A probabilistic approach to determining the length of schedule works

Abstract

Summary.  Research is devoted to scheduling and solving scientific  problem  of improving the reliability  of the planning and schedule  implementation.  This  article  discusses  the  factors  that  affect  the  estimated  time  of  performing  any  construction  and assembly operations  in the schedule. The method is based on the approach that the time of completion of a scope of work  is derived from the performance of a performer and the intensity of the work. To use this approach  it is required to apply statistical data of  kib performer at similar facilities in the past. So work completion time is not determined value but range (interval) of possible values that can be described by a normal distribution, β-  and α-distributions.  According this method  any planned work in the schedule is shown as "fan" function, which allows to see  an increase of uncertainty over time. The thesis  describes a method of calculation of the work organization taking into account such uncertainty. We describe the most typical cases of interlinkages between works in the calendar plan, performed by the proposed method. The proposed approach allows  to neutralize the contradiction between the probabilistic nature of real processes and deterministic methods of their description.  The  presence of the of the determined value of a parameter only  is  not  enough,  because  it  is  uncertain  its  position  in  the  possible  range  of  values  from  the  minimum  to  the  maximum. Manufacturing processes and conditions defining the performance indicators are  in the future, and prediction (forecasting) of their values  can  be  done  only  with  a  certain  level  of  probability.  Therefore,  schedules  based  on  real,  probabilistic  nature  of  their performance, allows you to get even more pessimistic values, but at the same time more reliable probability of achievement.