|
Denitrifying Bioreactor In Situ Woodchip Bulk Density
Laura Christianson1,*, Reid Christianson2, Carolina Díaz-García1, Gabriel Johnson3, Bryan Maxwell4, Richard Cooke5, Niranga Wickramarathne1, Lowell Gentry1
Published in Journal of the ASABE 66(3): 723-734 (doi: 10.13031/ja.15364). Copyright 2023 American Society of Agricultural and Biological Engineers.
1University of Illinois, Urbana, Illinois, USA.
2Minnesota Dept. of Agriculture, St. Paul, Minnesota, USA.
3Department of Natural Resource Ecology and Management, Iowa State University, Ames, Iowa, USA.
4University of Illinois Urbana Champaign Prairie Research Institute, Champaign, Illinois, USA.
5Agricultural and Biological Engineering, University of Illinois, Urbana, Illinois, USA.
*Correspondence: lauraechristianson@gmail.com
The authors have paid for open access for this article. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License https://creative commons.org/licenses/by-nc-nd/4.0/
Submitted for review on 22 September 2022 as manuscript number NRES 15364; approved for publication as a Research Article and as part of the “Advances in Drainage: Selected Works from the 11th International Drainage Symposium” Collection by Associate Editor Dr. Mohamed Youssef and Community Editor Dr. Zhiming Qi of the Natural Resources & Environmental Systems Community of ASABE on 8 February 2023.
Highlights
Abstract. Woodchip bulk density in a denitrifying bioreactor governs system hydraulics, but this prime physical attribute has never been estimated in situ. The objectives were twofold: (1) to establish estimates of in situ woodchip bulk density at bioreactors in the field, and (2) evaluate causal factors for and resulting impacts of these estimates. Proof-of-concept bulk density methods were developed at a pilot-scale bioreactor using three ways to estimate volume: surveying the excavated area, pumping the excavation full through a flow meter, and using iPhone Light Detection and Ranging (LiDAR). These methods were then further tested at two new and three old full-size bioreactors. Additional ex situ (off-site) testing with the associated woodchips included analysis of bulk density along a moisture gradient and particle size, particle density, wood composition, and hydraulic property testing. In situ dry bulk densities based on the entire volume of the new bioreactors (206-224 kg/m3) were similar to values from previous lab-scale studies. In situ estimates for woodchips at the bottom of aged bioreactors (22-mo. to 6-y) were unexpectedly low (120-166 kg/m3), given that these woodchips would presumably be the most compacted. These low moisture-content corrected dry bulk densities were influenced by high moisture contents in situ (>70% wet basis). The impacts of particle size and particle density on bulk density were somewhat mixed across the dataset, but in general, smaller woodchips had higher dry bulk densities than larger, and several woodchips sourced from the bottom of bioreactors had low particle densities. Although dry bulk densities in the zone of flow in bioreactors in the field were shown to be relatively low, the resulting permeability coefficients under those packing conditions did not differ from those of the original woodchips. The LiDAR-based volume estimation method was the most practical for large-scale, full-size evaluations and allowed high precision with small features (e.g., vertical reactor edges, drainage fittings).
Woodchip bulk density in a denitrifying bioreactor fundamentally governs bioreactor flow properties and thus underpins nitrate removal performance. Bulk density is an ostensibly simple ratio of media mass divided by the volume the media occupies (Eisenbies et al., 2019). However, the bulk density of woodchips in operational denitrifying bioreactors in the field is unknown. Woodchip bulk density is not an input parameter in current design models, but an improved understanding of this physical attribute in situ would better guide flow and contaminant transport assessments for these engineered systems.
Current knowledge of woodchip bulk density for bioreactor applications stems from laboratory permeameter studies, which often report around 200 ± 25 kg/m3 dry bulk density (Ghane et al., 2014; Feyereisen and Christianson, 2015; Johnson et al., 2022). A variety of packing procedures, such as shaking the permeameter, tamping, or not tamping, have been used to simulate field compaction in the lab, although it is unknown how well these procedures mimic in situ conditions (Feyereisen and Christianson, 2015; Ghane et al., 2016; Cameron and Schipper, 2010). Johnson et al. (2022) recently documented that most woodchip types reach maximum compacted dry bulk density at 120% to 145% of their uncompacted dry bulk density. It would be useful to know (1) if woodchip bulk densities in operational bioreactors differ from those previously measured in the lab, and, if so, (2) if those differences are sufficient to impact bioreactor design model input parameters like drainable porosity.
The wood fuel industry provides significant insight on the topic of woodchip bulk density because this parameter impacts material transport, storage, and fuel quality (Kofman and Kent, 2007; Eisenbies et al., 2019). Wood fuel assessments routinely report both bulk density as received (or, wet bulk density; eq. 1) and dry bulk density (eq. 2; ISO, 2015):
(1)
(2)
where
BDar = bulk density as received (kg/m3)
m2 = mass of the filled container (kg)
m1 = mass of the empty container (kg)
V = volume of the measuring container (m3)
BDd = bulk density of the sample on a dry basis (kg/m3)
Mar = moisture content of the as-received sample on a wet basis percentage, expressed by mass (%).
Woodchip bulk density is impacted by a number of factors, including tree species and age; type of chipper and blade sharpness, particle size and shape as well as compressibility and bridging; settling, shocking, or tamping, and moisture content (Bohm and Hartman, 2005; Kofman and Kent, 2007; Mozammel et al., 2006; Gendek et al., 2016; Eisenbies et al., 2019).
Standard methods for bulk density determination of solid biofuels specify the use of a cylindrical shock-resistant container that is filled from a given height and dropped from a set distance several times (CEN, 2009; ASABE Standards, 2012; ISO, 2015). Container volumes of at least 50 L are recommended (Bohm and Hartman, 2005), but the International Organization for Standardization (ISO) and European standards also have a 5-L small container option for media with a nominal size up to 12 mm (CEN, 2009; ISO, 2015). A key difference between the main application for these bulk density methods (e.g., wood fuel transport) and the denitrifying bioreactor application is that bioreactor woodchips are intentionally saturated when operational. Thus, bulk density testing for woodchips in bioreactors needs to reflect those high moisture conditions.
Freshly felled trees have a moisture content of 40% to 60%, and wood storage piles generally dry to around 30% moisture (wet basis; Kofman, 2006). Wood in air-conditioned laboratories comes to equilibrium at much lower moisture contents of 6% to 10% (Glass and Zelinka, 2010). The concept of wood fiber saturation point occurs at moisture contents of approximately 18%-26% (wet basis; Bohm and Hartman, 2005). This is the point during drying where all the free water in the wood has been removed and only water in the cell walls remains (Glass and Zelinka, 2010). Below this point, wood shrinks and swells based on its moisture content, but above this point, wood is dimensionally stable. Equations 1 and 2 disregard moisture-content related shrinkage and expansion, and thus accurate comparisons are possible only when bulk density is measured on woodchip samples that are at similar moisture contents (ISO, 2015). This sensitivity of wood to both very low moisture contents (e.g., woodchips being tested in air-conditioned labs) and very high moisture contents (e.g., woodchips that best simulate bioreactor saturation) has not been explored in bioreactor applications.
Our main objective was to develop estimates of woodchip bulk density in situ in denitrifying bioreactors. Secondary objectives were to (a) explore factors influencing these in situ bulk densities (moisture content, particle size and density, age, and wood composition) and (b) assess if these in situ bulk densities change understanding of appropriate bioreactor design parameters.
Materials and Methods
Site Descriptions
Proof-of-concept bulk density estimation methods were developed at a pilot-scale bioreactor that was originally built in 2014 on the University of Illinois South Farms (table 1; LWD: 6.1 x 1.5 x 1.5 m; Urbana, Illinois, USA; Rendall, 2015). The bioreactor was recharged with new chipped woody debris from the campus Facilities and Services department in September 2019 (tables S1 and S2). In July 2021, this bioreactor was excavated in three lifts to trial in situ bulk density estimation methods.
Two new and three old full-size bioreactors (table 1: Bioreactors #9-11) were used to further test the bulk density estimation methods in situ. Woodchips from a third new bioreactor (table 1: Bioreactor #8) were used to additionally test related hypotheses in the lab. These bioreactors were systematically named to retain landowner privacy while maintaining consistency across publications.
The three new full-size bioreactors were constructed in Illinois between Nov. 2020 and Sept. 2021. Bioreactors #8 and #9 were designed by the United States Department of Agriculture’s Natural Resources Conservation Service (NRCS) according to the Conservation Practice Standard for denitrifying bioreactors (USDA NRCS, 2020). Bioreactor #9 was built as a part of the Illinois Farm Bureau Bioreactor Partnership in collaboration with the Illinois Land Improvement Contractors Association, the Illinois NRCS, and the University of Illinois. Bioreactor #10 was a unique trapezoidal design with the research aim of maximizing the flow treated (table 1). The entire volumes of Bioreactors #9 and #10 were examined for the in situ bulk density analysis (table 1).
Small portions (approximately 70 L) at the bottom of three six-year-old bioreactors were evaluated for in situ dry bulk density at private farm Bioreactors #11a, #11b, and #11c (table 1). This farm was unique in that it had six total NRCS-designed bioreactors constructed from 2015 to 2016. By 2021, these bioreactors exhibited slumping of more than ˜30 cm from the ground surface, so recharge of the bioreactors provided an ideal opportunity to evaluate in situ bulk density of woodchips at the bottom. During the recharge process, an excavator dug to the bottom of the bioreactors. During a pause in excavation, a person standing at bottom of the reactor was able to hand excavate an undisturbed portion of the bottom woodchips using 19 L buckets.
In Situ Bulk Density Testing
Pilot-Scale Bioreactor Proof-of-Concept Methods Development
The pilot-scale bioreactor was excavated by hand in three lifts; see the accompanying symposium proceedings paper for full details of excavation and volume estimation methods (Christianson et al., 2022). The volume of each lift decreased with depth to minimize the risk of side wall slumping. The depths of the three lifts relatively equally divided the bioreactor, with the top and bottom lifts representing predominantly unsaturated and saturated conditions, respectively. Woodchip moisture content inside the bioreactor was not documented during bioreactor operation, but the middle lift was intended to provide an intermediate data point between the unsaturated and saturated portions due to assumed capillary action. Woodchips were excavated in 19 L buckets, which were weighed immediately on-site (Ohaus Catapult 1000 floor scale).
| Table 1. Bioreactor study site descriptions with testing specifics. Not every test was performed at every bioreactor. L x W is bioreactor constructed length x width, where the excavated depth was 0.9-1.5 m. | |||||||||
| Bioreactor | Date Built | Date of In Situ Testing | L x W (m) | Drainage Area (ha) | Testing | ||||
| Particle Size, Particle Density, and Nutrient Analysis [a] | Bulk Density | Permeability Coefficient and Porosity | |||||||
| In situ: Portion of Bioreactor | In situ: Entire Bioreactor | Ex Situ | |||||||
| Pilot-scale bioreactor | 2014, recharged Sept. 2019 | Jul. 2021 | 6.1 x 1.5 | 7.3 | Original and 22-mo. chips | Survey, flow meter, LiDAR; cone penetrometer | --- | 22-mo. chips from top and bottom across moisture contents | Original and 22-mo. woodchips |
| Bioreactor #8 | Nov. 2020 | --- | 16.5 x 4.9 | 8.1 | Original chips | --- | --- | Original chips across moisture contents | --- |
| Bioreactor #9 | Aug. 2021 | Aug. 2021 | 18 x 4.6 | 19 | Original chips | --- | Survey, LiDAR | --- | --- |
| Bioreactor #10 | Sept. 2021 | Sep. 2021 | 9.1 x 13.7-7.6 (trapezoid) | 13 | Original chips | --- | Survey, LiDAR | --- | --- |
| Bioreactor #11a | 2015, recharged Sept. 2021 | Sep. 2021 | 15 x 7.9 | 31 | 6-y old chips | LiDAR | --- | --- | --- |
| Bioreactor #11b | 15 x 7.3 | 28 | |||||||
| Bioreactor #11c | 15 x 6.1 | 24 | |||||||
[a] Nutrient analyses were performed on all woodchips in this study; fiber analyses were performed on all except the pilot bioreactor middle lift and Bioreactor #10 woodchips. | |||||||||
Average oven-dried moisture content (wet basis) was determined gravimetrically based on three representative woodchip samples from each lift. The samples were delivered to the lab in air-tight bags within 6 h of excavation. For these and all the woodchips in this study, moisture content was determined by drying 25-70 g (resulting dry weight) in triplicate at 70°C until reaching a consistent weight. The final weight was measured after cooling the pans in a desiccator.
The three volume estimation methods used for each of the three lifts were: laser level surveying; pumping the excavated volume full through a flow meter; and capturing the excavation volume with an iPhone using Light Detection and Ranging (LiDAR; see full details in Christianson et al., 2022). The survey method consisted of surveying an elevation every 15 cm along six cross sections for each lift. The flow meter method involved laying plastic in the excavation and pumping it full of water through a flow meter (Pulsafeeder model MJ-SDC). The LiDAR method involved using the built-in LiDAR capability on an iPhone 12, a free image capture app (SiteScape, version 1.0.10), and opensource volume software (CloudCompare, version 2.11.3). A timelapse video illustrating the excavation and volume estimation process is online [https://youtu.be/hukm76gpgC4].
A cone penetrometer (Humboldt Digital Static Cone Penetrometer) was used on the surface of each lift (i.e., prior to it being excavated) as an additional indicator of compaction. Twelve penetrometer measurements were made on the surface of each of the three lifts.
Full-Size Bioreactor Testing
The woodchip supplier dump truck weigh-scale tickets were used as estimates of the total woodchip mass for the new Bioreactors #9 and #10. Samples of the woodchips were collected from the temporary on-site woodchip piles in air-tight bags between 1-3 days after woodchip delivery. Using moisture content-corrected weigh-scale masses at these two full-size bioreactors was a practical approach that merited assessment for these large bioreactor structures (Kofman and Kent, 2007), but it introduced some inherent uncertainty. Firstly, although moisture content varies across woodchip piles, it was not feasible to dig to the middle of the large piles for sample collection. Secondly, moisture content can change due to rainfall and air humidity between when the trucks are weighed at the supplier and when the woodchip samples were collected.
The entire excavation volume was measured at new Bioreactors #9 and #10 during construction using the laser level survey and LiDAR-based methods. The bioreactors were surveyed along three transects for both the length and width in a 3 m grid both before and after woodchip filling. To ensure adequate coverage and to capture bioreactor features, four LiDAR scans were completed before and two scans after the woodchip filling of each bioreactor. The flow-meter method was not used because it would have caused unacceptable contractor delay and there was no readily available water source.
Woodchips from the three aged bioreactors (Bioreactors #11a-c) were collected from the bottom of each using four 19 L buckets (˜90% full) during the process of recharging the woodchips. The woodchips were collected from the bottom 30 cm of the bioreactor after access had been gained with help from an excavator. The woodchips were transported to the lab in lidded buckets, where they were weighed the same day to start moisture content analysis. Only the LiDAR-based volume method was performed at these three bioreactors given the ease and speed of this method on-site and the non-uniform shapes of the small excavation areas.
Ex Situ Woodchip Testing
Particle Size and Nutrient/Fiber Composition
Particle size distribution of all woodchips, in triplicate, was assessed following standard methods using a sieve shaker (W.S. Tyler RX-812 Sieve Shaker; ASABE Standards, 1992). For each test, 200-300 g of air-dried woodchips were shaken for five minutes using a series of seven sieves ranging from 1.7 to 37.5 mm mesh sizes. Parameters of D10, D50, D60, D90 were interpolated from the particle size graphs, and uniformity coefficient (UC) was calculated as the D60 divided by the D10. Dx is the size at which x% of woodchip particles are smaller by mass.
All woodchips were analyzed, in triplicate, for total C and N using a combustion method at an external lab (Brookside Laboratories, Inc., New Bremen, OH, USA). Fiber analysis was performed, in triplicate, on three of the four pilot-scale bioreactor woodchip types (original, top lift, and bottom lift; not middle lift) and five of the six types from the full-size bioreactors (not from Bioreactor #10). Cellulose, hemicellulose, and lignin concentrations were determined on ground samples using sequential neutral and acid detergent fiber and acid detergent lignin methods (NDF, ADF, ADL; Ankom200 Fiber Analyzer and Daisy Incubator II Digestor; ANKOM Technology Corp., Fairport, NY). This method is commonly used for forages but is valid for any fiber-bearing material and previously has been used for wood (e.g., Musule et al., 2016; Cahyanti et al., 2021). Hemicellulose was calculated as the difference between NDF and ADF, and cellulose as the difference between ADF and ADL.
Bulk and Particle Density
Dry bulk density was estimated ex situ for the pilot bioreactor woodchips as a routine part of the permeability testing described in the next section. Dry bulk density was calculated by dividing the total woodchip mass, weighed by layer, and corrected for moisture content, by the filled volume of the permeameter (20 cm diameter; 77 cm total column height). The woodchips had an initial moisture content of 6.8% to 12% (wet basis) when packed, which was when the woodchip packed volume was measured for the bulk density calculation.
The 16 to 23 L woodchip-filled volume of the permeameter (20 cm diameter; 53-70 cm woodchip depth) exceeded the 5 L volume specified in the ISO and European bulk density standards for use with media smaller than 12 mm (CEN, 2009; ISO, 2015). Woodchip median diameters of the five types measured ex situ in the permeameter (table 1) ranged from 6.6-14 mm (table S1). Only the new Bioreactor #8 woodchip type was larger than the 12 mm noted in the standards. The 20 cm permeameter diameter was approximately ten times the largest particle size for four of the five types tested, which was generally consistent with the ASABE standard for bulk density (D90s: 16, 20, 20, 21, and 29 mm; ASABE Standards, 2012). However, the permeameter’s diameter to filled-height ratio of less than 0.5 was well below the recommended minimum of 1.25.
Both dry and as received (wet) bulk densities were assessed across a range of initial moisture contents for woodchips excavated from the pilot bioreactor top and bottom lifts as well as for the woodchips used to build Bioreactor #8 (table 1). Bioreactor #8 woodchips were selected as a reference standard (or “control”) for these moisture content gradient tests because they were NRCS-approved. During testing, 2.8 to 5.9 kg of woodchips (corrected oven dried weight) were hand mixed with 0-11.8 kg of water and left to equilibrate for 24 h. In other words, several batches of woodchips were mixed with varying amounts of water to produce woodchips with a gradient of moisture contents.
The range of initial woodchip moisture contents was intended to span from woodchips air-dried in an air-conditioned lab (<10% moisture content) to fully saturated woodchips excavated from a bioreactor (>70% moisture; achieved range: 6.9% to 76% wet basis). Woodchips for the wettest tests were fully submerged under water prior to packing in the permeameter. At least seven initial moisture contents for the three woodchip types were used to develop a relationship between initial moisture content versus both dry and as received bulk density. For these tests, individual moisture contents were not replicated (n = 1). For each test across the moisture content gradient, the permeameter was packed in five layers, and each layer was tamped ten times. This compaction was selected to simulate an expected maximum for the three woodchip types (Johnson et al., 2022) to allow a consistent comparison for the moisture content/bulk density evaluation.
To estimate particle density, 1 L beakers were filled, in triplicate, by dropping a given air-dried woodchip type from 30 cm above the rim and gently knocking the beaker bottom on the lab bench five times. The full beakers were saturated with water for 24 h and then topped up with water. Total porosity and beaker-based bulk density were estimated based on weight and packed volumes. Dry particle density was calculated as the moisture-content corrected beaker-based bulk density divided by the difference of one minus the total porosity expressed as a decimal.
Permeability Coefficient and Drainable Porosity
Permeability coefficient testing and drainable porosity measured in the permeameters (20 cm diameter; 77 cm total column height) followed standard methods and were illustrated by Johnson et al. (2022) (ASTM, 2019). In the previous study, permeability coefficient was referred to as saturated hydraulic conductivity, and the same method is used here for consistency in comparing results, although the new term (permeability coefficient) is used. This parameter was measured in triplicate on four woodchip types: the original woodchips used in the 2019 pilot-scale bioreactor recharge and woodchips excavated from the top, middle, and bottom lifts in 2021 (table 1). The woodchips were packed in three to five layers, and when tamping was performed, a 2.5 kg manual soil hammer was dropped onto the layer from a 31 cm height. The original woodchips were tested at three compaction levels of 0, 10, and 25 tamps per layer (following Johnson et al., 2022). The aim for the excavated woodchips was to estimate hydraulic properties resembling in situ field conditions, so the woodchips from the top, middle, and bottom lifts were each packed at a dry bulk density that simulated their estimated in situ dry bulk density. In other words, permeability coefficient was estimated at three compaction levels for the original woodchips but was only estimated at one unique compaction level for each of the three excavated woodchip types.
Once packed, the woodchips were saturated from the bottom for at least 24 hours prior to testing. Outflow from the permeameter was measured at ten increments of hydraulic head loss, ten times each. Permeability coefficient (referred to in some previous works as saturated hydraulic conductivity, Ksat) was calculated using Darcy’s Law following methods by Feyereisen and Christianson (2015) and Ghane et al. (2016) (i.e., natural log correction; viscosity correction). Darcian Law is assumed for flow through woodchips due to simplicity of practical bioreactor design. Drainable porosity was calculated from the mass of water drained from the columns over 24 h.
Analysis and Statistics
Assumptions of data normality and equality of variance were assessed using Shapiro-Wilk and Browne-Forsyth tests, respectively. Treatments or woodchip types were compared using Kruskal-Wallis analysis of variance (ANOVA) on ranks, followed by Dunn’s all pairwise multiple comparison procedure when equal variance assumptions were not met. When a particular dataset met the assumption of equal variance, a regular ANOVA followed by the Holm-Sidak all pairwise multiple comparison procedure was used. All statistical tests and regression modeling were performed in Sigma Plot version 14.0 (Systat Software, Inc., San Jose, CA) and assessed with a = 0.05.
Results and Discussion
Proof-of-Concept In Situ Bulk Density (Pilot-Scale Bioreactor)
Woodchip dry bulk densities for the top lift excavated from the pilot-scale bioreactor ranged from 237 to 268 kg/m3 across the three methods, whereas the middle and bottom lifts ranged from 135 to 166 kg/m3 across methods (table 2). The low values for the bottom lift were unexpected given that these woodchips would presumably have been the most compacted. The dry bulk densities estimated for the top lift trended higher, and the middle and bottom lift dry bulk densities trended lower than ranges published in bioreactor lab studies. For example, Ghane et al. (2014) measured dry bulk densities of 182-233 kg/m3 across new and 26-month-old woodchips. Johnson et al. (2022) showed a variety of commonly available woodchips generally had uncompacted dry bulk densities of 160-220 kg/m3 and reached maximums of approximately 190-260 kg/m3.
| Table 2. Woodchip excavation mass, volume, and in situ dry bulk density estimated using three methods for three lifts at a pilot-scale bioreactor. | ||||||||||
| Moisture Content In Situ (%) | Oven Dry Mass (kg) | Volume | Dry bulk density | |||||||
| Survey | Flowmeter | LiDAR | Survey | Flowmeter | LiDAR | |||||
| (m3) | (kg/m3 -) | |||||||||
| Top lift | 59 | 242 | 1.02 | 0.90 | 0.91 | 237 | 268 | 265 | ||
| Middle lift | 73 | 73 | 0.51 | 0.44 | 0.45 | 144 | 165 | 161 | ||
| Bottom lift | 75 | 44 | 0.32 | 0.27 | 0.26 | 135 | 164 | 166 | ||
Comparing the estimation methods showed the survey method consistently resulted in the highest volume (and thus lowest dry bulk density) for each lift, whereas the flow meter and LiDAR method were within 3% of each other (Christianson et al., 2022). Nevertheless, the three estimates for each lift were no more than 20% different in this proof-of-concept evaluation. This relative consistency between methods supported the accuracy of this overall in situ approach for bulk density, despite the surprising trend of the lowest dry bulk densities at the bioreactor middle and bottom. While each volume estimation method had limitations and advantages (see Christianson et al., 2022), the LiDAR method was the most practical for full-size evaluations given its speed in the field. The flow meter method was the most direct method of estimating volume, making it the easiest for off-site data processing, but it would be impractical for full-size bioreactor assessments.
The cone penetrometer provided supporting evidence that woodchip compaction decreased with depth, with means of 5.6 ± 1.7, 3.3 ± 1.9, and 1.3 ± 0.6 kg/cm2 for the top, middle, and bottom lifts, respectively. The bottom lift had significantly less resistance than the top and middle lifts (p = 0.037). Soil compaction assessment is the most common use of cone penetrometers, and for context, a general guiding value is that heavily compacted soil has a resistance of about 21 kg/cm2 (Arriaga, 2017). The woodchip resistances were much lower than that, but regardless, corroborated the in situ bulk density estimates with reduced compaction at deeper lifts.
Full-Size Bioreactor In Situ Bulk Density
The aged woodchips excavated from the bottom of the three six-year-old bioreactors had notably low dry bulk densities of 120-143 kg/m3 (table 3). These values ranged even lower than estimates from the middle and bottom lifts of the pilot-scale bioreactor (table 2). However, the dry bulk densities using the entire volumes of the newly constructed bioreactors of 206-224 kg/m3 closely aligned with the expected range from the literature of 200 ± 25 kg/m3 (table 3: Bioreactor #9 and #10; Ghane et al., 2014; Feyereisen and Christianson, 2015; Christianson et al., 2020). The survey and LiDAR-based volume estimations performed at both new bioreactors provided volumes within 8% of each other, which further confirmed the relative consistency between these two field methods. The LiDAR method provided the benefit of capturing small features with high precision (e.g., the drainage tee at Bioreactor #9 in fig. 1a) and easily capturing the unique trapezoidal shape of Bioreactor #10 (fig. 1b).
| Table 3. Woodchip mass, volume, and in situ dry bulk densityat five full-size bioreactors in Illinois, USA. Truck weigh-scale tickets supplied the wet mass for Bioreactors #9 and #10. Approximately 70 L portions at the bottom of Bioreactors #11a-c were excavated (not the entire bioreactor). | |||||||||||
| Full-Size Bioreactors | Moisture Content % | Wet Mass | Oven Dry- Corrected Mass | Volume | Dry Bulk Density | ||||||
| Survey | LiDAR | Survey | LiDAR | ||||||||
| kg | m3 | kg/m3 | |||||||||
| New | Bioreactor #9 | 33 | 46,350 | 31,240 | 144 | 139 | 217 | 224 | |||
| Bioreactor #10 | 31 | 40,040 | 27,590 | 134 | 124 | 206 | 222 | ||||
| Old | Bioreactor #11a | 76 | 38.6 | 9.1 | --- | 0.064 | --- | 143 | |||
| Bioreactor #11b | 77 | 38.1 | 8.9 | --- | 0.070 | --- | 127 | ||||
| Bioreactor #11c | 76 | 37.5 | 9.0 | --- | 0.075 | --- | 120 | ||||
 |
| Figure 1. iPhone Light Detection and Ranging (LiDAR) images of Bioreactors (a) #9 and (b) #10 created during construction. Panels are at different scales to show (1) the drain tee and (2) the black/white reference point at Bioreactor #9 and (3) inlet pipe manifold and (4) outlet control structure and manifold at Bioreactor #10. LiDAR images for the filled bioreactors were also required to calculate volume. |
Factors Influencing Bulk Density
Moisture Content
Woodchip bulk densities, both as received (or, wet) and dry, for the three woodchip types tested were significantly related to initial moisture content, albeit with directionally different trends (fig. 2; three types: new woodchips used in Bioreactor #8 and woodchips excavated from the top and bottom of the pilot bioreactor). The increase in bulk density as received with increasing moisture content was more intuitive than the decrease in dry bulk density at those same increasing moisture contents (fig. 2a vs. fig. 2b). For example, as increasingly wet woodchips were packed in the permeameter (moving from left to right in fig. 2a), the mass in the numerator of the as-received bulk density equation increased due to water weight. The nearly parallel slopes in figure 2a across wood types of varying age and size (table S1) illustrated this impact as a characteristic of water.
In the calculation for dry bulk density, the mass in the numerator would be relatively consistent when moving from left to right in figure 2b because this increased water mass was corrected for using the moisture content. Thus, the decreasing regressions in figure 2b were driven by the increasing volume required by water at increasing moisture contents. The volume of the wood would have reached maximum swelling (i.e., maximum volume) due to moisture at the fiber saturation point of approximately 18%-26% (Bohm and Hartman, 2005). However, water, as an incompressible fluid, requires space (i.e., volume) in the water/woodchip bulk matrix. As more volume was required by the water in the bulk matrix, the dry bulk density decreased correspondingly.
Regression analysis confirmed the lowest dry bulk densities would be expected at the highest moisture content for each woodchip type (fig. 2b; R2: 0.52-0.76). This was an important explanatory factor for the unexpectedly low dry bulk densities for the middle and bottom lifts excavated from the pilot bioreactor and for the woodchips excavated from the bottom of Bioreactors #11a-c. All five of those woodchip types had moisture contents greater than 70% (tables 2 and 3).
Particle Size
The D10 and D50 of woodchips excavated from the relatively unsaturated top of the bioreactor were significantly smaller than those excavated from the bottom (fig. 3a; table S1; p = 0.014 and 0.003 for D10 and D50, respectively, for top vs. bottom). Those top woodchips were also significantly less uniform than those from the bottom, with UCs of 7.0 ± 2.0 vs. 2.5 ± 0.3, respectively (p = 0.029). The woodchips from the middle and bottom lifts, where the treated flow occurred, were not significantly different from the original woodchips in size or uniformity (fig. 3a; table S1; p > 0.40).
The woodchips collected from the bottom of Bioreactors #11a-c were relatively small (fig. 3b; D50: 7.8-8.9 mm).
 |
| Figure 2. Moisture content (wet basis) of packed permeameters versus bulk density (a) as received and (b) dry for the new NRCS-approved woodchips used in Bioreactor #8 and the woodchips excavated from the top and bottom of the pilot-scale bioreactor. Regression slopes were all significant at a = 0.05. All points reflect consistent permeameter packing in 5 layers with 10 tamps per layer (n = 1). |
 |
| Figure 3. Particle size distribution for pilot bioreactor woodchips (a: original and three lifts after 22 mo.) and for (b) six full-size bioreactors in Illinois. The medium horizontal dash at 50% indicates the D50; small dashes at 10% and 90% indicate D10 and D90, respectively. Means of n = 3 shown. |
Their original particle size distribution was unknown, but these 6-year-old woodchips were notably smaller than the new woodchips used to build Bioreactors #8, #9, and #10 (D50: 14-18 mm) and were also less uniform (UC: 5.0-6.5 vs. 2.3-3.1 for old vs. new woodchips; table S1). Woodchip particle size decreases over time in bioreactors (Ghane et al., 2018; Christianson et al., 2020), as was likely for the Bioreactor #11 woodchips and the woodchips excavated from the aerobic top of the pilot bioreactor. However, Schaefer et al. (2021) reported the particle size distribution of woodchips deployed in a bioreactor for two years was similar to the original woodchips, and that was consistent with the lack of change in size here of the pilot bioreactor middle and bottom lifts after 22 months.
Mozammel et al. (2006) showed that 10-25 mm woodchips had a higher bulk density than woodchips greater than 25 mm in size. Consistent with that finding, the small top-excavated woodchips had higher dry bulk densities than the larger bottom woodchips when compared both in situ (table 2) and across the moisture content gradient (fig. 2; D50: 6.6 and 11 mm, respectively). Moreover, a given estimation method (survey or LiDAR) for the two new bioreactors consistently produced higher estimates for Bioreactor #9 versus #10 (table 3). Bioreactor #9 woodchips were slightly smaller than those of Bioreactor #10, with D50s of 16 vs. 18 mm and D90s of 25 vs. 32, respectively (fig. 3; table S1). Both initial woodchip batches were at relatively similar moisture contents, so this provided additional evidence supporting the role of particle size in bulk density.
 |
| Figure 4. Mean woodchip particle densities (a), lignin concentrations (b), and carbon: nitrogen ratios (c) from a pilot-scale bioreactor initially and after 22 months and three new (#8-#10) and three 6-year-old (#11a-c) full-size bioreactors (n = 3). Columns with the same lower case black letter (pilot bioreactor comparison), uppercase red letter (new media comparison), or uppercase blue italics letter (aged media comparison) are not significantly different. Pilot-bioreactor Middle lift and Bioreactor #10 woodchips were not analyzed for fiber (“na”). |
The particle size impact was not entirely consistent across the full dataset, however. The very small, 6-year-old woodchips from Bioreactors #11a-c exhibited the lowest dry bulk densities estimated in this study (<150 kg/m3, table 3). While their high moisture content undoubtedly contributed to these low values, the small woodchips excavated from the top of the pilot-scale bioreactor also had a similarly high in situ moisture content, at least relative to the fiber saturation point (moistures of >75 and 59%, respectively). Thus, given these two types’ similar small sizes and moisture contents but very different in situ bulk densities, other factors such as age and position within the bioreactor may have been at play.
Particle Density and Wood Composition
Woodchips excavated from the bottom of the pilot bioreactor had significantly lower particle density than the original woodchips and woodchips from the top and middle lifts (fig. 4a, lowercase black letters; bottom vs. original: 695 ± 8 vs. 749 ± 35 kg/m3). This corroborated the qualitative observation that these woodchips felt light like Styrofoam after air drying in the lab. However, like particle size, the effect of particle density on in situ bulk density was mixed. For example, the pilot bioreactor middle lift and the bottom of Bioreactor #11b both had low in situ bulk densities, but these woodchips had relatively high particle densities. Bioreactor #11b woodchips appeared “dirtier” than #11a or #11c woodchips, so it was likely that sediment on the Bioreactor #11b woodchips contributed to the high particle density of 962 ± 5 kg/m3 (e.g., soil particle density ~ 2600 kg/m3).
Woodchip bulk density is influenced by the basic density of the tree species (Kofman and Kent, 2007). Woodchips for Bioreactor #8 were a hickory (genus Carya)/maple (Acer L.) blend, whereas Bioreactor #9 woodchips were a maple/walnut (genus Juglans) blend; the species of Bioreactor #10 woodchips was not available. Hickory has a very high wood density, and here Bioreactor #8 woodchips had a significantly higher particle density than Bioreactor #9 or #10 woodchips (means: 951, 868, and 860 kg/m3, respectively).
Bioreactor #8 and the top-excavated woodchips had similar maximum dry bulk densities at low moisture contents (i.e., similar y-intercepts in fig. 2b), despite Bioreactor #8 woodchips being twice as large as the top woodchips (D50: 14 and 6.6 mm, respectively; table S1). A notable similarity between these two woodchip types was their high lignin concentrations (fig. 4b; table S2). The divergence in regression lines at increasing moisture content indicated other differences existed in the bulk matrix of the aged, small top lift woodchips versus the new, large, hickory Bioreactor #8 woodchips.
Wood nutrient composition (e.g., C:N ratio) may be related to in situ woodchip bulk density, given the media with the lowest C:N ratios were sourced from bioreactor locations that exhibited the lowest bulk densities (pilot bottom and Bioreactors #11a-c; fig. 4c; table S2). The Styrofoam-like woodchips excavated from the bottom of the pilot bioreactor had been presumably “hollowed out,” but the C content only decreased from 47.5%C in the original woodchips to 44.3%C (table S2). Thus, the decrease in C:N ratio for this wood type from 114 to 39 was driven by changes in N content (table S2). Wood nutrient composition likely plays a correlation rather than a causation role in bioreactor bulk density, but this relationship, along with age and moisture content, is worth noting for this biological material.
Bioreactor Design Parameters Influenced by Bulk Density
Woodchips from the pilot-scale bioreactor bottom lift, which was the predominant zone of flow when in situ, had a mean permeability coefficient of 2.7 ± 0.85 cm/sec when they were packed at 162 ± 2.2 kg/m3 dry bulk density (fig. 5a; attempting to simulate 135-166 kg/m3, table 2). This was very similar to the 2.9 cm/s currently used in the Illinois NRCS bioreactor design model (Illinois NRCS, 2021). Comparing the values here with the design value used by the NRCS is the most valid comparison for design purposes, but other reported woodchip conductivities most commonly range from approximately 2-8 cm/s depending on particle size and compaction (recently summarized by Johnson et al. [2022]). High variability around each permeability coefficient mean (i.e., low testing precision) meant the only significant difference when all six means were compared was that the original woodchips at their lightest compaction (3.6 ± 0.74 cm/sec at 184 ± 6.8 kg/m3) had a significantly higher permeability coefficient than the three lowest values (means =1.2 cm/sec at >210 kg/m3; p = 0.014).
 |
| Figure 5. Mean dry bulk density versus permeability coefficient (a) and drainable porosity (b) for the pilot-scale bioreactor original and 22-month-old excavated woodchips (n = 3 for each point). Drainable porosity versus permeability is shown in (c). Original woodchips were packed at three compaction levels (0, 10, and 25 tamps per layer), whereas the 22-month-old woodchips were packed to simulate the average in situ dry bulk density for each of the three lifts (i.e., attempted to simulate values from table 2). |
Drainable porosity in the permeameters averaged 38 ± 0.8%, 42 ± 1.7%, and 45 ± 1.6% for the top, middle, and bottom lifts, respectively, when woodchips from each were packed to simulate their in situ dry bulk densities (fig. 5b). The measured drainable porosities trended lower than the Illinois NRCS’ default design values of 47 and 53% in the presence or absence of a soil cover, respectively (Illinois NRCS, 2021). This supported previous findings that compacted woodchips may have drainable porosities lower than 47% (e.g., a mean of 44%, Johnson et al., 2022). However, the lower measured values may also reflect the impact of media type, particle size, and age. Others have reported drainable porosity percentages ranging from the mid-forties to the mid-fifties for fresh woodchips (Cameron and Schipper, 2010; Feyereisen and Christianson, 2015).
Permeability coefficients and drainable porosities of media with similar particle sizes packed at similar densities would be expected to be relatively similar. This was generally shown to be the case here, with the permeabilities and drainable porosities of the bottom and middle lift woodchips not significantly different from those measured for the original woodchips when packed at their lowest dry bulk density. However, given the nearly identical particle size of those three chip types, it was somewhat surprising that the older chips did not better align with the original woodchips (fig. 5). It is also important to recognize that statistical testing additionally indicated the hydraulic properties of the aged woodchips did not significantly differ from the most compacted woodchips. This reflected methodological limitations with these naturally heterogenous media and highlighted that there may be an incorrectly assumed precision in the use of a single permeability value in design models.
Summary and Conclusions
This study presented the first estimates of in situ dry bulk density for woodchips in denitrifying bioreactors. Bottom portions of 22-mo. to 6-y old bioreactors had low bulk densities of 120-166 kg/m3, which shed new light on this prime physical attribute of bioreactors in the field. New bioreactors exhibited dry bulk densities of 206-224 kg/m3, which validated the most common range previously reported in lab-scale permeameter tests. The LiDAR-based volume estimation method paired with either weigh-scale tickets at new bioreactors or weighed buckets during bioreactor excavation was a fast and relatively easy field method for in situ bulk density determination.
Moisture content was an explanatory factor in the low in situ dry bulk densities of wet excavated woodchips. Particle size and particle density helped explain some additional bulk density differences between woodchip types. However, because these factors did not explain all the variability across the dataset, future in situ and ex situ tests are recommended for some of the supplemental hypotheses presented (e.g., impact of wood composition, woodchip placement within a bioreactor, bioreactor age).
This study was not specifically designed to explore relationships between bioreactor longevity and in situ woodchip bulk density. Although aged woodchips were studied, the same assessments were not made using the original woodchips used to construct those bioreactors; this precluded true systematic comparisons over time. Nevertheless, the development of these in situ methods, especially the relatively easy and reliable LiDAR-based method, could be viewed as a first step toward better understanding. It is hoped future testing of these methods is undertaken, but it is important to note these methods are necessarily destructive. That is, woodchip volume estimation is best suited to be performed either during initial bioreactor construction or during a recharge. Consecutive annual bioreactor excavations to develop a time-series relationship with in situ bulk density would likely create preferential flow paths, which themselves could impact woodchip physical properties.
This new understanding of in situ bulk density did not result in a recommendation for modifying the default woodchip bioreactor design values. The permeability coefficients of woodchips excavated from the middle and bottom of the 22-mo. old pilot-scale bioreactor did not significantly differ from either the highest or lowest measured (3.6 and <1.2 cm/s, respectively). This illustrated the limited testing precision for woodchips, which may be a weakness in current design models that are sensitive to this parameter.
Acknowledgments
The authors thank the many funders who made this large-scale work possible, with primary funding from the USDA NRCS (NR185A12XXXXC004 CESU under the Great Rivers Umbrella Agreement 68-3A75-18-518 504). Additional funding for specific study locations was provided by: Illinois Nutrient Research and Education Council (#2017-4-360498-302; #2022-GENTRY: “Do soil caps enhance bioreactor performance and increase woodchip longevity?”); USEPA grant number 00D87719; USDA NCR-SARE FNC21-1279; and the Illinois Farm Bureau Bioreactor Partnership (Illinois Farm Bureau, Illinois Land Improvement Contractors Association, Illinois NRCS). CDG’s time was funded by a USDA NRCS CIG (NR213A750013G038) and IL NREC (#2021-3-360498-144). This product was developed with support from the Sustainable Agriculture Research and Education (SARE) program, which is funded by the USDA-NIFA. Any opinions, findings, conclusions, or recommendations expressed within do not necessarily reflect the view of the SARE program or the USDA. Mention of companies and manufacturers’ names does not imply endorsement by the USDA. The USDA is an equal opportunity provider and employer. Most importantly, we thank the four private farmers involved who graciously allowed us to do this work.
Supplementary Information
| Table S1. Mean ± stdev woodchip particle size distribution parameters illustrated in figure 3 (n = 3). Pilot-scale bioreactor parameters were statistically compared where means within a given column with the same letter are not significantly different. UC is Uniformity Coefficient. | ||||||
| Particle size parameters (mm; UC is unitless) | ||||||
| D10 | D50 | D90 | UC | |||
| Pilot-scale bioreactor | Original chipped woody debris used in 2019 recharge | 4.2 ± 0.8 a | 10 ± 0.7 a | 21 ± 2.4 | 2.8 ± 0.4 b | |
| Top lift | 1.3 ± 0.5 b | 6.6 ± 1.0 b | 16 ± 0.7 | 7.0 ± 2.0 a | ||
| Middle lift | 3.1 ± 1.5 ab | 9.5 ± 1.0 a | 20 ± 1.1 | 4.1 ± 1.9 ab | ||
| Bottom lift | 4.8 ± 0.9 a | 11 ± 0.8 a | 20 ± 4.0 | 2.5 ± 0.3 b | ||
| Full-size – New | Bioreactor #8 | 5.3 + 0.5 | 14 + 2.0 | 29 + 4.5 | 3.1 + 0.2 | |
| Bioreactor #9 | 7.2 + 0.2 | 16 + 0.2 | 25 + 0.3 | 2.5 + 0.1 | ||
| Bioreactor #10 | 8.9 + 0.7 | 18 + 1.5 | 32 + 4.5 | 2.3 + 0.2 | ||
| Full-size – Old | Bioreactor #11a | 1.7 + 0.7 | 7.8 + 1.6 | 22 + 3.7 | 6.5 + 1.8 | |
| Bioreactor #11b | 2.1 + 0.6 | 8.3 + 1.3 | 21 + 4.7 | 5.0 + 1.1 | ||
| Bioreactor #11c | 2.0 + 0.4 | 8.9 + 1.2 | 26 + 4.2 | 5.5 + 0.7 | ||
| Table S2. Mean ± stdev woodchip nutrient and fiber content (n = 3). These data were previously included in a wood media database survey analysis by Christianson et al. in 2022. | ||||||||||
| Nutrient Concentration (%) | Fiber Concentration (%) | |||||||||
| C | N | P | C: N | Lignin | Cellulose | Hemicellulose | ||||
| Pilot-scale bioreactor | Original chipped woody debris used in 2019 recharge | 47.5 ± 0.2 | 0.42 ± 0.02 | 0.026 ± 0.002 | 114 ± 5.0 | 23 ± 1.9 | 46 ± 2.0 | 13 ± 0.4 | ||
| Top lift | 39.6 ± 1.7 | 0.51 ± 0.10 | 0.027 ± 0.003 | 80 ± 20 | 39 ± 3.6 | 42 ± 4.6 | 5.7 ± 2.5 | |||
| Middle lift | 44.4 ± 0.9 | 0.72 ± 0.09 | 0.028 ± 0.003 | 62 ± 8.4 | Not available | |||||
| Bottom lift | 44.3 ± 0.9 | 1.15 ± 0.14 | 0.066 ± 0.017 | 39 ± 5.2 | 29 ± 2.8 | 35 ± 5.9 | 18 ± 4.4 | |||
| Full-size – New | Bioreactor #8 | 43.7 ± 2.2 | 0.21 ± 0.01 | 0.015 ± 0.001 | 212 ± 14 | 50 ± 7.8 | 24 ± 9.0 | 15 ± 1.2 | ||
| Bioreactor #9 | 45.3 ± 0.2 | 0.35 ± 0.08 | 0.018 ± 0.003 | 138 ± 36 | 29 ± 1.2 | 45 ± 1.7 | 18 ± 1.1 | |||
| Bioreactor #10 | 46.0 ± 0.3 | 0.33 ± 0.05 | 0.015 ± 0.001 | 140 ± 18 | Not available | |||||
| Full-size – Old | Bioreactor #11a | 41.5 ± 1.3 | 1.29 ± 0.08 | 0.067 ± 0.009 | 32 ± 1.1 | 25 ± 1.8 | 40 ± 0.9 | 14 ± 2.0 | ||
| Bioreactor #11b | 36.6 ± 1.5 | 1.53 ± 0.26 | 0.075 ± 0.006 | 24 ± 3.2 | 25 ± 1.9 | 38 ± 1.8 | 12 ± 0.4 | |||
| Bioreactor #11c | 41.6 ± 2.6 | 1.19 ± 0.13 | 0.046 ± 0.007 | 35 ± 4.0 | 24 ± 1.8 | 49 ± 2.0 | 14 ± 0.6 | |||
References
Arriaga, F. J. (2017). Proper use of cone penetrometers for detecting soil compaction. College of Agricultural and Life Sciences, University of Wisconsin-Madison, and University of Wisconsin-Extension, Cooperative Extension.
ASABE Standards. (1992). S424.1: Method of determining and expressing particle size of chopped forage materials by screening. St. Joseph, MI: ASABE.
ASABE Standards. (2012). S269.5: Densified products for bulk handling - Definitions and methods. St. Joseph, MI: ASABE.
ASTM. (2019). D2343-19: Standard test method for permeability of granular soils (constant head). West Conshohocken, PA: ASTM Int.
Böhm, T., & Hartmann, H. (2005). Bulk density determination of solid biofuels. Landtechnik, 60(3), 158-159.
Cahyanti, M. N., Doddapaneni, T. R., Madissoo, M., Pärn, L., Virro, I., & Kikas, T. (2021). Torrefaction of agricultural and wood waste: Comparative analysis of selected fuel characteristics. Energies, 14(10), 2774. https://doi.org/10.3390/en14102774
Cameron, S. G., & Schipper, L. A. (2010). Nitrate removal and hydraulic performance of organic carbon for use in denitrification beds. Ecol. Eng., 36(11), 1588-1595. https://doi.org/10.1016/j.ecoleng.2010.03.010
CEN. (2009). EN 15103:2010 - Solid biofuels - Determination of bulk density. European Committee for Standardization.
Christianson, L. E., Christianson, R. D., Díaz-García, C., Johnson, G. M., Maxwell, B., Cooke, R. A., & Gentry, L. (2022). Digging in to denitrifying bioreactor in situ bulk density. Proc. 11th Int. Drainage Symp. (pp. 1-6). St. Joseph, MI: ASABE. https://doi.org/10.13031/ids.202200020
Christianson, L. E., Feyereisen, G. W., Hay, C., Tschirner, U. W., Kult, K., Wickramarathne, N. M.,... Soupir, M. L. (2020). Denitrifying bioreactor woodchip recharge: Media properties after nine years. Trans. ASABE, 63(2), 407-416. https://doi.org/10.13031/trans.13709
Eisenbies, M. H., Volk, T. A., Therasme, O., & Hallen, K. (2019). Three bulk density measurement methods provide different results for commercial scale harvests of willow biomass chips. Biomass Bioenergy, 124, 64-73. https://doi.org/10.1016/j.biombioe.2019.03.015
Feyereisen, G. W., & Christianson, L. E. (2015). Hydraulic flow characteristics of agricultural residues for denitrifying bioreactor media. Appl. Eng. Agric., 31(1), 89-96. https://doi.org/10.13031/aea.31.10552
Gendek, A., Aniszewska, M., & Chwedoruk, K. (2016). Bulk density of forest energy chips. Ann. Warsaw Univ. Life Sci. - SGGW, Agric., 67. Retrieved from https://www.researchgate.net/publication/322643959
Ghane, E., Fausey, N. R., & Brown, L. C. (2014). Non-Darcy flow of water through woodchip media. J. Hydrol., 519, Part D, 3400-3409. https://doi.org/10.1016/j.jhydrol.2014.09.065
Ghane, E., Feyereisen, G. W., & Rosen, C. J. (2016). Non-linear hydraulic properties of woodchips necessary to design denitrification beds. J. Hydrol., 542, 463-473. https://doi.org/10.1016/j.jhydrol.2016.09.021
Ghane, E., Feyereisen, G. W., Rosen, C. J., & Tschirner, U. W. (2018). Carbon quality of four-year-old woodchips in a denitrification bed treating agricultural drainage water. Trans. ASABE, 61(3), 995-1000. https://doi.org/10.13031/trans.12642
Glass, S. V., & Zelinka, S. L. (2010). Chapter 4: Moisture relations and physical properties of wood. In Wood handbook wood as an engineering material. Centennial ed. General Technical Report FPL-GTR-190. (pp. 4.1-1.19). Madison, WI: USDA, Forest Service, Forest Products Laboratory.
Illinois NRCS. (2021). Denitrifying bioreactor design Microsoft Excel fillable spreadsheet. Retrieved from https://www.nrcs.usda.gov/resources/guides-and-instructions/field-office-technical-guides
ISO. (2015). ISO 17828:2015 Solid biofuels - Determination of bulk density. Retrieved from https://www.iso.org/standard/60687.html
Johnson, G. M., Christianson, R. D., Cooke, R. A., Díaz-García, C., & Christianson, L. E. (2022). Denitrifying bioreactor woodchip sourcing guidance based on physical and hydraulic properties. Ecol. Eng., 184, 106791. https://doi.org/10.1016/j.ecoleng.2022.106791
Kofman, P. D. (2006). Quality wood chip fuel. COFORD Connects, Harvesting / Transportation, 6. Retrieved from http://www.woodenergy.ie/media/coford/content/publications/projectreports/cofordconnects/finalfuelquality.pdf
Kofman, P. D., & Kent, T. (2007). Harvesting and processing forest biomass for energy production in Ireland: The ForestEnegy 2006 Programme. Dublin, Ireland: COFORD.
Mozammel, H., Shahab, S., Tony, B., Sudhagar, M., Ladan, J., Lim, J., & Afzal, M. (2006). Interaction of particle size, moisture content and compression pressure on the bulk density of wood chip and straw. Proc. CSBE/SCGAB 2006 Annu. Conf. St. Joseph, MI: ASABE. https://doi.org/10.13031/2013.22060
Musule, R., Alarcón-Gutiérrez, E., Houbron, E. P., Bárcenas-Pazos, G. M., del Rosario Pineda-López, M., Domínguez, Z., & Sánchez-Velásquez, L. R. (2016). Chemical composition of lignocellulosic biomass in the wood of Abies religiosa across an altitudinal gradient. J. Wood Sci., 62(6), 537-547. https://doi.org/10.1007/s10086-016-1585-0
Rendall, T. J. (2015). Effect of passive and active heating on the performance of denitrifying bioreactors. MS thesis. Urbana, Illinois: University of Illinois at Urbana-Champaign, Department of Agricultural & Biological Engineering.
Schaefer, A., Werning, K., Hoover, N., Tschirner, U., Feyereisen, G., Moorman, T. B.,... Soupir, M. L. (2021). Impact of flow on woodchip properties and subsidence in denitrifying bioreactors. Agrosyst. Geosci. Environ., 4(1), e20149. https://doi.org/10.1002/agg2.20149
USDA, NRCS. (2020). Conservation Practice Standard Denitrifying Bioreactor (Code 605). Washington, DC: NSDA, NRCS. Retrieved from https://www.nrcs.usda.gov/