This thesis aims at analyzing the completion time of a product development (PD) process. A PD process is modeled as a stochastic project network, such as PERT networks, in addition to allowing probabilistic repetition of activities (known as activity iteration-rework). When stochastic activity iteration (rework) is allowed, analyzing the process completion time becomes difficult. Most existing literature refers to the use of simulation in such scenarios; however, few analytical methods exist to solve this problem. One such method is called the Reward Markov Chain (RMC) (Smith and Eppinger, 1997a) which we use as a starting point for our proposed method. Modifications for the RMC are made to accommodate rework proportions when solving sequential networks and to find the variance. The method is also extended to solve mixed networks (combination of parallel and sequential activities) and more complicated practical issues that may arise in PD environments. Specifically, these complications are a result of coupled activities in the network, allowing parallel rework, the inclusion of dynamic rework probabilities and proportions, forward probabilities, and multiple dependency relationships between activities. Numerical examples are solved for every idea discussed, along with four case studies (three real cases and one illustrative case). All are compared to simulations and gave very close results.
Description:
Thesis (M.E.M.)-- American University of Beirut, Engineering Management Program, 2013.
Advisor : Dr. Ali Yassine, Associate Professor, Engineering Management Program--Members of Committee : Dr. Bacel Maddah, Associate Professor, Engineering Management Program ; Dr. Walid Nasr, Assistant Professor, Suliman S. Olayan School of Business.
Includes bibliographical references (leaves 62-63)
