WSC 2004 Final Abstracts
Tuesday 3:30:00 PM 5:00:00 PM
Construction Engineering I
Chair: Dan Halpin (Purdue University)
This paper presents research that led to the design and implementation of practical 3D animation methods to visualize multiply-articulated construction equipment in 3D animations of discrete-event construction process models. Using principles of forward and inverse kinematics, we designed and implemented generic pieces of multiply-articulated virtual construction equipment that accept task-level instructions from external software processes. Discrete event simulation models can configure and instantiate specific pieces of such equipment and instruct them to perform common construction tasks using simple, parametric statements of text. Once instructed to perform specific tasks (e.g. Load dirt), these “smart” pieces of equipment (e.g. Backhoes) automatically decipher the sequence and amplitudes of the elemental motions their components (e.g. Boom) must undergo to accomplish those tasks. The animation methods are implemented in a software tool called KineMach that integrates as an add-on with the VITASCOPE visualization system.
Simulation Input Updating Using Bayesian Techniques
Tae Hwan Chung, Yasser Mohamed, and Simaan M AbouRizk (University of Alberta)
Simulation built on assumption and approximation has been traditionally utilized to make predictions prior to construction. Although there are many benefits of simulation such as its capability of multiple experiments with various scenario assumptions, it may lead to erroneous predictions when simulation input data are not accurate. Long-term repetitive projects such as tunnel construction provide opportunities to fine-tune the simulation input parameters based on real project progress. Bayesian updating techniques represent a very effective approach to enhance the quality of the estimates based on the observed data. This paper outlines some benefits that can be achieved using Bayesian updating techniques. The major benefits of these techniques includes more accurate simulation outcome even at the early stage of the project.
Implementing a General Purpose Framework Using Multi-Agents for Construction Management Education
Amlan Mukherjee, Eddy M. Rojas, and William D. Winn (University of Washington)
The need for contextually rich educational experiences in construction education has led researchers to explore alternatives using gaming and simulation environments. The work done so far has concentrated on special and general purpose simulations of construction operations with limited interactivity. We claim that it is necessary to have an extensible general purpose framework, for developing a wide variety of situational simulations of construction management processes, that can be used to create interactive environments for training managers. We have introduced such a framework and developed it using multiple autonomous agents. The Virtual Coach is an implementation of the developed framework. It was tested on a class of 19 construction management senior students, and proved to be an useful educational tool.
Practical 3D Animation of Multiply Articulated Construction Equipment
Vineet Rajendra Kamat (University of Michigan) and Julio Cesar Martinez (Virginia Polytechnic Institute and State University)
Wednesday 8:30:00 AM 10:00:00 AM
Construction Engineering II
Chair: Simaan Abourizk (University of Alberta)
From the perspective of flow analysis, the more complex the project, the more wastes are prone to build up, due to the increasing number of interfaces between activities. When processes are further expanded beyond the jobsite to the scale of a supply chain, the complexity is usually beyond human perception. In order to provide a systematic approach to help management make correct and timely decisions, an analytical approach capable of efficiently and economically modeling the complicated processes and to address various managerial questions is necessary.
An analytical framework combining simulation, design of experiment, regression analysis, and mathematical programming is proposed to facilitate management to determine the optimal design for a construction process under various constraints. A simplified concrete delivery process is used as an example to illustrate how the framework is implemented. Some implementation issues and other potential applications will be discussed.
Modeling Concurrent Operational Interruptions
in Construction Activities with Simplified Discrete Event Simulation Approach
Ming Lu and Wah-Ho Chan (Hong Kong Polytechnic University)
Numerous previous applications have demonstrated the high potential of discrete simulation methodologies in designing and analyzing construction systems. Still, construction simulation applications largely remain software exercises at the academic level because of the time and effort as required in constructing a valid simulation model that matches the real processes and operations. This paper introduces an enhanced version of the Simplified Discrete Event Simulation Approach (SDESA) by incorporating a concurrent interruptions model, which is intended to make construction simulations more realistic without compromising the simplicity of the original SDESA. Particular emphasis is placed on how to model the effects of operational interruptions upon the system performance by applying SDESA. The well-established CYCLONE method acts as a cross-validation tool to check the results given by SDESA. And the validity and simplicity of SDESA modeling are illustrated with a simple earthmoving operation simulation and a real hoist and barrow concreting operation simulation.
Decision Tree Module Within Decision Support Simulation System
Mohamed Moussa, Janaka Yasantha Ruwanpura, and George Jergeas (University of Calgary)
Decision trees are one of the most easy to use tools in decision analysis. Problems where decision tree branches are based on random variables have not received much attention. This paper introduces a successful application of a Special Purpose Simulation (SPS) program in developing a Decision Tree module that is part of a unified Decision Support System (DSS) template. The DSS template consists of three modules: Decision Tree (DT), shortest and longest path Dynamic Programming (DP) Network, and Cost / Time (CT) Estimate network. The DT module integrates with other modules and allows users to model decision trees with variables that are based on probabilistic random numbers. This paper introduces the DSS-DT module and shows its advantages.
Simulation Experiment for Improving Construction Processes
Shihyi Wang and Daniel W. Halpin (Purdue University)
Wednesday 10:30:00 AM 12:00:00 PM
Construction Engineering III
Chair: Anil Sawhney (Arizona State University)
Project Risk Simulation Under Uncertain Construction Duration
Kun-jung Hsu (Leader University)
This paper establishes a two-phase model to explore the financial risk of construction project under uncertain con-struction duration. Treated as a random variable in the model, the construction duration affects the random present worth of cash flow in both construction and operation phases. An illustration of BOT project under uncertain construction duration was utilized for simulation. The simulation result was applied to reveal how uncertain con-struction duration affects project risk. The paper also shows how the standard deviation of the NPV increases whenever the variance of construction duration rises, thus accentuating the contingency of the project. The resultsˇ¦ analysis extended to the discussion of its effect on the probability of negative NPV and its theoretical meanings.
Statistical Analyses for Simulating Schedule Networks
Dong-Eun Lee and Jonathan Jingsheng Shi (Illinois Institute of Technology)
This paper introduces the Stochastic Project Scheduling Simulation (SPSS) system and two additional useful statistical analysis tools. SPSS integrates CPM, PERT, and Discrete Event Simulation (DES) scheduling methods into one system. It can generate a CPM-based deterministic schedule, a PERT-based probabilistic schedule, and a simulation-based stochastic schedule for the project under study. Two additional statistical analysis tools are added to assist users in determining the number of simulation runs needed for a given schedule network and testing the significance of difference between two independent simulation experiments.
Development of a Parametric Estimating Model for Technology-Driven Deployment Projects
Rudy J. Watson and Young Hoon Kwak (The George Washington University)
The purpose of this paper is to define the process used to develop a parametric estimating model and to explain the use of the model as it allows a non subject matter expert to predict the cost of deployment of a technology-driven project with improved accuracy. The overall objective is to define criteria for organizations to use as a rule of thumb estimating model for determining potential resource requirements during the pre-conceptual and conceptual phase of a technology-driven project. The use of the model will determine the complexity level of a project and by using a matrix simulating the task complexity knowledge of a deployment subject matter expert, the organization deployment resource required will be estimated based upon responses to certain inquiries. The user of the model will then have the capability to modify the output to derive their specific deployment estimates.