Lubinpla

Salt spray test panels for a coating trial: how many you need

Summary

A coating trial's panel count depends on the specification minimum, the panel scatter you expect and the margin to the pass line. ASTM B117 and ISO 9227 set no panel count. The count comes from the governing document, such as ISO 12944-6:2018 or the customer specification, plus replicates for scatter and control panels in the same cabinet run.

  • ASTM B117-26a, last updated on 2026-09-22, does not prescribe the type of test specimen or the exposure period and asks for enough replicates to establish the variability of results (ASTM, 2026).

  • The superseded MIL-C-5541E (1990) specified five specimen panels for 168 hours of 5 percent salt spray per ASTM B117 as its corrosion resistance test (DLA, 1990).

  • To detect a one-standard-deviation shift in a mean with a one-sided test at α = 0.05 and β = 0.10, NIST works an example to about 11 measurements when the standard deviation is estimated (NIST/SEMATECH, accessed 2026).

  • ASTM D610-25 rates rusting on a zero to ten scale based on the percentage of visible surface rust (ASTM, 2025).

This guide covers laboratory salt spray trials that qualify a new coating, pretreatment or paint system against a current one on flat test panels. It does not set counts for field exposure racks, for cabinet corrosivity checks, or for qualification under a contract that already fixes the panel count, where that document governs.

When should you set the salt spray panel count for a coating trial?

Set the panel count in the written trial plan, before any panel is coated, together with the pass line (the result a panel or set must reach) and the rating method. The three depend on each other. A pass line of "no blistering" needs fewer panels to show a clear failure than a pass line on mean scribe creep, the distance corrosion spreads from a line scribed through the coating. The rating method decides what each panel contributes to the result. A count chosen after the first panels come out of the cabinet invites dispute.

ASTM B117 leaves the count to you. The practice "does not prescribe the type of test specimen or exposure periods to be used for a specific product, nor the interpretation to be given to the results." (ASTM B117-26a, clause 1.2). ISO 9227:2022 takes the same line for its neutral, acetic acid and copper-accelerated tests: it does not specify the dimensions or types of test specimens, the exposure period or the interpretation of results, which are left to product specifications (ISO 9227:2022, scope).

The trial plan therefore has to settle the count, the controls, the spares and the rating method as one decision. The general structure of such a plan, from purpose to stop criteria, is covered in How to write a plant trial plan for a new chemical product. Write the salt spray part into it as numbers, so the laboratory and the customer can check the panels they receive against it.

What decides the number of replicate panels in a salt spray coating trial?

Replicates are repeat panels of the same coating system, and three inputs decide how many a salt spray trial needs: the minimum the governing specification sets, the scatter you expect (how far results spread from panel to panel) for the rating you will use, and how close the coating system is likely to land to the pass line. ASTM states the principle: "In any testing program, sufficient replicates should be included to establish the variability of the results." (ASTM B117-26a, clause 3.3).

Once the specification minimum is met, scatter drives the count. NIST shows that the sample size needed to detect a shift in a mean from a fixed value depends on the standard deviation and on the risk levels you choose. The handbook works an example in which detecting a one-standard-deviation increase with α = 0.05 and β = 0.10 needs about 11 measurements when the standard deviation is estimated from data (NIST/SEMATECH, accessed 2026). The same reasoning carries straight over to measured ratings such as scribe creep in millimeters. With pass or fail ratings, extra panels mainly lower the chance that one panel decides the trial.

The table turns these inputs into a decision for each part of the panel set.

Criterion

Threshold (source)

If met

If not

Specification minimum per test

The count in ISO 12944-6:2018 or the customer specification; MIL-C-5541E used five panels per corrosion test (DLA, 1990)

Build at least that count per system and test

Get the count in writing before coating

Replicates from expected scatter

About 11 measurements to detect a one-standard-deviation shift from a fixed pass line at α = 0.05, β = 0.10 (NIST/SEMATECH, accessed 2026); more when comparing against control panels

Keep the count

Add panels or limit the claim to pass or fail

Margin to the pass line

Prior results on the system sit well clear of the pass line (illustrative judgment)

Specification minimum can be enough

Add replicates near the line

Control panels in the same cabinet run

Variability has been seen between fog chambers under nominally similar conditions (ASTM B117-26a, clause 3.3)

Read trial against control

Result cannot separate coating from cabinet

Spares for interim pulls

One panel per planned interim inspection (illustrative)

Final set stays complete

Final set shrinks

Rating method and pass line fixed

ASTM D1654-24e1, D714-25, D610-25 or ISO 4628 parts named in the plan

Ratings are comparable

Results are open to dispute

The control row is the one that most often changes a trial. Without panels of the current system in the same run, nobody can tell whether a weak result comes from the new coating or from the cabinet.

How many panels do ISO 12944-6 and customer specifications require per test?

ISO 12944-6 and customer specifications each require the panel count written in their own text. The trial should copy that number per system and per test instead of estimating it. Public summaries of ISO 12944-6:2018 do not state the count: ISO describes laboratory test methods and conditions for assessing paint systems for carbon steel (ISO 12944-6:2018, abstract), and a coatings testing laboratory describes exposing "replicate test panels" to each exposure separately (KTA-Tator, accessed 2026). Read the number from the purchased standard and quote its clause in the trial plan.

ISO 12944-6:2018 also limits what its results mean: "The test results are intended to be considered as an aid in the selection of suitable paint systems and not as exact information for determining durability." (ISO 12944-6:2018, abstract). As of 2026-10-23, ISO lists the 2018 edition at stage 90.92, to be revised (ISO, 2018), so check which edition the customer cites.

Customer and military specifications often set both a count and a pass line per panel. The superseded MIL-C-5541E specified that "Five specimen panels prepared in accordance with paragraph 4 3 3 shall be used for corrosion resistance testing" (MIL-C-5541E, 1990, clause 4.5.1). When a specification is ambiguous, get the customer's reading in writing before coating. On one quality forum, a user asked whether five dual-labeled panels or ten coupons met MIL-DTL-5541F and received advice to ask the customer and the NADCAP adviser rather than a count (Elsmar Cove, accessed 2026).

Do you need control panels in a coating qualification trial?

Yes. A coating qualification trial needs control panels: panels of the current system, coated on the same substrate and exposed in the same cabinet run as the trial panels. Salt spray results are relative. ASTM describes the practice as one that produces "relative corrosion resistance information for specimens of metals and coated metals exposed in a given test chamber" (ASTM B117-26a, clause 3.1), and warns that "Variability has been observed when similar specimens are tested in different fog chambers even though the testing conditions are nominally similar and within the ranges specified in this practice." (ASTM B117-26a, clause 3.3).

Judged only against a fixed pass line, a trial result answers two questions at once: is the new coating good, and was the run normal? Control panels pull them apart. If the controls perform as they usually do and the trial panels fail, the coating is the likely cause. If the controls fail too, the run itself is suspect, and the trial should not be judged on it.

Match the control panels to the trial panels in substrate, surface preparation, film thickness range, cure, scribe and position pattern in the cabinet. Give the controls the same count as each trial system, because a single control panel cannot show its own scatter.

Keep cabinet verification separate. ISO 9227:2022 describes a method for evaluating the corrosivity of the cabinet environment (ISO 9227:2022, scope). The reference specimens in that check test the cabinet, so they do not count toward the coating trial panel set.

How should coating test panels be prepared and labeled?

Prepare coating test panels to a written panel standard or to the production process, and label them so each panel traces back to its substrate lot, pretreatment, coating batch, film thickness and cabinet position. For cold-rolled steel, ASTM D609-17(2022) gives procedures designed "to provide steel panels with a uniform and reproducible surface for testing of paint, varnish, lacquer, conversion coatings, and related products" (ASTM D609-17(2022), clause 4.1).

When the trial must show production performance, run the panels through the line. MIL-C-5541E required that "The specimen panels shall be processed with the hardware during an actual production run, including all pre- and post-treatment processes such as cleaning and rinsing" (MIL-C-5541E, 1990). Panels prepared more carefully than production can pass a trial that production parts later fail.

Measure before exposure. A coatings testing laboratory measures the coating thickness and the adhesion of the coating system before exposure and scribes salt spray panels mechanically before they go in (KTA-Tator, accessed 2026). Record dry film thickness for every panel, so a thin panel shows up on its own instead of disappearing into the average.

Mark each panel on the back or on an edge zone excluded from rating, with a label that survives the exposure. Number trial and control panels in one series, so the label does not tell the rater which is which. Record each panel's cabinet position, and keep one unexposed panel per system as a reference for color and gloss.

How do you fix the panel rating method before salt spray results come in?

Fix the panel rating method in the trial plan before any salt spray result comes in. The plan names the standard for each rating, the inspection times, the pass line per panel and per set, and who rates.

Scribe creep, blistering and rusting each have their own method. ASTM D1654-24e1 "provides a means of evaluating and comparing basic corrosion performance of the substrate, pretreatment, or coating system, or combination thereof, after exposure to corrosive environments" (ASTM D1654-24e1, clause 4.1). ASTM D714-25 rates the size and density of blisters, and ASTM D610-25 rates rusting on a zero to ten scale (ASTM, 2025). ASTM D610-25 does not cover rust creep from a scribe (ASTM D610-25, clause 1.1), so scribe results need D1654 or ISO 4628-8:2012, which assesses delamination and corrosion around a scribe (ISO 4628-8:2012, scope). ISO 4628-3:2024 assesses the degree of rusting by comparison with pictorial standards (ISO 4628-3:2024, scope).

The pass line has to be agreed, not inferred. For blistering, ASTM states: "When the reference standards are used as a specification of performance, the permissible degree of blistering of the paint system shall be agreed upon by the purchaser and the seller." (ASTM D714-25, clause 1.1).

Decide too how the set result is formed: the mean of the panels, the worst panel, or a count of panels that pass. Write the rule down with the method. A worst-panel rule needs fewer panels to fail a system and more panels to pass one with confidence.

How do you present a coating panel trial result?

Present a coating panel trial result as the planned decision applied to the measured panels: the count and controls planned, the count tested, each panel's rating, the set result by the rule fixed in advance, and the pass or fail call. Show trial and control systems side by side with every panel listed, so the reader sees the scatter and not only an average.

List deviations from the trial plan separately. A panel dropped for low film thickness, a panel pulled early or a cabinet interruption changes what the remaining panels can show, and a reviewer needs to see that next to the result rather than discover it later. Keep the original ratings even when a re-rate is done, and state who rated and whether the rater knew which panels were trial and which were control.

State the limits that travel with the result. ISO 12944-6:2018 calls its results an aid to selection, not exact durability information (ISO 12944-6:2018, abstract), and ASTM notes that correlation and extrapolation of salt spray results to natural environments are not always predictable (ASTM B117-26a, clause 3.2.1). The report should therefore say what the coating trial supports, such as approval for a further field trial, and what it does not.

Lubinpla, an applied AI company for specialty chemical technical reasoning, prepares a test plan that fixes the samples, the controls and the analysis method before any results come in, and returns it as one cited document for the user to review and approve.

Worked example: salt spray panel count for a coating trial (illustrative)

The example below is illustrative: the specification minimum, scatter and margin are assumed values, not data from any standard. A plant wants to replace its current primer with one of two candidate primers on cold-rolled steel, judged by mean scribe creep after neutral salt spray. The customer specification asks for 3 panels per system per test. Earlier runs of the current primer show a scribe creep standard deviation of about 0.8 mm (illustrative), and the plant wants to detect a 1.0 mm shift in mean creep from the customer limit (illustrative).

Step 1, specification minimum: 3 panels per system. Step 2, scatter: the shift to detect is 1.0 / 0.8 = 1.25 standard deviations. The NIST one-sided formula with α = 0.05 and β = 0.10, after one iteration on the t distribution, gives about 8 panels per system (illustrative calculation). The scatter count governs, because it is larger than 3. Comparing two systems' means, each estimated from its own panels, needs roughly twice as many panels per system as this one-sample count, because the two-sample formula carries a factor of 2 (Chalmers University of Technology, accessed 2026). Step 3, margin: the current primer sits close to the customer pass line, so the plant keeps 8 rather than cutting back.

Step 4, controls and spares: two candidates plus the current primer as control make 3 systems of 8 panels, or 24. One spare per system for a mid-exposure inspection adds 3, and one unexposed reference panel per system adds 3, for 30 panels in all (illustrative). Step 5, rating: scribe creep by ASTM D1654-24e1, mean of the 8 panels per system, a pass line of mean creep no greater than the customer limit, with the control panels used to confirm the run was normal (illustrative), rated by one person without knowing which system is which.

Where these coating panel count rules do not apply

The panel count rules in this guide do not apply where a contract, specification or certification scheme already fixes the panel count, the exposure and the acceptance rule. There the document governs and the trial plan copies it. They also do not apply to the reference specimens used to check cabinet corrosivity under ISO 9227:2022, which test the cabinet, not the coating (ISO 9227:2022, scope).

Salt spray itself has limits that no panel count fixes. ISO 9227:2022 states that its salt spray methods are not intended for comparative testing as a means of ranking different materials relative to each other with respect to corrosion resistance, or for predicting long-term corrosion resistance (ISO 9227:2022, scope). ASTM adds that "Prediction of performance in natural environments has seldom been correlated with salt spray results when used as stand-alone data." (ASTM B117-26a, clause 3.2). A trial that compares a new coating with the current one in the same run is a screening step, and long-term claims need corroborating exposure.

ISO 12944-6:2018 covers paint systems on uncoated steel, hot-dip galvanized steel and thermal-sprayed metal coatings, and it does not apply to protective paint systems for electroplated or painted steel (ISO 12944-6:2018, abstract). Coatings in those groups take the count from their own product specification. Cyclic tests and immersion tests have their own procedures, and their panel counts should come from those documents rather than from the neutral salt spray plan.

Key takeaways

  • ASTM B117 and ISO 9227 set no panel count. The specification, the scatter and the margin to the pass line set it.

  • Treat the specification minimum as a floor, and raise it when scatter or a close pass line calls for more panels.

  • Expose control panels of the current system in the same cabinet run, with the same count as each trial system.

  • Fix the rating method, the set rule and the pass line in the trial plan before exposure.

Related posts