DEGREE PROJECT IN MECHANICAL ENGINEERING,

SECOND CYCLE, 30 CREDITS

STOCKHOLM, SWEDEN 2018

Evaluation of current welding

robot and improvement proposals
for future use

JOAKIM NILSSON

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ENGINEERING SCIENCES
 Degree program in Mechanical Engineering

KTH Royal Institute of Technology, Stockholm

MASTER THESIS

Evaluation of current welding robot and improvement

proposals for future use

Joakim Nilsson

Thesis project for the degree Master of Science

School of engineering sciences, Naval architecture

Supervisors: Zuheir Barsoum (Academic, KTH)

Jyri Palmu (Company, Valmet)

Abstract

Welding of elements, a heat exchanging component, is one of the most essential processes during the
manufacturing of Tubel evaporators. Weld joints consists of fillet welds of thin tubes against a plate
on one end, and on the other end against a thicker pipe with entry holes for the tubes. Since earlier a
welding robot station has been invested to the production line to improve manufacturing quality and
productivity. The results from the robot welding hasn’t been reliable enough though, and have led to
additional operations to fix the problems that has occurred.

This thesis project has the objective to investigate and analyse the problems occurred from earlier use
of the robot station. The main challenges of the process have been identified and key factors that
influence the welding performance are recognised to fully understand the problem. A list of
improvement actions has then been proposed by first hand considering in terms of quality and
productivity. The improvement actions are then evaluated by test welding.

Materials used in the component and during test welding are either duplex or austenitic stainless
steels, which weldability differs significantly between these. Related discontinuities and defects are
similar for both material, but occurrences of these are higher when welding duplex steels.

Improvement actions has included positioning of the electrode, parameter optimization, modification
of robot equipment, base material improvements and welding sequence. Results of welding austenitic
steels was considerably better and showed good results after the test welding, while welding of duplex
steel likely needs design modification of the component to satisfy quality requirements and standards.

Sammanfattning

Svetsning av element, en värmeväxlande komponent, är en av de viktigaste processerna under

tillverkningen av Tubel-indunstare. Svetsningen som förekommer är kälfogar av tunna tuber mot en
plåt på ena änden, och på andra ändra änden mot ett tjockare rör med ingångshål för tuberna. Sedan
tidigare har en svetsrobotstation investerats i produktionslinjen för att förbättra tillverkningen
gällande kvalitet och produktivitet. Däremot har inte resultaten från denna varit tillfredsställande
nog, som lett till extra arbete för att åtgärda de problem som förekommit.

Detta examensarbete har syftet att undersöka och analysera de svetsrelaterade problem som funnits
sedan den tidigare användningen av robotstationen. Huvudutmaningarna av processen har blivit
identifierade och nyckelfaktorerna som påverkar svetsprestanda är undersökta för att så fullt som
möjligt förstå sig på problemet. En lista av förbättringsförslag har tagits fram som i första hand
betraktar förslag gällande kvalitet och produktivitet. Dessa förbättringsförslag har sedan utvärderats
genom testsvetsning.

Material som använts i komponenterna och under testsvetsningen har antingen varit duplex- eller
austenitiskt rostfritt stål, vilkets svetsbarhet skiljer sig väsentligt från varandra. Relaterade
diskontinuiteter och defekter är lika mellan båda dessa stål, men förekomsten är betydligt högre då
duplexa stål svetsas.

Förbättringsåtgärder har omfattat svetspositionering av elektroden, parameteroptimering,

modifikation av robotutrustning, åtgärder av basmaterial och svetssekvenser. Testsvetsning av
austenitiskt rostfritt har varit betydligt enklare och gett goda resultat, medans svetsning av duplexstål
förmodligen behöver förbättrande åtgärder av komponentdesignen för att klara av de krav som
kvalitetskrav och standarder ställer.

1
Abbreviations

GMAW - Gas Metal Arc Welding

MIG - Metal Inert Gas

MAG - Metal Active Gas

CMT – Cold Metal Transfer

TIG - Tungsten Inert Gas

LBW – Laser Beam Welding

RSW - Resistance Spot Welding

ABB - ASEA Brown Boveri

TCP – Tool Centre Point

NDT – Non-Destructive Testing

PT- Penetrant Testing

VT - Visual Inspection

PED – Pressure Equipment Directive

ISO - International Organization for Standardization

EN – European Norms

2
Contents

1 Introduction to project ........................................................................................... 5

1.1 Background ............................................................................................................................. 5
1.1.1 Valmet production ............................................................................................................... 5
1.1.2 Evaporators ......................................................................................................................... 5
1.1.3 Manufacturing process ........................................................................................................ 5
1.2 Project assignment .................................................................................................................. 6
1.2.1 Problem ............................................................................................................................... 6
1.2.2 Overall goal and purposes ................................................................................................... 6
1.3 Method .................................................................................................................................... 7
1.3.1 Strategies ............................................................................................................................. 7
1.3.2 Activities .............................................................................................................................. 7
1.4 Limitations ..............................................................................................................................8
1.5 Requirements ..........................................................................................................................8
2 Theory Background ................................................................................................ 9
2.1 About Robot welding............................................................................................................... 9
2.2 Suitable welding processes for robot .................................................................................... 10
2.3 Programming ........................................................................................................................ 10
2.4 Work area and safety.............................................................................................................. 11
2.5 Tolerances and Seam finding/tracking .................................................................................. 11
2.6 Quality requirements and standards .................................................................................... 13
3 Robot welding at Gothenburg workshop and previous results ............................. 14
3.1 Robot welding station and work object................................................................................. 14
3.2 Previous results ..................................................................................................................... 14
4 Main challenges and currents state analyse.......................................................... 16
4.1 Defects and discontinuities ................................................................................................... 16
4.2 Laser searching problems ..................................................................................................... 19
4.3 Main challenges.....................................................................................................................20
5 Improvement proposals ....................................................................................... 21
5.1 Investment of a new weld torch ............................................................................................ 21
5.2 Root protection flux .............................................................................................................. 22
5.3 Short arc length when welding at the small space between tubes ........................................ 22
5.4 Using slope down at switch off and higher starting current ................................................. 23
5.5 Inspection of ingoing components ........................................................................................ 23
5.6 Use of an optimal welding sequence that spreads out heat more evenly ............................. 23
6 Results ................................................................................................................. 24

3
 6.1 Impact of root protective flux ............................................................................................... 24
6.2 Welding of a complete lower header against tubes............................................................... 25
6.3 Welding against upper header .............................................................................................. 27
6.4 Welding of austenitic stainless steel .....................................................................................28
7 Alternative design for effective production .......................................................... 29
7.1 Use of a “dig out” around the tubes ...................................................................................... 29
8 Discussion, conclusions and future work ............................................................. 30
8.1 Discussion .............................................................................................................................30
8.2 Conclusions ........................................................................................................................... 31
8.3 Future work ........................................................................................................................... 31
9 References ........................................................................................................... 32

4
 Chapter 1

1 Introduction to project
This chapter describes a short background of the company’s business and the manufacturing at the
workshop. It’s also presenting the project assignment, problem and overall goals. Finally it also
contains a part describing the methods, and a section showing the project limitations and
requirements demanded.

1.1 Background

1.1.1 Valmet production

Valmet is Finnish company that is developing and providing technologies, services and automation
systems for the pulp, paper and energy industries. In the pulp and energy business line, Valmet is a
leading supplier of evaporation plants and boilers worldwide.

The workshop located in Gothenburg has for several years been manufacturing high performance
evaporators called Tubel. As a step of a new production strategy, it has been decided to make a
technology transfer from the Gothenburg workshop to Lapua workshop. The main target of this
decision is to make improvements to Tubel manufacturing process in order to assure competitive
production in the future.

1.1.2 Evaporators
Evaporators are used as a recovery process to reuse the remaining secondary product from the
sulphate process, called black liquor. The black liquor contains a great energy content, but since the
black liquor mostly contains of water it needs to be evaporated so it later can be used for combustion.
Black liquor is evaporated from having around 15% solids up to 60-80% solids. It’s important to reach
as high dry content as possible, since the heat efficiency gets higher and reduces pollution by
decreasing emissions of sulphur-containing gases.

To make the process more effective, and reducing the amount of required steam, the black liquor is
refined by letting it evaporate through a number of (usually 6-7) evaporators linked in series. The
concentrated black liquor is then used as fuel in recovery boilers, which energy can be reused in the
pulping process or by generating power for other purposes [1].

Tubel evaporators are constructed with the design of a pressure vessel. The process of evaporating
black liquor in a Tubel, is by letting the black liquor passing a system of heat exchangers (called
elements) that is connected in groups. Each element itself consists of several tubes which is heated by
letting hot steam passing inside of them. The black liquor gets heated, and the water boils off and
passes away throughout an outlet.

1.1.3 Manufacturing process

The main process of the Tubel manufacturing is by assembling components manufactured by
subcontractors. The workshop also performs metal forming and machining as sawing, drilling and
bending. Welding is the joining technique used in the workshop for assembling the components. An
element of a heat exchanger is manufactured by joining several tubes into a steam inlet, called lower
header, and an outlet, called upper header. Each upper header of an element is also connected by a
steam header, were the steam is distributed to the different elements. The lower headers are similarly
joined into a condensate header, where the condensate water is collected and let out from the system.

5
1.2 Project assignment

1.2.1 Problem
Welding between tubes and headers has mostly been made by manual welding, but for the last years
a welding robot have been used for occasions during the production. Since welding between tubes and
headers is the most essential part of the manufacturing, one of the focus areas of improvement actions
regarding the technology transfer is a functional robot welding. The results of using the robot welding
in the manufacturing has not been reliable enough though, and have led to manual welding phases
afterwards to add an extra support to get the required quality.

The main assignment is to understand what the real challenges of robot welding are. To success in
that there is need to study the current state with operators, supervisor, welding engineer, quality
control manager and the supplier of the welding robot. After the evaluations, it is important to create
a list of improvement proposals so that the improvement actions could be started as fast as possible
when taking use the robot station in Lapua.

1.2.2 Overall goal and purposes

The goal of the project is to deliver a report that is supposed to be a decision support concerning the
improvement actions regarding the welding robot and future manufacturing between tubes and
headers.

One purpose of the technology transfer is to get understanding what the real challenges of the
manufacturing process is. The intention of the project is to identify and examine these, according to
implementing automation technology and robot welding. When the examination is done, a written
document of the evaluation is reported. The main parts of the report will consist of:

• A background of the previous results with robot welding. To understand the problem, one
should know the former development of the robot station. Difficulties with using robot welding
and problems that have occurred, but also improvements that has been made.

• List of improvement proposals of the current state operation, with the intent that these can be
implemented into the production in Lapua.

• Evaluation of the proposals. Test welding and inspection of the results.

• Ideas of alternative designs for a more effective production

The purpose of the project for the student is to deepen and broaden his knowledge about robot
welding and its implementation in the industry. The student will also get a deeper insight of other
welding related issues that a welding engineer can bump into, such as welding planning, design issues
and logistics.

6
1.3 Method

The project will be divided into the following parts

1. Information gathering of the previous use and problems occurred, current state analyses of
the welding process capability and main challenges. (Location: Gothenburg workshop)
2. Definition of the improvement proposals and testing of these. New constructive ideas of
designs to improve the production. (Location, Lapua workshop)

1.3.1 Strategies
• To have a proper background about the current situation, the project will start with gathering
as much information about the existing and present use of the robot welding. Literature study
will also be done in the start-up phase, so that the theoretical knowledge about the subject is
satisfied enough to perform the project.
• When the information is gathered, it will be analysed and the main challenges of the process
will be identified. The theoretical knowledge learned from the literature study will be used to
study the current situation and examine the process capability. This part of the project will be
done in consensus with the current employees at Valmet, Gothenburg who has greater insight
in the current situation.
• When the current process is examined, proposals for improvement will be listed. When the
proposals are listed, they should be checked if they are realistic according to the production in
Lapua. The proposals should for example be checked with production capability and economic
factors. The proposals will then be examined by test welding of the product and the results will
be gathered to the report. After the test welding is performed, concepts of new component
designs are ought to be made which could simplify and improve the robot welding of elements.

1.3.2 Activities
1. Start of project
2. Literature study
3. Information gathering from staff and employees in Gothenburg about their previous results
of robot welding
4. Identification of main challenges
5. Current state analyse
6. List of proposals
7. Check if the proposals are realistic
8. Test welding
9. Evaluation from the testing
10. Ideas of alternative component designs.

7
1.4 Limitations

The focus of the project is to deliver a technical report specifically about the welding phase between
headers and tubes. To fulfil a reasonable scope of the thesis, limitations are needed to get a suitable
depth of the assignment. These limitations are needed due to time boundaries and initial knowledge
considering the field of study. The following limitations might need to be investigated and evaluated
by similar side projects.

• Robot operation and programming

• Robot station logistics

• Component material choice, already decided at design phase

1.5 Requirements

Requirements for accepted weld qualities and performance of the robot welding are stated below. If
these requirements are not fulfilled, the application of the robot welding should be questioned if its
suitable at all for the current product/design.

• Weld beads accepted by EN 13445 – Unfired Pressure Vessels.

• Welding proposals should not make any negative effect on the products performance.

• Welds should withstand the corrosive media of black liquor.

• Higher production rate of welded elements compared to manual welding.

• Quality of the welds are more consistent compared to manual welding.

8
 Chapter 2

2 Theory Background
This chapter gives the reader a theory background that is relevant for the project assignment, which
improves the understanding of the problem.

2.1 About Robot welding

To remain on a competitive level, manufacturing companies always needs to find a way to progress,
or at least maintain, the level of quality and productivity of its manufactured products. When a
company is turning towards to automate its fabrication, the main goal is to increase the production
flow and keeping defects as low as possible, while in the same time keeping the costs for labour and
materials down.

Investing in robot welding cell can be of a great deal for company that wish to achieve those goals.
Although, implementing such a robot welding cell requires comprehensive preparation and planning
to make it as much efficient as possible.

Advantages of using robot welding according to [2] are

• Higher productivity: Arc time can be raised from 30-40 % up to 60-80% and arc travel speed
can as well be higher.

• A robot with one operator can replace 2-4 welders.

• Weld quality is more consistent and usually higher.

• Better working environment, since the operator doesn’t need to stand close to the arc.

• Necessary reorganisation and new thinking of surrounding activities associated to the

acquisition of the robot helps to raise efficiency.

• The workshop doesn’t need to deal with difficulties of recruiting and keeping qualified welders.

There are some things although that needs to be considered when implementing a robot station.
Ingoing parts requires high tolerances, as well as the robot station fixtures. If the company are
inexperienced with robots from before, wide training for the involved personnel are highly necessary.
Other things needed to be considered before deciding robot welding according to [3] are

• Accessibility: Can the robot reach all the positions necessary to make all the welds?

• Welding deformations: If deformations occur during welding, can the robot compensate for
these so it’s not welding incorrectly?

• Robot cell fixtures: Are the fixtures installed in the cell suitable for robot welding?

• Welding sequence: Can the best possible weld sequences be obtained by robot welding?

Generally, one can say that robot welding is suitable for products that are continuously identical and
contains repetitive seams. If the robot is programmed correctly and auxiliary equipment installed
properly, then welds will stay perfectly the same with equal properties and dimensions on every new
product welded.

9
2.2 Suitable welding processes for robot

Usually a robot welding cell consists of a robot with a control system, fixtures and/or manipulators
for the workpiece, power sources and associated welding equipment’s. A normal robot used in
industry often features 6 axes. A six axes robot allows greater flexibility than a robot with fewer axes.
To achieve greater accessibility for welding, manipulators can be used which positions the workpiece.

The most used welding process used for robots are GMAW processes, i.e. MIG/MAG welding. Other
usable processes are TIG, LBW and RSW. In this report, MIG welding will be the process examined.
This is due to the materials of the product (stainless steel). TIG is also a possibility according to the
material, but due to the design of the product it could be difficult to accomplish welding with that
process [4].

2.3 Programming

Programming of the welding motions on the workpiece is done by using a device containing a joy-
stick which is used to jog the robot from point to point. The operator is controlling the welding
motions by considering the position of the tip of the stick out against the seam, angle of the electrode,
arc travel speed and weaving patterns. The point which is oriented and moved around by the operator
is called TCP (tool centre point). Each point is saved into the memory of the control system, together
with instructions of the welding motions. TCP is exclusively always located at the stick-out of the
welding torch, resulting in that the stick-out doesn’t move when rotating the tool (i.e. angle of the
electrode is easily adjusted). TCP must always be calibrated now and then to assure that the tool
always operates in the correct position, or updated if the robot runs in to something which and the
axes are dislocated.

Figure 1: ABB flexpendant is used to perform different task when operating the robot [5].

Welding parameters (such as wire feed, voltage, current etc.) are often modified through the power
source, and are saved there as welding program which the robot then downloads according to the
welding instructions from the motion programming.

A deeper and a more advance programming must be made if the welding process is sought to be more
autonomous and motion programming is desired to be more simplified. A well thought software
architecture can make a comprehensive solution of the problem turned into a much easier one. In
case the robot also has aided tools, programming of these into the system also needs to be made
carefully made so that you can get the best out of it.

If the workpiece has repetitive seams, but located in different positions, it is therefore possible to
program the robot using the welding motions of one seam, and automatically move those

10
programmed motions to another seam without the need of creating welding motions for that one as
well. A well thoughtful programming leads most often to time saving in a later stage.

Another alternative to directly program the welding motions on the workpiece is instead to do this
“off line” by computer. This requires a 3D-graphical software where CAD-models of robot, fixtures
and workpiece is used to simulate the process. The reason to use off line programming is to avoid
downtime of the production. Welding motions can be adjusted without the need of stopping the robot
station [4].

2.4 Work area and safety

To extend the work area for bigger work objects, there are possibilities to install the robot on rail going
carriers or hanging them on cranes and portals. Besides the 6 axes the robot, 3 more axes (height-,
length- and sidewise of the work object) could be added to further increase the work area to make it
possible for the robot to reach larger work objects. The external axes are usually integrated to the
robots control system and can coordinate its motions with the robot.

Figure 2: Robot on a rail track [6].

Safety arrangement of a robot station is highly important since the robot moves with high speeds and
forces. When the robot is welding during production its therefore forbidden to access the within the
boundaries of the station. By rule, fences are surrounding the area while light beams are equipped at
the entrances which shuts down the robot if someone enters the area during production. Emergency
stop buttons are also placed around different positions so that easily can be stopped if something goes
wrong [4].

2.5 Tolerances and Seam finding/tracking

Robot arc welding requires narrow tolerances of the locations of the joint, and it cannot usually exceed
+/- 0.5 mm [4]. While the robot can repeat its weld to perfection, it cannot adjust its motions and
parameters if the joint is deviated somewhat. For example, if the joint is to some extent a bit wider or
the material slightly crooked, a human can adjust current, weave, travel speed or bend the plates for
compensation to get a good result. One other important thing to think about is that the filler material
regularly is to a certain degree a little bit bent. To not exceed the tolerances, one should therefore
consider the magnitude of the pre-bending of the filler metal and the impact of the stick out when
welding with MIG/MAG.

11
Narrow tolerances lead to high requirements of fixtures and incoming material/components which
are to be welded. The fixtures need to be very stiff and strong enough to withstand forces from
deformations of the work object. Incoming material should include as few shape deviations as possible
and if there are any, these needs to corrected.

In many cases it is possible to use robot welding without any aided-tools for seam finding and
tracking, by keeping the tolerances low. But if the work objects are large and already been processed,
there is a good chance that tolerances are impossible to sustain on a level that robot welding requires.
Seam finding and tracking are methods that can counteract against the problems upcoming from
deviations of tolerances.

In the easier case seam finding is a good solution. Seam finding works in the way that it localizes the
work objects surfaces (in 3 dimensions), for example by electrical contact through the gas nozzle or
by using laser equipment, and then using the information to solve the location of the joint. The start
position is thereby located and the programmed welding motion is updated with new coordinates
according to actual joint position.

Figure 3: Seam finding by using the gas nozzle as a sensor to find the exact position of the joint [4].

Seam tracking allows the robot to continuously update the joint position while welding. This is
particularly good if the joint is long, which can be deviated by deformations while welding. A common
way to this is by using the arc itself as sensor. The arc weaves perpendicular along the seam, and
compares voltage or current between the centreline of the joint and the edges. Another more complex
solution of seam tracking is by using optical devices, which involves cameras and laser [4].

Figure 4: Seam tracking by measuring the differences of arc length when weaving perpendicular to seam [4].

12
2.6 Quality requirements and standards

When welding with robot one can assume that the welding motions are thoroughly reproducible.
Instead it is important that one can control the conditions and keep these constant so that the robot
can weld with exactly the same results over and over again. A basic principle is that the deviations
from the surrounding equipment, fixtures and material needs to be within the tolerances from the
procedure. High quality requirements most often lead to smaller marginal between deviations and
tolerances [4].

Examples of following actions needed to fulfil the quality requirements are

• A suitable creative design and choice of joint

• Using the most favourable weld position

• Weld joint preparation

• Continuous maintenance of the welding equipment

• Use of monitoring systems

• Choose welding equipment that can regulate and hold constant welding parameters

Using visual aids as screens to monitor welding parameters such as current, voltage and wire feeding
can be a good idea in order to easily follow the parameters and see if deviations occur. Connecting
these tools to a computer makes it also easy for documentation and further analysis.

The quality requirements for the welds of a product is based on ISO 3834. ISO 3834 (Quality
requirements for fusion welding of metallic materials) doesn’t solely indicate on the quality properties
of a weld, but also surrounding activities as routines and quality control. The quality requirements are
consisting of three levels, 3834-2 (Comprehensive), -3 (Standard), -4 (Elementary) [7]. Since ISO
3834 is very general and corresponds to the whole quality management and surrounding activities of
welding processes, it will not be handled in this report more thoroughly.

ISO 5817 (Quality levels for imperfections) is however a standard that relates to the quality levels
required from 3834. This standard also consists of three quality levels which is denoted by B, C and
D. Quality level B corresponds to the highest requirements. The standard is used to determine the
acceptance level of discontinuities and shape deviations of welds to define the quality level of a
product, but can also be used as a tool for visual control during the regular production [8]. Even
though the standard is good to use to determine the quality level of a weld, it’s important to
understand that it shouldn’t be used as a general rule to determine the quality of all types of welds.
Deviations of welds can be differently critical between different products, although the weld and
quality level is the same. Customer and product requirements are therefore also important to take in
consideration.

Figure 5: Example of shape deviation (incorrect weld toe), taken from ISO 5817 [8].

13
 Chapter 3

3 Robot welding at Gothenburg workshop and previous

results
This chapter describes the previous use of the robot station in Gothenburg workshop. The chapter
presents a short overview of the robot station itself, and how the welding has been done. It’s also
presenting previous results from the welding. Further and more specific information about the robot
station and component design can be found in the appendix.

3.1 Robot welding station and work object

The robot station consists of a travelling gantry carrying the robot. The gantry can be moved along a
rail, where the different headers of the element can be welded. Within the gantry, there is also a carrier
axis which can move the robot sideways.

The welding consists of joining tubes (magnitude of mm) against headers (5-10 times thicker than
tubes), which are either a pipe with drilled holes or a plate with holes against the tubes. Each element
contains of dozens of tubes, with narrow space between these. Material of the components are either
austenitic stainless steel or duplex.

The robot itself is an ABB robot, and the welding method used is CMT. Additionally, a laser searching
seam finder is installed, assigned to locate the exact positioning of the joint.

3.2 Previous results

Most of the welding problems from before occurred because of the small space between the tubes, but
also because of the thickness difference between tubes and headers. Especially the welding of tubes
against upper header is where the problems occur the most. The thick hole plate requires more heat
input, and it’s difficult to melt the plate locally at the seam since the heat tends to spread out all along
the component. The thin tube in the same time needs only a small amount of heat to melt, and a too
large heat input will lead to a burn through of the tubes. The only technique to make a successful bond
between a thinner material against a thicker, is to concentrate the arc more on the thicker material so
that it gets a higher heat input. The weld must be positioned very accurately. If it’s placed too much
on the hole plate, the weld bead will simply be placed on the hole plate with no bond against the tube.
On the other hand, if the arc is exactly placed at the joint, there is a big chance of burning through the
tube.

Figure 6: Torch direction when welding at the gap between tubes

14
The small space between the tubes gives a huge accessibility problem. When welding around a tube,
the welding torch is supposed to always be perpendicular to curvature. To make it even possible to
weld the area between the tubes, the torch must be pointed much more vertically. The vertical welding
position creates a bonding problem, but also many other weld discontinuities. These will be furtherly
explained in Chapter 4.

The previous results have therefore been very varying, and many of the welds have been insufficient.
NDT is used during the production, and all the welds after the element welding is examined with the
purpose of finding occurring defects. NDT methods used is PT, VT and leak testing (where the element
is pressurised, and a spray is applied on the welds to find leaks). In the worst cases there has been up
to 50% welds of the tubes that has not met the quality requirements, and therefore needed to be
repaired afterwards. Consequences of this have led to extra work, with the costs of productivity,
manpower and expenses. Robot welding of an element has thereby been slowed down, and almost
have had a production rate similar to manual welding of elements. The biggest advantages, i.e.
reliability and productivity, of using an automated welding technique has thereby been lost.

15
 Chapter 4

4 Main challenges and currents state analyse

This chapter presents the current state analyse of the robot welding process. The difficulties from
the robot welding are stated and explained, in consideration to relevant standards. The main
challenges are then formulated of different actions to overcome the problems.

4.1 Defects and discontinuities

Discontinuities are abnormities that occurs various amounts depending on which type and execution.
Discontinuities are defined as defects if the desired functionality of the welded part is reduced. Limits
for quality levels are stated by ISO 5817 [8].

The pictures presented in this section are on test pieces from robot welding. They might not all be
fully representative for when the robot had the most optimized settings, but have been occurring
during the development and one should be aware of these when making improvements of the robot
welding in the future.

• Lack of fusion

One of the main problems of welding the element is lack of fusion in the position of the small space
between the tubes. When the robot starts from the upper position above the tubes, the downward
pointing of the welding torch and long stick out leads to cold metal flows ahead of the arc and not
attaching to the base metal. This have led to lack of fusion and the tubes is therefore not properly
welded into the headers.

Consequences of this is that manual repairing with TIG afterwards have been necessary to meet the
requirements of leak test, quality levels etc. The penalties of this extra repairing have led to additional
labour and in the end, longer production times.

Figure 7: Incomplete weld with lack of fusion, repaired afterwards with TIG

16
  Burning through of tubes
If the weld torch gets too close to the tubes, it easily burns through them. It is therefore very
important that the points programmed during the welding positioning must be exactly done and no
disturbances should occur during the laser searching.
Even though the points are carefully chosen, very often marks on the inside of the tubes is made.
Depending on the quality level this might be seen as a defect (Check ISO 5817).

Figure 8: Tubes affected by too much heat impact

 Crater pipes

Pipes most often occurs in the end position of the weld, and has been a problem when welding with
the robot. It’s caused by metal shrinkage while cooling down, and is due to rapid solidification when
the arc is turned off. Crater pipes are related to poor technique when switching of the arc, impurities
in materials and too much heat input.

Using the slope down function in the CMT is a method to counteract the upcoming of pipes. Crater
pipes is not allowed for quality levels B and C in ISO 5817

Figure 9: Crater pipes at the end of weld bead

17
  Large weld reinforcement
If the weld reinforcement gets to large, it has caused problems of welding the tube besides. When
the reinforcement is large, the arc has had tendencies to “jump back” to the already welded tube and
weld the joint again resulting in missed welds. Acceptance levels are stated by ISO 5817 as
measurements depending on the quality levels.
Large weld reinforcements also give sharp edges

Figure 10: A large weld reinforcement can disturb the welding.

 Sharp corner of weld toe

Sharp corners can lead to cracks if the constructed parts are under fatigue conditions. Limits of angle
are stated by ISO 5817 to responding quality level.

Figure 11: Weld toes having sharp edges against the hole plate.

18
 • Spatter
Spatter is a regular problem of MIG/MAG welding, but shouldn’t occur at all with CMT-welding if
using correct synergic lines and parameters. Spatter tends to land on the tubes as well on the header
and needs to be removed after welding. Spatter doesn't impact on the strength of material, but can
reduce the corrosion resistance. ISO 5817 states that all spatter needs to be removed from pressure
and load carrying parts.
Spatter is related to improper weld gas and flow, technique, welding parameters, process settings
and filler metal

Figure 12: Spatter on the hole plate.

4.2 Laser searching problems

Many of the occurred problems and welding defects has its source from wrong weld positioning due
to laser searching difficulties. It is very important that the components in the upper header is clean
from abnormalities and that the upper header is straightened well enough from the pre-fabrication
welding.

The upper header is especially affected by torsion from the pre-fabrication, which is not as easy to
straighten as for example bending. The torsion in this case then leads to that the ends of the of the
upper header are crooked, and the hole plate is not exactly vertically oriented. The welding positioning
are programmed for an exactly vertical aligned weld, and the laser search does not adjust (rotate) the
welding positioning for a crooked seam. Same problem occurs if the headers are positioned crooked
in the fixtures. Therefore, one has to be very careful to position the components correctly and as well
check that these are good enough to weld.

Figure 13: Laser searching disturbed by spatter, and consequently not welding exactly on the programmed position.

19
4.3 Main challenges

To overcome the weld problems previously assessed, the following main challenges listed below are
carefully considered during the test welding.

• Pre-fabrication of components should be carefully processed, with the special mind on small
tolerances. Post treatment of the fabricated headers are most essential, since impurities and
spatter can disturb the robot welding and laser searching. Exact positioning of the element
components is also very important

• Welding positioning during programming must be very accurate. How the welding torch is
pointed against the seam as well as electrode angle should be considered very well during the
testing.

• Parameter optimization. To get the most optimized welding parameters, CMT fundamentals
and welding technique should be fully understood.

• Lack of fusion in the small space between tubes must be avoided.

• Heat needs be spread out evenly throughout the whole element to avoid burn through of tubes
when welding.

20
 Chapter 5

5 Improvement proposals
This chapter presents proposals how to overcome the previous problems, when taking use the robot
station in Lapua workshop. The improvements proposals are reflected so that they are realistic for
the process and production in the workshop.

5.1 Investment of a new weld torch

A new welding torch is invested to increase the accessibility to get better welding positioning in the
small space between tubes. Fronius CMT Braze+ weld torch is equipped with a narrow gas nozzle that
suits better for welding of the product [9]. The new weld torch is illustrated in Figure 14.

Figure 14: Difference between a "normal" weld nozzle and CMT Brace+ nozzle [9].

The old gas nozzle was cut in the front part to make it possible for welding between tubes. This led to
that the contact tube was exposed and the long distance from the end of the gas nozzle to the contact
tip is giving bad gas protection. The new weld torch also gives a much more focused arc with gas
protection to the exact welding area. Gas flow will be much lower, reducing the consumption of it.
Welding positioning will also be simplified, increasing the accessibility of the torch. Finding a better
welding angle will be easier.

Figure 15: Comparison of the old nozzle (left) and the new welding torch (middle and right) used in the robot.

21
5.2 Root protection flux

The use of root protection is important when welding stainless steels and protects the weld against
defects. According to [10], backing gases provides protection against oxidation, porosity and bad
shape of the weld on the root side. Backing gases are however not suitable for the element welding,
since that would require enormous amounts of gas inside these because of the size.

Root protection flux is then another option that can be used. It doesn’t fulfil as high protection as
backing gas, but still gives much better results than without any protection. Root protection flux is
blended with water or industrial alcohol and applied on the root side of the seam before welding. Root
protective flux also gives good protection against burn through, since its helps spreading the heat
evenly on the root side. One thing to consider though, is that the flux remains after the welding. It’s
therefore important to find out if that would impact the application, for example clogging [11].

Figure 16: TA FLUX, a root protective flux that is blended with industrial alcohol and brushed on the root side [11].

5.3 Short arc length when welding at the small space between tubes
Arc length is a parameter chosen on the power source, and affects the shape of the weld and its
penetration. A long arc length gives warmer and broader weld, while a short arc length gives vice versa
but with a slightly larger weld reinforcement. Using a long arc length makes a problem though when
welding at the specific point between the tubes. Due to the bad welding positioning depending on the
up straight pointing of the weld torch, the arc sometimes hits the seam incorrectly.

For example; when welding the joints on the underside of a tube, a long arc might sometimes hit the
tube lower than it should when starting the arc. This results in an incomplete weld, without
successfully adding a weld at that point leading to leakages in the elements. A solution to avoid this
problem is to use a short arc length instead. The arc should thereby hit the seam more accurately,
leading to fewer defects of lack of fusion. The problem and solution is illustrated in Figure 17.

22
 Figure 17: When welding the lower left joint on the underside of a tube, a long arc might sometimes hit the tube lower
than it should when starting the arc. Using a short arc length, results in welds more accurate at the desired point
between tubes.

5.4 Using slope down at switch off and higher starting current

To avoid the problems of crater pipes, slope down is a method that can be helpful. Slope down is when
the welding current goes from the initial value down to a lower current so that the current doesn’t stop
so abruptly. A smoother end of the weld can therefore be finished since the arc can stay longer at the
end point and fil out the crater better. The slope down parameter value is easily changed on the power
source and is shown as a percentage of the initial current.

The starting of a weld can as well be adjusted through the power source in the same way as the slope
down. A high starting current gives an easier ignition of the welding and a higher wire feed at the start
fills the weld better as well at the beginning.

5.5 Inspection of ingoing components

Since the laser search easily gets disturbed by impurities and dirt from pre-fabrication, it’s important
that the operator or other personnel continuously checks the quality of the in going components used
for welding. It’s also important the operator places the components carefully in the fixtures, so that
the robot can weld correctly without any defects. To accomplish this, there should be well made
procedure to follow how to inspect the components.

5.6 Use of an optimal welding sequence that spreads out heat more evenly

A high heat impact when welding leads to deformations in the steel and residual stresses, which can
cause problem during the robot welding or other manufacturing processes afterwards. A well
thoughtful welding sequence spreads out heat more effectively, leading to that the heat doesn’t
concentrate at one welded area with high impact of residual stresses and deformations.

23
 Chapter 6

6 Results
This chapter presents the results of testing the improvement proposals stated in the previous
chapter. The tests are simulated by using short tubes not connected between from upper to lower
header and the headers are only including the vital components needed for a successful weld
between header and tube.

6.1 Impact of root protective flux

The use of root protective flux shows great results against the problems of burn through mentioned
earlier. A test was made where tubes were weld with and without root protective flux. The test was
made so that the robot was exactly welding the same with identical welding parameters. Duplex steel
was used for both components.

The tube without root protection is clearly affected by oxidation and as well a small dig out is shaped
so that the tube thickness is smaller at that area. The root of the tubes without root protection is also
much darker, which also indicates occurrence of oxidation. Oxidation can lead to that the corrosion
resistance is significantly reduced.

Figure 18: The tubes welded to the right and middle are welded with root protective flux while the left tube is without.

Tubes welded with the root protective flux is considerably less affected by the heat impact. The root
side is somewhat affected, but there is no dig out created by oxidation. Colour of the root side is not
as dark as before, and more evenly spread out. In addition to the root side colour, the acceptance
colour should be checked so that corrosion resistance doesn’t get to low according to the quality
requirements.
The use of root protection also gives a higher allowed span of the high input, so that more heat can
be used on locations where it’s needed or if the heat fluctuates which may can occur due to
disturbances etc.

24
Figure 19: Tube welded left without root protection affected by oxidation which has created a dig out with a dark colour
at the root. Tube to the right is welded with root protection flux and looks visually much better.

6.2 Welding of a complete lower header against tubes

Test welding of tubes against lower header was made to get an impression how the results would be
when welding multiple tubes. Until the test welding of a complete lower header against tubes, only
single or up to three tubes has been welded. That previous test welding was solely to learn the robot
welding, and also to find the most optimal welding positioning and parameters. But to understand
how heat and deformations etc. impacts the welding, a setup of several tubes against a lower header
was made to simulate the welding of elements. The tests were primarily evaluated by a leak test, to
find if there was any leakage at the welds. The test pieces were plugged after the welding, and
thereafter pressurised with air. All the welds were then sprayed with a liquid which bubbles if there
are any leak. A visual inspection was also done to see if there were any other discontinuities. The
material used in these tests was duplex.

The test was made two times, with the first test including welding of 35 tubes and the second test 20
tubes. The two different tests were welded with different welding sequences, trying to find the best
solution. Used welding sequences are shown in Figure 20.

Figure 20: Welding sequence of element including 35 tubes to the left, and the sequence of 20 tubes to the right.

25
The welding sequences are made so that the heat should spread out as much as possible all over the
workpiece. The robot starts welding two tubes at the edge, then jumps to the other edge where it
continues welding two tubes. The following tubes welded are thereafter the ones furthest away from
the already welded tubes, i.e. just in the middle of these. The difference between the welding
sequences of the tests are that the first test is welding every other tube, i.e. having an unwelded tube
between. The other test is welding two tubes exactly next to each other.

Figure 21: The first test containing a lower header with 35 tubes

The results from the first test showed that there were some weld positioning errors, were the robot
was welding slightly aside from the joint. This was due to that the laser search was disturbed by
already existing welds. It was found out that a tube cannot be searched by the laser if a tube on the
right-hand side (viewed from the header) is already welded. The welding sequence used in the first
test cannot therefor be used. The leak test showed that there were 5 leaks in total. All the holes were
located at the tight spot just between tubes.

The other test (2o tubes) showed improved results, since the laser search was modified to only search
for one of the paired tubes, i.e. the robot search for one tube and welds that tube and the one next to
on the right side. Then it continues and follows that procedure on the next pair of tubes, see Figure
20. The second test showed better results than the first one, since all the welds were located correctly.
No leaks were detected in the second test.

However, both test showed other types of discontinuities. Spatter was occurring regularly during
welding. Sharp corners of weld toes can also be a problem depending on the quality level.

Welding of a tube approximately takes a minute. Welding of the first test took 35 minutes while the
second took no longer than 20 minutes.

26
 Figure 22: A hole occurred during the welding precisely at the spot between tubes.

6.3 Welding against upper header

Getting good results of welding tubes against upper header has been significantly more difficult.
Especially the appearance of the welds has been of great problem with many imperfections. Many of
the surface deviations and related discontinuities has not passed the NDT made afterwards and it’s
very probable that the welds wouldn’t pass the requirements from ISO 5817.

The deviations from the pre-fabrication has also been to some extent hard control, and even the
smallest deviations have affected the outcomes. The welding positioning is required to be very precise,
due to heat impact depending on dissimilar component thicknesses, and very small deviances easily
leads to incorrect welds. Other discontinuities such as spatter and sharp corners, as similar to welding
against lower header, is also often recurring.

Figure 23: Welds of tubes against upper header.

27
6.4 Welding of austenitic stainless steel

Welding of austenitic stainless materials was also tested, but only in the earlier stages of optimizing
parameters. The joint consists of mixed metals, since the lower header is always of duplex stainless
steels. The filler material was also changed to austenitic stainless steel (EN 14301), the same material
as tubes. It was also very important to change the pre-set configurations in the power source so that
the material was met to these. It was noticed that the voltage was dropped greatly compared to the
duplex settings, even though the wire feed speed was the same (as mentioned earlier, voltage and
current can’t be changed manually as they are fixed to the wire feed speed depending on the settings).
The parameters presented below are used for the welding shown in Figure 24

Table 1: Welding parameters when welding austenitic stainless steels.

Welding parameters Lower header

Wire feed speed 3.0 – 3.4 𝑚/𝑚𝑖𝑛

Voltage 11.7 - 11.9 𝑉

Electrical current 72 - 77 𝐴

Arc travel speed 10-20 𝑚𝑚/𝑠

Starting current 100-160 %

Starting current duration 0.1 𝑠

Slope down current 40 - 60 %

Slope down duration 0.5 𝑠

During the welding, it was noticed that it went much smoother compared to the previous welding of
duplex steels. The results as well was much better, with the appearance and shape being a lot more
improved. The weld reinforcement was able to be formed quite smaller, more suitable for the joint.
The earlier sharp corners are reduced and spatter is almost eliminated. Crater pipes has occurred
occasionally, but it can probably be solved by working a bit more on the slope down and welding
positioning during the switch off.

Figure 24: Tubes of austenitic stainless steels welded against duplex lower header.

28
 Chapter 7

7 Alternative design for effective production

7.1 Use of a “dig out” around the tubes

To lower the heat impact and to get a gentler joint, a dig out can be milled around the inlets of the
hole plate. The material thickness will thereby be smaller locally at the joints, and as well the angle of
the joint will be larger. The intention is to point the arc at the top of the dig out, were the hole plate
meets the tube and has smaller material thickness locally at that point. The hole plate can then melt
more easily, and create the bond against the tubes. The melted metal is then supposed to fill out the
dig out to compensate the material milled away, and thereby recreate the original material thickness.

Figure 25: A dig out in the hole plate will create a smaller thickness locally at the joint and a larger angle.

Using this type of design, it is important to consider how the laser search is impacted. Since the laser
searches the edges of the hole plate precisely between the tube, the edges can’t be machined at that
spot. Welding positioning is though needed to be reprogrammed, so that the arc hits the joint as
intended. Depth of the dig out and as well the width needs to be considered and tested to get the
best suited. As a suggestion, several prototypes can be manufactured with different dimensions and
tested.

Figure 26: A prototype of the proposed design.

29
 Chapter 8

8 Discussion, conclusions and future work

This chapter includes a discussion concerning the methodology of the project and the results of the
test welding. Conclusion made regarding how well the robot welding is suited for the current
manufacturing is as well taken. It also contains recommendations for future work, i.e. significant
issues that needs to be further investigated and other improvement solutions that would
complement this project for a better robot station.

8.1 Discussion

The results made from the robot welding is based on a welding method (CMT) that has never been
used in Lapua workshop before. Previous experience and knowledge is thereby quite small. To
improve the welding results and increase the probability of achieving the requirements from
standards and quality, a good idea would be to make an investment in basic training and education
for the people involved in the project. A big amount of the time spent in this project has been to
optimize the parameters without really understanding the welding method from the beginning.
Having proper knowledge about the equipment’s (power source etc.) from the start would had
simplified the test welding, since much time went for learning of all the parameter settings and
understand the impact of these.

The robot station itself requires rather higher programming skills if the functionality is desired to be
developed. Programming of the welding positioning however doesn’t require the same knowledges,
where it’s enough with an operator education for running the robot and using the functions. But to
increase the automation and develop additional functions, fundamental coding knowledge is needed.
For example, one can develop a code which can calculate the angular misalignment if the upper header
is crooked due to torsion or bad position of the component. The code could then recalculate the
welding positioning, by using a coordinate transformation, and rotate the programmed weld position
to compensate for the misalignment and to fit the crooked seam. The robot could thereby weld exactly
where the seam is.

The results made during the course of the project have steadily been improving. When the robot
station was moved and mounted in Lapua, most of the previous settings were lost. During the initial
testing, the welding was completely unsatisfying with parameters and positioning misconfigured.
When the CMT welding was starting to be understood, the parameters started to be correct and the
results getting better. The use of root protective flux gave great results of protection against oxidation
and burn through. The new weld nozzle gave further improvements of accessibility, weld precision,
and gas protection. The gas flow was also reduced from 21 𝑙/𝑚𝑖𝑛 down to 7 𝑙/𝑚𝑖𝑛 leading to cost
saving of shielding gasses to one third.

The bonding and lack of fusion problems at the small space between tubes was controlled better by
using a short arc length at that location. The disturbances which could fluctuate the arc was reduced,
and the welding got more accurate.

One occurring problem, which has been existing for the whole project, is the method of searching and
determine the exact position of the seam (i.e. the laser search). Although, the laser search is very vital
to succeed at all with the robot welding, it should be considered if this is the best solution. The existing
method, and the current implementation of it (programming, application etc.), is highly sensitive for
disturbances (deviations, impurities etc.) of the ingoing components. If these problems can be solved
by development of the existing method, either on a software level (programming) or improving/
adding equipment’s in the robot station, or investing in a complete new solution is important to
investigate.

30
8.2 Conclusions

If the main problems of lack of fusion and burn through entirely can be avoided, or that the amount
of these defects occurs below a reasonable acceptance level, the issue of the robot welding will be how
it can fulfil the standards regarding weld qualities and imperfections. Since the product is designed
against the EN 13445 standard, which is connected to EU directive PED, the manufacturing is
mandatory to follow the requirements stated. Concerning the robot welding, all the welds needs
therefore to be fulfilled by the acceptance level stated by ISO 5817. The welds are a subject that will
be inspected and tested by both the customer and third party (i.e. an accredited inspection authority,
such as Dekra or Inspecta, will control that the welding is done as supposed regarding to the standards
and meeting the requirements).

According to [12] part 6.6.3.2, the lowest acceptance level is C referring to ISO 5817. Additional
requirements are removal of stray arc and spatter, and grounding of torn surfaces, chipping- and
grinding marks. However, if the “pressure vessel” is under cyclic loads, fatigue becomes a problem
which will increase the acceptance level to B to meet fatigue class (FAT) requirements. Acceptance
level B will also be applied if parts of the product are subjected to creep. In addition to acceptance
level B, the absence of surface deviations and smooth transitions are vital.

If its determined that the product is due to fatigue and/or creep conditions, and therefore the
requirement of acceptance level B, it will be very demanding to fulfil the high requirements of
imperfections. This is because CMT welding as a method can’t perform high quality welds by the mean
of appearance, in comparison to TIG that is a much smoother and calmer method. The challenge will
on that account be to try to replicate the characteristics of TIG welding, like smooth transitions and
small weld reinforcements.

Concerning productivity aspects, robot welding with CMT can lower welding times up to 4 times
comparing to manual TIG. Shielding gas flow rate is, as mentioned earlier, lower as well. Heat input
of CMT is also quite much lower, leading to smaller deformations, which will reduce the labour of
correcting these in later stages of production.

8.3 Future work

Future work needs still to be concentrated on improving the welds so they can meet ISO 5817
requirements. Regarding ISO 5817, the acceptance level needs to be determined in collaboration with
the accredited inspection authority.

An attempt to improve the seam searching and positioning could for instance be to develop the
programming so that it can make the robot welding to compensate crooked and twisted components.
For example, as mentioned earlier, it might be possible to rotate the welding positions using a code to
replicate the real directions of the two-dimensional plane that the seam (hole plate) constitutes. This
is a solution that adapts mostly for the upper header.

A more advance training of the CMT welding would also be of interest, to find the best synergic lines
(welding properties, parameters) suited. The current synergic lines are probably mismatched and
needs to be updated. The power source should as well be checked through, since its over 10 years old.
The CMT technology has been improved a lot during the years, and a new one could be more proper.

When the welds finally seem to be ok, and test welding of whole elements starts, it would be
appropriate to determine an acceptance level of the highest accepted defects/discontinuities that will
occur during the welding. Welding errors will always be a factor, but needs to be found and repaired,
up to certain amount. For example, 2-3% tubes welded are allowed to have discontinuities. When
performing the testing, one should therefore inspect the tubes and note the errors and keep statistics
of how often these occur.

31
9 References

[1] SkogsSverige, "Sulfatmassafabrikens kemikalieåtervinning," [Online]. Available:

https://www.skogssverige.se/papper/fakta-om/massa-och-
papperstillverkning/sulfatmassafabrikens-kemikalieatervinning. [Accessed 20 March 2018].

[2] K. Weman and H. Norinder, Vanliga svetsmetoder - Metodbeskrivningar och utrustningar,

Kungliga Tekniska Högskolan - Industriell produktion, 2000.

[3] C. Olsson, Konstruktionshandbok för svetsade produkter i stål, 5 ed., Techstrat Publishing,
2014.

[4] K. Weman, Karlebo Svetshandbok, 5 ed., Liber AB, 2013.

[5] ABB, "IRC5 Controller," [Online]. Available: http://new.abb.com/products/3HAC020536-

014/irc5-controller. [Accessed 21 March 2018].

[6] Direct Industry, "KUKA Roboter," [Online]. Available:

http://www.directindustry.com/prod/kuka-roboter-gmbh/product-17587-899277.html.
[Accessed 21 March 2018].

[7] SS EN ISO 3834-1:2005, "Quality requirements for fusion welding of metallic materials - Part
1," Swedish Standards Institute, 2005.

[8] SS-EN ISO 5817:2014, "Welding – Fusion-welded joints in steel, nickel, titanium and their
alloys (beam welding excluded) - Quality levels for imperfections," Swedish Standards
Institute, 2014.

[9] Fronius, "CMT Braze+," [Online]. Available: http://www.fronius.com/en/welding-

technology/our-expertise/welding-processes/cmt-braze. [Accessed 21 March 2018].

[10] Outokumpu Stainless AB, "Welding Handbook," First ed., Outukumpu Oyj, 2010.

[11] TA Chemistry AB, "TA Flux Rotskydd," [Online]. Available:

http://www.tachemistry.com/Rotflux.htm. [Accessed 21 March 2018].

[12] Swedish Standards Institute, "SS-EN 13445-5:2014+C4:2017, Part 5: Inspection and testing,"
Swedish Standards Institute, 2017.

32
www.kth.se