Development of Edge-Preparing Plasma Arc Cutting (EPPAC) System

Naoki Osawa Akira Furujo

Dept. N.A. & O.E., Osaka University Koike Sanso Kogyo Co., Ltd.
2-1 Yamadaoka, Suita, 2-3-1 Shinden, Ichikawa City,
Osaka 565-0871 Japan Chiba 272-0035 Japan

Masayuki Nagahori Toshitake Onishi

Nissan Tanaka Corporation Mitsui Engineering & Shipbuilding Co., Ltd.
11, Chikumazawa, Miyoshi-cho, Iruma-gun, 1 , Yawatakaigandori, Ichihara-shi,
Saitama 354-8585 Japan Chiba 290-8601 Japan

Hiroyuki Yajima Takeshi Miyamoto

IHI Marine United Inc. Japan Ship Technology Research Association
12,Shinsugita-cho,Isogo-ku,Yokohama-shi, 2-10-9, Akasaka, Minato-ku,
Kanagawa 235-8501 Japan Tokyo 107-0052 Japan

ABSTRACT

Edge-Preparing Plasma Arc Cutting (EPPAC) systems have been developed. Top edges of
kerfs generated by these systems have bevel angle of θ >155°, which is equal to or larger than that of
3-pass grinding edges. Cutting qualities except for melting of cut shoulder equivalent to or better than
those of kerfs generated by conventional machines, and there is no slowdown on the cutting speed.
The developed systems are tested under the practical condition in Japanese shipyards, and it has
been demonstrated that there is no special problem in workplace safety, and kerfs which satisfy the
requirements specification can be generated with high stability. The lower bound of 95% confidence
interval of edge retention ratio (ERR) of EPPAC kerfs is about the same as that of 3-pass grinding
edges. This means that the top edge shape of kerfs generated by developed EPPAC systems
conforms to IMO/PSPC, and there is no need of additional hand grinding. We can cut the man-hours
for edge treatment in half by using developed EPPAC systems.

Keywords: IMO/PSPC, edge retention, plasma arc cutting

 INTRODUCTION

On December 2006 the International Maritime Organization (IMO) has adopted a Performance
Standard for Protective Coatings (PSPC). This new coating standard applies to ballast water tanks
on newbuildings in all types of ships and is settled in the Resolution MSC.215(82). Hereafter, this
new regulation is referred to as IMO/PSPC.

Ship structures often come with complex geometric configurations, having large surface and
highly stressed areas, such as corners, edges, and weld seams areas. Coating defects such as
insufficient film thickness are often observed, resulting early coating failure and corrosion in these
areas. It is well known that secondary surface preparation (SSP) is effective in preventing these
coating defects. It is mandated that edges to be treated to a rounded radius of min. 2mm (2R), or
subjected to 3 pass grinding or at least equivalent process by IMO/PSPC.

Mechanical grinding of edges will retain paints of liquid phase for a longer duration than the
sharper edge does and improve edge retention behaviour. However, it takes a great amount of labor
to mechanically grind all edges to be painted in ships. Most steel plates which comprise ballast tanks
are cut by plasma arc cutting (PAC). If edges are prepared during PAC process, we can cut off the
amount of manual labor required for edge preparation.

PAC machines with automatic edge preparation capability are developed by an

industry-government-academia research group organized by Japan Ship Technology Research
Association. Hereafter, this new PAC system is called ‘Edge-Preparing PAC (EPPAC)’ system. In
this project, research and development of PAC technologies for realizing EPPAC machines are
carried out. Method for edge retention assessment of kerfs generated by developed systems is also
examined. This report presents the outcome of EPPAC development project.

AUTOMATIC EDGE PREPARATION DURING PLASMA ARC CUTTING

Requirements Specification

It would be ideal if edges on the front-back both sides were treated to rounded shapes during
PAC process. However, the edge treatment on the back face requires a major modification on the
plasma torch and the cutting table. It is also anticipated that the edge treatment on the back face
cause slowdowns on the cutting speed. These spoil the cost-reducing effectiveness of the developed
cutting machine.

It is well known that top rounding on the front edge is produced when the operating parameters
(especially, the stand off distance) of PAC machine are not fully-tuned. This leads us to an
assumption that a rounded edge can be formed on the front edge easily and it is possible to achieve
the required level of cut qualities (edge bevel, flatness of the cut face, etc.) with minor modifications
of plasma torch and gas control system. It is also expected that the slowdowns on the cutting speed
will be almost negligible. Though the back side edge is not treated, we can cut the man-hours for
edge treatment in half. That is, it is relatively easy to develop a PAC machine with front side edge
preparation capability (without back side edge preparation capability), and its cost-reducing
effectiveness is significant. In this research project, we carried out a study aimed at developing
computer numerical controlled (CNC) PAC system with front side edge preparation capability.

The quality of cuts is evaluated by means of a set of geometry attributes of the kerf. The
attributes taken into consideration in this research are those suggested by WES2801:1980
standard10: melting of cut shoulder, M; flatness of cut surface, F; roughness of cut surface, R; notch,
N; bevel accuracy, A,H; straightness, P; slag adhesion, S.

In Japanese shipyards, the control data for CNC-PAC process contains no information on the
distinction between free edge and weld line. The number of man-hours needed to create such
information becomes enormous, and this counteracts the reductions in the hand grinding cost
achieved by the development of EPPAC systems. Therefore, it is needed to use the rounded kerfs
for weld lines, and the top rounding generated by EPPAC systems should produce no ill effect on the
weld quality. This leads to the following requirements: a) qualities of F/R/N/A/H/P/S of
WES2801:1980 equivalent to or better than those of kerfs generated by conventional CNC-PAC
systems (Grade 1 of WES2801:1980); b) edge retention of the rounded top edge is equal to or better
than those of 2R or 3 pass grinding edge.

Development of EPPAC systems

Geometric features of the kerf generated by PAC is shown in Figure 1. The top edge is
generated at the boundary of the area on which plasma jet impinges, and it is impossible to prevent
the formation of this edge. The radius of curvature of the kerf shows its minimum at the top edge.
Hereafter, the angle which the tangent forms with the front face is called ‘edge bevel angle θ’, and the
radius at the top edge is called ‘the minimum radius ρ‘. θ is not equal to the bevel angle Δθ1 defined in
WES2801:1980, which is illustrated in Figure 1.

Figure 1: Geometric features of the kerf generated by PAC and the bevel angle defined in
WES2801:1980.
 Kharlamov et al.2 examined the relation between a coating performance and edge shape. They
showed that the larger the radius of curvature or the edge bevel angle, the better the edge protection.
3 pass grinding (3-pass) and a rounded radius of 2mm (2R) are required for edge preparation
process in IMO/PSPC. 3-pass is characterized by θ≅155° (≅180°−2×22.5°). 2R shape is
characterized by ρ≥2mm. Because it is impossible to prevent the formation of the top edge, the
objective of this research is accomplished by developing a PAC machine which generates kerfs with
θ≥155°.

The relation between the operating parameters and the top edge bevel angle θ was examined
using two existing CNC-PAC systems: Koike Sanso Kogyo SUPERGRAPH with plasma torch
434V-OPS and Nissan Tanaka KT-790PMX with plasma torch FY-500. Stand off distance, cutting
speed, current, plasma gas flow rate, inclination of torch and orifice diameter were changed in
experiments. SS400 mild steel plates with 12mm thickness coated with inorganic zinc rich primer
were cut by PAC. As a result, followings are found:
• The stand off distance h has the most powerful influence on θ. θ increases with h, and θ≅145° is
achieved when h>25mm but θ≥150° cannot be achieved.
• Increases in current and orifice diameter have a influence, but their impact is much smaller than
that of h.

h=8.5mm is recommended by manufacturers for the conditions chosen. The arc light intensity
increases as h increases, and h>25mm brings about an inadmissible worsening in workplace safety.
Furthermore, it also deteriorates the cut quality. The maximum bevel angle generated by existing
PAC machines, θ ≅145°, is below the IMO/PSPC requirement. It is needed to develop a new PAC
system which can achieve θ>155° with a minimum increase in h.

In our project, the participating cutting machine makers developed new PAC systems which
satisfy the above requirements. Top edge bevel angle larger than 155° was made possible by the
development of new plasma torches and gas control subsystems. Kerfs generated by these systems
are shown in Figure 2. Top edge bevel angle θ >155° is achieved when h=17mm. Qualities
F/R/N/A/H/P/S of WES2801:1980 are Grade 1 (the best grade). The developed systems were tested
under the practical condition in two Japanese shipyards for more than two weeks. In these tests, the
followings were demonstrated:
• The increased stand off distance (h=17mm) poses no special problem in workplace safety in
shipyards.
• Kerfs which satisfy the requirements specification can be generated with high stability.
• There is no slowdown on the cutting speed in the developed systems in comparison to
conventional systems..

EDGE RETENTION OF KERFS GENERATED BY EPPAC SYSTEMS

Edge retention test suitable for asymmetric edge shape

 (a) PCA machine maker A (b) PCA machine maker B
Figure 2: Kerfs generated by the developed EPPAC systems (Left: kerf, Up: front face).

Hereafter, the edge generated by an EPPAC system is called ‘EPPAC edge’. The top edge
bevel angles of EPPAC edges are approximately equal to those of 3-pass edges. However, both
sides of the edge are flat surfaces in a 3-pass edge, while one side is a flat surface and the other is a
rounded surface in an EPPAC edge. It is not clear that we can judge the superiority or inferiority in
edge retention of EPPAC and 3-pass edges just on the value of θ. It is needed to carry out edge
retention tests for EPPAC and 3-pass edges in order to demonstrate that a EPPAC edge meets
IMO/PSPC criteria.

Standards concerning edge retention of coating systems for ship structures include
MIL-PRF-23236C:20033. The edge retention test procedure of this standard is shown below:
• The retention specimen uses a structural angle (90º) section. Both exterior surfaces are milled, and
the edge is left as a sharp edge.
• Specimen is grit blasted.
• Topcoat is spray applied maintaining correct pump pressure and distance from substrate. Paint is
applied to both sides (first) and the edge (last) within one minute. After paint application, cure the
specimens at ambient laboratory conditions before sectioning.
• The test sample is cut, and the dry film thickness (DFT) is measured using an optical microscope.
Measurements are made at sufficient distance from the edge; this is DFT(flat). The lowest
thickness at the edge is also measured to determine DFT(edge).
• Edge Retention Rate (ERR) is calculated from:
DFT ( edge )
%ERR = × 100 (1)
DFT ( flat )

The test procedure of MIL-PRF-23236C:2003 is valid when the top edge shape is symmetric
about the axis at 45º to the front face. Hereafter, this axis is called ‘45º axis’. In cases where the top
edge is not symmetric about the 45º axis, it is difficult to decide the spray direction in the last
application. If paint is applied to an asymmetric edge from the 45º axis, the angle between the spray
jet and the kerf tangent at the top edge varies from shape to shape, and it is difficult to examine the
effect of the top edge radius on the edge retention.
 Figure 3: Paint application in Nagano’s retention test method.

Nagano et al.4 proposed a new edge retention test procedure. In their proposal, paint is applied
to the front face twice (first and second) and the kerf (last) (See Figure 3). In the first and second
applications, the angle between spray direction and the normal of the front face is 20º. The paint is
sprayed from a direction perpendicular to the kerf in the last application. This spray procedure
imitates the skilled worker’s practice in Japanese shipyards. In this test, the effect of the angle
between the spray jet and the kerf tangent on the edge retention can be weakened because the kerf
is on the leeward side. The edge retention is measured by ‘Side ERR’ defined as
DFT ( top edge )
%Side ERR = × 100 (2)
DFT ( front face )
where, DFT(top edge) is the lowest thickness at the top edge of the kerf, and DFT(front face) is DFT
on the front face.

Takada et al.5 applied Nagano’s test method to examine the edge retentions of chamfered
rounded edges, and they demonstrated that Side ERR defined by Eq. (2) is highly sensitive to the
difference in top edge shape parameters (radii and bevel angles). Because EPPAC edges are similar
to the edges examined in Takada’s study, we selected Nagano’s test method.

Yun et al.6 showed that blast cleaning can reduce the sharpness of edges and increase the
actual bevel angle. However, in actual blast cleaning process in shipyards, often an inadequate
number of grits hits the free edge. This means that we should carry out edge retention tests using
specimens without blast-cleaning if we want to assess edge retentions conservatively. Therefore, we
applied the topcoat to non-blasted substrate.

It is known that at least two coat at edge to build up the film thickness is important for the long
coating life, and a multi-coat system is recommended in IMO/PSPC. However, the difference in the
edge retention of EPPAC and 3-pass edges becomes negligible when a two-coat system is applied.
The objective of this report is the comparison of the edge retention. For this reason, we applied a
single-coat system in edge retention tests.
Materials and sample preparation

A KA32 steel plates with thicknesses 8, 12 and 16mm were blasted with steel grits to a
commercial finish (ISO 8501-1, Conditions SIS Sa 2), then coated with inorganic zinc rich primer.
Dry film thickness of the primer coat was 15μm.
Table 1: The information of applied coating material, painting variables and spray equipment.

Applied coating material

Type of binder modified epoxy
Viscosity 1.7 Pa·s
automatic painting apparatus
Airless pump pressure (the 2nd stage) 90~120 MPa(g)
Working distance 370 mm
Trip transfer velocity 450 mm/sec
Spray nozzle tip #521

Test specimens were created by cutting rectangular boards (100mm length x 25mm width)
using conventional CNC-PAC and EPPAC systems. EPPAC systems were developed by two cutting
machine makers (maker A and B). For EPPAC systems, plates were cut at three cutting speeds: the
manufacturers recommendation (V2); up 10% compared with V2 (V3); down 10% compared with V2
(V1). Part of EPPAC specimens was subjected to 1-pass grinding, and the rest was not subjected to
grinding. Hereafter, the i-th specimen with XXmm thickness cut at speed V3 by the system of maker
A is referred to as ‘AXXV3-i’. The j-th specimen with YYmm thickness cut by the system of maker B
at speed V1 with 1-pass grinding is referred to as ‘BYYV1G-j’. Edges of specimens cut by
conventional PAC systems were subjected to 3-pass grinding in two Japanese shipyards (shipyard C
and D). The thickness of 3-pass specimens was 12mm. The m-th section of the n-th 3-pass
specimen created in shipyard C is referred to as ‘C3P-m-n’.

All specimens were coated with anticorrosive paint systems. The shop primer was not removed
before the application of the top coat. The applied systems were modified epoxy resin system
(Chugoku Marine Paints, NOVA-2000QD). Automatic painting apparatus was used as the spraying
equipment in this test. Table 1 shows the information of applied coating material, painting variables
and spray equipment. After the top coat was applied, the painted specimens were dried for more
than 24 hours at room temperature (about 15ºC). The mean value of DFT(front face) was 136.3μm.

Analysis of edge retention behaviours

DFTs on the front face were measured at a distance of 10mm from the edge and those on the
kerf were measured on the mid-plane of the plate using an electromagnetic coating thickness tester
ElectoPhysik MiniTest 2100. The average of four measurements was taken as the representative
values of DFT(front face) and DFT(kerf).
Experimental results

Figure 4: Examples of cross-section views of 3-pass specimens (specimen C3P-2-1 ~ 4, Up: kerf,
Right: front face).

Figure 5: Examples of cross-section views of EPPAC specimens (specimen B12V3-1, Up: kerf,
Right: front face).
 Table 2: Side ERRs of EPPAC specimens.

DFT(top edge) DFT(front face)

Name Side ERR
[micro] [micro]

A12V2-1 80.6 134.1 60.1%

A12V2-2 83.5 130.8 63.8%

A12V2-3 97.1 125.2 77.5%

B08V1-1 119.7 134.0 89.3%

B08V2-1 111.5 143.0 78.0%

B08V2-2 113.8 143.0 79.6%

B08V2-3 109.0 152.0 71.7%

B08V3-1 111.8 134.0 83.4%

B12V1-1 101.6 133.0 76.4%

B12V2-1 119.1 140.0 85.1%

B12V2-2 97.4 147.0 66.2%

B12V2-3 106.8 142.0 75.2%

B12V3-1 107.5 142.0 75.7%

B12V3-2 112.9 142.0 79.5%

B12V3-3 115.6 148.0 78.1%

B16V1-1 127.6 149.0 85.7%

B16V2-1 110.1 158.0 69.7%

B16V2-2 107.9 156.0 69.2%

B16V2-3 119.4 150.0 79.6%

B16V3-1 103.0 146.0 70.5%

B16V3-2 114.0 155.0 73.6%

B16V3-3 115.6 162.0 71.4%

Table 3: Side ERRs of EPPAC and 1-pass grinding specimens.

DFT(top edge) DFT(front face)

Name Side ERR
[micro] [micro]

B12V1G-1 100.5 123.0 81.7%

B12V1G-2 106.0 134.0 79.1%

B12V1G-3 100.5 126.0 79.8%

B16V1G-1 135.0 136.0 99.2%

B16V1G-2 126.9 132.0 96.1%

B16V1G-3 116.0 143.0 81.1%

 Table 4: Side ERRs of EPPAC and 3-pass grinding specimens.

DFT(top edge) DFT(front face)

Name Side ERR
[micro] [micro]

C3P-1-3 100.3 128.0 78.4%

C3P-2-3 119.7 129.0 92.8%

C3P-3-1 109.2 135.0 80.9%

C3P-3-2 107.0 135.0 79.3%

C3P-3-3 86.9 135.0 64.4%

C3P-3-4 118.3 135.0 87.6%

D3P-1-1 122.1 138.0 88.4%

D3P-1-3 130.5 138.0 94.6%

D3P-1-4 88.4 138.0 64.1%

D3P-2-4 125.8 136.0 92.5%

D3P-3-4 107.4 138.0 77.8%

Table 5: Statistics of each specimen group.

Name μ σ C.V. μ−2σ

EPPAC 75.4% 7.3% 9.6% 60.9%

3pass grinding 81.9% 10.6% 13.0% 60.6%

EPPAC + 1pass grinding 86.2% 9.0% 10.5% 68.2%

The cross-sections of a specimen were ground using a grindstone, and the coating thickness
was measured at the edge using a light microscope (KEYENCE VH-800) of 50 magnifications. The
coating thicknesses were measured on the sections at distances of 20, 40, 60 and 80mm from the
plate end, and side ERR of Eq. (2) was calculated.

Examples of cross-section views for 3-pass and EPPAC (without 1-pass grinding) specimens
are shown in Figures 4 and 5. These figures show that 3-pass edge shape varies widely in the same
specimen while the edge shape difference is negligible in EPPAC specimens. Therefore, the
average of side ERRs of four sections is taken as the representative values for EPPAC specimens
while side ERR of each section is treated as a separate entity for 3-pass specimens.

For 3-pass specimens, there are times when DFT(top edge) becomes larger than DFT(front
face). Such situation occurs when the top edge bevel angle is close to 180º. We exclude these cases
from consideration because DFT does not show its minimum at the top edge in these cases. The
number of data is 22 for EPPAC specimens, 11 for 3-pass specimens and 6 for EPPAC+1-pass
grinding specimens.
 0.4

EPPAC specimens

Num. of specimen / Total num.

0.3

0.2

0.1

0.0
50 60 70 80 90 100
Side ERR [%]

Figure 6: Statistical analysis of edge retention behavior of EPPAC specimens.

0.4

EPPAC + 1-path grinding

Num. of specimen / Total num.

0.3

0.2

0.1

0.0
50 60 70 80 90 100
Side ERR [%]

Figure 7: Statistical analysis of edge retention behavior of EPPAC and 1-pass grinding specimens.

0.4

3-path specimens
Num. of specimen / Total num.

0.3

0.2

0.1

0.0
50 60 70 80 90 100
Side ERR [%]

Figure 8: Statistical analysis of edge retention behavior of 3-pass grinding specimens.

 Edge retention behaviors of EPPAC, EPPAC+1-pass grinding and 3-pass grinding specimens
are shown in Tables 2, 3 and 4. Statistical analysis concerning side ERRs of each specimen group
are carried out. Statistics of each specimen group are shown in Table 5. Figures 6, 7 and 8 show
histograms of side ERRs. The bin with of histograms is 2.5%, and the vertical scale is the number of
specimens in the interval per total. Probability densities of normal distributions estimated from the
statistics of measured data are also shown in these figures.

These results show the followings:

• The side ERR distribution of EPPAC specimens shows approximately a normal distribution which is
almost symmetrical about the mean value μ=75.4%. The lower bound of 95% confidence interval,
μ−2σ, is 60.9%.
• The mean value of side ERRs of 3-pass specimens is 81.9% and it is larger than that of EPPAC
specimens. However, side ERRs are polarized between high and low groups, and side ERR of
lower group closes to μ−2σ of EPPAC specimens. μ−2σ of 3-pass specimens is 60.6%, and it is
slightly smaller than that of EPPAC specimens.
• The mean value of side ERRs of EPPAC+1-pass grinding specimens is 86.2% and it is the largest
in three specimen groups. μ−2σ of this group is 68.2% and it is the largest.

Confomity of EPPAC edges for IMO/PSPC

Insufficient film thickness results in early coating failure and corrosion. The minimum DFT(top
edge), not the mean value, should be examined when edge retention performances are compared.
The lower bound of 95% confidence interval, μ−2σ, of side ERRs is a good criterion for such
purpose.

IMO/PSPC states that “edges to be treated to a rounded radius of min. 2mm, or subjected to 3
pass grinding or at least equivalent process before painting.” 3-pass specimens examined in this
research conform to IMO/PSPC. Therefore, it can be considered that EPPAC edge is ‘3-pass
equivalent’ and it conforms to IMO/PSPC if μ−2σ of side ERRs of EPPAC specimens is as well as or
better than that of 3-pass specimens.

μ−2σ of side ERRs of EPPAC specimens is 60.9%, and it is slightly larger than that of 3-pass
specimens (60.6%). This means that the top edge shape of kerfs generated by EPPAC systems
developed in this project conforms to IMO/PSPC, and there is no need of additional hand grinding.
We can cut the man-hours for edge treatment in half (from 6 pass for both front and back edges to 3
pass for the back edge) by using developed EPPAC systems.

The mean value and μ−2σ of side ERRs of EPPAC+1-pass grinding specimens are much larger
than those of 3-pass specimens. This means that the edge retention of this edge shape exceed
IMO/PSPC requirement substantially. Only 1-pass grinding is needed to attain this superior
performance.
 CONCLUSIONS

Edge-Preparing Plasma Arc Cutting (EPPAC) systems have been developed. Top edges of
kerfs generated by these systems have bevel angle of θ >155°, which is equal to or larger than that of
3-pass grinding edges. Cutting qualities except for melting of cut shoulder equivalent to or better than
those of kerfs generated by conventional machines, and there is no slowdown on the cutting speed.
The developed systems are tested under the practical condition in Japanese shipyards, and it has
been demonstrated that there is no special problem in workplace safety, and kerfs which satisfy the
requirements specification can be generated with high stability. The lower bound of 95% confidence
interval of edge retention ratio (ERR) of EPPAC kerfs is about the same as that of 3-pass grinding
edges. This means that the top edge shape of kerfs generated by developed EPPAC systems
conforms to IMO/PSPC, and there is no need of additional hand grinding. We can cut the man-hours
for edge treatment in half by using developed EPPAC systems.

REFERENCES

1 WES2801:1980 Standard (1980), “Quality Standard for Gas Cut Surface”, The Japan Welding
Engineering Society.

2 Kharlamov, IV, Koshin, II, (1976) “Corrosion Protection of the Members (Elements) of Steel
Structures with Different Bevel Angles and Radii of Curvature of the Edges”, Izvesiya VUZ, No.
8, pp. 17-22.

3 MIL-PRF-23236C:2003 Standard (2003), “Performance Specification: Coatings Systems for

Ship Structures”.

4 Nagano, M, Yao, Y, Takada, A, Osawa, N, (2009) “Development of Edge Retantion Rate

Measurement Method That Adapted to the Real Edge Shape”, Conference Proceedings
JASNAOE, Vol. 8, pp. 175-178 (in Japanese).

5 Takada, A, Nagano, M, Narayama, M, Osawa, N, et al., (2009) “Effect of shape parameter on

Edge Retention Rate of defective R shape edge”, Conference Proceedings JASNAOE, Vol. 8,
pp. 179-182 (in Japanese).

6 Yun, JT, Kwon, TK, Kang, TS, Kim, KL, Kim, TK, Han JM, (2005) “A Critical Study on Edge
Retention of Protective Coatings for a Ship Hull”, Proc. Corrosion2005, Paper No. 05016.

ACKNOWLEDGEMENT

This study was carried out within the activity of an industry-government-academia research
project ‘Research and Development of Advanced Anticorrosion Technology’ organized by Japan
Ship Technology Research Association (JSTRA) financially supported by Nippon Foundation. The
authors would like to thank Dr. Shinya Hayashi and Mr. Jiro Fujiwara for their discussions throughout
the work. The edge retention tests were carried out at Chugoku Marine Paint, Ltd. Technical Centre.
The authors gratefully acknowledge to Mr. Satoru Furumoto (Chugoku Marine Paint, Ltd.) for his
cooperation in the experiments.