Video based Process Observations of the Pulse Electrochemical Machining Process at High Current Densities and Small Gaps

Available online at www.sciencedirect.com

ScienceDirect Procedia CIRP 14 (2014) 418 – 423

6th CIRP International Conference on High Performance Cutting, HPC2014

Video based process observations of the pulse electrochemical machining process at high current densities and small gaps Andreas Rebschlägera,*, Rouven Kollmannspergera, Dirk Bähreb a

Center for Mechatronics and Automatization, Gewerbepark Eschberger Weg Geb. 9, D-66121 Saarbrücken b Institute of Production Engineering, Saarland University, D-66123 Saarbrücken

* Corresponding author. Tel.: +49 681 85787 22; fax: +49 681 85787 11. E-mail address: [email protected]

Abstract Pulse Electrochemical Machining (PECM), a nontraditional process, using pulse-lengths in the low millisecond range as well as feed overlaid mechanical vibration, allows more precise tolerances and geometric precision through narrowing the working gap compared to conventional sinking ECM. With small working gaps in ranges down to 10μm, the anodic shape evolution during machining is getting difficult to monitor. Therefore understanding the shaping phenomena during the PECM process is key factor in achieving precision during the manufacturing of dies and molds, as well as precision parts in e.g. automotive or aircraft industry. In this contribution an experimental approach towards visual in-process observations of the PECM shaping process during the use of mechanical vibrations up to 50Hz and high pulsed current densities will be presented. Recording the process with a precisely clocked high speed camera system allowing precise μs shutter times, visual observations are conducted and being used as input for detailed downstream data analysis. The experimental study incorporates one of the most widely used flushing conditions in PECM as well as an outlook into the comparison between recorded in-process data and a static FEM simulation based on the monitored shape are given. In all experiments stainless steel of type AISI 304 (X5CrNi18–10) is used as anode and cathode material and for all PECM experiments a commercially available PEMCenter8000 with sodium nitrate as electrolyte was used. The concept presented will help to better link experiment and modelling of the PECM process, by simultaneously providing process relevant electrochemical data as well as the directly corresponding geometric shaping information during experiments.

© 2014 Elsevier B.V. This is an open access article under the CC BY-NC-ND license

© 2014 The Authors. Published by Elsevier B.V. (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the International Scientific Committee of the 6th CIRP International Conference on High Selection and peer-review under responsibility of the International Scientific Committee of the 6th CIRP International Conference Performance Cutting.

on High Performance Cutting

Keywords: Identification, Precision, Electro chemical machining (ECM)

1. Introduction Precision in Pulse Electrochemical Machining (PECM) - as a process development of the Electrochemical Machining (ECM) [1] - is dependent on detailed information about the material dissolution behavior under a variety of electrochemical conditions. Especially for modelling the process, detailed datasets are necessary [2, 3]. Conventionally, this knowledge is acquired through vast sets of experiments under different voltage, electrolyte, current density, feed etc. conditions [4]. This effort to understand the electrochemical behavior of a single material is very time consuming and requires multiple experiments, mainly under laboratory rather

than actual production conditions. A good example of the amount of experiments necessary to investigate a single material can be found in the work of Altena [4]. Yet, herein no actual geometric shaping experiments were in the focus of the investigations. The main aim of this contribution is to present a new approach to combine aspects of determining the material dissolution behavior and at the same time also allowing geometric shaping experiments under industrial boundary conditions. Starting with a reference set of conventionally acquired material dissolution data the experimental setup, boundary conditions and downstream data analysis possibilities are presented. Unlike usual geometric shaping experiments, which most of the time only allow

2212-8271 © 2014 Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). Selection and peer-review under responsibility of the International Scientific Committee of the 6th CIRP International Conference on High Performance Cutting doi:10.1016/j.procir.2014.03.105

419

Andreas Rebschläger et al. / Procedia CIRP 14 (2014) 418 – 423

geometric measurements before and after the experiment, this contribution will show the possibility to acquire geometric data during the experiments. 2. Experimental setup and equipment 2.1. The PECM process The PECM process, schematically shown in Fig. 1, is a variation of the ECM process. During this process, the feed towards the workpiece (anode) is overlaid with a mechanical oscillation of the tool (cathode). The amplitude of the oscillation in this contribution is 200 μm, which results in two different process phases. During the minimum gap size, a pulsed current with a pulse duration ranging from 0.1-5 ms can be applied. The small gap size, achievable through the oscillation of the cathode, and the short current pulses of up to 8,000 A lead to an effective material removal process resulting in good surface quality and precise copying accuracy [1]. The upward movement during the oscillation results in the phase of maximum gap size, which enables enhanced flushing possibilities and consequently a better removal of the processed material as compared to the conditions at minimum gap size. Cathode Cathode

Cathode

Anode

t2

Anode

t3

t4

Δz

I(t) s(t)

Current I [A]

Tool position s [mm]

Vibrator position z [μm]

In this contribution, a stainless steel (1.4301, austenitic) is used as workpiece material. The chemical composition is listed in Table 2 in terms of minimum and maximum alloying element allowance by norm and as ICP-OES analysis result. Fig. 2 represents the experimental results of the specific mass removal (SMR) in milligram per Coulomb of the workpiece material in a water based NaNO3 electrolyte under different current densities acquired in experiments using a custom build PECM setup. Fig. 3 shows the frontal gap as well as feed rate dependencies of the used material for different current densities. The fit in Fig. 2 is done according to [4]. The experimental boundary conditions under which the data in Fig. 2 and Fig. 3 were determined are as follows: x x x x x x x

NaNO3 concentration Electrolyte conductivity Temperature pH number average flow rate pulse on time Voltage

flushing max. gap

machining min. gap

flushing max. gap

machining min. gap

75 g/l (technical pure) σ = 70.5 mS/cm (± 1mS/cm) T = 21°C (± 1°C) 7.1 pH (± 0.2pH) 4.7 l/min (± 0.2 l/min) ton = 2.5ms (at f=50Hz) 10 V

Table 2. Material investigated [in Weight-%]

Electrolyte

Anode

t1

(1)

2.2. Material investigated

Name Anode

 Pshift[%] ˜ ton[ms]

1.4301

X5CrNi18-10

AISI 304

Element

C

Si

Mn

P

S

Cr

Ni

N

Cu

min.

0.00

0.00

0.00

0.000

0.000

17.0

8.00

0.00

0.0

75

max.

0.07

1.00

2.00

0.045

0.015

19.5

10.5

0.11

1.0

66.885

ICPOES

-

0.83

1.92

0.13

-

17.72

10.16

-

0.33

68.89

t[ms]

0.2 Tanh-fit Removal vs. J

0.18

0.027

100

10

50

50

75

2.5

Initial gap [μm]

ton [ms]

Pshift [%]

felectric [Hz]

fmechanical [Hz]

Voltage U [V]

Pressure p [kPa]

Feed rate vfeed [mm/min]

Total feed [mm] 4

110

The phase shift Pshift [%] - as mentioned in Table 1 - relates to the shift of the pulse on-time ton in relation to the bottom dead center of the mechanical vibrator. The starting time tshift [ms] of the rising flank of the pulse on-time ton [ms] can be calculated in relation to the point in time when the vibrator reaches the bottom dead center according to formula (1).

Specific Mass Removal [1 mg/C]

0.16

Table 1. Experimental PECM parameters

Value

100 90

Fig. 1. PECM process schematic

Parameter

Fe

80 0.14 70 0.12 60 0.1

50

0.08

40

0.06

30

0.04

20

0.02

10

0

0

20

40

60

80

100

Removal Efficiency [1 %]

Cathode

t shift[ms]

0 120

Current density J [1 A/cm 2]

Fig. 2. Specific Mass Removal and Removal Efficiency for material 1.4301 at different current densities (R²=0.9809)

500

0.2

450

0.18

400

0.16

350

0.14

300

0.12

250

0.1

200

0.08

150

0.06

100

0.04

50

0.02

0

0

20

40

60

80

100

Feed [1 mm/min]

Andreas Rebschläger et al. / Procedia CIRP 14 (2014) 418 – 423

Frontal gap [1 μm]

420

0 120

Current density J [1 A/cm 2]

Fig. 3. Frontal gap (+) and feed rate (·) at different current densities

2.3. Experimental setup The experimental setup used is schematically shown in Fig. 4. It consists of a high speed camera of type Olympus iSpeed TR with an extension tube in front of the objective (specifications see Table 3), a flushing chamber made by additive manufacturing technology (EOS Formiga P110, Material: PA2200 infiltrated), two transparent PMMA plates (extruded, 80x70x10 mm) and the two electrodes contacted to the PECM machine, company PEMTec SNC, via custom metal fittings. The lighting was done using three IP65 protected 12 W LED spotlights (light colour 6400 K, 960 lm each).

The electrodes used in the experiments consist of precision sheet metal with a thickness of 1 mm (see Fig. 5). The electrode shape was laser cut and the cathode was additionally laser structured with a 500x500 μm grid to provide a size reference in each picture using a 3D-Micromac/Lumera 355 nm picosecond laser (see grid structure in Fig. 6 and Fig. 11). The frame rate in all experiments was chosen to 1 fps (at a shutter time of 100 μs) to enable a complete recording of the PECM process over a time interval of over 152 min using the best available resolution. In addition, at a given feed-rate of only 27 μm/min a higher frame-rate would only be useful to observe fast movements, as for example the fluidic flow or tool oscillation, which was not in the focus of this investigation. Electrodes allowing the reproduction of the most commonly used flushing cases in PECM: flushing through the cathode, flushing from side to side through the frontal gap and all variations of changing the flushing direction are already available and under investigation. However, in this contribution only experimental results derived from the example of flushing through the anode towards a cathode surface as shown in Fig. 5 are presented.

Fig. 5. Electrode dimensions [mm] and schematic flushing through the anode

2.4. Software procedure and analysis The complete image processing is done using Matlab R2012a, from the company Mathworks. Some features of the Image Processing Toolbox, for example the implemented ‘bwtraceboundary’ algorithm were used in combination with analysis specific coding. The code itself will not be part of this paper, however inaccuracies arising from the image processing steps will be discussed.

Fig. 4. Experimental setup schematic Table 3. Settings and properties of the High Speed Camera Type Resolution [pixel] 636 x 512 (Fig. 6 - Fig. 13)

1280 x 1024 (Fig. 11, Fig. 14, Fig. 15)

3. Results and discussion

Olympus i-Speed TR (CMOS) macro objective f=80mm, diaphragm setting 4 Frame rate [fps]

Shutter time [μs]

Bit depth [Bit]

Conversion factor [μm/pixel]

1

100

10

11.11

1

100

10

11.11

3.1. Experimental results The results and possibilities of the experimental setup will be exemplarily discussed using the data acquired during one experiment (Table 1). Due to a feed rate of 0.027 mm/min and a total machining depth of 4 mm, the video made during the experiment consists of 9,174 individual frames, including forerun and overshoot. The presented figures will focus on the evaluation using a time stepping of 900 seconds, as shown in

421

Andreas Rebschläger et al. / Procedia CIRP 14 (2014) 418 – 423

Fig. 6. The images recorded in a resolution of 636 x 512 pixels show the dissolution of the anode in nine successive steps. The individual pictures are then processed in Matlab to acquire the anode shape using an adapted boundary trace function after converting the pictures from RGB to grey scale and enhancing the contrast for better software detection of the edges. Two different representations of the shape analysis using the software are shown in Fig. 7 and Fig. 8.

0s 0A

1

900s 11.80A

2

1,800s 11.84A

3

4

2,700s 11.87A

5

3,600s 11.91A

6

4,500s 11.95A

7

5,400s 11.98A

8

6,300s 12.02A

9

7,200s 12.05A

Fig. 6. Image sequence (900s stepping) with an image size of 636x512 pixels respective 7065.96μm x 5688.32μm

3.2. Software analysis results Fig. 7 shows a graphical representation of the anode boundary in x and y dimensions of nine individual 900 s steps determined from the images in Fig. 6. Fig. 8 shows a quasicontinuous shape evolution for the selected time steps.

Fig. 8. Anode shape evolution in μm for 9 frames (total 8*900 s = 7,200 s)

Based on Fig. 7 different analyses can be conducted using a downstream software based approach to evaluate material dissolution and geometric shaping. Exemplarily the side gap evolution over time is shown in Fig. 9. The figure represents the shaping process for the time of the complete experiment tracing the boundary of the horizontal pixel row number ‘-97’ respective height ‘-1077.67 μm’ in Fig. 7. Similar to this, the complete shape in horizontal or vertical direction can be investigated towards dissolution rate and behavior for both simple and complex geometries. In addition to geometric analyses the presented experimental procedure also allows volumetric analyses without having to weigh the material during individual experiments. Fig. 10 shows the total material dissolution in mm³, whereas the surface area of one pixel is calculated to 123.43 μm². Taking into account the sheet metal thickness of 1 mm the volume and with help of the material density, the mass removal can be calculated. Through parallel recording of the electrical parameters during the image acquisition [5], the specific removal rate in mg/C can be recalculated according to Fig. 2. Gap evolution at the height -1078 μm 800

Anode shape evolution 0

700

-1000

Width +6899.31μm

-2000

Height [1 μm]

Horizontal feed [1 μm]

600

Height -1077.67μm

-3000

-4000

500 400 300 200

Width +1999.80μm

-5000

100 0

-6000

0

1000

2000

3000 4000 Width [1 μm]

Fig. 7. Result of the boundary trace function

5000

6000

7000

0

900

1800

2700

3600 Time [1 s]

4500

5400

6300

Fig. 9. Side gap evolution in pixel-row ‘-97’ respective at height ‘1077.67μm’ (as indicated in Figure 7)

7200

422

Andreas Rebschläger et al. / Procedia CIRP 14 (2014) 418 – 423

Further investigations on tolerances of the used setup and evaluation methods resulting for example from the conversion of grey scale images to binary images as well as other sources according to [6,7] will be part of future investigations.

Total material dissolution in mm³

10

8

Optical microscopy image

High speed camera image

Binary image

6

500μm

Volume of material dissolution [1 mm³]

12

4

2

500μm 900

1800

2700

3600 Time [1 s]

4500

5400

6300

7200

Fig. 10. Calculated total (accumulated) material dissolution of the anode at individual time steps [mm³]

Tool position s [mm]

0

3.3. Tolerances and accuracy

B, max. velocity Δz s(t)

flushing max. gap

shutter time 100μs

flushing max. gap

machining min. gap

t[ms]

Fig. 11. Inaccuracy through shutter time and image conversion Vertical camera movement at the width +6900 μm 100

Vertical camera movement [1 μm]

80 60 40 20 0 -20 -40 -60 -80 -100

0

900

1800

2700

3600 Time [1 s]

4500

5400

6300

7200

Fig. 12. Inaccuracy in vertical direction tracing pixel-column ‘+621’ respective at width ‘+6899.31μm’over time (as indicated in Figure 7) Vertical feed at the width +2000 μm 3500

3000

2500 Vertical feed [1 μm]

In a first step to evaluate the accuracy of the presented experimental setup and the downstream software analysis two main factors are discussed: the image acquisition and the image processing. The accuracy of the image acquisition is mainly affected by the mechanical vibration of the cathode, here f = 50 Hz. This mechanical vibration inevitably causes a vibration in the complete experimental setup. Hence, a calculation based on the frequency, feed rate and amplitude of the mechanical vibrator is necessary. Taking into account the shutter time of 100 μs the minimum and maximum blur in each picture during the vibration can be calculated. As indicated in Fig. 11 two points (A and B) are of special interest, the time period of 100 μs at the point in time when the feed rate equals the upward moving speed of the vibrator and the point in time when the feed rate and vibrator movement superimpose to a maximum velocity. The calculation of these time periods lead to a respective minimum blur of 4.5e-05 μm and a maximum of 6.283 μm during the time period of 100 μs shutter time. The displacement due to the effect of the vibration on the anode is shown in Fig. 12. The movement of the anode in column ‘+621’ over 7,200 s reveals a maximum inaccuracy of 3 pixels (33.33μm) during the complete experiment. However an overlaying dissolution effect cannot be ruled out at this point, since the stray current of the uninsulated cathode could still have an effect at this distance - see calculation of the current density distribution in Fig. 15. Using the laser marked cathode as a size reference for different optical zoom levels in the images, a comparison of the used zoom and imaging distance to the measurement using an optical microscope was conducted. For all presented pictures the 500 μm grid corresponds to 45 pixels in horizontal and vertical image direction. Consequently a conversion factor of 11.11 μm/pixel can be calculated. Yet unsuitable to investigate gaps in the region of 10 μm given the current experimental setup, these observations should be feasible using macro objectives with a higher magnification in future experiments. The vertical dissolution velocity of the anode presented in Fig. 13 also corresponds to the feed-rate set in the experiment. The evaluation of the incline calculates exactly to a feed-rate of 0.027(034) mm/min which was set in the experiment.

Vibrator position z [μm]

A, min. velocity 0

2000

1500

1000

500

0

0

900

1800

2700

3600 Time [1 s]

4500

5400

6300

7200

Fig. 13. Feed-rate calculated for frontal gap in vertical direction tracing pixelcolumn ‘+180’ respective at width ‘+1999.80μm’over time (as indicated in Figure 7)

3.4. Transfer to FEM Simulation In addition to the already presented possibility to investigate the shaping process at discrete points in time, a transfer of the anode shape information in a commercially available FEM Program, COMSOL Multiphysics 4.2a,

Andreas Rebschläger et al. / Procedia CIRP 14 (2014) 418 – 423

according to [8] was done. A stationary simulation using the ‘electric current’ module enables the investigation of the electric potential (see Fig. 14) and the current density distribution (see Fig. 15) during the experiment at any given point in time. The current density in the frontal gap calculates to approximately 49.21 A/cm² at a simulated gap size of 117.52 μm at half ton (ton=2.5ms, Pshift=75%) and assuming an experimentally determined overpotential of 1.81 V. According to the preliminary experiments given in Fig. 3, the expected current density for a given feed-rate of 0.027 mm/min is 46.69 A/cm². This mismatch of 2.52 A/cm² could be explained in the neglecting of the Joule heating and the flushing conditions in the FEM simulation as well as inaccuracies in the image, as discussed before. Yet, this possibility to link the actual geometry to FEM simulation offers the possibility to validate and improve simulations, by allowing comparisons at equidistant times during the simulation.

423

acquisition setup. The equipment, the experimental boundary conditions, including preliminary investigations of the used material to validate the downstream data analysis as well as the link to FEM simulations is presented. A short discussion on results, possibilities and accuracy of the video based process observations of the pulse electrochemical machining process is given. 5. Conclusion The link between electrochemical in-process and geometric shaping information offers possibilities to investigate a variety of different and complex geometries as well as process settings. A close look into the overall copying accuracy, as already conducted in [9], using this setup and the reduction of experiments necessary to gather material dissolution data based on the image informations will be part of future investigations. Acknowledgements The authors thank the European Regional Development Fund for the financial support. References

Fig. 14. Electric potential for geometry recorded in frame 942

Fig. 15. Current density distribution for geometry recorded in frame 942

4. Summary Focus in this contribution was the presentation and first basic validation of an experimental in-process image

[1] Rajurkar KP, Zhu D, McGeough JA, Kozak J, De Silva A. New Developments in Electro-Chemical Machining. Annals of the CIRP; Vol. 48/2/1999; 567-579 [2] Hinduja S, Kunieda M. Modelling of ECM and EDM processes. CIRP Annals – Manufacturing Technology, Volume 62, 2013, Pages 775-797 [3] Kozak J, Rajurkar KP, Wei B. Modelling and analysis of pulse electrochemical machining (PECM). Journal of engineering for industry; Vol. 116/3/1994; 316-323 [4] Altena S.J. Precision ECM by process characteristic modelling. Dissertation; Glasgow Caledonian University; Glasgow; 2000 [5] Rebschläger A, Weber O, Bähre D. In-situ process measurements for industrial size Pulse Electrochemical Machining. Proceedings of the 8th International Symposium on Electrochemical Machining Technology INSECT2012, Institute of Advanced Manufacturing Technology (IOS), Krakow,2012, Editor: Maria Zybura-Skrabalak, ISBN 978-83-931339-5-6 [6] Tiwari V, Sutton MA, McNeill SR. Assessment of High Speed Imaging Systems for 2D and 3D Deformation Measurements: Methodology Development and Validation. Experimental Mechanics (2007) 47:561– 579 [7] Versluis M. High-speed imaging in fluids. Exp Fluids (2013) 54:1458 [8] Weber O, Kollmannsperger R, Bähre D. Simulation of the Current Density Distribution and the Material Removal Behavior on the Graphite/Iron-Matrix Interface in Cast Iron under Pulse Electrochemical Machining Conditions. Proceedings of the European COMSOL Conference, Milan, 2012, ISBN 978-0-9839688-7-0 [9] Bähre D, Rebschläger A, Weber O, Steuer P. Reproducible, fast and adjustable surface roughening of stainless steel using Pulse Electrochemical Machining. Proceedings of the Seventeenth CIRP Conference on Electro Physical and Chemical Machining (ISEM), Procedia CIRP, Volume 6, 2013, Pages 362–367