AQUAPLAN-SDDP

AQUAPLAN-SDDP is a stochastic hydro-economic model representing the water economy of a regional water system (typically a river basin). The main features are:

  • Generic formulation that can easily be tailored to most water resources systems
  • Spatially-distributed water demands: municipal and industrial, irrigated agriculture, energy generation, in stream flow requirements, etc.
  • Spatially-distributed water supplies: river network, reservoirs, water transfers
  • Arc-node representation of the water system to hydraulically connect water demands with supplies
  • Determines weekly or monthly water allocation decisions over a large number of hydrologic scenarios (dry, …, wet)
  • Water allocation decisions are chosen so as to maximize expected basin-wide net benefits subject to operational constraints (physical, environmental, legal, …)
  • Can handle complex systems with a large number of reservoirs and demand sites
  • Considers the hydrologic uncertainty through a probabilistic description of future inflows
  • Can handle different hydrologic information: previous inflows, snowpack, sea surface temperature, seasonal forecasts as well as climate states (e.g. dry, normal, wet)
  • The optimization problem is solved using the SDDP algorithm (Stochastic Dual Dynamic Programming), which is extensively used in the power sector.

The code is written in Python and relies on the GUROBI solver. Input data are stored in two Excel files. A version of the code is also available as a MATLAB toolbox.

The outputs consist of:

Per node j, per hydrologic scenario m, per time period t:

  • Allocation decisions: water abstractions, downstream releases, storage volumes, spillage losses, evaporation losses, etc.
  • Marginal water values ($/m3)
  • Shadow prices ($/m3) associated with the constraints: max/min storage capacity (reservoirs); max/min water diversion capacity (irrigation, industrial, municipal and domestic); max/min turbining capacity (hydropower); min flow requirements (e-flows, navigation, etc.); max flow (flood control)

In general, those outputs are further processed to answer typical planning questions such as:

  • To what extent are the objectives achieved in terms of food production, energy generation, municipal and industrial water supply, navigation, ecological flows, recreation, etc.?
  • How much water is available in the basin and how it varies in space and time?
  • What are the trade-offs between competing uses in the basin?
  • What are the potential impacts of climate change on water availability and uses?
  • How vulnerable are the water users to climate change?
  • How can we best adapt the allocation policies to a changing climate?
  • …
Trade-offs between six water-using activities (objectives) in a river basin. Each line corresponds to a particular hydrological condition, from very wet (blue) to very dry (red).
Optimized drawdown-refill cycle of Lake St-Jean in Quebec

Want to know more?

Modelling details on SDDP-AQUAPLAN can be found in the following articles:

  • Espanmanesh V. and A. Tilmant, 2022. Optimizing the Management of Multireservoir Systems Under Shifting Flow Regimes. Water Resources Research. DOI: 10.1029/2021WR030582
  • Pina J., Tilmant A. and P. Côté, 2017. Optimizing Multireservoir System Operating Policies Using Exogenous Hydrologic Variables. Water Resources Research. DOI: 10.1002/2017wr021701
  • Goor Q., Kelman R. and A. Tilmant, 2010.  Optimal multipurpose-multireservoir operation model with variable productivity of hydropower plants, Journal of Water Resources Planning and Management(ASCE), DOI:10.1061/(ASCE)WR.1943-5452.0000117  
  • Tilmant A., Pinte D. and Q. Goor, 2008.  Assessing marginal water values in multipurpose multireservoir systems via stochastic programming, Water Resources Research, 44, W12431, DOI:10.1029/2008WR007024