Hec Ras w/ lateral flows

[sam74]

I'll try to keep this short. Initially we have modeled (steady state)what appears to be an inadequate culvert system on a blue line stream. It is at an older government facility. Culverts procede downstream with two 36" to two 30" to two 30" to two 36" RCPs.
The initial model worked out okay and showed flooding on most storm events at most all culverts from top to bottom. It was a simple model assuming all flooding went over road deck and back into the stream. The bad part is that over 30 years of observation shoots holes in our model.
An "expert " reviewed our model and input and saw no issues. Stream monitoring and rain gauges are set to be placed to aid future model adjustment. However, the expert suggested we look at lateral flows into a nearby parallel street to see if this would reduce flooding at road decks. We have three such situations occuring with two 30", two 30", and two 36" RCPs. We sectioned off approximate lateral weir dimensions and developed a rating curve and used this as output from the system for these three occurences slightly upstream of the road decks. We also played with road deck min/max WSE settings to prevent runoff across them.
I finally have everything adjusted so that the program will run and pass all storms optimizations. The first culvert looks good upstream and has WSE close to original model. The downstream hits the channel and goes crazy. 1 and 2 yr storms look reasonable (previously the only flows not crossing the road deck). Then every higher storm event's flow and WSE gets progressively lower. My input in the flow table gets higher but the program begins to ignore the table and ramps the flow of the 50, 100, and 500 yr storm to a constant value of 0.14 cfs. The remainder of the model is a mess after this.
Ever heard of the program ignoring the flow table? Any suggestions out there?

[sam74]

I thought my original problem might be a bug so I emailed the files to the HEC RAS team. I actually got a reply the same evening. It took us a while to get on the same page but I think it is a bug. It appears after several manipulations and scenarios that HEC RAS has trouble with new steady flow input downstream of an optimized lateral structure. I originally thought it might have something to do with the fact that I had the three lateral structures so close but if I delete the steady flow inputs in the channels downstream of them and allow only the flow calculated by the optimization process to pass through I get pretty good looking numbers (constants). The channel continues on to a bridge crossing and on the other side of this (which is about 300 ft downstream of the last lateral structure) I had another steady flow input. Looking at the profile table on this section it did the same thing as it did in the midst of the lateral weirs. For instance my steady flow input for 1,2,5,10,25,50,100, and 500 yr events was 154, 164, 166, 171, 186, 200, 199, and 213. The flows listed at this same cross section in the profile table were 154, 160, 120, 80, 20, 0, 0, 0. I couldn't get the HEC RAS guy to admit to anything but he said they were working on this in their next release. He offered the advice that I break the area with multiple lateral weirs into three separate reaches and optimize them by hand. I did and this worked. Hope this helps someone