How do you write a plant trial plan for a new chemical product?

Who reads a chemical plant trial plan, and what do they decide?

A chemical plant trial plan is usually read by five groups at the customer and the supplier, and each one decides something before the trial starts or after it ends. Production decides whether to give up line time and operators for the trial. Quality decides whether the measurements and sampling will show the result and whether product made during the trial can be released. Engineering or maintenance decides whether the equipment can run the new chemical and at which readings to stop. The safety and environmental reviewer decides whether the safety data sheet, permits and change procedures allow the trial. The supplier's technical service owns the product's conditions of use.

The Australian competency unit for managing trials lists these steps as performance criteria: draft trial plans, confirm them with all relevant personnel, and negotiate conflicts as required (training.gov.au PMC557001, 2016). The same unit asks the trial manager to define the purpose of the trial and confirm it with all stakeholders, as the unit's first element, ahead of its data steps (training.gov.au PMC557001, 2016).

Write the chemical plant trial plan for the reader who was not in the planning meeting. Each of the five groups should find, without asking, what the trial will measure, what they must provide and what they will be asked to approve afterwards. A plan that only the supplier understands tends to produce results that only the supplier accepts.

Which sections does a chemical plant trial plan need?

A chemical plant trial plan needs seven sections, and each section fixes one thing before the trial starts so that nobody can change it after the results arrive. The table below decides whether a chemical plant trial plan will produce a clear approve or reject decision: each row names what the section fixes and the evidence it needs.

The key row of the chemical plant trial plan is section 6, pass and stop criteria. If it is empty or written after the results, the other six sections still produce data, but no decision. Write sections 1 and 2 first, because they set the scope that sections 3 to 6 must measure, and write section 7 last, once the people who must sign know what they are signing.

Keep the seven sections of a chemical plant trial plan separate from the trial log. The plan says what will happen and is fixed once signed; the log records what did happen, shift by shift, including every deviation from the plan and its reason. Sample points, containers and the people who receive samples belong in the plan, so the log only has to confirm them. When a section cannot be filled, such as a test method with unknown precision, write the gap into the plan and agree how it will be handled before the trial, not after.

Which open items from the product recommendation should a chemical plant trial close?

A chemical plant trial should close only the open items that the product recommendation could not settle from documents and lab data, and that could reverse the decision to approve the product. Items the safety data sheet, technical data sheet, test reports or customer records already answer stay closed, and the trial plan should say so.

The Australian competency unit for managing trials describes the same work in order: determine the information required from the trial, identify relevant data which is available or could be made available, and identify gaps in the data (training.gov.au PMC557001, 2016). The gaps are the open items. Copy them from the accepted recommendation into section 2 of the plan, one line each, and write four things beside every open item: the question, the measurement that answers it, the test method, and the result that would close it in either direction.

Two kinds of item tend to creep into chemical plant trials and should be kept out. The first is a preference that would not change the decision, such as a faster drying time when drying time is not limiting; record it as an observation, not a criterion. The second is a property already proven in a documented test under comparable conditions. Re-testing it on the line uses trial time without reducing uncertainty. A trial that closes two or three open items with clear results gives the customer a decision. A trial asked to answer every possible question tends to answer none of them firmly.

How do you plan samples, controls and handover for a chemical plant trial?

You plan samples, controls and handover for a chemical plant trial by writing down, before the trial, every sample point, the baseline on the current product, and each transfer of product, samples and results between people. The trial log then records what happened against that plan instead of inventing it on the shift.

Controls come first. Take a baseline on the current product on the same line, with the same test method and laboratory, over a period long enough to show day-to-day variation. The NIST/SEMATECH e-Handbook shows two measurement processes with the same short-term variability over 6 days where one has large between-day variability (NIST/SEMATECH e-Handbook, section 2.4.4). In flotation reagent trials, Cytec described a statistical approach to differentiate between daily operational variation and actual process improvement (Morgan-Jones and Eichorn, AusIMM, 2000).

For samples, fix the point, frequency, container, label format, retained portion and the person who receives each one. ASTM D4840-99(2018)e1 describes chain-of-custody procedures that document sample possession during collection, shipment, storage and analysis; it was written for water analysis, but the same record suits trial samples (ASTM, 2018). For handover, name who receives the trial product, its safety data sheet and certificate of analysis, where it is stored, and who sends lab results to whom and by when.

Which approvals does a chemical plant trial plan need before the trial starts?

A chemical plant trial plan needs the site's own safety, change and environmental approvals before the first drum reaches the line, and the plan should list each one with its owner and status. Which approvals apply depends on the site and the jurisdiction, so the plan states them rather than assuming them.

On a US process covered by OSHA process safety management, introducing a new chemical into that process is a change: the employer shall have written procedures to manage changes to process chemicals, except replacements in kind, and those procedures must address the technical basis, the impact on safety and health, and the authorization requirements before the change (US OSHA, 29 CFR 1910.119(l)). Employees whose job tasks the change affects must be informed and trained before start-up (US OSHA, 29 CFR 1910.119(l)(3)). Sites in other jurisdictions apply their own change and permit rules.

Environmental permits can apply to a trial as well. In 2017, a chemical maker asked the West Virginia Department of Environmental Protection for a permit determination before a 35-batch trial of a new product, attaching a facility map, a process flow diagram, a process description, safety data sheets and potential-to-emit estimates (Addivant USA, WV DEP, 2017). The trial plan should also confirm that the new chemical's safety data sheet is on site and that handling, storage and disposal of trial material follow it and the site's waste rules.

What does an illustrative chemical plant trial plan look like?

An illustrative chemical plant trial plan for a water-based rust preventive, accepted for trial on a machining line, closes two open items from the recommendation and fits on two pages. Every product, value, period and role in this example is illustrative and comes from no real case.

The table below shows the content of each section in the illustrative trial plan.

The key row of this illustrative plan is section 6. Because the pass rule compares the new product with a baseline taken on the same line in the week before, a result can be read without arguing about the season or the parts mix. Drying time, a preference in the recommendation, is recorded but not judged, because in this illustrative case it would not change the decision. Each measurement in section 3 maps to one open item in section 2, so the plan asks nothing that the recommendation did not leave open.

Why do chemical plant trials end without a decision?

Chemical plant trials often end without a decision because the plan left something open that the results then had to settle, and the people involved disagree on how. Five gaps are worth checking for before the trial starts.

• Criteria chosen after the data. Clinical trial guidance says the primary variable "should be specified in the protocol" and that its redefinition after unblinding "will almost always be unacceptable" (ICH E9, 1998). A plant trial has no blinding, but a metric picked after the results can always be picked to fit them.

• No baseline. Without data on the current product under the same conditions, any change can be blamed on the season, the raw material lot or the operator.

• Samples without a record. If nobody can show which bath or shift a sample came from, its result cannot carry weight.

• Unplanned changes during the trial. ICH E9 states that "Deviations from the planned procedure always bear the potential of invalidating the trial results" (ICH E9, 1998). Record each deviation, with the reason, in the trial log.

• No decision owner. A trial that ends with data and no named person to approve or reject drifts into a second trial.

The Australian competency unit for managing trials ends the process with interpreting data in terms of the trial purpose, determining trial outcomes and drafting a trial report with recommendations (training.gov.au PMC557001, 2016). Each of those steps needs the plan's sections in place before the trial.

What should a chemical plant trial plan checklist cover?

A chemical plant trial plan checklist should cover each of the seven sections plus the signatures, and every line should be true before the first trial sample is taken.

• The objective names the one decision the trial informs and the accepted recommendation it starts from.

• Every open item comes from the recommendation's open checks, with its question, measurement, method and closing result.

• Preferences that would not change the decision are listed as observations, not criteria.

• Each measurement names the metric, test method, sampling point, frequency and number of measurements.

• The baseline on the current product uses the same line, method and laboratory over a comparable period.

• Sample labels, retained samples and the custody record are defined, with the person who receives each sample.

• Handover names who delivers and receives the trial product, its safety data sheet and certificate of analysis, and who sends results to whom by when.

• Site safety, change and environmental approvals are listed with owner and status.

• Pass criteria and stop readings have limits, units and the person who stops the line.

• Production, quality, engineering and the supplier's technical owner have signed and dated the plan.

• The person who decides after the trial is named.

Lubinpla's trial preparation starts from the recommendation the customer accepted and its remaining open checks, and issues a trial plan with pass and stop criteria, samples and controls for review.