Analyzing linear recursive projects as an absorbing chain英文文献资料.docVIP

ANALYZINGLINEARRECURSIVEPROJECTSAS ANABSORBINGCHAIN DOL.MINHANDR.BHASKAR Received15October2004;Accepted20June2005 Hardie (2001) used general Markov chains to study the linear “recursive projects” in which some activities mayberevisited afteralateractivity iscompleted. Inthispaper, weproposethatitisbettertotreattheprojectasanabsorbing chain.Thisallowsusto calculatetheexpectedvalueandthevarianceoftheprojectduration. Copyright?2006D.L.MinhandR.Bhaskar. This isanopen access article distributed undertheCreativeCommons Attribution License,whichpermits unrestricted use,dis- tribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited. 1.Introduction Project management isusually calculated asaset ofstatic activities, whereas inreality these activities tendtobedynamic involving multiple feedbacks (ParkandPe?na-Mora [5]). This results in many late project completions. Recently research has been done to formulate and analyze various methods to improve project prediction and control. These include concurrent engineering (Carter and Baker [1]) and stage-gate systems (Cooper[2])inwhichprojectsmustmeetspeci?cations beforetheypassfromonestage toanother.Theactivities within aprojectaregoverned byconstraints aswellasprece- dence relationships intraditional network scheduling likeCPM (critical path method, DupontInc.andRemingtonRand,1958),PDM(procedurediagramming method,IBM Co.,1964)andPERT(programevaluationandreviewtechnology,USNavy,Booz-Allen HamiltonandLockheedCo.1958). Normally,inaCPM/PERTproject,weassumethattheactivities arecarriedoutonly once. Hardie[3]observed that thisisnottrue inmanyprojects, inwhich some activ- itiesmayhavetoberevisited afteralateractivity iscompleted. Hecalled suchmodels the“recursivenetworks.”Forexample, consider aproject having thefollowing ?veac- tivities: (1) speci?cation, (2) design, (3) testand integration, (4) manufacture and (5) marketing. Weassume that they arelisted inthepredecessor order; that i