Odour, dust and non-methane volatile organic-compound emissions from tunnel-ventilated layer-chicken sheds: a case study of two farms

Additional keywords: aerial dust, emission rate, olfactometry, particulate matter.

The Australia egg industry annually produces 357 million eggs, with the majority being produced in caged housing systems (AECL 2011). More layer farms will be required in the future as egg production grows at a rate of ~7–10% per year (AECL 2009, 2010, 2011) to meet consumer demands for fresh eggs.

Odour and dust emissions are a normal part of layer-chicken sheds but may lead to environmental impacts and amenity issues for nearby residents. The potential for odour and dust nuisance has increased in recent years due to larger egg production farms, increasing population densities near layer farms and greater air-quality requirements by the community. The development of new layer farms, or the expansion of existing farms, requires careful consideration and planning to ensure compatibility with adjacent land uses (EPA Qld Government 2004).

Layer hens are housed in specially designed buildings where they lay eggs that are collected and sold for human consumption. Hens enter the egg-laying stage of their life at 18 weeks and lay eggs for 12–15 months. Sheds may be naturally or mechanically ventilated and can be configured and managed for caged, barn or free-range production. The focus of the present paper is the housing configuration with tunnel ventilation and multi-tiered cages with manure belts. Tunnel-ventilated sheds provide a high degree of control of the in-shed environment in terms of temperature, fresh air exchange and manure conditions, to provide optimal conditions for hen comfort and egg production. Manure belts positioned under each tier of cages allow manure to be removed at regular intervals, e.g. weekly or semi-weekly (twice per week).

In layer sheds, there are many interdependent factors that may influence the generation and emission of odour and dust, including ventilation-system design and ventilation rate, shed design, manure properties (composition, moisture content, surface crusting, quantity, age and removal frequency), feed properties, stage of production (bird numbers, age, stocking density), bird activity, bird health and shed microclimate (e.g. temperature, humidity) (Navaratnasamy and Feddes 2004; Nimmermark and Gustafsson 2005; Angel et al. 2008; Modini et al. 2010; Mostafa and Buescher 2011; Xin et al. 2011; Costa et al. 2012; Ni et al. 2012a). The interaction of these many factors creates challenges for measuring ‘typical’ emission rates, and also for comparing odour emission rates from discrete studies.

Unpleasant odours originating from layer sheds are primarily produced from the microbial decomposition of manure (Mackie et al. 1998; Hobbs et al. 2004), although some odour may also be emitted from the birds themselves (Lacey et al. 2004). Where decomposing manure has surfaces exposed to the atmosphere, odorous compounds are released and exhausted from the shed through the fans. Once exhausted from the shed, the odour plume is subjected to dispersion, and is diluted as it travels downwind. When planning for a new or enlarged production facility, odour dispersion modelling may be used to estimate the likelihood of odour nuisance. Prediction of odour impacts with dispersion modelling requires accurate odour emission-rate values. There is therefore a requirement for reliable odour emission rate data.

Dust originates from the mechanical breakdown and subsequent entrainment of mineral and organic material from birds, manure, feed and manure into the air, or the conversion of gases to the particle phase (Modini et al. 2010). Only tunnel-ventilated layer sheds with manure belts will be considered in the present paper because, in general, layer sheds with manure belts have been found to have lower emissions than other shed styles (e.g. high-rise barns, deep pit barns, aviary systems) (Fabbri et al. 2007; Costa et al. 2012; Ni et al. 2012a).

Odour is formed by combinations of hundreds of odorants, some of which are classified as volatile organic compounds (VOCs), while others are gases such as ammonia or hydrogen sulfide (O’Neill and Phillips 1992; Schiffman et al. 2001; Hobbs et al. 2004; Lacey et al. 2004; Cai et al. 2006). VOCs are compounds with a molecular structure that contains at least one carbon and one hydrogen atom, and readily evaporates under normal indoor temperature and pressure conditions (Ni et al. 2012b). When investigating VOCs for odour research, it is useful to exclude methane (which is an odourless gas) because it can dominate the sample analysis. The remaining non-methane VOCs (NMVOCs) have a range of odour detection thresholds (concentration at which they can be smelt), although many are non-odorous or make very little contribution to the whole odour.

There are few examples of studies where NMVOCs in poultry housing have been identified (Cai et al. 2007; Trabue et al. 2010), and even fewer where the odorous NMVOCs have been specifically investigated. Early research into the identification of odorants from poultry housing determined that the ‘poultry’ odour was a mixture of compounds that individually had characters that could be described with the terms such as rotten eggs, cabbage, onion-like, butter-like and garlic (Burnett 1969; Miner 1977). Ongoing research has focussed on identifying and quantifying the abundance of NMVOCs and odorants. Investigation of NMVOCs from poultry farms has been limited due to technical challenges and requirements for a multiplicity of techniques to detect most of the NMVOCs present due to the range of compound reactivity and volatility (Trabue et al. 2010). Despite the challenges, identification, measurement and prioritisation of NMVOCs provides a useful and important dimension to odour research that will lead to improved methods of odour measurement as well as development and assessment of odour abatement strategies.

The aims of the present observational study were to measure odour and dust emission rates and identify NMVOCs in the air emitted from modern tunnel-ventilated layer sheds where manure is regularly removed with a manure-belt system. We anticipate that these data will be used to support and improve planning and assessment of new or expanding layer farms and will contribute to the development of improved odour measurement techniques and odour reduction strategies.

Odour, dust and NMVOC samples were collected from two tunnel-ventilated layer sheds (Table 1). Farm A was located in south-eastern Queensland and Farm B was located in the West Gippsland region of Victoria. Both farms had Hyline Brown hens (http://www.hyline.com.au, verified 1 May 2013), with average bird weight 1.85–1.93 kg on the days when samples were collected. Manure belts were used at each of these farms to regularly remove manure from beneath multi-tiered cages. At Farm A, manure accumulating on the belts was dried using normal shed ventilation, whereas at Farm B, a forced-air drying system was used continuously to dry manure on the manure belts. Both sheds were constructed using concrete floors, metal framework and insulated panel walls (foam laminated with a metal skin, similar to materials used in commercial cold rooms).

Table 1.  Specifications of the layer sheds included in the present study

Tunnel-ventilation systems were configured differently at each farm. At Farm A, 11 tunnel-ventilation fans were installed on one of the narrow ends walls of the shed, with air being drawn the full length of the shed in a single direction. At Farm B, 16 tunnel-ventilation fans were installed in the centre of one of the long side walls, with air being drawn from both ends of the shed towards the middle.

At each farm, odour, dust and NMVOC samples were collected during summer and winter over a series of days spanning the manure collection cycle, to assess the influence of manure accumulation on emissions. For Farm A, samples were collected on four sequential days, beginning on the day following removal of manure from the manure belts. For Farm B, samples were collected on Days 1–3, 5 and 6 following removal of manure from the manure belts.

Summer sampling at Farm A was conducted on 10–13 December 2007 and winter sampling at Farm A was conducted on 23–26 July 2007. Summer sampling at Farm B was conducted 25–29 February 2008 and winter sampling at farm B was conducted 16–20 June 2008.

Methods for the collection and analysis of odour, dust and NMVOC samples have previously been described by Dunlop et al. (2010), Modini et al. (2010), Parcsi et al. (2011) and Murphy et al. (2012). The following is a brief summary of these methods and also describes variations from the previously described methods.

To facilitate sample collection, a temporary polyethylene duct ~14.0 m long and 1.3 m in diameter, similar to the duct previously described by Sohn et al. (2008), was attached to one of the tunnel-ventilation fans, with the other end opened to the atmosphere. This duct was designed to provide a sampling position in accordance with AS 4323.1:1995 (Standards Australia 1995a) and to prevent interference at the sampling position due to cross-winds. Samples were drawn from the duct at a distance of eight duct diameters from the fan face.

Concentration of odour was determined using dilution olfactometry, which involved presenting a series of diluted odorous air samples to a panel of trained human assessors using a standardised process. In Australia, odour concentration is assessed according to the Australian/New Zealand Standard AS/NZS 4323.3:2001 (Standards Australia/Standards New Zealand 2001). Odour samples were collected into customised 120-L drums lined with a specially prepared polyethylene terephthalate bag (PET, Melinex, 15 µm, Dupont Teijin Films, Chester, VA, USA). Bags were filled using negative pressure (lung principle) in an average sampling time of 10 min. Odour samples were transported to the olfactometer and analysed within 8 h of collection. It was necessary to use two different olfactometry laboratories because the two farms were located in different states. On each sampling day, three or four pairs of duplicate samples were collected (six to eight sample bags analysed by the olfactometer). The geometric mean was calculated for each pair of duplicate samples to produce a single value of odour concentration for each set of sampling conditions.

Dust samples were collected isokinetically in accordance with AS4323.2:1995 (Standards Australia 1995b) from within the temporary polyethylene duct that was attached to one of the layer-shed exhaust fans. Isokinetic sampling was used to obtain representative dust samples independently of the dust particle-size distribution. No measurements of shed input air were conducted during the study because background dust concentrations in these rural areas are generally only a small fraction of the expected in-shed and exhaust dust concentration. It was judged more important to use the limited number of sampling instruments to comprehensively characterise the total dust emission from the sheds.

Dust concentrations were characterised in terms of PM₁₀ and PM_2.5 particle mass concentration (mass of dust per m³ of air) and particle number concentration (number of suspended particles in the 0.5–20-µm size range per m³ of air). Particle mass concentration was measured using two TSI Model 8520 DustTrak instruments with appropriate inlets (TSI Inc., http://www.tsi.com, verified 1 May 2013). Particle number concentration was measured using a TSI Model 3320 Aerodynamic Particle Sizer (TSI Inc.). All three instruments were operated in parallel, downstream of the isokinetic sampling probe, and recorded mass concentrations every 30 s and number concentration and size distributions every 20 s.

NMVOC samples were collected from within the temporary polyethylene duct that was attached to one of the layer-shed exhaust fans. Odorous air was drawn into sorbent tubes (Markes International Ltd, Pontyclun, UK) that were filled with Tenax TA or Carbotrap 300 sorbent (both types of sorbent tubes were used simultaneously for all sampling). A vacuum pump with low-flow tube holders (SKC Universal Pump 224-PCXR8 and 224-26-01, respectively, SKC Inc., Eighty Four, PA, USA) was used to induce a 100 mL per minute flow rate through each tube for a 30-min sampling period (total sample volume 3.0 L). Flow through individual tubes was set using a flow meter (Model 4143, TSI Inc.). Air drawn into the sorbent tubes was filtered to prevent contamination of the sorbent material by airborne particulate matter. Following sample collection, sorbent tubes were sealed and stored in a refrigerator before analysis.

Sorbent tubes were analysed using a thermal desorber (TD), gas chromatograph (GC), mass spectrometer (MS) and odour detection port (ODP), which is a series of instruments commonly referred to as TD–GC–MS/O. The TD was a Markes Unity Thermal Desorber (Markes International) and was connected to the GC using a heated transfer line. The GC and MS were Agilent 6890N gas chromatograph and Agilent 5973N mass spectrometer (Agilent Technologies, Sydney, Australia). The TD method is summarised in Table 2 and GC–MS/O methods are summarised in Table 3.

Table 2.  Instrument controlling parameters for the thermal desorption of Tenax TA and Carbotrap 300 sorbent tubes

Table 3.  Gas chromatograph–mass spectrometry/olfactometry (GC-MS/O) analysis specifications
ODP, odour detection port

Odour, dust and NMVOC samples were collected concurrently from each of the layer sheds. Samples were collected between 0530 hours and 1400 hours on each sampling day. This sampling window was chosen to enable the odour samples to be analysed on the same day as sample collection, to comply with AS/NZS 4323.3:2001 (Standards Australia/Standards New Zealand 2001) and to minimise potential olfactometry error due to sample decay (van Harreveld 2003). In addition, collecting odour and NMVOC samples within this time frame enabled emission rates to be measured over a range of ventilation rates on each sampling day (targeting sample collection at 25%, 50%, 70% and 100% of daily maximum ventilation rate if possible). This was because ventilation rate generally increased with ambient temperature throughout the morning, as determined by the in-shed environmental control system. Dust was measured continuously over this period.

Ventilation rate was recorded during each sampling event to enable calculation of emission rate. Ventilation rate was determined using fan-performance information supplied by the fan manufacturer, the shed static pressure at the time of sampling (measured in the shed ~6 m upwind of the fans) and the number of active fans. Ventilation rate was regularly checked by measuring air-speed through each active fan with a hot-wire anemometer using the traverse method (adapted from AS 4323.1:1995; Standards Australia 1995a) on the inlet side of each fan. Ventilation rate was found to be similar with both methods. Ventilation rates were adjusted for standard conditions (0°C and 101.3 kPa), as required by AS/NZS 4323.3:2001 (Standards Australia/Standards New Zealand 2001). In addition to measuring ventilation rate, fan activity and static pressure at the fans was continuously monitored at Farm A, by using a customised logging system, mercury tilt switches on fan shutters, similar to the method described by Wilhelm et al. (2001), and differential pressure transducer (Setra model 264, range –63 Pa to +63 Pa, part number 2641R25WB11T1C, Setra Systems, Boxborough, MA, USA). Fan-activity monitoring was undertaken to confirm that odour samples were collected across the normal daily range of shed-ventilation rates.

Odour and dust emission rates were calculated by multiplying the concentration of odour (ou/m³) or dust (mg or particles per m³) by the standardised ventilation rate (m³/s). Emission rate values were then normalised in terms of emission rate per 1000 birds placed in the shed at the start of the production cycle (ou/s per 1000 birds for odour and mg/s per 1000 birds or particles/s per 1000 birds for dust) or per kilogram of total liveweight (ou/s.kg for odour).

Manure on the belts was sampled for moisture content analysis, which was determined gravimetrically according to AS 4454-2003 (Standards Australia 2003). Ambient temperature and humidity were measured using a Kestrel Pocket Weather Tracker (Model 4500, Nielsen-Kellerman, Boothway, PA, USA). Production information, including number of birds placed, number of birds present on each sample collection day and average daily liveweight, were supplied by the farm manager.

Odour emission rates varied during each sampling day and on each day of additional manure accumulation on the manure belts. Odour emission rates (OER) ranged from 58 to 512 ou/s per 1000 birds (0.03–0.27 ou/s.kg, see Fig. 1). Daily average odour emission rates were not calculated because odour emission rates vary throughout each day due to a variety of influences, and each measurement is representative of different conditions.

Fig. 1.  Odour emission rates per 1000 birds placed (in the shed at the start of the batch) and per kilogram liveweight.

Ventilation rates varied with season (Fig. 2), with ventilation rates at Farm A tending to be greater during summer (as would be expected), while at Farm B, seasonal variation of ventilation rate was less obvious. One explanation for this was unseasonably cooler weather during the summer sampling period (mean daily maximum temperature for February 2008 was 22.9°C, compared with the long-term value of 25.2°C) and unseasonably warm weather during the winter sampling period (mean daily minimum temperature for June 2008 was 6.5°C compared with the long-term value of 4.0°C). These seasonal effects (and arguably unseasonal effects at Farm B) on ventilation rate had a similar influence on shed odour emission rate, with odour emissions from Farm A being greater in summer than in winter, while at Farm B, emission rates were similar for both seasons, with slightly higher emission rates being measured during winter. To complicate matters, the higher odour emission rates measured at Farm B in winter may have also been related to higher moisture content of the manure.

Fig. 2.  Ventilation rate (m³/s) recorded during the collection of each odour sample.

Odour samples were collected over a range of ventilation rates on each sampling day, in an attempt to measure daily variations of odour emission rate. Fig. 3 displays the shed-ventilation rate on each of the sampling days at Farm A, with the ventilation rate at the collection time of each odour sample highlighted. It can be seen that, on most days, odour samples were collected at a range of ventilation rates spanning the normal range of ventilation rate for that day. This figure also shows that odour samples were collected during the earlier parts of the day when the rate of ventilation was increasing.

Fig. 3.  Continuous ventilation rate at Farm A during summer and winter, displaying the ventilation rate when the samples were collected (‘Sample VR’).

PM₁₀ mass concentrations ranged from 0.029 to 0.124 mg/m³ (Fig. 4), PM_2.5 concentrations ranged from 0.005 to 0.061 mg/m³ (Fig. 5) and particle-number concentrations ranged from 1.45 × 10⁶ to 1.95 × 10⁸ particles/m³, although the higher value corresponded with a short-term spike, with the majority of the number concentrations being less than 2.00 × 10⁷ particles/m³.

Fig. 4.  PM₁₀ concentration at Farms A and B.

Fig. 5.  PM_2.5 concentration at Farms A and B.

PM₁₀ emission rates ranged from 0.014 to 0.184 mg/s per 1000 birds (Fig. 6), PM_2.5 emission rates from 0.001 to 0.190 mg/s per 1000 birds (Fig. 7) and particle-number emission rates ranged from 7.90 × 10⁵ to 5.93 × 10⁸ particles/s per 1000 birds, although the majority of particle-number emission rates were below 3.00 × 10⁷ particles/s per 1000 birds.

Fig. 6.  PM₁₀ emission rate per 1000 birds placed (in the shed at the start of the batch).

Fig. 7.  PM_2.5 emission rate per 1000 birds placed (in the shed at the start of the batch).

At Farm B, a single dust measurement was performed while farm staff used a petrol-powered blower (similar to a garden blower) to blow dust out of the shed, which is an operation performed approximately once weekly on this farm. While staff used the blower, the ventilation rate in the shed was increased to ~62.5% of the maximum tunnel-ventilation rate of the shed. During this cleaning operation, PM₁₀ concentration was 0.192 mg/m³, PM_2.5 concentration was 0.030 mg/m³ and particle number concentration was 1.01 × 10⁷ particles/m³. Emission rates were 0.293 mg/s per 1000 birds for PM₁₀, 0.046 mg/s per 1000 birds for PM_2.5, and the particle number emission rate was 1.53 × 10⁷ particles/s per 1000 birds. The research team did not consider these values to be representative of normal emission rates due to manual over-riding of the ventilation system and re-entrainment of settled dust caused by the blower. These values were not included in Figs 4–7).

Twenty NMVOCs were identified in the emissions from the layer sheds (listed in Table 4). Some NMVOCs were present at both farms such as 1-butanol, 2-butanone, 2-ethyl-1-hexanol, acetic acid, benzaldehyde and toluene, while others were found at only one of the farms, such as dimethyl sulfide, 3-hydroxy-2-butanone, 2,3-butanedione and cyclohexanone.

Table 4.  Non-methane volatile organic compounds (NMVOCs) identified in the emissions from Farms A and B

Many of the NMVOCs were measured in such low abundance that they did not elicit an olfactory response at the odour port during TD–GC–MS/O analysis. The only odorants identified during analysis were 2,3-butanedione, 2-butoxy-ethanol and cyclohexanone.

At Farm A, manure moisture content during the winter sampling period was 59.8%. Data from the summer sampling period at Farm A were unfortunately lost. At Farm B, manure moisture content during the summer sampling period was 54.6% and during winter it was 66.9%.

Odour and dust emissions need to be individually considered, along with environmental and in-shed conditions at the time of measurement. Comparison of the odour and dust emission rates and identified NMVOCs reported in the literature with those measured in the present study need to be undertaken cautiously due to differences in building design, ventilation system, shed management practices, manure management practises, bird breed and feed.

OERs measured in the present project ranged from 58 to 512 ou/s per 1000 birds or from 0.03 to 0.27 ou/s/kg (Fig. 1) and are generally within the range of previously reported OERs from layer sheds; however, the minimum values were lower than those reported in the literature (refer to Table 5).

Table 5.  Reported odour emission rates (OERs) for tunnel-ventilated layer sheds with manure belts
Values given in parentheses are the geometric mean of the OERs

OERs were observed to be higher on the day following removal of manure from the manure belts and again after 4–6 days of manure accumulation, and tended to be lower on Days 2–4 following removal of manure from the manure belts. This may have occurred because manure-belt scrapers freshly smeared a thin layer of manure over the surface of each manure belt, increasing emission surface area and specific emission rate, which resulted in increased emissions during the first 24 h. Drying and crusting of the manure film during the first 24 h may have then reduced emission rates for a couple of days, until accumulation of manure after 4 days led to increased odour emissions. OERs reported in the literature were measured in sheds with a range of manure-removal intervals and manure drying methods, but it is unclear when in the manure cycle those OERs were measured and whether active manure drying had any influence. Further studies should focus on quantifying the effect of the manure-removal interval and active manure-drying systems on OERs. On the basis of the limited observations in the present study, removing manure from the belts every 4 days may be an effective odour reduction strategy, but this requires further investigation.

Seasonal variability in odour emission rates was observed at Farm A, where high OERs coincided with higher summer ventilation rates and, conversely, low OERs coincided with lower winter ventilation rates. Seasonal variability was less obvious at Farm B, presumably due to unseasonable ambient temperatures during the winter and summer sampling campaigns that resulted in uncharacteristic ventilation rates and OERs in both summer and winter. Seasonal effects on OERs from intensive animal housing, similar to those seen at Farm A, have previously been reported, with higher odour emissions being measured during summer (Hayes et al. 2006; Cai et al. 2010). In contrast, findings by Sun et al. (2010) supported the OERs measured at Farm B because odour emission rates did not correlate with season and were high and low throughout the year due to a variety of reasons. In contrast to OERs, they found that odour concentration within an intensive animal house increased during winter due to low ventilation rates and decreased in summer due to high ventilation rates. Emission rates should be measured throughout the year to reduce uncertainty when the data are to be used in odour dispersion modelling.

Odour emission measurements on each day were spread fairly zevenly across the normal range of ventilation rate for that day (as shown in Fig. 3 for Farm A). The sampling procedure was applied equally at Farm B, where ventilation rate was not continuously monitored outside the odour-sample collection period, but it is assumed that measuring OERs over the normal daily range of ventilation rate was achieved with equal success. The importance of measuring the emission rates over the full daily spread of ventilation rates cannot be understated, especially when quantifying daily emission rates at mechanically ventilated poultry sheds where there can be significant diurnal variation in ventilation requirements.

In the present study, odour emission rates were measured only in the mornings as daily ventilation requirements were increasing. In future studies, OER measurements using olfactometry or instrumental techniques should be undertaken to quantify emission rates throughout the entire day, especially targeting times that may be more likely to correspond with odour impacts. Sun et al. (2008) also recommended monitoring emissions from intensive animal housing throughout the entire day when the intention is to use the data in odour dispersion modelling. This will reduce uncertainty in the prediction of odour impacts that may be associated with using randomly measured data.

Dust concentration and emission rates (PM₁₀ and PM_2.5) measured in the present study ranged from 0.029 to 0.124 mg/m³ for PM₁₀, and from 0.005 to 0.061 mg/m³ for PM_2.5. PM₁₀ emission rates were 0.014–0.184 mg/s per 1000 birds and PM_2.5 emission rates were 0.001–0.190 mg/s per 1000 birds. These measured concentration and emission rates were similar or lower than values reported in the literature for caged layer sheds with manure belts (refer to Table 6). It is likely that some of the observed variability in dust concentrations can be attributed to in-shed activities that increase bird activity such as feeding, lighting changes, egg collection and farm staff activity (Mostafa and Buescher 2011; Costa et al. 2012).

Table 6.  Reported PM10 and PM2.5 concentrations and emission rates for tunnel-ventilated layer sheds with manure belts
Values given in parentheses are mean values from multiple or continuous measurement

There were no obvious relationships between PM₁₀ and PM_2.5 dust concentrations and manure accumulation on the manure belts (Fig. 4, 5). Shed-ventilation rate also appeared to have little influence on dust concentration, although slightly higher dust concentrations were measured at low ventilation rates (Fig. 8), presumably due to reduced dilution with fresh air. Because PM₁₀ and PM_2.5 dust concentrations were insensitive to ventilation rate, dust emission rates tended to increase with ventilation rate (Fig. 9). A short spike in dust concentrations and emission rates was observed while the farm staff were using a petrol-powered blower to blow dust out of the shed. Activities such as this should be scheduled for times when they are less likely to cause impacts.

Fig. 8.  Dust concentration (PM₁₀ and PM_2.5) vs shed-ventilation rate.

Fig. 9.  Dust emission rate (PM₁₀ and PM_2.5) vs shed-ventilation rate.

NMVOCs identified in the present study that have been previously identified in layer sheds included 1-butanol, acetic acid, butanoic acid and phenol (Hobbs et al. 1995; Mårtensson et al. 1999; Cai et al. 2007). NMVOCs identified in the present study that have not previously been reported included 2,3-butanedione, 2-butoxy-ethanol, cyclohexanone, 2-butanone, dimethyl sulfide, 2-ethyl-1-hexanol, toluene, styrene, benzaldehyde, acetophenone, 3-methyl-butanal, 3-hydroxy-2-butanone, o-xylene, p-xylene and nonanal. Many of these NMVOCs were measured in such low abundance that identification of odorants was challenging. The only odorants identified using the odour port on the TD–GC–MS/O were 2,3-butanedione, 2-butoxy-ethanol and cyclohexanone. Future NMVOC studies on layer-shed emissions may benefit from a refinement of sampling methods such as collecting a larger sample volume in the sorbent tubes. Additional research to identify NMVOCs and odorants on poultry farms should be undertaken to improve understanding of the relationships between primary odorants and odour concentration and to enable the development of instrumental odour/odorant assessment techniques. Such techniques will complement or be able to replace dilution olfactometry, improving odour measurement, enabling continuous odour measurement, especially at times when dilution olfactometry is impractical and will improve confidence in the assessment of odour-reduction strategies (as previously demonstrated by Cai et al. (2007)).

Odour and dust emission rates measured in the present study were observed to vary with diurnal and seasonal ventilation requirements. Odour emission rate also varied with accumulation of manure on the manure belts. Ventilation rate appeared to have a strong influence on emission rates and, therefore, requires close attention in future emission rate studies in mechanically ventilated poultry sheds. The expected range of ventilation rates at the time of sampling should be understood before sampling, to enable planning of emission rate measurements across the full, normal range of ventilation rates for that time of day and season. Future studies in layer sheds with manure belts should also consider the manure-removal cycle as well as manure moisture content, and measure emission rates throughout the cycle.