Final Report Summary - ACCENT (Adaptive control of manufacturing processes for a new generation of jet engine components)
The manufacture of rotating aero engine parts is subject to special designation, such as critical. The designation is intended to convey the need for special controls. Either the part is evaluated against the design intent (production is controlled to deliver product consistent with evaluation), or manufacturing methods identified as sensitive are controlled (specifications and / or validated parameter limits). However, once validated, the process is seen as 'frozen' and no changes are allowed without costly and time consuming re-validation. However, described methods and validation procedures do not guarantee repeatable quality by 100 %. The reasons why manufacturing processes cannot be fully fixed and still have uncertainties are many. This results in a range of variable part quality, which must be taken into account during design and stress calculations. To compensate, the manufacturers must apply more conservative manufacturing process parameters and change tools more frequently. For safety reasons, production and manufacture leads to non-optimised machining processes with high manufacturing costs. ACCENT allows the European aero engine manufacturers to improve their competitiveness by applying adaptive control techniques to the manufacture of their components. By operating in a multi-dimensional 'approved process window', processes are optimised to the prevailing conditions and no longer 'frozen'. The use of an advanced adaptive control loop ensures that component integrity is maintained.
Work package (WP) 2 has delivered all the information necessary to develop the ACCENT methodology and commenced with the definition of various process input descriptions for the range of materials (Inconel 718, Udimet 720, Ti 64 and Ti-6-2-4-6) and processes under investigation (turning, milling, drilling, and broaching). Defining a standard testing procedure provided the correlation between the process inputs and the process monitored manufacturing process outputs and various machining strategies for the expansion of process windows were proposed and developed. Test and validation of the new standardised procedure was completed and test reports collated. For each process and material combination, the trials for the process window tests are explicitly described, and from the analysis of the generated data, process validated windows are generated. The validated process windows consist of process parameter data, process monitoring data, and surface integrity information.
WP3 delivered an understanding of how to use process monitoring output in a closed loop adaptive control system (ACS), such that the machining process is monitored against a validated process window. Hence, for each material process combination, both the ability of the machine tools and the chosen process monitoring systems were tested. Signal analysis, feature extraction and development of algorithms are covered. In this task, algorithms should be understood as a methodology or strategy on how to react on certain features in the monitoring signals. Early analysis of the generated data showed that different levels of sophistication were required dependent upon the process under investigation, the methodology of testing, and the applied process monitoring strategies used. Customisation of the ACS to suit end-user trial conditions was a main focus for the demonstrators and was enabled for a milling / drilling process. In this case, the ACCENT control loop has been implemented on a real production machine tool.
A significant task of WP4 was to complete the survey of surface integrity definitions from the Industrial partners. Thus, design proven surface integrity specifications had to be documented and a definition of the desired surface integrity values are elaborated. The information is critical to the linkage between adaptive control signals and defined component surface quality. However, these tasks proved to be more complex than expected as it was difficult to converge to a common surface integrity understanding. It was required be able to define which features and anomalies characterise the surface integrity, what kind of common inspection methodologies characterise each anomaly and allow definition of acceptable values, and what are acceptable values for each feature that delimit the acceptable surface integrity windows.
Project context and objectives:
It is accepted that it is impossible to design modern gas turbine engines with total redundancy, i.e. that the failure of any single component can be accommodated by alternative load paths, containment. Rotating parts, the failure of which may hazard the airframe, are subject to special controls and designated as critical or some other suitable designation, such as flight safety part or life controlled part. The part designation is intended to convey the need for special controls to all parties who manufacture or handle the part. Hence, the part designation is systematic and may go beyond the drawing i.e. controlled source material, validated and controlled manufacturing processes and procedures. The Pensacola event in 1996 was the result of a rotating part failure. Subsurface damage caused by a drilling defect resulted in the loss of two lives and one serious injury. If this failure occurred in flight the consequences would have been disastrous.
Two approaches to process validation are used in the industry. The first approach is defined as part specific process validation (SPV) where the part is evaluated against the design intent and subsequent production is controlled to deliver product consistent with the evaluation. The second method is known as generic manufacturing process validation (GMPV) whereby those manufacturing methods that are identified as being sensitive, i.e. needing a high level of control if manufactured product is to meet the design intent, are controlled by specifications and / or validated parameter limits. Sometimes both methods are used in combination to validate the manufacturing process for a part. However, once validated, no changes are allowed to the process without costly and time consuming re-validation. Hence the process is seen as 'frozen' and no changes to process parameters are permitted without time consuming and costly re-validation. Even then, the use of the above described methods and validation procedures do not guarantee repeatable quality by 100 %.
The reasons, why manufacturing processes cannot be fully fixed and still have uncertainties are:
- variation in material properties;
- variation in part stability and fixture design;
- variation in tool quality and life;
- variation in coolant flow / cooling efficiency;
- random and unexpected process anomalies;
- deviation due to human factors.
These result in a particular range of variable part quality, which must then be taken into account for design and stress calculations. To compensate, the manufacturers are required to apply a more conservative choice of manufacturing process parameters and change tools more frequently to ensure part surface integrity. Out of safety reasons, production and manufacture is still kept very conservative and leads to a non-optimised machining process with high manufacturing costs. The above situation is also repeated each time new cutting tool materials, new machine tool capabilities, and especially new high strength materials replace existing technologies. Validation of new manufacturing methods (or even an adaptation of an existing method) can easily exceed a timeframe of two to four years until it is approved and introduced into production. There are several areas of the current operating procedure for critical parts where recurring costs are incurred.
These may be summarised as:
- The iterative loop where process parameters are changed to meet the needs of process improvement, change of tool design, tool manufacturer. At one level, changes to machining parameters to achieve acceptable surface finish and dimensional control, followed by costly laboratory examination of the surface integrity are repeated until an acceptable surface integrity that meets the design life is achieved. At this point the process may be considered as 'frozen', i.e. fixed for an indefinite period. However, at a second level, the process may again be changed to meet the next level of process improvement, or newly developed tooling and the whole circle of incurred costs are repeated. It is very common for this activity to be repeated many times throughout the manufacturing life of a given component design.
- Another feature in the production of critical components is that critical processes or critical design features may call for the need of special process controls. In this case, the recurring costs described above will apply to each critical process or each critical feature. Therefore the recurring cost may be multiplied may times. In the case where the whole component is subject to critical process controls, each and every process must undergo the time consuming, labour intensive and hence high cost of validation.
- Assuming the above points are satisfied, the next areas of concern where costs are incurred are as a consequence of variable part quality. Unpredictable tool wear and unpredictable tool quality, variation in material properties, variation in component stability and fixturing, coolant delivery, and the ever present threat of process malfunctions and process anomalies, requires the process parameters to be set at a very conservative and hence 'safe' level. Therefore manufacturing costs of critical components are always higher than they would be if the above variability could be catered for.
- Finally, the design is not fully optimised in terms of weight, performance, and life as a consequence of the need to take all the other variable factors into consideration. Whilst conservative in nature because of safety issues, the aero engine business is nevertheless a highly competitive multi-billion market. Sales margins are often stripped to a bare minimum in order to secure that all-important first application on a new aircraft, as the ongoing revenue for new engines and spares may last more than 30 years. In order to win and maintain engine sales, engine manufacturers have to design market leading products that incorporate low weight, high performance, low emissions, and low maintenance costs, which in an ever more demanding marketplace forces designers to use more exotic materials and novel designs.
With the above stated manufacturing issues making this job even more demanding, it is easy to understand that there is a clear need to improve the manufacture of these components. A review of the European aero engine manufacturers indicates that the costs associated with the introduction of a manufacturing change i.e. changing a cutting tool insert from one edge geometry to another for example could cost in the region of EUR 135 thousand. If multiplied by the number of features / processes that may be constantly changing for a given component and the possibility for process improvement every 2 years for example, then the typical costs incurred during the manufacturing life of one component could be as much as EUR 13.5 million. It is considered difficult to put a figure on the total cost of manufactured critical parts produced each year, but it is considered realistic that processes are operating at approximately 60 % of what they could be if the need to back off the process parameters to be 'safe' could be resolved. Therefore a 40 % saving in the manufacturing costs of all critical parts could be realised. Finally, the ability to optimise component weight, performance, and service life could be worth several billions of Euro in new engine sales and spares revenue. Hence, ACCENT will allow the European aero engine manufacturers to improve their competitiveness by applying adaptive control techniques to the manufacture of their components. By operating in a multi-dimensional 'approved process window', processes will be optimised to the prevailing conditions and no longer 'frozen'. The use of an advanced adaptive control loop will ensure that component integrity is maintained. Benefits will be seen in terms of reduced part manufacturing process time, more consistent part quality in terms of geometry, surface and sub-surface properties, tool usage optimisation, elimination of costly part re-validation due to small process changes, and the possibility to improve component design due to optimised machined surfaces.
The ACCENT project has been organised into five distinct but interconnected WPs.
WP1 is an all-encompassing WP defined to ensure that the project is planned correctly and managed effectively, that all WPs are coordinated and deliver their defined objectives and deliverables, that effective exploitation of the project results is achieved, that ethical, gender, and health, safety and environmental aspects are given due consideration, and that the consortium will foster good working relationships at both a personal and organisational level. A key component of the project concerns the standardisation of multi-dimensional processes.
WP2 is defined to ensure that the project is capable of generating multi-dimensional machining processes for the range of process and material combinations under investigation and is focused on establishment of the required process windows and related tasks such as definition, test and validation of testing procedures. Within the project, different teams of end users and universities are working together on several process material combinations, and different strategies will be selected for the respective process material groups. Hence, information in terms of material property variation, the range of cutting tools and associated tool life and failure modes, coolants, and machine tools. is required. Also, the connection of machining process parameter windows and surface integrity is examined.
Main areas to be addressed are:
1. definition and standardisation of process inputs descriptions;
2. definition of a testing plan for evaluation of manufacturing situation;
3. definition of a testing procedure;
4. test and validation of procedure;
5. establishing process validation windows;
6. characterisation of process operation dependencies.
WP3 will deliver an understanding of how to use process-monitoring systems for a closed loop ACS that will keep the machining process within a defined process window based on the definitions and specified requirements of WP2. Supportive tasks, such as the definition of adaptive control inputs for each machining process, software tools for use throughout the project, and algorithms for the adaptive control strategy will be developed. In the scope of ACCENT, four typical machining processes which are routinely involved in the manufacturing process chain of safety critical rotating engine parts are considered: these are turning, hole-making, milling and broaching. Correlating the machining process with the produced product quality and surface integrity and review of the machining experiments must be done regarding detectable and typical features representing anomalies or failures that occur during the machining operation which threaten the part quality. The outcome of this task will be fed into the adaptive control strategy development, and developed algorithms can be considered the link between the input to the adaptive control strategy and the component surface integrity. Consequently, processes such as signal analysis, the extraction of certain features and their correlation to physical events occurring during the machining processes are inputs to the algorithm development. Once the algorithms are developed and tested, strategies will be developed to feed decisions based on the algorithms back to the process.
WP4 will study the interaction between the surface integrity generated as a result of the machining process parameters, cutting tool and machine tool condition, and component material characteristics. The linkage between manufacturing process window parameters, desired surface integrity requirements, detection of process variation and correction within the control loop, and the final inspected surface integrity, will necessitate a review of current data storage capabilities and the development of new data management architecture. This knowledge will allow the design function to better understand the effect of machining processes on the part quality and subsequent component service, and also offer the opportunity for the component design to be optimised. In defining the component surface integrity requirements and the dependencies between surface integrity and manufacturing processes, the development of a surface integrity based part validation procedure can be realised. When validated, this procedure will allow better decision-making on part acceptability and hence significantly reduce inspection costs. In a similar manner, the effective communication of acceptable surface quality limits and technology to ensure consistent surface quality will allow production engineers to reduce new component introduction / development costs.
The primary objective of WP5 is to ensure that a framework exists whereby the consortiums industrial partners can directly exploit the results of the project, and that scientific research generated by the project can be exploited and disseminated by the consortium research organisations. Within this framework, intellectual property rights, rules regarding exploitation, and licensing of project deliverables will be formulated within the consortium agreement.
Project results:
The standardisation of multi-dimensional processes (WP2) has delivered all the information necessary to develop the ACCENT methodology. The WP is divided into different tasks focusing on partial aspects which are described below.
The project consortium commenced work with the definition of the various process input descriptions for the range of materials (Inconel 718, Udimet 720, Ti 64 and Ti-6-2-4-6) and processes under investigation (turning, milling, drilling, and broaching). The aero engine manufacturing partners (Industrial partners) made freely available their chosen material specifications, heat treatment requirements and material property variations to the consortium. And in a similar way, information regarding cutting tools, machining parameters, tool wear data, coolant application and machine tool descriptions were provided. This valuable information was collated and openly exchanged between the project partners in the form of a deliverable.
With the supplied input descriptions, testing plans for evaluation of the Industrial partners manufacturing conditions were developed for use at the university partners laboratories. It was important to verify that the results obtained at the laboratories would match those obtained in the Industrial partner companies, hence fully industrial machine tools were employed (with the exception of large Broaching machines), and production standard material was employed. This strategy was defined in order to raise the possibility that the newly developed technology was readily transferrable. For each laboratory, plans for testing tool life and tool failure modes were developed along with acceptable and unacceptable tool failure modes.
Within the project, most of the partners developed a method to carry out the tool failure test along with the first window trials to save machining time as well as additional batches of test tools. A major task for the Industrial partners was dedicated to the timely delivery of appropriate test material to the university partners. Here, it was vital for the testing campaign, that test material could be provided having the same specification as the original work pieces used in component production, so that the transferability of the achieved results could be guaranteed.
A task focused on process monitoring capabilities (a catalogue of available hardware and software for tool life and failure mode testing) was created and distributed to the partners. A final subtask focused on procedures for determining tool wear for each of the process and material combinations, and testing of the typical tool lives and failure modes. Basically, two tool wear categories could be determined: abrasive wear and tool chipping or major cutting edge breakouts. The latter one, if occurring in the window trials, could be regarded as uncontrolled tool wear. Such forms of tool wear can be usually related back to insufficient tool quality. Within the window trial, this occurred for one process material combination. The related tests were stopped and not regarded for the further establishment of the process window. The finalising deliverable covers the measurements of tool wear as it is applied in industry as well as for the university partners.
A task to define a standard testing procedure provided the correlation between the process inputs and the process monitored manufacturing process outputs. A test parameter matrix was completed for the various processes, tools and material combinations which were specified by the Industrial partners. Thereby, it was important that the test cases of the Industrial partners could be transferred to the university machine tools.
The tools and parameters to be considered in the tests for every process material combination were developed within a dedicated subtask. During testing, most of the defined tools and selected parameters could be used without any problems. However some modification of parameters was required where chipping and breakage of a particular tool occurred. The applicable process monitoring system hardware, software (see WP3), and methodologies on how to perform process window trials have been evolved and reported in a project deliverable. Within the project, various machining strategies for the expansion of process windows were proposed and developed, i.e. a tool material pair (TMP) approach (AFNOR E66-520), a design of experiment method, step by step trials, and a non-dimensional Péclet number based on cutting temperature.
In order to collate the trial data, a definition of a data base software tool (which was a main WP4 activity), accessible by each partner in a common format, was proposed for the collection of test documentation and associated measurement signals. As a further step, the data acquisition software tools had to be tested in combination with the selected process monitoring hardware in order to ensure an operational system for the window trials. With the testing phase complete, the results were summarised in a project deliverable in form of a document on the standardised testing procedure. A project deliverable documents the standardised testing procedure. With the definition of the standard testing procedure complete, test and validation of the new standardised procedure commenced. In this task, there were two activities: trials for testing and validating the standardised process window procedure, and the creation of the data base information for use in WP4. On completion of the trials, test reports were issued and collated into a project deliverable. For every process and material combination, the trials for the process window tests are explicitly described. This includes the entire testing procedure and testing methodology, the process monitoring output and how to process the generated data, how the surface integrity data is gathered (i.e. surface roughness results, tool wear evolution, photographic evidence of material surfaces, boundary zone structure and residual stress measurements).
Furthermore, information about the process window axes are included in the deliverable as the window axes have to be clear from a qualitative point of view in this task in order to focus on the right process parameters while performing the tests. With the amount of raw data generated, a solution for how the information could be shared was needed. It was decided that the raw data would be stored at the universities, and a software tool developed in WP3 would be used to condense the data and transform it into a common format for sharing on the project data base. The foremost information in the data base is the data triple of cutting parameters, process monitoring data (i.e. signal features) and the surface integrity information. It was agreed on having the information at three stages of complexity: the 'simple' window information that distinguishes the surface integrity on just acceptable and not acceptable, a second 'more complex' window information including tables with extracted features linked to certain surface integrity parameters and cutting conditions and finally the complete raw data, extracted signals features and calculated values out of signal features and the test description which offers experts the possibility to work on the data again. All the activities were summarised in a project deliverable. Furthermore, the completion of the trials for test and validation of the standard procedure was an important milestone of the project.
From the analysis of the generated data, process validated windows were generated. The validated process windows consist of the process parameter data, process monitoring data, and the surface integrity information. With reference to previously defined axis definition, the data was analysed regarding process and surface integrity sensitive values that can be determined from the process monitoring signals. By putting all the information together, windows can be identified by setting frames at the axes. This was considered an ongoing process until the end of the project as further tests ensured a statically more accurate result. The window frames for process signals, parameters and finally the creation of the multi-dimensional process window can be divided into different steps. In the first step, factors such as machine tool and process limits can be specified in advance. As an example, too low cutting speeds result in an unacceptable roughness profile that comes from a kinematic point of view. On the next level are factors that cannot be determined in advance. This includes, e.g. excessive cutting speed or unpredictable tool wear which leads to unacceptable surface integrity. In order to make a conclusion about the actually generated surface, laborious investigations were performed. A related deliverable, including pictures of the process windows for the process material combinations has been issued.
The characterisation of process operation dependencies is the main focus of a further task, and describes the differences between the real situation (as found in industrial production) and the trials being performed under laboratory conditions. Under laboratory conditions, many factors can be controlled for a distinct number of experiments. This, however, is more difficult in the production situation. The first part of the task covered the transferability of the process windows to situations which are different to the ACCENT trials. This can on the one hand be similar experiments being carried out under different conditions, but also machining operations in the production environment. The approach here was to relate or convert all measures to a physical level, such as specific cutting forces, material properties, temperatures. The question being discussed in this task was how it is possible to scale results in order to make them usable for different situations. The second question related to the transfer of results is about the so called inconsistent factors and parameters. These are factors that have not been taken into account within the ACCENT trials but possibly have an influence on the results. Such factors can either emerge on a very infrequent basis and are therefore difficult to investigate, or parameters that are just not known by the manufacturers and therefore could not be specified. A scientific discussion on these issues (which is the output of this task) is collated into a project deliverable.
The knowledge that has been collated and developed within the consortium (within the scope of the project) could help to build the basis for a potential future European standard. For that reason, the most important insights from the work being carried out within WP2 is considered on a on a generic level which could be applied for other applications and processes in the regarded field. Consequently, this document contains the most important steps for a generic process window approach and could serve for creation of a standard in the future.
'Adaptive control and monitoring' (WP3), was designed to deliver an understanding of how to use process monitoring output in a closed loop ACS, such that the machining process is monitored against a validated process window. The WP is divided into different tasks focusing on partial aspects which are described below.
A first task related to sensor system definition, was the clarification that all universities involved in machining experiments had the capability to run trials at their test facilities. Initially, sensor system specifications were elaborated for use in the various machining experiments at the university partner laboratories. The only exception to this is for the broaching process. Trials were undertaken at the Industrial partner sites due to the unavailability of representative broaching machines at the University partner sites. Following trials, elaboration of the process monitoring equipment and strategies for use in the process window trials were defined and reported. Hence for each material process combination, both the ability of the machine tools and the chosen process monitoring systems were tested and verified as suitable for ongoing trials. Furthermore, the industrial partners' capability to use the strategies developed within the scope of the project were evaluated and reported. The efficient communication and understanding of the demands of the industrial partners was vitally important to the development of algorithms and control strategies. Moreover, it was important for the development to understand the limitations for the use of process monitoring systems in an industrial environment. In contrast to the university laboratories where sophisticated systems are used in a test set-up, industrial production has a very strong need for robustness and easy handling. In a further subtask, the definition of the process monitoring equipment for the operation window trials was elaborated. This is the equipment used by the university partners (except for the case of broaching in which an industrial machine tool has been equipped with a university sensor system). Due to the limitations of systems in an industrial environment, the customisation of the systems used in the window trials is a further activity within this WP. The final sensor systems for each process material combination within the project are defined in a final deliverable of the task.
In this task, the signal analysis, feature extraction and development of algorithms is covered. Thereby, the results being elaborated in former project contributing to this field were reviewed e.g. the work done within the MANHIRP project undertaken in the Sixth Framework Programme (FP6) (in which some of the ACCENT partners were involved). A major output from the MANHIRP project was an increased understanding of how to reduce the probability of component failure caused by the presence of manufacturing induced anomalies. However, it also provided an insight into the strategies and algorithms used when attempting to process monitor the formation of the identified machining induced surface anomalies. Within the ACCENT project, the review of signal analysis, feature extraction and development of algorithms was changed from a short duration task into one that continued throughout the project. In this task, algorithms should be understood as a methodology or strategy on how to react on certain features in the monitoring signals. The link between features in the signals and the quality measures from the component often require true 'algorithms' such as mathematical operations e.g. transformation from a time to frequency domain, and in these cases they are included in 'feature extraction' activities. Early analysis of the generated data showed that different levels of sophistication were required dependent upon the process under investigation, the methodology of testing, and the applied process monitoring strategies used. Hence the analysis phase and algorithm development continued to run in parallel until the end of the project. A final subtask merges all the information gathered in the previous subtasks and summarises the algorithms which are used for the different processes. The outcome of the task is a deliverable.
Software programming is a task that continued throughout the project as software tools were required at different levels and at different stages. Firstly, tools to be programmed for the acquisition of process monitoring data at either the university or industrial partners' shop floors were required. This necessitated the availability of not only a process monitoring solution, but appropriate data acquisition software. Ideally, each of the groups working on different process and material combinations would have used the same solutions, but inevitably this was not possible. Hence a software tool was developed that was able to convert existing data files and provide the data with uniform header information for storage into the data base. In this way, partners were able to collect their data with whatever system they had available, but the collected data was available and exchangeable throughout the consortium. The header information also made it possible to search for data related to a specific partner, process, sensor type, parameter set.
Secondly, software tools had to be developed for proving the ACCENT closed loop. The concept is that the quality produced by the machining process is monitored against the defined process window that is proven to produce acceptable surface quality. The process window is therefore a mathematical multi-dimensional parameter 'room' with the acquired process monitoring sensor data functioning as the basis for correlation. In order to compute the values, however, features in the monitoring signals and the link to product quality or surface integrity had to be precisely understood. Even though process monitoring is meant to protect components in production, it has to be ensured that false alarms do not cause additional costs. For this reason, the extent to which a closed loop is connected to a production system has a significant impact on its effectiveness. The main question is how the process monitoring system being used for the university trials has to be customised when being transferred to an industrial situation? And then, how the related software is modified? A decision for this dilemma could be made on the basis of several factors such as:
- the quality of the sensor signals acquired (e.g. depending on machine tool, sensor system, production environment), hence difficult to control;
- the controllability of the input parameters (e.g. material, tool, variance of machining parameters);
- confidence in evaluating features in the signals (e.g. amplitude, curve shape);
- correlation of surface features with signal features (e.g. probability of detection of surface anomalies).
The extent to which the closed loop is applied is an important point of the project, not least because it determines the extent to which process monitoring is used in the production environment in a post ACCENT industrial application. Within the project, this task has not been performed for every process material combination, but was completed for the demonstrators at both the university lab stage and the industrial stage. A further sub-task covers the tools which have to be programmed for the tests on the sample components, again on both university lab and industrial stage. The related deliverable includes the details to the task.
The central action of WP3 is the development of an adaptive control strategy in order to realise the ACCENT idea. The task is divided into three subtasks. Within the first, the control strategy was developed on different levels: First of all in a very generic way being universally valid for all regarded ACCENT process material combinations. In later steps, the concept is more and more broken down and developed into specific solutions for the single cases as mentioned in the introduction. The control strategy is described in form of a control loop and shows where inputs from tasks and subtasks as well as outputs are directed. The control concept is discussed in the final deliverable in a very detailed way.
The second and third subtask were addressed in a combined activity and cover efforts to prove the control concept and further show, that boundary conditions for the established process windows are met. The boundary conditions are provided by WP2 ('Process window data') and WP4 ('Surface integrity data'). At the beginning of the project, it was planned that demonstrators would be located in the end-user production environment. Within the second project period however, it was decided within the consortium to accept industrial demonstrators at university laboratories, as well. During the project, it turned out that in some cases the further implementation into the production environment would have caused a far more extended project schedule. On the other hand, the industrial demonstrators at the universities being implemented on industrial machine tools rather than test beds show very similar results and the transfer to the end users will not give further scientific output. Besides the generic control loop concept which is applicable for every process material combination covered in ACCENT, the related deliverable shows the demonstrator cases on both a university laboratory level (which here means: fully industrial machine tool at a university facility with laboratory sensor equipment), as well as a demonstrator in a production environment of one of the industrial partners.
In a task dedicated to the evaluation of closed loop strategies for the real manufacturing situation, the development of strategies to react to process malfunctions was completed. In the former tasks, the control strategies were developed on a very scientific basis mainly by university partners. Within the test series, a wide range of monitoring equipment was used for data acquisition in order to get the best possible process data. When it comes to implementation in an industrial environment, many restrictions exist regarding additional equipment to the machine tools, systems interacting with the machine tool controller or the companies' production planning systems. The questions here are how deep may a control system interfere with the processes in production? Are there limits out of reasons such as disturbance of other processes, violations of any machine tool components, when additional sensor or control systems are introduced? All such restrictions for a potential ACCENT system beyond demonstration level are discussed with the Industrial partners in the consortium and collected in a document.
A further task covers the customisation of the ACS. The customisation of ACS to suit end-user trial conditions as discussed which within the project has its main focus on the demonstrators. As mentioned before, within ACCENT different demonstrators will be realised: on the one hand demonstrators on industrial machine tools (and industrial application) at the universities and on the other hand demonstrators in a production environment at the Industrial partners. For both kinds of demonstrators, the equipment used is different from the one used within the window trials. Furthermore, results being attained during the window trials with laboratory equipment have to be transferred to a commercial monitoring solution available at the Industrial partner or even further programmed to the machine tool controller itself. Since the demonstrator at university laboratory level could be used with the equipment employed during the process window trials, there was no need for further customisation. Only some adjustments had to be made to the software tools being used. In terms of the demonstrator on the industrial level at one of the original equipment manufacturer (OEM) partners, the entire control strategy had to be customised in order to have the control strategy working in combination with a commercial monitoring system. The related efforts are explained in a corresponding deliverable to the task. Finally, a test and verification of adaptive control strategy performance was made. This was mainly tested on the industrial demonstrator at one of the partners which was enabled for a milling / drilling process. In this case, the ACCENT control loop has been implemented on a real production machine tool. The system was designed to control the tool feed force over the cutting speed in order to have a higher productivity and be able to still operate in the process window in terms of the achieved quality and surface integrity, respectively. A related subtask covers the test and verification on sample components. Within the tests, the measures for a comparison to the former state-of-the-art process were the tool wear, the surface roughness as well as an optical analysis of the achieved residual surface layers. The related work is presented in a deliverable of the task.
A significant first task of WP4 was to complete the survey of surface integrity definitions from the Industrial partners. Thus, the design proven surface integrity specifications had to be documented and a definition of the desired surface integrity values had to be elaborated. This information is critical to the linkage between adaptive control signals and the defined component surface quality. However, these tasks proved to be more complex than expected, and in fact, it appears that the reason it was difficult to converge to a common surface integrity understanding was to be able to define:
- Which features and anomalies characterise the surface integrity?
- What kind of common inspection methodologies characterise each anomaly and allow definition of acceptable values?
- Which are the acceptable values for each feature that delimit the acceptable surface integrity windows.
Hence, a new subtask was added to the project that examined the methodologies used to categorise each anomaly. With the four manufacturing processes, and three typical materials under investigation, it was necessary to determine each type of feature and anomaly that can occur in each process and material combination. With reference to the work of FW6 MANHIRP project, in which anomalies were classified into two categories (geometrical and non-geometrical), a proposed solution in ACCENT was to define two new categories:
- Surface topography: features or anomalies on the manufactured surface.
- Microstructural changes: features or anomalies in the manufactured sub-surface.
During the first period, each feature / anomaly has been described with their corresponding non-destructive and destructive inspections, and a list of features / anomalies which may occur in each process / material combination has been issued. To complete this task, partners have defined which anomaly / feature had occurred during their trials and what kind of examination was required to detect them.
Another fundamental requirement of the work has been to establish the link between component surface integrity and the manufacturing dependencies that produce it, i.e. cutting conditions, tool wear. A methodology was to start from key features, e.g. roughness, and to correlate the feature to the cutting conditions. However when reviewing the task, the partners agreed that the original concept did not match the real world situation.
The following points were noted:
- Manufacturing dependencies can only be determined from trials using the defined parameters from the process windows and dependencies may not be found for all anomalies as the trials are not defined to provoke them. Anomalies that are produced will be as a consequence of the parameters used.
- Different post machining operations have been identified, but the effectiveness of post-machining operations (e.g. shot peening) largely depends on the surface condition produced by the machining process. Hence it is important to control the previous machining operation.
- Lifing policies will have an influence by adherence to currently applied quality standards.
For the development of a surface integrity part validation procedure, the definitions of surface integrity defined in a previous task led the partners to a methodology of surface integrity examination that was consistent for all trials. Consequently, a minimum requirement for examinations to be performed is proposed and tested.
Two steps of examination are distinguished:
- The minimum set of examinations (MSEs) that have to be performed to characterise acceptable surface integrity windows having the same observational scales.
- An additional set of examinations which has to be performed for some trials: The aim is to validate if the MSE is sufficient to characterise the surface integrity, and to complete the correlation between Phase modulation (PM) signals and surface integrity examinations.
A key task completed by the partners was to validate their process inputs in relation to their component quality requirements. For each process and material combination, acceptable surface integrity windows have been defined, and based on trials, a list of generated anomalies and their process parameter tendencies are known. Thus the dependencies between process inputs and the achieved surface integrity results are validated.
A survey on the data storage demands for surface integrity values and manufacturing process signals led to the definition of a new data storage requirement. Within the project framework, and consistent with the project aims, the task was to define what is logically possible to identify as the basic requirements for the data structure to be employed. This data structure was created on the requirements and outputs from WPs 3, 4. To develop the data management architecture for the surface integrity values and PM signals, the most important consideration was to build the infrastructure that is effective and available during the experimental process window machining trials. Different architectures were investigated with the aim of finding tools that directly focussed on the architecture and not the informatical development. In the early stages of the project, data was collected in line with the requirements defined in a data management deliverable, and sent to the WP4 manager for evaluation and to estimate the required resource storage requirement. The new data storage requirement has been described in a project deliverable. In order to provide a possible European standard based on the results generated in WP4, a task that collected and collated all the surface integrity results generated in the project was performed. The main objective of the task was to make guidelines for surface integrity analysis after machining based on the trials performed by the partners. The deliverable D4.7 constitutes a guideline based on a basic understanding of surface integrity anomalies extracted for the project deliverables and a critical review of the methodology used in the project.
The deliverable presents:
- a list of features and anomalies which define the surface integrity with their corresponding examinations to be characterised;
- a feedback on the methodology used for examination by all partners through their investigations of anomaly characterisation;
- a review of best practices for surface integrity analysis methodology.
Potential impact:
The ACCENT project has brought together the leading European aero engine manufacturers and European universities with the common aim to improve the competiveness of critical aero engine component manufacture, whilst at the same time ensuring that the safety and reliability of these components is improved verses current manufacturing methods.
As a direct consequence of the involvement in the ACCENT project, the European aero engine manufacturers have the opportunity to implement the improved process capability demonstrated in the project. Based on the machining process and component material combinations investigated, strategies for developing operating process windows, new process monitoring strategies, and novel adaptive control methods are now realised. The benefits, both in terms of reduced manufacturing costs and improved component quality control are already being achieved within some of the end user companies, whilst others are putting internal development programmes in place in order to implement ACCENT methodologies.
For the academic partners involved in the ACCENT project, opportunities to enter research into novel machining research has given greater awareness of their capabilities within the project consortium, and into the wider community with the publication of papers, journals, and representation at conferences and other events. This has also raised the awareness of the needs to control the manufacture of critical aero engine components to the wider manufacturing community and the general public.
The primary objective of ACCENT WP5 was to ensure that a framework exists whereby the consortiums industrial partners could directly exploit the results of the project, and that scientific research generated by the project could be exploited and disseminated by the consortium research organisations. Associated work was structured into four tasks:
Exploitation of project results by industrial partners
Based on the technical results from WPs 2 - 4, the task was defined to derive a thorough understanding of the issues associated with transferring ACCENT technology into an industrial environment, and to give the Industrial partners guidelines for implementing the project results into their production plants. Through this task, the industrial partners have used their ACCENT experience for supporting machine tool and process monitoring device investment as a short term exploitation activity. Most of them have already implemented individual process monitoring devices, or machines equipped with PM sensors and / or updated their procurement specification for future machine investment. Also, initial use of process monitoring systems has been implemented into industrial partner's production workshops. These implementations have been focused on critical rotating parts and are mostly applied for hole-making operations for tool wear control or special cause events detection, but implementation on broaching or milling has also commenced.
In addition, a longer term action plan has also been defined by the Industrial partners. It includes the application of process monitoring devices and associated data storage system implementation, the definition of new engineering specifications, training of personnel and shop floor implementation requirements. These activities have been identified as the most critical for industrial partners in order to be able to implement ACCENT overall control. In line with an aviation authority recommendation (AC FR33.70-2) some Industrial partners have started to implement machining process control with process monitoring use through updated or new engineering specifications, especially for hole-making processes.
Exploitation of academic research
This task was defined to promote the reputation of the university based partners, the ACCENT project as a whole, and the support of the European research framework. The technical expectations of the six academic partners were mainly focused on the development and expansion of technical knowledge in the research fields of process monitoring and machining processes related to the aero engine industry, expanding knowledge in the machining of difficult to machine materials, and developing new research on the correlation between process monitoring signals and component surface integrity characteristics. Due to their university status, the academic partners also considered the project an opportunity to confirm and improve the reputation of their respective universities, share and receive knowledge from other European laboratories, and to cooperate with the European aircraft aero engine manufacturers.
As part of this task, a total of nine doctorate (PhD) theses and nine Master theses, either partially or totally dedicated to ACCENT project, have been implemented by the academic partners. Researchers work has been focused on the ACCENT methodology, tool wear behaviour analysis, process monitoring strategies for machining, or the developed surface integrity obtained after machining. In addition, associated junior training and scientific dissemination has been performed by academic partners.
Exploitation and dissemination to the wider community
This task was defined to ensure that project results would be disseminated to the wider community and to evaluate possible exploitation opportunities in other Industrial sectors. A significant number of publications of ACCENT work have been published through European and international journals and conferences. These documents deal mainly with tool wear prediction methodologies, acceptable surface integrity machining domain definition, and process monitoring of machining operations to ensure acceptable surface integrity. Also, attendance and presentation of ACCENT work to national and international conferences has been undertaken by the consortium members. In taking the opportunity to organise such conferences, ACCENT dedicated sessions were included in the agenda of the following events events (CIRP ICME, Italy, 23 - 25 June 2010; INTERCUT Cluny - France, 14 October 2010). As a consequence, promotion of the reputation of individual partners and the support of the European research framework has been assured. A dedicated project website has been implemented: http://www.accent.wzl.rwth-aachen.de/en/default.html(öffnet in neuem Fenster)
Definition of an ACCENT standard
Based on the collation of technical WP output, the objective of this task was to evaluate the possibility of creating an ACCENT standard. The standard would describe generic requirements for the definition of process windows, methods describing process window validation, and descriptions of how the adaptive control of manufacturing processes with respect to surface integrity may be achieved. To give importance to the standard, the European Aviation Safety Agency (EASA) was contacted to see if they were able to support the issue of an ACCENT standard document. The project coordinators organised a dedicated meeting with EASA propulsion certification team and gave a detailed presentation of the ACCENT project. However, according to the current EASA position and strategy, the following statements have been concluded:
- EASA is not in a position to certify generic manufacturing sequences for critical aero engine components. Each application of a company's manufacturing strategy will be dealt with on a case by case basis.
- Results of the ACCENT project should be passed over to the propulsion certification group in order to strengthen the awareness of monitoring and control needs in the manufacture of aero engine components.
- EASA wishes to be involved in future activities concerning the dissemination of ACCENT results to a wider community and informed of future follow up activities in the field of adaptive machining in critical aero engine components.
Therefore, an ACCENT standard document has not been requested by EASA.
Project website: http://www.accent.wzl.rwth-aachen.de/en/default.html(öffnet in neuem Fenster)
Work package (WP) 2 has delivered all the information necessary to develop the ACCENT methodology and commenced with the definition of various process input descriptions for the range of materials (Inconel 718, Udimet 720, Ti 64 and Ti-6-2-4-6) and processes under investigation (turning, milling, drilling, and broaching). Defining a standard testing procedure provided the correlation between the process inputs and the process monitored manufacturing process outputs and various machining strategies for the expansion of process windows were proposed and developed. Test and validation of the new standardised procedure was completed and test reports collated. For each process and material combination, the trials for the process window tests are explicitly described, and from the analysis of the generated data, process validated windows are generated. The validated process windows consist of process parameter data, process monitoring data, and surface integrity information.
WP3 delivered an understanding of how to use process monitoring output in a closed loop adaptive control system (ACS), such that the machining process is monitored against a validated process window. Hence, for each material process combination, both the ability of the machine tools and the chosen process monitoring systems were tested. Signal analysis, feature extraction and development of algorithms are covered. In this task, algorithms should be understood as a methodology or strategy on how to react on certain features in the monitoring signals. Early analysis of the generated data showed that different levels of sophistication were required dependent upon the process under investigation, the methodology of testing, and the applied process monitoring strategies used. Customisation of the ACS to suit end-user trial conditions was a main focus for the demonstrators and was enabled for a milling / drilling process. In this case, the ACCENT control loop has been implemented on a real production machine tool.
A significant task of WP4 was to complete the survey of surface integrity definitions from the Industrial partners. Thus, design proven surface integrity specifications had to be documented and a definition of the desired surface integrity values are elaborated. The information is critical to the linkage between adaptive control signals and defined component surface quality. However, these tasks proved to be more complex than expected as it was difficult to converge to a common surface integrity understanding. It was required be able to define which features and anomalies characterise the surface integrity, what kind of common inspection methodologies characterise each anomaly and allow definition of acceptable values, and what are acceptable values for each feature that delimit the acceptable surface integrity windows.
Project context and objectives:
It is accepted that it is impossible to design modern gas turbine engines with total redundancy, i.e. that the failure of any single component can be accommodated by alternative load paths, containment. Rotating parts, the failure of which may hazard the airframe, are subject to special controls and designated as critical or some other suitable designation, such as flight safety part or life controlled part. The part designation is intended to convey the need for special controls to all parties who manufacture or handle the part. Hence, the part designation is systematic and may go beyond the drawing i.e. controlled source material, validated and controlled manufacturing processes and procedures. The Pensacola event in 1996 was the result of a rotating part failure. Subsurface damage caused by a drilling defect resulted in the loss of two lives and one serious injury. If this failure occurred in flight the consequences would have been disastrous.
Two approaches to process validation are used in the industry. The first approach is defined as part specific process validation (SPV) where the part is evaluated against the design intent and subsequent production is controlled to deliver product consistent with the evaluation. The second method is known as generic manufacturing process validation (GMPV) whereby those manufacturing methods that are identified as being sensitive, i.e. needing a high level of control if manufactured product is to meet the design intent, are controlled by specifications and / or validated parameter limits. Sometimes both methods are used in combination to validate the manufacturing process for a part. However, once validated, no changes are allowed to the process without costly and time consuming re-validation. Hence the process is seen as 'frozen' and no changes to process parameters are permitted without time consuming and costly re-validation. Even then, the use of the above described methods and validation procedures do not guarantee repeatable quality by 100 %.
The reasons, why manufacturing processes cannot be fully fixed and still have uncertainties are:
- variation in material properties;
- variation in part stability and fixture design;
- variation in tool quality and life;
- variation in coolant flow / cooling efficiency;
- random and unexpected process anomalies;
- deviation due to human factors.
These result in a particular range of variable part quality, which must then be taken into account for design and stress calculations. To compensate, the manufacturers are required to apply a more conservative choice of manufacturing process parameters and change tools more frequently to ensure part surface integrity. Out of safety reasons, production and manufacture is still kept very conservative and leads to a non-optimised machining process with high manufacturing costs. The above situation is also repeated each time new cutting tool materials, new machine tool capabilities, and especially new high strength materials replace existing technologies. Validation of new manufacturing methods (or even an adaptation of an existing method) can easily exceed a timeframe of two to four years until it is approved and introduced into production. There are several areas of the current operating procedure for critical parts where recurring costs are incurred.
These may be summarised as:
- The iterative loop where process parameters are changed to meet the needs of process improvement, change of tool design, tool manufacturer. At one level, changes to machining parameters to achieve acceptable surface finish and dimensional control, followed by costly laboratory examination of the surface integrity are repeated until an acceptable surface integrity that meets the design life is achieved. At this point the process may be considered as 'frozen', i.e. fixed for an indefinite period. However, at a second level, the process may again be changed to meet the next level of process improvement, or newly developed tooling and the whole circle of incurred costs are repeated. It is very common for this activity to be repeated many times throughout the manufacturing life of a given component design.
- Another feature in the production of critical components is that critical processes or critical design features may call for the need of special process controls. In this case, the recurring costs described above will apply to each critical process or each critical feature. Therefore the recurring cost may be multiplied may times. In the case where the whole component is subject to critical process controls, each and every process must undergo the time consuming, labour intensive and hence high cost of validation.
- Assuming the above points are satisfied, the next areas of concern where costs are incurred are as a consequence of variable part quality. Unpredictable tool wear and unpredictable tool quality, variation in material properties, variation in component stability and fixturing, coolant delivery, and the ever present threat of process malfunctions and process anomalies, requires the process parameters to be set at a very conservative and hence 'safe' level. Therefore manufacturing costs of critical components are always higher than they would be if the above variability could be catered for.
- Finally, the design is not fully optimised in terms of weight, performance, and life as a consequence of the need to take all the other variable factors into consideration. Whilst conservative in nature because of safety issues, the aero engine business is nevertheless a highly competitive multi-billion market. Sales margins are often stripped to a bare minimum in order to secure that all-important first application on a new aircraft, as the ongoing revenue for new engines and spares may last more than 30 years. In order to win and maintain engine sales, engine manufacturers have to design market leading products that incorporate low weight, high performance, low emissions, and low maintenance costs, which in an ever more demanding marketplace forces designers to use more exotic materials and novel designs.
With the above stated manufacturing issues making this job even more demanding, it is easy to understand that there is a clear need to improve the manufacture of these components. A review of the European aero engine manufacturers indicates that the costs associated with the introduction of a manufacturing change i.e. changing a cutting tool insert from one edge geometry to another for example could cost in the region of EUR 135 thousand. If multiplied by the number of features / processes that may be constantly changing for a given component and the possibility for process improvement every 2 years for example, then the typical costs incurred during the manufacturing life of one component could be as much as EUR 13.5 million. It is considered difficult to put a figure on the total cost of manufactured critical parts produced each year, but it is considered realistic that processes are operating at approximately 60 % of what they could be if the need to back off the process parameters to be 'safe' could be resolved. Therefore a 40 % saving in the manufacturing costs of all critical parts could be realised. Finally, the ability to optimise component weight, performance, and service life could be worth several billions of Euro in new engine sales and spares revenue. Hence, ACCENT will allow the European aero engine manufacturers to improve their competitiveness by applying adaptive control techniques to the manufacture of their components. By operating in a multi-dimensional 'approved process window', processes will be optimised to the prevailing conditions and no longer 'frozen'. The use of an advanced adaptive control loop will ensure that component integrity is maintained. Benefits will be seen in terms of reduced part manufacturing process time, more consistent part quality in terms of geometry, surface and sub-surface properties, tool usage optimisation, elimination of costly part re-validation due to small process changes, and the possibility to improve component design due to optimised machined surfaces.
The ACCENT project has been organised into five distinct but interconnected WPs.
WP1 is an all-encompassing WP defined to ensure that the project is planned correctly and managed effectively, that all WPs are coordinated and deliver their defined objectives and deliverables, that effective exploitation of the project results is achieved, that ethical, gender, and health, safety and environmental aspects are given due consideration, and that the consortium will foster good working relationships at both a personal and organisational level. A key component of the project concerns the standardisation of multi-dimensional processes.
WP2 is defined to ensure that the project is capable of generating multi-dimensional machining processes for the range of process and material combinations under investigation and is focused on establishment of the required process windows and related tasks such as definition, test and validation of testing procedures. Within the project, different teams of end users and universities are working together on several process material combinations, and different strategies will be selected for the respective process material groups. Hence, information in terms of material property variation, the range of cutting tools and associated tool life and failure modes, coolants, and machine tools. is required. Also, the connection of machining process parameter windows and surface integrity is examined.
Main areas to be addressed are:
1. definition and standardisation of process inputs descriptions;
2. definition of a testing plan for evaluation of manufacturing situation;
3. definition of a testing procedure;
4. test and validation of procedure;
5. establishing process validation windows;
6. characterisation of process operation dependencies.
WP3 will deliver an understanding of how to use process-monitoring systems for a closed loop ACS that will keep the machining process within a defined process window based on the definitions and specified requirements of WP2. Supportive tasks, such as the definition of adaptive control inputs for each machining process, software tools for use throughout the project, and algorithms for the adaptive control strategy will be developed. In the scope of ACCENT, four typical machining processes which are routinely involved in the manufacturing process chain of safety critical rotating engine parts are considered: these are turning, hole-making, milling and broaching. Correlating the machining process with the produced product quality and surface integrity and review of the machining experiments must be done regarding detectable and typical features representing anomalies or failures that occur during the machining operation which threaten the part quality. The outcome of this task will be fed into the adaptive control strategy development, and developed algorithms can be considered the link between the input to the adaptive control strategy and the component surface integrity. Consequently, processes such as signal analysis, the extraction of certain features and their correlation to physical events occurring during the machining processes are inputs to the algorithm development. Once the algorithms are developed and tested, strategies will be developed to feed decisions based on the algorithms back to the process.
WP4 will study the interaction between the surface integrity generated as a result of the machining process parameters, cutting tool and machine tool condition, and component material characteristics. The linkage between manufacturing process window parameters, desired surface integrity requirements, detection of process variation and correction within the control loop, and the final inspected surface integrity, will necessitate a review of current data storage capabilities and the development of new data management architecture. This knowledge will allow the design function to better understand the effect of machining processes on the part quality and subsequent component service, and also offer the opportunity for the component design to be optimised. In defining the component surface integrity requirements and the dependencies between surface integrity and manufacturing processes, the development of a surface integrity based part validation procedure can be realised. When validated, this procedure will allow better decision-making on part acceptability and hence significantly reduce inspection costs. In a similar manner, the effective communication of acceptable surface quality limits and technology to ensure consistent surface quality will allow production engineers to reduce new component introduction / development costs.
The primary objective of WP5 is to ensure that a framework exists whereby the consortiums industrial partners can directly exploit the results of the project, and that scientific research generated by the project can be exploited and disseminated by the consortium research organisations. Within this framework, intellectual property rights, rules regarding exploitation, and licensing of project deliverables will be formulated within the consortium agreement.
Project results:
The standardisation of multi-dimensional processes (WP2) has delivered all the information necessary to develop the ACCENT methodology. The WP is divided into different tasks focusing on partial aspects which are described below.
The project consortium commenced work with the definition of the various process input descriptions for the range of materials (Inconel 718, Udimet 720, Ti 64 and Ti-6-2-4-6) and processes under investigation (turning, milling, drilling, and broaching). The aero engine manufacturing partners (Industrial partners) made freely available their chosen material specifications, heat treatment requirements and material property variations to the consortium. And in a similar way, information regarding cutting tools, machining parameters, tool wear data, coolant application and machine tool descriptions were provided. This valuable information was collated and openly exchanged between the project partners in the form of a deliverable.
With the supplied input descriptions, testing plans for evaluation of the Industrial partners manufacturing conditions were developed for use at the university partners laboratories. It was important to verify that the results obtained at the laboratories would match those obtained in the Industrial partner companies, hence fully industrial machine tools were employed (with the exception of large Broaching machines), and production standard material was employed. This strategy was defined in order to raise the possibility that the newly developed technology was readily transferrable. For each laboratory, plans for testing tool life and tool failure modes were developed along with acceptable and unacceptable tool failure modes.
Within the project, most of the partners developed a method to carry out the tool failure test along with the first window trials to save machining time as well as additional batches of test tools. A major task for the Industrial partners was dedicated to the timely delivery of appropriate test material to the university partners. Here, it was vital for the testing campaign, that test material could be provided having the same specification as the original work pieces used in component production, so that the transferability of the achieved results could be guaranteed.
A task focused on process monitoring capabilities (a catalogue of available hardware and software for tool life and failure mode testing) was created and distributed to the partners. A final subtask focused on procedures for determining tool wear for each of the process and material combinations, and testing of the typical tool lives and failure modes. Basically, two tool wear categories could be determined: abrasive wear and tool chipping or major cutting edge breakouts. The latter one, if occurring in the window trials, could be regarded as uncontrolled tool wear. Such forms of tool wear can be usually related back to insufficient tool quality. Within the window trial, this occurred for one process material combination. The related tests were stopped and not regarded for the further establishment of the process window. The finalising deliverable covers the measurements of tool wear as it is applied in industry as well as for the university partners.
A task to define a standard testing procedure provided the correlation between the process inputs and the process monitored manufacturing process outputs. A test parameter matrix was completed for the various processes, tools and material combinations which were specified by the Industrial partners. Thereby, it was important that the test cases of the Industrial partners could be transferred to the university machine tools.
The tools and parameters to be considered in the tests for every process material combination were developed within a dedicated subtask. During testing, most of the defined tools and selected parameters could be used without any problems. However some modification of parameters was required where chipping and breakage of a particular tool occurred. The applicable process monitoring system hardware, software (see WP3), and methodologies on how to perform process window trials have been evolved and reported in a project deliverable. Within the project, various machining strategies for the expansion of process windows were proposed and developed, i.e. a tool material pair (TMP) approach (AFNOR E66-520), a design of experiment method, step by step trials, and a non-dimensional Péclet number based on cutting temperature.
In order to collate the trial data, a definition of a data base software tool (which was a main WP4 activity), accessible by each partner in a common format, was proposed for the collection of test documentation and associated measurement signals. As a further step, the data acquisition software tools had to be tested in combination with the selected process monitoring hardware in order to ensure an operational system for the window trials. With the testing phase complete, the results were summarised in a project deliverable in form of a document on the standardised testing procedure. A project deliverable documents the standardised testing procedure. With the definition of the standard testing procedure complete, test and validation of the new standardised procedure commenced. In this task, there were two activities: trials for testing and validating the standardised process window procedure, and the creation of the data base information for use in WP4. On completion of the trials, test reports were issued and collated into a project deliverable. For every process and material combination, the trials for the process window tests are explicitly described. This includes the entire testing procedure and testing methodology, the process monitoring output and how to process the generated data, how the surface integrity data is gathered (i.e. surface roughness results, tool wear evolution, photographic evidence of material surfaces, boundary zone structure and residual stress measurements).
Furthermore, information about the process window axes are included in the deliverable as the window axes have to be clear from a qualitative point of view in this task in order to focus on the right process parameters while performing the tests. With the amount of raw data generated, a solution for how the information could be shared was needed. It was decided that the raw data would be stored at the universities, and a software tool developed in WP3 would be used to condense the data and transform it into a common format for sharing on the project data base. The foremost information in the data base is the data triple of cutting parameters, process monitoring data (i.e. signal features) and the surface integrity information. It was agreed on having the information at three stages of complexity: the 'simple' window information that distinguishes the surface integrity on just acceptable and not acceptable, a second 'more complex' window information including tables with extracted features linked to certain surface integrity parameters and cutting conditions and finally the complete raw data, extracted signals features and calculated values out of signal features and the test description which offers experts the possibility to work on the data again. All the activities were summarised in a project deliverable. Furthermore, the completion of the trials for test and validation of the standard procedure was an important milestone of the project.
From the analysis of the generated data, process validated windows were generated. The validated process windows consist of the process parameter data, process monitoring data, and the surface integrity information. With reference to previously defined axis definition, the data was analysed regarding process and surface integrity sensitive values that can be determined from the process monitoring signals. By putting all the information together, windows can be identified by setting frames at the axes. This was considered an ongoing process until the end of the project as further tests ensured a statically more accurate result. The window frames for process signals, parameters and finally the creation of the multi-dimensional process window can be divided into different steps. In the first step, factors such as machine tool and process limits can be specified in advance. As an example, too low cutting speeds result in an unacceptable roughness profile that comes from a kinematic point of view. On the next level are factors that cannot be determined in advance. This includes, e.g. excessive cutting speed or unpredictable tool wear which leads to unacceptable surface integrity. In order to make a conclusion about the actually generated surface, laborious investigations were performed. A related deliverable, including pictures of the process windows for the process material combinations has been issued.
The characterisation of process operation dependencies is the main focus of a further task, and describes the differences between the real situation (as found in industrial production) and the trials being performed under laboratory conditions. Under laboratory conditions, many factors can be controlled for a distinct number of experiments. This, however, is more difficult in the production situation. The first part of the task covered the transferability of the process windows to situations which are different to the ACCENT trials. This can on the one hand be similar experiments being carried out under different conditions, but also machining operations in the production environment. The approach here was to relate or convert all measures to a physical level, such as specific cutting forces, material properties, temperatures. The question being discussed in this task was how it is possible to scale results in order to make them usable for different situations. The second question related to the transfer of results is about the so called inconsistent factors and parameters. These are factors that have not been taken into account within the ACCENT trials but possibly have an influence on the results. Such factors can either emerge on a very infrequent basis and are therefore difficult to investigate, or parameters that are just not known by the manufacturers and therefore could not be specified. A scientific discussion on these issues (which is the output of this task) is collated into a project deliverable.
The knowledge that has been collated and developed within the consortium (within the scope of the project) could help to build the basis for a potential future European standard. For that reason, the most important insights from the work being carried out within WP2 is considered on a on a generic level which could be applied for other applications and processes in the regarded field. Consequently, this document contains the most important steps for a generic process window approach and could serve for creation of a standard in the future.
'Adaptive control and monitoring' (WP3), was designed to deliver an understanding of how to use process monitoring output in a closed loop ACS, such that the machining process is monitored against a validated process window. The WP is divided into different tasks focusing on partial aspects which are described below.
A first task related to sensor system definition, was the clarification that all universities involved in machining experiments had the capability to run trials at their test facilities. Initially, sensor system specifications were elaborated for use in the various machining experiments at the university partner laboratories. The only exception to this is for the broaching process. Trials were undertaken at the Industrial partner sites due to the unavailability of representative broaching machines at the University partner sites. Following trials, elaboration of the process monitoring equipment and strategies for use in the process window trials were defined and reported. Hence for each material process combination, both the ability of the machine tools and the chosen process monitoring systems were tested and verified as suitable for ongoing trials. Furthermore, the industrial partners' capability to use the strategies developed within the scope of the project were evaluated and reported. The efficient communication and understanding of the demands of the industrial partners was vitally important to the development of algorithms and control strategies. Moreover, it was important for the development to understand the limitations for the use of process monitoring systems in an industrial environment. In contrast to the university laboratories where sophisticated systems are used in a test set-up, industrial production has a very strong need for robustness and easy handling. In a further subtask, the definition of the process monitoring equipment for the operation window trials was elaborated. This is the equipment used by the university partners (except for the case of broaching in which an industrial machine tool has been equipped with a university sensor system). Due to the limitations of systems in an industrial environment, the customisation of the systems used in the window trials is a further activity within this WP. The final sensor systems for each process material combination within the project are defined in a final deliverable of the task.
In this task, the signal analysis, feature extraction and development of algorithms is covered. Thereby, the results being elaborated in former project contributing to this field were reviewed e.g. the work done within the MANHIRP project undertaken in the Sixth Framework Programme (FP6) (in which some of the ACCENT partners were involved). A major output from the MANHIRP project was an increased understanding of how to reduce the probability of component failure caused by the presence of manufacturing induced anomalies. However, it also provided an insight into the strategies and algorithms used when attempting to process monitor the formation of the identified machining induced surface anomalies. Within the ACCENT project, the review of signal analysis, feature extraction and development of algorithms was changed from a short duration task into one that continued throughout the project. In this task, algorithms should be understood as a methodology or strategy on how to react on certain features in the monitoring signals. The link between features in the signals and the quality measures from the component often require true 'algorithms' such as mathematical operations e.g. transformation from a time to frequency domain, and in these cases they are included in 'feature extraction' activities. Early analysis of the generated data showed that different levels of sophistication were required dependent upon the process under investigation, the methodology of testing, and the applied process monitoring strategies used. Hence the analysis phase and algorithm development continued to run in parallel until the end of the project. A final subtask merges all the information gathered in the previous subtasks and summarises the algorithms which are used for the different processes. The outcome of the task is a deliverable.
Software programming is a task that continued throughout the project as software tools were required at different levels and at different stages. Firstly, tools to be programmed for the acquisition of process monitoring data at either the university or industrial partners' shop floors were required. This necessitated the availability of not only a process monitoring solution, but appropriate data acquisition software. Ideally, each of the groups working on different process and material combinations would have used the same solutions, but inevitably this was not possible. Hence a software tool was developed that was able to convert existing data files and provide the data with uniform header information for storage into the data base. In this way, partners were able to collect their data with whatever system they had available, but the collected data was available and exchangeable throughout the consortium. The header information also made it possible to search for data related to a specific partner, process, sensor type, parameter set.
Secondly, software tools had to be developed for proving the ACCENT closed loop. The concept is that the quality produced by the machining process is monitored against the defined process window that is proven to produce acceptable surface quality. The process window is therefore a mathematical multi-dimensional parameter 'room' with the acquired process monitoring sensor data functioning as the basis for correlation. In order to compute the values, however, features in the monitoring signals and the link to product quality or surface integrity had to be precisely understood. Even though process monitoring is meant to protect components in production, it has to be ensured that false alarms do not cause additional costs. For this reason, the extent to which a closed loop is connected to a production system has a significant impact on its effectiveness. The main question is how the process monitoring system being used for the university trials has to be customised when being transferred to an industrial situation? And then, how the related software is modified? A decision for this dilemma could be made on the basis of several factors such as:
- the quality of the sensor signals acquired (e.g. depending on machine tool, sensor system, production environment), hence difficult to control;
- the controllability of the input parameters (e.g. material, tool, variance of machining parameters);
- confidence in evaluating features in the signals (e.g. amplitude, curve shape);
- correlation of surface features with signal features (e.g. probability of detection of surface anomalies).
The extent to which the closed loop is applied is an important point of the project, not least because it determines the extent to which process monitoring is used in the production environment in a post ACCENT industrial application. Within the project, this task has not been performed for every process material combination, but was completed for the demonstrators at both the university lab stage and the industrial stage. A further sub-task covers the tools which have to be programmed for the tests on the sample components, again on both university lab and industrial stage. The related deliverable includes the details to the task.
The central action of WP3 is the development of an adaptive control strategy in order to realise the ACCENT idea. The task is divided into three subtasks. Within the first, the control strategy was developed on different levels: First of all in a very generic way being universally valid for all regarded ACCENT process material combinations. In later steps, the concept is more and more broken down and developed into specific solutions for the single cases as mentioned in the introduction. The control strategy is described in form of a control loop and shows where inputs from tasks and subtasks as well as outputs are directed. The control concept is discussed in the final deliverable in a very detailed way.
The second and third subtask were addressed in a combined activity and cover efforts to prove the control concept and further show, that boundary conditions for the established process windows are met. The boundary conditions are provided by WP2 ('Process window data') and WP4 ('Surface integrity data'). At the beginning of the project, it was planned that demonstrators would be located in the end-user production environment. Within the second project period however, it was decided within the consortium to accept industrial demonstrators at university laboratories, as well. During the project, it turned out that in some cases the further implementation into the production environment would have caused a far more extended project schedule. On the other hand, the industrial demonstrators at the universities being implemented on industrial machine tools rather than test beds show very similar results and the transfer to the end users will not give further scientific output. Besides the generic control loop concept which is applicable for every process material combination covered in ACCENT, the related deliverable shows the demonstrator cases on both a university laboratory level (which here means: fully industrial machine tool at a university facility with laboratory sensor equipment), as well as a demonstrator in a production environment of one of the industrial partners.
In a task dedicated to the evaluation of closed loop strategies for the real manufacturing situation, the development of strategies to react to process malfunctions was completed. In the former tasks, the control strategies were developed on a very scientific basis mainly by university partners. Within the test series, a wide range of monitoring equipment was used for data acquisition in order to get the best possible process data. When it comes to implementation in an industrial environment, many restrictions exist regarding additional equipment to the machine tools, systems interacting with the machine tool controller or the companies' production planning systems. The questions here are how deep may a control system interfere with the processes in production? Are there limits out of reasons such as disturbance of other processes, violations of any machine tool components, when additional sensor or control systems are introduced? All such restrictions for a potential ACCENT system beyond demonstration level are discussed with the Industrial partners in the consortium and collected in a document.
A further task covers the customisation of the ACS. The customisation of ACS to suit end-user trial conditions as discussed which within the project has its main focus on the demonstrators. As mentioned before, within ACCENT different demonstrators will be realised: on the one hand demonstrators on industrial machine tools (and industrial application) at the universities and on the other hand demonstrators in a production environment at the Industrial partners. For both kinds of demonstrators, the equipment used is different from the one used within the window trials. Furthermore, results being attained during the window trials with laboratory equipment have to be transferred to a commercial monitoring solution available at the Industrial partner or even further programmed to the machine tool controller itself. Since the demonstrator at university laboratory level could be used with the equipment employed during the process window trials, there was no need for further customisation. Only some adjustments had to be made to the software tools being used. In terms of the demonstrator on the industrial level at one of the original equipment manufacturer (OEM) partners, the entire control strategy had to be customised in order to have the control strategy working in combination with a commercial monitoring system. The related efforts are explained in a corresponding deliverable to the task. Finally, a test and verification of adaptive control strategy performance was made. This was mainly tested on the industrial demonstrator at one of the partners which was enabled for a milling / drilling process. In this case, the ACCENT control loop has been implemented on a real production machine tool. The system was designed to control the tool feed force over the cutting speed in order to have a higher productivity and be able to still operate in the process window in terms of the achieved quality and surface integrity, respectively. A related subtask covers the test and verification on sample components. Within the tests, the measures for a comparison to the former state-of-the-art process were the tool wear, the surface roughness as well as an optical analysis of the achieved residual surface layers. The related work is presented in a deliverable of the task.
A significant first task of WP4 was to complete the survey of surface integrity definitions from the Industrial partners. Thus, the design proven surface integrity specifications had to be documented and a definition of the desired surface integrity values had to be elaborated. This information is critical to the linkage between adaptive control signals and the defined component surface quality. However, these tasks proved to be more complex than expected, and in fact, it appears that the reason it was difficult to converge to a common surface integrity understanding was to be able to define:
- Which features and anomalies characterise the surface integrity?
- What kind of common inspection methodologies characterise each anomaly and allow definition of acceptable values?
- Which are the acceptable values for each feature that delimit the acceptable surface integrity windows.
Hence, a new subtask was added to the project that examined the methodologies used to categorise each anomaly. With the four manufacturing processes, and three typical materials under investigation, it was necessary to determine each type of feature and anomaly that can occur in each process and material combination. With reference to the work of FW6 MANHIRP project, in which anomalies were classified into two categories (geometrical and non-geometrical), a proposed solution in ACCENT was to define two new categories:
- Surface topography: features or anomalies on the manufactured surface.
- Microstructural changes: features or anomalies in the manufactured sub-surface.
During the first period, each feature / anomaly has been described with their corresponding non-destructive and destructive inspections, and a list of features / anomalies which may occur in each process / material combination has been issued. To complete this task, partners have defined which anomaly / feature had occurred during their trials and what kind of examination was required to detect them.
Another fundamental requirement of the work has been to establish the link between component surface integrity and the manufacturing dependencies that produce it, i.e. cutting conditions, tool wear. A methodology was to start from key features, e.g. roughness, and to correlate the feature to the cutting conditions. However when reviewing the task, the partners agreed that the original concept did not match the real world situation.
The following points were noted:
- Manufacturing dependencies can only be determined from trials using the defined parameters from the process windows and dependencies may not be found for all anomalies as the trials are not defined to provoke them. Anomalies that are produced will be as a consequence of the parameters used.
- Different post machining operations have been identified, but the effectiveness of post-machining operations (e.g. shot peening) largely depends on the surface condition produced by the machining process. Hence it is important to control the previous machining operation.
- Lifing policies will have an influence by adherence to currently applied quality standards.
For the development of a surface integrity part validation procedure, the definitions of surface integrity defined in a previous task led the partners to a methodology of surface integrity examination that was consistent for all trials. Consequently, a minimum requirement for examinations to be performed is proposed and tested.
Two steps of examination are distinguished:
- The minimum set of examinations (MSEs) that have to be performed to characterise acceptable surface integrity windows having the same observational scales.
- An additional set of examinations which has to be performed for some trials: The aim is to validate if the MSE is sufficient to characterise the surface integrity, and to complete the correlation between Phase modulation (PM) signals and surface integrity examinations.
A key task completed by the partners was to validate their process inputs in relation to their component quality requirements. For each process and material combination, acceptable surface integrity windows have been defined, and based on trials, a list of generated anomalies and their process parameter tendencies are known. Thus the dependencies between process inputs and the achieved surface integrity results are validated.
A survey on the data storage demands for surface integrity values and manufacturing process signals led to the definition of a new data storage requirement. Within the project framework, and consistent with the project aims, the task was to define what is logically possible to identify as the basic requirements for the data structure to be employed. This data structure was created on the requirements and outputs from WPs 3, 4. To develop the data management architecture for the surface integrity values and PM signals, the most important consideration was to build the infrastructure that is effective and available during the experimental process window machining trials. Different architectures were investigated with the aim of finding tools that directly focussed on the architecture and not the informatical development. In the early stages of the project, data was collected in line with the requirements defined in a data management deliverable, and sent to the WP4 manager for evaluation and to estimate the required resource storage requirement. The new data storage requirement has been described in a project deliverable. In order to provide a possible European standard based on the results generated in WP4, a task that collected and collated all the surface integrity results generated in the project was performed. The main objective of the task was to make guidelines for surface integrity analysis after machining based on the trials performed by the partners. The deliverable D4.7 constitutes a guideline based on a basic understanding of surface integrity anomalies extracted for the project deliverables and a critical review of the methodology used in the project.
The deliverable presents:
- a list of features and anomalies which define the surface integrity with their corresponding examinations to be characterised;
- a feedback on the methodology used for examination by all partners through their investigations of anomaly characterisation;
- a review of best practices for surface integrity analysis methodology.
Potential impact:
The ACCENT project has brought together the leading European aero engine manufacturers and European universities with the common aim to improve the competiveness of critical aero engine component manufacture, whilst at the same time ensuring that the safety and reliability of these components is improved verses current manufacturing methods.
As a direct consequence of the involvement in the ACCENT project, the European aero engine manufacturers have the opportunity to implement the improved process capability demonstrated in the project. Based on the machining process and component material combinations investigated, strategies for developing operating process windows, new process monitoring strategies, and novel adaptive control methods are now realised. The benefits, both in terms of reduced manufacturing costs and improved component quality control are already being achieved within some of the end user companies, whilst others are putting internal development programmes in place in order to implement ACCENT methodologies.
For the academic partners involved in the ACCENT project, opportunities to enter research into novel machining research has given greater awareness of their capabilities within the project consortium, and into the wider community with the publication of papers, journals, and representation at conferences and other events. This has also raised the awareness of the needs to control the manufacture of critical aero engine components to the wider manufacturing community and the general public.
The primary objective of ACCENT WP5 was to ensure that a framework exists whereby the consortiums industrial partners could directly exploit the results of the project, and that scientific research generated by the project could be exploited and disseminated by the consortium research organisations. Associated work was structured into four tasks:
Exploitation of project results by industrial partners
Based on the technical results from WPs 2 - 4, the task was defined to derive a thorough understanding of the issues associated with transferring ACCENT technology into an industrial environment, and to give the Industrial partners guidelines for implementing the project results into their production plants. Through this task, the industrial partners have used their ACCENT experience for supporting machine tool and process monitoring device investment as a short term exploitation activity. Most of them have already implemented individual process monitoring devices, or machines equipped with PM sensors and / or updated their procurement specification for future machine investment. Also, initial use of process monitoring systems has been implemented into industrial partner's production workshops. These implementations have been focused on critical rotating parts and are mostly applied for hole-making operations for tool wear control or special cause events detection, but implementation on broaching or milling has also commenced.
In addition, a longer term action plan has also been defined by the Industrial partners. It includes the application of process monitoring devices and associated data storage system implementation, the definition of new engineering specifications, training of personnel and shop floor implementation requirements. These activities have been identified as the most critical for industrial partners in order to be able to implement ACCENT overall control. In line with an aviation authority recommendation (AC FR33.70-2) some Industrial partners have started to implement machining process control with process monitoring use through updated or new engineering specifications, especially for hole-making processes.
Exploitation of academic research
This task was defined to promote the reputation of the university based partners, the ACCENT project as a whole, and the support of the European research framework. The technical expectations of the six academic partners were mainly focused on the development and expansion of technical knowledge in the research fields of process monitoring and machining processes related to the aero engine industry, expanding knowledge in the machining of difficult to machine materials, and developing new research on the correlation between process monitoring signals and component surface integrity characteristics. Due to their university status, the academic partners also considered the project an opportunity to confirm and improve the reputation of their respective universities, share and receive knowledge from other European laboratories, and to cooperate with the European aircraft aero engine manufacturers.
As part of this task, a total of nine doctorate (PhD) theses and nine Master theses, either partially or totally dedicated to ACCENT project, have been implemented by the academic partners. Researchers work has been focused on the ACCENT methodology, tool wear behaviour analysis, process monitoring strategies for machining, or the developed surface integrity obtained after machining. In addition, associated junior training and scientific dissemination has been performed by academic partners.
Exploitation and dissemination to the wider community
This task was defined to ensure that project results would be disseminated to the wider community and to evaluate possible exploitation opportunities in other Industrial sectors. A significant number of publications of ACCENT work have been published through European and international journals and conferences. These documents deal mainly with tool wear prediction methodologies, acceptable surface integrity machining domain definition, and process monitoring of machining operations to ensure acceptable surface integrity. Also, attendance and presentation of ACCENT work to national and international conferences has been undertaken by the consortium members. In taking the opportunity to organise such conferences, ACCENT dedicated sessions were included in the agenda of the following events events (CIRP ICME, Italy, 23 - 25 June 2010; INTERCUT Cluny - France, 14 October 2010). As a consequence, promotion of the reputation of individual partners and the support of the European research framework has been assured. A dedicated project website has been implemented: http://www.accent.wzl.rwth-aachen.de/en/default.html(öffnet in neuem Fenster)
Definition of an ACCENT standard
Based on the collation of technical WP output, the objective of this task was to evaluate the possibility of creating an ACCENT standard. The standard would describe generic requirements for the definition of process windows, methods describing process window validation, and descriptions of how the adaptive control of manufacturing processes with respect to surface integrity may be achieved. To give importance to the standard, the European Aviation Safety Agency (EASA) was contacted to see if they were able to support the issue of an ACCENT standard document. The project coordinators organised a dedicated meeting with EASA propulsion certification team and gave a detailed presentation of the ACCENT project. However, according to the current EASA position and strategy, the following statements have been concluded:
- EASA is not in a position to certify generic manufacturing sequences for critical aero engine components. Each application of a company's manufacturing strategy will be dealt with on a case by case basis.
- Results of the ACCENT project should be passed over to the propulsion certification group in order to strengthen the awareness of monitoring and control needs in the manufacture of aero engine components.
- EASA wishes to be involved in future activities concerning the dissemination of ACCENT results to a wider community and informed of future follow up activities in the field of adaptive machining in critical aero engine components.
Therefore, an ACCENT standard document has not been requested by EASA.
Project website: http://www.accent.wzl.rwth-aachen.de/en/default.html(öffnet in neuem Fenster)
