ELSEVIER Journal of Materials Processing Technology 51 (1995) 131-149

Journal of Materials Processing Technology

Physical modelling of cut welding

P. D o r p h * , L. D e Chiffre ,

Institute of Manufacturin9 Enoineering, Technical University of Denmark, Buildin9 425, 2800 Lyngby, Denmark

Received 26 November 1993

Industrial Summary

Cut welding is a newly developed cold welding process by which the parts to be joined are shaved and immediately thereafter the clean surfaces are forced to slide one against the other and finally upset to obtain bonding. The process is of potential interest as an easy-to-automize and clean welding process to join metals like aluminium, copper, stainless steel and others. The present work describes experiments carried out to analyze the mechanisms involved in cut welding using wax as a model material. The experiments have shown that it is possible to simulate a solid phase welding process using a model material, this being the first work of this kind reported in the literature. The model material technique gives a very good illustration of the mechanisms involved in cut welding, and the experiments have emphasized that the flow pattern in cut welding is a governing factor with respect to the obtainable bonding area. Experiments with varying different parameters have shown that it is possible to influence the flow pattern during cut welding and in that way increase the area with good contact between the parts.

1. Introduction

Cut welding is a new cold pressure welding process proposed by De Chiffre at the Technical Universi ty of De nm a rk [-1-3]. The process was developed as an a t tempt to solve the problem of surface prepara t ion in cold pressure welding, a topic which has been extensively treated by Bay [4, 5].

Cut welding is based on the fact that cutt ing provides perfectly clean surfaces. Dur ing the process, the parts to be cut welded are shaved and immediately thereafter the clean surfaces are forced to slide one against the other. Finally, one of the parts is upset to produce the normal pressure necessary to obtain bonding. In this way, the clean surfaces created by cutt ing are not in contact with air before they are welded together.

* Corresponding author

0924-0136/95/$09.50 © 1995 Elsevier Science S.A. All rights reserved SSDI 0 9 2 4 - 0 1 3 6 ( 9 4 ) 0 1 5 9 7 - T

132 P. Dorph, L. De Chilfre ,'Journal ffl' Materials Processimj Technology 51 (1995) 13l 149

With a variant of the process a copper stud was cut welded into a hold in an aluminium strip. Tests reported in [1] have shown that a weld strength approaching that of the parent material can be consistently obtained at as low surface expansion as 6% using cut welding, while conventional cold pressure welding techniques require deformations of the order of 60-70%.

De Chiffre has also made some preliminary experiments concerning a variant of the process where a rectangular aluminium block is cut welded to an aluminium strip having the same cross section [2]. The working principle of the tool used during these experiments is shown in Fig. 1. As the punch (a) moves downwards from the starting position, the cutting tools (b) and (c) simultaneously shave the surfaces of the two parts (d) and (e). At the same time the two parts slide against each other. When the tool

(a)

Strip " ~ ( e ) i i ~ ( f )

O.

b.

Fig. 1. Tool for plane cut welding in aluminium: a: Shaving and sliding, (al punch, (b) upper cutting tool, (el lower cutting tool, (d) Al-block, (e} Al-strip; b: upsetting of the block.

P. Dorph, L. De Chiffre / Journal of Materials Processing Technology 51 (1995) 131 149 133

reaches the bot tom stop, the punch upsets the aluminium block and produces a normal pressure on the weld surface. During these experiments De Chiffre obtained a weld strength of approximately 35% of full strength.

This process variant has been investigated in detail in Ref. [6]. During this work, metallographic investigations across the weld of cut welded aluminium specimens have shown that the dividing line between the parts features a characteristic wave form which seems to be the pr imary factor determining the obtainable weld strength. In order to be able to illustrate the process and to obtain a deeper understanding of the mechanisms governing the flow pattern, simulation of cut welding using a model material was performed. The main results of the experiments in aluminium are published elsewhere [7]. The present work summarizes the results of the model material experiments described in detail in [6].

2. The model material technique

At the Technical University of Denmark, a large amount of work by Wanheim has been concentrated on physical modelling of metal working processes using model materials [8, 9].

The model material technique takes advantage of the fact that some wax materials behave more or less like metals during plastic deformation. Therefore it is possible to obtain relevant information about a given metal forming process by carrying out the same process in model material. Since the yield stress of wax is about thousand times lower than that of metals, physical modelling in wax gives many advantages:

The tool materials required to simulate a metal forming process in model material can be easy to machine and cheap such as wood or plastic. One or more of the sides of the tool can be made of a transparent material such as glass or plexiglass for example, and in this way the deformation during the process can be observed. The dimensions of the tool can be enlarged several times compared with the real process, still the force required to perform the process will be much smaller than for the real process. In this way small details may be investigated on a larger scale.

- Since the force needed to perform a process in wax is much smaller than the force needed to perform the real process, the requirements as to the model press are less demanding. This means that unexpensive presses can be used for model material simulation. The experiments described in this work were performed on a plane model material

press described in detail in [10]. In Fig. 2 a photograph of the press used to investigate extrusion is shown. The workpiece is a plane wax plate (a) which is placed on the plane of the press (b) together with the tool (c) having exactly the same thickness as the workpiece. To make it possible to determine details of the deformation, a grid can be painted on the workpiece. The deformation pattern can then be estimated from measurements of the displacement of each node of the grid [10]. During the process, a glass plate is clamped over the workpiece. The piston (e) is displaced with constant velocity by a motor. In this way the process takes place between two planes and as

134 P. Dorph, L. De Ch![J~'e/Journal of Materials Processin 9 Technology 51 (1995) 131 149

Fig. 2. Plane model material press, a wax specimen with a grid painted on the surface; b plane of the press: c tool for simulating extrusion; d force transducer: e piston [8].

long as there are no tensile stresses in the workpiece, the deformation will be under plane strain. The deformat ion of the workpiece can be observed through the glass plate during the process and the process force evaluated by readings on the dial gauge of the dynamomete r (d).

3. Experimental details

A tool for cut welding in model material was designed to give a plane simulation of the process and the was moun ted on the plane model material press. The tool and the experimental set-up can be seen in Fig. 3. The tool is composed of seven different parts: three of the parts (a, b and c), all made of steel, are fastened on the steel plane of the model press (d). Between those three parts, a wax specimen (e) is placed. This specimen simulates the s tat ionary part in the metal experiments (the strip). On part (a) a brass cutting edge (f) is mounted. Fur thermore the tool consists of two moving parts: a punch (g) and a holder for the cutt ing tool (h). These two parts are made of

P. Dorph, L. De Chiffre / Journal of Materials Processing Technology 51 (1995) 131 149 135

m

m

h m .

I I -

I .

J

9 . d .

k .

e. b . ~ , ~ •

I ( ( J • C . K | l J

Ill Fig 3. Experimental set up for plane cut welding in model material, a tool holder; b back plate; c bot tom plate; d plane of the press; e wax specimen; ~ cutting edge; g punch, h tool holder: i cutting edge; k wax-specimen; 1 force transducer; m -- motor; n position transducer.

aluminium to prevent them from scratching the plane of the press. On the moving part is mounted a cutting edge of brass (i), and another model material specimen (k), simulating the block in the metal process, is placed between the two aluminium parts. When the motor moves the punch, this latter will press directly on the block. To avoid friction between the mobile parts of the tool (part h, g, and i), these, as well as the plane of the press and the glass plate, are lubricated with vaseline on both sides. Also the wax specimens are lubricated on the planes that are in contact with the plane of the press, since there should be no friction here. On the other planes, the model material parts are not lubricated. The tool is placed in the start position, and the material parts are placed in the tool. A glass plate is fixed covering the whole set-up, and the motor of the press is started, moving the piston, whereafter the process begins. During and after the process, photographs are taken, and from these the flow pattern can be studied.

A basic parameter with cut welding is the interference. In practice it will never be possible to place the two cutting edges with a zero mutual displacement, but normally

136 P. Dorph, L. De ChiJyke/Journal ~?]' Materials Processin,q Technolo,qy 51 (1995) 131 149

I[ In.

g . d,

© ©

¢"

© ©

M

<~Z30.

Fig 4. Experimental set-up for application of sidewards pressure. Lctter a n refer to Fig. 4: b is free to move; o pneumatic cylinder; p spring mounted on a guide pin; q screw.

there will be a small gap between the cutting edges. This gap ;is called the interference, the reason for this name being that it controls the amount of material overlap between the two parts during sliding. The waveformed flow pattern observed during cut welding in a luminium [7] is assumed to be a result of this interference. The overlap- ping edges will push each other aside to make room for the addit ional material. The influence of the interference was investigated in this work making experiments with different values.

A basic precondit ion for obtaining reliable results from model material experiments is that the model material behaves more or less like the real material during the process. As discussed in [6] it is difficult to ensure that all the different correlation criteria between the real process and the model process that may be drawn up be satisfied. In order to consider the model material as a parameter, experiments during this work were made with two model materials showing different stress- strain curves.

In order to influence the flow pattern, during some experiments a pressure perpen- dicular to the sliding direction was applied behind the s tat ionary part of the tool. The

P. Dorph, L. De Chiffre/Journal of Materials Processina Technology 51 (1995) 131 149 137

pressure was applied by the means of a pneumatic cylinder, and the set-up for these experiments is shown in Fig. 4. In order to obtain a constant cutting depth during these experiments in spite of the increased cutting pressure, a spring was mounted behind the cutting edge of the mobile part of the tool.

Another attempt to influence the flow pattern was made by mounting a wedge behind the mobile part as illustrated in Fig. 5. Here experiments were made varying the wedge angle. In analogy to the experiments with application of a pressure perpendicular to the sliding direction, a spring was mounted behind the cutting edge on the stationary part of the tool in order to assure a constant cutting depth.

The tool was constructed in such a way that makes it possible to vary different parameters such as: - clearance,

width of cutting edges,

m, ~

I.

9,

e.

d,

p.

e .

Fig 5. Experimental set-up utilizing a wedge behind the block. Letters a n refer to Fig. 4; o wedge; p spring mounted on a guide pin; q screw.

138 P. Dorph, L. De Chil:fre/Journal of Materials Processin9 Technolo,qy 51 (1995) 13l 149

- cutting depth, - length of sliding between the two workpieces, - distance between the two cutting edges (interference),

dimensions of the specimens.

4. Experiments with varying the material

In Fig. 6 four steps of the cut welding process in model material are shown. The material used in this experiment was a particular wax composition (93% Filia wax, 5% indramic, 2% FeO) which behaves approximately rigid-plastic [11]. It is seen from Fig. 6 how a wave formed curve between the two parts occurs in model material. This wave-formed flow pattern appears to be very similar to the flow pattern observed during experiments in aluminium reported in [7]. During similar experiments in a strain-hardening wax-material (88% Filia, 2% indramic, 9% kaolin, 1% FeO) the same flow pattern was observed.

In Figs. 7(a) and (b) are shown two cut welded specimens in an rigid-plastic model material and a strain-hardening model material respectively. In both cases bonding seems to be obtained only on the wave, in full agreement with what has been observed in the metal experiments reported in [7]. On the top of the wave the strains have been high, and the pressure between the two surfaces has apparently been high enough to ensure bonding, Outside the wave, the material adjacent to the weld is not deformed to the same degree, and the two parts have split, showing no bond at all. The pressure between the two parts during sliding has been smaller here than on the wave, and too small to ensure bonding.

With the rigid-plastic model material, it seems that the two parts bind together in the upper half of the cross section while in the bot tom half of the cross section, the two parts have split, showing no bonding at all. With the strain-hardening material on the other hand, is seen a gap in the upper part of the cross section, covering about 20% of the length of the cross section, and a gap in the lower part of the cross section, covering about 60% of the length of the cross section and leaving only 20% in the middle where the two parts seem to bind together.

Comparing the two specimens in Fig. 7, it appears furthermore that the deforma- tion of the material adjacent to the weld is more pronounced with the rigid-plastic material than with the strain-hardening material. It appears that the stationary part (the strip in Fig. 1) is much more deformed when the process is carried out with the rigid-plastic material than with the strain-hardening material. This can be explained by the difference in the strain-hardening index. Strain-hardening materials are able to distribute the deformations, while rigid-plastic materials will show higher and more localized strains. The weld performed with the rigid-plastic material seems to bond over a larger part of the cross section than the weld performed with the strain- hardening material.

The welds were broken up by bending to study the fracture surface, and from this testing it appeared that the welds performed with the rigid-plastic material showed bond of good quality over about 40% of the cross section, the bonded area being

c 0

, m

"0

P , m

"0

O0

P. Dorph, L. De Chiffre / Journal of Materials Processing Technology 51 (1995) 131-149

B l o c k S t r i p

139

(a) (b)

(c) (d) Fig 6. Four steps of cut welding in rigid-plastic model material. Grid size: 5 x 5 mm.

140

t -

O = n

t _

= m

10

O) e-

1 0 m m

m

P. Dorph, L. De Ch!ff're/ Journal of Materials Processing Technolo.ay 51 (1995) 131 149

B l o c k S t r i p

(a)

(b) Fig 7. Two cut welded model material specimens. (a) rigid-plastic (b) strain-hardening. Grid size: 5 × 5 mm.

off-set a bit from the middle towards the top of the weld. The strain-hardening material on the other hand showed only good bond formation in the middle of the cross section, covering about 14% of the length of the cross section. In the rest of the cross section, there was either no bonding or bonding of rather poor quality. From this it seems that materials with small strain-hardening indexes are more suitable for cut welding than those with large strain-hardening indexes. However this has only a practical relevance as long as the material with a small strain-hardening index is

P. Dorph, L. De Chiffre / Journal of Materials Processing Technology 51 (1995) 131 149 141

ductile enough to withstand the large local deformation that occurs during cut welding.

With both materials a flow pattern was found similar to that seen with the experiments in aluminium i.e. a waveformed curve between the two welded parts. On the wave, high strains have been obtained in both the "strip" and the "block", and almost no deformation at both ends of the weld. In particular, at the bottom of the weld the strip has practically not been deformed at al.

5. Experiments with varying the interference

Experiments varying the interference in both a rigid-plastic model material and a strain-hardening model material have indicated an interesting relationship between the amount of interference and the obtainable bond strength. When the interference is small, the flow pattern is very regular, the dividing line between the two parts being approximately straight, and the deformations adjacent to the weld being rather small and quite regularly distributed along the cross section. The bond strength of the weld seems to be rather poor: obviously, the pressure between the two parts has not been high enough to create bonding. When the interference increases, the flow pattern changes, the strains being higher and more irregularly distributed. The dividing line between the two specimens has a waveform, the wave being more pronounced with increasing interference, and a bond of good quality occurs on the wave. Here the pressure between the parts has been high enough to ensure bonding. Outside the wave on the other hand, the parts are parted, showing no bonds at all.

It seems that for small values of the interference, the pressure during sliding is too low to ensure bonding. High values of the interference on the other hand give an uneven pressure distribution, so bonding is only obtained in a part of the cross section.

All in all, these experiments have confirmed that the flow pattern is very important with plane cut welding, and that an attempt to improve the obtainable weld strength should result in a more equal distribution of the normal pressure in order to obtain bonding over the whole surface. The experiments have furthermore emphasized the influence of the model material, however all the following experiments were carried out in the rigid-plastic wax material.

6. Experiments with applying a pressure perpendicular to the sliding direction

A number of experiments were made applying a pressure perpendicular to the sliding direction and mounting a spring behind the cutting edge, as illustrated in Fig. 4.

In Figs. 8(a) (d), four steps of the process are shown. In Fig. 9 the cut welded specimen is shown.

It appears clearly from Figs. 8 and 9 that it is possible to change the flow pattern and suppress the formation of the wave with cut welding in model material by

142 P. Dorph, L. De Ch!ffre/Journal qf Materials Processin9 Technolo,qy 5l (1995) 131 149

applying a pressure on the strip during the process. From Fig. 9 it appears that there are no longer cracks at the ends of the weld, and breaking apart the weld showed that good bond was obtained over the entire surface.

From Fig. 8 it appears that, in spite of the spring, the cutting edge does not move backwards at all, and instead the cutting depth increases during the process. So the spring does not work according to the intention. Pressing a model material specimen against the cutting edge had shown that the cutting edge can move backwards when it is pushed directly by the model material, showing that the spring force was low enough to avoid an indentation of the cutting edge into the model material specimen. However, as soon as the cut welding process is initiated, the cutting edge remains fixed. One reason for this could be that the cutting force causes a small inclination of the cutting edge, resulting in self-locking of the tool.

c- o

o ~

* m

c

"o

B l o c k S t r i p

(a) (b)

Fig 8. (a) (d): Cut welding in rigid-plastic model material with pressure perpendicular to the sliding direction. Four steps of the process Grid size: 5 x 5 mm.

P. Dorph, L. De Chiffre/Journal of Materials Processin9 Technology 51 (1995) 131-149 143

(c) (d) Fig. 8. (Continued)

B l o c k S t r i p c-

o m l

¢9

L _

| m

C = n

"O = B

m

00

Fig 9. Model material specimen cut welded using pressure perpendicular to the sliding direction. Grid size 5 × 5 m m .

144 P. Dorph, L. De Ch!ffre /Journal o[ Materials Proeessinq Teehno/o,qy 5l (1995) 131 149

7. Experiments with a wedge behind the block

Experiments with a wedge behind the block, as illustrated in Fig. 5, were made with wedge angles equal to 0.8 °, 1.6 ° and 2.5 '~. In Fig. 10 four steps of the process are shown with a wedge angle of 2.5 °. In Fig. 11, the welded specimen is shown.

During the experiments it appeared that the formation of the wave is to some degree suppressed already by using a 0.8 ::~ wedge. There seems furthermore to be a more evenly distributed material flow with the 0.8 ~' wedge than without the wedge. It also appeared that, as the angle of the wedge increases, the wave formation is gradually suppressed, and with a 2 .5 wedge the whole surface seems to be welded, as can be seen in Figs. 10 and 11.

The welds were broken apart, and the "weld strength" was estimated. From the fracture surfaces it seems that the fraction of the surface where good bonding is obtained increases as the wedge angle increases. In that way it appeared that with

t -

o 0 m

L~

t _

"0

C , u

"0 , B

B l o c k S t r i p

(a) (b) Fig 10. Four steps of cut welding in rigid-plastic model material with a 2.5' wedge behind the block. Grid

size 5 x 5 mm.

P. Dorph, L. De Chiffre/Journal of Materials Processing Technology 51 (1995) 131 149 145

(c) (d)

Fig. 10. (Continued).

B l o c k S t r i p C

O m m

¢9

i . _ _

i m

"O

e" i n

"O i m

m

O0

Fig 11. A model material specimen cut welded with a 2.5 ° wedge behind the block. Grid size 5 × 5 mm.

146 P. Dorph, L. De Chi~ke / Journal qf Materials Processin,q Technoloyy 51 (1995) 131 149

a 0.8 ° wedge behind the block, the bonded area covers aprox. 60% of the cross section and is placed in the middle of the surface. With a 1.6 ° wedge, the bonded section seems to cover about 70% of the surface, and with a 2.5 ° wedge angle, good bonding seems to be obtained almost over the entire surface.

Measurements on photographs taken during the process have confirmed that, similar to the experiments with application of a sidewards pressure, the cutting edge with a spring behind does not move backwards at all, even though the force on the cutting edge is large enough to cause an increased cutting depth. Also this time the reason is probably that the friction between the cutting edge and the guide pin on which the spring is mounted increases considerably due to an inclination of the cutting edge during the process.

8. Experiments with the drawing in of contaminant layers

An essential problem with the plane variant of cut welding might be that a portion of the uncleaned surfaces can be drawn into the weld because of the interference. This mechanism is illustrated in Fig. 12. In order to investigate this, a layer of yellow paint was sprayed on the surface of the block while a layer of white paint was sprayed on the strip, as illustrated in Fig. 13. The material used for these experiments was rigid- plastic and of approximately the same composition as the material used during the above described experiments. In order to make it easier to distinguish between the parts, one part material was colored by a small amount of FeO, while the other wax material was uncoloured. According to [11], this should not influence the plastic behaviour of the model material significantly.

The parts were cut welded using the experimental set-up shown in Fig. 3. The paint was sprayed on the parts, and three different layer thicknesses were chosen. (a) Thin, (b) Medium, (c) Thick. In one of the experiments black FeO-powder was applied to the

ConLaminated I surfaces BLOCK, 0, h /~ ~ / BLOCK~,~

BLOCK.['[/~TRIP Fragmer~ts of oxide from v r - s;urfT;o igh ,dbe

Fig 12. Illustration of how the uncleaned surfaces might be drawn into the weld during sliding.

P. Dorph, L. De Chiffre/ Journal of Materials Processing Technology 51 (1995) 131 149 147

J

Sliding direction

BLOCK

/ YELLOW PAINT

l J

;TRIP

WHITE PAINT

/

J

Fig 13. Illustration of painting of the model material parts.

top of the paint layer in an attempt to model the behaviour of the ductile liquid and gaseous layer on the top of the oxide layer described in [4].

After cut welding the parts were broken up to investigate if there was any paint present in the weld. It appeared here that the paint seems to break up by brittle fracture and can be found as small pieces more or less regularly distributed along the entire cross section. Pieces of white paint are found at the "bot tom" of the weld, while yellow pieces of paint are found at the "top" of the weld. This shows that the paint has been pushed through the weld, the white paint being pushed by the strip, and the yellow paint by the block. On the specimen with a layer of FeO applied to model the ductile liquid and gaseous layer, traces of FeO are found, located in small islands spread over the entire cross section. Some pieces of paint are found in the zone of the cross section where the "best bond" occurs. In all the pictures it seems that the white paints generally is attached to the strip, while the yellow paint is attached to the block, also after welding.

The experiments in model material have confirmed that the uncleaned surfaces caught by the interference are drawn into the weld during sliding, and that this contaminant effect might prevent bonding in a zone of the cross section. It seems that with cut welding in model material, the contaminant layers can spread more or less over the entire cross section. However experiments described in [-6] with cut welding of anodized aluminium-parts and inspection in a SEM have indicated that no contaminants are present in the weld after cut welding.

It is difficult from these experiments to estimate the extension of the problem, however the conclusion seems to be that at the present stage of the process the uneven pressure distribution caused by the flow pattern is the most basic problem associated with the process.

148 P. Dorph, L. De Chi~i'e/Journal of Materials Processiny Technolo~ly 5l (1995) 131 149

9. Conclusions

The experiments have shown that simulation of cut welding in model material gives a very good illustration of the different mechanisms of the process since it is possible to watch the process as it goes on. In particular the experiments have shown that it is possible to simulate the wave-formed flow pattern seen with plane cut welding in aluminium.

The influence of the wax-material on the process has been investigated. This investigation showed that the material with the smallest strain-hardening index was most suitable for cut welding.

Experiments varying the interference have shown that at small values of the interference the pressure between the parts is not high enough to ensure bonding. At higher values of the interference, the pressure between the parts increases, but because of the more pronounced wave-formed flow pattern, the pressure distribution becomes irregular, and bonding is only obtained in a part of the cross section.

Experiments with applying a sidewards pressure on the strip during cut welding in model material have shown that it is possible to influence the flow pattern of the process, and to get an increased material flow and a larger bonding area. Furthermore the experiments have shown that the application of a sidewards pressure also influen- ces the cutting process, giving a cutting depth that increases during the process. Mounting of a spring behind the lower cutting edge did not seem to solve the problem with the increasing cutting depth. This is probably due to special conditions because of the design of the tool.

Analogous to this, experiments have shown that it is possible to influence the flow pattern with cut welding in model material by placing a wedge behind the block. The results obtained have shown that with a wedge angle of 2.5", the wave formed flow pattern seen from cut welding in model material without a wedge, is almost com- pletely suppressed, and the "weld strength" is improving considerably. Also during the experiments where a wedge is mounted behind the block, the cutting depth incrases because of the increased normal pressure between the parts. In analogy to the experiments with applying a sidewards pressure this problem was attempted solved by placing a spring behind the cutting edge. Also this time the attempt was not very successful, probably due to the design of the tool.

Experiments investigating the probability of contaminants being drawn into the weld during sliding have indicated that this might be a problem. However similar experiments in aluminium published elsewhere have indicated that this is not the most important problem with the process at the present stage of development.

All in all, the experiments have shown that it is possible to obtain relevant information about a cold welding process by performing it in wax. This is interesting since it is the first work of this kind reported in the literature.

Acknowledgements

The authors wish to thank professor Niels Bay for valuable discussions throughout this work. The work was carried out with grants from the Technical University of Denmark, the M'chrwold Foundation and the Ib Henriksen Foundation.

P. Dorph, L. De Chiffre/Journal of Materials Processing Technology 51 (1995) 131 149 149

References

[1] L. De Chiffre, Cut welding, CIRP Ann. 38 (1) (1989) 125. [2] L. De Chiffre, Cut welding variants and application, Welding Rev. Int. 12(3) (1991). [3] L. De Chiffre, Method for joining of metals or metal alloys by cold welding, European Patent no. 277

183 B1 (1991). [4] N. Bay, Friction and adhesion in metal forming and cold welding, D.Sc. Thesis, Technical University

of Denmark, PI 1986. [5] N. Bay, Cold welding, Parts 1, 2 and 3 in: Metal Construction, Vol. 18, The Welding Institute, Redhill,

UK, 1986, pp. 369, 486 and 625. [6] P. Dorph, Experimental analysis of cut welding, Ph.D. Thesis, Technical University of Denmark, PI,

1993. [7] P. Dorph, L. De Chiffre and N. Bay, Experimental analysis of cut welding in aluminium, CIRP Ann.

42(1) (1993) 357 360. [8] T. Wanheim, Physical modelling of metal processing, Technical University of Denmark, PI, 1988. [9] T. Wanheim, The SIMON-project, J. Mech. Working Techn., 12 (1985) 137 147.

[10] J. Danckert, Modelmaterialeteknik, Ph.D. Thesis, Technical University of Denmark, PI, 1977. [11] M.P. Schreiber, Atlas over arbejdskurver for voksmaterialer, SIMON, Technical University of

Denmark, PI, 1982.