Flange leakage Calculation acc to Blick method

[DavidL]

Thanks Tony,
It's a really good book you have started to write, because it explains the underlying theoretical approach which I have missed, then I have tried to understand the basic flange calculations. It's always good to understand the limitations/assumptions of the formulas you use. I'm just one of those guys who are a bit confused about which approach I should go for. So far I have first used the Kellogg method and if the flange leakage check fails (normally I end up with same problems as you describe) then I just go for Blick method. But maybe I should try to learn/understand some other methods like Koves, Tschiersch and Blach? Which approach do you use? I run Caesar II so it doesn't provide any "automatic flange check", I think Caepipe have some flange check built in. Normally I don't check the structural integrity of the flange because our pipe spec normally just use WN-flanges and I think this is the best flange in ASME B16.5. As you also point out the weakest element seems to be the weld between the pipe and the WN-flange so maybe the weld should be investigated as well? Do you normally investigate the capacity of the welds?

D

[tony_black]

I don't understand - CAESAR II has a flange check.

Because of the complexity of this problem, there is no "catch-all" solution. The exact methodology must be laid out in specifications/purchase orders with all involved. It may even include progressive checks, for example.

If (x) fails, move onto (x+1)

(1) Equivalent pressure method
(2) Blick Theory
(3) EN 1591
(4) FEA
(5) bite the bullet and install next higher flange class/move flange out of problem area.

It seems you misinterpreted the part on welds:

".....even under unusually severe bending stresses, flange assemblies did not fail in the flange proper, or by fracture of the bolts, or by leakage across the joint face. Structural failure occurred almost invariably in pipe adjacent to the flange, and in rare instance, across an unusually weak attachment weld;....."

The welds, when performed properly in good practice, with good NDE, are not necessarily a weak point. See [4] Markl, A.R.C., and George, H.H., 1950, "Fatigue Tests on Flanged Assemblies," Transactions of the ASME, 72(1), pp. 77-87

[DavidL]

You are correct Caesar has a flange check (implemented 1991!) but this check is more or less a standalone program. I found some background on their homepage:

[link Point to another website Only the registered members can access] But then you model in Caesar you choose "rigid elements" from Cadworx’s flange/valve database so if you could input some info about flange material/type, gasket bolt material I think that Caesar could be able to perform an "automated flange check". But on the other hand I realize the more I learn that there are so many methods so it can be hard. And as you mentioned a flange leakage evaluation method should be specified by/together with the client.
Thanks for clarifying the welds. I found a small introduction to EN1591-1 which seems to be a more modern (complex?) approach:

[link Point to another website Only the registered members can access] D

[tturit]

din 2505...

[A.Venugopal]

It is a great pleasure watching a good exchange of ideas/ information on this forum. It is a great benefit to all . Thanks to Mr Tony and Mr David

Venugopal

[Ali366]

Tony and David,

Thanks for sharing your ideas.

Apart from EN 1591 (and some CORRECT finite element models), all of the above mentioned methods do not actually address the leakage in the joint!

ASME BPV method (also known as Taylor Forge method) is only concerned about mechanical strength of the joints and "m" and "Y" factors do not represent the leak tightness of the joint. In other words, if a flange design passes ASME Sec.VIII Div.1 (or Div.2) criteria, it does not necessarily mean it will not leak! It only means that the stresses in the flange and bolts are within allowable limits. Flange rigidity check (which became mandatory from 2003-2004) was a step forward in ensuring that rotation of the flange does not exceed a certain limit (empirically determined limits). Hence, the rotation of the flange under bolt loads and leakage was partly addressed. But the Code still does not provide any method for evaluating the effect of gasket, which is the SEALING element! In the contrary in the European standard (EN-1591) and previous DIN standards joint leakage is recognized through defining tightness classes.
Recent efforts in the last 12-15 years have been made by PVRC to define new gasket factors which enable the designer to predict leakage as well. Some of the results of such investigations have been published in WRC bulletins.

Furthermore, ASME Sec. VIII method does not address the effect of piping loads on the flanged joint (although ASME BPV codes for nuclear components - ASME Sec. III - requires the equivalent pressure method to be used in case of flanges under loads other than pressure). So, the ASME Sec.VIII designer needs to use another method to evaluate the joint under bending and axial loads (if applicable).

Equivalent pressure method (also known as Kellogg's method) tries to calculate a pressure whose effect on the flange is equivalent to that of axial and bending loads. Then the equivalent pressure is added to the internal design pressure and the total pressure is compared with the pressure-temperature rating (for standard flanges) or is used in ASME VIII Div.1 App.2 calculations (for non-standard and sometimes standard flanges) to work out the bolt loads. But the method is believed to be overly conservative, as the calculated equivalent pressure is "artificially" high. Well explained in Tony's attached article.

Bolt stress limit (the method which has been in ASME B31 Mechanical Design Committee) is even less accurate.

Blick's method (also known as equivalent axial force) is a simpler method and gives reasonable results, but is still conservative.

Koves, on the other hand, suggests another method which seems to be more accurate than the above methods. Koves' method recognizes the non-uniform load distribution due to bending moments and takes the flange flexibility into account to some extent. The method gives more realistic results while it is conservative enough and complies with FEA. Beware that tortional flexibility and effect of bolt holes are neglected in this method.
Before going for an FE analysis, I would evaluate the joint by Koves' method, if it is allowed to do so as per project specifications.

FEA can be unnoticeablly risky and incorrect. In many cases I have seen engineers apply the method with wrong boundary conditions, specially the behaviour of gaskets is not often modeled accurately. Axisymmetric models do not accurately model the flange behaviour, as well. Most of the FEA softwares are not able to consider contacts (gaskets) in axisymmetric models. It depends how precise the designer would expect the results to be. And there are lots and lots of other pitfalls in performing FEA or interpreting the results (choosing correct SCL, stress singularities, stress linearization, etc).

As my last point, I have to take issue with generalization of the test results of 4-in 300# flanges to other classes and sizes. Especially when thermal effects such as creep, stress relaxations, differential thermal expansion, etc. is involved. Also, as Tony pointed out, as the size of flanges increases the safety margin (or inherent reserve capacity) decreases. There are some reports of cases where large diameter flanges (ASME B16.47) have leaked simply due to flange excessive rotation, in the absence of bending and axial loads. So, because of versatility of the affecting factors, care should be taken before any generalization of experimental results.

[tturit]

atention!: if somebody try to calculate with AD2000Merkblatt (according to DIN2505), before read that Linde Standard...(according the industrial experience, the DIN2505 flange valvulation is not perfect, so the design safety factor must be increased doubled)
(I forget it last time)

[link Point to another website Only the registered members can access]

[mariog]

Thanks for sharing your ideas.
Tony_Black, special thanks for sharing Blick articles. I’ve used one of Blick formulas for quite long time having no idea it is Blick work but "rediscovering" it myself!

I would like to add my thoughts on the subject.

First- a simplification- it is very useful to follow ASME approach where forces appear to be "global" but in fact this is just a convention. Let me explain the case of a moment M applied on a circular plate. The reaction on G diameter is 4M/G/(PI*G)*cos (theta) where theta is a polar-angular variable. On theta=0 the peak reaction is 4M/G/(PI*G), on opposite side we say it is -4M/G/(PI*G).
We can apply this local distributed forces (depending on theta) on a "infinitesimal slice" of circular plate and after writing the equilibrium conditions we can say the forces (that must be in equilibrium with M) appears as maximum -4M/G on tensile loaded part and 4M/G on compressed part (signs depends on reference system). So such forces can be written conventionally as "global" while they are local and dependent on theta defined direction- this is the ASME convention.

According to this convention, gasket load on "tensile" loaded part of flange –as circular plate- is:
HG = W -p*PI/4*G^2 -4*M/G-F, where W is Bolts Load and F,M are the external loads, under the assumption that gasket is carrying-on the external loads.

Both Blick method and Kellogg (EQP) method are counting in this way the gasket reaction.
In my opinion none of them counts the stress in flange, rather they are dedicated to count leakage danger.
This opinion may be in contradiction with the common opinion and I try to argue on this subject.

Taylor Forge and ASME VIII are methods counting flange stress and "J" index is a last hour improvement. The only problem would be the way in which we are counting "J" index under pressure and external loads.


A. Kellogg EQP method counts the gasket reaction in terms of equivalent pressure.
For tensile force this "transform" procedure is easy: a tensile F force has the same effect on gasket as a pressure:
P1=F/(PI/4*G^2)=4*F/(PI* G^2).
For a bending moment, the peak effect on gasket is a force of 4M/G magnitude and that force has the same effect on gasket reaction as a pressure:
P2=(4M/G)/(PI/4*G^2)=16*M/(PI* G^3).

So the total pressure is done by:
HG= W - Pt *PI/4*G^2
and the result is
Pt= operating pressure+ 4*F/(PI* G^2)+ 16*M/(PI* G^3), under the assumption the external F and M act on gasket, while bolts are not participating

The condition for "non leakage" is to maintain this gasket reaction "big enough".
Kellogg fellows have considered this condition as "gasket load big enough means over the gasket load in rating pressure", i.e
W - Pt *PI/4*G^2 > W – P_rating*PI/4*G^2

So Pt< P_rating.

I guess the simplicity of this condition explains the success of this approach, while the method is very conservative. In rating pressure status, the gasket can be considerable tight, not near the leakage! Rating pressure is not a real "leakage" limit for gasket; instead it is a conservative one.


B. Blick method follows the same approach but the gasket "leakage" condition is more realistic.
For the case the tensile force F is missing, the gasket reaction is
HG = W -p*PI/4*G^2 -4*M/G
and is considered greater than
2*b*PI*G*m*p
(which is the condition to have at least "mp" stress on gasket, the same as considered by Taylor Forge- ASME VIII)

This condition leads to:
M < G/4*W -PI/16* p*G^3 -PI/2*b* G^2*m*p
or
M < G/4*Sop*Ab-PI/2* p*G^2*(G/8+m*b)
with Sop "operating" bolt stress.

Many quotations of Blick formula replace Sop with Sb.
The truth is that Sop (I’ve maintained the original notation from article!) is the central point of the problem, not only here, but in every serious method -including FEA, where the user must provide a bolt initial stress, or -alternatively- the initial bolt load. Maybe a value near 1.5Sb is more realistic as stress after bolt-up, but also realistic is to think we may have a variation for an individual bolt load, let's say of plus or minus 25% for one bolt load by manual bolt-up procedure, so finally for a single bolt it would be as 0.75*1.5Sb, let’s say Sb!

I would appreciate Blick method as realistic; however it may be still conservative because the gasket is counted as reacting under loads while the bolt load remains the same.

The formula explains qualitatively why we haven't leakage for most of practical cases; it is likely that we have in practice Sop more than Sb (remember ASME evaluation Sop=45000/sqrt(dia_bolt) in psi) and an experienced worker shall apply an uniform gasket tightening....

C. ASME VIII Div1 is the Taylor-Forge method. ASME VIII Div2 is also Taylor-Forge method modified by Koves. Both of them are specifically written for flange design case and are intended to count stress in flange.
Codes take into consideration two limit states: seating- bolt-up status and "design pressure status" and consider two minimum gasket load HG1 and HG2 for each case. Scope is to evaluate the flange stress in these cases.

When using Code for flange checking, the common perception is that for every pressure the gasket reaction is 2*b*PI*G*m*p. In other words, the Code requirement to have at least 2*b*PI*G*m*p_design -as gasket reaction for p_design- it is currently interpreted as valid for every pressure as 2*b*PI*G*m*p.
If the last were true, in p=0 case HG would be zero, while the actual gasket load for p=0 is that done by bolt-up procedure!
So I would consider that the real meaning of 2*b*PI*G*m*p is that the minimum required gasket load under pressure criteria can be taken as pressure proportional, since the minimum stress on gasket must be at least "mp" to have no leakage.

Probably the best description of this problem has been done by Blick himself, on the first page of Part I: "It should be noted that despite the fact that the code goes to great lengths to avoid defining the gasket load, the implication is there"

So my personal remark is an warning about applying Code for checking flange purpose by using a gasket load which is not in equilibrium with the rest of forces; if you think "this is the Code and I don't care about what it is" you should remember no Code can ask you to consider something out of basics. And one basic aspect is the gasket load is decreasing with pressure (p) and external loads (M and F) for most of practical applications and this is the main reason for which the leakage appears!
One possibility to interpret correctly the Code for flange checking case is to consider a gasket load expression.
It may be considered as HG = W -p*PI/4*G^2 -4*M/G-F, with the same W in bolt-up and under loads; it can be proved that loads and reactions are now in mechanical equilibrium.
This is in fact Blick's approach.
However one can keep in mind that this is "worse gasket scenario", while the actual bolts-gasket behavior can be caught with more accurate methods as EN or FEA.

Best regards.

[Priyoyo]

tony black thank's

[mcburns]

Thanks a lot for the links and explanation posts

[mariog]

Hi, Tony!

Thanks for the posted articles and for your work.

Do you have also an article detailing a flange calculation model by considering "compression set"?

Thank you in advance and my best regards.

[mkhurram79]

thanks