AN MILP FORMULATION FOR LOAD-SIDE DEMAND CONTROL
Electric machines and power systems
This paper presents a mixed integer linear programming formulation for load-side control of electrical energy demand. The formulation utilizes demand prediction to determine if control actions are necessary, and it schedules both shedding and restoration times based on an optimization model that minimizes the net cost of load shedding. Operational constraints are satisfied through the use of minimum/maximum uptimes/downtimes, which depend upon the current state of the system. The algorithm is
... The algorithm is evaluated using a simulation model of an underground coal mining operation where, (i) its performance is compared with a traditional static, prioritybased, load-shedding schedule, and, (ii) its potential is established for producing net savings through demand control. Introduction Electricity charges for industrial customers are generally based on total energy consumption, power factor, and maximum demand. The demand component is determined from the maximum average power flow during short, fixed-duration time intervals, typically 15, 30, or 60 minutes, and it is measured in kVA, kW, or other suitable units. This demand charge is levied primarily because of the cost of peaking generation required for occasions when the utility experiences high coincident demand; and it is a large component of the customer's total bill (a demand charge of 25% of the total is typical for most industrial users, Pongia and Battish). Accordingly, a very effective means of reducing utility bills is through control of the demand component, especially for customers with large, uneven loading patterns that make their operations prone to high demand charges. In fact, demand control has been used in residential power systems (Gustafson; Hsu and Su), manufacturing systems (Pongia and Battish; Su, et al), coal mines (Croyle, et al), and even the utility companies themselves (Cohen and Wang; Halevi and Kottick). These applications show that demand control can not only reduce costs, but also increase the productivity and the life of equipment (Pongia and Battish; Effler and Wagner; Hsu and Su). Demand control is likely to become of greater importance in the future as the gap between electrical demand and supply decreases. For example, utilities are currently offering large industrial customers interruptible rates (which allows the utility to shed customers on short notice during periods of high coincident demand). In addition, utilities are also moving towards shorter demand intervals, e.g., five minutes, and even rolling demand intervals. * These authors contributed equally to the preparation of this paper. Supply-side Versus Load-side Demand Control Demand control systems can be divided functionally into supply-side and load-side control/management systems. Supply-side demand management systems, implemented by the utilities, choose between maintaining the demand below a certain level, or permitting an increase in the demand level, in which the choice depends on the current generating costs. If demand is to be limited, most systems attempt to determine the best strategies to shed customer load groups during peak periods. Customers participating in such an arrangement are usually given special rates. Load-side management systems are used by power consumers and are motivated by a desire to reduce the demand component of their power bills. Control objectives are to minimize the electric demand charges by switching loads off and on while minimizing production losses caused by load shedding. Load shedding schedules are established according to the structure of the demand charge, operational priorities of each load, and power consumption of each load. Control decisions should also consider load operational status, guard against equipment damage from frequent switching, and prevent safety hazards caused by automatic load switching.