any good book for modeling bridge? 1

[koo]

This site has been very helpful. I searched past posts for modeling methods, but couldn't find much details.

I just started learning modeling bridge in SAP or Staad, mostly by trial and error. Are there some good books/manuals describing computer modeling of deck and bridges?

I know there's a book named Bridge Modeling by Microcomputer, but it's a 1989 ver. and out of print. SAP's demo is too simple and with no deck.

Any input? How did you pick it up?

 

[VoyageofDiscovery]

koo,

Any slope in either direction might do this. Typically, for flat grillages, I do not get lateral loads. I don't typically analyse them on a grade, unless I feel I should.

I suspect your skew is between 30 and 35 degrees with 1/5 larger values for shear.

I also think you should look at your 1/3 less moment. It doesn't seem right. From what I can remember the simplified AASHTO method and the simplified Canadian method yielded different results for two design lane bridges.

You should check your skewed bridge grillage against a straight bridge grillage to verify for moment. If the AASHTO method gives you less moment, use the grillage moment, as this is for sure.

Regards

VOD

 

[koo]

VOD,

My model is based on a straight bridge though. It's the simplest form of a composite steel bridge. So I'm quite supprised with the results. I can't find anything I did wrong in the model. I don't know how sap or other programs move its self-generated AASHTO load, only along the bridge or transversely within the lane as well. But I would think it's reasonable to have some discrepancy with the AASHTO simplfied method due to its "simplied" nature, but not this big though.

Koo

 

[VoyageofDiscovery]

koo,

Not to say your analysis is wrong, but to correctly model your bridge, do not use shell elements since you do not know how SAP is distributing the loads.

Do the grillage first, then as you get results from the grillage, compare them with your "shell between girders" model. As you get more similar results with different bridges you will then decide which one is easier based on your trust of the particular program. I have never done this "shell between girders" as some programs have problems mixing shells and members.

At the end of it all, check it against the "simplified " method in AASHTO. One should not use the AASHTO method without verifying it each time. This is why I always do a grillage and the "simplified" method together. Estimate your time for the job with this in mind.

Regards

VOD

 

[koo]

VOD,

sorry that I still need to ask you several questions about grillage modeling. I started building a simple steel bridge in StaadPro. It has 4 beams at 9.67' spacing and 38 feet span, 2lanes for HS20 truck. I put 3 transverse frame members with concrete deck properties (9.5' wide-19'w-9.5'w). After load it with 2 moving HS20 trucks, I observed that 1)wheel load is distributed correctly to girders by simple beam assumption; 2) the resulting reaction and max. momemnt on the most heavily loaded interior girder is way less than a AASHTO line beam analysis would give (234 vs. 400 k-ft), but a nearby exterior girder is overloaded (200 vs. 128 k-ft). Apparently, load is redistributed by transverse members.

My questions are:
1) Is the result reasonable or correct at all?
2) How would you change the torsional propertie of the transvere member (deck) in Staad? Will it affect the outcome by much? (I tried to change the wide conc. deck property to a week steel beam, the resulting forces barely changed at all.)
3) How would you model the composite action in the girders? Use the CM section option in Staad or transform it into a steel I section?
4) Do you have the Staad.Bevea package? Will it give different results?

If you have the patience and interest, I could put my input file here for some critique.

Thanks alot.

Koo

 

[VoyageofDiscovery]

Hi koo,

First, your inputs should be like this.

When you put in your transverse section properties. Torsional properties (J or R) should be for concrete cross section. This does make a difference.

For longitudinal beams, transform the composite section into steel. For its sectional area in Staad, use the steel girder only otherwise you will have additional area of the transformed concrete into steel as dead load that is accounted for in the transverse sections.

I would stay away from the options in Staad for now and just use the edit screen.

Finally, for the live load and superimposed dead load analysis, which we are discussing, the exterior long. girders should have all the section properties of any barrier wall and sidewalk if any, including its transformed torsional properties.

Once you have done these, compare your results again, it appears that your load distribution is correct but the analysis is going wrong.

Regards

VOD

 

[koo]

VOD,

I specified trans. member to be concrete and gave it a demension (9" by 9.5'). Isn't J a property of the section and generated automatically by Staad based on my input geometry? I found nowhere I can edit this property.

You also mentioned earlier that "the torsional properties of the deck within this combined section should be halfed to account for the two way nature of bridge decks." where would you define it Staad? I assume this is in the long. direction and how would you bring this 1/2J of deck into a combined section?

Lastly, how would you transform a barrier into a composite section? just like a deck, shrink it in width? then the N.A. of the combined section would be brought very high. Is this composite action from barrier supposed to be accounted for?

One answer usually brings out more quetions...sorry

Koo

 

[VoyageofDiscovery]

Hi Koo,

There was an edit screen pre-windows version, it should still be there, check with REI on this.

I do not know how the input geometry interacts with section properties, but I doubt it.

J also known as R is the sum of individual R. Check Omar Blogett's "Design of Steel Structures". Typically for b/d for 20 to infinity the beta value is 0.333, but for smaller values interpolate from the provided Table 1 on pg 2.10-2

As you sum the individual values you make sure to half the slab's contribution not the haunch's or anything else concrete.

Yes, shrink the width, yes the NA rises. Yes this should be accounted for, as the stiffness of this girder increases, affecting the structural response to live load.

Regards

VOD

 

[VoyageofDiscovery]

Sorry koo,

That should be "Design of Welded Structures" by Blodgett published by James F. Lincoln Arc Welding Foundation.

This book is found in most structural engineering offices.

Regards

VOD

 

[edward1]

Koo, If you want to know how bridge systems are designed, you should download a trial version of MDX. It uses strudl
to analyze the system as a finite element structure or you can choose the grid option. Comparing the results from MDX and STAADPRO you must you OFFSET BEAMS from the ELEMENTS which are used to model the deck. The supports are put on the ELEMENTS in this case. By using MDX you don't need to mess with these things; MDX does it automatically.

 

[koo]

ed1,

glad you are here. I was looking for you. I modeled a simple steel composite bridge in both Staad and SAP with the method you described before( plate for deck, frame for beam and bolster). The results are good at supports, but midspan bending moment is much less than AASHTO values from App.A and D.F. tables (25%, 15% less for each). I know there will be some discrepency but most comments I read, including yours, said the FE model gives very good results. (I first thought DF in AASHTO is too conservative, but a more accurate LRFD fomular gives similar value.) Can you give me some comments on this?

I had Staad and Sap people checked my model, and they said it looks good. Do you compare your model results to AASHTO?

Koo

 

[edward1]

Koo, In order to get the correct girder + slab moments you must use MEMBER RELEASE START MPZ .999 END MPZ .999 on connecting beams. Without RELEASE you get correct deflections. On curved girders you use MEMBER RELEASE START MZ END MZ on these connecting members. I think by RELEASE you are making sure all the longitudinal moment is either induced to the deck or girder. Ed

 

[koo]

ed,
1. Can you clarify if it's MPZ or FPZ? Your two posts gave them differently. Are you sure it's correct command? Member release has FX(full release), KFX(spring) and MP(partial), which one are you trying to tell me? Partial momement release in Z?

2. Is 0.999 an arbitrary number? how did you discover it?

3. Among FE model, AASHTO code values, and field tests, how are they compare to each other, any percentage range?

Thanks. I've been fighting this for a while...

Koo

 

[edward1]

Koo, On straight girders use MPZ .999; on curved girders use MZ which is full release. These are things which I have learned by trial and error.
Ed

 

[koo]

Ed,
I tried MPZ, MPY, but no effect. Then I used MZ, MY, indeed, the links become a connection rod taking only compression. But the total moments at beam and slab still didn't add up to AASHTO value.

I think by using MPZ or MP, the deck and girder effectively become non-composite, right? The moments in girder and deck don't reflect the actural composite condition, then what's the point of obtaining them at all? Can you use them for anything?

Maybe it makes sense that FE model gives a reduced moment since sharing among girders is factored in vs. a single beam approach in AASHTO. Any comment?

Koo

 

[VoyageofDiscovery]

Hi koo,

Not to get involved in the specifics of your FE model with Ed, but I don't think you can make that assumption about the extent of load sharing. Usually the single beam approach, for applicable bridges, and FE models should be within a few percent.

Regards

VOD

 

[koo]

VOD,
That's what I thought first, but after checked my models and have Staad, Sap tech supports checked them. I couldn't find any problem with them. But I am not sure Ed's end release method is correct in modeling a composite section. (there should be large axial load since NA of the section is up in the deck.) Staad/sap will always give correct moment if it's a non-composite section. I think what makes the deal is to produce good result in composite action. So far I have no luck. I do hope people here like you and Ed can help me have this solved. Thanks.

Koo