Final Report Summary - ADMAP-GAS (Unconventional (Advanced) Manufacturing Processes for Gas-Engine Turbine Components)
Executive Summary:
The European project ADMAP-GAS succeeded in presenting alternatives to the critical broaching process for the manufacturing of fir tree structures in gas turbine discs by the alternative processes of Abrasive Water Jet Cutting (AWJC) and High-Speed Wire-EDM (HS-WEDM). Eight partners effectually combined their efforts to develop robust manufacturing processes for the European engine manufacturing industry with the objective to drastically decrease failure risks and machining costs as well as preserve energy and environment through the project.
Until the last project month research and development work was carried out and successfully finished. Both developed manufacturing processes (AWJC and HS-WEDM) are able to enter into competition with the broaching process. This showed the evaluation process which was done during the last project months. It can be stated, that both manufacturing technologies underwent an improvement through the project. Especially cutting rates, surface integrity aspects and process monitoring systems have been increased or even completely new developed. The developed AWJC process for cutting fir tree structures was proven to be a manufacturing process for roughing operations which requires a very economic technology to process hard to machine alloys. Through a process monitoring system, a multi axis cutting head, increased accuracy and a new developed nozzle geometry (all objects have been developed within ADMAP-GAS) the manufacturing technology is on a very good way to be able to machine fir tree geometry profiles. With further effort and development this technology will be an alternative to broaching.
In terms of HS-WEDM several developments improved the standard manufacturing process to a competitive manufacturing technology to the state of the art broaching process. Improved cutting rates which have been achieved through technology optimisation, nozzle development and generator tuning together with very good surface integrities (minimal surface layer and contamination as well as no crack occurrences) indicates that the developed process is a serious alternative to broaching. On the one hand this was shown regarding economic aspects and on the other hand this was shown regarding functionality through fatigue tests.
To summarize the technical work done within ADMAP-GAS and to critically assess the project regarding the objectives of the complete project it can be stated, that the failure risks of engine discs have been decreased through the developed process monitoring systems. Additionally, the machining costs to produce a fir tree with the challenging requirements have been decreased by 25%. This was shown through a Life Cycle Cost analysis. Concerning the energy consumption and the environmental impact a Life Cycle Assessment showed an enormous decrease (up to 60%) by using one of the two developed processes. Furthermore it was shown, that the developed processes are able to be implemented into other industries like the motorsport sector.
Regarding the work concerning public relation dissemination and exploitation activities have been done. For the future a dissemination and exploitation plan has been set up to spread the very good project results after the project end. During the project end phase several turbine manufacturers like MTU, Rolls Royce and Pratt & Whitney contacted the consortium to receive detailed project results. All of them are very interested in the research which has been done during the project and they asked the partners for cooperation to evaluate the processes with their own standards.
Overall it can be summarized, that the project ADMAP-GAS has come to a very good end. Two special manufacturing processes to machine fir tree geometry profiles have been developed which are able to substitute the critical broaching process in the future.
Project Context and Objectives:
In the project ADMAP-GAS new “Unconventional Advanced Manufacturing Processes for Gas Engine Turbine Components” have been developed in order to substitute the critical conventional broaching process for fir tree profiles between rotating blades and discs.
Presentation of the project and consortium
The ADMAP-GAS project is a European collaborative project of the 7th Framework Programme within Theme 7, Transport (including Aeronautics). It is a three and a half year project (36 month + 6 month cost neutral prolongation) that started on August 2009 and will last until January 2013. The scientific and technological methodology applied in ADMAP-GAS is kept at the highest level due to the combined talents of the consortium partners (see Figure 1). Two world leading industrial partners (CHARMILLES and BEDRA), three SMEs (OELHELD, TEKS and DIAD), one research institution (AMRC) and two outstanding higher education partners (WZL-RWTH Aachen and University of Sheffield) from five European countries form the consortium.
Figure 1: Consortium of ADMAP-GAS
In order to provide examples of issues that ADMAP-GAS project is able to address, the basic information and problem allocation were defined by the industrial partners in correlation to their daily production needs and challenges. The proposed new unconventional processes in particular for ADMAP-GAS Wire-EDM and AWJM for the substitution of broaching will not only have a benefit for the turbine industry but also have a significant impact on other transport industries which need special manufacturing processes like the car sector.
Project objectives
Due to a high increase of market demand for gas engine turbines in the aircraft industry, efficient manufacturing processes for the turbine components are increasingly necessary. The rising standard for individual and adapted product solutions leads to an increased demand for efficient, reliable and additional flexible manufacturing technologies according to the turbine parts.
The ADMAP-GAS project develops new “Unconventional Advanced Manufacturing Processes for Gas-Engine Turbine Components” in order to substitute the critical conventional broaching process for fir tree profiles that connect rotating blades and discs. Especially the good thermal and mechanical properties of nickel-based super alloys which are used for disc material complicate the machining by conventional processes.
Therefore, the two manufacturing technologies “High Speed Wire-EDM Slotting” and “Abrasive Water Jet Slotting” (ADMAP-GAS Processes) have been developed, optimised and evaluated, highlighting their individual performance and effectiveness for the machining task.
Based on the achieved results ADMAP-GAS will further explore its possible impact on the manufacturing of blisk geometries. Here a large amount of expenses results from high tool costs due to high tool wear. Therefore in this project the process capabilities of fast roughing of the blisk structures with unconventional machining processes was also investigated to reduce the volume of material which has later to be machined by milling.
Project methodology and implementation of the work
In order to accomplish the project objectives and goals, an appropriate structure has been established. Figure 2 shows the holistic build-up of the work package structure, wherein all partners have their defined tasks. All work packages are surrounded by work package one the Project Management (WP1, MGT). WP1 is in this project essential in order to ensure the adequate communication between project partners and the correct check of the project’s progress. Furthermore in WP1 the project co-ordinator is positioned who is responsible for the communication with the European Commission (EC). The main work on Research and Technical Development (RTD) is done inside the work packages with the titles Wire-EDM Process Development (WP2) and Water-Jet Process Development (WP3). The main tasks herein address the research to achieve the specific objectives of the two unconventional processes. The special turbine manufacturing requirements which can be seen as the standards to be reached by the core project work (WP2+WP3) are gathered within work package 4 (WP4). Furthermore WP4 has the function to design a technique to integrate the new processes into the existing process chain of disc manufacturing. This action includes the demonstration (DEM) of the developed ADMAP-GAS demonstrator set-ups which show the developments and improvements achieved in the project to the industrial target group. Work package five (WP5) is responsible to incorporate the exploitation and dissemination work of the consortium. Furthermore the construction of a web-based internet platform is the content of WP5.
Figure 2: ADMAP-GAS work package structure
ADMAP-GAS periods and tasks
ADMAP-GAS is a 36 + 6 month project and is divided into two periods. The tasks of each period are as following:
• First period (month 1 - 18): Analysis of current broaching operation and development of special unconventional processes for fir tree production.
• Second period (month 19 – 36 + 6): Optimisation and monitoring of developed processes and evaluation of unconventional processes regarding fir tree production.
Project Results:
The main scientific and technical results respectively foregrounds of ADMAP-GAS were gained within the research and development work of the two processes AWJC and HS-WEDM. This work was mainly done in the work package 2 and 3. Following the work done is summarized.
WP 2: Development of “High Speed Wire-EDM Slotting”
WP 2 mainly focusses on the development of Wire-EDM process for fir tree production. Charmilles as a machine tool manufacturer, Bedra as a wire manufacturer, oelheld as a supplier of dielectrics and the WZL as a research organisation are forming the main partners of this WP dealing with process technology development. In addition DIAD and TEKS provide information from the turbine industries in terms of dimensional and surface integrity requirements and evaluation criteria to compare the “High Speed Wire-EDM Slotting” against standard (broaching) machining operations. The following major objectives have been set up before the start of the project:
• Increase the cutting rate to reduce machining time by 30% compared to state-of-the-art Wire-EDM.
• Maintain a sufficient parallelism in the machined geometry of the parts at targeted cutting rates.
• Minimise heat affected zone, so that the demands of the cut parts regarding fatigue strength are met.
• Achieve a surface roughness smaller than Ra = 0.8 µm.
• Develop wire to allow reliable threading and workpiece touch-off.
The main research task within WP2 is the Task 2.2 “Development of High Speed Wire-EDM”. To achieve the overall WP objectives these subtasks are dealing with developments of machine tools, wire electrodes, dielectrics and process monitoring systems.
Development of new wires
The main procedure determined for the development of new wire types is depicted in Figure 3. Prior to wire design itself the detailed performance targets and requirements had to be rendered. This included a basic alignment with the design or use of test machines and dielectrics. In addition procedures for testing the EDM wires needed to be defined. Wire design followed two main routes according to two different material groups, nickel and titanium based alloys, considered in the project. Main focus, however, was dedicated to nickel based alloys (Inconel 718).
Figure 3: Wire development process
In a first step a preliminary wire specification was derived from the detailed process conditions and requirements. In Table 1 the specifications and requirements are listed.
New wire design to avoid undesired contamination
Strategies for avoidance or minimisation of contamination of the wire cut surface were developed (see Figure 4). Contamination by wire material is a critical issue since the deposition of zinc can lead to the formation of brittle phases in the Inconel material which can dramatically downgrade the material performance. The selected strategies are marked in grey. In general the approach combines a wire with an outermost nickel coating and a cutting technology adapted to the new wire design. The use of oil-based dielectrics may also support the minimisation of contamination. This has not been tested so far.
Table 1: Wire specification derived from process conditions and requirements
Figure 4: Strategies for avoidance of undesired work piece contamination
In order to achieve both, high productivity and avoidance of contamination the new wire design features a first layer of zinc rich brass and an outermost layer of pure nickel (see Figure 5). The function of this layer is to generate only such contamination which is already a parent material of Inconel. Since nickel is rather unsuitable for WEDM in terms of cutting speed the challenge is to keep the thickness of the outermost layer as small as possible. Here the interaction of thickness and cutting technology i.e. number of cuts and machining parameters is crucial. For the layer providing high speed two main alternatives beta-brass and gamma-brass were chosen. Usually, the thickness of gamma-brass layers can be significantly thinner compared to beta brass layers while achieving the similar cutting speeds. Thus a gamma phase layers may provide a more economic solution. Core material for type AG-N2 is brass due to better availability of the pre-material.
Besides the wire design of type AG-N3 which was chosen with focus on Charmilles machines two further designs were developed to supply wire solutions for the whole range of conceivable requirements and conditions. One wire type consists of a standard brass core (CuZn36) and a 1.5 – 3.0 µm thick nickel layer (type AG-N4). Due to the smoother surface of the uncoated brass wire it requires less effort to generate a homogeneous nickel layer. Moreover, the overall production effort is less compared to type AG-N3. Application of such wire is favourable in a WEDM process which can operate with 2 different wire types, one for those cuts with a relatively high material removal i.e. potential for speed increase and the other for those cuts with a relatively small material removal. The rough cut and 1st finishing cut would be performed with a speed wire and the final finishing cuts with the wire type mentioned above. In the same sense this wire type is a solution for the process combination of AWJM (roughing) and WEDM (finishing).
A further wire design (AG-N5, AG-N6) consists of a brass core, a first layer of beta brass (s = 15 - 25 µm) and an outermost nickel layer (s = 1.5 – 3.0 µm). This type would be applicable on all WEDM machinery with a wire threading system requiring a hard wire with a high straightness, both, features a copper core wire does not present. Compared to a copper core wire this wire type can enhance the cost efficiency since production effort and material costs are less while cutting rates are in the same range.
Figure 5: Wire designs for cutting Inconel fir trees
In addition to the above mentioned wire designs more simple designs were proposed. Composition, approach and drawbacks of those designs are displayed in Figure 6.
Figure 6: Simple wire designs for cutting Inconel fir trees
In the following steps the new wires types were produced, analysed and tested. In numerous WEDM test series cutting time, precision, surface roughness and contamination were analysed.
Figure 7 shows the productivity results in comparison to reference wires. In the main cut a time saving of 18-20% vs. brass wire is confirmed for the new series with copper core (AG-N3). Series AG-N2 yields similar time savings. The total time saving after 4 cuts with AG-N3 is also about 18-20% vs. brass wire. The standard X type wire technology for trim cuts does not fit to series AG-N3. As a result the time for the 2nd trim cut is too high and so the total machining time is. The standard brass wire technology which was applied for series AG-N2 (brass core) fits better in terms of productivity. With both series however slight grooves were generated after the final trim cut. The Ra values measured on the workpiece were therefore slightly higher as compared to brass wire or BCX/TPX. The precision obtained with series AG-N2 and N3 is of the same quality as with brass wire or BCX/TPX. Technology optimizations towards speed and surface quality in the trim cuts will be performed once the favourite wire type has been identified.
Figure 7. Machining time reduction with sample series AG-N2 and AG-N3
The analysis of the sample series AG-N1, AG-N2 and AG-N3 revealed that the outermost nickel layer presents discontinuities and larger variations in thickness. These defects are regarded as a drawback towards the avoidance of contamination by copper and zinc. The defects are due to the relatively high roughness of the wire pre-material prior to electro plating with nickel. The high roughness is a result of the formation of the zinc rich brass layer through a diffusion annealing process. After coating with nickel at an intermediate diameter the wire is drawn to its final diameter. Also this drawing step affects the quality of the outermost layer. For economic reasons, however, nickel plating at the final diameter is not an option. For the following sample series the process chain was modified in order to improve the homogeneity of the nickel layer. For sample series AG-N2, however, an improvement could not be achieved. The very brittle gamma brass layer generates cracks during final drawing which inevitably resulted in discontinuities of the nickel layer. In contrast the improved sample series AG-N3 displays a more continuous outermost layer (see Figure 8).
Figure 8: Cross sections of sample AG-N3b and AG-N3c (improved samples)
Workpiece contamination has first been analysed for reference samples made of tool steel. The influence of cutting regime (rough / finish) and wire type (plain, coated) was studied. Electron Micro Beam Analysis (EBMA) shows that roughing regime generates more contamination of Cu and Zn than finishing regime (about factor 3). After finishing with 2 cuts contamination is however still clearly detectable. The thickness of the deposition of wire material was found to be in average 0.3 µm for roughing regime and 0.1 µm for finishing regime. EBMA analyses of Inconel samples machined with 4 cuts are depicted in Figure 9, Figure 10 and Figure 11. Here it is visible that both coated wire types generate less contamination of Cu and Zn. However, the Ni coated sample does not further reduce contamination.
Figure 9: Work piece contamination after 4 cuts with brass in Inconel
Figure 10: Work piece contamination after 4 cuts with BCXcc in Inconel
Figure 11: Work piece contamination after 4 cuts with AG-N1 in Inconel
As already stated before, the Ni layer of the first series AG-N1 and AG-N3 is too discontinuous over the circumference. In order to investigate another potential reason for remaining contamination cross sections of the wire types in new and worn condition have been prepared. As microscope analysis shows, the depth of wear after 4 cuts is ca. 2-3 µm which is exceeding the Ni layer thickness of the sample AG-N1 and AG-N3a. Thus, samples AG-N3b and AG-N3c are regarded as preferred solutions.
The improved wire series AG-N3 was tested with 4 cuts and the Inconel samples were again analysed by EBMA regarding contamination. Figure 12 shows the resulting thickness of the deposited layer in comparison to the former results with reference wires. The wire BCXcc which is similar to type AG-N3 but without Ni layer already provides a drastic reduction of Cu and Zn deposition. The AG-N3 wires further reduce contamination. With a nickel layer thickness of 2.5 µm the contamination becomes in average less than 20 nm and in maximum less than 30 nm. In order to analyse reasons for the still remaining contamination a test was conducted leaving the work piece submerged in water dielectric for about 40 min without spark erosion. As Figure 12 reveals deposition of Cu and Zn already occurs without the EDM process. Typically Cu and Zn ions or fine particles are in solution respectively dispersed in the water dielectric. This debris material origins from prior EDM operations with ordinary wires or even from a prior rough cut with type AG-N3 where the wear of the wire exceeds the thickness of the nickel layer. The contamination layer without erosion has a similar thickness as in the test with AG-N3b or AG-N3c. Thus these wire types already represent favourable solutions.
Figure 12: Work piece contamination with different wire types
Parameters considered to be relevant are the electro plating process for zinc deposition, the diffusion annealing process in order to generate the beta brass layer and the calibration (drawing) process prior to the nickel plating process. These processes were each modified and the resulting layer quality was assessed. The investigations revealed that a well defined intermediate drawing step would be the most efficient way to improve the surface uniformity of the first layer (see Figure 13).
With the improved samples AG-N3b as well as samples AG-N4b and AG-N5b cutting tests and contamination analyses were performed. As depicted in Figure 14, however, no further improvements regarding contamination were obtained.
Figure 13: Improvement of surface prior to nickel plating
Figure 14: Workpiece contamination: overview of results with additional samples
At this stage the results and challenges could be summarised as follows:
• A preferred wire design comprises a copper or brass core, a 20 to 40 µm thick beta brass layer and a 1.5 to 3 µm thick nickel layer.
• Cutting speed in the rough cut is about 25% over standard brass wire.
• With a 4 cut technology the workpiece contamination is reduced by 3 times versus brass wire (down to 30 nm).
• Deposition of Cu and Zn within a thickness of 20 nm already occurs without the EDM process just by submerging the workpiece in water dielectric.
• Standard cutting technology for the trim cuts needs to be adapted to the new wire type since cutting time and surface quality are partly suffering.
• The potential for speed increase in the rough cut needs to be analysed.
• The new wire types have to be tested in the defined fir treecutting test.
Finally, Figure 15 gives an overview of the potential process set-ups for fir tree slotting together with the preferred wire solutions to be used in each set-up. If a post treatment was to be considered to eliminate surface contamination wire solutions without a special nickel coating could be applied. Those wire types have been identified within the early test series of the project.
Figure 15: Wire solution for cutting Inconel depending on the process set-up
The approach for cutting titanium alloys is different: Since an economic production of a wire with a titanium coating would not be feasible, wires with a reasonably thick layer of pure zinc are proposed (see Figure 16). The aim is to avoid contamination by copper. The subsequent chemical cleaning processes could be shortened since the less noble zinc can easily be dissolved.
Figure 16: Wire types AG-T1 / T2 / T3
In a first step samples of AG-T1 were produced and tested. Figure 17 and Figure 18 show the Electron Beam Micro Analysis (EBMA) of surfaces of TiAl6V4 parts cut with brass and AG-T1. Similar to the results with AG-N wires in Inconel the average layer thickness is drastically reduced. But more importantly the contamination with copper becomes practically zero when using AG-T1.
Figure 17: Work piece contamination after 3 cuts with brass in TiAl6V4
Figure 18: Work piece contamination after 3 cuts with AG-T1 in TiAl6V4
According to the wire solutions proposed for Inconel a table has been rendered displaying the recommended wire types for titanium alloys as a function of the process set-up (see Figure 19).
Figure 19: Wire solution depending on the process set-up
New dielectrics and additives
New development in wire-EDM made machine manufacturers to design new machines which do not work with deionised water as a dielectric but with hydro-carbon fluids. The big advantage is that oil dielectrics produce much better surfaces with lower thermally influenced layers and less micro cracks, which reduces the typical caused white layer to a minimum. The white layer can cause a reduction in lifetime of work pieces (compare project “EDM2005” (ref. GRD1-2001-40502)).
Using hydro-carbon fluids as a dielectric allows using better off-set parameters with smaller gaps, better surface finish but worse flushing properties. Further the heat conductivity of the oil is not as good as water, which normally results in a lower cutting speed when cutting with oil dielectrics. Normally several cuts (roughing and finishing) are needed for the desired surface finish, but cutting with oil dielectrics allows to skip one of the finishing cuts due to more effective roughing cuts (compare project “EDM2005” (ref. GRD1-2001-40502)). The optimisation of the cutting strategy with oil compensates the disadvantage which comes from the slower speed by cutting in oil. Therefore the total work time is similar to water, but the surface finish is much better. Another advantage of the use of oil as a dielectric is that there is no contamination of the surface with melted eroded wire particles on the surface of the work pieces and that hydro-carbon fluids avoid leaching effects on materials e.g. cobalt leaching compared to water.
For ADMAP-GAS, alternative ways for the manufacturing of turbine blades were analysed and developed. Besides water jet cutting the EDM wire cutting was selected as one alternative, due to reducing the EDM caused heat influenced layer (white layer) to a minimum and improving the surface finish by using hydro-carbon dielectrics (compare project “EDM2005” (ref. GRD1-2001-40502)). Further with special additives it was possible to increase the cutting speeds by using oil as a dielectric. For steel increases of the cutting rate in the range of 20-30% are possible. For cutting in Inconel the results received in the last six months period show improvements of up to 15%. During the project ADMAP-GAS roughly 80 additives have been tested in Inconel. These additives were tested in 24 different dielectrics. The dielectrics differ in hydro-carbon based fluids and additive contents. Some additives have been tested in the dielectric, but most additives have been tested through a special applicator which oelheld has patented. This additive applicator allows applying additives directly onto the EDM wire in high concentrations. Since October 2011 (project month 27) the dielectrics and additives have been tested on an AgieCharmilles RoboFil 440 Wire-EDM machine tool at WZL.
In one test procedure additives were applied through the applicator - “impregnating the EDM wire” - straight on the EDM wire and the additives were brought along the EDM wire directly to the erosion zone. The main goal of the trials was to test different additives, which were similar in their molecular chemical structure, but showed different behaviours under wire-EDM cutting conditions.
In another test procedure hydro-carbon fluids were used as base fluids and the additives were mixed into the whole fluid in varying concentrations. Here the main goal was to find additives which were 100% soluble in oil and shouldn’t be filtered out.
Further the same test procedures were repeated in deionised water used as dielectric, in which pre-tests had been made in the past, but more tests were done together with the project partners Charmilles, BEDRA and WZL.
The last test procedure then was a summary of all of the above, in which the new developed know-how and the best performing additives were used to dilute them into dielectrics of varying viscosities. The finishing cuts in oil dielectrics were done together with the project partners WZL and the machine tool manufacturer CHAR. While oelheld used an old analog generator for their cuts, WZL and CHAR used both new digital generators. Due to the fact that there is no existing technology for EDM cutting with hydro-carbon dielectrics the two partners WZL and CHAR mainly cut in de-ionized water, while oelheld cut mainly with hydro-carbon fluids. So, standard technologies of cutting steel with de-ionized water as dielectric were used, even with different wires. To compare in the end the cutting speeds and feeds with the existing technology the percentage ratio to the value of the standard technology table based on deionised water was used. So, the values of the standard technology table were seen as 100% and the received values were compared to them. The measured values are given in % technological value. The tested oil dielectrics differ in viscosity and are doped later on with different concentrations of an additive. In Figure 20 the test set up which has been used for the cutting tests is shown.
Figure 20: Test set up of cutting tests
To compare the cutting rates on a RoboFil 240ccS with water dielectric external tests were made at GF AgieCharmilles in Schorndorf. There is the same generator (digital) as located at WZL. The priority of the test was the speed. For the test deionised water was used as dielectric. The work piece was of Inconel 718 with a height of 40 mm. The following different wires were used: Brass 900, Topas plus H and Topas plus X. The results show that the highest cutting rate was achieved with the Topas plus X wire (see Figure 21). In each test the wire-additive OH 2377 could improve the cutting rate. With this experience for all tests the wire Topas plus X was used.
Figure 21: Cutting rates with different wire with and without wire-additive
In Figure 22 further test cut results are shown. The stated graph shows the different effect of the additives in the different dielectrics. So the abscissa shows the wire additives and the single blue bars represent the different dielectrics. In the graph which is shown in Figure 23 the dielectric OH 2286 is doped with different concentrations (given in percent values) of a performance improving additive. The percentage addition of the dielectric OH 2286 varies from 0.1 to 20%. The continuous line shows the averages of the results of each dielectric. At the end 100% of the technological value is achieved.
Figure 22: Improvements of different wire-additives in different dielectrics
Figure 23: Results of the dielectric OH 2286 doped with different amounts of additives
For the comparison of different dielectrics not only different base dielectrics were tested but also the base dielectrics were added with additive packages. So the base dielectric OH 2788 was added with additives to the dielectric OH 3521. As it is seen in Figure 24 with the additives the cutting rate of the dielectric OH 2788 could be improved. The result of the OH 3521 is even better than the other dielectrics. The test conditions were the following: as the wire-EDM-machine a RoboFil 230/240 located at oelheld with an analog generator (old generator generation); work piece Inconel 718, wire Topas plus X, no additive on the wire. The dielectric OH 2788 and OH 3521 were selected for the tests at the partner WZL in Aachen.
Figure 24: Cutting rates in different oil dielectrics
In Figure 25 the results of the cutting tests at WZL are mapped. A test series with three different dielectric fluids (water, CH-based dielectric and CH-based dielectric with additive) has been set up. Based on the developed technology table (more details about the technology are documented in the mid-term report) a technology has been tuned for the base dielectric (CH-based) as well as for the CH-based dielectrics with additives. The resulting cutting rates of the tests can be taken from the mapped graphs. It can be seen that the cutting rate in water is the highest one and the cutting rate in the basis dielectric is the lowest one. Due to the additives an increase of 66% in the cutting rate was measured. The measurement of the surface roughness shows that all tests resulted in a roughness minor Ra = 0.8 µm and therefore all processes meet the project requirements.
Figure 25: Results of cutting tests at WZL
Summarized it can be stated that using hydro-carbon dielectrics, mixing additives into these dielectrics and to apply special additives in high concentrations directly on the EDM wire showed tremendous improvements in the wire-EDM cutting process. Both the additives on the wire and in the dielectric had remarkable positive effects on cutting speeds and surface finish. The best results could be gained with additives mixed in high concentrations into the hydro-carbon based dielectric. One of the best performing mixture unfortunately caused handling and health problems and also was economical not an option.
No long term filter tests, no long term corrosion tests, no long term seals and gaskets tests or no long term machine tests were possible to run, due to so many different additives and base fluids to test. Also the stability (decomposition) of the dielectrics was not examined related to a long term test. Another problem is that the additives which had positive effects on the wire could not be used in the dielectric. This was because they were not soluble in deionised water or hydro-carbon fluids and were filtered out or taken out by the deionising resin cartridge. But even when deionised water as dielectric was used, some additives increased the cutting rate. As previous tests of the fir tree profile had shown, the effect of additives on the wire would cause an angel correction of the wire which means it would need a lot of technology adaption. So for the time after the project oelheld will start long term studies to examine the dielectrics and additives under production conditions. The main focus will be the stability of the dielectrics and the influence of the dielectrics on other wire-EDM machines.
Technology optimisation and process monitoring
Within the research regarding technology optimization and process monitoring several tests concerning technology tuning have been carried out during the whole project. To observe the surface integrity according to technology aspects metallographic observations were performed. The observation included inspections in terms of recast layer, heat affected zone dimension and cracks. Basis for the technology comparison was the developed technology table (more details about the technology are documented in the mid-term report).
Following the four available settings are listed:
• E2: roughing setting with triangular current iso-pulses.
• E3: roughing setting with trapezoidal current iso-pulses.
• E7: finishing setting with triangular current iso-pulses.
• E21: surface-finishing setting in HF mode.
The samples were prepared on the metallographic preparation laboratory (metallographic saw, enrobing, polishing material, chemical etching) which is available at CHAR. The surface observations of different positions on the test piece (top, middle and bottom – related to wire unwinding direction) were made on an optical microscope.
Following the results of four test cuts with different settings will be presented. Firstly the rough cut settings E2 and E3 are analyzed and after that the observations of the finishing setting E7 and finally the sufacing setting E21 are documented.
In Figure 26 the rim zone views of the E2 setting (M52 - triangular) are mapped. The recast layer is not continuous, its optical thickness vary from 0 up to 15 μm. Furthermore no heat affected zone is observed. Additionally, no major differences from top to bottom are observed. The resulting removal rate during the tests was VW = 156 mm²/mim.
Figure 26: Surface integrity of roughing E2 on Inconel 718
In Figure 27 the rim zone views of the E3 test cuts are pictured. The observed recast layer is not continuous. Its optical thickness varies from 0 up to 15 μm and no heat affected zone is visually observed. Furthermore no major differences from top to bottom were detected. The resulting cutting rate was VW = 92 mm²/min.
Figure 27: Surface integrity of roughing E3 on Inconel 718
In Figure 28 the rim zone views of the E7 setting cuts are mapped. The recast layer observed is not continuous and it’s optical thickness varies from 0 up to 8 μm. No heat affected zone is visually observed. Furthermore, no major differences are observed from top to bottom.
Figure 28: Surface integrity after finishing E7 on Inconel 718
In Figure 29 the rim zone views of the E21 setting cuts are mapped. The recast layer is not continuous, its optical thickness varies from 0 up to 4 μm and no heat affected zone is observed. Furthermore, no major differences from top to bottom.
Figure 29: Surface integrity after surfacing E21 on Inconel 718
To summarize the rim zone view inspections the following aspects can be mentioned. Additionally in Table 2 the results are listed:
• The recast layer, formed by re-solidification of the re-deposited and melted material during the EDM process is not continues and homogeneous.
• The thickness of the layer is correlated to the power of the last setting used. As the machining power decreases, the recast layer thickness also decreases.
• The resilience to high temperature of the Nickel based super alloy (as Inconel 718) shows a good behavior on EDM machining process.
Table 2: Overview of rim zone view analysis results
Finally, at the end of the project a good working technology table has been developed. This table includes several machining settings and sequences (including two roughing: E2, E3; two finishing: E7, E9; two surfacing settings: E21, E23; special pocketing setting for cleaning AWJ machined surfaces: E4) to meet the required surface integrity of the turbine industry for the fir tree application. The core of the technology is based on hard brass wires with a diameter of 0.250 mm and is able to cut several Nickel based alloys like Inconel 718, Udimet 720, etc. In addition to the fulfillment of the surface integrity requirements this table meets further on the surface roughness and accuracy requirements (Ra < 0.8 µm and accuracy ± 5 µm). Within the table machining parameters for the heights between 5 mm and 60 mm are stored. In detail the following settings and sequences are available:
• Settings:
o E2: H10-60 mm, Fast roughing with sealed nozzle M52,
o E3: H05-60 mm, Reliable roughing M54, for various nozzle configurations,
o E4: H05-60 mm, pocketing or cleaning setting,
o E7: H05-60 mm, finishing std U,
o E9: H10-60 mm, finishing HFBF,
o E21: H5-60 mm, surfacing M30 (200/200) Ra 0.7 and
o E23: H5-60 mm, surfacing M23 (400/400) Ra 0.8.
• Sequences:
E2, E7, E21. E3, E7, E21. E4, E7, E21.
E2, E7, E23. E3, E7, E23. E4, E7, E23.
E2, E9, E21. E3, E9, E21. E4, E9, E21.
E2, E9, E23. E3, E9, E23. E4, E9, E23.
In addition to the standard brass wire technology table further research and development concerning machining with the new nickel coated wires has been carried out. With regard to the productivity target the standard cutting technology for coated wire (X-type) was successively optimised by conducting standard EDM tests during the last project months. The result is shown in Figure 30. With wire type AG-N3b and optimised technology the time saving vs. brass wire is about 30 % while also precision and roughness targets are met.
Figure 30: Results from standard EDM test with standard and optimised technology
The optimised technology was then applied in a fir tree cutting test. Here slight parameter changes still were necessary. The results are displayed in Figure 31 and Figure 32. AG-N3b yields a time saving versus brass wire of 22% with standard technology and 30% with optimised technology. AG-N4b which has just a nickel coating performs similar like brass. With AG N5b about 7% time saving is reached under standard conditions. While there are no objections to the surface roughness the contour precision target is still not met. This also applies to the contour check of parts cut with samples AG-N4b and AG-N5b (not displayed here). However, the deviations are not that big that the wire type itself has to be considered as non-approriate. This is also supported by the fact that the new wire type AG-N3b reveals a higher precision compared to brass wire. It is therefore expected that a duly adapted cutting technology will yield the required contour precision.
Figure 31: Results from fir tree cutting tests
Figure 32: Contour check of wire-cut fir tree profiles
Evaluation of the results
The economic evaluation of the results can be seen in the chapter “Potential impact and main dissemination activities and exploitable results”. Here the two ADMAP-GAS processes are evaluated concerning the feasibility of the substitution of the broaching process of fir tree profiles in terms of costs and environmental impact.
Concerning the surface integrity comparison fatigue tests have been carried out. Three test series were tested which have been produced with the following manufacturing technologies:
1. Wire-EDM
2. Broaching
3. Grinding
The Wire-EDMed series has been produced at WZL with the developed Inconel 718 machining technology. The broached series was machined at WZL too. In Figure 33 the broaching machine located in WZL is located. This machine is a vertical working machine and equipped with a force measuring platform to monitor the tool wear. On the right in the figure the clamping device with the test specimen is sketched. According to the finishing process of the turbine manufacturing to produce fir tree geometries a tool and machining parameters were designed and test sample were produced. The third series which has been produced by grinding was manufactured by an OEM.
Figure 33: Machine set up to manufacture fatigue test samples by broaching
After the manufacturing of the specimen a surface integrity inspection of a sample of each series has been carried out. In Figure 34 results of a surface roughness measurement are mapped. All surface roughness measurements are minor Ra = 0.8 and fulfill the projects requirements.
Figure 34: Agenda and surface roughness measurement results
Additionally to the surface roughness measurements rim zone analysis of a specimen of each series has been set up. In Figure 35 the rim zone view of an eroded specimen is pictured. A very thin white layer is visible. No micro cracks or thick thermally influenced zones have been detected.
Figure 35: Rim zone view of series 1
In Figure 36 the rim zone view of a broached specimen is pictured. Two different analyses according to the cutting direction have been prepared. On the one hand a view in cutting direction and on the other hand a view perpendicular to the cutting direction has been prepared. The view in cutting direction (upper two views) shows a slightly rough surface. Furthermore a small thermo-mechanical layer is visible. The other view shows a flat surface topography. Also here a small thermo-mechanical affected zone can be seen. Due to the deformation of the grains a mechanical detraction of the manufacturing process is obvious. Approximately 5 – 8 µm under the surface the parent and non-affected material can be seen.
Figure 36: Rim zone view of series 2
Figure 37 maps the rim zone views of the ground series. Here, nearly the same aspects as in the broached series can be seen. A zone which was affected by a thermo-mechanical and a zone which was affected by mechanical forces are obvious.
Figure 37: Rim zone view of series 3
In Figure 38 the test set up and the results of the tests are shown. The specimen is clamped between two chucks. One chuck is moving and one is fixed. The moving chuck induces the bending moment into the specimen. The test conditions include a limitation of the cycle numbers of 5 x 106 cycles. The bending frequency is 70 – 75 Hz.
Figure 38: test set up and results of fatigue tests
Looking at the results a very good project outcome can be summarized. It can be seen, that the Wire-EDMed and broached series fail in the same surface stress area. The ground series is the best of the three series. Nevertheless it can be stated that in terms of fatigue behaviour the developed Wire-EDM process is able to substitute the broaching process to manufacture fir tree profiles.
Flushing and nozzle design studies and developments
Flushing is one of the most important parameters within Wire EDM and especially within the rough cut. A high quality flushing improves the speed in a very high range, due to the fact that flushing removes dirt and scraps off the machined slot and cools down the wire allowing higher sustainable electrical power. In opposite to the flushing during the rough cut the rinsing within the finishing cut has a minor role. So, the main research activities concerning nozzle design were given to the rough cut flushing.
The dielectric flow out of the nozzles modifies the position of the wire due to vibration. The results are geometrical defects which have to be corrected by the following cuts. These effects are influencing the achievable global speed and accuracy. To achieve the best process outcome an optimization of the flushing should improve efficiency and reliabilities of the WEDM process. Hydrodynamics specialists from a partner institute (HES-SO) assist CHAR to simulate dielectric flows into the machining gap and optimize nozzles design.
Several new nozzles (12) were designed and manufactured. These nozzles were tested and their performances were compared to existing and standard nozzles. Within former work of the project some “short” nozzles were tested and their performances analyzed (see former project reports for more details). The performances of the so called “long” nozzles and diffuser were analyzed, tested and quantified during the last project months.
As stated above the productivity and the efficiency of the rough cut varies significantly with the flushing quality (nozzles dimensions/positions or machined disk geometries constrain). For a substantial comparison a range of eight upper head nozzles were tested and their performances were compared (see Figure 39). The lower head’s nozzle configuration has been the same in all tests.
The tests were carried out on the test bed machine tool at CHAR. The first tests contain a static set up. The upper head was placed over a machined slot with a wire but without any erosion. Starting from a sealed nozzle, the distance between the work piece surface and the lower nozzle surface has been increased for each nozzle. As seen in Figure 40, the main drop of pressure occurs in the first millimeter. The nozzle with the highest pressure was the nozzle L3. Furthermore the pressure gradient of this nozzle seems to be the best within these tests.
Figure 39: Drawing and characteristics of tested nozzles
Figure 40: Static pressure measurements for different upper nozzle types
To quantify the improvement made by the nozzles comparative tests were performed. The test sequence included machining a slot and visualizing the water flow on the work piece by analyzing the coloration of the work piece surface. The cuts were made on steel test pieces with the height h = 125 mm. This height was chosen to have a better visualization of the coloration. The used technology table was the standard technology LT25A, HM66. The upper nozzle was placed ~ 0.2 mm over the upper work piece surface. The lower nozzle geometry was always a standard geometry. In Figure 41 the machined surface and the corresponding nozzles are pictured. The different coloration can be observed in the picture. The differences in the coloration are on the top of the surface. Here the upper nozzles differ from the lower one which has been the same for all tests. It can be seen in the figure, that for the nozzles L1, L2, L3, L4, F2 and BR in the upper region (approx. 10 mm from the upper edge) a different coloration due to the flushing appears than in the standard nozzle geometries Standards and PM. A flushing dead zone can be examined in the middle of the surface. In this zone the flushing is very bad which means, that not enough fresh dielectric fluid is present for good process conditions. This results in e.g. not enough wire cooling in this region. For minor work piece heights this issue is not as relevant, because the dead zone should be smaller.
Figure 41: Machined surfaces and nozzles corresponding
Additional to the coloration tests, machining speed tests were carried out. For these tests the standard project height h = 40 mm was chosen. In Figure 42 the resulting removal rate can be seen. In this case, the ability to renew dielectric in the slot has measurable effects on the machining speed. The best result is made with auto adjustable nozzle (PM). The PM nozzle is partly mobile, so the dielectric pressure sticks the nozzle surface on the work piece surface. This maximizes the dielectric flow into the machined slot. Unfortunately, this nozzle is not suitable for ADMAP-GAS application because it induces mechanical forces onto the work piece that lead to loss of geometrical accuracy. Furthermore, its adjustability is limited and this kind of nozzle is suitable only on flat grinded surface.
The tested nozzles did not show better results in terms of speed and accuracy than actual standard nozzle. Some differences are measured between nozzles but no new designed nozzles over take the actual standard nozzle performances.
Figure 42: Removal rate achieved with different upper nozzle geometries for H: 40 mm
The numerical simulations by finite elements modeling which have been presented in the last report show a bad dielectric flow in front of the wire. One conclusion out of the simulation was that a spinning dielectric flow around a symmetrical axis (around the wire) is expected to increase the dielectric flow into the working gap. To achieve a rotating dielectric flow the standard diffusor which is placed in the upper head was redesigned. In Figure 43 the standard and two redesigned diffuser are mapped. These diffuser geometries were tested on the test bed machine tool at CHAR.
Figure 43: Three different diffuser geometries
The tests and results of these diffusers did not get measurable improvements on the machining performances. Some concepts to decreases the dielectric flow diameter and to increases the pressure in the gap are ongoing. The work done on nozzles and diffuser show and confirm that the actual material available on the machine is performing quite well and to improve significantly their performance it will require more work. The nozzles have strong effect when they are sealed (as close as possible at the surface). During fir tree machining, external geometry of the disk is often not flat, therefore it is not possible to machine with the nozzles always close enough and sealed to the machined surface, then nozzles effects are greatly reduced.
Machining technology optimization
A new idea to improve efficiency and reliabilities in machining fir tree are correlated to the actual way of machining with the 3 cuts sequences. 1s t cut roughing, 2nd cut finishing and 3rd cut surfacing. The roughing setting is working in the bulk of the material, its performances is tightly correlated to the capabilities to put as much power as possible into the machining area sustainable by the wire. The 2nd cut is a finishing cut, it corrects and improves the geometry of the machined surfaces. The following aims have to be met by the cut:
• The corrections are necessary to correct the straightness of the surface. This defect has its origin on the soft wire electrode which is tight in position by the upper and lower guides. Between the guides the wire vibrates, that affect its straightness on the machined surface.
• It must also correct dissymmetry defect. This dissymmetry is caused by the deviation of the dielectric flow against the surface of the disk. A turbine disk is not flat and infinite.
• It also corrects machining geometrical defects produced during the roughing operation. They are linked to the machining strategies and protection resulting of loss of accuracy induced by the wire drag phenomenon and the way to minimise it.
This finishing cut corrects all these defects at a reasonable machining speed. Unfortunately it did not improve sufficiently the surface roughness and surface integrity up to the project’s objective of Ra ≤ 0.8 µm. The 3rd cut is a surfacing cut; the purpose of this cut is to put the surface into the objective of roughness and surface integrity. It is a constant speed cutting operation that is not controlled. The cutting is “blind” and no actions are possible if something is not into its range of functionalities. The two cuts studies aimed to replace the 2nd and the 3rd cuts by a controlled single cut that should give the project objectives in geometry, surface quality and integrity at a resulting machining speed equivalent or faster to the 3 cuts existing sequences.
To reach these objectives several solutions were implemented and tested to produce a setting capable to give in one cut the performance of two cuts (E7 and E21). The following boundary conditions were given out of the project for the development:
• The surface quality and geometry have to meet the ADMAP-GAS requirements:
Ra ≤ 0.8 μm, minimum surface contamination and geometry precision within tolerance band ± 5 μm.
• Stay competitive in terms of time, consumables and costs:
Speed of one cut ≥ actual resulting speed of two cuts E7 and E21.
• To improve reliabilities and productivity: Controlled machining cut.
The expected improvements of the implementation of the new two cut sequence are on the one hand efficiency aspects (all controlled settings) and on the other hand productivity issues (less threading operations…). Within the scope of the research several solutions were tested on the standard generator of the machine test bed at CHAR:
• New current pulse shapes: “Isopulses” without ignition.
• “Isofrequence” modes using UP power supply instead of ignition UAL power supply.
• Shorten the current pulses width (delay line, faster transistors, diode to minimize current damping, etc.).
The Table 3 summarizes the results. The achieved results do not match the ADMAP-GAS specifications in terms of surface roughness (Ra < 0.8 µm). Nevertheless, a real improvement is made showing a new way to increase the machine performances in improving the efficiency and stabilities of the 1st finishing cuts. Some more development and improvements are required to fulfill the objectives. Innovation developments are ongoing on these topics.
Table 3: Achieved results with several prospected solutions
Critical assessment concerning WP2 objectives
Looking on the overall WP2 objectives it can be said that the research on Wire-EDM within the project ADMAP-GAS was very successful. In Figure 44 the objectives, which have been set up for WP2 and the status of the achievements of the aims are listed.
Figure 44: Achieved objectives
WP 3: Development of “Abrasive Water Jet Slotting”
This subchapter focuses on the main S&T results/ foregrounds concerning Abrasive Water Jet Machining (AWJM). The AMRC, USFD, TEKS and DIAD investigated the performance of AWJM for titanium and nickel based alloys to benchmark the process against broaching in turbine engine disc manufacture.
This work package was set up to develop AWJM technologies as an alternative to broaching in aerospace; particularly for fir tree slots. The following objectives have been set up before the project start:
• Improve process control to compensate for jet-deflection in fir trees and blisks; particularly blow-out and curves on hard alloys.
• Reduce kerf size to less than 0.3 mm in order to be able to cut inner fir tree diameters.
• Develop methods of precisely controlled depth cutting, including the complex curves and tapers found on blisks (depths 50 – 100 mm); positional accuracy will be improved to that of the nozzle (± 30 μm).
• Improve nozzle-life by 50%.
Definition and process analysis towards the targeted application
The work concerning the definition and process analysis towards the targeted application was finished successfully during the first period of the project. The following work was carried out:
• A survey of aerospace discs and fir tree features was carried out. The information collected was used to design a generic fir tree geometry, which contains the most difficult features to machine from all the disks. This has been used throughout the project by all partners in the consortium.
• A specification of tolerances and surface integrity requirements was produced. The specifications have been used as a target for developing the technology and methods to achieve the requirements - e.g. tolerances of 5 -10 µm, subsurface deformation < 51 µm (0.002 in).
• Additionally, analysis of existing mathematical models for modelling AWJ system was carried out; as well as a review of AWJ cutting mechanisms.
Fir tree features are complex geometry with tight tolerances up to 5 µm in the pressure flanks and 10 µm in non-pressure flanks. The subsurface condition is restricted to 51 µm deformation, no white/amorphous/recast layer and no foreign material present. Aerospace discs can have a diameter as big as 900 mm and their thickness is around 30 mm. Fir tree features can have from 2 to 4 teeth and have a minimum internal/external diameter of 0.6 mm. The test-piece design has 4 teeth and a minimum internal/external diameter of 0.6 mm.
Development of Abrasive Water Jet Machining
The main research and development work was split into three sub-tasks with certain overlap in the development of nozzle geometries and improvement of hydraulic jet kinematics.
Analytical work has been directed towards developing a Precision Abrasive Water-jet system (PAW). The PAW is composed of a precision cutting head, a precision table, machine frame, and benchmark and calibration system.
An analysis of the current system uncertainty and sources of variation led to USFD proposing the PAW system. The analysis was based on the available AWJ machine ZX-613 located at USFD. The cutting head is a standard Dialine™ head from Accustream. However, the analysis, methodology and concepts developed are suitable for all other water jet systems including 5-axis systems. The main error sources in AWJ machining are due to uncertainty caused by:
• u1: cutting head vibration;
• u2: locating components in the machine bed;
• u3: flow deflection;
• u4: machine frame and system vibration;
• u5: other elements.
The research and development within ADMAP-GAS focuses on the avoidance and improvement of the first three errors. Other sources of machine uncertainty such as machine frame and system vibration (u4) and other elements (u5) were not in scope for this project. The concepts developed in u1 to u3 can be applied to any waterjet machine regardless of machine frame configuration.
Development of different nozzle geometries control angles (avoid the errors u1 and u2)
Within the work on the different nozzles mainly the two error sources u1: cutting head vibration and u2: location components in the machine bed have been analysed and improved. In detail research on a precision cutting head and different new components for a more efficient process were carried out. Following the work is summarised.
Precision Cutting head - u1
The aim of the precision head is to increase stiffness, reduce vibration and to provide probing capability which will help to locate parts in the machine bed (u2). It also allows the nozzle to be easily changed, reducing set-up errors.
Suppose the z axis is the axis where the component is to be cut, then the machining errors can be classified as position errors (Δx, Δy, Δz), orientation errors (Δθx, Δθy), and timing errors (Δθz) with respect to 3 axes, x, y, z.
Machining uncertainties will be produced in all 6 freedom degrees, (x, y, z, θx, θy, θz), correspondingly machining uncertainty for each error source i and its components in 6 freedom degrees are U {Ux, Uy, Uz, Uθx, Uθy, Uθz}.
Hence, the general machining uncertainty (U) of the system is the sum of all main error sources (ui) as Equation 1
[1]
The machining frame and motion system produce machining uncertainties depending on the machining system configuration used for various applications. The other elements that produce machining uncertainty are not as significant as the first three items, and are assumed to be negligible to simplify the expression.
In order to control and minimise the machining uncertainty (u1), a precision cutting head has been developed. A comparison of the conventional and precision heads is shown in Figure 45. The precision cutting head has higher stiffness, a connector which fits the novel nozzle developed in ADMAP-GAS, the capability to interchange tools and is constructed to allow the nozzle to be easily changed.
Figure 45: Conventional (left) and precision (right) abrasive waterjet cutting head
The combined standard position machining uncertainty (σ=1) produced by the cutting head in x, y directions are as Equations [2] and [3].
[2]
[3]
The combined standard orientation machining uncertainty (σ=1) produced by the cutting head in x, y directions are as Equations [4] and [5].
[4]
[5]
The position machining uncertainty (0.0185 mm) produced by the precision cutting head is predicted to be 7.5% of the error produced by the conventional head (0.2476 mm) in both x and y directions. The orientation machining uncertainty (0.000185 rad) produced by the precision cutting head is predicted to be 41% of the conventional head (0.000238 rad) in both x and y directions. Manufacture of the precision head was carried out at the University of Sheffield facilities. Figure 46 provides a picture of the precision head.
Figure 46: Precision water-jet head manufactured
The precision water-jet head was installed in the z carriage of the machine, Figure 48. The head was tested in an experimental framework using Design of Experiments in order to quantify the improvement in terms of surface roughness, taper angle and error in producing internal and external radii. Fir tree-like test-pieces were cut from Inconel 718 plate. These had the same geometry as that used to characterise the system with the conventional head and included internal and external radii of 0.3 and 0.5 mm as shown in Figure 47.
Figure 47: Fir tree like test-piece
The precision head produced superior surface roughness and taper angles when compared to the conventional head. However, the error in internal and external radii was increased. It is hypothesized that this was due to a reduced kerf size, resulting from the decreased vibration in the precision head.
Figure 48: Precision Waterjet head installed in the z-carriage
The validated gains made in precision due to the new head developed in ADMAP-GAS are shown in Table 4. The taper ratio was reduced by 10% due to the lower level of vibration and higher nozzle stiffness. Surface roughness showed an decrease of ~50%. It is postulated that the increased error when producing radii was due to the conventional controller working with different kerf geometry. The new kerf geometry needs to be calculated experimentally and fed into the AMRC Limited controller to reduce errors in the radii.
Table 4: Validated gains made in precision due to the new head
Locating Components in the machine bed - u2
In order to control the component location (u2), a precision AWJ table was designed. Standard waterjet machine beds consist of a series of grids with low location precision where it is difficult to clamp parts. Moreover, the grids are flexible and induce unwanted vibration while cutting. The precision table developed is specifically designed to accommodate turbine disc components. It can be calibrated by means of a novel mechanism to high micron precision. It consists of a matrix of Ф12 mm location holes at intervals of 50 mm as shown in Figure 49. The fixture table can be adjusted by ± 2° in all axes, to compensate for variation in the machine bed and component.
The new table is based on a general component location concept used in most CNC machines, and will enable the AWJM to reach the accuracies required for the ADMAP-GAS components. The standard machining uncertainty of such a system is approximately ± 0.010 mm.
Figure 49: Precision abrasive water jet table customised for discs
The table consists of two main components made of stainless steel and three adjustable mechanisms. The fixture rests on an aluminium frame. All washers, bolts and nuts were specially ordered as they need to be made of stainless steel.
Manufacture of the precision table took place during this project month 31 and 36. The table was manufactured by TEKS. The precision table was set up on the 3-axis machine waterjet tank, Figure 50. It was calibrated using high precision clocks clamped onto the z-carriage. Calibration was done for the x, y and z planes. The achieved calibration was 0.015 /100 mm which satisfies the objectives of precision location for the production of fir tree features in ADMAP-GAS.
Figure 50: Precision abrasive water jet table set up on the waterjet tank
As the project was initially due to finish in project month 36, USFD had committed to acquiring a 5-axis waterjet machine to take forward the ADMAP-GAS developments. This was a result of the high interest in the ADMAP-GAS research demonstrated by USFD’s mainly aerospace industrial partners which resulted in a number of feasibility studies being carried out for these partners. One result of these studies was that to realise the full benefits of the research and to take it into industry, a 5-axis machine would be required. As a result of this, the 3-axis machine was replaced by a larger 5-axis version with the capability of holding and machining very large parts. The machine has a working envelope of 6 x 4 x 1.5 meters. The waterjet head has a completely different configuration than the 3-axis having the capability of rotating the A axis ±180° and the C axis indefinitely. Unfortunately, this meant that the parts designed for the 3-axis machine needed to be modified to fit the 5-axis in a very short time-period in order to complete the validation for reporting. The results show definite improvements over the conventional head, but further developments are required to take full advantage of the 5-axis machine and USFD are committed to refining the precision table to fit with the 5-axis machine as part of their commercialisation strategy.
The precision table bracket was modified in order to make it fit into the USFD new waterjet machine. The precision table was set-up and calibrated to the 0.015 / 100 mm. The table was used to produce the aerospace demonstrator (a quarter of a gas-turbine test-piece) and the automotive demonstrator (a flange).
Improvement of hydraulic jet kinematics (avoid flow deflection - u3)
The work carried out on the improvement of hydraulic jet kinematics is focused on redesigning the nozzle orifice and building a nozzle assembly. The new design will improve the jet kinematics and improve nozzle life. The new nozzle will equip the waterjet system with a more powerful jet minimising flow deflection (u3) which will make it capable of machining to higher tolerances or higher productivity. A nozzle assembly consists of a connector and an orifice as shown in Figure 51.
Figure 51: Conventional Orifice; Din, supply diameter; Dout, outlet diameter; α, jet taper; Do, orifice diameter; Lo, orifice length
Flow deflection (u3), (Figure 52), is another machining uncertainty to be controlled, if precision cutting is to be achieved. The flow deflection heavily depends on component thickness and process parameters (different pre-programmed cut qualities, Qi). Improvements can be achieved by increasing output power.
Figure 52: Flow deflection in different cut qualities (Qi) (source: WARDjet)
In order to reduce the flow deflection, kinetic jet optimization has been carried out with the aim of increasing the cutting head power by developing a novel nozzle consisting of a new head connector and nozzle orifice.
Computer Fluid Dynamics (CFD) simulation has been used extensively in order to analyse the flow of the conventional orifice and to develop a new one. Multi-phase (water and air) models of the conventional orifice and prototype orifices were developed. Analysis in terms of the water/air volume fraction, velocity, pressure (static and dynamic) distribution and Reynolds number was carried out.
The Reynolds number gives the ratio of inertial forces to viscous forces and describes the type of fluid flow (turbulent, laminar, etc.). A low Reynolds number (<2100) implies laminar flow and will improve nozzle efficiency, power and lifetime.
CFD models of an ‘ideal orifice’ were also developed and normalised non-dimensional parameters calculated based on the ideal orifice, for comparative purposes. The ideal orifice is defined with a length / diameter ratio of 1.0. The ideal orifice assembly would have completely laminar flow, but would not be useful for AWJ applications as there is no build-up of pressure and hence the machine would not cut. The conventional orifice assembly has very turbulent flow; cutting power and efficiency would be improved by making the flow more laminar. The optimal orifice assembly would be a compromise between totally laminar and turbulent flow, where power is optimised. The three types of orifice assembly are summarised in Table 5.
Table 5: Summary of orifice assembly types
Simulations were carried out with 24 suggested geometries based on four different acceleration tapers and six different outlet tapers using both the ideal and conventional orifices. The computational work provides an estimate for the improvements in performance and tool life using the redesigned nozzle assembly. A summary of the analytical analysis is provided below, for a full report see Deliverable 3.2.
The 24 suggested orifice designs, conventional and ideal were simulated and compared in terms of maximum velocity, maximum total pressure, flux and output power. A comparison of the output power grouped by type of taper for different outlet tapers can be seen in Figure 53. The optimal geometry was selected according to its power performance.
Figure 53: Normalised output power for different orifice geometries
Simulations were carried out for the optimal orifice and results obtained in terms of orifice water fraction; velocity; total pressure; and Reynolds number for both transient and steady states. A comparison based on simulation results for the distribution of the Reynolds number in steady state for i) the ideal, ii) conventional and iii) optimal orifices is shown in Figure 54, where the white space surrounding the coloured stream represents the orifice material.
Figure 54: Comparison of distribution of Reynolds Number for i) ideal, ii) conventional and iii) optimal orifices; white space represents nozzle geometry
Table 6 provides a comparison of the theoretical conventional and optimal orifices in terms of the percentage increase in efficiency, flux, output power and maximum velocity.
Table 6: Normalised comparison of conventional and optimal orifices
Another objective of ADMAP-GAS is to increase life of the nozzle. Theoretical analysis has been carried out in order to consider tool life while designing the new orifice. A summary of the work carried out follows:
Orifices are damaged through two mechanisms: wear or breaking. Accordingly, two nozzle orifice life definitions are used: wear life and fatigue life. The shorter one will decide the orifice life.
Output power was used to calculate the orifice contribution per time unit, regardless of whether the water jet machine is cutting or not. Another important issue is the maximum water speed at the orifice where a higher speed will produce faster wear. The third item to consider is that a unit orifice radius change produces a volume increase (worn).
An orifice anti-wear index, ξ, can be defined as its output power, wout, divided by the maximum velocity, vmax, at the orifice centre times the unit orifice radius increment produced by volume induction, dV/dr, Equation [6], normalized as Equation [7]. Calculations of the anti-wear index for different ratio dr/r0 are presented in Table 7.
[6]
[7]
Table 7: Anti-wear indexes of the ideal, conventional, and optimal orifices
The normalised anti-wear index comparison of ideal, conventional, and optimal orifices is shown in Figure 55 with the radius change from 0 to 10%. Compared to a conventional orifice, the theoretical anti-wear index of the optimal orifice increases by 267% (dr/r0=0) and 447% (dr/r0=0.1).
Figure 55: Theoretical normalized anti-wear index comparison of ideal, conventional, and optimal orifice
The supply tube is full of air at the start of the process. When the water valve is turned on, the air inside the supply tube will be pressed out through the orifice. This happens very quickly because of the low viscosity of air. However, when water arrives at the orifice the speed reduces immediately because of the high viscosity of water. This happens in such a short time, Δt, that it produces a big impact, like a hammer. This phenomenon is produced by the viscosity difference, so it is called the ‘viscosity hammer’. The impact force produced is expressed in Equation [8].
[8]
In the optimal nozzle, an acceleration taper is applied before the orifice and the acceleration distance and time, Δt, are extended. Using the optimal orifice geometry, the viscosity hammer effect will be decreased and the fatigue life will be improved.
Using the optimal geometry, the Reynolds number will be reduced to 58% that of the conventional orifice, and a laminar flow will be achieved upstream of the orifice. The Reynolds number is less than 2100; hence it is assumed that a laminar flow is produced. That is the main reason why the optimal orifice is so efficient.
Based on the theoretical analysis carried out and the comparison of the optimal, conventional and ideal orifices, the following conclusions can be drawn. Using the theoretical optimal orifice increases:
• The orifice efficiency to 148%;
• The orifice flux to 220%;
• The orifice output power to 325%;
• The orifice anti-wear index to 267% (dr/r0=0) and 447% (dr/r0=0.1).
Considering the improvement of the orifice anti-wear index and the orifice impact, one can safely conclude that the orifice life will be improved more than 50% which is one of the objectives of ADMAP-GAS.
These improvements need to be validated by creating the optimal orifice geometry and connecting it to the AWJM. This was a challenging task, as the parts needed to be produced and assembled to high accuracy and tolerances, and the production and alignment of the various components was not straightforward. Initially, in order to make the nozzle orifice outlet taper an ultrasonic drill was considered. Time was spent trying to source a drill from China (the only suppliers of this equipment), acquiring drawings and making drawings to design a fixture for the ultrasonic drill (Figure 56), so small holes could be drilled in rubies. A method to drill the rubies and produce the outer tapers was devised. Equipment to this end was ordered (cleaning acid, rubies, fine wire, needles, materials, etc.). In the end the ultrasonic drill could not be bought due to good financial practice as required by University regulations, and USFD did not want to subcontract the manufacture of the rubies as they wish to patent the nozzle and are concerned to control the intellectual property, and more prosaically, there is no subcontract budget in the project. Therefore, a second plan was considered which would allow them to manufacturing the nozzle in-house. A similar ruby orifice was found – this was not an optimal orifice, but with some modification, it was expected to produce good results in conjunction with the rest of the nozzle assembly. While this will allow the theoretical results to be validated, there is less flexibility to experiment with different geometries, so a true optimal could not be tested.
Work was also carried out on the manufacture of nozzle inserts for the acceleration taper. Three different inset orifice sizes were considered - 0.10 0.20 and 0.35mm. Micro tools with a diameter of 0.066 mm were acquired in order to machine the small size inserts. Different manufacturing routes were tested. In the first one, metallic cylinders to size were drilled in steps from both ends (Figure 57) and then wire fed through them and a cone produced using a WEDM machine. There were issues with the concentricity of the hole and cylinder as the tolerances needed are tight. As a second approach, a disc was firstly made out of brass, a threading hole made and both the cone and the cylinder wired out using a WEDM. This method produced a better concentricity but changing set-ups still meant that the tolerances were not achieved.
Figure 56: Utrasonic drill fixture (approximately 120 mm x 96 mm x 40mm)
Figure 57: Drilling of a 0.07 mm threading hole, where Øi are section diameters and Li are section depths
However, the precision head connector was manufactured and tested. This is a critical component as it joins the high pressure water line to the precision head. Initially, the connector consisted of two sections with one interchangeable part where nozzle inserts with different geometries could be fitted. After testing such a design it was evident that it was prone to leaking. A second design was made which consisted of a single piece where the insert was press-fit. This design also experiences leakage in the top section which joins a seat to the actuator. The seat and actuator system close the valve when the system is not cutting.
During the last project period nozzle inserts and nozzle connectors were manufactured and tested. Assembling the different components to high concentricity and accuracy was not trivial and a number of assembly routes were also tested. There were persistent problems with the nozzles leaking and exploding under the high pressures produced by the jet. The assembly consists of three parts, seat, connector and insert. Initially, all three components were leaking. The leakage of each was addressed separately. A new design for the interface between the actuator and the seat alleviated the problem in that part of the assembly. Different designs of the connector – insert interface were tested. Most of them still leaked and the ones that did not produced a flow with high turbulence. It was believed that tolerance errors in concentricity in the axis of the assembly caused turbulence in the flow producing leaking. Different combinations of connectors an inserts were tested, however, no combination was found which did not produce turbulence or leakage.
During the last period of the project the design of the insert and connector was changed in an attempt to reduce turbulence due to the assembly and facilitate manufacturing. A form-tool was manufactured for producing the inserts using a lathe. The cylinders were turned and drilled in the same set-up allowing tighter tolerances to be achieved. The new inserts were tested (see Figure 58) with the new connector; although leakage was stopped, the system was still experiencing high turbulence producing a fan-like kerf.
From observations made during cutting, the new nozzle is significantly more powerful than the conventional one. If it can be properly manufactured, it will reduce cutting times and increase the accuracy of the jet. Even without the nozzle, the PAW system shows impressive improvements over the conventional head, and is accurate enough to perform rough-cuts on 70% of fir trees, for later finishing by wEDM or broaching. A new design team has been set up at USFD and the waterjet team are working closely with them to redesign the nozzle insert for manufacture; they expect to manufacture and validate a prototype nozzle insert within 6 months of the project end. Once the nozzle has been validated, USFD plan to patent the system.
Figure 58: Testing of the nozzle insert
Technology optimisation and process monitoring
This work has been split into two sections, technology optimisation, carried out by USFD and process monitoring, carried out by AMRC.
Technology optimisation
Work has been carried out to benchmark and characterise the AWJ system as well as investigating of process monitoring technologies. The following work has been completed:
Benchmarking: A benchmark of the pre-programmed process settings for Inconel 718 was carried out. The benchmark consisted of assessing 5 different ‘Quality settings’ which are generated with different travel rates. Relationships between travel rate, surface roughness, taper angle and level of affected layer of the sub-surface were developed. The best quality setting was considered as ‘state-of-the-art’ and the improvement in the project assessed against this.
It was found that the quality setting (based on traverse rate) has a significant effect on taper angle, sub-surface deformation and surface roughness. An interesting finding was the cutting mechanism for Inconel is by micro-cutting and ploughing. The subsurface of water-jet cut Inconel has craters. Figure 59 shows some micrographs showing the difference in surface conditions.
Figure 59: Sub-surface condition of water jet cut Inconel
Process Characterisation: Characterisation of the process of water jetting straight edges and radii in Inconel 718 was carried out. Experiments were performed in fir tree-like test-pieces; six different critical process parameters (CPP) were considered (water pressure, nozzle configuration, traverse rate, abrasive mass flow, abrasive grit and nozzle stand-of distance). The results were analysed and presented in mathematical models that relate all six CPPs to critical quality attributes (CQA); specifically: taper ratio, surface roughness and relative error of producing an internal and external radius.
An interesting finding was that the pressure has a quadratic effect on surface roughness and the relative error of producing an internal/external radius. A summary of the effect of the CPPs on CQAs is presented in Table 8. The arrow indicates the direction of change, the asterix the most important CPP and 2 a quadratic effect. For instance, the main effect for taper ratio is the nozzle configuration (Noz), which reduces with nozzle diameter. Reducing the feed rate (Feed) as well as the nozzle stand-off distance (StDis) and abrasive grit (AbTpe) reduced taper ratio; whereas reducing the water pressure (Pre) increases the taper ratio. Abrasive mass flow rate (AbRte) did not have an effect on taper ratio.
Table 8: Summary of influence of CPPs on CQAs
An example of a mathematical model showing the quadratic effect on pressure is shown in Figure 60. The model relates surface roughness at the bottom of the sample with pressure at a stand-off distance of 2 mm, abrasive rate of 0.55 kg/min. It can be seen that for example at a pressure of around 2800 and 4000 bar the same surface condition can be achieved at the same traverse rate. This has implications for the wear of the system as well as energy consumption.
Figure 60: Contour plots for surface roughness
The characterisation results were used to produce the first fir tree test-piece. This test-piece was used to assess the development of the technology and methods developed throughout the project. This was produced with non-optimal parameters, no new controller or nozzle or head.
Figure 61 presents a summary of the results in the form of a scan. The maximum error in the non-pressure flank for the front scan was 0.072 mm, 0.115 mm for the pressure and 0.034 mm for the bucket groove. The error in the non-pressure flank for the back scan was 0.179 mm, 0.276 mm for the pressure flank and 0.1136 mm for the bucket groove. The blue line shows the nominal profile whereas the green lines show the tolerance bands; any points outside these bands are shown in red.
a) Front scan b) Back scan
Figure 61: Scans of fir tree test-piece; blue lines are nominal profiles, green lines are tolerance bands and red parts are deviations
The process characterisation work was extended to incorporate sub-surface condition in waterjet surfaces. Experiments were carried out on Inconel 718 test-pieces using a Design of Experiments approach (Rechtschaffner design). Four critical process parameters were considered: traverse rate, abrasive mass flow, abrasive grit and pressure.
Measurements of surface roughness, taper angle and most important sub-surface deformation (plastic deformation and crater depth) were taken. Mathematical models were produced relating process parameters to level of deformation. Figure 62 presents the models for maximum level of plastic deformation.
The main effect controlling the level of surface deformation is the abrasive grit and mass rate. The interactions between abrasive and pressure, traverse rate and mass flow are significant. Increasing the mass flow, abrasive size reduces the maximum level of deformation. Pressure has a non-linear behaviour and needs to be optimised to reduce the level of deformation.
Figure 62: Contour plot for maximum level of deformation, µm
A model for the maximum level of crater depth is shown in Figure 63. The main effect controlling the level of depth is the abrasive mass flow followed by its interaction with pressure and abrasive type. Increasing the level of abrasive mass flow and traverse speed decreases the depth of craters; decreasing the pressure decreases crater depth.
Figure 63: Contour plot for maximum level of crater depth, µm
The level of sub-surface deformation and crater depth has been quantified and related to process parameters. These need to be optimised in order to minimise the effect of parameters on sub-surface condition. SEM work is required to quantify the level of material embedment.
USFD also investigated whether ‘washing’ the machined surfaces with the waterjet (no abrasive) has a cleaning effect on the sub-surfaces, to remove any embedded particles and ensure that the part could be finished on the WEDM. Experiments were carried out on Inconel 718 samples. In preliminary trials, a couple of samples were washed with fixed parameters (water pressure, traverse speed and jet offset). No significant effect was observed in crater depth, see Figure 64.
Figure 64: Effect of washing cuts on maximum crater depth for different grit sizes
Process Monitoring
There are several different sensing methods which could be applied to provide an input to an adaptive control system for AWJ cutting. A literature review was carried out into sensors which could be used for process monitoring. Investigations into three candidate sensors were conducted; microphones, accelerometers and acoustic emissions sensors. These sensors were analysed for their applicability to the AWJM process.
Cuts were performed at parameters known to give unsatisfactory finishes and tolerances as well as those that give good results. For the microphone, the results showed little to no correlation between the signals being gained and the quality of the cuts performed. When using the Accelerometer, the frequencies present in the signal showed a pattern in one frequency where the magnitude increased and decreased as the parameters were changed. However there was lack of repeatability on identical cuts and the results did not justify building a monitoring system based around accelerometer sensing. After a number of trials using AE sensors, it was demonstrated that machining conditions could be identified in the signals. As a result of these initial trials, the control of the AWJM used in the ADMAP-GAS is based around AE monitoring.
Continuous sampling of the AE signal will provide the feedback from the AWJ system. The signal generated during machining is used to analyse the part remotely and give vital information if losses in accuracy are detected. The monitoring provides information from the cutting process which can be fed into a signal analyser to calculate the adjustment in feed rate required to increase the machining accuracy. This is then fed back into the program and overrides the feed rate stated.
The output from the sensor feeds back into the optimal parameter generator. The information gained is analysed and put through a fuzzy logic system. This system has been developed to determine the accuracy of the AWJM process from information detected in the AE signal. If change is required, a new feed rate is generated, overriding the set feed rate.
Figure 65: System flow diagram for process monitoring
The flow diagram (Figure 65) shows the output from the sensor being fed back into the optimal parameter generator. This updates the parameters which led to poor cuts. Each time a new part is cut, more information is fed into the system, updating it and increasing the accuracy of the parameters provided.
This system will not improve the best-cut capabilities of the AWJM, however it does ensure the best process is applied to a component so that the best possible cut can be achieved each time.
The signal from the AE sensor was investigated. A fast fourier transform was used to analsye the frequencies present. The patterns from the different signals were used to detect variations in the cutting process.
Analysis of the frequency domain from the signals was carried out. The signals show clear patterns. To define these patterns, features such as peak frequency were analysed and compared with feed rate. There are two useful peaks between approximatly 50 kHz and 200 kHz. As the feed rate is adjusted, these peaks shift; at a higher feedrate the peaks have a higher frequency. In addition to the peak frequency shift there are other noticable differences in the frequency response from the signal for different feed rates. Between 200 kHz and 600 kHz there are smaller peaks which vary in size.
Each signal analysed had one or two large peak frequencies present. The largest peak frequency was analysed by measuring the area under a section of the curve 10 KHz to either side of the peak. Figure 66 shows how the different feed rates resulted in noticeable variations in the maximum frequency area.
Figure 66: Relationship between the feed rate and the peak frequency
The second pattern is the change in magnitude of the three smaller frequencies in the range of 200 – 400 kHz. It was noted that a cut which resulted in an out of tolerance feature also reduced the magnitude at these smaller frequency peaks. Figure 67 shows how the feed rate affects the magnitude of these peaks. As with the maximum peak, these have also been normalised by the distance from the sensor.
Figure 67: Graph showing relationship between the areas under the triple peak frequencies
The third pattern identified was that of the RMS of the frequency response from the AE signal. The relationship is not as strong as the previous two, however it can supply a third monitoring factor to ensure that the signal is correctly analysed.
It was found that with the introduction of a database for the target component application, geometric accuracy could be predicted from the AE measurements. This allows the system to detect whether the AWJM cut is within the target tolerance band of the root slot feature.
Optimisation of the signal analyser was carried out. This optimisation looked at the initial relationships between the cutting process and component accuracy and refined them through additional experiments. Four test pieces, as shown in Figure 68, were machined using different traverse rates for each of the slots. This resulted in slots of varying accuracy. The results were measured and the relationship between the traverse rate and the signal form the AE sensor was analysed as described previously (Figure 66 and Figure 67) Figure 67 during the initial system development. This second set of experiments gave a much more accurate relationship between the traverse rate and accuracy as more traverse rates were tested with a greater number of repeated results.
Figure 68: Firtree test piece.
These experiments used the system to predict process accuracy and compared this to the measured results from traditional methods. This information allowed the fuzzy logic analyser to become more robust. The main functionality of the system remains the same as previously demonstrated. This work concluded the development of the process monitoring system.
The process monitoring system was fully developed in the previous periods (project month 32 - 36). Validation of the system is discussed in Section Evaluation of controller.
Development of control strategies
The control of the waterjet tool path can be used to compensate for the known errors in the system. In addition, feed rate regulation ensures the optimal cutting parameters are used throughout the complete tool path. Figure 69 shows how these values vary for an example tool path. In a root form slot, the variation in parameters as the jet travels around the tight radii is more extreme than in this example. It is because of this relationship that changes in the feed rate playing a large role in the tolerances that an AWJM cut part achieves. Details can be found in D3.3.
Figure 69: Example of Curvature for a Sinusoidal Tool Path.
An optimal parameter generator was developed. This system used a rate of curvature algorithm to assess the rate of change of the linear feed. This allows tool paths to be created which are more accurate than is currently possible with state of the art machines. The aim of this was to apply optimal feed rates to the different features found in the fir tree profile.
The system calculates an optimal feed rate which is based on the curvature and the rate of change of curvature of the geometry at each point. The system was developed to detect the following three situations and to analyse the feed rate:
• Zero Curvature: Areas of zero curvature such as the short straights on each flank and at the base of the profile allow a constant and optimised feed;
• Entrance/Exit into Corner: These areas have zero curvature just before the radius and a constant curvature once in the radius. Between these two points, an instantaneous change in the curvature occurs. The instantaneous change results in a sudden acceleration of the jet and a change of direction. It is this area which results in the worst loss of accuracy from the jet;
• Main Radii: These have a constant curvature value and as a result, zero rate of curvature. This means a constant feed rate can be applied as the acceleration of the jet is constant.
A look-ahead function was also applied. This is done to compensate for the acceleration of the machine and analyses the tool path and applies the adjusted feed rates in advance.
Experiments were carried out to characterise variations in the cutting jet. This was done indirectly by cutting slots into a test piece through a 0.5 mm nozzle, where a linear relationship was assumed to exist between the slot width produced and the jet size.
Figure 70: Jet Variation Results
As can be seen from Figure 70, the average slot width (and therefore jet size) centres around the orifice diameter of 0.5 mm. However the range of the results shows a large distribution. The variation in the jet width is from 0.4 mm - 0.69 mm, a spread of 0.39 mm. It is necessary to compensate for this variation if the AWJM is to continually achieve tolerances of +/- 0.3 mm.
The monitoring and control system was tested on the WARDjet ZX-613 at the AMRC. For each experiment, a full test piece, as defined in D3.1 was machined, as described in D3.3.
It was shown that increases in the accuracy could be achieved using the optimal parameters along the tool path. This increase in accuracy is different for the different features and is summarised in Table 9.
Table 9: Improvements to the state-of-the-art accuracies achieved by the controller
The experiments conducted previously showed potential for further improvements. The average error seen was higher than the nominal measurement. To compensate for this and to improve the accuracy of the system, additional experiments were carried out. The experiments consisted of cutting simple representative geometries of firtree profiles as shown in Figure 68. Four of these test pieces were machined. The parameters for the machining were different for each radius on the firtree profiles. The firtree profiles were measured after machining; this was used to link the accuracy achieved to the parameters used.
The resultant relationships between parameters and accuracy were used in the control system to enhance the parameter generation algorithms. The results of these additional experiments were used to improve the controller’s accuracy by applying optimised parameters for features to the firtree profile. The process monitoring system was used to measure test pieces machined on the AWJM. These results were fed into the control system to improve the self-learning algorithms. The system was validated by cutting a final test piece to determine the accuracy achievable.
The system was modified to accommodate the new 5-axis machine implemented at USFD. Modifications were made by calibrating the new nozzle for use on the 5-axis machine (as described previously for the 3-axis machine). Test cuts were made for known dimensions, measured, and the error of the machine was calculated. This error was then applied to the system as a correction.
Once the nozzle on this new machine was calibrated, the test piece shown in Figure 68 was machined to confirm the accuracy; the results demonstrated that the implementation of the controller with the 5-axis machine matched the results obtained on the 3-axis machine. This confirmed that the controller was correctly integrated.
Evaluation of Results
Evaluation of PAW system
Comparing the conventional AWJ system to the theoretical PAW, machining uncertainty, Ui, has been significantly reduced, as demonstrated in Table 10.
Table 10: Machining uncertainty comparison
Evaluation of controller
A Demonstration component was then machined using the new 5-axis machine at USFD. The fir trees were measured and the results are shown in Table 11.
Table 11: Fir tree tolerances achieved using the ADMAP-GAS controller, precision table and the 5-axis machine
The demo piece was a stage model of the final part with a constant stock level of 60 microns on all surfaces. A stage model is representative of a roughing process, with the resultant part to be finished on a secondary process, in this case, WEDM. Figure 71 shows the stage model drawing to which this final demonstration piece was machined.
Figure 71: Drawing of fir tree part machined for demonstration
Table 11 shows that by using the controller and the precision system developed in T3.2.1 and T3.2.2 but not the nozzle, the accuracy of the water jet has been improved by 61% for radii on fir tree profiles. Previously a tolerance of +/- 0.235 mm was possible; the new system delivers +/- 0.09mm. The improvement in accuracy in the pressure flank is 22%. These improvements do not reach the requirements for finishing the fir trees using a single EDM cut. With the addition of the PAW nozzle developed by USFD, as well as compensation of the linear taper by the 5-axis machine, it should be possible to achieve this target.
During the development phase, the control system was tested on the Wardjet machine at USFD. Implementation on multiple platforms is part of AMRC’s commercialisation strategy.
This system has been developed as a standalone package which can be integrated with industry-standard CADCAM packages which interface with AWJ machines. After the project, the system requires further testing before it can be developed into a commercial piece of software. This will be carried out during the first year and AML will deploy the new controller within their manufacturing department.
Environmental Impact of the Abrasive WaterJet Machining
Ecoprofile data and assumptions
The ecoprofile of the developed AWJM process has been prepared on the basis of the experimental data produced in the WP3 and WP4 cutting tests. Additional inputs have been received from WardJet, who manufacture the machine used by USFD, in particular the data on the functional life of the waterjet machine consumable parts (filters, orifice, focusing tube) and state-of-the-art methods for recycling water and abrasive. In term of process inputs and outputs, the AWJM roughing and the AWJM finishing are very similar, therefore the considerations reported here are valid for both the processes.
The most relevant process data are summarized below:
• The measured cutting time per fir tree is 5 minutes. This value has been obtained from real cutting tests using the selected fir tree geometry and turbine disk thickness.
• The energy consumption when cutting one fir tree is 2.708 kWh. It is important to observe that the energy for roughing and finishing is assumed to be the same, because it depends largely on the pump pressure, which is 6000 bar in both cases. The compressed air consumption is negligible.
• The abrasive adopted is a garnet (Ca3Fe2(SiO4)3), mesh 80 corresponding to a particle diameter of 300-150 µm, and a hardness which is 6.5 - 7.5 on the Mohs scale. The abrasive rate is 0.55 (kg/min), corresponding to a consumption of 2.5 kg of garnet per fir tree: 1.25 kg is virgin abrasive, whereas 1.25 kg is recycled abrasive.
• With the ADMA-GAS developments, the maximum the softened water flow rate (while cutting) is 4.163 kg/min, corresponding to a maximum water requirement of 20.819 kg per fir tree. After cutting, the water is recovered and recycled.
Some maintenance consumables have been considered for the waterjet machine:
• The filters are made in polypropylene (130 g). The functional life of the filters depends on many factors, but an average value of 2000 h can be used.
• The focusing tube is made of boron carbide (43 g) and lasts 100 h.
• The orifice is made of ruby or diamond; a diamond orifice is guaranteed for 500 h, whereas a ruby orifice should last about 80 h. In the experiments a ruby orifice has been used (3 g).
AWJM of fir trees produces the by-products and waste described below:
• Two accessory systems are available for recycling the water and the abrasive: the first system extracts the garnet from the cutting fluid, the second one recovers, filters and treats the water. Almost all the garnet is recovered, but it can be only partially reused for the fir tree cutting because the particles break when cutting Inconel, reducing their mesh. For this reason, only 50% – 60% of the garnet can be reused directly, while the remaining amount is only suitable for cutting softer materials; this garnet at reduced mesh size is considered in the LCA as a by-product of the AWJM. When the garnet has been pulverized and it is too fine to reuse for waterjet cutting, it can sometimes be used in other industries. There have been cases where people have sold their used abrasive for use in road beds. Other uses could potentially be found: as long as the material that is being cut is not hazardous, options like this could be viable. All the water recovered can be recycled. It has been assumed that there is negligible refill due to evaporation or component wetting. A simplified LCA calculation of recycling systems has been done, considering the energy consumption (0.183 kWh per fir tree) and neglecting the consumption of compressed air and consumables.
• Unlike broaching, the AWJM does not produce a mixture of oil and different metal chips, but a monolithic scrap of Inconel 718. This can be directly recycled by the foundries without any treatment, as it not polluted.
• The boron carbide, the ruby and the polypropylene from the end of life of the focusing tube, orifice and filters, can be sold as by-products and recycled by specialized industries.
Life Cycle Inventory
Adopting one turbine disk with 48 fir trees as the functional unit, a Life Cycle Inventory (LCI) of the eco profile “AWJM single cut” has been prepared, describing the significant inputs and outputs. The LCI data are summarized in Figure 72.
Figure 72: Life Cycle Inventory for the Ecoprofile “AWJM single cut”
Impact Assessment and Interpretation
The data collected in the Life Cycle Inventory have been elaborated using SimaPro7 Analyst software (PRé Consultants), based on Ecoinvent database and customized with DIAD’s databases. The Ecoinvent database represents the result of a large effort by Swiss institutes to update and integrate the well-known ETH-ESU 96, BUWAL250 and several other databases.
The LCA has been carried out, considering a 48 slot fir tree and adopting the LCA method CLM 2 as the baseline method conforming to ISO 14000. According to this method the main impact categories considered are:
Climate change - Climate change can result in adverse effects upon ecosystem health, human health and material welfare. Climate change is related to emissions of greenhouse gases to air. The characterization model as developed by the Intergovernmental Panel on Climate Change (IPCC) is selected for development of characterization factors. Factors are expressed as Global Warming Potential for time horizon 100 years (GWP100), in kg carbon dioxide/kg emission. The geographic scope of this indicator is at global scale. Some characterization factors were added from the IPCC 2001 GWP 100a method: Methane, bromodifluoro, Halon 1201, Methane, dichlorofluoro, HCFC-21, and Methane, iodotrifluoro. Stratospheric Ozone depletion: the characterisation model is developed by the World Meteorological Organisation (WMO) and defines ozone depletion potential of different gases (kg CFC-11 equivalent/ kg emission).
Stratospheric Ozone depletion - Because of stratospheric ozone depletion, a larger fraction of UV-B radiation reaches the earth surface. This can have harmful effects upon human health, animal health, terrestrial and aquatic ecosystems, biochemical cycles and on materials. This category is output-related and at global scale. The characterization model is developed by the World Meteorological Organisation (WMO) and defines ozone depletion potential of different gasses (kg CFC-11 equivalent/ kg emission). The geographic scope of this indicator is at global scale. The time span is infinity.
Acidification - Acidifying substances cause a wide range of impacts on soil, ground-water, surface water, organisms, ecosystems and materials (buildings). Acidification Potentials (AP) for emissions to air are calculated with the adapted RAINS 10 model, describing the fate and deposition of acidifying substances. AP is expressed as kg SO2 equivalents/ kg emission. The time span is eternity and the geographical scale varies between local scale and continental scale. Characterization factors including fate were used when available. When not available, the factors excluding fate were used (In the CML baseline version only factors including fate were used). The meth-od was extended for Nitric Acid, soil, water and air; Sulphuric acid, water; Sulphur trioxide, air; Hydrogen chloride, water, soil; Hydrogen fluoride, water, soil; Phosphoric acid, water, soil; Hydrogen sulphide, soil, all not including fate. Nitric oxide, air (is nitrogen monoxide) was added including fate.
Human toxicity - This category concerns effects of toxic substances on the human environment. Health risks of exposure in the working environment are not included. Characterization factors, Human Toxicity Potentials (HTP), are calculated with USES-LCA, describing fate, exposure and effects of toxic substances for an infinite time horizon. For each toxic substance HTP’s are expressed as 1,4-dichlorobenzene equivalents/ kg emission. The geographic scope of this indicator determines on the fate of a substance and can vary between local and global scale.
Figure 73 presents the result of the LCA analysis. In particular, the energy consumption results as the main factor in all the impact categories of AWJM, except for the fresh water aquatic eco toxicity, where the consumption of abrasive garnet also has a considerable impact.
Figure 73: Characterization of impact assessment of the AWJM single cut, where red is the abrasive garnet, light blue is the focusing tube, dark blue is the electricity mix, yellow are filters and turquoise of the orifice
Figure 73 shows that energy consumption is the main factor in all the impact categories of AWJM, except for the fresh water aquatic eco toxicity, where the consumption of abrasive garnet also has a considerable impact. To reduce the environmental impact of AWJM, it will be important to reduce the energy consumption which depends on the pump power, cutting time and water-garnet recycling systems. Table 12 shows the values of the estimated environmental impacts of AWJM roughing.
In order to have a more general view of the environmental impact of AWJM, it is important to not focus only on the process itself, but to extend the boundaries to the integrated W2 manufacturing chain. Figure 74 shows the impact of the integration of AWJM roughing and WEDM finishing is shown. The impact of WEDM is higher than AWJM for all the categories considered, in particular for fresh water and marine aquatic toxicity. In these categories the incidence of the production of the copper/zinc wire is very high.
Table 12: Characterization of impact assessment of the AWJM single cut: detailed data table
Figure 74: Characterization of impact assessment of the Ecoprofile “AWJM roughing + WEDM finishing”, where blue is AWJM and green is WEDM
It is important to remember that the WEDM process is composed of 3 separate cuts (2 precision cuts and 1 surfacing cut) to compensate for the taper angle and abrasive contamination from the AWJM. Moreover the environmental impact of WEDM and the W2 process chain would be significantly reduced by increasing the precision of the AWJM. Table 13 gives detailed values for the estimated environmental impacts of the integrated process AWJM roughing + WEDM finishing.
Table 13: Characterization of impact assessment of the Ecoprofile “AWJM roughing + WEDM finishing”: detailed data table
The Life Cycle Assessment carried out was focused on the AWJM cutting and in particular the manufacture of 48 fir trees for an aeroengine turbine disk. On the basis of the data collected and assumptions made, it has been observed that:
• Energy consumption is the main factor in all the impact categories of AWJM, except on the fresh water aquatic ecotoxicity, where the consumption of abrasive garnet has a relevant impact.
• The energy consumption of AWJM depends mainly on 3 elements:
o the pump power,
o the cutting time,
o the recycling systems (water and abrasive),
• The abrasive impact is mainly seen in energy use for recovering and recycling.
These are the main factors to be taken into account for reducing the environmental impact of AWJM (more details are documented in WP4 summary). Considering the integration of the AWJM in a process chain, it is clear that its precision allows a reduction of finishing operations, therefore:
• The quality of the AWJM is an additional factor relevant for the reduction of the environmental impact of the overall chain.
• The LCA must be used to define, case by case, the best environmental compromise between the cutting time and the cutting precision.
Potential Impact:
Within this section of the final report the potential impact and main dissemination activities and exploitable results are presented. Based on an ADMAP-GAS demonstrator geometry, which has been defined during the project, three potential new process chains are introduced:
1. Wire-EDM roughing and finishing
2. AWJM roughing and finishing
3. AWJM roughing and Wire-EDM finishing
Data which have been measured during the production of the demonstrators are used to set up a Life Cycle Cost Analysis and a Life Cycle Assessment to compare the new process chains and especially the new developed processes with the state of the art broaching process. Here, costs, the environmental impact and social aspects are the main assessment criteria. Finally, the main dissemination activities in terms of publications and the exploitable results of the partners are summarized.
Definition of the section of turbine disk
The demonstration of turbine disk manufacturing has been done on a quarter of turbine disk to save costs and time. The slot profile adopted does not correspond to any drawing of specific aircraft engine, but it has been conceived as an integration of critical geometries and requirements from different aero-engines manufacturers (Rolls Royce, MTU, Avio). In Figure 75 the design of the demonstrator is presented.
Figure 75: Design of the section of turbine disk demonstrator
Demonstration of turbine disk manufacturing
In Figure 76 the demonstration of AWJM fir tree manufacturing is shown. The measurements carried out show that a tolerance of +/- 20 μm has been achieved for the geometries, which was the goal to finally finish the profile by WEDM.
Figure 76: AWJM cutting of fir trees on the section of turbine disk
The main conclusions are:
• It has been demonstrated that AWJM fulfils the quality requirements for an efficient integration with WEDM finishing.
• At the present, the nozzle geometry limits the value of radii that can be cut, but shortly the new nozzle geometry will overtake this limit.
• Further improvement of AWJM quality is expected from 5 axis machines with postprocessor compensating tapper angle and new nozzle increasing the cutting power. This will reduce the WEDM finishing cycle, resulting in a significant increase of productivity and a decrease of energy consumption (costs and environmental impact).
• Tolerances of +/- 5 μm can be achieved with the precision table and precision head (PAW - Precision Abrasive Waterjet).
• A limiting factor for water-jetting of a fir tree to a finishing stage is the material embedment. In fact the turbine disk is contaminated by abrasive particles. It is expected to reduce this effect by the adoption of the new nozzle geometry and optimised cutting.
Figure 77 shows different phases of fir trees cutting by WEDM including positioning of the demonstrator on the WEDM machine working area, rough / precision cut and surfacing cut:
Figure 77: WEDM cutting of fir trees in the section of turbine disk
The conclusions of this approach are:
• It has been demonstrated that WEDM fulfils the quality requirements for the substitution of broaching,
• WEDM can be applied both as stand-alone slotting or it can be integrated as finishing of AWJM slotting and
• the integration with AWJM slotting can increase the productivity with respect to WEDM, but this aspect is strictly dependent to the AWJM precision.
LCC-modelling of complete turbine disk manufacture
For each process combination, three phases of development have been considered during a LCC analysis. The phases correspond to a different degree of development of technologies: Intermediate results, final achievements and further developments after the end of the project:
• Phase 1: Achieved progress at Month 30 (phase 1)
• Phase 2: Achieved progress at month 42 (phase 2)
• Phase 3: Expected progress month 54 – 12 months past the project (phase 3).
AWJM fir tree manufacturing
The cutting time is average 5 min per fir tree. The different phases of AWJM development are focused on the improvement of the accuracy of cutting and not on the cutting rate, because the geometry and surface quality are critical, whereas the cutting time is already comparable to broaching.
• Phase 1: The initial quality of AWJM was not sufficient, as it was completely impossible to achieve the required tolerance range for fir tree geometry (+/- 5 μm) and the surface roughness was outside the required specification up to a factor of 14.
• Phase 2: The optimisation of the AWJM and the integration of a precision table allowed achieving a tolerance of +/- 20 μm. This value is sufficient for the integration with WEDM finishing, but it is still not enough for the substitution of broaching. The nozzle geometry (0.3 mm) limits the value of radii that can be cut.
• Phase 3: In next industrialisation phase, the adoption of the new nozzle geometry (0.1 mm) will allow to machine all the radii of the fir tree, also increasing the cutting power. A reduction of about 40% of energy consumption is expected. The implementation of the 5 axis precision head with postprocessor compensating of the tapper angle will constitute the further step in the technology. PAW will then achieve a geometrical tolerance of +/- 5 μm which is finally comparable to broaching.
WEDM fir tree manufacturing
The optimized WEDM slotting process chain is composed of the sequence of three cuts: “E2”, “E7” and “E21”. “E2” is a rough cut, “E7” is a precision cut and “E21” is a surfacing cut.
• Phase 1: The initial mean cutting speed of the sequence was 0.99 mm/min, corresponding to a cutting time per slot around 78 min.
• Phase 2: Thanks to further optimisation of the rough cut, the cutting speed has been increased to 1.8 mm/min, corresponding to a cutting time of 43.3 min per slot.
• Phase 3: Further increases of cutting rate are expected from next industrialisation of the process, but at the present it is not possible to quantify this. As example, an option would be the modification of the cut settings, achieving a reduction of the sequence.
AWJM + WEDM fir tree manufacturing
• Phase 1: Surface quality after AWJM necessitates the WEDM sequence “E4”, “E4”, “E7” and “E21” resulting in a mean WEDM speed of only 0.8 mm/min. The cutting time is therefore 97 min per fir tree slot in average. In this condition the integrated process shows a productivity lower that the WEDM slotting.
• Phase 2: The latest cutting trials indicate that it is possible to only use a WEDM cutting sequence “E7” and “E21”, resulting in a mean WEDM speed of 3.4 mm/min. The cutting time is then 22.9 min per fir tree in average.
• Phase 3: Better AWJM process control (5-axis machine, AMRC post processor and monitoring) allows the elimination of the “E7” cut, leaving only one ”E21” cut. In this case the cutting speed reaches 8.8 mm/min, resulting in an average cutting time per slot of 9 min. The energy consumption of the AWJM process is considerably reduced by the application of the new nozzle geometry.
Life Cycle Cost analysis
In Figure 78 the Life Cycle Costs are presented of the different W2 process chains proposed. The analysis clearly shows that the LCC/part of the integrated process AWJM+WEDM will become competitive against the state of the art broaching process within the phase 3 (estimated 12 months after the project finish).
Figure 78: LCC of different slotting processes for a turbine disk (48 fir trees)
Life Cycle Assessment
In Figure 79 the evolution of the impact of W2 processes on the global warming (GWP100) and in comparison with broaching process impact (considered constant) is presented. Similar results have been obtained for all the environmental impact categories of CLM 2 baseline method (conform to ISO 14000). All the details concerning the LCA methodology, inventory data and assumptions are reported in the deliverable D4.2. It can be concluded that:
• All the developed W2 processes achieved a reduction of environmental impact with respect to broaching.
• WEDM fir tree manufacturing is the technology with higher degree of readiness at the end of the project and this is reflected by the corresponding low environmental impact.
• AWJM slotting will express higher environmental benefits after the end of the project, when all the technologies developed will be fully implemented.
• AWJM roughing + WEDM finishing, as summary of the improvements of both of the technologies, is expected to represent at the month 54 the most environmental friendly solution for the substitution of broaching.
Figure 79: Evolution of the impact of W2 processes (GWP100) with respect to broaching
Adopting the “Damages Evaluation” of Eco-indicator 99 method, it is possible to obtain a comparison of the different turbine disk slotting processes more oriented to social aspects. In particular in Figure 80 the evolution of the human health damage of developed slotting processes is presented. The diagrams clearly show that all the W2 processes have achieved a significant reduction of the impact on human health at the end of the project and further improvements are expected by the month 54.
Figure 80: Evolution of the impact of W2 processes on Human Health respect to broaching
Assessment of W2 fir tree manufacturing respect to broaching
On the basis of the demonstrations done, in Figure 81, Figure 82 and Figure 83 the achievements of W2 slotting technologies with respect to state of the art broaching are summarised.
Figure 81: Achievements of W2 slotting with respect to broaching
Figure 82: Assessment of the application of W2 technologies for the fabrication of turbine disks
Figure 83: Assessment of the application of W2 technologies for the fabrication of turbine disks
Publications
• Klocke, F.; Welling, D.; Veselovac, D.; Perez, R.; Nöthe, T.: Wire-EDM developments for manufacturing safety critical turbine components features in Inconel 718, Veranstaltung: "6th Int. Conference “Supply on the wings” Aerospace – The global innovation driver". Frankfurt, 04.11.2011. 2011.
• Klocke, F.; Zeis, M.; Welling, D.; Garzon, M.; Klink, A.; Veselovac, D.: Aktuelle Forschungsschwerpunkte im Bereich Funkenerosion am Werkzeugmschinenlabor WZL der RWTH Aachen, in: Der Schnitt- & Stanzwerkzeugbau 18 (2011), 5, S. 8-14.
• Klocke, F.; Zeis, M.; Welling, D.; Garzon, M.; Klink, A.; Veselovac, D.: Aktuelle Forschungsschwerpunkte im Bereich Funkenerosion am Werkzeugmaschinenlabor WZL der RWTH Aachen, in: Der Stahlformenbauer 28 (2011), 5, S. 6 – 12.
• Klocke, F.; Welling, D.; Dieckmann, J.: Comparison of grinding and Wire EDM concerning fatigue strength and surface integrity of machined Ti6Al4V components, in: Procedia Eng. 3 (2011), 19, ISSN 1877-7058, S. 184–189.
• Welling, D.: Drahtfunkenerosion im Turbinenbau, in: Tagungsband zur 8. Fachtagung Funkenerosion, Hrsg.: Klocke, F., Apprimus Verlag/ WZL-Forum Aachen 2011, S. Kap. 7, S 1 – 25.
• Klocke, F.; Zeis, M.; Klink, A.; Welling, D.; Baumgärtner, M.: Technological and Economical Comparison of Pre-manufacturing Titanium- and Nickel-based Blisks via Milling from solid and EDM, Veranstaltung: "EUCOMAS 2012". Hamburg, 07.-08.02.2012. 2012.
• Klocke, F.; Welling, D.; Dieckmann, J.; Klink, A.: Titanium Parts for Medical Sector Made By Wire-EDM, in: Proceedings of the 1st International Conference on Design and Processes for Medical Devices: Padenghe sul Garda (Brescia), Italy, 2 - 4 May 2012 , Hrsg.: Ceretti, E.; Fiorentino, A.; Giorleo, L.; Giardini, C., Neos Edizioni Brescia/ Italy 2012, ISBN 9788866080589, S. 163-166.
• Klocke, F.; Welling, D.; Dieckmann, J.; Veselovac, D.; Perez, R.: Developments in Wire-EDM for the manufacturing of fir tree slots in turbine discs made of Inconel 718, in: Key Engineering Materials 1665 (2012), 504-506, ISSN 1013-9826, S. 1177-1182.
• Trauth, D.; Klocke, F.; Welling, D.: Auswirkungen des elektro-mechanischen Festklopfens auf den Randschichtzustand und die Dauerfestigkeit einer Nickel-Basis-Legierung, Veranstaltung: "53. DGM Fachauschusstagung "Mechanische Oberflächenbehandlungen". Rastatt/Malsch, 18.04.2013. 2013.
• Escobar G, Gault RS & Ridgway K, 2011, “Characterisation of water-jet process for the production of fir trees in Inconel 718”, 6th International conference Supply on the Wings (AIRTEC), Frankfurt Germany, Nov 2-4.
• Escobar G, Gault RS & Ridgway K, 2012, “Characterisation of abrasive water-jet process for pocket milling in Inconel 718”, CIRP High Performance Cutting, Zurich 4-7 June.
• GA Escobar-Palafox, RS Gault, K Ridgway, 2012, “The Effect of Abrasive Water Jet Process Variables on Surface and Subsurface Condition of Inconel 718”, Advanced Materials Research vol 565, pp 351-356.
• Paper submitted to INTED 2012 - “Knowledge transfer: an EC funded research project case study”.
Conferences and presentations:
• 09.-10.10.2012 Presentation of ADMAP-GAS demonstrator at BRAGECRIM, Meeting at WZL Aachen (http://www.bragecrim.rwth-aachen.de/)
• 09.11.2012 Visit and getting insight into ADMAP-GAS project for Mechanical Engineering students (150 students) of RWTH Aachen University
• 28.-29.11.2012 Discussion about ADMAP-GAS results at MTU
• 20.-21.02.2013 Presentation of ADMAP-GAS demonstrator at ICTM in Aachen (http://www.ictm-aachen.com/)
• ESAFORM Conference 2012 – Title: Developments in Wire-EDM for the manufacturing of fir tree slots in turbine discs made of Inconel 718, 09.2012
Activities in 2011:
• ICTM at WZL/IPT RWTH Aachen University 02.2011
• Presentation at Aerodays, Madrid 03.2011
• AWK at WZL RWTH Aachen University 05.2011
• Presentation at the Supply on the Wings in Franktfurt 11.2011
• Presentation at 8. Fachtagung Funkenerosion, WZL RWTH Aachen, 11/2011.
• Rolls-Royce waterjet Community of Practice meeting, University of Sheffield, 2nd November 2012.
• Technical Fellows conference, University of Sheffield, 30th October 2012.
• Water jet Open Day, University of Sheffield, 30th and 31th January 2013, with accompanying article on the AMRC website. This event was reported by a number of publications, which also mentioned ADMAP_GAS.
Exploitation of results
Through the ADMAP-GAS project, 12 exploitable results have been developed. A summary of these results can be seen in Table 14. It is evident from the results that the exploitable results developed during the ADMAP-GAS project have the potential to impact many industries across the world. More pertinently, the refinement and adaptation of AWJC and HS-WEDM during the project could lead to large economic and environmental improvements in manufacturing processes.
• Table 14: List of exploitable results developed through the ADMAP-GAS project
Number Exploitable Result Timetable to Commercial Use Partners Involved
1 NEW FAMILY OF WEDM WIRES Sep-14 BEDRA, WZL,
AGIE-CHARMILLES
2 NEW WEDM MACHINE INTEGRATING INNOVATIVE MONITORING SYSTEM Jan-13 AGIE-CHARMILLES, WZL
3 NOZZLE GEOMETRY AND GENERATOT TECHNOLOGY Jan-13 AGIE-CHARMILLES, WZL
4 NEW BASE FLUID Mar-12 BEDRA,
AGIE-CHARMILLES, OELHELD, WZL, DIAD, TEKS
5 NEW ADDITIVES Mar-12 BEDRA,
AGIE-CHARMILLES, OELHELD, WZL, DIAD, TEKS
6 OPTIMIZED WATERJET MACHINE
(CUTTING HEAD AND TABLE) Nozzle: July-13; Head: Dec-13; Table: July-13 USFD
7 PROCESS MONITORING SUITE AWJM Mar-13 AMRC LTD
8 WEDM MONITORING SYSTEM AGIE-CHARMILLES, WZL
9 TEST BED DEVELOPMENT WEDM 5 years after project end AGIE-CHARMILLES
10 NEW PROCESS ROUTE AUTOMOTIVE SECTOR 1 year after project end AGIE-CHARMILLES, USFD, WZL, DIAD
11 NEW PROCESS ROUTE MEDICAL APPLICATION 1 year after project end AGIE-CHARMILLES, USFD, WZL, DIAD, TEKS
12 NEW PROCESS AEROSPACE APPLICATION Within 10 years after project end All project partners
List of Websites:
http://www.admapgas.com/
Project’s Coordination:
Organisation Name: WZL – RWTH Aachen
Representative Name: Dražen Veselovac
Tel: +49-241-80-27432
Fax: +49-241-80-22293
E-mail: d.veselovac@wzl.rwth-aachen.de
The European project ADMAP-GAS succeeded in presenting alternatives to the critical broaching process for the manufacturing of fir tree structures in gas turbine discs by the alternative processes of Abrasive Water Jet Cutting (AWJC) and High-Speed Wire-EDM (HS-WEDM). Eight partners effectually combined their efforts to develop robust manufacturing processes for the European engine manufacturing industry with the objective to drastically decrease failure risks and machining costs as well as preserve energy and environment through the project.
Until the last project month research and development work was carried out and successfully finished. Both developed manufacturing processes (AWJC and HS-WEDM) are able to enter into competition with the broaching process. This showed the evaluation process which was done during the last project months. It can be stated, that both manufacturing technologies underwent an improvement through the project. Especially cutting rates, surface integrity aspects and process monitoring systems have been increased or even completely new developed. The developed AWJC process for cutting fir tree structures was proven to be a manufacturing process for roughing operations which requires a very economic technology to process hard to machine alloys. Through a process monitoring system, a multi axis cutting head, increased accuracy and a new developed nozzle geometry (all objects have been developed within ADMAP-GAS) the manufacturing technology is on a very good way to be able to machine fir tree geometry profiles. With further effort and development this technology will be an alternative to broaching.
In terms of HS-WEDM several developments improved the standard manufacturing process to a competitive manufacturing technology to the state of the art broaching process. Improved cutting rates which have been achieved through technology optimisation, nozzle development and generator tuning together with very good surface integrities (minimal surface layer and contamination as well as no crack occurrences) indicates that the developed process is a serious alternative to broaching. On the one hand this was shown regarding economic aspects and on the other hand this was shown regarding functionality through fatigue tests.
To summarize the technical work done within ADMAP-GAS and to critically assess the project regarding the objectives of the complete project it can be stated, that the failure risks of engine discs have been decreased through the developed process monitoring systems. Additionally, the machining costs to produce a fir tree with the challenging requirements have been decreased by 25%. This was shown through a Life Cycle Cost analysis. Concerning the energy consumption and the environmental impact a Life Cycle Assessment showed an enormous decrease (up to 60%) by using one of the two developed processes. Furthermore it was shown, that the developed processes are able to be implemented into other industries like the motorsport sector.
Regarding the work concerning public relation dissemination and exploitation activities have been done. For the future a dissemination and exploitation plan has been set up to spread the very good project results after the project end. During the project end phase several turbine manufacturers like MTU, Rolls Royce and Pratt & Whitney contacted the consortium to receive detailed project results. All of them are very interested in the research which has been done during the project and they asked the partners for cooperation to evaluate the processes with their own standards.
Overall it can be summarized, that the project ADMAP-GAS has come to a very good end. Two special manufacturing processes to machine fir tree geometry profiles have been developed which are able to substitute the critical broaching process in the future.
Project Context and Objectives:
In the project ADMAP-GAS new “Unconventional Advanced Manufacturing Processes for Gas Engine Turbine Components” have been developed in order to substitute the critical conventional broaching process for fir tree profiles between rotating blades and discs.
Presentation of the project and consortium
The ADMAP-GAS project is a European collaborative project of the 7th Framework Programme within Theme 7, Transport (including Aeronautics). It is a three and a half year project (36 month + 6 month cost neutral prolongation) that started on August 2009 and will last until January 2013. The scientific and technological methodology applied in ADMAP-GAS is kept at the highest level due to the combined talents of the consortium partners (see Figure 1). Two world leading industrial partners (CHARMILLES and BEDRA), three SMEs (OELHELD, TEKS and DIAD), one research institution (AMRC) and two outstanding higher education partners (WZL-RWTH Aachen and University of Sheffield) from five European countries form the consortium.
Figure 1: Consortium of ADMAP-GAS
In order to provide examples of issues that ADMAP-GAS project is able to address, the basic information and problem allocation were defined by the industrial partners in correlation to their daily production needs and challenges. The proposed new unconventional processes in particular for ADMAP-GAS Wire-EDM and AWJM for the substitution of broaching will not only have a benefit for the turbine industry but also have a significant impact on other transport industries which need special manufacturing processes like the car sector.
Project objectives
Due to a high increase of market demand for gas engine turbines in the aircraft industry, efficient manufacturing processes for the turbine components are increasingly necessary. The rising standard for individual and adapted product solutions leads to an increased demand for efficient, reliable and additional flexible manufacturing technologies according to the turbine parts.
The ADMAP-GAS project develops new “Unconventional Advanced Manufacturing Processes for Gas-Engine Turbine Components” in order to substitute the critical conventional broaching process for fir tree profiles that connect rotating blades and discs. Especially the good thermal and mechanical properties of nickel-based super alloys which are used for disc material complicate the machining by conventional processes.
Therefore, the two manufacturing technologies “High Speed Wire-EDM Slotting” and “Abrasive Water Jet Slotting” (ADMAP-GAS Processes) have been developed, optimised and evaluated, highlighting their individual performance and effectiveness for the machining task.
Based on the achieved results ADMAP-GAS will further explore its possible impact on the manufacturing of blisk geometries. Here a large amount of expenses results from high tool costs due to high tool wear. Therefore in this project the process capabilities of fast roughing of the blisk structures with unconventional machining processes was also investigated to reduce the volume of material which has later to be machined by milling.
Project methodology and implementation of the work
In order to accomplish the project objectives and goals, an appropriate structure has been established. Figure 2 shows the holistic build-up of the work package structure, wherein all partners have their defined tasks. All work packages are surrounded by work package one the Project Management (WP1, MGT). WP1 is in this project essential in order to ensure the adequate communication between project partners and the correct check of the project’s progress. Furthermore in WP1 the project co-ordinator is positioned who is responsible for the communication with the European Commission (EC). The main work on Research and Technical Development (RTD) is done inside the work packages with the titles Wire-EDM Process Development (WP2) and Water-Jet Process Development (WP3). The main tasks herein address the research to achieve the specific objectives of the two unconventional processes. The special turbine manufacturing requirements which can be seen as the standards to be reached by the core project work (WP2+WP3) are gathered within work package 4 (WP4). Furthermore WP4 has the function to design a technique to integrate the new processes into the existing process chain of disc manufacturing. This action includes the demonstration (DEM) of the developed ADMAP-GAS demonstrator set-ups which show the developments and improvements achieved in the project to the industrial target group. Work package five (WP5) is responsible to incorporate the exploitation and dissemination work of the consortium. Furthermore the construction of a web-based internet platform is the content of WP5.
Figure 2: ADMAP-GAS work package structure
ADMAP-GAS periods and tasks
ADMAP-GAS is a 36 + 6 month project and is divided into two periods. The tasks of each period are as following:
• First period (month 1 - 18): Analysis of current broaching operation and development of special unconventional processes for fir tree production.
• Second period (month 19 – 36 + 6): Optimisation and monitoring of developed processes and evaluation of unconventional processes regarding fir tree production.
Project Results:
The main scientific and technical results respectively foregrounds of ADMAP-GAS were gained within the research and development work of the two processes AWJC and HS-WEDM. This work was mainly done in the work package 2 and 3. Following the work done is summarized.
WP 2: Development of “High Speed Wire-EDM Slotting”
WP 2 mainly focusses on the development of Wire-EDM process for fir tree production. Charmilles as a machine tool manufacturer, Bedra as a wire manufacturer, oelheld as a supplier of dielectrics and the WZL as a research organisation are forming the main partners of this WP dealing with process technology development. In addition DIAD and TEKS provide information from the turbine industries in terms of dimensional and surface integrity requirements and evaluation criteria to compare the “High Speed Wire-EDM Slotting” against standard (broaching) machining operations. The following major objectives have been set up before the start of the project:
• Increase the cutting rate to reduce machining time by 30% compared to state-of-the-art Wire-EDM.
• Maintain a sufficient parallelism in the machined geometry of the parts at targeted cutting rates.
• Minimise heat affected zone, so that the demands of the cut parts regarding fatigue strength are met.
• Achieve a surface roughness smaller than Ra = 0.8 µm.
• Develop wire to allow reliable threading and workpiece touch-off.
The main research task within WP2 is the Task 2.2 “Development of High Speed Wire-EDM”. To achieve the overall WP objectives these subtasks are dealing with developments of machine tools, wire electrodes, dielectrics and process monitoring systems.
Development of new wires
The main procedure determined for the development of new wire types is depicted in Figure 3. Prior to wire design itself the detailed performance targets and requirements had to be rendered. This included a basic alignment with the design or use of test machines and dielectrics. In addition procedures for testing the EDM wires needed to be defined. Wire design followed two main routes according to two different material groups, nickel and titanium based alloys, considered in the project. Main focus, however, was dedicated to nickel based alloys (Inconel 718).
Figure 3: Wire development process
In a first step a preliminary wire specification was derived from the detailed process conditions and requirements. In Table 1 the specifications and requirements are listed.
New wire design to avoid undesired contamination
Strategies for avoidance or minimisation of contamination of the wire cut surface were developed (see Figure 4). Contamination by wire material is a critical issue since the deposition of zinc can lead to the formation of brittle phases in the Inconel material which can dramatically downgrade the material performance. The selected strategies are marked in grey. In general the approach combines a wire with an outermost nickel coating and a cutting technology adapted to the new wire design. The use of oil-based dielectrics may also support the minimisation of contamination. This has not been tested so far.
Table 1: Wire specification derived from process conditions and requirements
Figure 4: Strategies for avoidance of undesired work piece contamination
In order to achieve both, high productivity and avoidance of contamination the new wire design features a first layer of zinc rich brass and an outermost layer of pure nickel (see Figure 5). The function of this layer is to generate only such contamination which is already a parent material of Inconel. Since nickel is rather unsuitable for WEDM in terms of cutting speed the challenge is to keep the thickness of the outermost layer as small as possible. Here the interaction of thickness and cutting technology i.e. number of cuts and machining parameters is crucial. For the layer providing high speed two main alternatives beta-brass and gamma-brass were chosen. Usually, the thickness of gamma-brass layers can be significantly thinner compared to beta brass layers while achieving the similar cutting speeds. Thus a gamma phase layers may provide a more economic solution. Core material for type AG-N2 is brass due to better availability of the pre-material.
Besides the wire design of type AG-N3 which was chosen with focus on Charmilles machines two further designs were developed to supply wire solutions for the whole range of conceivable requirements and conditions. One wire type consists of a standard brass core (CuZn36) and a 1.5 – 3.0 µm thick nickel layer (type AG-N4). Due to the smoother surface of the uncoated brass wire it requires less effort to generate a homogeneous nickel layer. Moreover, the overall production effort is less compared to type AG-N3. Application of such wire is favourable in a WEDM process which can operate with 2 different wire types, one for those cuts with a relatively high material removal i.e. potential for speed increase and the other for those cuts with a relatively small material removal. The rough cut and 1st finishing cut would be performed with a speed wire and the final finishing cuts with the wire type mentioned above. In the same sense this wire type is a solution for the process combination of AWJM (roughing) and WEDM (finishing).
A further wire design (AG-N5, AG-N6) consists of a brass core, a first layer of beta brass (s = 15 - 25 µm) and an outermost nickel layer (s = 1.5 – 3.0 µm). This type would be applicable on all WEDM machinery with a wire threading system requiring a hard wire with a high straightness, both, features a copper core wire does not present. Compared to a copper core wire this wire type can enhance the cost efficiency since production effort and material costs are less while cutting rates are in the same range.
Figure 5: Wire designs for cutting Inconel fir trees
In addition to the above mentioned wire designs more simple designs were proposed. Composition, approach and drawbacks of those designs are displayed in Figure 6.
Figure 6: Simple wire designs for cutting Inconel fir trees
In the following steps the new wires types were produced, analysed and tested. In numerous WEDM test series cutting time, precision, surface roughness and contamination were analysed.
Figure 7 shows the productivity results in comparison to reference wires. In the main cut a time saving of 18-20% vs. brass wire is confirmed for the new series with copper core (AG-N3). Series AG-N2 yields similar time savings. The total time saving after 4 cuts with AG-N3 is also about 18-20% vs. brass wire. The standard X type wire technology for trim cuts does not fit to series AG-N3. As a result the time for the 2nd trim cut is too high and so the total machining time is. The standard brass wire technology which was applied for series AG-N2 (brass core) fits better in terms of productivity. With both series however slight grooves were generated after the final trim cut. The Ra values measured on the workpiece were therefore slightly higher as compared to brass wire or BCX/TPX. The precision obtained with series AG-N2 and N3 is of the same quality as with brass wire or BCX/TPX. Technology optimizations towards speed and surface quality in the trim cuts will be performed once the favourite wire type has been identified.
Figure 7. Machining time reduction with sample series AG-N2 and AG-N3
The analysis of the sample series AG-N1, AG-N2 and AG-N3 revealed that the outermost nickel layer presents discontinuities and larger variations in thickness. These defects are regarded as a drawback towards the avoidance of contamination by copper and zinc. The defects are due to the relatively high roughness of the wire pre-material prior to electro plating with nickel. The high roughness is a result of the formation of the zinc rich brass layer through a diffusion annealing process. After coating with nickel at an intermediate diameter the wire is drawn to its final diameter. Also this drawing step affects the quality of the outermost layer. For economic reasons, however, nickel plating at the final diameter is not an option. For the following sample series the process chain was modified in order to improve the homogeneity of the nickel layer. For sample series AG-N2, however, an improvement could not be achieved. The very brittle gamma brass layer generates cracks during final drawing which inevitably resulted in discontinuities of the nickel layer. In contrast the improved sample series AG-N3 displays a more continuous outermost layer (see Figure 8).
Figure 8: Cross sections of sample AG-N3b and AG-N3c (improved samples)
Workpiece contamination has first been analysed for reference samples made of tool steel. The influence of cutting regime (rough / finish) and wire type (plain, coated) was studied. Electron Micro Beam Analysis (EBMA) shows that roughing regime generates more contamination of Cu and Zn than finishing regime (about factor 3). After finishing with 2 cuts contamination is however still clearly detectable. The thickness of the deposition of wire material was found to be in average 0.3 µm for roughing regime and 0.1 µm for finishing regime. EBMA analyses of Inconel samples machined with 4 cuts are depicted in Figure 9, Figure 10 and Figure 11. Here it is visible that both coated wire types generate less contamination of Cu and Zn. However, the Ni coated sample does not further reduce contamination.
Figure 9: Work piece contamination after 4 cuts with brass in Inconel
Figure 10: Work piece contamination after 4 cuts with BCXcc in Inconel
Figure 11: Work piece contamination after 4 cuts with AG-N1 in Inconel
As already stated before, the Ni layer of the first series AG-N1 and AG-N3 is too discontinuous over the circumference. In order to investigate another potential reason for remaining contamination cross sections of the wire types in new and worn condition have been prepared. As microscope analysis shows, the depth of wear after 4 cuts is ca. 2-3 µm which is exceeding the Ni layer thickness of the sample AG-N1 and AG-N3a. Thus, samples AG-N3b and AG-N3c are regarded as preferred solutions.
The improved wire series AG-N3 was tested with 4 cuts and the Inconel samples were again analysed by EBMA regarding contamination. Figure 12 shows the resulting thickness of the deposited layer in comparison to the former results with reference wires. The wire BCXcc which is similar to type AG-N3 but without Ni layer already provides a drastic reduction of Cu and Zn deposition. The AG-N3 wires further reduce contamination. With a nickel layer thickness of 2.5 µm the contamination becomes in average less than 20 nm and in maximum less than 30 nm. In order to analyse reasons for the still remaining contamination a test was conducted leaving the work piece submerged in water dielectric for about 40 min without spark erosion. As Figure 12 reveals deposition of Cu and Zn already occurs without the EDM process. Typically Cu and Zn ions or fine particles are in solution respectively dispersed in the water dielectric. This debris material origins from prior EDM operations with ordinary wires or even from a prior rough cut with type AG-N3 where the wear of the wire exceeds the thickness of the nickel layer. The contamination layer without erosion has a similar thickness as in the test with AG-N3b or AG-N3c. Thus these wire types already represent favourable solutions.
Figure 12: Work piece contamination with different wire types
Parameters considered to be relevant are the electro plating process for zinc deposition, the diffusion annealing process in order to generate the beta brass layer and the calibration (drawing) process prior to the nickel plating process. These processes were each modified and the resulting layer quality was assessed. The investigations revealed that a well defined intermediate drawing step would be the most efficient way to improve the surface uniformity of the first layer (see Figure 13).
With the improved samples AG-N3b as well as samples AG-N4b and AG-N5b cutting tests and contamination analyses were performed. As depicted in Figure 14, however, no further improvements regarding contamination were obtained.
Figure 13: Improvement of surface prior to nickel plating
Figure 14: Workpiece contamination: overview of results with additional samples
At this stage the results and challenges could be summarised as follows:
• A preferred wire design comprises a copper or brass core, a 20 to 40 µm thick beta brass layer and a 1.5 to 3 µm thick nickel layer.
• Cutting speed in the rough cut is about 25% over standard brass wire.
• With a 4 cut technology the workpiece contamination is reduced by 3 times versus brass wire (down to 30 nm).
• Deposition of Cu and Zn within a thickness of 20 nm already occurs without the EDM process just by submerging the workpiece in water dielectric.
• Standard cutting technology for the trim cuts needs to be adapted to the new wire type since cutting time and surface quality are partly suffering.
• The potential for speed increase in the rough cut needs to be analysed.
• The new wire types have to be tested in the defined fir treecutting test.
Finally, Figure 15 gives an overview of the potential process set-ups for fir tree slotting together with the preferred wire solutions to be used in each set-up. If a post treatment was to be considered to eliminate surface contamination wire solutions without a special nickel coating could be applied. Those wire types have been identified within the early test series of the project.
Figure 15: Wire solution for cutting Inconel depending on the process set-up
The approach for cutting titanium alloys is different: Since an economic production of a wire with a titanium coating would not be feasible, wires with a reasonably thick layer of pure zinc are proposed (see Figure 16). The aim is to avoid contamination by copper. The subsequent chemical cleaning processes could be shortened since the less noble zinc can easily be dissolved.
Figure 16: Wire types AG-T1 / T2 / T3
In a first step samples of AG-T1 were produced and tested. Figure 17 and Figure 18 show the Electron Beam Micro Analysis (EBMA) of surfaces of TiAl6V4 parts cut with brass and AG-T1. Similar to the results with AG-N wires in Inconel the average layer thickness is drastically reduced. But more importantly the contamination with copper becomes practically zero when using AG-T1.
Figure 17: Work piece contamination after 3 cuts with brass in TiAl6V4
Figure 18: Work piece contamination after 3 cuts with AG-T1 in TiAl6V4
According to the wire solutions proposed for Inconel a table has been rendered displaying the recommended wire types for titanium alloys as a function of the process set-up (see Figure 19).
Figure 19: Wire solution depending on the process set-up
New dielectrics and additives
New development in wire-EDM made machine manufacturers to design new machines which do not work with deionised water as a dielectric but with hydro-carbon fluids. The big advantage is that oil dielectrics produce much better surfaces with lower thermally influenced layers and less micro cracks, which reduces the typical caused white layer to a minimum. The white layer can cause a reduction in lifetime of work pieces (compare project “EDM2005” (ref. GRD1-2001-40502)).
Using hydro-carbon fluids as a dielectric allows using better off-set parameters with smaller gaps, better surface finish but worse flushing properties. Further the heat conductivity of the oil is not as good as water, which normally results in a lower cutting speed when cutting with oil dielectrics. Normally several cuts (roughing and finishing) are needed for the desired surface finish, but cutting with oil dielectrics allows to skip one of the finishing cuts due to more effective roughing cuts (compare project “EDM2005” (ref. GRD1-2001-40502)). The optimisation of the cutting strategy with oil compensates the disadvantage which comes from the slower speed by cutting in oil. Therefore the total work time is similar to water, but the surface finish is much better. Another advantage of the use of oil as a dielectric is that there is no contamination of the surface with melted eroded wire particles on the surface of the work pieces and that hydro-carbon fluids avoid leaching effects on materials e.g. cobalt leaching compared to water.
For ADMAP-GAS, alternative ways for the manufacturing of turbine blades were analysed and developed. Besides water jet cutting the EDM wire cutting was selected as one alternative, due to reducing the EDM caused heat influenced layer (white layer) to a minimum and improving the surface finish by using hydro-carbon dielectrics (compare project “EDM2005” (ref. GRD1-2001-40502)). Further with special additives it was possible to increase the cutting speeds by using oil as a dielectric. For steel increases of the cutting rate in the range of 20-30% are possible. For cutting in Inconel the results received in the last six months period show improvements of up to 15%. During the project ADMAP-GAS roughly 80 additives have been tested in Inconel. These additives were tested in 24 different dielectrics. The dielectrics differ in hydro-carbon based fluids and additive contents. Some additives have been tested in the dielectric, but most additives have been tested through a special applicator which oelheld has patented. This additive applicator allows applying additives directly onto the EDM wire in high concentrations. Since October 2011 (project month 27) the dielectrics and additives have been tested on an AgieCharmilles RoboFil 440 Wire-EDM machine tool at WZL.
In one test procedure additives were applied through the applicator - “impregnating the EDM wire” - straight on the EDM wire and the additives were brought along the EDM wire directly to the erosion zone. The main goal of the trials was to test different additives, which were similar in their molecular chemical structure, but showed different behaviours under wire-EDM cutting conditions.
In another test procedure hydro-carbon fluids were used as base fluids and the additives were mixed into the whole fluid in varying concentrations. Here the main goal was to find additives which were 100% soluble in oil and shouldn’t be filtered out.
Further the same test procedures were repeated in deionised water used as dielectric, in which pre-tests had been made in the past, but more tests were done together with the project partners Charmilles, BEDRA and WZL.
The last test procedure then was a summary of all of the above, in which the new developed know-how and the best performing additives were used to dilute them into dielectrics of varying viscosities. The finishing cuts in oil dielectrics were done together with the project partners WZL and the machine tool manufacturer CHAR. While oelheld used an old analog generator for their cuts, WZL and CHAR used both new digital generators. Due to the fact that there is no existing technology for EDM cutting with hydro-carbon dielectrics the two partners WZL and CHAR mainly cut in de-ionized water, while oelheld cut mainly with hydro-carbon fluids. So, standard technologies of cutting steel with de-ionized water as dielectric were used, even with different wires. To compare in the end the cutting speeds and feeds with the existing technology the percentage ratio to the value of the standard technology table based on deionised water was used. So, the values of the standard technology table were seen as 100% and the received values were compared to them. The measured values are given in % technological value. The tested oil dielectrics differ in viscosity and are doped later on with different concentrations of an additive. In Figure 20 the test set up which has been used for the cutting tests is shown.
Figure 20: Test set up of cutting tests
To compare the cutting rates on a RoboFil 240ccS with water dielectric external tests were made at GF AgieCharmilles in Schorndorf. There is the same generator (digital) as located at WZL. The priority of the test was the speed. For the test deionised water was used as dielectric. The work piece was of Inconel 718 with a height of 40 mm. The following different wires were used: Brass 900, Topas plus H and Topas plus X. The results show that the highest cutting rate was achieved with the Topas plus X wire (see Figure 21). In each test the wire-additive OH 2377 could improve the cutting rate. With this experience for all tests the wire Topas plus X was used.
Figure 21: Cutting rates with different wire with and without wire-additive
In Figure 22 further test cut results are shown. The stated graph shows the different effect of the additives in the different dielectrics. So the abscissa shows the wire additives and the single blue bars represent the different dielectrics. In the graph which is shown in Figure 23 the dielectric OH 2286 is doped with different concentrations (given in percent values) of a performance improving additive. The percentage addition of the dielectric OH 2286 varies from 0.1 to 20%. The continuous line shows the averages of the results of each dielectric. At the end 100% of the technological value is achieved.
Figure 22: Improvements of different wire-additives in different dielectrics
Figure 23: Results of the dielectric OH 2286 doped with different amounts of additives
For the comparison of different dielectrics not only different base dielectrics were tested but also the base dielectrics were added with additive packages. So the base dielectric OH 2788 was added with additives to the dielectric OH 3521. As it is seen in Figure 24 with the additives the cutting rate of the dielectric OH 2788 could be improved. The result of the OH 3521 is even better than the other dielectrics. The test conditions were the following: as the wire-EDM-machine a RoboFil 230/240 located at oelheld with an analog generator (old generator generation); work piece Inconel 718, wire Topas plus X, no additive on the wire. The dielectric OH 2788 and OH 3521 were selected for the tests at the partner WZL in Aachen.
Figure 24: Cutting rates in different oil dielectrics
In Figure 25 the results of the cutting tests at WZL are mapped. A test series with three different dielectric fluids (water, CH-based dielectric and CH-based dielectric with additive) has been set up. Based on the developed technology table (more details about the technology are documented in the mid-term report) a technology has been tuned for the base dielectric (CH-based) as well as for the CH-based dielectrics with additives. The resulting cutting rates of the tests can be taken from the mapped graphs. It can be seen that the cutting rate in water is the highest one and the cutting rate in the basis dielectric is the lowest one. Due to the additives an increase of 66% in the cutting rate was measured. The measurement of the surface roughness shows that all tests resulted in a roughness minor Ra = 0.8 µm and therefore all processes meet the project requirements.
Figure 25: Results of cutting tests at WZL
Summarized it can be stated that using hydro-carbon dielectrics, mixing additives into these dielectrics and to apply special additives in high concentrations directly on the EDM wire showed tremendous improvements in the wire-EDM cutting process. Both the additives on the wire and in the dielectric had remarkable positive effects on cutting speeds and surface finish. The best results could be gained with additives mixed in high concentrations into the hydro-carbon based dielectric. One of the best performing mixture unfortunately caused handling and health problems and also was economical not an option.
No long term filter tests, no long term corrosion tests, no long term seals and gaskets tests or no long term machine tests were possible to run, due to so many different additives and base fluids to test. Also the stability (decomposition) of the dielectrics was not examined related to a long term test. Another problem is that the additives which had positive effects on the wire could not be used in the dielectric. This was because they were not soluble in deionised water or hydro-carbon fluids and were filtered out or taken out by the deionising resin cartridge. But even when deionised water as dielectric was used, some additives increased the cutting rate. As previous tests of the fir tree profile had shown, the effect of additives on the wire would cause an angel correction of the wire which means it would need a lot of technology adaption. So for the time after the project oelheld will start long term studies to examine the dielectrics and additives under production conditions. The main focus will be the stability of the dielectrics and the influence of the dielectrics on other wire-EDM machines.
Technology optimisation and process monitoring
Within the research regarding technology optimization and process monitoring several tests concerning technology tuning have been carried out during the whole project. To observe the surface integrity according to technology aspects metallographic observations were performed. The observation included inspections in terms of recast layer, heat affected zone dimension and cracks. Basis for the technology comparison was the developed technology table (more details about the technology are documented in the mid-term report).
Following the four available settings are listed:
• E2: roughing setting with triangular current iso-pulses.
• E3: roughing setting with trapezoidal current iso-pulses.
• E7: finishing setting with triangular current iso-pulses.
• E21: surface-finishing setting in HF mode.
The samples were prepared on the metallographic preparation laboratory (metallographic saw, enrobing, polishing material, chemical etching) which is available at CHAR. The surface observations of different positions on the test piece (top, middle and bottom – related to wire unwinding direction) were made on an optical microscope.
Following the results of four test cuts with different settings will be presented. Firstly the rough cut settings E2 and E3 are analyzed and after that the observations of the finishing setting E7 and finally the sufacing setting E21 are documented.
In Figure 26 the rim zone views of the E2 setting (M52 - triangular) are mapped. The recast layer is not continuous, its optical thickness vary from 0 up to 15 μm. Furthermore no heat affected zone is observed. Additionally, no major differences from top to bottom are observed. The resulting removal rate during the tests was VW = 156 mm²/mim.
Figure 26: Surface integrity of roughing E2 on Inconel 718
In Figure 27 the rim zone views of the E3 test cuts are pictured. The observed recast layer is not continuous. Its optical thickness varies from 0 up to 15 μm and no heat affected zone is visually observed. Furthermore no major differences from top to bottom were detected. The resulting cutting rate was VW = 92 mm²/min.
Figure 27: Surface integrity of roughing E3 on Inconel 718
In Figure 28 the rim zone views of the E7 setting cuts are mapped. The recast layer observed is not continuous and it’s optical thickness varies from 0 up to 8 μm. No heat affected zone is visually observed. Furthermore, no major differences are observed from top to bottom.
Figure 28: Surface integrity after finishing E7 on Inconel 718
In Figure 29 the rim zone views of the E21 setting cuts are mapped. The recast layer is not continuous, its optical thickness varies from 0 up to 4 μm and no heat affected zone is observed. Furthermore, no major differences from top to bottom.
Figure 29: Surface integrity after surfacing E21 on Inconel 718
To summarize the rim zone view inspections the following aspects can be mentioned. Additionally in Table 2 the results are listed:
• The recast layer, formed by re-solidification of the re-deposited and melted material during the EDM process is not continues and homogeneous.
• The thickness of the layer is correlated to the power of the last setting used. As the machining power decreases, the recast layer thickness also decreases.
• The resilience to high temperature of the Nickel based super alloy (as Inconel 718) shows a good behavior on EDM machining process.
Table 2: Overview of rim zone view analysis results
Finally, at the end of the project a good working technology table has been developed. This table includes several machining settings and sequences (including two roughing: E2, E3; two finishing: E7, E9; two surfacing settings: E21, E23; special pocketing setting for cleaning AWJ machined surfaces: E4) to meet the required surface integrity of the turbine industry for the fir tree application. The core of the technology is based on hard brass wires with a diameter of 0.250 mm and is able to cut several Nickel based alloys like Inconel 718, Udimet 720, etc. In addition to the fulfillment of the surface integrity requirements this table meets further on the surface roughness and accuracy requirements (Ra < 0.8 µm and accuracy ± 5 µm). Within the table machining parameters for the heights between 5 mm and 60 mm are stored. In detail the following settings and sequences are available:
• Settings:
o E2: H10-60 mm, Fast roughing with sealed nozzle M52,
o E3: H05-60 mm, Reliable roughing M54, for various nozzle configurations,
o E4: H05-60 mm, pocketing or cleaning setting,
o E7: H05-60 mm, finishing std U,
o E9: H10-60 mm, finishing HFBF,
o E21: H5-60 mm, surfacing M30 (200/200) Ra 0.7 and
o E23: H5-60 mm, surfacing M23 (400/400) Ra 0.8.
• Sequences:
E2, E7, E21. E3, E7, E21. E4, E7, E21.
E2, E7, E23. E3, E7, E23. E4, E7, E23.
E2, E9, E21. E3, E9, E21. E4, E9, E21.
E2, E9, E23. E3, E9, E23. E4, E9, E23.
In addition to the standard brass wire technology table further research and development concerning machining with the new nickel coated wires has been carried out. With regard to the productivity target the standard cutting technology for coated wire (X-type) was successively optimised by conducting standard EDM tests during the last project months. The result is shown in Figure 30. With wire type AG-N3b and optimised technology the time saving vs. brass wire is about 30 % while also precision and roughness targets are met.
Figure 30: Results from standard EDM test with standard and optimised technology
The optimised technology was then applied in a fir tree cutting test. Here slight parameter changes still were necessary. The results are displayed in Figure 31 and Figure 32. AG-N3b yields a time saving versus brass wire of 22% with standard technology and 30% with optimised technology. AG-N4b which has just a nickel coating performs similar like brass. With AG N5b about 7% time saving is reached under standard conditions. While there are no objections to the surface roughness the contour precision target is still not met. This also applies to the contour check of parts cut with samples AG-N4b and AG-N5b (not displayed here). However, the deviations are not that big that the wire type itself has to be considered as non-approriate. This is also supported by the fact that the new wire type AG-N3b reveals a higher precision compared to brass wire. It is therefore expected that a duly adapted cutting technology will yield the required contour precision.
Figure 31: Results from fir tree cutting tests
Figure 32: Contour check of wire-cut fir tree profiles
Evaluation of the results
The economic evaluation of the results can be seen in the chapter “Potential impact and main dissemination activities and exploitable results”. Here the two ADMAP-GAS processes are evaluated concerning the feasibility of the substitution of the broaching process of fir tree profiles in terms of costs and environmental impact.
Concerning the surface integrity comparison fatigue tests have been carried out. Three test series were tested which have been produced with the following manufacturing technologies:
1. Wire-EDM
2. Broaching
3. Grinding
The Wire-EDMed series has been produced at WZL with the developed Inconel 718 machining technology. The broached series was machined at WZL too. In Figure 33 the broaching machine located in WZL is located. This machine is a vertical working machine and equipped with a force measuring platform to monitor the tool wear. On the right in the figure the clamping device with the test specimen is sketched. According to the finishing process of the turbine manufacturing to produce fir tree geometries a tool and machining parameters were designed and test sample were produced. The third series which has been produced by grinding was manufactured by an OEM.
Figure 33: Machine set up to manufacture fatigue test samples by broaching
After the manufacturing of the specimen a surface integrity inspection of a sample of each series has been carried out. In Figure 34 results of a surface roughness measurement are mapped. All surface roughness measurements are minor Ra = 0.8 and fulfill the projects requirements.
Figure 34: Agenda and surface roughness measurement results
Additionally to the surface roughness measurements rim zone analysis of a specimen of each series has been set up. In Figure 35 the rim zone view of an eroded specimen is pictured. A very thin white layer is visible. No micro cracks or thick thermally influenced zones have been detected.
Figure 35: Rim zone view of series 1
In Figure 36 the rim zone view of a broached specimen is pictured. Two different analyses according to the cutting direction have been prepared. On the one hand a view in cutting direction and on the other hand a view perpendicular to the cutting direction has been prepared. The view in cutting direction (upper two views) shows a slightly rough surface. Furthermore a small thermo-mechanical layer is visible. The other view shows a flat surface topography. Also here a small thermo-mechanical affected zone can be seen. Due to the deformation of the grains a mechanical detraction of the manufacturing process is obvious. Approximately 5 – 8 µm under the surface the parent and non-affected material can be seen.
Figure 36: Rim zone view of series 2
Figure 37 maps the rim zone views of the ground series. Here, nearly the same aspects as in the broached series can be seen. A zone which was affected by a thermo-mechanical and a zone which was affected by mechanical forces are obvious.
Figure 37: Rim zone view of series 3
In Figure 38 the test set up and the results of the tests are shown. The specimen is clamped between two chucks. One chuck is moving and one is fixed. The moving chuck induces the bending moment into the specimen. The test conditions include a limitation of the cycle numbers of 5 x 106 cycles. The bending frequency is 70 – 75 Hz.
Figure 38: test set up and results of fatigue tests
Looking at the results a very good project outcome can be summarized. It can be seen, that the Wire-EDMed and broached series fail in the same surface stress area. The ground series is the best of the three series. Nevertheless it can be stated that in terms of fatigue behaviour the developed Wire-EDM process is able to substitute the broaching process to manufacture fir tree profiles.
Flushing and nozzle design studies and developments
Flushing is one of the most important parameters within Wire EDM and especially within the rough cut. A high quality flushing improves the speed in a very high range, due to the fact that flushing removes dirt and scraps off the machined slot and cools down the wire allowing higher sustainable electrical power. In opposite to the flushing during the rough cut the rinsing within the finishing cut has a minor role. So, the main research activities concerning nozzle design were given to the rough cut flushing.
The dielectric flow out of the nozzles modifies the position of the wire due to vibration. The results are geometrical defects which have to be corrected by the following cuts. These effects are influencing the achievable global speed and accuracy. To achieve the best process outcome an optimization of the flushing should improve efficiency and reliabilities of the WEDM process. Hydrodynamics specialists from a partner institute (HES-SO) assist CHAR to simulate dielectric flows into the machining gap and optimize nozzles design.
Several new nozzles (12) were designed and manufactured. These nozzles were tested and their performances were compared to existing and standard nozzles. Within former work of the project some “short” nozzles were tested and their performances analyzed (see former project reports for more details). The performances of the so called “long” nozzles and diffuser were analyzed, tested and quantified during the last project months.
As stated above the productivity and the efficiency of the rough cut varies significantly with the flushing quality (nozzles dimensions/positions or machined disk geometries constrain). For a substantial comparison a range of eight upper head nozzles were tested and their performances were compared (see Figure 39). The lower head’s nozzle configuration has been the same in all tests.
The tests were carried out on the test bed machine tool at CHAR. The first tests contain a static set up. The upper head was placed over a machined slot with a wire but without any erosion. Starting from a sealed nozzle, the distance between the work piece surface and the lower nozzle surface has been increased for each nozzle. As seen in Figure 40, the main drop of pressure occurs in the first millimeter. The nozzle with the highest pressure was the nozzle L3. Furthermore the pressure gradient of this nozzle seems to be the best within these tests.
Figure 39: Drawing and characteristics of tested nozzles
Figure 40: Static pressure measurements for different upper nozzle types
To quantify the improvement made by the nozzles comparative tests were performed. The test sequence included machining a slot and visualizing the water flow on the work piece by analyzing the coloration of the work piece surface. The cuts were made on steel test pieces with the height h = 125 mm. This height was chosen to have a better visualization of the coloration. The used technology table was the standard technology LT25A, HM66. The upper nozzle was placed ~ 0.2 mm over the upper work piece surface. The lower nozzle geometry was always a standard geometry. In Figure 41 the machined surface and the corresponding nozzles are pictured. The different coloration can be observed in the picture. The differences in the coloration are on the top of the surface. Here the upper nozzles differ from the lower one which has been the same for all tests. It can be seen in the figure, that for the nozzles L1, L2, L3, L4, F2 and BR in the upper region (approx. 10 mm from the upper edge) a different coloration due to the flushing appears than in the standard nozzle geometries Standards and PM. A flushing dead zone can be examined in the middle of the surface. In this zone the flushing is very bad which means, that not enough fresh dielectric fluid is present for good process conditions. This results in e.g. not enough wire cooling in this region. For minor work piece heights this issue is not as relevant, because the dead zone should be smaller.
Figure 41: Machined surfaces and nozzles corresponding
Additional to the coloration tests, machining speed tests were carried out. For these tests the standard project height h = 40 mm was chosen. In Figure 42 the resulting removal rate can be seen. In this case, the ability to renew dielectric in the slot has measurable effects on the machining speed. The best result is made with auto adjustable nozzle (PM). The PM nozzle is partly mobile, so the dielectric pressure sticks the nozzle surface on the work piece surface. This maximizes the dielectric flow into the machined slot. Unfortunately, this nozzle is not suitable for ADMAP-GAS application because it induces mechanical forces onto the work piece that lead to loss of geometrical accuracy. Furthermore, its adjustability is limited and this kind of nozzle is suitable only on flat grinded surface.
The tested nozzles did not show better results in terms of speed and accuracy than actual standard nozzle. Some differences are measured between nozzles but no new designed nozzles over take the actual standard nozzle performances.
Figure 42: Removal rate achieved with different upper nozzle geometries for H: 40 mm
The numerical simulations by finite elements modeling which have been presented in the last report show a bad dielectric flow in front of the wire. One conclusion out of the simulation was that a spinning dielectric flow around a symmetrical axis (around the wire) is expected to increase the dielectric flow into the working gap. To achieve a rotating dielectric flow the standard diffusor which is placed in the upper head was redesigned. In Figure 43 the standard and two redesigned diffuser are mapped. These diffuser geometries were tested on the test bed machine tool at CHAR.
Figure 43: Three different diffuser geometries
The tests and results of these diffusers did not get measurable improvements on the machining performances. Some concepts to decreases the dielectric flow diameter and to increases the pressure in the gap are ongoing. The work done on nozzles and diffuser show and confirm that the actual material available on the machine is performing quite well and to improve significantly their performance it will require more work. The nozzles have strong effect when they are sealed (as close as possible at the surface). During fir tree machining, external geometry of the disk is often not flat, therefore it is not possible to machine with the nozzles always close enough and sealed to the machined surface, then nozzles effects are greatly reduced.
Machining technology optimization
A new idea to improve efficiency and reliabilities in machining fir tree are correlated to the actual way of machining with the 3 cuts sequences. 1s t cut roughing, 2nd cut finishing and 3rd cut surfacing. The roughing setting is working in the bulk of the material, its performances is tightly correlated to the capabilities to put as much power as possible into the machining area sustainable by the wire. The 2nd cut is a finishing cut, it corrects and improves the geometry of the machined surfaces. The following aims have to be met by the cut:
• The corrections are necessary to correct the straightness of the surface. This defect has its origin on the soft wire electrode which is tight in position by the upper and lower guides. Between the guides the wire vibrates, that affect its straightness on the machined surface.
• It must also correct dissymmetry defect. This dissymmetry is caused by the deviation of the dielectric flow against the surface of the disk. A turbine disk is not flat and infinite.
• It also corrects machining geometrical defects produced during the roughing operation. They are linked to the machining strategies and protection resulting of loss of accuracy induced by the wire drag phenomenon and the way to minimise it.
This finishing cut corrects all these defects at a reasonable machining speed. Unfortunately it did not improve sufficiently the surface roughness and surface integrity up to the project’s objective of Ra ≤ 0.8 µm. The 3rd cut is a surfacing cut; the purpose of this cut is to put the surface into the objective of roughness and surface integrity. It is a constant speed cutting operation that is not controlled. The cutting is “blind” and no actions are possible if something is not into its range of functionalities. The two cuts studies aimed to replace the 2nd and the 3rd cuts by a controlled single cut that should give the project objectives in geometry, surface quality and integrity at a resulting machining speed equivalent or faster to the 3 cuts existing sequences.
To reach these objectives several solutions were implemented and tested to produce a setting capable to give in one cut the performance of two cuts (E7 and E21). The following boundary conditions were given out of the project for the development:
• The surface quality and geometry have to meet the ADMAP-GAS requirements:
Ra ≤ 0.8 μm, minimum surface contamination and geometry precision within tolerance band ± 5 μm.
• Stay competitive in terms of time, consumables and costs:
Speed of one cut ≥ actual resulting speed of two cuts E7 and E21.
• To improve reliabilities and productivity: Controlled machining cut.
The expected improvements of the implementation of the new two cut sequence are on the one hand efficiency aspects (all controlled settings) and on the other hand productivity issues (less threading operations…). Within the scope of the research several solutions were tested on the standard generator of the machine test bed at CHAR:
• New current pulse shapes: “Isopulses” without ignition.
• “Isofrequence” modes using UP power supply instead of ignition UAL power supply.
• Shorten the current pulses width (delay line, faster transistors, diode to minimize current damping, etc.).
The Table 3 summarizes the results. The achieved results do not match the ADMAP-GAS specifications in terms of surface roughness (Ra < 0.8 µm). Nevertheless, a real improvement is made showing a new way to increase the machine performances in improving the efficiency and stabilities of the 1st finishing cuts. Some more development and improvements are required to fulfill the objectives. Innovation developments are ongoing on these topics.
Table 3: Achieved results with several prospected solutions
Critical assessment concerning WP2 objectives
Looking on the overall WP2 objectives it can be said that the research on Wire-EDM within the project ADMAP-GAS was very successful. In Figure 44 the objectives, which have been set up for WP2 and the status of the achievements of the aims are listed.
Figure 44: Achieved objectives
WP 3: Development of “Abrasive Water Jet Slotting”
This subchapter focuses on the main S&T results/ foregrounds concerning Abrasive Water Jet Machining (AWJM). The AMRC, USFD, TEKS and DIAD investigated the performance of AWJM for titanium and nickel based alloys to benchmark the process against broaching in turbine engine disc manufacture.
This work package was set up to develop AWJM technologies as an alternative to broaching in aerospace; particularly for fir tree slots. The following objectives have been set up before the project start:
• Improve process control to compensate for jet-deflection in fir trees and blisks; particularly blow-out and curves on hard alloys.
• Reduce kerf size to less than 0.3 mm in order to be able to cut inner fir tree diameters.
• Develop methods of precisely controlled depth cutting, including the complex curves and tapers found on blisks (depths 50 – 100 mm); positional accuracy will be improved to that of the nozzle (± 30 μm).
• Improve nozzle-life by 50%.
Definition and process analysis towards the targeted application
The work concerning the definition and process analysis towards the targeted application was finished successfully during the first period of the project. The following work was carried out:
• A survey of aerospace discs and fir tree features was carried out. The information collected was used to design a generic fir tree geometry, which contains the most difficult features to machine from all the disks. This has been used throughout the project by all partners in the consortium.
• A specification of tolerances and surface integrity requirements was produced. The specifications have been used as a target for developing the technology and methods to achieve the requirements - e.g. tolerances of 5 -10 µm, subsurface deformation < 51 µm (0.002 in).
• Additionally, analysis of existing mathematical models for modelling AWJ system was carried out; as well as a review of AWJ cutting mechanisms.
Fir tree features are complex geometry with tight tolerances up to 5 µm in the pressure flanks and 10 µm in non-pressure flanks. The subsurface condition is restricted to 51 µm deformation, no white/amorphous/recast layer and no foreign material present. Aerospace discs can have a diameter as big as 900 mm and their thickness is around 30 mm. Fir tree features can have from 2 to 4 teeth and have a minimum internal/external diameter of 0.6 mm. The test-piece design has 4 teeth and a minimum internal/external diameter of 0.6 mm.
Development of Abrasive Water Jet Machining
The main research and development work was split into three sub-tasks with certain overlap in the development of nozzle geometries and improvement of hydraulic jet kinematics.
Analytical work has been directed towards developing a Precision Abrasive Water-jet system (PAW). The PAW is composed of a precision cutting head, a precision table, machine frame, and benchmark and calibration system.
An analysis of the current system uncertainty and sources of variation led to USFD proposing the PAW system. The analysis was based on the available AWJ machine ZX-613 located at USFD. The cutting head is a standard Dialine™ head from Accustream. However, the analysis, methodology and concepts developed are suitable for all other water jet systems including 5-axis systems. The main error sources in AWJ machining are due to uncertainty caused by:
• u1: cutting head vibration;
• u2: locating components in the machine bed;
• u3: flow deflection;
• u4: machine frame and system vibration;
• u5: other elements.
The research and development within ADMAP-GAS focuses on the avoidance and improvement of the first three errors. Other sources of machine uncertainty such as machine frame and system vibration (u4) and other elements (u5) were not in scope for this project. The concepts developed in u1 to u3 can be applied to any waterjet machine regardless of machine frame configuration.
Development of different nozzle geometries control angles (avoid the errors u1 and u2)
Within the work on the different nozzles mainly the two error sources u1: cutting head vibration and u2: location components in the machine bed have been analysed and improved. In detail research on a precision cutting head and different new components for a more efficient process were carried out. Following the work is summarised.
Precision Cutting head - u1
The aim of the precision head is to increase stiffness, reduce vibration and to provide probing capability which will help to locate parts in the machine bed (u2). It also allows the nozzle to be easily changed, reducing set-up errors.
Suppose the z axis is the axis where the component is to be cut, then the machining errors can be classified as position errors (Δx, Δy, Δz), orientation errors (Δθx, Δθy), and timing errors (Δθz) with respect to 3 axes, x, y, z.
Machining uncertainties will be produced in all 6 freedom degrees, (x, y, z, θx, θy, θz), correspondingly machining uncertainty for each error source i and its components in 6 freedom degrees are U {Ux, Uy, Uz, Uθx, Uθy, Uθz}.
Hence, the general machining uncertainty (U) of the system is the sum of all main error sources (ui) as Equation 1
[1]
The machining frame and motion system produce machining uncertainties depending on the machining system configuration used for various applications. The other elements that produce machining uncertainty are not as significant as the first three items, and are assumed to be negligible to simplify the expression.
In order to control and minimise the machining uncertainty (u1), a precision cutting head has been developed. A comparison of the conventional and precision heads is shown in Figure 45. The precision cutting head has higher stiffness, a connector which fits the novel nozzle developed in ADMAP-GAS, the capability to interchange tools and is constructed to allow the nozzle to be easily changed.
Figure 45: Conventional (left) and precision (right) abrasive waterjet cutting head
The combined standard position machining uncertainty (σ=1) produced by the cutting head in x, y directions are as Equations [2] and [3].
[2]
[3]
The combined standard orientation machining uncertainty (σ=1) produced by the cutting head in x, y directions are as Equations [4] and [5].
[4]
[5]
The position machining uncertainty (0.0185 mm) produced by the precision cutting head is predicted to be 7.5% of the error produced by the conventional head (0.2476 mm) in both x and y directions. The orientation machining uncertainty (0.000185 rad) produced by the precision cutting head is predicted to be 41% of the conventional head (0.000238 rad) in both x and y directions. Manufacture of the precision head was carried out at the University of Sheffield facilities. Figure 46 provides a picture of the precision head.
Figure 46: Precision water-jet head manufactured
The precision water-jet head was installed in the z carriage of the machine, Figure 48. The head was tested in an experimental framework using Design of Experiments in order to quantify the improvement in terms of surface roughness, taper angle and error in producing internal and external radii. Fir tree-like test-pieces were cut from Inconel 718 plate. These had the same geometry as that used to characterise the system with the conventional head and included internal and external radii of 0.3 and 0.5 mm as shown in Figure 47.
Figure 47: Fir tree like test-piece
The precision head produced superior surface roughness and taper angles when compared to the conventional head. However, the error in internal and external radii was increased. It is hypothesized that this was due to a reduced kerf size, resulting from the decreased vibration in the precision head.
Figure 48: Precision Waterjet head installed in the z-carriage
The validated gains made in precision due to the new head developed in ADMAP-GAS are shown in Table 4. The taper ratio was reduced by 10% due to the lower level of vibration and higher nozzle stiffness. Surface roughness showed an decrease of ~50%. It is postulated that the increased error when producing radii was due to the conventional controller working with different kerf geometry. The new kerf geometry needs to be calculated experimentally and fed into the AMRC Limited controller to reduce errors in the radii.
Table 4: Validated gains made in precision due to the new head
Locating Components in the machine bed - u2
In order to control the component location (u2), a precision AWJ table was designed. Standard waterjet machine beds consist of a series of grids with low location precision where it is difficult to clamp parts. Moreover, the grids are flexible and induce unwanted vibration while cutting. The precision table developed is specifically designed to accommodate turbine disc components. It can be calibrated by means of a novel mechanism to high micron precision. It consists of a matrix of Ф12 mm location holes at intervals of 50 mm as shown in Figure 49. The fixture table can be adjusted by ± 2° in all axes, to compensate for variation in the machine bed and component.
The new table is based on a general component location concept used in most CNC machines, and will enable the AWJM to reach the accuracies required for the ADMAP-GAS components. The standard machining uncertainty of such a system is approximately ± 0.010 mm.
Figure 49: Precision abrasive water jet table customised for discs
The table consists of two main components made of stainless steel and three adjustable mechanisms. The fixture rests on an aluminium frame. All washers, bolts and nuts were specially ordered as they need to be made of stainless steel.
Manufacture of the precision table took place during this project month 31 and 36. The table was manufactured by TEKS. The precision table was set up on the 3-axis machine waterjet tank, Figure 50. It was calibrated using high precision clocks clamped onto the z-carriage. Calibration was done for the x, y and z planes. The achieved calibration was 0.015 /100 mm which satisfies the objectives of precision location for the production of fir tree features in ADMAP-GAS.
Figure 50: Precision abrasive water jet table set up on the waterjet tank
As the project was initially due to finish in project month 36, USFD had committed to acquiring a 5-axis waterjet machine to take forward the ADMAP-GAS developments. This was a result of the high interest in the ADMAP-GAS research demonstrated by USFD’s mainly aerospace industrial partners which resulted in a number of feasibility studies being carried out for these partners. One result of these studies was that to realise the full benefits of the research and to take it into industry, a 5-axis machine would be required. As a result of this, the 3-axis machine was replaced by a larger 5-axis version with the capability of holding and machining very large parts. The machine has a working envelope of 6 x 4 x 1.5 meters. The waterjet head has a completely different configuration than the 3-axis having the capability of rotating the A axis ±180° and the C axis indefinitely. Unfortunately, this meant that the parts designed for the 3-axis machine needed to be modified to fit the 5-axis in a very short time-period in order to complete the validation for reporting. The results show definite improvements over the conventional head, but further developments are required to take full advantage of the 5-axis machine and USFD are committed to refining the precision table to fit with the 5-axis machine as part of their commercialisation strategy.
The precision table bracket was modified in order to make it fit into the USFD new waterjet machine. The precision table was set-up and calibrated to the 0.015 / 100 mm. The table was used to produce the aerospace demonstrator (a quarter of a gas-turbine test-piece) and the automotive demonstrator (a flange).
Improvement of hydraulic jet kinematics (avoid flow deflection - u3)
The work carried out on the improvement of hydraulic jet kinematics is focused on redesigning the nozzle orifice and building a nozzle assembly. The new design will improve the jet kinematics and improve nozzle life. The new nozzle will equip the waterjet system with a more powerful jet minimising flow deflection (u3) which will make it capable of machining to higher tolerances or higher productivity. A nozzle assembly consists of a connector and an orifice as shown in Figure 51.
Figure 51: Conventional Orifice; Din, supply diameter; Dout, outlet diameter; α, jet taper; Do, orifice diameter; Lo, orifice length
Flow deflection (u3), (Figure 52), is another machining uncertainty to be controlled, if precision cutting is to be achieved. The flow deflection heavily depends on component thickness and process parameters (different pre-programmed cut qualities, Qi). Improvements can be achieved by increasing output power.
Figure 52: Flow deflection in different cut qualities (Qi) (source: WARDjet)
In order to reduce the flow deflection, kinetic jet optimization has been carried out with the aim of increasing the cutting head power by developing a novel nozzle consisting of a new head connector and nozzle orifice.
Computer Fluid Dynamics (CFD) simulation has been used extensively in order to analyse the flow of the conventional orifice and to develop a new one. Multi-phase (water and air) models of the conventional orifice and prototype orifices were developed. Analysis in terms of the water/air volume fraction, velocity, pressure (static and dynamic) distribution and Reynolds number was carried out.
The Reynolds number gives the ratio of inertial forces to viscous forces and describes the type of fluid flow (turbulent, laminar, etc.). A low Reynolds number (<2100) implies laminar flow and will improve nozzle efficiency, power and lifetime.
CFD models of an ‘ideal orifice’ were also developed and normalised non-dimensional parameters calculated based on the ideal orifice, for comparative purposes. The ideal orifice is defined with a length / diameter ratio of 1.0. The ideal orifice assembly would have completely laminar flow, but would not be useful for AWJ applications as there is no build-up of pressure and hence the machine would not cut. The conventional orifice assembly has very turbulent flow; cutting power and efficiency would be improved by making the flow more laminar. The optimal orifice assembly would be a compromise between totally laminar and turbulent flow, where power is optimised. The three types of orifice assembly are summarised in Table 5.
Table 5: Summary of orifice assembly types
Simulations were carried out with 24 suggested geometries based on four different acceleration tapers and six different outlet tapers using both the ideal and conventional orifices. The computational work provides an estimate for the improvements in performance and tool life using the redesigned nozzle assembly. A summary of the analytical analysis is provided below, for a full report see Deliverable 3.2.
The 24 suggested orifice designs, conventional and ideal were simulated and compared in terms of maximum velocity, maximum total pressure, flux and output power. A comparison of the output power grouped by type of taper for different outlet tapers can be seen in Figure 53. The optimal geometry was selected according to its power performance.
Figure 53: Normalised output power for different orifice geometries
Simulations were carried out for the optimal orifice and results obtained in terms of orifice water fraction; velocity; total pressure; and Reynolds number for both transient and steady states. A comparison based on simulation results for the distribution of the Reynolds number in steady state for i) the ideal, ii) conventional and iii) optimal orifices is shown in Figure 54, where the white space surrounding the coloured stream represents the orifice material.
Figure 54: Comparison of distribution of Reynolds Number for i) ideal, ii) conventional and iii) optimal orifices; white space represents nozzle geometry
Table 6 provides a comparison of the theoretical conventional and optimal orifices in terms of the percentage increase in efficiency, flux, output power and maximum velocity.
Table 6: Normalised comparison of conventional and optimal orifices
Another objective of ADMAP-GAS is to increase life of the nozzle. Theoretical analysis has been carried out in order to consider tool life while designing the new orifice. A summary of the work carried out follows:
Orifices are damaged through two mechanisms: wear or breaking. Accordingly, two nozzle orifice life definitions are used: wear life and fatigue life. The shorter one will decide the orifice life.
Output power was used to calculate the orifice contribution per time unit, regardless of whether the water jet machine is cutting or not. Another important issue is the maximum water speed at the orifice where a higher speed will produce faster wear. The third item to consider is that a unit orifice radius change produces a volume increase (worn).
An orifice anti-wear index, ξ, can be defined as its output power, wout, divided by the maximum velocity, vmax, at the orifice centre times the unit orifice radius increment produced by volume induction, dV/dr, Equation [6], normalized as Equation [7]. Calculations of the anti-wear index for different ratio dr/r0 are presented in Table 7.
[6]
[7]
Table 7: Anti-wear indexes of the ideal, conventional, and optimal orifices
The normalised anti-wear index comparison of ideal, conventional, and optimal orifices is shown in Figure 55 with the radius change from 0 to 10%. Compared to a conventional orifice, the theoretical anti-wear index of the optimal orifice increases by 267% (dr/r0=0) and 447% (dr/r0=0.1).
Figure 55: Theoretical normalized anti-wear index comparison of ideal, conventional, and optimal orifice
The supply tube is full of air at the start of the process. When the water valve is turned on, the air inside the supply tube will be pressed out through the orifice. This happens very quickly because of the low viscosity of air. However, when water arrives at the orifice the speed reduces immediately because of the high viscosity of water. This happens in such a short time, Δt, that it produces a big impact, like a hammer. This phenomenon is produced by the viscosity difference, so it is called the ‘viscosity hammer’. The impact force produced is expressed in Equation [8].
[8]
In the optimal nozzle, an acceleration taper is applied before the orifice and the acceleration distance and time, Δt, are extended. Using the optimal orifice geometry, the viscosity hammer effect will be decreased and the fatigue life will be improved.
Using the optimal geometry, the Reynolds number will be reduced to 58% that of the conventional orifice, and a laminar flow will be achieved upstream of the orifice. The Reynolds number is less than 2100; hence it is assumed that a laminar flow is produced. That is the main reason why the optimal orifice is so efficient.
Based on the theoretical analysis carried out and the comparison of the optimal, conventional and ideal orifices, the following conclusions can be drawn. Using the theoretical optimal orifice increases:
• The orifice efficiency to 148%;
• The orifice flux to 220%;
• The orifice output power to 325%;
• The orifice anti-wear index to 267% (dr/r0=0) and 447% (dr/r0=0.1).
Considering the improvement of the orifice anti-wear index and the orifice impact, one can safely conclude that the orifice life will be improved more than 50% which is one of the objectives of ADMAP-GAS.
These improvements need to be validated by creating the optimal orifice geometry and connecting it to the AWJM. This was a challenging task, as the parts needed to be produced and assembled to high accuracy and tolerances, and the production and alignment of the various components was not straightforward. Initially, in order to make the nozzle orifice outlet taper an ultrasonic drill was considered. Time was spent trying to source a drill from China (the only suppliers of this equipment), acquiring drawings and making drawings to design a fixture for the ultrasonic drill (Figure 56), so small holes could be drilled in rubies. A method to drill the rubies and produce the outer tapers was devised. Equipment to this end was ordered (cleaning acid, rubies, fine wire, needles, materials, etc.). In the end the ultrasonic drill could not be bought due to good financial practice as required by University regulations, and USFD did not want to subcontract the manufacture of the rubies as they wish to patent the nozzle and are concerned to control the intellectual property, and more prosaically, there is no subcontract budget in the project. Therefore, a second plan was considered which would allow them to manufacturing the nozzle in-house. A similar ruby orifice was found – this was not an optimal orifice, but with some modification, it was expected to produce good results in conjunction with the rest of the nozzle assembly. While this will allow the theoretical results to be validated, there is less flexibility to experiment with different geometries, so a true optimal could not be tested.
Work was also carried out on the manufacture of nozzle inserts for the acceleration taper. Three different inset orifice sizes were considered - 0.10 0.20 and 0.35mm. Micro tools with a diameter of 0.066 mm were acquired in order to machine the small size inserts. Different manufacturing routes were tested. In the first one, metallic cylinders to size were drilled in steps from both ends (Figure 57) and then wire fed through them and a cone produced using a WEDM machine. There were issues with the concentricity of the hole and cylinder as the tolerances needed are tight. As a second approach, a disc was firstly made out of brass, a threading hole made and both the cone and the cylinder wired out using a WEDM. This method produced a better concentricity but changing set-ups still meant that the tolerances were not achieved.
Figure 56: Utrasonic drill fixture (approximately 120 mm x 96 mm x 40mm)
Figure 57: Drilling of a 0.07 mm threading hole, where Øi are section diameters and Li are section depths
However, the precision head connector was manufactured and tested. This is a critical component as it joins the high pressure water line to the precision head. Initially, the connector consisted of two sections with one interchangeable part where nozzle inserts with different geometries could be fitted. After testing such a design it was evident that it was prone to leaking. A second design was made which consisted of a single piece where the insert was press-fit. This design also experiences leakage in the top section which joins a seat to the actuator. The seat and actuator system close the valve when the system is not cutting.
During the last project period nozzle inserts and nozzle connectors were manufactured and tested. Assembling the different components to high concentricity and accuracy was not trivial and a number of assembly routes were also tested. There were persistent problems with the nozzles leaking and exploding under the high pressures produced by the jet. The assembly consists of three parts, seat, connector and insert. Initially, all three components were leaking. The leakage of each was addressed separately. A new design for the interface between the actuator and the seat alleviated the problem in that part of the assembly. Different designs of the connector – insert interface were tested. Most of them still leaked and the ones that did not produced a flow with high turbulence. It was believed that tolerance errors in concentricity in the axis of the assembly caused turbulence in the flow producing leaking. Different combinations of connectors an inserts were tested, however, no combination was found which did not produce turbulence or leakage.
During the last period of the project the design of the insert and connector was changed in an attempt to reduce turbulence due to the assembly and facilitate manufacturing. A form-tool was manufactured for producing the inserts using a lathe. The cylinders were turned and drilled in the same set-up allowing tighter tolerances to be achieved. The new inserts were tested (see Figure 58) with the new connector; although leakage was stopped, the system was still experiencing high turbulence producing a fan-like kerf.
From observations made during cutting, the new nozzle is significantly more powerful than the conventional one. If it can be properly manufactured, it will reduce cutting times and increase the accuracy of the jet. Even without the nozzle, the PAW system shows impressive improvements over the conventional head, and is accurate enough to perform rough-cuts on 70% of fir trees, for later finishing by wEDM or broaching. A new design team has been set up at USFD and the waterjet team are working closely with them to redesign the nozzle insert for manufacture; they expect to manufacture and validate a prototype nozzle insert within 6 months of the project end. Once the nozzle has been validated, USFD plan to patent the system.
Figure 58: Testing of the nozzle insert
Technology optimisation and process monitoring
This work has been split into two sections, technology optimisation, carried out by USFD and process monitoring, carried out by AMRC.
Technology optimisation
Work has been carried out to benchmark and characterise the AWJ system as well as investigating of process monitoring technologies. The following work has been completed:
Benchmarking: A benchmark of the pre-programmed process settings for Inconel 718 was carried out. The benchmark consisted of assessing 5 different ‘Quality settings’ which are generated with different travel rates. Relationships between travel rate, surface roughness, taper angle and level of affected layer of the sub-surface were developed. The best quality setting was considered as ‘state-of-the-art’ and the improvement in the project assessed against this.
It was found that the quality setting (based on traverse rate) has a significant effect on taper angle, sub-surface deformation and surface roughness. An interesting finding was the cutting mechanism for Inconel is by micro-cutting and ploughing. The subsurface of water-jet cut Inconel has craters. Figure 59 shows some micrographs showing the difference in surface conditions.
Figure 59: Sub-surface condition of water jet cut Inconel
Process Characterisation: Characterisation of the process of water jetting straight edges and radii in Inconel 718 was carried out. Experiments were performed in fir tree-like test-pieces; six different critical process parameters (CPP) were considered (water pressure, nozzle configuration, traverse rate, abrasive mass flow, abrasive grit and nozzle stand-of distance). The results were analysed and presented in mathematical models that relate all six CPPs to critical quality attributes (CQA); specifically: taper ratio, surface roughness and relative error of producing an internal and external radius.
An interesting finding was that the pressure has a quadratic effect on surface roughness and the relative error of producing an internal/external radius. A summary of the effect of the CPPs on CQAs is presented in Table 8. The arrow indicates the direction of change, the asterix the most important CPP and 2 a quadratic effect. For instance, the main effect for taper ratio is the nozzle configuration (Noz), which reduces with nozzle diameter. Reducing the feed rate (Feed) as well as the nozzle stand-off distance (StDis) and abrasive grit (AbTpe) reduced taper ratio; whereas reducing the water pressure (Pre) increases the taper ratio. Abrasive mass flow rate (AbRte) did not have an effect on taper ratio.
Table 8: Summary of influence of CPPs on CQAs
An example of a mathematical model showing the quadratic effect on pressure is shown in Figure 60. The model relates surface roughness at the bottom of the sample with pressure at a stand-off distance of 2 mm, abrasive rate of 0.55 kg/min. It can be seen that for example at a pressure of around 2800 and 4000 bar the same surface condition can be achieved at the same traverse rate. This has implications for the wear of the system as well as energy consumption.
Figure 60: Contour plots for surface roughness
The characterisation results were used to produce the first fir tree test-piece. This test-piece was used to assess the development of the technology and methods developed throughout the project. This was produced with non-optimal parameters, no new controller or nozzle or head.
Figure 61 presents a summary of the results in the form of a scan. The maximum error in the non-pressure flank for the front scan was 0.072 mm, 0.115 mm for the pressure and 0.034 mm for the bucket groove. The error in the non-pressure flank for the back scan was 0.179 mm, 0.276 mm for the pressure flank and 0.1136 mm for the bucket groove. The blue line shows the nominal profile whereas the green lines show the tolerance bands; any points outside these bands are shown in red.
a) Front scan b) Back scan
Figure 61: Scans of fir tree test-piece; blue lines are nominal profiles, green lines are tolerance bands and red parts are deviations
The process characterisation work was extended to incorporate sub-surface condition in waterjet surfaces. Experiments were carried out on Inconel 718 test-pieces using a Design of Experiments approach (Rechtschaffner design). Four critical process parameters were considered: traverse rate, abrasive mass flow, abrasive grit and pressure.
Measurements of surface roughness, taper angle and most important sub-surface deformation (plastic deformation and crater depth) were taken. Mathematical models were produced relating process parameters to level of deformation. Figure 62 presents the models for maximum level of plastic deformation.
The main effect controlling the level of surface deformation is the abrasive grit and mass rate. The interactions between abrasive and pressure, traverse rate and mass flow are significant. Increasing the mass flow, abrasive size reduces the maximum level of deformation. Pressure has a non-linear behaviour and needs to be optimised to reduce the level of deformation.
Figure 62: Contour plot for maximum level of deformation, µm
A model for the maximum level of crater depth is shown in Figure 63. The main effect controlling the level of depth is the abrasive mass flow followed by its interaction with pressure and abrasive type. Increasing the level of abrasive mass flow and traverse speed decreases the depth of craters; decreasing the pressure decreases crater depth.
Figure 63: Contour plot for maximum level of crater depth, µm
The level of sub-surface deformation and crater depth has been quantified and related to process parameters. These need to be optimised in order to minimise the effect of parameters on sub-surface condition. SEM work is required to quantify the level of material embedment.
USFD also investigated whether ‘washing’ the machined surfaces with the waterjet (no abrasive) has a cleaning effect on the sub-surfaces, to remove any embedded particles and ensure that the part could be finished on the WEDM. Experiments were carried out on Inconel 718 samples. In preliminary trials, a couple of samples were washed with fixed parameters (water pressure, traverse speed and jet offset). No significant effect was observed in crater depth, see Figure 64.
Figure 64: Effect of washing cuts on maximum crater depth for different grit sizes
Process Monitoring
There are several different sensing methods which could be applied to provide an input to an adaptive control system for AWJ cutting. A literature review was carried out into sensors which could be used for process monitoring. Investigations into three candidate sensors were conducted; microphones, accelerometers and acoustic emissions sensors. These sensors were analysed for their applicability to the AWJM process.
Cuts were performed at parameters known to give unsatisfactory finishes and tolerances as well as those that give good results. For the microphone, the results showed little to no correlation between the signals being gained and the quality of the cuts performed. When using the Accelerometer, the frequencies present in the signal showed a pattern in one frequency where the magnitude increased and decreased as the parameters were changed. However there was lack of repeatability on identical cuts and the results did not justify building a monitoring system based around accelerometer sensing. After a number of trials using AE sensors, it was demonstrated that machining conditions could be identified in the signals. As a result of these initial trials, the control of the AWJM used in the ADMAP-GAS is based around AE monitoring.
Continuous sampling of the AE signal will provide the feedback from the AWJ system. The signal generated during machining is used to analyse the part remotely and give vital information if losses in accuracy are detected. The monitoring provides information from the cutting process which can be fed into a signal analyser to calculate the adjustment in feed rate required to increase the machining accuracy. This is then fed back into the program and overrides the feed rate stated.
The output from the sensor feeds back into the optimal parameter generator. The information gained is analysed and put through a fuzzy logic system. This system has been developed to determine the accuracy of the AWJM process from information detected in the AE signal. If change is required, a new feed rate is generated, overriding the set feed rate.
Figure 65: System flow diagram for process monitoring
The flow diagram (Figure 65) shows the output from the sensor being fed back into the optimal parameter generator. This updates the parameters which led to poor cuts. Each time a new part is cut, more information is fed into the system, updating it and increasing the accuracy of the parameters provided.
This system will not improve the best-cut capabilities of the AWJM, however it does ensure the best process is applied to a component so that the best possible cut can be achieved each time.
The signal from the AE sensor was investigated. A fast fourier transform was used to analsye the frequencies present. The patterns from the different signals were used to detect variations in the cutting process.
Analysis of the frequency domain from the signals was carried out. The signals show clear patterns. To define these patterns, features such as peak frequency were analysed and compared with feed rate. There are two useful peaks between approximatly 50 kHz and 200 kHz. As the feed rate is adjusted, these peaks shift; at a higher feedrate the peaks have a higher frequency. In addition to the peak frequency shift there are other noticable differences in the frequency response from the signal for different feed rates. Between 200 kHz and 600 kHz there are smaller peaks which vary in size.
Each signal analysed had one or two large peak frequencies present. The largest peak frequency was analysed by measuring the area under a section of the curve 10 KHz to either side of the peak. Figure 66 shows how the different feed rates resulted in noticeable variations in the maximum frequency area.
Figure 66: Relationship between the feed rate and the peak frequency
The second pattern is the change in magnitude of the three smaller frequencies in the range of 200 – 400 kHz. It was noted that a cut which resulted in an out of tolerance feature also reduced the magnitude at these smaller frequency peaks. Figure 67 shows how the feed rate affects the magnitude of these peaks. As with the maximum peak, these have also been normalised by the distance from the sensor.
Figure 67: Graph showing relationship between the areas under the triple peak frequencies
The third pattern identified was that of the RMS of the frequency response from the AE signal. The relationship is not as strong as the previous two, however it can supply a third monitoring factor to ensure that the signal is correctly analysed.
It was found that with the introduction of a database for the target component application, geometric accuracy could be predicted from the AE measurements. This allows the system to detect whether the AWJM cut is within the target tolerance band of the root slot feature.
Optimisation of the signal analyser was carried out. This optimisation looked at the initial relationships between the cutting process and component accuracy and refined them through additional experiments. Four test pieces, as shown in Figure 68, were machined using different traverse rates for each of the slots. This resulted in slots of varying accuracy. The results were measured and the relationship between the traverse rate and the signal form the AE sensor was analysed as described previously (Figure 66 and Figure 67) Figure 67 during the initial system development. This second set of experiments gave a much more accurate relationship between the traverse rate and accuracy as more traverse rates were tested with a greater number of repeated results.
Figure 68: Firtree test piece.
These experiments used the system to predict process accuracy and compared this to the measured results from traditional methods. This information allowed the fuzzy logic analyser to become more robust. The main functionality of the system remains the same as previously demonstrated. This work concluded the development of the process monitoring system.
The process monitoring system was fully developed in the previous periods (project month 32 - 36). Validation of the system is discussed in Section Evaluation of controller.
Development of control strategies
The control of the waterjet tool path can be used to compensate for the known errors in the system. In addition, feed rate regulation ensures the optimal cutting parameters are used throughout the complete tool path. Figure 69 shows how these values vary for an example tool path. In a root form slot, the variation in parameters as the jet travels around the tight radii is more extreme than in this example. It is because of this relationship that changes in the feed rate playing a large role in the tolerances that an AWJM cut part achieves. Details can be found in D3.3.
Figure 69: Example of Curvature for a Sinusoidal Tool Path.
An optimal parameter generator was developed. This system used a rate of curvature algorithm to assess the rate of change of the linear feed. This allows tool paths to be created which are more accurate than is currently possible with state of the art machines. The aim of this was to apply optimal feed rates to the different features found in the fir tree profile.
The system calculates an optimal feed rate which is based on the curvature and the rate of change of curvature of the geometry at each point. The system was developed to detect the following three situations and to analyse the feed rate:
• Zero Curvature: Areas of zero curvature such as the short straights on each flank and at the base of the profile allow a constant and optimised feed;
• Entrance/Exit into Corner: These areas have zero curvature just before the radius and a constant curvature once in the radius. Between these two points, an instantaneous change in the curvature occurs. The instantaneous change results in a sudden acceleration of the jet and a change of direction. It is this area which results in the worst loss of accuracy from the jet;
• Main Radii: These have a constant curvature value and as a result, zero rate of curvature. This means a constant feed rate can be applied as the acceleration of the jet is constant.
A look-ahead function was also applied. This is done to compensate for the acceleration of the machine and analyses the tool path and applies the adjusted feed rates in advance.
Experiments were carried out to characterise variations in the cutting jet. This was done indirectly by cutting slots into a test piece through a 0.5 mm nozzle, where a linear relationship was assumed to exist between the slot width produced and the jet size.
Figure 70: Jet Variation Results
As can be seen from Figure 70, the average slot width (and therefore jet size) centres around the orifice diameter of 0.5 mm. However the range of the results shows a large distribution. The variation in the jet width is from 0.4 mm - 0.69 mm, a spread of 0.39 mm. It is necessary to compensate for this variation if the AWJM is to continually achieve tolerances of +/- 0.3 mm.
The monitoring and control system was tested on the WARDjet ZX-613 at the AMRC. For each experiment, a full test piece, as defined in D3.1 was machined, as described in D3.3.
It was shown that increases in the accuracy could be achieved using the optimal parameters along the tool path. This increase in accuracy is different for the different features and is summarised in Table 9.
Table 9: Improvements to the state-of-the-art accuracies achieved by the controller
The experiments conducted previously showed potential for further improvements. The average error seen was higher than the nominal measurement. To compensate for this and to improve the accuracy of the system, additional experiments were carried out. The experiments consisted of cutting simple representative geometries of firtree profiles as shown in Figure 68. Four of these test pieces were machined. The parameters for the machining were different for each radius on the firtree profiles. The firtree profiles were measured after machining; this was used to link the accuracy achieved to the parameters used.
The resultant relationships between parameters and accuracy were used in the control system to enhance the parameter generation algorithms. The results of these additional experiments were used to improve the controller’s accuracy by applying optimised parameters for features to the firtree profile. The process monitoring system was used to measure test pieces machined on the AWJM. These results were fed into the control system to improve the self-learning algorithms. The system was validated by cutting a final test piece to determine the accuracy achievable.
The system was modified to accommodate the new 5-axis machine implemented at USFD. Modifications were made by calibrating the new nozzle for use on the 5-axis machine (as described previously for the 3-axis machine). Test cuts were made for known dimensions, measured, and the error of the machine was calculated. This error was then applied to the system as a correction.
Once the nozzle on this new machine was calibrated, the test piece shown in Figure 68 was machined to confirm the accuracy; the results demonstrated that the implementation of the controller with the 5-axis machine matched the results obtained on the 3-axis machine. This confirmed that the controller was correctly integrated.
Evaluation of Results
Evaluation of PAW system
Comparing the conventional AWJ system to the theoretical PAW, machining uncertainty, Ui, has been significantly reduced, as demonstrated in Table 10.
Table 10: Machining uncertainty comparison
Evaluation of controller
A Demonstration component was then machined using the new 5-axis machine at USFD. The fir trees were measured and the results are shown in Table 11.
Table 11: Fir tree tolerances achieved using the ADMAP-GAS controller, precision table and the 5-axis machine
The demo piece was a stage model of the final part with a constant stock level of 60 microns on all surfaces. A stage model is representative of a roughing process, with the resultant part to be finished on a secondary process, in this case, WEDM. Figure 71 shows the stage model drawing to which this final demonstration piece was machined.
Figure 71: Drawing of fir tree part machined for demonstration
Table 11 shows that by using the controller and the precision system developed in T3.2.1 and T3.2.2 but not the nozzle, the accuracy of the water jet has been improved by 61% for radii on fir tree profiles. Previously a tolerance of +/- 0.235 mm was possible; the new system delivers +/- 0.09mm. The improvement in accuracy in the pressure flank is 22%. These improvements do not reach the requirements for finishing the fir trees using a single EDM cut. With the addition of the PAW nozzle developed by USFD, as well as compensation of the linear taper by the 5-axis machine, it should be possible to achieve this target.
During the development phase, the control system was tested on the Wardjet machine at USFD. Implementation on multiple platforms is part of AMRC’s commercialisation strategy.
This system has been developed as a standalone package which can be integrated with industry-standard CADCAM packages which interface with AWJ machines. After the project, the system requires further testing before it can be developed into a commercial piece of software. This will be carried out during the first year and AML will deploy the new controller within their manufacturing department.
Environmental Impact of the Abrasive WaterJet Machining
Ecoprofile data and assumptions
The ecoprofile of the developed AWJM process has been prepared on the basis of the experimental data produced in the WP3 and WP4 cutting tests. Additional inputs have been received from WardJet, who manufacture the machine used by USFD, in particular the data on the functional life of the waterjet machine consumable parts (filters, orifice, focusing tube) and state-of-the-art methods for recycling water and abrasive. In term of process inputs and outputs, the AWJM roughing and the AWJM finishing are very similar, therefore the considerations reported here are valid for both the processes.
The most relevant process data are summarized below:
• The measured cutting time per fir tree is 5 minutes. This value has been obtained from real cutting tests using the selected fir tree geometry and turbine disk thickness.
• The energy consumption when cutting one fir tree is 2.708 kWh. It is important to observe that the energy for roughing and finishing is assumed to be the same, because it depends largely on the pump pressure, which is 6000 bar in both cases. The compressed air consumption is negligible.
• The abrasive adopted is a garnet (Ca3Fe2(SiO4)3), mesh 80 corresponding to a particle diameter of 300-150 µm, and a hardness which is 6.5 - 7.5 on the Mohs scale. The abrasive rate is 0.55 (kg/min), corresponding to a consumption of 2.5 kg of garnet per fir tree: 1.25 kg is virgin abrasive, whereas 1.25 kg is recycled abrasive.
• With the ADMA-GAS developments, the maximum the softened water flow rate (while cutting) is 4.163 kg/min, corresponding to a maximum water requirement of 20.819 kg per fir tree. After cutting, the water is recovered and recycled.
Some maintenance consumables have been considered for the waterjet machine:
• The filters are made in polypropylene (130 g). The functional life of the filters depends on many factors, but an average value of 2000 h can be used.
• The focusing tube is made of boron carbide (43 g) and lasts 100 h.
• The orifice is made of ruby or diamond; a diamond orifice is guaranteed for 500 h, whereas a ruby orifice should last about 80 h. In the experiments a ruby orifice has been used (3 g).
AWJM of fir trees produces the by-products and waste described below:
• Two accessory systems are available for recycling the water and the abrasive: the first system extracts the garnet from the cutting fluid, the second one recovers, filters and treats the water. Almost all the garnet is recovered, but it can be only partially reused for the fir tree cutting because the particles break when cutting Inconel, reducing their mesh. For this reason, only 50% – 60% of the garnet can be reused directly, while the remaining amount is only suitable for cutting softer materials; this garnet at reduced mesh size is considered in the LCA as a by-product of the AWJM. When the garnet has been pulverized and it is too fine to reuse for waterjet cutting, it can sometimes be used in other industries. There have been cases where people have sold their used abrasive for use in road beds. Other uses could potentially be found: as long as the material that is being cut is not hazardous, options like this could be viable. All the water recovered can be recycled. It has been assumed that there is negligible refill due to evaporation or component wetting. A simplified LCA calculation of recycling systems has been done, considering the energy consumption (0.183 kWh per fir tree) and neglecting the consumption of compressed air and consumables.
• Unlike broaching, the AWJM does not produce a mixture of oil and different metal chips, but a monolithic scrap of Inconel 718. This can be directly recycled by the foundries without any treatment, as it not polluted.
• The boron carbide, the ruby and the polypropylene from the end of life of the focusing tube, orifice and filters, can be sold as by-products and recycled by specialized industries.
Life Cycle Inventory
Adopting one turbine disk with 48 fir trees as the functional unit, a Life Cycle Inventory (LCI) of the eco profile “AWJM single cut” has been prepared, describing the significant inputs and outputs. The LCI data are summarized in Figure 72.
Figure 72: Life Cycle Inventory for the Ecoprofile “AWJM single cut”
Impact Assessment and Interpretation
The data collected in the Life Cycle Inventory have been elaborated using SimaPro7 Analyst software (PRé Consultants), based on Ecoinvent database and customized with DIAD’s databases. The Ecoinvent database represents the result of a large effort by Swiss institutes to update and integrate the well-known ETH-ESU 96, BUWAL250 and several other databases.
The LCA has been carried out, considering a 48 slot fir tree and adopting the LCA method CLM 2 as the baseline method conforming to ISO 14000. According to this method the main impact categories considered are:
Climate change - Climate change can result in adverse effects upon ecosystem health, human health and material welfare. Climate change is related to emissions of greenhouse gases to air. The characterization model as developed by the Intergovernmental Panel on Climate Change (IPCC) is selected for development of characterization factors. Factors are expressed as Global Warming Potential for time horizon 100 years (GWP100), in kg carbon dioxide/kg emission. The geographic scope of this indicator is at global scale. Some characterization factors were added from the IPCC 2001 GWP 100a method: Methane, bromodifluoro, Halon 1201, Methane, dichlorofluoro, HCFC-21, and Methane, iodotrifluoro. Stratospheric Ozone depletion: the characterisation model is developed by the World Meteorological Organisation (WMO) and defines ozone depletion potential of different gases (kg CFC-11 equivalent/ kg emission).
Stratospheric Ozone depletion - Because of stratospheric ozone depletion, a larger fraction of UV-B radiation reaches the earth surface. This can have harmful effects upon human health, animal health, terrestrial and aquatic ecosystems, biochemical cycles and on materials. This category is output-related and at global scale. The characterization model is developed by the World Meteorological Organisation (WMO) and defines ozone depletion potential of different gasses (kg CFC-11 equivalent/ kg emission). The geographic scope of this indicator is at global scale. The time span is infinity.
Acidification - Acidifying substances cause a wide range of impacts on soil, ground-water, surface water, organisms, ecosystems and materials (buildings). Acidification Potentials (AP) for emissions to air are calculated with the adapted RAINS 10 model, describing the fate and deposition of acidifying substances. AP is expressed as kg SO2 equivalents/ kg emission. The time span is eternity and the geographical scale varies between local scale and continental scale. Characterization factors including fate were used when available. When not available, the factors excluding fate were used (In the CML baseline version only factors including fate were used). The meth-od was extended for Nitric Acid, soil, water and air; Sulphuric acid, water; Sulphur trioxide, air; Hydrogen chloride, water, soil; Hydrogen fluoride, water, soil; Phosphoric acid, water, soil; Hydrogen sulphide, soil, all not including fate. Nitric oxide, air (is nitrogen monoxide) was added including fate.
Human toxicity - This category concerns effects of toxic substances on the human environment. Health risks of exposure in the working environment are not included. Characterization factors, Human Toxicity Potentials (HTP), are calculated with USES-LCA, describing fate, exposure and effects of toxic substances for an infinite time horizon. For each toxic substance HTP’s are expressed as 1,4-dichlorobenzene equivalents/ kg emission. The geographic scope of this indicator determines on the fate of a substance and can vary between local and global scale.
Figure 73 presents the result of the LCA analysis. In particular, the energy consumption results as the main factor in all the impact categories of AWJM, except for the fresh water aquatic eco toxicity, where the consumption of abrasive garnet also has a considerable impact.
Figure 73: Characterization of impact assessment of the AWJM single cut, where red is the abrasive garnet, light blue is the focusing tube, dark blue is the electricity mix, yellow are filters and turquoise of the orifice
Figure 73 shows that energy consumption is the main factor in all the impact categories of AWJM, except for the fresh water aquatic eco toxicity, where the consumption of abrasive garnet also has a considerable impact. To reduce the environmental impact of AWJM, it will be important to reduce the energy consumption which depends on the pump power, cutting time and water-garnet recycling systems. Table 12 shows the values of the estimated environmental impacts of AWJM roughing.
In order to have a more general view of the environmental impact of AWJM, it is important to not focus only on the process itself, but to extend the boundaries to the integrated W2 manufacturing chain. Figure 74 shows the impact of the integration of AWJM roughing and WEDM finishing is shown. The impact of WEDM is higher than AWJM for all the categories considered, in particular for fresh water and marine aquatic toxicity. In these categories the incidence of the production of the copper/zinc wire is very high.
Table 12: Characterization of impact assessment of the AWJM single cut: detailed data table
Figure 74: Characterization of impact assessment of the Ecoprofile “AWJM roughing + WEDM finishing”, where blue is AWJM and green is WEDM
It is important to remember that the WEDM process is composed of 3 separate cuts (2 precision cuts and 1 surfacing cut) to compensate for the taper angle and abrasive contamination from the AWJM. Moreover the environmental impact of WEDM and the W2 process chain would be significantly reduced by increasing the precision of the AWJM. Table 13 gives detailed values for the estimated environmental impacts of the integrated process AWJM roughing + WEDM finishing.
Table 13: Characterization of impact assessment of the Ecoprofile “AWJM roughing + WEDM finishing”: detailed data table
The Life Cycle Assessment carried out was focused on the AWJM cutting and in particular the manufacture of 48 fir trees for an aeroengine turbine disk. On the basis of the data collected and assumptions made, it has been observed that:
• Energy consumption is the main factor in all the impact categories of AWJM, except on the fresh water aquatic ecotoxicity, where the consumption of abrasive garnet has a relevant impact.
• The energy consumption of AWJM depends mainly on 3 elements:
o the pump power,
o the cutting time,
o the recycling systems (water and abrasive),
• The abrasive impact is mainly seen in energy use for recovering and recycling.
These are the main factors to be taken into account for reducing the environmental impact of AWJM (more details are documented in WP4 summary). Considering the integration of the AWJM in a process chain, it is clear that its precision allows a reduction of finishing operations, therefore:
• The quality of the AWJM is an additional factor relevant for the reduction of the environmental impact of the overall chain.
• The LCA must be used to define, case by case, the best environmental compromise between the cutting time and the cutting precision.
Potential Impact:
Within this section of the final report the potential impact and main dissemination activities and exploitable results are presented. Based on an ADMAP-GAS demonstrator geometry, which has been defined during the project, three potential new process chains are introduced:
1. Wire-EDM roughing and finishing
2. AWJM roughing and finishing
3. AWJM roughing and Wire-EDM finishing
Data which have been measured during the production of the demonstrators are used to set up a Life Cycle Cost Analysis and a Life Cycle Assessment to compare the new process chains and especially the new developed processes with the state of the art broaching process. Here, costs, the environmental impact and social aspects are the main assessment criteria. Finally, the main dissemination activities in terms of publications and the exploitable results of the partners are summarized.
Definition of the section of turbine disk
The demonstration of turbine disk manufacturing has been done on a quarter of turbine disk to save costs and time. The slot profile adopted does not correspond to any drawing of specific aircraft engine, but it has been conceived as an integration of critical geometries and requirements from different aero-engines manufacturers (Rolls Royce, MTU, Avio). In Figure 75 the design of the demonstrator is presented.
Figure 75: Design of the section of turbine disk demonstrator
Demonstration of turbine disk manufacturing
In Figure 76 the demonstration of AWJM fir tree manufacturing is shown. The measurements carried out show that a tolerance of +/- 20 μm has been achieved for the geometries, which was the goal to finally finish the profile by WEDM.
Figure 76: AWJM cutting of fir trees on the section of turbine disk
The main conclusions are:
• It has been demonstrated that AWJM fulfils the quality requirements for an efficient integration with WEDM finishing.
• At the present, the nozzle geometry limits the value of radii that can be cut, but shortly the new nozzle geometry will overtake this limit.
• Further improvement of AWJM quality is expected from 5 axis machines with postprocessor compensating tapper angle and new nozzle increasing the cutting power. This will reduce the WEDM finishing cycle, resulting in a significant increase of productivity and a decrease of energy consumption (costs and environmental impact).
• Tolerances of +/- 5 μm can be achieved with the precision table and precision head (PAW - Precision Abrasive Waterjet).
• A limiting factor for water-jetting of a fir tree to a finishing stage is the material embedment. In fact the turbine disk is contaminated by abrasive particles. It is expected to reduce this effect by the adoption of the new nozzle geometry and optimised cutting.
Figure 77 shows different phases of fir trees cutting by WEDM including positioning of the demonstrator on the WEDM machine working area, rough / precision cut and surfacing cut:
Figure 77: WEDM cutting of fir trees in the section of turbine disk
The conclusions of this approach are:
• It has been demonstrated that WEDM fulfils the quality requirements for the substitution of broaching,
• WEDM can be applied both as stand-alone slotting or it can be integrated as finishing of AWJM slotting and
• the integration with AWJM slotting can increase the productivity with respect to WEDM, but this aspect is strictly dependent to the AWJM precision.
LCC-modelling of complete turbine disk manufacture
For each process combination, three phases of development have been considered during a LCC analysis. The phases correspond to a different degree of development of technologies: Intermediate results, final achievements and further developments after the end of the project:
• Phase 1: Achieved progress at Month 30 (phase 1)
• Phase 2: Achieved progress at month 42 (phase 2)
• Phase 3: Expected progress month 54 – 12 months past the project (phase 3).
AWJM fir tree manufacturing
The cutting time is average 5 min per fir tree. The different phases of AWJM development are focused on the improvement of the accuracy of cutting and not on the cutting rate, because the geometry and surface quality are critical, whereas the cutting time is already comparable to broaching.
• Phase 1: The initial quality of AWJM was not sufficient, as it was completely impossible to achieve the required tolerance range for fir tree geometry (+/- 5 μm) and the surface roughness was outside the required specification up to a factor of 14.
• Phase 2: The optimisation of the AWJM and the integration of a precision table allowed achieving a tolerance of +/- 20 μm. This value is sufficient for the integration with WEDM finishing, but it is still not enough for the substitution of broaching. The nozzle geometry (0.3 mm) limits the value of radii that can be cut.
• Phase 3: In next industrialisation phase, the adoption of the new nozzle geometry (0.1 mm) will allow to machine all the radii of the fir tree, also increasing the cutting power. A reduction of about 40% of energy consumption is expected. The implementation of the 5 axis precision head with postprocessor compensating of the tapper angle will constitute the further step in the technology. PAW will then achieve a geometrical tolerance of +/- 5 μm which is finally comparable to broaching.
WEDM fir tree manufacturing
The optimized WEDM slotting process chain is composed of the sequence of three cuts: “E2”, “E7” and “E21”. “E2” is a rough cut, “E7” is a precision cut and “E21” is a surfacing cut.
• Phase 1: The initial mean cutting speed of the sequence was 0.99 mm/min, corresponding to a cutting time per slot around 78 min.
• Phase 2: Thanks to further optimisation of the rough cut, the cutting speed has been increased to 1.8 mm/min, corresponding to a cutting time of 43.3 min per slot.
• Phase 3: Further increases of cutting rate are expected from next industrialisation of the process, but at the present it is not possible to quantify this. As example, an option would be the modification of the cut settings, achieving a reduction of the sequence.
AWJM + WEDM fir tree manufacturing
• Phase 1: Surface quality after AWJM necessitates the WEDM sequence “E4”, “E4”, “E7” and “E21” resulting in a mean WEDM speed of only 0.8 mm/min. The cutting time is therefore 97 min per fir tree slot in average. In this condition the integrated process shows a productivity lower that the WEDM slotting.
• Phase 2: The latest cutting trials indicate that it is possible to only use a WEDM cutting sequence “E7” and “E21”, resulting in a mean WEDM speed of 3.4 mm/min. The cutting time is then 22.9 min per fir tree in average.
• Phase 3: Better AWJM process control (5-axis machine, AMRC post processor and monitoring) allows the elimination of the “E7” cut, leaving only one ”E21” cut. In this case the cutting speed reaches 8.8 mm/min, resulting in an average cutting time per slot of 9 min. The energy consumption of the AWJM process is considerably reduced by the application of the new nozzle geometry.
Life Cycle Cost analysis
In Figure 78 the Life Cycle Costs are presented of the different W2 process chains proposed. The analysis clearly shows that the LCC/part of the integrated process AWJM+WEDM will become competitive against the state of the art broaching process within the phase 3 (estimated 12 months after the project finish).
Figure 78: LCC of different slotting processes for a turbine disk (48 fir trees)
Life Cycle Assessment
In Figure 79 the evolution of the impact of W2 processes on the global warming (GWP100) and in comparison with broaching process impact (considered constant) is presented. Similar results have been obtained for all the environmental impact categories of CLM 2 baseline method (conform to ISO 14000). All the details concerning the LCA methodology, inventory data and assumptions are reported in the deliverable D4.2. It can be concluded that:
• All the developed W2 processes achieved a reduction of environmental impact with respect to broaching.
• WEDM fir tree manufacturing is the technology with higher degree of readiness at the end of the project and this is reflected by the corresponding low environmental impact.
• AWJM slotting will express higher environmental benefits after the end of the project, when all the technologies developed will be fully implemented.
• AWJM roughing + WEDM finishing, as summary of the improvements of both of the technologies, is expected to represent at the month 54 the most environmental friendly solution for the substitution of broaching.
Figure 79: Evolution of the impact of W2 processes (GWP100) with respect to broaching
Adopting the “Damages Evaluation” of Eco-indicator 99 method, it is possible to obtain a comparison of the different turbine disk slotting processes more oriented to social aspects. In particular in Figure 80 the evolution of the human health damage of developed slotting processes is presented. The diagrams clearly show that all the W2 processes have achieved a significant reduction of the impact on human health at the end of the project and further improvements are expected by the month 54.
Figure 80: Evolution of the impact of W2 processes on Human Health respect to broaching
Assessment of W2 fir tree manufacturing respect to broaching
On the basis of the demonstrations done, in Figure 81, Figure 82 and Figure 83 the achievements of W2 slotting technologies with respect to state of the art broaching are summarised.
Figure 81: Achievements of W2 slotting with respect to broaching
Figure 82: Assessment of the application of W2 technologies for the fabrication of turbine disks
Figure 83: Assessment of the application of W2 technologies for the fabrication of turbine disks
Publications
• Klocke, F.; Welling, D.; Veselovac, D.; Perez, R.; Nöthe, T.: Wire-EDM developments for manufacturing safety critical turbine components features in Inconel 718, Veranstaltung: "6th Int. Conference “Supply on the wings” Aerospace – The global innovation driver". Frankfurt, 04.11.2011. 2011.
• Klocke, F.; Zeis, M.; Welling, D.; Garzon, M.; Klink, A.; Veselovac, D.: Aktuelle Forschungsschwerpunkte im Bereich Funkenerosion am Werkzeugmschinenlabor WZL der RWTH Aachen, in: Der Schnitt- & Stanzwerkzeugbau 18 (2011), 5, S. 8-14.
• Klocke, F.; Zeis, M.; Welling, D.; Garzon, M.; Klink, A.; Veselovac, D.: Aktuelle Forschungsschwerpunkte im Bereich Funkenerosion am Werkzeugmaschinenlabor WZL der RWTH Aachen, in: Der Stahlformenbauer 28 (2011), 5, S. 6 – 12.
• Klocke, F.; Welling, D.; Dieckmann, J.: Comparison of grinding and Wire EDM concerning fatigue strength and surface integrity of machined Ti6Al4V components, in: Procedia Eng. 3 (2011), 19, ISSN 1877-7058, S. 184–189.
• Welling, D.: Drahtfunkenerosion im Turbinenbau, in: Tagungsband zur 8. Fachtagung Funkenerosion, Hrsg.: Klocke, F., Apprimus Verlag/ WZL-Forum Aachen 2011, S. Kap. 7, S 1 – 25.
• Klocke, F.; Zeis, M.; Klink, A.; Welling, D.; Baumgärtner, M.: Technological and Economical Comparison of Pre-manufacturing Titanium- and Nickel-based Blisks via Milling from solid and EDM, Veranstaltung: "EUCOMAS 2012". Hamburg, 07.-08.02.2012. 2012.
• Klocke, F.; Welling, D.; Dieckmann, J.; Klink, A.: Titanium Parts for Medical Sector Made By Wire-EDM, in: Proceedings of the 1st International Conference on Design and Processes for Medical Devices: Padenghe sul Garda (Brescia), Italy, 2 - 4 May 2012 , Hrsg.: Ceretti, E.; Fiorentino, A.; Giorleo, L.; Giardini, C., Neos Edizioni Brescia/ Italy 2012, ISBN 9788866080589, S. 163-166.
• Klocke, F.; Welling, D.; Dieckmann, J.; Veselovac, D.; Perez, R.: Developments in Wire-EDM for the manufacturing of fir tree slots in turbine discs made of Inconel 718, in: Key Engineering Materials 1665 (2012), 504-506, ISSN 1013-9826, S. 1177-1182.
• Trauth, D.; Klocke, F.; Welling, D.: Auswirkungen des elektro-mechanischen Festklopfens auf den Randschichtzustand und die Dauerfestigkeit einer Nickel-Basis-Legierung, Veranstaltung: "53. DGM Fachauschusstagung "Mechanische Oberflächenbehandlungen". Rastatt/Malsch, 18.04.2013. 2013.
• Escobar G, Gault RS & Ridgway K, 2011, “Characterisation of water-jet process for the production of fir trees in Inconel 718”, 6th International conference Supply on the Wings (AIRTEC), Frankfurt Germany, Nov 2-4.
• Escobar G, Gault RS & Ridgway K, 2012, “Characterisation of abrasive water-jet process for pocket milling in Inconel 718”, CIRP High Performance Cutting, Zurich 4-7 June.
• GA Escobar-Palafox, RS Gault, K Ridgway, 2012, “The Effect of Abrasive Water Jet Process Variables on Surface and Subsurface Condition of Inconel 718”, Advanced Materials Research vol 565, pp 351-356.
• Paper submitted to INTED 2012 - “Knowledge transfer: an EC funded research project case study”.
Conferences and presentations:
• 09.-10.10.2012 Presentation of ADMAP-GAS demonstrator at BRAGECRIM, Meeting at WZL Aachen (http://www.bragecrim.rwth-aachen.de/)
• 09.11.2012 Visit and getting insight into ADMAP-GAS project for Mechanical Engineering students (150 students) of RWTH Aachen University
• 28.-29.11.2012 Discussion about ADMAP-GAS results at MTU
• 20.-21.02.2013 Presentation of ADMAP-GAS demonstrator at ICTM in Aachen (http://www.ictm-aachen.com/)
• ESAFORM Conference 2012 – Title: Developments in Wire-EDM for the manufacturing of fir tree slots in turbine discs made of Inconel 718, 09.2012
Activities in 2011:
• ICTM at WZL/IPT RWTH Aachen University 02.2011
• Presentation at Aerodays, Madrid 03.2011
• AWK at WZL RWTH Aachen University 05.2011
• Presentation at the Supply on the Wings in Franktfurt 11.2011
• Presentation at 8. Fachtagung Funkenerosion, WZL RWTH Aachen, 11/2011.
• Rolls-Royce waterjet Community of Practice meeting, University of Sheffield, 2nd November 2012.
• Technical Fellows conference, University of Sheffield, 30th October 2012.
• Water jet Open Day, University of Sheffield, 30th and 31th January 2013, with accompanying article on the AMRC website. This event was reported by a number of publications, which also mentioned ADMAP_GAS.
Exploitation of results
Through the ADMAP-GAS project, 12 exploitable results have been developed. A summary of these results can be seen in Table 14. It is evident from the results that the exploitable results developed during the ADMAP-GAS project have the potential to impact many industries across the world. More pertinently, the refinement and adaptation of AWJC and HS-WEDM during the project could lead to large economic and environmental improvements in manufacturing processes.
• Table 14: List of exploitable results developed through the ADMAP-GAS project
Number Exploitable Result Timetable to Commercial Use Partners Involved
1 NEW FAMILY OF WEDM WIRES Sep-14 BEDRA, WZL,
AGIE-CHARMILLES
2 NEW WEDM MACHINE INTEGRATING INNOVATIVE MONITORING SYSTEM Jan-13 AGIE-CHARMILLES, WZL
3 NOZZLE GEOMETRY AND GENERATOT TECHNOLOGY Jan-13 AGIE-CHARMILLES, WZL
4 NEW BASE FLUID Mar-12 BEDRA,
AGIE-CHARMILLES, OELHELD, WZL, DIAD, TEKS
5 NEW ADDITIVES Mar-12 BEDRA,
AGIE-CHARMILLES, OELHELD, WZL, DIAD, TEKS
6 OPTIMIZED WATERJET MACHINE
(CUTTING HEAD AND TABLE) Nozzle: July-13; Head: Dec-13; Table: July-13 USFD
7 PROCESS MONITORING SUITE AWJM Mar-13 AMRC LTD
8 WEDM MONITORING SYSTEM AGIE-CHARMILLES, WZL
9 TEST BED DEVELOPMENT WEDM 5 years after project end AGIE-CHARMILLES
10 NEW PROCESS ROUTE AUTOMOTIVE SECTOR 1 year after project end AGIE-CHARMILLES, USFD, WZL, DIAD
11 NEW PROCESS ROUTE MEDICAL APPLICATION 1 year after project end AGIE-CHARMILLES, USFD, WZL, DIAD, TEKS
12 NEW PROCESS AEROSPACE APPLICATION Within 10 years after project end All project partners
List of Websites:
http://www.admapgas.com/
Project’s Coordination:
Organisation Name: WZL – RWTH Aachen
Representative Name: Dražen Veselovac
Tel: +49-241-80-27432
Fax: +49-241-80-22293
E-mail: d.veselovac@wzl.rwth-aachen.de
