Hi All,

In older days people were used to A0 size stress isometrics; nowadays, A3 size is generally preferred, the number of them running into near 100 figure in a single stress package sometimes, the number of elements as evident from caesar II element listing reaches near to 1000. We know people are getting smarter, with robust softwares available with them. Does not the reliability of result of piping stress analysis suffer due to such a size ? If some critical restraint is not built as designed, it can affect the whole system compising many equipments connected through the piping. What size is a large size today, beyond which we should break it up by putting inline anchors ?

regards,

sambhu

What kind of systems are you modeling to required that many nodes to complete? I think the largest model I've ever done has run to a few hundred and that was pretty big.

Dear Klein,

Thanks for your response.

In a typical process plant, steam piping of various low/medium pressures/ temperatures going over pipe rack can be one such instance where we may have such big caesar II models. In a critical application of safety related nature, we do not
allow to have such large size.

regards,

sambhu

We occasionally see models upwards of 4,000 elements. While we definitely do not encourage this model size, it is possible from a computational point of view to create and analyze much larger models with no loss of accuracy in the results.

But there are some very real problems with large model sizes. First, it is difficult to render such large models in the plot utility. The new HOOPS Graphics engine is quite memory intensive and for large models you may need a serious upgrade to your computer's video card to speed up the rendering. Also in the plot, with so many elements, it is difficult to display all the information such as restraints and node numbers. You will be forced to zoom in on small sections of your model to view any significant detail. However there is a search feature in the graphics that allows you to find specific elements rather quickly. I think as future improvements to our graphics capabilities come online you will be able to more effectively view these larger models in the plot.

The second concern with large model sizes is human limitation. It is simply easier to make a mistake on a larger model and not catch it during your review process. Fatal errors, especially the dreaded coordinate mismatch can be very difficult to track down in a large model. And although everyone knows they should back-up their work, I have had at least one case where a 4,000 element model had become corrupted and there was nothing that could be done to recover the data. This engineer had to start from scratch re-creating this model!

One mathematical limitation occurs with huge models containing many nonlinear restraints (such as all the +Ys I assume you use in modeling your pipe rack). The more nonlinear restraints and friction you have in your model, the more likely it will be to not converge on a solution. Then you will have the task of changing friction tolerances or even modifying restraints (linearizing a few of them usually does the trick). In a large model this modification of restraints can become a real pain.

One more note: It would be far better to break your model into two or more smaller models at hard anchor points. Breaking at an anchor will not affect your results. If you truly have no anchors then you can model this in sections and connect them together during the analysis level by using the piping include feature. Currently you can create and include 10 piping models together. If when you include the piping files you leave the ReadNow flag to N, then these files will only be included during the analysis itself. They will not be available in a combined plot. This is nice for larger models because you can review each section separately (you must error check each of the included models before running the analysis on the main file when the ReadNow flag is set to N), avoiding many of the problems mentioned in the previous post.