Biogas Potential of Coffee Processing Waste in Ethiopia

Primary coffee processing is performed following the dry method or wet method. The dry method generates husk as a by-product, while the wet method generates pulp, parchment, mucilage, and waste water. In this study, characterization, as well as the potential of husk, pulp, parchment, and mucilage for methane production were examined in biochemical methane potential assays performed at 37 ◦C. Pulp, husk, and mucilage had similar cellulose contents (32%). The lignin contents in pulp and husk were 15.5% and 17.5%, respectively. Mucilage had the lowest hemicellulose (0.8%) and lignin (5%) contents. The parchment showed substantially higher lignin (32%) and neutral detergent fiber (96%) contents. The mean specific methane yields from husk, pulp, parchment, and mucilage were 159.4 ± 1.8, 244.7 ± 6.4, 31.1 ± 2.0, and 294.5 ± 9.6 L kg−1 VS, respectively. The anaerobic performance of parchment was very low, and therefore was found not to be suitable for anaerobic fermentation. It was estimated that, in Ethiopia, anaerobic digestion of husk, pulp, and mucilage could generate as much as 68 × 106 m3 methane per year, which could be converted to 238,000 MWh of electricity and 273,000 MWh of thermal energy in combined heat and power units. Coffee processing facilities can utilize both electricity and thermal energy for their own productive purposes.


Introduction
Coffee production is a livelihood for about 125 million people worldwide, particularly from developing countries [1]. Ethiopia is known to be the origin of and gene pool for coffee Arabica [2]. In the last decade, Ethiopia has been the largest coffee producer in Africa, and it remains fifth in the world, contributing a share of about 4.5% to the world production. Annual coffee production increased from 273,400 Mg in 2007 to 469,091 Mg in 2016, while the cultivation area increased from 407,147 ha to 700,475 ha ( Table 1). The annual green bean production has increased in the last 4 consecutive years, but productivity (yield per harvest area) has declined in the same period. The amount of coffee by-products is directly related to coffee production. Coffee is Ethiopia's leading export commodity; and the livelihood of more than a million households depends on coffee production [3]. Coffee made up about 24% of the country's total export earnings for the fiscal period 2012/13 [4]. Ethiopian coffee is produced under forest, semi-forest, garden, and plantation production systems contributing 10,35,50, and 5% of the country's coffee production, respectively. Thus, about 95% of Ethiopia's coffee is produced by small holder farmers [2,4]. Adapted from www.faostat.org database [5].
In Ethiopia, the main regional states involved in coffee production are Oromia, Southern Nations Nationalities & People (SNNP), and Gambella. As of 2013/14, there were 1026 wet and 696 dry milling stations in these regions. About 60% of the wet milling and 79% of dry milling stations were located in SNNP and Oromia regions, respectively. Coffee milling stations are owned by private companies, cooperatives, or states. About 75% of the wet and 96% of the dry mills are owned by private firms. Cooperatives own 23% of wet and 3% of dry stations (Table 2) [5]. Unlike the dry milling stations, the majority of the wet mills are not connected to the electric grid; they are run on diesel engines for pulping and further processing steps. Replacing fossil fuels by bioenergy could be an option with ecologic and economic advantages.  Oromia  297  95  15  407  524  20  6  550  957  SNNP  470  146  616  136  4  -140  756  Gambella  3  --3  6  --6  9  Total  770  241  15  1026  666  24  6  696  1722 Coffee cherries are collected from the coffee trees by selective harvesting (picking only the ripe fruits by hand) or strip harvesting (fruits striped at once with different maturity levels). The cherries are then processed to green beans following either the dry method or the wet method ( Figure 1). In the wet method, the cherries undergo pulping, fermentation/washing, and peeling/polishing one after the other before producing the green beans in 7-12 days. The respective by-products are pulp (43% w/w), mucilage (12% w/w), and parchment (6.1% w/w) on the intrinsic fresh weight basis of coffee cherries [6]. On the other hand, the dry method uses hulling after drying the cherries to produce green beans. Husk is the major single by-product of the dry method. It represents about 50% w/w of the dry cherry. The dry method often requires up to 4 weeks from harvest to green beans [7]. Despite the higher water requirement (about 80 L kg −1 green beans) and disposal problems, green coffee beans from wet processing have a superior aroma and are sold with higher premiums [8]. Coffee by-products are dumped within the hosting community with virtually no economic benefits, thus causing severe environmental problems. The by-products are known to be rich in organic pollutants (e.g., proteins, sugars, and pectin), tannins, and phenolic compounds harmful to plants, humans, and aquatic biota [9,10]. Globally, coffee processing generates about 15 × 10 6 Mg (dry weight basis) of coffee residues, of which 9.4 × 10 6 Mg is pulp [11]. Biogas technology was introduced to Ethiopia five decades ago. The technology is envisaged as a de-centralized energy source for improved livelihood and a source of organic fertilizer. The Ethiopian national biogas program installed about 8,000 household biogas units (4 to 10 m 3 ) in 2009-2013 [12,13]. Most of the biogas digesters are fed with cow dung or latrine waste. By-products of coffee processing, which are currently discarded as waste, could be an alternative feedstock.
The rates and extent of bio-degradation are crucial in anaerobic fermentation of agricultural residues, which in turn depend on lignocellulose contents and properties [14]. Higher cell contents (protein, carbohydrates, and fats) tend to ferment easily and result in higher methane production [15]. The complex interaction of hemicellulose and lignin often leads to a reduced cellulose hydrolysis. However, in a first approximation, lignin content in substrates has been proven to be a good predictor of methane potential from agro-industrial wastes [15,16].
The biogas yield from coffee processing by-products has been investigated by several researchers, but the bio-methane yield potentials reported vary substantially [17]. Ulsido, et al. [18] found a methane production potential of 132 L kg −1 VS from husk, while Kivaisi et al. [19] reported 650 L kg −1 VS of Robusta and 730 L kg −1 VS of Arabica mixed solid wastes (husk and pulp). Similarly, Baier and Schleiss [20] determined a biogas potential of 380 L kg −1 VS (57-66% methane) and 900 L kg −1 VS of pulp using batch assay and semi-continuous digesters, respectively. Adams and Dougan [21] reported the suitability of anaerobic conversion of coffee pulp and the waste water for waste treatment, and the generation of useful fuel, as high as 66 m 3 per ton of pulp digested. Variation in the bio-methane potentials of substrates could be attributed to the variety of coffee waste investigated and mode of fermentation. Chanakya and De Alwis [1] summarized several research and field trials on coffee processing wastes. Different types of reactors (batch, CSTR, UASB, and BIBR) were tested at varying scales of operation.
On the other hand, Franca and Oliveira [7] pointed out that, despite the bio-methane potential, coffee by-products are comparable to common agricultural residues, and that the lack of scientific Biogas technology was introduced to Ethiopia five decades ago. The technology is envisaged as a de-centralized energy source for improved livelihood and a source of organic fertilizer. The Ethiopian national biogas program installed about 8,000 household biogas units (4 to 10 m 3 ) in 2009-2013 [12,13]. Most of the biogas digesters are fed with cow dung or latrine waste. By-products of coffee processing, which are currently discarded as waste, could be an alternative feedstock.
The rates and extent of bio-degradation are crucial in anaerobic fermentation of agricultural residues, which in turn depend on lignocellulose contents and properties [14]. Higher cell contents (protein, carbohydrates, and fats) tend to ferment easily and result in higher methane production [15]. The complex interaction of hemicellulose and lignin often leads to a reduced cellulose hydrolysis. However, in a first approximation, lignin content in substrates has been proven to be a good predictor of methane potential from agro-industrial wastes [15,16].
The biogas yield from coffee processing by-products has been investigated by several researchers, but the bio-methane yield potentials reported vary substantially [17]. Ulsido, et al. [18] found a methane production potential of 132 L kg −1 VS from husk, while Kivaisi et al. [19] reported 650 L kg −1 VS of Robusta and 730 L kg −1 VS of Arabica mixed solid wastes (husk and pulp). Similarly, Baier and Schleiss [20] determined a biogas potential of 380 L kg −1 VS (57-66% methane) and 900 L kg −1 VS of pulp using batch assay and semi-continuous digesters, respectively. Adams and Dougan [21] reported the suitability of anaerobic conversion of coffee pulp and the waste water for waste treatment, and the generation of useful fuel, as high as 66 m 3 per ton of pulp digested. Variation in the bio-methane potentials of substrates could be attributed to the variety of coffee waste investigated and mode of fermentation. Chanakya and De Alwis [1] summarized several research and field trials on coffee processing wastes. Different types of reactors (batch, CSTR, UASB, and BIBR) were tested at varying scales of operation.
On the other hand, Franca and Oliveira [7] pointed out that, despite the bio-methane potential, coffee by-products are comparable to common agricultural residues, and that the lack of scientific studies hinders the wider utilization of the by-products for technical or economic reasons. Jayachandra, et al. [22] reported that the acidic pH and polyphenols in the coffee husk makes it resistant to anaerobic conversion. However, the treatment of husk with thermophilic fungus (Mycotypha) resulted in suitable pH levels for anaerobic conversion and thus increased the bio-methane yield considerably. Ensiled coffee pulp/husk presents an ideal method to preserve the substrate for longer periods of use in anaerobic digesters and reduces caffeine content (13-63%), total polyphenols (28-70%), and condensed polyphenols (51-81%) [23]. Seasonal availability of coffee by-products could be a limitation for continuous anaerobic fermentation.
Summary of bio-methane yields from typical anaerobic substrates is listed in Table 3. Municipal wastes tend to produce more methane than other substrate types. Despite frequent use of animal manure as biogas substrate, the methane production rate was relatively low. Table 3. Methane yield potential of different substrates common to anaerobic digesters.

Substrates
Methane Yield (L kg −1 VS) Reference In 2015, Ethiopia had an annual electricity production of 10.08 TWh. The overall installed capacity was 2.7 GW, in which hydro, fossil fuel, and other renewables contributed 79.5%, 7.5%, and 13% of the installed capacity, respectively [27]. As of 2016, 85.4% of the urban and 26.5% of the rural population of Ethiopia had access to electricity [28].
To the authors' knowledge, no comprehensive study on the characterization and bio-methane potential of coffee husk, pulp, parchment, and mucilage has been published in the scientific literature. Electrical and thermal potentials (in Ethiopian context) of the bio-methane from husk, pulp, and mucilage were not reported either.
The objective of this study was to examine the physico-chemical characteristics and determine the anaerobic bio-methane potential of coffee husk, pulp, parchment, and mucilage generated from primary coffee processing in Ethiopia. Furthermore, the potential of the bio-methane from husk, pulp, and mucilage to generate electrical and thermal energy was estimated.

Raw Materials
Coffee husk, pulp, parchment, and mucilage were collected from Gomma-2 estate coffee farm (7 o 55 16.32" N, 36 o 37 06.62" E) about 400 km South-West of Addis Ababa, Ethiopia. The pulp was collected right after pulping of the cherries, from consecutive processing days of a month. The pulp then was spread on a plastic sheet for 3-4 days of sun drying. The pulp was shuffled in regular intervals, in order to dry evenly. The mucilage was fetched from a second fermentation pit, poured on a plastic sheet, and left to dry under the sun for 5-6 days. The parchment and husk were collected from the same farm, without the need for further drying. All samples underwent a size reduction to pass 1 mm pores, packed in polyethylene bags, and shipped to the University of Hohenheim, Germany. The samples were stored in a cool and dry place until further use.

Inoculum
The inoculum was obtained from the State Institute of Agricultural Engineering and Bioenergy, University of Hohenheim. It consists of dairy manure and energy crops pre-digested in laboratory reactors, cultivated under the institute's standard procedure for bio-methane potential (BMP) assays [29]. Prior to the BMP assay, the inoculum was digested at 37 • C to degas and allow temperature adaptation of the anaerobic bacteria for further experimental assays. The inoculum had a total solids and volatile solids content of 5.0% and 61.6% (of TS), respectively.

Anaerobic Batch Digestion Tests
The Hohenheim Biogas Yield Test (HBT) is a laboratory batch method used to determine the methane and biogas yield potential of substrates [30], which is mentioned in the German Engineers Association (VDI) Guideline 4630 [31], as one of six recommended methods for BMP assays. The trial was carried out using 100 mL calibrated glass syringes as fermenters (Figure 2). The syringes were fitted to a motorized rotor to enable continuous mixing of the samples. The fermentation was operated for duration of 35 days at a temperature of 37 • C. At the beginning of the process, 30 g of inoculum was placed inside a glass syringe with 0.4 g of ground and dried samples [29,31]. Reference samples (hay and concentrate) of known methane yield were used in order to verify the quality of the inoculum and the reliability of the digestion process [29]. The blank (zero variant) fermenter was filled with about 50 g of inoculum alone (Table 4). Biogas and methane yields were obtained by deducting the proportional biogas and methane yields from the zero variant. The biogas volume was measured according to the volume difference before and after emptying the gas inside the syringe. While emptying the syringes, biogas was injected into the methane sensor (Advanced Gasmitter, Pronova, Berlin, Germany), which was calibrated with ambient air and gas (~60% CH 4 ). The fermenter temperature (mostly 2 • C less than the digestion temperature) and ambient air pressure were measured during the gas measurement and used to normalize the gas yield at STP (101.325 kPa, 273 K). The specific methane yield (L CH 4 kg −1 VS added) is the net normalized methane yield divided by the amount of organic dry matter of substrate added.  The degree of degradation (η ) was evaluated according to the mass ratio of gas produced m to volatile solids loaded (Equation (1)).
in which m is the dry mass of substrate added and c is the concentration of volatile solids in the substrate. The molar masses of methane (M and carbon dioxide (M were used to determine the mass of the biogas (Equation (2)) based on their respective concentrations in the biogas, assuming the biogas was composed of methane and carbon dioxide only [31,32].
in which Vbiogas is the volume of biogas measured, c is the concentration of carbon dioxide in the biogas, and c is the concentration of methane in the biogas. The energy recovery (ER) rate was determined by the ratio of calorific values of the methane generated to the calorific value of input substrates in the batch assay. The gross energy (GE) of substrates can be estimated using results from Weende feed analysis and applying the Equation ( in which XP is crude protein, XL is crude fat, XF is crude fiber, NfE is nitrogen free extract (NfE) (g kg −1 , dry mass basis), and GE is gross energy (MJ kg −1 ). The specific energy content of component parameters has certain differences, as shown in the fixed coefficients of the equation. The theoretical energy potential of by-products from coffee processing was calculated under the following scenario: (1) that all the husk, mucilage, and pulp generated in a harvest season is recovered and utilized in biogas digesters (the parchment was excluded from calculation), (2) the electric conversion efficiency in a CHP is 35% electricity and 40% thermal energy that could be used for coffee drying, and (3) the lower methane heating value of 9.968 kWh m −3 is used to calculate the primary energy yield.  The degree of degradation (η oDM ) was evaluated according to the mass ratio of gas produced m biogas to volatile solids loaded (Equation (1)). η oDM = m biogas m·c VS (1) in which m is the dry mass of substrate added and c VS is the concentration of volatile solids in the substrate. The molar masses of methane (M CH 4 ) and carbon dioxide (M CO 2 ) were used to determine the mass of the biogas (Equation (2)) based on their respective concentrations in the biogas, assuming the biogas was composed of methane and carbon dioxide only [31,32]. 4 22.4·100 (2) in which V biogas is the volume of biogas measured, c CO 2 is the concentration of carbon dioxide in the biogas, and c CH 4 is the concentration of methane in the biogas. The energy recovery (ER) rate was determined by the ratio of calorific values of the methane generated to the calorific value of input substrates in the batch assay. The gross energy (GE) of substrates can be estimated using results from Weende feed analysis and applying the Equation (3) [33]. GE = 0.0239 MJ g −1 · XP + 0.0398 MJ g −1 ·XL + 0.0201 MJ g −1 ·XF + 0.0175 MJ g −1 ·NfE (3) in which XP is crude protein, XL is crude fat, XF is crude fiber, NfE is nitrogen free extract (NfE) (g kg −1 , dry mass basis), and GE is gross energy (MJ kg −1 ). The specific energy content of component parameters has certain differences, as shown in the fixed coefficients of the equation.
The theoretical energy potential of by-products from coffee processing was calculated under the following scenario: (1) that all the husk, mucilage, and pulp generated in a harvest season is recovered and utilized in biogas digesters (the parchment was excluded from calculation), (2) the electric conversion efficiency in a CHP is 35% electricity and 40% thermal energy that could be used for coffee drying, and (3) the lower methane heating value of 9.968 kWh m −3 is used to calculate the primary energy yield.
Statistical analysis was made using Microsoft Excel (2013) to determine the mean and standard deviation of SMY of substrates and other parameters related to physico-chemical characterization.

Chemical Composition
There were certain differences in the chemical composition between coffee husk, pulp, parchment, and mucilage ( Table 5). The parchment showed the highest proportion (76.9%) of crude fiber contents, followed by husk (39.9%), pulp (24.8%), and mucilage (19.4%). NDF, ADF, and ADL contents in parchment were also considerably higher than in husk, pulp, and mucilage. Non-fiber carbohydrates (sugar, starch and pectin) in mucilage were about 28.8%, while husk and pulp showed comparable NFC of 16.2% and 17.6%, respectively. Hemicellulose contents in parchment, husk, pulp, and mucilage were 20%, 15%, 8%, and 1%, respectively. However, cellulose contents in pulp, husk, and mucilage were roughly the same (32%), while the parchment showed higher values (45%). The lignin in the parchment was about two times higher than for husk and pulp, each. Parchment exhibited the highest lignin content (32%) among the by-products. Samples were also examined for their sugar (sucrose, glucose, and fructose) and organic acid (lactic, formic, acetic, and propionic) contents ( Table 5). The highest sugar content was obtained from husk, followed by pulp and mucilage. Fructose was the main component in the sugar analysis. Mucilage had the highest content in organic acids, while husk and pulp contained little amounts of the acid fractions. Lactic acid was the main component in the acid analysis. There was neither detectable sugar nor organic acids in the parchment. Carbon to nitrogen ratio (C:N) in husks showed an ideal value for anaerobic digestion of 25. Mucilage, however, showed lower than the recommended optimum value of 20-30, which might need co-digestion with other substrates with a fairly higher C:N ratio. The parchment, on the other hand, showed the highest C:N ratio (190) [34].

Elemental Analysis
The elemental analysis of husk, pulp, parchment, and mucilage showed that the substrates might exhibit a deficiency in some important trace elements required for optimal and stable biogas production, as illustrated in Table 6, compared to the range of values suggested by Oechsner, et al. [35]. Mucilage was deficient in Mo, Se, and W, pulp in Mn and Co., and husk in all analyzed trace elements besides Co. The deficiency of Mo in mucilage, pulp, and husk was as high as 70%, 80%, and 90%, respectively, compared to minimum trace element requirements. Ni and Se were also deficient with 60% and 55%, respectively, both in husk and pulp. Fe, which is required in a higher amount than other trace elements, was available 29% in husk and 65% in pulp compared to the minimum requirements. Shortage of trace elements often causes a decline in biogas production due to the loss of a stable digestion process and eventually leads to a lower substrate feeding rate [36][37][38][39]. Full scale anaerobic digestion of coffee by-products thus requires supplementation of trace elements through co-digestion with animal manure or commercial formulas.

HBT Analysis
The mean SMY from husk, pulp, parchment, and mucilage was examined at 37 • C for 35 days following the HBT batch assay protocol ( Table 7). The SMY from husk was 159 L kg −1 VS, and the average methane content of biogas was 56.8%. The energy recovery rate of husk was 34%, which indicates a fairly low anaerobic performance. Mönch-Tegeder, et al. [33] demonstrated similar results from batch fermentation of horse dung. The pulp showed SMY of 245 L kg −1 VS with a methane quality of 51.5%. Parchment exhibited the lowest SMY of 31 L kg −1 VS. The result suggested that coffee parchment is not suitable for anaerobic conversion. The highest recalcitrant content in the parchment could be attributed to the lowest SMY. Previous research demonstrated that a higher lignin content was one of the main reasons for lower bio-methane conversion [26,[40][41][42]. Mucilage yielded SMY of 294 L kg −1 VS with about 55.4% of methane from the biogas. Thus, the mucilage showed the highest SMY methane yield among the coffee by-products, followed by pulp and husk. The degradability of pulp and mucilage was 63% and 68%, respectively. Higher SMY recovered from mucilage among all examined substrates can be attributed to higher soluble cell-contents (62.3%), as well as lower lignin (5%) and hemicellulose (0.8%) contents. Moreover, the organic acid content of mucilage was much higher than that of the other substrates [43]. Furthermore, the SMY and quality of the biogas from husk, pulp, and mucilage was comparable to common agro-industrial wastes and some energy crops [25] as depicted in Table 3  The energy recovery rate ranges from 4.6% (parchment) to 66% (mucilage), while pulp and husk showed 56% and 33%, respectively. The lowest recovery rate was expected from parchment due to the very high recalcitrant contents in the biomass. Since the ash content of the parchment is very low, it might be suitable for other types of energy conversion technologies, like pelleting and briquetting. The husk is basically a combination of pulp, mucilage, and parchment; hence, it reflects intermediate values of the components.

Energy Potential of Coffee by-Products in Ethiopia
About 70% of Ethiopian coffee is produced following the dry method and 30% by the wet method. The average green coffee bean production for the past 3 harvest years (2014-2016) was 448,695 Mg year −1 ( Table 1) (Table 8). Since the pulp and mucilage are both available in a single wet processing station, their respective energy potential could be combined. The aggregate methane estimate can produce 238,000 MWh of electricity and 272,586 MWh of thermal energy, provided a CHP unit (total conversion efficiency 75%) is applied. The diesel equivalent of the total bio-methane estimated from the husk, pulp, and mucilage was 68,365 m 3 , which costs about 40.3 million USD, based on the current fuel retail price in Ethiopia. The bio-methane could displace fossil fuel, which is often used by coffee processing facilities. The diesel is mainly used to run pulping machines and pump water for coffee processing. In Ethiopia, the average diesel consumption by farmers' cooperatives to process 1 Mg of fresh coffee cherry into parchment coffee was 2.8 L. However, in large estate coffee farms, the consumption was very low (0.9 L Mg −1 fresh cherry). Furthermore, the thermal energy from the CHP units could be utilized in drying parchment coffee or fresh cherries. Therefore, it would be a supplement for open-air drying and reduce labor costs and weather risks faced by traditional coffee drying. Fischer, et al. [17] estimated a potential of 18MW el from anaerobic fermentation for the Kenyan coffee sector, with an installed capacity of 50 kW el for coffee cooperatives and 250 kW el for large estate farms.

Conclusions
Ethiopia's coffee processing sector generates a huge amount of by-products, both in liquid (mucilage) and solid (pulp, parchment, and husk) forms. The current electricity shortage coupled with the environmental issues associated with coffee waste disposal make anaerobic conversion technology a benign intervention. Coffee by-products are rich in lignocellulosic contents with different proportions of cell contents and cell wall (cellulose, hemicellulose, and lignin) contents. The by-products exhibited promising bio-methane potential (BMP) comparable to common agro-industrial residues and some energy crops. The specific methane yield was highest in mucilage followed by pulp and husk. The BMP from parchment was very low, thus implying that it is not suitable for anaerobic conversion. It was estimated that anaerobic fermentation of coffee processing by-products generated by the Ethiopian coffee sector has significant potential to generate methane as high as 68×10 6 m 3 per year, which can produce 238,000 MWh el of electricity and 272,586 MWh th thermal energy. Most coffee processing facilities could recover bio-methane from their own by-products, which would be sufficient to run the processing activities. Seasonal availability of coffee by-products, particularly from the wet method, could limit a year-round utilization. Ensiling, however, enables longer storage periods. Policy instruments like the feed-in tariff (FIT) are suggested to encourage private investors to produce electricity/thermal energy from coffee by-products and also promote renewable energy production. The FIT offers long-term (15-25 years) guaranteed purchase contract and cost-based compensation to renewable energy producers. Further research on the performance of coffee by-products in combination with other biomass sources should be investigated.