Professional Documents
Culture Documents
PVP2006-ICPVT-11-94028: Comparison of Design by Analysis Methods
PVP2006-ICPVT-11-94028: Comparison of Design by Analysis Methods
PVP2006-ICPVT-11-94028
MJP/06/05
Sebastian Schindler∗
TUV Austria
Krugerstrasse 16
A-1015 Vienna, Austria
Email: seb@tuv.at
S1el
S1pl
cycles 9–20
σ1
σ2
S2pl S2el
Direct Route
Figure 4. Example 1: Stress Categorisation: membrane plus bending
GPD-DC: The check was carried out using Mises yield con-
stresses at 1 MPa internal pressure
dition with reduced design yield stress. The design√ material
strength parameter is given by: RMd = R p1.0,180◦C · ( 3/2)/γR =
138.5 MPa, with partial safety factor for the material γR = 1.25 to the limit of 5% gave a maximum allowable cycle number of
according to EN 13445-3 Annex B Table B.8-2. The Finite El- 1440. But a further (more time consuming) cyclic analysis up
ement Analysis showed that the maximum absolute value of the to 20 full cycles showed that the structure shakes down to alter-
principal structural strain reached the 5% limit at a pressure of nating plasticity after approximately 10 cycles. Figure 2 shows
0.451 MPa. According to EN 13445-3 Annex B Table B.8-1 the stress path in a deviatoric map [5] of the critical node, with
the partial safety factor for pressure (without natural limit) γP the typical trajectory for alternating plasticity which is symmet-
is 1.2. Therefore, the maximum allowable pressure according to ric with regard to the origin [6]. Therefore a fatigue analysis has
the GPD design check is PGPD = 0.451/γP = 0.375 MPa. to be performed to obtain the allowable number of cycles. There-
PD-DC: A detailed analysis, using Melan’s shakedown the- fore, PAP = 0.375 MPa results.
orem with RMd = 200 MPa and all partial safety factors equal For more details of the GDP- and PD-DC see [3, 4].
to 1 according to EN 13445-3 Annex B, rendered a shakedown
limit pressure of PSD = 0.289 MPa. Since the limit pressure ac-
cording to the GPD-check is larger than the one resulting from Stress Categorisation Route
the PD-check further investigation of the cyclic behaviour of the The design stress according to EN 13445-3 Clause 6 is given
structure, with a pressure cycling between a minimum value of by fd,180◦C = 133.3 MPa for an austenitic material with a rup-
0 MPa and a maximum value of 0.375 MPa was performed. In ture elongation larger 35%. The evaluation of the membrane and
the first step 4 full action cycles were calculated. An extrapo- membrane plus bending stresses can be performed very easily
lation of the maximum principle strain of the determining node with shell elements. Stresses results in the middle plane of the
6 (3 + ν) p · R2
σ plate (p = 1.6 MPa) = ± = ±47.6 MPa. (1)
16 h2 EXAMPLE 3: AIR COOLER HEADER
The geometry of the air cooler header with rectangular cross
If the bending stress at the connection is treated as secondary, section is shown in Fig. 9. Neglecting the beneficial influence of
then the bending stress calculated above, in the center of the sim- the end plates and also the influence of nozzles in the short side
ply supported plate is governing the maximum allowable pres- of the frame, only a slice of the middle portion has been mod-
sure: elled as shown in the side view in Fig. 9. The tubes are modelled
to include the stiffening influence on the tube side plate, but their
1.5 · 163.3 · 1.6 admissibility according to the design checks is not further inves-
Pm + Pb ≤ 1.5 · fd,End ⇒ PCAT 1 = = 8.2 MPa. tigated in this paper. It has to be checked whether the required
47.6
(2) maximum pressure of 7.2 MPa at a temperature of T=120°C is
0.6
0.4
0.2
0
0 4.32 8.64
Pressure (MPa)
Direct Route
GPD-DC: The check was carried out using Mises yield tial safety factors equal to 1 and Mises yield condition has been
condition with reduced design yield stress. The design mate- calculated. According to EN 13445-3 Annex B Section 8.3.4
rial
√
strength parameters are given by: RMd,plate = √R p02,plate · for the design material parameter RMd the value of the 0.2% -
( 3/2)/γR = 205.1 MPa and RMd,tube = R p02,tube · ( 3/2)/γR =
proof stress (for ferritic material) at a temperature equal trE =
126.1 MPa, with partial safety factor for the material γR = 1.25 0.25tmin + 0.75tmax = 95◦C may be used. Figure 11 shows the
according to EN 13445-3 Annex B Table B.8-2. Figure 10 shows Mises equivalent stress of the plates (tubes unselected) after un-
the maximum absolute value of the principal strain in the critical loading. It can be seen that all stresses are lower than the yield
node, located in the center bore of the plug plate on the outer sur- strength (R p0.2,plate,95◦C = 307 MPa). Therefore, the plates have
face. The maximum principal strain is below the 5% limit with a shaken down to linear-elastic behaviour and the principle is ful-
value of εmax = 1.2% at a pressure of 7.2·γ p = 8.64 MPa, consid- filled. There follows: PPD = 7.2 MPa.
ering the partial safety factor for the pressure γ p = 1.2 according
to EN 13445-3 Annex B Table B.8-1, and therefore, the pressure
of PGPD = 7.2 MPa is allowable. Stress Categorisation Route
PD-DC: To check the admissibility against progressive plas- The design stress according to EN 13445-3 Clause 6 is given
tic deformation one full pressure cycle between a minimum by fd,plate = 197.3 MPa for the Plates.
value of 0 MPa and a maximum value of 7.2 MPa with all par- The determining cross-section of the structure is the center bore
section in the plug plate, as can be seen in Fig.12 where the
Tresca equivalent stress at a pressure of 1 MPa is shown. As
1 For the tube material hot tensile properties are not given in the material stan-
prescribed in EN 13445-3 Annex C the stress linearization has
dard. An interpolated value has been chosen, with interpolation between the to be performed with the mean values over the ligament width.
values for EN 10216-2: P195GH and P235GH
50
Stress (MPa)
Figure 13. Example 3: Stress categorisation: membrane (dashed line) simple and needs often further investigation or lots of practice,
and membrane plus bending (solid line) equivalent stresses and could easily lead to misinterpretation.
In comparison the direct route gives a more suitable, adapted
tool with a much larger and better output of information. The
This is a very time consuming procedure, because it is not imple- need of a plastic solution results in a larger calculation time, but
mented in the standard Finite Element software and the averaging with modern computer hardware this argument does not count
and linearization has to be performed by hand or a mathemat- any more. The design check against gross plastic deformation
ical software. First, for the single stress components the mean is quite simple and straightforward in most cases. The check
value over the ligament width has to be calculated for a number against progressive plastic deformation is the more complicated
of points over the plate thickness. Then the mean values of the one, but there are some good guidelines given nowadays with
stress components have to be linearised over the plate thickness, lots of examples [3–5,7]. Especially if thermal stresses occur this
and with these values the equivalent stress of the membrane and design check gives unique solutions and lots of information about
bending stress can be calculated. Figure 13 shows, for a pressure the behaviour of the structure, useful for design improvements
of 1 MPa, the membrane and membrane plus bending stresses in and specification of dedicated in-service inspection intervals.
the considered cross section over the thickness oft the plate, mea-
sured from the outer surface, calculated in this way. The maxi-
mum of the membrane plus bending stress occurs at the outer sur- REFERENCES
face. The bending stress has to be categorised as primary bending [1] ASME, 2004. Boiler and Pressure Vessel Code. Section VIII,
stress, and this gives Pm = 6.5 MPa and Pm + Pb = 59.6 MPa. Division 2: Alterntive Rules. ASME, New York.
Therefore: [2] CEN, 2002. EN 13445-3, European Standard for Unfired
Pressure Vessels. European Committee for Standardization.
1.5 · 197.3 [3] N. Taylor, e., 1999. The design-by-analysis manual. Tech.
Pm + Pb ≤ 1.5 · fd,End ⇒ PCAT = = 4.97 MPa (4)
59.6 Rep. EUR 19020 EN, European Commission Joint Research
Centre, Petten - The Nederlands.
All other cross-sections give much lower stress values, and are [4] Preiss, R., “Design by Analysis: The Racheting Check in
therefore not determining. EN 13445-3 Annex B”. J.L. Zeman, ed., ICPVT-10 Vienna,
2003.
[5] Zeman, J.L., Preiss, R. “The deviatoric map - a simple tool
in design by analysis.”. Int. J. Pres. Ves. & Piping, 76 (1999),
CONCLUSION
pp. 339–334.
The considered examples show that for simple geometries
[6] Zeman, J.L. “The european approach to design by analysis”.
the stress categorisation route is a quite easy to apply tool, es-
Pressure Vessel and Piping Codes and Standards, ASME
pecially if the structure has been modelled using shell elements.
PVP 439, 2002.
Additionally a Finite Element software supporting elastic mate-
[7] Zeman, J.L., et. al., 2006. Pressure Vessel Design: The Di-
rial only is required. Nevertheless the stress categorisation route
rect Route. Elsevier.
gives much lower allowable loads for the considered examples
compared to the ones calculated according to the direct route. Ta-
ble 2 shows the result summary of the investigated examples. As
soon as the geometry of a structure gets more complex, the stress
categorisation becomes much more time consuming. The way in
which categories the stresses have to be classified is not always