Improving bioaerosol exposure assessments of composting facilities — Comparative modelling of emissions from different compost ages and processing activities

Abstract

We present bioaerosol source term concentrations from passive and active composting sources and compare emissions from green waste compost aged 1, 2, 4, 6, 8, 12 and 16 weeks. Results reveal that the age of compost has little effect on the bioaerosol concentrations emitted for passive windrow sources. However emissions from turning compost during the early stages may be higher than during the later stages of the composting process. The bioaerosol emissions from passive sources were in the range of 10³–10⁴ cfu m⁻³, with releases from active sources typically 1-log higher. We propose improvements to current risk assessment methodologies by examining emission rates and the differences between two air dispersion models for the prediction of downwind bioaerosol concentrations at off-site points of exposure. The SCREEN3 model provides a more precautionary estimate of the source depletion curves of bioaerosol emissions in comparison to ADMS 3.3. The results from both models predict that bioaerosol concentrations decrease to below typical background concentrations before 250 m, the distance at which the regulator in England and Wales may require a risk assessment to be completed.

Introduction

Composting is reliant on the presence of various microorganisms, such as fungi and bacteria, which may become airborne and pose a risk to human health (Douwes et al., 2003). In response to perceived public health concerns, the Environment Agency of England and Wales (the environmental regulator) requires a risk assessment for licensed composting facilities that have a sensitive receptor within 250 m of their boundaries (Environment Agency, 2001). This must examine, among other hazards, the dispersal of airborne microorganisms or bioaerosols from the site (Wheeler et al., 2001; Environment Agency, 2002; Pollard et al., 2006). In this context, sensitive receptors may include a residence, school or office building. The aim of the risk assessment is to inform risk management. However, the quality of the risk assessment is dependant on the availability and quality of the bioaerosol source term data employed (Pollard et al., 2006). This data is frequently limited, in part because of the practical difficulties of microbiological analyses, but also due to cost constraints.

There is a growing body of research that examines the concentrations of bioaerosols in and around composting facilities (e.g. Danneberg et al., 1997; Swan et al., 2002; Sanches-Monedero and Stentiford, 2003; Taha et al., 2005, Taha et al., 2006). Many studies that attempt to predict the dispersal of bioaerosols from facilities use simple methods (e.g. Millner et al., 1980, Lighthart and Mohr, 1987; Swan et al., 2003; ADAS/SWICEB, 2005) and so are not able to take into account the full range of atmospheric and local conditions that could affect bioaerosol dispersion. Air dispersion models were developed to predict dispersion of pollutants, such as the nitrogen oxides, based on emissions and meteorological inputs, and have successfully been used for assessing odour dispersion from industry (e.g. Environment Agency, 2002). However, the use of these models for bioaerosol dispersion has been limited, as dispersion models may not be able to fully take into account the mechanisms of release of bacteria and fungi.

According to McCartney (1994), environmental factors such as wind speed, turbulence, humidity and water availability will influence when spores are released. Although these parameters are taken into account by dispersion models when predicting downwind concentrations, there are very few ‘at source’ measurements of bioaerosol concentrations that link these parameters to the concentrations measured. Furthermore, certain characteristics of bioaerosols complicate the use of dispersion models; in particular, their ability to form aggregates or clumps once released and the loss of viability that may occur as they are emitted from the compost windrow (Wheeler et al., 2001).

Our own research (Taha et al., 2004, Taha et al., 2005, Taha et al., 2006, Taha et al., 2007) focuses on improving the quality of regulatory risk assessments for composting by providing:

• accurate source term data at the point of release (Taha et al., 2006); and

• developing new methods for improving sampling and enumeration of bioaerosols (Taha et al., 2007).

Here, we propose further improvements by presenting the refinement of air dispersion modelling for the prediction of downwind bioaerosol concentrations at off-site points of exposure.

We present results from measuring on site and predicting the dispersal of bioaerosols from a green waste composting research facility in southern Wales. The advantage of sampling at this facility was that the processes on site could be adjusted to aid the sampling activities; for example, timing of shredding and screening. The objectives of the study were (i) to characterise the source term bioaerosol emissions, taking into consideration storage properties, compost age and dispersal during agitation using a windrow turner, front-end loader, screener and shredder; and (ii) to compare the predicted downwind concentrations modelled by two separate dispersion models. We seek to improve bioaerosol exposure assessments by refining the methods currently used to estimate downwind dispersal of compost emissions.

In order to estimate the static pile emission flux rate and bioaerosol active dispersal emission rate during agitation, the single source Gaussian plume model SCREEN3 (USEPA, 1995) was used to initially estimate bioaerosol depletion curves (Taha, 2005). SCREEN3 is a screening-level model that adopts steady-state Gaussian plume algorithms and meteorological scenarios to estimate worst-case dispersal. This was followed by the application of advanced modelling using the ADMS 3.3 air dispersion model (Carruthers et al., 1994; CERC, 2003). ADMS is an advanced steady state, Gaussian-like dispersion model, capable of modelling continuous plumes, short duration releases and transport over complex terrain. The model simulates point, line, area and volume sources, and can estimate pollutant concentrations at a number of user defined receptors. The model has been shown to perform in a comparable manner to similar new generation models (Hanna et al., 2000). Using the model results, we infer the possible influences that bioaerosol properties such as inactivation and microbial agglomeration may have on the depletion curves (decrease in concentration with distance) produced by the models.