ENTERPRISE AI ANALYSIS
Exploring Artificial Intelligence as a Tool for Logistics Process Simulation
This study evaluates generative large language models, Perplexity and ChatGPT, for discrete-event simulation in ExtendSim, focusing on a complex manufacturing system (9721 tons output). Three scenarios were assessed: autonomous model creation, output estimation, and copilot-guided building. LLMs couldn't autonomously build models due to API limitations. Output estimation required iterative prompt refinement (errors 0.1% for Perplexity, 1.2-22.8% for ChatGPT). The copilot approach was viable, reducing development time from days to 8-10 hours, but required human expertise to resolve errors (28-32% hallucinations). This paradigm offers acceleration for experienced users but needs API integration and RAG enhancements.
Executive Impact
Key metrics from the study highlighting the potential for AI in logistics simulation.
8-10 Hours
Development Time Reduction
0.1% MAPE
Model Output Accuracy
28-32% of errors
Hallucination Rate
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
The study evaluated LLM's ability to create, estimate, and assist in DES modeling. Key findings include limitations in autonomous model creation and the necessity of human oversight.
9721
Tons of output from benchmark model
| Feature | Perplexity | ChatGPT |
|---|---|---|
| Syntactic Errors | 24% (6 errors) | 18% (4 errors) |
| Logical Fallacies | 24% (6 errors) | 23% (5 errors) |
| Hallucinations | 28% (7 errors) | 32% (7 errors) |
| Parameter Misinterpretations | 24% (6 errors) | 27% (6 errors) |
| Total Errors | 25 | 22 |
The research used a complex manufacturing system in ExtendSim 10 Pro 2024. Prompt engineering involved iterative refinement across three experimental tasks.
Enterprise Process Flow
Independent Model Creation
→
Output Estimation (DES)
→
Copilot-Assisted Development
→
Iterative Prompt Refinement
→
Model Validation & Results
Benchmark Case Study: Construction Material Production
The benchmark model is a complex Discrete-Event Simulation (DES) model of construction material production, developed in ExtendSim. It features 13 operations, feedback waste loops, mass loss during drying, and variable work shifts. The simulation duration is 30 days, producing 9721 tons of products. This complexity highlights LLM limitations in advanced, real-world scenarios, contrasting with simpler models in prior research. Material flow unit is 1 ton per time unit.
Generative AI excels as a copilot but has limitations. API integration and RAG enhancements are crucial for future advancements.
LLMs cannot autonomously create ExtendSim models due to the lack of direct API access.
Output estimation only reached benchmark levels after iterative prompt refinement, highlighting the need for human intervention. Perplexity achieved 0.1% MAPE, ChatGPT 1.2–22.8%.
The copilot approach is the most promising, reducing development time from days to 8-10 hours for experienced users, despite initial errors and hallucinations.
Hallucinations (28–32% of errors) and logical fallacies necessitate rigorous human validation, making it less feasible for inexperienced users without additional support (e.g., RAG knowledge bases).
Generative AI performs best as a copilot rather than autonomously. No LLM is unequivocally superior. Practical deployment requires structured human-AI workflows, with validation essential to ensure model accuracy and reliability.— Study Conclusion
Advanced ROI Calculator
Estimate the potential return on investment for integrating AI into your logistics processes.
Estimated Annual Savings
$0
Hours Reclaimed Annually
0
Implementation Roadmap
Our phased implementation roadmap ensures a smooth transition and maximizes the value of AI in your operations.
Phase 1: Discovery & Strategy
Assess current logistics processes, identify AI opportunities, and define strategic goals. This involves detailed data analysis and stakeholder workshops.
Phase 2: Pilot & Proof-of-Concept
Develop and test AI models on a small scale, validating their effectiveness and refining algorithms. Focus on a high-impact, low-risk area.
Phase 3: Integration & Scaling
Integrate validated AI solutions into existing systems, expand deployment across relevant departments, and establish performance monitoring.
Phase 4: Optimization & Future-Proofing
Continuously monitor and optimize AI performance, iterate on models, and explore new AI capabilities to maintain competitive advantage.
Ready to Transform Your Logistics with AI?
Schedule a personalized consultation with our AI specialists to discuss how generative AI can optimize your supply chain and drive efficiency.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat