Equivalence testing of filter-based, beta-attenuation, TEOM, and light-scattering devices for measurement of PM10 concentration in animal houses

• We performed EN 12341 equivalence tests on five PM10 samplers inside animal houses.
• Candidate and reference samplers failed the comparability tests at high PM10 levels.
• Four out of five samplers showed substantial bias from the reference sampler.
• Duplicate sampling, correction factors and more frequent servicing are suggested.
• When aforementioned measures are taken, some samplers could be regarded acceptable.

Emissions of particulate matter (PM) from poultry and pig houses may contribute significantly to ambient concentrations. Yet, our knowledge on the accuracy and comparability of samplers available for measuring the high PM10 concentrations (>100 μg m–3) in the inside air directly upstream of the ventilation exhausts of these buildings is very limited. The aim of this study was to provide insight into this matter for five candidate samplers: a filter-based cyclone sampler (CYS), the Thermo Scientific FH 62 I-R beta-attenuation sampler (BAS), the Thermo Scientific Tapered Element Oscillating Microbalance, model 1400ab (TEOM), the TSI DustTrak model 8520 (DT), and the GRIMM Portable Aerosol Spectrometer model 1.109 (PAS). Equivalence tests were carried out following European Standard EN 12341 using two devices for each candidate sampler (CAS) and four filter-based low-volume reference samplers (RES). Measurements were performed inside three major animal housings (a fattening pig house, a laying hen house, and a broiler house) and inside an office room. Our key results and conclusions are: (1) neither one of the five CASs, nor the RES itself, met the EN 12341 requirement for comparability between devices of the same sampler type. Using a less strict boundary for this aspect – in concert with performing duplicate sampling – may be appropriate. (2) The CYS met the EN 12341 accuracy requirements in pigs and layers, but overestimated the RES concentration in broilers. The BAS, TEOM, and DT underestimated, and the PAS overestimated, RES concentrations in a systematic manner. The use of correction factors seems to be a promising method to calibrate measured values to RES concentrations. (3) The BAS, TEOM, DT, and PAS started to show scattered regression after 432–500 h of sampling, which stresses the need for shortened time intervals between full services. In conclusion, some of the samplers tested could be regarded acceptable when appropriate measures (such as duplicate sampling, correction factors, and more frequent servicing) are applied.