Grillage vs Shell model

[Dakka]

Hi.
Please consider a typical I girder superstructure.
Common practice in the industry is to use a Grillage model for analysis and design.
In longitudinal direction, we model the effect of deck slab in the form of effective width of each girder (Girder modeled as beam element) to account for the effect of shear lag and transversely the long. members are connected by transverse members (beam elements).
What I am interested is to calculate the reactions due to Dead load of a curved span model.
If I model the girder alone (without the effective width) as a Beam element and model a shell element to simulate the Deck slab, is the effect of changing effective width along the girder inherently considered in the Shell? Are my reactions accurate?
(In order to accurately model the Inertia, I will consider my axis at top of the deck slab, model the shell element as Top line model, model the girder along this axis but eccentric to this axis such that the top of girder coincides with the bottom of deck slab. Supports assigned at the ends of axis. This is as per Sofistik model)
Please respond.
Thanks in advance.
Best.

 

[skewl]

Hi Dakka,

Your question is answered in Section 8.2.3 of the 2019 FHWA Manual for Refined Analysis in Bridge Design and Evaluation. I have attached the relevant section below in case you don't have a copy.

Essentially, to make your model account for "effective width", you have to refine your slab grid/mesh. When you do the "section cut" to integrate the stresses for equivalent beam forces (M and V), the "section cut" has to be approximately what the "effective width" is. To do that you will need a refined slab mesh.