1.1 Goal and scope
The LCA methodology was chosen not only considering its relevance as a scientific assessment, but also the current significant impact of LCA in other issues, such as policy development, system, and component design, authorization and permissions as well as consensus-building [20]. Hence, the goal of this LCA is to provide a consistent understanding of the biobutanol production process related to environmental issues. This LCA model was developed together with biologists, technical and process engineers. Figure 1 shows the system boundaries and the considered process flow.
Emerging technologies have considerable potential for improvements, e.g. by applying economies of scale [21], as already successfully shown for other biofuel technologies [22]. The depicted biobutanol production process begins with the continuous cultivation process, which was upscaled considering three consecutive production volumes: Pre-inoculation (PBR 1), Inoculation (PBR 2), and Production (PBR 3). Following this, the product is filtrated and separated into two phases. Whereas, the biobutanol-rich medium undergoes a pervaporation and distillation step to be recovered. The residual algae biomass is then valorized through hydrothermal liquefaction (HTL), separation, and centrifugation processes.
Microalgae biofuel technology is still at an early stage of development compared to other biofuels [4,23]. According to the literature, a significant pending issue for algal biofuel LCA prevails on the input data and its characteristics, detail, and quality [20]. Consequently, for comprehensive data acquisition from lab and pilot scale installations, a systematic and elaborated questionnaire was developed to collect the required data and information on process design, technology, and equipment to model and calculate the LCA from well-to-tank. Intense and close personal exchange, virtual meetings and email correspondence complemented the discussion on the constructed data framework and proposed assumptions to bridge the data gaps. By this, as much information and original data for the LCA inventory as possible were gathered, analyzed, and adapted. Based on this inventory completed by data from other projects, literature, and the Ecoinvent database 3.2 [24], the environmental impact assessment (EIA) for a pilot-scale system (pre-study) was conducted. The production process was modeled with the open-source software OpenLCA 1.9.
The Ecoinvent database comprises raw material extraction as well as the assembly to pre-products, e.g. borosilicate tubes. However, transport, storage, and end-of-life options are excluded and not considered, as these processes were assumed to play only a minor role within the overall impacts. In addition to that, the delivery of inputs and raw materials to the production site, as well as machine abrasion and disposal, are defined to be outside this system’s boundaries (see Figure 1).
The functional unit is defined as the production of 1 kg of “engine-ready” biobutanol, which can be used directly as fuel for a combustion engine. To evaluate a specific process, it must refer to conventional fossil-based production pathways. Here, we selected the Ecoinvent database process named “Butanol production from hydroformylation of propylene” to compare and interpret the results of the biobutanol produced by the cyanobacteria.
For quantifying the life cycle impacts, the recommendations of the ILCD Handbook 2011 were followed [25]. This impact assessment method was developed and promoted by the European Commission and is already implemented in the OpenLCA software. There are 16 midpoint categories, classified into three types, and ranked according to their reliability. Thereof, we selected six categories, which are considered the most relevant ones for the process analyzed (Table 1).
Table 1 Selected ILCD midpoint impact categories [25]
ILCD Midpoint application
|
Impact category
|
default LCIA method
|
Indicator
|
Classification
|
Climate Change
|
ILCD 2011: Baseline model of 100 years of IPCC
|
Radiative forcing as Global Warming Potential (GWP 100)
|
I
|
Particulate matter/ Respiratory inorganics
|
RiskPoll model (Rabl and Spadaro, 2004) and Greco et al. 2007
|
Intake fraction for fine praticles (kg PM2.5-eq/kg)
|
Resource depletion, mineral, fossil, and renewable (Depletion of renewable resources is included in the analysis, but none of the analyzed methods is mature for the recommendation)
|
CML 2002
(Guinée et al. 2002)
|
Scarcity
|
II
|
Eutrophication, aquatic
|
EUTREND model
(Struijts et al., 2009b) as implemented in ReCiPE
|
Fraction of nutrients reaching freshwater end compartment (P) or marine end compartment (N)
|
Land use
|
Model-based on Soil Organic Matter (SOM)
(Milà i Canals et al., 2007b)
|
Soil Organic Matter
|
III
|
Resource depletion, water
|
Model for water consumption as in Swiss Ecoscarcity
(Frischknecht et al., 2008)
|
Water use related to the local scarcity of water
|
Midpoint categories classified as “I” deliver recommended and satisfactory results. Among these, we selected well-known and commonly used categories like climate change and particulate matter formation. Freshwater eutrophication and mineral resource depletion were chosen from category “II”, because of the expected impact of materials’- and fertilizers’ upstream production inputs used in a large-scale production system. Since this category is recommended but needs to be improved, the reliability of the results is not as high as those of the category “I”. The indicators considered from category “III” are water resource depletion and land use. They were selected based on the relevance of water demand for microalgae cultivation and industrial land occupation in the upscaled system [4]. However, these “III”-ranked categories are less recommended and need to be applied with caution.
1.2 Prospective process upscaling
Following the typical methodology of ex-ante prospective LCA [26], the insights are drawn from the lab and pilot-scale results were used to design and model a large-scale production plant (20 ha). Therefore, the best productivities achieved at lab scale and upscaling the process were used, to analyze whether an optimized, large-scale system has lower environmental impacts per unit (kg biobutanol) produced than the pilot scale. (Table 2).
Table 2 Assumptions for the prospective upscaling of the LCA model
|
Lab
|
Upscale
|
Assumptions
|
Productivity (mg/l/d)
|
50 (weighted average of two strains)a
|
600d
|
Highest productivity achieved under lab conditions
|
Biobutanol concentration in PBR [g/l]
|
1
|
2.65
|
Measured Biobutanol concentration
|
Biomass/ biobutanol ratio
|
1:1 [own assumption]
|
0.35:0.65d
|
|
Energy use [kWh/ kg biobutanol]
- Cultivation
- Separation/ Pervaporation:
- Heating and cooling
- Pumping
Hydrothermal liquefaction [kWh/kg biobutanol]
|
483a
86 [17]
18 [17]
n.a.
|
23
109
23
0.84
|
Upscale by flowb
Upscale by flowc, energy saving appliede
Upscale by flowc, energy saving appliede
Upscale by flowb [27,28]
|
Biobutanol productivities of Synechocystis PCC 6803 were extrapolated from lab-scale experiments by Uppsala University (Sweden) and Imperial College London (UK). Like this, we applied an optimistic year-round production of 360 days with average butanol productivity of 600 mg/l/d (Boatman, T., Zemichael, F., Wang, X., Harun, I., Vachiraroj, N., & Hellgardt, K. WP3 Activities and Outcomes. Unpublished Work. 2018). Since genetically modified cyanobacteria were used in a closed photobioreactor (PBR) system, a continuous production process was assumed to keep the risk of leakage low and prevent exposure and cross-breeding, as shown in the literature [11,12].a Guerra, T. (personal communication, March 28, 2017). b Guerra, T. (personal communication, September 19, 2018). c According to Lauersen and Kruse (2017). d Boatman, T., Zemichael, F., Wang, X., Harun, I., Vachiraroj, N., & Hellgardt, K. WP3 Activities, and Outcomes. Unpublished Work. 2018. e According to Liu et al. (2013)
The system boundaries of the pilot-scale are based on a PBR installation of the existing pilot plant in Lisbon. After analyzing different reactor types, the unilayer horizontal tubular (UHT) PBR was chosen as the best available cultivation technology, since this reactor proved to be more efficient in terms of productivity as well as material and energy inputs than a multilayer horizontal tubular (MHT) PBR (Guerra, T., personal communication, January 23, 2019).
To assess the environmental impacts prospectively and to improve comparability, scenarios with upscaled industrial production systems for algae-based biobutanol production on 20 ha were modeled and calculated.
2.2.1 Cultivation
Commercial large-scale cultivation of microalgae is almost entirely performed using open raceway ponds in the batch mode and to produce other products than biofuels [12]. The theoretical process design for this prospective and upscaled system is based on the concept and design of a cultivation unit at a pilot scale, with a continuous process and daily product (ethanol) harvesting. For such a case, continuous cultivation is considered to be a more efficient path for producing biofuels [29]. Based on this process, confidential information on the upscaling of the UHT-PBR system for ethanol was provided by Guerra, T. (personal communication, July 3, 2018). The land occupied by the reactor was assumed to be classified as an industrial area and used for 20 years to match the system and equipment lifespan. For a final cultivation scale, three consecutive production volumes have to be achieved: Pre-inoculation (PBR 1), Inoculation (PBR 2), and Production (PBR 3). The occupied area as well as the associated volume and number of units are listed in Table S 1 in the supplement.
Production and preparation steps providing less than 1 m3 of culture were neglected and considered as lab work outside the system boundaries. The main materials of the production system were taken into account without assembling, forming, and construction processes. Whenever catalog data were used on electric devices like the blower, 70% of the total mass was assumed to be stainless steel only. The main materials used within the cultivation phase are listed in Table S 2. Pump work and culture bubbling were applied, too. Optimal pumps were selected using the flow rates (30 m3/h, 83 m3/h, 500 m3/h) per reactor size given by Guerra, T. (personal communication, September 19, 2018) to ensure a culture speed of 0.5 m/s. No power for thermoregulation is considered since only spray water cooling is assumed to be used during the summer period (no additional pumping, tap water). The energy consumption for the 360 days of production can be depicted in Table S 3. Sensors and controlling equipment, as well as connecting pipes between the different production steps, were neglected. Values for operational materials like fertilizer, freshwater for cleaning purposes, thermoregulation, or fresh culture supply were calculated based on information provided by Guerra, T. (personal communication, February 4, 2019). Like this, we considered 107 g N/kg DM biomass as NaNO3 and 15 g P/kg DM biomass as P2O5. According to literature, we consider that 62.5% of nitrogen and 90% of phosphate can be recycled within a hydrothermal liquefaction (HTL) process [30] significantly reducing the nutrient demand from primary sources, which are either limited (phosphate rock) or linked to energy-demanding production processes (Haber-Bosch process).
Bioenergy systems are assumed to be carbon-neutral since all carbon stored in the biomass is taken by the plants from the atmosphere. Consequently, the biomass has effectively removed carbon from the atmosphere (short-term perspective). Therefore, any thermochemical or biological conversion of biomass, which releases carbon dioxide into the atmosphere, does not contribute to any net additional greenhouse gases [31]. However, to achieve high productivities, microalgae and cyanobacteria need to be supplied with higher CO2 concentrations than available in the atmosphere. CO2 sources providing such higher concentrations can be supplied for example by coal-fired power or biogas plants. Nevertheless, we did not consider a technical supply of CO2, but only CO2 taken up during photosynthesis and assimilation by the algae. As it will be released while burning the biobutanol in a combustion engine, we did not include the process in the LCA.
As it is assumed that the system is running continuously, only one cleaning per year takes place, flushing the tubes with twice the water volume of the reactors and a solution with chlorine (7.0 kg) and thiosulfate (5.6 kg). To keep the freshwater demand low, recycling of the culture medium is assumed to reduce the water consumption, which totals to 433 m3 including the water demand for the cooling system per year caused by evaporation. As 90% of the culture broth harvested can be recycled and fed back to the PBR system, the total freshwater volume per year can be reduced to 83 m3. The yearly biobutanol production was considered to be about 2,000 t and 1,080 t of biomass as co-product since daily harvesting of around 30% of the culture was assumed. Differences in densities of biomass, biobutanol, and culture medium were not considered for technical configuration and processing.
2.2.2 Separation and harvesting
The recovery of biobutanol from dilute mixtures represents a bio-technical challenge in the photoautotrophic production of excreted biofuels which has not yet been addressed satisfyingly. The most suitable and cost-effective microalgal harvesting method is a constant matter of research [12]. Recently, four of the most promising butanol separation technologies (distillation, pervaporation, gas stripping, and ionic liquid extraction) were assessed concluding that, at present, it is necessary to make a compromise between energy requirement and operating costs [17]. Based on these findings we selected for our prospective upscaled LCA model the technique of pervaporation to separate the biobutanol from the cyanobacteria culture broth.
Following the cultivation process in the UHT-PBR, the product flow was separated using a polypropylene microfilter (Boatman, T., personal communication, January 7, 2019) into two phases: a biobutanol rich medium (ca. 65 Vol.%) and the biomass slurry (ca. 35 Vol. %) (Boatman, T., Zemichael, F., Wang, X., Harun, I., Vachiraroj, N., & Hellgardt, K. WP3 Activities and Outcomes. Unpublished Work. 2018). Consequently, the pervaporation system was implemented with a pervaporation temperature of 60 °C, including energy savings of 42.9 % related to energy integration (compared to a system without energy recovery) [17]. To match with the industrial scale, the data for the equipment for heating and cooling (Table S 4) and their energy demand were upscaled based to match a continuous process design [27]. For this, electricity input savings in large-scale production of 15% were considered [27]. The energy demand for the pervaporation pump was scaled up linearly without any economies of scale, due to a lack of information on flow rates.
Based on the data from the pilot scale, the biobutanol concentration was determined to be 2.65 g/l, therefore the pervaporation process in our model was built accordingly. However, the pervaporation model showed that a minimal biobutanol concentration of 10 g/l has to be achieved to reach a break-even point in terms of energy input and output [17]. Based on the productivity and concentration of biobutanol in the PBRs, a daily partial harvest has been calculated with a fixed volume of 30 % of the total culture, to maintain a stable biological system. The energy requirements for pervaporation were considered in the LCA model assuming a yearly operation of 360 days and a daily full-time 24-hour operation (Table S 5). Simultaneously to the harvesting process, fresh medium including fertilizer was added to achieve constant cultivation conditions.
Following the process simulations by Wagner, Lee-Lane, Monaghan, Sharifzadeh & Hellgardt (2019), in the LCA it is considered that the butanol-rich flow in the system is treated with a two-step pervaporation process and a final distillation column to purify the product. A process scheme can be depicted in Figure 2.
The pervaporation process is used to increase the biobutanol levels above the spontaneous butanol-water phase separation point. After this, it is possible to separate and recycle the aqueous phase back to the separation system and recover the biobutanol from the organic phase through distillation. Each pervaporation unit requires a heater, a condenser, and a vacuum pump to reduce the outlet pressure. The vacuum pumps are defined to be outside the system boundaries of this study, as no data was available. All aqueous streams are being recycled to minimize the amount of required fresh water and to recover the entire residual butanol.
2.3 Scenario development
Three different scenarios for the production of biobutanol were developed and analyzed. The specifications for the three scenarios and their major input parameters are shown in Table 3. The first scenario “Upscale” is regarded as the baseline scenario for a 20 ha system on which the other scenarios were built on. The specifications of the second and third scenarios remain the same while only the energy source was changed.
Table 3 Major input parameters for the scenarios “Upscale” and “Upscale + HTL”
Key parameters
|
Upscale
|
Upscale + HTL
|
Productivity [mg/L/d]
|
600
|
Biomass/biobutanol ratio
|
0.35:0.65
|
Electricity demand [kWh/kg biobutanol]
|
155
|
156
|
CO2
|
Flow is not considered in the LCA due to the biogenic source
|
Nitrogen fertilizer
[kg/kg biobutanol]
|
0.35
|
0.13
(62.5% recycling)
|
Phosphorus fertilizer
[kg/kg biobutanol]
|
0.04
|
0.004
(90% recycling)
|
By-product credit
[kWh/kg biobutanol]
|
n.a.
|
3
|
Electricity mix (2012)
|
European mix (24.2% renewable) [32]
|
HTL: Hydrothermal liquefaction
2.3.1 HTL and nutrient recovery
The scenario “Upscale + HTL” is based on the “Upscale” scenario, but supplemented by an HTL process, a further downstream step to valorize the biomass as well as to recycle nutrients (Figure 3). Pre-studies showed that biomass valorization should be included to improve the overall efficiency of the process, e.g. by recycling nutrients as well as increasing the energy output [33]. For this LCA, the HTL process has been considered as the most suitable technology to convert the residual algae biomass, since HTL, in general, is appropriate for the conversion of wet feedstocks [30]. Besides, HTL delivers a liquid energy carrier, so-called bio-crude oil, which can be upgraded and used as fuel as well.
The majority of HTL research has been performed using small-batch reactors, typically a few hundred milliliters in volume. The present LCA study comprises an upscaled HTL process operating 24 hours a day, with a daily feed of about 15,000 L which was designed according to Jones, Zhu, Anderson, Hallen & Elliot (2014) and Zhang et al. (2017) [34], complemented by experimental data provided by Wagner et al. (2019). As a result, residual algal slurries with 20 % DM content were processed and converted by a high temperature (350°C) and pressure (210 bar) reaction into four streams (Boatman, T., personal communication, January 7, 2019). A generic organic co-solvent (1,1 dimethyl cyclopentane was chosen as a reference) is also included to support the separation of the bio-oil from the other products. The solvent flow rate was set to 10 % of the total flow entering the HTL process (Boatman, T., personal communication, January 7, 2019). The process of solvent recycling itself was not considered to be within the system boundaries of this LCA study. However, a solvent recycling rate of 99,9% is applied as a credit according to Liu et al. (2013).
Figure 3 shows the HTL process as used in our model, the product yields, and the nutrient amount according to data from a case model [30]. This so-called “Aspen Design Case Model” was developed based on experimental results for Nannochloropsis and Chlorella as well from three other algae types from unpublished works that make this model applicable for fresh and saline water algae [30]. Table 4 shows elementary compositions and ash content of Synechocystis PCC6803 and the reference algae used in this study for HTL [35].
Table 4 Elementary composition and ash content of Synechocystis PCC6803
|
Synechocystis PCC6803 [35]
|
Aspen Design Case Model [30]
|
Component
|
(Wt. %)
|
(Wt. %)
|
C
|
49.8
|
52
|
H
|
6.7
|
7.5
|
O
|
26.8
|
22
|
N
|
12.5
|
4.8
|
S
|
0.7
|
0.61
|
P
|
1.5
|
0.6
|
Ash
|
2.7
|
13
|
A list of equipment used in the HTL phase is given in Table S 6. For the full-scale model, all equipment inputs were dimensioned according to the upscaled flows that have to pass the system. After passing the HTL reactor, the solid phase is being removed from the product flow by a ceramic filter as a first step. Subsequently, a 3-phase separator is needed to isolate the other product phases [30].
According to the Aspen design case model [30], the HTL product contains 51 wt.% biocrude (dry), 43 wt.% of the aqueous phase, 4 wt.% of product gas, and 2 wt.% of solids, based on dry algae. Streams refer to the biomass flow as reported by Jones et al. (2014). Figure 3 shows the nitrogen balance in the product streams as estimated from experimental results based on Jones et al. (2014). The nitrogen and phosphorus bound in the solid and aqueous phases are internally recycled and used as a credit for the substitution of inputs [36] in the model, reducing the external nutrient demand during cultivation. The solid fraction from HTL requires a conversion step (such as acid digestion) to make the phosphorous bound bioavailable, before re-using it for algae cultivation [30]. However, any further processing of nutrients to enhance bioavailability as well as the separation of the solvent and bio-crude were considered to be outside the system boundaries. For the HTL process, the energy inputs are contemplated (Table S 7).
With every kg of biobutanol, 0.27 kg of bio-crude oil (HHV 39 MJ/kg) [37] are produced and considered in the LCA model as energy credit along with the biobutanol production. It is assumed that the biocrude oil is used to produce electricity with a conversion efficiency of 40 % [38]. The produced gas is not considered to be recycled or used as its amount is negligible (4 wt. %).
2.3.2 Renewable energy supply
The results of the pre-study LCA on data from small-scale pilot plant operation showed that the electricity demand for algae cultivation and harvest has a major impact on the LCA results because of the environmental impacts of the non-renewable sources-based electricity supply. Therefore, the standard European electricity mix has been changed to a renewable energy mix for the third scenario based on the conditions of the energy supply in Norway, where hydropower is dominating the electricity market with a share of 96.2 % [39]. This electricity mix scenario was chosen as an approximation to the time after the energy transition, whereas there is no real intention to import electricity from Norway. As the biobutanol separation also requires energy in terms of heat, the heat supply was changed to renewable sources (biogas).