knowledge transfer


Taming the Bullwhip Effect with Fuzzy Optimisation Methods

Submitted by Christer Carlsson, 18.12.2001, IBA E


The logistical chain between producer and customer is highly integrated in the fine papers market. Information about the true demand of the end-customer market is collected routinely and is then passed up the supply chain by the actors step-by-step. This is, however, not a pure information transition as the actors also take note of and both plan and act on the basis of the information they handle. The actors are rational agents and try to make optimal decisions. There is some friction in the information transition, which causes the actors to take precautions and build buffer inventories against unforeseen demand changes. This will induce variations in the demand estimates, which will grow as the information works its way up the supply chain. This is called the bullwhip effect and causes annual costs of 50-60 M for a paper mill with an annual production capacity of 300 000 ton.


IAMSR proved that the bullwhip effect is caused by (i) the assumption of rational actors, (ii) optimisation methods building on rational actions, and (iii) the assumption that stochastic models are suited for handling imprecise information. Thus the classical methods used for dealing with the bullwhip effect do not work properly, they may even contribute to the problem. The stochastic elements of the bullwhip models were replaced with fuzzy parameters and new solution methods were worked out and proved to converge to optimal solutions. Then the methods were implemented as models on a hyperknowledge platform and introduced in a corporate intranet.

Status and results

The fuzzy bullwhip models worked surprisingly well. The method introduced a new ordering policy for the fine paper markets, which the paper mills accepted after some trials. IAMSR also worked out a new policy for the actors of the supply chain to share information over secure inter-/intranet links. This policy was quickly abandoned and later on effectively sabotaged by the whole-sellers, which found out that their possibilities to build good negotiation positions were reduced by the information sharing. The fuzzy bullwhip models are now further developed for the steel industry.

Adaptivity and portability

The fuzzy bullwhip system (a system of 4 interactive models) adapts to changing market conditions and to changes in the variance of demand. In terms of the EUNITE levels of adaptivity the fuzzy bullwhip system is of Level I.

More information


technical reports & case studies

executive summaries

scientific papers




search eunite with