CSI SAFE Beam Deflection (Cracking, Creep, Rebar etc)

[bookowski]

Calling JoshPlum from the ether.... I've let our support subscription run out so no longer able to pester csi with questions.

As far as I can tell (from the safe manual and help) safe calculates cracking based on the required steel from analysis. There is an option to consider the cracking based on user input steel (which is fairly laborious in safe but that's a separate issue) - but what about beams? There is no way to input beam steel so I assume beam cracked properties are still based on the analysis demand? And creep in both cases is accounted for in the manual input of the long term factor? If you jam a beam with top and bottom steel you can make significant improvements to the transformed section and the creep - but is there no way to reflect this in safe? If that's correct this seems to be a fairly annoying shortcoming.

 

[JoshPlumSE]

Bookowski -

SAFE is probably the program I've used least. And, I'm under a bit of a deadline today. But, I'll look into this in more detail.

My first guess is that there is NOT the advanced ability in beam stiffness / deflection calculation. SAFE has really has focused on concrete slabs more than the beams.

However, I have to believe that we give you the opportunity to provide some "overwrite" for our beam stiffness assumptions. I just don't have the time to dig into it right now. At least not in the type of detail I'd like to give you.

 

[bookowski]

I think I've found a partial answer but if you can gather any more info it would be great. It looks like the only option is to create beams as thick slab elements vs line elements. You then have to manually define the rebar over the entirety of your floor (i.e. including slabs, not just the beam in question). From what I can tell this is working to change the cracked stiffness (I'm able to vary deflection results by varying the rebar). As far as I can tell though there is no way to vary the creep coefficient per element/area. The input is global so I can't jam a few beams with top steel and see a benefit unless I change the global coefficient, in which case the balance of the results are using an incorrect coefficient. If this is the case then the easiest workaround I can think of is to manually input the rebar, manually change the global creep coefficient, view results at the beam only, change the global coefficient back to an appropriate value, look at the new results, then scale the property modifiers within the individual beam until the results match the earlier reduced creep model. That seems like a lot of work.

If you find there's an easier way it'd be very useful. I agree that most people, at least in the US, are looking at flat plate. But outside the US beam systems are more common - and personally i'm stuck with a beam system right now which is why I went down this path.

 

[Mike Mike]

Bookowski, I'm analyzing an old concrete beam building here in the US, and I was hoping Josh could give us his two cents on this as well. In your experience does SAFE give reasonable long term deflection output? It seems to do a decent job in my experience, but I am new to SAFE.

 

[bookowski]

@MikeMike - I've been using safe for a long time, if you're new to it I would summarize by saying it's generally a very good program (at least within the world it's competing in) but you do need to make sure you play with it quite a bit, test some things, and read what is available to understand how it does things (their documentation is not great, online they have wiki pages which tend to be better). The design side of safe, vs analysis, is terrible in my opinion. ram concept is about 100x better for ease of design.

I could go write a book but i'll try to be short
- look at their wiki page on calculating long term deflections (google it & you'll find it)
- they handle long term deflection differently than say ram concept, but safe is very widely used so it seems to work well
- note that deflections are very dependent upon the inputs, make sure you understand their defaults. for example it defaults aci 435 rupture stress for cracking vs 318 (so it takes the more conservative) but strangely uses the lt creep factor from 318 fs 435 (doesn't make too big of a difference usually). the shrinkage strain default - make sure it makes sense for you. the percent sustained live load. etc etc
- beams are tricky in safe in my opinion: in a new building with beams you can do ok if you're careful, for an existing building I think you'll need to combine hand calcs
- note the analysis option to "ignore vertical offsets" which defaults to yes. This makes a big difference in results with beams. The default treats beams as centerline on the slab, i.e. everything is in one plane. This is of course not true. If you uncheck the default it accounts for the offset centroid (also check the insertion point). This is more realistic but you need to be careful about interpreting beam designs/results. When the beam offset is accounted for your top forces are in the slab, and they come from fea - not a code defined effective width. This is not a bug, it's just the nature of fea, but it can be hard to square the results with hand calcs. It's also easy to accidentally not provide enough steel - because you essentially have to manually combined the beam result steel with the local slab steel that is axial from the beam flange force.
- for hand calcs in beam systems you are usually assuming a one way slab spanning to beams. in fea this isn't as clean as simple hand calcs, the 2 way action is still very present so again your results may be quite different. in a new building you can work out how to deal with this but in an existing it's tricky, i.e it's not so easy to see in safe if what was originally provided "works"
- note that deflections are based on analysis required rebar by default, not user input steel. so if you run your existing building in safe and view deflections they may not be accurate, as you'd have to manually input the actual reinf. and then change the default to consider this input
- safe has a terrible interface for manually drawing in rebar so good luck if you go that route, you may want to test a small area to see what you're in for
- for an existing building you may be better off to skip safe altogether. it depends what you mean by old but it was most likely designed by hand as 1 way slabs supported by beams. you may want to start out by trying to replicate the original design to see if it makes sense, and then adjust for your new condition. you may spend more time battling safe than it's worth.

 

[Mike Mike]

Been battling SAFE the last 3 days I have to stay the course now lol, good learning experience for me anyway. I've used a lot of Ram Concept and that seems like a better program all around IMO but maybe that's because I'm used to it. I'm using SAFE version 20.3.0 and I think SAFE did away with the differentiation between "Designed slab rebar" and "User specified rebar" under Cracking Analysis Options. They're now lumped together in one option.