Methane emissions from sugarcane vinasse storage and transportation systems: Comparison between open channels and tanks

Atmospheric Environment 159 (2017) 135e146

Contents lists available at ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Methane emissions from sugarcane vinasse storage and transportation systems: Comparison between open channels and tanks ~o Luís Nunes Carvalho b, Mateus Ferreira Chagas b, Bruna Gonçalves Oliveira a, *, Joa Carlos Eduardo Pellegrino Cerri c, Carlos Clemente Cerri a, Brigitte Josefine Feigl a ~o Paulo, Av. Centena rio 303, Piracicaba, SP, Brazil Center for Nuclear Energy in Agriculture, University of Sa Brazilian Bioethanol Science and Technology Laboratory (CTBE), Brazilian Center for Research in Energy and Materials (CNPEM), Campinas, Sao Paulo, Brazil c ~o Paulo, Av. Pa dua Dias 11, Piracicaba, SP, Brazil Department of Soil Science, “Luiz de Queiroz” College of Agriculture, University of Sa a

b

h i g h l i g h t s
Vinasse is the main residue of sugarcane ethanol production.
We quantified CH4 emissions in the main systems of vinasse storage and transportation.
Higher CH4 emissions are associated with open channel systems.
We estimated an emission of 1.36 kg CO2 eq m3 of vinasse in open channels.
Tanks and pipes is a good strategy to mitigate CH4 emission.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 18 November 2016 Received in revised form 31 March 2017 Accepted 5 April 2017 Available online 7 April 2017

Over the last few years the brazilian sugarcane sector has produced an average of 23.5 million liters of ethanol annually. This scale of production generates large amounts of vinasse, which depending on the manner that is disposed, can result significant greenhouse gas emissions. This study aimed to quantify the methane (CH4) emissions associated with the two most widespread systems of vinasse storage and transportation used in Brazil; open channel and those comprising of tanks and pipes. Additionally, a laboratory incubation study was performed with the aim of isolating the effects of vinasse, sediment and the interaction between these factors on CH4 emissions. We observed significant differences in CH4 emissions between the sampling points along the channels during both years of evaluation (2012e2013). In the channel system, around 80% of CH4 emissions were recorded from uncoated sections. Overall, the average CH4 emission intensity was 1.36 kg CO2eq m3 of vinasse transported in open channels, which was 620 times higher than vinasse transported through a system of tanks and closed pipes. The laboratory incubation corroborated field results, suggesting that vinasse alone does not contribute significant emissions of CH4. Higher CH4 emissions were observed when vinasse and sediment were incubated together. In summary, our findings demonstrate that CH4 emissions originate through the anaerobic decomposition of organic material deposited on the bottom of channels and tanks. The adoption of coated channels as a substitute to uncoated channels offers the potential for an effective and affordable means of reducing CH4 emissions. Ultimately, the modernization of vinasse storage and transportation systems through the adoption of tank and closed pipe systems will provide an effective strategy for mitigating CH4 emissions generated during the disposal phase of the sugarcane ethanol production process. © 2017 Elsevier Ltd. All rights reserved.

Keywords: Greenhouse gas emission Effluents Methanogenesis Sedimentation

* Corresponding author. E-mail address: [email protected] (B.G. Oliveira). http://dx.doi.org/10.1016/j.atmosenv.2017.04.005 1352-2310/© 2017 Elsevier Ltd. All rights reserved.

136

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

1. Introduction Replacing fossil fuels with sugarcane-based ethanol can reportedly reduce emissions of greenhouse gas (GHG) by around 85% (Cavalett et al., 2013; Macedo et al., 2008). Nevertheless, the magnitude of GHG reductions that can be achieved through the use of sugarcane ethanol is strongly dependent on the management practices associated with agricultural production (Davis et al., 2013) and the disposal of residues (Moraes et al., 2014). The process of obtaining sugarcane ethanol generates several residues, which depending on the manner that is disposed can result significant environmental pollution. Vinasse, also called stillage, is the most significant of these residues, both in terms of quantity and its potential environmental impacts. This residue is generated during the distillation phase of production, at an average proportion of 13 L (ranging from 10 to 15 L) for each liter of ethanol produced. Regardless of its chemical composition, vinasse contains high concentrations of organic matter, potassium and sulfates (Fuess and Garcia, 2014) and may cause serious environmental impacts, depending on the manner in which it is disposed (Filoso et al., 2015). About 30 years ago, the ethanol industry in Brazil regulated the disposal of vinasse, specifying that it must be recycled back into agricultural fields (Filoso et al., 2015). This solution was adopted mainly because of its implicity in creating algal blooms and increasing biological oxygen demand when vinasse is discarded into watercourses (Christofoletti et al., 2013), but also because it represents the cheapest and simplest solution (Cortez et al., 1992). Thus, at present, the main way to dispose of vinasse is through direct application onto sugarcane fields. There have been several studies addressing the agronomic benefits of vinasse fert-irrigation of sugarcane fields (Macedo, 2005; Resende et al., 2006; Parnaudeau et al., 2008; Smeets et al., 2008). More specifically, there are other studies that have emphasized the environmental impacts of this vinasse application practice (Resende et al., 2006; Carmo et al., 2013; Christofoletti et al., 2013; Oliveira et al., 2013; Fuess and Garcia, 2014; Paredes et al., 2014). However, all of these studies focus on the consequences of vinasse ferti-irrigation, but little is known about the environmental consequences resulting from the vinasse storage and transportation phases occurring prior to its field application. In Brazil, the most common system of vinasse storage and transportation involves open channels (either uncoated or coated), in which vinasse is conveyed by gravity and active pumping (Macedo et al., 2004). Increasingly for some sugarcane mills, this system is being replaced by one composed of closed pipes and tanks. However, there is a lack of conclusive information comparing the impact of these alternative methods in terms of their respective GHG emissions. In a pioneering study, Oliveira et al. (2015) quantified the GHG emissions from vinasse flowing through an open ~o Paulo State. The authors concluded that the storage channel in Sa and transportation phase is an important source of methane (CH4) emissions to the atmosphere and should be included in future GHG inventories for sugarcane ethanol production. According to the authors, CH4 represents about 98% of total GHG emissions released during vinasse storage and transportation, while N2O emissions contributed on average less than 2%. However, these results represent a case study and to our knowledge, no other efforts have been made to compare GHG emissions from the two most common vinasse storage and transportation systems used in Brazil. This study based on the hypothesis that improvements to vinasse storage and transportation systems will alter the conditions for methanogenesis and may significantly reduce GHG emissions. This research aimed to quantify emissions of CH4 over two agricultural years, while monitoring changes in the physico-chemical

characteristics of vinasse in the most widespread storage and transportation systems within the south-central region of Brazil. Additionally, we performed a laboratory incubation study aimed at isolating the individual and synergistic effects of vinasse and sediment on CH4 emissions. 2. Methods 2.1. Field experiment Assessments of CH4 emissions were carried out at two commercial sugarcane mills; each representing one of the two most widespread vinasse storage and transportation systems used in south-central Brazil; i) a system of open coated and uncoated channels, and ii) a system composed of closed pipes and coated tanks. The first case study was located in the region of Bauru (22 290 3200 S, 48 460 5700 W) while the second was located in Piraci~o Paulo. caba (22 380 0500 S 474100900 O), both within the state of Sa The regions in which the mills are located have similar climatic conditions (Cwa, humid subtropical) and altitudes of 520 m and 554 m, respectively in Piracicaba and Bauru regions. 2.1.1. System 1 - Coated and uncoated open channels This system consisted of a main vinasse transportation channel, the first 40 km of which is coated with cement while the last 20 km is uncoated and vinasse is subsequently in direct contact with the soil (Fig. 1a). Both comprise of a furrow dug with a backhoe, but differ in that: i) the coated section is 1.5 m wide  0.6 m deep and comprises the section closest to the mill; while ii) the uncoated section is 1.5 m wide  0.7 m deep and comprises the end section of the vinasse transportation system. Throughout the two-year evaluation period, each section of the channel system (coated and uncoated), was represented by four equidistant sampling points where CH4 emissions were quantified. 2.1.2. System 2 - Tanks and closed pipes In this system vinasse is stored in three coated tanks with the flow between them facilitated through closed pipes (Fig. 1b). The three tanks were ellipsoidal in shape and presented the following dimensions: Tank 1 was 25 m  20 m; Tank 2 was 45 m  25 m, and Tank 3 was 56 m  30 m. The three tanks were coated with a polyethylene membrane. The vinasse is stored temporarily in these tanks and can be pumped out at high speed to the field through the closed pipes. In order to accurately quantify GHG emissions, we considered each tank as a replicate experimental unit. 2.2. Sampling and measurements of CH4 fluxes Methane fluxes from both vinasse storage and transportation systems were measured by obtaining gas samples as follows. For system 1, eight sampling points were established along the main channel, four along the coated section (10 km apart from each other) and the other four along the uncoated section (5 km apart from each other). The sampling points were established at equidistant intervals along each section of the channel to better account for spatial heterogeneity and permit more accurate extrapolation of CH4 emissions. Gas samples were obtained over two sugarcane crop seasons starting in June 2012 and finishing in December 2013, with a total of 22 sampling events. For system 2, gas samples were taken from the center of each tank to avoid the influence of edge effects. Gas sampling in the tanks was achieved using an apparatus comprising pulleys and cables that enabled chambers to be positioned at the desired location (Fig. 2b). The sampling started during October 2012 and finished in December 2013, with a total of 16 sampling events. The

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

delay in initial measurements from this system, when compared to the channel system, was due to the assembly of the apparatus and necessary adjustments to the sampling methodology. To our knowledge, this was the first time that automatic static chambers had been used as a method to collect CH4 emissions from vinasse storage tanks in Brazil. It is important to note that this study did not take into account possible fugitive CH4 emissions at the interface between closed pipes and tanks. For both vinasse storage and transportation systems, CH4 emissions were measured using floating chambers, with an internal volume of 20 L. At each sampling point, three chambers were used to collect gas samples (Fig. 2a and b). Samples were collected automatically into four 40 mL evacuated vials using battery -operated solenoid valves. The solenoid valves opened the inlet for the vials for a period of 30 s after 0, 5, 10 and 20 min. During each sampling event, the physico-chemical attributes of vinasse including temperature, redox potential (Eh) and carbon (C) content were measured. Methane concentrations were analyzed using a gas chromatograph (SRI-GC 8610®) equipped with a FID (flame ionization detector), operating at a temperature of 250  C. The CH4 fluxes were calculated using the linear change in the concentration inside the chamber as a function of the incubation time. Daily CH4 fluxes were estimated by linear extrapolation, since the sampling done in the morning (9:00e11:00 a.m.) provided a valid estimate of daily average GHG emissions (Reeves et al., 2016). Daily flux (F, mass of gas m2 h1) was calculated according to the equation used by Hutchinson and Livingston (1993) presented below:


F¼

Dg Dt


 V k A

(1)

where: F is the CH4 flux (mg m2 h1); Dg/Dt is the rate of change in CH4 concentration inside the analyzed chamber (mg CH4 h1); V is the chamber volume (L); A is the surface area delimited by the chamber (m2) and k is the time conversion factor (24 h day1). Average emissions and standard deviation of CH4 fluxes were calculated from triplicate observations for each sampling point along the different vinasse storage and transportation systems. The annual cumulative CH4 emissions were calculated by linear interpolation obtained between two successive sampling days, with the numerical integration of the curve realized using the trapezoidal rule (Whittaker and Robinson, 1967). To convert CH4 to CO2 eq (equivalent in CO2), the global warming potential of 25 was used, as proposed by IPCC (2007). To calculate the total emissions of CH4 in CO2 eq from each vinasse distribution system, we used the average emissions from each evaluated point or tank, and the corresponding area, as proposed by Oliveira et al. (2015). To perform the extrapolations for the systems composed by channels and tanks we used Eq. (2) and (3), respectively.

137

Etank ¼ total emission in tank system (kg CO2 eq day1); EP1, EP2, EP3, EP4, EP5, EP6, EP7, EP8 ¼ emission of CH4 at each sampling point (mg CO2 eq m2 h1); ET1, ET2, ET3 ¼ emission of CH4 at each sampling tank (mg CO2 eq m2 h1); AP1, AP2, AP3, AP4, AP5, AP6, AP7, AP8 ¼ area (m2) represented by each sampling point; AT1, AT2, AT3 ¼ area (m2) represented by each tank; f ¼ factor to convert mg of CO2 eq h1 into kg CO2 eq day1. After these calculations and considering the average amount of vinasse produced daily, we estimated the intensity of CH4 emissions for each m3 of vinasse transported along each system. The annual volumes of vinasse used for the calculations were 9340 and 10,547 m3 day1 in system 1, respectively for 2012 and 2013. For system 2, we considered 6500 m3 of vinasse day1, for both the years 2012 and 2013. According to information provided by the sugarcane mill, the period of vinasse storage and transportation in the channel system was 244 and 238 days in 2012 and 2013 respectively, while in tanks a period of 210 days were observed for both years. Information on the production of vinasse and the harvest period was obtained from the database maintained by the sugarcane mill.

2.3. Laboratory incubation experiment Based on the assumption that the CH4 emissions from vinasse storage and transportation systems result from the anaerobic decomposition of organic material deposited on the bottom of channels or tanks, a laboratory incubation study was performed to isolate the contribution of vinasse (V), sediment (S) and the interaction of both of these factors (V þ S) to CH4 emissions. A 16-day incubation study was carried out using 1.5 L Kilner jars into which we added the equivalent of 500 mL of vinasse to the treatment ‘V’, 500 mL of sediment to the treatment ‘S’ and 500 mL of sediment plus vinasse to the treatment ‘V þ S’. It is important to highlight that same headspace (1 L) was maintained for all jars. The experiment was set-up as a fully randomized design, with three treatments and five replicates (jars). The material used for the incubation (vinasse and sediment) was collected from the open channel system in June 2013 and the physico-chemical attributes are presented in Table 1. During the evaluation period, the jars were maintained open at 20  C. The jars were closed only during the incubation period to collect gas emission samples. Gas samples were collected with syringes after 0, 30 and 60 min following closure of jars in order to quantify the CH4 emissions during the incubation period. Samples were collected daily for the first nine days and every two days after that to give a total of 12 sampling events. The gas samples were analyzed on the same day of collec-

Echannel ¼ ððEP1*AP1Þ þ ðEP2*AP2Þ þ ðEP3*AP3Þ þ ………… þ ðEP8*AP8Þ*fÞ

(2)

Etank ¼ ððET1*AT1Þ þ ðET2*AT2Þ þ ðET3*AT3ÞÞ*f

(3)

where: Echannel ¼ total emission in channel system (kg CO2 eq day1);

tion with analysis and calculations performed using the same methodology described above.

138

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

Fig. 1. Illustration of systems of vinasse storage and transportation involving open channels (a) and tanks and pipes (b).

2.4. Statistical analysis Data describing the physico-chemical attributes of vinasse were analyzed using boxplot graphics. Daily CH4 emissions were presented as means and standard deviation (mean ± SD) for each vinasse storage and transportation system. Cumulative CH4 emissions (mean ± uncertainty) were analyzed considering a fully randomized design with three replicates (chambers) from each of the eight and three sampling points representing the systems composed by channels and tanks and pipes, respectively. Despite efforts to enhance the accuracy of field data measured in situ, uncertainties and limitations of data quality are inevitable (Chen et al., 2017). According to IPCC (2000), uncertainty estimates are an essential element when monitoring CH4 emissions. This study presented both emission estimates and uncertainty ranges derived from site-specific data, in accordance with the best practices for emission factor estimation (IPCC, 2000). Uncertainty analyzes on the cumulative CH4 emissions over each year and the emission intensity for each storage and transportation system were quantified according to the distribution of daily CH4 fluxes. Triangular distributions were fitted for each sampling point, and the software @Risk 6.2 was used to run stochastic simulations with 10,000 iterations for each system and each year. Latin Hypercube Sampling was used as propagation errors method. For the laboratory incubation study, the statistical analysis was

performed using a randomized design with five replicates (jars). One-way ANOVA was used to assess variation in CH4 fluxes, with analysis of significant differences among treatments conducted using Statistical Analysis System (SAS) software, v.9.

3. Results 3.1. Physico-chemical attributes of vinasse In system 1, the values of redox potential (Eh) varied significantly among sampling points (Fig. 3a and d). For both years of evaluation, a similar tendency of reduction in Eh values was observed along the system. The values ranged from positive for the coated section of the channel to negative in the uncoated section, indicating a general increase in anaerobic conditions. Comparing both sections of channel over the entire two-year evaluation period, showed an overall tendency for more negative Eh values in 2013. In general, the temperature of vinasse decreased significantly along the length of the channel during both years of evaluation. For most of the evaluated days, the vinasse reached the coated section of the channel (P1) with a temperature ranging from 50 to 60  C, and at the end of the coated section had decreased to values of between 33 and 41  C (Fig. 3b and e). For the uncoated section, the temperature of the vinasse was relatively uniform. Throughout the

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

139

Fig. 2. Automatic CH4 sampling units deployed at vinasse storage and transportation systems defined by open channels (a) and by tanks and pipes (b).

Table 1 ePhysico-chemical attributes of vinasse and sediment used at the incubation experiment. Data represent the mean of five replicates. Attributes

Redox potential

pH

a b

225 ± 39 233 ± 18

Total N

BODa

CODb

0.1 ± 0.0 0.02 ± 0.0

18.6 ± 0.1 20.3 ± 0.0

35.5 ± 0.1 39.0 ± 0.0

1

mV Vinasse Sediment

Total C gL

4.4 ± 0.1 4.5 ± 0.0

6.2 ± 0.1 0.2 ± 0.0

Biochemical oxygen demand. Chemical oxygen demand.

evaluation period, temperatures in the uncoated channel ranged from 20 to 35  C. With respect total carbon (C) content, there was a tendency for reduction along the entire length of channel in 2012, indicating losses of this element from both the coated and uncoated sections (Fig. 3c). In 2013, no significant variation in C content was observed along the channel (Fig. 3f). In system 2, the first tank (T1) presented a tendency for higher Eh, temperature and C content during 2012 and 2013 (Fig. 4). In T2 and T3, negative Eh values were observed, indicating an increase in anaerobic conditions along the consecutive tanks used to store and transport vinasse. In 2013, the same patterns were observed

(Fig. 4d). Although the temperature in the T1 (first tank where vinasse is stored) was higher when compared to T2 and T3, the variation was lower in comparison to the channel system and did not vary across the two years of evaluation. A similar trend was observed for C content.

3.2. Emissions of CH4 from different systems of vinasse storage and transportation 3.2.1. System 1 CH4 emissions were consistently higher from uncoated than coated sections of the channel during both evaluation years

140

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

(Fig. 5a). In the coated section, emissions ranged from 1 to 23; 2 to 17; 11 to 152 e 1 a 353 mg CH4 m2 h1 in 2012; and from 1 to 5; 5 to 166; 8 to 70 e 79e176 mg CH4 m2 h1 in 2013, respectively for P1, P2, P3 and P4. The fluxes of CH4 from the uncoated section ranged from 48 to 1288; 217 to 1505; 184 to 1768; 348e1005 mg m2 h1 in 2012; and from 1 to 5; 5 to 166; 8 to 70 e 79e176 mg CH4 m2 h1 in 2013, respectively for P5, P6, P7 and P8. Higher CH4 emission peaks were observed for P6 and P7 during

both years of evaluation. The cumulative emissions of CH4, considering daily flux and the area covered by each evaluated point and across both years are presented in Fig. 5b and c. We observed significant differences in CH4 emissions between the sampling points along the channels during both years. Higher emissions were observed from the uncoated section of channel. In 2012, the average emission from uncoated sampling points was 28.7 Mg CH4, six fold higher than

Fig. 3. Redox potential (Eh), temperature and carbon (C) content of vinasse along the system of storage and transportation comprising of channels. Boxplots indicate median and distributions of the values over the two years of evaluation. Values represent the average of three replicates ± standard deviation. P1, P2, P3 and P4 represent the sampling points from the coated section, and P5, P6, P7 and P8 represent the sampling points from the uncoated section of the channel.

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

coated points (4.7 Mg CH4), and similarly, in 2013 the average emission was 21.8 ± 1.3 Mg CH4, from uncoated points almost five fold higher than coated points (5.0 Mg CH4). 3.2.2. System 2 The CH4 emissions from the tank system ranged from 0.4 to 19; 4 to 21 and 0.5e9 mg CH4 m2 h1, respectively for T1, T2 and T3, in 2012, and from 0.9 to 36; 0.3 to 38 e 0.1e10 mg CH4 m2 h1 in 2013 (Fig. 6a). Cumulative emissions of CH4 in 2012 were 16, 50 and 46 kg for T1, T2 and T3, respectively (Fig. 6b). In 2013, cumulative emissions of 15, 54 and 32 kg of CH4 were observed from T1, T2 and T3, respectively. (Fig. 6c). 3.2.3. Intensity of CH4 emission For the vinasse storage and transportation system comprised of coated and uncoated channels, the total observed emissions of CH4 were 141.6 and 117.3 Mg, respectively during the 2012 and 2013 crop years (Table 2). The highest percentage of these total CH4 emissions (ranging from 85 to 90%), were obtained from the uncoated section of channel during both 2012 and 2013, respectively. The average emission intensities were 62 and 47 g CH4 m3 of vinasse transported through open channels, respectively in 2012 and 2013, which were equivalent to 1.55 and 1.17 kg CO2 eq m3. Considering the two-year evaluation period, the average emission intensity was 1.36 kg CO2 eq m3 of transported vinasse. The total CH4 emissions observed from the tank and pipe system were 0.12 and 0.10 Mg, in 2012 and 2013. The emission intensity for the tank and pipe system was consistently lower when compared to channels, with the results given as 0.0022 and 0.0020 kg CO2 eq m3 of vinasse, respectively for 2012 and 2013. Fig. 7 presents the probability density functions that describe the range and relative likelihood of possible values of CH4 emission intensities for channels and tanks in 2012 and 2013, resulting from uncertainty assessment that considered the variability in CH4 fluxes from each sampling point over the crop year. The curves show a normal distribution of CH4 emission intensities around the mean values presented in Table 2 along with the confidence interval giving a 90% probability of containing the true value. The uncertainties in cumulative CH4 emissions from Table 2 reflect data variability presented in Fig. 5. The uncertainties in CH4 emissions are associated with variation in vinasse attributes, sampling frequency, climatological conditions and load fluctuations over the year. Lower variability over time of daily CH4 emissions for open channels in 2013 compared to 2012 relates to a narrower confidence interval. While in 2012 the spread considering all the sections of the channels was 23% of the mean, in 2013 this figure declined to 17%. This effect can also be observed in Fig. 7a by examining the wider curve for CH4 emission intensity in 2012. For the tank and pipe system, beyond the effect of daily CH4 emissions over time, the standard deviation of daily emission (Fig. 6a) demonstrate the uncertainty in emission intensity estimates. In some cases, the standard deviation was even bigger than the CH4 fluxes. Although they present a very low probability of occurrence, minimum and maximum values for CH4 emission intensity in open channels are around 20 and 100 g m3 vinasse. For the tank system, the minimum and maximum values for CH4 emissions intensity were 0 and 0.15 g m3 vinasse, respectively. Effects of uncertainty analysis in CH4 fluxes measured for each sampling point based on the emission intensity for channel system are shown in Fig. S1 (supplementary material). This figure demonstrates how the average values of CH4 emission intensity vary in their stochastic calculation considering the uncertainty of each single point individually.

141

3.3. Emission of CH4 from incubation experiment Considering the evaluation period of 16 days, the average daily CH4 emissions ranged from 4 to 11, 43 to 1084 and 221 to 1701 mg m2 day1, respectively for the treatments V, S and S þ V (Fig. 8a). Higher CH4 emissions were observed for the treatment S þ V while an opposite trend was observed for V. CH4 emissions from treatment V remained close to zero throughout the evaluation period, indicating that vinasse alone does not provide optimal conditions for methanogenesis and CH4 production. Cumulative CH4 emissions during the experimental period were 0.14, 5.02 and 8.93 g m2, respectively for treatments V, S and S þ V (Fig. 8b), and significant differences (p < 0.05) were observed. 4. Discussion Overseeing the expansion of biofuel production, while ensuring the proper management of residues generated along the production chain are some of the major concerns of the bioenergy sector. The production of sugarcane ethanol generates large amount of vinasse, which depending on the way in which it is disposed, can result in significant CH4 emissions. In this study we quantified the CH4 emissions associated with the two most widespread systems of vinasse storage and transportation used in Brazil, and use this data to provide information to support GHG inventories for the sugarcane-based ethanol industry. 4.1. Role of the physico-chemical attributes of vinasse in the generation of CH4 Two successive years of measurements indicated that the CH4 emissions from vinasse storage and transportation systems comprising of open channels were significantly higher than those observed from tanks and closed pipes. Although it is well known that numerous factors can influence CH4 emissions from different residue disposal approaches, our results highlight the importance of redox potential (Eh), temperature and direct contact between vinasse, sediment and soil as the main regulators of CH4 emissions from vinasse during the storage and transportation phases. Higher CH4 emissions are associated with open channel systems, especially in uncoated sections. This seems to relate to a combination of ideal Eh and temperature conditions as well as direct contact between vinasse, sediment and soil at the bottom of the channel. Indeed, corroborating this assumption, the results of the incubation study show that contact between vinasse and sediment can result in favorable conditions for CH4 formation that consequently generate higher emissions. These findings support the hypothesis that vinasse in isolation it is not a major source of CH4 emissions. Vinasse provides nutrients and anaerobic conditions for CH4 formation at the bottom of the channel, where sediment and soil (in uncoated sections) are decomposing anaerobically. According to Le Mer and Roger (2001) methanogenesis generally occurs when Eh values are less than 150 mV. The higher Eh values observed along the coated section of channel may have been a limiting factor for CH4 emissions, as well as the higher temperatures recorded for the first sampling points along the coated section. The optimum temperature for methanogenesis ranges from 30 to 40  C and values outside of this range are likely to inhibit methanogenesis and, consequently reduce the production of CH4 (Le Mer and Roger, 2001; Wood et al., 2014). With the exception of the first two sampling points along the coated section of channel, which are closer to the mill and where the vinasse not only presented higher temperatures but also higher levels of oxygenation, the vinasse attributes were favorable to CH4 production. The

142

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

Fig. 4. Redox potential (Eh), temperature and carbon (C) content of vinasse from the tanks and pipes storage and transportation system. Boxplots indicate median and distributions of the values obtained over two years of evaluation. T1, T2 and T3 represent the evaluated tanks.

consistently lower CH4 emissions measured from the last uncoated point (P8) may be related to a reduced and intermittent vinasse flow. These conditions may prolong direct contact between sediment deposited on the bottom of the channel and atmospheric oxygen, which can alter the anaerobic conditions necessary for methanogenesis. The process of methanogenesis throughout the vinasse storage and transportation phase seems to be dynamic and may be limited by the composition of vinasse. For example, the

observed values of Eh from the uncoated section of channel were similar to those of tanks 2 and 3, but higher CH4 emissions were observed from the uncoated channel. In this case, the Eh values did not appear to be a limiting factor and the interaction between vinasse, sediment and soil may increase CH4 formation and emission to the atmosphere. In the tank system, vinasse is stored temporally in coated tanks and transported under pressure through closed pipes that

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

143

Fig. 5. Daily (a) and cumulative in 2012 and 2013 (b and c, respectively) CH4 emissions from vinasse in the open channel storage and transportation system. Values represent the average of three replicates ± standard deviation (a). Bars (b, c) represent the uncertainty of the results. P1, P2, P3 and P4 represent the sampling points from the coated section, and P5, P6, P7 and P8 represent the sampling points from the uncoated section of the channel.

Fig. 6. Daily (a) and cumulative in 2013 and 2013 (b and c, respectively) fluxes of CH4 from vinasse in the tank and pipe storage and transportation system. Values represent the average of three replicates ± standard deviation (a). Bars (b, c) represent the uncertainty of the results. T1, T2, and T3 represent the evaluated tanks.

Table 2 Cumulative emissions of CH4 and the intensity of emission per m3 of vinasse for both systems of storage and transportation. Data were obtained in the years of 2012 and 2013. Values represent the average ± uncertainty. Emission

Year 2012 Cumulative emission Emission intensity Year 2013 Cumulative emission Emission intensity

Unit

Channel

Tanks

Coated

Uncoated

Total

Mg CH4 kg CH4 m3 of vinasse kg CO2eq m3 of vinasse

21.7 ± 10.1 0.009 ± 0.004 0.237 ± 0.110

120 ± 31 0.053 ± 0.014 1.325 ± 0.350

141.6 ± 32.8 0.062 ± 0.014 1.550 ± 0.361

0.12 ± 0.089 0.00009 ± 0.00003 0.00225 ± 0.00007

Mg CH4 kg CH4 m3 of vinasse kg CO2eq m3 of vinasse

23.5 ± 5.3 0.011 ± 0.0024 0.275 ± 0.06

93.8 ± 19.5 0.042 ± 0.009 1.050 ± 0.023

117.3 ± 20.3 0.047 ± 0.009 1.175 ± 0.227

0.10 ± 0.35 0.00008 ± 0.00003 0.00200 ± 0.0007

discharge onto the sugarcane fields. Therefore, with this type of system, the vinasse is not in direct contact with soil and the period of sedimentation is short, which consequently reduced the potential for CH4 formation. Although lower overall CH4 emissions were observed from the tank system, there were notable peaks during the evaluation period. In both systems, our measurements

indicated a high variation in CH4 emissions throughout the year. Seasonal fluctuations in CH4 emissions from liquid surfaces can be caused by environmental factors linked to the substrate composition (Chen et al., 2013; Wood et al., 2014). Vinasse contains compounds, such as sugars and volatile acids, which can be easily degraded to reduce the amount of organic C as well as the

144

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

Fig. 7. Probability density of methane emission intensity, for channels (A) and tanks (B).

Fig. 8. Daily (a) and cumulative (b) emissions of CH4 during the incubation period. Values represent the average of five replicates ± standard deviation. Means followed by the same letter do not differ according to Tukey test (p < 0.05). “V” represents vinasse, “S” represents sediment and “S þ V” represents vinasse þ sediment.

production of CH4 (Oosterkamp et al., 2016). Segers (1998) showed that high levels of organic matter in eutrophic systems increase the dissolved oxygen demand and thereby release large amounts of substrate for methanogenesis. High variability in CH4 emissions from liquid surfaces are expected due to the release mechanism for CH4 molecules, which involves a boiling process, and can be influenced by water turbulence and changes to the hydrostatic , 2007). The relative level of vinasse in pressure (Marani and Alvala tanks can also vary significantly throughout the year, which may influence CH4 emissions, altering its solubility, oxygenation rates and cause variations in temperature (Joyce and Jewell, 2003; Chen et al., 2013). 4.2. What is already known about vinasse storage and transportation systems and its associated GHG emissions Currently, Brazil produces 30.2 billion liters of ethanol per year (UNICA, 2017). Considering an average production of 13 L of vinasse per liter of ethanol, is it estimated that 393 billions liters of vinasse are produced annually, with most of this residue applied to fields through ferti-irrigation. Recent estimates suggest that 65% of the total vinasse produced in Brazil is transported through open channels (Macedo et al., 2004; IDEA, 2010). According to Oliveira et al. (2015) the system of vinasse storage and transportation by open channels results in significant CH4

emissions to the atmosphere, and most of the emissions comes from the uncoated section of channel. Yet, in accordance with this study, the inclusion of these emissions increase in 4.4% the total GHG emissions from sugarcane ethanol production in Brazil. However, it is important to highlight that the results of Oliveira et al. (2015) are derived from a relatively short-term study (48 days) which did not permit evaluations of possible emissions across an entire year. In the absence of other studies evaluating GHG emissions from vinasse storage and transportation, comparisons to evaluate if these estimates were high or low could not be performed. The emissions of CH4 reported by Oliveira et al. (2015) ranged from 394 to 1092 mg CH4 m2 h1 and were on similar magnitude to those obtained for other agricultural residues. For example, Sharpe et al. (2002) evaluated the storage of liquid swine slurry and reported average CH4 emissions of 458 mg m2 h1 and Wood et al. (2014) evaluated liquid dairy manure disposal and observed emissions ranging from 242 to 708 mg m2 h1. Other studies also evaluating swine slurry obtained higher rates of CH4 emissions, ranging from 9666 to 12,000 mg m2 h1 (Dinuccio et al., 2008; Sard a et al., 2010). The discrepancies in these results indicate that there is large variation in the CH4 emissions from agricultural residue storage and transportation systems.

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

145

4.3. Lessons and implications of the current study

5. Conclusions

In the present study, we estimated emission intensities ranging from 1.550 to 1.175 kg of CO2 eq m3 of vinasse transported through an open channel, which was similar than that observed in Oliveira et al. (2015). In the tank system, the intensity of CH4 emissions was 620 times lower than the channel system. Over the last decade, a number of studies have estimated total GHG emissions from sugarcane ethanol production (Macedo et al., 2008; Seabra et al., 2011; Cavalett et al., 2013; Chagas et al., 2016). However, none have focused on the emissions generated through the storage and transportation of vinasse. The inclusion of CH4 emissions from vinasse storage and transportation on life cycle analyzes performed by Macedo et al. (2008) would increase by 4.3% and 0.006% the total emissions attributed to ethanol production in Brazil, respectively from open channels and tanks and pipes systems. It is important to note, however, that the CH4 emissions from tanks might represent an underestimate, since we did not account for eventual fugitive emissions resulting from the faster transportation of vinasse under pressure. A number of studies have indicated that fugitive emissions of CH4 (ranging from 3 to 15% of the total) can occur during the transportation of agricultural wastes through pipelines (CDM, 2005; IPCC, 2006; Flesch et al., 2011). To our knowledge, no similar study has yet been performed on sugarcane vinasse transportation and we suggest further research to quantify such CH4 emissions along the system of vinasse transportation through closed pipes. It is relevant to emphasize that open channels, especially uncoated sections, are an older approach to the storage and transport of vinasse in Brazil and can thus be considered as a baseline scenario when improvements in the systems are realized. Based on our findings, we estimate that the replacement of channel system with tanks and closed pipes will result in an annual reduction of 348,000 Mg CO2 eq. These findings imply that systems comprising tanks and closed pipes present an excellent strategy for mitigating CH4 emissions, since no significant sedimentation and no direct contact with the soil occurs along the length of the system. Additionally, the removal of sediments from the bottom of channels and tanks can be another effective strategy to mitigate CH4 emissions. This finding is aligned with the previous results of Wood et al. (2014) who observed that aged liquid dairy sediments act as an inoculum for CH4 production and that its removal from the bottom of tanks reduced CH4 emissions by 56%. However, there is a current lack of information in the literature describing the potential effects of removing sediment from systems of vinasse storage and transportation and its consequences for CH4 emissions. Another possible GHG mitigation strategy involves the anaerobic digestion of vinasse, which, besides adding to the production of biogas in ethanol production plants, could reduce CH4 emissions during the storage and transport of vinasse (Barrera et al., 2016; Junqueira et al., 2016; Moraes et al., 2017). Moraes et al. (2017) compared CH4 emissions from fresh and biodigested vinasse and showed that anaerobic biodigestion was effective in mitigating GHG emissions during storage compared to untreated vinasse. Additionally, Bernal et al. (2017) highlighted that the biogas produced from vinasse could represent an important source of energy with a high potential for GHG mitigation. The application of available technologies to improve vinasse storage and transportation systems is likely to significantly reduce the potential environmental impacts of sugarcane ethanol production and increase the revenues of the sucronergetic sector.

In summary, our results indicate that CH4 emissions from vinasse originate through the anaerobic decomposition of organic material deposited on the bottom of channels and tanks and should be included in the sugarcane based-ethanol GHG inventories. Significantly greater emissions were shown from vinasse stored and transported using a low technology open channel system, where the mill had not upgraded to a full coating to prevent direct contact between vinasse, sediment and soil. The increased adoption of channel coating presents an effective and affordable means of reducing CH4 emissions. Ultimately, the modernization of vinasse storage and transportation systems through the adoption of tanks and closed pipes constitutes an even greater strategy for mitigating CH4 emissions from the disposal phase of the sugarcane ethanol production. The adoption of new technologies and subsequent improvements to vinasse storage and distribution systems should significantly reduce GHG emissions and thereby make sugarcanebased ethanol a cleaner biofuel. Acknowledgements ~o Paulo Research FounThe authors would like to thank the Sa dation (FAPESP) for financial support (Process 2013/05597e5) and for the graduate scholarship (Process 2012/05735e6) provided to B. G. Oliveira. We also thank anonymous reviewers for their constructive suggestions. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.atmosenv.2017.04.005. References Barrera, E.L., Rosa, E., Spanjers, H., Romero, O., De Meester, S., Dewulf, J., 2016. A comparative assessment of anaerobic digestion power plants as alternative to lagoons for vinasse treatment: life cycle assessment and energy analysis. J. Clean. Prod. 113, 459e471. Bernal, A.P., Santos, I.F.S., Silva, A.P.M., Barros, R.M., Eruin Martuscelli Ribeiro, E.R., 2017. Vinasse biogas for energy generation in Brazil: an assessment of economic feasibility, energy potential and avoided CO2 emissions. J. Clean. Prod. http:// dx.doi.org/10.1016/j.jclepro.2017.03.064. Carmo, J.B., Filoso, S., Zotelli, L.C., De Sousa Neto, E.R., Pitombo, L.M., DuarteNeto, P.J., Vargas, V.P., Andrade, C.A., Gava, G.J.C., Rossetto, R., Cantarella, H., Neto, A.E., Martinelli, L.A., 2013. Infield greenhouse gas emissions from sugarcane soils in Brazil: effects from synthetic and organic fertilizer application and crop trash accumulation. Glob. Change Biol. Bioenergy 5, 267e280. Cavalett, O., Seabra, J.E.A., Bonomi, A., Chagas, M.F., 2013. Comparative LCA of ethanol versus gasoline in Brazil using different LCIA methods. Int. J. Life Cycle Assess. 18, 647e658. CDM. Clean Development Mechanism, 2005. Forced Methane Extraction from Organic Waste-water Treatment Plants for Grid-connected Electricity Supply And/or Heat Production. Draft Revision to Approved Baseline Methodology. AM0013 (AM0013/Version 02, Sectoral Scope: 1, 15 April 2005). United Nations Framework Convention on Climate Change. Chagas, M.F., Bordonal, R.O., Cavalett, O., Carvalho, J.L.N., Bonomi, A., La Scala Junior, N., 2016. Environmental and economic impacts of different sugarcane production systems in the ethanol biorefinery. Biofuels Bioprod. Biorefinig 10, 89e106. Chen, G., Yang, S., Lv, C., Zhong, J., Wang, Z., Zhang, Z., Fang, X., Li, S., Yang, W., Xue, L., 2017. An improved method for estimating GHG emissions from onshore oil and gas exploration and development in China. Sci. Total Environ. 574, 707e715. Chen, H., Wu, N., Wang, Y., Zhu, D., Zhu, Q., Yang, G., Gao, Y., Fang, X., Wang, X., Peng, C., 2013. Inter-annual variations of methane emission from an open fen on the Qinghai-Tibetan Plateau: a three-year study. PLoS One 8. http:// dx.doi.org/10.1371/journal.pone.0053878. Christofoletti, C.A., Escher, J.P., Correia, J.E., Marinho, J.F.U., Fontanetti, C.S., 2013. Sugarcane vinasse: environmental implications of its use. Waste Manag. 33, 2752e2761. Cortez, L., Magalhaees, P., Happi, J., 1992. Principais subprodutos da agroindustria canavieira e sua valorizaçaeo. Rev. Bras. Energ. 2, 1e17. Davis, S.C., Boddey, R.M., Alves, B.J.R., Cowie, A.L., George, B.H., Ogle, S.M., Smith, P.,

146

B.G. Oliveira et al. / Atmospheric Environment 159 (2017) 135e146

Van Noordwijk, M., Van Wijk, M.T., 2013. Management swing potential for bioenergy crops. Glob. Change Biol. Bioenergy 5, 623e638. Dinuccio, E., Berg, W., Balsari, P., 2008. Gaseous emissions from storage of untreated slurries and the fractions obtained after mechanical separation. Atmos. Environ. 42, 2448e2459. Filoso, S., Carmo, J.B.d., Mardegan, S.F., Lins, S.R.M., Gomes, T.F., Martinelli, L.A., 2015. Reassessing the environmental impacts of sugarcane ethanol production in Brazil to help meet sustainability goals. Renew. Sustain. Energy Rev. 52, 1847e1856. Flesch, T.K., Desjardins, R.L., Worth, D., 2011. Fugitive methane emissions from an agricultural biodigester. Biomass Bioenergy 35, 3927e3935. Fuess, L.T., Garcia, M.L., 2014. Implications of stillage land disposal: a critical review on the impacts of fertigation. J. Environ. Manag. 145, 210e229. Hutchinson, G.L., Livingston, G.P., 1993. Use of chamber systems to measure trace gas fluxes. In: Harper, L.A. (Ed.), Agricultural Ecosystem Effects on Trace Gases and Global Climate Change, vol. 55. ASA Spec. Publ., Madison, pp. 63e78. ASA, CSSA e SSSA. IDEA. Instituto de Desenvolvimento Agroindustrial, 2010. Indicadores de desempenho da agroindústria canavieira: Safras 2009/2010. Grupo Idea, 2010, ~o Preto. Ribeira IPCC. Intergovernmental Panel on Climate Change, 2000. In: Penman, J., Kruger, D., Galbally, I., Hiraishi, T., Nyenzi, B., Emmanuel, S., Buendia, L., Hoppaus, R., Martinsen, T., Meijer, J., Miwa, K., Tanabe, K. (Eds.), Good Practice Guidance and Uncertainty Management in National Greenhouse Gas Inventories. IGES; Published, Japan. IPCC. Intergovernmental Panel on Climate Change, 2006. In: Eggleston, H.S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K. (Eds.), IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by the National Greenhouse Gas Inventories Programme. IGES; Published, Japan. IPCC, Intergovernmental Panel on Climate Change, 2007. Climate Change: the Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). Cambridge University Press, Cambridge. Joyce, J., Jewell, P.W., 2003. Physical controls on methane ebullition from reservoirs and lakes. Environ. Eng. Geosci. 9, 167e178. Junqueira, T.L., Moraes, B.S., Gouveia, V.L.R., Chagas, M.F., Morais, E.R., Watanabe, M.D.B., Zaiat, M., Bonomi, A., 2016. Use of VSB to plan research programs and public policies. In: Bonomi, A., et al. (Eds.), Virtual Biorefinery, 1ed, vol. 1. Springer International Publishing, Switzerland, pp. 257e282. Le Mer, J., Roger, P., 2001. Production, oxidation, emission and consumption of methane by soils: a review. Eur. J. Soil Biol. 37, 25e50. Macedo, I.C., Leal, M.R.L.V., Silva, J.E.A.R., 2004. Assessment of Greenhouse Gas Emissions in the Production and Use of Fuel Ethanol in Brazil,” Report. Gov~o Paulo and Secretariat of the Environment, Sa ~o Paulo, ernment of the State of Sa Brazil. Macedo, I.C., 2005. Sugarcane's Energy: Twelve Studies on Brazilian Sugarcane ~o da Agroindústria Canavieira de S~ Agribusiness and its Sustainability. Unia ao Paulo, S~ ao Paulo. Macedo, I.C., Seabra, J.E.A., Silva, J.E.A.R., 2008. Greenhouse gases emissions in the production and use of ethanol from sugarcane in Brazil: the 2005/2006 averages and a prediction for 2020. Biomass & Bioenergy 9, 582e595. Marani, L., Alval a , P.C., 2007. Methane emissions from lakes and floodplains in Pantanal, Brazil. Atmos. Environ. 41, 627e1633. Moraes, B.S., Junqueira, T.L., Pavanello, L.G., Cavalett, O., Mantelatto, P.E., Bonomi, A.,

́

Zaiat, M., 2014. Anaerobic digestion of vinasse from sugarcane bio refineries in Brazil from energy, environmental, and economic perspectives: profit or expense? Appl. Energy 113, 825e835. Moraes, B.S., Petersen, S.O., Zaiat, M., Sommer, S.G., Triolo, J.M., 2017. Reduction in greenhouse gas emissions from vinasse through anaerobic digestion. Appl. Energy 189, 21e30. http://dx.doi.org/10.1016/j.apenergy.2016.12.009. Oliveira, B.G., Carvalho, J.L.N., Cerri, C.E.P., Cerri, C.C., Feigl, B.J., 2013. Soil greenhouse gas fluxes from vinasse application in Brazilian sugarcane areas. Geoderma 200e201, 77e84. Oliveira, B.G., Carvalho, J.L.N., Cerri, C.E.P., Cerri, C.C., Feigl, B.J., 2015. Greenhouse gas emissions from sugarcane vinasse transportation by open channel: a case study in Brazil. J. Clean. Prod. 94, 102e107. ndez-García, C., Kim, C.-H., Bauer, S., Ib ~ ez, A.B., Oosterkamp, M.J., Me an Zimmerman, S., Hong, P.-Y., Cann, I.K., Mackie, R.I., 2016. Lignocellulose-derived thin stillage composition and efficient biological treatment with a high-rate hybrid anaerobic bioreactor system. Biotechnol. Biofuels 9, 120. http:// dx.doi.org/10.1186/s13068-016-0532-z. Paredes, D.S., Lessa, A.C.R., Sant’Anna, S.A.C., Boddey, R.M., Urquiaga, S., Alves, B.J.R., 2014. Nitrous oxide emission and ammonia volatilization induced by vinasse and N fertilizer application in a sugarcane crop at Rio de Janeiro, Brazil. Nutr. Cycl. Agroecosyst. 98, 41e55. Parnaudeau, V., Condom, N., Oliver, R., Cazevieille, P., Recous, S., 2008. Vinasse organic matter quality and mineralization potential, as influenced by raw material, fermentation and concentration processes. Bioresour. Technol. 99, 1553e1562. Reeves, S., Wang, W., Salter, B., Halpin, N., 2016. Quantifying nitrous oxide emissions from sugarcane cropping systems: optimum sampling time and frequency. Atmos. Environ. 136, 123e133. Resende, A.S., Xavier, R.P., Oliveira, O.C., Urquiaga, S., Alves, B.J.R., Boddey, R.M., 2006. Long-term effects of pre-harvest burning and nitrogen and vinasse applications on yield of sugar cane and soil carbon and nitrogen stocks on a plantation in Pernambuco, N.E. Brazil. Plant soil 281, 339e351. Sard a, L.G., Higarashi, M.M., Muller, S., Oliveira, P.A., Comin, J.J., 2010. Reduç~ ao da s da compostagem de dejetos suínos. Rev. emiss~ ao de CO2, CH4 e H2S atrave Brasil. Eng. Agríc. Ambient. 14, 1008e1013. Seabra, J.E.A., Macedo, I.C., Chum, H.L., Faroni, C.E., Sarto, C.A., 2011. Life cycle assessment of Brazilian sugarcane products: GHG emissions and energy use. Biofuels Bioprod. Biorefining 5, 519e532. Segers, R., 1998. Methane production and methane consumption: a review of processes underlying wetland methane fluxes. Biogeochemistry 41, 23e51. Sharpe, R.R., Harper, L.A., Byers, F.M., 2002. Methane emissions from swine lagoons in southeastern US. Agric. Ecosyst. Environ. 90, 17e24. Smeets, E., Junginger, M., Faaij, A., Walter, A., Dolzan, P., Turkenburg, W., 2008. The sustainability of Brazilian ethanoldan assessment of the possibilities of certified production. Biomass Bioenergy 32, 781e813. UNICA, Uni~ ao da Indústria de Cana-de-Açúcar, 2017. Unicadata e Produç~ ao:  rico da produç~ Histo ao e moagem por produto. Available at: www.unicadata. com.br [Accessed 12 January 2017]. Whittaker, E.T., Robinson, G., 1967. The Calculus of Observations: a Treatise on Numerical Mathematics, 4th.ed (New York: Dover). Wood, J.D., VanderZaag, A.C., Wagner-Riddle, Smith, C.E.L.R.J., Gordon, R.J., 2014. Gas emissions from liquid dairy manure: complete versus partial storage emptying. Nutr. Cycl. Agroecosyst. 99, 95e105.