Evaluation of queuing systems for knowledge-based simulation of construction processes

a r t i c l e i n f o During the course of a construction project, there are many situations in which formation of waiting lines or queues is inevitable. The effect of resource delays in queues on the overall project completion time and cost has motivated researchers to employ simulation for analysis of queuing systems in order to identify the best operational strategies to reduce the time wasted in queues. Providing proper and timely input data with high spatial and temporal accuracy for queuing systems simulation enhances the reliability of decisions made based upon the simulation output. Hence, the presented paper describes a methodology for collecting and mining of spatio-temporal data corresponding to the interactions of queue entities to extract computer interpretable knowledge for simulation input modeling. The developed framework was validated using empirical datasets collected from a series of experiments. The extracted relevant knowledge from the queuing system entities was used to update corresponding simulation models. Waiting lines or queues exist in almost all industrial and manufacturing processes. In all such queuing systems, there are entities that need to be repetitively processed by other entity(s). The entity waiting in a line to receive service is called a client and the entity that processes clients is called a server. Similar to queuing systems in manufacturing settings, in many construction systems, clients (or resources) can be delayed in waiting lines when a server (or processor) is already captured by a previously arrived client and thus is busy. A classic example of a construction queuing system is the arrival of dump trucks in a loading area where excavators or front end loaders load them with soil. As shown in Fig. 1, cyclic activities of an earthmov-ing operation consist of load, haul, dump, and return processes. A part of this cycle that embraces the waiting line and server is considered as the queuing system. Therefore, it is clear that the boundaries of the system are not necessarily spatially fixed and can dynamically change depending on the length of the queue and the efficiency of the server. As soon as a client arrives inside the boundaries of the system, depending on the state of the server (i.e. idle or busy), it either waits in the queue or proceeds to be served immediately. Once the service is completed, the client leaves the system and its state, attributes, and other properties will no …