DETERMINATION OF OPTIMAL DIRECT LOAD CONTROL STRATEGY USING LINEAR PROGRAMMING Zeljko Popovic Distribution engineer Rade Koncara 57, 24300 Backa Topola, Yugoslavia Phone: +381 24 714 220 Fax: +381 24 714 898 E-mail: zeljko.popovic@su.ev.co.yu SUMMARY Direct load control (DLC) is one of the most widely used techniques of load management. This technique is usually applied at residential sector. In most cases DLC is being applied to minimize peak load or production costs in power system. In this paper is exposed a model for determining an optimal DLC strategy of devices in households with the aim of reducing system peak load. The proposed model enables optimal (maximal) reduction of the system peak load in cases when various types of devices in household are controlled at the same time. The presented model is based on linear programming and tested on a real power system where water heaters and air conditioners are controlled at the same time. INTRODUCTION Many load management (LM) programs have been applied by the utilities in recent years. Load management objectives for different utilities may be different. In the power systems with a little reserve in power production and transmission load management is primarly applied for reducing system peak load. Thus, the needs for building new capacities are postponed (reduced). Systems with relatively high reserve are applying load management for reducing production costs and dependence of expensive fuels [1,2] . For achieving load management objectives, it is usually used DLC of some devices in households. Commonly controlled devices are the following: water heaters, air conditioners, electric space heaters and storage heaters. In the literature, a considerable attention is dedicated to DLC and especially to determination of optimal DLC strategy. Direct load control strategy, which determines turning on/off diary schedule of controlled devices, is composed of several elements, the most important of which are: (i) type of controlled devices and the way of their control, (ii) beginning and duration of control period during a day, (iii) number of groups of controlled devices and their mutual time shift, i.e. turning on/off moments of some groups and (iv) number of devices according to groups. There is a series of models for determining an optimal DLC strategy [2-13] . Majority of the proposed models is developed with the aim of reducing production costs of the power system [2-9], and a smaller part with the aim of reducing system peak load [10-13]. This paper is dedicated to determination of optimal DLC strategy with the aim of reducing system peak load. Cohen [10] has developed a model for determining elements of DLC with the aim of reducing system peak load. The model is based on dynamic programming. In that model, total number of controlled devices is in advance divided in determined, beforehand known (assigned), number of groups of the same size (the same number of devices in each group). During optimization procedure, the exposed model treats each group of devices separately. In this way, proper (optimal) coordination of load control (connection and disconnection) between different device groups is missing and maximum reduction of the system peak load is not obtained. Lee [11] has developed a model based on linear programming which enables determination of number of controlled groups and number of devices in each group as a result of optimizing procedure. In the model, turning on/off moments of some groups of devices are determined in advance. A set of these moments forms so-called control shame. Thus, the problem of coordination of load control between some groups of devices is partially surpassed, but maximal reducing of system peak load is not possible because of limits imposed by the method of forming control scheme. We can find similar drawbacks in the model exposed in the literature [12] , which is based on the combination of linear and dynamic programming. The model proposed in the literature [13] is based on linear programming. This model enables determination of optimal DLC strategy when devices, which can be turned off for a long period of time (for example, water heaters), are controlled, while it is not possible for devices which can be turned off only for a brief periods of time (periodically) (for example, air conditioners). In this paper is developed a model, based on linear programming, which enables determination of most important elements of DLC strategy, and consequently realizing an appropriate (optimal) coordination of load control (turning on/off moments) between different groups of controlled devices. In this way, proposed model ensures maximal reduction of system peak load (maximal utilization of available DLC resources) even in case of applying DLC to various types of devices at the same time. This paper is composed of five parts. In the second part is exposed mathematical model for determining an optimal DLC strategy. Appliance of the proposed model to peak load reduction of the real power system, when water heaters and air conditioners are controlled simultaneously is exposed in the third part. The most important conclusions are given in the fourth and the list of cited works in the fifth part. MATHEMATICAL MODEL Direct load control is generally realized in two ways: