Just been using the flange checker as part of some work I'm doing (not used CAESAR II in a bit) and I think there might be an error in it.
My first reaction was there is something wrong here - no way are those flanges overloaded by that much, but a bit of digging revealed the following (I think)
It appears when you're working in SI units the Peq module pulls through the gasket diameter in inches but still evaluates the loads, especially the moment, in the user units. So in my case we have Nm being evaluted against inches and not mm.
So for one typical node the output was
Node----F-------M-------G/C-----Peq-------Temp---Pallow
--------N-------Nm------mm------bar-------C------bar

1270----980-----984-----13.82---19151.87--130----91.49

Note the Peq is over 19000bar (277690psi!!) on a 600# flange!
A quick glance and check showed that the G dimension was still in inches and not the correct equivalent in mm. since this figure gets cubed or squared this is a big big difference. The correct figures I think are
G---------P_bend----------P_axial-------P_oper----P_eq
mm--------MPa-------------MPa-----------MPa-------bar
351.028---O.1158----------0.01012-------8.1-------82.2

Thats more like it!
Can anybody spot any flaws in the logic and I'm not sure if the bug has been fixed yet.

So as ever it's always helpfull to know when the answer is wrong and don't beleive everything the 'puter tells you!!

That may explain why Peq results are always much higher than the NC-3658.3 method. I always use the NC-3658.3 method, and even then I find that is conservative. If I am ever under 150% of failing with the NC-3658.3 method, I usually use the flange module built into Ceasar to further evaluate it, and it always passes. It is a pain though to put all of that info. in, so I usually just change my pipe slightly so I don't have to do that lol.

What version of CAESAR II are you using? Can you send us the job file so we can review this.

Mr. Threeouts,

Can you please provide the formula to calculate the total bolt area calculation for the NC-3658.3 method.

I take the root area of the bolt, and multiply that times the number of bolts in the flange. So for instance, a 6" 150# flange will have eight 3/4" bolts. A 3/4" bolt has a root area of 0.302 in^2. Multiplied by 8, you get a total bolt area of 2.416 in^2.

I make it routine to put a flange check at every flange in my model. It is easy to check, and you never miss a possible high bending moment that could cause leak.

And remember, flange checks propagate to downstream elements just like everything else does in Caesar, so it is very easy to carry over values to other flanges, but watch out and be aware when you change the number of a flange upstream it changes all flanges downstream, unless you have manually changed those numbers.

Hi Richard
It's Ver 5.30.0 (build 101122) so I appreciate it may not be bang up to date. Part of the post was to raise the issue (if it hadn't been already addressed) and part to be a general warning about knowing when the results are wrong (engineering experince and all that)

You should update as first build had some bugs that were corrected later in the next SP.

Regards,

One thing that I noticed is that regardless the versions, many problems were due to the use of custom units file.

Regards,

Thank you very much threeouts. One final request, do you have the formula to calculate the bolt root area. If so, can you please provide or the reference of it...

http://www.mp-sealing.com/assets/files/corr_handbook.pdf

see page 40 of 60

Yep danb. The equation for the term 'Ar' Root Diameter Area matches the numbers I use. I also see they provide a calculation for 'As' the Tensile Stress Area, which gives a bigger area. Not sure where they use this term at, but I always used the Root Diameter Area. It is more conservative anyways. But to reiterate in case the link gets broken:

Ar = 0.7854 x (D - (1.3 / n))^2

Where,

Ar = Root Diameter Area (in^2)
D = Nominal diameter of bolt (in)
n = Number of threads per inch

So for a UNC 1-8 bolt, Ar = 0.551 in^2

The NC-3658.3 equation takes no account of flange body stiffness or gasket type and pre-load requirements. How can it be considered a 'leakage check' ?

I was surprised to realize the formula has been establish by Rodabaugh and Moore in a work with title "EVALUATION OF THE BOLTING AND FLANGES OF ANSI B16.5 FLANGED JOINTS"

It can be seen in various sections of their work that this equation "is a reasonably good approximation of the moment limits calculated using the computer program FLANGE."

I'll quote again from their work dated September,1976:

The development of Equation (15) is described in Appendix B, where it is identified as Equation (B5). As discussed in Appendix B, Equation(15)
(a) is conservative with respect to available test on joint leakage,
(b) is a good approximation of the moment capability (as limited by joint leakage) as determined by the computer program FLANGE with an assumed bolt prestress of 40,000 psi,
(c) except for the 150 class, gives permissible moments essentially equal to or higher than permitted by the rules in the present NB-3647.1,
(d) tends to discourage the use of the larger sizes of 150 class joints by giving a small permissible moment.

So it seems to be a result of a serious work (performed with 70s tools) however presented as an empirical simple formula.

My best regards.

Thankyou very much Danb and Threeouts. For the valuable information.

Use of the NC-3658.3 equation is probably ok provided ALL the other contingent factors in ASME III are also complied with. Cherry picking equations and applying out of context is always dangerous.

There is no reference to flange stiffness or gasket performance in NC-3658.3 because one assumes, ASME B16.5 components are anticipated. There is no warning anywhere that it may not be appropriate to apply to much weaker and thinner flanges, DIN for instance or to custom designed flanges, compact flanges etc.

Also, one wonders about the computer program used in 1976. Bearing in mind that API's (1980's I think) first attempt at decent stress analysis of flanges, API 6AF1 left much to be desired being axi-symmetrical. Their later work, 6AF2 used 3D models, accepting that moment loading of a flange is far from symmetrical. What basis was the 1976 analysis ?

Thank you all for this information.
Would you please tell me where I can find the term n=Number of threads per inch ?

http://www.mp-sealing.com/wp-content/uploads/2015/04/CORR-A-SEAL-Handbook.pdf

the old link is not working anymore.

I guess that you can find a lot of links on internet as this number is standardized.

Thanks a lot Dan,