Proceedings of PVP2006-ICPVT-11

Eleventh
2006 ASME Pressure Vessels International
and Piping Division Conference
Conference on
Pressure Vessel Technology-2006
July 23-27, 2006, Vancoucer, BC, Canada
ICPVT-11

PVP2006-ICPVT-11-94027
MJP/06/04

DESIGN BY ANALYSIS - DIRECT ROUTE FOR CASES

WITH PRESSURE AND THERMAL ACTION

Franz Rauscher
Institute for Pressure Vessel & Plant Technology
Vienna University of Technology
Gusshausstr. 30/329, A-1040, Vienna, Austria
Email: f.rauscher+e329@tuwien.ac.at

check (PD-DC) and the Instability design check (I-DC), a

ABSTRACT non-linear analysis based with linear-elastic ideal-plastic
This paper focuses on the usage of Direct Route for constitutive laws is performed.
Design by analysis (DBA), which is included in the new During the development of EN13445-3 [1] and
European standard for unfired pressure vessels (EN 13445- afterwards, numerous demonstration examples were analysed
3 [1]). The Direct Route addresses failure modes directly, and published [3-10]. Guidance on the usage of the method is
having, therefore, advantages in comparison to the traditional given in [3, 9].
method, which is stress categorization. This paper addresses the Direct Route for cases with
Special attention is given to the Progressive Plastic combined pressure and thermal action. In general thermal
Deformation design check (PD-DC) with space and time- actions vary with time and are not in phase with the pressure
dependent temperature distribution and the Fatigue design action - thus resulting in non-proportional loading. The so
check (F-DC) with stress components which do not vary obtained time-dependent temperature distribution causes
simultaneously. temporal and spatial varying material parameters and thermal
As a simple demonstration example, a nozzle with cold stresses. In such cases, progressive plastic deformation and
media injection is used to show how the Direct Route can be fatigue are critical failure modes.
applied in such cases. This example is analyzed with abruptly Addressing the failure modes directly allows actions
changing injection temperature, which makes a transient which are nearer to the limits. Therefore, all the design
thermal analysis necessary. checks, including the Fatigue design check (F-DC), have to
In the case of the Progressive Plastic Deformation design be considered. Especially, in cases with large thermal
check (PD-DC), Melan’s shakedown theorem and cycling of stresses, the Fatigue design check (F-DC) may be the
a Finite Element (FE) model with a linear-elastic ideal-plastic governing design check. Some phenomena in connection
material model are used. In the case of the Fatigue design with fatigue under variable temperature and complex stress
check (F-DC) some problems are discussed: One of them is history – varying temperature, non-proportional loading,
cycle counting in the case of non-proportional loading, time-dependent principal axis – are not well understood. A
another one is the use of structural stresses and stress general approach (e.g. EN13445-3) which is based on
concentration factors. specified material parameters has to be conservative and
relatively simple to apply in the relevant cases. Although the
INTRODUCTION first point should be continuously evaluated, this paper
All design checks required for DBA according to Direct addresses the second point - the application of the prescribed
Route are based on specified material parameters – some are rules (Chapter 18 of EN13445-3 at the current Rev. 14) for
specified in material standards or approvals, others in the cases with combined pressure and thermal actions. Some
design standard. Like the traditional method, which is stress possibilities to overcome problems in the application of the
categorisation [1, 2], this method permits limited plastic given procedure are described.
deformation, and, therefore, sufficient ductility of the
materials is required. EXAMPLE SPECIFICATION
Stress categorisation is based on linear-elastic analyses A simple nozzle in a sphere with cold media injection
and, according to the relevance to failure modes, the stresses was chosen as demonstration example (Figure 1). The
are classified as primary, secondary, or peak stresses. In the example can be modeled by a simple rotational symmetric
case of Direct Route, design checks address the failure modes FE-model, resulting in reasonable computer times for non-
directly. In the case of the Gross Plastic Deformation design linear analyses and making parameter variations simpler.
check (GPD-DC), the Progressive Plastic Deformation design
1 Copyright © 2006 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 06/15/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 The example, with some differences in action history
and dimensions, had originally been discussed in the DBA
Manual [3], and DBA of this example with slow variations of
the temperature field is demonstrated in [9].
The geometry of the model, a spherical end of a vessel
with a nozzle connected to a pipe, is shown in (Figure 1). In
this drawing allowances, which are dimension tolerances and
corrosion allowance, are already considered.
The following operating history is assumed (Figure 2,
Table 2): During startup pressure and temperature are slowly
increased to the operating pressure and operating
temperature. During the operation period one cold media
injection cycle takes place. In case of a cold media injection
cycle the temperature of the fluid in the nozzle (TN) is
assumed to change instantly to the injection temperature
(TNi). Due to the fluid flow in the nozzle the heat transfer in
the nozzle changes to the one for forced convection (h Ni). As
approximation, temperature and heat transfer coefficient in Figure 2: Operation history
the sphere do not change. After some time (here 600s) the Material parameters according to EN13445 and the
cold injection stops – the temperature and the heat transfer material standards are used.
coefficient in the nozzle change back to the original values.
During shutdown, the temperature and the pressure decrease TEMPERATURE DISTRIBUTION
slowly. A FE model was meshed in such a way that the same
mesh could be used for the thermal transient and for the
various static analyses. The time-dependent temperature
distribution was evaluated first, independently from the static
45°
analyses. Afterwards the element type was switched to
structural elements, and the static analyses were performed
with nodal temperatures read from the result file of the
45.
5

Ø134.1 (133.1)
TS h S
R1061.5 (R1060)
thermal analysis. Due to the simple geometry a fine mesh
with rotational symmetric elements (8 nodes) could be used.
R13
For the transient thermal analysis, temperature-
dependent thermal conductivity and specific heat as specified
42.5
TN h N
30
° in Annex O of EN 13445-3 were used as input to the FE
174

program. At the inside surface thermal convection was

modeled. The injection was modeled by a simple step change
13.5 of the bulk temperature and the heat transfer coefficient in
167

the nozzle.
Ø168.3
At the beginning of the cold media injection the material
is cooled from the inside of the nozzle, building a layer with
Figure 1: Geometry of nozzle in sphere large temperature gradient (Figure 3). After the specified
600s of injection, a temperature distribution near to the
Table 1: Material specification stationary one (Figure 4) is reached. During the reheating of
Pos. Nr. Material Standard the nozzle after the injection, the temperature gradients are
1 10CrMo9-10 + NT EN 10028-2 smaller because of the smaller heat transfer coefficient due to
2 11CrMo9-10 +QT EN 10216-2 free convection.
3 P265GH EN 10216-2

Table 2: Pressures and temperatures

Max. allowable PS = 80 bar, TS = 295°C
Pressure/Temperature
Stationary operation: TS = TN = 295°C; p = 80 bar,
hS = hN = 1.16 kW/(m2K)
Cold media injection TSi = 295°C; TNi = 60°C;
p = 80 bar, hSi = 1.16 kW/(m2K);
hNi = 10.8 kW/(m2K)

2 Copyright © 2006 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 06/15/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 A1

Figure 3: Temperature distribution 10s after Figure 5: Pressure only, liner elastic analysis –
injection (transient) Mises’ equivalent stress

A2

A3

Figure 6: Pressure and thermal action 10s after

Figure 4: Stationary temperature distribution during injection, linear-elastic analysis – Mises’ equivalent
injection stress
LINEAR-ELASTIC STRESS DISTRIBUTION
To obtain a first insight, stress distributions from linear-
elastic analyses are presented. In the Figures 5 – 7 Mises’
equivalent stress of the most important operating conditions –
pressure only, 10s after the cold media injection, and 600s
after starting the injection - are plotted. In case of these linear
elastic analyses, the temperature-independent modulus of
elasticity and thermal expansion coefficient, as given in the
section of the PD-DC, were used.
During pressure action only (Figure 5), the maximal
stress occurs at the crotch corner (point A1). At the beginning
of cold media injection, large stresses arise at the inside of
the nozzle (Figure 6). At this condition, at the connection
between the pipe and the nozzle reinforcement, a large stress
concentration takes place (point D). After 600s of cold media Figure 7: Pressure and temperature distribution
injection, the equivalent stress distribution of Figure 7 is after 600s of injection, linear-elastic analysis –
reached. Mises’ equivalent stress
Excluding the stress concentration in Point D, the
following locations with maximal equivalent stresses were GROSS PLASTIC DEFORMATION DESIGN CHECK
identified: When only considering thermal actions, the (GPD-DC)
maximal equivalent stress occurs in point A3 (Figure 6). For the GPD-DC thermal strains need not be considered.
Combined pressure and thermal actions lead to a stress field For the minimum Design Material Parameters (RM) occuring
with the maximum in A2. at the maximum allowable temperature, only the condition
3 Copyright © 2006 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 06/15/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 with PS and constant temperature distribution TS has to be superposition, but showing shakedown by superposition of
considered. In [9] it was shown for this example that the stress fields avoids other problems (see below) and gives a
thermal stresses have practically no influence on the 5% stain better understanding.
limit. The superposition was performed with the post processor
For the GPD-DC, RM (Table 3) has to be divided by the (ANSYS), but it was difficult to compare the equivalent
partial safety factor for the resistance (1.25), and because of stress of the resulting stress field with the temperature-
using Mises’ yield condition, an additional factor of √3/2 is dependent RM. By usage of a macro within the ANSYS
required. program, the quotient of the equivalent stress and the
Using the design model (295°C), Mises’ yield condition temperature-dependent value of RM, which is called
and a partial safety factor for the pressure action of 1.2, a compatibility ratio, was evaluated and plotted
maximum for the admissible pressure of 1.33·PS was (Figures 8 and 9). When using the ANSYS program,
evaluated. incompatibilities in the stress fields can be avoided by
copying, not extrapolating, the stresses from the integration
PROGRESSIVE PLASTIC DEFORMATION DESIGN points to the nodes in the used elastic analyses as well as in
CHECK (PD-DC) the elasto-plastic analysis for the evaluation of the self-
It has to be shown that progressive plastic deformation equilibrium stress field.
does not take place in the design model for the PD-DC. In the residual stress field (Figure 8), the maximum
In the PD-DC, all partial safety factors are equal to one, compatibility ratio is 1 near point A2 (critical point location
and Mises’ yield condition can be used. Therefore, the values see Figure 6). At this point a compatibility ratio of 1 is also
of RM, given in Table 3, are directly used as yield strength reached during the injection (Figure 9), showing that the limit
for the Mises’ yield condition. The operating cycle, as for shakedown to pure elastic behavior is reached in this
specified in Table 2 and Figure 2, is directly applied. region. At point D the compatibility ratio reaches a value of
1.7. This is caused by the stress concentration modeled at the
Table 3: Design material parameters weld root. The region where the value exceeds unity extends
Pos. Material strength parameter RM only over a very small fraction of the cross-section – being
Nr. 20°C 100°C 150°C 200°C 250°C 300°C not relevant for the shakedown of the stress-concentration-
1 290 249 238 232 227 221 free structure.
2 355 323 312 304 296 289
3 265 226 213 192 171 154

According to EN 13445-3 Annex B, it is allowed to use

time-dependent material parameters as well as material
parameters at a so-called reference temperature. When using
a FE-program which enables temperature-dependent material
parameters, like the used ANSYS program, it is no problem
to input the temperature-dependent values. Problems with the
time-dependent modulus of elasticity arise only if Melan’s
shakedown theorem shall be used.
Therefore, for the PD-DC, time-independent modulus of
elasticity (E = 196.6 MPa) and thermal expansion coefficient
(β20°,t = 12,74e-6) were used. Both values were evaluated for
a reference temperature of 226°C, and the thermal expansion Figure 8: Residual stress field – compatibility ratio
coefficient is a secant coefficient relating to 20°C. (RM for 20°C)
There are various methods to show that shakedown
occurs [e.g. 3, 6-8], but not all of them are applicable for this
case. Clear results can be obtained by using Melan’s
shakedown theorem. In this case, to the time-dependent
linear-elastic stress field, a constant self-equilibriating stress
field has to be superposed. If the resulting stress field lies
within the yield surface at all points of the structure, the
structure will shakedown to fully elastic behavior.
Shakedown with alternating plasticity cannot be shown with
this method.
The problem of finding an appropriate self-equilibriating
stress field was solved in this case by simulating a few cycles
with the PD-model (linear-elastic ideal-plastic material). The
stress field evaluated in this simulation after cycling, for the Figure 9: Superposition of linear-elastic stress field
unloaded model, is an appropriate self-equilibriating stress 10s after injection and residual stress field –
field. In this way the elasto-plastic cycling of the model is not compatibility ratio
avoided, which is usually the reason for using this type of
4 Copyright © 2006 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 06/15/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 A second method of showing that progressive plastic
deformation does not occur is to investigate the stress and/or
strain history during cycling of the PD-model. Therefore, a
number of operating cycles were simulated with the PD-
model, and stress and/or strain histories were investigated.
If the structure shakes down, the stresses, the strains, and
the deformations converge to a stationary cycle. If cycling
without plastic deformation is reached, the structure shakes
down to fully elastic behavior.
If the cycle is symmetric, which means that unloading
takes place in the reversed way as loading, no plasticity
during unloading shows that shakedown to elastic behavior a) b)
occurs. Figure 11: Accumulated plastic strain during action
In the considered case the cycle is not symmetric, and, cycle with increased actions a) original RM, b)
therefore, the full cycle has to be considered. reduced RM
To see where plastic deformation occurs during the last
simulated action cycle, the accumulated plastic strain, which 400
is the accumulated plastic equivalent strain increment, is

Circumferential stress
used. The accumulated plastic strains at the beginning of a
200
cycle are subtracted from the ones at the end of the cycle.
Locations where the resulting value is zero had only elastic
deformation during the cycle. Values larger than zero show 0

that plastic deformation took place during the cycle. If the 0,00E+00 2,00E-03 4,00E-03

resulting value is near zero, it is difficult to differentiate -200

between touching the yield surface and plastic deformation.
For the considered cycle, the resulting difference in the -400
accumulated plastic strain (Figure 10) shows that the yield Circumferential strain
surface is touched in the region around the crotch corner.
Considerable plastic deformation takes place at a very small 2,00E-04 4,00E-04
region at point D.
To come nearer to the limit the same model was cycled -280
with the pressure increased to 100 bar and the injection
temperature decreased to 20°C. The accumulated strain -300
during such an operating cycle is considerable larger (Figure Figure 12: Stress vs. strain cycle in point A2
11a) but a considerable fraction of the cross-section shows
elastic deformation only, indicating shakedown. To see the The development of progressive plastic deformation may
failure mechanism, the yield strengths were decreased to the take many cycles and complicated phenomena may occur
one of the model for the GPD-DC, and the previous action [11]. Additionally, very small changes in the plastic strain
cycle was taken. In this case, cyclic plastic straining takes may be important. For example, for 2000 admissible cycles,
place at the whole cross-section (Figure 11b), and the the admissible principal strain of 5% is reached if strain
displacements indicate a failure mechanism as indicated in increments of 5%/2000=0.0025% are accumulated per cycle.
Figure 11b. In the cross-section around point D, some The last two simulations with a few cycles give only an idea
fraction of the cross-section stays elastic indicating on the failure mechanism, a clear answer whether or not
shakedown. shakedown occurs is not given.
With the original action cycle the global structure, with
the original material parameters, reaches a stable cycle very
soon. The stress strain history for the critical circumferential
direction (Figure 12) shows practical no change in the stress-
strain curve after the third cycle. In a very small local region
around point D, the cycling does not become stationary soon.
It is not expected that progressive plastic deformation
develops from such a small region, if the rest of the cross-
section shows elastic behavior.

FATIGUE DESIGN CHECK (F-DC)

Chapter 18 of EN13445-3, which is used for this design
Figure 10: Accumulated plastic strain during one check, uses a peak stress concept with S-N curves based on
cycle tests with polished samples for non-welded regions. For
welded regions, a structural stress concept with S-N curves
based on tests with welded components is applied.
5 Copyright © 2006 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 06/15/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 The model for the F-DC is a simple linear-elastic one. elementary cycles. Usage of the maximum distortion energy
The maximum stress being important, the model has to be (Mises) for the evaluation of the stress ranges is not
mashed for peak stress evaluation. Here the extrapolation of appropriate in connection with critical plane concepts.
the stresses from the integration points, which was switched The critical plane concept is only one option to deal with
off for the PD-DC, was switched on. Another fact is that the this type of stress history, e.g. a method from Wang and
temperature-dependent modulus of elasticity and the Brown described in [12, 15] extends the widely used rain
temperature-dependent thermal expansion coefficients (as flow counting algorithm to work with equivalent stress
specified in Annex O of EN13445-3) lead to larger stresses ranges.
than the constant ones which were used for the PD-DC. The stress variations in Figure 13 are directly used as
These are the reasons why Figure 6 shows considerably equivalent stress range for the evaluation of the fatigue life,
smaller stresses in point A2 than Figure 13 and 14. The which is shown in more detail for cycle 1-2 in Table 4. A
difference between Mises’ and Tresca’s equivalent stress, equivalent linear stress range can be used for the evaluation
which is here near to the possible maximum of 16%, is of the theoretical stress concentration factor, which is the
another reason. basis for the evaluation of the effective stress concentration
At first the critical points for the fatigue analysis have to factor. In point A2 no classical stress concentration occurs.
be identified. It has to be distinguished between non-welded For the thermal stress here, the evaluated theoretical stress
and welded regions. concentration factor is time dependent (Figure 15). Of
Point A2 (for the critical point location see Figure 6) is course, it is possible to evaluate the equivalent linear stress
chosen for demonstrating the fatigue analysis in a non- range, which is also an equivalent value for the considered
welded region: cycle, and to determine a stress concentration factor (Kt) with
The S-N-curves, which are used for the fatigue analysis the equivalent stress range. The resulting Kt = 1.50 leads to
in EN13445-3, are for sinusoidal single-amplitude cycling. Keff = 1.36. The evaluated reduction of the maximum stress
For complex stress histories, elementary cycles, for which the range is questionable, and, therefore, here it is proposed to
allowable number of cycles is evaluated, have to be use the equivalent stress range instead of the equivalent linear
identified. Afterwards, the damage of the different stress range, which is always on the conservative side.
elementary cycles within the complete operating cycle are According to EN13445-3, a correction of the equivalent
summed up (Miner’s rule). In the case of non-proportional stress range is required for equivalent linear stress ranges
loading, like in the considered example, for the prescribed larger than 2*Rp0.2/t*. No correction would be necessary when
reservoir counting algorithm, no appropriate action history is using the equivalent linear stress range. Here the approach
available. Because loading and unloading can take place at chosen above is followed and the equivalent linear stress
different locations of the structure and at different phases of range is set equal to the equivalent stress range, which is also
the action cycle, the cycle counting has to be done at the level conservative here. Two different correction factors are given
of the stresses or strain, not at the level of the actions. The in EN13444-3, one for mechanical stress and one for thermal
problem is now to find an appropriate stress or strain value stress. Because it is not specified, how to use these factors for
for cycle counting. combined thermal and mechanical actions, here the larger
For non-welded regions, selecting the plane with the correction factor, which is the one for mechanical action, is
critical shear stress history and performing cycling counting used, resulting in a plasticity correction factor of 1.114.
based on this shear stress history is in accordance with For the evaluation of the mean stress effect, EN13445-3
critical plane concepts [e.g. 12-14]. uses the stress normal to the critical plane. In the case of
Because the evaluation is performed on the surface of a constant principal stress direction, it is the sum of the two
rotational symmetric model, the principal axes do not rotate principal stresses which built the relevant shear stress.
in this example. The principal stresses are the stress normal Because this stress value may exceed the yield stress,
to the surface (x-direction), the one in longitudinal direction reducing the mean stress when the mean stress plus half of
(y-direction), and the one in circumferential direction (z- the equivalent stress range exceeds the yield stress – as it is
direction). The largest stress is in z-direction and the stress in specified in EN13445-3 - is not really consistent. In the
y-direction has the same sign (Figure 13). Therefore, the considered case with equivalent stress variation larger than
relevant shear stress is σz- σx. Cycle counting with this shear 2*Rp0.2/t*, the simulations performed for the PD-DC showed
stress history (Figure 14) results in one cycle (1-2) having the that the mean stress is really reduced to zero (Figure 12). In
maximum stress variation between 10s after starting of this case where σz-σx built the equivalent stress range and
injection and the unloaded structure. A second small cycle (3- σz+σx built the mean stress, the difference between the shear
4) is identified during reheating after injection. stress, which is limited to the yield stress, and σz+σx is small
In accordance with the critical plane concept, the shear - the radial stress σx is small in comparison to the
stress differences evaluated during cycle counting are directly circumferential stress σz.
used for the F-DC. The evaluation of the maximum The evaluation leads to an allowable number of fatigue
equivalent stress of all possible stress tensor differences cycles 1-2 N = 3527. This is only allowable for an
within the identified cycles, as specified in EN13445-3, environment where conservation of the magnetite protective
would be another possibility. When using the maximum shear layer is not necessary – the stress variation in point A2
stress theory (Tresca) for the evaluation of the equivalent exceeds the one allowed for water conducting parts made
stress ranges, different results would be possible if the from non-austenitic steels.
relevant shear stress direction were not the same for all

6 Copyright © 2006 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 06/15/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Table 4: Fatigue point A2, non-welded, cycle 1-2
800 Injection
Equivalent stress range ∆σeq = 765 MPa
700 Temperature range 20 – 295°C,
600 t*= 227°C
Shutdown Material parameters Rm20° = 540 MPa,
500 Startup
Rp0.2/t* = 300 MPa
Stress [MPa]

400 Plasticity correction ke = 1.114

300 Stress concentration factors Kt = Keff = 1
200
sy Thickness correction en = 44 mm,
fe = 0.964
sz
100 Temperature correction ft*(t*) = 0.92
0 Surface finish correction Rz = 200µm, fs = 0.87
-100
0 1000 2000 3000 4000 5000 Mean stress correction σeq, m = 375 MPa,
Time [s] fm = 1
Allowable number of N = 3527
Figure 13: Stress history at point A2 cycles 1 - 2
900
800 1 Point D was chosen for demonstrating the F-DC in a
1
700 welded region. For the used structural stress concept,
Stress σz - σx [MPa]

600
EN13445-3 allows usage of structural equivalent stress
ranges or structural principal stress ranges. Structural stresses
500
have to be determined by extrapolation. Linearization over
400
the cross-section would partially eliminate the influence of
300 the thermal stress and is, therefore, not appropriate for cases
200 4 3 with thermal stresses anyway [16].
100 Because in welded regions small defects are usually
2
0 present, the phase of crack propagation dominates the fatigue
0 1000 2000 3000 4000 5000 6000 life time [16]. Therefore, the F-DC should be based on an
Time [s] assumption for the direction of most critical initial defects. If
the principal stress directions is time-invariant, defects
Figure 14: Cycle counting at point A2 normal to the critical principal stress are the most critical,
crack initiation in this directions reflects the critical failure
Kt = σz/σlz

3 mechanism. In point D the longitudinal stress σy was

σz identified as the critical principal stress (Figure 16) -
2,5
assuming an initial crack in circumferential direction.
σz
l
Consequently, cycle counting is performed with the
2
stress σy. Two elementary cycles were identified, the thermal
1,5 cycle 1-2 and the startup cycle 3-4. Recognizing that the
cycles are different to the one identified for the non-welded
1 point A2 (Figure 13) clearly shows that cycle counting has to
be performed on the level of stresses or strains, not on the
0 1000 2000 3000 4000
0,5 level of actions.
Time
σl
With a structural principal stress range of 474 MPa, the
0 σz fatigue analysis for the elementary cycle 1-2 results in 2411
admissible cycles (Table 5). As in the non-welded region,
here the plasticity correction factor for mechanical actions
was used too.
This cycle 1-2 shows that the evaluation of the mean
cycle temperature t*, which is used for the evaluation of
Figure 15 Time varying Stress concentration factor material parameters and temperature correction, may be
questionable. If one considers the cycle 1-2 as the full cycle
and cycle 3-4 as a sub-cycle, the minimum temperature in
this full cycle, including shutdown, would be 20°C.
Considering the thermal cycle only, it is 63°C. This results in
less damage for the thermal cycle only than for the whole
cycle. Here in a conservative way the minimum temperature
for the cycle 1-2 is set to 63°C.

7 Copyright © 2006 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 06/15/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 [2] ASME, 2004, “Boiler und Pressure Vessel Code. Section
600
1 Injection Shutdown 1
VIII, Division 2: Alternative Rules,” ASME, New York.
500 Startup [3] The Design-by-Analysis Manual, 1999, European
400
Comission, DG-JRC/IAM Petten – The Netherlands.
[4] Preiss, R., 2000, “CEN's DBA applied to axisymmetrical
300 structures - a storage tank as an example,” ICPVT 9,
200 sy
April 2000, Sidney, Vol.1, pp.405-417.
sz [5] Rauscher, F., 2000, “Design by analysis (DBA) – the
100
3 direct route applied to some nozzles,” ICPVT-9, April
2000, Sidney, Vol. 1, pp.487-495.
4
0 2
0 1000 2000 3000 4000 5000 6000 [6] Preiss, R., 1999, “On the shakedown analysis of nozzles
-100
using elasto-plastic FEA,” International Journal of
Figure 16: Stresses and cycle counting at D1 Pressure Vessels and Piping 76 421-434.
[7] Preiss, R., 2003 “Design-By-Analysis: The Ratcheting
Table 5: F-DC in point D, non-welded, cycle 1-2 Check in EN 13445-3, Annex B,” ICPVT-10, July 2003,
Structural principal stress ∆σstruct = 474 MPa Vienna, pp.79-86.
range (normal to weld) [8] Muscat M., Mackenzie D., Hamilton R.,
Temperature range 62 – 295°C, Makulsawatudom P., 2003 “Elastic Shakedown in
t*= 237°C Pressure Vessel Components under Proportional and
Material parameters Rm20° = 410 MPa, Non-Proportional Loading,” ICPVT-10, July 2003,
Rp0.2/t* = 177 MPa Vienna, pp.51-57.
Plasticity correction ke = 1.137 [9] Zeman J.L., et. al., 2006, „Pressure Vessel Design: The
Thickness correction en = 16 mm, few = 1 Direct Route,” Elsevier.
Temperature correction ft*(t*) = 0.91 [10] Schindler S., 2006, ”Comparison of Design by Analysis
Class of weld detail 63** Methods”, ICPVT-11, July 2006, Vancouver.
Allowable number of N = 2411 [11] Preiss R., 2000, „Ratcheting and shakedown analysis of
cycles 1 - 2 pressure equipment using elasto-plastic Finite-Element-
**) Full penetration butt weld, principal stress normal to Analysis”, Doctoral Thesis, Vienna University of
weld, full penetration assured, testing group 1 o. 2 Technology, Institute of Pressure Vessel and Plant
Technology.
CONCLUSIONS [12] Wang C.H., Brown M.W., 1996, “Multiaxial random
For the usage of Direct Route in cases with combined load fatigue: Life prediction techniques and
thermal and pressure actions the following can be concluded: experiments”, Multiaxial Fatigue and Design, ESIS 21,
The GPD-DC is clearly defined and replaces the non- Pineau A. et al., eds., Mechanical Engineering
unique identification of primary stresses, local primary Publications, London, pp. 513-527.
stresses, and primary bending stresses in stress [13] Han C., Chen X. Kim K.S., 2002, “Evaluation of
categorization. multiaxial fatigue criteria under irregular loading”,
The requirements and models of the PD–DC are also International Journal of Fatigue 24, pp. 913-922.
clearly specified and applicable to such cases. In comparison [14] Zenner H., Heidenreich R., Richter I., 1985,
to the limitation of the secondary stress range in stress Dauerschwingfestigkeit bei nichtsynchroner
categorization, it definitely addresses the failure mode of mehrachsiger Beanspruchung“, Z. Werkstofftechnik 16,
progressive plastic deformation in a conservative way. pp. 101-112.
Shakedown to purely elastic behavior can clearly be shown, [15] Wang C.H. Brown M.W., 1995, “On plastic deformation
e.g. using Melan’s shakedown theorem. Because small strain and fatigue under multiaxial loading”, Nuclear
increments may be important, shakedown analysis by cycling Engineering and Design 162, pp. 75-84.
of the elasto-plastic model may be not that clear. If a large [16] Zeman J.L, 2006, ”Fatigue Design Check of Welded
portion of all the cross-sections shakes down to elastic Regions: A Proposal for Combining EN 13445 and
behavior, progressive plastic deformation is improbable. ASME Division VIII Section 2 Rewrite Concepts”,
More problems arise with the F-DC. Cycle counting with ICPVT-11, July 2006, Vancouver.
artificial stress histories is not really prescribed. Based on the
critical plane theory an appropriate method for many cases
can be found. Also the evaluation of structural stresses and
mean stresses poses sometimes problems or leads to results
which are not plausible. Here the user has to find an
appropriate, conservative approach for the specific case.

REFERENCES
[1] CEN 2002, cur. issue 14 – 2005, EN 13445-3 “Unfired
pressure vessels - Part 3: Design,” European Committee
for Standardization.
8 Copyright © 2006 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 06/15/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use