TS 50625-3-1: How to correctly sample output fractions

Technical Specification TS 50625-3-1 remains the reference document for sampling and testing of output fractions resulting from WEEE treatment. However, in practice, most treatment operators in Romania apply the methodology incompletely or with systematic errors that invalidate the results. This guide details step by step how to correctly perform sampling in accordance with CENELEC requirements, based on experience gained from over 40 audits at treatment facilities.

Why correct sampling matters

Sampling of output fractions serves two fundamental purposes: demonstrating compliance with depollution requirements and calculating actual recycling and recovery rates. A treatment operator who cannot prove that the plastic fraction sent for recovery contains less than 2,000 ppm total bromine (the threshold for brominated flame retardants per EN 50625-3-1 Annex B) risks not only losing WEEELABEX certification, but also legal liability for sending hazardous waste to unauthorised facilities.

From experience, approximately 60% of treatment operators audited in Romania between 2023 and 2025 had deficiencies in sampling, ranging from a complete lack of procedures to errors in sample sizing.

Fundamental principles of sampling

TS 50625-3-1 is based on the principle of stratified representative sampling. This means that the final sample must reflect the actual composition of the entire lot, not just an accessible portion. The key principles are:

• Representativeness: every particle in the lot must have the same probability of being included in the sample
• Minimum sample size: determined by the maximum particle size in the fraction (the d95 rule — the size below which 95% of particles fall)
• Number of increments: a minimum of 10 increments per composite sample for standard output fractions
• Frequency: at least one sampling per homogeneous lot or at a defined interval (for example, every 50 tonnes processed)

Step-by-step sampling procedure

Step 1: Defining the lot

A lot is a quantity of material that is homogeneous in terms of source and treatment process. For example, the ferrous fraction resulting from shredding large household appliances in a single production campaign constitutes a lot. Do not mix fractions from different campaigns or from sources with fundamentally different compositions (e.g., iron from cooling equipment with iron from large cooking appliances).

The maximum recommended lot size is 500 tonnes for ferrous and non-ferrous fractions, and 100 tonnes for plastic fractions.

Step 2: Taking increments

Increments are taken from points systematically distributed across the entire surface and depth of the lot. TS 50625-3-1 recommends:

• Lots on conveyor belts: sampling at regular time intervals (for example, every 5 minutes of operation)
• Lots stored in piles: sampling from a minimum of 10 points distributed on an imaginary grid, at different depths
• Lots in containers: sampling from at least 3 depth levels per container, from different points

The mass of each increment depends on the maximum particle size. For fractions with a d95 of 50 mm, the minimum mass of an increment is approximately 2 kg. For fine fractions (d95 below 10 mm), 0.5 kg per increment is sufficient.

Step 3: Preparing the composite sample

All collected increments are combined into a composite sample. This must be homogenised through mechanical or manual mixing, then reduced to the size required by the laboratory using the successive quartering method or a rotary splitter. It is essential that the reduction does not introduce segregation errors — larger or heavier fragments tend to separate from finer ones.

Step 4: Preparing the laboratory sample

The reduced sample is sent to an ISO 17025 accredited laboratory. Preparation includes drying (at 105°C until constant mass), grinding to the size required by the analytical method and, for chemical analyses, acid digestion. The parameters analysed depend on the type of fraction:

• Plastic fractions: total bromine content (XRF or ICP-OES after digestion), lead, cadmium, and mercury content
• Ferrous fractions: proportion of non-ferrous metals, contamination with plastic materials
• Non-ferrous fractions: purity, content of foreign materials
• Dust and residues: heavy metal concentrations, SVHC

Step 5: Interpreting results

Results are compared against the limit values defined in EN 50625-1 and TS 50625-3-1. If a parameter exceeds the limit, a non-conformity is raised that requires corrective actions: adjusting process parameters, re-sorting, or redirecting the fraction to a facility authorised for hazardous waste.

Common mistakes and how to avoid them

Surface-only sampling

The most widespread error: the operator takes increments only from the top layer of the pile. Fine, heavier materials migrate towards the base of the pile through gravity. A sample taken only from the surface will systematically underestimate heavy metal and dust content.

Insufficient sample size

A 500 g sample from a 200-tonne lot of shredded iron is not representative. The minimum mass of the composite sample must be calculated based on the d95 of the fraction — for coarse materials, samples need to reach 20-30 kg before reduction.

Lack of documentation

Undocumented sampling is non-existent sampling from an audit perspective. Each sampling event must be accompanied by a report that includes: date, lot identification, number of increments, sampling points, sample mass, operator name, and storage conditions of the sample until dispatch.

Sampling frequency in practice

TS 50625-3-1 does not prescribe a universal fixed frequency, but provides guidelines. In practice, a reasonable regime entails:

• Plastic fractions: one sampling event every 20-50 tonnes processed or for each new lot, whichever comes first
• Metal fractions: one sampling event every 50-100 tonnes or per production campaign
• Dust and residues: at least quarterly or at each evacuation of the filtration system

Sampling is not a bureaucratic exercise — it is the only way a treatment operator can objectively demonstrate that its output fractions are what it claims them to be. Without rigorous sampling, reported recycling rates are mere estimates.