Optimizing the exchange of partial information in integrated product development with multiple activities.

Abstract

Integrated product development (IPD) is the overlapping of nominally sequential activities while considering downstream concerns, in an effort to reduce the time-to-market of the product. The IPD practice helps companies to either pre-empt the competition and beat them to market, or respond quickly to changes in the market and quickly align their products with evolving customer needs. Working with incomplete information, downstream activities are subject to the risk of rework; thus, entailing more development costs. In extreme cases these costs could surpass expected benefits. Prior research has shown that it may not be optimal to always consider partial information. On the other hand, research also showed that never considering partial information could entail high rework costs. Therefore, there is a need for an optimal policy to manage exchange of partial information in an IPD environment. In this thesis, we formulate a dynamic programming model to manage upstream partial information flow in a multi-activity IPD process (m 2 activities). The multi-activity aspect is novel as all previous research considers two-activity models. We then resort into solving the dynamic program directly and performing an extensive Monte Carlo simulation study to analyze the behavior of the optimal policy. The simulation results suggest several important insights regarding the timing and frequency of considering partial information in an IPD environment. The study showed that upstream activities would consider more information in IPD environments, and they do so earlier. Most notably, we observe a reverse bullwhip effect in IPD environments where the effect of variability of information is dampened rather than amplified downstream. Finally, we present a decomposition heuristic to easily approximate the decision policy, by solving a sequence of two-activity models. The heuristic performs very well yielding near-optimal results at significantly lower computer storage requirement. This enhances the applicability of the researc