Dynamic model for ammonia volatilization in housing with partially slatted floors, for fattening pigs

Livestock Production Science 53 Ž1998. 153–169

Dynamic model for ammonia volatilization in housing with partially slatted floors, for fattening pigs A.J.A. Aarnink ) , A. Elzing DLO–Institute of Agricultural and EnÕironmental Engineering, P.O. Box 43, NL-6700 AA Wageningen, Netherlands Received 20 March 1997; accepted 13 October 1997

Abstract A dynamic model was developed to simulate the ammonia volatilization from pig housing with partially slatted floors, where no litter is used. Simulated ammonia emission levels were compared with measured levels for 1 day in each 3-week period during two fattening periods of 15 weeks Žone in winter and one in summer. in each of three fattening compartments. The overall mean ammonia emission measured was 6.84 g dy1 pigy1, while the mean overall simulated emission was 6.36 g dy1 pigy1, with 1.96 g dy1 pigy1 simulated to volatilize from the floor and 4.40 g dy1 pigy1 from the slurry pit. The mean relative difference between the daily simulated and measured ammonia emissions was 16.9%. This was 15.0% at the low and moderate emission levels Ž- 9 g dy1 pigy1 . and 27.9% at the high emission levels Ž) 9 g dy1 pigy1 .. Simulated effects of different factors corresponded satisfactorily with the measured effects reported in the literature. It is concluded that the ammonia emission from housing for fattening pigs in which the floors are partially slatted can be reasonably well predicted at the low and moderate levels of emission, but is poorly predicted at high emission levels when the solid pen floor is severely fouled at high ambient temperatures. q 1998 Elsevier Science B.V. Keywords: Pigs; Ammonia; Model; Housing

1. Introduction Ammonia is one of the main factors determining the air quality inside pig houses. Its concentration in these buildings should be at a low level, for reasons of animal and human health ŽDrummond et al., 1980; Preller, 1995.. Furthermore, the drastic increase in livestock production in recent decades in parts of Europe has led to concern about the detrimental effects on the environment of ammonia emitted from

)

Corresponding author. Tel.: q31-317-476554; fax: q31-317425670; e-mail: [email protected]

these production systems ŽSoveri, 1992; Fangmeier et al., 1994.. In pig houses ammonia volatilizes mainly from urine-fouled floor areas and from the surface area of slurry in the slurry pit. The ratio between the emission from both sources depends on the fouling of the solid floor, the ratio of slatted to solid floor area and the type of slatted floor ŽAarnink et al., 1996.. Total emission levels are related to pH ŽStevens et al., 1989., ammoniacal nitrogen content ŽElzing and Kroodsma, 1993., emitting surface area ŽHartung and Buscher, 1995; Aarnink et al., 1996., air temper¨ Ž ature Muck and Richards, 1983. and air velocity ŽOlesen and Sommer, 1993; Zhang et al., 1994..

0301-6226r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 0 3 0 1 - 6 2 2 6 Ž 9 7 . 0 0 1 5 3 - X

154

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

Most of the ammoniacal nitrogen is formed from urea present in urine ŽAarnink et al., 1993.. The rate at which the urea is converted depends on the urease activity ŽMuck and Steenhuis, 1981.. The effects of the various factors involved in the volatilization of ammonia can be described in quantitative relationships. A model comprising these relationships can be used to estimate and evaluate the effects of single and combined measures to reduce the ammonia release in pig houses. Ammonia volatilization from stored pig slurry has been modelled by Olesen and Sommer Ž1993. and by Zhang et al. Ž1994.. However, these static models are less suitable for the dynamic situation of a pig house, with frequent addition of fresh urine and faeces on the floor and in the slurry pit. The excreting behaviour of the animals is important in this respect. Anderson et al. Ž1987. have also suggested that the ammonia volatilization from the floor might have been an important factor causing their model to underestimate the ammonia release. The basis of our model was described and validated by Elzing and Monteny Ž1997. for a model system of a cow house. The objective of our study was to extend that basic model to the dynamic situation of a house for fattening pigs. We validated the model with an independent data set from a real house for fatteners and performed a sensitivity analysis of the different factors influencing ammonia emission.

Ammonia is very soluble in water, therefore the velocity of transfer is mainly determined by diffusion from the gas boundary layer to the atmosphere ŽHashimoto, 1972.. For urine puddles on the floor and the slurry in the pit the ammonia volatilization can be described as follows ŽOlesen and Sommer, 1993.: EXNH 3 s k P Ž Cg y Ca . EXNH 3

where is the ammonia volatilization per m of the ammonia solution Žmol my2 sy1 .; k is the mass transfer coefficient Žm sy1 .; Cg is the ammonia concentration in the gas boundary layer Žmol my3 . and Ca is the ammonia concentration in the atmosphere. Ammonia concentration in the atmosphere is low compared with the concentration in the gas boundary layer and is neglected in our model. The concentration in the gas boundary layer can be calculated from Henry’s law equation: Cg s

Cl

Ž 2.

H

where C l is the ammonia concentration in the liquid boundary layer Žmol my3 .; H is the Henry constant Ždimensionless.. Ammonia concentration in the liquid boundary layer equals the un-ionized fraction Ž f . of the total ammoniacal nitrogen ŽTAN.. Multiplying EXNH 3 by the area of the urine puddle or the area of the slurry in the pit Ž A. enables the ammonia volatilization from that source to be calculated ŽElzing and Monteny, 1997.:

2. Model description The model simulates the ammonia volatilization in a partially slatted house for fattening pigs, where no litter is used. The schematic diagram in Fig. 1 shows the various components and the variables in the model. The main components of the model are the urine puddles on the floor and the slurry in the pit. These are the main ammonia sources in a pig pen. Ammonia volatilization from faeces on the floor is neglected. Volatilization per pen is calculated by summating the emission from each puddle on the floor and adding it to the emission from the slurry in the pit. Ammonia emission from a pig house is calculated by summating the emissions from each pen. Ammonia emission is a process of mass transfer from the ammonia solution to the free atmosphere.

Ž 1. 2

E NH 3 s

k P A P f P w TAN x H

Ž 3.

2.1. Henry constant (H) On the basis of results from different investigations, Hashimoto Ž1972. found the following empirical relationship between the Henry constant, expressed in g cmy3 atmy1 , and the temperature: H s a1 P 1.053Ž293yT .

Ž 4.

where a1 is a regression coefficient and was estimated by Hashimoto Ž1972. to be 1.013; T is the temperature of the emitting surface ŽK.. Elzing and Monteny Ž1997. converted this relationship into a dimensionless Henry constant, by using the ideal gas

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

law at 283 K, obtaining an a1 of 1384. In our model it was converted at a more relevant temperature for pig houses, 293 K, giving an a1 of 1431. 2.2. Mass transfer coefficient (k) From an aqueous ammonia solution, Haslam et al. Ž1924. found the following empirical relationship between the mass transfer coefficient, expressed in g hy1 cmy1 atmy1 , and the air velocity Ž n . and temperature ŽT .: k s a2 P Ž n .

0 .8

P Ty1.4

Ž 5.

where a 2 is a regression coefficient and was estimated by Haslam et al. Ž1924. to be 1.62 = 10 4 . The relationship of the mass transfer coefficient to the 0.8 power of the gas velocity is very plausible. This relationship is generally found for mass or heat transfer. The negative power of the temperature can be explained by the increasing viscosity of the gas at higher temperatures, which is inversely related to

155

Fig. 1. Schematic diagram showing the components and variables of the model. The driving variables in the model are: T, n , pH, TAN, I; the state variables are: A, V U , D, U0 , r; the auxiliary variables are: f, C l , Cg , pH e , ENH3 ; the rate variables are: Sm , k; H is a constant. Notation a Regression coefficient A Area of ammonia solution Žm2 . b Constant c Constant Ca Ammonia concentration in the atmosphere Žmol my3 . Cg Ammonia concentration in the gas boundary layer Žmol my3 . Cl Ammonia concentration in the liquid boundary layer Žmol my3 . d Depth of urine puddle Žm. D Distance between urination site on solid floor and slurry pit Žm. ENH 3 Ammonia volatilization from the ammonia solution Žmol sy1 . X ENH 3 Ammonia volatilization from the ammonia solution Žmol sy1 my2 . f Un-ionized ammonia fraction in solution Ždimensionless. H Henry’s constant Ždimensionless. I Urinating interval Žs. k Mass transfer coefficient Žm sy1 . Km Michaelis constant Žmol my3 . L Urination site r Number of urinations in ‘fouling area’ relative to total urinations R Mean roughness of the floor surface Ž m m. Sm Maximum conversion rate of urea at high urea concentrations Žmol my3 sy1 . T Temperature of the emitting surface ŽK. TAN Total ammoniacal nitrogen Žmol my3 . U0 Urea concentration in excreted urine Žmol my3 . Ut Urea concentration at time t after excretion Žmol my3 . n Air velocity Žm sy1 . VU Volume of urine excreted per urination Žm3 . V Ventilation rate Žm3 sy1 my2 .

a e F g l m mix P S so t U

Subscripts Atmosphere Effective Floor Gas Liquid Maximum Mixed slurry Slurry pit Slurry Solid floor Time Urine puddle

156

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

mass transfer ŽHaslam et al., 1924.. To express k in m sy1 , Eq. Ž5. was converted by using the ideal gas law at 293 K giving an a 2 of 50.1. The air velocities above the floor and above the slurry in the pit were assumed to be linearly related to the ventilation rate per m2 floor area of the compartment:

n s a3 P V q b

Ž 6.

where n is the air velocity above the floor or in the slurry pit Žm sy1 .; a3 is a regression coefficient; V is the ventilation rate Žm3 sy1 per m2 floor area. and b is a constant. Randall et al. Ž1983. measured the air velocities near to the pigs for different ventilation systems at different ventilation rates. In modern pig houses ventilation systems are designed in such a way that air velocities near to the pigs are low. The air velocities measured in Randall’s system ‘A’ are most comparable with those measured in modern pig houses. Up to 3.9 = 10y2 m3 sy1 per m2 floor area Žequivalent to approximately 120 m3 hy1 per pig., in their system ‘A’, an almost linear relationship was found between the ventilation rate and air velocity near to the pigs. The regression coefficient a 3 was estimated to be 4.62 Žs.e., 0.30. and the constant b 0.106 Žs.e., 0.007.. For the slurry pit a minimum air velocity above the slurry was assumed to be 0.02 m sy1 Žs b in Eq. Ž6... On the basis of the statistically estimated regression coefficient of the ammonia emission on ventilation rate ŽAarnink et al., 1996., a change of 21.2% in ammonia emission could be calculated when the ventilation rate changes by 0.01 m3 sy1 per m2 floor area. This effect was also calculated by our model when the regression coefficient was fixed at 2.3 Žs a 3 in Eq. Ž6... 2.3. Emitting area (A) The emitting area constitutes the surface area of the slurry in the pit or the area of a urine puddle. The total number of emitting urine puddles at time t is determined by the urinating intervals Ž I ., and the locations of urination ŽL.. When fresh urine is deposited in the same place as a former urination, the old puddle is completely replaced by the new one. To save computer memory a urine puddle stops emitting when the puddle is older than 10 h and the volatilization is less than 10y8 mol sy1 . The loca-

tions of urinations are Poison distributed over space by the random generator of the computer. The intervals between urinations are input into the model. The number of urinating locations is calculated by dividing the total area potentially used for urination by the area of one urine puddle. Pigs deposit most of the urine and faeces on a small part of the pen floor. Observations showed that this ‘excreting area’ is an area of approximately 1.75 m2 on the pen floor ŽAarnink et al., 1997., generally the slatted floor. A fouling factor Ž r . is introduced between 0 and 1, which gives the number of urinations deposited outside the ‘excreting area’ relative to the total number of urinations. This fouling factor is an input in the model. From visual observations it is arbitrarily assumed that half of the pen floor area outside the ‘excreting area’ is ‘fouling area’. The area of a urine puddle on the slatted floor depends mainly on the slatted floor type and on animal weight. This area was measured on different types of slatted floors at different moments during the fattening period ŽAarnink et al., 1997.. The results are given in Table 1. The puddle area on the solid floor was expected to be related to the volume urinated and with the Table 1 Area of a urine puddle on the slats depending on animal weight and slatted floor type Žcalculated from Aarnink et al., 1997. and the depth of the urine puddle on the different slatted floors Weight a

30 45 60 80 100

Area urine puddleb Žm2 . S1

S2

S3

S4

S5

0.072 0.082 0.092 0.101 0.111

0.046 0.056 0.065 0.075 0.085

0.035 0.045 0.054 0.054 0.073

0.013 0.023 0.032 0.042 0.052

0.017 0.026 0.036 0.046 0.055

(mm) 0.69 1.61

0.56 1.66

0.61 2.25

0.79 2.00

Mean depth urine puddle c Clean slats d 0.58 Fouled slats e 1.36 a

Estimation based on mean initial and final weights of the pigs and average growth Ždata from Aarnink et al., 1997.. b Wetted area excluding the gaps; s.e.s81; S1s concrete, 10 cm slats, 2 cm gaps; S2 s concrete, 7 cm slats, 1.8 cm gaps; S3s cast iron, 2.5 cm slats, 1.5 cm gaps; S4 s metal, 1 cm slats, 1 cm gaps; S5s metal with naps, 1 cm slats, 1 cm gaps. c Calculated on the assumption that all urine remained on top of the slats. d s.e.s 0.07. e s.e.s 0.23.

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

distance from the urination site to the slurry pit. This relationship was established experimentally Žsee Section 3.. 2.4. Total ammoniacal nitrogen concentration ([TAN]) The TAN concentration in the liquid boundary layer depends on the rate of ammonia formation, ammonia transport and ammonia volatilization. Ammonia is mainly formed from nitrogen compounds in urine, with urea as the main compound ŽPfeiffer and Henkel, 1991; Nasi, ¨ 1993.. Urea in urine puddles on the floor is converted into two molecules of ammonia and one molecule of carbon dioxide. This conversion process is catalysed by the enzyme urease, which is produced by different microorganisms ŽBremner and Mulvaney, 1978.. The Michaelis– Menten equation is used to describe the rate of conversion ŽMoore, 1972.: dUt dt

s

ySm P Ut

Ž K m q Ut .

Ž 7.

where Ut is the urea concentration of the urine puddle at time t Žmol my3 .; t is time Žs.; Sm is the maximum conversion rate at high urea concentrations Žmol my3 sy1 . and K m is the Michaelis constant Žmol my3 .. The maximum conversion rate at high urea concentrations Sm , also called the urease activity, is dependent on the floor surface characteristics and on the fouling of the floor surface with faeces. For bare concrete floors and concrete floors with different coatings, Braam and Van den Hoorn Ž1996. found the following relationship between the mean roughness of the floor surface Ž R in m m. and the urease activity on the floor Ž Sm, F . expressed in mg NH 3 –N Žmy2 hy1 .: Sm ,F s 2738 y 2665 P 0.989 R

Ž 8.

In this relationship, 68% of the variance in urease activity on the floor could be explained by differences in roughness of the surface of the different floor types. Dividing Sm, F by the depth of the urine puddle enables Sm to be calculated. The Michaelis constant K m equals the urea concentration at which the conversion rate is half of its maximum. The

157

value of 2.0 mol my3 determined by Elzing et al. Ž1992. is used in the model. In general, the amounts of urine added to the slurry in the pit are relatively small, compared with the bulk of slurry already present. Therefore, only a minor urease activity of the slurry is sufficient for a fast conversion of urea. Results from Olesen and Sommer Ž1993. showed that during the first few days after mixing there is only a minor vertical gradient of TAN in slurry stored in a tank Žless than 10% at the lowest wind speed.. After a longer storage period the gradient increases Žto more than 50%, according to results of Zhang et al., 1994.. In the dynamic situation of a pig house, fresh urine is regularly added on top of the slurry. In such a situation TAN is not only formed by anaerobic digestion, as in the research of Zhang et al. Ž1994., but mainly by the conversion of urea from fresh urine. The formation of TAN in the bottom layers as well as in the top layers will prevent a large gradient in the slurry. The gradient in the shallow urine puddle on the floor may even be smaller. Therefore, it is assumed that the ammoniacal nitrogen concentration of the urine puddles on the floor and the slurry in the pit are homogeneous. The change in TAN concentration in the urine puddles with time can be calculated as follows: d w TAN x t dt

s2P

dUt dt

y ENH 3 ,t

Ž 9.

The volume of the urine puddles is calculated from their mean area and mean depth. The mean depth of the urine puddles on the solid floor and on different types of slatted floors was found experimentally Žsee Section 3.. 2.5. Fraction of un-ionized ammonia in solution (f) The fraction of un-ionized ammonia in aqueous solutions can be calculated from the pH and the acid ionization constant for ammonia Ž K a ., which depends on the temperature. Zhang et al. Ž1994. found a K a in pig manure Ž1% total solids. of about one-fifth of the value in aqueous solutions. In chicken manure Ž3.5 to 8.5% total solids., Hashimoto and Ludington Ž1971. found an acid ionization constant

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

158

for ammonia of about one-sixth of the value in a water solution. In a chemical sense, both slurry and urine are considered as concentrated solutions. In our model, the fraction of un-ionized ammonia in the liquid boundary layers of slurry and urine puddles was calculated according to Zhang et al. Ž1994., as follows: fs

10 pH 10 pH q 5 P 10 Ž0 .0897qŽ2729r T ..

Ž 10 .

where T is the temperature of the emitting surface ŽK.. The pH of fresh urine from normally fed fattening pigs has a value of about 7 ŽCanh et al., 1997.. The pH increases rapidly after the conversion of urea has started. The pH was found to be 8.5 when 11% of the urea in urine was converted into ammonia and 9.1 when 95% of the urea was converted ŽElzing and Aarnink, 1996.. A mean pH of 8.8 is used in our model. The mean slurry pH, measured in mixed slurry, is an input variable in the model. However, due to the settling of solids in the slurry and due to different volatilization rates of ammonia and carbon dioxide, the pH of the top layer of the slurry is higher than the pH of the mixed slurry ŽHusted et al., 1991; Olesen and Sommer, 1993.. Therefore an ‘effective’ pH was introduced, which is calculated from the pH of the mixed slurry as follows: pH e s pH s q c

Ž 11 .

where pH e is the ‘effective’ pH of the top layer of the slurry; pH S is the pH of the mixed slurry and c is a constant determined in this study Žsee Section 4.. The urine added to the pit is to some extent buffered by the ‘old’ slurry in the pit. This means that the pH of the top layer of the slurry will not be higher than

the pH of the urine. Therefore, pH e has a maximum of 8.8 in our model.

3. Experiments 3.1. Puddle area solid floor The puddle area on the solid floor was measured by simulating the urination of the pigs. At distances of 1, 2 and 3 m from the slurry pit 0.25, 0.50 and 1.0 l, of urine Žfreshly collected from sows. was poured from a height of 30 cm to a clean and dry, solid concrete floor. The floor had a common slope of 2% and a mean surface roughness of 91.0 m m Žs.e., 4.0. ŽNEN procedure 3632, 1986.. The urine was evenly watered over an area of 10 = 10 cm Žframed by a mobile frame., using a watering can Ž1.0 l content; orifice 6 mm.. Linear regression revealed the following relationship between the area of the urine puddle on the solid floor Ž A U, so , in m2 . versus the volume of urine deposited Ž VU , in m3 . and the distance from the centre of the urinating ‘spot’ to the slurry pit Ž D, in m.: AU ,so s 0.038 q 162 P VU q 0.043 P D Ž R 2 s 0.74 . Ž 12 . 3.2. Depth of urine puddles The mean depth of urine puddles was determined on five types of slatted floors in a laboratory set-up. Half a kg of urine Žfreshly collected from sows. was poured evenly from a height of 30 cm over an area of 10 = 10 cm of clean slats, using a watering-can Ž1

Table 2 The mean floor surface roughness for five types of slats and fora concrete solid floor and the calculated urease activities Ua and Sm Slatted floor a

Roughness Ž m m.

Ua Žmg NH 3 –N my2 hy1 .

Sm Žmmol urea ly1 sy1 .

S1 S2 S3 S4 S5 Solid floor

43.8 Ž3.1. 39.8 Ž3.1. 34.9 Ž6.0. 4.8 Ž0.3. 4.8 Ž0.3. 91.0 Ž4.0.

1078 1022 927 211 211 1747

0.0184 0.0147 0.0164 0.0034 0.0026 0.0615

a

For description of the slatted floors see Table 1.

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

l content; orifice 6 mm.. This was repeated nine times on randomly assigned spots on each slatted floor. The amount of urine poured on the slats and the amount of urine runoff from the slats was measured by weighing ŽMettler balance, max. 60 kg type 60.2r32, error s 0.01 kg.. The whole experiment was repeated twice. After every simulated urination a rectangular area around the top of the slats wetted with urine was measured. The mean depth of the urine puddle was then calculated by dividing the total volume Žcalculated from weightrspecific weight. of the urine on the slats by the total wetted area on top of the slats ŽTable 1.. This experiment was repeated with fouled slats. Fouling was simulated by trapping 5 kg of faeces between the slats of the floors, once a day for four consecutive days. The results are given in Table 1. It was observed that the sides of the slats were partially wet with urine. The sides of the triangular metal slats were wetter and there were more urine drops hanging underneath these slats compared with the concrete and cast iron slats. The depth of the urine puddle on the solid floor was determined on a clean and dry, concrete floor with a common slope of 2% and a mean surface roughness of 91.0 m m Žs.e., 4.0.. Half a litre of urine was spread over the solid floor in such a way that urine did not run into the slurry pit. The area of the floor wetted in this way was measured; the procedure was repeated four times. The depth of the urine puddle was determined by dividing the volume poured by the area wetted. The mean depth of the urine puddle was 0.282 mm Žs.e., 0.012..

159

houses. To do this, simulated ammonia emissions were compared with measured emissions from a real house for fattening pigs. The levels and patterns during the day, during the growing period and during different seasons were simulated on the basis of the following input variables: number of pens, area and type of slatted floor per pen, area of solid floor per pen, temperatures of the floor and of the top layer of the slurry ŽT F and TP in K., ventilation rate Ž V in m3 sy1 per m2 floor area., intervals between urinations Ž I in s., fouling factor Ž r ., urine produced per urination Ž V U in m3 ., initial urea concentration ŽUo in mol my3 ., pH ŽpH S . and TAN concentration of mixed slurry. The material and methods for the collection of the data used in the validation have been extensively described elsewhere ŽAarnink et al., 1997. and will therefore only be outlined here. In the earlier study, the effect of five types of slatted floors on the ammonia emission was determined. Three compartments were used during two fattening periods of 15 weeks each, in which the pigs grew from 26 to 112 kg. The following were measured every 10 min for 1 day in each 3-week period: ammonia concentration of incoming and outgoing air; slurry temperature, measured 1 to 2 cm beneath the surface; ambient temperature, 1 m above the solid floor; ventilation rate. The ambient temperature was assumed to equal the floor temperature. The frequency pigs urinated, during every 10 min period, on the solid and slatted floors was observed from video recordings on tape. After every 3-week period the slurry was removed completely, mixed and sampled. Slurry samples were

3.3. Floor surface roughness The floor surface roughness of five types of slatted floors and of a concrete solid floor was measured according to Dutch standards ŽNEN 3632, 1986.. The results are shown in Table 2. Sm, F was converted into Sm by dividing it by the depth of the urine puddles. 4. Model validation The model was validated with an independent data set to determine its accuracy in predicting the levels and patterns of ammonia volatilization in pig

Fig. 2. Effect of different values of the constant c on the relative difference between simulated and measured ammonia volatilizations. The constant c is added to the pH of the mixed slurry to calculate the effective pH ŽpH e . of the upper layer of the slurry.

160

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

Fig. 3. Measured and simulated ammonia emission during a winter and a summer fattening period. Points give the mean ammonia emission for each day in the validation.

analyzed for NHq 4 –N and pH. Total feed and water intake were measured at the end of every 3-week period. Because of incomplete data sets five out of the 30 measuring days could not be used for the validation. The urine production per pig per day and the urea concentration of fresh urine were calculated with the MESPRO model ŽAarnink et al., 1992., on the basis of the measured feed and water intake and compartment temperature and the estimated animal growth. The urine volume per urination and the urea concentration were assumed to be constant during the day and equal for all the pigs in the compartment. The ‘effective’ pH ŽpH e . was calculated by adding a constant c to the pH of the mixed slurry ŽpH S ; see Eq. Ž11... The constant c was determined by minimizing the relative differences between simulated and measured ammonia volatilizations in a grid analysis with steps of 0.1. The best fit was obtained with: c s 1.1. An example of the effect of different

Fig. 4. Typical measured and simulated diurnal patterns of ammonia emission during Ža. a day in the winter fattening period with moderate levels of ammonia emission and Žb. a day in the summer fattening period with high levels of ammonia emission.

values of c on the relative differences between simulated and measured ammonia volatilizations are given in Fig. 2. Model calculations showed that an iteration interval of a few minutes was suitable to enable the daily pattern of ammonia emission to be predicted accu-

Table 3 Mean input values in the model for the sensitivity analyses Žconverted to commonly used unities., with standard deviations between brackets Variable

Unit

Winter

Summer

Ventilation rate Ambient temperature Slurry temperature pH mixed slurry NHq 4 -N mixed slurry Urinating frequency on slatted floor Urinating frequency on solid floor Urine volume per urination wUreax fresh urine

m3 hy1 pigy1 8C 8C

25.2 Ž0.8. 21.4 Ž0.3. 18.2 Ž0.3. 7.55 Ž0. a 5.61 Ž0. a 3.4 Ž1.9. 0.8 Ž0.6. 0.81 Ž0. a 14.9 Ž0. a

67.1 Ž2.2. 23.0 Ž0.3. 21.9 Ž0.1. 7.33 Ž0. a 5.47 Ž0. a 3.3 Ž2.0. 1.1 Ž0.7. 0.76 Ž0. a 14.0 Ž0. a

a

g kgy1 dy1 pigy1 dy1 pigy1 l g kgy1

These variables did not change during the simulation run.

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

rately by the computer in a reasonable computable time. Ammonia emission rates in mol sy1 were recalculated by the model to give g dy1 pigy1 . The overall mean measured ammonia emission was 6.84 g dy1 pigy1 , while the mean overall simulated emission was 6.36 g dy1 pigy1 , with 1.96 g dy1 pigy1 volatilized from the floor and 4.40 g dy1 pigy1 volatilized from the slurry pit. The relative difference between the 10-min simulations and measurements was 19.1%. The relationship between the daily mean simulated and the daily mean measured ammonia emissions is given in Fig. 3. The mean of the daily simulated ammonia emissions was

161

on average 3.6% lower than the mean of the measured values. This was 1.2% in the winter fattening period and 5.8% in the summer fattening period. The mean relative difference between the daily simulated and measured ammonia emissions was 16.9%. This was 16.5% during the winter fattening period and 17.3% during the summer fattening period. Fig. 3 shows that differences between simulated and measured ammonia emissions became clearly larger at higher levels of emission. On average, the emission levels were underestimated at high levels of ammonia emission. On average, at low and moderate emission levels Ž- 9 g dy1 pigy1 ., the simulated emis-

Table 4 Sensitivity analysis for the different factors affecting the ammonia emission from the floor Factor

Value

Ammonia emission Ž%. Winter

Summer

Floor

Total

Floor

Total

Urea urine Žmol ly1 . a 0.2 s 100%

0.10 0.30

52 143

79 119

52 143

82 116

Air velocity floor Žm sy1 . 0.16 s 100%

0.08 0.24

86 107

93 103

88 106

95 102

83 113

92 106

87 110

95 104

8.6 9.0

89 108

95 104

93 105

97 102

2.5 10.0

98 100

99 100

98 100

99 100

Temperature floor Ž8C. 20 s 100% pH urineb 8.8 s 100% ‘Fouling area’ c Žm2 . 5.0 s 100%

15 25

‘Excreting area’ d Žm2 . 1.75 s 100%

0.88 3.50

89 121

95 110

87 115

94 106

Depth of urine puddle on solid floor e Žmm. 0.40 s 100%

0.20 0.60

89 110

95 105

87 112

94 105

Depth of urine puddle on slatted floor e Žmm. 1.40 s 100%

0.70 2.10

65 126

83 113

64 130

85 112

Urinations in lying area Ž% of total. 25 s 100%

0 50

90 101

95 100

91 99

96 100

Slatted floor type S1 s 100%

S2 S3 S4 S5

91 81 53 45

96 90 77 73

91 81 57 49

96 92 82 79

a

It is assumed that the TAN concentration of the slurry in the pit remained the same. It is assumed that the pH of the slurry remained the same. c The maximum pen area outside the excreting area that can be fouled. d The pen area where pigs usually excrete. e The depth was varied assuming the same puddle area. b

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

162

sions were 1.0% higher and at high levels Ž) 9 g dy1 pigy1 . they were 11.0% lower than the measured values. The mean relative difference between the daily simulated and measured ammonia emissions was 15.0% at the low and moderate emission levels and 27.9% at the high emission levels. As with the daily mean levels, simulated and measured diurnal patterns agreed well at low and moderate levels of ammonia emission Ž- 9 g dy1 pigy1 ., but were different at high levels of emission Ž) 9 g dy1 pigy1 .. In Fig. 4, typical diurnal patterns are shown, one at a moderate emission level and one at a high emission level.

5. Sensitivity analysis The effects of the different variables on the ammonia emission were determined in the range that might be expected in a real situation. The mean 10-min values of the summer and winter days of the input data set as used in the model validation were used in the sensitivity analyses as the starting values. The mean of these starting values for the input variables with their standard deviation during the day are given in Table 3. The sensitivity of the different factors was determined when using a concrete slatted floor with 10 cm wide slats and 2 cm wide gaps. Results are shown in Table 4 Žfor the floor factors., Table 5 Žfor the slurry factors. and in Figs. 5, 6 and 7. In these calculations, one factor changed to the

Table 5 Sensitivity analysis for the different factors affecting the ammonia emission from the slurry pit Factor

Value

Winter

Summer

Pit

Total

Pit

Total

0.20 0.60

50 150

75 126

50 150

70 130

Air velocity pit Žm sy1 . 0.06 s100%

0.03 0.09

57 138

74 123

57 138

76 122

55 179

75 144

55 180

76 143

a

15 25

values given, while all others were kept at their starting values. The relationship between the natural logarithm of the urease activity and ammonia emission from the slatted and solid floors is given in Fig. 5. Changes at low and high levels of urease activity Ž- 0.001 and ) 0.05. only slightly influence ammonia emission. At medium levels Ž) 0.001 and - 0.05., a change

Ammonia emission Ž%.

TAN slurry Žmol ly1 . a 0.40 s100%

Temperature slurry Ž8C. 20 s100%

Fig. 5. Model calculations of the effect of the natural logarithm ŽLOG. of the urease activity of Ža. the slatted floor and Žb. the solid floor on the ammonia emission.

It is assumed that the urea concentration of the urine remained the same.

Fig. 6. Model calculations of the effect of the pH of mixed slurry on the ammonia emission.

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

in urease activity has a larger effect. The urea concentration of urine and TAN concentration of slurry had an almost linear proportional effect on the ammonia emission from the floor and from the slurry pit respectively. Air velocity and temperature had a stronger effect on the ammonia emission from the slurry pit than on the ammonia emission from the floor. The effect of the urinary pH on the emission is less pronounced than the effect of the pH of the slurry ŽTable 4 and Fig. 6.. The surface area of a urine puddle, at constant depth, on the slatted floor showed a nonlinear relationship with ammonia emission, the ammonia level peaking at approximately 0.4 m2 . At larger puddle areas the levels of ammonia emission decreased ŽFig. 7a.. The puddle area on the solid floor, at constant depth, showed an almost linear relation with ammonia emission ŽFig. 7b.. The magnitude of the ‘fouling area’ had almost no effect on the ammonia emission in the range of the calculations. The effect of the magnitude of the ‘excreting

163

area’ was much larger. The depth of the urine puddle on the slatted floor, and to a lesser extent on the solid floor, also significantly influenced the ammonia emission. Urinations on the solid floor rather than on the slatted floor only slightly increased ammonia emission. The ammonia emissions from the different types of slats were simulated. Emissions from the concrete slatted floor with narrow slats Ž7 cm wide slats. and from the cast iron slatted floor were estimated to be respectively approximately 4% and 9% lower than from the concrete slatted floor with wide slats. The emissions from the metal triangular slats were estimated to be 21% ŽS4. and 24% ŽS5. lower than the emission from the concrete slatted floor with wide slats. The ratio of floor emission to total emission was calculated to be 0.45 for S1, 0.43 for S2, 0.40 for S3, 0.31 for S4 and 0.28 for S5.

6. Discussion

Fig. 7. Model calculations of the effect on ammonia emission of the area of urine puddles on Ža. the slatted floor in the ‘excreting area’ and Žb. the solid floor in the ‘fouling area’. The depth of the urine puddle remained constant in these calculations.

In this study, the ammonia volatilization from housing for fattening pigs where no litter is used and the floor is partially slatted was modelled. Zhang et al. Ž1994. attempted to model this volatilization, but they did not account for the ammonia release from the floor and for the regular addition of fresh urine, which is the main ammonia source, into the slurry pit. Anderson et al. Ž1987. did not account for the floor emission either, and used a pH of 6.5 for the slurry. In our study it was shown that the floor emission can account for more than 40% of the total ammonia release. Further it was shown that the pH of the mixed slurry, as used by Anderson et al. Ž1987., is not representative for the pH of the top layer of the slurry. Generally the pH of the top layer is much higher. The validation of our model showed that the mean of the daily simulated ammonia emissions was on average 3.6% lower than the mean of the measured values. The mean of the relative differences between simulated and measured values was 16.9%. Differences between simulated and measured values were clearly larger at the higher levels of ammonia emission. At these higher levels, emissions were generally underestimated. The main factors influencing the

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

164

ammonia emission and which could possibly explain the differences between simulated and measured values are discussed below. In Table 6, a summary is given of the sensitivity analyses of the various factors influencing ammonia emission. 6.1. pH of urine and slurry The pH is one of the main factors influencing the ammonia emission ŽEq. Ž10... When the pH of the mixed slurry changes by 0.1 unit a change in ammonia emission of 9.0% was calculated ŽTable 6.. Elzing and Aarnink Ž1996. found reductions varying from 7% to 18% when the pH measured in the top layer of the slurry was lowered by 0.2 unit in a scale model of a pig house. Aarnink et al. Ž1996. found a regression coefficient of 1.0 Žs.e., 0.3. for the influence of the pH of mixed slurry on the natural logarithm of ammonia emission from a real pig house. This means that ammonia emission changed by about 10% when the pH changed by 0.1 unit. In

our study a relationship was estimated between the pH of the mixed slurry, which is an input in the model, and the ‘effective’ pH of the surface layer. The calculated ‘effective’ pH was 1.1 units higher than the pH of the mixed slurry, with a maximum of 8.8. Olesen and Sommer Ž1993. also found a clearly higher pH in the top layer compared with the pH of the mixed slurry. Three days after mixing they found the pH of the surface to be 0.84 unit higher than the initial pH of the mixed slurry. Sommer and Husted Ž1995. developed a model to predict pH changes in mixed slurry on the basis of NH q 4 rNH 3 , 2y CO 2rHCOy rCO and concentrations of volatile 3 3 fatty acids. Higher concentrations of CO 2 and volatile fatty acids generally lower the pH, while higher concentrations of NH 3 increase the pH of slurry. CO 2 is less soluble than NH 3 . Husted et al. Ž1991. found CO 2 volatilization to be 5.5 times higher than NH 3 volatilization during the first 4 h after mixing. A slow transport rate from CO 2 in the slurry ŽOlesen and Sommer, 1993. caused the CO 2 to NH 3 ratio to

Table 6 Mean change in ammonia emission in the given range for the given change of the model variable Žcalculated from Table 4, Table 5 and Fig. 5, Fig. 6 and Fig. 7. Model variable

Unit

Range

Change in model variable Žunits.

Change in ammonia emissiona Ž%.

pH mixed slurry b pH urine c wUreax fresh urine d Sm slatted floor Sm solid floor Ventilation rate e

g kgy1 mol urea my3 sy1 mol urea my3 sy1 m3 hy1 pigy1

7.0–7.8 8.6–9.0 6.0–9.0 0.0025–0.0498 0.0025–0.0498 20–80

0.1 0.1 0.1 0.01 0.01 10

9.0 1.8 1.8 6.9 0.7 6.5

Temperature Floor Pit

8C 8C

15–25 15–25

1.0 1.0

1.2 6.8

Area urine puddle f Slatted floor Solid floor

m2 m2

0.01–0.1 0.01–0.1

0.02 0.02

7.6 0.6

a

Changes are calculated relative to the mean of the ammonia emissions at the minimum and maximum levels of the given range. With the assumption that the pH of the urine does not change. c With the assumption that the pH of the slurry does not change. d It is assumed that the ammonia concentration in the slurry is changed at 2r3 of the change of the urea concentration Žthe factor 2r3 corrects the dilution of the urine with faeces, which is assumed to be produced in half of the amount of the urine.. e These calculations are based on the relation between ventilation rate and air velocities on the floor and in the slurry pit, as given in Eq. Ž6.. f At constant depth of the urine puddle. b

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

be lower in the surface layer than in the rest of the slurry. This seem to be the main cause of the higher pH of the surface layer compared with the rest of the slurry. Model calculations performed by Ni et al. Ž1996. also show a clear increase of the pH of the top layer of the slurry when CO 2 was volatilized from the slurry. They estimated a maximum steady state pH of approximately 8.85, which agrees well with our supposed maximum pH of 8.8. Although the literature seems to support our estimated relationship between the pH of the mixed slurry and the ‘effective’ pH of the top layer, the basis for this relationship is still very weak and further research is needed in this area. The inaccuracy in determining the ‘effective’ pH is probably one of the main reasons for the differences between simulated and measured emission levels in the lower range of ammonia emissions Ž- 9 g dy1 pigy1 .. The pH of urine has only a small effect on the ammonia emission from the floor ŽTable 4. and on the total ammonia emission ŽTable 6.. It would seem that irrespective of the urinary pH, almost all nitrogen in the urine left on the floor is volatilized. 6.2. Urea and TAN concentration Calculations of the urea concentration in urine and the ammonia emission from the floor demonstrated an almost linear proportional effect ŽTable 4.. This linear relationship was experimentally proven by Elzing and Kroodsma Ž1993. in a scale model of a dairy cow house. The sensitivity analysis also shows a linear proportional effect of the TAN concentration on the ammonia emission from the slurry. Because almost all ammoniacal nitrogen originates from urea in urine ŽAarnink et al., 1993. there is close relationship between the urea concentration in the urine and the TAN concentration in the slurry. Table 6 shows the estimated total effect of a change in urea concentration on the ammonia emission. A decrease of 8.3% in the urea concentration resulted in ammonia emission falling by 6.2%. Measurements are in agreement with this calculated effect. Smits et al. Ž1995. reported a reduction in ammonia emission in a dairy cow house of 39% when the urea concentration decreased by 42%. Van der Peet-Schwering et al. Ž1996. found a mean reduction of 10.7% in

165

ammonia emission when at a fixed water to feed ratio the nitrogen excreted via the urine was lowered by 14.7%, although there seemed to be an interaction effect between feeding and housing treatments in their study. For the validation the urea concentrations were calculated with the MESPRO model ŽAarnink et al., 1992.. The validation of that model showed a mean error of 11.4% for the total nitrogen excreted in faeces and urine by groups of pigs. This error might also have added to the differences between simulated and measured ammonia emissions. 6.3. Urease actiÕity and floor type The urease activity is determined in the model via input of the mean roughness of the floor surface ŽEq. Ž8... Fig. 5 shows that changes in urease activities above levels of 0.05 mol my3 sy1 or below levels of 0.001 have only a minor influence on ammonia emission. At high urease activity levels the factor limiting ammonia emission is not the urease but the urea. At low levels of urease activity, ammonia volatilization from the floor is almost zero and is negligible compared with the release from the slurry pit. Urease activity on the concrete slats and the cast iron slats was much higher than the urease activity on the metal triangular slats ŽTable 2 and Fig. 5., resulting in less ammonia emission from the latter slats. The calculations show that differences in ammonia volatilizations from different types of slatted floors are not only the result of differences in the area and the depth of the urine puddles on the slats, but also of the smoothness of the material of the slats. The simulated effects of the different slatted floors on ammonia volatilization agree well with the measured effects ŽAarnink et al., 1997.. The urease activity of the solid floor gave almost maximum ammonia volatilization from the floor ŽTable 2 and Fig. 5.. The calculated ratios between floor and total emissions for the different types of slatted floors show lower ratios for the metal slatted floors Žapproximately 30%. compared with the concrete and the cast iron slatted floors Žapproximately 43%.. The estimated floor emission of 43% for the concrete slatted floor with narrow slats is in good agreement

166

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

with the 40% estimated in earlier research ŽAarnink et al., 1996..

to obtain more basic knowledge about the dependence of air flow pattern in animal houses on temperature differences and animal behaviour.

6.4. Air Õelocity and air flow pattern

6.5. Temperature

The air velocities above the emitting surfaces of the floor and slurry pit were calculated from a linear relationship with ventilation rate ŽEq. Ž6... The sensitivity analyses showed that air velocity on the floor had only a minor effect on the total ammonia emission, while the effect of the air velocity above the slurry in the pit was substantial. Air velocities on the floor and in the slurry pit may be significantly influenced by the air flow pattern. Randall et al. Ž1983. showed a clear effect of the ventilation system, creating different air flow patterns, on the air velocity near the pigs. Such an effect of ventilation systems may also be expected on the air velocity in the slurry pit. However, Jungbluth and Buscher ¨ Ž1996. and Aarnink and Wagemans Ž1997. found only little effect on ammonia emissions when the air was extracted just beneath the ceiling compared with extraction underneath or just above the slatted floor. When there is sufficient space between the slurry surface and the slatted floor and between the slurry surface and the inlets to the pit ventilation ducts, only little air movement above the slurry was observed ŽJungbluth and Buscher, 1996.. ¨ Changing air flow patterns during the day might have been an important factor influencing the diurnal pattern of ammonia emission at high levels of ammonia emission. During periods of high emissions the temperatures and ventilation rates were generally high and the pigs were heavy. When temperatures are high and pigs are heavy, the pigs will lie on the slatted floor ŽAarnink et al., 1996, 1997.. This behaviour of the pigs might cause changes in the air flow through the slurry pit and influence the air velocity above the slurry. In a cattle house it was observed that temperature differences between the air inside and outside the slurry pit is a main factor determining the air flow in and the ammonia emission from the slurry pit ŽMonteny, unpublished results.. In houses for rearing pigs, Aarnink et al. Ž1993. also suggested an effect of temperature differences on the ammonia emission. Research is needed

In Table 6, the floor temperature is shown to have only a minor effect on the total emission of ammonia. Our assumption that the floor temperature equalled the compartment temperature Žmeasured at 1.0 m height. may be simplistic; this error, however, has only a small effect on the total ammonia emission. The effect of the slurry temperature is much greater. Model calculations show that the ammonia emission changes by an average of 6.8% for each 1.08C change in slurry temperature ŽTable 6.. Aarnink et al. Ž1996. estimated a regression coefficient of 0.10 Žs.e., 0.02. for the natural logarithm ŽLog. of ammonia emission on slurry temperature. This means that ammonia emission changed by approximately 10% when the slurry temperature changed by 1.08C. Although the simulated effect of slurry temperature is lower, it falls within the calculated confidence interval of the measured regression coefficient. 6.6. Size of ‘ fouling’ and ‘excreting’ areas The size of the ‘fouling area’ has only a minor influence on ammonia emission, while the effect of the size of the ‘excreting area’ is significant. The large ‘fouling area’ in combination with the relatively small number of urinations in this area causes only very few replacements of ‘old’ urine puddles by fresh puddles. In contrast, many urinations in a relatively small ‘excreting area’ result in a frequent replacement of ‘old’ puddles. Enlarging the ‘excreting area’ would prolong the life time of the ‘old’ puddles, giving a larger emitting area and a higher ammonia emission. 6.7. Depth and area of urine puddles The depth of the urine puddle has a significant effect on ammonia emission ŽTable 4.. The effect of the depth of the urine puddle is less on the solid floor than on the slatted floor. This can simply be

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

explained by the smaller number of urinations on the solid floor than on the slatted floor, resulting in less emission from the solid floor. The depth of a urine puddle on the different types of slats was determined by dividing the amount of urine left on the slats by the top area of the slats wetted with urine. However, it was observed that the sides of the slats were also partially wetted with urine, especially the metal triangular slats. As mentioned before, almost all the nitrogen left on the solid or slatted floor volatilizes. This means that the volume of the urine puddle on the solid or slatted floor is more important for the total ammonia emission than the depth or the size of the puddle. In the range of the calculations, errors in total emission levels made by calculations with depth rather than area are therefore small. When the areas of urine puddles, in the ‘excreting area’, are larger than measured for the different types of slatted floors, the error will increase. Then, urine puddles will be replaced even before maximum emissions from the puddles have been reached. Fig. 7 shows that at puddle areas larger than approximately 0.4 m2 the ammonia emission even decreases, while the puddle depth remains the same. The diurnal pattern may be significantly influenced by depth and area of the urine puddle. When the puddle area enlarges at constant puddle volume, volatilization rate is initially higher, but decreases faster due to depletion of ammonia in the thin layer. The effect on ammonia emission when pigs urinate on the solid floor instead of on the slatted floor was found to be slight. The puddle area is larger on the solid floor, but the depth of the urine puddle is less than on the slatted floor. The result was a smaller volume of the urine puddle on the solid floor than on the concrete slatted floor with wide slats. Still, ammonia emission was slightly higher when urine was deposited on the solid floor, because of its higher rate of replacement of ‘old’ puddles on the slatted floor of the ‘excreting area’ than on the solid floor of the ‘fouling area’. It should be noted that the depth of the urine puddle on the slatted floor was determined after the slats had been fouled with faeces, whereas the depth of the urine puddle on the solid floor was determined on a clean floor. During a fattening period there will always be a thin layer of dust on the solid floor or, in the case of fouling, also some faeces. The depth of the urine puddles on the

167

solid floor used in our model might have been too shallow, especially for fouled floors. This might have contributed to the underestimation of the ammonia emission at high emission levels. On the slatted floors it was also found that the puddles were much deeper on fouled than on clean slats.

7. Conclusions The dynamic model, described in this paper, could predict the ammonia emission from a house for fattening pigs with a partially slatted floor reasonably well. The mean of the daily simulated values was 3.6% lower than the mean of the measured values. The mean relative difference between daily simulated and measured ammonia emissions was 16.9%. Levels and patterns were simulated more accurately at the low and moderate emission levels Ž- 9 g dy1 pigy1 . than at the high emission levels Ž) 9 g dy1 pigy1 .. The inaccurate prediction at the high levels was probably caused by an underestimation of the depth of urine puddles on a fouled solid floor and by an interactive effect between climatic and behavioural factors on the air flow pattern in the pig house. The pH of the top layer of the slurry in the pit is generally much higher than the pH of the mixed slurry. This seems to be caused by the lower CO 2 to NH 3 ratio in the surface layer than in the rest of the slurry. Simulated effects of slurry pH, urea concentration, slatted floor type and slurry temperature corresponded satisfactorily with the measured effects reported in literature. Also, the simulated division between floor and pit emissions agreed well with the division experimentally measured. The model may be improved by determining the depth of the urine puddle on the solid floor at different degrees of fouling and by a more accurate measurement or model estimation of the pH of the top layer of the slurry in the pit. A better understanding of the climate and behavioural factors influencing the air flow pattern is also recommended for further refinement of the model.

168

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169

References Aarnink, A.J.A., Wagemans, M.J.M., 1997. Ammonia volatilization and dust concentration as affected by ventilation systems in houses for fattening pigs. Trans. ASAE 40 Ž4., 1161–1170. Aarnink, A.J.A., Van Ouwerkerk, E.N.J., Verstegen, M.W.A., 1992. A mathematical model for estimating the amount and composition of slurry from fattening pigs. Livest. Prod. Sci. 31, 133–147. Aarnink, A.J.A., Hoeksma, P., Van Ouwerkerk, E.N.J., 1993. Factors affecting ammonium concentration in slurry from fattening pigs. In: Verstegen, M.W.A., den Hartog, L.A., van Kempen, G.J.M., Metz, J.H.M. ŽEds.., Nitrogen Flow in Pig Production and Environmental Consequences. EAAP-publication No. 69, Pudoc-DLO, Wageningen, pp. 413–420. Aarnink, A.J.A., Van den Berg, A.J., Keen, A., Hoeksma, P., Verstegen, M.W.A., 1996. Effect of slatted floor area on ammonia emission and on the excretory and lying behaviour of growing pigs. J. Agric. Eng. Res. 64, 299–310. Aarnink, A.J., Swierstra, A., Van den Berg, D.A.J., Speelman, L., 1997. Effect of type of slatted floor and degree of fouling of solid floor on ammonia emission rates from fattening piggeries. J. Agric. Eng. Res. 66, 93–102. Anderson, G.A., Smith, R.J., Bundy, D.S., Hammond, E.G., 1987. Model to predict gaseous contaminants in swine confinement buildings. J. Agric. Eng. Res. 37, 235–253. Braam, C.R., Van den Hoorn, C.J. ŽEds.., 1996. Ammoniakemissie-arme betonnen stalvloeren; Resultaten van experimenteel en toegepast onderzoek wConcrete floors for animal houses with low ammonia emission; Results of experimental and applied researchx. Report 96-12, IMAG-DLO Wageningen, pp. 207. Bremner, J.M., Mulvaney, R.L., 1978. Urease activity in soils. In: Burns, R.G. ŽEd.., Soil Enzymes. Academic Press, London, pp. 149–196. Canh, T.T., Verstegen, M.W.A., Aarnink, A.J.A., Schrama, J.W., 1997. Influence of dietary factors on nitrogen partitioning and composition of urine and faeces of fattening pigs. J. Anim. Sci. 75, 700–706. Drummond, J.G., Curtis, S.E., Simon, J., Norton, H.W., 1980. Effects of aerial ammonia on the growth and health of young pigs. J. Anim. Sci. 50, 1085–1091. Elzing, A., Aarnink, A.J.A., 1996. Invloed van de zuurtegraad van varkensurine op de ammoniakemissie. Report 96-02, IMAGDLO Wageningen, pp. 27. Elzing, A., Kroodsma, W., 1993. Relatie tussen ammoniakemissie en stikstofconcentratie in urine van melkvee. Report 93-3, IMAG-DLO Wageningen, pp. 22. Elzing, A., Monteny, G.J., 1997. Modeling and experimental determination of ammonia emission rates from a scale model dairy cow house. Trans. ASAE 40 Ž3., 721–726. Elzing, A., Kroodsma, W., Scholtens, R., Uenk, G.H., 1992. Ammoniakemissiemetingen in een modelsysteem van een rundveestal: Theoretische beschouwingen. Report 92-3, IMAG-DLO Wageningen, pp. 25. Fangmeier, A., Hadwiger-Fangmeier, A., Van der Eerden, L.,

Jager, H.-J., 1994. Effects of atmospheric ammonia on vegeta¨ tion—a review. Environ. Pollut. 86, 43–82. Hartung, E., Buscher, W., 1995. Ammoniakfreisetzung. ¨ Einflußfaktoren auf den Stoffubergang aus Flussigmist. ¨ ¨ Landtechnik 50 Ž3., 164–165. Hashimoto, A.G., 1972. Ammonia desorption from concentrated chicken manure slurries. PhD Thesis, Cornell University, Ithaca, NY, pp. 224. Hashimoto, A.G., Ludington, D.C., 1971. Ammonia desorption from concentrated chicken manure slurries. In Livestock Waste Management and Pollution Abatement. ASAE, St. Joseph, MI, pp. 117–121. Haslam, R.T., Hershey, R.L., Keen, R.H., 1924. Effect of gas velocity and temperature on rate of absorption. Ind. Eng. Chem. 16, 1224–1231. Husted, S., Jensen, L.S., Jørgensen, S.S., 1991. Reducing ammonia loss from cattle slurry by the use of acidifying additives: the role of the buffer system. J. Sci. Food Agric. 57, 335–349. Jungbluth, T., Buscher, W., 1996. Reduction of ammonia emis¨ sions from piggeries. ASAE Paper No. 96-4091. ASAE, St. Joseph, MI. Moore, W.J. 1972. Physical Chemistry, 5th edn. Longman, London. Muck, R.E., Richards, B.K., 1983. Losses of manurial nitrogen in free-stall barns. Agric. Wastes 7, 65–79. Muck, R.E., Steenhuis, T.S., 1981. Nitrogen losses in free stall dairy barns. In Proceedings 4th International Symposium on Livestock Waste: A Renewable Resource. ASAE, St. Joseph, MI, pp. 406–409. Nasi, ¨ M. 1993. Nitrogen utilization responses of growing pigs to supplemental lysine in barley–barley protein diet. In: Verstegen, M.W.A., den Hartog, L.A., van Kempen, G.J.M., Metz, J.H.M. ŽEds.., Proceedings of Congress on Nitrogen Flow in Pig Production and Environmental Consequences. Pudoc Wageningen, pp. 224–229. NEN 3632. 1986. Dutch Standards, Dutch Organization for Standardization, Delft, The Netherlands. Ni, J., Hendriks, J., Berckmans, D., Vinckier, C., 1996. Ammonia release CO2 factor in commercial swine house. ASAE Paper No. 96-4094. ASAE, St. Joseph, MI. Olesen, J.E., Sommer, S.G., 1993. Modelling effects of wind speed and surface cover on ammonia volatilization from stored pig slurry. Atmos. Environ. 27A Ž16., 2567–2574. Pfeiffer, A., Henkel, H., 1991. The effect of different dietary protein levels on water intake and water excretion of growing pigs. In: Verstegen, M.W.A., Huisman, J., den Hartog, L.A. ŽEds.., Proceedings of the 5th International Symposium on Digestive Physiology in Pigs. EAAP-publication no. 54, Pudoc Wageningen, pp. 126–131. Preller, L., 1995. Respiratory health effects of pig farmers. Assessment of exposure and epidemiological studies of risk factors. PhD thesis, Agricultural University Wageningen, pp. 173. Randall, J.M., Armsby, A.W., Sharp, J.R., 1983. Cooling gradients across pens in a finishing piggery: II. Effects on excretory behaviour. J. Agric. Eng. Res. 28, 247–259.

A.J.A. Aarnink, A. Elzingr LiÕestock Production Science 53 (1998) 153–169 Smits, M.C.J., Valk, H., Elzing, A., Keen, A., 1995. Effect of protein nutrition on ammonia emission from cubicle house for dairy cattle. Livest. Prod. Sci. 44, 147–156. Sommer, S.G., Husted, S., 1995. A simple model of pH in slurry. J. Agric. Sci. 124, 447–453. Soveri, J., 1992. Acidifying effects on groundwater. In: Schneider, T. ŽEd.., Acidification Research: Evaluation and Policy Applications. Studies in Environmental Science 50. Elsevier, Amsterdam, pp. 135–143. Stevens, R.J., Laughlin, R.J., Frost, J.P., 1989. Effect of acidifica-

169

tion with sulphuric acid on the volatilization of ammonia from cow and pig slurries. J. Agric. Sci. 113, 389–395. Van der Peet-Schwering, C.M.C., Verdoes, N., Voermans, M.P., Beelen, G.M., 1996. Effect van voeding en huisvesting op de ammoniakemissie uit vleesvarkensstallen. Research report P 1.145, Pigs Research Station, Rosmalen, pp. 40. Zhang, R., Day, D.L., Christianson, L.L., Jepson, W.P., 1994. A computer model for predicting ammonia release rates from swine manure pits. J. Agric. Eng. Res. 58, 223–229.