Finite Element Analysis and BIM modelling

[pmok]

Hi All,

The BIM space is growing stronger and stronger these days. The early faults/problems there were are getting solved in a fast pace. Hence, it has come to my attention that the so-called "BIM workflow" will become a necessity in the near future. I am currently evaluating the right software(s) with respect to BIM for my employer.
I understands on a global scale the virtue of BIM.

I am a tunnel engineer who does both structural and geotechnical analysis and design on a day-to-day basis. My primary focus is in soil-structural interaction modelling. I never believe in the separation between the modeller and the designer. From the art of the "black box", we all know we don't always analyse the structure to how it actually looks or its true dimensions. We often analyse a local piece of a structure rather the full scale/entire structure. More importantly the modeller needs to know what the analysis is supposed to give to facilitate the design. Hence, the designer is always the best modeller. I have come across a number of engineers who rely on drafter to draw the structure in AutoCAD to a specific format such that it can be imported into the numerical software for analysis (though I really disagree with this). The question was then "what if the structure drawn on CAD to its true dimensions can be automatically converted to 'engineering' dimensions and imported into the numerical software directly?" "If so, the conversion process can be reversed and permit detailing and drawings to be produced from the re-import of the analysed model?" From my understandings, the BIM concept is the answer to these questions in a telescopic scale of our day-to-day work activities.

My understandings of BIM in a global scale is the integration and control of multidiscipline information in a single space which is expansive in nature. However, this applies in general to large scale major civil projects from my point of view. I am more interested in the day-to-day workflow. From my research done to-date, I realised most of the BIM tools have been built around the interface development with AutoCAD or Revit (both are Autodesk's products; Tekla excluded of course). This does make sense to me since the industry has been using AutoCAD for a long time and it is the reason why I think Tekla is out of consideration. Though I have heard many problems/complaints on Revit regarding its limitations. I have had a bad experience of my own with Revit with the Revit Modeller not able to produce useful drawings and I was not able to recover the cost from the Client because of this. What good can a fancy 3D model does if it can't produce a 2D drawing on a piece of paper that construction labours can use on site to build things? Though from my research it does appear that it may be able to produce proper 2D drawings. If anyone can shed some light on this it would be great?

Autodesk offers a structural package called Robot Structural Analysis which is full integrated with Revit. However, it seems to be frame base type analysis only. It doesn't seem to offer the FEA capabilities I am after. I have learned that SAP2000 now offers Revit integration but is uncertain of the limitations or the extent of what it can do. Though I understand that SAP2000 is one of the more popular or recognised structural package in Europe.

I have also recently discovered the softwares family Sofistik. It basically built its CAD capabilities as a interface products to either AutoCAD or Revit. It also offers a Structural package called the Sofistik Structural Desktop which appear to include a series of FEA softwares that is capable to carry out soil-structural analysis to the extent like Strand7 or SAP2000 but may be a bit more advance.

From the above, it appears Revit is well ahead in the BIM space. But given the bad experience I have had with Revit. I am uncertain if Revit is the right package to invest in. Given that developers for SAP2000 and likely other well known softwares are also building their BIM integration ability, should I invest money in a more BIM ready structural analysis package like Sofistik Structural Desktop?

 

[glass99]

What options are there for BIM other than Revit?

I think Revit has Robot built into it if you get the option. Personally, I don't like excessive automation of structural analysis. If every beam is not looked at by a human being, the computer will miss something.

 

[Brad805]

We use Tekla Structures for our BIM modeling. We own Revit as well, but I find it limiting to create an accurate model. Plus I get annoyed by all the add-ons you need if you want to create an accurate model. We are looking at a wood framing add on for Revit right now that is $6k, plus annual maintenance.

We own SAP2000 and ETABS. CSI and Tekla have been working well to keep the analysis links up to date. Bentley bought RAM a few years ago, so I doubt the link to Tekla will be updated anytime soon. I suspect the ROBOT link will be about the same since Autodesk purchased that. Risa is has a baisc link, but it is not the best for analysis. Using the current Risa link you can import the basic frame geometry from tekla, but not the Tekla analysis model. Once you have used Tekla for any period of time you will find the view port controls make analysis model editing a bit easier. S-Frame has a link as well, but I have not tested it.

Regardless of the BIM solution you decide on, do not underestimate the time to implement this into your project flow. Regardless of the choice you make each option has a ton of settings. We did not heed to the advice when we started into BIM in 2007, and it was been a struggle at the start. The other pitfall with BIM is where do you go to get modeling staff? How many technical schools do you know teaching it so far?

 

[glass99]

brad: i think of tekla as a package better suited for steel and concrete drawings, including production of shop drawings. I hear about fabricators using it. Does it do architectural bits like partition walls/light fixtures and HVAC? I have heard good things about TEKLA as a structural engineers tool though.

I agree that Revit is hugely overrated. I see it way more as a bureaucratic hurdle than an exciting technology. In principle its awesome because it does coordination, but the reality I see is it makes it harder to do structural engineering.

 

[Brad805]

glass, yes, I agree Tekla is best suited with concrete and steel. We have modeled wood buildings using Tekla as well, but it would take quite a bit of time to create all of the custom components that one would need if you want to model all of the framing accurately and quickly. The Revit package we are looking at for wood construction is quite flexible and will fit into our work flow better.

Tekla is not great for architectural. We have endless problems with Architects curtain walls, overhead doors, and other misc items. We typically show these items as simple polygons where they are critical to our structure. We had started creating custom components for those as well, but it was a bit too too time consuming for our purpose. If there is something critical we want to show we can import a sketchup file, IFC, Solidworks, or 3D cad file.

Tekla is not the best for mechanical aspects. If you want to create fittings or anything other than simple shapes, this is not the tool. We have added plenum openings, water closet penetrations, duct mains, and a few other items in the past, but the Revit MEP package is far more convenient. We did create a number of grain handling parts recently as well to check motor fit and alignment issues next to grain bins and handling equipment, but it was a bit time consuming (solidworks has made this easier). For our commercial projects we are starting to get mechanical Revit models so we typically create an IFC file in Revit and import that. That works great (the attached pic is for a small portion of a current job).

I think Revit can be made to create a good model, but most clients do not demand it or know what is truly critical to ensure things fit correctly when they arrive. Revit has quite a few options that allows the user to make up for modeling shortcomings. I understand why this is, but at the end of the day I am not sure it is the best route to improve construction. Tomas Thornsetti (completed the NY Stadium) works in Revit and Tekla. I was speaking with one of their lead modelers and he mentioned it took them years to train their architectural modelers to model things accurately enough to allow them to start the detailed steel model downstream using the architectural model as a base in Tekla.

Returning our projects to a Revit team when we are acting as the EOR is an added step, but I think it is worthwhile.

 

[glass99]

I have heard of Tekla being used for curtain wall type projects. A colleague designed a large domed skylight with a steel structure in an arabic pattern recently. Span ~200ftx200ft or so. All custom milled nodes, not just standard moment connections and gusset plates. I know Thornton Tomasetti made a huge deal out of using Tekla for that stadium.

We have been scaling up on Solidworks for intricate stand along objects like structural glass hardware, though have seen whole buildings modeled in SW. It does work for sure, and the true parametric nature and part based model structure make it super for mechanical component design. The reality of SW for me is that because no one else in the construction world uses it, we can only use it for stand alone objects.

Brad: is it more effort to do structural engineering for a conventional building structure in Revit than in 2d ACAD?

 

[Brad805]

If the workflow is setup correctly I think the engineering can be more efficient. The problem we have is the lack of modeling staff. If we had more technicians to complete the modeling end of things it would work very well. The advantage is the technicians actually create usable information for the engineering staff. We can easily create the basic frame geometry from the model provided it is done correctly. After analyzing in SAP or ETABs it automatically reports the changes back to the model and then the technicians can complete the work. Now the way they talk about the sharing of data back and forth does have a couple limitations, but it will work. I see one of the biggest problems with BIM right now is the staffing.

 

[JoshPlumSE]

The term "BIM" is really just 3D multi-discipline modeling. We had been doing it in the petrochem world (with Integraph) for years before it became a catch phrase in the commercial building world.

If you're using the term BIM, then you're probably talking about Revit, Bentley or Tekla and the FEM programs that work with them will all be targeted towards buildings or bridges. Not much mechanical FEM packages will interact with them (at least not that I'm aware of). Robot, RISA (my employer), and CSC Fastrak have all put some effort into their Revit interaction. But, they're all Frame based or building based FEM packages. In a similar line, I imagine that Bentley's building / frame packages (RAM, STAAD) have put a lot of effort into working with the Bentley BIM system.

To my knowledge SAP / CSI / ETABs has not made much of an effort to be Revit friendly. The Revit guys certainly haven't made it easy with their constant re-working of the API functions and such. So, I don't blame them. But, if you look into SAP then you should check with them to see which versions of Revit they support.

If you're doing a heavy duty soil analysis, I don't know what type of FEM program you'd be doing for. If you're doing mechanical work, I'd imagine that there are FEM programs that work directly with AutoDesk's Inventor program. Not sure if those models can be pulled directly into Revit or not.

 

[glass99]

My experience with CAD to FEM interface is that it takes quite a bit of effort to clean up the CAD geometry before you can use it in FE. I'm not super clear on how it would be "automated" without the person doing the drawing being the analyst. I could see Revit being smart enough to know that for steel framing for a really ordinary floor structure that it had to connect the beams into nodes, but its never really that simple. Even for a simple building, the calculations always have some degree of fudge to them at analysis time.

 

[glass99]

Experienced BIM people: how do you guys deal with "fuzzyness" in information? My experience with Solidworks is that it works well in a manufacturing environment because everything is very tightly defined geometrically, and then manufactured to tolerances that are frequently measured in 1/1000ths. SW relies on this precision, otherwise the whole thing falls apart, for example stacked relationships between parts get nutty if with tolerances. Construction is critically different to manufacturing in the sense that:
- There are field conditions, many of which are unknown, especially for a renovation project
- Concrete and foundation tolerances frequently exceed 2"
- There are a bunch of design build elements which are not fully defined at construction document time.
- Can you ever write "Verify in Field"?

 

[Brad805]

With BIM you need to decide your goals with respect to accuracy at the start of the project. One cannot really compare Solidworks with typical BIM applications. With SW one can model things perfectly. You can even go one step farther by adding Abaqus and analyze it to death. This level of detail would not work on a large scale project with a team of consultants and owners that do not know what they want. In BIM projects the Architect tends to start the project by defining the model level. 1/4" or 1/8" accuracy is not uncommon. One can still add all the notes about verification unless the contract defines otherwise. If that is the case 3D laser scanning can be appropriate, but 3D laser scanning leads to a whole host of new problems.

It seems you might be thinking BIM is more than it actually is in reality. One still delivers the same type of drawings with all the typical disclaimers one would use in a 2D drawings. The usefulness of the models boils down to the qualifications of the people creating them. It really is no different than typical 2D drafting, but with BIM your giant file cabinet of data is embedded in one nice neat model as attributes that others can access.

CSI did develop a link to Revit not long ago. I have not tested it, but I noticed this option recently when I was downloading updates.