JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT / JULY/AUGUST 1999 / 265 GENERAL-PURPOSE SYSTEMS FOR EFFECTIVE CONSTRUCTION SIMULATION By Julio C. Martinez 1 and Photios G. Ioannou 2 , Members, ASCE ABSTRACT: This paper examines the characteristics of discrete-event simulation systems in terms of their application breadth (general or special purpose), modeling paradigm (process interaction versus activity scan- ning), and flexibility (programmable or not). Several construction simulation systems are examined with primary emphasis on CYCLONE and STROBOSCOPE as representatives of the wide range of tools that are currently available. CYCLONE is a well-established, widely used, and simple system that is easy to learn and effective for modeling many simple construction operations. STROBOSCOPE is a programmable and extensible simu- lation system designed for modeling complex construction operations in detail and for the development of special-purpose simulation tools. The characteristics of these systems, as well as other recent developments, illustrate that an effective general-purpose simulation tool for construction is in essence one based on extended forms of activity cycle diagrams and the activity scanning modeling paradigm. As explained through several examples, these representations are indeed the most convenient and intuitive for construction simulation systems. Furthermore, the programmability of such a system is the principal factor that determines its power, flexibility, and ease of learning and use. INTRODUCTION Discrete-event simulation has been used to analyze and de- sign construction operations for over three decades. This ex- tensive research activity has created an implicit but limited notion of what constitutes a ‘‘construction simulation system.’’ It has also created an implicit understanding that, conceptually, these systems are indeed better suited to construction opera- tions modeling than systems commonly used in other indus- tries. A clear and explicit explanation of the essence of a con- struction simulation system and the reasons for which these systems are a natural choice for construction modeling, how- ever, are not available. This paper explains in detail the es- sence of a construction simulation system as well as the rea- sons that make such systems intuitively ideal for construction operations modeling. This sets the background for a review of the various systems that have been designed for modeling construction operations. Although a representative set of tools is covered, the emphasis of this paper is on CYCLONE (Halpin and Riggs 1992) and STROBOSCOPE (Martinez 1996). CYCLONE is a well-es- tablished, widely used, and simple system that is easy to learn and effective for modeling relatively simple construction op- erations. STROBOSCOPE is a programmable and extensible simulation system designed for modeling complex construc- tion systems in detail and for the development of special-pur- pose simulation tools. SIMULATION TOOL CHARACTERISTICS The scope, flexibility, and suitability-to-task of a simulation tool depends on its application breadth, modeling paradigm (simulation strategy), and flexibility. Application breadth is the scope of models for which the tool is designed. General-purpose simulation tools and lan- guages target a very broad domain and can be used to model almost any type of operation. General-purpose simulation tools 1 Asst. Prof. of Civ. Engrg., Virginia Polytechnic Inst. and State Univ., 200 Patton Hall, Blacksburg, VA 24061-0105. 2 Assoc. Prof. of Civ. and Envir. Engrg., Univ. of Michigan, Ann Arbor, MI 48109. Note. Discussion open until January 1, 2000. To extend the closing date one month, a written request must be filed with the ASCE Manager of Journals. The manuscript for this paper was submitted for review and possible publication on July 7, 1998. This paper is part of the Journal of Construction Engineering and Management, Vol. 125, No. 4, July/ August, 1999. ASCE, ISSN 0733-9634/99/0004-0265–0276/$8.00 + $.50 per page. Paper No. 18732. for construction modeling include those described in Halpin (1976), Chang (1986), Paulson et al. (1987), Ioannou (1989), Mohieldin (1989), Liu (1991), Odeh (1992), and Sagert (1995). In contrast, special-purpose simulators are tools that target a narrow domain such as ductile iron pipe installation. Special-purpose simulators designed for specific construction tasks include those described in McCahill and Bernold (1993), Shi and AbouRizk (1997), Oloufa and Ikeda (1997), and Mar- tinez (1997, 1998). Modeling paradigm or simulation strategy is the conceptual framework that guides model development and determines how the modeler views the system being modeled (Hooper 1986; Balci 1988). The two main simulation strategies are pro- cess interaction (PI) and activity scanning (AS). Event sched- uling (ES) is a third simulation strategy that is often combined with PI or AS. (Some authors also consider a transaction-based approach that is very similar but subtly different to PI. In the following, all statements about PI also apply to the transaction- based approach.) Flexibility reflects the capability of the tool to model com- plex situations and to adapt to a wide range of application requirements. Advanced simulation systems typically involve computer programming and are flexible enough to model very complex operations in detail. Simpler nonprogrammable tools are typically easier to learn and can be used to model many simple operations effectively. However, they often require as- sumptions that prevent the effective analysis of many complex or detailed operations. Simulation Strategy By far the most significant characteristic of any general- purpose discrete-event simulation system is its simulation strategy. The main simulation system strategies in use today for modeling construction processes are PI and AS. A PI model is written from the point of view of the entities (transactions) that flow through the system. These entities typ- ically arrive, undergo some processing where they capture and release scarce resources, and then exit. This strategy is partic- ularly suited to modeling operations where the moving entities are differentiated by many attributes and where the machines or resources that serve these entities have few attributes, a limited number of states, and do not interact too much (Hooper 1986). Most operations in manufacturing and the industrial and service industries are of this type. Consequently, a large num- ber of commercial simulation tools are based on the PI para-