A Novel Method for the Real-time Force Losses Detection in Servo

Welding Guns

D. Ibáñez

, E. García

, J. Martos

and J. Soret

Dept. of Electrical and Electronic Engineering, University of Valencia, Burjassot, Valencia, Spain

Ford Valencia, 46440, Valencia, Spain

Keywords: Resistance Spot Welding, Bending, Real-time, Automatization, Force Control, Welding Gun.

Abstract: Nowadays, real-time detection methods are increasingly necessary for predictive maintenance in production

processes. Specifically, in the metal joining production processes that use the resistance welding process, the

optimization of maintenance programs is sought to improve both the quality of the body and its manufacturing

cost. In this novel paper a new method is presented for the sensorless detection of pressure losses in welding

lines. The proposed system bases its operation on the measurement of the existing variables in the resistance

welding process carried out using servo guns. This paper also shows the proposed system for the data

acquisition, data sending and visualization in real-time of the health of the welding gun. This results in a

system with a low installation cost but with great performance in reducing problems associated with force

losses in welding guns.

1 INTRODUCTION

The resistance welding process is currently the most

widely used metal joining process, especially in the

automotive industry, where this process represents

around 90% of the joints made in a car body.

(Koskimäki et al., 2007; Yu et al., 2014; Hwang et al.,

2013). Despite the great use of this process, the

resistance welding process is highly complex, since

this process involves different fields such as

electromagnetism, thermodynamics, metallurgy and

mechanics. (Li et al., 2007). Due to the importance of

this process in the metallurgical industry, specifically

in the automotive manufacturing industry, it is

essential to guarantee the welding quality of each of

the joints carried out during the production process.

The quality of the joint of each of the joints that

are made through the resistance welding process is

determined by the diameter of the weld nugget. This

diameter is directly reflected in the joint design shear

load requirement.

Depending on the specific characteristics of the

joint, type and thickness of the metal, the type of

https://orcid.org/0000-0002-3917-9875

https://orcid.org/0000-0002-4210-9835

https://orcid.org/0000-0002-8455-6369

https://orcid.org/0000-0001-8695-6334

welding transformer and electrode from the welding

gun, several different parameters are required to form

the desired weld nugget that meets the quality

requirements.

Figure 1: Spot welding process.

The three main parameters that can be controlled to

obtain adequate quality in RSW are welding current,

electrode force and welding time (Willian and Parket,

2004).

Ibáñez, D., García, E., Martos, J. and Soret, J.

A Novel Method for the Real-time Force Losses Detection in Servo Welding Guns.

DOI: 10.5220/0010549006690676

In Proceedings of the 18th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2021), pages 669-676

ISBN: 978-989-758-522-7

669

Figure 2: Influence of the electrode force on the temperature distribution (Kobayashi & Mihara, 2014).

Resistance welding is carried out according to the

following process: the electrodes are closed at a

determined force by squeezing the metal sheets to be

welded. Once the programmed force is reached, it is

maintained for a short period to eliminate any gaps

that may exist between the sheets.

The welding current then flows through the

electrodes supplying energy to contacting workpieces.

Through the Joule effect, electrical energy is

transformed into heat gradually spreading through the

junction. Upon reaching the melting temperature, a

liquid nugget is formed that grows as a function of

welding time. Once the pre-set time has been reached,

the welding current stops, and then the electrodes are

released. Finally, the melting nugget solidifies, and the

weld is formed. (Kang & Lilong, 2014).

Consequently, as long as it is not influenced by

other external parameters, the welding current, the

force applied by the electrodes and the welding time

determine the heat given to the joint and therefore,

they directly influence the final quality of the welds.

Different authors speak of the importance of force

in the final quality of the weld, emphasizing the need

to guarantee a certain force. (Lucas, 2003; Norrish,

2006; Brijesh, 2014).

As mentioned, resistance welding bases its

operation on the Joule heat represented by equation 1.

𝐻=



𝐼



𝑡





𝑅



𝑡



𝑑𝑡





(1)

where H is the amount of heat energy generated

during the process, I(t) is the welding current, R(t) is

the dynamic resistance of the sheet metals, T1 and T2

are the beginning and ending times of the operation,

respectively (Henry,1910).

Resistivity for metals is usually considered to be

independent of force. In contrast, in resistance

welding, the contact resistance has a relationship with

the force distribution, in addition to the variation of

the contact surface conditions.

In general, due to irregularities in the electrode

and the metal sheets to be welded, only a small part

of the apparent contact area is in actual contact.

During resistance spot welding the force created by

the compression of the electrode breaks these

irregularities and causes a decrease in contact

resistance. this was demonstrated during the studies

Dickinson’s research and can be clearly seen in figure

1. Hence, not high enough electrode force cannot be

able to create enough electrical contact at the

interfaces and can cause concentrated heating and

possibly local melting or even expulsions.

Figure 3: Relationship between dynamic resistance and

electrode force (Dickinson, 1981).

Expulsion is a phenomenon that occurs when the

expansion of the weld nugget exceeds the force of the

supplied electrode, this causes an ejection of the

molten metal from the weld zone during welding. due

to force caused by the expansion of the weld nugget

exceeds supplied electrode force. Furthermore, as the

electrode force decreases, the diameter of the welding

nugget increases and the resistance of the weld until

ejection occurs (Zhang et Al.2009).

The contact resistance is also influenced by the

surface condition of the metal sheets. The presence of

Oil, dirt or any other foreign content causes a change

in the resistance. The effects on contact resistance of

these foreign contents decrease rapidly after the force

of an electrode is applied. The contact surfaces have

peaks and valleys. When subjected to low force,

metal-to-metal contact will be only the peaks. The

resulting contact area is less than that produced by an

appropriate force. Therefore, the contact resistance

will be greater, causing a greater amount of heat to be

generated.

ICINCO 2021 - 18th International Conference on Informatics in Control, Automation and Robotics

670

In short, it could be summarized as the lower the

force, the greater the interfacial resistance, which

results in a greater generation of heat at the tip-to-

sheet and sheet-to-sheet interfaces.

Then, it can be stated that the welding force plays

a crucial role in the quality of the final joint. (Zhang

& Senkara,2005) (Wang & Wei,2001).

Figure 4: Schema piezoelectric sensor.

Currently, the measurement of force loss in real

welding lines presents a series of drawbacks

depending on the method used for this detection.

First, the use of piezoelectric sensors or strain

gauges gives good results in welding lines but

represents too high a cost, which, in most cases,

makes their installation unfeasible (Podržaj et al,

2011) as seen in the figure 4.

Secondly, this measurement can be carried out

with manual dynamometers, in the same way, this

method provides a reliable result and more versatility

since the same dynamometer can be used for all

welding guns because this function is done through

an operator. The main problem of this method is that

the data is not available in real-time, so it is necessary

to carry out preventive maintenance, generating

personnel costs.

In short, there is currently no truly implantable

method for detecting force losses in welding guns for

high production factories.

For this reason, this research proposes the use of

stiffness measurement to create a real-time

measurement system integrated into the welding lines

for the detection of force losses.

2 EXPULSION AND FORCE IN

REAL WELDING LINES

For a first evaluation of the relationship between

force and the presence of projections in a real welding

line.

For this, a welding gun that due to force problems

has begun to present a high number of expulsions is

evaluated. Specifically, this welding gun performs 13

welding points on a specific part of the body, the tests

are carried out on an electric welding gun, ServoGun.

The phase difference between the real force of the

gun and the nominal force programmed is measured

with a dynamometer. This makes it possible to assess

the state of the welding gun and the influence on

expulsions.

Table 1: Nominal and Real Forde before calibration.

Nominal Force

[DaN]

Real Force

[DaN]

Difference

[DaN]

150 77 73

200 93 107

250 111 139

300 131 169

350 157 193

400 192 258

Once the state of the welding gun has been

determined, the data of average projections during

production are taken, obtaining the data measured in

figure 6.

Figure 5: Expulsion rate before force calibration.

As can be seen, the expulsion data is above the

usual and adequate values for optimum quality. Thus,

a force calibration is performed to reduce the

difference between the nominal force and the actual

force.

It is assumed that initially the weld was

programmed with an adequate force and that,

therefore, when the force returned to its original

value, a notable reduction in expulsions will be

observed.

After performing the calibration, the real and

nominal force data and the difference between them

are taken again, in order to observe the improvement

in force and the relationship with the improvement in

the expulsion rate. These data are shown in table 2.

It can be seen that when carrying out the

calibration, the difference between the nominal force

and the real force is significantly reduced so that the

A Novel Method for the Real-time Force Losses Detection in Servo Welding Guns

671

Table 2: Nominal and Real Forde after calibration.

Nominal Force

[DaN]

Real Force

[DaN]

Difference

[DaN]

150 152 2

200 202 2

250 262 12

300 310 10

350 355 5

400 417 17

force applied to the weld joint will be more similar to

the programmed force.

In the same way, as in the previous case, the data

on the expulsion rate is taken during the manufacture

of four parts of the body by the same welding gun,

obtaining the results shown in figure 7.

Figure 6: Expulsion rate after force calibration.

As can be seen from the comparison between

figures 6 and 7. The problems due to lack of force

cause an increase in expulsions in the production

lines, causing an increase in the expense for the repair

and cleaning of the welded parts. For this reason, the

interest in finding a system for the real-time detection

of force losses increases.

3 FORCE AND BENDING IN

REAL WELDING LINES

In order to obtain a viable method for the detection of

lack of force, the measurement of the bending of the

arms of the welding guns is considered.

Different authors have studied the measurement

of the encoder of the servomotors and the influence

of the bending on the real value of the electrode

position.

Specifically, some authors established a

relationship between the measurement of the position

of the electrodes and the indentation of the weld. (Yu-

Jun et al.,2020).

In welding guns, the fixed electrode arm of a

servo gun can be considered as a cantilever beam with

limited stiffness, this means that a non-negligible

deviation can occur in the arm of the fixed electrode

(Tang et al., 2003).

As Yun et al. show in their research, the fixed arm

can be simplified and reduced into three parts and P3,

as shown in figure 7 each considered as a beam of

uniform cross-section. This means that the bending

moment M

is distributed evenly along the direction

length of the part P3, this happens in the same way

with the axial force F

. Therefore, the longitudinal

surface deformation of the piece P3 can be considered

constant Fig 8.

If the theory of elasticity is followed, the

longitudinal surface deformation ε

and deflection D

of the electrodes can be calculated following equation

(2)(3). where F is the force of the electrode, E refers

to the elastic modulus of the beam, A1 and A3 are the

cross-sectional areas of part G1 and G3, l2 is the

length of the piece G2, w3 is the width or diameter of

the cross-section of the part G3, I2 and I3 are the

moments of inertia of part G2 and part G3,

respectively. γ and λ are the proportionality

coefficients between ε3x and F, Df and F,

respectively.

























=𝛾𝐹

𝛾=

























(2)



𝐷





























=λ𝐹

λ=



























(3)

Figure 7: Schema Force in fixed arm.

From the previous equations (2)(3) a relevant

conclusion can be drawn for this study, and this

conclusion is that there is a proportional relationship

between the force and the deflection marked by the

ICINCO 2021 - 18th International Conference on Informatics in Control, Automation and Robotics

672

proportionality variable λ. Therefore, if the

relationship between deflection and force is

established, the pressure variations can be determined

based on the variations in deflection. To see the real

behaviour in a welding line, the electrode

displacement data is taken in a welding gun installed

in a production line.

For this, the electrodes are closed without any

metal sheet between them, in such a way that if it is

considered that there is no influence of the force, this

position should remain stable despite the changes in

force. However, as it has been explained if this

relationship exists, it is expected to observe how as

the force of the electrodes increases, therefore, their

closing position increases.

This test is carried out on a ServoGun, in such a

way that the position value can be obtained directly

from the encoder rotation value, shown in degrees

Fig9. The behaviour of two different welding guns

located on different production lines is analysed. In

the first one (Gun 1) it can be seen how as the force

increases, the position of the encoder increases,

behaving almost proportionally (R

=0.988).

Similarly, if the encoder position curve generated

by the change in force for gun 2 is analysed, it can be

stated that it also presents a linear behaviour, but, in

this case, there is a greater slope of the curve, that is,

this welding gun features a greater displacement for

the same force range.

In both cases, the behaviour between the

displacement of the electrode and the force is affine.

Since, as it was deduced from equations 2 and 3, the

displacement is proportional to the force, but also the

encoder adjustment component must be taken into

account, since the position of 0 degrees should be

equal to a force of 0 kN, but sometimes due to a bad

configuration, the zero of the encoder does not

coincide with the real zero of the welding gun, when

the electrodes are in contact without applying

pressure.

As result, two main conclusions can be drawn

from this real experiment: First, in a real welding gun

the elasticity law is fulfilled and there is a relationship

between the bending and the displacement of the

electrodes. Second, the ratio between electrode

displacement and force is not necessarily the same for

all welding guns, depending on the mechanical and

physical characteristics of each one.

4 METHOD FOR DETECTING

LOSS OF FORCE

The method investigated in this paper bases its

operation on the measurement of the displacement of

the electrode for the detection of force losses. As

explained in the previous section, there is a linear

relationship between the bending of the arms of the

welding electrodes and the force applied to them. This

method is specially designed for the detection of

power losses in servo welding guns.

Since there is a real relationship between the

position of the electrode and the force applied on

them, a system for measuring the position of the

electrodes at different forces is designed. In such a

way that possible variations in position could be

monitored in real-time.

The measurement system is carried out as follows:

First, the PLC sends the order to change the

electrodes to the robot. After carrying out this change

of electrodes, the gun is sent to carry out the first

milling to avoid shear problems when making the

measurements.

Once the milling is finished, the wear of the

electrodes stored in the robot is reset. When all this

Figure 8: Real relationship between force and encoder position.

A Novel Method for the Real-time Force Losses Detection in Servo Welding Guns

673

process has been carried out, the electrodes are in the

optimal state to begin with the measurement of the

force loss following the newly proposed method.

Figure 9: Schema measure operation.

To measure force wear, the electrodes are closed

by pressing a 20mm thick metal piece Fig 10. In such

a way that if the real position reached by the

electrodes is measured, it will represent how much

they have bent.

Specifically, two measures are carried out, the

first of them executing a low force and the second of

them executing a force at the maximum capacity of

the welding gun. This is due to the fact that, on

occasions, the welding guns lose force in a certain

range, that is, the wear of the force does not occur

equally in all the command forces.

Figure 10: Real-Time data collection Schema (Garcia &

Montes, 2019).

Once these measurements are obtained, they are

sent to the PLC. The PLC, upon receiving the position

measurements, sends them labelled to the Rslinx and

through the MQTT protocol they are stored in a

database to be analysed.

To send and display these data, the same system

used by Garcia & Montes (2019) for data acquisition

from PLCs in real-time in factories is used.

5 REAL CASE

IMPLEMENTATION

This method is being tested in a real welding line on

which real-time data of force wear is being collected.

In this specific case, the programming was done

on a Kuka KRC4 robot with an ARO ServoGun 3G

pistol. For this, the check-up was programmed as

described in the previous section and began with the

acquisition and sending of data for its analysis.

In the first place, it can be observed in figure 11

how when working at a higher pressure (3.5kN), the

virtual thickness detected by the method is less than

in the case of low pressure (1kN) since the deflection

of the arms is directly proportional to force.

Figure 11: Historical data of the force check position

measurements.

After a period of time of correct operation of the

method, the pressure was checked according to

traditional methods, using a dynamometer. This

check showed that there was a lag of -300DaN

between the nominal pressure and the real pressure so

the calibration was carried out. As can be seen in

figure 11, as of April 19, once the calibration was

carried out, it was verified that when adjusting the

pressure by increasing the pressure, a change was

experienced in the virtual thickness measured. That is

to say, in short, the method is capable of detecting

pressure changes in real welding guns.

This leads to the expectation that when the high

force begins to wear out, the value of the position will

begin to approach the value of the low force position.

Approaching the minimum position value

established, which in this case is 20 mm due to the

thickness of the part on which the electrodes are

closed.

In short, data monitoring has begun, hoping that

as wear occurs in the welding gun, the system will be

able to detect them.

ICINCO 2021 - 18th International Conference on Informatics in Control, Automation and Robotics

674

With all that is exposed throughout this section, it

can be affirmed that the method is capable of

detecting power loss problems in welding guns

without the need for an external sensor, only using the

existing variables in robotized servo guns.

6 CONCLUSION

Throughout the paper, the importance of the force

parameter in the resistance welding process has been

shown, emphasizing the influence of force losses and

the appearance of expulsions.

The need to measure force losses and the

difficulty in applying current methods to carry out

force detection in large production factories has also

been discussed throughout the paper.

For all this, a novel method has been presented for

the real-time detection of force losses based on the

relationship between bending and the applied force.

This method provides the following advantages over

the previous methods:

• Reduced implementation costs: It is only

necessary to program the robot and the data

collection system. As it is a method based on the

analysis of variables of the servo gun welding

gun, it is not necessary to acquire any type of

additional sensor.

• Real-time measurement system: The system

proposed for the implementation of the method

makes it possible to view the data at any time,

being possible to generate alarms as soon as the

loss of force occurs.

• Maintenance cost reduction: since the force

value is acquired directly from the welding line,

it is not necessary to carry out any other type of

maintenance to determine force losses, so

preventive maintenance plans carried out can be

reduced.

In short, throughout this paper, a response has

been given to a problem that exists in factories that

produce metal joints that use servo guns for resistance

welding. It is expected that future research will

present optimizations to the method and success cases

for the final validation of the method.

ACKNOWLEDGEMENTS

The authors wish to thank Ford España S.A, in

particular, the Almussafes factory for their support in

the present research. The authors also wish to extend

their thanks to the "Foundation for Development and

Innovation of the Valencian Community (FDI)".

REFERENCES

Koskimäki, P. Laurinen, E. Haapalainen, L. Tuovinen, and

M. Juha Röning, “Application of the extended knn

method to resistance spot welding process

identification and the benefits of process information,”

IEEE Trans. Indus. Electron. 54(5), 2823–2830 (2007).

J. Yu, D. Choi, and S. Rhee, “Improvement of weldability

of 1 GPa grade twin-induced plasticity steel,” Welding

J. 93(3), 78s–84s (2014).

G. Hwang, P. Podrzaj, and H. Hashimoto, “Note:

Resistance spot welding using a microgripper,” Rev.

Sci. Instrum. 84(10), 106105-1–3 (2013).

Y. B. Li, Z. Q. Lin, S. J. Hu, and G. L. Chen, “Numerical

analysis of magnetic fluid dynamics behaviors during

resistance spot welding,” J. Appl. Phys. 101(05),

053506 (2007)

N. T. Williams and J. D. Parker, “Review of resistance spot

welding of steel sheets Part 1 Modelling and control of

weld nugget formation,” Int. Mater. Rev. 49(2), pp. 45–

75, (2004).

Zhou, Kang & Cai, Lilong. (2014). Study on effect of

electrode force on resistance spot welding process.

Journal of Applied Physics. 116. 084902-084902.

10.1063/1.4893968.

W Lucas Ph.D., DSc, CEng, FIM, FWeldl, EWE, S

Westgate BSc(Hons), in Electrical Engineer's

Reference Book (Sixteenth Edition), (2003)

John Norrish, in Advanced Welding Processes, (2006).

Brijesh, Makwana, M. Desai, Mr. Pradhyuman Parmar and

P. Student. “Optimization of process parameters for

Resistance Spot Welding process of Austenitic SS 304

using Response Surface Method – A Review.” (2014).

Crew, Henry. General Physics: an elementary textbook

FOR colleges, 2nd Edition. The University of

Michigan: The Macmillan Company. pp. 402–404.

(1910).

Zhang, H. & Senkara, J. Resistance welding: Fundamentals

and applications. (2005).

Wang, S. & Wei, P., Modeling Dynamic Electrical

Resistance During Resistance Spot Welding. Journal of

Heat Transfer-transactions of The Asme - J HEAT

TRANSFER. 123. 10.1115/1.1370502. (2001)

X. Zhang, G. Chen, Y. Zhang and X. Lai, “Improvement of

resistance spot weldability for dual-phase (DP600) steel

using servo gun”, Journal of Materials Processing

Technology, pp 2671-2675, (2009).

Kobayashi T, Mihara Y (2014) Numerical simulation of

nugget formation in spot welding: Mechanical Design

& Analysis Corporation. SIMULIA Community

Conference 2014:1–15

Dickinson, D. W., "Welding in the Automotive Industry",

American Iron and Steel Institute. Research Report SG

81-5. August (1981).

A Novel Method for the Real-time Force Losses Detection in Servo Welding Guns

675

Podržaj, Primož & Simončič, Samo. Resistance spot

welding control based on fuzzy logic. The International

Journal of Advanced Manufacturing Technology. 52.

959-967. 10.1007/s00170-010-2794-0. (2011).

Xia, Yu-Jun & Su, Zewei & Lou, Ming & Li, Yongbing &

Carlson, Blair. Online Precision Measurement of Weld

Indentation in Resistance Spot Welding Using Servo

Gun. IEEE Transactions on Instrumentation and

Measurement.69.4465-4475. 10.1109 /TIM.2019.2943

981. (2020).

H. Tang, W. Hou, S. J. Hu, H. Y. Zhang, Z. L. Feng, and

M. Kimchi, "Influence of welding machine mechanical

characteristics on the resistance spot welding process

and weld quality," Welding Journal, vol. 82, no. 5, pp.

116S-124S, 2003.

E. Garcia, N. Montes, Mini-term, a novel paradigm for fault

detection, IFAC-PapersOnLine, Volume 52, Issue 13,

Pages 165-170, ISSN 2405-8963. (2019).

ICINCO 2021 - 18th International Conference on Informatics in Control, Automation and Robotics

676