Torrefaction of almond shell as a renewable reinforcing agent for plastics: techno-economic analyses and comparison to bioethanol process

In 2016, the US state of California alone produced nearly 3.5 billion kilograms of almonds, accounting for approximately 84% of the world’s almond production. This generated about 2.58 million metric tons (MTs) of almond residues. Almond shells are currently either burned to generate power or disposed of in landfill. Valorizing almond shells and hulls provides an opportunity to replace petroleum-derived products and divert organic material from landfill. Here we demonstrate a detailed techno-economic analysis (TEA) of an almond shell torrefaction process capable of utilizing the 520 000 MTs of almond shells produced annually in California. Our process also includes preprocessing the torrefied biomass to exploit it as a reinforcing agent for plastics. We further compared the revenue generated from the torrefied biomass and bioethanol derived from the same quantity of almond shells. We considered three different torrefaction facility scales to evaluate trade-offs between economies of scale at the facility and trucking costs to deliver almond shells. A facility that takes in 200 000 MT yr–1 of almond shells results in lower per-unit-output basis capital and operating cost relative to other smaller-scale torrefaction facilities, including 10 000 MT yr–1 and 50 000 MT yr–1, considered for analysis in this study. The large-sale facility results in a minimum selling price (MSP) of the torrefied biomass of $311.4 MT–1. An analogous TEA on converting almond residues into bioethanol is also investigated. The MSP of almond shell derived ethanol ($1.71 kg−1) is higher than that of corn ($0.48 kg−1) or cellulosic biomass ($0.88 kg−1) derived ethanol. Compared with the bioethanol route, the torrefied almond shells result in three times more revenue if utilized as a reinforcing agent for plastics.


Introduction
Almond is one of the most valuable agricultural commodities. California alone accounts for approximately 84% of the world's almond production [1,2]. In 2016, 3.52 million metric tons (MTs) of almonds were produced, out of which 19% and 54% were shells and hulls, respectively [3]. The almond post-harvest processing facilities could sell the hulls and shells directly to the dairy industry. According to the California Food Processing Residue Assessment in 2010, the market prices of almond hulls and shells are in the range of $120-$160 MT -1 for hulls and $15-$25 MT -1 for shells [4]. However, moisture or rain affect the quality and application of the stored almond hulls and shells [5]. These negative factors could significantly reduce their market values and consequently result in a large quantity of almond residues being disposed of as waste each year [6]. Ideally, almond hulls and shells should be utilized or preprocessed immediately after harvesting, which minimizes exposure to the environment, reduces dry matter losses and helps to maintain their quality.
The abundance of almond hulls and shells has attracted many researchers to identify economic and practical ways of utilizing them. Aktas et al [7] characterized the properties of commercially available almond hulls and shells for thermochemical conversion. Using these renewable residues in co-generation facilities could support state-level renewable portfolio standards and reduce greenhouse gas (GHG) emissions from fossil fuel-based power plants. Burning 2.7 million MTs of almond hulls and shells generates 475 MW of electricity which could displace 0.19 million MTs of coal [7]. However, this converting pathway is still economically unfavorable because the cost for almond shell-generated electricity is 18 cents kWh −1 [8], compared with 5.7 cents kWh −1 for solar photovoltaics and 3.9 cents kWh −1 for onshore wind [9].
Almond shells and hulls could also be used for bioethanol production through fractionation or pretreatment followed by saccharification, fermentation and ethanol recovery [6,10,11]. Gong et al [12] evaluated an almond shell fractionation process consisting of hot water pretreatment coupled with organic solvent (a mixture of ethanol and water) pretreatment. This two-stage fractionation process released nearly 100% of the glucose available in almond shells, which could be further converted into ethanol. Kacem et al [13] reported a multistage almond shells-to-ethanol production process, including acid and alkaline pretreatments, enzymatic saccharification, laccase detoxification and Saccharomyces cerevisiae yeast fermentation. Their process results in 5.8 g of ethanol from 100 g of almond shells.
In recent years, torrefaction, a thermal process that treats biomass in the absence of oxygen at a temperature of 200 • C-300 • C has gained substantial interest for the production of high-density fuels or materials [14]. By removing moisture and volatile compounds, the torrefied biomass would be rich in carbon and contain less oxygen and hydrogen than the raw biomass. Torrefaction has been investigated for valorizing low-value agricultural byproducts such as rice straw, corn stover and nutshells [14][15][16][17]. Luo et al [18] developed a one-dimensional model that considers internal and external heat transfer, particle shrinkage, moisture evaporation and reaction kinetics to predict the torrefaction behavior of biomass. Their modeling result indicated that the torrefaction process is controlled by both heat transfer and kinetics. The intra-particle temperature gradient in the torrefied process has also been studied by Krochmalny et al [19]. They demonstrated that the moisture content and particle size are important factors in thermal runaway of torrefaction. For thermally thick particles of wood the temperature gradient can exceed 50 • C. Chiou et al [20] used response surface methodology to examine the effects of torrefaction temperature and time on the mass and energy yields of the torrefied biomass. The condensate yields and gross calorific values increased with torrefaction severity. They also examined the effect of inorganic species on the torrefaction kinetics of almond shells using thermogravimetric analysis, and demonstrated that the kinetic model with two consecutive parallel reactions exhibited the best fit to experimental data [21]. Lin and Zheng [22] reported that GHG emissions were reduced by 58.5%-82.5% with the combustion of the biochar obtained from the torrefaction process compared with the combustion of coal.
Torrefied biomass can also be used as a biodegradable and low-cost filler in composite materials [11,23,24]. Incorporating torrefied biomass into a polymer matrix can improve its mechanical and thermal properties [25]. Lashgari et al [26] studied the effect of the composition of nano-clay and almond shell powder on the strength and properties of polypropylene. They observed that the tensile modulus of pure polypropylene (0.92 GPa) is enhanced by at least 29.8% when almond shell powder is added. Altay et al [27] investigated mechanical and thermal properties of linear low-density polyethylene (LLDPE)-based composite with different proportions of almond shell powder. Filling LLDPE with 40 wt% almond shell increased the flexural strength of LLDPE by 100%. McCaffrey et al [28] revealed that torrefied almond shells (TASs) can be used as a bio-reinforcing agent for recycled polypropylene-polyethylene blends without any compatibilizers.
In this study, the preliminary almond shell torrefaction process was modeled considering production-scale data from California's almond processing facilities and farms. We performed a detailed techno-economic analysis (TEA) and identified the key operating factors that affect the process economics. This study provides guidance for the design and optimization of the almond shell valorization process. As a comparison, we performed an analogous TEA on converting almond residues into bioethanol.

Methods
SuperPro Designer v12.0 process modeling software developed by Intelligen, Inc., Cambridge, MA, USA was used to develop TEA models of the selected production processes. The data used in modeling almond shell torrefaction were provided by the United States Department of Agriculture (USDA) and the Almond Board of California. Modeling data inputs for the almond shell-to-bioethanol production process were collected from prior studies [13,29]. The ethanol production model was developed based on prior TEA models developed at the Lawrence Berkeley National Laboratory (LBNL) [30][31][32] and National Renewable Energy Laboratory (NREL) [33]. The results from material and energy balances obtained from the SuperPro model serve as the main input for economic analysis.

Process design and modeling descriptions
The almond shells are assumed to be collected locally from farms. After milling into small pieces, shells are then conveyed to large-scale biorefineries for torrefaction. The torrefied shells are designed to sell as a filler for plastic composites or a colorant for recycled plastics. The flowsheet for the almond shell torrefaction process considered for analysis in this study is shown in figure 1. The whole torrefaction process has been divided into three sections: feedstock transportation and handling, torrefaction and milling torrefied biomass. The torrefaction parameters assumed for process simulation are based on literature reports [17,20,21]. The energy of the byproduct (torgas) has been recovered and considered in the calculation of utility costs by the software.
Almond production in California spans 680 km throughout the Central Valley, the geographic location map is shown in the supporting information (SI) figure S1. For the baseline analysis, it was assumed that three satellite centers (North, Central Valley and South) could process 6 000 000 dry MTs of almond shells per year in California. The design capacity covers the current available almond shells in California (520 000 MT) and considers the losses during collection and delivery as well as the potential annual production growth. The nameplate capacity of the satellite plant is designed to utilize 200 000 MT of dry almond shells per year. Moreover, we also explore scenarios with smaller almond shell torrefaction plants (50 000 MT yr −1 and 10 000 MT yr −1 ) to evaluate the tradeoff between a lower trucking cost and a larger number of satellite torrefaction facilities. The major input parameters associated with the almond shell torrefaction process are listed in table 1. Within each scenario, there are sources of uncertainty that will impact the minimum selling price (MSP) of the torrefied biomass. Addressing this limitation, our analysis includes single-point sensitivity analysis and stochastic uncertainty analysis using Monte Carlo simulation. A single-point sensitivity analysis was performed by modeling all inputs with a uniform probability distribution and varying them individually. For the Monte Carlo simulations, probability distributions were assigned to each parameter including materials cost, selling price, utility cost, labor cost and the total capital investment (TCI). The simulation ran for 5000 Monte Carlo trials.

Economic analysis
The economic analyses were performed by estimating the total capital and operating costs. Similar TEA models developed previously at LBNL [30][31][32] and NREL [33] and the chemical engineering handbook [34] were used to estimate process equipment costs and direct/indirect costs of the torrefaction and ethanol production facilities. Consumption of process chemicals and utilities was estimated from the process simulations and is listed in table S2. The construction time was set to be 24 months, and the start-up time was assumed to be 6 months. The annual operating time was assumed to be 7920 h (330 days/year and 24 h/day). This study considered the life of a plant to be 15 years. The materials and utilities required for the torrefaction process under different scenarios are summarized in table 2.
The detailed TCI is listed in table 3. The TCI was calculated using the following equations, where the fixed capital investment (FCI) includes the cost of equipment and facilities: The annual operating cost (AOC) is divided into seven categories: materials cost, utility cost, labor cost, fixed charges, plant overhead cost, other production costs and general expenses. The fixed charges cover the cost for maintenance and repairs, operating supplies fee and patents and royalties. The local taxes and insurance are included in the other production costs category. The underlying economic assumptions and sources are shown in table 4. The production costs of torrefied biomass and ethanol were estimated using the discounted cash flow rate of return analysis that accounts for capital investment, AOC and time value of money. The unit production cost is estimated by dividing the AOC by the annual production. We assumed an internal rate of return after taxes of 10%, a plant lifetime of 15 years, plant operating hours of 7920 h (330 days/year and 24 h/day) and income tax of 21%. The MSP was calculated by setting net present value to zero. The detailed TEA financial parameters are listed in table S4. We further estimated gross profit and return on investment (ROI) using the following equations [35]: return on investment (%) = gross profit/TCI × 100.
(4)  Figure 2 shows the TCI for small-, medium-and large-scale torrefaction facilities considered in this study to fully utilize the available almond shells in California (520 000 MT yr −1 ). Details of the contribution from each conversion stage to the TCI are listed in table 3. The results show that building several small-scale torrefaction facilities with a nameplate capacity of 10 000 MT yr −1 requires ten times higher capital investment relative to a proposed large-scale facility with a nameplate capacity of 200 000 MT yr −1 . The differences in the TCI with small-and large-scale facilities are due to the economy of scale as the equipment cost is not linearly dependent on the equipment size. As shown in table 3, the purchased equipment cost is the main contributor (~10% of the TCI) among all the nine categories in direct cost.

Annual operating cost
The itemized AOC of the torrefied biomass with three different facility sizes is shown in figure 3. Details of raw material/utility costs and the breakdown of the various operating costs is provided in tables 4 and S2. The material cost is the largest cost component. This is consistent with the results of most TEAs that raw materials are often the largest contributor. As expected, the increase in plant scale leads to a decrease in unit production cost. The production costs of the torrefied biomass are $270.93 MT -1 , $362.32 MT -1 and $1197.50 MT -1 , respectively for the large-, medium-and small-scale facilities considered in this study. The raw material and utility costs are almost identical under the different scales of the facility. The transportation costs for small-scale facilities ($15.2 MT -1 ) are less than 50% of that the larger-scale facilities ($30.5 MT -1 ), which is due to the shorter transport distances. But there is no significant difference between 50 000 MT yr −1 ($15.1 MT -1 ) and 10 000 MT yr −1 ($15.2 MT -1 ) plants since transport distance is no longer the dominant parameter (both <60 km). The saving in trucking costs due to the shorter transport distance could not compensate for the higher capital and labor costs required for building and operating several small-scale facilities to fully process the available almond shells. Our results highlight that building three satellite torrefaction facilities would be cost-effective to fully process the almond shells in California.

Sensitivity analysis results
A single-point sensitivity analysis on the MSP of torrefied biomass was performed for the selected large-scale torrefaction facility as a representative case. The uncertainties related to materials cost, labor cost, utility cost and TCI are considered for the sensitivity analysis. As illustrated in figure 4, the labor cost was found to be the most sensitive variable to the production cost of torrefied biomass among the selected inputs. If the labor cost is increased by 20%, the production cost will increase by 7.5%. The delivered cost of almond shells was another sensitive input variable. A 20% variation in the delivered almond shell cost alters the production cost by 6.8%. A Monte Carlo simulation was performed to demonstrate uncertainties associated with the estimated ROI. The probability distributions were assigned to several input parameters, including the delivered cost of almond shells, the selling price of torrefied biomass, utility cost, labor cost and TCI. The detailed assumptions are summarized in table 5. The simulation ran for 5000 Monte Carlo trials and the results are shown in figure 5. Based on the Monte Carlo simulation, the likelihood of achieving a positive ROI is 88.5%. The probability of achieving an ROI greater than 20% is 77.5%. These likelihoods can be further increased by optimizing and/or fixing the most sensitive cost drivers ( figure 4).

Comparison of torrefied biomass and bioethanol derived from almond shells
For comparison, an almond shell-to-bioethanol production process is modeled that utilizes the same quantity of almond shells (200 000 MT yr −1 ) considered for the baseline torrefaction facility. The bioethanol biorefinery collects available almond shells from the local farms around the biorefinery. At the biorefinery, feedstock (almond shells) is first pretreated with dilute sulfuric acid and then goes through a series of subsequent processes, including pH adjustment, enzymatic hydrolysis, fermentation and distillation to recover ethanol. The major input parameters and corresponding flowsheet for the ethanol production process are documented in table S1 and figure S2. The results from a comparative TEA of the torrefaction of almond shells and almond shells-to-bioethanol production process are summarized in table 6. We observed that the utility cost for the torrefaction process is 1.8 times higher relative to the ethanol process, which is attributed to its harsh operating temperature. However, the bioethanol production process requires higher capital and operating costs due to its comprehensive pretreatment and bioconversion processes. Furthermore, although the unit market value of bioethanol is higher, the production rate for bioethanol is only 21% of torrefied biomass. Furthermore, the MSP of the almond shell derived ethanol ($1.71 kg −1 ) is much higher than the reported values for corn ($0.48 kg −1 ) [36] and cellulosic biomass ($0.88 kg −1 ) [37] derived ethanol. The comprehensive biorefinery process, low biomass-to-ethanol conversion rate and low energy density relative to gasoline makes it challenging for almond shell derived ethanol to gain a competitive advantage in the market. Our findings demonstrate that bioethanol production is less profitable than producing TASs as a reinforcing agent for plastics.

Conclusions
The importance of almond shell valorization increases with increasing demand for almonds. Here, we present the production cost of TASs at different scales and compare the value of TASs with almond shell derived bioethanol. Our results show that commercial deployment of three satellite large-scale torrefaction facilities with a nameplate capacity of 200 000 MT yr −1 reduces the total capital and operating costs compared with other medium-and small-scale torrefaction facilities with a nameplate capacity of 50 000 MT yr −1 and 10 000 MT yr −1 , respectively. Labor and the delivered almond shell costs are the most influential parameters affecting the production cost of torrefied biomass. Monte Carlo simulation indicates that the likelihood of achieving a positive ROI for the torrefaction process is 88.5%. Compared with the bioethanol route, conversion of almond hulls to torrefied biomass results in three times more revenue.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary information files).