Those of you who are already familiar with simulation modeling certainly know that it can be used in supply chain (to design and plan operations) or in manufacturing (to improve production lines and find bottlenecks). Nothing unusual there. However, its application is much wider than that.
Our users have conducted studies in epidemiology investigating how a disease is spread. They’ve modeled social and economic issues, explored sustainability-related challenges, and developed marketing strategies using simulation.
The application of simulation modeling also includes mining operations and oil well optimization, rail logistics and port throughput increase. So, if you’re still wondering if simulation can be used in your industry, the answer is most probably “yes”.
In this blog post, we give an overview of three different projects demonstrating the versatility of simulation application. Based on the conference presentation, it also introduces the session and Q&A recording. For more technical details about the projects, watch the video at the end of this blog post.
1. Extensions to Schelling’s social segregation model
A research team from the University of Tennessee conducted a study based on one of the earliest agent-based social simulation models — Schelling's model of segregation.
The Schelling’s model contains two social groups (represented by agents), for example, boys and girls, smokers and non-smokers, who are all living in a city. A person is happy if a certain percent of people from the same social group lives in their neighborhood. Happy people stay where they are, unsatisfied ones move to a free location.
Schelling found that even when agents didn't mind living nearby agents from another social group, they would still segregate themselves from other agents over time.
After the research team recreated the Schelling’s unidimensional model, they decided to extend it by adding more dimensions as people are multidimensional by nature. The goal was to address one of the United Nations challenges and investigate if people can live happily as a diverse society.
For the multidimensional agent model, the team defined two types of variables:
- Similarity — how many attributes two agents should have in common to consider them similar.
- Utility — how many similar neighbors an agent wants to have in its neighborhood to feel happy.
The results showed that a society where people are viewed as multidimensional beings is more likely to become stable and non-segregated if they are happy to accept some differences among their neighbors and focus more on common interests.
The research finding can be used for infrastructure and transportation planning, studying social dynamics and other tasks.
2. Joint installation of solar panels and green roofs
The same research team used simulation modeling to evaluate the potential of green technologies. They studied how to increase viability of installing green roofs with solar panels and evaluated promotional strategies.
Green roofs are basically roofs covered with plants and trees. They encourage biodiversity and keep water from running off. Green roofs also purify the air and regulate the ambient temperature. Integrated with solar panels, together they improve the panels’ performance. Additionally, green roofs reduce the dust and air pollutants surrounding the building facilitating panels’ maintenance.
For installing solar panels on their roof, a person gets a tax reduction — Solar Investment Tax Credit (ITC). It’s a reduction in the income taxes that a person would otherwise pay the federal government. In the end, ITC helps increase the efficiency of green technology implementation.
Using stochastic optimization, agent-based modeling, sensitivity analysis and other tools, the researchers found that running promotional campaigns and solar tax credit for 20 years gives the best projection for green roof and solar panel popularization.
3. Stand-alone microgrid location problem
Natural disasters (hurricanes, floods, etc.) can cause widespread electric power outages within grids. They also interrupt power supply to critical public service buildings, for example, grocery stores, hospitals, and civil services.
When there is a power outage and the main grid doesn’t work, communities can get power from microgrids. While the main power grid connects homes and buildings to central power sources, microgrids can operate independently (known as the island mode) using distributed generators, batteries, or solar panels.
This study used optimization and simulation to investigate a microgrid design problem: where to put distributed solar panel-based generators considering their operation, maintenance, efficiency, costs, and weather uncertainties.
These projects are detailed in the presentation given by Xueping Li from the University of Tennessee at the AnyLogic Conference 2021. They’re followed by a quick overview of a machine learning research study and Q&A session. Here are the session slides and recording:
Research studies mentioned in the video:
- On the extension of Schelling's segregation model (read)
- An agent-based model for joint placement of PV panels and green roofs (read)
- Multi-stage stochastic optimization of islanded utility-microgrids design after natural disasters (read)
Read more about AnyLogic in academia in our white paper Why Use Simulation Modeling in Education and Research.
