This case study was based on the dairy farm and manure processing facility described in VanderZaag and Baldé (2022). The farm was located in British Columbia, Canada, and utilized an anaerobic digester, SLS, and NR system. Emissions of CH₄ from the liquid storages were measured over one year. The present case study does not exactly reflect the real farm conditions; rather, the facility was used as a reference point providing general farm information as inputs for building the case-study. The farm had 375 milking cows and 350 replacement animals on site. The manure management system included an anaerobic digester (AD) receiving the manure and food wastes (source separated organics, fats-oils-grease) and producing upgraded and compressed natural gas. Digestate was first processed through a solid-liquid separator (SLS) to produce bedding for the cows. The liquid fraction was then processed into a nutrient rich “cake” and a liquid fraction “tea”, using a nutrient recovery (NR) system including a dissolved air flotation (DAF) system and a moving disc separator (Trident Processes Inc., Abbotsford, BC).

Three scenarios with different manure treatment systems were developed and analyzed for nutrients budgets, and GHG emissions. In Scenario A (SA), manure is processed by AD and the digestate is stored until field application. In scenario B (SB), the manure processing includes both an AD and a screw press (SLS). In this case digestate passes through a SLS, and the liquid fraction is stored, while the solid fraction is recycled back to the barn as bedding material. In Scenario C (SC), manure is processed by an AD, a SLS and a nutrient recovery (NR) system in series. The NR system includes a dissolved air flotation system and press that produces a solid fraction (“cake”) and the liquid fraction (“tea”) has very low solids and is stored. The flow chart of treatment scenarios is shown in Figure 1 .

Four transportation scenarios were considered for potential nutrient redistribution strategies, including a baseline reference scenario. The scenarios consider a farm that owns 250 ha of cropland adjacent to the barns, and has 100 ha of land located 30 km away. This is similar to the case study farm in the Fraser Valley, BC where available farmland is rare due to agricultural, commercial, and urban competition. Surplus manure in excess of crop requirements is sold to other farmers located at least 30 km away. The digestate application rate is limited by the N and P requirement for the crop, for which we used values of 300 kg N/ha and 60 kg P/ha (Kowalenko and Bittman, 2000; Sawyer and Mallarino, 2008). In the reference scenario (T0), unprocessed digestate of treatment SA is transported through a dragline pipe and directly applied on a local field of 250 hectare within 1 km distance. The remaining liquid manure is transported by truck to a 100-hectare field that is 30 km away from the farm for application.

The remaining three transportation scenarios are based on the nutrient flows from manure treatment scenario SC ( Figure 2 ) in which solid and liquid fractions are transported separately. In transportation scenario 1 (T1), the liquid fraction is transported and applied by dragline to the same local field as described in T0, in this case the local 250 ha field receives less than 58.7 kg P/ha and is assumed to utilize soil P reserves. The solid fraction is transported by truck to a distant fields located 30 km away. Surplus solid manure is sent to additional fields at the same distance (30 km) by truck. Local transportation and application of the liquid fraction (1 km) and distant transportation and application of the solid fraction (30 km) of transportation scenarios 2 (T2), and scenario 3 (T3), are the same as T1. For T2, surplus solid manure is transported to fields as far as possible by truck (farther than 30 km) while keeping the total transportation cost the same as T0. For T3, the surplus solid manure is transported the same distance as for T2, however, solid manure is transported by train rather than by truck. Transport costs were given in tonne × km, or tkm, the cost to transport one tonne a distance of one km. The unit transportation price (application cost not included) for dragline (liquid), truck for solid, truck for liquid, and train (solid) were $0.06/tkm for dragline, $0.28/tkm for truck, $0.04/tkm for train. Due to inflation these values are higher than those in previous studies (Fleming et al., 1998; Ribaudo et al., 2003; Mahmudi & Flynn, 2006).

Effects of the management scenarios on gaseous emissions and nutrient budgets were modeled using Dairy-CropSyst, a comprehensive dairy farm and manure management modeling tool developed by Khalil et al. (2019). The model was developed based on standards such as IPCC and ASABE, mass balance, and empirical equations which can be found in the Appendix A of Khalil et al. (2019). The Dairy-CropSyst model considers the impacts from manure treatment processes including AD, SLS, and NR to separate nutrients and fine solids. The model was developed based on integrating existing equations describing the biochemical transformations (e.g., C, N, P) and gas emissions (CO₂, CH₄, N₂O, NH₃), as well as performance parameters of manure treatment technologies. In this study, the default model settings for weather input were modified (R. Nelson, personal communication), to represent the target site described in VanderZaag and Baldé (2022). Site-specific inputs to the model included site location and dimensions, dairy characteristics, diet components, manure digestion, treatment, and storage processes ( Table 1 ).

At the facility described in VanderZaag and Baldé (2022), the manure stream was mixed with off-farm feedstocks (e.g. food waste) and the mixture was fed to the AD. The Dairy-CropSyst model, however, can only consider manure produced on the farm as the feedstock for AD. Therefore, to produce a similar mass flow of digestate, the number of dairy cows was doubled in the model simulations, to 750 head (N1; Table 1 ). This assumption of a manure-only digester likely changes the C, N, and P allocations compared to a digester receiving food-waste as co-feedstocks. The model simulated biogas production of 980 kg/d (CH₄), which is about 1/3 the amount produced at the case-study farm. The differences is likely due to the lower biogas potential of manure compared to off-farm co-feedstocks used at the case-study farm.

The model calculated urine and manure production and nutrients in manure slurry as functions of milk production and body weight (Khalil et al., 2019). Manure produced in the barn was assumed to be periodically scraped and flushed out from the floor with water in milking parlors (Khalil et al., 2019). Manure mass flow rates, total solid contents of manure (TS), nutrient concentrations of total nitrogen (TN), total phosphorous (TP), and total potassium (TK) of each manure treatment process were predicted for each liquid and solid stream. Default separation efficiencies (nutrient content in the solid fraction/liquid influent) of the coarse SLS separator are 40%, 12%, 10%, 5%, and 9% for TS, TKN, TON, TP, and TK, respectively. Nutrient concentrations of TN, TP, and TK in the liquid “tea” were estimated for the NR system in treatment SC using the separation efficiency of 36% for N, 85% for P, and 11% for K, respectively based on a previous study (Porterfield et al., 2020).

The Dairy-CropSyst model simulates emissions from the animals and barn surfaces including CH₄, N₂O and NH₃ (biogenic CO₂ is simulated but was excluded in our analysis as it is not a net GHG). Biogas produced by the AD was predicted by the model, and fugitive emissions were assumed to be 1% of the total CH₄ production (IPCC, 2019). Literature values of volatile solid (VS) to TS ratios are used to estimate VS contents for liquid and solid storage which are 71% and 91%, respectively (Fillingham et al., 2017; VanderZaag & Baldé, 2022). Direct and indirect N₂O (via NH₃) emissions from lagoon, bedding materials, and solid “cake” were predicted following the IPCC procedures (IPCC, 2019), specifically the Tier 2 method of methane emissions from solid and liquid manure management was used for estimating CH₄ emissions. Methane conversion factors (MCF) for solids (bedding and cake) was 2% (IPCC, 2019). The maximum methane-producing capacity of dairy manure (B₀) was assumed to be 0.24 m³ CH₄/kg VS (IPCC, 2019). Nitrogen loss due to volatilization of NH₃ (and NOx) as a fraction of total N excreted for uncovered anaerobic lagoon systems and solid storage systems are 0.35 and 0.30, respectively (IPCC, 2019). Emission factors (EFs) for direct N₂O emissions from solid and liquid manure storage are 0.02 and 0.001 kg N₂O-N/kg N, respectively (Kebreab et al., 2006).

According to estimates from Dairy-CropSyst the daily manure produced in the barn was 97.6 Mg/d with 8.12% total solids (TS) content. This is approximately double the observed manure production rate at the case study farm (VanderZaag and Baldé, 2022) which makes sense because we assumed the herd size was doubled to make-up for the lack of off-farm feedstocks. Simulated water flow out of the barn (89.65 Mg/d) was slightly greater than the total water excreted by cows (41.69 Mg/d) and water used in the parlor (44.18 Mg/d) and the difference was the amount of water consumed for other cleaning. The liquid manure was subject to further processing using different technologies ( Table 2 ). The initial nutrients in the out-flow stream leaving the barn contained 0.37% N, 0.06% P, and 0.17% K, and the N/P ratio was 6.1:1, which is similar to the observed ratio of 6.7:1 (VanderZaag and Baldé, 2022). Therefore, a total of 2890 Mg of TS per year was generated in the barn with annual production of N, P, and K of 131 Mg/yr, 22 Mg/yr, and 62 Mg/yr, respectively.

As manure was processed in the AD, organic matter (OM) was decomposed to produce biogas (CH₄, CO₂) and the mass of total solids in manure dropped from 7.92 Mg/d in barn out-flow to 4.45 Mg/d in AD effluent which was beneficial for reducing CH₄ emission during digestate storage ( Table 2 ). Parts of the organic N (TON) was converted to ammoniacal N (TAN) in the digester, but the total N (TN) was unchanged. Total P (TP) and total K (TK) were all retained in the digestate and flowed into the screw press separator.

After passing through SLS, 8.95 Mg/d solid mass was separated from the liquid fraction accounting for 9.5% of the total mass input and the TS content in the liquid fraction was decreased to 3.1%. According to the separation efficiencies of the screw press separator, the TN, TP, and TK remaining in the liquid fraction were 315.5, 56.5, and 154.3 kg/d, respectively. The N/P ratios of liquid and solid fractions were 16:1 and 5.6:1, respectively.

Finally, the nutrient recovery system separated 19.7 Mg/d of mass from the liquid manure to the solid fractions ( Table 2 ). The TS content in the liquid was further reduced to 1.4%, which was somewhat higher than the observed value at the case-study farm (0.8%, VanderZaag and Baldé, 2022). As a result of NR, the liquid “tea” contained only 64% of TN and 15% of TP, compared to the unseparated digestate. Conversely, the solid fraction contained 85% of the TP with only 8% of the mass of water. The majority of K remained in the liquid fraction and only 11% of TK was recovered into the solid “cake” by NR. In general, these simulated results are comparable to the observed values at the case-study farm (VanderZaag and Baldé, 2022). The simulated N/P ratio in the liquid “tea” was 23.8:1 which was significantly higher than that in the solid “cake” of 2.4:1, and this was because most of the P was recovered in the solid “cake” by NR treatment.

Simulated GHG emissions for each treatment scenario is listed in Table 3 . Emissions from the dairy barn were primarily in the form of CH₄ from enteric fermentation, which are constant across all manure treatment scenarios. In SA, enteric emissions account for less than half of the total emissions from the whole facility, owing to higher emissions from the manure processing and storage. In SC, enteric emissions account for the over 75% of the total due to lower emissions from manure processing. Fugitive emissions from AD are also constant across all scenarios and contribute <2% of the total CO2e in all scenarios.

In SA, the majority (59%) of the total GHG emissions were from the storage of digestate in the lagoon ( Table 3 ). These emissions were primarily due to CH₄, with indirect N₂O from ammonia emissions having a small contribution and direct N₂O being negligible ( Figure 3 ). Guest et al. (2017) also simulated that emissions from conventional manure storage tanks on a farm in Ontario were high and contributed more than half of the carbon footprint of the farm.

In SB, the addition of SLS led to reduced overall GHG emissions by 24% (including barn emissions), and reduced lagoon emissions by 45%. These reductions were due to less solids (carbon) entering the lagoon resulting in lower CH₄ emissions, offset by small increases in emissions from solid storage. Emissions from the solid fraction were mainly direct N₂O emissions; however, the contribution of N₂O from the solid fraction was small relative to the CH₄ emitted from the liquid fraction ( Figures 3 , 4 ). Thus, overall liquid and solid manure storage GHG emissions were 42% lower in SB compared to SA ( Figure 4 ).

In SC, the addition of NR technology further reduced overall GHG emissions by 44%, and lagoon emissions by 87%. The simulated CH₄ emissions from the lagoon were 83.8 kg/d on average, which is somewhat higher than the observed average of 56 kg/d at the case-study farm (VanderZaag and Baldé, 2022). This difference is reasonable given the differences in feedstocks to the AD, and the higher simulated TS in the liquid tea compared to observed. The simulated CH₄ emissions from the lagoon in SC were 75% lower than SB, and 87% lower than SA. Reduced emissions from the liquid fraction were due to much greater carbon and nitrogen content in the solid fraction. This, in-turn, led to higher emissions from the solid fraction. In fact, the solid fraction contributed 44% of the total storage emissions ( Figure 4 ), owing primarily to N₂O emissions from the solid cake ( Figure 3 ).

Annual cost of manure transportation and field application for separated liquid and solid fractions in scenario SC was evaluated and compared to transporting unseparated liquid digestate in SA. In the reference transportation scenario, T0, liquid was transported to a nearby 250-ha field through a dragline, and transported 30 km by truck to a 100-ha field to meet the P requirement. Surplus manure in excess of the P requirements of those 300-ha was applied to other fields 30 km-away ( Table 4 ). In this case the total cost of transporting manure for field application was $97,253, due to the relatively high cost of transporting water in the liquid fraction 30 km by truck.

The first transport scenario with separated fractions, T1, the liquid “tea” fraction stored in the lagoon was transported by dragline to the nearby 250 ha field, while the solid “cake” fraction was transported 30 km by truck. The total applied area of T1 was 30 ha greater than T0 to distribute the mass of P in the solid fraction according to P requirements of the soil ( Table 4 ). Nevertheless, despite transporting P to more land, the total transportation cost of T1 dropped more than 72% to $26,638 because the mass of water transported by truck was reduced by 8,071 tonnes compared to T0 (from 10,479 to 2,408). The dramatic cost reduction shows that dewatered solid manure offers a promising opportunity to reduce manure transportation cost by distributing water via dragline rather than by truck.

In the second transportation scenario, T2, the parameters of T1 were kept the same except that the transportation distance for the solid fraction was increased so that the total transportation cost was equal to T0. In this case the solid manure in excess of the requirements for 350 ha was transported by truck to 198 ha of land located over 150 km away.

In the third transportation scenario, T3, the parameters of T2 were kept the same, except that long distance transportation of the excess solid fraction was done by train instead of by truck. Since the transport by rail has a much lower cost than truck, the overall cost in this scenario was more than 75% lower than T0. Further calculation were done to maximize transport distance while maintaining the same transport cost from T0. In this case, the surplus solid fraction could be transported 1,100 km by train for the same overall cost as T0. This illustrates the potential for nutrient recovery to enable transport of manure P out of a manureshed with a high level of manure P, to another agricultural landscape where there is a P deficit (Reid et al., 2019).