🍺 Beer Distribution Game Simulator
This interactive simulation showcases the bullwhip effect in supply chains by leveraging advanced Large Language Models (LLMs) as autonomous decision-makers. Each participant in the supply chain— Retailer, Wholesaler, Distributor, and Factory—is controlled by a sophisticated LLM (currently powered by GPT-4o-mini), which independently processes available data, interprets demand signals, and makes ordering decisions under real-world constraints.
How it works:
- Set your simulation parameters below
- Click "Run Simulation" to start
- Use "Stop Simulation" to interrupt long-running simulations. For fast comparison of scenarios, GPT-4o-mini is recommended.
- Watch as AI agents make decisions and the bullwhip effect emerges
- Toggle information sharing to see its impact on the bullwhip effect
Simulation Parameters
Decision Support Features
All players receive information about actual customer demand
Provide players with volatility analysis of customer demand
Provide players with pipeline inventory and inventory position analysis
Provide players with downstream facility inventory position to predict future orders
Demand Pattern
AI model for decision making
Ready to run simulation
Results
Facility Inventory Position
About the Decision Support Features
Information Sharing: When enabled, all players receive actual customer demand information. You can configure how many historical weeks to share (0 = current week only, >0 = current week + historical demand), reducing information distortion that causes the bullwhip effect.
Volatility Signals: Provides players with statistical analysis of demand volatility over a configurable time window, including standard deviation, coefficient of variation, and volatility level (Low/Medium/High/Limited). You can adjust the analysis window from 0 weeks (current week only) to 10 weeks of historical data.
Pipeline Inventory Signals: Provides players with visibility into their pipeline inventory (orders in transit) and inventory position (pipeline + on-hand - backlog), along with strategic guidance to maintain optimal inventory levels and minimize costs.
Downstream Inventory Visibility: Provides players with visibility into their downstream facility's inventory position to predict future order patterns. High downstream inventory suggests fewer future orders, while low downstream inventory indicates potential for increased orders.
About the Bullwhip Effect
The bullwhip effect occurs when small changes in customer demand cause increasingly larger fluctuations in orders upstream in the supply chain. This simulation demonstrates how this effect emerges naturally when each player makes rational decisions based on limited local information.
Key factors contributing to the bullwhip effect:
- Order batching and lead times
- Demand forecast updating
- Price fluctuations and forward buying
- Shortage gaming
Benefits of decision support features:
- Reduces demand forecast errors
- Improves coordination across supply chain
- Decreases order variability upstream
- Lowers total supply chain costs
- Enables proactive inventory management
Observe how:
- Customer demand is relatively stable
- Retailer orders show some variability
- Wholesaler orders are more volatile
- Distributor orders are even more volatile
- Factory production shows the highest variability
- Decision support features should reduce this amplification
Stop Functionality
New Feature: You can now stop long-running simulations using the "Stop Simulation" button that appears when a simulation is running.
- The simulation will complete the current week before stopping
- Partial results will still be displayed and analyzed
- This prevents wasted compute resources on unwanted runs