The distribution factors could be still applicable, if the spacing of the girders and their stiffness were the same.
Otherwise, the grillage analogy model will suffice.

Hi Wiktor!
Thanks for your reply.
AASHTO distribution factors aside; do you think lever rule is still applicable if the girders were of considerably different (20%) stiffnesses.

The lever rule is generally applicable, but in some cases is very conservative. It assumes the deck as simple spans between girders, so it ignores the contribution of deck continuity and torsional stiffness of the girders.

Rod Smith, P.E.

I believe that for each girder, the highest distribution factor (DF) calculated for that girder or either of the adjacent girders, should be correct or conservative.

I think you could add together .25DF for each of the adjacent girders and .5DF for the girder under consideration to calculate the DF for each girder. However, my understanding of the mechanics involved is not advanced enough to say with any certainty that this approach would be accurate.

A detailed finite element model is likely the only way to get close to the real distribution.

Rod Smith, P.E.

The lever rule is typically used for calculations of the reactions, and is on the very conservative side for the moment distribution for a typical 40'-60'span bridge. With the stiffness of the beams varying by 20% you may still use AASHTO distribution factors, with some amplification factor - I think 10% will suffice. The full F.E. model would be an overkill, and the results should not differ from the grill analyses.

Use a grillage model.