LRFD Curved Girder Bridge Design

[KuJayHawk]

I am about to use DESCUS I to design a curved girder bridge with 2005 LRFD spec. Is the program able to handle the new code? Thanks ahead for any input!

 

[Qshake]

Descus has some limitations; please refer to March/April Journal of BRidge Engineering (ASCE): Examination of Level of Analysis Accuracy for Curve I-Girder Bridges through Comparisons to Field Data.

Regards,
Qshake

Eng-Tips Forums:Real Solutions for Real Problems Really Quick.

 

[TampaBridgeDesign]

Let me know how you like DESCUS I. We've been using MDX to size our curved girders and then have been using BSDI for final design using the curved girder guide specs. FDOT issued Temporary Design Bulletin C06-04 on June 26 and it states "Once curved steel girder design software is available that complies with Section 6 of the 2005 and 2006 Interim Revisions, and the design results are verified, the Department will issue a revised policy on both curved and straight steel girder bridges. This process is expected to take more than one year."

 

[jimaitken]

We have been using DESCUS-I to design a bridge, but not to the new LRFD specs. I would suggest contacting Dr. Fu at U of M to ask him if the program has been updated or when it will be out. He has worked with us in the past to resolve some issues that were project specific. Also check the website

http://best.umd.edu/

 

[KuJayHawk]

Thanks everyone for the input! I just got the newest version of DESCUS I on my computer now. Since this is also the first time I use this program, I have so many questions about it. The one I am having right now is the Live Load Distribution Factor. For example, if Girder 2 is the primary girder, why do I need to also provide DFs for the other girders? Is that used for diaphragm force calculation? Or can I just go ahead zero the rest of the girders. I tried AUSTO-DF and noticed that it actually combines the load intensity reduction factor with the lateral distribution factor calculated by lever rule. The truck locations, however, may not be the critical location in certain situations, because it always keeps 6' wheel spacing between the two adjacent lanes. If I have a bridge that allows four lanes in between the roadway, it makes 15% difference if I consider maximizing the effect with 2-lane condition vs. with 4-lane conditions (1.0 vs. 0.75 load intensity reduction).

What did you do when specifying the DFs? Maybe I should just let the program handle it with S/11 over all the girders?

 

[jimaitken]

LRFD LL distribution is more complex than the standard specifications. S/11 is no longer valid in the LRFD spec and has been replaced by multiple calculations that must be checked for both interior and exterior girder locations (Moment and Shear). Section 4 of the AASHTO LRFD spec is going to be very helpful. Live load distribution and the way it is calculated is one of the biggest differences in the LRFD spec vs. the Standard specs.

If this is your first LRFD bridge design project, it may be worth it to go ahead and do calculations for ALL of the beams on your bridge. LRFD is more calculation intensive than the standard specs. You will spend more time to get loads than you would have to calculate loads under the standard specs. But in the end you will have a good example set of calculations to refer to when you do your next LRFD bridge design. Once you have some experience you can begin to weed out some of the un-neccesary calculations.
Good Luck.

 

[KuJayHawk]

Well, I've had LRFD bridge designs before but this is the first one with curved girder. You know the 2005 interim has combined both straight and curved girder designs into one unified code, which I really don't know if I should appreciate or not. It's my first curved girder design no matter what code I am going to use. Folks in my office have done a couple of designs with ASD using DESCUS I. So I am trying to run an ASD design to get a rough estimate first. Then check with LRFD. However, the DESCUS user manual really did not tell me how the program runs. All I was told is that the program uses Influence Surface combined with Lateral Load Distribution Factors. The user manual says that for an interior girder, it actually creates only 3 influence lines, and for an exterior girder, 2 influence lines. So my guess is that, the program only looks at load cases when the truck is on the primary girder and the girders right/left to it. The effect due to other girder farther away from it won't be accounted. Or even it is actually calculated, it won't govern. The reason? Say Girder 2 is your primary girder. To find out the maximum/minimum live laod moment, the program will run the truck on top of G2, multiple the influence line values (when the unit load is on G2, say IN22), and the lateral distribution factor when G2 is the primary girder (say DF22). Then the program runs the truck on girder 3, multiple the truck axle loads by inluence line values (when the unit load is on G3, say IN23), and the lateral distribution factor of G2 when G3 is the primary girder (say G23). Then the program will run the truck on G1, G4, .... G10, and do the same type of work. So when the truck is away from the girder under consideration, sicen the influence line value is very small, even the DF is large (if you try to load the whole roadway width even G1 isn't affected by truck on G10 in terms of lateral distribution).

As to the lateral load distribution factor calculation, the policy in our office is to use lever rule for curved girders and neglect the AASHTO Tables in chapter 4. This is consistent with DESCUS user input method for card 0301. The confusion I had for a while is to whether to include multiple presence factor (LRFD) or load intensity reduction factor (Std. Spec). I ran Auto-DF yesterday and noticed the porgram gives me options about whether to include it or not. For average girder spacing, the effect on DF calcs is just one (exterior girder) or two (interior girder) trucks. To include more trucks or lanes will only reduce the effect due to the less than 1.0 multiple presence factor. Things are a little easy with ASD design since this factor is 1.0 for both one and two lane loaded conditions. In LRFD, this is 1.2 for signle lane and 1.0 for two lane loaded condition. So I probably need to be very careful when I start to calculate the DFs in LRFD to maximize the force effect.

Well, I am gonna stop here. I guess you guys probably have figured that out long time ago. I don't know if I made things clearer or not. I just finally got to point a minute ago There are a couple of things in this program that I don't really feel comfortable with. That's why I decided to spend sometime to figure out how the program actually works. I will post more questions or findings on my way to finish this project.

Keep letting me know if you have any thoughts/ideas .... Thanks ahead!

 

[KuJayHawk]

Thanks everyone for your help! I've contacted Dr. Fu as some of you suggested during the analysis of the bridge using Descus-I. He was very helpful and responded quickly for the questions I raised. That made me feel comfortable with the program. Overall, the program is very good, but you need to understand the code well to use it. Due to the transition to LRFD, I believe, some tables do not report the results appropriately. So my suggestion is that: try to track the numbers in between different tables and check by hand. You may wind up making little adjustment here and there. But I actually learned the behavior and understood the code by going through the tables and checking the numbers myself. Hopefully these issues will be solved in the future version of the program.

 

[edward1]

I designed a three span continuous composite steel girder with a relative small radius. The bridge was analyzed using DESCUS, MDX, and STAADPRO 2005 using plates to model the deck and offset members for the girders. The bridge was completed in fall of 2006 and VDOT Research Council in Charlottesville, Va has just finished field testing the bridge. We will know which software is closest to the field studies. I am betting on MDX.