Procurement planning to maintain both short-term adaptiveness and long-term perspective

We study the fuel procurement problem for electrical utilities under uncert ain demand and market price. Long-term contractual supply commitments are m ade at a set price with fuel suppliers at the beginning of each year. Each month the procurement planner can use fuel from these contracts or purchase fuel at the current market price. Motivated by practical insights from thi s market, we propose a two-phase dynamic procedure to determine a procureme nt plan. In the first phase, the minimum contract purchases for each month are determined at the beginning of the year. In the second phase, given the minimum contract purchases, the more detailed procurement decisions are de termined at the beginning of each month with the most up-to-date informatio n. We perform intensive computational experiments that show that this proce dure produces high-quality solutions comparable to a rolling-horizon stocha stic-programming heuristic, is easier to maintain and generalize, is comput ationally faster, and is robust to random fluctuations in demand requiremen ts, spot market prices, and other sources of uncertainty.