home / skills / a5c-ai / babysitter / thermodynamic-model-selector

thermodynamic-model-selector skill

safe

/plugins/babysitter/skills/babysit/process/specializations/domains/science/chemical-engineering/skills/thermodynamic-model-selector

This skill helps select thermodynamic property methods based on components and conditions, enabling accurate model choice, parameter fitting, and uncertainty

npx playbooks add skill a5c-ai/babysitter --skill thermodynamic-model-selector

Review the files below or copy the command above to add this skill to your agents.

Files (1)

SKILL.md

1.8 KB

---
name: thermodynamic-model-selector
description: Automated thermodynamic property method selection based on component characteristics and operating conditions
allowed-tools:
  - Read
  - Write
  - Glob
  - Grep
  - Edit
  - Bash
metadata:
  specialization: chemical-engineering
  domain: science
  category: Process Simulation
  skill-id: CE-SK-003
---

# Thermodynamic Model Selector Skill

## Purpose

The Thermodynamic Model Selector Skill guides selection of appropriate thermodynamic property methods based on component characteristics, operating conditions, and accuracy requirements.

## Capabilities

- Component analysis (polarity, association, electrolytes)
- Operating condition assessment
- Property method recommendation
- Binary interaction parameter fitting
- VLE/LLE data regression
- Model validation against experimental data
- Uncertainty quantification

## Usage Guidelines

### When to Use
- Selecting property methods for simulation
- Fitting interaction parameters
- Validating thermodynamic models
- Assessing model uncertainty

### Prerequisites
- Component list defined
- Operating ranges specified
- Experimental data available
- Accuracy requirements known

### Best Practices
- Consider all phase equilibria
- Validate with experimental data
- Document model selection rationale
- Assess sensitivity to parameters

## Process Integration

This skill integrates with:
- Process Simulation Model Development
- Distillation Column Design
- Crystallization Process Design

## Configuration

```yaml
thermodynamic-model-selector:
  model-categories:
    - equation-of-state
    - activity-coefficient
    - specialized
  data-sources:
    - DECHEMA
    - NIST
    - DIPPR
```

## Output Artifacts

- Model selection reports
- Parameter fitting results
- Validation comparisons
- Uncertainty assessments

Overview

This skill automates selection of thermodynamic property methods based on component characteristics, operating conditions, and accuracy targets. It recommends equations of state, activity-coefficient models, or specialized correlations and produces model selection reports and validation summaries. The goal is faster, more consistent selection for simulation and design workflows.

How this skill works

The skill inspects component lists to detect polarity, associating species, electrolytes, and other key traits. It evaluates operating ranges (T, P, composition), cross-checks available experimental data, and matches those inputs to model categories (EOS, activity-coefficient, specialized). It can fit binary interaction parameters to VLE/LLE data, validate models against experiments, and quantify uncertainty to support decision-making.

When to use it

Choosing a property method for process simulation or flowsheeting
Fitting and updating binary interaction parameters from VLE/LLE data
Validating thermodynamic models against experimental datasets
Assessing model uncertainty and sensitivity before scale-up
Design tasks involving phase equilibria, distillation, or crystallization

Best practices

Provide a complete component list and realistic operating range before selection
Include experimental VLE/LLE data for parameter fitting and validation when available
Consider all relevant phase equilibria (vapor, liquid, solid, and associative behavior)
Document the selection rationale and validation metrics for traceability
Perform sensitivity analysis on fitted parameters and report uncertainty bounds

Example use cases

Select an EOS (PR, SRK, or cubic-plus-association) for high-pressure hydrocarbon streams
Choose an activity-coefficient model (NRTL, UNIQUAC) for polar solvent separations
Fit binary interaction parameters to experimental VLE for azeotrope prediction
Validate models and produce uncertainty estimates for distillation column design
Recommend specialized models for electrolytes or strongly associating species

FAQ

What inputs are required for a reliable recommendation?

At minimum: a complete component list, temperature/pressure ranges, and your accuracy requirements. Experimental VLE/LLE data and phase labels improve fitting and validation.

Can it fit interaction parameters automatically?

Yes. The skill supports regression of binary interaction parameters from supplied VLE/LLE datasets and returns fit quality metrics and uncertainty estimates.