FEA of fiberglass tower

[HPsteam]

Hi.

I am working for a small engineering company where we design steel tanks and exchanges on daily basis. All thermodynamic and mechanical calculations are done in house. Every once in a while, we will make an FRP tank for use in highly corrosive environments. At that point mechanical calculations are part of fabricators scope (do not have our own shop) and we go by whatever they say. No known structural issues on equipment supplied to us since we started working with them.

We recently fabricated packed tower and supplied hand calcs with it. We are in a situation now, that customer is asking for FEA analysis of the design as well, i guess as a proof of hand calculations. I am reading everything i can find online regarding this issue, and it seems that the consensus is that every layer would need to be modeled separately and its properties properly assigned. For hand calculations, material has been assigned with certain properties and was taken as isotropic material. Calculations were pretty much straight forward after that. I am assuming that the mechanical properties of the material were taken from the actual test of their process.

My question is following: During FEA analysis, how close can this approximation be to the results obtained by modeling the right way? I understand all of the issues with composite materials and actual fabrication.

IN any case, we are going to hire third party engineering company to do this for us as we do not have experience in this kind of FEA analysis. I am planing on performing FEA (SolidWorks) on my end as a learning tool, so i am wondering if i should go ahead and design the tower in all the little details, or just take it as isotropic and go with it?

any help is appreciated. thank you

 

[rb1957]

each ply should be present in the model, but that does not mean modelling each ply. Codes have laminate element properties that allow you to express the "grain" direction of each ply. Of course, this means this is specified on your drawings in the first place.

another day in paradise, or is paradise one day closer ?

 

[ShellsPlatesMeshes]

In order to accurately evaluate all the properties of such object, you definitely need to use FEA of multi-layered shell (for very accurate and flexible modeling you may want to consider some in-house software development). Going with just isotropic model would, most likely, lead to incorrect evaluation of all properties, including deflections and maximum stresses.

 

[adampar]

I agree with the other answers. FE programs have the capability of modelling laminate materials both in terms of 2D and 3D elements (if the out-of-plane stresses are of importance to you). You can just enter the material properties of a lamina in its reference orientation and the sequence of the plies, and the software automatically calculates the [A],[B},[D] matrices and solves the problem. Meshing also tends to be a bit difficult in the case of complec design, so use your intuition and make your life easier by making preservative simplifications in your model.

 

[HPsteam]

Thank you for your replies. There are several difficulties with this project:
1) orientation of layers is not monitored and seems not important in this case. It is basically a cylindrical tank that is put on "rotisserie" and cloth is being rolled onto it.
2) the layers are not made out of same material. They seem to use orthotropic band (i assume it is typical woven fiberglass cloth" and something that has fibers running in every direction, i guess, trying to simulate isotropic material. location and sequence of layers depend on the location on the tank.
3) There is not mechanical report for each layer separately. The best what we can have is tests after all layers have been applied and resin hardened.
4) fabricator explained that good number of calculations have been approximated with material being isotropic. And some parts calculated using Tsai-Wu criterion. I know that this is not the right way, but we are not making airplanes, just tanks to hold water. I assume that there is a pretty big factor of safety included in this calculations. Obviously, the method is proven, as they do not have any failures, as far as i can remember.

not connected to my issues above, i have a question about transitions. in this case, were two parts are joined (for example, straight tube and conical head), and certain number layers are overlapping both parts. how do you take care of that during FEA analysis?


thanks