Subject: LP infeasibility in Maximum Hydraulic Outflow Posted: 6/13/2018 Viewed: 129 times
I am trying to model the effect of reservoir drawdown prior to flood events in WEAP. I have the Maximum Hydraulic Outflow (MHO) as a daily variation timeseries for four different reservoirs. For two of them, it works perfectly, but for the two remaining ones I get an error message saying the MHO was removed for such and such days due to an LP infeasibility. It typically happens only in the first day of my simulation period (which is a problem if I want to start the simulation 0-7 days before the flood event, for example). Furthermore, it only happens with certain reservoir levels. If the initial reservoir filling is less than 50%, I avoid the error, but if it is greater than 50% the error returns. I don't understand what could be causing the error, since it doesn't happen in all the reservoirs even though they have similar inputs (with different magnitudes). I have tried to put fixed MHO values to experiment, and if the MHO values are low enough the error does not occur. There are no demands in the system, just rivers, diversions, headflows, and reservoirs.
Does anyone have experience with the LP infeasibility error and know any reason why it might be occurring?
Mr. Bendik Hansen
Subject: Re: LP infeasibility in Maximum Hydraulic Outflow Posted: 6/14/2018 Viewed: 118 times
I managed to solve the problem by deleting the streamflow gauges downstream of the reservoirs. As far as I know there should be no connection in WEAP between the gauged flow and the simulated flow (headflows in this case), so I don't understand why it was causing a problem. Anyway, if anyone else gets a similar problem, this might solve it.
Mr. Jack Sieber
Subject: Re: LP infeasibility in Maximum Hydraulic Outflow Posted: 6/14/2018 Viewed: 117 times
I am also very surprised that deleting the streamflow gauges resolved the errors with the MHO's. Gauge are not used in the calculation of streamflow, so they should not have any effect on these errors.
The MHO calculations involve integer variables in the LP (linear program), which can sometimes cause instability in the LP solver. By the way, you might be able to get a free academic license for the Gurobi solver, which can be used in WEAP and might be more stable. See http://www.weap21.org/WebHelp/LPFormulation.htm for more details.
Mr. Bendik Hansen
Subject: Re: LP infeasibility in Maximum Hydraulic Outflow Posted: 6/14/2018 Viewed: 111 times
I just installed the Gurobi solver (7.5.2 32bit for Windows) with a free academic license and WEAP ran the simulation without an error (with the exact same setup that yields an error with LPSolve).
Thanks a lot!
Topic "LP infeasibility in Maximum Hydraulic Outflow"