Evaluation of the detection of glass contamination in horticultural peat and compost using microwave reflection by Monte Carlo simulation

Contamination of bagged peat by glass fragments needs to be detected before point-of-sale to protect users from cut injuries. In order to evaluate the potential for designing a microwave detection system, a bootstrap Monte Carlo simulation of transverse-electric plane microwave reflection from glass-contaminated random flat strata of wet peat and compost was carried out. Measured complex permittivity data for peat, compost, and soda-lime glass were used in conjunction with calculated water vapour pressure for the simulation. The simulation was for polyethylene bag material on both faces of randomly generated peat (or compost) strata on a supporting stainless-steel platform. This was to simulate bagged peat or compost. Generally, a dual-frequency analysis resulted in more consistent discrimination between pure and contaminated peat and compost samples than single-frequency phase-shift or amplitude ratio. Best performance was obtained with dual-frequency corrected phase difference for peat (difference=0.15 rad, p<0.0005) and compost (difference=0.23 rad, p<0.0005). The most influential factors in the detection of glass were bag thickness, high frequency used and interactions of bag thickness and high frequency used with other factors. The results suggest that a reliable microwave glass detection system is not technically feasible for use with peat/compost bags.