Instrumentation for precise quantification ofmethane emissions from dairy herds

F.J.B. Tremblay and D.I. Masse*

Agriculture and Agri-Food Canada, Dairy and Swine Research and Development Centre, 2000 College Street, P.O. Box 90,Sherbrooke, Quebec, Canada, J1M 1Z3.*Email: massed@agr.gc.ca

Tremblay, F.J.B. and Masse, D.I. 2008. Instrumentation forprecise quantification of methane emissions from dairy herds.Canadian Biosystems Engineering/Le genie des biosystemes auCanada 50: 7.21�7.28. The objective of this study was to developa method for accurately quantifying methane emissions fromdairy herds under conditions similar to those that exist incommercial dairy operations. A mass balance approach appliedto a barn was utilized. The barn was maintained under negativepressure by means of mechanical ventilation. The calibrationcurves of each fan were precisely determined on site.The instrumented barn continuously measures methane emis-

sions from a herd of 21 cows, 24 h a day, all year round. Thebarn is also equipped to regularly measure ambient conditionsoutside and inside the barn, water consumption per herd, milkproduction and quantity of rations dispensed and consumed percow. This instrumented barn presents an environment similar tothat generally found on commercial farms. The system isaccurate to within 4.5%. Keywords: Methane, entericfermentation, dairy cattle, greenhouse gas, dairy barn.

Cette etude visait a developper une technologie capable dequantifier avec une grande precision les emissions de methaneproduites par un troupeau laitier dans des conditions d’exploita-tion similaires aux pratiques de l’industrie. Une approche debilan massique sur une etable a ete utilisee. L’etable estmaintenue sous pression negative par une ventilation mecaniquedont les courbes de calibration de chaque ventilateur ont eteetablies sur le site avec precision.L’etable instrumentee permet de mesurer en continu les

emissions de methane produitent par un troupeau de 21 vaches24 heures par jour et a l’annee. L’etable est egalement equipeepour mesurer de facon reguliere les conditions climatiques al’exterieur et a l’interieur de l’etable, la consommation d’eau parle troupeau, la production de lait et la quantite d’aliment servieet ingeree par vache. Cette etable instrumentee presente unenvironnement similaire a celui que l’on retrouve generalementen ferme commerciale. La precision du systeme est de4,5%. Mots-cles: methane, fermentation enterique, bovinlaitier, gaz a effet de serre, etable laitiere.

INTRODUCTION

The ratification of the Kyoto Protocol by Canada in 2005requires all sectors of the Canadian economy to reducetheir current level of greenhouse gas (GHG) emissions by6% below their 1990 level. According to EnvironmentCanada (2001), GHG emissions from the agriculturalsector totaled approximately 60 Mt CO2 equivalents in2000. Agricultural emissions represent about 10% ofCanada’s total GHG emissions. Livestock production

accounts for 42% of total GHG emissions from agricul-ture, two-thirds of which are associated with entericfermentation emissions from Canadian cattle (Boadiet al. 2003).

Canada uses the emission factors of the Intergovern-mental Panel on Climate Change (IPCC) to establish itsinventory of GHG emissions from cattle. However,methane emissions from cattle vary depending on therations consumed by the animals, their age and weightand their environment (Sutton et al. 1986; Mathisonet al. 1998; Harper et al. 1999; Shabi et al. 1999; Dohmeet al. 2000; Moss et al. 2000). As indicated by Boadi et al.(2003), the Canadian data are insufficient to relate GHGemission to all agricultural practices used by the dairyindustry across Canada.

METHODS of METHANE QUANTIFICATION

The objective of this study is to develop a method forquantifying the impact of feeding practices on methaneemissions from dairy cows. The methods must be suffi-ciently accurate to detect variations in methane produc-tion of 5%. In addition, the instrumented infrastructuremust offer an environment similar to the business as usualpractices that prevail in Canadian commercial dairyoperations.

Several methods can be used to quantify methaneemissions from a dairy cow or herd. Johnson and Johnson(1995) present various methods and discuss the advantagesand disadvantages of each. The respiration chambermethod is a particularly interesting approach because itprovides a high level of accuracy and makes it possible tocontrol all production parameters relating to the animal.However, the disadvantage of this method is that itconfines and isolates the animal in a restricted environ-ment that can modify its behavior, reduce its feed intakeand induce variance factors. Methods making use of openpath laser are not accurate. They do not provide con-tinuous monitoring because of frequent changes in winddirection and of periods of low wind velocity or withoutwind. Methods based on tracer gas have been used toquantify methane emission from mechanically and natu-rally ventilated buildings (Phillips et al. 1998; Kaharabataet al. 2000). Fisk and Faulkner (1992) have found that thetracer gas does not mix uniformly inside the building andthe best precision to quantify the building ventilation ratewas around920%. Powell et al. (2007) used small

Volume 50 2008 7.21CANADIAN BIOSYSTEMS ENGINEERING

chambers containing up to four cows to quantify ammo-nia emission with a precision of920%. The precisemeasurement of the air flow rate across a livestockbuilding is a major challenge (Lefcourt et al. 2001a,2001b). Singh et al. (2003) and Zhang et al. (2001) foundthat gas concentration at the outlets is not homogenous.Therefore, precise quantification of gaseous emissionsrequired accurate monitoring at each outlet. Accordingto Ozcan et al. (2007), air flow rates in a mechanicallyventilated building can be measured with a precision of 5to 16%. The precision is lower (15 to 40%) with naturallyventilated buildings. Some air flow measurements arecarried out at the air plenium level. This approach isapplicable to some swine and poultry buildings where theinlet air is preheated in a plenium (Osada et al. 1998; Hoffet al. 2000). Anemometers are frequently used to establishthe ventilation rate of dairy buildings, but this methodlacks accuracy because the anemometer does not cover theentire ventilator surface, and assumes that the air flow rateis constant across the ventilator opening (Heber et al.2005).

In this study, the respiration chamber approach wasapplied to an entire livestock building. The approachconsists of establishing a mass balance on a barn betweenthe quantity of methane entering the building and thequantity of methane leaving the building. The difference inthe methane masses entering and leaving the buildingrepresents the quantity of methane produced by theanimals housed in the building. The parameters used toestablish the mass balance are methane concentrations atthe air inlets and outlets of the building and the air massflow rate at the outlets. Given that there is no accumula-tion of air mass in the building and that the building ismaintained under negative pressure, the air mass at theinlets is therefore equal to the air mass evacuated by theexhaust fans. Although the system developed in this studyhas some similarities with the systems developed byJackson et al. (1993) and Kinsman et al. (1995), severalmajor improvements were made to improve the estimationof air flow from the exhaust fans, to the treatment of theair sample prior to analysis and, more specifically, to theautomation of the system in order to increase its accuracyand reliability. All ventilators were calibrated very pre-cisely. A unique aspect of this study is that the calibrationequipment was pre-calibrated in the laboratory. Calibra-tion curves were developed for each ventilator to estimatethe ventilator flow rate as a function of the ventilatorRPM and the static pressure across the building envelopeat the ventilator location. Kinsman et al. (1995) used astatic pressure for a bank of ventilators, in contrast to thisstudy, in which the static pressure was measured at eachventilator. The accuracy of the systems developed byKinsman et al. (1995) was around915%; in this study theobjective is to achieve a precision of95%.

Figure 1 illustrates the floor plan of the instrumentedbarn used in this project. The barn section is 34 m long by7 m wide. It can accommodate a row of 21 cows housed intie stalls. Negative pressure is continuously maintained inthe building by means of seven 40 cm wall exhaust fansinstalled on the south wall of the barn. Fresh air is

provided by a 24-m inlet located at the top of the northwall. The inlet opening and the number of ventilators inoperation are regulated by a temperature controller. Adifferential pressure meter (Modus T40, General Electric,USA) is used to measure the pressure difference betweenthe building and the attic to ensure that the building isalways under negative pressure.

Figure 2 illustrates the operation of the measurementsystem. Air samples were continuously collected at 12locations in the building (see Fig. 1 for sampling locations)and were accumulated in collection reservoirs. Sampleswere taken at each exhaust fan location to determine theconcentration of methane leaving the building via theoutlets. Two samples were also continuously collected atthe air inlet to determine the concentration of methane infresh air entering the building. One air sample wascontinuously taken in the attic to verify that there wasno methane escaping from the ceiling of the barn. One airsample was also continuously taken near the cover of theholding tank to verify that the feces stored in the transferstructure do not contribute to the methane emissions.

The methane mass balance on the instrumented dairybuilding was calculated as follows:

�mCH4� inlets ��mCH4

� cows��mCH4� outlets (1)

�mCH4� cows��mCH4

� outlets ��mCH4� inlets (2)

�mCH4� cows

�X7

i�1

[[mair]i �[CH4]i]��X7

i�1

[mair]i �1

2

X2

j�1

[CH4]j

�(3)

where�mCH4� cows is the mass of methane produced by

dairy cows inside the building, [mair]n is the air mass flowrate at the ventilator n, [CH4]n is the methane concentra-tion at location n, i is the number of ventilator locations, jis the number of inlet sample locations.

The collection reservoirs in which the air samplesaccumulate are made of PVC cylinders measuring1350 mm in length and 305 mm in diameter (Fig. 3). Thereservoirs allow the continuous collection of air sampleover a measurement cycle. The representative air samplesare analyzed during the subsequent measurement cycle.The cylinders are hermetically sealed at both ends bymeans of an acrylic cover. A 0.075 mm thick polypropy-lene bag placed inside the reservoir and connected to twoports of the reservoir is completely isolated from thereservoir. A system of two three-way solenoid valves(ASCO Model 8320G184, Florham Park, NJ, USA) isused to alternately empty the reservoir and the bag. Thevacuum created by the aspiration of air contained in oneof the chambers (reservoir or bag) fills the other chamber(reservoir or bag) with ambient air and vice versa. Airsampling is performed continuously for the entire periodof the measurement cycle and is analyzed in the followingcycle. Each measurement cycle lasts 40 min, including 2min for purging the sampling tubes connected to the gasanalyzer, 2 min for the gas analysis of each of the 12

7.22 TREMBLAY and MASSELE GENIE DES BIOSYSTEMES AU CANADA

samples, and 4 min to completely empty the gas samplingreservoir.

A partial vacuum is created with a vacuum pump(Busch model SV1003, Maulburg, Germany) in the 12sampling tubes connecting the sampling reservoirs to thevalve system (Fig. 4). The air flow in the sample tubes iscontrolled by a volumetric flow meter (Dwyer VFB-66-BV, Michigan City, IN, USA). Twelve three-way solenoidvalves (ASCOModel 8320G184, Florham Park, NJ, USA)are mounted on two separate manifolds; one manifoldisolates one of the air samples sequentially and directs it tothe analyzer, whereas the other manifold allows circula-tion of the 11 other air samples into the atmosphere. Adiaphragm pump (GAST DAA-V111, Benton Harbor,MI, USA) draws the air sample from the isolated duct anddelivers it to a sample conditioning system for analysis.The sample conditioning system consists of a thermo-electric dryer, which lowers the sample dew point toapproximately 58C, and a 1-mm TeflonTM filter, whichreduces the amount of dust in the sample (UniversalAnalyzers 620SS, Carson City, NV, USA). The sample isthen analyzed by means of an infrared analyzer(MGA3000, ADC Gas Analysis Ltd., Herts, England).

Continuous measurement of static pressure at the fan(Modus T40, General Electric, USA) and fan rotationspeed (SRF-3A, Microswitch, Honeywell, USA) at eachfan, and the measurement of temperature (ThermocoupleType T) at each sampling location, make it possible toestimate the air mass flow rates in the building.

Interior and exterior temperature and humidity levels,wind speed and direction, and barometric pressure in thebuilding are measured by means of a weather station(Davis Instruments Weather Monitor II, Hayward, CA,USA). The barometric pressure is used to correct theestimated air mass flow rates.

The readings from the analog sensors (thermocouple,pressure sensor), the frequency of the signal (proximitysensor), the opening of the ports and the operation of thevalves are all controlled by a Series 200 data acquisitionand control system (Sciemetric Instruments, Ottawa, ON,Canada) connected to a computer. The computer alsocommunicates with four peripheral devices: the weatherstation; the methane analyzer; the balance and the massflow meter, which are used to release methane intobuilding via a RS232 series communication port.

The computer system that controls the entire system isconstructed of several layers of software, all of which weredeveloped in a LabVIEWTM environment (National In-struments, Houston, TX, USA). The first layer consists ofdevice drivers specially developed to be compatible inLabVIEWTM in order to communicate with the peripheraldevices (weather station, methane analyzer, mass flowmeter, digital balance and data acquisition and controlsystem). This layer regularly refreshes the display of thevarious readings at a rate determined by the communica-tion speed of the peripheral device. The second softwarelayer consists of a shared memory space to whichinstantaneous data are transferred from the various

Fig. 1. Diagram of the building layout.

Volume 50 2008 7.23CANADIAN BIOSYSTEMS ENGINEERING

peripherals supported by the drivers. The third softwarelayer constitutes the body of the system. This programsynchronizes the system and ensures that data areacquired at significant times to establish the variouscalculations. A fourth layer processes and records thedata, evaluates the quality of the data, sends warning e-mails if necessary and updates the various displaysintended for the user.

The building used is equipped with various pieces ofequipment for monitoring the production parameters.Four water meters (Les compteurs Lecomte ltee, Saint-Hyacinthe, QC, Canada) provide daily measurements of

the herd’s total water consumption. Lactometers (Tru-Test WB EziTest, New Zealand) permanently hooked upto each milking unit are used to weigh the milk of eachcow and, as required, to take milk samples for analysis.The total mixed ration and hay fed as well as refusals weremeasured for each animal.

The seven fans in the building were calibrated. Figure 5illustrates the nozzle used in the building to establish a setof calibration curves for each fan for different RPM andstatic pressures. The nozzle measures 457 mm in diameterand is equipped with a Wilson Flow Grid (AirFlowInstruments, Buckinghamshire, England) and a T40

Fig. 2. Schematic of the measurement system.

Reservoir inlet Reservoir outlet

Bag inlet Bag outlet

COM COM

NO

NC

Air Sample

NC

NO

To analyser

3/2 solenoid valves

Fig. 3. Schematic of sampling reservoirs.

7.24 TREMBLAY and MASSELE GENIE DES BIOSYSTEMES AU CANADA

pressure sensor to determine the flow in the nozzle. Avariable-speed backup fan measuring 450 mm in diameter(Multifan 4E45) is installed at the nozzle inlet andconnected to an analog controller (Varifan VSD-1,Monitrol, Montreal, QC, Canada). This nozzle was firstcalibrated in a chamber equipped with an inlet thatconforms to AMCA (1999). The ventilator installationsetup in the chamber was identical to the setup in the dairybarn. A calibration curve linking the air flow in thechamber and the pressure difference at the boundaries ofthe Wilson Flow Grid was constructed in accordance withthe manufacturer’s procedures.

Each component of the measurement method has alevel of inaccuracy. Because it would be very difficult toestablish the overall method accuracy from the compo-nents’ accuracy, we have established the accuracy of thewhole system (instrumented dairy barn) by quantifying therecovery rates of known quantities of methane released inthe dairy building. Pure methane (grade 2.0-99%, Praxair,Canada) was released into the building in the absence ofanimals. The mass of released gas was compared with themethane mass measured by the system. The flow of therelease was controlled by a digital mass flow meter(FMA-2608, Omega Engineering, USA). To increase the

reliability of the mass of released gas, the cylindercontaining the methane was placed on a balance (GSEScale Systems GSE465, Novi, MI, USA), which isaccurate to91 g.

RESULTS AND DISCUSSION

Sensitivity and reproducibility

Fifteen methane release tests were performed on thesystem over an 18-month period. The quantities ofmethane released were 250 to 500 g over periods rangingfrom 1 h 20 min to 3 h, for a methane release rate rangingbetween 2 and 5 L/min. This range of release ratessimulated a herd of about 10 cows with an average entericmethane production of 450 L per head per day (Kinsmanet al. 1995). The mean ambient methane concentration inthe building during the tests ranged between 5 and 70ppmv depending on ventilation rate. Eight tests wereconducted during winter ventilation conditions (a singlefan operating), five tests were conducted during summerventilation conditions (seven fans operating), and two testswere conducted during transition ventilation conditions(four fans operating). Figure 6 shows the history of the

Common pumpBUSCH SV1003

Analyser pumpGAST DDA-V111

V1

V2

V3

V12COM NO

NO

NO

NO

NC

NC

NC

NC

...

...

To atmosphere

To analyzer

Flow meter

Air sample from location 1

Air sample from location 2

Air sample from location 3

Air sample from location 12

...

12 x 3/2 solenoid valves mounted on 2 mainfolds

COM

COM

COM

Fig. 4. Diagram of solenoid valves and pumps.

Fig. 5. Fan calibration nozzle.

Volume 50 2008 7.25CANADIAN BIOSYSTEMS ENGINEERING

ratios of the methane mass measured by the system to theknown mass of methane released.

The statistical average of the ratios of the methanereleased over the methane recovered for the 15 calibrationtests is 99.4%, with a standard deviation of 2.1%. If weassume that the error of the system follows a t-studentdistribution with 14 degrees of freedom, the system isaccurate to within94.6% (95% confidence interval). Thevariation between the tests with a different number of fansis not significant.

In order to estimate uncertainty of the methanereading and its variation in time, the system tookmeasurements for a period of just over 3 days withoutanimals and without any source of methane in thebuilding. The implicit error of the equipment that makesup the system introduces noise into the measurement ofthe difference in methane between the exits and the inlets.Figure 7 shows the change in the measurement over the

period observed. The average of the methane readings is�0.03 g, and the standard deviation is 2.66 g, whichshows that the distribution of the error is similar to anormal distribution (Fig. 8). Given that the averagemeasurement is very close to zero, no offset is observed.Therefore, the uncertainty of the reading is less than thereproducibility and can therefore be disregarded in a dailymethane balance in the building.

Continuous measurement of methane productionmakes it possible to measure variations in productionrate as a function of the time of the day. Figure 9illustrates the profile of methane production measuredby the system over a 24-h period. The herd in the buildingconsisted of 10 lactating cows. Daily methane productionis determined by integrating the area under the curve. Asseen in Fig. 9, there is a variation of 80% between theminimum methane production, which occurs early in themorning, and the daily maximum, which occurs late in theevening. This variation demonstrates the importance ofcontinuously monitoring the air from the barn to accu-rately estimate daily methane production.

CONCLUSION

The quantification system developed in this study has theaccuracy of the respiration chambers for quantifyingmethane emissions, while reproducing an operating en-vironment that is more representative of commercial dairyfarms. The instrumented building makes it possible tomeasure emissions from a herd of up to 21 cows, whichreduces the variability of readings associated with onecow. The system is reproducible to94.5%, and thedeviation of the methane reading is zero over a 3-dayperiod. The variation of absolute error follows a virtuallynormal distribution, and the sum of the errors over a78-h period was �0.032 g of methane. The system is

Fig. 6. Building calibration.

Fig. 7. Variation in readings of the system without animals

in the building.

7.26 TREMBLAY and MASSELE GENIE DES BIOSYSTEMES AU CANADA

completely automated and provides methane measure-ments distributed over the entire daily period.

In addition to quantifying methane for different cowherds, the system is sufficiently sensitive to validate certainmethane emission reduction strategies in a dairy produc-tion environment similar to that found in the Canadiandairy industry.

ACKNOWLEDGEMENTS

The authors wish to acknowledge the financial support ofthe Dairy Farmers of Canada.

REFERENCES

AMCA. 1999. ANSI/AMCA Standard 210-99 (AN-

SIASHRAE Standard 51-1999), Laboratory Meth-

ods of Testing Fans for Aerodynamic Performance

Rating. Arlington Heights, IL: Air Movement and

Control Association International, Inc.

Boadi D., C. Benchaar, J. Chiquette and D.I. Masse.

2003. Mitigation strategies to reduce enteric meth-

ane

emissions from dairy cows: Update review. Cana-

dian Journal of Animal Science 84: 319-335.

Dohme F., A. Machmuller, A. Wasserfallen and M.

Kreuzer. 2000. Comparative efficiency of various

fats rich in medium-chain fatty acids to suppress

ruminal methanogenesis as measured with RUSI-

TEC. Canadian Journal of Animal Science 80: 473-

482.

Environment Canada. 2001. Canadian Greenhouse Gas

Inventory for 1990-2000, p159. Ottawa, ON: En-

vironment Canada.

Fisk W.J. and D. Faulkner. 1992. Air exchange effective-

ness in office buildings: Measurement techniques and

results. In International Symposium on Room Air

Convention and Ventilation Effectiveness. Tokyo,

Japan. July 22-24.

Harper L.A., O.T. Denmead, J.R. Freney and F.M. Byers.

1999. Direct measurements of methane emissions from

grazing and feedlot cattle. Journal of Animal Science

77: 1392-1401.

Heber, A.J., P.-C. Tao, J.-Q. Ni, T.T. Lim and A.M.

Schmidt. 2005. Air emission from two swine finishing

building with flushing: Ammonia characteristics. In

Fig. 8. Distribution of absolute errors on mass balances calculated by the system.

Fig. 9. Daily profile of methane production of a herd of 10

dairy cows.

Volume 50 2008 7.27CANADIAN BIOSYSTEMS ENGINEERING

Proceedings of the Seventh International Symposium onLivestock Environment, ASAE publication number701P0205. St. Joseph, MI: ASAE.

Hoff, S.J., D. Van Utrecht, J.D. Harmon and D.W.Mangold. 2000. A general purpose laboratory forevaluating livestock ventilation systems. Applied Engi-neering in Agriculture 16: 701-710.

Jackson, H.A., R.G. Kinsman, D.I. Masse, J.A. Munroe,F.D. Sauer, N.K. Patni, D.J. Buckley, E. Pattey, R.Desjardins and M. Wolynetz. 1993. Measuringgreenhouse gas emissions in a controlled environmentdairy barn. ASAE Paper No. 94-4521. St. Joseph, MI:ASAE.

Johnson, K. A. and D. E. Johnson. 1995. Methaneemissions from cattle. Journal Animal Science 73:2483-2492.

Kaharabata, S.K., P.H. Schuepp and R. L. Desjardins.2000. Estimating methane emissions from dairy cattlehoused in a barn and feedlot using an atmospherictracer. Environmental Science and Technology 34:3296-3302.

Kinsman, R.F, F.D. Sauer, H.A. Jackson and M.S.Wolynetz. 1995. Methane and carbon dioxide emis-sions from dairy cows in full lactation monitored over asix-month period. Journal of Dairy Science 78: 2760-2766.

Lefcourt, A.M. 2001a. Large environmental chamber:Ammonia recovery calibration. Applied Engineering inAgriculture 17: 683-689.

Lefcourt, A.M., B. Buell and U. Tasch. 2001b. Largeenvironmental chamber: Design and operating char-acteristics. Applied Engineering in Agriculture 17: 691-701.

Mathison, G.W., E.K. Okine, T.A. McAllister, Y. Dong,J. Galbraith and O.I.N. Dmytruk. 1998. Reducingmethane emissions from ruminant animals. Journal ofApplied Animal Research 14: 1-28.

Moss, A.R., J.P. Jouany and J. Newbold. 2000. Methaneproduction by ruminants: Its contribution to globalwarming. Annales de Zootechnie 49: 231-253.

Osada, T., H.B. Rom and P. Dahl. 1998. Continuousmeasurement of nitrous oxide and methane emission in

pig units by infrared photoacoustic detection. Transac-

tions of the American Society of Agricultural Engineers

41: 1109-1114.

Ozcan, S.E., E. Vranken and D. Berckmans. 2007. A

critical comparison of existing ventilation rate

determination techniques for building opening and

prospect for future. In Proceeding of the International

Symposium on Air Quality and Waste Management for

Agriculture. ASABE Publication number 701P0907cd.

St-Joseph, MI: ASABE.

Phillips, V.R., M.R. Holden, R.W. Sneatha, J.L. Short,

R.P. White, J. Hartung, J. Seedorf, M. Schroder, K.H.

Linkert, S. Pedersen, H. Takai, J.O. Johnsen, P.W.G.

Groot Koerkamp, G.H. Uenk, R. Scholtens, J.H.M.

Metz and C M. Wathes. 1998. The development of

robust methods for measuring concentrations and

emission rates of gaseous and particulate air pollutants

in livestock buildings. Journal of Agricultural Engineer-

ing Research 70: 11-24.

Powell, J.M., P.R. Cusick, T.H. Misselbrook and J.B.

Holmes. 2007. Design and calibration of chambers for

measuring emissions from tie-stall dairy barns. Trans-

action of the ASAEB 50: 1045-1051.

Shabi Z., I. Bruckental, S. Zamwell, H. Tagari and A.

Arieli. 1999. Effects of extrusion of grain and feeding

frequency on rumen fermentation, nutrient digestibil-

ity, and milk yield and composition in dairy cows.

Journal of Dairy Science 82: 1252-1260.

Singh, A., T.L. Richard and V. Stout. 2003. Methane and

nitrous oxide emissions from swine hoop structures

using a tracer gas technique. In Proceeding of the Third

Internal Symposium on Air Pollution from Agricultural

Operation III, ASAE publication number 701P1403,

St-Joseph, MI: ASAE.

Sutton J.D., I.C. Hart, W.H. Broster, R.J. Elliot and E.

Schuller. 1986. Feeding frequency for lactating cows:

Effect on rumen fermentation and blood metabolites

and hormones. British Journal of Nutrition 56: 181-192.

Zhang, Y., X. Wang, G.L. Riskowski and L.L. Christian-

son. 2001. Quantifying ventilation effectiveness for air

quality control. Transactions of the ASAE 44: 385-390.

7.28 TREMBLAY and MASSELE GENIE DES BIOSYSTEMES AU CANADA