Concrete Finite Element Procedure - Explicit Rebar vs. Flat Plate 2

[MegaStructures]

I have a unique reinforced concrete non-building structure that I will analyze using an FEA model. The structure is basically a rectangular tube with two side walls a slab roof and a floor. I have one design example to follow and the design uses what appears to be beam elements for reinforcing bar and solid quad4 elements for the concrete. I think this is very complicated and introduces a higher probability for user error and requires the use of a more powerful FEA program other than a SAP-2000 or Risa that is a little easier to use for us structural engineers.

My question to all of you is: Are you aware of the research behind SAP-2000 and Risa's method of plate modeling for reinforced concrete slab design and is there any research available that directly compares the accuracy of this method over the more detailed method of explicitly modelling each reinforcing bar?

 

[JoshPlumSE]

I'm not aware of any research justifying the use of a plate model over a FEA model that includes reinforcement.

But, I am aware of what the standard industry practice is for MOST structures. You model concrete slabs with plate elements. Maybe with some "equivalent" reduced thickness or such to account for cracking or such.

If you were doing a PhD thesis project, then maybe you would use the complex reinforcement model. But, there is no guarantee that would give you model behavior that is closer to reality.

Also, the goal (as a structural designer) isn't to produce a model that is closest to the actual behavior you'd get if you tested the structure to failure. Instead, the goal is to get a safe structural design that is compliant with the governing code (ACI-318?). I'm not sure that ACI has any guidance on modeling reinforcement in an FEA model, but they do have guidance on how to model an equivalent elastic stiffness for your various structural elements.

 

[MegaStructures]

That's a very helpful response JoshPlumSE. I was hoping for a comparative analysis of flat slab models made with 1) plate elements 2) explicitly modeled rebar and 3) physical design results, but I did a pretty thorough literature review and found nothing even close. I went to Risa's website and also found no research on the topic. It seems that there is a plethora of research on the use of explicitly modeled rebar and complex nonlinear properties of concrete, but no research comparing these complicated models to simpler ones. Maybe if I do a PhD someday that could be a possible area of research.

Unless I'm missing something it seems that the design of this structure using plate elements, which SAP and Risa offer would be allowed per ACI 318. Although, really almost any FEA analysis is allowed, since ACI's only guidance on the topic is "the licensed design professional should ensure that an appropriate analysis model is used for the problem of interest. This includes selection of computer software program, element type, model mesh, and other modeling assumptions." per ACI R6.9.1. But really I mean that the ACI nominal design capacities are based on calculated shears and moments, which a plate analysis would provide. The more detailed FEM would give stresses in rebar, which ACI provides no guidance to interpret these stress (I assume a simple von misses stress comparison would work).

JoshPlumSE, I remember seeing in past posts that you used to work at Risa. I'm not sure if you have also done design engineering, but if so how did you enjoy your time at Risa compared to design work? I've always had a strong interest in computer applications in engineering and wonder how it would be to work on the development of some of these programs.

 

[JoshPlumSE]

I remember seeing in past posts that you used to work at Risa. I'm not sure if you have also done design engineering, but if so how did you enjoy your time at Risa compared to design work?

Yes, I worked for RISA for something like 16 years. For the most part I loved my time there, until close to the end when the company was in the process of getting acquired by Nemetschek.

Before I worked for RISA, I spent 5 or 6 year working for Fluor Daniel working on heavy industrial projects (as well as Fluor's in-house programs). Then after I left RISA, I worked for a year at a company that does many OSHPD / healthcare projects. Now, I'm back with another analysis / design software company (CSI, the maker of SAP and ETABS).

Regarding your structure, I would be surprised if the use of plate elements were NOT allowed. If it's an odd situation (like Table Top Foundations for Turbines in the power industry), it might be more common to use solid elements. But, that's largely an industry specific practice for a very specialized structure.

I suppose there could be another very specialized industry where they really want to model the reinforcement as well. Though I haven't seen it. Even with nuclear work.

 

[MegaStructures]

This work happens to be it's own niche and there is no standard. That is why I'm taking such a hard look at the one research paper I can find and making sure I don't miss anything important when deviating from the steps the researchers took when completing their model. I think all logic dictates that rebar doesn't need to be modeled. Even if solid elements are necessary principle stress outputs can still be used to check rebar by hand

 

[IDS]

I think this is very complicated and introduces a higher probability for user error and requires the use of a more powerful FEA program other than a SAP-2000 or Risa that is a little easier to use for us structural engineers.

I totally agree, and even if the model was completed and analysed without error, there is no guarantee that the results would be closer to the actual behaviour of the structure than a plate model with stiffness adjusted for post-cracking behaviour. In fact unless the reinforced model was almost impossibly complex a plate model with smeared crack behaviour would probably give better results.

But unless accurate prediction of deflections is important even that is over complicated. A linear elastic model with adjustment of peak stresses in accordance with the code would be both simpler and more relaible.

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[WARose]

My question to all of you is: Are you aware of the research behind SAP-2000 and Risa's method of plate modeling for reinforced concrete slab design and is there any research available that directly compares the accuracy of this method over the more detailed method of explicitly modelling each reinforcing bar?

Me personally, I've never seen reinforced concrete modeled with the rebar. (In a normal, non-research project.) SOP everywhere I've worked is: you model it with plate/beam elements.....and you vary the modulus of elasticity to get the stiffness you want. (For plates. For beam/stick elements, you just used a reduced "I/K".)