Most of the research on algae-to-energy systems carried out to date has been at the bench or demonstration scale [18]. This makes it difficult to say with much certainty what a full-scale algae-to-energy industrial facility would look like and herein lies one of the fundamental challenges of developing reliable LC estimates for algae production. Using best engineering judgment, it is possible to design hypothetical algae-to-energy facilities, but naturally, there is variability among these designs (Fig. 1). For example, one modeler might assume that algae should be cultivated in ponds, while another could assume photobioreactors [6]. Similarly, a belt filter press could be modeled as means to separate algae from the growth medium, whereas self-cleaning bowl centrifuges might be a viable alternative [24]. Both unit opera­tions carry out the same dewatering function but with different requirements in terms of inlet and outlet concentration, demand for chemical flocculants used to accelerate the settling of the algae out of solution, and energy use profiles. Similarly, there are several technically viable options for extraction of oil from algae biomass, namely: sonication [28], bead mills [7], and enzymatic processes [11]. For conver­sion of algae biomass into biodiesel, one might choose decarboxylation of fatty acids [29] and digestion of non-fatty acid fraction [27] or the conventional transesterification route. Finally, the end-product of the algae-to-energy facility can

ICIarcns 2010|

— solar drying

I Sander20I0|

— autoflocculation (Lardon 2009]

Fig. 1 In selected system boundaries for an algae LCA study, one must typically select from (a) or capture all of (b) a large number of possible unit operations also vary, because biodiesel is not the only energy carrier that can be produced from alga biomass [3]. It can be dried and combusted directly to generate electricity or it can be separated such that the carbohydrate fraction may be fermented to produce ethanol [24]. Naturally these two systems would have very different impacts.

As an example of the way in which systems boundaries selection can impact LCA results and conclusions, it’s informative to consider two of the more thoroughly documented algae LCA studies that have been published to date: Clarens et al. [8] and Stephenson et al. [31]. Clarens et al. [8] used an energy-basis functional unit and only modeled cultivation-phase burdens for open pond systems. They did not account for the possibility that energy production from algae might also create valuable coproducts since they argue that it is still unclear whether there will be tenable mar­kets for these coproducts. In contrast, Stephenson et al. [31] utilized a functional unit of 1 ton algae biodiesel to compare between open pond cultivation systems and pho­tobioreactor cultivation systems. These authors included two types of valuable coproducts: electricity, as produced via combustion of natural gas generated during anaerobic digestion of residual (non-lipid) algae biomass, and glycerin. In light of these dramatically different sets of systems inputs, it’s not surprising that each study reached different types of conclusions. Clarens et al. found algae-derived biomass energy to be generally more environmentally burdensome than corn, canola, or switchgrass alternatives. In contrast, Stephenson et al. found algae-derived biodiesel to be more environmentally beneficial than fossil-derived diesel.

Once an algae-to-energy process has been specified there is the additional uncer­tainty associated with setting system boundaries. LCA is typically intended to cap­ture all of the environmental impacts of an engineered system. Naturally, in a highly interconnected technical world, system expansion results in models that become impossibly large and complex. For example, to produce carbon dioxide for use in industrial processes, it is necessary to model ammonia production since most of the carbon dioxide in this country comes from the steam reforming of hydrocarbons to produce hydrogen, most of which is used to produce ammonia via the Haber-Bosch process [21]. This in turn requires that we understand something about the way
natural gas is produced and transported in this country and the countless unit operations that allow us to purchase a canister of relatively pure carbon dioxide for the factory. To cope with this complexity, many LCA practitioners have set arbitrary boundaries around their processes of interest. For example, one study might state that any process contributing less that 5% of the total mass or energy or other impact to the final total is neglected. In this way the problem can be distilled down to some­thing that is not computationally expensive and still yields good approximations of a process’ impact.

Beyond system design and boundary setting, LCA analysts may chose to focus on specific pieces of a larger system to provide a desired level of resolution. For example, in their work, Clarens et al. considered only the cultivation of algae argu­ing that the uncertainties with that first step in the algae-to-energy life cycle should be addressed [8]. By focusing only on cultivation, the authors were able to explore the full implications of that important LC stage including crucial upstream impacts such as fertilizer production and carbon dioxide generation and delivery. In fact, a sensitivity analysis included in this chapter suggests that these two impacts are among the most important factors driving the overall life cycle burdens of algae production. Many of the other studies assume that the upstream impacts of deliver­ing fertilizers and carbon dioxide should not be included. In Sander and Murthy, a cut off of 5% was assigned to LC contributions that would be neglected in the analy­sis [24] (Fig. 2). This represented the most rigorous treatment of boundaries from any of the studies published to date. However, this study also made certain assump­tions, notably, that the effluent from a secondary wastewater treatment plant would contain enough nutrients to sustain a community of algae [4] . This assumption is not supported by stoichiometry or by the bench-scale research and as a result their estimates for algae life cycle impacts are most likely low.