The Beer Distribution Game
Computational notebooks, System Dynamics Models, Agent-based Models and a Reinforcement Learning Algorithm
Keywords
agent-based modeling, system dynamics,abm, bptk, bptk-py, python, business simulation, reinforcement learning, q-learning
The Beer Distribution Game
Computational notebooks, System Dynamics Models, Agent-based Models and a Reinforcement Learning Algorithm
The Beer Game was developed by Jay Forrester’s group at MIT in the early years of System Dynamics, to illustrate how difficult it is to manage dynamic systems – in this case a supply chain that delivers beer from a brewery to the end consumer.
Have a go at playing the Beer Game (on your own or – much more fun – in a group) before you read these pages: you can play it online at beergame.transentis.com. The companion blog posts are on the transentis blog.
Contents
- Understanding the Beer Game. The best place to get started: the rules of the game, the dynamics governing it and how to improve on a typical game. It uses the SD DSL version of the model.
- Simulating the Beer Game. The stock and flow model behind the game and its implementation in the SD DSL.
- The Beer Game in Stella. The same model built in Stella Architect, loaded from its XMILE file and run on both engines.
- An Agent-based Approach To Modeling the Beer Game. An agent-based simulation of the Beer Game that can be used to test policies.
- Training AI to play the Beer Game – A Reinforcement Learning Approach. Reinforcement learning, applied to training agents to play the Beer Game - from the training to the stored q-tables of 50,000 episodes.