You are on page 1of 7

Proceedings of PVP2008

2008 ASME Pressure Vessels and Piping Division Conference


July 27-31, 2008, Chicago, Illinois, USA

PVP2008-61708

Selecting the Optimum Bolt Assembly Stress: Influence of Flange Type on Flange Load Limit

Warren Brown
The Equity Engineering Group
Shaker Heights, OH, USA
iwbrown@equityeng.com

ABSTRACT conservative. In this paper both the limits and results have been
In previous papers, practical limits on the maximum applied appropriately adjusted to represent the correct material that was
load for standard ASME B16.5 and B16.47 carbon steel, weld neck analyzed. In addition to the maximum stress limit, a gasket may have
pipe flanges were examined. A new code equation for the tangential a maximum flange rotation that it can accept prior to leakage. This
(hoop) stress at the small end of the hub for a weld neck flange was limit must also be assessed to ensure that the specified bolt load will
developed to facilitate calculation of the limits using elastic analysis. not cause excessive flange rotation, as outlined in the previous
The results were verified against elastic-plastic Finite Element papers.
Analysis (FEA). In this paper, the work is extended to include other
In the previous papers by this author on the topic, the concept
flange configurations, including loose ring flanges, slip-on flanges
of maximum allowable bolt load for both standard pipe flanges
and flat plate flanges. This paper is a continuation of the papers
presented during PVP 2006 and PVP 2007 (Brown [1,2]) and it (based on Elastic-Plastic FEA and code calculation) and limitations
extends the scope of the proposed methodology for determining for non-standard flanges (based on code calculation) were examined.
flange stress limits in determining the maximum allowable bolt load The FEA used to confirm the work included all standard ASME
for any given flange size and configuration. B16.5 flanges greater than 2 in. nominal bore and ASME B16.47,
Series A flanges up to 48 in. nominal bore. The goal of this analysis
INTRODUCTION was to determine the point of Gross Plastic Deformation (GPD) of
In the previous papers, the importance of knowing two basic each of the flanges, which is suggested as the point of maximum
values for the joint; minimum acceptable assembly bolt load and, allowable bolt load for flange assembly. GPD is defined as the point
more importantly, maximum acceptable assembly bolt load was where the rotation of the flange ring under applied bolt load becomes
highlighted. The reason that the maximum acceptable bolt stress is non-linear. This signifies that, at that point, the flange is taking on a
more important is that an optimal bolt load for joint assembly should permanent set and will, upon release of load, have residual flange
always be biased towards the maximum acceptable load, as this gives rotation. In addition, the onset of GPD results in a rather rapid
the largest buffer against joint leakage. It could even be argued that increase in flange rotation with applied load and so, therefore, it is
one need not know the minimum acceptable bolt load, because by likely that such rotation will cause joint leakage due to mal-
assembling to the maximum possible bolt load then we have given distribution of gasket stresses. This paper will focus on determining
ourselves the greatest chance of sealing the joint. The maximum the maximum acceptable bolt load for other standard (slip-on and
acceptable assembly bolt load is determined from examining the loose) and non-standard (flat plate) piping flanges based on both
maximum load that can be taken by any one of the three joint flange rotation and GPD limits. Only the assembly bolt load case is
components (bolts, gasket or flange). Unfortunately, the upper limit considered in this work, as this is generally considered sufficient due
for both gasket and flange are not well defined. A selection of to the fact that gasket relaxation will reduce the bolt load during
maximum gasket stress values were presented in a previous paper operation to a point below the flange or bolt material yield at
(Brown [1]), based on both laboratory test and field experience. This temperature (during operation). If gasket relaxation is less than the
original work was expanded on and new code equations and elastic reduction in material yield due to temperature then the specified
stress limits were proposed in a following paper (Brown [2]). In that assembly bolt load will need to be reduced by the ratio of yield at
paper, Elastic-Plastic (E-P) FEA results were presented for SA105 temperature divided by yield at ambient. This adjustment must also
flanges that were actually based on lower strength material. Due to be made if the joint is to be tightened while hot.
this, the subsequently developed code equation limits were overly

1 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms


ELASTIC-PLASTIC FEA membrane stress. It was found that this worked well for the flanges
Axisymmetric elastic-plastic FEA was performed on the where yield occurred at the shell to hub junction, however for flanges
selected flanges with material properties for SA-105 or SA182-F304 with long hubs, the yield location was at the ring to hub junction and
and an elastic-plastic stress-strain function as shown in Fig. 1. In all therefore further limits on the other stress components were imposed
cases a spiral wound gasket (dimensions per ASME B16.20) was in order to establish overall limits for the determination of the
modelled using the full stress vs. deflection curve obtained from test maximum permissible flange assembly load (as limited by the
with the uniaxial gasket elements (GK4AXN) available in the flange). The following revised limits were found to give the best
agreement when compared to the elastic-plastic FEA results.
ABAQUS commercial software program. The elastic modulus in the
ST0 < Sy
bolt hole region of the flange ring was adjusted to account for the
ST < 1.5 Sy
reduction of elasticity in the tangential direction due to the presence SH < 2.0 Sy
of the bolt holes. The bolt was modelled using plane stress elements (ST + SH/f) < 3.0 Sy
with an area equivalent to the cross-section of all of the bolts and the (1)
(ST0 + SH) < 3.0 Sy
width was adjusted, such that the bending moment of inertia for the (SR + ST) < 3.0 Sy
plane stress model was equal to the actual bolt case. Both the gasket
and nut surface were modelled with contact, which allowed Using these limits, the maximum assembly bolt stress limits
separation of the surfaces as the flange rotated (which reduces the shown in Table 1 were calculated for the range of standard SA-105
effective moment arm as the flange rotates). The bolt load was weld-neck flange sizes and ratings. There was generally good
applied using the ABAQUS *PRELOAD command, which reduces agreement found between the FEA results and the closed-form
the length of the mid-plane bolt elements until the desired load is solution, with less than 10% difference between most values.
obtained. The preload was applied in 10% of bolt yield increments to
a final yield value of 105ksi (for SA193-B7 bolting material < 2-1/2 Table 1 – Calculated GPD Bolt Loads; Weld-Neck Flanges
in. diameter). No adjustment was made for the reduction in yield for Flange Class
150 300 600 900 1500 2500

Nominal Pipe Size (in.)


larger diameter studs, as the goal was to determine the maximum
2 65.2 44.9 74.7 48.1 59.9 64.8
permissible bolt load prior to GPD, rather than to determine the 2.5 83.5 41.2 56.3 54.7 64.0 71.9
maximum % of bolt yield. 3 105.0 57.1 79.0 75.0 62.7 77.0
4 64.6 81.3 91.8 60.5 71.4 65.8
ANALYSIS METHOD AND RESULTS 5 58.3 105.0 96.2 67.9 76.6 72.6

The pipe wall thicknesses and material yield properties used 6 78.5 86.0 91.4 78.8 87.7 77.6
8 105.0 89.0 95.3 67.1 83.5 80.8
in the analysis are listed in Brown [2]. The gasket type used in the
10 73.0 92.7 82.1 64.4 91.0 78.8
analysis was a spiral wound gasket in accordance with ASME 12 103.3 88.0 81.6 71.6 80.3 86.2
B16.20. For the weld-neck flanges, the elastic-plastic FEA models 14 84.5 65.9 74.4 76.3 70.3
were run for each case and the flange rotation (defined as the rotation 16 81.6 57.7 73.7 77.1 70.6
occurring at the mid-plane of the flange ring) versus bolt load 18 89.0 68.5 86.2 77.4 75.6
information from the FEA output was analysed to determine the bolt 20 82.4 65.4 69.9 79.1 72.7
load prior to the change in the slope of the load vs. deflection curve 24 69.4 53.0 65.3 79.2 69.7
26 31.6 35.1 52.1 65.0
becoming two times greater than the initial slope. This is a nominal
28 28.0 38.3 51.4 57.8
limit selected by the author that allows a standard determination of 30 33.0 42.0 64.8 67.4
the onset of GPD, although in many cases the accuracy of this 32 25.1 39.5 57.5 66.7
determination for the FEA was only within a 10ksi range, due to the 34 23.2 43.0 67.1 60.6
rather large increments chosen. The large FEA load step increments 36 30.0 37.8 58.6 63.2
were chosen in order to minimize the computer time required to 38 30.6 80.8 90.3 79.9
obtain the results. 40 28.9 77.7 91.9 77.2
42 31.6 84.2 90.8 84.9
44 32.0 98.0 92.6 82.6
The resulting maximum assembly bolt loads from the
46 34.5 105.0 99.6 81.6
elastic-plastic FEA prior to onset of GPD were used to develop 48 32.2 76.0 87.8 90.6
closed-form solutions to determine the maximum permissible Note: the shaded cells represent values that are lower than 50ksi bolt stress.
assembly bolt load limits for individual joint geometries (including
different gaskets, pipe wall thicknesses and flange materials). The Since the closed-form solution is based on standard code
closed-form solutions used the current code (ASME [3]) flange stress equations, it is a relatively simple step to apply these formulae to
equations and also included the development of an equation for alternative flange configurations, using the rules presently defined in
tangential (hoop) stress at the small end of the hub (ST0). The ASME [3]. It should be noted, that presently the code instructs the
development of the new equation followed the original approach used user to analyse optional type flanges (slip-on flanges) as integral
for development of the ASME code equations (Waters [4]), and flanges. This requirement is generally more stringent than allowing
resulted in a new “f” factor that relates tangential stress at the shell to them to be analysed as loose flanges and will result in a more robust
hub junction to longitudinal stress at the ring to hub junction, in the design. In addition, the code instructs the value of g0 to be equal to
same way that the existing “f” factor relates longitudinal stress at the the vessel wall thickness for slip-on type flanges, which is not
hub to shell junction to longitudinal stress at the ring to hub junction. appropriate, as this is the equivalent of specifying a weld-neck
(tapered) hub. In this paper, the code approach and the approach of
In order to determine the maximum permissible load on the using the actual dimension for the small end of the hub is used, which
flange, the value of ST0 was limited to material yield (Sy), since it is a means that for non-tapered hubs g0 equals g1. In order to verify the

2 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms


applicability of the method to the different geometries being analysed thickness case are shown in Fig. 10 to Fig. 12. It can be seen that
an Elastic-Plastic FEA of a selection of geometries was conducted there is excellent agreement when the slip-on flange is treated as a
and the results were compared to the closed-form solution to verify loose ring. The point of GPD appears to be accurately predicted by
accuracy. the new stepped closed form solutions (explained following) for both
wall thickness cases, however there is an initial amount of plasticity
The first geometry case that was run, was actually a weld- that occurs in the FEA results that is not reflected in the closed form
neck flange, but with a modified hub dimension. This case highlights solution and therefore there is a significant difference in the
the wide variation in flanges that are possible using ASME B16.5, calculated flange rotations. One interesting result from the analysis
where the hub height is not defined, only the overall flange height. on the connected flange cases is that the extent of connection appears
The specification on hub dimensions in ASME B16.5 allows a hub to make a relatively minor impact on the rotation and point of GPD.
angle of up to 45 degrees. This configuration is, obviously, a much This is encouraging, as it means that the same limits should be
different flange than one where the hub extends the full length of the applicable to most commonly found applications of connected slip-on
flange height. The GPD results, both FEA and closed form, are flanges. The second aspect that is apparent from the analysis is that
outlined in Fig. 2 for a NPS 4, cl.600 flange and in Fig. 3 for a NPS the representation of the slip-on flange as an integral flange, per the
20, cl.150 flange for SA182-F304 material. It can be seen that in both present requirements of the code (ASME [3]) or by assuming that the
cases GPD occurs much earlier for the 45 degree hub case as flange is integral and using g0 = g1 is not accurate. For the present
compared to the full height hub. This is particularly apparent in the code case the point of GPD is under predicted and for the fully
NPS 20 joint, where GPD occurs at 17ksi for the 45 degree hub, as integral case, the flange rotation is grossly under predicted and the
compared to 70ksi for the full hub height case. In both cases, it can be point of GPD is not over predicted (non-conservative). When the
seen that the closed form solutions gave good agreement with the original work is examined (Waters [4]) it is apparent why this is the
FEA results. Plots of the flange tangential (hoop) stresses for both case. The original development included two types of flanges.
cases for the 4in. and 20in. cases are shown in Fig. 4 and Fig. 5, Integrally connected flanges, for which factors “F” and “V” were
respectively. developed and loose ring flanges, for which the factors “FL” and
“VL” were developed. The reason why the equations do not work for
The second case run was the case of a flat plate flange, having the welded slip-on flange case, is that the original development
identical dimensions to a NPS 20, cl.150 ASME B16.5, SA182-F304 assumes that the pipe or vessel wall thickness is equal to the hub
flange, but without the hub. This flange geometry was run with three small end thickness for an integral flange. In the case of a welded
cases, a loose hub (not connected to the pipe wall, similar to a lap slip-on, this is quite obviously not the case, as the pipe or vessel wall
joint flange) and two cases connected to the pipe wall, one with a is generally considerably thinner than the hub thickness.
4.78mm wall thickness and the other with a 7.62mm wall thickness.
The objective was to compare the three cases with FEA results and In order to accurately assess slip-on flanges, it was necessary
the closed form solution in order to confirm that the closed form to go back to the original development of Waters [4] and extend that
solution gave acceptable results. In the case of the closed form to the case of a stepped flange, where the pipe or vessel wall
solutions, the flange is analysed as a loose ring flange when not thickness (hereafter referred to a g2) is different from the hub small
connected to the wall and an integral flange when connected to the end thickness (g0). By manipulating the equations for the connected
wall. The GPD results from both FEA and closed form are shown in shell in the development of the integral flange factors, it is possible to
Fig. 6. It can be seen that the flat plate flange, without a hub, is a insert the value of g2 into the calculation and therefore generate a
relatively weak flange and that welding it to the pipe wall stiffens the third set of flange factors (“FS” and “VS”) for stepped flange
arrangement significantly. In the case of the loose flange, the limit configurations. These factors can be accurately used to represent both
state is tangential (hoop) stress in the ring. For the connected flanges, the stresses and deformation in a flange where the pipe or vessel wall
the limit state is hoop stress in the connected pipe wall due to the thickness (g2) is significantly different to the hub thickness (g0).
flange bending. The closed form solution for the loose flange agrees These new flange factors are used in place of the original factors (“F”
well with the FEA results. The integral flanges appear to and “V”) in the integral flange stress equations. At the extremes of
conservatively predict GPD, as compared to the FEA results. geometry; g2/g0 ratio tending to 0.0 or g2/g0 ratio tending to 1.0, the
However, the limit state of the closed form solution is the pipe wall results of this equation tend to the integral flange (“F” and “V”) and
stresses, and when the FEA stress plots are examined (Fig. 7) it can loose ring flange (“FL” and “VL”) factors respectively, as would be
be seen that a lower GPD value (limited by pipe wall stresses) would expected. The equations for the flange factors were developed for a
seem appropriate, as the wall is yielded through the cross section. range of g2/g0 within the range or 0.1 to 0.6, which is expected to
describe most flange configurations. Flanges with ratios above 0.6
The third case that was examined was a slip-on flange, both in can be treated as integral flanges with g2 = g0 and below 0.1
the loose (unattached to the pipe wall) and welded configurations. extrapolation can be performed between the stepped solution and the
Once again, a NPS 20, cl.150, SA182-F304 flange was used as an loose ring solution if required. However, it is likely that sufficient
example, with hub dimensions as per ASME B16.5. The hub was accuracy would be obtained by simply using the value of 0.1 for
connected to two different wall thickness pipes (4.78mm and ratios below that value and 0.6 for ratios above this value.
7.62mm) with two different insertion lengths into the slip-on flange.
The first case modelled a either a minimally inserted pipe case or the The equations for FS and VS were developed for the case of
case where the slip-on is welded only at the hub location and not near g0/g1 equal to 1.0 (i.e.: a straight hub). A similar development can be
the flange face. The second case would represent when the pipe was performed for other cases, and when this is done there appears to be a
fully inserted into the flange and a fillet weld made at both the hub relatively minor influence on the results for small hub angles. It is
and flange bore locations. The GPD results for the thin and thick therefore suggested that the factors determined for g0/g1 equal to 1.0
walled pipe cases are shown in Fig. 8 and Fig. 9, respectively. The can be used for most hub configurations, and if a check is required
corresponding FEA component stress results for the 4.78mm wall for configurations that are significantly different from this ratio, then

3 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms


it would be appropriate to compare the results obtained between the reduced if applied to a gasket where it is appropriate to limit the
integral (F,V) and stepped (FS, VS) equations. The developed flange rotation during operation. The bolt load in Table 2 should be
equations were generated by determination of a range of FS and VS linearly reduced by the ratio of the maximum desired flange rotation
values using the system of equations used in Waters [4] for the given divided by the flange rotation listed in Table 3. A comparison
range of the three different ratios (g0/g1, g2/g0, and h/h0) that describe between the GPD loads for weldneck and slip-on flanges are shown
the shape of the hub. These values were then input into curve fitting in Table 4, which lists the ratio of the weldneck load to the slip-on
software to generate an equation that accurately (within the given load for each flange size and class. It can be seen that the maximum
range) describes the value of FS and VS for any given geometry. assembly load for the slip-on flanges can be up to 60% less when
These factors can also be represented in the traditional graph format, compared to weldneck flanges. Given the range of values and
and these graphs are shown in Fig. 13 and Fig. 14. It should be noted inconsistency across sizes and classes, it does not appear that it is
that for values of h/h0 above 1.0, the result is identical to the value at possible to use one table for both types of flanges even with a
1.0 and is the same as the standard integral flange solution, indicating multiplier to reduce the load for the slip-on case.
that the hub length is sufficiently long that it is bending and passing
little effect onto the shell wall. Table 2 – Calculated GPD Bolt Loads; Slip-On Flanges
Flange Class
Once the new factors are used in both the stress and rotation 150 300 600 900 1500

Nominal Pipe Size (in.)


2 105.0 52.2 83.0 61.4 59.9
equations, it can be seen in Fig. 8 and Fig. 9 that the results are in 2.5 77.4 46.5 59.5 54.7 64.0
good agreement with the FEA results and give a sufficiently accurate 3 103.5 64.7 81.6 75.1
prediction of both GPD and rotation. 4 57.2 86.2 87.1 67.7
5 64.7 98.3 73.5 71.4
6 87.4 66.4 71.8 77.7
The equation used to define both factors is: 8 105.0 78.1 74.7 66.2
⎡ ⎛ h ⎞ ⎛g ⎞ ⎛ h ⎞
2
⎛ ⎛ g ⎞⎞
2
⎛ h ⎞ ⎛ g ⎞⎤ 10 69.2 68.4 62.4 62.2
⎢ a +c ⎜ ⎟ +e.ln ⎜ 2 ⎟ +g ⎜ ⎟ +i. ⎜ ln ⎜ 2 ⎟ ⎟ +k ⎜ ⎟ .ln ⎜ 2 ⎟ ⎥ 12 97.8 69.1 61.0 67.9
⎢ ⎝ h0 ⎠ ⎝ g0 ⎠ ⎝ h0 ⎠ ⎝ ⎝ g0 ⎠ ⎠ ⎝ h 0 ⎠ ⎝ g 0 ⎠ ⎥⎦ 14 64.6 41.1 49.9 73.1
FS , VS = ⎣ 16 65.7 46.4 53.6 73.8
⎡ ⎛ h ⎞ ⎛g ⎞ ⎛ h ⎞
2
⎛ ⎛ g ⎞⎞
2
⎛ h ⎞ ⎛ g ⎞⎤ 18 81.4 54.5 79.2 74.6
⎢1+b ⎜ ⎟ +d .ln ⎜ 2 ⎟ +f ⎜ ⎟ +h. ⎜ ln ⎜ 2 ⎟ ⎟ +j ⎜ ⎟ .ln ⎜ 2 ⎟ ⎥ 20 70.6 62.1 72.3 76.0
⎢ ⎝ h0 ⎠ ⎝ g0 ⎠ ⎝ h0 ⎠ ⎝ ⎝ g0 ⎠ ⎠ ⎝ h 0 ⎠ ⎝ g 0 ⎠ ⎥⎦ 24 77.6 57.3 72.5 76.6

Table 3 – Flange Rotation (deg.) at GPD; Slip-On Flanges
Where the values “a” through “k” are defined per the below table: Flange Class
FS VS 150 300 600 900 1500
Nominal Pipe Size (in.)

2 0.03 0.28 0.21 0.14 0.10


a 0.982634253 0.802498966 2.5 0.35 0.29 0.24 0.12 0.10
b -1.349190596 -0.41880463 3 0.40 0.32 0.27 0.21
c -1.544346815 -1.515821747 4 0.52 0.38 0.27 0.21
5 0.64 0.43 0.30 0.20
d 0.570487533 0.884947634
6 0.73 0.49 0.33 0.20
e 0.262651782 -0.480427025 8 0.84 0.57 0.38 0.22
f 1.260508765 5.95650358 10 1.02 0.59 0.40 0.27
12 1.09 0.66 0.47 0.33
g 1.384917314 4.336219615
14 1.14 0.70 0.50 0.33
h 0.093155462 0.206726608 16 1.26 0.76 0.52 0.33
i 0.090531969 0.049751454 18 1.34 0.80 0.52 0.34
j -0.644972722 0.141528162 20 1.38 0.86 0.55 0.33
24 1.52 0.91 0.58 0.33
k -0.330072273 0.912545573

Table 4 – Ratio of GPD Bolt Loads; Weldneck/Slip-On


The results of the new equations and their prediction of
Flange Class
rotation and GPD were also examined for a NPS 18, cl.300 flange 150 300 600 900 1500
Nominal Pipe Size (in.)

(Fig. 15 and Fig. 16). It can be seen that in this case there was 2 0.6 0.9 0.9 0.8 1.0
significant difference between the partially inserted and fully inserted 2.5 1.1 0.9 0.9 1.0 1.0
pipe cases. However, by using the new factors, the stepped solution 3 1.0 0.9 1.0 1.0
4 1.1 0.9 1.1 0.9
predicts the point of GPD conservatively for the fully inserted case, 5 0.9 1.1 1.3 1.0
and therefore is sufficiently close to the partially inserted case that it 6 0.9 1.3 1.3 1.0
would be considered equally applicable. In the case where the wall 8 1.0 1.1 1.3 1.0
thickness gets significantly larger, the effect of partial versus full 10 1.1 1.4 1.3 1.0
12 1.1 1.3 1.3 1.1
insertion could be studied by using the stepped solutions with a 14 1.3 1.6 1.5 1.0
flange ID equal to the pipe ID for the fully inserted case and equal to 16 1.2 1.2 1.4 1.0
the flange ID for the partially inserted case. 18 1.1 1.3 1.1 1.0
20 1.2 1.1 1.0 1.0
24 0.9 0.9 0.9 1.0
The closed-form GPD results for the welded slip-on, SA105
flanges are in general lower than the weldneck flanges (Table 2). It
can be seen that the majority of them are still above the nominal CONCLUSIONS
value of 50ksi (shaded yellow). The associated flange rotation at the Allowing a 45 degree hub in ASME B16.5 does not appear to
calculated GPD bolt load is shown in Table 3, in order that these may be a good idea, as it results in a much weaker flange. Luckily (or by
be used in determining if the maximum assembly bolt load should be design), most flange manufacturers use the full hub height when

4 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms


fabricating a standard flange and so therefore this has not been an REFERENCES
issue within industry. However, given the results contained in this
[1] Brown, W., Reeves, D., 2006, “Considerations for Selecting the
paper and the fact that this hub configuration is rarely used, it would Optimum Bolt Assembly Stress For Piping Flanges”, Proceedings of
seem logical to tighten the allowable hub dimensions in ASME B16.5 the ASME PVP 2006, ASME, Vancouver, Canada, PVP2006-
to exclude the 45 degree hub option. ICPVT11-93094

Improved joint leakage performance can be expected if the [2] Brown, W., Reeves, D.., 2007, “An Update on Selecting the
bolt assembly stress is set by calculation of the actual gasket stress, Optimum Bolt Assembly Stress For Piping Flanges”, Proceedings of
with an upper limit that avoids damage to the joint components. The the ASME PVP 2007, ASME, San Antonio, Texas, PVP2007-26649
newly introduced calculation of the stepped flange factors, coupled [3] ASME. 2007, ASME VIII, Div 1, Boiler and Pressure Vessel
with limits on the magnitude of the existing code calculated stresses, Code, Appendix 2, American Society of Mechanical Engineers, NY,
have been shown to accurately predict the actual maximum USA
acceptable assembly bolt load for several standard flange
[4] Waters, E.O., Rossheim, D.B., Wesstrom, D.B., Williams, F.S.G.,
configurations. The same methodology and stress limits have been
1949, “Development of General Formulas For Bolted Flanges”,
shown to be applicable to both connected (weld-neck, slip-on and flat Taylor-Forge & Pipe Works, Southfield, Michigan, Reprinted by the
plate) flanges and loose (lap joint) flanges. PVRC in 1979.

The results presented in this paper will be, in general, 500


conservative by comparison to the actual case, in part due to the fact
450
that minimum material properties are used and actual material
properties tend to exceed minimum by 10% or more. The calculated 400

results presented in Table 2 and 3 may, therefore, be used directly to 350

Stress (MPa)
set an upper limit on acceptable assembly bolt load for standard 300
SA105, ASME B16.5 slip-on flanges. The only risk with using these 250
values would be if the actual wall thickness was significantly less
200 SA182 F304
than the values presented in Brown [2] or if the gasket dimensions SA105
150
were significantly different from the values used, resulting in a
significantly larger value of hg. Alternatively, for a given piping 100
specification or, based on the minimum wall thickness across all 50
piping specifications on a individual sites, more accurate maximum 0
allowable bolt loads may be calculated using the equations presented 0.00E+00 2.00E-02 4.00E-02 6.00E-02 8.00E-02 1.00E-01
in ASME [3] and this paper and using the stress limits outlined in this Strain
paper. In specifying these values, it is implicit that the load be
Figure 1 – SA105 and SA182-F304 Elastic-Plastic Curve
accurately applied, otherwise flange GPD may occur if the actual
achieved bolt load is significantly above the specified value. An 1
FEA - Full Hub
example of this would be the use of an incorrect nut factor when 0.9
FEA - 45deg. Hub
determining the assembly bolt torque. 0.8 Code Equations - Full Hub
Flange Rotation (degrees)

Code Equations - 45deg. Hub


0.7
The final goal of this work is to define assembly bolt load
tables for flange assembly within a plant environment. An assembly 0.6

bolt load table is constructed by considering all maximum allowable 0.5


bolt stress limits, including the limits due to the flange and bolt 0.4
outlined in this paper, maximum bolt load allowed for the gasket and
0.3
maximum allowable flange rotation. Depending on the selected limits
0.2
for maximum allowable gasket stress, it may be possible to construct
one table for all flanges in a plant or it may be necessary to construct 0.1

several tables for different gasket types. 0


0 0.2 0.4 0.6 0.8 1 1.2
Fraction of Bolt Yield (105ksi)
Further work is required to extend this analysis across a wider
variety of flange dimensions and materials, such as custom designed Figure 2 – 4in., cl.600 45deg. Weldneck Results
pressure vessel flange configurations. Confirmation of the proposed
limits for those types of flanges, versus elastic-plastic FEA, would
demonstrate the broad applicability of the method. In addition, the
effect of high bolt loads on fatigue and creep life of flanges should be
examined in order to demonstrate the broad applicability of these
results, which are based on static assembly stress results only.

5 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms


3 Figure 6 – 20in., cl.150 Flat Plate Flange GPD Results
FEA - Full Hub
FEA - 45deg. Hub
2.5 Code Equations - Full Hub
Code Equations - 45deg. Hub
Flange Rotation (degrees)

1.5

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2
Fraction of Bolt Yield (105ksi) Figure 7 – 20in., cl.150 Flat Plate Hoop Stress Results
Figure 3 – 20in., cl.150 45deg. Weldneck Results
NPS 20, cl.150, Slip-On Flange, 4.78mm w.t.
3

2.5

Flange Rotation (°)


84ksi Bolt 73ksi Bolt 2 Fully Inserted
Stress Stress Partially Inserted
1.5 Non-Welded
Code - Integral

1 New Code - Stepped


Code - Loose
Code - Present
0.5

0
0 0.2 0.4 0.6 0.8 1
Fraction of Bolt Yield (105ksi)
Figure 4 – 4in., cl.600 Hoop Stress
Figure 8 – 20in., cl.150, 4.78mm w.t. Slip-On Flange GPD

84ksi Bolt 53ksi Bolt NPS 20, cl.150, Slip-On Flange, 7.62mm w.t.
Stress Stress 3

2.5
Flange Rotation (°)

2 Fully Inserted
Partially Inserted
1.5 Non-Welded
Code - Integral

1 New Code - Stepped


Code - Loose
Code - Present
0.5

Figure 5 – 20in., cl.150 Hoop Stress


0
0 0.2 0.4 0.6 0.8 1
NPS 20, cl.150, Flat Plate Flange with No Hub Fraction of Bolt Yield (105ksi)
2

1.8
Figure 9 – 20in., cl.150, 7.62mm w.t. Slip-On Flange GPD
1.6
Flange Rotation (°)

1.4
Welded (7.62mm wt)
1.2
Welded (4.78mm wt)
1 Not Welded

0.8 Code - Integral (4.78mm)


Code - Integral (7.62mm)
0.6
Code - Loose
0.4

0.2

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

Fraction of Bolt Yield (105ksi)

6 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms


Figure 10 – 20in., cl.150, 4.78, Slip-On FEA – Hoop Stress
NPS 18, cl.300, Slip-On Flange, 9.53mm w.t.
2.5

Flange Rotation (°)


1.5
Fully Inserted
Partially Inserted
1
Non-Welded
Code - Integral
0.5 New Code - Stepped
Code - Loose
Code - Present
0
0 0.2 0.4 0.6 0.8 1 1.2
Figure 11 – 20in., cl.150, 4.78, Slip-On FEA – Long. Stress Fraction of Bolt Yield (105ksi)

Figure 15 – 18in., cl.300, 9.53mm w.t. Slip-On Flange GPD

Figure 12 – 20in., cl.150, 4.78, Slip-On FEA – Radial Stress

0.1 g2/g0 1 Figure 16 – 18in., cl.300, Slip-On FEA – Hoop Stress


h/(sqrt(B.g0)
10
0.005
0.05
0.1
0.2
0.3
FS 0.35
1 0.4
0.5
0.6
0.7
0.8
0.9
1
0.1

Figure 13 – Graph of FS

0.1 g2/g0 1
1000
h/(sqrt(B.g 0)
0.005
0.05
100
0.1
0.2
0.3
VS
10 0.35
0.4
0.5
1 0.6
0.7
0.8
0.9
0.1 1

Figure 14 – Graph of VS

7 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/17/2015 Terms of Use: http://asme.org/terms

You might also like