Approximations for non-stationary stochastic lot-sizing under (s,Q)-type policy

This paper addresses the single-item single-stocking location non-stationary stochastic lot-sizing problem under a reorder point -- order quantity control strategy. The points and quantities are chosen at beginning of planning horizon. allowed to vary with time we consider either be series time-dependent constants or fixed value; this leads two variants policy: (st,Qt) (st,Q) policies, respectively. For both present dynamic programs (SDP) determine optimal policy parameters introduce mixed integer non-linear programming (MINLP) heuristics that leverage piecewise-linear approximations cost function. Numerical experiments demonstrate our solution method efficiently computes near-optimal for broad class instances.