Water flows and financial viability of electricity generation in Manitoba

The chapter contributes to the international literature on the management of droughts. Predicting and managing droughts are major challenges for utilities dependent on hydro-generation. Manitoba Hydro in Canada (MH) is unique in that over 95% of its total power generation is hydro, and MH sells a large share of its generation outside Manitoba; exceeding on average 31% under long-term contracts that at one time did not have curtailment clauses. In 2004 an unanticipated drought almost brought the company to bankruptcy. Variability in water flows translates into variability in electricity production, revenues, and exports. It is crucial for the utility to predict accurately and in good time the probability of drought occurrences. Several statistical techniques are used to provide reliable forecasts of water variability in the short, medium, and long terms. These techniques included Autoregressive (AR) time series and Extreme Value Theory (EVT) models. The main results derived from using these models suggest that AR and EVT approaches provide very similar results at the 95% intervals, which fact can be considered an important check on the sensitivity and reliability of the prediction model used in the management of water variability predictions.