IDAES carbon-capture parameter estimation
A chemical-engineering parameter-estimation study for a reduced MEA carbon-capture absorber, with held-out validation, bootstrap uncertainty, and operating sensitivity analysis.

Project definition
Problem statement
Reduced absorber models contain empirical corrections for transfer, equilibrium loading, and thermal response. These parameters can compensate for one another unless campaigns provide enough independent information.
The engineering problem is to estimate the corrections from controlled campaign responses, test prediction on held-out conditions, quantify uncertainty, and state where the reduced model remains valid.
Project objectives
- Model counter-current carbon dioxide absorption across twenty-four axial segments.
- Estimate mass-transfer, loading-offset, and heat-response corrections from twelve campaigns.
- Predict four held-out campaigns without using them in the fitting objective.
- Quantify uncertainty with 240 campaign bootstrap refits and local Jacobian conditioning.
- Evaluate eighty operating cases across lean loading, liquid-to-gas ratio, and inlet carbon dioxide fraction.
Project structure
Project components
Absorber model
Marches gas composition, solvent loading, and temperature through the segmented column.
IDAES parameter layer
Declares bounded model parameters through an IDAES and Pyomo compatible container.
Estimator
Fits three corrections with weighted nonlinear least squares and reports Jacobian conditioning.
Uncertainty study
Resamples complete campaigns and refits the model 240 times.
Evidence pipeline
Writes retained CSV, JSON, PNG, and SVG results for validation and reporting.
Methodology
Project workflow
- 01Prepare campaigns
The study loads sixteen deterministic synthetic campaigns with declared response uncertainty.
- 02Fit parameters
Only the twelve K campaigns contribute to the weighted objective.
- 03Validate
The four V campaigns are predicted after fitting and retain separate error metrics.
- 04Estimate uncertainty
Campaign bootstrap intervals and parameter correlations expose weak information and bound contact.
- 05Map operation
The eighty-case matrix compares capture, rich loading, outlet temperature, and a duty proxy.
Demonstration scenario
The released study fits twelve absorber campaigns and predicts four held-out campaigns. It then performs 240 campaign bootstrap refits and maps eighty combinations of solvent loading, liquid-to-gas ratio, and inlet carbon dioxide fraction.
Engineering
Tools and method
- Tools
- The project uses Python, IDAES, Pyomo, NumPy, SciPy, Pandas, Matplotlib, Jupyter for subject analysis, simulation, and results.
- Process model
- Python and NumPy for segment balances, equilibrium opposition, transfer, solvent loading, and heat response.
- Estimation
- SciPy nonlinear least squares with explicit response weights, bounds, residuals, and Jacobian analysis.
- IDAES integration
- IDAES and Pyomo provide the declared engineering parameter container and compatibility layer.
- Engineering evidence
- Pandas and Matplotlib generate campaign tables, uncertainty evidence, sensitivity matrices, and eight figures.
- Verification
- Automated tests cover model behavior, estimation, IDAES integration, study outputs, and commands.
Testing
Evaluation
Evaluation measures
- Estimation and held-out RMSE for capture, rich loading, and outlet temperature
- Recovery of the three generating corrections
- Bootstrap percentile intervals and parameter correlation
- Jacobian singular values and condition number
- Directional response across the eighty-case operating matrix
- Automated tests, coverage, resolver checks, dependency audit, and artifact validation
Project boundaries
- The prepared observations are deterministic synthetic data.
- The model is steady state, reduced order, and intended for parameter-estimation education.
- The regeneration value is a transparent duty proxy, not a measured reboiler duty.
- Hydraulics, rigorous electrolyte thermodynamics, degradation, corrosion, emissions, equipment sizing, safety, and plant economics require separate specialist work.
Included
- 01Complete Python source code
- 02Reduced twenty-four-segment MEA absorber model
- 03Twelve estimation and four held-out validation campaigns
- 04240 bootstrap refits and eighty sensitivity cases
- 05Eight result figures and complete CSV evidence
- 0671-page project report in PDF and editable Word formats
- 0715-page setup and usage guide in PDF and editable Word formats
- 0836 annotated references and two sourced literature figures
- 0921 automated tests with 100 percent statement coverage
Project record
No information is collected on this page.
- Permanent project ID
- GP-CH-121TMRX
- Catalogued
- 21 Aug 2026
- Completed
- 24 Aug 2026
- Verified
- 24 Aug 2026
- Demonstration
- Included in repository
Handover
After purchase
- 01Payment is confirmed
The project is marked unavailable and cannot be purchased again.
- 02Repository access is granted
The buyer's submitted GitHub account receives access to the private repository.
- 03The purchase record is delivered
The certification sheet is prepared from the reviewed buyer details and sent privately by email.