Final Report Summary - CONTEMP (Self-Learning Control of Tool Temperature in Cutting Processes)
The CONTEMP project was funded under the European Commission (EC)'s Seventh Framework Programme (FP7) Theme 4: Nanosciences, nanotechnologies, materials and new production technologies, NMP2008-3.2.2: Self Learning production systems, grant agreement: CP-FP 228585. The main objectives were the development of a self-learning platform for the analysis and estimation of process conditions and self-optimisation of different manufacturing tasks and the development of a new tool system that is easy to integrate into existing manufacturing systems incorporating a micro-structured high performance cooling device. The expected results are higher stability of the cutting process, longer tool life by minimised thermal shock, higher quality of the produced parts, reduction of sub-surface damage, less ecological impact by avoiding cooling lubricant and economical machining of new work piece materials.
High-temperature alloys and composites put the tools under high thermal and mechanical strain during the machining process. At the same time, manufacturing technology must put a strong emphasis on environmental issues and sustainability, which challenges conventional cooling methods. With a closed internal cooling circuit, contamination of the environment and the cooling fluid can be avoided. By using a self-learning adaptive process control, it is possible to stabilise the machining conditions and control the tool temperature. This will not only improve part accuracies and productivity but also enable the economical cutting of a larger range of materials by minimising tool wear.
The CONTEMP project focused on controlling and stabilising the tool temperature. To make an effective temperature control possible it was necessary to develop a self-learning platform that analyses the process as well as a cutting tool that enables the machining system to monitor and influence the temperature of the work piece and cutting tool to prevent part damage and tool wear. The control system considers its knowledge of the material's cutting behaviour and the geometrical and kinematical parameters of the process to estimate dynamic process conditions. The knowledge base is fed during the process by constantly monitoring the measured temperatures. By estimating the process conditions the system is especially well suited for small batch production where time-consuming optimisation procedures can be reduced by the self-optimising control.
The tool system is based on a novel closed circuit internal micro-cooling device that enables an effective temperature control of the tool as well as the measurement of temperatures. In conventional processes with cooling lubricants, the temperature difference between the hot chips and the cooling lubricant leads to wear on the cutting edge of the tool caused by micro-cracks. This thermal shock damage is avoided by the internally cooled system and thus tool life and part accuracy are increased. Due to different overlapping wear mechanisms, a wear minimum can be observed at a certain tool temperature. The combination of the internal cooling system as a sensor / actor system and a self-learning control allows the maximisation of tool life.
The CONTEMP system constitutes the development of a new generation of high performance intelligent and environmentally friendly tools for turning operations. The optimisation of machining parameters such as cutting speed permits the reduction of machining times and costs through a decrease of manufacturing times. With a closed internal coolant circuit, the cooling system is almost maintenance free and avoids external cooling lubrication. The prototype tool and cooling system developed in the project have demonstrated longer tool life and better surface qualities than state of the art coolant lubrication systems and tools.
Project context and objectives:
For the specification of the CONTEMP system, it was essential, as a first step, to determine the process conditions with which the system must be in a position to adequately cope with. The procedure to validate the CONTEMP system in the industrial environment was defined. Industrial demands regarding evaluation criteria of the system and legal aspects, such as industrial safety standards and machine security, were defined. The case studies, including work piece materials, cutting conditions and tool requirements were pre-selected by the end users DIAD and CRF. The academic partners - BU and TUB - measured the thermal and mechanical loads on the cutting tool during the machining of the different work piece materials defined by the end-users.
According to the process conditions and specifications analysed and stated in WP 1, the cooling process was modelled and simulated by computational fluid dynamics (CFD). Different design alternatives of the internal cooling structure were compared in order to achieve the best possible heat flow into the cooling liquid in task 2.1. The design process in task 2.2 was supported by the results from task 2.1. The finite element method (FEM) and CFD were also applied to optimise the configuration / layout and design for the integration of the cutting tool, cooling device and tool holder. It was vitally important to maintain the mechanical stability and dynamics of the cutting tool while achieving optimal heat dissipation from the tool tip. In task 2.3 an optimisation was undertaken on the basis of the experiments conducted with the first prototype tool. The objectives of task 2.1 and task 2.2 were fulfilled by the evaluation of the pre-production prototype in month 35.
The different designs of the cooling process were discussed and evaluated during the first period of the project. The most suitable designs were defined and taken as input to WP 3. The objectives of WP 3 include the design and manufacturing of the cooling periphery as well as the manufacturing of the prototype and pre-production tool system. The results of task 2.1 were used for the design of the cooling system. Thereafter, the cooling system was constructed and implemented in the turning machine. This configuration was used in task 5.2. where cutting tests to verify the cutting tools, the cooling system and the self-learning control system were undertaken. The cooling system prototype is described in deliverable 3.1B. A conclusion of WP 2 was that the CONTEMP tool shall consist of a tool holder and a cutting insert. Thus, in task 3.2 the development of the tool manufacturing process was divided into two parts. A laser-sintering process was adapted to manufacture the tool holder and a coating process adapted to coat the cutting inserts with diamond. During the process development, it transpired that the adhesion between the substrate and the diamond coating was very low. In order to establish the reason for this low coating adhesion, detailed technological investigations were undertaken. The key finding was that the substrate quality and the cemented carbide properties were not consistent. This was due to the fact that different material batches and thus different material compositions were used. This problem was solved by using a substrate material from one batch. Thus, the work programme of this subtask was completed within the expected timeframe. Task 3.2 was planned from month 11 until month 16. Unfortunately, the completion of the task was delayed by six months due to the additional investigations, the definition of an action plan and the execution of this plan. This plan was defined during an extraordinary meeting with all partners in Milan in July 2011 (Italy). It was initially planned to begin task 3.3 immediately following completion of task 3.2. A respective delay of task 3.3 by six months was deemed to be unacceptable by the consortium however and thus, task 3.3 was begun concurrent to task 3.2. As a result, task 3.3 was four months behind schedule and was completed by month 21. Task 3.3 constituted the manufacturing of the tool holder and the cutting insert. The design, as defined in task 2.2 was used. Due to the delays described above, task 5.2 was also behind schedule as the first prototype was needed in order to begin with the simplified laboratory cutting tests. The cutting tool and the tool holder are described in deliverable 3.3.1B.
The first 18 months of the project covered task 4.1 task 4.2 and task 4.3 which collectively define preparatory work and the development of a self-learning control system for enhancing the life of internally cooled tools utilising a tool temperature measurement. Such a control system has the benefits of:
(a) enabling the user to find the best machining conditions for their application;
(b) preventing the tool from experiencing the excessive temperatures that damage its material properties and hence produce dramatic reductions in life; and
(c) detecting the onset of tool wear.
Thus, a self-learning control system increases the overall performance of the complete CONTEMP tool and coolant system. As a first step for the development of the control system the requirements from task 1.2 were taken to build up an experimental set-up as part of task 4.1. This experimental system was constructed to collect first data for designing the adaptive software algorithm. In addition to this, a study was undertaken to find suitable sensor types for temperature measurement purposes. A study of suitable software platforms was performed at the beginning of task 4.2 to gain an overview of relevant software designs that could be used for the CONTEMP control system. The findings of this study and the information from task 1.2 and task 4.1 led to the view that the control system software should have an artificial neuronal network (ANN) structure based upon multilayer perceptions (MLP). After determining the ANN structure, the software code was programmed and tested in task 4.3. The ANN software was then trained in task 4.4 with data generated during task 5.2. As mentioned above, task 5.2 was subject to a delay of three months. As a consequence of this, task 4.4 was postponed by three months. The major technical objectives in WP 4 were reached. During the mid-term review meeting, it was pointed out that deliverable 4.1 should be strongly focused on the requirements and specification of the working structure. Therefore, deliverable 4.1 was updated by month 24, including a more comprehensive description of task 4.1 activities and the design information relevant to deliverable 4.3.
WP 5 was focused on the preparation of the industrial evaluation in task 5.1 (preliminary roadmaps and case studies definition) and on testing and evaluation of the cutting tool, cooling system and control system under laboratory conditions in task 5.2 (simplified laboratory cutting tests and optimisation). A pre-screening of relevant applications and a definition of case studies was undertaken and reported in deliveravle 5.1. In task 5.1 preliminary roadmaps for the case studies were defined in order to provide continuity of the laboratory cutting tests with the project demonstrator. The first prototype tools were tested in task 5.2. Due to the delays in task 3.3 the planned cutting tests were delayed by four months (tool manufacturing, first prototype). The results of these tests were analysed in task 5.3 (results analysis and selection of project demonstrators) and together with deliverable 5.1 provided input for WP 6 (industrial assessment, validation, and environmental implications).
The objectives of WP 6 (industrial assessment, validation, and environmental implications) was the assessment of the tool and control system and validation of the laboratory tests and industrial case studies. Also, the economic and environmental impact will be analysed. WP 6 was completed on schedule. The major outcome of task 6.1 was that the internally cooled tool shows a high potential to decrease the energy consumption of common cutting processes. It was determined that the energy demand can be decreased by factor of four.
In WP 7 (dissemination and exploitation), public awareness was focused on task 7.1 (dissemination). For this, a plan for use and dissemination of foreground was created and continuously updated during the project. It was possible to bring an awareness of the CONTEMP project to the tool producing industry. Further objectives were to define all exploitable results and their respective exploitation strategies in task 7.2 (exploitation) and to create an intellectual property rights (IPR) agreement that defines how to handle arising IPR issues following completion of the project (IPR). As output from task 7.1 a dissemination plan was established. In task 7.2 and task 7.3 an exploitation strategy seminar was held on 25 November 2010.
Project results:
The main functions of the cooling system and the cutting tool with the internal micro-cooling device is described below.
Mobile cooling system
The function of the mobile cooling system is, as mentioned previously, to supply the tool with cooling fluid, to analyse the measured temperature values and to adapt the cutting parameters of the machine tool according to the tool temperature. An overloading of the tool and thereby resulting early tool failure can thus be avoided. From a software point of view two optimisation options exist: one option offers the reduction of the energy consumed by adapting the cutting speed to the current tool wear state. The tool lifetime of the cutting insert can be significantly increased by implementing this method. The other optimisation option is an increase in productivity by maximising the volume of material machined. This productivity increase is realised by maximising the cooling performance of the mobile cooling system. The cutting speed can be increased if the cutting insert is substantially cooled during the cutting process.
The CONTEMP system was designed such that the control of temperature and volume flow rate is possible simultaneously. Current systems available on the market do not offer this feature. In order to realise this complex function, the mobile cooling system is divided into two separate cooling circuits.
Cooling circuits
One circuit cools the liquid to the designated temperature. The second circuit pumps the cooled medium to the tool and supplies the micro-cooling device with a controllable volume flow rate.
By using a membrane pump it is possible to regulate the liquid flow from 0 to 2.5 l/min. A safety valve is installed behind the outlet of the pumping head to protect the rubber membrane into the pumping head from overpressure. If the outlet pressure rises above a certain threshold the safety valve opens independently of the delivery rate of the pump. Additionally, a pressure sensor is installed between the pump outlet and the overpressure valve to measure the fluid pressure. Thus, leaks or blockages in pipes, valves or tool holders can be detected independently of the delivery rate of the pump.
Metal particles and impurities will circulate in the machine tool system whilst the cooling system is operating. All systems, especially the membrane, could be damaged by the particles. A fine mesh filter was thus installed between the tank and the pump inlet to filter the particles. The circulation cooling system is a closed system, in which 5 l of coolant is pumped in a closed circuit through two large heat exchangers. Both heat exchangers work in parallel to one another. The reason for this is that parallel-connected exchangers have lower pressure drops than series-connected exchangers. Because of this, the coolant-flow increases and the energy consumption of the cooling circuit pump decreases.
Both heat exchangers are installed in the coolant tank to cool the liquid from the supply circuit. The coolant can be cooled to temperatures below -10 degrees Celsius. The cooling output then reaches 150 W, which means that a heat source of 150 W can be cooled down to -10 degrees Celsius continuously. In practice this value decreases due to thermal losses, e.g. imperfect thermal insulation and the size of the heat exchanger limits the heat transmission to output temperatures of 0 degrees Celsius.
The software developed within the CONTEMP project calculates the required values for the cooler and the membrane pump. In order to feed this calculation with information, the software continuously gathers different process parameters from the machine tool, the cooler, the pump and the tool.
Data acquisition and control system
The following measurement data is collected by the CONTEMP software:
- pressure of the cooling liquid;
- volume flow of the pump;
- temperatures beneath the cutting insert tip;
- coolant temperatures at the inlet and outlet of the micro-cooling device;
- cutting speed, feed rate and depth of cut.
A pressure transducer is installed to measure the pump outlet pressure. The setup of the temperature acquisition system was arranged as with the pressure measurement setup. Type K thermocouples were connected to an I/O terminal. The terminals convert the analogue sensor signals to 16-bit digital signals and send them via ethernet to the industrial PC. The digital data is then processed by the Labview 2009b software. This software is used as a data recording and controlling tool.
The sampling rate is given by the different dead times of the thermocouples and the pressure transducer. The dead time for thermocouples is mainly given by their mass. Small thermocouples with small diameters have fast response times, whereas thermocouples with larger diameters react slowly to temperature changes. The CONTEMP system has inbuilt thermocouples with response times lower than 250 ms to react quickly to temperature changes. The flow of the coolant is measured by using the internal pump signal and the signal of the pressure transmitter.
Due to manufacturing tolerances, the sensor signals include imperfections. Measurement errors that are caused by these imperfections are usually determined by comparing the sensor signal with signals of a calibrated sensor. However, in doing so sensors signals can be equilibrated by the user. It is not necessary for the CONTEMP system to calibrate the sensor signals. The reason for this is that the measurement error consists of systematic and stochastic errors. Systematic errors can be eliminated by a comparison between calibrated and non-calibrated equipment. Due to the fact that it is not necessary to exchange the measurement equipment during the lifetime of the system, the systematic error will remain almost constant (drifts will occur due to aging of the sensors and components).
Cutting tool
Two crucial design criteria for an efficient cooling of the cutting area were validated during the project:
- the distance between the heat sink and the heat source should be minimal;
- the difference between the temperatures in the cutting zone and the temperature of the coolant should be as high as possible.
Tool holder
Both design aspects were considered in the development of the tool holder and the cutting insert. The tip of a customary tool holder was substituted and the inlet and outlet for the coolant and a channel for the temperature sensors were implemented in the new tip. The seat of the insert and the underlying cooling structure were manufactured subsequently. Apart from the realisation of a minimal distance between heat source and heat sink, the second design criterion requires a difference in temperature as high as possible between the cutting area and the coolant.
During the design process of the final CONTEMP tool, a number of different designs with direct cooling of the cutting insert were tested. The cutting performance of these tools was evaluated in machining tests. The tests showed that the geometrical orientation of the cooling channels has a significant impact on the tool lifetimes of the cutting inserts. The tool testing in an industrial environment led to further insights for the optimisation of tool handling and manufacturing. These insights were implemented in subsequent design stages.
The pre-production tool implements all the design aspects learned throughout the project and is the final CONTEMP tool design. This tool has, in contrast to previous models, a closed cooling circuit. The cooling fluid flows through a cooling device, which has replaced the open canal system. This change means that the cutting insert is indirectly cooled, meaning the cutting insert does not come into contact with the cooling fluid. This has the following advantages: a tool change while the cooling system is running is possible without any safety hazards and in the worst-case scenario of a tool breakage the cooling fluid does not enter the machine environment.
The pre-production tool has, besides the closed cooling circuit, three integrated temperature sensors. These temperature sensors make it possible to determine the temperature of the cooling fluid and the micro-cooling device at any time. The sensors were integrated into the tool holder to protect them from the machining environment and the sensors have shown to be damaged only in the case of a tool breakage.
A further goal of the consortium was to develop a tool whose tool holder geometry can be altered according to demand without a re-design process being necessary. This is for example necessary for a series production of the CONTEMP system: the machining of hardened steel will require a different geometry compared to the machining of soft alloys. In order to leave this option open for a possible series manufacturer, the tip of the tool is therefore individually manufactured using a laser sintering process. The cutting inserts and the micro-cooling device can be adapted to any tool holder geometry. The temperature sensors and the cooling fluid supply channels in the tool holder are standardised and can be implemented into any tool holder geometry.
Micro-cooling device
As previously described, the first prototypes implemented open cooling structures and these were shown to be highly effective in their cooling. Unfortunately, the open cooling structures also were also found to be user-unfriendly (due to safety hazards and tool change times). For this reason, the subsequent design implemented cooling in the form of a closed copper micro-cooling device. The cooling fluid thus circulates continuously in a closed circuit without any cooling fluid entering the environment.
The innovation of the micro-cooling device is that micro-channels were manufactured into a copper device made of a high-strength copper alloy. These channels were designed using computer simulations such that the fluid in the channels undergoes maximum turbulence whilst minimising flow resistance. The heat transfer for a given heat transfer area can thus be maximised. The computer simulations also allowed the design team to position the cooling channels as close as possible to the hot spots, whilst also taking the stability and density of the device into account.
Cutting inserts
The development of the cutting inserts was undertaken in parallel to the development of the tool holder. In order to achieve efficient cooling of the insert, it was necessary to move away from ISO standardised cutting inserts and re-design the insert according to the demands of the CONTEMP system.
As previously mentioned, one design goal was to keep the distance between the cutting zone, where the heat is generated, and the cooling medium as small as possible. In order to assess this, a conventional cutting insert was machined using a die sinking process and tested in machining trials. It was shown that this design has a very short tool lifetime and high production costs.
The consortium then tested the implementation of cooling structures in the cutting inserts. Cutting inserts with two different geometries were thus manufactured using an initial sintering process. The two parts were then fused together in a further sintering step. Fully functional cutting inserts with integrated cooling channels were thus manufactured. It was shown during the evaluation of these prototypes however, that a tool design with integrated cooling channels has negative implications on the manufacturing process. The required manufacturing tolerances and the adapted manufacturing processes increase the consumption of energy and raw materials. These concepts were thus not further investigated.
A significant design step was the reduction of the height of the cutting insert from over 4 mm to less than one mm. Cutting tests with steel, titanium and aluminium showed that such thin cutting inserts have adequate strength and tool lifetimes. A height of one mm was found to be the most advantageous in finding a compromise between stability and cooling capacity of the cutting edge. The distance between the cutting edge and cooling area is 1.1 mm in average. The maximum distance is 1.6 mm. The simple cutting insert design had significant implications for the design of the micro-cooling device and the tool holder.
In order to further reduce the temperatures in the cutting zone the cutting inserts were coated with a heat transfer coating. Diamond coatings were implemented due to their high thermal conductivity. Subsequent cutting tests showed that the coating has a positive influence on the lifetime of the cutting tool, due to the fact that the heat is spread away from the cutting edge effectively.
The performance of the thin cutting inserts was demonstrated in a large number of cutting trials. The tests showed that these cutting inserts had tool lifetimes equal to standard ISO cutting insert geometries when used under the same testing conditions. This was particularly unexpected in the roughing longitudinal turning tests, but demonstrated that the design of the CONTEMP system does not require thicker inserts.
The thin cutting inserts, when used in combination with the cooling system, demonstrated significant tool lifetime or productivity increases in comparison to standard insert geometries, due to the higher effectiveness of the cooling. The temperature at the cutting edge can be substantially reduced and the wear rate is closely related to the cutting tip temperature. At high temperatures, the work piece material softens and abrasive wear reduces, however thermally induced wear increases. At low cutting temperatures the dominating wear mechanism is opposite to that acting at high temperatures, i.e. abrasive and adhesive wear alternate their dominance. The CONTEMP system validated the findings of other researchers that a certain optimum temperature range exists for the machining of different materials in which the wear is minimised. This 'optimal' range often cannot be reached with a non-regulated internally cooled tool, or a non-cooled tool. The machine user must find this operating window through extensive testing. Through an automated and in the CONTEMP system integrated temperature regulation, the setting of optimal parameters, e.g. cutting speed and necessary cooling fluid supply becomes possible without such tests being necessary.
The CONTEMP software (UPM)
The software resides on a PC external to the machine tool's CNC controller. Communication with the machine tool and the tool's coolant system occur via a Beckhoff Ethercat™ interface and a Profibus™ interface. The parameters cutting speed, feed rate and depth of cut are transmitted by the CNC controller via Profibus to the PC whilst the tool temperature sensor signal and coolant flow rate control signal are transmitted via Beckhoff Ethercat terminals.
This system includes an ANN design algorithm that allows the user to design a range of multilayer perceptron networks that, once trained, can be used to monitor or control tool temperature. A data acquisition and editing function allows the user to create experimental data files for training. Training is performed off line before process monitoring or control is required. Note that monitoring, control, ANN design, training, data acquisition and editing are separate functions that cannot be run in parallel and are selected by the user.
The monitoring module
The monitoring algorithm is intended for use in situations where the coolant flow rate to the tool is fixed. The user selects the tool and material type which has the action of loading the appropriate ANN weighting factors and the permissible tool operating temperature threshold. The latter is factored to account for the difference in temperature between the cutting edge of the tool and that at the sensor position. Manually selecting the cutting conditions enables tool temperature to be predicted before the machining operation is started. During monitoring, the algorithm displays the measured tool temperature and the tool temperature predicted by the ANN in real time. The predicted temperature is based upon cut data that is received from the machine tool's CNC controller, updated at the chosen sampling interval whilst the measured temperature is a continuously sampled analogue signal. Both temperatures are continuously displayed along with two performance parameters - an instantaneous relative temperature and a wear index.
The control module
As for the monitoring module the user selects the tool and material type which has the action of loading the appropriate ANN weighting factors and temperature threshold. Manually selecting the cutting condition then enables tool temperature to be estimated and cutting conditions optimised before the machining operation is started. In this algorithm, the ANN is used to determine the set point control temperature from the known cutting conditions and a reference coolant flow rate. In general these will be changing in real time. The difference between set point and measured tool temperature (error) is used to determine the adjustment to coolant flow rate using a standard PID (proportional, integral, derivative) control algorithm. Differences between the controlled flow rate level and a reference flow rate are used to determine the state of tool wear. PID parameters are set manually by trial and error.
Data acquisition and editing
The purpose of this module is to allow experimental data files for training the ANN to be built offline. It comprises of three functions:
- acquire temperature and time data, in building an experimental data file it is first necessary to set a constant cutting condition and acquire tool temperature - time data for a time interval sufficient for it to reach a stable value. The data can be saved for subsequent inspection;
- view existing data, this function allows the temperature - time data of the preceding section to be recalled and displayed. The data is displayed in the form of a static oscilloscope trace. A high resolution X-Y cursor allows precise values of temperature and time to be determined;
- edit existing data file, this function allows data to be changed or added to in an existing experimental data file.
ANN design
The ANN is confined to the MLP type and is principally defined by the number of layers, the number of neurons in each layer and the activation function type.
However, an ANN defined by these parameters alone is insufficient to determine the values of the weighting factors essential for applying the ANN to a control system. For this, an experimental data file and the parameters that determine a training process also need to be defined. ANN design is closely linked to training in as much as there is little point in defining an ANN design without training it.
Training
All weighting factors are assigned random initial values that enable the ANN to be used to predict values for the experimental target data. For all data entries the difference between predicted and experimental values is stored and used to calculate an overall RMS error. If this is greater than the error required and providing the number of iterations is within a specified limit the weighting factors are adjusted using a back propagation criterion and the RMS error recalculated. Once convergence is achieved the weighting factors are saved in a data file.
Data files
When coolant is not present or is controlled, as for internally cooled tools, the relationship between tool temperature and the parameters depth of cut, feed / rev, cutting speed and coolant flow rate may be expected to vary with tool type and work piece material. It follows that to segregate the experimental training data in terms of tool and material type results in a minimum number of input variables for any given machining condition and hence a minimum training time as now for each tool and material combination selected there are only three input variables in the case of dry machining and four input variables for internally cooled tools. This is the structure used for the control system developed, however the implementation requires either three separate databases or a single database partitioned to account for:
- tool design and material type;
- experimental data for training the ANN: tool temperature versus cut depth, feed/rev, speed and flow rate for each category in 'a' above;
- for control and monitoring, weighting factors and a maximum permissible tool temperature threshold for each category.
System validation
In particular titanium, aluminium and steel alloys were used for the validation of the CONTEMP system. These materials cover a broad spectrum of those used in the manufacturing industry, e.g. in the automotive, aeronautical and motorsport industries.
At the beginning of the CONTEMP project reference processes were defined. It was thus possible to derive indicators, with which the different tool cooling systems could be characterised and compared, e.g. dry machining with the CONTEMP system, wet machining and machining with minimum-quantity lubrication. The consortium concentrated on the tool wear and the energy consumption of the system.
Tool wear
The tool wear emerged as the most important indicator for comparison purposes. The tool wear was analysed in the machining of different materials and the parameters cutting speed, feed rate and depth of cut have significant influences on the tool wear mechanism. As a measure of the tool wear, the maximum width of flank wear land was chosen. This was optically measured following machining trials. In the machining of steel and titanium alloys, the crater wear was also measured.
It was found during the testing phases of the CONTEMP project that the design of the cooling structures in the tool holder and the micro-cooling device is the most significant influencing factor for the performance of the tooling system. The first designs achieved tool lifetimes between those of dry machining and wet machining. With the subsequent design iteration, the tool wear could be reduced to the same level as that occurring during wet machining. The final iteration step within the project, the implementation of the copper micro-cooling device, allowed tool wear progression to be reduced further and also allowed a closed cooling fluid circuit to be implemented. This design achieved tool life increases of over 100 % in comparison to wet machining, in a temperature-controlled process.
The adaptive control adapts the tool cooling strategy and the cutting speed online to the tool temperature. The temperature of the tool, and as such the thermal load acting on the cutting insert, is measured and stored by the system. An algorithm, consisting of an artificial neural network, is fed with the process parameters, the tool temperatures and the cooling fluid parameters. As previously mentioned, if the temperature is low, the cutting speed and thus the productivity of the cutting process is increased. If the temperature is too high, the cooling system performance is increased and the cutting speed reduced, until an optimal temperature level is reached.
The testing of the CONTEMP tools showed that the temperature on the underside of the cutting insert can be brought to a constant level. A continuous increase of the tool temperature was measured in dry cutting processes, although the tool wear remained constant. The reason behind this is the heating up of the tool holder. The thermal energy released from the cutting process is led through the cutting insert into the tool holder and thus leads to this increased temperature. The speed at which the tool holder temperature increases is mainly dependent on the work piece material and the process parameters.
Cutting trials with the aluminum alloy AlSi7Cu4Mg included tests with and without cooling fluid. It was shown that it is possible to significantly reduce the tool wear progression when using the CONTEMP tool with internal cooling.
If wet machining is considered to be the reference, a decrease in flank wear of 25 % can be achieved by the internal cooling. If dry machining is used as the reference level, the wear can be reduced by 230 % with coolant temperatures at 20 degrees Celsius.
List of websites: http:// www.CONTEMP.org
Dipl.-Ing. Martin Roeder
Tel: +49-303-1423473
Fax: +49-303-1425895
E-Mail: roeder@iwf.tu-berlin.de
MSc. Dipl.-Ing. (FH) Paul Fürstmann
Tel: +49-303-1421791
Fax: +49-303-1425895
E-Mail: fuerstmann@iwf.tu-berlin.de
High-temperature alloys and composites put the tools under high thermal and mechanical strain during the machining process. At the same time, manufacturing technology must put a strong emphasis on environmental issues and sustainability, which challenges conventional cooling methods. With a closed internal cooling circuit, contamination of the environment and the cooling fluid can be avoided. By using a self-learning adaptive process control, it is possible to stabilise the machining conditions and control the tool temperature. This will not only improve part accuracies and productivity but also enable the economical cutting of a larger range of materials by minimising tool wear.
The CONTEMP project focused on controlling and stabilising the tool temperature. To make an effective temperature control possible it was necessary to develop a self-learning platform that analyses the process as well as a cutting tool that enables the machining system to monitor and influence the temperature of the work piece and cutting tool to prevent part damage and tool wear. The control system considers its knowledge of the material's cutting behaviour and the geometrical and kinematical parameters of the process to estimate dynamic process conditions. The knowledge base is fed during the process by constantly monitoring the measured temperatures. By estimating the process conditions the system is especially well suited for small batch production where time-consuming optimisation procedures can be reduced by the self-optimising control.
The tool system is based on a novel closed circuit internal micro-cooling device that enables an effective temperature control of the tool as well as the measurement of temperatures. In conventional processes with cooling lubricants, the temperature difference between the hot chips and the cooling lubricant leads to wear on the cutting edge of the tool caused by micro-cracks. This thermal shock damage is avoided by the internally cooled system and thus tool life and part accuracy are increased. Due to different overlapping wear mechanisms, a wear minimum can be observed at a certain tool temperature. The combination of the internal cooling system as a sensor / actor system and a self-learning control allows the maximisation of tool life.
The CONTEMP system constitutes the development of a new generation of high performance intelligent and environmentally friendly tools for turning operations. The optimisation of machining parameters such as cutting speed permits the reduction of machining times and costs through a decrease of manufacturing times. With a closed internal coolant circuit, the cooling system is almost maintenance free and avoids external cooling lubrication. The prototype tool and cooling system developed in the project have demonstrated longer tool life and better surface qualities than state of the art coolant lubrication systems and tools.
Project context and objectives:
For the specification of the CONTEMP system, it was essential, as a first step, to determine the process conditions with which the system must be in a position to adequately cope with. The procedure to validate the CONTEMP system in the industrial environment was defined. Industrial demands regarding evaluation criteria of the system and legal aspects, such as industrial safety standards and machine security, were defined. The case studies, including work piece materials, cutting conditions and tool requirements were pre-selected by the end users DIAD and CRF. The academic partners - BU and TUB - measured the thermal and mechanical loads on the cutting tool during the machining of the different work piece materials defined by the end-users.
According to the process conditions and specifications analysed and stated in WP 1, the cooling process was modelled and simulated by computational fluid dynamics (CFD). Different design alternatives of the internal cooling structure were compared in order to achieve the best possible heat flow into the cooling liquid in task 2.1. The design process in task 2.2 was supported by the results from task 2.1. The finite element method (FEM) and CFD were also applied to optimise the configuration / layout and design for the integration of the cutting tool, cooling device and tool holder. It was vitally important to maintain the mechanical stability and dynamics of the cutting tool while achieving optimal heat dissipation from the tool tip. In task 2.3 an optimisation was undertaken on the basis of the experiments conducted with the first prototype tool. The objectives of task 2.1 and task 2.2 were fulfilled by the evaluation of the pre-production prototype in month 35.
The different designs of the cooling process were discussed and evaluated during the first period of the project. The most suitable designs were defined and taken as input to WP 3. The objectives of WP 3 include the design and manufacturing of the cooling periphery as well as the manufacturing of the prototype and pre-production tool system. The results of task 2.1 were used for the design of the cooling system. Thereafter, the cooling system was constructed and implemented in the turning machine. This configuration was used in task 5.2. where cutting tests to verify the cutting tools, the cooling system and the self-learning control system were undertaken. The cooling system prototype is described in deliverable 3.1B. A conclusion of WP 2 was that the CONTEMP tool shall consist of a tool holder and a cutting insert. Thus, in task 3.2 the development of the tool manufacturing process was divided into two parts. A laser-sintering process was adapted to manufacture the tool holder and a coating process adapted to coat the cutting inserts with diamond. During the process development, it transpired that the adhesion between the substrate and the diamond coating was very low. In order to establish the reason for this low coating adhesion, detailed technological investigations were undertaken. The key finding was that the substrate quality and the cemented carbide properties were not consistent. This was due to the fact that different material batches and thus different material compositions were used. This problem was solved by using a substrate material from one batch. Thus, the work programme of this subtask was completed within the expected timeframe. Task 3.2 was planned from month 11 until month 16. Unfortunately, the completion of the task was delayed by six months due to the additional investigations, the definition of an action plan and the execution of this plan. This plan was defined during an extraordinary meeting with all partners in Milan in July 2011 (Italy). It was initially planned to begin task 3.3 immediately following completion of task 3.2. A respective delay of task 3.3 by six months was deemed to be unacceptable by the consortium however and thus, task 3.3 was begun concurrent to task 3.2. As a result, task 3.3 was four months behind schedule and was completed by month 21. Task 3.3 constituted the manufacturing of the tool holder and the cutting insert. The design, as defined in task 2.2 was used. Due to the delays described above, task 5.2 was also behind schedule as the first prototype was needed in order to begin with the simplified laboratory cutting tests. The cutting tool and the tool holder are described in deliverable 3.3.1B.
The first 18 months of the project covered task 4.1 task 4.2 and task 4.3 which collectively define preparatory work and the development of a self-learning control system for enhancing the life of internally cooled tools utilising a tool temperature measurement. Such a control system has the benefits of:
(a) enabling the user to find the best machining conditions for their application;
(b) preventing the tool from experiencing the excessive temperatures that damage its material properties and hence produce dramatic reductions in life; and
(c) detecting the onset of tool wear.
Thus, a self-learning control system increases the overall performance of the complete CONTEMP tool and coolant system. As a first step for the development of the control system the requirements from task 1.2 were taken to build up an experimental set-up as part of task 4.1. This experimental system was constructed to collect first data for designing the adaptive software algorithm. In addition to this, a study was undertaken to find suitable sensor types for temperature measurement purposes. A study of suitable software platforms was performed at the beginning of task 4.2 to gain an overview of relevant software designs that could be used for the CONTEMP control system. The findings of this study and the information from task 1.2 and task 4.1 led to the view that the control system software should have an artificial neuronal network (ANN) structure based upon multilayer perceptions (MLP). After determining the ANN structure, the software code was programmed and tested in task 4.3. The ANN software was then trained in task 4.4 with data generated during task 5.2. As mentioned above, task 5.2 was subject to a delay of three months. As a consequence of this, task 4.4 was postponed by three months. The major technical objectives in WP 4 were reached. During the mid-term review meeting, it was pointed out that deliverable 4.1 should be strongly focused on the requirements and specification of the working structure. Therefore, deliverable 4.1 was updated by month 24, including a more comprehensive description of task 4.1 activities and the design information relevant to deliverable 4.3.
WP 5 was focused on the preparation of the industrial evaluation in task 5.1 (preliminary roadmaps and case studies definition) and on testing and evaluation of the cutting tool, cooling system and control system under laboratory conditions in task 5.2 (simplified laboratory cutting tests and optimisation). A pre-screening of relevant applications and a definition of case studies was undertaken and reported in deliveravle 5.1. In task 5.1 preliminary roadmaps for the case studies were defined in order to provide continuity of the laboratory cutting tests with the project demonstrator. The first prototype tools were tested in task 5.2. Due to the delays in task 3.3 the planned cutting tests were delayed by four months (tool manufacturing, first prototype). The results of these tests were analysed in task 5.3 (results analysis and selection of project demonstrators) and together with deliverable 5.1 provided input for WP 6 (industrial assessment, validation, and environmental implications).
The objectives of WP 6 (industrial assessment, validation, and environmental implications) was the assessment of the tool and control system and validation of the laboratory tests and industrial case studies. Also, the economic and environmental impact will be analysed. WP 6 was completed on schedule. The major outcome of task 6.1 was that the internally cooled tool shows a high potential to decrease the energy consumption of common cutting processes. It was determined that the energy demand can be decreased by factor of four.
In WP 7 (dissemination and exploitation), public awareness was focused on task 7.1 (dissemination). For this, a plan for use and dissemination of foreground was created and continuously updated during the project. It was possible to bring an awareness of the CONTEMP project to the tool producing industry. Further objectives were to define all exploitable results and their respective exploitation strategies in task 7.2 (exploitation) and to create an intellectual property rights (IPR) agreement that defines how to handle arising IPR issues following completion of the project (IPR). As output from task 7.1 a dissemination plan was established. In task 7.2 and task 7.3 an exploitation strategy seminar was held on 25 November 2010.
Project results:
The main functions of the cooling system and the cutting tool with the internal micro-cooling device is described below.
Mobile cooling system
The function of the mobile cooling system is, as mentioned previously, to supply the tool with cooling fluid, to analyse the measured temperature values and to adapt the cutting parameters of the machine tool according to the tool temperature. An overloading of the tool and thereby resulting early tool failure can thus be avoided. From a software point of view two optimisation options exist: one option offers the reduction of the energy consumed by adapting the cutting speed to the current tool wear state. The tool lifetime of the cutting insert can be significantly increased by implementing this method. The other optimisation option is an increase in productivity by maximising the volume of material machined. This productivity increase is realised by maximising the cooling performance of the mobile cooling system. The cutting speed can be increased if the cutting insert is substantially cooled during the cutting process.
The CONTEMP system was designed such that the control of temperature and volume flow rate is possible simultaneously. Current systems available on the market do not offer this feature. In order to realise this complex function, the mobile cooling system is divided into two separate cooling circuits.
Cooling circuits
One circuit cools the liquid to the designated temperature. The second circuit pumps the cooled medium to the tool and supplies the micro-cooling device with a controllable volume flow rate.
By using a membrane pump it is possible to regulate the liquid flow from 0 to 2.5 l/min. A safety valve is installed behind the outlet of the pumping head to protect the rubber membrane into the pumping head from overpressure. If the outlet pressure rises above a certain threshold the safety valve opens independently of the delivery rate of the pump. Additionally, a pressure sensor is installed between the pump outlet and the overpressure valve to measure the fluid pressure. Thus, leaks or blockages in pipes, valves or tool holders can be detected independently of the delivery rate of the pump.
Metal particles and impurities will circulate in the machine tool system whilst the cooling system is operating. All systems, especially the membrane, could be damaged by the particles. A fine mesh filter was thus installed between the tank and the pump inlet to filter the particles. The circulation cooling system is a closed system, in which 5 l of coolant is pumped in a closed circuit through two large heat exchangers. Both heat exchangers work in parallel to one another. The reason for this is that parallel-connected exchangers have lower pressure drops than series-connected exchangers. Because of this, the coolant-flow increases and the energy consumption of the cooling circuit pump decreases.
Both heat exchangers are installed in the coolant tank to cool the liquid from the supply circuit. The coolant can be cooled to temperatures below -10 degrees Celsius. The cooling output then reaches 150 W, which means that a heat source of 150 W can be cooled down to -10 degrees Celsius continuously. In practice this value decreases due to thermal losses, e.g. imperfect thermal insulation and the size of the heat exchanger limits the heat transmission to output temperatures of 0 degrees Celsius.
The software developed within the CONTEMP project calculates the required values for the cooler and the membrane pump. In order to feed this calculation with information, the software continuously gathers different process parameters from the machine tool, the cooler, the pump and the tool.
Data acquisition and control system
The following measurement data is collected by the CONTEMP software:
- pressure of the cooling liquid;
- volume flow of the pump;
- temperatures beneath the cutting insert tip;
- coolant temperatures at the inlet and outlet of the micro-cooling device;
- cutting speed, feed rate and depth of cut.
A pressure transducer is installed to measure the pump outlet pressure. The setup of the temperature acquisition system was arranged as with the pressure measurement setup. Type K thermocouples were connected to an I/O terminal. The terminals convert the analogue sensor signals to 16-bit digital signals and send them via ethernet to the industrial PC. The digital data is then processed by the Labview 2009b software. This software is used as a data recording and controlling tool.
The sampling rate is given by the different dead times of the thermocouples and the pressure transducer. The dead time for thermocouples is mainly given by their mass. Small thermocouples with small diameters have fast response times, whereas thermocouples with larger diameters react slowly to temperature changes. The CONTEMP system has inbuilt thermocouples with response times lower than 250 ms to react quickly to temperature changes. The flow of the coolant is measured by using the internal pump signal and the signal of the pressure transmitter.
Due to manufacturing tolerances, the sensor signals include imperfections. Measurement errors that are caused by these imperfections are usually determined by comparing the sensor signal with signals of a calibrated sensor. However, in doing so sensors signals can be equilibrated by the user. It is not necessary for the CONTEMP system to calibrate the sensor signals. The reason for this is that the measurement error consists of systematic and stochastic errors. Systematic errors can be eliminated by a comparison between calibrated and non-calibrated equipment. Due to the fact that it is not necessary to exchange the measurement equipment during the lifetime of the system, the systematic error will remain almost constant (drifts will occur due to aging of the sensors and components).
Cutting tool
Two crucial design criteria for an efficient cooling of the cutting area were validated during the project:
- the distance between the heat sink and the heat source should be minimal;
- the difference between the temperatures in the cutting zone and the temperature of the coolant should be as high as possible.
Tool holder
Both design aspects were considered in the development of the tool holder and the cutting insert. The tip of a customary tool holder was substituted and the inlet and outlet for the coolant and a channel for the temperature sensors were implemented in the new tip. The seat of the insert and the underlying cooling structure were manufactured subsequently. Apart from the realisation of a minimal distance between heat source and heat sink, the second design criterion requires a difference in temperature as high as possible between the cutting area and the coolant.
During the design process of the final CONTEMP tool, a number of different designs with direct cooling of the cutting insert were tested. The cutting performance of these tools was evaluated in machining tests. The tests showed that the geometrical orientation of the cooling channels has a significant impact on the tool lifetimes of the cutting inserts. The tool testing in an industrial environment led to further insights for the optimisation of tool handling and manufacturing. These insights were implemented in subsequent design stages.
The pre-production tool implements all the design aspects learned throughout the project and is the final CONTEMP tool design. This tool has, in contrast to previous models, a closed cooling circuit. The cooling fluid flows through a cooling device, which has replaced the open canal system. This change means that the cutting insert is indirectly cooled, meaning the cutting insert does not come into contact with the cooling fluid. This has the following advantages: a tool change while the cooling system is running is possible without any safety hazards and in the worst-case scenario of a tool breakage the cooling fluid does not enter the machine environment.
The pre-production tool has, besides the closed cooling circuit, three integrated temperature sensors. These temperature sensors make it possible to determine the temperature of the cooling fluid and the micro-cooling device at any time. The sensors were integrated into the tool holder to protect them from the machining environment and the sensors have shown to be damaged only in the case of a tool breakage.
A further goal of the consortium was to develop a tool whose tool holder geometry can be altered according to demand without a re-design process being necessary. This is for example necessary for a series production of the CONTEMP system: the machining of hardened steel will require a different geometry compared to the machining of soft alloys. In order to leave this option open for a possible series manufacturer, the tip of the tool is therefore individually manufactured using a laser sintering process. The cutting inserts and the micro-cooling device can be adapted to any tool holder geometry. The temperature sensors and the cooling fluid supply channels in the tool holder are standardised and can be implemented into any tool holder geometry.
Micro-cooling device
As previously described, the first prototypes implemented open cooling structures and these were shown to be highly effective in their cooling. Unfortunately, the open cooling structures also were also found to be user-unfriendly (due to safety hazards and tool change times). For this reason, the subsequent design implemented cooling in the form of a closed copper micro-cooling device. The cooling fluid thus circulates continuously in a closed circuit without any cooling fluid entering the environment.
The innovation of the micro-cooling device is that micro-channels were manufactured into a copper device made of a high-strength copper alloy. These channels were designed using computer simulations such that the fluid in the channels undergoes maximum turbulence whilst minimising flow resistance. The heat transfer for a given heat transfer area can thus be maximised. The computer simulations also allowed the design team to position the cooling channels as close as possible to the hot spots, whilst also taking the stability and density of the device into account.
Cutting inserts
The development of the cutting inserts was undertaken in parallel to the development of the tool holder. In order to achieve efficient cooling of the insert, it was necessary to move away from ISO standardised cutting inserts and re-design the insert according to the demands of the CONTEMP system.
As previously mentioned, one design goal was to keep the distance between the cutting zone, where the heat is generated, and the cooling medium as small as possible. In order to assess this, a conventional cutting insert was machined using a die sinking process and tested in machining trials. It was shown that this design has a very short tool lifetime and high production costs.
The consortium then tested the implementation of cooling structures in the cutting inserts. Cutting inserts with two different geometries were thus manufactured using an initial sintering process. The two parts were then fused together in a further sintering step. Fully functional cutting inserts with integrated cooling channels were thus manufactured. It was shown during the evaluation of these prototypes however, that a tool design with integrated cooling channels has negative implications on the manufacturing process. The required manufacturing tolerances and the adapted manufacturing processes increase the consumption of energy and raw materials. These concepts were thus not further investigated.
A significant design step was the reduction of the height of the cutting insert from over 4 mm to less than one mm. Cutting tests with steel, titanium and aluminium showed that such thin cutting inserts have adequate strength and tool lifetimes. A height of one mm was found to be the most advantageous in finding a compromise between stability and cooling capacity of the cutting edge. The distance between the cutting edge and cooling area is 1.1 mm in average. The maximum distance is 1.6 mm. The simple cutting insert design had significant implications for the design of the micro-cooling device and the tool holder.
In order to further reduce the temperatures in the cutting zone the cutting inserts were coated with a heat transfer coating. Diamond coatings were implemented due to their high thermal conductivity. Subsequent cutting tests showed that the coating has a positive influence on the lifetime of the cutting tool, due to the fact that the heat is spread away from the cutting edge effectively.
The performance of the thin cutting inserts was demonstrated in a large number of cutting trials. The tests showed that these cutting inserts had tool lifetimes equal to standard ISO cutting insert geometries when used under the same testing conditions. This was particularly unexpected in the roughing longitudinal turning tests, but demonstrated that the design of the CONTEMP system does not require thicker inserts.
The thin cutting inserts, when used in combination with the cooling system, demonstrated significant tool lifetime or productivity increases in comparison to standard insert geometries, due to the higher effectiveness of the cooling. The temperature at the cutting edge can be substantially reduced and the wear rate is closely related to the cutting tip temperature. At high temperatures, the work piece material softens and abrasive wear reduces, however thermally induced wear increases. At low cutting temperatures the dominating wear mechanism is opposite to that acting at high temperatures, i.e. abrasive and adhesive wear alternate their dominance. The CONTEMP system validated the findings of other researchers that a certain optimum temperature range exists for the machining of different materials in which the wear is minimised. This 'optimal' range often cannot be reached with a non-regulated internally cooled tool, or a non-cooled tool. The machine user must find this operating window through extensive testing. Through an automated and in the CONTEMP system integrated temperature regulation, the setting of optimal parameters, e.g. cutting speed and necessary cooling fluid supply becomes possible without such tests being necessary.
The CONTEMP software (UPM)
The software resides on a PC external to the machine tool's CNC controller. Communication with the machine tool and the tool's coolant system occur via a Beckhoff Ethercat™ interface and a Profibus™ interface. The parameters cutting speed, feed rate and depth of cut are transmitted by the CNC controller via Profibus to the PC whilst the tool temperature sensor signal and coolant flow rate control signal are transmitted via Beckhoff Ethercat terminals.
This system includes an ANN design algorithm that allows the user to design a range of multilayer perceptron networks that, once trained, can be used to monitor or control tool temperature. A data acquisition and editing function allows the user to create experimental data files for training. Training is performed off line before process monitoring or control is required. Note that monitoring, control, ANN design, training, data acquisition and editing are separate functions that cannot be run in parallel and are selected by the user.
The monitoring module
The monitoring algorithm is intended for use in situations where the coolant flow rate to the tool is fixed. The user selects the tool and material type which has the action of loading the appropriate ANN weighting factors and the permissible tool operating temperature threshold. The latter is factored to account for the difference in temperature between the cutting edge of the tool and that at the sensor position. Manually selecting the cutting conditions enables tool temperature to be predicted before the machining operation is started. During monitoring, the algorithm displays the measured tool temperature and the tool temperature predicted by the ANN in real time. The predicted temperature is based upon cut data that is received from the machine tool's CNC controller, updated at the chosen sampling interval whilst the measured temperature is a continuously sampled analogue signal. Both temperatures are continuously displayed along with two performance parameters - an instantaneous relative temperature and a wear index.
The control module
As for the monitoring module the user selects the tool and material type which has the action of loading the appropriate ANN weighting factors and temperature threshold. Manually selecting the cutting condition then enables tool temperature to be estimated and cutting conditions optimised before the machining operation is started. In this algorithm, the ANN is used to determine the set point control temperature from the known cutting conditions and a reference coolant flow rate. In general these will be changing in real time. The difference between set point and measured tool temperature (error) is used to determine the adjustment to coolant flow rate using a standard PID (proportional, integral, derivative) control algorithm. Differences between the controlled flow rate level and a reference flow rate are used to determine the state of tool wear. PID parameters are set manually by trial and error.
Data acquisition and editing
The purpose of this module is to allow experimental data files for training the ANN to be built offline. It comprises of three functions:
- acquire temperature and time data, in building an experimental data file it is first necessary to set a constant cutting condition and acquire tool temperature - time data for a time interval sufficient for it to reach a stable value. The data can be saved for subsequent inspection;
- view existing data, this function allows the temperature - time data of the preceding section to be recalled and displayed. The data is displayed in the form of a static oscilloscope trace. A high resolution X-Y cursor allows precise values of temperature and time to be determined;
- edit existing data file, this function allows data to be changed or added to in an existing experimental data file.
ANN design
The ANN is confined to the MLP type and is principally defined by the number of layers, the number of neurons in each layer and the activation function type.
However, an ANN defined by these parameters alone is insufficient to determine the values of the weighting factors essential for applying the ANN to a control system. For this, an experimental data file and the parameters that determine a training process also need to be defined. ANN design is closely linked to training in as much as there is little point in defining an ANN design without training it.
Training
All weighting factors are assigned random initial values that enable the ANN to be used to predict values for the experimental target data. For all data entries the difference between predicted and experimental values is stored and used to calculate an overall RMS error. If this is greater than the error required and providing the number of iterations is within a specified limit the weighting factors are adjusted using a back propagation criterion and the RMS error recalculated. Once convergence is achieved the weighting factors are saved in a data file.
Data files
When coolant is not present or is controlled, as for internally cooled tools, the relationship between tool temperature and the parameters depth of cut, feed / rev, cutting speed and coolant flow rate may be expected to vary with tool type and work piece material. It follows that to segregate the experimental training data in terms of tool and material type results in a minimum number of input variables for any given machining condition and hence a minimum training time as now for each tool and material combination selected there are only three input variables in the case of dry machining and four input variables for internally cooled tools. This is the structure used for the control system developed, however the implementation requires either three separate databases or a single database partitioned to account for:
- tool design and material type;
- experimental data for training the ANN: tool temperature versus cut depth, feed/rev, speed and flow rate for each category in 'a' above;
- for control and monitoring, weighting factors and a maximum permissible tool temperature threshold for each category.
System validation
In particular titanium, aluminium and steel alloys were used for the validation of the CONTEMP system. These materials cover a broad spectrum of those used in the manufacturing industry, e.g. in the automotive, aeronautical and motorsport industries.
At the beginning of the CONTEMP project reference processes were defined. It was thus possible to derive indicators, with which the different tool cooling systems could be characterised and compared, e.g. dry machining with the CONTEMP system, wet machining and machining with minimum-quantity lubrication. The consortium concentrated on the tool wear and the energy consumption of the system.
Tool wear
The tool wear emerged as the most important indicator for comparison purposes. The tool wear was analysed in the machining of different materials and the parameters cutting speed, feed rate and depth of cut have significant influences on the tool wear mechanism. As a measure of the tool wear, the maximum width of flank wear land was chosen. This was optically measured following machining trials. In the machining of steel and titanium alloys, the crater wear was also measured.
It was found during the testing phases of the CONTEMP project that the design of the cooling structures in the tool holder and the micro-cooling device is the most significant influencing factor for the performance of the tooling system. The first designs achieved tool lifetimes between those of dry machining and wet machining. With the subsequent design iteration, the tool wear could be reduced to the same level as that occurring during wet machining. The final iteration step within the project, the implementation of the copper micro-cooling device, allowed tool wear progression to be reduced further and also allowed a closed cooling fluid circuit to be implemented. This design achieved tool life increases of over 100 % in comparison to wet machining, in a temperature-controlled process.
The adaptive control adapts the tool cooling strategy and the cutting speed online to the tool temperature. The temperature of the tool, and as such the thermal load acting on the cutting insert, is measured and stored by the system. An algorithm, consisting of an artificial neural network, is fed with the process parameters, the tool temperatures and the cooling fluid parameters. As previously mentioned, if the temperature is low, the cutting speed and thus the productivity of the cutting process is increased. If the temperature is too high, the cooling system performance is increased and the cutting speed reduced, until an optimal temperature level is reached.
The testing of the CONTEMP tools showed that the temperature on the underside of the cutting insert can be brought to a constant level. A continuous increase of the tool temperature was measured in dry cutting processes, although the tool wear remained constant. The reason behind this is the heating up of the tool holder. The thermal energy released from the cutting process is led through the cutting insert into the tool holder and thus leads to this increased temperature. The speed at which the tool holder temperature increases is mainly dependent on the work piece material and the process parameters.
Cutting trials with the aluminum alloy AlSi7Cu4Mg included tests with and without cooling fluid. It was shown that it is possible to significantly reduce the tool wear progression when using the CONTEMP tool with internal cooling.
If wet machining is considered to be the reference, a decrease in flank wear of 25 % can be achieved by the internal cooling. If dry machining is used as the reference level, the wear can be reduced by 230 % with coolant temperatures at 20 degrees Celsius.
List of websites: http:// www.CONTEMP.org
Dipl.-Ing. Martin Roeder
Tel: +49-303-1423473
Fax: +49-303-1425895
E-Mail: roeder@iwf.tu-berlin.de
MSc. Dipl.-Ing. (FH) Paul Fürstmann
Tel: +49-303-1421791
Fax: +49-303-1425895
E-Mail: fuerstmann@iwf.tu-berlin.de