Masters Theses


"The method of Stochastic Dynamic Programming is applied to the problem of Long-term hydrothermal coordination for a five-reservoir system in the Midwest. The model that has been developed will optimize the monthly reservoir elevations to produce the lowest expected energy production cost. The production cost calculations are decoupled from the Dynamic Programming loop, resulting in shorter execution times. Constraints are included for maximum and minimum reservoir elevations, maximum monthly reservoir rise and drop, and maximum and minimum monthly energy production. These constraints help to reduce the state space and shorten execution time. An additional constraint is modeled to control the reliability of the reservoir management schedule that is produced. Schedules with higher reliability also have higher costs.

The model is demonstrated on a typical planning problem: to optimize the lake elevations over the fall, winter, and spring. The stochastic dynamic programming model reduced the expected energy production cost by $680,000 over this period, as compared to the rule curve-based methods currently in use.

This research shows that, with the computing power now available, the direct stochastic solution for a five-reservoir system is now feasible. Execution times were under two hours on a 486-computer running at 33 MHz"--Abstract, p. iii


Omurtag, Yildirim Bill
Raper, Stephen A.

Committee Member(s)

Kincaid, John B.


Engineering Management and Systems Engineering

Degree Name

M.S. in Engineering Management


University of Missouri--Rolla

Publication Date

Fall 1991


viii, 83 pages

Note about bibliography

Includes bibliographical references (pages 78-82)


© 1991 Duane David Highley, All rights reserved.

Document Type

Thesis - Open Access

File Type




Thesis Number

T 6284

Print OCLC #