DOI 10.1007/s10661-011-2397-8 Methane emission from naturally ventilated livestock buildings can be determined from gas concentration measurements Bjarne Bjerg & Guoqiang Zhang & Jørgen Madsen & Hans B. Rom Received: 19 November 2010 / Accepted: 4 October 2011 # Springer Science+Business Media B.V. 2011 Abstract Determination of emission of contaminant gases as ammonia, methane, or laughing gas from natural ventilated livestock buildings with large opening is a challenge due to the large variations in gas concentration and air velocity in the openings. The close relation between calculated animal heat production and the carbon dioxide production from the animals have in several cases been utilized for estimation of the ventilation air exchange rate for the estimation of ammonia and greenhouse gas emissions. Using this method, the problem of the complicated air velocity and concentration distribution in the openings is avoided; however, there are still some important issues remained unanswered: (1) the precision of the estimations, (2) the requirement for the length of measuring periods, and (3) the required measuring point number and location. The purpose of this work was to investigate how estimated average gas emission and the precision of the estimation are influenced by different calculation procedures, measuring period length, measure point locations, measure point numbers, and criteria for excluding B. Bjerg (*) : J. Madsen Department of Large Animal Sciences, University of Copenhagen, Groennegaardsvej 2, 1870 Frederiksberg C, Denmark e-mail: bb@life.ku.dk G. Zhang : H. B. Rom Department of Engineering, Aarhus University, Aarhus, Denmark measuring data. The analyses were based on existing data from a 6-day measuring period in a naturally ventilated, 150 milking cow building. The results showed that the methane emission can be determined with much higher precision than ammonia or laughing gas emissions, and, for methane, relatively precise estimations can be based on measure periods as short as 3 h. This result makes it feasible to investigate the influence of feed composition on methane emission in a relative large number of operating cattle buildings and consequently it can support a development towards reduced greenhouse gas emission from cattle production. Keywords Methane. Ammonia. Laughing gas. Carbon dioxide. Emission. Naturally ventilated buildings Introduction Determination of emission of contaminant gases as ammonia, methane, and laughing gas from natural ventilated livestock buildings with large opening is a challenge due to the large variations in gas concentration and air velocity in the openings. The close relation between calculated animal heat production (CIGR 2002) and the carbon dioxide production from the animals (Pedersen et al. 2008) have, in several cases, been utilized in the estimation of the emission of ammonia and other gases (Zhang et al. 2005;
Ngwabie et al. 2009). The calculations in published articles have used the average concentration of carbon dioxide in the livestock building over a specific time span, but the precision may potentially be increased considerably by using the method proposed by Madsen et al. (2010), where the relation between the concentration of the gases in question, e.g., methane and carbon dioxide, is calculated on the individual air samples instead of using mean values for carbon dioxide concentrations. Using this method, the problem of the complicated air velocity and concentration distribution in the openings is avoided, but still there is considerable doubt about (1) the precision of the estimations, (2) the requirement for the length of measuring periods, and (3) the required measuring point number and location. The purpose of the work was to investigate how estimated average gas emission and the precision of the estimation are influenced by different calculation procedures, different measuring period length, different measure point locations, different measure point numbers, and different criteria for excluding s. Methods The analyses in this work were based on gas concentration data recorded for 6 days in a naturally ventilated milk production building (see Fig. 1), which were included in a larger investigation by Zhang et al. (2005). The building housed 150 cows and was equipped with cubicles in resting area. Aisle areas consisted of slatted floor above 40 cm deep Fig. 1 Building layout and sensor positions slurry channels with scrapers. The manure was scraped to a transverse channel wherefrom it was pumped to a storage tank. The roof and walls were uninsulated and the ventilations system consisted of an open ridge and large openings in side walls. Gas concentrations were measured with an INNOVA gas monitors 1312 and a multiplexer 1303 (INNOVA 2002). Sampling air was drawn from the measurement points to the centrally placed gas monitor. Each air-sampling channel had a pump in Teflon with flow capacity of 7 l/min with a filter and a 20-m sampling tube of 6 mm of internal diameter. A three-way connection was applied at each channel to the multiplexer of the gas monitor to ensure the excessive sampled air could exit via the third way and thus avoid overloading of the internal pump in the gas monitor. Air temperatures were measured with Testo 174 temperaure sensors with built-in data loggers (Testo 2011). Further details on the measurements are given by Zhang et al. (2005), where the building included in this work was named building 5. The investigated data contain measured air temperature and measured carbon dioxide, ammonia, methane, and laughing gas concentration at five locations inside the building and one location outside the building (see Fig. 1). Measure point 1 was located in the middle of the building about 1 m below the ridge opening and two measure points were located 0.25 m from the side wall openings and 0.25 m below the eaves with 20 m distance at each side wall. Air temperatures were recorded every 2 min, and Table 1 shows measured daily air temperature at each measuring point. It appears that the average outdoor
Table 1 Average value and standard deviation for measured indoor and outdoor temperature at different measuring days Location of measuring point appears from Fig. 1 Day Indoor temperature, C Outdoor temperature, C Point 1 Point 2 Point 3 Point 4 Point 5 Point 6 1 11.5±1.2 10.2±1.1 10.0±1.0 10.6±1.4 10.8±1.5 9.9±1.6 2 13.4±1.5 12.4±1.3 12.0±1.1 12.2±1.4 12.1±1.3 12.1±1.2 3 11.6±1.5 10.1±1.2 10.0±2.1 10.5±1.5 10.9±1.6 10.2±1.6 4 11.9±2.6 10.2±2.0 9.9±2.1 11.2±2.8 11.7±2.8 9.5±4.0 5 9.2±2.6 7.3±2.0 7.2±2.3 7.7±2.1 8.6±2.6 5.5±4.5 6 8.3±3.4 6.4±3.0 5.9±3.4 7.4±3.3 8.2±3.3 4.6±5.3 All 11.0±2.9 9.4±2.8 9.2±2.9 9.9±2.8 10.4±2.8 8.6±4.3 temperature during the whole period was 8.6 C, and that the temperatures decreased in the last part of the period. Parallel with the decreased temperature level, the difference between day and night outdoor temperatures increased from around 6 K in the beginning period to around 14 K at days 5 and 6. The indoor temperatures were highest at the measure point (1) in the middle of the building, and at that point, the daily average temperature was between 1.3 and 3.7 K above the outdoor level. The temperature difference between inside and outside was larger at days 5 and 6 than at the first 4 days. Gas concentrations were measured with a frequency of nine to ten measurements per hour, and Tables 2, 3, 4, and 5 show the daily average gas concentration in each measure point for carbon dioxide, ammonia, methane, and laughing gas, respectively. The tables show that the concentration generally increased from day 4 to day 5 and it is assed that both increase concentrations and the simultaneous increase in temperature difference between indoor and outdoor most likely can be explained with a possible decrease in outdoor wind speed even though this parameter was not recorded. Figures 2, 3, and 4 show the relation between measured values for carbon dioxide, ammonia, methane, or laughing gas concentrations for all measurements in all points. The determination of ammonia, methane, and laughing gas emission can be based on an assumed strong correlation between carbon dioxide production and calculated animal heat production (Pedersen et al. 2008). Calculation of total heat production from livestock is generally expressed in heat production unit (HPU), where one HPU is defined as a total heat production of 1,000 W at 20 C. A discussion of the exact value of the expected carbon dioxide production is not an objective of this work and consequently the value of 0.185 m 3 HPU 1 h 1, stated by CIGR (2002), is used in all calculations in this work. Using this value, air change per HPU can be calculated as: 0:185 Q HPU ¼ ð1þ c CO2 ; outlet c CO2 ; inlet 10 6 Table 2 Average value and standard deviation for measured indoor and outdoor carbon dioxide concentration at different measuring days Location of measuring point appears from Fig. 1 Day Indoor carbon dioxide concentration, ppm Outdoor carbon dioxide concentration, ppm Point 1 Point 2 Point 3 Point 4 Point 5 Point 6 1 558±104 469±165 430±115 449±77 501±107 387±12 2 566±52 487±147 443±107 404±39 388±12 381±4 3 546±35 379±20 376±7 426±61 485±101 383±8 4 696±147 425±103 406±58 614±175 722±173 404±14 5 771±92 521±174 504±127 611±156 725±165 419±19 6 763±87 442±91 432±76 716±117 764±130 418±16 All 650±135 454±136 432±98 537±163 598±192 399±21
Table 3 Average value and standard deviation for measured indoor and outdoor ammonia concentration at different measuring days Location of measuring point appears from Fig. 1 Day Indoor ammonia concentration, ppm Outdoor ammonia, concentration, ppm Point 1 Point 2 Point 3 Point 4 Point 5 Point 6 1 1.2±0.5 1.0±0.8 0.8±0.6 0.9±0.4 1.1±0.5 0.6±0.2 2 1.0±0.3 0.9±0.7 0.8±0.6 0.6±0.3 0.5±0.2 0.4±0.2 3 2.4±0.2 0.6±0.1 0.5±0.1 0.6±0.3 0.8±0.4 1.1±0.2 4 2.8±1.0 1.2±0.6 0.9±0.3 1.9±1.1 2.5±1.2 1.1±0.3 5 2.5±0.7 2.0±1.0 1.7±0.9 2.1±0.6 2.8±0.8 1.1±0.2 6 1.8±0.51 1.2±0.4 1.0±0.6 2.3±1.4 2.8±0.8 1.1±0.2 All 1.8±1.0 1.4±0.8 1.0±0.7 1.4±1.0 1.8±1.2 0.8±0.4 where Q HPU is air change, in cubic meter per hour per heat production unit, and c CO2 is the concentration of CO 2, in parts per million. Production of ammonia, methane, or laughing gas can be calculated as: P gas ¼ 0:185 c gas; outlet c gas; inlet ð2þ c CO2 ; outlet c CO2 ; inlet where P gas is the production of ammonia, methane, or laughing gas, in cubic meter per hour per heat production unit, and c CO2 is concentration of CO 2,in parts per million. Zhang et al. (2005)assumed that the average of the five indoor located measure points were representative for concentration in outlet. In this work, how estimated emission is influenced by using one measure point only and by the different location of measure point is investigated. Zhang et al. (2005) also used daily average values to calculate standard deviations of each data set. In this work, how different periods for time averaging of concentrations influence the precision of the estimated emissions, expressed as interval of confidence, are investigated. The accuracy of carbon dioxide measurements has a large influence on estimated gas production if the difference in carbon dioxide concentration in outlet and inlet is small. In this work, is generally excluded if this difference is less than 6 ppm, corresponding to specified accuracy of the instrument, and how larger thresholds influence the estimated emission are investigated. Zhang et al. (2005) assumed that the outdoor measure point was representative for concentrations in inlets. To simplify measuring procedure, it could be considered to use fixed values for outdoor concentration based on measurement from a period before or after the indoor measurement. This simplification makes it possible to conduct measurements with one single channel measuring device only. To simulate that possibility, the average concentrations in the outdoor measurements at days 1 and 6 were used as fixed values for inlet concentration (404 ppm CO 2, 0.91 ppm NH 3, 0.96 ppm CH 3, and 0.45 ppm N 2 O), and the used indoor was restricted to Table 4 Average value and standard deviation for measured indoor and outdoor methane concentration at different measuring days Location of measuring point appears from Fig. 1 Day Indoor methane concentration, ppm Outdoor methane concentration, ppm Point 1 Point 2 Point 3 Point 4 Point 5 Point 6 1 14.6±8.6 7.6±13.7 4.3±9.2 5.6±6.0 10.0±8.5 0.7±0.9 2 15.8±4.1 9.1±11.7 5.5±8.3 2.8±2.9 1.5±1.1 0.8±0.5 3 13.2±2.9 0.2±1.5 0.1±0.5 2.2±3.1 8.8±8.3 0.3±0.5 4 24.2±11.4 3.0±8.1 1.6±4.3 17.8±13.7 26.5±13.8 1.0±1.1 5 29.5±6.1 9.5±12.7 8.3±9.5 16.6±11.6 26.2±12.1 1.2±1.5 6 30.0±7.2 3.8±7.4 3.4±6.2 26.5±9.6 30.5±11.0 1.4±2.6 All 21.2±10.1 5.5±10.6 3.9±7.4 12.1±12.5 17.2±14.7 0.9±1.2
Table 5 Average value and standard deviation for measured indoor and outdoor laughing gas concentration at different measuring days Location of measuring point appears from Fig. 1 Day Indoor laughing gas concentration, ppm Outdoor laughing gas concentration, ppm Point 1 Point 2 Point 3 Point 4 Point 5 Point 6 1 0.45±0.03 0.44±0.04 0.43±0.03 0.43±0.03 0.44±0.04 0.42±0.02 2 0.44±0.07 0.41±0.05 0.40±0.03 0.39±0.03 0.40±0.03 0.39±0.03 3 0.46±0.06 0.42±0.03 0.42±0.03 0.43±0.04 0.45±0.10 0.42±0.03 4 0.49±0.11 0.43±0.04 0.43±0.03 0.47±0.07 0.49±0.08 0.44±0.03 5 0.54±0.08 0.49±0.06 0.49±0.07 0.51±0.06 0.53±0.08 0.47±0.04 6 0.55±0.11 0.48±0.05 0.48±0.04 0.54±0.07 0.54±0.08 0.48±0.04 all 0.49±0.09 0.44±0.05 0.44±0.05 0.46±0.07 0.47±0.09 0.44±0.04 measurement at point 1 in the 4-day period in between. A further simplification of the measuring procedure would be to replace the outdoor measurement with outdoor concentration data from local or international environmental monitoring organizations. The quality of this method will naturally be dependent on how well such general data reflect the local conditions in the measuring period and therefore it is not feasible to assess the method on data from one building only, and consequently the method is not included in this work. Another possibility is to use indoor measurement only. In this approach, it is assumed that the inlet concentration of ammonia, methane, or laughing gas is 0 ppm, and for each of those gases, linear regression is used to determine the carbon dioxide concentration that corresponds to that assumption. For the entire measure period, the estimated carbon dioxide concentration corresponding to zero ammonia concentration was 433 ppm (R 2 =0.74), the corresponding value for zero methane concentration was 371 ppm (R 2 =0.97), and for zero laughing gas concentration was 277 ppm (R 2 =0.28). Subsequently, the ammonia, methane, and laughing gas production was calculated as: 0:185c gas P gas ¼ ð3þ c CO2 c CO2 ; 0 Fig. 2 Ammonia concentration as function of carbon dioxide concentration in all measure points
Fig. 3 Methane concentration as function of carbon dioxide concentration in all measure points where P gas is the production of ammonia or methane, in cubic meter per hour per heat production unit, and c CO2 (0 is the concentration of CO 2, where c gas =0 ppm. Zhang et al. (2005) measured in 6-day periods. The influence of reducing the measuring period to 24, 3, or 1 h is investigated in this work. To conduct the analyses on different s and to simulate measurement at a convenient time of the day, it was chosen to let the 3-h periods start at 9 a.m. and the one 1-h periods start at 2 p.m. Fig. 4 Laughing gas concentration as function of carbon dioxide concentration in all measure points
Results and discussion Influence of averaging period Table 6 shows that the interval of confidence for gas emission measurements for the entire 6-days period can be reduced for methane and ammonia by using 1- h averaging periods. An opposite change was seen for laughing gas emission, and further analyses revealed that this was caused by a frequent occurrence of approximately 1-h long periods with low or high emission resulting in relative high standard deviation of the hourly emission data for that gas. For all three gases, the interval of confidence was found to be narrowest if each single measurement was used in the calculation, and thereby considering the composition of each individual sample as a certain portion of air from the cows or from the manure and a certain portion of outside air (Madsen et al. 2010). Consequently, the following calculation is based on each single measurement. Influence of number and location of measure points used for estimation of outlet concentration Table 7 shows that using point 1, as the only point for assumed outlet condition, results in similar methane and laughing gas emission but an about 18% lower ammonia emission than using average value from points 1 to 5. For the points located near to the sidewall openings (two to five), a larger number of s were excluded due to low carbon dioxide concentration difference. Consequently, point 1 is used as assumption for outlet condition in the following calculations. Table 7 also shows that the coefficient of variation for estimated emission between the single measure points was about ten times higher for ammonia than for methane. The used threshold that excluded s where the indoor carbon dioxide concentration was less than 6 ppm above the outdoor level resulted in relatively few s for points 2 and 3, and it is calculated that exclusion of those two points would reduce the coefficient of variation between different measure points for methane emission to less than 0.01. Influence of different criteria for excluding s Generally, in this work, s were excluded if the difference between carbon dioxide concentration in outlet and inlet was less than 6 ppm. This was to obtain a higher concentration of air from the cows or from the manure and thereby minimizing the influence of the composition of the outside air. Table 8 shows that increasing the limit to 100 ppm has no significant influence on calculated emissions of any of the gasses. If the limit is increased to 400 ppm, 85% of the is excluded but the estimated methane emissions are still at the same level, and also for ammonia and laughing gas emissions, the changes are relatively small. Influence of using assumptions of fixed values for inlet condition Using the assumption of fixed values for inlet conditions makes it possible to measure emissions with a single channel instrument. Table 9 shows that using the indoor measurement (point 1) for determining inlet air condition results in a large deviation in estimated ammonia and laughing gas emissions. This is mainly caused by the poor correlation between concentration of these two gases and carbon dioxide concentration, which results in assuming a to high carbon dioxide concentration in the inlet air. For methane, the correlation with carbon dioxide concentration was much better and, consequently, this method resulted in a reduction of estimated methane Table 6 Average value and interval of confidence for estimated ammonia, methane, and laughing gas emission using different averaging periods Length of averaging period, h NH 3 emission, N 2 O emission, 24 6 14.4±2.31 243±4.41 1.46±0.20 1 144 14.4±0.89 243±2.54 1.46±0.26 No averaging 1,270 14.4±0.41 243±1.33 1.46±0.12
Table 7 Average value and interval of confidence for estimated ammonia, methane, and laughing gas emission using different points for assuming outlet conditions Measure points used as outlet conditions NH 3 emission, N 2 O emission, Average of 1 5 1,269 14.4±0.41 243±1.33 1.46±0.20 1 1,288 12.3±0.27 241±0.94 1.51±0.12 2 369 30.4±4.64 260±10.1 1.29±0.87 3 319 30.9±5.24 263±9.37 1.44±0.76 4 998 23.1±3.87 248±3.87 1.47±0.47 5 1,034 19.4±1.44 250±3.39 1.53±0.49 Coefficient of variation between points (n=5) 0.34 0.036 0.064 emission of 7% only. For both methane and ammonia emissions, the deviations were smaller if the fixed values for inlet conditions were estimated from the outdoor measurement (point 6) at days 1 and 6, and the used s for indoor measurements were restricted to the 4 days in between. Influence of using data from reduced measuring periods Table 10 compares estimated emission for each day with the estimation for the entire period. It appears that the relatively deviation in emission estimations between different days was more than ten times larger for ammonia than for methane. The relative large difference in ammonia emission between days could be related to differences in climatic conditions, but in this study, an expected positive correlation between indoor air temperature and ammonia emission was not observed. The outdoor wind conditions might also cause differences in ammonia emission between days, and in the absence of data for outdoor wind condition, it was assumed that low difference between in- and outdoor temperature could be used as an indication of high outdoor wind speed. But the data showed no clear correlation between this temperature difference and the daily ammonia emission. Therefore, the large deviation in estimated ammonia emission between different days is assumed to reflect that the used method is less precise for ammonia than for methane. The small difference between daily estimations of methane emission makes it relevant to investigate even shorter measuring periods for methane. Table 11 shows estimated methane emission based on 3-h periods started at 9 a.m. on each day, and 1-h periods started at 2 p.m. The comparison of results in Tables 10 and 11 shows that reducing the measuring periods from 1 day to 3 h had a very limited influence on estimated methane emission and on the deviation between different days. Table 11 shows that a further decrease of the measure period to 1 h only results in a significantly larger, but still relatively low, variation between days. Table 8 Average value and interval of confidence for estimated ammonia, methane, and laughing gas using different limits for excluding s Observation excluded if c CO2;outlet c CO2;inlet <, (ppm) NH 3 emission, g day -1 HPU -1 g day -1 HPU -1 N 2 O emission, g day -1 HPU -1 6 1,288 12.3±0.27 241±0.94 1.51±0.12 50 1,283 12.3±0.28 241±0.94 1.51±0.12 100 1,227 12.3±0.28 240±0.94 1.53±0.12 200 726 13.2±0.36 241±1.14 1.56±0.15 400 195 13.2±0.36 240±1.77 1.45±0.19 Measurements at point 1 was used as outlet concentrations
Table 9 Average value and interval of confidence for estimated ammonia and methane emission using different assumption for inlet conditions Assumed inlet condition NH 3 emission, N 2 O emission, Point 6 1,288 12.3±0.27 243±4.41 1.46±0.20 Fixed values calculated from point 1 a 1,316 30.3±0.83 225±1.17 10.8±0.21 Fixed values calculated from point 6 b 716 c 11.6±0.60 244±1.78 0.72±0.22 a 0 ppm of ammonia, methane, and laughing gas. Carbon dioxede concentrtation calculated by linear regression to correspond to 0 ppm for each of the three gases, respectively b Average values of meausurements at days 1 and 6 c Measurements from days 2 to 5 Influence of estimating inlet concentration using indoor measurements for 3-h periods The method illustrated in Table 12 is based on data from 3-h periods at point 1 only. By assuming a linear relationship between methane and carbon dioxide concentration, linear regression was used to calculate the carbon dioxide concentrations that correspond to no methane concentration. This carbon dioxide concentration and no methane were subsequently used as assumption for the condition in the inlet air. For the entire period and on days 1, 3, and 4, there was a high correlation between methane and carbon dioxide concentration and, consequently, the estimated methane emission was at the same level as if other methods were used. On days 2, 5, and 6, large discrepancies were found and the analyses did not reveal special circumstances that could explain the poor methane emission determination on these days, but from Table 12, it appears that discrepancies are associated with overestimation of the carbon dioxide concentration that correspond to no methane content. Influence of using fixed value assumption based on outdoor measurements for 3-h periods For 3-h measuring periods, Table 13 shows the influence of using fixed values for inlet condition based on outdoor measurement in point 6 from 9 a.m. to 10 a.m. Due to the 1 h spent on outdoor measurement, the utilized indoor s were restricted to measurement in point 1 in the 2-h period from 10 a.m. to 12 a.m. The comparison of Tables 11 and 13 shows that the coefficient of variation between estimated daily methane emissions increase nearly four times, if the first hour of a 3-h measuring periods is used for determining inlet concentration, compared with simultaneous indoor and outdoor measurements. But, Table 10 Average value and interval of confidence for estimated ammonia and methane emission at different days Measurements at point 1 was used as outlet concentrations Day NH 3 emission, N 2 O emission, 1 6 1,288 12.3±0.27 241±0.94 1.46±0.20 1 193 10.5±0.50 242±1.83 1.34±0.21 2 220 10.6±0.75 242±3.17 1.67±0.37 3 219 10.3±0.50 238±2.25 1.54±0.42 4 218 13.6±0.60 236±1.91 1.39±0.31 5 219 15.7±0.82 241±2.07 1.55±0.22 6 219 13.0±0.47 246±1.98 1.54±0.28 Coefficient of variation between days (n=6) 0.17 0.013 0.050
Table 11 Average value and interval of confidence for estimated methane emission at different days at a 3-hour period beginning at 9 a.m. on each day and a 1-h period beginning at 2 pm Measurements at point 1 was used as outlet concentrations Day 3-h periods beginning at 9 a.m. 1-h periods beginning at 2 p.m. 1 6 139 243±3.60 55 243±5.40 1 3 238±32.7 10 244±20.3 2 28 242±8.88 9 266±12.1 3 27 246±1.42 9 232±16.1 4 27 242±4.97 9 229±9.87 5 27 247±9.35 9 246±9.05 6 27 236±10.7 9 241±7.29 Coefficient of variation between days 6 0.017 6 0.054 nevertheless, the level of coefficient of variation between days is still relatively low. Influence of diurnal variation Estimated diurnal variation in the emission of ammonia and methane is shown in Fig. 5. These diurnal fluctuations are likely to be related to the activities in the livestock building. For ammonia, it appears that the largest deviation occurs before noon where the emission is 45% above average. This peak is likely related to a high frequency of animals that urinate at that part of the day. For methane, the fluctuations are much less pronounced and the largest deviation from the average level is found early in the morning, where it is up to 9% below the average level. The diurnal variation in methane emission is undoubtedly connected to the animals eating activity, but the exact relationship cannot be interpreted in this study because no data are included that allow the determination of a possible correlation between diurnal variation in methane and carbon dioxide emission. Kinsman et al. (1995) reported that the diurnal variation in methane and carbon dioxide emissions from a mechanical ventilated dairy cow building were close connected and the fluctuation in carbon dioxide emission varied between approximately 12% below and above the average level. For methane, the fluctuation was nearly twice as high and that suggests that a coincident diurnal variation in the emission of methane and carbon dioxide will contribute to the relatively small diurnal variation in estimated methane emissions if the estimation is based on an assumed constant carbon dioxide production, as it is in this work. Even though the diurnal variation in estimated methane emissions is relative small, it will constitute Table 12 Average value and interval of confidence for estimated methane emission at different days at a 3-h period beginning at 9 a.m., using data from point 1 to calculate inlet concentrations Day CO 2 concentration where CH 4 concentration is 0, ppm R 2 1 6 371 0.97 158 237±5.40 1 375 0.99 27 236±29.7 2 410 0.75 27 307±12.1 3 376 0.87 27 243±5.60 4 392 0.98 27 255±4.01 5 640 0.25 27 678±49.4 6 496 0.78 27 327±6.8 Coefficient of variation between days 6 0.49
Table 13 Average value and interval of confidence for estimated methane emission at different days using 3- h periods beginning at 9 a.m. on each day and using the first hour to determine fixed values for inlet conditions Day a certain source of error if the measuring period is reduced to less than 24 h, e.g., the estimated methane emission would deviate up to 7% from average level if any 3-h continuous period was selected from the data used in Fig. 5. For 12-h periods, the largest deviations from the average level would be around 2%. Conclusions 1 6 97 256±5.51 1 18 244±8.77 2 18 253±6.06 3 18 246±4.11 4 18 273±13.7 5 18 233±12.2 6 6 275±17.3 Coefficient of variation between days 6 0.066 The analyses based on gas concentration data from a 6-day measuring period in a naturally ventilated, building for milking cows show that methane emission can be determined with much higher precision than ammonia or laughing gas emission. Ammonia, methane, and laughing gas emissions were determined with significant higher precision if each single measurement was used instead of average values for 24-h measuring periods. Estimated daily methane emission was not significantly influenced: & & & & By reducing the number of measure point for determination of outlet concentrations from 5 to 1; By the location of measure point for determination of outlet concentration, if locations close to inlet were avoided; By excluding where the difference between carbon dioxide concentration in outlet and inlet was below different thresholds in the interval between 6 and 400 ppm, and By reducing the measure period from 6 days to 1 day. Even shorter measure periods than 1 day can be considered but may result in a certain deviation from daily average due to the diurnal variation. Measuring procedures can be simplified by using fixed values for assumed inlet concentration. If suitable values are used, the influence on estimated gas emission is limited, and in this way, it becomes possible to conduct measurement with a single one channel measuring system. Fixed values for inlet concentrations can be based on outdoor or indoor measurement. If indoor measurement is used for the determination of inlet conditions, the quality of the estimation is highly dependent on the correlation between the concentration of carbon dioxide and the concerned gas. The relatively moderate measuring requirements for methane emission estimations make it feasible to investigate the influence of feed composition in a relative large number of operating cattle buildings, and consequently, it can support a development Fig. 5 Diurnal variation in estimated ammonia and methane emissions. Error bars show interval of confidence calculated from 46 to 55 single measurements for each 1-h interval
towards reduced greenhouse gas emission from cattle production. For ammonia or laughing gas, the analyses did not reveal any shortcuts to precise and simple determination of emission from naturally ventilated buildings. References CIGR. (2002). 4th Report of Working Group on Climatization of Animal Houses. CIGR. International Commission of Agricultural Engineering, Section 2. ISBN 87-88976-60-2. 45 pp INNOVA. (2002). Technical descriptions of gas monitor 1312 and multipoint sampler 1303. Denmark: INNOVA AirTech Instruments A/S. Kinsman, R., Sauer, F. D., Jackson, H. A., & Wolynetz, M. S. (1995). Methane and carbon dioxide emissions from dairy cows in full lactation monitored over a six-month period. Journal of Dairy Science, 78(12), 2760 2766. Madsen, J., Bjerg, B. S., Hvelplund, T., Weisbjerg, M. R., & Lund, P. (2010). Methane and carbon dioxide ratio in excreted air for quantification of the methane production from ruminants. Livestock Science, 129(1 3), 223 227. doi:10.1016/j.livsci.2010.01.001. Ngwabie, N. M., Jeppsson, K.-H., Nimmermark, S., Swensson, C., & Gustafsson, G. (2009). Multi-location measurements of greenhouse gasses and emission rates of methane and ammonia from a naturally-ventilated barn for dairy cows. Biosystems Engineering, 103(1), 68 77. Pedersen, S., Blanes-Vidal, V., Jørgensen, H., Chwalibog, A., Haeussermann, A., Heetkamp, M. J. W., & Aarnink, A. J. A. (2008). Carbon dioxide production in animal houses: a literature review. Agricultural Engineering International: CIGR Ejournal Manuscript BC 08 008, Vol. X. December, 2008. Testo. (2011). Testo 174 mini temperature datalogger. Testo AG, Germany http://www.testosites.de/datalogger2011/ en_uk/#/174t Zhang, G., Strøm, J. S., Li, B., Rom, H. B., Morsing, S., Dahl, P., et al. (2005). Emission of ammonia and other contaminant gases from naturally ventilated dairy cattle buildings. Biosystems Engineering, 92(3), 355 364.