The effect of time delays in make-to-order information on the dynamics of supply chains

While the literature suggests many causes for the dynamics of supply chains, the so-called "bullwhip effect", this paper focuses only on the effect of time delays in the flow of ordering information. We examine this effect in a supply chain in which its members, aiming at increasing their responsiveness to customer demand, have the option of coordinating their transactions using one of two canonic topologies of information structure; one serial with attendant local transfer of ordering information, and the other parallel with attendant global information transfer. Using computer simulation experiments on each topology of the supply chain, we analyse the effects of time delays in the transfer of information between chain members, on the material flow moving downstream in the chain. The effects are quantified in terms of the variability of material flows along the chain, the individual response of chain members to customer demand, and the overall performance of the chain in meeting customer demand. We characterise both models by the same production and shipping lead times, but with varying time delays in inter-organisational information transfer, while controlling for time delays in the transfer of information within the organisation of each member of the chain. Our results show that in the serial system, there is a time delay below which the behaviour of the system is damped, and above which it oscillates. This is termed the crossover delay. Its value decreases as one moves upstream from customer to suppliers. To ensure damped behaviour, the lower tier suppliers must receive information with less delay than the upper tier suppliers. If even one supplier suffers a time delay exceeding the crossover delay, the entire system oscillates. For a parallel system, with global ordering information, the crossover time delay is at a much higher value than for the serial system. Furthermore, the parallel structure shows a fundamentally more damped behaviour than the serial system. It is thus an easier management task to maintain a parallel system in damped mode, compared with a serial system.