means for European officers in charge of IREMO to be informed about the advancement of the project. Furthermore, a mailist containing all consortium member e-mail address has been established to facilitate exchange between IREMO partners
Press Releases
Four press releases have been sent to specialized press in the field of composite materials, like technical magazine, online technical website. They have generated news published in several support.
- 1st Press Release has been disseminated the 2nd of April 2010 in order to announce the beginning of the project and to detail the consortium, the context and the technical objectives of the project.
- The 2nd Press Release has been disseminated the 06th of January 2011 in order to announce the participation of IREMO consortium as speaker to the Symposium PPE RTM & Infusion in St Avold, France (9th and 10th February 2011).
- The 3rd Press Release has been disseminated the 15th of February 2012 in order to announce the organisation of the 2nd Technical Workshop of the IREMO project during a session at SEICO 12 International Conference in Paris, France, the 27th of March 2012. (Session 4B: IREMO from 08:30 to 09:50).
- The 4th Press Release has been disseminated the 11th of May 2012 with the objective to announce the organization of the 3rd and Final Workshop and Demonstration Sessions of the IREMO project at SOTIRA plant in Saint-Méloir-des-Ondes, France, the 18th of July 2012.
List of Websites:
http://www.iremo.eu