ENStarting point
The sampling plan should connect the intended decision, the population represented and the way samples are collected and prepared. Material quantity, particle distribution, origin and variability influence this design. Repeating the analysis of the same portion does not correct an error in selecting the batch.
The plan follows the sample
Follow the sequence and open each stage to see its criteria.
- 01
Objective
Which result allows the batch to be accepted, rejected or investigated?
- 02
Population
Which containers, period and mass are represented?
- 03
Selection
How should sampling be distributed without favouring the most accessible portion?
- 04
Preparation
How can the sample be reduced and homogenised without changing the characteristic of interest?
- 05
Record
How can each result be linked to the sample and batch?
- 06
Retained sample
Which material will be retained, and under what conditions?
Write down the decision before sampling
A sample may be used to identify polymers, negotiate price, check a chemical restriction or release production. These purposes require different information. Before sampling, state what decision will be made, which characteristics will be measured and what happens if the result is inconclusive. An exploratory test may guide the next study, but should not be presented as full supply approval.
The EPA organises waste sampling around a technically defensible plan and sample chain-of-custody records. This methodological reference helps structure the reasoning, without replacing applicable Brazilian standards and requirements. The technical lead should select procedures suited to the material, test and risks of accessing the batch. [1]
Define the population the sample represents
“Supplier material” is too broad a population for a decision about a delivery. Define the date, process, origin, containers and quantity. If the inventory combines different campaigns, assess whether they should be treated as separate batches. The division should reflect relevant differences, not just loading convenience.
Also identify the unit of assessment. A batch of equipment may be classified by item; a flake stream may be assessed by mass and particle distribution. A percentage by number of pieces is not equivalent to a percentage by mass. This distinction affects interpretation of composition, expected yield and commercial settlement. State the unit at every stage, including in the test report and proposal.
Recognise heterogeneity before choosing the method
Particles of different sizes and densities may be distributed unevenly within a container. The accessible surface of a bulk bag may not represent its interior. Sampling should account for this structure and safe access conditions, without improvising entry into piles, equipment or containers. In some processes, planned sampling during material movement is more feasible than accessing static stock.
The Eurachem guide distinguishes the sampling contribution to uncertainty from the analytical contribution. Its approach reminds us that observed variation begins before the instrument. In practice, identify which differences belong to the material and which may have been introduced by sampling or preparation. This distinction helps determine whether the next effort should expand sampling or revise the measurement method. [2]
Distribute sampling and preserve its traceability
The plan must specify where and when to sample, how to select points or intervals and how to form the samples intended for each test. The required mass depends on particle size, heterogeneity, the parameter and the laboratory procedure. No single quantity is representative of all waste or plastics.
Record who collected the samples, container identification, conditions encountered and changes to the procedure. If a point could not be accessed, document the limitation. Sample identification must survive transport, preparation and analysis. Use containers and preservation conditions specified for the test, avoiding loss of fractions or contamination introduced by the packaging itself.
| Element | Question to resolve |
|---|---|
| Objective | Which result allows the batch to be accepted, rejected or investigated? |
| Population | Which containers, period and mass are represented? |
| Selection | How should sampling be distributed without favouring the most accessible portion? |
| Preparation | How can the sample be reduced and homogenised without changing the characteristic of interest? |
| Record | How can each result be linked to the sample and batch? |
| Retained sample | Which material will be retained, and under what conditions? |
Use composite samples only when they answer the question
Combining increments can help assess an average characteristic, provided the sample composition preserves the basis defined by the plan. However, this combination may conceal very different sub-batches. If the decision depends on finding a problematic fraction or checking units individually, an overall average may be insufficient.
Illustrative example: eight tonnes contain 2% of an impurity and one tonne contains 10%. The average composition by mass is (8 × 2% + 1 × 10%) ÷ 9, approximately 2.89%. If the eight tonnes are in eight containers and the remaining tonne is in two smaller containers, the simple average of the ten percentages is 3.6%. The calculations use different bases; the correct result depends on the intended population and weighting.
Distinguish repeat analysis from fresh sampling
Analysing the same portion several times helps investigate test repeatability. Collecting new samples from the batch helps investigate variation in sampling and material. Both types of information can be useful, but they are not interchangeable. A report with many decimal places or repeats may still be unrepresentative if the material was selected with bias.
When investigating a discrepancy, preserve the history: initial sample, retained sample, method, conditions and responsible parties. If supplier and buyer obtained different results, first compare what each actually measured. Sometimes the tests used different temperatures; in other cases, samples came from different containers. Repeating negotiations without resolving this methodological difference tends to reproduce the conflict.
Agree on a decision rule for acceptance
The specification should state limits and how results are interpreted. Where uncertainty matters to the decision, how it is considered must be agreed with the laboratory and the parties. A value close to the limit may require a more careful decision rule than simple rounding. The procedure should not be created only after the buyer obtains an unfavourable result.
Also record the process for an inconclusive result: holding, further investigation, additional assessment or rejection, as appropriate. Conditional release requires defined authority and cannot override a mandatory requirement. For chemical or safety parameters, the decision must comply with relevant technical and regulatory criteria, without using batch blending as a shortcut to resolve non-conformity.
Use historical data to improve the plan
After several deliveries, compare variability between batches, between containers and between repeated results. This information can guide a review of the plan and segregation at source. If the largest difference is between production campaigns, increasing analyses of a single sample will have little effect on supply predictability.
The sampling plan should keep pace with changes in origin, preparation, particle size and application. In recycled material projects, its role is to bring declared quality closer to received quality. A well-identified batch, with documented sampling and pre-agreed criteria, provides a more useful basis for deciding price, destination and industrial approval.
Sources and references
References accessed on 30 September 2026. The numbering in the text indicates the concepts supported by each source.
Circuvia corporate content to support project discussions. Methods, acceptance criteria and applicable requirements are defined within the scope of each operation.