Lecture 28: Simulation Modeling

Lecture 28: Simulation Modeling#

Note

Simulation modeling is a powerful analytical tool for studying, experimenting with, and predicting the behavior of complex systems. Instead of experimenting directly with the real system—which can be costly, risky, or impractical—simulation creates a controlled, virtual environment that mimics system dynamics and allows analysts to explore “what-if” scenarios, test alternative designs, and make informed decisions.

Introduction#

The primary purpose of simulation is to mimic the behavior of a real-world system under various conditions without directly interfering with the actual system, but rather through a model that defines the relationships between endogenous (internal) variables and exogenous (external) variables. Note that the need for simulation often arises from the intrinsic uncertainty in exogenous variables, as well as from the complexity and cost of experimenting with real systems. By systematically altering these variables within the model, we can explore a wide range of possible scenarios, assess system robustness, identify potential failures or inefficiencies, and evaluate alternative strategies before committing resources in the real world.

This enables us to:

Understand how the system typically operates under real-world conditions
Experiment by making changes to system inputs, configurations, or rules and observing their effects on outputs
Identify bottlenecks in processes, information flow, or resource utilization that may cause delays or inefficiencies

However, simulation modeling has certain limitations:

It requires substantial domain knowledge, specialized training, and practical experience to build and interpret models effectively
It is inherently random, making it difficult to intuitively predict outcomes without running the simulation
It can be logistically and computationally expensive, especially for high-fidelity or large-scale models

Applications of Simulation Modeling in Transportation Engineering include

Traffic Flow Simulation to understand vehicle movement including car-following and lane-changing
Signal Timing Optimization Simulation for adaptive traffic control systems
Transit Operations Simulation to optimize scheduling and departures
Urban Freight Delivery Simulation to evaluate last-mile delivery strategies
Behavior Simulation to replicate how travelers decide between available transportation modes

Classification#

Discrete Event Simulation#

Discrete Event Simulation models systems in which changes in state occur at discrete points in time, triggered by specific events. In this approach, time advances by jumping from one event to the next through event scheduling, and state variables change instantaneously when those events occur. In transportation engineering, DES is widely used to capture queueing behavior, resource allocation, and operational logic. Examples include simulating vehicle queues at signalized intersections, modeling passenger boarding and alighting at bus stops or train platforms, and representing aircraft takeoff and landing sequences at airports.

Continuous Simulation#

Continuous simulation models systems in which state variables change smoothly over time, typically described by differential or difference equations. In this approach, time progresses continuously, and changes in system state occur gradually rather than in sudden jumps. In transportation engineering, continuous simulation is used to capture the evolution of physical variables and system performance over time. Examples include modeling vehicle acceleration and deceleration profiles along a roadway segment, simulating traffic flow dynamics on highways using fluid-flow models, and analyzing variations in train speed and braking patterns along a rail corridor.

Agent-based Simulation#

Agent-based simulation models systems as collections of autonomous entities, or agents, each with their own attributes, decision rules, and behaviors. These agents interact with one another and with their environment, and the overall system behavior emerges from these individual interactions rather than being centrally controlled. In transportation engineering, agent-based simulation is often used to represent travelers, vehicles, or infrastructure elements as agents, allowing for the study of complex, adaptive behaviors. Examples include simulating pedestrian movements through transit stations, modeling ride-hailing driver-passenger matching in urban areas, and representing traveler mode choice behavior in response to changes in travel time, cost, or service quality.

Simulation Methodology#

A structured simulation study typically follows these steps:

Problem Definition – This first step involves clarifying exactly why the simulation is being undertaken and what it is intended to achieve. The scope of the system is defined, along with the specific questions the simulation should answer. At the same time, clear and measurable performance criteria—such as throughput, cost, or delay—are established so that results can be evaluated against well-defined objectives.
System Conceptualization – Once the problem is clear, the next step is to create a coherent representation of the system that captures its boundaries, the main entities involved, and the variables that drive its behavior. This conceptual model also describes the processes and interactions within the system, while distinguishing between endogenous variables, which are determined internally, and exogenous variables, which are influenced by external factors.
Data Collection and Input Modeling – With the conceptual model in place, attention turns to gathering accurate, relevant data to support the simulation. This involves identifying reliable sources for parameters and inputs, and for stochastic systems, fitting probability distributions to capture inherent variability. The quality of this step is critical, as the accuracy of the simulation depends heavily on the validity of its inputs.
Model Translation – The conceptual model is then transformed into an operational simulation using an appropriate programming language or specialized simulation software. During this process, the focus is on ensuring that the logic, processes, and interactions defined in the conceptual stage are faithfully and efficiently implemented in a computational form.
Verification and Validation – Once implemented, the model must be rigorously checked. Verification ensures that the simulation has been built correctly and behaves according to its specifications, while validation confirms that its outputs reflect real-world behavior. If discrepancies are found, adjustments to the model’s structure or parameters are made to improve its accuracy and reliability.
Experimentation and Analysis – A verified and validated model can then be used to explore different scenarios. This may involve running simulations under baseline conditions, testing alternative configurations, or performing sensitivity analysis to determine how changes in key parameters affect system performance. The results of this step help to identify potential improvements and assess trade-offs.
Documentation and Reporting – Throughout the study, all assumptions, model logic, data sources, and results are documented clearly and comprehensively. The findings are then presented in a structured and accessible format, enabling stakeholders to interpret and act on the results with confidence.
Implementation – Finally, the insights gained from the simulation are applied to the real system. This might involve making design modifications, refining operational strategies, or guiding policy decisions, with the ultimate goal of improving system performance based on evidence from the simulation.