Technoeconomic feasibility of a sustainable charcoal industry to reduce deforestation in Haiti

Sustainable Energy Technologies and Assessments 29 (2018) 131–138

Contents lists available at ScienceDirect

Sustainable Energy Technologies and Assessments journal homepage: www.elsevier.com/locate/seta

Technoeconomic feasibility of a sustainable charcoal industry to reduce deforestation in Haiti A. Balogun Mohammeda, S. Vijleeb, E. Belmonta, a b

T

⁎

Department of Mechanical Engineering, The University of Wyoming, Laramie, WY, USA Donald P. Shiley School of Engineering, The University of Portland, Portland, OR, USA

A R T I C LE I N FO

A B S T R A C T

Keywords: Technoeconomic analysis Pyrolysis Elephant grass Charcoal Haiti Monte Carlo

This study evaluates the technoeconomic viability of utilizing a fast-growing crop, Pennisetum purpureum (elephant grass), to sustainably produce charcoal in Haiti, thereby reducing the harvest of trees, slowing deforestation and driving local economy. The objective of the analysis is to determine the potential for a profitable investment in sustainable charcoal, with profitability defined by net present value (NPV) modeling over a 10year period. Monte Carlo statistical simulation is employed in computing NPV based on variability in model inputs. Results include a probabilistic NPV that incorporates uncertainty in model inputs and yields an oddsbased assessment of the likely profitability of the plant. Results indicate that the plant is likely to yield positive return on capital investment, with a 91% likelihood of the plant breaking even and an 84% likelihood of achieving a 25% return on investment within 10 years. Sensitivity analysis of the model results to model input variability show that charcoal sale price and feedstock cost are the most significant variables that affect plant profitability and profitability uncertainty. Based upon model results, suggestions for improved economic feasibility are presented. Additionally, the broader social and environmental impacts of the proposed venture to reduce deforestation and create local jobs is assessed.

Introduction The economic value of natural resources, such as forests, has been shown to be a major cause of deforestation and forest degradation in developing countries [1]. Specifically, the demand for wood and charcoal in many countries has been shown to put significant pressure on forests, leading to degradation or deforestation without proper management [2,3]. It is estimated that about three billion people use biomass as fuel for cooking and heating worldwide, including most of the ten million people of Haiti [4]. Haiti's population growth and accompanying increase in proportion of people living in urban settings has led to an increase in demand for fuel wood and charcoal, as the latter is used extensively for cooking [5]. While other contributing factors to forest degradation have also been identified, such as agricultural techniques, the demand for fuelwood and charcoal has been confirmed to be a significant contributor which is poised to grow as Haiti’s population continues to increase [6,7]. Recent estimates are that 3.4 × 107 kg of wood charcoal was produced in Haiti in 2014, which was a 22% increase in production over the previous decade [8]. Conversion of native forests for resource utilization has led to extensive forest degradation and economic and political instability [9]. Although

⁎

a 2% forest cover estimate is widely circulated in media and in discourse concerning the country, a recent analysis that utilized high-resolution satellite imagery estimates forest cover at 30% [10]. Nevertheless, deforestation has been severe, and ongoing degradation poses a threat to Haiti’s social stability and energy security [11]. There is significant interest in reducing the use of wood-derived charcoal, but alternative cleaner and more sustainable energy options to wood-derived charcoal, such as liquified petroleum gas (LPG), have remained limited in use due to high cost, tradition and social barriers [6]. Thus, substantial efforts focused on improving cookstoves for higher fuel efficiency [12–16] and replacing wood with other biomass sources for charcoal production in developing countries in order to slow deforestation [17–19]. Any viable alternative to wood-derived charcoal will need to be affordable to the population and satisfy the drivers of wood-derived charcoal production, however, including the economic foundation of charcoal whereby a significant fraction of a poor population derives income. The work of Luoga et al. [20] examined Tanzania with a focus on the lack of alternative livelihoods for the rural poor. Traditional charcoal-making was shown to provide income when agriculture and livestock wane. Harvesting of trees and thermal conversion in simple earthbound kilns require little cash or skilled labor, so

Corresponding author at: 1000 E. University Ave., Dept. 3295, Laramie, WY 82071, USA. E-mail address: [email protected] (E. Belmont).

https://doi.org/10.1016/j.seta.2018.08.001 Received 15 March 2018; Received in revised form 29 June 2018; Accepted 22 August 2018 2213-1388/ © 2018 Elsevier Ltd. All rights reserved.

Sustainable Energy Technologies and Assessments 29 (2018) 131–138

A. Balogun Mohammed et al.

Nomenclature a Ct F HTG l L NPV OpEx r

R ROI S t T TCI TIC TOC USD x X

local fee multiplication factor net cash flow feedstock expense Haitian Gourde (Currency) land lease labor net present value operating expense discount rate

revenue return on investment secondary expenses current time period total time period total capital investment total installed cost total operating cost United States Dollar (Currency) annual feedstock produced tax

Monte Carlo modeling, which draws from ranges of potential values for input variables to calculate probabilistic model outputs [34]. Over many iterations, a probability distribution of model outcomes is achieved. Paap et al. [35] used a Monte Carlo model to assess environmental and economic costs and benefits of producing ethanol or fatty acid ethyl esters from switchgrass based upon uncertainty in model inputs such as process yields and enzyme loading. Trivedi et al. [36] used a Monte Carlo simulation to assess the costs and benefits of using corn stover for production of liquid fuels and power, and results included potential ranges of greenhouse gas reductions and value of societal benefits as functions of variability in model inputs such as feedstock supply and fuel conversion costs. Di Lorenzo et al. [37] used a Monte Carlo simulation to guide financial investment in carbon reduction technologies for power production by evaluating probabilistic potential NPV, internal rate-of-return, and payback period with consideration of uncertainty in factors such as future capital, operating, and emissions costs. Hsu [38] used Monte Carlo uncertainty analysis to evaluate life cycle greenhouse gas emissions and net energy values of gasoline and diesel produced by pyrolysis of forest residue biomass, accounting for uncertainty in fuel product yield and properties. Shahrukh et al. [39] used Monte Carlo analysis to assess biomass pellet production costs, accounting for uncertainties in transportation cost, field cost, and material loss during processing. These studies demonstrate the substantial advantage of Monte Carlo modeling over deterministic modeling due to the additional insight that is gained by the inclusion of input variability to assess output uncertainty. The present study evaluates the technical logistics and economics of using a dedicated bioenergy crop to replace trees as feedstock for charcoal production in Haiti. Elephant grass is selected as the crop of interest in this study, but the model can be readily extended to other energy crops and feedstock sources, such as food crop byproducts or waste. The objective of the analysis is to determine the potential for a self-sustaining entrepreneurship venture based on sustainable charcoal, where self-sustenance is defined as a positive return on capital investment as determined by net present value (NPV) modeling over a 10year period. Monte Carlo statistical simulation is employed in computing NPV based on variability in model inputs. Thus, the model provides a probabilistic NPV that incorporates uncertainty in model inputs. Sensitivity analysis gives insight into plant profitability based upon each of the input variables, and strategies for improving economic viability based upon model results are discussed. Additionally, potential environmental and societal impacts of the proposed venture are evaluated. In particular, the extent to which future deforestation may be reduced by replacement of tree-based charcoal with crop-based charcoal is assessed and the potential for local job creation is discussed.

charcoal making will bring positive economic returns to the individual despite the costs of degradation to the local ecosystem. Therefore, it is important when seeking alternative options to acknowledge that it might not be possible for renewable charcoal to completely replace traditional charcoal. Fast growing perennial bioenergy crops, such as Pennisetum purpureum (commonly called Napier grass or elephant grass, and henceforth referred to as elephant grass), may be a viable and sustainable alternative to trees as a source of charcoal, thereby slowing deforestation in countries such as Haiti. These crops have gained attention worldwide as an energy feedstock to replace fossil fuels for carbon emissions reduction because of their rapid growth, high yield, substantial energy density, and ability to grow in marginal soils [21]. Reviews of energy crop utilization, both from dedicated farmland and permanent grassland, in terms of environmental impacts and economics have shown the potential for substantial reduction of atmospheric carbon stores by combustion of crops instead of fossil fuels, as well potential economic viability under careful planning [22,23]. As perennial grasses, these crops do not need replanting after each harvest, reducing the energy input requirements as compared to annual crops. Of these grasses, elephant grass is particularly interesting because it grows quickly, reaching 3–5 m in height and 2 cm in diameter within 180 days [24]. Elephant grass can typically be harvested up to four times within a year in warm climates with a ratio of energy output to energy input of around 25:1, thereby making it one of the best potential energy crops for development of efficient and economical bioenergy systems [25]. Furthermore, elephant grass has minimal supplementary nutrient requirements and has been shown to grow well in similar climates to Haiti’s [26]. Elephant grass and other energy crops have been shown to behave similarly to wood in the production of charcoal by pyrolysis, in which biomass is heated in an inert environment, thereby removing moisture and volatiles while concentrating carbon and increasing energy density [27]. Strezov et al. [28] studied the thermal conversion of elephant grass to produce charcoal and showed that gaseous byproducts contained sufficient caloric energy to provide the required energy for pyrolysis, suggesting minimal additional energy needs for thermal processing. Technoeconomic analyses (TEAs) have been used to assess a wide variety of renewable energy systems, such as torrefaction of biomass residue briquettes in Brazil [29], fermentation of dedicated energy crops to produce bioethanol [30], production of other liquid fuels via upgrading of biomass pyrolysis oils [31], gasification of torrefied biomass to produce dimethyl ether fuel [32], and combustion of forest waste biomass for power generation [33]. These analyses frequently assess technical and economic feasibility in order to assess profitability, and therefore likelihood of adoption, of renewable energy systems. The majority of TEAs are deterministic, utilizing mean representative costs in economic modeling, which result in model outputs, such as net present values (NPVs), which are single point values based upon these mean model inputs. Far fewer TEAs have incorporated uncertainty and variability in model inputs into the assessment of model outputs. One such approach to incorporating model input variability is the use of

Model analysis Plant characteristics and configuration The technical and economic feasibility of a production plant that outputs 7.6 × 105 kg·yr−1 of elephant grass-derived charcoal is 132

Sustainable Energy Technologies and Assessments 29 (2018) 131–138

A. Balogun Mohammed et al.

Economic model

assessed. This output scale is sized based upon the smallest commercially available equipment as quoted by a manufacturer, and the semiautomated plant in this model is assumed to have scalability of the process and costs such that model results are expected to approximately scale for larger plants. Based upon on a 6-day workweek and 8-hour workday, the daily charcoal production output for the plant is 2.4 × 103 kg·day−1. The total charcoal market of Port-Au-Prince is estimated to be 6.1 × 107 kg·yr−1 of charcoal consumed based upon a Port-Au-Prince population of 733,840 people and an average household size of 5 people, approximately 50% of whom utilize an average of 2.3 kg·day−1 of charcoal for cooking [40,41]. Based upon comparable heating values of carbonized mesquite and energy crops [42,43], it is assumed that crop-derived and wood-derived charcoal are directly substituted. Thus, the output of the plant corresponds to approximately 1.3% of the Port-Au-Prince charcoal market. The plant functions as a vertically integrated business in that it utilizes biomass feedstock grown on leased dedicated farmland that is co-located with the plant. Elephant grass is selected as the feedstock for this analysis due to its rapid growth and suitability for the climate of Haiti [44]. Based upon the target output of the plant and assuming a conversion yield of 35% during the pyrolysis process [45], feedstock requirements for the plants are 2.2 × 106 kg·yr−1 of raw feedstock, or 7.0 × 103 kg·day−1, on a dry basis. Elephant grass is expected to yield 3.6 × 106 kg·km−2 per year on a dry basis [46]. Therefore, 0.61 km2 of farm land is required to produce the needed feedstock to meet the plant output target. The harvested elephant grass is moved to the pyrolysis plant for processing into charcoal. Since the plant and farm are co-located, this transportation consists of manual labor and short distance trucking by the laborers. The plant is designed to be a low cost, semi-automated plant that might be installed in a developing country. It chiefly consists of pre-processing, pyrolysis, and briquette machines that chip, dry, carbonize and briquette the charcoal for manual packaging by laborers. The feedstock is first chopped into approximately 10 cm pieces by a chopper, then dried with exhaust gas heat from an internal combustion engine, which may be used to power a generator for electricity. The pretreated feedstock then undergoes pyrolysis, again using combustion heat, after which the processed feedstock is mixed with a binder, such as corn starch, in a mixer. Product gases and liquids from the pyrolysis process are burned to provide process heat as well, although there may potential to monetize pyrolysis liquids in other ways as implementation is scaled. Finally, machine presses are used to compact the mixture into dense blocks which are packaged in bags by laborers. The packaged charcoal is sold to local wholesalers, who pick up the charcoal at the plant site for distribution in markets. Therefore, there are no transportation costs for the final product incurred by the plant. Fig. 1 shows the flow diagram for the charcoal production process analyzed in this study, as well as the system boundary for the study.

The economic model utilizes NPV modeling to determine whether a positive return on investment can be achieved, which makes it possible to develop and operate a sustainable charcoal production business in Haiti. Economics are assessed for the production plant, with farm capital and operating expenses built into the plant NPV model via feedstock costs. Eq. (1) gives the NPV of the plant T

NPV =

C

∑ (1 +t r )t −TCIplant t=1

(1)

where Ct is net cash flow during year t, TCIplant is the total capital investment, r is discount rate, and T is the number of analyzed years, which is 10 years in this study. The discount rate in this study is set to 3% based on the rate used Moon et al. [47], which is comparable to the 3.5% discount rate recommend by Moore et al. [48] for projects focused on societal benefits. It is assumed that the plant is financed by equity and therefore no interest charges are incurred. Net cash flow is the difference between the plant’s cash inflows and outflows in a given year, and Ct is defined by Eq. (2)

Ct = R−TOCplant

(2)

where R is the revenue derived from the sale of crop-derived charcoal and TOCplant is the total operating cost, as given by Eq. (3)

TOCplant = OpEx plant + Lplant + F + X

(3)

where TOCplant is the sum of operating expenses (OpEx plant ), labor (Lplant ), feedstock expenses (F), and payroll taxes assessed on labor wages ( X ). Finally, the total capital investment of the plant (TCIplant ) is included in the calculation of NPV in Eq. (1). TCIplant is comprised of all initial permanent plant expenses and is divided into total installed costs (TICplant ) and secondary expenses (STIC , plant ), as shown in Eq. (4)

TCIplant = a∙TICplant + STIC , plant

(4)

where TICplant represents the expenses associated with plant equipment and site development, and STIC , plant are indirect costs, field expenses, office and construction fees, and project contingency. The coefficient, a, is a multiplication factor applied to equipment expense which encompasses local costs that are difficult to anticipate, such as additional taxes, transportation costs, and local fees which might be incurred upon delivery of equipment to Port-Au-Prince and during transport to the plant site. Monte Carlo simulation NPV is calculated for a 10-year operating period of the plant (T = 10 yr). This time period is selected because investors are expected to desire payback within this time due to the volatility of the region. Additionally, the system will incur some additional costs after this time

Fig. 1. An overview of the modeled charcoal production process is shown, including the boundary of the analyzed system, which includes a production plant and a farm for feedstock and processing operations. 133

Sustainable Energy Technologies and Assessments 29 (2018) 131–138

A. Balogun Mohammed et al.

0.063, and high and low values set to +/- 15% from the base value.

period; for example, replanting of the elephant grass crop is anticipated within 15 years of initial planting [21]. A Monte Carlo simulation is used to calculate NPV while accounting for variability and uncertainty in inputs. This is accomplished by modeling NPV model inputs as distributions, rather than point estimates. The result of Monte Carlo modeling is that this method delivers weighted possible outcomes of NPV based on the likely distributions of model inputs, and provides insight of outcome probability to support investment decisions. The model inputs that are assigned variable distributions are charcoal sale price (R), feedstock cost (F), operating expenses of the plant (OpEx plant ), payroll taxes ( X ), plant labor costs (Lplant ), and the multiplication factor applied to the total installed cost of the plant (a). The ranges in values of these inputs are modeled as Gaussian distribution functions. Estimates of reasonable input value ranges have been gathered where available, and are specified. In cases where an estimate of variable value range is not available, standard deviation and high and low variable values are set as percentages of the mean; the high and low variable values are then the upper and lower limits of the Gaussian distribution. The NPV model and model input value ranges are programmed in @Risk software by Palisade Decision Tools, and 100,000 Monte Carlo iterations are run to calculate a probability-weighted plant NPV. At each iteration, values of each variable are randomly selected based upon the possible range of values and their likelihood of occurrence. The model output NPV is then logged as a single occurrence and, upon the completion of 100,000 iterations, a histogram of likely NPV values is attained. Sensitivity analysis is performed to determine the dependence of NPV on each of the variable inputs. This analysis is realized by running iterations of each variable input, within the 90% confidence interval of their values, while holding all other variable inputs constant at their mean values, to compute the NPV output. The relative range of NPV output calculated for each isolated variable input corresponds to the effect of the input variable on NPV output and is thus indicative of the sensitivity of NPV to the inputs.

Total operating cost The components of total operating cost, TOCplant , which include OpEx plant , Lplant , F, and X , are modeled as follows: OpEx plant includes recurring annual expenses due to normal operations, such as equipment maintenance, generator fuel, and packaging expenses in terms of binding agents and bags. These expenses are estimated to be 10% of TICplant [49]. The plant is staffed with a team of seven laborers, one technician, one assistant manager, and one manager. Labor expenses are estimated based on the 2016 government minimum wage regulation in Haiti, which dictates a daily wage for unskilled laborers of USD 5 per day [50]. The skilled labor rate for a technician and supervisor is estimated to be 145% of unskilled labor rate, and the managerial wage is estimated to be 200% of the unskilled rate based upon correspondence with personnel of Carbon Roots International, a Haiti-based NGO that is developing charcoal from feedstock other than trees. The plant feedstock cost is evaluated separately as a sub-model of the NPV model due to the numerous factors that contribute to feedstock cost. This sub-model analysis approach also permits the comparison of feedstock costs in this model to other scenarios in which different feedstocks are grown or purchased. The sum of expenses to establish the farm and supply feedstock over the period of analysis, T, is divided by the annual biomass feedstock produced, x, and the period of analysis, T, to obtain the cost per kg of feedstock based on the plant requirement. Thus, feedstock expenses are defined in Eq. (5)

F=

T (Lfarm + X + lfarm + OpEx farm) + TCIfarm x∙T

(5)

where the farm subscript indicates the labor costs (Lfarm ), land lease (lfarm ), operating expenses (OpEx farm ), and total capital investment (TCIfarm ) that are specific to the farm operation of the business. The plant requires 0.61 km2 of farm land to meet the annual plant charcoal output requirement, which translates to 2.0 × 10−3 km2 of elephant grass that is harvested daily. It is estimated that three unskilled laborers, utilizing two mini harvesters and wheelbarrows, are sufficient to harvest the daily feedstock. Mean feedstock cost is assessed based upon mean input variable values, and then ranges of variable values are assigned for Monte Carlo analysis. A 6% land lease rate is assessed on land cost of 1,000,000 USD·km−2, based upon correspondence with Carbon Roots International personnel, and equipment cost of USD 7,761 according to a manufacturer quote, is used to determine a mean dry feedstock cost of 0.047 USD·kg−1. Variability in feedstock cost is estimated as a Gaussian distribution bounded by ± 25% of the mean value, with standard deviation of 0.0038 USD·kg−1 for Monte Carlo modeling. Finally, payroll taxes vary between 30 and 40% depending on labor wage bracket, and an additional 5% payroll tax for social security is applied based upon correspondence with Carbon Roots International personnel. Thus, the payroll tax is estimated at a mean and standard deviation value of 40% and 3.2% respectively, with variability of ± 25% of the mean value is assigned to a Gaussian distribution for Monte Carlo modeling.

Equipment costs Plant equipment includes the pyrolysis process components, including pre- and post-processing, and a power generator. While future implementations may consider the local construction of equipment to lower costs, particularly as nth plant scale is anticipated, this study examined procurement of equipment from outside of Haiti. The pyrolysis unit includes the chopper, dryer, pyrolyzer, mixer, and charcoal briquette machine components needed to process raw elephant grass feedstock to charcoal. This equipment, sized to process 300 kg·hr−1 of charcoal, was quoted at USD 94,760 by an equipment manufacturer and the quoted price includes sea freight delivery to Port-Au-Prince. A power generator is required to power the 11 kW pyrolysis unit because power grid supply in Haiti is intermittent and generally not reliable. An equipment supplier quote of USD 4,173 was obtained for a 15 kW generator suitable to power the pyrolysis unit. This cost also includes sea freight delivery to Port-Au-Prince. Warehouse storage and other site development expenses are estimated to be 1.5% and 9%, respectively, of the total equipment costs [49]. The secondary TCIplant expenses(STIC , plant ) considered in the present analysis are indirect costs, field expenses, office and contingency fee, and project contingency. Indirect costs and field expenses are estimated to be 20% and 25% of TICplant , and office and construction fee and project contingency are estimated to be 3% and 5% of TICplant [49]. Equipment purchase costs and development expenses are fixed, and uncertainty in overall equipment costs are reflected in the multiplication factor that is applied to the total installed cost of the plant, a. This factor encompasses local expenses upon equipment delivery to Port-Au-Prince are not easily estimated due to unpredictability of local factors. Thus, the value of this coefficient is estimated as a variable Monte Carlo input with a mean value of 1.25, standard deviation of

Revenue Revenue is derived from the sale of crop-derived charcoal at a market wholesale price of HTG 500 per 30 kg sack of charcoal based upon correspondence with Carbon Roots International personnel. This translates to a charcoal price of 0.25 USD·kg−1 based on a 0.015 USD/ HTG currency conversion rate. However, the price of charcoal can fluctuate by as much as 25% from month to month depending on weather, the political situation in Haiti, and other factors according to Carbon Roots International personnel. Thus, the charcoal sale price is 134

Sustainable Energy Technologies and Assessments 29 (2018) 131–138

A. Balogun Mohammed et al.

assessment of the likely profitability of the plant. Based upon mean values alone, the conclusion from NPV modeling would be that the plant is profitable under the given scenario and no further information about the likelihood of profitability would be gleaned. With Monte Carlo modeling, however, the model indicates not only the likelihood of profitability, but also the potential gain or loss that an investor might enjoy or incur. Therefore, the Monte Carlo approach to TEA modeling is found to give far more nuanced insight into the economics of the system as compared to mean-based NPV modeling.

modeled as a Gaussian distribution with a mean value of 0.25 USD·kg−1, standard deviation of 0.02 USD·kg−1 and ± 25% of the mean value as high and low limits for Monte Carlo modeling. Results and discussion The economics of the charcoal plant are evaluated in terms of NPV by Eq. (1), using a Monte Carlo modeling approach that accounts for uncertainty in the inputs. Results include the uncertainty in input variables, a histogram distribution of NPV for the plant, and sensitivity of model results to inputs. Though the presented results are from the 100,000 Monte Carlo simulations, iterative convergence was tested and observed for 10,000, 20,000, 50,000 and 100,000 Monte Carlo simulations. Only at Monte Carlo iterations of 10,000 were NPV likelihood differences of 0.2% observed.

Sensitivity to Monte Carlo inputs The sensitivity of plant NPV to variation in model inputs is assessed in order to appraise the factors that most substantially affect NPV, which can then guide system design and modification. Fig. 4 shows the results of this sensitivity analysis by summarizing the effect of variable input values on the mean NPV of the plant at USD 204,500. Variability in charcoal sale price (R) is observed to have the most significant impact on plant NPV, with the 90% confidence interval values of sale price translating to a positive NPV is USD 438,300 and a negative NPV of USD -30,000, respectively. In fact, charcoal sale price is the only model variable that singlehandedly produces a negative plant NPV over its 90% confidence interval values. All other variables produce a range of plant NPV, but the NPV is positive over the 90% confidence interval values. Following charcoal sale price, the next most impactful variable on plant NPV is feedstock cost (F). Though the high end of the 90% confidence interval of feedstock cost does not result in a negative NPV, it does nearly eliminate investor return by lowering the plant NPV by more than half, to USD 76,000. Thus, while the feedstock cost may not singlehandedly cause a negative plant NPV, it does, in conjunction with variability in other input values, contribute significantly to overall uncertainty in profitability. It is observed that other model input values, including equipment costs, taxes, labor costs, and other expenses have significantly smaller impacts on the plant NPV as compared to charcoal sale price and feedstock cost. Nevertheless, as noted for feedstock cost, these factors may contribute to a plant NPV above or below the mean, or even a negative NPV in conjunction with variability in other factors. The analyses of NPV variability due to model input values may be used to guide the plant, process and business model design. For example, based upon the strong impact of charcoal sale price on plant NPV, the plant operator might investigate selling charcoal on a contractual basis to hedge against potential fluctuation in charcoal sale price. Feedstock cost, which is the second most impactful variable for plant profitability, might be lowered by finding alternative feedstock options.

Variability in input parameters While mean, high and low values of model input variables are input to NPV modeling, there is a smaller range within which a large percentage of variable values fall. A 90% confidence interval analysis is used to highlight the likely ranges of variable values that result from the input ranges. As an example of how model input variability is modeled and accounted for in Monte Carlo modeling, Fig. 2 shows the histogram of charcoal sale price, based upon the assumed mean cost of 0.25 USD·kg−1, standard deviation of 0.02 USD·kg−1, 25% fluctuation in price, and a Gaussian distribution. The horizontal axis indicates the range of possible prices, while the vertical axis is the probability of occurrence during 100,000 iterations. A 90% confidence range of the variable is highlighted, which indicates where 90% of charcoal prices are expected to occur if sampled over time. Thus, while the charcoal sale price has the potential to fluctuate between 0.19 USD·kg−1 and 0.31 USD·kg−1, which is 75% and 125% of the mean price of 0.25 USD·kg−1, respectively, the price is expected to be between 0.21 USD·kg−1 and 0.28 USD·kg−1 90% of the time, less than 0.21 USD·kg−1 5% of the time, and greater than 0.28 USD·kg−1 5% of the time, with all values fitting a Gaussian distribution. The importance of the 90% confidence interval in all model variables is most significantly realized in the outcome of the NPV model. Table 1 summarizes the full value ranges for the variables included in the model, and the 90% confidence intervals of those values. The results in Fig. 2 and Table 1 highlight the Gaussian distribution applied to the data and the likely values, within 90% confidence, of these variables Plant profitability Fig. 3 is a histogram of the evaluated NPV for the modeled sustainable charcoal plant. The figure is annotated to indicate the likelihood of breaking even, or achieving an NPV of zero after 10 years. The likelihood of positive revenue, with NPV equal to 25% ROI, after 10 years is also indicated, where ROI is the sum of TCI . There is a 90.8% likelihood of breaking even or earning positive NPV, and an 83.9% likelihood of earning a 25% ROI or greater, after 10 years of plant operation. These likelihoods are determined by dividing the number of occurrences of zero or positive NPV values by the total iterations. Correspondingly, there is a 9% chance that the plant will have a negative NPV, and a 16% chance of less than 25% ROI (including the possibility of negative NPV), after 10 years. The 90% probability range of NPV for the plant ranges from USD -48,900 to USD + 457,100, with a mean of USD + 204,500. Furthermore, based on model inputs and NPV values, the payback period of the plant is 5 years of operation based upon mean variable values. Whether or not the plant is profitable depends on the values of the variables in Table 1 and the Gaussian distributions of these variables as shown in Fig. 2; thus, the results shown in Fig. 3 give an odds-based

Fig. 2. The probability distribution of charcoal sale price is shown, including the upper and lower values within a 90% confidence interval. 135

Sustainable Energy Technologies and Assessments 29 (2018) 131–138

A. Balogun Mohammed et al.

cane bagasse availability is dependent upon the growth of sugar cane as a separate primary crop. Therefore, considering the fuel flexibility of the pyrolysis process and the high cost of elephant grass feedstock, a blended feedstock of elephant grass and bagasse or other agricultural byproducts or wastes may be an approach that, when available, lowers the average feedstock cost while maintaining feedstock source reliability. When considering different feedstock options for the charcoal plant in this study, the modeled feedstock cost based upon a co-located farm provides a reasonable basis for comparison for purchase prices of feedstock from third-party suppliers, if needed to augment feedstock supply to the plant.

Table 1 Summary of model inputs, including mean, and high and low input values as well as high and low values with a 90% confidence interval (CI). Input

90% CI

Variable

Units

Mean

High

Low

High

Low

Charcoal sale price (R) Plant labor cost (Lplant) Feedstock cost (F) Local fee factor (a) Payroll tax (X)

USD·kg−1 USD·yr−1 USD·kg−1 n/a %

0.25 14,352 0.047 1.25 40.0

0.31 17,940 0.059 1.45 45.0

0.19 10,764 0.035 1.05 35.0

0.28 15,697 0.053 1.35 44.0

0.21 13,031 0.041 1.15 36.0

Environmental and societal benefits The plant modeled in this study has the potential to slow deforestation by substituting elephant grass for trees as charcoal feedstock. There are numerous types of trees of importance to Haiti, one of which, Prosopis juliflora is a mesquite that is extensively used for charcoal production [51]. Given a conversion efficiency of mesquite to charcoal of 30% [42], replacement of wood-derived charcoal with crop-derived charcoal has the potential to eliminate the harvest of 2.5 × 106 kg of tress annually. Assuming a medium density stand of 19.4 metric tons of mesquite per hectare [52], this equates to 1.3 km2 of wood that is not harvested per year. While the single plant modeled in this study serves only 1.3% of the Port-Au-Prince charcoal market, the plant is scalable and can be adopted on larger scales, particularly because of the flexibility to process bagasse and other feedstock for blending with elephant grass feedstock. Furthermore, similar plants can be adopted around the country to increase the production of sustainable charcoal and, in the process, further reduce deforestation and create employment while putting marginal lands not suitable for agriculture farming to use. Under such scenarios in which widespread implementation is achieved, a cooperative structure might then be considered to share equipment, lower production costs and provide an improved means of product distribution. While policy and local incentives or tax credits were not a focus of the present study, it is worth noting that tax credits based upon the environmentally favorable nature of the proposed venture should be considered to make the economics of the business venture, and uptake of crop-derived charcoal by consumers, even more favorable. Whereas the NPV model indicates a high likelihood of business profitability and

Fig. 3. Probability distribution of modeled plant NPV indicating breakeven and 25% ROI likelihoods after 10 years of operation.

Fig. 4. Effect of variable model input values on plant NPV.

Breakdown of total operating costs The total operating cost, TOCplant , which is the sum of OpEx plant , Lplant , F, and X , is often targeted for cost savings because it is a recurring, and typically large, expenditure that significantly affects business profitability. Therefore, the magnitudes of contributing factors to TOCplant are assessed with the goal of targeting cost reduction and profitability improvement. Fig. 5 is a breakdown of the TOCplant expenses incurred annually during operation of the plant. It is evident from Fig. 5 that feedstock cost is the most significant recurring expense at USD 102,000 per year. As previously discussed, uncertainty in this substantial expense also contributes significantly to uncertainty in the profitability of the plant. Thus, it is worth recalling that, while the present analysis is focused on a dedicated elephant grass feedstock, the modeled plant is capable of processing a variety of other fast-growing energy crops as feedstock such as switchgrass, or food crop byproducts, such as sugar cane bagasse. While elephant grass feedstock can be up to four times more expensive than locally sourced bagasse according to Carbon Roots International, elephant grass offers the benefits of reliable availability when grown for the plant, whereas sugar

Fig. 5. Boxplot showing components of TOCplant , including maximum, minimum, and first and third quartiles for feedstock cost (F), plant labor cost (Lplant ), operating expense of the plant (OpEx plant ), and payroll tax (X). 136

Sustainable Energy Technologies and Assessments 29 (2018) 131–138

A. Balogun Mohammed et al.

[8] FAOStat, “Forestry Statistics – Forestry Production and Trade,” Food and Agriculture Organization of the United Nations, [Online]. Available: http://www. fao.org/. [9] Michel G, Kendall MD. Charcoal production through distillation of wood, perhaps the key to the deforestation of Haiti. J Haitian Stud 2013;19(1):282–7. [10] Churches C, Wampler P, Sun W, Smith A. Evaluation of forest cover estimates for Haiti using supervised classification of Landsat data. Int J Appl Earth Obs Geoinf 2014;30:203–16. [11] Stevenson G. The production, distribution, and consumption of fuelwood in Haiti. J Dev Areas 1989;24:59–76. [12] Grimsby L, Rajabu H, Treiber M. Multiple biomass fuels and improved cook stoves from Tanzania assessed with the Water Boiling Test. Sustainable Energy Technol Assess 2016;14:63–73. [13] Pesa I. Sawdust pellets, micro gasifying cook stoves and charcoal in urban Zambia: understanding the value chain dynamics of improved cook stove initiatives. Sustainable Energy Technol Assess 2017;22:171–6. [14] Lask K, Booker K, Gadgil A. Lessons learned from a comparison study of charcoal stoves for Haiti. Sustainable Energy Technol Assess 2017;22:188–93. [15] Lask K, Booker K, Han T, Granderson J, Yang N, Ceballos C, et al. Performance comparison of charcoal cookstoves for Haiti: laboratory testing with water boiling and controlled cooking tests. Energy for Sustainable Devel 2015;26:79–86. [16] MacCarty N, Still D, Ogle D. Fuel use and emissions performance of fifty cooking stoves in the laboratory and related benchmarks of performance. Energy Sustainable Dev 2010;14(3):161–71. [17] de Dieu Hakizimana J, Kim H. Peat briquette as an alternative to cooking fuel: A techno-economic viability assessment in Rwanda. Energy 2016;102:453–64. [18] Mwampamba T, Owen M, Pigaht M. Opportunities, challenges and way forward for the charcoal briquette industry in Sub-Saharan Africa. Energy Sustainable Dev 2013;17:158–70. [19] Okoko A, Reinhard J, von Dach S, Zah R, Kiteme B, Owuor S, et al. The carbon footprints of alternative value chains for biomass energy for cooking in Kenya and Tanzania. Sustainable Energy Technol Assess 2017;22:124–33. [20] Luoga E, Witkowski E, Balkwill K. Economics of charcoal production in miombo woodlands of eastern Tanzania: some hidden costs associated with commercialization of the resources. Ecol Econ 2000;35:243–57. [21] Lemus R, Lal R. Bioenergy crops and carbon sequestration. Crit Rev Plant Sci 2005;24(1):1–21. [22] Prochnow A, Heiermann M, Plochl M, Hobbs P. Bioenergy from dedicated farmland and permanent grassland – A review: 2. Combustion. Bioresour Technol 2009;100:4945–54. [23] Rahman M, Mostafiz S, Paatero J, Lahdelma R. Extension of energy crops on surplus agricultural lands: A potentially viable option in developing countries while fossil fuel reserves are diminishing. Renew Sustain Energy Rev 2014;29:108–19. [24] Fontoura C, Brandao L, Gomes L. Elephant grass biorefineries: towards a cleaner Brazilian energy matrix? J Cleaner Prod 2015;96:85–93. [25] Mohammed I, Abakr Y, Kazi F, Yusup S, Alshareef I, Chin S. Comprehensive characterization of Napier grass as a feedstock for thermochemical conversion. Energies 2015;8(5):3403–17. [26] Osava M, Elephant Grass for Biomass. In: Inter Press Service News Agency, 10 October 2007. [27] Matali S, Rahman N, Idris S, Yaacob N, Alias A. Lignocellulosic biomass solid fuel properties enhancement via torrefaction. Procedia Eng 2016;148:671–8. [28] Strezov V, Evans T, Hayman C. Thermal conversion of elephant grass (Pennisetum Purpureum Schum) to bio-gas, bio-oil and charcoal. Bioresour Technol 2008;99(17):8394–9. [29] Felfli F, Luengo C, Rocha J. Torrefied briquettes: technical and economic feasibility and perspectives in the Brazilian market. Energy Sustainable Dev 2005;9(3):23–9. [30] Zhang J, Osmani A, Awudu I, Gonela V. An integrated optimization model for switchgrass-based bioethanol supply chain. Appl Energy 2013;102:1205–17. [31] Sharifzadeh M, Richard C, Liu K, Hellgardt K, Chadwick D, Shah N. An integrated process for biomass pyrolysis oil upgrading: a synergistic approach. Biomass Bioenergy 2015;76:108–17. [32] Clausen L, Elmegaard B, Houbak N. echnoeconomic analysis of a low CO2 emission dimethyl ether (DME) plant based on gasification of torrefied biomass. Energy 2010;35(12):4831–42. [33] Beagle E, Belmont E. Technoeconomic assessment of beetle kill biomass co-firing in existing coal fired power plants in the Western United States. Energy Policy 2016;97:429–38. [34] Geraili A, Salas S, Romagnoli J. A decision support tool for optimal design of integrated biorefineries under strategic and operational level uncertainties. Ind Eng Chem Res 2016;55:1667–76. [35] Paap S, West T, Manley D, Steen E, Beller H, Keasling J, et al. Biochemical production of ethanol and fatty acid ethyl esters from switchgrass: a comparative analysis of environmental and economic performance. Biomass Bioenergy 2013;49:49–62. [36] Trivedi P, Malina R, Barrett S. Environmental and economic tradeoffs of using corn stover for liquid fuels and power production. Energy Environ Sci 2015;8(5):1428–37. [37] Di Lorenzo G, Pilidis P, Probert D. Monte-Carlo simulation of investment integrity and value for power-plants with carbon-capture. Appl Energy 2012;98:467–78. [38] Hsu D. Life cycle assessment of gasoline and diesel produced via fast pyrolysis and hydroprocessing. Biomass Bioenergy 2012;45:41–7. [39] Shahrukh H, Oyedun A, Kumar A, Ghiasi B, Kumar L, Sokhansanj S. Techno-economic assessment of pellets produced from steam pretreated biomass feedstock. Biomass Bioenergy 2016;87:131–43. [40] Joseph M, Wang F. Population density patterns in Port-au-Prince, Haiti: a model of

a sales price for crop-derived charcoal was set equal to the price that customers currently pay for wood-derived charcoal, the difficulties in breaking into charcoal markets has been acknowledged by other research [18]; thus, tax incentives for the production and use of sustainable charcoal can help encourage its success. Tax credits and other incentives might also be sought for economic development, as the proposed business would employ 14 people directly, which is reflected in the labor cost portion of TOCplant , where labor costs are the second most significant component of TOCplant at USD 18,600 per year. Conclusions Deforestation is a serious and ongoing issue for Haiti, in part due to widespread usage of trees for production of charcoal, which is widely used as a cooking fuel throughout the country. A technoeconomic assessment of the potential for dedicated bioenergy crops to replace trees as feedstock for charcoal production in Haiti was conducted. The objective of the analysis was to determine the potential for a profitable entrepreneurship venture to produce and sell sustainable, crop-derived charcoal. A net present value (NPV) model was developed which accounted for installation costs to develop a charcoal production plant with a co-located farm for feedstock growth. Operating costs, which included feedstock, labor, taxes, and other operating expenses were accounted for, as well as revenue from the sale of charcoal. A Monte Carlo analysis approach was used to account for uncertainty and variability in model inputs in assessment of plant NPV. As a result, a probabilistic NPV output is achieved which gives not only the NPV based upon mean model input values, but also the probability of deviation from this mean-derived NPV. The proposed charcoal production plant was sized for an annual output of 7.6 × 105 kg·yr−1, which would serve approximately 1.3% of the Port-Au-Prince market. The results indicate that the plant is likely to yield positive return on capital investment, with a 91% likelihood of the plant breaking even and an 84% likelihood of achieving a 25% ROI within 10 years. Further analysis of the model results and the impact of model input variability revealed that charcoal sale price and feedstock cost are the most significant variables that affect plant profitability and profitability uncertainty. Based upon model results, suggestions for improved economic feasibility are presented, including the possibilities of selling charcoal on a contractual basis to hedge against the potential fluctuation in charcoal sale price, using a different feedstock such as bagasse for blending with elephant grass, and getting tax credits based on the environmentally and societally beneficial nature of this venture. Acknowledgements The authors thank Carbon Roots International for their valuable insights on the study, including the provision of economic and logistic information pertaining to Haiti. References [1] Munasinghe M. Environmental economics and biodiversity management in developing countries. Ambio 1993;22(2/3):126–35. [2] Aabeyir R, Adu-Bredu S, Agyare W, Weir M. Empirical evidence of the impact of commercial charcoal production on Woodland in the Forest-Savannah transition zone, Ghana. Energy Sustainable Dev 2016;33:84–95. [3] Rembold F, Oduori S, Gadain H, Toselli P. Mapping charcoal driven forest degradation during the main period of Al Shabaab control in Southern Somalia. Energy Sustainable Dev 2013;17(5):510–4. [4] International Energy Agency (IEA), “World energy outlook 2004: energy and development,” International Energy Agency (IEA), Paris, 2004. [5] Dolisca F, McDaniel JM, Teeter LD, Jolly CM. Land tenure, population pressure, and deforestation in Haiti: the case of Forêt Des Pins reserve. J Forest Econ 2007;3(4):277–89. [6] Haiti: Strategy to Alleviate the Pressure of Fuel Demand on National Woodfuel Resources, The World Bank - Energy Sector Management Assistance Program, Washington, DC; 2007. [7] Ghilardi A, Tarter A, Bailis R. Potential environmental benefits from woodfuel transitions in Haiti: geospatial scenarios to 2027. Environ Res Lett 2018;13:1–11.

137

Sustainable Energy Technologies and Assessments 29 (2018) 131–138

A. Balogun Mohammed et al.

Latin American city? Cities 2010;27(3):127–36. [41] Ponsar F, Ford N, Van Herp M, Mancini S, Bachy C, “Mortality,. violence and access to care in two districts of Port-au-Prince, Haiti,“ Conflict and. Health 2009;vol:3. [42] Khristova P, Khalifa A. Carbonization of some fast-growing species in Sudan. Appl Energy 1993;45(4):347–54. [43] De Conto D, Silvestre W, Baldasso C, Godinho M. Performance of rotary kiln reactor for the elephant grass pyrolysis. Bioresour Technol 2016;218:153–60. [44] Mwendia S, Yunusa I, Whalley R, Sindel B, Kenney D, Kariuki I. Use of plant water relations to assess forage quality and growth for two cultivars of Napier grass (Pennisetum purpureum) subjected to different levels of soil water supply and temperature regimes. Crop Pasture Sci 2013;64:1008–19. [45] Lin J-C. Development of a high yield and low cycle time biomass char production system. Fuel Process Technol 2006;87(6):487–95. [46] Walmsley D, Sargeant V, Dookeran M. Effect of fertilizers on growth and composition of elephant grass (“Pennisetum purpureum”) in Tobago, West Indies. Tropical Agric 1978;55(4):329. [47] Moon J, Lee J, Lee U. Economic analysis of biomass power generation schemes

[48]

[49]

[50]

[51] [52]

138

under renewable energy initiative with Renewable Portfolio Standards (RPS) in Korea. Bioresour Technol 2011;102(20):9550–7. Moore MA, Boardman AE, Vining AR. More appropriate discounting: the rate of social time preference and the value of the social discount rate. J Benefit-Cost Anal 2013;3(4):401–9. Garcia-Perez M, Nunez J, Pelaez-Samaniego M, Flora G. Sustainability, business models, and techno-economic analysis of biomass pyrolysis technologies. Innovative Solutions in Fluid-Particle Systems and Renewable Energy Management. IGI Global; 2015. p. 298–342. HaitiLibre, Haiti - FLASH : Full details on the new minimum wage, HaitiLibre, 25, Online Available: http://www.haitilibre.com/en/news-17541-haiti-flash-fulldetails-on-the-new-minimum-wage.html May 2016 Accessed 03 January 2017. Timyan J, Yo Bwa. Important trees of Haiti, South-East consortium for. Int Dev 1996. Whisenant S, Burzlaff D. Predicting green weight of mesquite (Prosopis glandulosa Torr.). J Range Manag 1978;31(5):396–7.