Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com

ScienceDirect
ScienceDirect
Available online at www.sciencedirect.com
Procedia Manufacturing 00 (2020) 000–000
ScienceDirect
Procedia Manufacturing 00 (2020) 000–000 www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia

Procedia Manufacturing 51 (2020) 302–308

30th International Conference on Flexible Automation and Intelligent Manufacturing (FAIM2021)

30th International Conference on Flexible Automation and Intelligent Manufacturing (FAIM2021)
15-18 June 2021, Athens, Greece.
15-18 June 2021, Athens, Greece.
A
A knowledge-based multipass welding
knowledge-based multipass welding distortion
distortion estimation
estimation method
method for
for aa
multi-robot welding
multi-robot welding off-line
off-line programming
programming and
and simulation
simulation software
software
Hannu Lundaa*, Sakari Penttiläaa, Tuomas Skrikoaa
Hannu Lund *, Sakari Penttilä , Tuomas Skriko
LUT University, P.O. Box 20, FI-53851 Lappeenranta, Finland
LUT University, P.O. Box 20, FI-53851 Lappeenranta, Finland

* Corresponding author. Tel.: 358 409101920; E-mail address: hannu.lund@lut.fi

* Corresponding author. Tel.: 358 409101920; E-mail address: hannu.lund@lut.fi

Abstract
Abstract
The current robotic welding off-line programming software packages do not consider the effect of welding distortions although the welding
The currentare
distortions robotic weldingphenomena.
well-known off-line programming software packages
Due to the distortions, off-line do not consider
programmed the programs
robot effect of welding
will havedistortions although
inaccuracies thepositions.
in robot welding
distortions are well-known phenomena. Due to the distortions, off-line programmed robot programs will have inaccuracies in
The purpose of this study is to develop a method for estimating the welding distortions during the jigless multi-robot welding production using a robot positions.
The
robotpurpose
weldingofoff-line
this study is to develop
programming a method
software as for estimatingenvironment.
a simulation the welding distortions
The appliedduring the jigless
methodology is multi-robot
to measure welding production
distortions of welded using
S700a
robot welding off-line programming software as a simulation environment. The applied methodology is to measure distortions
workpieces with laser scanner and by manual measurement, and create knowledge-based welding distortion estimation method in the simulation of welded S700
workpieces
software. Thewith laserof
results scanner and by
this study manual
indicate measurement,
that the distortionandin create
weldedknowledge-based
workpiece can bewelding
measureddistortion estimation
with laser scannermethod in the and
or manually simulation
during
software.
simulationThe
the results of this
robot was ablestudy indicate
to edit that theaccording
its position distortiontointhe
welded workpiece
measured can beLaser
estimation. measured with was
scanning laserfound
scanner
to or
be manually
preferred and during
measuring
simulation the data
method as the robotcanwas
be able to edit
collected its position according to the measured estimation. Laser scanning was found to be preferred measuring
digitally.
method as the data can be collected digitally.
© 2020 The Authors. Published by Elsevier Ltd.
©
© 2020The
2020 TheAuthors.
Authors. Published by ElsevierLtd. Ltd. license https://creativecommons.org/licenses/by-nc-nd/4.0/)
This is an open accessPublished by Elsevier
article under the CC BY-NC-ND
This
This isan
is anopen
openaccess
access article
article under
under the the BY-NC-ND
CC CC BY-NC-ND licenselicense (https://creativecommons.org/licenses/by-nc-nd/4.0/)
https://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the scientific committee of the FAIM 2020.
Peer-review underresponsibility
Peer-review under responsibility ofof
thethe scientific
scientific committee
committee of FAIM
of the the FAIM 2021.
2020.
Keywords: Welding distortions; Laser line triangulation; Off-line programming; Robot positioning; Digital Twin
Keywords: Welding distortions; Laser line triangulation; Off-line programming; Robot positioning; Digital Twin

Nomenclature testing and tuning before the off-line programs fully work
Nomenclature testing and
during the tuning
actual before
roboticthewelding.
off-line [1]
programs
Welding fullycauses
work
α Plate angle during the actual robotic welding. [1] Welding
distortions in the welded workpieces, causing positional errors causes
α
FEA Plate
Finiteangle
Element Analysis distortions in thewelding
weldedstation
workpieces, causingtopositional errors
in the physical and leading inaccuracies in
FEA Finite Element Analysis in the physical welding station and
GMAW Gas Metal Arc Welding workpiece nominal dimensions. The leading to inaccuracies
main cause for distortionsin
GMAW
MotoNIS Gas Metal Arc Welding
Motoman Network Information Server workpiece nominal dimensions. The main cause for distortions
is the heat brought to the workpiece. The heat causes the joint
MotoNIS
WPS Motoman Network Information
Welding Procedure SpecificationServer is theinheat brought to material
the workpiece. The heat causes theshrink
joint
area the workpiece first to expand and then to
WPS
QT Welding
QuenchedProcedure Specification
and Tempered area in the workpiece material first to expand and then to shrink
during cooling. Welding distortions cause changes to the
QT Quenched and Tempered during cooling. Welding distortions cause changes to off-
the
workpiece position and therefore add uncertainties to the
workpiece position and therefore add uncertainties
line program. [2] Traditionally in robotic welding production to the off-
1. Introduction line workpieces
the program. [2]are Traditionally
held in place in robotic
duringwelding production
tack welding and
1. Introduction the workpieces are held in place during tack welding anda
welding by jigs in order to minimize the misaligning of
Off-line robotic welding programming has proved to be a workpiece due to welding distortions. The use of jigs is time-a
welding by jigs in order to minimize the misaligning of
Off-line
useful robotic
tool for weldingtheprogramming
decreasing time requiredhas
forproved to be a
programming workpiece due
useful tool for decreasing the time required for programming consuming andto welding
requires distortions.
manual labor, The use of jigs
which is time-
makes it
the welding robot in mass production. However, in low-volume consuming toand
expensive use.requires
[1] In manual labor, whichwelding
jigless multi-robot makes the it
the welding robotrobotic
high-variation in mass welding
production.production,
However, inthe
low-volume
off-line expensive to use. [1] In jigless multi-robot welding the
production costs can be significantly decreased, but the
programming has not been often adapted. One reasonoff-line
high-variation robotic welding production, the being production
problem willcosts
be how canthebe significantly
welding decreased,
distortions but the
can be controlled
programming
that has not been often
off-line programming adapted.
requires One reason
significant being
amount of problem will be how the welding distortions can be controlled
that off-line programming requires significant amount of
2351-9789©©2020
2351-9789 2020
TheThe Authors.
Authors. Published
Published by Ltd.
by Elsevier Elsevier Ltd.
2351-9789
Thisisisanan
This ©open
2020
open The
access Authors.
access Published
article
article under under
the by
theElsevier
CC BY-NC-ND Ltd.
CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
license https://creativecommons.org/licenses/by-nc-nd/4.0/)
This is an open
Peer-review
Peer-review access
under
under article under
responsibility
responsibility the scientific
of the CC BY-NC-ND
of the license
scientific
committee https://creativecommons.org/licenses/by-nc-nd/4.0/)
committee
of the FAIMof the FAIM 2021.
2020.
Peer-review under responsibility of the scientific committee of the FAIM 2020.
10.1016/j.promfg.2020.10.043
 Hannu Lund et al. / Procedia Manufacturing 51 (2020) 302–308 303
2 H. Lund et al. / Procedia Manufacturing 00 (2020) 000–000

beneficially so that acceptable quality can be achieved. grippers. Workpieces are positioned and measured on the
Therefore, there is a need for estimating welding distortions turning L-table MT1-1000 S2X equipped with the magnetic
reliably and to be able to adjust robot programs to estimate positioning system. The whole robotic system is controlled by
welding distortions during positioning. a pair of NX100 controllers and MotoNIS (Motoman Network
Finite element analysis (FEA) has been the traditional Information Server) is used for communication with
method for the estimation and simulation of welding distortions workstations when running custom computer code. Fig. 1
and the simulated distortion values have been accurate. presents a simulation of the experimental set up of the laser
However, the welding distortions FEA is a complex process, as scanning procedure.
the welding process itself has multiple nonlinear characteristics
involved in thermo-elastic-plastic behavior of weld and base
metal and in structural analysis, therefore a high-level of
expertise is required from the designer/off-line programmer to
be able to realistically estimate welding distortions. [3, 4].
Some attempts to control the amount of distortions by using
FE-simulations and off-line programming have been made, but
the focus has been on the controlling of the heat input to reduce
the distortions, instead of estimating the right setup angle of the
workpiece plate which would be aligned in to right final
position by the welding distortions [2, 5].
With the FEA approach the actual realized welding
distortions are not known. Therefore, there is a need for Fig. 1. Experimental setup of scanning the angle between two plates,
simulated in Delfoi Arc 4.1
developing a system that can estimate welding distortions
during the off-line programming of robots based on a known
The experimental setup for measuring the position of fillet
amount of realized welding distortions. An alternative to
weld joint and measuring the amount of deformation consisted
simulating welding distortions with FEA could be measuring
of laser line scanner attached to a robot 1 and workpiece made
test specimens angular distortion either manually or with a laser
of two S700 grade quenched and tempered steel plates with
scanner and create a database based on the measurements. A
thickness of 5 mm. The workpieces are held in place
database can be utilized in making of WPS (welding procedure
perpendicularly to each other by a magnetic positioning system
specification) and further on welding robots can be off-line
on a turning L-table MT1-1000 S2X and a magnetic gripper
programmed to take into account welding distortions based on
attached to robot 2. The nominal joint geometry is a T-joint
the information given in either WPS or measurements in the
with a fillet weld, a 90° angle between plates and a root gap of
database. Laser scanning produces accurate depth
0 mm. The laser line scanner used was a Micro-Epsilon
measurement data, which can be analyzed online. Laser
ScanControl 2950, which was held in 45° angle in relation to a
scanning has been widely used for measuring object features,
base plate during the measuring. The measuring line width of
such as weld joint geometry, and therefore it could be a useful
the scan is 58 mm. The angle between the plates was scanned
method to measure the angular distortions caused by welding.
for the whole length of the workpiece. The measured angle data
[6, 7] From the measured data, it is possible to calculate the
was analyzed in Winteria weld geometry analysis software.
required correction that needs to be made to the robots’
Fig. 2 presents the schematic of the laser scanning procedure
program during positioning.
and Fig. 3 presents the laser scanning setup, in which the laser
The research problem of this study is to create a welding
scanner is held in the desired angle in relation to plate 1 and 2
distortion estimation method, that is based on the known
and moved along the length of the workpiece. 5 manual
amount of measured distortion, in multi-robotic welding
measurements were taken in the middle of the workpiece with
simulation and off-line programming software. To solve the
an angle gauge. The median of each measurement was used to
research problem, the following research questions were set:
prevent false measurements.
what is the angle that the handling robot needs to adjust the
second plate of the assembly, in order to compensate the
misaligning caused by welding distortions; how to simulate the
measured welding distortion compensation process?

2. Methods and materials

The experiment environment is a jigless multi-robot

welding cell with two Motoman robots and a turning L-table.
Robots used are welding robot MA 1900 (Robot 1) with 3 kg
handling capacity and handling robot ES165N (Robot 2) with
165 kg handling capacity. Robot 1 is used to weld and measure
the workpieces and Robot 2 is used to manipulate the
workpieces. Robot 1 is equipped with the welding torch and a Fig. 2. Schematic of the laser scanning procedure of the plate angle α
laser line scanner and the robot 2 is equipped with magnetic
304 Hannu Lund et al. / Procedia Manufacturing 51 (2020) 302–308
H. Lund et al. / Procedia Manufacturing 00 (2020) 000–000
Fig. 3. Laser scanning of the joint
The digital twin of multi-robot jigless welding cell
experiment environment was modeled in the simulation
environment in Delfoi Arc 4.1 software. The digital twin of
multi-robot welding cell consisted of welding robot MA 1900
equipped with welding torch a laser line scanner, handling
robot ES165N equipped with magnetic gripper and the turning
L-table MT1-1000 S2X equipped with magnetic positioning
system. The CAD-models of magnetic gripper, magnetic
positioning system, workpiece and laser scanner were modeled
in Solidworks and the functionalities of components were
implemented in Delfoi Arc 4.1. Robot programs for estimating
the effect of welding distortions on a workpiece were made in
the simulation environment. The robot 2 was programmed to
have a nominal position for the positioning of the workpiece,
which was in the centerline of the first workpiece plate and
perpendicularly against the plate at a 90° angle. The position of
workpiece and robot positions were corrected according to the
measurements of the welded workpiece, in order to estimate the
deformations and distortions caused by welding. The
integration of a knowledge-based welding distortion estimation
method to an off-line programming and simulation software is
presented as a function diagram in Fig. 4.
Before the welding experiments, robot 2 assembled the plate
2 perpendicularly on top of plate 1 to a nominal 90° angle and
the plates were tack welded on both ends of the plates. After
tack welding, the plate angle was measured. The welding
experiments were performed in three passes in full length of the
workpiece, in PB position with GMAW (gas metal arc
welding) process. One pass was equivalent to throat thickness
of 4 mm. The welding parameters for each pass were as
follows: welding current 218 A, voltage 26.2 V and welding
speed 7 mm/s. The shielding gas composition was Ar + 8 %
CO2 + 0,03 % NO, shielding gas flow was 15 l/min and a
forehand technique with a 15° travel angle was applied. The
workpiece was let to cool down to room temperature before the
laser measurement and producing the next pass to prevent
inconsistencies in measurement. The plate angle was measured
before and after each weld to get reliable data for the analysis.
Plate angular distortions were evaluated between the different
passes and the reference. Total deformation after each weld was
analyzed and compared with the manual measurement of each
welded pass. The results of the distortion analysis were used as
3
a basis of the simulation model to determine the required preset
plate angle to reach perpendicular joint as a welded outcome.
Fig. 4. Integration of knowledge-based welding distortion estimation to an
off-line programming and simulation software.
3. Results and analysis
The angle between plates was measured with the laser line
scanner after plates were tack welded together and before the
fillet weld was made. The results of the laser scanning of the
joint before welding shows that the median value of the angle
between the plates is 90.487° and the standard deviation is
0.03°. The minimum angle value measured was 90.373° and
the maximum angle value measured was 90.556°. The results
of laser scanning are presented in Fig. 5, in which the plate
angle is shown in a vertical axis and the measuring length is
shown in a horizontal axis. The plate angle was also measured
manually with angle gauge and the measured median plate
angle before welding was 90.6° when measured in the middle
of the plate.
After welding, the T-joint was measured with the laser
scanner after each weld. The results of laser scanning after the
welding shows that the median value of angle between plates
was 89.731° in weld 1, 88.505° in weld 2 and 88,029° in weld
3. Scanning results after each weld are presented in Fig. 6. The
plate angle is shown in vertical axis and the measuring length
is shown in horizontal axis. The manual measurement of the
workpiece showed 89.6°, 88.5° and 88.2° after weld 1, 2 and 3,
respectively. The differences between measurement methods
results can be explained by the measuring methods accuracy
and also by the possible human error during manual measuring.
In Fig. 6, it can be seen that the welding distortions are smaller
at the end of the weld seam. However, the difference in
distortion is relatively small and does not have effect in practice
in this case.
 Procedia Manufacturing 00 (2020) 000–000
www.elsevier.com/locate/procedia

Hannu Lund et al. / Procedia Manufacturing 51 (2020) 302–308 305

Fig. 5. Result of the pre-weld scan of the weld joint.

Fig. 6. Results of laser scanning of the weld joint.

The mean, median, standard deviation, minimum and
maximum angle values of the laser scans are presented in Table
1. Table 1 also provides the mean values of the manual
measurements after each weld. From Table 1 it can be seen that
for each weld scan the standard deviation is between 0.118 and
0.213.
The resulting angle after the third weld pass was 88.029°
when measured with the laser scanner. Therefore, the required
amount of angle correction to robot 2 positioning during the
assembly of the second plate, can be calculated by extracting
the measured plate angle after welding from the angle before
welding, as shown in equation 1.

2351-9789 © 2020 The Authors. Published by Elsevier Ltd.

This is an open access article under the CC BY-NC-ND license https://creativecommons.org/licenses/by-nc-nd/4.0/)
Peer-review under responsibility of the scientific committee of the FAIM 2020.
306 Hannu Lund et al. / Procedia Manufacturing 51 (2020) 302–308
H. Lund et al. / Procedia Manufacturing 00 (2020) 000–000 5

Table 1. Measurements of the T-joint before and after welding. The program, which edits robot programs of a simulation
Measurement Mean Median Std Min Max model, was implemented as a script in the model, which is run
[°] [°]
Dev
[°] [°] during each simulation run. The position of robot 2 was edited
in the following way, first the program reads the measured
Laser scanner Pre- 90.481 90.487 0.030 90.373 90.556
angle, which is required to compensate the welding distortions.
weld
Then the program finds the workpiece, workpiece bottoms
Laser scanner Weld 1 89.736 89.731 0.118 89.519 89.989
coordinates and robot 2 tool point coordinates in the simulation
scan
environment and calculates the distance between workpiece
Laser scanner Weld 2 88.600 88.505 0.213 88.339 89.187
bottom and a tool point. In the next, step program reads all the
scan
statements in a robot program and finds call for estimation of
Laser scanner Weld 3 88.111 88.029 0.206 87.760 88.717
welding distortions, which it uses to infer the correct
scan
positioning point. Finally, the program creates four target
Manual measurement 90.62 90.6 - 90.6 90.7
points for departing from the positioning point, rotating the
pre-weld
workpiece, moving robot 2 so that the bottom of the workpiece
Manual measurement 89.64 89.6 - 89.6 89.8
is in the same position as it was before rotation and lowering
post-weld 1
the workpiece against the first plate to a positioning point. The
Manual measurement 88.52 88.5 - 88.4 88.6
program which edits the robot 2 position based on the measured
post-weld 2
distortion angle is presented in Fig. 8 as a function diagram and
Manual measurement 88.20 88.2 - 88.2 88.2
in figures to illustrate the main steps of the program.
post-weld 3

𝛼𝛼 = 𝛼𝛼𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 − 𝛼𝛼𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 (1)

Where α is the required angle to be considered in robot 2

position, αbefore welding is the measured angle before welding and
αafter welding is the measured angle after welding.
When the measured angle values are inserted to equation 1,
the required angle that robot 2 needs to adjust its position in the
pre-welding position can be calculated. For laser-scanned
measurement equation 1 solves as follows:

𝛼𝛼 = 90.487° − 88.029°
𝛼𝛼 = 2.458°

For manual measurement equation 1 solves as follows:

𝛼𝛼 = 90.6° − 88.2°
𝛼𝛼 = 2.4°

From equation 1, it can be calculated that the required angle

that robot 2 needs to adjust its position in pre-welding position
is 2.458° when measured with laser scanner and 2.4° when
measured manually. This means that, if the plates are ideally
placed perpendicularly to each other, the robot 2 needs to offset
the plate 2 a 2.458°, or 2.4° if measured manually, to the Fig. 8. Illustration of steps in welding distortion estimation program in off-
opposite side where the weld will be made in order to predict line programming and simulation software.
the misaligning caused by welding distortions. The result of the
robot 2 position adjustment is presented in Fig. 7. 4. Discussion and conclusions

In this study, the research problem was to create a welding

distortion estimation method, which is based on a measured
distortion, in multi-robotic welding simulation and off-line
programming software. To solve the research problem the
research questions were set to find how the angle measurement
data can be used to predict the welding distortion in a simulated
environment. To answer the research questions the angle
between two plates in T-joint was measured with a laser
scanner and manually with an angle gauge. The required offset
Fig. 7. The situation in the simulator a) before and b) after adjustment to to the handling robots’ position during assembly of the second
robot 2 position. plate was calculated based on the measurements. For the
 Hannu Lund et al. / Procedia Manufacturing 51 (2020) 302–308 307
6 H. Lund et al. / Procedia Manufacturing 00 (2020) 000–000

practical use of the method studied in the article, a knowledge FEA approach. Therefore, FEA results could be used to
database for different throat thicknesses, number of passes, estimate the welding distortions of the test sample and the
plate thicknesses and other variables should be created. Also, methodology presented in this paper could be used to correct
the practical tests after the corrected position are a subject of the errors of the calculated distortions in positioning. Fig. 9
future research. presents how the FEA results could be integrated into a
Regarding measurements, it was noticed that at the end of knowledge-based welding distortion estimation method.
the weld seam the distortions were smaller than at the start of
the weld seam. This might be caused by the pre-heating of the
end-joint area due to the weld heat, which might cause more
equal thermal gradient through the thicknesses of the plates and
thermal expansion in the plates, which then decreases the effect
of welding distortions at the end of the weld joint. Another
explanation might be that at the time when the end of the seam
is being welded, the weld seam has already solidified and
cooled at the beginning of the joint, which then supports the
plates so that the distortions will be smaller in the end. Due to
this phenomenon, manual measurements should be taken in the
beginning of the plate, in the middle and at the end of the plate
to verify the laser triangulation measurement.
According to the research results, the calculated angle, in
which the handling robot needs to offset its position during
assembly of the second plate to compensate the effect of
welding distortion, was 2.458° when measured with the laser
scanner and 2.4° when manually measured with an angle
gauge.
The simulation results show that handling robot is capable
to make corrections to its nominal position, during the
assembly of the second plate, according to the measured and
calculated data. Both methods are suitable for operating in the
limits of robots positioning accuracy. The differences between
the results of both measuring methods can be explained by the
effect of human error and measurement inaccuracies of manual
measuring. The laser scanning method is quite accurate, but
still the laser scanning can have measurement inaccuracies if
the laser light is illuminated from the highly reflective metal
surface or when the laser light is blocked by some object Fig. 9. Schematic of how FEA can be integrated into knowledge-based
geometry, such as particles in the air or on the workpiece welding distortion estimation method
surface. Sometimes the amount of data gathered with a laser
scanner might exceed the capability of the analyzing software The results are utilizable in the robotic welding production
or even hardware, which then might cause measurement and off-line programming of welding robots with a condition
inaccuracies. The laser scanning has an advantage over manual that only the position of the workpiece is changed during
measurement, as the measured angle values can be given to the production. The welding parameters and workpiece materials
welding distortion estimation program directly after scanning must remain constant or a new measurements of welding
without a need of the human user to input the angle values distortions are required. Furthermore, the results are a step
manually. towards a jigless multi-robot welding production, as the
The developed methodology presented in this paper can be simulation shows that the handling robot is capable of adjusting
applied in quality control of production and adaptation to its position according to the calculated angle, which reduces the
production errors, which differs from the current state of art of need for using expensive and time-consuming jigs to fix the
estimating the welding distortions with FEA. FEA can be plate into its position [1]. The methodology presented in this
accurate, but it cannot verify the realized amount of welding paper can be used to reduce the robotic welding production
distortions and therefore cannot be used in quality control. delays and post-weld work, as the welding distortions can be
Furthermore, the FEA requires high-level of expertise in considered during off-line programming, which should reduce
structural analysis and thermo-elastic-plastic behavior to get or even prevent the occurrence of distortion related errors.
accurate results in estimating welding distortions, but with the As the methodology used in this research can be generalized
developed knowledge-based estimation methodology the to be used with other materials and weld joints, however, the
adjustments to robot 2 position can be made automatically direct results of this research are applicable only for the
based on measurements results from the test workpiece, and material used in the research, which was S700 QT steel.
therefore the required expertise in the field of welding Similarly, the results are applicable to a single-sided fillet
distortions from the off-line programmer is a lot less than in welded T-joint case only. The results are not directly applicable
308 Hannu Lund et al. / Procedia Manufacturing 51 (2020) 302–308
H. Lund et al. / Procedia Manufacturing 00 (2020) 000–000
to different size weld throat thicknesses, as the welding
distortion is dependable on the throat thickness. In case the
different throat thickness is used, a new welding distortion
measurement is required. Similarly, the results are applicable
only to the plate thickness used in a applied workpiece. Still,
for other plate thicknesses, throat thicknesses, materials and
other weld joint types, the procedure presented in this paper can
be applied for estimating and preventing the effect of welding
distortions.
For future research, it is suggested to test if the angular
distortion remains constant for all plate set angles or does the
set angle have an effect on angular distortion amount. Further
research is required on testing the methodology in real
industrial multi-robot welding cell environment to verify that
the simulated distortion estimation method works on actual
jigless multi-robot welding cell. Further research is suggested
for testing the presented methodology on other materials and
weld joints to create a database, which can be utilized in off-
line programming and robotic welding production. Further
research could be applied to make similar kind of estimation
method based on FE-calculations, therefore reducing the need
for making welding test specimens. Another research topic
could be a measuring of welding distortions of long test
specimens, to study if the same phenomenon in which the
welding distortions are smaller at the end of the weld seam
occurs.
7
Acknowledgements
The research work was completed during the project:
Digitalized Robotic Welding for Demanding Applications
“DigRob” financed by the Business Finland.
References
[1] M. Bejlegaard, T.D. Brunoe, K. Nielsen, A changeable jig-less welding
cell for subassembly of construction machinery, in: IFIP Adv. Inf.
Commun. Technol., 2018. doi:10.1007/978-3-319-99704-9_37.
[2] W. Liang, D. Deng, Influences of heat input, welding sequence and
external restraint on twisting distortion in an asymmetrical curved
stiffened panel, Advances in Engineering Software, 115, (2018), 439-
451.
[3] A.A. Bhatti, Z. Barsoum, H. Murakawa, I. Barsoum, Influence of
thermo-mechanical material properties of different steel grades on
welding residual stresses and angular distortion, Materials & Design
(1980-2015). 65 (2015) 878-889.
[4] P.C. Adamczuk, I.G. Machado, J.A.E. Mazzaferro, Methodology for
predicting the angular distortion in multi-pass butt-joint welding, Journal
of Materials Processing Technology, 240 (2017) 305-313.
[5] F. Sikström, & M. Ericsson, Integration of finite element analysis and
computer aided robotics for advanced programming of robotized
welding. ASM Proceedings of the International Conference: Trends in
Welding Research. (2009) 454-460. doi:10.1361/cp2008twr454.
[6] J. Ratava, S. Penttilä, M. Lohtander, P. Kah, Optical measurement of
groove geometry, Procedia Manuf. 25 (2018) 111–117.
doi:https://doi.org/10.1016/j.promfg.2018.06.064.
[7] J. Muhammad, H. Altun, & E. Abo-Serie, Welding seam profiling
techniques based on active vision sensing for intelligent robotic welding.
Int J Adv Manuf Technol (2017) 88: 127-145.
https://doi.org/10.1007/s00170-016-8707-0.