- Email: [email protected]
Model-based evaluation of ammonia removal in biological air scrubbers
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/issn/15375110
Research Paper
Model-based evaluation of ammonia removal in biological air scrubbers Caroline Van der Heyden a,b, Kimberly Solon a, Peter Demeyer b, Eveline I.P. Volcke a,* a
Biosystems Control Group, Department of Green Chemistry and Technology, Ghent University, Coupure Links 653, 9000 Ghent, Belgium b Unit Technology and Food, Flemish Research Institute for Agriculture and Fisheries, Burgemeester Van Gansberghelaan 115, Bus 1, 9820 Merelbeke, Belgium
article info
A mechanistic model for ammonia removal in a countercurrent biological air scrubber was
Article history:
set up. This model was used to study the effect of the influent characteristics e air tem-
Received 15 April 2019
perature, ventilation rate and ammonia load, on ammonia removal efficiency. Besides
Received in revised form
mass balances of the components participating in the biological conversions, the water
13 December 2019
mass balance and the heat balance were considered. The effects of the pH and the con-
Accepted 29 December 2019
centration of the nitrogen components on the driving force for mass transfer were examined. The model output was compared against experimental data from a pig housing facility. Simulations were performed to assess the usefulness of pH control and to inves-
Keywords:
tigate the effect of inflow air conditions on the ammonia removal efficiency. The study
Biofiltration
found out that although pH control affected the nitrogen component distribution in the
nitrification
washing water, it hardly affected the ammonia removal efficiency. Thus, pH control for
ammonia removal
biological air scrubbers is not recommended in practice, however, an on/off pH control
modelling
system adding only acid at critical moments (pH above 7.5) could be considered. The variations in the ammonia removal efficiency are mainly caused by a changing ventilation rate rather than air temperature fluctuations or ammonia load. © 2020 IAgrE. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Biofiltration by biological air scrubbers or so-called biotrickling filters has evolved to an off-shelf technique to treat exhaust air from industries but also from mechanically ventilated animal houses, as they have been successfully applied for large air streams with low concentrations of pollutants (Cox & Deshusses, 1998). The contaminated air is * Corresponding author. E-mail address: [email protected] (E.I.P. Volcke). https://doi.org/10.1016/j.biosystemseng.2019.12.011 1537-5110/© 2020 IAgrE. Published by Elsevier Ltd. All rights reserved.
passed through a packed bed where soluble and biodegradable gases such as ammonia or volatile compounds are absorbed into a biofilm and converted to harmless forms. Despite the relatively simple operating principle of biotrickling filters, biofiltration is a complex process involving several physical, chemical and biological interactions. In biological air scrubbers, regular discharge of the washing water is needed for proper operation. The moment of
86
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
Nomenclature h t m
Ammonia removal efficiency, % Biomass retention factor, Maximum specific growth rate with pHdependency, d1 mmax Maximum specific growth rate, d1 Abj stoichiometric component of component b and process j, Packing specific surface area, m2.m3 Aspec AP Packing cross sectional area, m2 b Decay rate, d1 CBT;TN Total nitrogen content in buffer tank, gN.m3 max CBT;TN Maximum allowed total nitrogen concentration in the buffer tank, gN.m3 min CBT;TN Minimum allowed total nitrogen concentration in the buffer tank, gN.m3 Cin Absorbed ammonia concentration, gN.m3 G;N CG;NH3 Ammonia concentration in the gas phase, gN.m3 in CG;NH3 Incoming ammonia gas concentration, gN.m3 out CG;NH3 Outgoing ammonia gas concentration, gN.m3 C Hþ Concentration of hydrogen ions in the liquid phase, mol.m3 CL,AOB/XAOB Concentration of ammonia-oxidizing bacteria, gCOD.m3 Cmax Maximum nitrogen concentration in the washing L;N water, gN.m3 CL;NH3 Free ammonia concentration in the liquid phase, gN.m3 * CL;NH3 Equilibrium ammonia concentration in the liquid phase, gN.m3 CL,NOB/XNOB Concentration of nitrite-oxidizing bacteria, gCOD.m3 CL;TAN Concentration of total ammoniacal nitrogen in the liquid phase, gN.m3 CL,TN Concentration of total nitrogen in the liquid phase, gN.m3 DF Driving force, gN.m3 Eact Activation energy, J.mol1 EC Electrical conductivity, mS.cm1 H Henry’s law constant, HP Packing height, m Acid dissociation constant for ammonia, mol.m3 Ka;NH3 KH Henry’s law volatility constant, Mass transfer coefficient, m.s1 KL KpH pH dependency parameter, Incoming ammonia load, gN.h1 LNH3 ;in nY Number of horizontal cells, pH pH in water buffer tank, Optimum pH, pHopt
discharge can be automatically controlled by measuring electrical conductivity (EC), as it is correlated to the total nitrogen (TN) concentration, including ammoniacal, nitrite and nitrate nitrogen (Melse, Ploegaert, & Ogink, 2012b). When the predefined EC setpoint value is reached or exceeded, a fixed volume of water is discharged from the buffer tank using a time-controlled valve and replaced by fresh water. At most
pKa;NH3 QD QG QL Qvent R rb RH RHout G Rj rT r293 T TBT TG TL Tout G TG,in TL,BT TL,in VBT VL,cell
Negative log of Ka;NH3 , Discharge rate, m3.d1 Gas flow rate, m3.h1 Liquid flow rate, m3.h1 Ventilation rate, m3.h1 Gas constant, J.K1.mol1 Conversion rate of component b, g.m3.d1 Initial relative humidity gas phase, % Outflow relative gas humidity, % Reaction rate for process j, g.m3.d1 Decay rate at temperature T, d1 Specific growth/decay rate at 20 C, d1 Temperature, C (in K for Eq. (5)) Temperature in buffer tank, C Temperature of gas phase, C Temperature of liquid phase, C Temperature of outflow gas Temperature of incoming gas, C Temperature of water in buffer tank, C Temperature of incoming liquid, C Buffer tank volume, m3 Volume of each cell, m3
Abbreviations AOB Ammonia-oxidizing bacteria CSTR Continuous stirred tank reactor EBRT Empty bed residence time FA Free ammonia FNA Free nitrous acid NOB Nitrite-oxidizing bacteria TAN Total ammoniacal nitrogen (ammonia þ ammonium) TIC Total inorganic carbon (carbon dioxide þ bicarbonate þ carbonate) TN Total nitrogen (ammonia þ ammonium þ nitrite þ nitrate) Total nitrite nitrogen (nitrite þ nitrous acid) TNO2 Chemical CO2 CO32 HCl HCO3HNO2 NaOH NH3 NH4þ NO2NO3O2
symbols Carbon dioxide Carbonate Hydrochloric acid Bicarbonate Nitrous acid Sodium hydroxide Ammonia Ammonium Nitrite Nitrate Oxygen
installed biological air scrubbers, the discharged volume is kept relatively small. In this way the EC or nitrogen content in the buffer tank decreases only slightly and will fluctuate close around the predefined setpoint. It can be said that the EC is kept nearly constant. Additionally, pH is measured as it must remain between 6.5 and 7.5 to maintain proper operation (InfoMil, 2015; MB31/05/2011). However, biological air
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
scrubbers installed at pig housing facilities in Flanders (Belgium) and the Netherlands most often do not have a pH control system. Ammonia absorption and nitrification have a counteracting effect on pH, keeping the pH rather constant under normal operating conditions. In Germany, biological air scrubbers mostly have a pH dosing system, applying acid or base when necessary (DLG, 2014). In this paper, particular attention was paid to the effect of pH and the concentration of nitrogen components in the washing water on the driving force for mass transfer. The nitrification of a batch test was simulated to study the effect of the decreasing pH and the decreasing nitrogen concentrations on the driving force for mass transfer. Additionally, the total behaviour of a biological air scrubber was described through a mechanistic model. Ammonia removal, temperature and evaporation dynamics were described through mass and energy balances. Simulations were carried out, comparing open-loop operation (i.e. without pH control) to that with closed-loop operation (i.e. with pH control). Continuous short-term and long-term ammonia measurements at biological air scrubbers applied in pig housing facilities showed a daily and seasonal pattern in the ammonia removal performance (Melse, Hofschreuder, & Ogink, 2012a; Van der Heyden, Brusselman, Volcke, & Demeyer, 2016a). The electrical conductivity was kept nearly constant. These changes could be attributed to a fluctuating ventilation rate, ammonia loading rate or air temperature. Because of the interrelations between ventilation rate, air inlet temperature and loading rate, it is not fully clear which are the primary variables responsible for the variation in biotrickling filter performances. The ventilation rate in a pig housing facility determines the air flow rate through the air scrubber and is controlled depending on the air temperature inside the housing facility (Van Gansbeke, Van den Bogaert, & Vettenburg, 2009). Besides, an increased ventilation rate results in a lower ammonia inlet concentration due to dilution if the emission rate is kept constant. Nevertheless, it is possible that with an increased ventilation rate, more ammonia is extracted from the building due to an increased ammonia emission by emitting surfaces or due to increased defecation and urination (Melse et al., 2012b). The air flow through an air scrubber is inversely proportional to the empty bed residence time (EBRT), which reflects the contact time through the air scrubber, and thus may negatively affect the ammonia removal efficiency. At higher air temperatures, the water temperature will increase, making ammonia less soluble (higher H) and in turn reducing the driving force to the liquid phase. Melse et al. (2012b) performed a long term measurement campaign on a biological air scrubber for a pig housing facility and hypothesized that air temperature is the main influencing disturbance variable, rather than the changing ventilation rate and ammonia load. However, it is difficult to test the relative importance of these variables in practice as they are interrelated. By using the biological air scrubber model, these variables can be tested independently. In this paper, the biological air scrubber model, with pH control, was used to study the effect of the influent characteristics e air temperature, ventilation rate and ammonia load, on ammonia removal efficiency.
2.
Theory
2.1.
Operating principles of a biological air scrubber
87
The physicochemical and biological conversions associated with a biological air scrubber for ammonia removal are schematically presented in Fig. 1. Ammonia is absorbed from the gas phase into the liquid phase, where a chemical equilibrium with ammonium is established based on the pH. The total ammoniacal nitrogen (TAN; ammonia and ammonium) concentration is reduced by nitrification, i.e. biological conversion of ammonia to nitrite and subsequently to nitrate. The total nitrogen (TN; ammonia, ammonium, nitrite and nitrate) content in the washing water is controlled by discharge to keep a nearly constant concentration. Sometimes the pH is controlled by dosing acid and/or base to keep a constant pH value. The essential concepts to describe the gaseliquid mass transfer are Henry’s law and the two-film model (Lewis & Whitman, 1924). The driving force (DF) for ammonia mass transfer from the gas to the liquid phase is determined by the difference between the equilibrium ammonia concentration in the liquid phase ðC*L;NH3 Þ; and the actual ammonia concentration in the liquid phase (CL;NH3 ). The former is determined by dividing the ammonia concentration in the gas phase ðCG;NH3 Þ by the Henry’s law volatility constant (KH). The latter is a function of the total ammoniacal nitrogen (TAN) concentration ðCL;TAN Þ; the pH and the acid dissociation constant, as described in Eq. (1): CG;NH3 CL;NH3 DF ¼ C*L;NH3 CL;NH3 ¼ KH ðTÞ CG;NH3 CHþ CL;TAN 1 ¼ KH ðTÞ CHþ þ Ka;NH3 T
(1)
where CL;NH3 is the free ammonia concentration in the liquid phase, CG;NH3 is the ammonia concentration in the gas phase,
Fig. 1 e Physicochemical and biological conversions in a biotrickling filter for ammonia removal. Nitrification by AOB (ammonia-oxidizing bacteria) and NOB (nitriteoxidizing bacteria) results in a reduction of the ammonia concentration (A) as well as in a pH decrease (B). Absorption of ammonia from the gas into the liquid phase results in a pH increase (C).
88
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
KH(T) is the Henry’s law volatility constant, CL;TAN is the concentration of total ammoniacal nitrogen (ammonia and ammonium) in the liquid phase, CHþ is the concentration of hydrogen ions in the liquid phase and Ka;NH3 ðTÞ is the acid dissociation constant for ammonia. Note that both KH(T) and Ka;NH3 ðTÞ are temperature-corrected and is detailed in Supplementary Information A e SA5. Both pH and the total ammoniacal nitrogen concentration thus have an effect on the driving force and are used in practice to control the operation of biological air scrubbers. To gain insight on the relative importance of both variables, the effect of pH and the total ammoniacal nitrogen concentration in the washing water on the mass transfer driving force was studied in more detail. Additionally, ammonia absorption and nitrification both have an influence on and will be determined by pH. The absorption of 1 mol of ammonia (indicated by C in Fig. 1) involves the consumption of 1 mol of protons, resulting in a pH increase. On the other hand, nitrification results in the production of 2 mol of protons per mole ammonia converted (indicated by B in Fig. 1), thus decreasing pH. Nitrification is influenced by the pH through a direct and indirect effect. The growth rate is directly affected by inhibition at both low and high pH values (Van Hulle et al., 2007). A narrow pH interval between 6.5 and 8 exists, where the growth rate is optimal. Additionally, nitrification stops when the pH decreases below a value where the substrate (i.e. ammonia) is no longer available (indirect effect). It is often stated that the role of nitrification (indicated by A in Fig. 1) in a biological air scrubber for ammonia removal is to increase the driving force for mass transfer from the gas phase to the liquid phase by reducing the concentration of total ammoniacal ammonium. However, nitrification also results in a pH decrease, shifting the chemical equilibrium to a relatively lower concentration of free ammonia (indicated by B in Fig. 1), additionally increasing the driving force for ammonia transfer from the gas phase to the liquid phase. To investigate which one has the largest impact on the driving force, the effect of nitrification through the reduction of total ammoniacal nitrogen and through the decrease of pH were considered separately in this paper.
stirred-tank reactor (CSTR), with the cells separated by a stagnant boundary layer at the interface (two film theory). Both mass transfer and heat transfer were considered in order to simulate the removal process, temperature and evaporation in the air scrubber. The individual gas and liquid mass balances can be found in the Supplementary Information A e SA1. The energy and water balances, including the temperature dependencies of the used parameters, were similar to the chemical air scrubber model of Van der Heyden et al. (2016b) and can be found in Supplementary Information A e SA2. The components considered in the gas phase were ammonia (NH3), oxygen (O2) and carbon dioxide (CO2). The components considered in the liquid phase were ammonia (NH3) and ammonium (NHþ 4 ), lumped as total ammonia nitrogen (TAN), nitrite (NO-2) and nitrous acid (HNO2), lumped as total nitrite nitrogen (TNO2), carbon dioxide (CO2), bicarbonate (HCO-3) and carbonate (CO2 3 ), lumped as total inorganic carbon (TIC), nitrate (NO-3), as well as oxygen (O2). In biotrickling filters, the discharge strategy of keeping a nearly constant EC value or total nitrogen concentration is frequently used in practice. This discharge strategy is implemented in the model by setting a maximum allowed total nitrogen concentration in the buffer tank ðCmax BT;TN Þ. In practice, when the setpoint is reached, a small liquid volume will be discharged from the buffer tank, resulting in a small drop in the total nitrogen concentration in the buffer tank. In the model, the nitrogen concentration reached after this discharge is called the minimum allowed total nitrogen concentration in the buffer tank ðCmin BT;TN Þ, which value is close to the maximum allowed nitrogen concentration ðCmax BT;TN Þ, simulating the small drop in the total nitrogen concentration. In the model, this minimum allowed total nitrogen concentration in the buffer tank was assumed to be immediately reached after discharge (i.e. in the next iteration step). This discharge strategy implies that the total amount of absorbed ammonia from the gas into the liquid phase is discharged to keep the total nitrogen at the desired setpoint. The discharge rate (QD) can thus be calculated taking into account the absorbed ammonia concentration ðCin G;N Þ and the maximum nitrogen concentration in the washing water ðCmax L;N Þ; according to following equation: . Cin G;N $Qvent $h 100
3.
Materials and methods
QD ¼
3.1.
Biological air scrubber model
where Qvent is the ventilation rate and h is the ammonia removal efficiency. In a biotrickling filter, the bacteria carrying out the biological conversions are mostly attached to the packing material. The retention of bacteria on the biotrickling filter packing was modelled by including a retention factor in the biomass mass balance, thereby neglecting spatial biofilm gradients. The applicability of this type of zero-dimensional models for the description of a moving bed biofilm reactor has been demonstrated before (Plattes, Henry, & Schosseler, 2008). Two types of bacteria were considered in the biofilm and in the liquid phase, namely ammonia-oxidizing bacteria (XAOB) and nitrite-oxidizing bacteria (XNOB).
A mechanistic model for a biological countercurrent air scrubber was set up and implemented in Matlab-Simulink®. The physical-chemical part of the model was based on the model of a chemical air scrubber by Van der Heyden, Vanthillo, Pieters, Demeyer, and Volcke (2016b). The packing material was divided into a number (nY ¼ 10) of horizontal cells, complying with the required accuracy while keeping a reasonable calculation time. A fixed-step discrete solver was used with a time step chosen smaller than the refreshment rate of each cell. Each cell consisted of a gas and liquid phase, each of which was modelled as an ideally mixed continuous
Cmax L;N
(2)
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
The individual biomass mass balance in cell i was implemented as dXi t,QL iþ1 ¼ , X Xi þ rb dt VL;cell
(3)
where t is the biomass retention factor, QL is the liquid flow rate and VL,cell is the volume of each cell. rb denotes the conversion rate of component b (Eq. (4)), which is calculated from the process rates and the stoichiometric matrix (Supplementary Information A e SA3, Table S1) rb ¼
X j
Abj ,Rj
(4)
where Abj is the stoichiometric component of component b and process j, and Rj is the reaction rate for process j. The process rates are expressed by a Monod-equation taking into account the affinity for O2 and NH3 during nitritation and for O2 and NO-2 during nitratation. Inhibition by free ammonia (FA) and free nitrous acid (FNA) was assumed to be negligible in this study as nitrite was not found to accumulate in the study of Melse et al. (2012b). For biomass decay, the decay rate is directly proportional with the biomass concentration. The stoichiometric and kinetic parameters related to AOB and NOB growth are summarized in Supplementary Information A e SA3 (Table S2). The effect of temperature (T) on the maximum specific growth rate and decay rate (rT) is taken into account, considering the Arrhenius equation: Eact ,ð293 TÞ rT ¼ r293 ,exp R,293,T
(5)
where r293 is specific growth/decay rate at 20 C, Eact is the activation energy, R is the gas constant and T is the temperature. Absorption of ammonia and the biological conversion reactions that involve proton consumption or production affect the pH of the medium. Vice versa, pH influences the biological conversion rates both through a direct effect on the maximum growth rates (Van Hulle et al., 2007) and indirectly through the chemical equilibria of the species involved. pH was therefore added in the model as a state variable and calculated by means of the charge balance method (Hellinga, van Loosdrecht, & Heijnen, 1999; Volcke, 2006), as detailed in Appendix Supplementary Information A e SA4. The bacteria are indirectly dependent on the pH through the equilibrium between ammonia and ammonium. A direct pH-dependency of the maximum specific growth rate (mmax) was added, according to Van Hulle et al. (2007), with a pH optimum (pHopt) of 7.23 and a pH dependency parameter (KpH) of 8.21: KpH m ¼ mmax , KpH 1 þ 10jpHopt pHj
3.2.
(6)
System under study & simulation set-up
Before the biological air scrubber model is used to simulate and study some scenarios, two main aspects of the operating principle of air scrubbers were tested separately. In the first place, the concept of the driving force and the impact of pH and the total ammoniacal nitrogen concentration of the
89
washing water on the driving force is presented in more detail, using a simple visual representation. Secondly, a batch test was considered to test the effect of nitrification on the driving force of ammonia mass transfer. Nitrification results both in a reduced total ammoniacal nitrogen concentration (indicated by A in Fig. 1) and a reduced pH (indicated by B in Fig. 1). To investigate which one has the largest impact on the driving force, the effect of nitrification through the reduction of total ammonium and through the decrease of pH were considered separately. For this purpose, a batch of washing water was considered, containing a total nitrogen content of 3.2 gN.L1 (i.e. the maximum value allowed in Flanders) (MB31/05/2011), distributed as 50% TAN and 50% nitrate. An ammonia gas concentration of 20 ppm was assumed; the corresponding equilibrium liquid concentration was considered in the calculation of the driving force for ammonia mass transfer (Eq. (1)). A temperature of 20 C was assumed. The change in pH was iteratively calculated through the charge balance method (Supplementary Information A e SA4, Eq. S9), considering also the bicarbonate/carbon dioxide equilibrium. A total inorganic carbon (TIC) concentration of 500 gC/m3 was considered, assuming that the washing water is saturated with CO2. The washing water then has a large buffer capacity resulting only in a pH change due to nitrification and not by CO2 absorption. Subsequently, the biological air scrubber model was used to simulate some scenarios. A reference case was defined based on the biotrickling filter studied by Melse et al. (2012b). It consists of a countercurrent biotrickling filter used to treat the exhaust gas of a pig housing facility of 600 sows, with a maximum of 40 000 m3.h1 ventilation air. Discharge was automatically controlled using an electrical conductivity (EC) setpoint around 16 mS.cm1 to limit the total nitrogen content in the washing water, which varied between 1775 and 4817 gN.m3. The discharge rate was on average 0.33 m3.d1. The inlet conditions were set according to constant values, representing the average values during the 141 days measurement period of Melse et al. (2012b) (Table 1). In the model, the maximum allowed total nitrogen content in the washing 3 water ðCmax BT;TN Þ was set at 3200 gN.m , the setpoint in Flanders (MB31/05/2011). Additionally, this value is comparable to the total nitrogen concentrations of the washing water in the study of Melse et al. (2012b). The minimal total nitrogen concentration in the buffer tank after discharge ðCmin BT;TN Þ was set to 3150 gN.m3. The discharge rate to keep the total nitrogen concentration below this setpoint amounts to 0.80 m3.d1 (Eq. (2)). The initial conditions were set according to the values in Table 2. The initial values of AOB and NOB were estimated based on prevailing concentrations of bacteria in activated sludge systems. As pH was not controlled in the study of Melse et al. (2012b), the reference case behaviour was therefore simulated with the model including pH dynamics. The initial pH was set to 6.6, which is the average pH value measured (Melse et al., 2012b). The simulation results of the reference case were compared to the measurement results of Melse et al. (2012b). As the measurement results were only used as input for the model simulation and were not used for parameter estimation, only an intuitive qualitative comparison was made between the model and the measurement results. The
90
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
Table 1 e Reference case - design and input parameters (Melse et al., 2012b). Symbol HP AP Aspec VBT TG,in TL,in Cin G;NH3 QG QL LNH3 ;in
Description
Value
Unit
Packing height Packing cross sectional area Packing specific surface area Buffer tank volume Incoming air temperature Incoming liquid temperature (assumed) Incoming ammonia concentration Gas flow rate Liquid flow rate Incoming ammonia load
1.1 3.8 100 20 20 13 14 13 900 15 396
m m2 m2.m3 m3 C C ppm m3.h1 m3.h1 gN.h1
Table 2 e Initial conditions. Variable TG TL TBT RH CG;NH3 CL,TN CBT,TN CL,AOB CL,NOB
Description Initial temperature gas phase Initial temperature liquid phase Initial temperature buffer tank Initial relative humidity gas phase Initial NH3 gas concentration Initial TN concentration liquid phase Initial TN concentration buffer tank Initial AOB concentration liquid phase Initial NOB concentration liquid phase
simulation results of the reference case (i.e. open-loop behaviour) was subsequently compared to the one of a biotrickling filter in which pH was controlled at a constant value (i.e. closed-loop behaviour). A pH setpoint of 6.6 was considered. This can be achieved in practice through dosing of a base, such as sodium hydroxide (NaOH), as was performed in zquez, Bezerra, Lafuente, and Gabriel (2017). the study of Bla This model with pH control (closed-loop behaviour) was then applied to study the influence of the influent characteristics and conditions on the ammonia removal efficiency, by independently varying the incoming air temperature, ventilation rate and incoming ammonia loading rate to test which of these has the largest effect on the ammonia removal efficiency of biological air scrubbers.
4.
Value
Results and discussion
4.1. Effect of pH and TAN concentration on mass transfer driving force e simple representation The total nitrogen content in the washing water is set through the discharge setpoint and corresponds with a certain TAN concentration. The TAN concentration and pH, on their turn, determine the amount of free ammonia in the water phase ðCL;NH3 ; Eq. (1)). Figure 2 presents both the free ammonia concentration in the liquid phase ðCL;NH3 Þin terms of pH for various total ammoniacal nitrogen concentrations ðCL;TAN Þ and the equilibrium concentration ðC*L;NH3 Þ for various ammonia gas phase concentrations ðCG Þ. This figure thus represents in which cases a mass transfer driving force is established (in case C*L;NH3 is higher than CL;NH3 for the same pH). The solid lines represent
Unit
20 13 13 65 0 0 0 5000 5000
C C C % gN.m3 gN.m3 gN.m3 gCOD.m3 gCOD.m3
CL;NH3 and were calculated using the second term in Eq. (1) (i.e., C Hþ CL;NH3 ¼ CL;TAN 1 C þ þK . The dashed lines represent a;NH H
3
C*L;NH3 and were calculated using the first term of Eq. (1) (i.e., C
* 3 C*L;NH3 ¼ G;NH KH ). As evident in Eq. (1), only when CL;NH3 is higher than CL;NH3 will a positive driving force exist. To illustrate, at a pH of 7, an ammonia gas concentration of 10 ppm can be removed in the biological air scrubber if the TAN concentration in the washing water is 1.6 gN.L1 but not anymore at 3.2 gN.L1. This shows that it is important to keep the total nitrogen setpoint, and in this way the total ammoniacal nitrogen content in the washing water, low enough. The lower the pH, the higher the total nitrogen concentration allowed in the washing water to still keep a high driving force for ammonia transfer. At a pH of 4 or lower, as applied in a chemical air scrubber, all total ammoniacal nitrogen is in the form of ammonium. Therefore, the total allowed nitrogen content in the washing water may be significantly higher than in biological air scrubbers (58 gN.L1 compared to 3.2 gN.L1; MB31/05/2011), reducing the discharge rate significantly (Eq. (2)). In a biological air scrubber, such high total nitrogen concentrations, and thus total ammoniacal nitrogen concentrations, cannot be allowed. As the pH lies only around 7, a slight increase can immediately result in a strongly reduced driving force.
4.2. Effect of nitrification on mass transfer driving force e batch test Figure 3 represents the nitrification of a batch of washing water with a total ammoniacal nitrogen concentration of 1600 gN.m3. The total effect of nitrification concerns both a decreased total ammonium concentration and a decreased pH
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
91
Fig. 2 e Free ammonia concentrations in the washing water for different pH and total ammoniacal nitrogen concentrations ðCL;NH3 at CL,TAN - solid lines) and the ammonia concentrations in equilibrium with different gas concentrations, based on Henry’s law ðC*L;NH3 - dashed lines).
Fig. 3 e Effect of nitrification on a batch of washing water containing 1600 gN.m-3 (CTAN) on the driving force (DF) for ammonia mass transfer, comparing the total effect of nitrification (A and B in Fig. 1, solid lines) with the effect of nitrification at constant pH (A in Fig. 1, dashed lines). DF and pH are displayed as a function of the percentage of the total ammonium concentration which is nitrified.
(solid lines). The driving force for ammonia transfer (DF) increases with increasing ammonium conversion, until it reaches the maximum value equalling C*L;NH3 : At this point, no free ammonia (FA) is present in the liquid as the pH is low (pH < 4) and the ammonia/ammonium equilibrium is shifted to ammonium ðpKa;NH3 ¼ 9.4 at 20 C). The effect of nitrification on the driving force for ammonia mass transfer (DF) was also studied for a constant pH,
considering only the decreased total ammonium concentration (dashed lines). In this case, the driving force increases linearly with an increasing ammonium conversion, at a gradual manner (i.e. small slope). Comparing the two scenarios, it is clear that the effect of nitrification on the driving force for ammonia mass transfer is mainly due the associated pH decrease rather than due to the reduction of the total ammonium concentration in the liquid.
92
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
Fig. 4 e Comparison of the experimental data of Melse et al. (2012b) with the simulation results of model without pH control and the model with regulated pH, for the nitrogen components in the washing water: total nitrogen (TN), total ammoniacal nitrogen (TAN), nitrite (NO-2) and nitrate (NO3).
Table 3 e Comparison between measured data (Melse et al., 2012b) and steady-state simulation results for the reference case without pH control and with pH control. Symbol h Cout G;NH3 QD RHout G Tout G TL;BT pH
Removal efficiency Outgoing NH3 gas concentration Discharge rate Outflow relative gas humidity Outflow gas temperature Temperature of water in buffer tank pH in water buffer tank
Unit
Measured (Melse et al., 2012b)
% ppm m3.d1 % C C e
82 2.4 0.33 >95 16 16 6.6
The effect of pH will even be more pronounced when the initial pH of the liquid batch is higher because the initial driving force is smaller. If 1% is nitrified at an initial pH of 7.5, this induces a decrease of pH by only 0.2, but this doubles the initial driving force. In practice, the pH will not drop below 4 as nitrification has an optimum pH and is inhibited at low pH (Van Hulle et al., 2007). These results only represent a batch test. During normal operation of an air scrubber, absorption of ammonia from the gas to the liquid phase will again increase the pH of the washing water.
4.3.
Reference case e open-loop operation
The scenario without pH control resulted in a steady state ammonia removal efficiency of 92%. The total ammoniacal nitrogen, total nitrite and nitrate concentrations were 1.75, 0.001 and 1.45 gN.L1, respectively, corresponding with a total nitrogen concentration of 3.2 gN.L1 (Fig. 4; Table 3). The cation (i.e. NHþ 4 ) has to be balanced by anions (NO2 and NO3) (Ottosen et al., 2011). As a result, the total ammoniacal nitrogen cannot be entirely converted. In this study, only half of the total ammonia concentration in the water was converted, as 54.5% of the total nitrogen is still total ammoniacal nitrogen. The observed incomplete ammonia conversion can be
Simulated no pH control
pH control
92 1.0 0.87 99.2 15 15 6.2
94 0.82 0.80 99.2 15 15 6.6
attributed to the effect of pH, as explained in Section 3.1. As nitrification produces two moles of protons, while absorption only consumes one mole of protons, the bacteria can only convert half of the absorbed ammonia. At steady state, an equilibrium between ammonia absorption and nitrification is reached at a pH of 6.2. The nitrate concentration is high (44.7% of the total nitrogen concentration) while the nitrite concentration in the washing water is very low (0.8% of the total nitrogen concentration, Fig. 4), indicating that almost all nitrite, which is formed by the AOB, is immediately converted by NOB. The AOB concentration is higher than the NOB concentration, as AOB have a higher specific yield than NOB (Supplementary material, Figure S1). The gas temperature inside the air scrubber decreased from 20 C at the inlet to 15 C at the outlet, thereby heating up the colder washing water to 15 C. The relative humidity increased from 65% to almost 100%. These temperature and relative humidity profiles are similar to the ones for the chemical air scrubber model (Van der Heyden et al., 2016b). The steady state simulation results agree well with the measurement data of Melse et al. (2012b) (Fig. 4, Table 3). The slightly higher simulated ammonia removal, compared to the average measured value over the entire measurement
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
93
period (efficiency: 92% simulated versus 82% measured), could be attributed to dynamics in the operation of the air scrubber of Melse et al. (2012b). For instance, the total nitrogen content was not constant during the measurement campaign. Therefore, the average total nitrogen content was also slightly higher than the used value in the simulations (Fig. 4). The distribution of the nitrogen components in the water was similar, with only half of the total nitrogen that could be converted to nitrate. This was also observed during other measurement campaign (Melse & Mosquera, 2014; Mosquera et al., 2011; Ottosen et al., 2011; Van der Heyden et al., 2016a). The simulated results thus agree well with the experimental data from a conventional pig housing facility.
4.4.
Effect of pH control e closed-loop operation
When the pH is controlled at a constant value, almost all (99%) absorbed ammonia is converted to nitrate (Fig. 4), instead of only 44.7% when pH is not controlled. With pH control, nitrification is not inhibited by the decreasing pH, resulting in full conversion of ammonium to nitrate. Through dosing of a base, the pH is kept around 7, thereby allowing nitrification to exceed ammonia absorption. Because of this, the biomass can have higher growth compared to the case without pH control, resulting in a higher AOB and NOB concentration (Supplementary Information B e Figure S1). Additionally, at the same pH, the driving force will be higher compared to the situation without regulated pH, as the TAN concentration is much lower (Fig. 4). However, only a slight increase in the ammonia removal efficiency is observed overall (94% compared to 92%). zquez et al. (2017), it was shown In a lab scale study by Bla that a biotrickling filter treating ammonia gas concentration of 100 ppmv could reach 100% ammonium conversion to nitrate (i.e. complete nitrification) by keeping the pH constant around 7. An on/off pH control system based on the addition of NaOH (0.5M) or HCl (0.5M) was used. For this high ammonia gas concentration, this type of control system could be beneficial. The question arises to which extent the operation of a biological air scrubber with constant pH is different from a chemical air scrubber and if it is worth applying in practice. In a biological air scrubber the pH is only controlled around 7 compared to 4 in a chemical air scrubber. For each ammonia molecule that is absorbed, a net 1 mol of proton is formed after complete nitrification, which must be neutralised with a base (Supplementary Information C). This means that the base consumption in a biological air scrubber is as high as the acid consumption in a chemical air scrubber, increasing the operational costs substantially. Additionally, the ammonia removal efficiency only increases slightly using the base dosing system in biological air scrubbers (94% compared to 92%). In a chemical air scrubber, at a pH of 4, all total ammoniacal nitrogen is in the form of ammonium. Therefore, the total allowed nitrogen content in the washing water may be significantly higher than in biological air scrubbers (58 gN.L1 compared to 3.2 gN.L1) (MB31/05/2011), reducing
Fig. 5 e Influence of the influent conditions: air temperature ( C), ventilation rate (m3.h¡1) and ammonia load (gN.h¡1) on the ammonia removal efficiency.
the discharge rate significantly (Eq. (2)). In a biological air scrubber, these high total nitrogen concentrations cannot be allowed as this would inhibit the bacteria. As the pH lies only around 7, a slight increase can immediately result in a reduced driving force. Controlling the pH at a constant value around 7 by neutralising all formed protons, thus seems not a viable option to implement in practice in pig housing facilities. However, an on/off pH control system adding only acid at critical moments (pH above 7.5) could be considered in practice. The system then only serves as a back-up if the pH increases too far. Only a high pH has a direct negative impact on the ammonia removal efficiency (Fig. 2).
4.5.
Effect of inlet conditions
Figure 5 shows the influence of the inlet conditions: air temperature, ventilation rate and ammonia load on the ammonia removal efficiency. The effect of an increased air temperature on the ammonia removal efficiency is negligible (Fig. 5A). At increased air temperatures, ammonia is less soluble (higher H), reducing the driving force to the liquid phase. The acid dissociation constant increases (higher Ka;NH3 Þ resulting in a higher concentration of ammonia compared to ammonium. This results in a decreased driving force as well as increased ammonia availability for the biomass. Besides, the specific growth rate and decay rate of the biomass increases (higher mmax and b). Additionally, the mass transfer coefficient will increase (higher KL), implying that more ammonia will be transferred from the gas phase to the liquid phase which has a positive effect on the removal efficiency. As a result of all the different effects of temperature, the total effect on the ammonia removal efficiency is less than 1% over the considered temperature range. Figure 5B shows a decreased ammonia removal efficiency at an increased ventilation rate. At a higher ventilation rate or
94
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
air flow rate through the air scrubber, the incoming ammonia load increases but the EBRT (empty bed residence time ¼ contact time) decreases. The net effect of both is a decreased ammonia removal efficiency. When keeping a constant ventilation rate but increasing the inlet ammonia load by increasing the inlet ammonia concentration (Fig. 5C), the outlet ammonia concentration increases but the removal efficiency remains constant. The effects of air temperature, ventilation rate and load on the ammonia removal efficiency are similar to those observed for a chemical air scrubber (Van der Heyden et al., 2016b). Overall, considering these simulation results, it is the ventilation rate rather than the inlet air temperature or influent load that influences the ammonia removal efficiency in biological air scrubbers.
5.
Conclusions A mechanistic model for a biotrickling filter for ammonia removal was set up, taking into account temperature, relative humidity and pH dynamics. The simulation results were in agreement with experimental data from a conventional pig housing facility. The positive impact of nitrification on the driving force for mass ammonia transfer was due to the associated pH decrease rather than to the reduction of the total ammoniacal nitrogen content as such. The total nitrogen concentration in the washing water still needs to be kept sufficiently low to sustain the driving force for ammonia from the gas phase to the liquid phase. The application of pH control affected the nitrogen component distribution in the washing water. Without pH control, only half of the ammonia absorbed in the washing water could be oxidized by the biomass. When the pH was controlled at a constant value which is sufficiently low to allow ammonia mass transfer to the liquid phase but still high enough for bacterial growth, almost complete ammonium conversion could be achieved. The ammonia removal efficiency as such was hardly affected by pH control, while operational costs for base dosing are substantial. Applying pH control at biological air scrubbers in practice is therefore not recommended. However, an on/off pH control system adding only acid at critical moments (pH above 7.5) could be considered in practice. Simulation results clearly demonstrated that variations in the ammonia removal efficiency are mainly caused by a changing ventilation rate rather than by fluctuations in air temperature or ammonia load.
Author contribution All authors have participated in (a) conception and design, or analysis and interpretation of the data; (b) drafting the article or revising it critically for important intellectual content; and (c) approval of the final version.
Acknowledgements The work of Caroline Van der Heyden was financially supported by the Special Research Fund (BOF) of Ghent University (project number 01N02012), the Research Institute for Agriculture, Fisheries and Food (ILVO) and through a Ph.D. fellowship of the agency for Innovation by Science and Technology (IWT, project number SB-131586). Kimberly Solon has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie Grant Agreement No. 846316 (WISEFLOW).
Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.biosystemseng.2019.12.011.
references
zquez, E., Bezerra, T., Lafuente, J., & Gabriel, D. (2017). Bla Performance, limitations and microbial diversity of a biotrickling filter for the treatment of high loads of ammonia. Chemical Engineering Journal, 311, 91e99. Cox, H., & Deshusses, M. (1998). Biological waste air treatment in biotrickling filters. Current Opinion in Biotechnology, 9(3), 256e262. DLG. (2014). DLG Expert Knowledge Series 403: Notes on operating waste air cleaning facilities for pig farming. Frankfurt, Germany: DLG e.V. Hellinga, C., van Loosdrecht, M. C. M., & Heijnen, J. J. (1999). Model based design of a novel process for nitrogen removal from concentrated flows. Mathematical and Computer Modelling of Dynamical Systems, 5(4), 351e371. InfoMil. (2015). Kenniscentrum InfoMil. Retrieved from https:// www.infomil.nl/onderwerpen/landbouw/ammoniak/rav-0/ bijlage-1/stalbeschrijvingen/ (2018 June 30). Lewis, W. K., & Whitman, W. G. (1924). Principles of gas absorption. Industrial & Engineering Chemistry Research, 16(12), 1215e1220. MB31/05/2011. (2011). Ministrieel besluit van 31 mei 2011 tot wijziging van bijlage I van het ministrieel besluit van 19 maart 2004 houdende vaststelling van de lijst van ammoniakemissiearme stalsystemen in uitvoering van artikel 1.1.2 en artikel 5.9.2. Brussels, Belgium: 1bis. Melse, R. W., Hofschreuder, P., & Ogink, N. W. M. (2012a). Removal of particulate matter (PM10) by air scrubbers at livestock facilities: Results of an on-farm monitoring program. Transactions of the ASABE, 55(2), 689e698. Melse, R. W., & Mosquera, J. (2014). Nitrous oxide (N2O) emissions from biotrickling filters used for ammonia removal at livestock facilities. Water Science and Technology, 69(5), 994e1003. Melse, R. W., Ploegaert, J. P. M., & Ogink, N. W. M. (2012b). Biotrickling filter for the treatment of exhaust air from a pig rearing building: Ammonia removal performance and its fluctuations. Biosystems Engineering, 113, 242e252. Mosquera, J., Hol, J. M. G., Melse, R. W., Winkel, A., Nijeboer, G. M., Ploegaert, J. P. M., et al. (2011). Dust emission from animal houses: Air scrubbing techniques. Wageningen, the Netherlands: Livestock Research (in Dutch). Report 295. Ottosen, L. D. M., Juhler, S., Guldberg, L. B., Feilberg, A., Revsbech, N. P., & Nielsen, L. P. (2011). Regulation of ammonia
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
oxidation in biotrickling air filters with high ammonium load. Chemical Engineering Journal, 167(1), 198e205. Plattes, M., Henry, E., & Schosseler, P. M. (2008). A zerodimensional biofilm model for dynamic simulation of moving bed bioreactor systems: Model concepts, Peterson matrix, and application to a pilot-scale plant. Biochemical Engineering Journal, 40(2), 392e398. Van Gansbeke, S., Van den Bogaert, T., & Vettenburg, N. (2009). Ventilatie en klimaatbeheersing bij varkensstallen - vlaamse Overheid Departement Landbouw en Visserij Afdeling Duurzame Landbouwontwikkeling (ADLO), Brussels, Belgium. Van Hulle, S. W. H., Volcke, E. I. P., Josefa, L., Donckels, B., van Loosdrecht, M. C. M., & Vanrolleghem, P. A. (2007). Influence of temperature and pH on the kinetics of the Sharon nitritation
95
process. Journal of Chemical Technology & Biotechnology, 82, 471e480. Van der Heyden, C., Brusselman, E., Volcke, E. I. P., & Demeyer, P. (2016a). Continuous measurements of ammonia, nitrous oxide and methane from air scrubbers at pig housing facilities. Journal of Environmental Management, 181(3), 163e171. Van der Heyden, C., Vanthillo, B., Pieters, J., Demeyer, P., & Volcke, E. I. P. (2016b). Mechanistic modelling of pollutant removal, temperature and evaporation in chemical air scrubbers. Chemical Engineering & Technology, 39(10), 1785e1796. Volcke, E. I. P. (2006). Modelling, analysis and control of partial nitritation in a SHARON reactor. PhD thesis. Belgium: Ghent University.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/issn/15375110
Research Paper
Model-based evaluation of ammonia removal in biological air scrubbers Caroline Van der Heyden a,b, Kimberly Solon a, Peter Demeyer b, Eveline I.P. Volcke a,* a
Biosystems Control Group, Department of Green Chemistry and Technology, Ghent University, Coupure Links 653, 9000 Ghent, Belgium b Unit Technology and Food, Flemish Research Institute for Agriculture and Fisheries, Burgemeester Van Gansberghelaan 115, Bus 1, 9820 Merelbeke, Belgium
article info
A mechanistic model for ammonia removal in a countercurrent biological air scrubber was
Article history:
set up. This model was used to study the effect of the influent characteristics e air tem-
Received 15 April 2019
perature, ventilation rate and ammonia load, on ammonia removal efficiency. Besides
Received in revised form
mass balances of the components participating in the biological conversions, the water
13 December 2019
mass balance and the heat balance were considered. The effects of the pH and the con-
Accepted 29 December 2019
centration of the nitrogen components on the driving force for mass transfer were examined. The model output was compared against experimental data from a pig housing facility. Simulations were performed to assess the usefulness of pH control and to inves-
Keywords:
tigate the effect of inflow air conditions on the ammonia removal efficiency. The study
Biofiltration
found out that although pH control affected the nitrogen component distribution in the
nitrification
washing water, it hardly affected the ammonia removal efficiency. Thus, pH control for
ammonia removal
biological air scrubbers is not recommended in practice, however, an on/off pH control
modelling
system adding only acid at critical moments (pH above 7.5) could be considered. The variations in the ammonia removal efficiency are mainly caused by a changing ventilation rate rather than air temperature fluctuations or ammonia load. © 2020 IAgrE. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Biofiltration by biological air scrubbers or so-called biotrickling filters has evolved to an off-shelf technique to treat exhaust air from industries but also from mechanically ventilated animal houses, as they have been successfully applied for large air streams with low concentrations of pollutants (Cox & Deshusses, 1998). The contaminated air is * Corresponding author. E-mail address: [email protected] (E.I.P. Volcke). https://doi.org/10.1016/j.biosystemseng.2019.12.011 1537-5110/© 2020 IAgrE. Published by Elsevier Ltd. All rights reserved.
passed through a packed bed where soluble and biodegradable gases such as ammonia or volatile compounds are absorbed into a biofilm and converted to harmless forms. Despite the relatively simple operating principle of biotrickling filters, biofiltration is a complex process involving several physical, chemical and biological interactions. In biological air scrubbers, regular discharge of the washing water is needed for proper operation. The moment of
86
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
Nomenclature h t m
Ammonia removal efficiency, % Biomass retention factor, Maximum specific growth rate with pHdependency, d1 mmax Maximum specific growth rate, d1 Abj stoichiometric component of component b and process j, Packing specific surface area, m2.m3 Aspec AP Packing cross sectional area, m2 b Decay rate, d1 CBT;TN Total nitrogen content in buffer tank, gN.m3 max CBT;TN Maximum allowed total nitrogen concentration in the buffer tank, gN.m3 min CBT;TN Minimum allowed total nitrogen concentration in the buffer tank, gN.m3 Cin Absorbed ammonia concentration, gN.m3 G;N CG;NH3 Ammonia concentration in the gas phase, gN.m3 in CG;NH3 Incoming ammonia gas concentration, gN.m3 out CG;NH3 Outgoing ammonia gas concentration, gN.m3 C Hþ Concentration of hydrogen ions in the liquid phase, mol.m3 CL,AOB/XAOB Concentration of ammonia-oxidizing bacteria, gCOD.m3 Cmax Maximum nitrogen concentration in the washing L;N water, gN.m3 CL;NH3 Free ammonia concentration in the liquid phase, gN.m3 * CL;NH3 Equilibrium ammonia concentration in the liquid phase, gN.m3 CL,NOB/XNOB Concentration of nitrite-oxidizing bacteria, gCOD.m3 CL;TAN Concentration of total ammoniacal nitrogen in the liquid phase, gN.m3 CL,TN Concentration of total nitrogen in the liquid phase, gN.m3 DF Driving force, gN.m3 Eact Activation energy, J.mol1 EC Electrical conductivity, mS.cm1 H Henry’s law constant, HP Packing height, m Acid dissociation constant for ammonia, mol.m3 Ka;NH3 KH Henry’s law volatility constant, Mass transfer coefficient, m.s1 KL KpH pH dependency parameter, Incoming ammonia load, gN.h1 LNH3 ;in nY Number of horizontal cells, pH pH in water buffer tank, Optimum pH, pHopt
discharge can be automatically controlled by measuring electrical conductivity (EC), as it is correlated to the total nitrogen (TN) concentration, including ammoniacal, nitrite and nitrate nitrogen (Melse, Ploegaert, & Ogink, 2012b). When the predefined EC setpoint value is reached or exceeded, a fixed volume of water is discharged from the buffer tank using a time-controlled valve and replaced by fresh water. At most
pKa;NH3 QD QG QL Qvent R rb RH RHout G Rj rT r293 T TBT TG TL Tout G TG,in TL,BT TL,in VBT VL,cell
Negative log of Ka;NH3 , Discharge rate, m3.d1 Gas flow rate, m3.h1 Liquid flow rate, m3.h1 Ventilation rate, m3.h1 Gas constant, J.K1.mol1 Conversion rate of component b, g.m3.d1 Initial relative humidity gas phase, % Outflow relative gas humidity, % Reaction rate for process j, g.m3.d1 Decay rate at temperature T, d1 Specific growth/decay rate at 20 C, d1 Temperature, C (in K for Eq. (5)) Temperature in buffer tank, C Temperature of gas phase, C Temperature of liquid phase, C Temperature of outflow gas Temperature of incoming gas, C Temperature of water in buffer tank, C Temperature of incoming liquid, C Buffer tank volume, m3 Volume of each cell, m3
Abbreviations AOB Ammonia-oxidizing bacteria CSTR Continuous stirred tank reactor EBRT Empty bed residence time FA Free ammonia FNA Free nitrous acid NOB Nitrite-oxidizing bacteria TAN Total ammoniacal nitrogen (ammonia þ ammonium) TIC Total inorganic carbon (carbon dioxide þ bicarbonate þ carbonate) TN Total nitrogen (ammonia þ ammonium þ nitrite þ nitrate) Total nitrite nitrogen (nitrite þ nitrous acid) TNO2 Chemical CO2 CO32 HCl HCO3HNO2 NaOH NH3 NH4þ NO2NO3O2
symbols Carbon dioxide Carbonate Hydrochloric acid Bicarbonate Nitrous acid Sodium hydroxide Ammonia Ammonium Nitrite Nitrate Oxygen
installed biological air scrubbers, the discharged volume is kept relatively small. In this way the EC or nitrogen content in the buffer tank decreases only slightly and will fluctuate close around the predefined setpoint. It can be said that the EC is kept nearly constant. Additionally, pH is measured as it must remain between 6.5 and 7.5 to maintain proper operation (InfoMil, 2015; MB31/05/2011). However, biological air
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
scrubbers installed at pig housing facilities in Flanders (Belgium) and the Netherlands most often do not have a pH control system. Ammonia absorption and nitrification have a counteracting effect on pH, keeping the pH rather constant under normal operating conditions. In Germany, biological air scrubbers mostly have a pH dosing system, applying acid or base when necessary (DLG, 2014). In this paper, particular attention was paid to the effect of pH and the concentration of nitrogen components in the washing water on the driving force for mass transfer. The nitrification of a batch test was simulated to study the effect of the decreasing pH and the decreasing nitrogen concentrations on the driving force for mass transfer. Additionally, the total behaviour of a biological air scrubber was described through a mechanistic model. Ammonia removal, temperature and evaporation dynamics were described through mass and energy balances. Simulations were carried out, comparing open-loop operation (i.e. without pH control) to that with closed-loop operation (i.e. with pH control). Continuous short-term and long-term ammonia measurements at biological air scrubbers applied in pig housing facilities showed a daily and seasonal pattern in the ammonia removal performance (Melse, Hofschreuder, & Ogink, 2012a; Van der Heyden, Brusselman, Volcke, & Demeyer, 2016a). The electrical conductivity was kept nearly constant. These changes could be attributed to a fluctuating ventilation rate, ammonia loading rate or air temperature. Because of the interrelations between ventilation rate, air inlet temperature and loading rate, it is not fully clear which are the primary variables responsible for the variation in biotrickling filter performances. The ventilation rate in a pig housing facility determines the air flow rate through the air scrubber and is controlled depending on the air temperature inside the housing facility (Van Gansbeke, Van den Bogaert, & Vettenburg, 2009). Besides, an increased ventilation rate results in a lower ammonia inlet concentration due to dilution if the emission rate is kept constant. Nevertheless, it is possible that with an increased ventilation rate, more ammonia is extracted from the building due to an increased ammonia emission by emitting surfaces or due to increased defecation and urination (Melse et al., 2012b). The air flow through an air scrubber is inversely proportional to the empty bed residence time (EBRT), which reflects the contact time through the air scrubber, and thus may negatively affect the ammonia removal efficiency. At higher air temperatures, the water temperature will increase, making ammonia less soluble (higher H) and in turn reducing the driving force to the liquid phase. Melse et al. (2012b) performed a long term measurement campaign on a biological air scrubber for a pig housing facility and hypothesized that air temperature is the main influencing disturbance variable, rather than the changing ventilation rate and ammonia load. However, it is difficult to test the relative importance of these variables in practice as they are interrelated. By using the biological air scrubber model, these variables can be tested independently. In this paper, the biological air scrubber model, with pH control, was used to study the effect of the influent characteristics e air temperature, ventilation rate and ammonia load, on ammonia removal efficiency.
2.
Theory
2.1.
Operating principles of a biological air scrubber
87
The physicochemical and biological conversions associated with a biological air scrubber for ammonia removal are schematically presented in Fig. 1. Ammonia is absorbed from the gas phase into the liquid phase, where a chemical equilibrium with ammonium is established based on the pH. The total ammoniacal nitrogen (TAN; ammonia and ammonium) concentration is reduced by nitrification, i.e. biological conversion of ammonia to nitrite and subsequently to nitrate. The total nitrogen (TN; ammonia, ammonium, nitrite and nitrate) content in the washing water is controlled by discharge to keep a nearly constant concentration. Sometimes the pH is controlled by dosing acid and/or base to keep a constant pH value. The essential concepts to describe the gaseliquid mass transfer are Henry’s law and the two-film model (Lewis & Whitman, 1924). The driving force (DF) for ammonia mass transfer from the gas to the liquid phase is determined by the difference between the equilibrium ammonia concentration in the liquid phase ðC*L;NH3 Þ; and the actual ammonia concentration in the liquid phase (CL;NH3 ). The former is determined by dividing the ammonia concentration in the gas phase ðCG;NH3 Þ by the Henry’s law volatility constant (KH). The latter is a function of the total ammoniacal nitrogen (TAN) concentration ðCL;TAN Þ; the pH and the acid dissociation constant, as described in Eq. (1): CG;NH3 CL;NH3 DF ¼ C*L;NH3 CL;NH3 ¼ KH ðTÞ CG;NH3 CHþ CL;TAN 1 ¼ KH ðTÞ CHþ þ Ka;NH3 T
(1)
where CL;NH3 is the free ammonia concentration in the liquid phase, CG;NH3 is the ammonia concentration in the gas phase,
Fig. 1 e Physicochemical and biological conversions in a biotrickling filter for ammonia removal. Nitrification by AOB (ammonia-oxidizing bacteria) and NOB (nitriteoxidizing bacteria) results in a reduction of the ammonia concentration (A) as well as in a pH decrease (B). Absorption of ammonia from the gas into the liquid phase results in a pH increase (C).
88
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
KH(T) is the Henry’s law volatility constant, CL;TAN is the concentration of total ammoniacal nitrogen (ammonia and ammonium) in the liquid phase, CHþ is the concentration of hydrogen ions in the liquid phase and Ka;NH3 ðTÞ is the acid dissociation constant for ammonia. Note that both KH(T) and Ka;NH3 ðTÞ are temperature-corrected and is detailed in Supplementary Information A e SA5. Both pH and the total ammoniacal nitrogen concentration thus have an effect on the driving force and are used in practice to control the operation of biological air scrubbers. To gain insight on the relative importance of both variables, the effect of pH and the total ammoniacal nitrogen concentration in the washing water on the mass transfer driving force was studied in more detail. Additionally, ammonia absorption and nitrification both have an influence on and will be determined by pH. The absorption of 1 mol of ammonia (indicated by C in Fig. 1) involves the consumption of 1 mol of protons, resulting in a pH increase. On the other hand, nitrification results in the production of 2 mol of protons per mole ammonia converted (indicated by B in Fig. 1), thus decreasing pH. Nitrification is influenced by the pH through a direct and indirect effect. The growth rate is directly affected by inhibition at both low and high pH values (Van Hulle et al., 2007). A narrow pH interval between 6.5 and 8 exists, where the growth rate is optimal. Additionally, nitrification stops when the pH decreases below a value where the substrate (i.e. ammonia) is no longer available (indirect effect). It is often stated that the role of nitrification (indicated by A in Fig. 1) in a biological air scrubber for ammonia removal is to increase the driving force for mass transfer from the gas phase to the liquid phase by reducing the concentration of total ammoniacal ammonium. However, nitrification also results in a pH decrease, shifting the chemical equilibrium to a relatively lower concentration of free ammonia (indicated by B in Fig. 1), additionally increasing the driving force for ammonia transfer from the gas phase to the liquid phase. To investigate which one has the largest impact on the driving force, the effect of nitrification through the reduction of total ammoniacal nitrogen and through the decrease of pH were considered separately in this paper.
stirred-tank reactor (CSTR), with the cells separated by a stagnant boundary layer at the interface (two film theory). Both mass transfer and heat transfer were considered in order to simulate the removal process, temperature and evaporation in the air scrubber. The individual gas and liquid mass balances can be found in the Supplementary Information A e SA1. The energy and water balances, including the temperature dependencies of the used parameters, were similar to the chemical air scrubber model of Van der Heyden et al. (2016b) and can be found in Supplementary Information A e SA2. The components considered in the gas phase were ammonia (NH3), oxygen (O2) and carbon dioxide (CO2). The components considered in the liquid phase were ammonia (NH3) and ammonium (NHþ 4 ), lumped as total ammonia nitrogen (TAN), nitrite (NO-2) and nitrous acid (HNO2), lumped as total nitrite nitrogen (TNO2), carbon dioxide (CO2), bicarbonate (HCO-3) and carbonate (CO2 3 ), lumped as total inorganic carbon (TIC), nitrate (NO-3), as well as oxygen (O2). In biotrickling filters, the discharge strategy of keeping a nearly constant EC value or total nitrogen concentration is frequently used in practice. This discharge strategy is implemented in the model by setting a maximum allowed total nitrogen concentration in the buffer tank ðCmax BT;TN Þ. In practice, when the setpoint is reached, a small liquid volume will be discharged from the buffer tank, resulting in a small drop in the total nitrogen concentration in the buffer tank. In the model, the nitrogen concentration reached after this discharge is called the minimum allowed total nitrogen concentration in the buffer tank ðCmin BT;TN Þ, which value is close to the maximum allowed nitrogen concentration ðCmax BT;TN Þ, simulating the small drop in the total nitrogen concentration. In the model, this minimum allowed total nitrogen concentration in the buffer tank was assumed to be immediately reached after discharge (i.e. in the next iteration step). This discharge strategy implies that the total amount of absorbed ammonia from the gas into the liquid phase is discharged to keep the total nitrogen at the desired setpoint. The discharge rate (QD) can thus be calculated taking into account the absorbed ammonia concentration ðCin G;N Þ and the maximum nitrogen concentration in the washing water ðCmax L;N Þ; according to following equation: . Cin G;N $Qvent $h 100
3.
Materials and methods
QD ¼
3.1.
Biological air scrubber model
where Qvent is the ventilation rate and h is the ammonia removal efficiency. In a biotrickling filter, the bacteria carrying out the biological conversions are mostly attached to the packing material. The retention of bacteria on the biotrickling filter packing was modelled by including a retention factor in the biomass mass balance, thereby neglecting spatial biofilm gradients. The applicability of this type of zero-dimensional models for the description of a moving bed biofilm reactor has been demonstrated before (Plattes, Henry, & Schosseler, 2008). Two types of bacteria were considered in the biofilm and in the liquid phase, namely ammonia-oxidizing bacteria (XAOB) and nitrite-oxidizing bacteria (XNOB).
A mechanistic model for a biological countercurrent air scrubber was set up and implemented in Matlab-Simulink®. The physical-chemical part of the model was based on the model of a chemical air scrubber by Van der Heyden, Vanthillo, Pieters, Demeyer, and Volcke (2016b). The packing material was divided into a number (nY ¼ 10) of horizontal cells, complying with the required accuracy while keeping a reasonable calculation time. A fixed-step discrete solver was used with a time step chosen smaller than the refreshment rate of each cell. Each cell consisted of a gas and liquid phase, each of which was modelled as an ideally mixed continuous
Cmax L;N
(2)
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
The individual biomass mass balance in cell i was implemented as dXi t,QL iþ1 ¼ , X Xi þ rb dt VL;cell
(3)
where t is the biomass retention factor, QL is the liquid flow rate and VL,cell is the volume of each cell. rb denotes the conversion rate of component b (Eq. (4)), which is calculated from the process rates and the stoichiometric matrix (Supplementary Information A e SA3, Table S1) rb ¼
X j
Abj ,Rj
(4)
where Abj is the stoichiometric component of component b and process j, and Rj is the reaction rate for process j. The process rates are expressed by a Monod-equation taking into account the affinity for O2 and NH3 during nitritation and for O2 and NO-2 during nitratation. Inhibition by free ammonia (FA) and free nitrous acid (FNA) was assumed to be negligible in this study as nitrite was not found to accumulate in the study of Melse et al. (2012b). For biomass decay, the decay rate is directly proportional with the biomass concentration. The stoichiometric and kinetic parameters related to AOB and NOB growth are summarized in Supplementary Information A e SA3 (Table S2). The effect of temperature (T) on the maximum specific growth rate and decay rate (rT) is taken into account, considering the Arrhenius equation: Eact ,ð293 TÞ rT ¼ r293 ,exp R,293,T
(5)
where r293 is specific growth/decay rate at 20 C, Eact is the activation energy, R is the gas constant and T is the temperature. Absorption of ammonia and the biological conversion reactions that involve proton consumption or production affect the pH of the medium. Vice versa, pH influences the biological conversion rates both through a direct effect on the maximum growth rates (Van Hulle et al., 2007) and indirectly through the chemical equilibria of the species involved. pH was therefore added in the model as a state variable and calculated by means of the charge balance method (Hellinga, van Loosdrecht, & Heijnen, 1999; Volcke, 2006), as detailed in Appendix Supplementary Information A e SA4. The bacteria are indirectly dependent on the pH through the equilibrium between ammonia and ammonium. A direct pH-dependency of the maximum specific growth rate (mmax) was added, according to Van Hulle et al. (2007), with a pH optimum (pHopt) of 7.23 and a pH dependency parameter (KpH) of 8.21: KpH m ¼ mmax , KpH 1 þ 10jpHopt pHj
3.2.
(6)
System under study & simulation set-up
Before the biological air scrubber model is used to simulate and study some scenarios, two main aspects of the operating principle of air scrubbers were tested separately. In the first place, the concept of the driving force and the impact of pH and the total ammoniacal nitrogen concentration of the
89
washing water on the driving force is presented in more detail, using a simple visual representation. Secondly, a batch test was considered to test the effect of nitrification on the driving force of ammonia mass transfer. Nitrification results both in a reduced total ammoniacal nitrogen concentration (indicated by A in Fig. 1) and a reduced pH (indicated by B in Fig. 1). To investigate which one has the largest impact on the driving force, the effect of nitrification through the reduction of total ammonium and through the decrease of pH were considered separately. For this purpose, a batch of washing water was considered, containing a total nitrogen content of 3.2 gN.L1 (i.e. the maximum value allowed in Flanders) (MB31/05/2011), distributed as 50% TAN and 50% nitrate. An ammonia gas concentration of 20 ppm was assumed; the corresponding equilibrium liquid concentration was considered in the calculation of the driving force for ammonia mass transfer (Eq. (1)). A temperature of 20 C was assumed. The change in pH was iteratively calculated through the charge balance method (Supplementary Information A e SA4, Eq. S9), considering also the bicarbonate/carbon dioxide equilibrium. A total inorganic carbon (TIC) concentration of 500 gC/m3 was considered, assuming that the washing water is saturated with CO2. The washing water then has a large buffer capacity resulting only in a pH change due to nitrification and not by CO2 absorption. Subsequently, the biological air scrubber model was used to simulate some scenarios. A reference case was defined based on the biotrickling filter studied by Melse et al. (2012b). It consists of a countercurrent biotrickling filter used to treat the exhaust gas of a pig housing facility of 600 sows, with a maximum of 40 000 m3.h1 ventilation air. Discharge was automatically controlled using an electrical conductivity (EC) setpoint around 16 mS.cm1 to limit the total nitrogen content in the washing water, which varied between 1775 and 4817 gN.m3. The discharge rate was on average 0.33 m3.d1. The inlet conditions were set according to constant values, representing the average values during the 141 days measurement period of Melse et al. (2012b) (Table 1). In the model, the maximum allowed total nitrogen content in the washing 3 water ðCmax BT;TN Þ was set at 3200 gN.m , the setpoint in Flanders (MB31/05/2011). Additionally, this value is comparable to the total nitrogen concentrations of the washing water in the study of Melse et al. (2012b). The minimal total nitrogen concentration in the buffer tank after discharge ðCmin BT;TN Þ was set to 3150 gN.m3. The discharge rate to keep the total nitrogen concentration below this setpoint amounts to 0.80 m3.d1 (Eq. (2)). The initial conditions were set according to the values in Table 2. The initial values of AOB and NOB were estimated based on prevailing concentrations of bacteria in activated sludge systems. As pH was not controlled in the study of Melse et al. (2012b), the reference case behaviour was therefore simulated with the model including pH dynamics. The initial pH was set to 6.6, which is the average pH value measured (Melse et al., 2012b). The simulation results of the reference case were compared to the measurement results of Melse et al. (2012b). As the measurement results were only used as input for the model simulation and were not used for parameter estimation, only an intuitive qualitative comparison was made between the model and the measurement results. The
90
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
Table 1 e Reference case - design and input parameters (Melse et al., 2012b). Symbol HP AP Aspec VBT TG,in TL,in Cin G;NH3 QG QL LNH3 ;in
Description
Value
Unit
Packing height Packing cross sectional area Packing specific surface area Buffer tank volume Incoming air temperature Incoming liquid temperature (assumed) Incoming ammonia concentration Gas flow rate Liquid flow rate Incoming ammonia load
1.1 3.8 100 20 20 13 14 13 900 15 396
m m2 m2.m3 m3 C C ppm m3.h1 m3.h1 gN.h1
Table 2 e Initial conditions. Variable TG TL TBT RH CG;NH3 CL,TN CBT,TN CL,AOB CL,NOB
Description Initial temperature gas phase Initial temperature liquid phase Initial temperature buffer tank Initial relative humidity gas phase Initial NH3 gas concentration Initial TN concentration liquid phase Initial TN concentration buffer tank Initial AOB concentration liquid phase Initial NOB concentration liquid phase
simulation results of the reference case (i.e. open-loop behaviour) was subsequently compared to the one of a biotrickling filter in which pH was controlled at a constant value (i.e. closed-loop behaviour). A pH setpoint of 6.6 was considered. This can be achieved in practice through dosing of a base, such as sodium hydroxide (NaOH), as was performed in zquez, Bezerra, Lafuente, and Gabriel (2017). the study of Bla This model with pH control (closed-loop behaviour) was then applied to study the influence of the influent characteristics and conditions on the ammonia removal efficiency, by independently varying the incoming air temperature, ventilation rate and incoming ammonia loading rate to test which of these has the largest effect on the ammonia removal efficiency of biological air scrubbers.
4.
Value
Results and discussion
4.1. Effect of pH and TAN concentration on mass transfer driving force e simple representation The total nitrogen content in the washing water is set through the discharge setpoint and corresponds with a certain TAN concentration. The TAN concentration and pH, on their turn, determine the amount of free ammonia in the water phase ðCL;NH3 ; Eq. (1)). Figure 2 presents both the free ammonia concentration in the liquid phase ðCL;NH3 Þin terms of pH for various total ammoniacal nitrogen concentrations ðCL;TAN Þ and the equilibrium concentration ðC*L;NH3 Þ for various ammonia gas phase concentrations ðCG Þ. This figure thus represents in which cases a mass transfer driving force is established (in case C*L;NH3 is higher than CL;NH3 for the same pH). The solid lines represent
Unit
20 13 13 65 0 0 0 5000 5000
C C C % gN.m3 gN.m3 gN.m3 gCOD.m3 gCOD.m3
CL;NH3 and were calculated using the second term in Eq. (1) (i.e., C Hþ CL;NH3 ¼ CL;TAN 1 C þ þK . The dashed lines represent a;NH H
3
C*L;NH3 and were calculated using the first term of Eq. (1) (i.e., C
* 3 C*L;NH3 ¼ G;NH KH ). As evident in Eq. (1), only when CL;NH3 is higher than CL;NH3 will a positive driving force exist. To illustrate, at a pH of 7, an ammonia gas concentration of 10 ppm can be removed in the biological air scrubber if the TAN concentration in the washing water is 1.6 gN.L1 but not anymore at 3.2 gN.L1. This shows that it is important to keep the total nitrogen setpoint, and in this way the total ammoniacal nitrogen content in the washing water, low enough. The lower the pH, the higher the total nitrogen concentration allowed in the washing water to still keep a high driving force for ammonia transfer. At a pH of 4 or lower, as applied in a chemical air scrubber, all total ammoniacal nitrogen is in the form of ammonium. Therefore, the total allowed nitrogen content in the washing water may be significantly higher than in biological air scrubbers (58 gN.L1 compared to 3.2 gN.L1; MB31/05/2011), reducing the discharge rate significantly (Eq. (2)). In a biological air scrubber, such high total nitrogen concentrations, and thus total ammoniacal nitrogen concentrations, cannot be allowed. As the pH lies only around 7, a slight increase can immediately result in a strongly reduced driving force.
4.2. Effect of nitrification on mass transfer driving force e batch test Figure 3 represents the nitrification of a batch of washing water with a total ammoniacal nitrogen concentration of 1600 gN.m3. The total effect of nitrification concerns both a decreased total ammonium concentration and a decreased pH
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
91
Fig. 2 e Free ammonia concentrations in the washing water for different pH and total ammoniacal nitrogen concentrations ðCL;NH3 at CL,TAN - solid lines) and the ammonia concentrations in equilibrium with different gas concentrations, based on Henry’s law ðC*L;NH3 - dashed lines).
Fig. 3 e Effect of nitrification on a batch of washing water containing 1600 gN.m-3 (CTAN) on the driving force (DF) for ammonia mass transfer, comparing the total effect of nitrification (A and B in Fig. 1, solid lines) with the effect of nitrification at constant pH (A in Fig. 1, dashed lines). DF and pH are displayed as a function of the percentage of the total ammonium concentration which is nitrified.
(solid lines). The driving force for ammonia transfer (DF) increases with increasing ammonium conversion, until it reaches the maximum value equalling C*L;NH3 : At this point, no free ammonia (FA) is present in the liquid as the pH is low (pH < 4) and the ammonia/ammonium equilibrium is shifted to ammonium ðpKa;NH3 ¼ 9.4 at 20 C). The effect of nitrification on the driving force for ammonia mass transfer (DF) was also studied for a constant pH,
considering only the decreased total ammonium concentration (dashed lines). In this case, the driving force increases linearly with an increasing ammonium conversion, at a gradual manner (i.e. small slope). Comparing the two scenarios, it is clear that the effect of nitrification on the driving force for ammonia mass transfer is mainly due the associated pH decrease rather than due to the reduction of the total ammonium concentration in the liquid.
92
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
Fig. 4 e Comparison of the experimental data of Melse et al. (2012b) with the simulation results of model without pH control and the model with regulated pH, for the nitrogen components in the washing water: total nitrogen (TN), total ammoniacal nitrogen (TAN), nitrite (NO-2) and nitrate (NO3).
Table 3 e Comparison between measured data (Melse et al., 2012b) and steady-state simulation results for the reference case without pH control and with pH control. Symbol h Cout G;NH3 QD RHout G Tout G TL;BT pH
Removal efficiency Outgoing NH3 gas concentration Discharge rate Outflow relative gas humidity Outflow gas temperature Temperature of water in buffer tank pH in water buffer tank
Unit
Measured (Melse et al., 2012b)
% ppm m3.d1 % C C e
82 2.4 0.33 >95 16 16 6.6
The effect of pH will even be more pronounced when the initial pH of the liquid batch is higher because the initial driving force is smaller. If 1% is nitrified at an initial pH of 7.5, this induces a decrease of pH by only 0.2, but this doubles the initial driving force. In practice, the pH will not drop below 4 as nitrification has an optimum pH and is inhibited at low pH (Van Hulle et al., 2007). These results only represent a batch test. During normal operation of an air scrubber, absorption of ammonia from the gas to the liquid phase will again increase the pH of the washing water.
4.3.
Reference case e open-loop operation
The scenario without pH control resulted in a steady state ammonia removal efficiency of 92%. The total ammoniacal nitrogen, total nitrite and nitrate concentrations were 1.75, 0.001 and 1.45 gN.L1, respectively, corresponding with a total nitrogen concentration of 3.2 gN.L1 (Fig. 4; Table 3). The cation (i.e. NHþ 4 ) has to be balanced by anions (NO2 and NO3) (Ottosen et al., 2011). As a result, the total ammoniacal nitrogen cannot be entirely converted. In this study, only half of the total ammonia concentration in the water was converted, as 54.5% of the total nitrogen is still total ammoniacal nitrogen. The observed incomplete ammonia conversion can be
Simulated no pH control
pH control
92 1.0 0.87 99.2 15 15 6.2
94 0.82 0.80 99.2 15 15 6.6
attributed to the effect of pH, as explained in Section 3.1. As nitrification produces two moles of protons, while absorption only consumes one mole of protons, the bacteria can only convert half of the absorbed ammonia. At steady state, an equilibrium between ammonia absorption and nitrification is reached at a pH of 6.2. The nitrate concentration is high (44.7% of the total nitrogen concentration) while the nitrite concentration in the washing water is very low (0.8% of the total nitrogen concentration, Fig. 4), indicating that almost all nitrite, which is formed by the AOB, is immediately converted by NOB. The AOB concentration is higher than the NOB concentration, as AOB have a higher specific yield than NOB (Supplementary material, Figure S1). The gas temperature inside the air scrubber decreased from 20 C at the inlet to 15 C at the outlet, thereby heating up the colder washing water to 15 C. The relative humidity increased from 65% to almost 100%. These temperature and relative humidity profiles are similar to the ones for the chemical air scrubber model (Van der Heyden et al., 2016b). The steady state simulation results agree well with the measurement data of Melse et al. (2012b) (Fig. 4, Table 3). The slightly higher simulated ammonia removal, compared to the average measured value over the entire measurement
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
93
period (efficiency: 92% simulated versus 82% measured), could be attributed to dynamics in the operation of the air scrubber of Melse et al. (2012b). For instance, the total nitrogen content was not constant during the measurement campaign. Therefore, the average total nitrogen content was also slightly higher than the used value in the simulations (Fig. 4). The distribution of the nitrogen components in the water was similar, with only half of the total nitrogen that could be converted to nitrate. This was also observed during other measurement campaign (Melse & Mosquera, 2014; Mosquera et al., 2011; Ottosen et al., 2011; Van der Heyden et al., 2016a). The simulated results thus agree well with the experimental data from a conventional pig housing facility.
4.4.
Effect of pH control e closed-loop operation
When the pH is controlled at a constant value, almost all (99%) absorbed ammonia is converted to nitrate (Fig. 4), instead of only 44.7% when pH is not controlled. With pH control, nitrification is not inhibited by the decreasing pH, resulting in full conversion of ammonium to nitrate. Through dosing of a base, the pH is kept around 7, thereby allowing nitrification to exceed ammonia absorption. Because of this, the biomass can have higher growth compared to the case without pH control, resulting in a higher AOB and NOB concentration (Supplementary Information B e Figure S1). Additionally, at the same pH, the driving force will be higher compared to the situation without regulated pH, as the TAN concentration is much lower (Fig. 4). However, only a slight increase in the ammonia removal efficiency is observed overall (94% compared to 92%). zquez et al. (2017), it was shown In a lab scale study by Bla that a biotrickling filter treating ammonia gas concentration of 100 ppmv could reach 100% ammonium conversion to nitrate (i.e. complete nitrification) by keeping the pH constant around 7. An on/off pH control system based on the addition of NaOH (0.5M) or HCl (0.5M) was used. For this high ammonia gas concentration, this type of control system could be beneficial. The question arises to which extent the operation of a biological air scrubber with constant pH is different from a chemical air scrubber and if it is worth applying in practice. In a biological air scrubber the pH is only controlled around 7 compared to 4 in a chemical air scrubber. For each ammonia molecule that is absorbed, a net 1 mol of proton is formed after complete nitrification, which must be neutralised with a base (Supplementary Information C). This means that the base consumption in a biological air scrubber is as high as the acid consumption in a chemical air scrubber, increasing the operational costs substantially. Additionally, the ammonia removal efficiency only increases slightly using the base dosing system in biological air scrubbers (94% compared to 92%). In a chemical air scrubber, at a pH of 4, all total ammoniacal nitrogen is in the form of ammonium. Therefore, the total allowed nitrogen content in the washing water may be significantly higher than in biological air scrubbers (58 gN.L1 compared to 3.2 gN.L1) (MB31/05/2011), reducing
Fig. 5 e Influence of the influent conditions: air temperature ( C), ventilation rate (m3.h¡1) and ammonia load (gN.h¡1) on the ammonia removal efficiency.
the discharge rate significantly (Eq. (2)). In a biological air scrubber, these high total nitrogen concentrations cannot be allowed as this would inhibit the bacteria. As the pH lies only around 7, a slight increase can immediately result in a reduced driving force. Controlling the pH at a constant value around 7 by neutralising all formed protons, thus seems not a viable option to implement in practice in pig housing facilities. However, an on/off pH control system adding only acid at critical moments (pH above 7.5) could be considered in practice. The system then only serves as a back-up if the pH increases too far. Only a high pH has a direct negative impact on the ammonia removal efficiency (Fig. 2).
4.5.
Effect of inlet conditions
Figure 5 shows the influence of the inlet conditions: air temperature, ventilation rate and ammonia load on the ammonia removal efficiency. The effect of an increased air temperature on the ammonia removal efficiency is negligible (Fig. 5A). At increased air temperatures, ammonia is less soluble (higher H), reducing the driving force to the liquid phase. The acid dissociation constant increases (higher Ka;NH3 Þ resulting in a higher concentration of ammonia compared to ammonium. This results in a decreased driving force as well as increased ammonia availability for the biomass. Besides, the specific growth rate and decay rate of the biomass increases (higher mmax and b). Additionally, the mass transfer coefficient will increase (higher KL), implying that more ammonia will be transferred from the gas phase to the liquid phase which has a positive effect on the removal efficiency. As a result of all the different effects of temperature, the total effect on the ammonia removal efficiency is less than 1% over the considered temperature range. Figure 5B shows a decreased ammonia removal efficiency at an increased ventilation rate. At a higher ventilation rate or
94
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
air flow rate through the air scrubber, the incoming ammonia load increases but the EBRT (empty bed residence time ¼ contact time) decreases. The net effect of both is a decreased ammonia removal efficiency. When keeping a constant ventilation rate but increasing the inlet ammonia load by increasing the inlet ammonia concentration (Fig. 5C), the outlet ammonia concentration increases but the removal efficiency remains constant. The effects of air temperature, ventilation rate and load on the ammonia removal efficiency are similar to those observed for a chemical air scrubber (Van der Heyden et al., 2016b). Overall, considering these simulation results, it is the ventilation rate rather than the inlet air temperature or influent load that influences the ammonia removal efficiency in biological air scrubbers.
5.
Conclusions A mechanistic model for a biotrickling filter for ammonia removal was set up, taking into account temperature, relative humidity and pH dynamics. The simulation results were in agreement with experimental data from a conventional pig housing facility. The positive impact of nitrification on the driving force for mass ammonia transfer was due to the associated pH decrease rather than to the reduction of the total ammoniacal nitrogen content as such. The total nitrogen concentration in the washing water still needs to be kept sufficiently low to sustain the driving force for ammonia from the gas phase to the liquid phase. The application of pH control affected the nitrogen component distribution in the washing water. Without pH control, only half of the ammonia absorbed in the washing water could be oxidized by the biomass. When the pH was controlled at a constant value which is sufficiently low to allow ammonia mass transfer to the liquid phase but still high enough for bacterial growth, almost complete ammonium conversion could be achieved. The ammonia removal efficiency as such was hardly affected by pH control, while operational costs for base dosing are substantial. Applying pH control at biological air scrubbers in practice is therefore not recommended. However, an on/off pH control system adding only acid at critical moments (pH above 7.5) could be considered in practice. Simulation results clearly demonstrated that variations in the ammonia removal efficiency are mainly caused by a changing ventilation rate rather than by fluctuations in air temperature or ammonia load.
Author contribution All authors have participated in (a) conception and design, or analysis and interpretation of the data; (b) drafting the article or revising it critically for important intellectual content; and (c) approval of the final version.
Acknowledgements The work of Caroline Van der Heyden was financially supported by the Special Research Fund (BOF) of Ghent University (project number 01N02012), the Research Institute for Agriculture, Fisheries and Food (ILVO) and through a Ph.D. fellowship of the agency for Innovation by Science and Technology (IWT, project number SB-131586). Kimberly Solon has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie Grant Agreement No. 846316 (WISEFLOW).
Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.biosystemseng.2019.12.011.
references
zquez, E., Bezerra, T., Lafuente, J., & Gabriel, D. (2017). Bla Performance, limitations and microbial diversity of a biotrickling filter for the treatment of high loads of ammonia. Chemical Engineering Journal, 311, 91e99. Cox, H., & Deshusses, M. (1998). Biological waste air treatment in biotrickling filters. Current Opinion in Biotechnology, 9(3), 256e262. DLG. (2014). DLG Expert Knowledge Series 403: Notes on operating waste air cleaning facilities for pig farming. Frankfurt, Germany: DLG e.V. Hellinga, C., van Loosdrecht, M. C. M., & Heijnen, J. J. (1999). Model based design of a novel process for nitrogen removal from concentrated flows. Mathematical and Computer Modelling of Dynamical Systems, 5(4), 351e371. InfoMil. (2015). Kenniscentrum InfoMil. Retrieved from https:// www.infomil.nl/onderwerpen/landbouw/ammoniak/rav-0/ bijlage-1/stalbeschrijvingen/ (2018 June 30). Lewis, W. K., & Whitman, W. G. (1924). Principles of gas absorption. Industrial & Engineering Chemistry Research, 16(12), 1215e1220. MB31/05/2011. (2011). Ministrieel besluit van 31 mei 2011 tot wijziging van bijlage I van het ministrieel besluit van 19 maart 2004 houdende vaststelling van de lijst van ammoniakemissiearme stalsystemen in uitvoering van artikel 1.1.2 en artikel 5.9.2. Brussels, Belgium: 1bis. Melse, R. W., Hofschreuder, P., & Ogink, N. W. M. (2012a). Removal of particulate matter (PM10) by air scrubbers at livestock facilities: Results of an on-farm monitoring program. Transactions of the ASABE, 55(2), 689e698. Melse, R. W., & Mosquera, J. (2014). Nitrous oxide (N2O) emissions from biotrickling filters used for ammonia removal at livestock facilities. Water Science and Technology, 69(5), 994e1003. Melse, R. W., Ploegaert, J. P. M., & Ogink, N. W. M. (2012b). Biotrickling filter for the treatment of exhaust air from a pig rearing building: Ammonia removal performance and its fluctuations. Biosystems Engineering, 113, 242e252. Mosquera, J., Hol, J. M. G., Melse, R. W., Winkel, A., Nijeboer, G. M., Ploegaert, J. P. M., et al. (2011). Dust emission from animal houses: Air scrubbing techniques. Wageningen, the Netherlands: Livestock Research (in Dutch). Report 295. Ottosen, L. D. M., Juhler, S., Guldberg, L. B., Feilberg, A., Revsbech, N. P., & Nielsen, L. P. (2011). Regulation of ammonia
b i o s y s t e m s e n g i n e e r i n g 1 9 1 ( 2 0 2 0 ) 8 5 e9 5
oxidation in biotrickling air filters with high ammonium load. Chemical Engineering Journal, 167(1), 198e205. Plattes, M., Henry, E., & Schosseler, P. M. (2008). A zerodimensional biofilm model for dynamic simulation of moving bed bioreactor systems: Model concepts, Peterson matrix, and application to a pilot-scale plant. Biochemical Engineering Journal, 40(2), 392e398. Van Gansbeke, S., Van den Bogaert, T., & Vettenburg, N. (2009). Ventilatie en klimaatbeheersing bij varkensstallen - vlaamse Overheid Departement Landbouw en Visserij Afdeling Duurzame Landbouwontwikkeling (ADLO), Brussels, Belgium. Van Hulle, S. W. H., Volcke, E. I. P., Josefa, L., Donckels, B., van Loosdrecht, M. C. M., & Vanrolleghem, P. A. (2007). Influence of temperature and pH on the kinetics of the Sharon nitritation
95
process. Journal of Chemical Technology & Biotechnology, 82, 471e480. Van der Heyden, C., Brusselman, E., Volcke, E. I. P., & Demeyer, P. (2016a). Continuous measurements of ammonia, nitrous oxide and methane from air scrubbers at pig housing facilities. Journal of Environmental Management, 181(3), 163e171. Van der Heyden, C., Vanthillo, B., Pieters, J., Demeyer, P., & Volcke, E. I. P. (2016b). Mechanistic modelling of pollutant removal, temperature and evaporation in chemical air scrubbers. Chemical Engineering & Technology, 39(10), 1785e1796. Volcke, E. I. P. (2006). Modelling, analysis and control of partial nitritation in a SHARON reactor. PhD thesis. Belgium: Ghent University.