Brownian-motion approximations for tankage assessment and stock control

This paper presents a model for refinery tankage assessment. Special characteristics covered are a hybrid demand process and a periodic-review target-stock policy for production control. The demand is assumed to be in two forms: as large parcels, collected at fixed intervals, and as many small parcels, modelled through a Brownian motion. Analytical approximations for the average stock level, the probability and expected volumes of overflow and stockout per period have been developed and compared with exact and simulation results. Using these approximations the problem of determining both the optimal tank capacity and the optimal target stock can be separated and solved rapidly on a PC. Finally, some sensitivity analyses to determine the most important model parameters have been carried out.