AnyLogic Conference 2026

Introduction to the Presentation

Overview of the Presentation

• Jaen Voslo introduces himself and Devin Cowling, mentioning confidentiality agreements that limit their presentation content.

• The focus will be on how simulation helped analyze a frozen food packaging system from production to dispatch.

• The presentation structure includes company background, process flow, simulation usage, AI in model development, analysis methods, results discussion, and Q&A.

Project Background

Client and Objectives

• The client is represented by a fictitious company called Golden Veged, a large frozen food manufacturer with an end-to-end process.

• Key project questions include identifying true system constraints and assessing changes for improved performance.

Complexity of Operations

• The operation involves multiple production areas with shared downstream networks leading to complex interactions among equipment.

• A blockage in one area can affect overall production flow; thus, the model must represent complete material flow rather than isolated machine evaluations.

Importance of Simulation

Reasons for Using Simulation

• Simulation helps determine achievable throughput and identify bottlenecks that static calculations cannot capture.

• It captures interactions across interconnected systems like packaging lines and conveyors which are critical for accurate performance assessment.

Benefits of Simulation

• Evaluates impacts of downtime and variability on material flow effectively through dynamic modeling.

• Visually demonstrates proposed changes using animated scenarios to enhance stakeholder understanding.

Model Development Tools

Features of AnyLogic Software

• AnyLogic's material handling library allows quick modifications to conveyor networks using standard objects.

• Java code integration provides flexibility for detailed logic implementation and automated testing capabilities.

Demonstration of Model Interaction

Single Run Demonstration

• Users interact with the model through a single run demonstration where they can view animations while data drives the scenario setup via Excel files.

Data Management Capabilities

• Detailed statistics during model runs help verify accuracy; users can visualize operations in both 2D and 3D formats.

Use of Artificial Intelligence in Model Development

AI Integration Process

• Issues are created in GitHub for tasks like generating charts; AI assists by planning work based on provided context within minutes.

Sensitivity Analysis Methodology

Conducting Sensitivity Analysis

• Users load previous scenarios to run predefined sensitivity analyses efficiently using parallel processing capabilities.

Automated Unit Testing Implementation

Ensuring Model Integrity

• Automated unit tests validate functionality by reading scenarios from Excel files; various assertions ensure correctness during execution.

Verification and Validation Techniques

Comparing Outputs with Historical Data

• Daily outputs from simulations are compared against historical operating data to assess accuracy.

Statistical Stability Assessment

• Multiple runs confirm statistically stable results necessary for reliable decision-making regarding capacity constraints.

Summary of Results

High-Level Adjusted Results

• Base case results show outputs from plants A, B, C along with packaging line availability metrics.

Scenario Testing Outcomes

• Distribution adjustments across bagging loads showed minor improvements but highlighted interdependencies affecting overall efficiency.

Discussion on Model Verification and Validation

Challenges with Historical Data

• The speaker discusses the difficulties of verifying models in Greenfield projects where historical data is absent, making it challenging to assess model accuracy.

• In such cases, reliance shifts to expert opinions, conducting experiments to predict outcomes based on system changes, like increasing a variable by 10%.

• Historical data is crucial for validating models; without it, proving that a model accurately represents reality becomes problematic.

Surprising Insights from the Model

• The project team anticipated bottlenecks in palletizing and sorting but discovered that packaging issues arose first when increasing plant capacity.

• This insight indicated that investments in improving palletizing would be ineffective if packaging problems were not addressed first.

Development Process and Tools Used

• The team does not utilize CI/CD or GitHub for verification due to licensing restrictions with AnyLogic software, relying instead on manual processes.

• Developers must remember to run custom experiments before committing changes or branching code, which complicates the workflow.

Timeframe for Model Development

• The total development time was approximately six to seven weeks, with significant effort spent analyzing limited historical data.

• A detailed analysis pack was created alongside the model scenarios, including Excel sheets and workbooks containing comprehensive information.

Importance of Testing During Development

• Emphasizes the necessity of unit testing during development to ensure any modifications do not compromise model integrity.

• Running unit tests after changes allows developers to identify potential issues early in the process.

• There are inherent risks associated with using AI tools for development; however, these risks are similar when working with human team members.

Conveyor System Challenges

• Discusses complexities within conveyor systems where different SKUs merge; this can lead to blockages requiring manual intervention.