Redesign of a Steam Strainer

Ann Jannesson
Solid Mechanics

Degree Project
Department of Management and Engineering
LIU-IEI-TEK-A--07/00239SE

Redesign of a Steam Strainer

Ann Jannesson
Solid Mechanics

Degree Project
Department of Management and Engineering
LIU-IEI-TEK-A--07/00239SE

Supervisor:

Michael Blomqvist
DATC, Siemens Industrial Turbomachinery AB

Examiner:

Sren Sjstrm
IEI, Linkping University
Finspng, November 22, 2007

iii

Preface
This report is the result of my masters thesis, the last part of my education to
achieve a Master of Science in Mechanical Engineering at Linkpings
University. The project was done during the fall of 2007 in Finspng, Sweden, at
Siemens Industrial Turbomachinery AB.
In order to complete this thesis I have had a great deal of help from a number of
persons. At the company I would first of all like to thank Mattias Tallberg at
DAC for taking the time to support me while constructing the FE-models. Then I
want to thank my instructor, Michael Blomqvist at DATC, Christer Svensson at
GTSL, Lennart Persson and Bo Andersson at STI and finally Bo Skoog at IEI,
Linkpings University.
This is an official version of the report, intended to be available on the Internet to
anyone with an interest in it.

Finspng 2007-11-22

Ann Jannesson

Abstract
This thesis was done at Siemens Industrial Turbomachinery AB in Finspng.
Placed in the inlet to a steam turbine is a filter, a steam strainer, which separates
particles and larger objects from the steam. These particles and objects will cause
solid particle erosion in the actual turbine if they pass by. The strainer is exposed
to large pressure drops when clogged, i.e., static loads which require a good creep
resistance in the material. The temperature of the steam in the turbines is
increased in order to deliver more energy; todays turbines are dimensioned for
almost 600 C. The material in parameters, such as the strainer, should also be
adjusted to the higher temperatures. Todays temperature is suspected to be the
cause of damage in the strainer because the present material might get brittle at
higher temperatures.
The purpose of the thesis is to find a new material for the strainers and also to
find a new concept for how to manufacture them. There are nine sizes of steam
strainers but only five of them are exposed to the highest temperatures and
pressure drops, which make only these five interesting to examine in this thesis.
The concepts were chosen according to the method of Ulf Liedholm (1999),
Systematic Concept Development. The thesis did not end up with only one
concept because not all possible methods were tested but the suggestions are all
based on a strainer built out of membranes as before. The discussed methods to
join the membranes are EB-welding, laser welding and brazing.
An investigation to find if it was possible to improve the strength of the strainer
by simple design changes and a calculation of what percentage of clogging the
strainer would hold for was also done.
The chosen material was a creep resistant, alloy special steel. Three suggestions
on concepts were presented. The improvements in strength from simple changes
in design were too small and too costly but are enclosed as an appendix in this
report. Calculations on the strength were done without regard taken to fatigue
caused by possible vibrations, so-called high cycle fatigue.
What would be interesting to do as a future work based on this thesis is, of
course, to test the three manufacturing methods and evaluate them thoroughly but
also to discuss other ways of improving the strength through design changes.
These should be done regarding the flow. Also high cycle fatigue should be
considered.

vii

Sammanfattning
Examensarbetet utfrdes i Finspng p Siemens Industrial Turbomachinery AB.
I inloppet till en ngturbin sitter ett filter, en ngsil, som silar bort partiklar och
strre freml frn ngan. Dessa partiklar och freml skadar turbinen genom
erosion om de tillts passera. ngsilen i sig utstts fr stora tryckfall nr den blir
igensatt, vilket krver god krypresistans i materialet. Fr att kunna leverera allt
mer energi utvecklas ngturbiner som kan arbeta vid allt hgre ngtemperaturer.
De som konstrueras idag dimensioneras fr nrmare 600 C. D mste ven
materialet i detaljer, som silen, anpassas. De temperaturer som anvnds idag
misstnks vara en orsak till skador p silarna eftersom nuvarande material kan bli
sprtt d temperaturen stiger.
Syftet med arbetet r att vlja ett nytt material till ngsilarna samt att finna nya
koncept fr hur ngsilen kan tillverkas. Nio storlekar p ngsilar finns men bara
fem av dessa anvnds vid hgsta temperatur och tryck och drfr har enbart dessa
fem anvnts vid berkningar i detta examensarbete.
Koncept valdes enligt Liedholms (1999) metod, Systematisk Konceptutveckling.
Istllet fr ett slutligt koncept lmnas istllet tre frslag. Detta grs d de olika
frslagen inte har testats ordentligt. Samtliga tre frslag r baserade p den typen
av sil uppbyggd av membran som anvnds idag men med nya metoder att
sammanfoga membranen med. Metoderna r EB-svetsning, lasersvetsning och
vakuumldning.
En underskning om det var ekonomiskt rimligt att frbttra hllfastheten genom
enkla designndringar samt en berkning ver hur stor igensttning silen klarar
gjordes.
Materialet som valdes var ett krypresistant, legerat specialstl. De tre frslagen p
koncept lades fram. Designndringarna gav inte det resultat som det hade hoppats
p och var framfrallt fr dyra att genomfra. Hllfasthetsberkningar gjordes
utan hnsyn till utmattning p grund av eventuella vibrationer, s kallad
hgcykelutmattning.
Som framtida arbete med detta examensarbete som sprngbrda rekommenderas i
frsta hand att testa de freslagna metoderna fr sammanfogning av membranen
men ven djupare diskussioner kring hur hllfastheten skulle kunna frbttras
genom designfrndringar borde tas. Dessa skulle kunna genomfras med
avseende p fldet. ven hgcykelutmattning, HCF, borde underskas.

Table of Contents
1

INTRODUCTION...................................................................................1
1.1

THE COMPANY ....................................................................................1

1.1.1 History of Turbine Production in Sweden in General and in

Finspng in Particular ..............................................................................1
1.1.2
1.2

STEAM TURBINE ..................................................................................2

1.2.1

SIT AB in Numbers ......................................................................2

Steam Strainer .............................................................................4

1.3

AIM AND PURPOSE ...............................................................................4

1.4

LIMITATIONS .......................................................................................4

1.5

PRESENT AND FORMER SOLUTIONS .....................................................5

1.5.1

Present Method............................................................................5

1.5.2

Former Solutions Discussed........................................................6

THEORY..................................................................................................7
2.1

THE MATERIAL....................................................................................7

2.2

TO BUILD THE STRAINER .....................................................................7

2.3

THE ENVIRONMENT OF THE STRAINER.................................................8

2.4

SYSTEMATIC CONCEPT DEVELOPMENT ...............................................9

2.5

PRINCIPLES OF THREE POSSIBLE PROCESSES .......................................9

2.5.1

Electronic Beam Welding ..........................................................10

2.5.2

Laser Welding............................................................................10

2.5.3

Brazing.......................................................................................11

METHOD...............................................................................................13
3.1

CHOICE OF CONCEPT .........................................................................13

3.1.1

Phase 1 Establishing the Design Criterion List.....................13

3.1.2

Phase 2 Establishing the Function Analysis ..........................14

3.1.3

Phase 3 Establishing the Concept ..........................................15

3.2

IMPROVING THE STRENGTH ...............................................................16

3.2.1
3.3

Influences on the Strength from Changes in Design .................16

EFFECTS OF CLOGGED STRAINER ......................................................18

3.3.1

Influence of Clogging ................................................................18

3.3.2
4

RESULTS...............................................................................................25
4.1

CHOICE OF MATERIAL .......................................................................25

4.2

CHOICE OF CONCEPT .........................................................................26

4.2.1

Recommendation on Percentage Clogging ...............................22

Shrink-fit Rings..........................................................................27

4.3

CHANGES IN DESIGN..........................................................................28

4.4

DIMENSIONING WITH RESPECT TO LOAD, VIBRATIONS EXCLUDED ...29

4.4.1

Ductile Fracture ........................................................................29

4.4.2

LCF Crack Initiation ..............................................................29

4.4.3

Creep .........................................................................................29

DISCUSSION ........................................................................................31
5.1

MANUFACTURING PROCESS ...............................................................31

5.2

THE FE-MODELS ...............................................................................31

5.3

CHANGES IN DESIGN..........................................................................33

5.4

DIMENSIONING AND HCF ..................................................................33

5.5

SUPPORTING RINGS ...........................................................................33

5.5.1

EB-Welding................................................................................33

5.5.2

Brazing.......................................................................................34

CONCLUSIONS....................................................................................35

RECOMMENDATIONS ON FURTHER WORK ............................37

REFERENCES .............................................................................................39
LITERATURE ................................................................................................39
ELECTRONIC SOURCES .................................................................................39
APPENDIX A METHOD USED TO DEVELOP CONCEPTS.............. I
A.1

SYSTEMATIC CONCEPT DEVELOPMENT ............................................... I

A.1.1

Phase 1 From Problem to Design Criterion List (KKL).......... I

A.1.2

Phase 2 Function Analysis .....................................................IV

A.1.3

Phase 3 Establish Concept .....................................................VI

A.2

CONCEPT EVALUATION ................................................................... VII

Words and abbreviations

EBW and EB-weld ......... Electron Beam Weld
EDM and ED-machined. Electron Discharge Machining
FE ................................... Finite Element, a method used calculating solid
mechanics.
F/M-tree.......................... Function/parameter-tree, this is a visual way of showing
the different solutions, sub-solutions and suggestions on
how to solve the problems in an explicit manner. This
tree is used when developing a new concept.
HCF ................................ High Cycle Fatigue, i.e., fatigue at a great number of
cycles, more than 105 cycles.
HP................................... High Pressure
IP .................................... Intermediate Pressure
KKL................................ Design criterion list, a list where the product
specifications are assembled. This list is used when
developing a new concept.
LCF................................. Low Cycle Fatigue, i.e., fatigue at a small number of
cycles, less than of 103 cycles.
LP ................................... Low Pressure
SIT AB ........................... Siemens Industrial Turbomachinery AB
VAX turbines ................. Axial steam turbine that was introduced in 1982. Based
on the steam moving in the axial direction of the turbine
in contrary to the predecessor, the radial steam turbine,
where the steam is introduced in the centre of the
turbine and travels in the radial direction.

1 Introduction
In this section Siemens Industrial Turbomachinery AB is presented. Also, the
steam turbine and steam strainer are introduced. The aim, purpose and limitations
are noted and finally the present and previous solutions are shortly described.
1.1 The Company
Siemens is an international company with nine different business areas;
Automation and Control, Power, Transportation, Medical, Information and
Communication, Lighting, Financing and Real Estate, Affiliates and finally Other
Activities.
The company where this thesis was done, Siemens Industrial Turbomachinery in
Finspng, 30 km west of Norrkping, is a part of the area Power and produces
gas and steam turbines.
1.1.1 History of Turbine Production in Sweden in General and in
Finspng in Particular
It all began in 1883 when Gustav de Laval took out a patent for his steam turbine
and then in 1893, De Laval ngturbiner (De Laval Steam Turbines) was founded
in Stockholm. Svenska Turbinfabriks AB Ljungstrm, STAL, was founded in 1913
in Finspng by the Ljungstrm brothers who had just constructed the
Ljungstrmsturbin.
STAL was in 1916 bought by what was then called ASEA and consolidated with
AB de Laval ngturbin. In 1959 it became Stal-Laval Turbin AB. Until this date
De Laval and STAL were in a sense competitors even though STAL was
specialized on stationary steam turbines and De Laval on turbines for warships
and fast merchant vessels. After the fusion marine steam turbines were produced.
In 1944 STAL began developing gas turbines. Svenska Flygvapnet (the Swedish
Air Force) was interested to buy three different jet engines and the response from
STAL was the engines Skuten, Dovern and Glan, all named after local lakes. But
when the jet engine Dovern was ready for mass production in 1951 Svenska
Flygvapnet chose one from abroad. Stal-Laval modified the engines and in 1955
the stationary gas turbine GT35 was complete, based on the jet engine Glan.
Over the last 25 years the company has changed its name several times due to
changes of owners but ended in 2005 up as todays Siemens Industrial
Turbomachinery AB (SIT AB).
(SIT AB, 2004)
Figure 1.1 illustrates this in a more explicit manner.

1 Introduction

1913: STAL is
founded by the
Ljungstrm bros.
who have just
invented the
Ljungstrmsturbine.

1880

1900

1883:
Gustav de
Laval took
patent for a
steam
turbine.

1920

1916:
STAL is
bought by
what was
ASEA.

2000: Alstom
Power
Sweden AB.
1984:
ASEA
Stal AB.

1940

1960

2005:
SIT AB.

1980

2000

1959: STALLAVAL
Turbin AB.
1982: VAX was
introduced on
the market

1988: ABB
Stal AB
2003: Demag
Delaval Industrial
Turbomachinery
AB (DDIT AB)

Figure 1.1: Timeline over the history of SIT AB.

1.1.2 SIT AB in Numbers

Year of establishment: 2005
Number of employees at Siemens worldwide (2007): just over 475 000 in 190
countries
Number of employees at SIT AB (2007): about 2200 where 85 is in Trollhttan.
Turnover (2006): about 6 billion SEK
(Siemens AB, 2007)
1.2 Steam Turbine
Steam turbines are used to transform steam into kinetic energy. Steam is heated
and put under high pressure. By letting it pass through the blades of the turbine
the steam is expanded which releases enough kinetic energy to make the blades
turn and produce mechanical energy in a generator.
There are mainly two kinds of steam turbines, axial and radial, Figure 1.2. In
Finspng the radial turbine was developed during the 19th century and was also
first to be developed. Not until in the 1980s the axial turbine took over the
market. The main difference is the direction of the steam flow. In the radial
turbine the steam is brought to the centre to flow out in the radial direction while
in the axial turbine the steam flows parallel to the axle passing the blades on its

1.2 Steam Turbine

way. It is the axial turbine that is dealt with in this report and will be referred to
as turbine in the following texts.

Figure 1.2:
a) Axial turbine (IP/LP)
b) Radial turbine
(Internal material SIT AB, 2007)
There are three types of turbines, high pressure (HP), intermediate pressure (IP)
and low pressure (LP). Most efficient are the last stage blades on the LP and the
IP. But when the steam is too hot and at too high pressure, an HP is connected to
use the steam at that state and then the steam is linked on in a closed circuit to an
IP or LP turbine as in Figure 1.3. In this way the efficiency is raised and more
energy is generated.

Figure 1.3: HP and IP/LP turbines connected in series. The place for the steam
strainer is marked. (Internal material SIT AB, 2007)

1 Introduction
1.2.1 Steam Strainer

The core of this thesis, the steam strainer, is placed together with an emergency
stop valve and a control valve in the pipe before the actual turbine to prevent
particles and larger objects to pass into the turbine and cause possible fatal
damage. The placing of the strainer is marked in Figure 1.3. The steam is very
hot and the strainer is therefore exposed to considerable temperatures. The
strainer is supposed to be cleaned annually since clogging causes pressure drop
and losses in performance. The strainer is exposed to large forces due to large
pressure differences when clogged. The design of the strainer is more described
in section 1.5.1.
1.3 Aim and Purpose
The steam strainer is supposed to last throughout the life of the turbine
(100 000 h) and since it does not, the purpose of this master thesis is to improve
its strength to achieve this request. Above all it is the material that is assumed to
be too brittle at the temperatures in question. But if the material is changed, the
design must be altered since a more temperature endurable material is not
possible to weld using conventional methods.
The aim is to form a new concept for material and design of a steam strainer for
steam axial turbines. The new concept is supposed to deal with temperatures up
to 585 C and pressure drops of at most 165 bars, which is the pressure drop
when the strainer is 100 % clogged. When the pressure exceeds 165 bars the
safety valves open and the pressure never increases more. A clean strainer has a
pressure drop of about 1.42 bars.
1.4 Limitations
The three types; HP, IP and LP are made in a range of different sizes within each
type. The extreme cases in dimension for HP and IP are presented in Table 1.1.
There is no reason to make calculations for the non-extreme versions, thus this
report will focus on the extreme cases. Also, the LP turbines handle much lower
pressures and temperatures and will therefore not be discussed at all.
Table 1.1: The extreme cases that will be managed. The pressure drop is given as
an absolute data.
HP/IP
DN 150

HP/IP
DN 400

Maximum difference in pressure

165 bar (a)

165 bar (a)

Maximum temperature

585 C

585 C

1.5 Present and Former Solutions

In service there are vibrations in the machine caused by the flow of steam, i.e.,
also in the steam strainer. These will be disregarded since they would make the
project too extensive. Instead only low cycle fatigue (LCF) will be assumed to
occur.
1.5 Present and Former Solutions
Different solutions of how to design the strainer in the best way according to
needs and economy have been discussed throughout the years.
1.5.1 Present Method
The design used today is a strainer built from a large number of membranes
milled into their shape. See Figure 1.4. These are placed in a circular shaped
structure by hand and welded and turned in the ends and heat treated still in the
structure. The shape of the channels is developed to lead the steam on its way
into the turbine.

a)

b)

Figure 1.4:
a) The present design of the membranes. Between the channels are the borders,
i.e., the elevations which support the next membrane.
b) The strainer as it is constructed today. The arrows represent the direction of
the steam.
(Internal material SIT AB, 2007)
Different shapes and dimensions of the channels in the strainer have been
examined over the years and the company prefers to keep the dimensions and
angle of the present shape if possible. The area of the channels is at most 0.8 mm
wide and tilted 30 to simplify the flow of the steam. See Figure 1.4a. Wider
channels of 1 mm have been used before but too large particles were then let into
the turbine and a limit of 0.8 mm was decided. The plant in Grlitz, Germany,

1 Introduction

produces a different type of strainer which does not have the 30 angle. This type
of strainer is cheaper to produce but not as efficient. Since the calculations on the
flow are not included in this project, no further research is done in this area and
the guidelines set at SIT AB are followed.
The advantages of the present method using membranes are the cost and that it is
relatively simple to manufacture. The disadvantages are that the material used
does not resist the heat of the steam. And with a change of material it might not
be possible to join the membranes by conventional welding.
1.5.2 Former Solutions Discussed
There are quite a number of solutions that have been discussed in the past. One of
them is the so-called dovetail profile where the ends of the membranes have got a
particular profile over which a ring is placed and squeezed to fit, Figure 1.5. This
method uses the force of friction to keep the membranes in their positions.

Figure 1.5: Example of what a cross section of the dovetail profile joint could
look like. The darker part is the ring.
Another way of fixing the membranes is to weld them together using electronic
beam welding (EB-welding) and a supporting ring. This method is more
described in section 2.5.
One totally different method discussed is to cut the whole strainer out of one
piece. This should be done using laser cutting, water jet cutting or Electron
Discharge Machining (EDM).
At the Siemens site in Grlitz a fourth method is used, namely membrane rings
on top of each other. Every second membrane is wavy, the others are flat and in
the space between the steam is allowed to flow.

Figure 1.6: A close-up on how the membranes in Grlitz strainer are oriented.

Theory

This section covers the theoretical background to the thesis. What material is
used today? How is a strainer built out of the membranes? What does the
environment of the strainer looks like? The method systematic concept
development by Ulf Liedholm (1999) is used to get a number of useable concepts
and the theory is shortly noted in this section. Also the theory behind EB-welding
and brazing is presented.
2.1 The Material
The material used today in the strainer is a ferrite stainless steel. The materials
Youngs modulus is dependent of the temperature and even at 300 C it is as low
as 164 GPa to compare with 217 GPa at 20 C. The steel has a good corrosion
resistance and the scaling temperature is 650 C.
2.2 To Build the Strainer
The strainer is built out of membranes and there are two variants of each
membrane, one plane and one conical, see Figure 2.1. The thickness of the plane
is 2.5 mm and the one of the conical varies linearly between 2.5 and 1.9 mm with
the thickest side out of the strainer, upwards on the figures below. When putting
them together to get the strainer, an algorithm is used in order to get the correct
diameter. The algorithm can for example look like PK or 3(PKK) + PK where P
is the plane and K is the conical.

a)

b)

c)

Figure 2.1: a) A membrane showing where the cross section is cut. b) Cross
sections of the plane membrane and c) the conical membrane.
The dimensions of the strainers are listed in Table 2.1. For the two biggest
strainers the number of channels are given as 8 (4+4) and 10 (5+5) which means

2 Theory

that there is a supporting weld on the middle and therefore the border on the
middle is wider.
Table 2.1: The dimensions of the strainers in the HP turbines.
Number
of
Din Dout
Ain
Aout height thickness tout
(mm)
(mm) channels
Dout / Ain (mm) (mm) (cm^2) (cm^2) (mm)
DN150

0,124

180

220

177

290

133 20 +0,06

1,6

DN200

0,124

230

270

218

371

133 20 +0,06

1,6

DN250

0,083

280

320

386

675

185 20 +0,06

1,6

DN300

0,062

330

370

595

1053

252

20 h11

1,6

8 (4+4)

DN400

0,050

430

470

944

1415

311

20 -0,25

1,2

10 (5+5)

2.3 The Environment of the Strainer

There are two types of strainers, one loose and one integrated. The integrated one
is placed at the inlet of the turbine integrated with the safety valves. The separate
is placed just before the safety valves. The integrated steam strainer is the one
that is exposed to the most loads and will therefore be discussed in this report and
further on will be called the strainer.
The strainer is placed together with an emergency stop valve and a control valve.
These work independently of each other and act as safety devices to stop the
steam flow in case of an accident. The strainer does therefore not need any safety
function. The valves near the strainer require the strainer to be in the shape of a
short pipe. In Figure 2.2 the casing in which the strainer is place together with the
safety valves is pictured. The arrows show the direction of the steam flow. Inside
the casing the strainer is surrounded by the steam which penetrates from all sides,
see Figure 1.4b.

2.4 Systematic Concept Development

a)

b)

Figure 2.2: The casing where the strainer and the valves are placed. The arrows
represent the direction of the steam flow and the strainer is marked with rings.
2.4 Systematic Concept Development
A systematic and structured way of establishing a new concept is used. It is
thoroughly described in Appendix A, but will be very briefly described here also.
The main parts come from Liedholm (1999) which divides the method into three
phases:
o Concept Phase 1:
List the demands and requests in a design criterion list, the KKL. This is
done by critically examining the problem and investigating the state of the
art.
o Concept Phase 2:
Use the KKL to make a function/parameter tree, the F/M tree. This is a tree
structure that describes the functions and sub-solutions of the product.
o Concept Phase 3:
From the previous two phases a number of concepts are generated. Since in
this case there are only three interesting choices of concepts no further
screening was done.
2.5 Principles of Three Possible Processes
Three possible processes, EB-welding, laser welding and brazing, are introduced
here. All of them are favourable to be used together with shrink-fit rings.

10

2 Theory
2.5.1 Electronic Beam Welding

An electron beam is focused on the parts to be joined together. The beam is

accelerated to approximately half the speed of light and melts the material into an
extremely concentrated melt. The method is performed in vacuum to avoid the
electrons being slowed down by molecules in the air and gives a deep penetration
into the material, see Figure 2.3. The advantages compared to conventional
welding are first of all the possibility to weld parts in difficult materials, for
instance, heat resistant materials such as titanium, tungsten and molybdenum, but
also in two different materials. This method leaves very little deformation in the
parts since the area exposed to the beam is small and concentrated. The greatest
disadvantage is the cost; EB-welding is expensive compared to conventional
welding. (Svetskommissionen, 2007)
In this case, there are a few additional disadvantages. The EB-weld on the site in
Finspng is designed to weld the actual turbines and has an oversized chamber.
The welding is done in vacuum which has to be re-established every time the
work piece is changed or replaced. The size of the chamber makes this a time
consuming part of the work.

Figure 2.3: Simplified sketch of EB-weld principles. (Svetskommissionen, 2007)

2.5.2 Laser Welding

Laser welding is, like EB-welding, a method with high energy density. A laser
beam is focused by systems of lenses or mirrors on the details to be welded. It
gives a concentrated heating which heats the surface and leads the heat into the
material. The material vaporizes and leaves a cavity deep inside the material. To
protect the lenses and the weld itself a protective gas is used, usually helium or
argon. The advantages for laser welding are that it is a fast method with a deep
penetration and small deformation. The disadvantage is the need for a high

2.5 Principles of Three Possible Processes

11

exactitude. The method also implies a major investment, but in this case, the
investment is already done. (Svetskommissionen, 2007)
2.5.3 Brazing
The basic principle is that two metal pieces are put together closely and a filler
metal is filling the cavity between them. This is all put in a furnace and heated
during vacuum. The filler metal is melted but the two pieces to be joined are still
solid. In this way an almost invisible joint is produced. The filler metal used at
SIT AB is consequently Ni-based because of its excellent corrosion resistance, its
resistance to high performance temperatures and its high strength. In order to give
a satisfying joint the joint clearance should be within the range of 0.10-0.15 mm.
If the clearance is wider precipitations of intermetallic phases will occur and the
hardness of these precipitations are extremely high and will therefore cause a
very brittle joint. A good way of getting the joint clearance small enough in this
case could be by shrinking on supporting rings, see section 4.2.1. There are a
number of different forms of the filler metal but the interesting forms in this case
are paste, tape or foil, Figure 2.4. The paste is a mixture of metal and binder and
would be applied as in the figure and drawn into the clearance by the capillarity
force. The tape is similar to the paste but with an adhesive layer which makes the
metal easier to apply. The foil has the advantage of being pure metal, no binders,
unfortunately this also makes it hard to handle since the nickel itself has very low
ductility. The brazing cycle is divided into six steps. First of all is a heating ramp
to a temperature below the melting temperature of the filler metal, the preheat
temperature. The second step is when the preheat temperature is reached and kept
constant in order to let the temperature gradients even out, the soaking time. If a
filler metal with binders or solvents is used, a soaking time at a lower temperature
can be used to let the additives evacuate. The third and fourth steps are a short
heating ramp followed by a soaking time at a higher temperature, the brazing
temperature. When the soaking time is elapsed a cooling step follows until a
temperature under the filler metals melting temperature is reached. Then, when
the filler metal is solid, a forced cooling starts. (Internal material, SIT AB)

a)

b)

Figure 2.4: Two ways of applying the filler metal: a) As a tape of foil or b) as a
paste both with shrink-fit rings.

Method

This section covers the method of how the thesis was performed. The concepts in
question and two ways of improving the strength are presented. Also shown are
calculations over how the material in the strainers is affected when the strainers
have been used and are getting clogged.
3.1 Choice of Concept
Liedholms (1999) method, Systematic Concept Development, is used.
3.1.1 Phase 1 Establishing the Design Criterion List
The first step in establishing the concept is to make a design criterion list, a
so-called KKL.
Critical Inspection of the Problem
What is the problem? With higher temperatures the material fails.
Who has the problem? The customers, and if the customers are not satisfied
they will not buy the turbines from SIT AB, and then SIT AB has got the
problem too. With a 5 year warranty SIT AB is keen to have a strainer fulfilling
the demands.
What is the purpose/goal? The product is supposed to manage 100 000 hours in
duty with a maximum temperature at 585 C or 10 000 temperature cycles with a
maximum temperature at 370 C and a minimum temperature at 180 C. At
normal duty the pressure drop is 0.86 % but as a request the product should
manage 100 % clogging, i.e., 165 bars of pressure drop. The product is
recommended to be cleaned once every year.
What side effects should be avoided? The flow through the strainer should not
be affected, i.e., the shape and angle of the channels through the strainer should
be kept the same.
What limits are there for solving the problem? The dimensions of the parts that
are connected to the parts around the turbine should not be changed. The
dimensions where the steam flows are calculated and should not be changed.
Establishing the Design Criterion List KKL
Function

Demand/request

To strain particles and larger objects from the steam

Function determining properties

Roughly the same dimensions as today
Dimensions of part in contact with other detail close by
13

r
d

14

3 Method

Properties for time of use

Manage steam at 585 C for 100 000 hours
Manage 10 000 cycles with Tmax=370 C and Tmin=180 C
Possible to remove the product at steam blow
Possible to change gaskets and other loose parts

d
d
d
d

Manufacturing properties
Produced using the existing machine park/suppliers
Possible to test

r
d

Distributing properties
Manage storage in direct sunlight
Manage outdoor storage in rain and snow
Manage storage at -40 C
Manage storage at 50 C

r
r
r
r

Delivery and planning properties

Will be manufactured in very small batches
Will be manufactured continuously

r
r

Economical properties
Manufacturing cost about the same as before

3.1.2 Phase 2 Establishing the Function Analysis

Starting from the KKL, an F/M-tree will be made.
The black box model
The first step in this phase is to determine the main function in a black box
model. See Figure 3.1.

Impure
steam,
585C

Cleaning
the steam

Clean steam,
585C
Slag

Figure 3.1: The black box model

Main function:

Cleaning the steam.

Operator:

Steam.

Input:

Impure steam, 585 C, containing particles and possibly larger

objects.

3.1 Choice of Concept

Output:

15

Clean steam, 585 C.

Slag.

Establish technical principles

There are a number of possible solutions on how to actually clean the steam and
in this second step they will be examined to find out the best combination. First
of all, three possible ways to separate the particles from the steam are
mechanically with a strainer, with a magnet and by chemically cleaning. The
magnet is found insufficient since there might be non-magnetic particles. The
chemical solution is also disregarded from since there is really no reason to use
chemicals un-called for. Remaining is the mechanical solution, the strainer.
Three possible methods to make a mechanical strainer is to cut it out of one
piece, build it up from membranes or finally use a container filled with some kind
of straining content. For the last suggestion the area of the flow was found to be
too complicated to control and therefore disregarded from. Basically there are
now three ways left to end up with a functional strainer; to make a cylindrical one
out of one piece, to use longitudinal membranes as it is done today and finally to
Figure 3.2.
Clean
steam
Strainer
One
piece
Cylindrical
EDmachining

Water
jet cut

Magnet

Chemical
Filled
container

Membranes

Flat

Laser

Longitudinal
EBwelding

Radial

Dovetail
Brazing
profile

Tangential
Laser
welding

Sand

Steel
wool

Grlitz
variant

Figure 3.2: The F/M-tree with the chosen options in the coloured fields

3.1.3 Phase 3 Establishing the Concept

The final concepts are the following eight:
o A cylindrical strainer ED-machined out of one piece into the desired shape.
o A strainer laser cut from one piece into desired shape.
o A strainer water jet cut from one piece into desired shape.

16

3 Method
o A strainer built out of the same kind of membranes used today but, of
course, in another material and joined by EB-welding.
o A strainer built out of the same kind of membranes used today but, of
course, in another material and joined by a dovetail profile.
o A strainer built out of the same kind of membranes used today but, of
course, in another material and joined by brazing.
o A strainer built out of the same kind of membranes used today but, of
course, in another material and joined by laser welding.
o A strainer built out of membranes shaped as rings placed on top of each
other. These are then held together by some kind of supporting
construction.
3.2 Improving the Strength

There are a number of ways in which the strength of the steam strainer can be
improved. This can be done first of all by changing the material but also by
changing the design to dissolve the stress concentrations. The change of material
is more discussed in section 4.1.
3.2.1 Influences on the Strength from Changes in Design
By small changes in the design major impacts on the strength can be achieved.
Here, the modification is done by adding a small angle to the borders to smoothen
out the stress concentrations, see Figure 4.7 and Figure 4.8.
The FE-models used in this section are based on the longest membrane without a
welding support on the middle and with six channels. This is the one used
making the strainer DN250, see Table 2.1. This one is chosen as a representative
membrane. The algorithm for building a strainer out of these membranes is
PKPK, i.e., every second membrane is plane and every second conical.
To achieve a reasonable model two membranes are modelled, one plane and one
conical, and joined together into one final model, see Figure 3.3. This is not
entirely correct since in reality only the ends are joined by welding or a similar
method and the borders are only leaning on each other. But since the steam flows
radially from the outside and into the strainer the membranes are squeezed
together and will probably act as if they were welded on all contact surfaces. This
is more discussed in section 5.2.
On the ends of the model, in Figure 3.3 marked 1 and 2, boundary conditions in
the second direction are prescribed on the whole line. In the end marked 1 a
boundary condition in the first and third direction is also applied. The one in the
third direction is applied to the whole line while the one in the first direction is
only applied to one node. The boundary condition in the first direction is to
prevent the model from spinning during the calculations. The model is also under

3.2 Improving the Strength

17

a cyclic boundary condition, i.e., the model is restrained in the surfaces where the
next membrane in a real strainer would have been if the whole strainer were
modelled.

Figure 3.3: The unloaded model.

To confirm that the models correspond enough to reality to use them in this
project a four-point bending test was performed and compared to a model loaded
in the same way. The result of the model can be seen in Figure 3.4 and the actual
test in Figure 3.5. The results are close enough to verify the other models.

Figure 3.4: The model of the bending test.

18

3 Method

Figure 3.5: The result of the actual bending test.

3.3 Effects of Clogged Strainer
In this section, first of all clogging is evaluated for different sizes of strainers.
Secondly a recommendation is given on at what percentage clogging the strainers
should be cleaned to keep the life of the turbine as long as possible. There are
three different FE-models in this section and they are based on the strainers
DN200, DN250 and DN400. Dimensions are found in Table 2.1.
3.3.1 Influence of Clogging
The load that the strainer is exposed to is directly proportional to the pressure
drop over the strainer. The pressure drop on the other hand is depending on the
percentage clogging of the strainer. The pressure on the outside of the strainer is
16.5 MPa, which is also the pressure drop if the strainer is 100 % clogged. No
higher pressures will be reached because of safety valves. At a high percentage
clogging the load will however not be the problem since the strainer will
probably be affected more because of vibrations as the flow velocity increases.
But up to what flow velocity is it the load that matters? The equations presented
in this section are gathered from SIT AB Internal material if nothing else is noted.
The pressure drop over the strainer, p, is calculated by
p =

a2
1
1

100[%]
2 V spec p 0

(3-1)

where p is given in percent, a is the flow velocity through the strainer in m/s, p0
is the maximum pressure drop over the strainer and also the pressure outside the
strainer, 16.5 MPa = 16,5105 N/m2, Vspec is the specific volume of steam flowing
through the strainer at p0, here given as 0.0212 m3/kg. The factor is the pressure
loss coefficient and is empirically determined to 2.4 from research in the 1950s.
is more discussed in section 5.4.

3.3 Effects of Clogged Strainer

19

The flow velocity with respect to percentage clogging is calculated by

a=

V&
V&
=
A Ain ,% 10 4

(3-2)

where V& is the volume steam flowing through the strainer in m3/s and Ain,% is the
area of the inlet through the strainer depending on the percentage clogging given
in cm2.
The relation between Ain,% and the inlet through a clean strainer, Ain is

Ain ,% = 1 cl Ain
100

(3-3)

where the cl is the percentage clogging in the strainer.

The inlet Ain for each DN size is listed in Table 3.1. In this table only strainers
used in HP turbines are noted, see section 1.4.
Table 3.1: Table over DN sizes and corresponding inlet through the strainer for
strainers used in HP turbines.
DN [mm] Ain [cm2]
150

177

200

218

250

386

300

595

400

944

A well balanced flow velocity of the steam, according to SIT AB, is a = 50 m/s.
When the strainers are clean the volume, V, is chosen to achieve this speed. Use
(3-2) and data from Table 3.1 with these values:
V& = 50

Ain,%
100 2

(3-4)

Equation (3-2) is plotted in Figure 3.6. The sound velocity in the steam at this
pressure and temperature, which has been calculated by Markus Jcker1,
c = 681 m/s, is also plotted. According to SIT AB, the flow velocity of the steam
will stabilize at this speed because of properties in the steam. This will not be
more discussed in this report.

Markus Jcker, Doctor of Engineering at KTH (Royal Institute of Technology), discussion November 13, 2007.

20

3 Method
700,00

600,00

a [m/s]

500,00

DN200
c [m/s]

400,00

300,00

200,00

100,00

0,00
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89
cl [%]

Figure 3.6: The change of flow speed depending on percentage clogging. The
horizontal line is c, speed of sound in the medium.
The velocity and pressure drop with respect to percentage clogging calculated
with equations (3-1), (3-2) and (3-4) are presented in Table 3.2 where p1 is the
pressure inside the strainer in MPa. The pressure drop is visualized in Figure 3.7
and the ratio between inner and outer pressure in Figure 3.8.
18,000
16,000
14,000

p [%]

12,000
10,000
8,000
6,000
4,000
2,000
0,000
1

11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77
cl [%]

Figure 3.7: How the pressure drop depends on the clogging in the strainer.

3.3 Effects of Clogged Strainer

21

18,000
16,000
14,000

p1 [MPa]

12,000
10,000
8,000
6,000
4,000
2,000
0,000
1

11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77
cl [%]

Figure 3.8: How the ratio between the pressures on the inner and outer side of
the strainer depends on the clogging in the strainer.
Table 3.2: Speed, pressure drop and ratio between the outer and inner pressure
presented with respect to percentage clogging.
cl [%]

a [m/s]

p [%]

p1/p0

50,00

0,86

0,9914

52,63

0,95

0,9905

10

55,56

1,06

0,9894

15

58,82

1,19

0,9881

20

62,50

1,34

0,9866

25

66,67

1,52

0,9848

30

71,43

1,75

0,9825

35

76,92

2,03

0,9797

40

83,33

2,38

0,9762

45

90,91

2,84

0,9716

50

100,00

3,43

0,9657

55

111,11

4,24

0,9576

60

125,00

5,36

0,9464

65

142,86

7,00

0,9300

70

166,67

9,53

0,9047

75

200,00

13,72

0,8628

80

250,00

21,44

0,7856

85

333,33

38,12

0,6188

90

500,00

85,76

0,1424

22

3 Method

3.3.2 Recommendation on Percentage Clogging

Previously the strainers have been guaranteed to manage 100 % clogging. Now,
an alarm at a certain pressure drop has been discussed. In this section a limit for
how much the strainer can take without respect to vibrations will be calculated.
This will be done with respect to ductile fracture, LCF and creep. The els used
are the effective stresses according to von Mises and calculated according to
Dahlberg (2001) as:
evM = x2 + y2 + z2 x y y z z x + 3 xy2 + 3 yz2 + 3 zx2

(3-5)

Ductile Fracture
First of all a safety factor, , is calculated
=

R p 0.2

el

(3-6)

1.5

where Rp0.2 is the yield limit.

LCF Crack Initiation
Regarding a cyclic load it is the stress difference that is important but in this case
the minimum stress min = 0 and therefore the difference in stress is equal to the
maximum stress; = max = el. The criterion is
el R p 0.2 =

R p 0.2
el

(3-7)

LCF Crack Propagation

If there is a crack initiation the safety factor for crack propagation, , is
=

ac
3.0
aN

(3-8)

Where ac is the critical crack depth and aN is the crack depth after N cycles.
Creep
Finally and maybe most important is to evaluate the creep. The safety factor
need to be bigger than 1.
=

where

St

el

1 .0

(3-9)

3.3 Effects of Clogged Strainer

2

S t = min R km ,t ; R k ,1%,t
3

23
(3-10)

Where Rkm,t is the creep rupture strength, i.e., the stress that gives creep rupture
after time t and Rk,1 %,t is creep strain limit, i.e., the stress that gives 1 % creep
strain after time t. Here the time t = 100 000 h is used and all data are at
temperature 585 C.
If this condition is not fulfilled the next step is to check the following condition:
=

St

dim

(3-11)

1 .0

Where

dim = max n ;

( n + b )
1.5

(3-12)

Where n is the mean stress over a load carrying cross section and b is the
bending stress over a load carrying cross section.

Results

In this section the result of the choices and calculations are presented and
somewhat analysed. No final choice of concept is done but the costs for the three
finalists are listed.
4.1 Choice of Material
The material chosen is a creep resistant, alloy special steel. Its main requirement
is the creep resistance under mechanical long-time stressing at temperatures
above 500 C. Looking at the properties at room temperature this material is as
good as, or worse, than other materials discussed during the progress of the
project, but at higher temperatures the development of the chosen material is
more uniform than the other. And since the strainer will be exposed to loads at as
high temperature as 585 C the properties at this level is the most important.
Some examples on the properties of this material compared to other can be seen
in Figure 4.1 and Figure 4.2 where the chosen material is called Mtrl A and the
other, Mtrl B-D, are other materials discussed throughout the project. In Figure
4.3 the elongation limit for the material chosen, three other materials discussed
and the present material is plotted.
Also, while dimensioning the strainer, this material managed quite high loads
relative the other materials evaluated.
350

Creep elongation limit [MPa]

300

250
Mtrl A
Mtrl B
Mtrl C
Mtrl D

200

150

100

50

0
470

480

490

500

510

520

530

540

550

560

570

580

590

600

610

620

Temp [C]

Figure 4.1: The creep elongation limit for the material chosen, Mtrl A, and three
other materials discussed. The materials Mtrl B-D are all creep resistant,
martensitic stainless steel.

25

26

4 Results
350

Creep rupture limit [MPa]

300

250
Mtrl A
Mtrl B
Mtrl C
Mtrl D

200

150

100

50

0
480

490

500

510

520

530

540

550

560

570

580

590

600

610

620

630

640

650

Temperature [C]

Figure 4.2: The creep rupture limit for the material chosen, Mtrl A, and three
other materials discussed. The materials Mtrl B-D are all creep resistant,
martensitic stainless steel.
700

600

Rp0.2 [MPa]

500
Mtrl A
Mtrl B
Mtrl C
Mtrl D
Mtrl E

400

300

200

100

0
0

100

200

300

400

500

600

700

Temperature [C]

Figure 4.3: The elongation limit for the material chosen, Mtrl A, three other
materials discussed and as Mtrl E, the material used today. The materials Mtrl
B-D are all creep resistant, martensitic stainless steel.
4.2 Choice of Concept
In 3.1 the choices of concept have been narrowed down to three versions of
membranes; the EB-welded profile, the laser welded profile and the brazed
profile. They all have advantages and disadvantages such as cost and simplicity

4.2 Choice of Concept

27

that do not make any of them the perfect choice. Four strainers have been
manufactured using EB-welding with satisfying results but tests of the other two
concepts are to prefer to know which one is the most suitable.
4.2.1 Shrink-fit Rings
In order to produce a safe EB-weld, laser weld or brazing joint while
manufacturing the strainer some kind of supporting ring is needed. The ends are
machined after the welding no matter what type of joining process is used.
Weld
To weld the strainer, a ring with a cross-section as the ones in Figure 4.4 could be
used. Figure 4.4a is welded twice in each end of the strainer with the welds
perpendicular to each other, as the arrows in the figure. In this way there is
non-welded material too to secure the joint but as the weld shrinks the material
there are some stresses built into the material in this way. Figure 4.4b solves the
problem with the stresses but leaves no non-welded material to support the load.
The dashed arrow in the figure represents a possible supporting weld which could
be done if needed. The advantage of the second alternative is that it will only
need to pump vacuum in the chamber once, see section 2.5.

a)

b)

Figure 4.4: Two suggestions on supporting ring to be used with EB- and laser
welding. The arrows show where the welding should be placed.

Braze
It is very important that the space between the surfaces to be brazed together is
small, not bigger than 0.1 mm, and in order to get such a fine tolerance shrunk on
rings are excellent. One suggestion on how this ring could be designed is shown
in Figure 4.5. With a v-shaped cavity as shown in the figure the filler metal as a
paste is easily placed. The ring in Figure 4.4a is also possible to use when
brazing, see section 2.5.3.

28

4 Results

Figure 4.5: Suggestion on how a supporting ring could be designed for brazing.
The rings symbolize the filler metal before it is suck into the joint by the
capillarity attraction.
4.3 Changes in Design
Two variants in design are considered. With the original straight borders, as can
be seen in Figure 4.6, it is possible to mill a number of membranes in the same
pass which makes the manufacturing quite cheap. The alternatives, as seen in
Figure 4.7 and Figure 4.8, are more complicated to manufacture. A more detailed
examination of the changes is done but will not be presented here in the official
version of the report. But briefly, they are improvements in strength but not worth
the change in cost.

Figure 4.6: The original design of the membranes.

Figure 4.7: The first variant in design, M1.

4.4 Dimensioning with Respect to Load, Vibrations Excluded

29

Figure 4.8: The second variant in design, M2.

4.4 Dimensioning with Respect to Load, Vibrations Excluded
If the strainers fail at 16.5 MPa pressure drop, it is interesting to know at which
percentage clogging they will pass. The sizes DN200, DN250 and DN400 are
used together with the same materials as shown in Figure 4.1, Figure 4.2 and
Figure 4.3. A lot of the results are censored due to corporate secrecy.
4.4.1 Ductile Fracture
=

R p 0.2

el

1.5

(4-1)

The elongation limit for the interesting materials varies as (4-2) at 585 C
206MPa R p 0.2 280MPa

(4-2)

Due to the secrecy the result will not be presented here.

4.4.2 LCF Crack Initiation
With the explanation in section 3.3.2 the following equations are used.
el R p 0.2 =

R p 0.2
el

(4-3)

Where
el = elmax elmin = el 0 = el

(4-4)

The same el as before is used and therefore the same data as in section 4.4.1 can
be used. Also the same results occur. The levels where crack initiation might
occur will not be used and therefore no calculations on crack propagation are
done. But due to censorship the result will not be presented here.
4.4.3 Creep
=

where

St

el

1 .0

(4-5)

30

4 Results
2

S t = min R km,t ; R k ,1%,t

3

(4-6)

For the different materials the St varies as:

(4-7)

44.67 S t 61.33

All of these St are equal to (2/3)Rkm,100

fulfilled a second condition is used:
=

St

dim

000.

In the case this condition is not

(4-8)

1 .0

Where

dim = max n ;

( n

+ b )

1 .5

(4-9)

where n is the mean stress over a load carrying cross section and b is the
bending stress over a load carrying cross section.
And as before, no results are presented here.

Discussion

The choices made in this thesis are discussed together with the results. Important
is to notice that the results of the dimensioning in section 4.4 are done with no
regard to possible vibration damages.
5.1 Manufacturing Process
The choice of process in not finished in this report. The suggested concepts
should be tested and more studies are required.
The changes in cost for each process are not to be ignored while making the
choice. It has to be considered if maybe an expensive process is to prefer because
of its advantages, such as in-house expertise.
The joints will have different size depending on manufacturing process. The
cross section area of the EB-welding joint is almost the same as the one done
with brazing. But if the filler metal in brazing has leaked into the clearance
between the membranes this joint will be bigger and therefore stiffer. The
membranes should not be able to move that much independently no matter what
size the joint is so it will probably not matter that much in that point of view. But
a stiff joint tends to be more brittle than a flexible one and that could be a
disadvantage. The larger an object is the bigger the probability is that there will
be imperfections in it. With this device it seems to be better with a small joint.
But, not to forget, a small joint might be too fragile.
The supporting welds on the middle that is used today on the larger strainers are
not discussed in the results. There will be a problem if shrink-fit rings are meant
to be used since it will not be possible to shrink them that much. This gives the
brazing method a disadvantage since a supporting ring is crucial. For the methods
with EB- and laser welding it is easier just to make a weld on the middle. The
direction of the weld might have to be angled or the border in the middle has to
be redesigned. An investigation on how much load this middle support is taking
could be done in order to choose an appropriate method.
5.2 The FE-Models
In this thesis models are modelled in I-DEAS (UGS, Corp. 2006) and analysed in
Abaqus (Abaqus, Inc. 2006). They where relatively easy to learn and did not
extend the thesis too much.
A model is a simplified version of reality. The closer it is to the truth, the more
complicated it gets and complicated models are time consuming both regarding
computation time and modelling time. The models used in this thesis are
simplified mostly regarding load, support and joints.

31

32

5 Discussion

First of all, the load itself is simplified. Instead of modelling a flow and a
pressure drop a uniformed load is used. This was done because it would be too
complicated to model the flow. Secondly; to model the load the end lines of the
channels are used, see Figure 5.1 with the load on the lines marked A. In this
particular model there are four lines on each membrane, therefore a fourth of
each membranes total load is put on each line. This simplification was done
because there are no suitable surfaces to load since the top of the model is
rounded. Because of these two simplifications no reactions cased by the flow of
steam are examined. And as has been remarked before, no vibrations are taken
into account and therefore the limits are a little high.
The supports are done as boundary conditions in two lines, see Figure 5.1 marked
B. The real strainer is hanging together with the valve cover in the casing which
can be seen with a closer look at Figure 2.2b. The bottom end of the strainer has
space to increase in length. Also, the casing is shaped to fit the strainer and will
therefore support it in the radial direction for a short distance in the longitudinal
direction. This is not taken into account at all in the models.
The membranes in the models are done as if they were stiffly joined in all contact
surfaces. This is compared with reality where most of the contact surfaces are
only supporting each other. But when the steam is flowing through the strainer it
is squeezing the membranes together because the strainer is circular and some of
the membranes are conical. And therefore this simplification is not unjustified.
Although it is hard to see what difference the supporting weld on the middle of
the two larger strainers did.

Figure 5.1: The model of the DN200 strainer in I-DEAS (UGS, Corp. 2006)

5.3 Changes in Design

33

5.3 Changes in Design

Since the surfaces of the channels are angled, membranes are milled with a shank
end mill after being arranged on a plate with holes for each membrane. Then they
are all milled in one pass. This worsens the possibilities to alter the design as
suggested in section 4.3. This far the modified versions have only been
investigated with a distributed load on the top edge and it is not known how the
membranes would react if loaded with the actual flowing steam. Probably the
second modified version, M2, will have some problems with the quite thin and
sharp edges on the bottom side.
Other changes in design could be discussed to improve the strength. Maybe the
radius between the surface in the channels and the edges could be enlarged.
5.4 Dimensioning and HCF
As been pointed out before, vibrations and any other kind of HCF is disregarded
from in the calculations in this report and therefore the limits in section 4.4 are
probably to be reduced. The material will hold for these loads if there are no
vibrations but when the flow velocity of the steam increases there will probably
be some sort of self excitation.
The factor used in Equation (3-1) in section 3.3.1 is a reason for discussion. The
factor is based on experiments in the 1950s and has been used for dimensioning
for years. But in this case it might not be totally correct. It is developed to work
with increase in flow velocity but because of increasing pressure and not because
of decreasing flow area. But to find an appropriate factor for this case would be
another thesis. And all the same, it has been used in this thesis because of lack of
options and that it will probably not make that much of a difference.
5.5 Supporting Rings
In the discussed concepts supporting rings are included but with various cross
sections, see Figure 5.2.
5.5.1 EB-Welding
It is not safe to EB-weld without a supporting ring since there is no extra material
added and if the membranes are not as closely packed as they should be there will
be a crack in the weld. Two ways of designing the ring for EB-welding are shown
in Figure 5.2 a and b.
The first, a, is the one method that has been tested with EB-welding. It is also the
one that at a first glance should have the best strength since the weld will not be
straight through the membrane. It is hard to do any computations for the strength
of the weld because it can vary a lot depending on a number of parameters such
as material and cleanliness but the untreated material act like a safety device. But
what have to be considered is that the part shrinks in the weld and two welds

34

5 Discussion

perpendicular to each other might leave the part with built-in stresses. Also this
type of supporting ring requires the work piece to be changed during the welding
cycle which extend the machine time, see section 2.5.1.
The second suggestion on the ring, b, does not have the problem with built-in
stresses since the weld is straight through the membrane but that is also a
disadvantage. As said before it is very hard to compute the strength of the weld
and in this case there is no non-welded material to trust. But compared to the
previous suggestion the largest advantage is that the work piece does not have to
be changed and the machine time will be shorter and therefore the cost would
decrease. The edge on the inside of the strainer will be machined away after the
welding. Also an extra supporting weld could be done perpendicular to the other,
although this would require the same change of work piece as for suggestion a.
5.5.2 Brazing
When brazing a very small clearance where the filler metal goes is important. A
supporting ring shrunk on is an excellent way of getting this small clearance.
Two of the supporting rings in Figure 5.2 are possible to use when brazing, a and
c.
The first suggestion requires filler metal in the form of tape or foil. The filler
metal is placed around the strainer before the ring is shrunk on. The
disadvantages are the cost and that the foil and tape are difficult to work with, see
section 2.5.3.
The third ring, Figure 5.2c, is designed to be used together with filler metal in the
form of a paste. There are two tracks on each end of the membrane which follows
the ring around the strainer. The paste is placed in these tracks and when melted
is it sucked in between the ring and the membranes by the capillarity force. As
long as the soaking is good and the clearance is small this joint will allegedly
work well.

a)

b)

c)

Figure 5.2: Three versions of possible supporting rings depending on choice of

joining process.
a) For EB-welding or brazing with tape or foil b) For EB-welding c) For brazing
with paste.

Conclusions

A new material is chosen. It has average properties at lower temperatures but

good properties at temperatures around 600 C which is what it is requested for.
No final concept is chosen but three suggestions are made. All of them are based
on the membrane system used today. The suggestions are EB-welding, laser
welding and brazing. Tests have been done on the EB-welding method but no
choice should be made before more tests are performed. EB-welding is expensive
but the cost depends on the machine time which could be shortened by different
designs of the supporting ring.
Different sizes of the strainer can handle loads of different percentage of
clogging. These calculations are done without any regard to possible vibrations
and as can be seen in section 5.4 there is an uncertainty of the factor .

35

Recommendations on further work

First of all, the investigation of the most appropriate joining concept should be
finished.
Secondly, a deeper examination on the strength improvements from design
changes could be done, not only from a solid mechanics point of view but also
fluid mechanics. How does the flow change because of the shape of the channels?

37

References
Literature
Dahlberg, Tore (2001), Teknisk hllfasthetslra, Studentlitteratur, Lund. ISBN:
91 44 01920 3
Liedholm, Ulf (1999), Systematisk konceptutveckling, Institutionen fr
konstruktions- och Produktionsteknik, Linkpings Universitet, Linkping
Roozenburg, N. F. M. and Eekels, J. (1991) Product design: Fundamentals and
Methods, John Wiley & Sons Ltd, West Sussex, England. ISBN: 0 471
94351 7
Electronic Sources
Abaqus, Inc., ( 2006), Abaqus/Viewer Version 6.6-5
Siemens Industrial Turbomachinery AB (2004), DDIT AB - Historia,
<http://www.sit-ab.se/en/s_nav53.html> (Collected 2007-07-06)
Siemens AB (2007), Siemens Sverige - Fretagsfakta, <http://www.siemens.com/
index.jsp?sdc_p=c174l17mn1148022o1148022ps6u>
(Collected 2007-07-06)
Svetskommisionen (2007), Svetskommisionen - Metoder <http://www.svets.se/te
kniskinfo/svetsning/metoder.4.ec944110677af1e8380009688.html>
(Collected 2007-10-24)
UGS Corp., ( 2006), I-DEAS 12 NX Series m4

39

Appendix A
A.1

Method Used to Develop Concepts

Systematic Concept Development

Liedholm (1999) divides the development of the concepts into three phases. In
short terms; the first provides a specification of the product properties from the
main problem; in the second phase the decision of what the product is supposed
to do and to produce is made and a function/parameter-tree (F/M-tree) is
established; and finally in the third phase parameters are chosen from the
F/M-tree and combined into sub-solutions. From theses combinations the final
concepts are chosen. (Liedholm 1999)
In his book Liedholm (1999) stresses that models are only simplifications of the
reality. Simple models are easy to understand but far from the truth, complex
models are closer to the truth but harder to understand. In the end there has to be
some kind of a balance between these two.
A.1.1 Phase 1 From Problem to Design Criterion List (KKL)
From the problem the product specification, including its properties, is collected.
It is compiled in a design criterion list (the KKL) where the properties and the
purpose of the product are defined. In the KKL the ground rules for the
development process are given and it is supposed to be useful at evaluations. In
the beginning the KKL will be vague because of the lack of knowledge but will
be more specified as the project moves on. The KKL is made with the help of
four steps described in Figure A. 1. In the first step the problem is carefully
reviewed by asking some simple questions.
- What is the problem? Formulated to be possible to solve.
- Who has the problem? One single person or a whole group?
- What is the purpose/goal? What does the one who has the problem
want?
- What side effects should be avoided?
- What limits are there for solving the problem? Supplies, time,
staff, etc.
(Liedholm, 1999, pp. 7-8 and Roozenburg & Eekels, 1991, pp. 132-136)
The second step is to do a background research to check the state of the art.
Maybe a similar problem has been solved before. In that case, how has that been
done and at what cost? This kind of information might be found in patents, with
competitors and in literature. It can also be useful to look for similar problems in
totally different areas. As a third step, the project should be evaluated to be sure it
is technically and economically viable. This check is not only supposed to be
done in this third step but also as a regularly feature during the projects course.
I

II

Appendix A Method Used to Develop Concepts

In case more knowledge or staffs is needed, an arrangement to acquire it will be

formed. The fourth and final step is to actually establish the KKL. In the
beginning of the project the KKL should be independent from the solution but
will turn dependent as the project proceeds and the new knowledge is added to
the KKL. (Liedholm, 1999)
Description of the problem
1.1 Critical inspection of the problem
What is the problem? Who has the problem? What is
the purpose? What side effects should be avoided?
What are the limits?
1.2 Examine state of the art
Search for patents, analyze the competitors, etc.
1.3 Examine technical/economical viability
Viable/non-viable
1.4 Establish the design criterion list
Formulate the properties correctly
List demands and requests

Function analysis
Figure A. 1: Concept phase 1, from problem to design criterion list, KKL
(Liedholm, 1999, p. 7)
There are a few things to consider while formulating the properties in order to
receive the best result in the end.
- Comparable/incomparable properties Choose comparable
properties that are possible to rank. A property is incomparable if the
only conclusion to be drawn is if it is fulfilled or not.
- Demands/requests There should be a distinct line between demands
and requests. Demands have to be fulfilled, if they are not, the solution
is not an option. Requests on the other hand are properties that should
be fulfilled if possible. Often, all the requests are not fulfilled by the
best solution.
- Standard Sometimes solutions are set by law or by standards within
the business area.

A.1 Systematic Concept Development

III

- Independent of the solution It is important to keep a clear mind and

not be stuck with what is feasible, in that case possible solutions might
be overlooked.
- Functional/measurable When designing a product the functions and
performance must be known. An objective judgment if one option
fulfils the properties at all or better than another is desirable. The
properties should be measurable in order to be objective.
- Non-redundant Product properties should be formulated so that
important goals, viewpoints or properties are not covered by several
product properties in the KKL. Otherwise this might affect the
evaluation in the end.
(Liedholm, 1999, pp. 9-12)
It is very important to fit all the product properties in the KKL and after it is
established all the properties are checked to be well formulated and the list itself
is checked. There are a number of checklists available to establish the KKL and
the one below is just one example.
1.
2.

3.

4.

5.

6.

Function
What is the purpose of the product, what is its task?
Etc.
Function determining properties
What performance should the product have?
Is the manufacturing, assembly, distribution or use going to have
an influence on the size or weight of the product?
What dimensions should the product have?
Etc.
Properties for time of use
In what environment will the product be used?
What is the products life?
Is maintenance necessary and possible?
Etc.
Manufacturing properties
Will the product be produced with the existing machine park?
Will it be economically profitable to invest in new machines?
Should the manufacturing be done at another factory?
Etc.
Distributing properties
Will the product be stored, and in that case, what demands are
put on the product?
Etc.
Delivery and planning properties

IV

Appendix A Method Used to Develop Concepts

7.
8.
9.

Safety/ergonomics properties
Aesthetics properties
Legal properties
What standards does the company have to follow?
Will the product imply patent infringement, and how will this be
avoided?
10. Economical properties
What cost of manufacturing is allowed?
11. Scrapping and recycling properties
12. Ecological properties
(Liedholm, 1999, pp. 10-12)
A.1.2 Phase 2 Function Analysis
In the second phase the products functions are listed together with the
parameters necessary to realize them. The purpose is to make it clear what the
product is supposed to do, find its functions and establish several different
parameters/principle. This phase is also divided into four steps, presented in
Figure A. 2. (Liedholm, 1999, p. 13)
First of all a black box is established, i.e., the main function with in- and output
are defined. Also, an operator, a main function, input and output will be chosen.
The operator is what transforms during the process. In this step the black box
should be abstract and not showing anything about the solution. Secondly, all
possible solutions to perform the main function are listed, those are the technical
principles. They are established by three steps: To generate the technical
principles, to evaluate them and to control the choice of operator. Most important
is to remember, do not make it harder than it has to be! As the third step
transformation systems are established, i.e., the transformations performed are
mapped, then sorted after preparing-, performing- and completion phase. Finally
they are ranked according to the order they are performed. The fourth and final
step is to establish the function/parameter-tree, the F/M-tree, which is to show the
different solutions, sub-solutions and proposals on how to solve the problems in
an explicit manner. The function describes the systems task while the parameters
describe how the task is to be performed. The parameters are searched through
ide generation and every new parameter forms its own new branch in the F/Mtree. A disadvantage with the F/M-tree is the size of it, it often tends to get too
big and then it might be a good idea to list the lowest level in the tree in a
morphological matrix instead, see Table A. 1 (Liedholm, 1999, pp. 13-20)

A.1 Systematic Concept Development

Construction criterion list

2.1 Establish black-box model

Input

Main function
(purpose)

Output

2.2 Produce technical principles (main parameters)

Generate, evaluate and choose
Technical principle 1

Technical principle 2

Technical principle 3

2.3 Establish transformation systems

2.4 Establish function/parameter-tree

Main function
Technical
principle 1

Technical
principle 2

Technical
principle 3

Establish concept
Figure A. 2: The four steps in the second phase, the function analysis
(Liedholm, 1999, p. 13)
Table A. 1: Example of a morphological matrix over how to boil potatoes
Functions

Parameters

Peel the potatoes

Potato
peeler

Knife

Brush

Heat the water

and boil the
potatoes

Stove

Boiler

Microwave

Strain the water

from the potatoes

Colander Lid

Potato peeler for

restaurant use

VI

Appendix A Method Used to Develop Concepts

A.1.3 Phase 3 Establish Concept

The object with this final phase is to establish the concepts that will make out the
foundation of the continuous work. This third phase is also divided into four steps
starting with the F/M-tree, see Figure A. 3. (Liedholm, 1999, p. 21)
Function analysis
3.1 Choose parameters/create concept
- rank parameters and
functions
- investigate compatibility

3.2 Inspect and improve the concepts

- Describe how the concepts work
- List advantages and disadvantages
- Investigate difficulties and possibilities

3.3 Evaluate/choose concept

- Are the demands fulfilled according to the KKL?
- Compare the concepts with each other
- Pick a few for further development

3.4 Correct and complement the KKL,

plan further work

Preliminary layout
Figure A. 3: The four steps in the third concept phase, establishing the concept
(Liedholm, 1999, p. 21)
First of all parameters have to be chosen and concepts have to be created. But
there are two issues: How to limit the number of solutions and how to create
suitable arrangement. It is very time consuming to examine all theoretically
possible solutions and to limit these there are some possible strategies, Liedholm
uses two of them. The first is to divide the parameters into three groups: good,
approved and bad. The second is to group the functions with reference to
importance. These strategies can be used one by one or together according to the
situation. It is important to use compatible parameters and functions, i.e., if the
parameter electrical motor is used in the function transform energy then the
parameter diesel tank can not be used with the function store energy. A fairly

A.2 Concept Evaluation

VII

large number of concepts should be generated at this point, at least 10-20 of

them. Now the concepts are to be inspected and improved. A way of
understanding them is to describe their function in words. List the advantages and
disadvantages. Most important are the functional properties but also possible
future difficulties and possibilities. Now the evaluation and choice of concept has
to be done. First of all, is the KKL fulfilled? Secondly, compare the remaining
concepts in a so called evaluation matrix. This is a matrix where the properties
are weighted according to importance. This matrix is described in the next
passage, A.2 Concept Evaluation. Remember that this is only a guide, and the
result is not obviously the best. Finally a few concepts are through to further
development. At last, the final step: To correct and complement the KKL and
plan further work. Investigate the KKL and make a few decisions: Should new
properties be added? Should properties be adjusted or made more detailed?
Should properties be cancelled? By now the KKL is probably dependent on the
solution, but remember to keep the old version independent of the solution. A few
concepts are now considered possible and further work can be planned. What
resources are available; staff, time economic? Establish a long-term plan, e.g. in a
Gantt diagram. Also establish a short-term plan for critical tasks such as; What
parameters are most critical? In what order should the sub-systems be developed,
sequentially and/or parallel? Which are the fundamental dimensions and
performance? The long-term plan is to be adjusted as the project moves on, a
short-term plan is to be established e.g. every week. (Liedholm, 1999, p. 22-28)
A.2

Concept Evaluation

There are a number of ways to evaluate the remaining concepts in a systematic

manner but since there are only a few left in this case they are all taken into
account in the further analyses.