Combining forward and backward process simulation for generating and analysing construction schedules

In the construction industry there is an essential requirement for efficient construction schedules that consider the availability of resources as well as constraints. It is standard practice in today’s construction process planning to use the Precedence Diagram Method (PDM), including wellestablished methods such as the Critical Path Method (CPM). A possible means of improving the mostly manually executed generation process is to apply computer-aided methods such as discreteevent simulation. By combining this method with resource-constrained scheduling methodology it becomes possible to not only verify schedules but also to generate schedules that take resources into consideration. In addition to discrete-event simulation CPM is able to analyse float times, but is not able to consider resources. Discrete-event simulation on the other hand can consider resources, but is unable to calculate float times. The calculation of float times for construction tasks is an important task. An awareness of float times makes it possible to identify whether a delayed task can be compensated for or if the entire construction project will be late. In this paper we propose using Backward Simulation methodology to develop an elaborated approach that extends discrete-event simulation by the ability to calculate float times. Here, the simulation starts at the virtual completion date of the construction project and runs backwards in time until the start date. To determine float times it is important that the task execution sequence is in the same order for forward and backward simulation. Consequently we present an extension to the simulation concept that controls the execution order of the tasks within the backward simulation. By combining the results of the forward and backward simulation, it is possible to determine float times for each task while taking into account resources. A comprehensive case study is described that illustrates the application of this new approach. One result of this is the determination of detailed float times for each task using discrete-event simulation.