Saaty's AHP is helpful in evaluating alternatives thanks to its effective procedure to determine the relative weights of several comparison criteria. Combining the results of expert interviews, AHP can be very useful for a company in choosing a third party logistics service provider (3PL). However, in the traditional AHP procedure, several results may be rejected when the consistency ratio (CR) of the respondent exceeds a certain threshold. As a consequence, AHP interviews may be repeated several times with a consequent waste of time. In many industrial domains, a faster way to choose a supplier would thus be appreciated. In this paper we propose a mathematical method that combines AHP, DEA and linear programming in order to support the multi-criteria evaluation of third party logistics service providers. The proposed model aims to overcome the limitation of the AHP method, merging experts' indications with objective judgments which originate from historical data analysis. Suppliers' past performance is thus used to correct eventual errors resulting from the acceptance of interviews where the consistency ratio is high. The proposed model has been validated on the real case of an international logistics service provider.

Supply Chain Network Design (SCND) deals with the determination of the physical configuration and infrastructures of the supply chain. Specifically, facility location is one of the most critical decisions: transportation, inventory and information sharing decisions can be readily re-optimized in response to changes in the context, while facility location is often fixed and difficult to change even in the medium term. On top of this, when designing a supply network to support a new product diffusion (NPD), the problem becomes both dynamic and stochastic. While literature concentrated on approaching SCND for NPD separately coping with dynamic and stochastic issues, we propose an integrated optimisation model, which allows warehouse positioning decisions in concert with the demand dynamics during the diffusion stage of an innovative product/service. A stochastic dynamic model, which integrates a Stochastic Bass Model (SBM) in order to better describe and capture demand dynamics, is presented. A myopic policy is elaborated in order to solve and validate on the data of a real case of SCND with 1,400 potential market points and 28 alternatives for logistics platforms.