Modelling hydro reservoirs with LDS

[filippopecci]

There is probably a bug in modelling of inter-period linkage for hydro resources. In hydro_res.jl, the constraint cHydroReservoir models the state of charge for all hydro resources and all time steps. This includes the start time steps of each representative periods. As a result, start and end of each period are linked:

	cHydroReservoir[y in HYDRO_RES, t in 1:T], EP[:vS_HYDRO][y,t] == (EP[:vS_HYDRO][y, hoursbefore(p,t,1)]
				- (1/dfGen[y,:Eff_Down]*EP[:vP][y,t]) - vSPILL[y,t] + inputs["pP_Max"][y,t]*EP[:eTotalCap][y])

However, when a hydro resource has LDS=1, the link between the state of charge at the start and the end of each period is modelled in hydro_inter_period_linkage.jl by constraint cSoCBalLongDurationStorageStart_H, which should be the only one to hold for start time steps for hydro resources with LDS.

This is why, for normal storage resources, we have these constraint definitions in storage_all.jl:

if representative_periods > 1 && !isempty(inputs["STOR_LONG_DURATION"])
		CONSTRAINTSET = STOR_SHORT_DURATION
	else
		CONSTRAINTSET = STOR_ALL
	end
	@constraint(EP, cSoCBalStart[t in START_SUBPERIODS, y in CONSTRAINTSET], EP[:vS][y,t] ==
		EP[:vS][y,t+hours_per_subperiod-1] - (1/dfGen[y,:Eff_Down] * EP[:vP][y,t])
		+ (dfGen[y,:Eff_Up]*EP[:vCHARGE][y,t]) - (dfGen[y,:Self_Disch] * EP[:vS][y,t+hours_per_subperiod-1]))

and

cSoCBalInterior[t in INTERIOR_SUBPERIODS, y in STOR_ALL], EP[:vS][y,t] ==
			EP[:vS][y,t-1]-(1/dfGen[y,:Eff_Down]*EP[:vP][y,t])+(dfGen[y,:Eff_Up]*EP[:vCHARGE][y,t])-(dfGen[y,:Self_Disch]*EP[:vS][y,t-1])

We should modify constraint cHydroReservoir to match the formulation used in storage_all.jl.