CYCLIC STOCHASTIC ALTERNATIVE NETWORK MODELS FOR PROJECT MANAGEMENT

摘要

Project implementation process modelling is the main active methodological body of the Project Management discipline. The efficiency of decisions made and the whole functioning of the PM system is determined by the adequacy of models for real processes and their meeting the requirements of project management tasks and goals.rnThe high degree of complexity and laboriousness of drawing up timing schedules for numerous activities performed by many project members using a great range of resources, strict requirements for the quality of plans, the need for regular control of their fulfilment and adjustment call for the proper methods of solving problems of such sophisticated nature.rnToday world market of Project Management Software presents products with network models and methods of scheduling based on the researches of the end of 50th- beginning of 70th with very limited possibilities. At the same time the current mathematical methods of modelling project processes (classical network models, generalised, probabilistic and stochastic network models) do not always appear adequate to the complex reality of the modelled process. It should be noted that it refers to each method taken separately and to some combinations of these methods.rnThe paper describes a new class of network models adequately reflecting the complex project realisation process that are used for stating and solving optimal management tasks for this project. This class of models is a synthesis of generalized network models (with their rich spectrum of means for equivalence conversion of models and describing the different logical and time interrelations between of project activities) with probabilistic and stochastic models to a considerable extent taking into account factors of risk and uncertainty the implementation of a project involves. These models (further referred to as cyclic alternative network models - CANMs) are the most flexible and adequate in the range of known tools for describing the process of managing and control over the development of a complex sophisticated project. CANMs offer all the advantages of generalised and stochastic models in comparison with traditional network models while at the same time involving just a slight complication of the language used for describing CANMs.rnA general description of CANM category models was given in a number of works.rnIn the present paper a detailed mathematical description of CANMs is given and supplied with the substantiation of the requisite conditions of consistency as well as problem statements and CANM time analysis algorithms.