Animation of Complex Construction Simulation Models

One of the major impediments in using computer simulation to model and optimize construction processes has been the fact that decision makers often do not have the training nor the time to check the validity of simulation models prepared by others and thus have little confidence in the results. Animation is an increasingly popular technique that can be used to verify, debug and validate simulation models. Perhaps most importantly, it can also show in an easy to understand manner the insights gleaned from building and running the underlying model, thus increasing its credibility significantly. This paper presents a simulation model for the movement of people inside a building served by a single elevator with fairly complex control logic that was verified using animation. The simulation model for this example was developed using Stroboscope and verified using Proof Animation. Animation of Simulation Models Recent advances in computer animation are a main contributor to the increased popularity of simulation modeling. Typically, the animation of a simulation model involves a graphical computer application that uses changes in the position, shape, or color of icons to illustrate changes in the state of the simulated system. These icons often represent interacting resources (labor, equipment, materials, space, etc.). The result is a precisely-described, smoothly-moving depiction of a complex system whose states are constantly changing. Animation of simulation models differs from cartoon-like animation in that it does not attempt to create photo-realistic sequences of frame-by-frame effects. It is the state of the system that the animation software has to depict faithfully. While showing changes in the system state, the animation produces a visual rendition (possibly to scale) that is meaningful to a human observer. Almost all major simulation languages on the market today support animation in one of two modes. In concurrent mode, animation is rendered while the system is being simulated (albeit slowly to allow for visual comprehension). In playback mode, animation is rendered 1Associate Professor, Civil & Environmental Engineering Department, University of Michigan, Ann Arbor, MI 48109-2125. 2Research Fellow, Civil & Environmental Engineering Department, University of Michigan, Ann Arbor, MI 48109-2125. Animation of Complex Construction Simulation Models 2 P.G. Ioannou & J.C. Martinez based on changes in the system that have been recorded earlier and saved to a disk file by a simulation model. The animation of simulation models can serve a variety of purposes. During model development process animation is useful for verifying that the simulation model behaves correctly and represents what the model developer had in mind. It is not possible to animate what is not already simulated. Thus, the simulation model has to provide a level of detail than can support realistic animation. At the same time it can be used to debug the model and find out why it does not behave as intended. Debugging a complex model is a difficult and arduous task. Visualizing the changing states of an entire system and tracing its behavior through a segment of simulation language code or a simulation trace output file can be extremely timeconsuming. In this respect, computer animation can serve as the ideal trace mechanism for revealing the complex interrelationships represented by the model. Often the developer of a simulation model is not an expert in the process being represented. In this case, animation can be used for model validation. Animations of the model can be shown to a process expert that can point out flaws not just in the model itself but also in way the model developer understands the behavior of the system. Thus, animation can provide feedback and increase everyone’s confidence that the model is indeed a realistic representation of the underlying system. Perhaps the most important use of animation is to communicate the simulation model to others and boost the credibility of its results. This is particularly important when the audience includes managers and other key personnel that are directly responsible for making decisions that affect system performance but do not understand simulation modeling. In this case, animation is an indispensable tool for increasing everyone’s confidence. Instead of taking the blind advice of others, managers can see for themselves the insights gleaned from building and running the underlying model and appreciate why some strategies make sense while others do not. Modeling Elevator Operations The remainder of this paper describes the simulation model for a complex system that was verified through animation. The example used is that of a building served by a single elevator. This fairly complex example has been adapted from (Law and Kelton 1991) and is used to illustrate the effectiveness of animation to verify that the model is an accurate representation of the elevator’s complicated control logic. The problem is of interest in high-rise building construction (50+ stories) because subcontractors add money to their bid to account for lost time in hoisting personnel. If at the bidding stage a general contractor can show the subcontractors a video and documentation that the proposed hoisting system has been optimized and that wait times will be as short as possible, there may be cost benefits to all parties (including the owner). In order to do this, project managers have expressed the need for models depicting the movement of labor that can provide queue-related statistics for different hoisting policies. In general, such models should be able to analyze and optimize the operation of multiple elevators and hoists (some of which may or may not span the entire building height) in very tall buildings. In addition to being interesting, this problem was also selected because the “classical” elevator problem (even with simplified control logic) is one of the most difficult problems for Animation of Complex Construction Simulation Models 3 P.G. Ioannou & J.C. Martinez simulation languages to model and verify. It has been estimated, for example, that the elevator problem (with much simpler control logic than the example described below) requires an expected modeling effort of 20-30 hours of work (Chisman 1996). Example Problem Statement A five-story building is served by a six-person elevator. People arrive to the ground floor (floor 1) with independent exponential interarrival times having a mean of 1 minute. A person will go to each of the four upper floors with probability 0.25. It takes the elevator 15 seconds to travel up or down one floor. For simplicity it will be assumed that the elevator loading and unloading time at any floor is zero. The length of stay of a person at a particular floor is distributed uniformly between 15 and 120 minutes. When a person leaves floor i = 2, 3, 4, 5, he or she will go to floor 1 with probability 0.7, and will go to each of the other three floors with probability 0.1. A person coming down to floor 1 departs from the building immediately. When the elevator is going up, it will continue in that direction if a current passenger wants to go to a higher floor or if a person on a higher floor wants to get on the elevator. When the elevator is going down, it will continue in that direction if it has at least one passenger or if there is a waiting passenger at a lower floor. When the elevator stops at floor i = 2, 3, 4 while going up (down), it picks up only those people at that floor that want to go up (down). If the arriving elevator does not have enough room to get all the people waiting at a particular floor, the excess remain in queue. The elevator decides at each floor what floor it will go to next. It does not change directions between floors. At the start of the simulation the elevator is at rest at its base floor. This is the same floor the elevator returns to whenever it is idle. The best choice for the base floor is made by minimizing the average of individual delays over all floors and all people.