TY - JOUR
T1 - Agent-embedded system dynamics (aeSD) modeling approach for analyzing worker policies
T2 - a research case on construction worker absenteeism
AU - Hwang, Sungjoo
AU - Ahn, Seungjun
AU - Lee, Sang Hyun
N1 - Funding Information:
This paper forms part of a special section ?Hybrid Simulation in Construction?, guest edited by Farnad Nasirzadeh, Susan Howick and SangHyun Lee.
Publisher Copyright:
© 2021, Emerald Publishing Limited.
PY - 2021
Y1 - 2021
N2 - Purpose: Both system dynamics (SD) and agent-based modeling (ABM) have been used in simulation-based group dynamics research. To combine the advantages of both simulation approaches, the concept of SD-ABM hybrid simulation has been proposed. However, research efforts to compare the effectiveness of modeling approaches between the hybrid and non-hybrid models in the context of group dynamics study are rare. Against this background, this study aims to propose an agent-embedded SD (aeSD) modeling approach and demonstrate its advantages when compared to pure SD or ABM modeling approaches, based on a research case on construction workers’ social absenteeism. Design/methodology/approach: The authors introduce an aeSD modeling approach to incorporate individual attributes and interactions among individuals in an SD model. An aeSD model is developed to replicate the behavior of an agent-based model previously developed by the authors to study construction workers’ group behavior regarding absenteeism. Then, the characteristics of the aeSD model in comparison with a pure ABM or SD model are demonstrated through various simulation experiments. Findings: It is demonstrated that an aeSD model can capture the diversity of individuals and simulate emergent system behaviors arising from interactions among heterogeneous agents while holding the strengths of an SD model in identifying causal feedback loops and policy testing. Specifically, the effectiveness of the aeSD approach in policy testing is demonstrated through examples of simulation experiments designed to test various group-level and individual-level interventions to control social absence behavior of workers (e.g. changing work groupings, influencing workgroup networks and communication channels) under the consideration of the context of construction projects. Originality/value: The proposed aeSD modeling method is a novel approach to how individual attributes of agents can be modeled into an SD model. Such an embedding-based approach is distinguished from the previous communication-based hybrid simulation approaches. The demonstration example presented in the paper shows that the aeSD modeling approach has advantages in studying group dynamic behavior, especially when the modeling of the interactions and networks between individuals is needed within an SD structure. The simulation experiments conducted in this study demonstrate the characteristics of the aeSD approach distinguishable from both ABM and SD. Based on the results, it is argued that the aeSD modeling approach would be useful in studying construction workers’ social behavior and investigating worker policies through computer simulation.
AB - Purpose: Both system dynamics (SD) and agent-based modeling (ABM) have been used in simulation-based group dynamics research. To combine the advantages of both simulation approaches, the concept of SD-ABM hybrid simulation has been proposed. However, research efforts to compare the effectiveness of modeling approaches between the hybrid and non-hybrid models in the context of group dynamics study are rare. Against this background, this study aims to propose an agent-embedded SD (aeSD) modeling approach and demonstrate its advantages when compared to pure SD or ABM modeling approaches, based on a research case on construction workers’ social absenteeism. Design/methodology/approach: The authors introduce an aeSD modeling approach to incorporate individual attributes and interactions among individuals in an SD model. An aeSD model is developed to replicate the behavior of an agent-based model previously developed by the authors to study construction workers’ group behavior regarding absenteeism. Then, the characteristics of the aeSD model in comparison with a pure ABM or SD model are demonstrated through various simulation experiments. Findings: It is demonstrated that an aeSD model can capture the diversity of individuals and simulate emergent system behaviors arising from interactions among heterogeneous agents while holding the strengths of an SD model in identifying causal feedback loops and policy testing. Specifically, the effectiveness of the aeSD approach in policy testing is demonstrated through examples of simulation experiments designed to test various group-level and individual-level interventions to control social absence behavior of workers (e.g. changing work groupings, influencing workgroup networks and communication channels) under the consideration of the context of construction projects. Originality/value: The proposed aeSD modeling method is a novel approach to how individual attributes of agents can be modeled into an SD model. Such an embedding-based approach is distinguished from the previous communication-based hybrid simulation approaches. The demonstration example presented in the paper shows that the aeSD modeling approach has advantages in studying group dynamic behavior, especially when the modeling of the interactions and networks between individuals is needed within an SD structure. The simulation experiments conducted in this study demonstrate the characteristics of the aeSD approach distinguishable from both ABM and SD. Based on the results, it is argued that the aeSD modeling approach would be useful in studying construction workers’ social behavior and investigating worker policies through computer simulation.
KW - Absenteeism
KW - Agent-based modeling
KW - Agent-embedded system dynamics (aeSD)
KW - Construction engineering management
KW - Construction management
KW - Construction workers
KW - Group dynamics
KW - HRM
KW - People
KW - System dynamics
KW - Team working
UR - http://www.scopus.com/inward/record.url?scp=85106335378&partnerID=8YFLogxK
U2 - 10.1108/CI-04-2020-0067
DO - 10.1108/CI-04-2020-0067
M3 - Article
AN - SCOPUS:85106335378
SN - 1471-4175
VL - 21
SP - 379
EP - 397
JO - Construction Innovation
JF - Construction Innovation
IS - 3
ER -