I hope every Chinese engineer here could read this post and talk something more about CII and Chinese local stress analysis software.

I did static analysis for a main steam pipeline with both CAESAR II and GLIF. The piping layout, material properties, every restraints, displacements, spring hangers, cold springs, and other boundry conditions are all same modeled, but the results of these two softwares are quite different.

I changed the elastic modulus to the hot value and applied every hot loads of spring hangers as the GLIF output displayed, but the CAESAR output of hot case(OPE) is still greatly different from GLIF.

Now I am really confused about this problem. I'd like to know what do you do if you have the same situation? Which one would be your choice of the analysis result?

I do belive in CAESAR II for its excellent performance all around the world. But GLIF has also been approved to be quite safe for engineeing practices over decades in China.

Could anybody be helpful? All kinds of response would be greatly appreciated.

Should I do an extra MANUAL calculation for every pipeline to get the last result?

Well, you have stumbled into the art of "benchmarking". This is difficult to do with a real model if you're not intimately familiar with both programs.

Step back and start simple. Here are a few points to get you started.

• Start with a simple cantilever. Apply a tip load and a tip displacment. Then setup three load cases, weight, force, and displacement. Do both packages yield the same results for all 3 cases? (They should.)
• Now add a bend and another piece of pipe. Repeat the above analysis. Are the results the same? Are the SIFs the same?
• Now add some significant pressure to the model (and to the load cases) and repeat the run. If the results differ, check how "pressure stiffening of elbows" is handled in both programs.
• Now add some significant temperature to the model (and to the load cases) and repeat the run. If the results differ, check the value of the expansion coefficients used.

If everything matches to this point, look at your real model. What does it have that these tests don't? Some things are handled differently by different programs. For example:

• In your real model, what results are different (displacments, element forces, or stresses)? If just stresses, then there is something different in the implementation of the piping code. If the displacments differ, then ...
• In CAESAR II, the stiffness of rigid elements is determined by keeping the ID constant and increasing the thickness by a factor of 10. Some other packages simply multiply the elastic modulus by 3. If the system is sensitive to rigid elements, you'll get different results.
• In CAESAR II, expansion joints have a finite length. In many other packages, expansion joints have zero length, so you may get different results in the area around the joint.
• What is the default stiffness used for restraints in the two packages?
• Does the model include non-linear boundary conditions? If so, you need to find out how each package treats them.

This should be enough to get you started. Also, there is the potential that someone made an input mistake in one of the packages. For example, corrosion in only one model can result in wildly varying results ...

Good luck. If you need to discuss this in detail, send me an e-mail.

Dear chiu,

It's not a new problem, CII and glifw will give
you a different results, In some modules, especially in some simple modules,the difference maybe small, in some complex modules, the difference maybe be quite great.

To do manual calculation is not a good way ,i think, too hard ,and manual calculation also are not very precise.

if you use CII 4.5, and you change the elastic modulus, applied every hot loads of spring hangers, Some difference maybe still there:
1)configure setup
2)the simulate of reducer,elbow,tee,
3)you may should not only use the same hot loads of spring hangers ,but the spring rate,
4)material, remember, the CII use the most new data, but glifw may use the data before 1998 B31.1. The expand coefficient have some difference,
....
Try to make all the things the same! not an easy things.

and ,if you change the elastic modulus to hot muduls, CII may use it not only in hot case, but also in cold case ,but gilfw only use hot muduls in hot case!!!
so, Try to make all the things the same! I think you may be can't succeed.
I think the elastic modulus may be one of the most big difference of this two software, when it come to 4.5 version and before, CII use cold modulus.

Best Reagrds,

The B31 codes require the use of a cold modulus of elasticity for displcement stress analysis, plain and simple.

CAESAR II follows these B31 code requirements (B31.1 & B31.3). If GLIFW does not use the cold modulus and yet claims to dfollow the B31 code(s) I would have to question its results. The bottom line is, before simply comparing programs first read what the code you are working to requires, then make sure the software follows those requirements.

As for hand calcs.... I am probably one of the last people around doing legacy "flexibilty analysis" I can still even do Spielvogel!

Thank you, Richard. Thank you, whm. And thank you, John. All of you gave me a lot of things to be considered for a detailed analysis. I know some boudary conditions like different elbow/tee/rigid-element/reducer treatments will lead to different analysis results. But I don't think the difference could be so great. The other boundary conditions seem to be the same for two models.

Someone told me before that the simulation of the two software may be different. I just want to know exactly what kind of difference between them. After the simulation it made me really confused about piping stress analysis, because the stress, restraints, displacements are significantly different, though both assumptions and output claims to be resonable.

To whm: I used the same spring rate for spring hanger design, and the material properties are inputed in Caesar II just as those values in GLIF. But it's not adequate for building same model for Caesar II and GLIF, because the springs they calculated are still different (one of the two elastic modulus is not proper under HOT and COLD cases). I don't think it's necessary to compare the two program output and tell which one is correct because the assumption and processing method are different.

To tell you the truth, our customer would be very pleased if I can finish the engineering design in a couple of days and give them the final drawings for construction. They don't care which computer program we used for drawing and computing. They just want to put their machines into operation. Each one of the two software could be my choice for analysis and can be easily approved after the model/result is checked by the technical stuff of our institute. It's not my time today to do detailed analysis and compare the results of the two computer program. So firstly I think I'd like to do just a typical analysis to get a resonable result and finish my design before the so-called deadline. Then maybe I could sit down and read some detailed books to make things more clear.

I'd like to contact you for some question, but unfortunately, I don't know how to send you email within this forum. I cannot find your email address.

I hope everybody could share their idea about this problem. Thank you in advance.

Are there spring hangers in the analysis? We added a few configuration switches to the Computational Control for a Chinese user. Perhaps he, too, wanted to "match" such a program. Maybe these switches come into play - ignore spring stiffness & add spring stiffness in hanger travel calculations.

I have to agree with Rich. Start small and then work in more and more possibilities for differences. It is a slow process.

My email can be obtained from either Rich or Dave... Richs Email is ray-at-coade.com Daves is ddiehl-at-@coade.com, Substitute @ for the -at-


If I was in your shoes and hand no ultimate repsonsibilty for code compliance I would use the program that gives the answers I like the most.

However if you absolutely MUST comply with B31.1 or B31.3 I would use CAESAR II, it is used on a world wide basis by thousands of users, it has a rich QA manual that can be purchased, and it has partcicpation of some of its people within the ASME codes thereby assuring its ever green relationship with the constantly renewed codes....

If GLIFW had those credentials ( I've never heard of it) it would be tougher choice.

So have fun!

Thank you for your kind reply. I don't think it's necessary to compare the two computer programs because they have different philosophy as I mentioned before. Additionally, I don't have much detailed documents for GLIF.

But I would be very happy to share your ideas for piping stress analysis.

Best Regards,