|M.Sc Student||Yevgenia Mikhaylidi|
|Subject||Scheduling Electricity-Consuming Operations:|
A Supply Chain Approach
|Department||Department of Industrial Engineering and Management||Supervisors||Dr. Yedidsion Liron|
|Dr. Naseraldin Hussein|
|Full Thesis text|
The electricity generation, transmission, distribution, and consumption can be characterized as a supply chain network. Within the distribution-consumption echelons, we consider a finite planning horizon with electricity-consuming operations that need to be completed and are available for processing at predetermined periods throughout the planning horizon. We assume a capacity constraint on the total power consumed in each period due to infrastructure and provider limitations. To benefit from variations in electricity prices, we integrate a storage device in the form of a rechargeable battery. We assume a known and given pricing scheme for the whole horizon. Each operation is unique and has its concave electricity consumption function. Preemptions of operations are allowed yet, postponing an operation incurs a cumulative penalty for each time period. In addition, each preemption is considered as a new operation. There is an exogenous electricity cost in each time period and the customer has to determine when to process each operation within the time horizon so as to minimize the total electricity consumption and operations postponement penalty costs.
We relate our model to the capacitated lot sizing area of research. We consider a model which includes fixed startup cost incurred for switching on the appliance and fixed reservation cost incurred for keeping the appliance On. That is, the startup cost for a given operation is incurred only if the appliance was Off in the previous period.
We consider several special cases of the model and determine when to charge and when to discharge the rechargeable battery so as to minimize the total electricity consumption and operational costs. We propose a polynomial-time algorithm for a special case of single-operation type with uniform capacity. For the other models that we develop, we prove the complexity and elaborate on the tractability of the solution algorithm.
We provide algorithm implementation for the most general case by a Matlab code. Finally, we present a numeric example that illustrates all of the steps of the proposed algorithm.