Safety, health, and environmental assessment of bioethanol production from sugarcane, corn, and corn stover

2.1 Alternative processes for producing bioethanol

The simplified flowsheets of bioethanol production processes from three types of feedstock are shown in Figure 1A–C (see also section 1, Table S1 in the Supporting Information for feedstock composition). Each process is divided into three main stages: pretreatment of input feedstock, fermentation, and purification of ethanol. All considered processes are described briefly below, and the operating conditions of the main process units are summarized in Table 1 (see also section 2, Tables S2–S10 in the Supporting Information for the considered reactions and the derived mass balances). Different methods and operating conditions have been recommended for the optimal performance of the considered process stages; however, only the widely studied ones are applied and evaluated in the present study. It should be noted that the agricultural steps for the production of the biomass feedstock as well as the waste treatment and byproduct recovery stages are not considered in this study for the hazard assessment of the bioethanol processes, and for this reason, the respective flowsheets are not presented in Figure 1A–C. This selection for the system boundaries corresponds to the common practice of analyzing the hazards “locally”, that is, for a certain plant facility, which, in this case, is defined as the bioethanol production section. In this context, the waste management of the process effluents is assumed to be done centrally in a dedicated plant facility for waste treatment together with effluents from other processes. Moreover, the bioethanol process does not implicate any special type of waste treatment. The byproduct recovery stages as well as the other mass and heat integration potential of the bioethanol processes are not included in the respective flowsheets due to lack of explicit information (i.e., process conditions and efficiencies) required for the mass balances and the hazard assessment. However, it should be noted that, for the LCIA included in the multicriteria process assessment of this study, aggregated information is available, including all relevant cradle-to-gate stages.

Figure 1

Simplified flowsheets of bioethanol production processes considering the main unit operations and process streams. (A) Sugarcane-derived process [10, 14]. (B) Corn-derived process [1, 14]. (C) Corn stover-derived process [1, 2].

2.2 Assessment methods

Various hazard assessment methods have been proposed and compared for process design [35–37]. In the present study, the ISI and the EHS methods were chosen to evaluate hazards for bioethanol production. These two methods have been widely applied to typical chemical production case studies, have different process data requirements, and highlight different aspects of hazard assessment. The basic concepts and settings of the EHS and the ISI methods are briefly summarized below (see also the study of Adu et al. [36] and Table A1 in the Appendix for more information about the similarities and differences between the ISI and the EHS methods).

3.1 ISI method

The evaluation of the bioethanol production processes using the ISI method is presented in Figure 2. According to this analysis, the same pattern appears for all processes as far as the ranking of process sections is concerned, namely, the fermentation stage is the most hazardous followed by the purification and the pretreatment stages. This is mainly due to the heat release from the exothermic reactions during the fermentation (e.g., the side reaction forming acetic acid from glucose) and the reactivity of the byproducts (e.g., acetic acid, lactic acid, and succinic acid). The latter also applies for the purification section, where ethanol is purified from these byproducts.

Figure 2

ISI of alternative bioethanol production processes.

The feedstock pretreatment of the sugarcane-derived process obtains the lowest ISI score mainly because it is performed at lower temperature compared with the corn-derived and the corn stover-derived processes. Essentially, the difference of operating conditions applied in the pretreatment step is the main factor that differentiates the investigated bioethanol production processes according to the ISI method. According to this ranking, the sugarcane-derived process is the safest one followed closely by the corn-derived and the corn stover-derived processes.

Finally, it is tested whether the ranking results are sensitive to boundary conditions of the analysis, as expressed by the ranges for biomass composition, and process operating conditions reported in Table 1. It is shown that neither the variations in process operating conditions nor the feedstock composition from different parts of the world distort the ranking results of the ISI method (see also section 3, Table S11 in the Supporting Information).

However, the marginal superiority of the sugarcane-derived process regarding safety aspects disappears when the whole process is taken as one section [i.e., comparing total (decomposed) and total (nondecomposed), respectively, in Figure 2]. The reason is that process decomposition highlights section-specific hazards, which are not revealed if no decomposition is performed. This indicates that the ISI method settings defined by the decision-maker can sometimes be more important than process operating parameters.

3.2 EHS method

The evaluation of the bioethanol production processes using the EHS method is presented in Figure 3. Again, the same patterns appear in all three processes regarding the ranking of the process sections, that is, fermentation is ranked as the least hazardous section followed by the pretreatment and the purification sections. Interestingly, this pattern is different from the one proposed by the ISI method (Figure 2). The EHS method penalizes the purification section mainly due to the higher values of persistency, arising from the existence of CO₂ in the purge of the absorber. The persistency category refers to the environmental effects not captured by the ISI method; therefore, the differentiation of the EHS method results is justified from this point of view. It should also be noted that some of the considered categories in the EHS method (e.g., water- and air-mediated effects), which are not included in the safety-oriented ISI method (see also Table A1 in the Appendix), do not play a major role in the evaluation. Others, such as the mobility category referring to substance vapor pressures, are considered in the holistic categories of process temperature and pressure in the ISI method, therefore not being substance specific. The mobility category of the EHS method is important for the score of both the fermentation and the purification sections compared with the pretreatment section mainly due to CO₂ produced during the fermentation.

Figure 3

EHS hazard metric of alternative bioethanol production processes.

Some other aspects of the EHS method, also responsible for the process section rankings, are the acute toxicity and fire/explosion categories, which are also considered by the ISI method in the category hazardous substance. These aspects severely penalize the pretreatment section of all production processes for the EHS method, whereas the differentiation effect for the ISI method is minimal. One reason for this lies in the fact that the EHS method is strongly influenced by the mass inventories of the considered substances, in this case of feedstock composition, which are multiplied with the substance intrinsic dangerous properties to quantify the respective hazards. On the other hand, the inventory cate­gory of the ISI method, which has some analogy to the mass of substances in the EHS method, is considered as a separate category without any multiplication effect. Moreover, as it is clear from Figure 2, the inventory cate­gory of the ISI method has a minor effect on the inferred rankings of the process sections for all feedstocks of the present study.

The potential for an agreement between the results of the ISI and the EHS methods is further investigated by assigning weights in the categories of the EHS method. The weights are let free to be optimized in various subranges between 0 and 1, the only other constraint being that they have to sum up to 1. The optimization goal is to maximize the Pearson’s correlation coefficient (R-Pearson) between the two methods for the assessment of the decomposed bioethanol production processes based on the three different feedstocks. To this end, two different modes of the EHS method are used. In the first mode, the categories persistency, water and air hazards, irritation, and chronic toxicity are discarded (i.e., their weight was set to zero), because, according to Table A1 in the Appendix, only the categories acute toxicity, mobility, fire/explosion, and reaction/decomposition are commonly shared between the ISI and the EHS methods. In the second mode, all the categories of the EHS method are included. Figure 4A and B depicts the respective results of these two modes, that is, the weights resulting in the maximum correlations and the respective correlation coefficient values. In the first mode (Figure 4A), it is clear that a significant agreement between the ISI and the EHS methods is achieved (i.e., with R-Pearson>0.9), when the categories of reaction/decomposition and mobility obtain higher weights. This is also true in the second mode (Figure 4B), adding the categories of acute toxicity and irritation. The importance of chemical reactivity has been already mentioned during the analysis of the results of the ISI method (Figure 2) and the impact of mobility and acute toxicity has been already identified in the analysis of the EHS results (Figure 3). However, the effect of the irritation category was difficult to foresee based on the results of Figure 3. The main effect seems to be that, by assigning a higher weight to this category, which plays a minor role in the pretreatment section, the hazard value of this section decreases, resembling more the ISI assessment (Figure 2). Of course, it should be noted that these results are case specific and more elaborate and diverse case studies have to be conducted for inferring an optimal weighting for maximal agreement between the ISI and the EHS methods.

Figure 4

Agreement between the ISI and the EHS methods according to Pearson’s correlation coefficient (R-Pearson) for the assessment results of the decomposed bioethanol production processes based on the three different feedstocks.

Despite the differences of the EHS and the ISI methods for the process section evaluation, the overall process ranking is the same, that is, the sugarcane-derived process is the least hazardous and the corn stover-derived process is the most hazardous. However, the EHS method appears to be more sensitive to the various boundary conditions of the analysis. In particular, when the whole process is considered as one section, the ranking is distorted in a greater effect compared with the ISI method, that is, the ranking of the sugarcane-derived process is completely changed from least to most hazardous. Here, it should be noted that the resolution of the endpoint indices for both the ISI and the EHS methods is not straightforward to infer, this aspect being currently a point of discussion for all hazard identification index-based methods [38, 39].

Finally, like for the ISI method, a sensitivity analysis with respect to process conditions and feedstock biomass has also been performed for the EHS method. Although the impact on the EHS scores was more evident, it was still not significant enough to cause overall process ranking changes (see also section 3, Table S12 in the Supporting Information).

3.3 Hazard vs. environmental impact

For a multicriteria assessment of the bioethanol production processes, more than one aspect should be typically considered. In this study, a bicriteria profile with respect to hazard identification, expressed by the EHS and the ISI methods, and CED is presented in Figure 5. CED has been shown to have strong correlation with other LCIA metrics estimating the environmental burden [26–28]. In this study, cradle-to-gate CED values were estimated using the Ecoinvent database 2010 [40] and literature sources (see also section 4, Table S13 in the Supporting Information). It should be noted that several articles with a lifecycle orientation have already been published regarding the environmental performance of bio-based products. Especially in the case of corn, wide variations can be observed in the net energy value due to different agricultural production data, yields, ethanol conversion technologies, fertilizer manufacturing efficiency, fertilizer application rates, byproduct evaluation, and the number of energy inputs. Besides data sets, methodological issues, including choices of the system boundaries and allocation procedures, also can play a role for these variations [41]. Clearly, the sugarcane-derived process is significantly superior to the rest of the processes regarding both metrics followed by the corn-derived and the corn stover-derived processes. However, a tradeoff may also exist between the corn-derived and the corn stover-derived processes due to the overlapping regarding the ranges of the CED values. Whereas only material and energy input/output per functional unit is considered by LCIA (e.g., CED), hazard identification methods can express both extensive and intensive process parameters, such as process temperatures and pressures. However, it should be noted that the LCIA results involve a cradle-to-gate analysis extending the system boundaries to include biomass production, byproduct recovery, and waste treatment of process effluents, whereas the hazard assessment refers to the local system of bioethanol production, whose boundaries are defined by the flowsheets in Figure 1A–C.

Figure 5

Hazard (EHS and ISI) vs. LCIA depicted by CED [LCIA (CED)].

Generally, the same type of multicriteria analysis presented here based on the simplified flowsheets can be repeated with more detailed process information. In such a case, it is expected that the ISI method will be mainly affected by additional substances in the form of chemical auxiliaries that may have been neglected here, whereas the EHS method will be additionally more sensitive to updated mass flow values. As far as LCIA is concerned, more detailed flowsheets could provide incentives for energy and mass integration approaches, therefore updating the calculation of gate-to-gate emissions, water, and energy consumption [42, 43].

3.4 Green bioethanol process design

Green process design for bioethanol production should consider safety, health, and environmental hazards as well as lifecycle impacts [17]. The three types of bioetha­nol production in this study, that is, ethanol derived from sugarcane, corn, and corn stover, refer to the most representative biomass resources for sucrose-containing materials, starch materials, and lignocellulosic biomass. Therefore, based on the results of this study, a new viewpoint, that is, process hazard assessment, can be implemented into bioethanol production in addition to economic aspects, LCIA, resource availability, and use of land and water discussed by other researchers [14–16]. In this context, the specific category of environmental impacts related with nutrients used in biomass cultivation, such as nitrogen, phosphorus, and potassium, is also important and their input/output balances and cycles among involved carriers such as animal and plant should be carefully considered [44]. However, this detailed analysis lies outside the scope of the present study, which mainly focuses on the hazard assessment of biomass technologies as one of the additional elements for sustainable bioethanol production.

The second-generation bioethanol technologies based on lignocellulosic biomass have been developed due to non-competition with food supplies. This study considered only corn stover from this category and demonstrated that the hazards of the respective process are higher, whereas its CED value might be lower than that of the corn-based process. This higher hazard is due to the pretreatment section as indicated by both the ISI and the EHS methods, pointing out to the necessity for further development of technologies for the pretreatment of corn stover considering safety, health, and environmental perspectives. A review of different pretreatment options can be found in scientific literature [1, 45]. From the ISI point of view, it would be beneficial to target at milder process conditions (i.e., lower temperatures and pressures), because this was the main reason for penalizing the corn stover pretreatment section. To this end, the technologies of ammonia fiber explosion (AFEX) and liquid hot water (LHW) could be considered. On the one hand, similar process pressure and milder process temperature are reported for AFEX (~90°C), but a system for ammonia recovery is required that complicates the process and may further penalize it from the EHS point of view. On the other hand, the LHW process is simpler, which becomes more relevant for the ISI method, if the subindices for equipment safety and process structure are considered in the calculation. However, the LHW process does not involve lower temperatures (170–230°C) or pressures (>5 MPa). The impact of these other pretreatment options according to the EHS method is more difficult to foresee because a mass balance is required. However, these methods have not been yet implemented in large scales, and the respective process data are not of the same accuracy. This is also true for other reported pretreatment options (e.g., involving supercritical fluids or irradiation). For this reason, these pretreatment options are not included in the current study but certainly constitute material for future research. Moreover, similar studies for other lignocellulosic biomass resources (e.g., switch grass) are necessary to provide a more complete evaluation profile of the second-generation bioethanol technologies, which are constantly gaining interest. It would also be interesting to compare the bioethanol production routes described here with conventional ethanol production methods to highlight the advantages as well as the challenges.

Finally, as mentioned above, the ISI and the EHS methods do not share a common analysis scope and can reveal different aspects of process hazards. From the process safety viewpoint of the ISI method, the highest hazard is allocated in the fermentation process of all biomass technologies, whereas the EHS method highlights the purification section as the most hazardous. Therefore, this study shows that it is important not only to consider hazard assessment methods for a multicriteria process design but also to comprehend the different basis and scope of such methods.