The decision dashboard, on real client work
Stage 06 is where numbers become decisions. Below: an anonymized extract from a paid industrial CO₂ capture study on flue gas with 15% CO₂ and 15% water. Identities and absolute figures are withheld under client confidentiality.
Paid industrial study · identities and absolute figures withheld
The pipeline, as applied
The funnel from the technology page, doing paid work. Four material families were assessed; the primary library is shown below, with a 1,000+ structure COF track and benchmarks alongside.
Every survivor on the purity-recovery plane
All 140 purity-gate survivors, placed by the two KPIs the client contracts on. The dashed box is the 90/90 design target: 18 clear it in the fast screen, and full cycle simulations then optimize each finalist to it.
One mark per candidate, no identities shown. Hover a mark for its KPI values. Estimates from an ML surrogate trained on 10,000 full cycle simulations; the shortlist below comes from the full model.
The shortlist, against the industrial benchmark
Each finalist ran through the full cycle model, optimized to 90/90 while minimizing regeneration energy. Energies are indexed to the zeolite benchmark, which the model reproduces against published experiments.
| Candidate | Case | Purity | Recovery | Energy (indexed) | Status |
|---|---|---|---|---|---|
| M-01 | Dry | 90.0% | 90.0% | 59 | Advance |
| M-02 | Dry | 89.9% | 90.0% | 61 | Advance |
| M-03 | Dry | 90.0% | 90.0% | 62 | Advance |
| W-01 | Wet (30% RH) | 89.9% | 90.0% | 66 | Advance |
| W-02 | Wet (30% RH) | 87.4% | 91.3% | 100 | Monitor |
| W-03 | Wet (30% RH) | 89.9% | 89.9% | 71 | Monitor |
| C-01 | Dry (COF) | 89.4% | 89.9% | 84 | Not recommended |
| Zeolite 13X | Benchmark | 83.2% | 89.1% | 100 | Benchmark |
| Mesoporous carbon | Benchmark | 69.2% | 86.4% | 142 | Benchmark |
Purity and recovery from the full cycle simulation at each candidate's optimum; energy indexed to the zeolite 13X benchmark = 100 (lower is better). Wet-case candidates were additionally screened on CO₂-over-water selectivity from ternary simulations at 30% relative humidity. COFs were not recommended for this duty; the best one is retained for reference. At their own optima the benchmarks fall short of the purity target.
What a client works with
Interactive, not static
The delivered dashboard is filterable and drillable: per-candidate isotherms, cycle behaviour, and the full audit trail from crystal structure to verdict.
Named materials, real units
Clients see the actual material identities, absolute energy in their units, and cost per tonne in their currency. This public preview indexes and anonymizes all of it.
Sensitivity built in
Each ranking ships with sensitivity analysis: how the verdict moves when feed composition, utility prices, or material costs shift.
Want this view of your separation problem?
Bring a feed composition and a performance target, and we'll run the pipeline against it.