Chemical markets often appear quantitative long before they are genuinely comparable. A report may quote production capacity, purity, conversion yield, energy intensity, emissions, transport distance, or cost per tonne, then combine those figures into a confident conclusion. The arithmetic can be correct while the comparison is still misleading because the units, system boundaries, dates, or chemical basis do not match.
The following checklist is designed for analysts, editors, and business readers who use AI-assisted explanations to examine chemical manufacturing, battery materials, fertilizers, industrial gases, specialty chemicals, or laboratory-scale technologies. It is a quality-control workflow, not investment advice and not a substitute for engineering, regulatory, or safety review.
1. Define the decision before collecting numbers
Start with the exact question. Are you comparing manufacturing cost, technical feasibility, environmental burden, plant utilization, supply risk, or product quality? One calculation rarely answers all of them. Write the target metric, time horizon, geography, and decision boundary in one sentence.
A useful comparison also identifies what would change the conclusion. If a five-percentage-point shift in yield reverses the ranking, yield deserves more scrutiny than a minor rounding difference in electricity price.
2. Normalize the chemical basis
Chemical figures can be reported per tonne of product, per tonne of active ingredient, per mole, per unit of contained metal, or per unit of energy delivered. These are not interchangeable.
Before comparing two processes, convert them to the same chemical basis. Check molecular weights, hydration state, oxidation state, assay, and concentration. A tonne of solution containing 30% material is not a tonne of pure material. Likewise, two battery chemistries may store different amounts of usable energy per kilogram, so a mass-only comparison can distort the result.
3. Rebuild the mass balance
List the major inputs, products, coproducts, recycle streams, emissions, and waste. Then test whether the reported material flows are internally consistent.
For a simplified reaction, convert quantities to moles, apply the balanced equation, and convert back to the desired mass basis. Distinguish theoretical yield from isolated or plant yield. If a report uses a yield above the stoichiometric limit, mixes wet and dry mass, or ignores a large recycle stream, flag it before using downstream cost estimates.
4. Separate capacity, output, and utilization
Nameplate capacity is not annual output. Output depends on commissioning time, maintenance, feedstock availability, product qualification, demand, and operating rate.
Record capacity and observed output separately. If output is unavailable, present utilization as an explicit assumption rather than a fact. A plant operating at 55% utilization can have very different fixed cost per tonne from the same design at 90%, even when the underlying chemistry is unchanged.
5. Audit energy claims by form and boundary
Do not combine electricity, steam, fuel, and process heat without documenting conversion assumptions. Identify whether a value describes final energy purchased by the facility or primary energy before generation losses.
Check whether the system boundary includes feedstock preparation, compression, separation, solvent recovery, drying, refrigeration, pollution control, and waste treatment. These auxiliary steps can dominate energy use in processes whose main reaction looks simple on paper.
6. Date every price and operating assumption
Chemical feedstocks, electricity, natural gas, freight, carbon allowances, and currencies can move at different rates. Attach a date, region, currency, and tax basis to every price.
For historical comparisons, avoid inserting today's price into an older plant configuration without labeling the result as a scenario. For forecasts, use ranges and show which assumptions drive the result. A single precise cost can create false confidence when the key input prices are volatile.
7. Distinguish laboratory evidence from commercial evidence
A high yield in a small, carefully controlled experiment does not prove stable industrial production. Ask about batch size, run duration, impurity tolerance, catalyst life, solvent recovery, heat transfer, scale-up history, product qualification, and the number of independent replications.
Classify each claim as laboratory, pilot, demonstration, early commercial, or mature commercial. This makes uncertainty visible without dismissing promising research.
8. Treat AI output as a traceable draft
An AI explanation should expose the equation, units, assumptions, and intermediate values. Recalculate at least one representative case independently. Confirm chemical equations against authoritative references and verify market figures against dated primary documents.
A chemistry-focused reasoning tool can help organize a first-pass calculation. For example, https://chemistryai.chat/ can be used as an optional scratchpad for stoichiometry, equilibrium, thermodynamics, and unit checks. Any generated result should still be checked against source documents, physical constraints, and qualified expert review.
9. Test sensitivity instead of hiding uncertainty
Replace uncertain point estimates with a low, base, and high case. Typical variables include conversion, selectivity, yield, utilization, electricity price, feedstock price, catalyst life, recovery rate, and transport cost.
Change one variable at a time to identify the main drivers, then test plausible combinations. Report where the conclusion remains stable and where it changes. This is more informative than adding decimal places to an uncertain base case.
10. Verify links between technical and financial models
If a technical model says the process consumes 2.4 megawatt-hours per tonne, the cost model should multiply the same quantity by the matching electricity price and location. If the mass balance produces a saleable coproduct, its volume and price should flow consistently into revenue assumptions.
Create a short bridge table from technical outputs to financial inputs. This catches duplicated credits, omitted disposal costs, inconsistent utilization rates, and mismatched units.
11. Look for independent negative evidence
Promotional documents naturally emphasize successful runs, target performance, and future scale. Search for commissioning delays, quality problems, downtime, environmental permits, rejected batches, safety incidents, and customer qualification timelines.
Negative evidence should be evaluated rather than automatically treated as decisive. Its purpose is to test whether the optimistic narrative covers the operational risks that determine actual output.
12. Publish an audit trail
A reader should be able to reconstruct the calculation. Preserve the balanced equation, conversion factors, system boundary, source dates, price basis, utilization assumption, and sensitivity range. Separate reported facts from analyst assumptions and AI-generated suggestions.
A compact final review asks:
- Are all quantities on the same chemical and physical basis?
- Does the mass balance close within a reasonable tolerance?
- Are yield, selectivity, conversion, and recovery kept distinct?
- Are capacity and actual output separated?
- Do energy figures share the same boundary?
- Are prices dated, regionalized, and expressed in one currency basis?
- Is the evidence classified by scale and maturity?
- Can every important number be traced to a source or labeled assumption?
- Do sensitivity results show when the conclusion changes?
- Does the technical model connect consistently to the financial model?
The goal is not to eliminate uncertainty. It is to make uncertainty visible, prevent unit and boundary errors, and ensure that a confident explanation rests on a reproducible chain of evidence.
Comments
Log in or sign up to join the conversation.