Doasa overview

One of the difficulties with decision making in a hydro-dominated electricity system such as New Zealand's is the uncertainty and variability associated with the inflows into hydro storage reservoirs. A hydro-thermal scheduling model can be used to analyse the management of hydrological resources.

Doasa is an implementation of the stochastic dual dynamic programming technique applied to the solution of hydro-thermal scheduling problems. The Doasa model can be used to formulate a policy of releasing water from reservoirs for electricity generation while satisfying demand over a fixed time horizon and minimizing the expected fuel cost of thermal generation. But the model will not necessarily satisfy demand in all scenarios. Given an appropriate penalty cost, some load may need to be shed.

A version of the the Doasa model has been developed for the Authority by Stochastic Optimization Limited (SOL) and is described in the Doasa documentation.

The Authority version of Doasa is restricted to the New Zealand electricity system. The algorithm divides a year into 52 weekly stages. All input data are deterministic except for weekly inflows that are assumed to be stagewise independent. However, a technique known as Dependent Inflow Adjustment (DIA) may be applied to the empirical inflow distribution to make it share some properties believed to be present in the true inflow distribution.

The Authority's version of Doasa is configured to use a commercial solver. We also have an earlier version of the model that has been compiled to use an open source solver, although we would not recommend using this version in anything but trial mode. Please email us at emi@ea.govt.nz to discuss options for getting started with Doasa.

Doasa documentation

Download the Doasa documentation to learn more about the workings of the Doasa model and the input data it requires.

Information paper

Last updated: 10th April 2017

Latest Doasa discussions
Aggregate demand in demand.csv file in DOASA

Category  - Wholesale 20 months ago

Latest Doasa discussions
Aggregate demand in demand.csv file in DOASA

Category  - Wholesale 20 months ago