distillation-shortcut-design
Skill7 starsChanged 6 months agoApache-2.0View on GitHub
Estimate distillation column theoretical stages, minimum reflux, feed-stage location, and McCabe-Thiele stair-stepping for binary or pseudo-binary cuts using Fenske, Underwood, Gilliland (Molokanov form), and McCabe-Thiele methods. Use when sizing a new column, screening separation feasibility, or sanity-checking a rigorous simulation. Don't use for rigorous tray-by- tray design (use a rate-based or equilibrium-stage simulator), azeotropic / extractive distillation, or batch distillation.
In this repository31 other files it gives its agents.
- absorption-stripping-designSkill
- bayesian-hierarchical-process-modelsSkill
- caustic-merox-extractionSkill
- censored-regressionSkill
- compositional-data-analysisSkill
- control-valve-sizing-isa75Skill
- convective-heat-transfer-correlationsSkill
- design-of-experimentsSkill
- dimensionless-numbersSkill
- distributed-lag-modelsSkill
- engg-skills-commonSkill
- engineering-skill-creatorSkill
- engineering-statisticsSkill
- equation-of-state-selectionSkill
- heat-exchanger-sizingSkill
- literature-search-engineeringSkill
- material-energy-balancesSkill
- pipe-flow-pressure-dropSkill
- process-causal-inference-dagsSkill
- process-units-conversionSkill
- reactor-sizing-and-kineticsSkill
- relief-valve-sizing-api520Skill
- separator-vessel-sizingSkill
- sour-gas-amine-treatingSkill
- statistical-process-controlSkill
- steam-tables-iapwsSkill
- thermo-process-propertiesSkill
- time-series-process-data-analysisSkill
- two-phase-flowSkill
- uvSkill
- vle-flash-calculationsSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 1 is distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 78
Use Gilliland to estimate N and the feed-stage ratio.
- Very specific: line 107
If the system is too non-ideal for constant-alpha (e.g.
- Very specific: line 111
If Underwood's theta is not bracketed, the alphas are likely too closely spaced (or the feed is too dilute in the keys).
- Very specific: line 117
Run the listed script with representative inputs and an --output file when a deterministic calculation is available.
- Distinctive: line 119
Check result magnitudes against the stated assumptions, references, and a hand calculation or known operating range before reporting them.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Distillation Shortcut Design
Overview
Implements the classical shortcut suite:
- Fenske: minimum number of equilibrium stages from
xD,xB, and an averaged relative volatilityalpha. - Underwood: minimum reflux ratio for a multicomponent feed at a given
thermal condition
q. - Gilliland (Molokanov): actual stages from
N/NminvsR/Rmin. - McCabe-Thiele: stage stepping on a constant-
alphabinary equilibrium curve, with operating-line intersection from the feed q-line.
The methods are textbook (Seader/Henley, Sinnott, McCabe/Smith/Harriott) and
implemented from first principles in engg_skills_common.separations.
Prerequisites
uvavailable.- On first use the script writes
LICENSE_NOTIFICATION.txtlisting the library terms.
When to Use
- Sizing a new column (number of stages, reflux ratio, feed stage).
- Screening separation feasibility before committing to a rigorous simulation.
- Sanity-checking a rigorous result with an independent calculation path.
Don't use for
- Rigorous tray-by-tray rating or design — use an equilibrium-stage or rate-based simulator.
- Azeotropic or extractive distillation — the
alphaassumption breaks down. - Reactive distillation, batch distillation, or columns with side draws and multiple feeds.
- Sub-ambient cryogenic columns where K-values change strongly with stage T.
Utility Scripts
uv run scripts/shortcut.py fenske --alpha 2.5 --xD 0.95 --xB 0.05 --output /tmp/fenske.jsonuv run scripts/shortcut.py underwood --alphas 4.0,2.0,1.0 --zs 0.4,0.4,0.2 --xds 0.95,0.04,0.01 --q 1.0 --output /tmp/underwood.jsonuv run scripts/shortcut.py gilliland --Nmin 12 --Rmin 1.3 --R 1.8 --output /tmp/gilliland.jsonuv run scripts/shortcut.py mccabe-thiele --alpha 2.5 --xD 0.95 --xB 0.05 --xF 0.40 --q 1.0 --R 2.0 --output /tmp/mt.json
Procedure
- Decide on light key (LK) and heavy key (HK). Compute or look up the
average relative volatility
alpha = K_LK / K_HKover the column. - Pick distillate (
xD) and bottoms (xB) compositions for the LK from product specs or recovery targets. - Compute
Nminwith Fenske. - For multicomponent feed, compute
Rminwith Underwood at the feed thermal conditionq(1.0 saturated liquid, 0.0 saturated vapor). - Pick operating reflux
R = 1.2-1.5 × Rmin(industry rule of thumb; higher gives fewer stages but more reboiler/condenser duty). - Use Gilliland to estimate
Nand the feed-stage ratio. - For binary screening with constant
alpha, runmccabe-thieleto sanity-check the integer stage count.
Pitfalls
- Using a feed-zone
alphainstead of an averagedalphaacross the column. Fenske and Underwood need a representative value; averaging top and bottom K-values is the usual workaround. - Underwood's theta is bracketed between successive component alphas;
picking the wrong root gives a useless
Rmin. The script bisects only the first valid interval — if your system has multiple distributed components, inspect the theta value carefully. - Treating
R/Rminof 1.05 or less as feasible; the Gilliland correlation is increasingly inaccurate in that range and the resulting column would have many stages and small driving forces. - Forgetting that Fenske returns total stages including the reboiler.
- Using McCabe-Thiele for a non-binary cut.
- Picking
R = 1.5 × Rminand reporting it as optimal without an energy-vs-capex trade-off check. - Assuming
q = 1for a partially vaporized feed; even ~10% vapor changesRminnoticeably. - Reporting "10 actual trays" when Gilliland returns "10 equilibrium stages" — actual tray count needs Murphree (or O'Connell-style) efficiency.
- Using the shortcut output for a tower-control or relief calculation. Shortcut methods do not give per-stage T/P profiles.
Fallback Strategies
- If the system is too non-ideal for constant-
alpha(e.g. methanol-water, ethanol-water), Fenske/Gilliland gives misleading numbers; switch to a pseudo-binary basis (LK vs HK at top and bottom alphas averaged) and flag the deviation in your report. - If Underwood's theta is not bracketed, the alphas are likely too closely
spaced (or the feed is too dilute in the keys). Pivot to a McCabe-Thiele
treatment of LK/HK at constant average
alphaand document the simplification.
Verification
- Run the listed script with representative inputs and an
--outputfile when a deterministic calculation is available. - Confirm the JSON result contains
ok: true, expected units, and no unhandled warnings. - Check result magnitudes against the stated assumptions, references, and a hand calculation or known operating range before reporting them.
References
references/methods_summary.md— equations and assumptions for each shortcut method.- Seader, Henley, Roper, Separation Process Principles (any recent edition).
- McCabe, Smith, Harriott, Unit Operations of Chemical Engineering.
Anti-Patterns
- Reporting Gilliland stages as the actual tray count without efficiency derating.
- Using shortcut methods to specify column internals (tray spacing, downcomer, weir height); those require flooding/loading correlations not in this skill.
- Hiding the
R/Rminratio in the final report — the optimization context matters for the reader.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

