w a t e r r e s e a r c h 7 1 ( 2 0 1 5 ) 2 9 4 e3 0 5

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/watres

Modeling nitrate removal in a denitrification bed Ehsan Ghane a,*, Norman R. Fausey b, Larry C. Brown a a b

Department of Food, Agricultural and Biological Engineering, Ohio State University, Columbus, OH 43210, USA USDA Agricultural Research Service, Soil Drainage Research Unit, Columbus, OH 43210, USA

article info

abstract

Article history:

Denitrification beds are promoted to reduce nitrate load in agricultural subsurface drainage

Received 13 June 2014

water to alleviate the adverse environmental effects associated with nitrate pollution of

Received in revised form

surface water. In this system, drainage water flows through a trench filled with a carbon

22 September 2014

media where nitrate is transformed into nitrogen gas under anaerobic conditions. The

Accepted 19 October 2014

main objectives of this study were to model a denitrification bed treating drainage water

Available online 30 January 2015

and evaluate its adverse greenhouse gas emissions. Field experiments were conducted at an existing denitrification bed. Evaluations showed very low greenhouse gas emissions

Keywords:

(mean N2O emission of 0.12 mg N m2 min1) from the denitrification bed surface. Field

Arrhenius

experiments indicated that nitrate removal rate was described by MichaeliseMenten ki-

Bromide tracer

netics with the MichaeliseMenten constant of 7.2 mg N L1. We developed a novel deni-

Drainage water

trification bed model based on the governing equations for water flow and nitrate removal

Forchheimer

kinetics. The model evaluation statistics showed satisfactory prediction of bed outflow

Greenhouse gas

nitrate concentration during subsurface drainage flow. The model can be used to design

Woodchip bioreactor

denitrification beds with efficient nitrate removal which in turn leads to enhanced drainage water quality. © 2014 Elsevier Ltd. All rights reserved.

1.

Introduction

For economic crop production, farmers in humid and semihumid regions install subsurface drainage systems to remove excess water from the soil profile. Farmers also apply nitrogen fertilizers on fields and some of the nitrate can be transported with the removal of excess water to surface water bodies. Consequently, nitrate pollution in downstream water bodies can cause adverse effects to aquatic ecosystems such as hypoxia and toxin production by harmful algal blooms (Horst et al., 2014). One emerging biotechnology to reduce nitrate in flowing water is a denitrifying bioreactor of which there are different types, namely wall, bed, and layer (Schipper et al., 2010b;

Bednarek et al., 2014). The bed type, also known as a woodchip bioreactor, is well suited for treating agricultural subsurface drainage water. This system is a trench filled with a carbon media (commonly woodchips) that can remove nitrate by transformation to nitrogen gas. One of the earliest implementations of denitrification beds was in California, USA, where drainage water was diverted through a trench filled with barley straw (Williford et al., 1971). In this system, denitrification is the main nitrate removal mechanism (Greenan et al., 2009; Warneke et al., 2011a), so knowledge of the denitrification process is essential to the modeling of denitrification beds. The reaction rate law governing the denitrification process describes the extent of nitrate removal in a denitrification bed. Thus, obtaining the correct rate law is important to the

* Corresponding author. Tel.: þ1 614 292 7246; fax: þ1 614 292 9448. E-mail addresses: [email protected], [email protected] (E. Ghane). http://dx.doi.org/10.1016/j.watres.2014.10.039 0043-1354/© 2014 Elsevier Ltd. All rights reserved.

w a t e r r e s e a r c h 7 1 ( 2 0 1 5 ) 2 9 4 e3 0 5

success of the model. According to Sparks (2003), the appropriate method for determining the reaction rate law is to employ multiple initial (inflow) nitrate concentrations that vary considerably (within the range of application) to investigate whether the nitrate removal rate constant is independent of concentration. However, nitrate removal rate law determination using this method is scarce. Some studies have determined nitrate removal rate law based on a single inflow nitrate concentration (Gibert et al., 2008; Leverenz et al., 2010). Gibert et al. (2008) concluded that a zero-order rate law described their data, since linear regression fitted most of the treatment data (i.e., hardwood, mulch, compost, leaves and soil) rather well when using a single inflow nitrate concentration of 32 mg N L1. Leverenz et al. (2010) reported a firstorder rate law for nitrate removal in a pilot-scale unplanted woodchip subsurface flow constructed wetland with a high single inflow nitrate concentration of about 70 mg N L1. Other studies have assumed nitrate removal in denitrifying bioreactors to follow a zero-order (Robertson et al., 2000, 2009), or first-order rate law (Camilo et al., 2013; Moorman et al., 2015). Literature review revealed that only Robertson (2010) has evaluated nitrate removal rate law by employing multiple inflow nitrate concentrations in a small-scale laboratory experiment with retention times of 1 to 1.3 days. However, these conditions may not adequately represent natural conditions present in a denitrification bed due to differences in scale, hydraulic retention time, denitrifying community, etc. Therefore, there is need for determining the nitrate removal rate law under natural conditions. In addition to enhancing drainage water quality, denitrification beds may also have inadvertent adverse effects. During the biological nitrate removal process, incomplete denitrification and woodchip decomposition might generate greenhouse gases. Literature review indicated that only Woli et al. (2010) have reported surface emissions of nitrous oxide (N2O), a potent greenhouse gas, from a denitrification bed treating drainage water. Also, methane (CH4) surface emission from this system has not been evaluated previously. Therefore, there is a need to evaluate these potential adverse effects. An optimally designed denitrification bed is important for providing efficient nitrate removal. Since denitrification beds have recently been re-introduced for treating subsurface drainage water, optimal design is still under development. A denitrification bed model can help optimize the design of these systems by enhancing our understanding of how input

295

design parameters affect nitrate removal. Only Christianson et al. (2013a) have developed a semi-empirical model of a denitrification bed, but they did not include the effect of temperature on nitrate removal. Furthermore, they assumed that flow is governed by Darcy's law. However, we proved that Darcy's law is inadequate for describing water flow through woodchips (Ghane et al., 2014). Therefore, there is a need for a denitrification bed model that incorporates temperature, and employs the governing water flow and reaction rate equations. The objectives of this study were to (1) evaluate and quantify greenhouse gas emissions, (2) determine the reaction rate law of nitrate removal, and (3) model nitrate removal in a denitrification bed treating drainage water. The applied significance of this research is that it provides a mechanistic model for optimizing denitrification bed design, which in turn leads to enhanced drainage water quality.

2.

Materials and methods

We presented a detailed description of the site in Ghane et al. (2014). In brief, the polyethylene lined trapezoidal denitrification bed, hereafter referred to as bed, was installed in October 2011 at the Waterman Agricultural and Natural Resources Laboratory, Ohio State University, USA (Supplementary Image 1). The bed receives subsurface drainage water from a 1.5 ha rainfed agricultural field planted in corn and soybean during the 2013 and 2014 growing seasons, respectively. Nitrogen fertilizer was applied for corn before planting and before flowering at a rate of 13 and 185 kg N ha1, respectively. Water flows in and out of the bed via perforated drain pipes placed on the bed bottom at each end (Fig. 1).

2.1.

Tracer test

We conducted a tracer test at the site using potassium bromide (KBr) on September 11, 2013. A total of 500 g of KBr (335.7 g Br) was dissolved in 10 L of water and instantaneously poured in the inlet structure at 09:00. Thereafter, samples were collected from the outlet structure until 20:00. Sampling was spaced at 10 min intervals during the rise of the hydrograph, and then increased to 15, 30, 60, and 120 min after the peak to capture a high resolution hydrograph rather than equal interval sampling. Flow rate during the tracer test was

Fig. 1 e Diagram of the trapezoidal denitrification bed at the Waterman Agricultural and Natural Resources Laboratory.

296

w a t e r r e s e a r c h 7 1 ( 2 0 1 5 ) 2 9 4 e3 0 5

kept constant using a flow controller (described in section 2.2). We used a woodchip total porosity of 0.85 in the calculation of the nominal retention time based on the companion study of Ghane et al. (2014).

2.1.1.

Residence time

The tracer residence time (t) (actual retention time) at constant flow rate is calculated as P ti BCi Dti ty P BCi Dti

(1)

where ti is the time at the ith sample, BCi is the outlet bromide concentration, and Dti is the time interval between sampling (Kadlec and Wallace, 2009). The nominal (theoretical) retention time (tn) of a bed is calculated as tn ¼ (Vs nt)/Q, where Vs is the saturated volume of the bed, Q is the flow rate, nt is the total porosity of the woodchip media (Kadlec and Wallace, 2009). By replacing Vs ¼ L A and Q ¼ q A in the equation for tn, another form of the nominal retention time in hours is derived as tn ¼

Lnt 3600q

(2)

where L is the length (cm) of the water flow path (i.e., bed length), A is the mean cross-sectional area (cm2) that water flows through, and q is the specific discharge (cm s1). To quantify the extent of non-ideal flow, Thackston et al. (1987) defined volumetric efficiency (ev) as the ratio of t to tn where an ev value of 1 would indicate ideal plug flow. Actual hydraulic retention time (AHRT) in a bed can also be calculated from the ratio of the water travel time along the length of a bed (L) at pore water velocity (v) as AHRT ¼ L/v. Pore water velocity (average linear velocity or seepage velocity) is the actual water velocity in the woodchip media and is calculated as the ratio of specific discharge (q) to effective porosity (ne). Effective porosity (drainable porosity or specific yield) is the portion of the total porosity that contributes to flow through woodchips (Bear, 1988; Fetter, 2001). Thus, AHRT in hours is calculated as AHRT ¼

Lne 3600q

2.2.

Reaction rate law experimental setup

(3)

To determine the in-situ nitrate removal rate law, we used groundwater to represent subsurface drainage flow into the bed during September 2013. A Mazzei Injector model 584 (Mazzei Injector Company, Bakersfield, California) was employed to inject a potassium nitrate solution into the flow entering the bed, creating five inflow nitrate concentration runs of 3.1, 10.1, 14.9, 20.5, and 25.5 mg N L1 which were conducted on September 8, 15, 11, 10 and 9, respectively. These values were chosen since they are within the common range of nitrate concentrations in agricultural drainage water (Adeuya et al., 2012). A Kates Flow Controller model GB11T-A (W.A. Kates Co., Clawson, Michigan) was used to maintain steady-state flow and in turn a constant inflow nitrate concentration during each run.

Following the flow controller, groundwater flowed through an A1 Series Flow Meter model N100 (Great Plains Industries, Inc., Wichita, Kansas) where instantaneous flow rate was recorded before flowing into the inlet structure (Supplementary Video 1). The flow meter recording was verified by measuring the time to fill a known volume before initiation of each run. The duration of each run was from about 08:00 to 16:00 to allow enough time for nitrate removal to reach steady-state, except the September 11 run which continued until 20:00 for tracer testing. Water flow through the bed was maintained from about 07:00 to 08:00 each day to flush the nitrate from the previous run.

2.2.1.

Nitrate removal kinetics

Enzyme catalyzed reactions, such as denitrification, can be described by the MichaeliseMenten equation which has been used to model denitrification in the soil (Grant, 1991; Laudone et al., 2011). With nitrate as the substrate, this equation takes the common form (Fogler, 2011)

rNO3 ¼

Vmax Ci KM þ Ci

(4)

where rNO3 is the nitrate removal rate (mg N L1 h1), Vmax is the maximum removal rate (mg N L1 h1), KM is the MichaeliseMenten (or affinity) constant (mg N L1) that is the nitrate concentration at which nitrate removal rate is half the maximum rate, and Ci is the inflow (initial) nitrate concentration (mg N L1). Nitrate removal rate is the difference between the bed inflow and outflow nitrate concentrations (DC) divided by the time (AHRT) it takes for the nitrate removal to occur (i.e., rNO3 ¼ DC=AHRT). In the case of water flow through porous media (e.g., denitrification beds and subsurface flow wetlands), including effective porosity and saturated volume in the calculation of AHRT (Kadlec and Wallace, 2009) results in nitrate removal rate as rNO3 ¼ ðDC QÞ=ðVs ne Þ. It is important to note that rNO3 was not calculated for sampling events where outflow nitrate concentrations were zero (i.e., below detection limit). This is due to the possibility of underestimating rNO3 by using a longer AHRT than needed to get zero outflow nitrate concentration. This situation occurs when nitrate is completely removed from water before it reaches the bed outlet. At high nitrate concentration (Ci >> KM), reaction is apparent zero-order rNO3 yVmax

(5)

At low nitrate concentration (Ci

Modeling nitrate removal in a denitrification bed.

Denitrification beds are promoted to reduce nitrate load in agricultural subsurface drainage water to alleviate the adverse environmental effects asso...
1MB Sizes 0 Downloads 25 Views