An LCA of the Pelamis wave energy converter

Purpose To date, very few studies have attempted to quantify the environmental impacts of a wave energy converter, and almost all of these focus solely on the potential climate change impacts and embodied energy. This paper presents a full life cycle assessment (LCA) of the first-generation Pelamis wave energy converter, aiming to contribute to the body of published studies and examine any potential trade-offs or co-benefits across a broad range of environmental impacts. Methods The process-based attributional LCA was carried out on the full cradle-to-grave life cycle of the Pelamis P1 wave energy converter, including the device, its moorings and sub-sea connecting cable up to the point of connection with the grid. The case study was for a typical wave farm located off the north-west coast of Scotland. Foreground data was mostly sourced from the manufacturer. Background inventory data was mostly sourced from the ecoinvent database (v3.3), and the ReCiPe and CED impact assessment methods were applied. Results and discussion The Pelamis was found to have significantly lower environmental impacts than conventional fossil generation in 6 impact categories, but performed worse than most other types of generation in 8 of the remaining 13 categories studied. The greatest impacts were from steel manufacture and sea vessel operations. The device performs quite well in the two most frequently assessed impacts for renewable energy converters: climate change and cumulative energy demand. The carbon payback period is estimated to be around 24 months (depending on the emissions intensity of the displaced generation mix), and the energy return on investment is 7.5. The contrast between this and the poor performance in other impact categories demonstrates the limitations of focussing only on carbon and energy. Conclusions The Pelamis was found to generally have relatively high environmental impacts across many impact categories when compared to other types of power generation; however, these are mostly attributable to the current reliance on fossil fuels in the global economy and the early development stage of the technology. Opportunities to reduce this also lie in reducing requirements for steel in the device structure, and decreasing the requirements for sea vessel operations during installation, maintenance and decommissioning. Electronic supplementary material The online version of this article (10.1007/s11367-018-1504-2) contains supplementary material, which is available to authorized users.


S1.2 Data from manufacturer and detailed life cycle
The process of calculating the Life Cycle Inventory is described in Figure S1.1. Table S1.1, Table S1.2, Table S1.3 and Table S1.4 summarise the input data derived from information provided by Pelamis Wave Power Ltd (PWP), along with the selected process from ecoinvent v3.3 and the uncertainty indicator scores. The last refer to ratings used to estimate the uncertainty according to the same pedigree matrix used in the ecoinvent database, and described in Section 3.5 of the main report (Weidema, et al., 2013). Figure S1.2 describes the life cycle flows included/excluded from the study.    Table S1.9. Table S1.4 -Input data for assembly and specialist sea vessel processes provided by PWP, with corresponding uncertainty indicator scores.

S1.3 Process approximations
The ecoinvent database does not contain detailed inventory information for some specialist materials, manufacturing processes and sea vessel operations. In order to assess the resource use and pollutant emissions associated with these, data on material quantities and fuel consumption were sourced elsewhere, and new processes were built using inventory data from ecoinvent. The selected materials, quantities and associated uncertainty are given in Table S1.5, Table S1.6, Table S1.7, Table S1.8 and  Table S1.9. Table S1.10 details the waste disposal processes selected from the ecoinvent dataset for each of the principal materials within the analysis.

S1.5 Alternative electricity generation
In order to compare the environmental impacts of the Pelamis with those from other types of power generation, selected average electricity generation data from a number of key energy sources was analysed with the ReCiPe Midpoint Hierarchist and Cumulative Energy Demand impact assessment methods. The processes selected from the ecoinvent database (v3.3) are detailed in Table S1.11.  Table S1.6. Table S1.5 -Details of new processes created from data within ecoinvent for manufacturing processes and glass-flake paint, with corresponding uncertainty indicators 9 From Jiven et al. (2004). 10 Quantity derived from Axxiom (2008). Pressure from Kalpakjian et al. (2008). Compressed air sourced locally, so European average data selected. 11 From data for abrasive blasting of aluminium in Classen et al. (2009). 12 The paint is applied with an airless spray at 250 bar, providing a coverage of 3.9 m 2 /l with a thickness of 200μm (Hempel, 2007). 13 Parker et al. (2007) estimated an overall 1mm paint thickness requiring a base coat of primer, three layers of paint and a topcoat. 14 The paint application process was approximated from manufacturer's data for an airless spray pump (Graco, 2010), powered by 200 m 3 /min of compressed air to provide paint coverage of 12 l/min. 15 From Hempel (2007) 16 Assumed to be the same as the glass flake paint, without the glass flakes.  Table S1.6 -Details of materials within glass-flake paint, selected ecoinvent data and corresponding uncertainty indicators 17 Uncertainty ranges taken from material data sheets 18 (Hempel, 2010b) 19 (Hempel, 2010a) Table S1.10 -Waste processing datasets selected from ecoinvent.

Type of generation Process Name
Coal Electricity, high voltage {GB}| electricity production, hard coal Gas (CCGT) Electricity, high voltage {GB}| electricity production, natural gas, combined cycle power plant Nuclear Electricity, high voltage {GB}| electricity production, nuclear, pressure water reactor 27 Hydro Electricity, high voltage {RoW}| electricity production, hydro, reservoir, non-alpine region Onshore Wind Electricity, high voltage {GB}| electricity production, wind, 1-3MW turbine, onshore Offshore Wind Electricity, high voltage {GB}| electricity production, wind, 1-3MW turbine, offshore Table S1.11 -Source data for comparison with other types of generation 27 Only one nuclear power station in the UK is a pressurised water reactor. The remainder are advanced gas-cooled reactors, but as this is an old technology that is rarely used elsewhere, data for it is not included in ecoinvent v3.3.

S2 Additional Numerical Results
This section contains additional results not presented in the main article.  The results of the uncertainty analysis are shown graphically in the paper, but for completeness, the numerical results are given in Table S2 Table S2.4 -Results of comparative uncertainty analysis of Pelamis with other types of generation. Values between 30 and 70% are highlighted, as these show a significant probability that the impacts of the Pelamis relative to the given type of generation may be reversed.

S3 Locational Adjustment Factors
The normalised impact potentials can be estimated for any given installation location, using the following equation: where: = Embodied impacts per kWh = Distance from Pelamis plant to steel fabrication yard (km) ℎ = Distance from dockyard to installation site (km) = Annual energy output (kWh) , and = Constants for each impact category (given in Table S3.1) Note that this formula is a simplification of the results of this analysis, and cannot be used to determine the effect of a change in other factors. Furthermore, this model has been developed for an installation scenario in the UK, and therefore installation in other countries may not have the same impacts.

S4.2 Approximating the end-of-life recycling method
Section 5.2 of the main article describes how the analysis was re-run using an approximation of the end-of-life method for allocating recycling credit within the foreground data, in order to replicate the method applied by Parker et al. (2007). Although this method is no longer considered appropriate for use in an attributional LCA, it was tested here to explain the discrepancy in results between the two studies.
The end-of-life recycling method (also known as the avoided burdens or closed-loop approximation method) is a method of allocating credit for the avoided production of primary material in the future by producing recyclable material (Schrijvers, et al., 2016). Recycled material consumed in the product life cycle, therefore, does not give an environmental credit so has the same burdens has primary material. The underlying mathematical expression for this method from Schrijvers et al. can be rearranged to form Equation 1, assuming that the impacts of the substituted primary material will be the same as the impacts of the consumed primary material and the quality correction factor is one (as for closed-loop recycling of a material such as steel): where is embodied impacts per unit of material, is embodied impact of primary material, is embodied impact of the recycling process, is embodied impact of recovery and transport of the recyclable material, is embodied impact of waste disposal and is recycling rate at end-of-life. It can be seen that the first term is the embodied impacts of all input material, which is considered to have the impacts of primary material. End-of-life impacts include the credit for recycling, described by ( + − ), which is a function of the difference between embodied impacts of the production of primary and recycled material. Disposal of non-recycled material is represented by (1 − ) .
In order to simulate the method applied by Parker et al., the above method was applied only to the foreground data for steel. All background data was still sourced from ecoinvent v3.3, using the recycled content allocation method, as with the main analysis. Modifications were made as follows: • A new input steel dataset was created by copying the ecoinvent v3.3 data for the global steel market, but replacing all flows of recycled steel with data for primary steel for the same region.
• Recycling credit was estimating by creating a waste flow with a global recycled steel market as input (as above, but with all primary steel replaced with recycled steel), and a global virgin steel market as avoided product.  The result of running the analysis with this modification is a reduction in all impacts. Of the factors relevant for comparison with Parker et al.: climate change was found to be 28 g CO2eq/kWh, cumulative energy demand 421 kJ eq/kWh and CO2 emissions 26 g/kWh. This reduction is likely due to the recycling rate of 90% being much higher than the average recycled content of the global steel mix in the ecoinvent data (43%) (ecoinvent, 2016).

Impact Category
Errors may have been introduced to this analysis by using a mixture of allocation methods, so use of the method described here is not recommended.