A pragmatic approach to the design of bullwhip controllers

It is well known that forecasting mechanisms can greatly increase "bullwhip" - demand variance amplification of orders as processed by both human and algorithmic decision makers. This paper is concerned with the application of the well-established APIOBPCS Decision Support System (a variant of the Order-Up-To Rule) in such circumstances. It has two feedback controls (based on the inventory and the orders-in-pipeline respectively) with gains set equal according to the Deziel-Eilon Rule. There is one feed-forward control based on exponential forecasting, although this is not a restriction on the application of this system. We consider the pragmatic role of APIOBPCS in the situation where the echelon decision maker may be handling a wide range of SKU's in a non-altruistic environment where upmarket information may either be withheld or simply unavailable. Under such circumstances it has been established via site-based studies that the decision makers output (the orders) reflect a wide range of strategies (or maybe ignorance). Three strategies may be regarded as "appropriate", i.e. Pass-orders-Along; Demand Smoothing; and Level Scheduling depending, on context. APIOBPCS can be adapted to each of these modus operandi. In the first case with the added capability of smoothing the "sharp edges" with a modicum of inventory variation, and in the last case with the advantage of built-in trend detection. "Players" in non-altruistic supply chains must be able to cope with added uncertainties due to lead-time variations. We show that APIOBPCS may be well matched to such situations and is hence "copable" as well as "capable". The paper includes recommended parameter settings according to desired decision-making policies.