General approach
The techno-economic assessment was performed in accordance with Zimmermann et al. (2020) in order to construct the methodology section as transparent and coherent as possible. This TEA guideline allows TEA to be conducted in parallel to life cycle assessment (LCA) which emerges in aligned vocabulary and assessment steps, and it applies many concepts from ISO 14,044 (ISO Organisation 2006). TEA should cover the process steps illustrated in Fig. 1. Similar to LCA, the construction of TEA is an iterative process.
Goal and scope clarified the context of the study and the reasons for carrying it out. Moreover, the methodological framework was assessed in this section (e.g., functional unit, system boundaries). The inventory was covered by the establishment of the microalgae cultivation model used in this study. The cost assessment examined investment cost and operating cost as well as the NPV and the ROI as significant indicators for the economic potential of the model. Price forecasts for the most important input materials during the whole lifetime of the facility were conducted. A following scenario analysis considered the variation of both relevant economic and technical system parameters.
Goal, scope and target product
The overall goal of this study was to evaluate the economic potential of microalgae biomass production for food in a tubular photobioreactor in a humid continental climate. For that matter, the costs for generic industrial scale microalgae cultivation were assessed in a techno-economic assessment (TEA) applying the NPV and the ROI. Furthermore, major economic and technical hurdles were identified in a scenario analysis where relevant alterations of system parameters were tested. System-boundaries comprised all relevant processes and input materials of the cultivation stage up to the dry biomass (target product) (Fig. 2). One scenario considered EPA-rich microalgae oil and residual protein-rich biomass as the final products. This scenario thus moreover comprised an oil extraction stage. The evaluation of the economic potential was conducted in EUR. The inputs in Fig. 2 refer to the baseline scenario and the scenario considering microalgae oil as the target product.
Microalgae cultivation model
Input flows of microalgae cultivation were based on the design model of a previous study of the authors which evaluated the environmental impacts of cultivation processes in a hypothetical 628 m3 tubular photobioreactor located in Halle/Saale, Central Germany (Schade and Meier 2020). Dry biomass for human nutrition was produced in borosilicate glass tubes with a diameter of 40 mm (baseline scenario) and a wall thickness of 2 mm. Borosilicate glass as the material for the tubes was contemplated as superior over other possible tube materials, such as polymethyl methacrylate or silicone as it has a long lifespan of at least 50 years (Schultz and Wintersteller 2016). Moreover, it has an excellent translucence, which does not degrade under solar radiation to enable the photoautotrophic process (Schultz and Wintersteller 2016), and it is a safe material to use for the production of edibles. Further construction materials (aluminum, steel, synthetic rubber) were designed in accordance with the study by Pérez-López et al. (2017). Cultivation took place from mid-April until mid-October, because an evaluation of the climatic characteristics of the location suggested this period to be most efficient for the cultivation of microalgae, taking into account maximum, minimum and mean daily temperatures as well as solar insolation.
The calculation of the yields and productivity was based on the climatic conditions of the site, which were drawn from detailed satellite data provided by the NASA Power Data Access Viewer (NASA—National Aeronautics and Space Administration 2019). The location shows temperatures slightly below − 3 °C in the coldest months of the year (January and February) and average temperatures between 14.5 and 20.3 °C from May until September. April and October portray mean temperatures of slightly below 10 °C which is why it was assumed that on average, half of these months, cultivation was feasible. Solar insolation is highest from May until August with 16.1–18.2 MJ/m2/day and still reaches 14.1 MJ/m2/day in April and 10.8 MJ/m2/day in September. The month with the lowest radiation during the cultivation season is October with 5.6 MJ/m2/day on average. Table 1 lists the essential system parameters and input materials of all scenarios, which were based on a technical variation in the system (for scenario overview see chapter 2.6 Sensitivity analysis). All other scenarios that are not portrayed in Table 1 relied on a variation of economic parameters and thus were equal to the baseline scenario. The plant was modeled with a lifespan of 30 years. The system model relied on a combination of sources. Resource consumptions of major processes during the cultivation were based on own calculations and literature data. Additionally, some input parameters relied on expert advice. More specifically, the nutritional values of the microalgae species considered in this study are mean values compiled from the literature. Likewise, the photoconversion efficiency (PCE) is a mean value, which was obtained from several studies that had all applied a similar cultivation system and taken into account all relevant restraining parameters. Those parameters comprised light saturation and photoinhibition, practical losses through reflection, inactive absorption, respiration, and high oxygen rates (Lundquist et al. 2010; Posten 2012; Skarka 2012; Benemann 2013; De Vree et al. 2015).
Table 1 Parameters and input materials in microalgae cultivation scenarios (the origin of the values presented are described in the text) The nutrient demand was calculated according to the protein content of the microalgae species under implementation of the nutrient-to-protein conversion factor, which was adopted from the literature (Templeton and Laurens 2015). The carbon dioxide demand was in accordance with the literature, too (Chisti 2007; Patil et al. 2008; Lardon et al. 2009; Ma et al. 2016). Usage of hydrogen peroxide and hypochlorite was estimated after Pérez-López et al. (2017). The electricity consumption and demand for natural gas resulted from different processes in the cultivation model, which were all based on own calculations. These processes comprised water pumping, aeration with CO2, mixing of the suspension in the tubes, centrifugation and drying of the algae biomass (Schade and Meier 2020). Total uncertainty of the electrical processes is 1.07, and total uncertainty of the remaining foreground processes (photobioreactor materials, nutrients, water use, land use) is 1.13. The complete system model can be accessed in the previous study of the authors (Schade and Meier 2020).
Cost assessment
For the compilation of the cost assessment, a net cash flow table for the construction phase and the first 30 years of operation was established in order to calculate the NPV (discounted cash flow) and the ROI from it. The net cash flow is obtained by subtracting all cost paid from the received benefits while including cost for a capital loan, such as capital payback and interest rate (Lauer 2008). The first two years were supposed to be the construction phase of the plant without incoming benefits. Investment costs were evenly spread over these two years. For the following first 6 years of cultivation, a reduced production was modeled to account for the starting phase of the plant with the benefits doubling in the second year of production and a growth of 6% for each of the subsequent five years until full production is achieved. A bank loan covered the investment cost with an interest rate of 2.05% (average interest rate of the months 01/2019 to 09/2019 for Germany) (Trading Economics 2019). The net present value of a given time period NPVn is calculated from Eq. 1 (Short et al. 1995) where NCF is the net cash flow of a time period n, and d is the nominal discount rate, which also covers inflation and is set here to 12% (Thomassen et al. 2016; Walsh et al. 2018).
$${\text{NPV}}_{n} = \frac{{{\text{NCF}}}}{{\left( {1 + d} \right)^{n} }}$$
(1)
For better comparability between the assessed scenarios, the ROI is calculated in addition to the NPV. The ROI indicates the percentage of a net value that is received from an investment over a given period of time and was calculated from the NPV of a given period of time n (Eq. 2) (Chen 2020a).
$${\text{ROI}} = \frac{{\mathop \sum \nolimits_{3}^{n} {\text{NPV}}_{n} - C_{{{\text{tot}}}} }}{{C_{{{\text{tot}}}} }}$$
(2)
The ROI is estimated for every year of plant operation, starting from the third year when facility operation begins. For every time period n, the sum of the net present values from year three to year n is calculated. The total costs Ctot represent the sum of payments assigned in the first and second year when all investment costs were paid. Since the investment project covers a timeframe of 30 years and the ROI does not consider length of time, the calculation of the annualized ROIA gives a more representative result. The ROI can be annualized following Eq. 3 (Chen 2020b) where n is the number of years for the investment.
$${\text{ROI}}_{A} = \left[ {\left( {1 + {\text{ROI}}} \right)^{\frac{1}{n}} - 1} \right] \times 100\%$$
(3)
Although the facility was modeled as an ‘nth’ plant which presupposes a mature state of facility design and technique, there can still be contingencies that have not been accounted for in the model. In order to avoid an underestimation of the costs in the TEA and get a too optimistic result, a contingency factor of 1.25 was put on the overall costs (Lauer 2008).
Price forecasts for input materials
Prices for input materials that are constantly required in a great amount over the lifetime of the facility were analyzed in a forecast. Since the facility is supposed to run with a 30-year-lifespan, analyses of future prices intended to portray—as accurately as possible—representative prices at a given point in time. Historical prices were obtained for a 15-year period from 2004 to 2019, except electricity, for which the annual price was observed from 2000 to 2018. Prices were recorded for the materials ammonium, phosphate fertilizer and natural gas (World Bank 2019), as well as electricity (Eurostat 2019). The prices for carbon dioxide were investigated using the commodity price index (U.S. Bureau of Labor Statistics 2019) and a reference year price from 2016 (Thomassen et al. 2016). Concerning electricity, location-specific prices for Germany were applied. For natural gas, European prices were utilized. When prices were given in US-Dollar, they were converted based on the respective monthly currency rate [UKForex Limited (OFX) 2019]. It was relied on the Geometric Brownian Motion (Eq. 4) in order to estimate future prices of required resources (Hilpisch 2014).
$$\Delta S\left( t \right) = S_{t - 1} \mu \Delta t + S_{t - 1} \sigma \varepsilon \sqrt {\Delta t}$$
(4)
ΔS(t) is the price change of a respective resource over a given time period which changes as a function of the drift (\(\mu\)), the volatility (\(\sigma\)) and a Wiener process of random values (ε). The time period t comprises one month. The calculation is based on a starting price St−1. Whereas the drift indicates the deterministic trend of the process, the volatility controls the influence that coincidence has on the process S(t). The drift and volatility are both constants and are estimated from the mean and standard deviation of historical returns (\(\mu_{t}\)), which in return are drawn from the price (st) of each time period (Eq. 5).
$$\mu_{t} = \frac{{s_{t} - s_{t - 1} }}{{s_{t - 1} }} \times 100$$
(5)
The simulation with the Geometric Brownian motion was repeated 1000 times for each commodity and future prices analyzed according to their mean, median and 90% confidence interval. A 90% confidence interval was considered appropriate, given that these results are meant to be used in the sensitivity analysis. Thus, they should not be characterized by extreme values at the lower and upper end. A timespan of 30 years in the future was portrayed by means of monthly prices (Fig. 3).
Sensitivity analysis
The sensitivity analysis was based on both changes in technic and economic parameters in order to analyze the volatility of the system model. Besides the earlier described baseline scenario, the following alterations were made to the model.
Min/max commodity prices
In the baseline scenario, the mean predicted commodity prices calculated in the forecast were used for the cost assessment. However, forecasts, especially over a long time period, are rather volatile and subject to a great amount of unpredictable parameters. In order to evaluate the influence of possibly higher or lower developments of the commodity prices, the upper and lower ends of the 90% confidence intervals were included in the assessment.
Long/short production periods
The selected location in Central Germany as part of the humid continental climate zone is subject to variations in seasonal temperatures. As a result, cultivation season lengths can fluctuate drastically with a focus on the months April and October. Both can be rather warm and thus suitable for microalgae cultivation, or cold on too many days of the month with minimum temperatures dropping below 0 °C. The system model was thus evaluated applying constantly longer and shorter cultivation periods. As can be seen in the inventory table (Table 1), the length of the production period influenced the yield tremendously. Even though productivity decreases in spring and fall due to a lower solar insolation, a longer cultivation season is favorable and expands the yield. This climatic dependence was considered a key element of the cultivation in a ‘cold-weather’ climate and was hence analyzed in the cost assessment, too.
Tube price
The borosilicate glass tubes used in the photobioreactor are one of the main materials needed for microalgae cultivation and constitute one of the biggest investments concerning the whole facility. Consequently, it was tested how a fluctuation in the cost of a rather big share of the investment affects the whole cost assessment of the plant.
Variations in selling price
The agreement on a specific reasonable selling price is a crucial step in the cost assessment as it here generates the only source for benefits. The selling price used in this study was determined according to values found in the literature. Yet, a sample analysis on microalgae biomass products on the German market was conducted which revealed great fluctuations in the selling price. Price variations within the same product were due to package size where a bigger size generated a discount. Among the different products, prices differed at times tremendously without evident reason. At the same time, it was not obvious where and how the biomass had been produced. However, even with the economy of buying the biggest package, industry net selling prices still outnumbered the selling price suggested in the literature to a great extent with the literature price of EUR 38 per kg dry mass hardly being in the low range of the 90% confidence interval (Fig. S1). However, the mean value of the market analysis resulted in a net selling price (EUR 109 per kg)—equaling 187% of the literature price. The results of the market analysis can be found in the supplementary material (Tab. S1, Fig. S1). Consequently, these possible variations were accounted for in the sensitivity analysis by altering the selling price of the biomass by 5% and 15%.
Alternative microalgae species
In a previous study of the authors, it has been shown how the choice of microalgae species can alternate the environmental impacts, depending on the target product (Schade et al. 2020). For that matter, the cost assessment was repeated using P. tricornutum as the cultivated species in the PBR. Phaeodactylum tricornutum also is an oleaginous microalga with a nutritional profile that is similar to Nannochloropsis sp. However, in comparison with the latter, P. tricornutum possesses a higher percentage of protein (36.4%) (Rebolloso-Fuentes et al. 2007) and slightly less EPA + DHA (3.25%) (Zhukova and Aizdaicher 1995; Ryckebosch et al. 2014). The cultivation of another microalga with a comparable nutritional profile not only illuminates possible changes in the outcome due to taxonomical reasons. Because of the similar profile, information can be obtained about how sensitively the system reacts if the composition of one and the same alga fluctuate like it naturally does.
Microalgae oil
In another scenario, it was tested how the choice of a different goal product affects the NPV. In the baseline scenario, the whole biomass was considered as the target product. However, whole microalgae biomass only procures a fraction of the selling price compared to more refined products such as extracted lipids or antioxidants. Thus, the production of microalgal oil from Nannochloropsis sp. as a source for EPA was modeled with the residual protein-rich biomass being sold as an alternative to soybean meal. It was assumed that the whole lipid content of the biomass was extracted using supercritical CO2. In order to extract 1 kg of microalgae oil, 11.7 kg CO2 (Wang et al. 2016) and 58.6 MJ electricity (Shimako et al. 2016) were needed. Based on the inventory data for microalgae cultivation used in this study, 1 kg DM contains 206 g lipids and 42 g EPA. The yearly microalgae oil yield would accordingly add up to 13,225.2 kg lipids with an EPA content of 20.39%. Microalgae omega-3 oil has been reported to realize a net market price between USD 80 and 160 (Borowitzka 2013; Matos 2017; Barkia et al. 2019) which roughly corresponds to EUR 72–144. Yet, a sample market analysis of microalgae oil products for omega-3 PUFAs for the German market resulted in much higher net prices between EUR 240 and 900 (mean: EUR 452) per kg microalgae oil with an EPA + DHA share of 40–50%. Remarkably, the prices for microalgae oil were not correlated with the actual concentration of EPA and DHA in the product which provoked a massive fluctuation of the EPA + DHA net price across the products with around EUR 530–2300 (mean: EUR 1025) per kg EPA + DHA. In order to obtain a reasonable selling price, the mean value of the EPA + DHA kilogram prices was multiplied with the percentage of EPA contained in the microalgae oil. Hence, a net selling price of EUR 209 per kg microalgae oil was estimated in this study. For each kilogram microalgae oil, 3.85 kg protein-rich residual biomass was generated. The unit price for the residual protein-rich biomass was assumed to amount to 0.44 EUR/kg DM (van der Voort et al. 2017). The results of the market analysis on microalgae oil products can be accessed in the supplementary material (Tab. S2, Fig. S2). In order to conduct a profound analysis, the cost assessment for microalgae oil as the target product was repeated using the highest price suggested in the literature (144 EUR/kg oil).