July 2001 .

/t'
COADE News

Modelírrg of Internøl Pressure F1

ønd Thrust load on Nozzles K1(x)

using WRC 368 P.A

(by: Månde€p 6ingh and Dåvê Oiehl)

The vessel-nozzle junction presents an unusual situation for stress K2(x)

analysis. Local a¡eas of high stress occur near the junction because
of the presence of the hole in shell wall and welds that attach the

#l
nozzle to the shell. The loads on the vessel-nozzle junction can be
external (such as from the piping system) or can be due to internals--
pressur€. /
The Welding Research Council (WRC) Bulletins 107, 297 nd368, <liF
".F2
provide empirical methods, for calculating stressesãt the
nozzle junction. Many have asked how to model the th¡ust on
the nozzle. WRC 368 addresses the intemal pressure tlrust
loadings on the nozzle. In PVElite Version 4.2 we implement
WRC 368, as it can be a useful design aid. In this article, we I
examine various aspects of WRC 368 and how it affe¿ts the local
stress calculations. To illustrate further, case of a nozzle attached to a
s assume
vessel on one system on the other side. Let P
WRC 107 and WRC 297 provide the formulae for süesses resulting be the internal and piping, and A be the inside
from external loading. WRC 107 has been discussed in two previous a¡ea of the nozzle. the load interest is P*A located on the
articles i¡ June 1997, June 2000 newsletters. In this article, we will elbow "upstream" the nozzle, away from the nozzle.
focus on strosses due to internal pressure, The balancing force A) acts the vessel wall opposite to the
nozzle and is shown Figure 1 balancing force is countered
Concepts by the vessel support, it from the nozzle; hence it is
not considered i¡ this
WRC 368 includes 2 loading components, the surface stress due to
internal pressu¡e and the pressure thrust load. Let's review the The load on the will be a function of the
Pressure thrust load.
stiffness between vessel load (including any nozzle
flexibilities) l), and the of the system beyond the
Pressure Thrust load (Spring 2). It be visualized two springs in series with the
applied load them.
Pressure th¡ust is the force exerted on the vessel-nozzle junction
due to the internal pressure. Figure I shows the artangement of a
F2 F1
typical vessel-nozzle junction.

Vessel side
õl
@,u,
The force F is in equilibrium with the two spring forces Fl and F2:

F=F1 +F2 (l)

The spring stiffness K and the displacement ô can be related as:

Kl=Fl/ôl
IV K2=F2l õ2
\
I
\ p
u 9
i
COADE MechanicalEngi News July 2001

So: str€sse$ are not self-limiting in nature and can cause ductile rupture
or a complete loss of load carrying capacity due to the plastic
F=ô1 *Kl+õ2*K2 collapse of the structure upon single application of load (ASMB).
Primary shess can be further sub-categorized as:
Since, ô1 - ô2, let's denote it by ô:
, General Prim.ary Membrane Stress (Pm)
So: This is the average primary stess across a solid section. It
excludes the effect of discontinuities and concentrations, An
F=ô*(Kl+K2) example is stress in a cylinder due to internal pressure given

[<zà &
õ=F/(Kl +K2) by Pdl2t.

on -nozzlejunction: * O
. Local Prirnary Membrane Stle.ss (PJ)
?t This is the average stress across a solid section. It is caused by
-Fl=F*Kl/(Kl+ (2) extemal edge resultants developerl because of the global
discontinuities. Examples include stresses developed at the
nozzle hole or at tlle small end of a conical reducor.
other side ofthe applied load (Spring 2)
is stiff, for example due to an anchor, then pressure th¡ust will be Secondary Stress (Q)
absorbed by the anchor. Thus, the nozzle will experience very little Secondary stess is developedas result ofimposed suain. Secondary
di¡ect axial stress. This can be seen from equation 2. Note that a shess is a global selfJimiting stress, Bending stressos and the
gfeater K2 results in a lower thrust force Fl. Therefore, in this stresses due to thermal expansion come under this category.
including all ofthe pressure th¡ust into
Peak Stresses (F)
Ifon the other hand the run of pipe denoted by 2 is flexible Poak st¡ess is a localized self-limiting stress. It causes no
the nozzle objectionable disto¡tion except rhar it may be a possible source of
thrust. we should add the fatigue failure. Fatigue analysis for the vesselattachment junction
appropriate is explained in the June 2000 newsletter.

There can be another extreme case; ifnozzle has a blind flange Nomenclatule
then it will experience the enti¡e force due to the pressure thrust, Following nomenclature is used i¡ this article:
must include whole thrust load for this case.
R Mea¡ Vessel Radius
Hence, the amount ofpressure th¡ust acting on a nozzle depends D Mean Vessel Diameter
of the T Vessel Thickness
pressure thrust are to the prprnS are analyzed, the d Nozzle Diameter
süuctural load at the nozzle due to pressure can be calculated. More t Nozzle Thickness
research is wa¡ranted in this direction, to determine the amount of
pressure thrust the vessel-nozzle junction experiences. Note: Except WRC 368, an lntroduction:
for the pressure effect on expansionjoints, the CAESAR II program
does not automatically include piping loads due to pressure. Instead, WRC-368, entitled "Stresses in Intersecting Cylinders Subjected to
the longitudinal pressure stess is simply added to the piping stresses Pressure" was released in 1991, WRC 368 provides an approxímate
where applicable as a scalar. method of calculating the maximum skess intensities due to internal
pressure at cylinder-nozzle intersections. It is based on the finite
If we cannot accurately determine the amount of pressure tlrust, element analysis progr¿ìm developed by Prof. C.R. Steele, FAST2.
there is a method that analyzes the thrust load more accurately. Here The same program was used in the development of WRC 297.
we will review WRC 368 and compa¡e it with other cur¡ent methods,
\VRC 368 applies the full load due to pressure thrust (P*A). The method for design ofnozzles, subjected to pressurç, is given in
many pressure vessel codes, A typical method is the area-replacement
Let's look at the va¡ious categories of stress caused by internal method. This method assures that the general primary membrane
plessure and pressure thrust load. süess near the opening remains below the level ofshess before the
hole was made, This method does not consider the local primary
himary Stress membrane stesses and bending stresses. The \VRC 3ó8 method
Primary $tress is necessary to satis$ the equilibrium conditions provides the rnaximum value of membrane sEess intensity (general
with the extemal imposed loading such as P*4, MZ. It may also be and local, Pm+PL) and the memb¡ane + bending stress intensity
called load-controlled stress (ASME Code Case N-47-28). Primaty (Pm+PL+Q). Moreover, these stresses are calculated in bofh the

l0
July 2001 COADE Meohanical Engineering News

shell and the nozzle. Therefore, WRC368 considers two additional

criteria of failure, in addition to the ca$e checked by the area-
replacement method,

The FAST2 program, used for creating this Bulletin, applies the full
pressure thrust force on the nozzle along with the internal pressure.
Therefore, it can be deduced that WRC 368, which is based on
FAST2 program, also includes the pressuro thrust force on the
nozzle. This was further confi¡med by one of the authors of WRC
368. It is importåntbccause IVRC 3ó8 provides much better modeling
of the pressure tb¡ust load than the other current methods. Let's Ftgure 3
compare the analysis methods WRC 107, IIEA and WRC 368.
4, FEA: The NozPro finitç element prograrn, developcd by
Comparntive Studyt Paulin Resea¡ch Group, is used to analyze the models, This
program also applies the whole pressure thrust load. Links to
Here we will compare the results from analysis performed using the this program a¡e conveniently provided in the WRC l0?
following methods: module in CodeCalc/PVElite,

l. Pd/2t This approach uses tho general primary membrane

shess equation (Pd/2t) for calculation of intemal pressure Intcrnal Pressure only and N<¡ P¡d:
shess. This method is used in tho tilRC rc7n97 module in
COADE's programs (CAESAR II, CodeCalc and PVElite), First, we will do a comparison with internal pressìre, no extprnal
asWRC 107/297 only addressextemalloads.Forthis approach loads and no reinforcement pad. However, the pressure thrust is an
we did not include the pressure th¡ust load, see Figure 2. external load, it is considered here because it occu¡s when the
system is prêssurized.
2. Pdl2t + tull Pressure Thrust, Pd/2t + PT(107): This merhod
uses the methodology ofìVRC 107. In addition to pressure, Vessel¡
the whole thrust load (P*A) is applied as a load along the axis Mean diameter: 70 inch
of the nozzle. Here wo would check the box to include the Thickness: I inch
pressure thrust load. Length; 220inch

Nozzle¡
Mean diameter: diffe¡ent runs at 14,21,28, 35 and 40 inch
Thickness:.875 inch
Length: 2Oinch
Pressure: 200 psi

Let's check if these models are within the geometric limit¿tion of

WRC 107/368, The models with nozzle mean diameters of 2l inch
to 40 inch oxceed the curves used fo¡ calculating the bending stress
due to radial load on the nozzle (in this case, the pressure thrust).
This becomes more pronounced as the nozzle diameter increases.
We will see later that this may have an effect on the accuracy of the
bending sFessas due to the thrust load.

The d/D ratio for the model with the mean nozzle diameær of 40
inches is 0.571, which exceeds the limitation of 0.5 in WRC 107/
X'lgure 2 297t368.

3, WRC 3681 Here we used the IùVRC 368 feaore implemented Figure 4 displays the finite element mesh and tÌ¡e contour of thc
in CodeCalc/PVElite, to activate it click on tho appropriate secondary sEess, for the model with nozzle mean diameter of 14
checkbox as ehowninFigure3.Tholoadings includeintemal inches.
prcssure and the full pressure th¡ust load on vessel-nozzle
junction.

1l
COADE Mechanícal News July 200'l

It ¡rjf,{f' (glf q¡tafla) gra 3 not properly address the pressure issues, especially for the bending
stress. Another point to note is that for this method, the curve used
for calculating the bending sFess due to the thrust load was exceeded,

ffi In other words, there was no data availablç in WRC 107 for this
case, Then program used the last value available on the curve,
which introduces an inaccuracy. Hence, the increa$e in sûess values
f¡om Pd/2t + PT(107) will also be affected by this.

The results from IWRC 368 and FEA are relatively close. Indicating
that, WRC 368 can be used as a design tool, if performing a finite
element analysis is not an option,

k*, Varlatlon of Memb¡¡ne + Bendlng St¡ec.cc wllh dlD

't20000

100000
Figure 4
Püâ + PT(107)
 80000
368
An impofant p¿uameter in this evaluation is the d./D ratio (nozzle ô 60000
FËA
E
mean diameter/Vessel mean diameter). Therefore, to see its effect !
ø 40000 PdIA
we varied the nozzle diameter from 14 to 40 inches, while keeping 200@
the rest of the geometry consta¡t. The variation of the primary 0
membrane stresses is shown in the Figure 5. The stresses from o,2 0.3 04 0.5 0.571
WRC 368 and from Pdl2t + PT(107) are close, the stresses f¡om d/D ratlo
FEA taper off with the increasing d/D ratio,

Figure 6
Varlatlon of M¡¡rûranc Strca¡er wlth d/D
Stesses f¡om the pd/2t method are much smaller than the other
30000
¡nethods that additionally include the pressure thrust effect. Pressure
25000 thrust load can make a significant effect on stress level a¡ound the
o 20000
 --¡-Pdzl + PT(f 07)
vessel-nozzle junction. Hence, it useful to check the system and
E 15000 --*-368 estimate if any pressure thrust load exists,
6
ó
ø 10000 -}-FEA
Pda Due to a more accurate analysis performed by FEA, this design still
5000 passes with the full pressr¡re thrust load. We can also see that the
0 accuracy of the r¡/RC methods decrease with an increasing d/D
0¿ 03 o.4 0-6 0.671 ratio. The points with maximum membrane + bending stress per the
diDÞilo FEA, are located in the longitudinal plane (shown in Figure 4),
coresponding o the points A and B in the WRC 107 couvention.
Figure 5 However, WRC 107 roports a¡eas of high stress near points C, D
along the circumferential plane. That again suggests that WRC 107
Figure 6 shows the variation of the membrane + bending stresses is not appropriate for modeling the pressure loadings,
compared to the dlD ratio. Notice the increase in the stress values
f:om the Pdl2t+ PT(107) method with the increasing d/D ratio. If Rcinforcemcnt pad
the allowable stess for this case is 60, 000 psi (3ts*u,, for SA-51ó
?0), thedesign fails miserably per Pdtzt+Yfç}7) methõd. However, \ryRC 107, 297 ud368 do not consider ¿ reinforcement pad. WRC
it still passes when analyzed with FEA and ìVRC 368 methods! 368 recommends a rule of thumb that has been used successfully
and provides somewhat accurate and generally conservative results.
The reason is simple, as the nozzle diameter increases; the thrust
load (P*A) increases by the square of that amount and becomes a If
significant number. The tcsts used for preparing WRC l0? did not
include inte¡aal pressure. Hence, the method Pd/2t + PI(107), does Pad width > 1.65*rft7 and > !
2

t2
July 2001 COADE Mechanical Engineering News

then the shell thickness can be increased by the amount of pad WRC 368 only addresses cylinder-to-cylinder intersections loaded
thickness. This ensures that the pad be at least as wide as the region under internal pressure. When these limits are exceeded then the
of discontinuity stress around the hole. If the pad does not results will not be as accurate.
these limit¿tions then it the
not because of this limitation, resultsfrom Conclusions
WRC 368 can be significantly conservative.
We have shown that for cylinder-nozzle junctions, under internal
Internal pressure and Dxternal loads pressure only, WRC 368 is a better tool than the pdlàt+ pT(107)
method, assuming that FEA is most accurate.JllgOyides..a-.m¡¡Ch
To get a complete analysis ofthe vessel-nozzlejunction, the sûesses of the thrust when the ftll
from external loads and ones f¡om internal pressure should be thrust acts on the such
combined. We considered using WRC 368 pressure sresses with is no option to
the lQ71297 stresses due to external loads in the section VIII Div 2 control the amount load, Hence, WRC 368 will be
stress summation. However, there are some obstacles to this conservative, in cases where only a portion ofthe th¡ust load acts on
approach. The main reason is thatWRC 368 provides the maximum the nozzle. However, because of better accuracy than pd/2t +
stress intensity, but lacks information about the location and the PT(107), the results may be more reasonable (as seen in the case
orientation. On the other hand, the equations given in WRC 107/ above).
297 calculate the süesses at different locations a¡ound the vessel-
nozzle junction and assign proper signs and directions to the stress Utilizing rù/RC 368 along with WRC 1071297 is nor very accurate
values. for calculating the combined stress from pressure and external
loads. This is because WRC 368 does not provide information
It is not possible to accurately calculate thestress intensity value about the location and the orientation of the stresses. However, if
due to the combined loads, using WRC 368 along with WRC 107/ the stress analyst has an estimate of the pressure thrust, then a
297. However, WRC 368 recommends that an upper bound on the feasible option is to use the pd/2t + PT(10?) method and instead of
combined stress can be obtainedby adding the absolute value ofthe tho full thrust load ente¡ the estimated value in the radial load input
maximum stress from external loads to the results from tilRC 368, (with proper signs). The analyst sbould also note that the results of
This resulting combined stress can be quite conservative depending if
WRC bulletins will be less accurate the model exceeds the
upon the stress distribution, as the maximum stress due to external geometrical limitations or if the curves used for calculating the
loads and pressure can occur at different locations. Moreover, the stresses are exceeded. If the analyst does not have an estimate of the
stresses from these 2 loading conditions can also act in opposite thrust load, he or she can put the whole thrust load and watch-out for
directions to reduce the combined effect. very high values of Membrane + Bending st¡esses. In those cases,
IIVRC 3ó8 can be used to check the pressure stress levels, or
Lhnifations of ìüRC 368 advanced analysis tools such as finite element method can be used
to obtain accurate combined stress.
WRC 3ó8 has geometric limitations simila¡ to those traditionally
applied ro WRC L07 and297: Overall, knowing the benefits and limitations of WRC 3ó8, it can be
a useful design aid.
10<Dtr<1000
4<d/r< 1000

0.1 <r/r<3
0.3<Dr/dT<6
0.3 < ¿tJu<6.5
Nozzle mustbe isolated (itmay notbe close to adiscontinuity)
-not within z.sJã] on ve$sel and not within 2.sJl on nozzle.

t Results are bæed on nozzles extending normal to the vessel,

on the outside only.

l3