The Base Stock Model 1 Assumptions Demand occurs continuously over time Times between consecutive orders are stochastic but independent and identically distributed (i.i.d.) Inventory is reviewed continuously Supply leadtime is a fixed constant L There is no fixed cost associated with placing an order Orders that cannot be fulfilled immediately from on-hand inventory are backordered 2 The Base-Stock Policy Start with an initial amount of inventory R. Each time a new demand arrives, place a replenishment order with the supplier. An order placed with the supplier is delivered L units of time after it is placed. Because demand is stochastic, we can have multiple orders (inventory on-order) that have been placed but not delivered yet. 3 The Base-Stock Policy The amount of demand that arrives during the replenishment leadtime L is called the leadtime demand. Under a base-stock policy, leadtime demand and inventory on order are the same. When leadtime demand (inventory on-order) exceeds R, we have backorders. 4 Notation I: inventory level, a random variable B: number of backorders, a random variable X: Leadtime demand (inventory on-order), a random variable IP: inventory position E[I]: Expected inventory level E[B]: Expected backorder level E[X]: Expected leadtime demand E[D]: average demand per unit time (demand rate) 5 Inventory Balance Equation Inventory position = on-hand inventory + inventory onorder – backorder level 6 Inventory Balance Equation Inventory position = on-hand inventory + inventory onorder – backorder level Under a base-stock policy with base-stock level R, inventory position is always kept at R (Inventory position = R ) IP = I+X - B = R E[I] + E[X] – E[B] = R 7 Leadtime Demand Under a base-stock policy, the leadtime demand X is independent of R and depends only on L and D with E[X]= E[D]L (the textbook refers to this quantity as q). The distribution of X depends on the distribution of D. 8 I = max[0, I – B]= [I – B]+ B=max[0, B-I] = [ B - I]+ Since R = I + X – B, we also have I–B=R–X I = [R – X]+ B =[X – R]+ 9 E[I] = R – E[X] + E[B] = R – E[X] + E[(X – R)+] E[B] = E[I] + E[X] – R = E[(R – X)+] + E[X] – R Pr(stocking out) = Pr(X R) Pr(not stocking out) = Pr(X R-1) Fill rate = E(D) Pr(X R-1)/E(D) = Pr(X R-1) 10 Objective Choose a value for R that minimizes the sum of expected inventory holding cost and expected backorder cost, Y(R)= hE[I] + bE[B], where h is the unit holding cost per unit time and b is the backorder cost per unit per unit time. 11 The Cost Function Y (R ) hE[ I ] bE[ B ] h( R E[ X ] E[ B ]) bE[ B ] h( R E[ X ]) ( h b) E[ B ] h( R E[ D ]L) (h b) E ([ X R ] ) h( R E[ D ]L) (h b) x R ( x R ) Pr( X x ) 12 The Optimal Base-Stock Level The optimal value of R is the smallest integer that satisfies Y (R 1) Y ( R) 0. 13 Y ( R 1) - Y ( R ) h R 1 E[ D ]L ( h b) x R 1 ( x R 1) Pr( X x ) h R E [ D ]L ( h b) x R ( x R ) Pr( X x ) h ( h b) x R 1 ( x R 1) ( x R ) Pr( X x ) h (h b) x R 1 Pr( X x ) h (h b) Pr( X R 1) h (h b) 1 Pr( X R ) b (h b) Pr( X R ) 14 Y ( R 1) - Y ( R ) 0 b (h b) Pr( X R ) 0 b Pr( X R ) bh Choosing the smallest integer R that satisfies Y(R+1) – Y(R) 0 is equivalent to choosing the smallest integer R that satisfies b Pr( X R) bh 15 Example 1 Demand arrives one unit at a time according to a Poisson process with mean l. If D(t) denotes the amount of demand that arrives in the interval of time of length t, then (l t ) x e l t Pr( D(t ) x ) , x 0. x! Leadtime demand, X, can be shown in this case to also have the Poisson distribution with ( l L) x e l L Pr( X x ) , E[ X ] l L, and Var ( X ) l L. x! 16 The Normal Approximation If X can be approximated by a normal distribution, then: R* E ( D ) L zb /( b h ) Var ( X ) Y ( R*) ( h b) Var ( X ) ( zb /( b h ) ) In the case where X has the Poisson distribution with mean lL R* l L zb /( b h ) l L Y ( R*) ( h b) l L ( zb /( b h ) ) 17 Example 2 If X has the geometric distribution with parameter , 0 1 P ( X x ) x (1 ). E[ X ] 1 Pr( X x ) x Pr( X x ) 1 x 1 18 Example 2 (Continued…) The optimal base-stock level is the smallest integer R* that satisfies Pr( X R* ) 1 R* 1 b bh b R* bh b ] b h 1 ln[ ] ln[ b ln[ ] bh R* ln[ ] 19 Computing Expected Backorders It is sometimes easier to first compute (for a given R), E[ I ] x 0 ( R x ) Pr( X x ) R and then obtain E[B]=E[I] + E[X] – R. For the case where leadtime demand has the Poisson distribution (with mean q = E(D)L), the following relationship (for a fixed R) applies E[B]= qPr(X=R)+(q-R)[1-Pr(X R)] 20