Batch-Mode Denitrifying Woodchip Bioreactors for Expanded Treatment Flexibility

Abstract

Denitrifying woodchip bioreactors successfully remove nitrates from reverse osmosis desalinization brine. On-farm desalination plants only operate for several hours per day in batch mode, meaning bioreactors should also operate in batch cycles, although this type of bioreactor operation is relatively unstudied. This study compared two tests of three cycles of 24 h per week with two treatments each (Test 1 8 vs. 24 h, and Test 2 8 vs. 12 h). Cylindrical pilot-scale bioreactors were filled with 130 kg of citrus woodchips and an average of 322 L of brine. The results show that the treatments with longer saturation periods of 24 and 12 h exhibited higher removal rates under operational conditions (i.e., 8 h flooding based on a 24 h cycle) than the 8 h treatment. However, the nitrate removal rates of the 8 h treatment were higher under fill cycle conditions (i.e., 8 h flooding based on an 8 h cycle). Dissolved organic carbon liberated from the woodchips was greater in treatments with longer drying periods (i.e., treatments with shorter saturation periods). Batch bioreactors should be considered under applicable conditions to increase nitrate removal rates.

1. Introduction

Campo de Cartagena in southeast Spain grapples with an intricate interplay of agriculture and water quantity and quality. This region is a primary producer of water-intensive horticultural crops that serve markets throughout the rest of Europe [1,2]. Nevertheless, water scarcity in the area means irrigation water must be sourced from one or a combination of four options: the Tagus-Segura aqueduct, desalinated seawater, reused wastewater, and groundwater from the Quaternary aquifer. Groundwater withdrawals are the most cost-effective option for local farmers, but crop irrigation requirements necessitate salinized groundwater (≈4 to ≈6.5 dS m⁻¹) be either blended with freshwater or desalinized prior to use.

Historically, on-farm desalinization was performed by small reverse osmosis plants [3]. Generally, 70% of the treated groundwater was fit for irrigation use, while the remaining 30% was a concentrated brine high in salts and nitrate (e.g., electrical conductivity, EC ≈ 18 mS/cm; ≈50 mg of NO₃⁻-N/L; NO₃⁻-N = nitrate–nitrogen). This brine was discharged from farms into the receiving waters of the Mar Menor Lagoon. The Mar Menor is a highly sensitive and important protected waterbody [4]. Due to its unique characteristics, several protection categories establish its preservation [5,6]. In 2016, pressures on the lagoon, including the high-nitrate brine discharge, resulted in a highly politicized phytoplankton bloom [3]. As a result, on-farm desalinization systems were banned if the treatment system did not include a denitrification unit to reduce the nitrate concentrations in the discharge [7]. Research into simple and inexpensive denitrification systems, such as denitrifying woodchip bioreactors that could be paired with shuttered reverse osmosis plants, began in this region at that time [8,9].

Denitrifying woodchip bioreactors are water treatment systems that use a carbon source (such as woodchips) to reduce nitrate concentrations through enhanced denitrification. The woodchip carbon serves as the terminal electron donor in the microbially mediated reduction of nitrate to di-nitrogen gas during the anoxic process of heterotrophic denitrification. Generally, the carbon source is used to fill trenches, which are designed in such a way to promote the anaerobic conditions that facilitate microbial denitrification [10,11]. These treatment systems are nearly always designed and operated in a continuous flow mode rather than a batch treatment mode. However, the reverse osmosis plants in Campo de Cartagena generally withdraw groundwater overnight when electrical rates are lower. Thus, denitrification treatment in this application would feasibly be performed in batches based on daily operational cycles. In other words, one realistic treatment design consists of pumping and desalinizing groundwater during the night, followed by denitrification and releasing the treated brine during the day.

Nearly all woodchip bioreactors studied to date have used continuous flow mode due to the nature of subsurface drainage, which is the most common application for this type of bioreactor [10]. Only a few studies have compared the performance of batch versus continuous-mode bioreactors, and the findings are conflicting. Díaz et al. [12] documented better nitrate removal performance from a continuous flow versus batch reactor when the carbonaceous fill was almond shells. Wrightwood et al. [13] found that sequencing batch processes achieved nitrate reduction more rapidly than continuous flow systems, but their scale of treatment was very small (8 L). Otherwise, Weigelhofer et al. [14] found negative effects in the nitrate reduction after desiccation periods, but with high variability among the bioreactors studied.

Maxwell et al. [15] thoroughly tested the impact of what they termed “drying-rewetting cycles” in bioreactors, which would be similar to what could be experienced during batch-mode operation. They determined that nitrate removal can be enhanced by unsaturated periods as short as 8 h. However, not only nitrate removal is affected by these dry periods. In a pilot study, Abusallout et Hua [16] found an increase in Dissolved Organic Carbon (DOC) immediately after rewetting the woodchips by carrying out cycles of 3 dry days over 10 wet days. That increase in DOC declined and reached a stable concentration after 6 days of operation. Furthermore, Mardani et al. [17] found a higher E. Coli removal rate in denitrifying bioreactors with wet–dry flow conditions than in a continuous flow, but without significant differences.

In Campo de Cartagena, treated brine must contain < 19 mg of NO₃⁻-N/L for it to be discharged [18]. Achieving > 60% concentration reductions from the initial brine containing ≈50 mg of NO₃⁻-N/L, along with the unique operational demands in this desalination application, requires new treatment approaches for woodchip bioreactors. This work compared nitrate treatment for desalination brine in bioreactors operated with batch-mode saturation periods of 8, 12, and 24 h. This study was not intended to be a comparison of bioreactors operated using batch versus continuous flow modes but rather an exploration of nitrate removal under batch-mode operation, along with associated drivers such as organic carbon release and temperature. It was hypothesized that bioreactors with longer dry periods between treatment batches would exhibit better nitrate removal because of greater DOC release during the aerobic period.

2. Material and Methods

2.1. Experimental Setup

A denitrifying bioreactor pilot plant consisting of six replicate reactors was operated in batch mode between October 2018 and February 2019. The experimental setup was located at the Agri-Food Experimental Station Tomas Ferro (ESEA) (37°41′17.6″ N and 0°57′04.4″ W) of the School of Agricultural Engineering, Technical University of Cartagena (ETSIA-UPCT), Cartagena, Region de Murcia, Spain, and it was previously described by [9]. A 50 m-deep well provided the pilot plant groundwater (18 mg of NO₃⁻-N/L; EC = 6 mS/cm), which was desalinized with a reverse osmosis plant producing 68.7% fresh water and 31.3% brine by volume (capacity: 80 m³, 48 mg of NO₃⁻-N/L; EC = 18 mS/cm) and then denitrified in the bioreactors.

The brine was stored in an opaque tank for 48 h at ambient temperature to ensure continuous availability for the batch experiments. That tank supplied water to the six woodchip bioreactors. Each bioreactor consisted of a cylindrical above-ground fiberglass container (120 cm height, 80 cm diameter) filled to a depth of ~100 cm with locally sourced citrus woodchips (citrus sp.) (130 kg of woodchip per tank, air-dried weight). The long and thin woodchips averaged 36 mm in length and 5.2 mm in diameter (Table S1; Figure S1). The wood medium was a mixture of chipped heartwood with small branches and twigs from local citrus trees that would be used for this application (between 3 and 6 m in height). As stated by [8], the citrus woodchips were selected because they were the most efficient in denitrification among the studied media in the area (almond shells, chopped carob, olive bone, and citrus woodchips). The average pore volume was 322 ± 5.8 L when the bioreactors were filled (average fill depth: 100 cm, consistent with the woodchip depth). The average drainable porosity of the woodchips measured in the lab was 57%, but the average filled pore volume divided by the woodchip volume was 64% (given the empty bed volume of 502 L). Desalinization brine was introduced into the bioreactors by several 4 cm-diameter pipes arranged in a herringbone pattern at the bottom of each bioreactor; the flow orientation was upflow. Each bioreactor contained a vertical PVC piezometer (63 mm diameter) placed within the woodchip media with holes at 68 cm from the bottom of the bioreactor to allow water to enter for sampling.

2.2. Experimental Design and Sampling

Six cylindrical woodchip bioreactors were operated in batch mode. The experiment was divided into three phases: Pre-test (32 days: 8 October to 8 November 2018), Test 1 (38 days: 12 November to 20 December 2018), and Test 2 (38 days: 14 January to 21 February 2019) (

Table 1. Characteristics of different parameters of the experiment.

	Pre-Test	Test 1	Test 2
Length period (days)	32	38	38
Woodchips age at test start (days)	0	36	99
Treatments	none	8-h and 24-h fill	8-h and 12-h fill
Water temperature (°C)	20.2 ± 3.0	16.6 ± 2.3	14.8 ± 1.6
Average inflow of NO3−-N (mg/L)	48.3 ± 1.6	47.4 ± 2.0	46.4 ± 2.7

). The Pre-test was used to flush the woodchips so that Tests 1 and 2 would be more representative of long-term performance. This flushing period consisted of 350 ± 6 L per bioreactor per day for 32 days. The DOC concentration was checked weekly to determine the end of this flush period. A change of <4% in the DOC concentration between days 23 and 30 was used as a turning point to finish the flushing period (day 30: 30.8 mg DOC/L). Gaps between the three testing phases were due to labor availability and to allow time for overall system maintenance.

Each treatment (n = 3) consisted of three 24 h batch cycles per week. The treatments differed based on the time of saturation (or being flooded or filled with brine) within the 24 h periods. In Test 1, the “8 h fill” treatment bioreactors were flooded for 8 h, drained, and then remained unsaturated for 16 h. In other words, this treatment provided one batch of water for 8 h of denitrification treatment within a 24 h operational period. The second treatment during Test 1 consisted of a flood of 24 h, which comprised the entire 24 h operational period. These bioreactors were drained and refilled within 1 h for the subsequent 24 h batch-mode cycles the next day. For Test 2, the “8 h fill” treatment was the same 8 h treatment described in Test 1. The “12 h fill” treatment bioreactors were flooded for 12 h, drained, and then remained unsaturated for 12 h. Of the six pilot-scale bioreactors, the same three replicates were used for the 8 h treatments in both Tests 1 and 2, whereas the same replicates were used for the 24 h treatments in Test 1 and then the 12 h treatments in Test 2. The woodchips were not replaced between Tests 1 and 2.

The three 24 h operational cycles per week for all the treatments were performed consecutively on Monday, Tuesday, and Wednesday, starting each day at approximately 8 AM. On Thursday morning, after completing the third weekly cycle, all treatments remained unsaturated without brine for 96 h until the next weekly cycle began the following Monday. This mode of operation was designed based on expected operational guidelines for farmers in Campo de Cartagena [9], where the desalination plants work in batch mode mainly during the night due to the most affordable price of electricity.

The volume of brine added to the bioreactors during each batch was measured using a flow counter (Abering R100) at the bioreactor inflow. During the three batches, the brine was sampled before filling the bioreactor and after 30 min, 8 h, and 24 or 12 h from the beginning of each batch. A polyethylene sampler collected the sample from the piezometer inside the bioreactor after 30 min and 8 h. Before collecting the sample, the brine was removed from the well to refill the pipe with brine in contact with the woodchips. The 24 or 12 h samples were collected from the outlet of the bioreactors. The temperature was measured by inserting a calibrated multi-parameter instrument (Hanna HI 98194 pH/EC/DO Multiparameter) within the piezometer. Samples were filtered through Microsart CN-Filter filters (0.45 μm pore size).

2.3. Analyses and Statistics

Brine samples were analyzed for NO₃⁻-N with a double-channel chromatographic system using the 850 Professional Ion Chromatography Metrohm. The DOC concentration was only measured for samples collected from the second batch of each week (Tuesday) after 8 h of flooding, independent of the length of the batch due to cost (carbon analyzer TOC-V CSH Shimadzu; TOC: total organic carbon). The initial DOC concentration was not usually measured because the brine came from a desalination plant with a reverse osmosis process. However, it was measured once at the beginning of the study and was verified to be 2 mg DOC/L. All water chemistry analyses were performed at the Technological Research Support Service (SAIT) of the Technical University of Cartagena.

The performance of the bioreactors was evaluated by measuring NO₃⁻-N removal rates (NRRs), which were calculated for each batch run.

$$\mathrm{N}\mathrm{R}\mathrm{R}\left(\mathrm{g} \mathrm{N} {\mathrm{h}}^{-1}\right)=\frac{\left({\mathrm{N}\mathrm{O}}_3^{-}-\mathrm{N} \mathrm{i}\mathrm{n}\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}-{\mathrm{N}\mathrm{O}}_3^{-}-\mathrm{N} \mathrm{e}\mathrm{f}\mathrm{f}\mathrm{l}\mathrm{u}\mathrm{e}\mathrm{n}\mathrm{t} \mathrm{c}\mathrm{o}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\right)\times {\mathrm{V}}_{\mathrm{b}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{e}}}{\mathrm{t}}$$

(1)

where NO₃⁻-N influent concentration was the NO₃⁻-N in the initial brine of each batch (mg L⁻¹), NO₃⁻-N effluent concentration was the NO₃⁻-N in the effluent (mg L⁻¹), V_brine (L) was the volume of brine in each bioreactor in each batch, and t was the time of the nitrate measurement (h).

Assumptions of data normality and equality of variance were verified using the Shapiro–Wilk and Brown–Forsythe tests, respectively. Treatment comparisons across both tests (i.e., comparing four treatments) were made using an analysis of variance (ANOVA). After conducting the ANOVA, a Tukey test was implemented to find the significant difference among the means of the different treatments. The Tukey test is one such method for making these multiple comparisons.

Student’s t-tests were performed to compare the two treatments inside a given test. The data distribution is reflected by reporting the standard deviation and accompanying means. All statistical tests were performed in Sigma Plot version 15 (Systat Software, Inc., San Jose, CA, USA) and assessed with α = 0.05.

3. Results and Discussion

3.1. Nitrate Removal

The cumulative inflow load during Test 1 (261 g of NO₃⁻-N) was reduced by 62% and 88% by the 8 h and 24 h saturation treatments, respectively (Figure 1A;

Table 2. Mean ± standard deviation of nitrate removal metrics for the two test periods and batch-mode bioreactor treatments (n = 3). Nitrate removal rates were calculated by dividing by either the operational period of 24 h or by the filled periods (8, 12, or 24 h) for each treatment. Capital letters in a given column indicate differences between all four treatments (across both tests) at α = 0.05 (ANOVA); lowercase letters show the student’s t-test comparison between the two treatments within a given test at α = 0.05.

Test			Test 1 (from Day 36 to 73)		Test 2 (from Day 99 to 136)
Treatment			8 h	24 h	8 h	12 h
Influent load		Cumulative g NO3−-N	260.7 ± 0.5	261.8 ± 3.0	259.6 ± 5.5	260 ± 12.7
Effluent load			98.9 ± 1.0	31.7 ± 2.6	151.1 ± 7.8	134.3 ± 12.1
N removal		%	62 ± 0.3 bB	88 ± 1.0 aA	42 ± 1.8 bD	49 ± 2.3 aC
NRR	Operational cycle (24 h)	g N removed/h	0.37 ± 0.05 Bb	0.53 ± 0.05 Aa	0.25 ± 0.04 Db	0.30 ± 0.05 Ca
	Fill cycle (8, 12 or 24 h)		1.12 ± 0.10 Aa	0.53 ± 0.05 Cb	0.75 ± 0.10 Ba	0.60 ± 0.10 BCb

). Test 2 had a very similar inflow load of 260 g of NO₃⁻-N and a similar, but lower, treatment effect for the two saturation periods. More specifically, the 8 h and 12 h saturation treatments in Test 2 provided efficiencies of 42 and 49% (Figure 1B; Table 2). It is possible for intermediate products (nitrite, nitrous oxide) to be produced, given that this is a denitrification technology, but nitrous oxide production is generally less than 6% of the nitrate removed in woodchip bioreactors [11].

When the NRRs were calculated based on the 24 h operational cycle, the treatments with longer saturation periods (i.e., 24 or 12 h) exhibited higher removal rates than the 8 h treatments. Not only did the treatments with longer saturation periods have more time to remove nitrate (and thus, they did remove more nitrate mass), but all treatments were divided by the same 24 h period for this operational cycle assessment. The 8 h treatments were only actively treating water for one-third of that operational window. Nevertheless, this 24 h period provides a realistic comparison basis for treatments given that desalination plant operators in Campo de Cartagena pump groundwater nightly.

While the treatments with longer saturation periods in both tests exhibited greater mass N removal and efficiency, the fill cycle removal rates showed the opposite trend (Table 2). When the removal rates were calculated to represent more mechanistic conditions (i.e., saturated conditions of 8, 12, or 24 h) rather than operational cycles, the removal rates were higher for the short 8 h treatment in both tests (Table 2; Figure 2). Another way to consider this is that, in Test 1, the 8 h treatment removed 162 g of N in cycles of only 8 h saturation (Figure 1A). The additional 16 h of saturation every day provided by the 24 h treatment only yielded an additional 67 g of N removed (i.e., 229 g of N removed by the 24 h treatment; Table 2). While the NNRs for the lower operational cycle for the 8 h treatment might have indicated that the 16 h rest period was “wasted” from a treatment perspective, the higher fillable cycle NRRs (i.e., mechanistic NRRs) for these short treatments demonstrated the benefit of dry periods for nitrate removal in bioreactors. The aerobic release of labile carbon during the resting period, explored below (see Section 3.2), was hypothesized to be the cause of the higher fill cycle NRRs for the 8 h treatments.

Bioreactor design and operation should be based on local water quality criteria and discharge permits. Some water quality programs are based on allowable loadings (e.g., a Total Maximum Daily Load, TMDL, in the USA), whereas other programs use discharge criteria based on effluent nitrate concentrations (e.g., the National Pollution Discharge Elimination System, NPDES, in the USA; the Ministry for the Ecological Transition and the Demographic, in Spain). The Spanish regulation has some ordinances to protect water courses and waterbodies against diffuse pollution produced by nitrates. These regulations are part of the European strategy called the “European Green Deal” [19,20].

The relatively high inflow nitrate concentration of the brine of 42.7–51.8 mg NO₃⁻-N/L in this experiment may require a relatively longer saturation period to achieve effluent concentrations of <19 mg NO₃⁻-N/L, as required in Campo de Cartagena [18,19]. In this experiment, when the soaking period was 24 h, the 19 mg NO₃⁻-N/L discharge criterion was always met. However, this criterion was not consistently met during the shorter treatments. During Test 1, for 8 h of flooding, the requirements were met a total of 36 times in the bioreactors (67% of the time). For Test 2, neither treatment met the nitrate discharge requirements. These differences between Tests 1 and 2 illustrate that the flooding time (i.e., saturation period) is important for meeting concentration-based discharge criteria, but the most appropriate flooding periods vary with water quality (e.g., DOC, DO) and temperature.

3.2. Dissolved Organic Carbon

In both tests, the DOC concentration measured in the bioreactor solution at 8 h was greater in the treatments with shorter saturation periods (Figure 3A). In other words, the DOC liberated from the woodchips was greater in the treatments with longer drying periods (drying period of 16 h vs. either 0 or 12 h) (Figure 3A). Moreover, the 25 d resting period between the two tests resulted in a release of DOC from all six replicates, regardless of treatment. For example, the 8 h treatment DOC concentration of 15.5 mg C/L at the end of Test 1 (day 37) increased to 24.6 mg C/L at the beginning of Test 2 (Day 2; Figure 3A). These data further support the notion that the release of organic carbon, which is required for denitrification, occurs under aerobic conditions [13]. This effect can be explained due to the fact that aerobic decomposition is usually more efficient than anaerobic decomposition [15]. Also, subjecting woodchips to temporary exposure in unsaturated, aerobic conditions can significantly enhance denitrification in woodchip bioreactors. This enhancement is attributed to the increased production of more easily degradable and bioavailable carbon resulting from aerobic breakdown during periods of unsaturation.

While the NRRs discussed above allow practically applied and mechanistic comparisons (i.e., operational cycle and fillable cycle NRRs; Table 2), comparing the nitrate removal after 8 h of saturation from all the treatments most clearly demonstrated the impact of the resting periods and associated DOC release and nitrate removal, as we can see in the similar trend that Figure 3A,B have. When the DOC has high values, the NRE reaches a higher denitrification. The 8 h treatment in both tests removed nitrate more efficiently over the first 8 h of saturation than the 12 or 24 h treatments (Figure 3B). Although this treatment effect was observed for both tests, the difference between the two paired treatments was more notable for Test 1. The greater impact of the 8 h treatment in Test 1 versus Test 2 may have been due, in part, to the much longer drying period that this treatment had compared to its comparison treatment (i.e., the comparison treatment had 0 h drying in Test 1 and 12 h drying in Test 2).

Another reason for the differences between Tests 1 and 2 could have been the temporal effect of relatively new woodchips (despite the Pre-test flushing; see Section 2.2). New woodchips supply greater access to labile carbon to the microorganisms, achieving higher NRE under the same conditions as older woodchips [9]. The DOC concentration of the bioreactor solution decreased as both Tests 1 and 2 progressed, but the decrease was more linear during Test 1 (Figure 3A). In other words, the two Test 1 treatments showed a relatively consistent DOC concentration decrease with time, which is likely because labile carbon was still being released from the relatively fresh woodchips. The DOC concentration decrease with time for the two Test 2 treatments was likely more driven by the 25 d resting period between tests. The DOC concentrations in the bioreactor solution from day 9 onward for both treatments in Test 2 varied less by than 2.0 mg C/L, stabilizing at approximately 12–13.5 mg C/L (Figure 3A).

3.3. Water Temperature

The lower removal efficiency of 42 versus 62% for the 8 h treatment during Test 2 versus Test 1 was due to both woodchip age (i.e., carbon release; see above discussion) as well as temperature. The water temperature averaged nearly two degrees warmer during Test 1 versus Test 2 (Table 1; 16.6 versus 14.8 °C), with temperatures below 14 °C for several sample events during Test 2. While this was presumably a small difference compared to the range of other studies (from 4 to 20 °C [21], from 10 to 20 °C [22], from 4 to 30 °C [23]), this might have been a sufficient temperature difference to still impact nitrate removal. Temperature is known to be a strong driver of denitrifying bioreactor performance and emerges as a restricting factor as woodchips age [24].

A factor called Q₁₀ describes the temperature sensitivity of biological and chemical processes, defining how dependent a rate is on a 10 °C change. This factor is often in the range of 2.0–3.0 for many biological processes [25,26,27], including nitrate removal in denitrifying bioreactors [22,28,29]. This means that nitrate removal rates roughly double or triple for every 10 °C increase in temperature.

A Q₁₀ of 2.0 equates to a 200% increase in the removal rate for a 10 °C increase in temperature, and assuming this Q₁₀ effect is linear means that every 1 °C increase would result in a 20% increase in the rate. It follows that a Q₁₀ of 2.0 applied to the 2 °C change in temperature observed between Test 1 and 2 could account for a 40% difference in removal rates between the tests. The removal rate of the 8 h treatment in Test 1 was 150% of the rate achieved by the 8 h treatment during Test 2 (1.12/0.75 g N/h; Table 2). This simple Q₁₀ modeling of a 40% rate increase for a 2 °C increase could account for much of that 50% increase between tests.

3.4. Daily NO₃⁻-N Concentration

The nitrate concentration reductions within the first 30 min of saturation for all the treatments were rapid (Figure 4). For example, during the first three batches of Test 1 (days 36 to 38, Figure 4A–C), the nitrate concentrations decreased from 46.0 mg NO₃⁻-N/L in the inflow solution to 34.4 mg NO₃⁻-N/L (a total of 11.6 mg NO₃⁻-N/L) within the first 30 min of bioreactor flooding. The concentration decreased another 11.5 mg NO₃⁻-N/L over the subsequent 7.5 h of treatment (final concentration: 22.8 mg NO₃⁻-N/L at 8 h). This rapid initial N removal trend was observed for all the batches of both tests. The first three batches from each test are illustrated as examples in Figure 4. To our knowledge, this rapid removal has not been reported previously, but not many other studies have sampled so quickly after bioreactor saturation.

An alternative explanation for the rapid nitrate concentration declines (other than denitrification) could be the dilution of the brine due to the small volume of previously treated water potentially remaining in woodchip micropores. Microporosity, or secondary porosity, for woodchips has been reported at as high as 30% [30,31]. If these micropores retained water over the drying periods, this could equate to as much as 151 L of treated water per batch (30% × 503 L bioreactor volume). Given that the treatment volume was 322 L (drainable porosity of 64% × 503 L bioreactor volume), this possible dilution could account for some of the rapid decline in the nitrate concentration. Nevertheless, mass transfer into and out of the micropores is a relatively slow process in most bioreactor dual-porosity models [32], meaning dilution to this degree is unlikely to be the sole cause of the rapid nitrate concentration declines.

3.5. Possibilities for Multi-Batch Operation

Given the higher mechanistic (“fill cycle”) removal rates in the 8 h treatments, a benefit of batch-mode operation within a 24 h operational cycle is that one could feasibly treat multiple shorter batches at lower N removal efficiencies to maximize the overall daily mass N loading reduction. For example, in this testing configuration, two batches, both treated using 8 h saturation periods, could have been performed within a 24 h operational window. Treating two 8 h batches would still allow two 4 h rest periods per day to release labile C. The fill cycle nitrate removal rates from the 24 and 8 h treatments of 0.5 and 1.1 g/h, respectively (Table 2), yielded daily removals of 12 and 8.8 g NO₃⁻-N, given that one cycle of each treatment was performed daily. Hypothetically, nearly 18 g NO₃⁻-N/d could be removed if the 8 h saturation treatment was performed twice during the daily operational period (1.1 g removed/h × 2 batches of 8 h each).

This simple modeling assumed the same removal rate would be achieved for the second 8 h treatment batch as the first batch, following only a 4 h drying period. This was not tested here; the 8 h saturation treatment had a 16 h drying period in the current work. The authors of [33] tested drying periods of 2, 8, and 24 h and determined that the longest drying period best optimized nitrate removal. Thus, the multiple-batch idea proposed here would need further testing to finetune the balance between aerobic DOC release, nitrate removal, and woodchip longevity. Moreover, if the discharge criteria are based on concentrations rather than mass loading (e.g., 19 mg NO₃⁻-N/L in the Murcia legislation [18,19]), the batch saturation periods must be calibrated to achieve the appropriate concentration regardless of single or multiple-batch operation.

The concept of multiple daily batches could provide a particular benefit during periods of high irrigation water use during the summer. Larger summer irrigation water withdrawals could mean that one 24 h saturation batch may not provide the necessary water treatment volume (depending on the overall bioreactor volume). Options for high water use periods would be to build additional bioreactors (with associated expenses) or to optimize treatment using this multiple-batch idea. The high water demand in the summer corresponds to higher air temperatures, which would improve treatment efficiency even for shorter saturation periods [29,34].

4. Conclusions

The batch operation of denitrifying bioreactors, particularly in on-farm desalinization plants in Campo de Cartagena, shows promise. This study underscores the benefits of increased DOC release during dry periods between treatment batches, enhancing nitrate removal. Further research is needed to refine batch mode for woodchip bioreactors, considering overall volume requirements and potential impacts on the bioreactor lifespan due to increased aerobic degradation. The study emphasizes the importance of explicit design and performance objectives. While longer saturation periods achieved higher removal efficiency and mass removal, shorter 8 h saturation periods showed the highest mechanistic fill-cycle removal rates. Fine-tuning saturation periods is crucial, given the rapid nitrate removal observed within 30 min. Temperature was confirmed as a significant driver, potentially becoming limiting as woodchips age.