A new analytical algorithm for computing probability distribution of project completion time

Abstract

An analytical algorithm was presented for the exact computation of the probability distribution of the project completion time in stochastic networks, where the activity durations are mutually independent and continuously distributed random variables. Firstly, stochastic activity networks were modeled as continuous-time Markov process with a single absorbing state by the well-know method of supplementary variables and the time changed from the initial state to absorbing state is equal to the project completion time. Then, the Markov process was regarded as a special case of Markov skeleton process. By taking advantage of the backward equations of Markov skeleton processes, a backward algorithm was proposed to compute the probability distribution of the project completion time. Finally, a numerical example was solved to demonstrate the performance of the proposed methodology. The results show that the proposed algorithm is capable of computing the exact distribution function of the project completion time, and the expectation and variance are obtained.

Keywords

stochastic activity networks / project completion time / distribution function / Markov process / supplementary variable technique

References

[1] HardieN.. The prediction and control of project duration: A recursive model [J]. International Journal of Project Management, 2001, 19(7): 401-409

[2] CoppendaleJ.. Manage risk in product and process development and avoid unpleasant surprises [J]. Engineering Management Journal, 1995, 5(1): 35-38

[3] ElmaghrabyS. E.. The estimation of some network parameters in the PERT model of activity networks: Review and critique [C]. Advances in Project Scheduling, 1989, Amsterdam, Elsevier: 371-432

[4] YaoM.-j., ChuW.-ming.. A new approximation algorithm for obtaining the probability distribution function for project completion time [J]. Computers & Mathematics with Applications, 2007, 54(2): 282-295

[5] MartinJ. J.. Distribution of the time through a directed acyclic network [J]. Operations Research, 1965, 13(1): 46-66

[6] LeemisL., DugganM., DrewJ., MallozziJ., ConnellK.. Algorithms to calculate the distribution of the longest path length of a stochastic activity network with continuous activity durations [J]. Networks, 2006, 48(3): 143-165

[7] KulkarniV. G., AdlakhaV. G.. Markov and Markov-Regenerative PERT networks [J]. Operations Research, 1986, 34(5): 769-781

[8] van SlykeR. M.. Monte Carlo methods and the PERT problem [J]. Operations Research, 1963, 11(5): 839-861

[9] FishmanG. S.. Estimating critical path and arc probabilities in stochastic activity networks [J]. Naval Research Logistic Quarterly, 1985, 32(2): 249-261

[10] AlfaA. S., RaoT. S.. Supplementary variable technique in stochastic models [J]. Probability in the Engineering and Informational Sciences, 2000, 14(2): 203-218

[11] HouZ.-t., LiuG.-xin.Markov skeleton processes and their application [M], 2005, Beijing, Science Press and International Press: 1-6

[12] HouZ.-ting.. Markov skeleton processes and application to queueing systems [J]. Acta Mathematicae Applicatae Sinica, 2002, 18(4): 537-552

[13] HouZ.-t., YuanC.-g., ZouJ.-z., LiuZ.-m., LuoJ.-w., LiuG.-x., ShiPeng.. Transient distribution of the length of GI/G/n queueing systems [J]. Stochastic Analysis and Applications, 2003, 21(3): 567-592

[14] YuZ., LiuZ.-m., HouZ.-ting.. Distribution of the transient queue length in SMAP/INID/1 [J]. Journal of Central South University: Science and Technology, 2004, 35(2): 341-344

[15] AzaronA., Tavakkoli-moghaddamR.. A multi-objective resource allocation problem in dynamic PERT networks [J]. Applied Mathematics and Computation, 2006, 181(1): 163-174

[16] AbdelkaderY. H.. Evaluating project completion times when activity times are Weibull distributed [J]. European Journal of Operational Research, 2004, 157(3): 704-715

[17] AbdelkaderY. H.. Erlang distributed activity times in stochastic activity networks [J]. Kybernetika, 2003, 39(3): 347-358