System dynamics, Discrete Event Simulation or Agent-Based Modeling
Dynamic simulation modeling methods are used to model processes and systems. They provide the ability to run scenarios and learn the consequences of different decisions before expensive interventions are made in practice. These modeling techniques are very useful in many fields (from logistic inventory purchasing decisions to healthcare delivery process improvement).
For me as a modeler, the main differences between the three most commonly used methods (system dynamics (SD), discrete event simulation (DES), and agent-based modeling (ABM)) were not always clear. In my experience, the choice of the dynamic simulation model is usually done according to the previous experience of the modeler. Many times I was tempted to utilize the known and familiar DES for different problems I encountered, but other modelers suggested using SD or ABM. Recently I stumbled upon a very good series of papers, published by the International Society For Pharmacoeconomics and Outcomes Research (ISPOR), which provides an excellent overview and differentiation criteria. The reports are looking at the techniques using a healthcare applications prism, but in my opinion, the discussed concepts can be useful for other domains as well. The first report, entitled "Applying Dynamic Simulation Modeling Methods in Health Care Delivery Research—The SIMULATE Checklist: Report of the ISPOR Simulation Modeling Emerging Good Practices Task Force", provides an overview of the SD, DES and ABM methods and examples of health care system interventions in which these are useful. It can be found here: ISPOR part 1
The second report, entitled "Selecting a Dynamic Simulation Modeling Method for Health Care Delivery Research—Part 2: Report of the ISPOR Dynamic Simulation Modeling Emerging Good Practices Task Force" reviews criteria for selecting the most suitable method depending on 1) the purpose— type of problem and research questions being investigated, 2) the object—scope of the model, and 3) the method to model the object to achieve the purpose.
I think that the most important conclusion, however, is presented in the commentary to these reports and verifies my initial gut feeling. "Dynamic Simulation in Health Care Comes of Age" commentary states that "the specific boundaries between the SD, DES, and ABM simulation methods are growing very hard to define, and to some degree, the differences between these modeling types started as different software systems to represent dynamic problems rather than truly different methodologies... There is more overlap than uniqueness in the criteria [for choosing a model type], and the relative superiority of one modeling type over another is hard to define explicitly" and "virtually any set of feedback loops typically described in an SD model can be represented by an appropriately constructed agent-based model, and entities within appropriately constructed DES models can contain algorithmic responses to the virtual environment that replicates the autonomous behavior of an agent in an ABM model. DES models can exhibit emergent behavior just as ABM models do, and accurate representations of complex realistic systems may require hybrid approaches that use components across multiple modeling types."
I would like to conclude with a recommendation to a free Simulation software, that supports all three modeling approaches, Anylogic, which can be downloaded here:
