How to Use the Simulator
- For the Riopele pitch, start with Presentation Mode and click each ribbon step at your own pace.
- Choose an industrial example or adjust the sliders manually.
- Keep fiber composition at 100%. The app auto-balances sliders, but the total is always shown.
- Review the executive summary strip for predicted class, target status, main driver, and trade-off.
- Use Set as Baseline before changing inputs to compare one fabric recipe against another.
- Review Path to Class ≥3 for practical improvement recommendations.
- Click Export Scenario Summary to download a standalone HTML report.
- Optionally paste an OpenAI API key and use the AI Engineering Assistant to explain the model output, compare scenarios, create a lab-trial plan, or draft a Riopele engineering brief.
What the Outputs Mean
- Predicted Class: estimated ISO 12945-2 pilling class on a 1–5 scale.
- Target Status: whether the scenario is below, meets, or exceeds the Class ≥3 target.
- Design Typicality: proximity to the current prototype envelope. It is not statistical certainty.
- Influence by Process Stage: relative contribution by raw materials, yarn, construction, and processing.
- Parameter-Level Importance: the largest specific drivers within the selected scenario.
- Trade-off Analysis: pilling resistance compared with handle, cost, processing speed, and sustainability.
- AI Engineering Assistant: optional AI explanation layer. The deterministic PillNot model remains the prediction source of truth.
Presentation Mode
The Presentation Mode ribbon is designed for a controlled sponsor demo rather than an auto-running tour. Use it to pause for questions or jump to the area Riopele wants to discuss.
- Value Frame: state the business and engineering value before showing features.
- Risk Scenario: load a vulnerable fabric recipe.
- Save Baseline: make the weak scenario the comparison point.
- Improve Recipe: move to a stronger Class ≥3 path.
- Explain Drivers: show process-stage and parameter-level influence.
- AI / Lab Brief: generate a practical explanation or validation plan.
- Validation Plan: show what Riopele data would calibrate the model.
- Platform Vision: explain how pilling can become the first vertical of broader textile engineering intelligence.
Using Baseline vs Scenario
Set a baseline using the current fabric settings. Then change one or more inputs to create an alternative. The comparison table shows whether the scenario improved pilling class and what happened to handle, cost, time, and energy.
Recommended workflow: change one stage at a time. For example, first compare Low vs Medium twist, then compare no finish vs anti-pilling finish. This makes the engineering explanation easier.
Scope Warnings
Boundary warnings appear when an input is outside the intended challenge scope or near the edge of the current prototype model. These warnings do not block the simulator; they flag cases that need stronger ISO lab validation.
Using the AI API Layer
- Choose Live OpenAI API to use your API key during the demo.
- Paste your OpenAI API key into the API Key field on the Simulator page.
- Click Save API Key. The key is stored locally in this browser so the static GitHub Pages prototype can call the API.
- Use Explain Scenario, Suggest Class ≥3 Path, Compare Baseline, Generate Lab Trial Plan, or AI Engineering Brief.
- Use Demo fallback when you want a reliable no-network presentation response.
- Use Manual prompt to copy structured scenario data and send it to an approved AI workflow yourself.
- The AI receives the current structured scenario data and explains it. It does not change or override the prediction.
- Do not use Live API mode on a shared or public computer unless you clear the key afterward.
Maintaining the App
The app remains static and GitHub Pages compatible. Key files are separated for maintainability:
model-config.js— editable model coefficients, examples, target class, and scope thresholds.prediction.js— deterministic calculation engine.recommendations.js— warnings, target status, and Class ≥3 guidance.charts.js— local canvas charts for stage influence, feature importance, and trade-offs. No CDN is required.report-export.js— downloadable scenario report generator.ai-assistant.js— optional ChatGPT API advisory/reporting layer.app.js— UI controller and event handling.