Dynamic-programming approach to the selection and timing of generation-plant additions

Abstract

SynopsisMany basically different types of generating units, such as nuclear, hydroelectric, steam-and gas-turbine units, could be added to a generating system to meet constantly growing installed-capacity requirements. Since each of these types may also vary considerably in capital cost, size, efficiency, fuel and site location, there are a very large number of possible choices each time a new generating unit is needed. Moreover, once a unit is purchased, its known capital fixed charges continue for the life of the plant, while its fuel and operating costs will initially depend on the composition and characteristics of the system prior to the installation of the new unit, later depending also on the relative efficiencies of units added subsequently, since these will directly affect the future power and energy supplied by the new unit to the system. In order to evaluate the economic merits of a particular type and size of unit to be added to a particular system, it is therefore necessary to determine the total costs associated with this choice over the projected life of the unit, in conjunction with each of the many thousands of combinations of future units that might be added during this lifetime period to be studied. At any stage in the process of making these decisions, the correct choice cannot be made without considering all possible decisions that might have been made during earlier stages and that might be made in future stages. The paper describes a dynamic-programming application to this particular problem, which, by eliminating many alternatives during each stage or year of a long-range study period, reduces the number of possible choices to proportions that can be handled and evaluated by a medium-sized digital computer, when using relatively simple reserve-capacity-requirement and production-cost logic, but which can become an exceptionally accurate tool in system-planning studies when programmed for use on a large computer.

A two‐stage stochastic integer programming model for a thermal power system expansion

“…Taking into consideration the many possible expansion choices and the high investment costs involved in a plan of such nature, it becomes appropriate to approach the formulation of an optimization model which provides an optimum investment policy. In the deterministic case, starting from the seminal work by Masse and Gibrat (1957) many applications and extensions arise; such as the articles by Dale (1966), Bessiere (1970), Peterson (1973), and Sawey and Zinn (1977), to mention a few.…”

Decision analysis of capacity expansion strategies for electrical utilities

“…However, the implied dis economies of scale that are associated with reliability con straints present computational difficulties in this class of models. The third category of procedures for dealing with large-scale capacity expansion problems makes use of dynamic programming [6], [ 7 ] . These algorithms can account for most aspects of actual problems, but the amount of computation and data storage necessary has restricted analyses to relative short planning periods.…”