Final Report for Defra Project FO0103, Comparative Life Cycle Assessment of Food Commodities Procured for UK Consumption through a Diversity of Supply Chains

Authors

A.G. Williams, Cranfield University.

E. Pell, J Webb, E Tribe, AEA.

D Evans, Beef Production Systems Ltd.

E. Moorhouse, Moorhouse Consulting.

P. Watkiss, Paul Watkiss Associates Ltd.

Provision of information on production techniques, processing and transport

M. Palmer, Agriculture and Horticulture Development Board, Meat Services (AHDBMS); P Cook, rlconsulting, poultry production in Brazil; Shipping emissions: Ø. Buhaug, Norwegian Marine Technology Research Institute (Marintek).

Scrutiny of procedures

J Bates, AEA

SCIENTIFIC OBJECTIVES

The overall objective of this project was to produce a comparative life-cycle inventory (LCI) of the environmental burdens and resource use arising from the production of seven key food commodities for which the UK-based and imported production and supply chains are significantly different. The specific objectives were to:

  1. agree the scope of study with Defra;
  2. produce comparative inventories for commodity production to the farm gate where they do not already exist in the Cranfield life cycle assessment (LCA) analyses;
  3. produce comparative inventories for post farm gate commodity storage, processing and transport;
  4. estimate comparative effects on wider eco-services;
  5. compare whole chains for commodities;
  6. draw interim conclusions and report these to Defra;
  7. publicise the work;
  8. draw final conclusions and report these and the whole project to Defra.

Extent to which the objectives have been met

The objectives have been met as follows.

  1. The scope of the study was agreed with Defra at the meeting held on 2 May 2007. A report to describe the Goal and Scope Definition and Production Flow Charts was submitted as Appendix 1 of the annual report submitted on January 2008.
  1. Comparative inventories have been produced for production of all the commodities to the farm gate and are presented in this report.
  1. Comparative inventories have been produced for production of all the commodities post farm gate and are presented in this report.
  1. Estimates of the comparative effects on wider eco-services have been developed and are included in Appendix 4.
  1. Comparisons of whole chains have been made and are presented in this report.
  1. Interim conclusions have been drawn and were reported in a draft final report to allow Defra to comment.
  1. Some publicity has been given to the work.
  1. Final conclusions are presented in this report.

METHODS

Appendix 2, 'Goal and Scope Definition and Production Flow Charts' outlines the methodology used in this project, consistent with LCA goal and scope definition. This task is consistent with the scope of the project, Objective O1. Appendix 2 also presents production flow charts for all commodities for all seasons of demand and for all the sources of production (Objective O2.1). The goal and scope phase is important as it determines why an LCA is being conducted, including the intended use of the results. While the project team has outlined the potential reasons and use, discussion and confirmation of these aspects with Defra was essential. The study was planned to undertake an LCI, and then progress to a LCA, i.e. associating inventory data with specific environmental impacts and understanding those.

A summary of the key approaches to meet the goal are given below. More detail is provided in Appendix 2.

Intended application.

The intended application here is a LCA of the environmental burdens and resource use arising from the production of seven key food commodities for which the UK-based and imported production and supply chains are significantly different. A major element of this was to make a comparison of the production of out-of-season goods in the UK against imports, as well complementary seasons (i.e. towards year-round supply).

Reasons for the study.

Food production and consumption has been identified as a high impact category of consumption, when all life cycle stages are taken into account. The increased global trade in food is leading to a greater diversity of food chains supplying the UK consumer. The move towards year-round supply is also leading to greater production in the UK out-of-season. This raises questions regarding the comparative life-cycle burdens of different food supply chains across the year and the extent to which some types of chains might be exporting our environmental burdens to countries outside the UK (through imports) or leading to additional UK burdens (through domestic out-of-season production). The study follows from a number of previous studies which have investigated this, but which have highlighted data gaps.

Intended audience.

There are a number of potential targets for the study. The first is Defra, with the aim of providing policy relevant information (to fill evidence gaps). Second, it is expected that the results would be useful for the companies across the food chain, in identifying hot spots in relation to environmental burdens, which could be addressed to reduce life cycle burdens. Finally, it is possible that consumers could be informed about the impact of their choices of out-of-season produce.

The key proposals for scope were that:

  • The study focused on a reference year of 2005, looking at current practice.
  • It focused on a representative picture of the national average. However, it differentiated for seasonality (i.e. production in season vs. out-of-season for UK production).
  • It undertook a sensitivity analysis to focus on a limited number of potentially important issues, for example the comparison between glasshouses heated using gas, combined heat and power (CHP) generation or industrial waste heat.
  • The system boundary was from the first point of pre-production through to the Regional Distribution Centre (RDC), rather than the consumer, as all steps post RDC will be common to UK and non-UK food. It needs to be noted that this does not provide information on the overall life cycle burdens of the final product (i.e. defined as requiring no additional transformation prior to use) as food will go through additional steps post RDC through to cooking and consumption.
  • The functional units are weight based (e.g. per kg or tonne at the RDC). Note that for meat products, to allow a comparable LCA, identical products were considered and used as the functional unit, i.e. frozen chicken breast, and frozen hind beef cut.
  • All direct steps were investigated. For indirect steps (e.g. capital burdens associated with inputs and outputs) a materiality rule of 10% was used (e.g. in the case of transport, the inclusion of capital burdens was included if these comprise 10% or more of the total transport burdens).
  • The study produced a life cycle inventory for each commodity. Each included established LCA criteria including:
  1. primary energy (including the proportion as fossil energy if possible)
  2. acidification
  3. eutrophication
  4. abiotic resource use
  5. pesticide use
  6. land requirement
  7. ozone depletion
  • In addition, the following were assessed, but only semi-quantitatively:
  1. loss of habitat and biodiversity
  2. loss of soil C
  3. sustainability of water supplies
  4. pollution of water-courses or reserves.

The study then progressed to a life cycle impact assessment, i.e. associating inventory data with specific environmental impacts, for example looking at greenhouse gas (GHG) emissions, and using the global warming potential (GWP) of each gas to estimate the total global warming burden for each system. It did not propose to proceed beyond this stage.

Models used

All the pre-farm gate LCA calculations used Excel, with some code in VBA. Some underlying values were derived with the SUNDIAL and RothC SUNDIAL simulation models. Most LCI data of inputs were developed during project IS0205 from various sources. A small amount of data also came from the Ecoinvent database. For the post farm gate analysis calculation spreadsheets were created specifically for the project. Emission factors were derived from a number of sources including IPCC (2006) and Simapro (CML, 1999).

GHG emissions from soils

Calculations of GHG emissions from soils were largely based on the slightly modified IPCC (2006) Tier 2 inventory reporting guidelines, as interpreted by Williams et al. (2006). The main items included nitrous oxide (N2O) from N supplied as fertiliser, managed manure, grazing animal N excretion and crop residue returns (NB N fixed by legumes was thus excluded, in contrast to the 1996 and 2001 guidelines) together with secondary emissions from leached nitrate and volatilised ammonia (NH3). The same emission factors were used for all countries. Emissions from polytunnels were assumed to behave in the same way as in open ground. Nitrous oxide emissions from recirculating nutrient solutions were calculated as by Williams et al. (2006), which was derived from data in the scientific literature.

Emissions of C (as CO2) from soil (or indeed sequestration in soil organic carbon – SOC) were generally not included on the basis that most of the systems being studied are likely to be in or close to steady state. In two cases, these are treated separately, i.e. potatoes in Israel and beef in Brazil.

Uncertainty

Before discussing any apparent differences, the topic of uncertainty must be addressed. All scientific measurements and models contain some uncertainty (or error). LCA is no exception and some areas are subject to greater uncertainty than others. Greenhouse gases from all agricultural systems are particularly awkward with very large uncertainties associated with N2O emissions (applying to virtually all soil N turnover) and large ones for enteric methane (CH4). These are central to any comparison. The way in which these are aggregated in LCA (or indeed carbon footprinting alone or many other environmental assessments) tends to reduce the errors, but the way in which these should be treated in a comparison is still the subject of debate. This is not simply an experimental comparison in which two normal distributions may be compared using a conventional statistical test. Cranfield University is currently addressing this matter for Defra in an assessment of PAS 2050:2008 (Publicly Available Specification PAS 2050:2008, Specification for the assessment of the life cycle greenhouse gas emissions of goods and services). Although PAS 2050 applies only to GHGs, the sources and types of errors and the ways in which these should be handled will be analogous. We thus feel it premature to offer a conclusive statement about statistical significance between the results reported here. As a guide, in other similar system models the uncertainty (quantified as the coefficient of variance – standard deviation/mean) may lie in the region of 25% to 35% for factors like NH3 emissions, but possibly over 50% for N2O. Given this and some understanding of the other factors, it seems likely that some criteria are significantly different up to the farm gate. These, with caution, are thought to include: energy, acidification, land occupation, pesticide use, abiotic resource use and volatile organic carbon. All the text that follows should, however, be read with qualification that calculated statistical significance is not implied in any statement.

Relationship between full LCA approach and carbon footprinting (PAS2050)

The work conducted here was not initiated with a view to being compliant with PAS2050

( PAS2050 was being developed during the course of this project and was only finalised after this project was completed.

The differences that occur include:

  1. We dealt with more environmental impacts than PAS2050.
  2. We include capital overheads, whereas PAS2050 does not. We considered this particularly important in this study for post farm gate anlaysis, which relies on an existing infrastructure (shipping etc.,) and also includes some crops (e.g. tomatoes) which are grown under protection and/or require storage (e.g. potatoes).
  3. PAS2050 has explicit limits on the amount of primary data needed to be compliant. This would apply for some commodities in some locations. Without checking every detail, it is reasonable to assume that the Israeli potatoes and Spanish tomatoes would be compliant. The main approach taken here has, however been to apply system modelling to generalise the specific, which is almost the opposite philosophical approach to that in PAS2050.
  4. The main effects of GHG estimates will be that our values are somewhat larger that those from the application of PAS2050, because of capital inclusion. We undertook a study of a legume for a commercial organisation and found that the GHG estimates using the PAS2020 approach were underestimated by 11% while the energy used was underestimated by 20%. That study is, we believe, the first to apply PAS2050 and LCA using the same set of data, but applying the different system boundaries and rules.

RESULTS

In this Appendix we report the results of the quantified LCA. Results of the estimates of impacts on wider ecosystem services are given in Appendix 4. For each product the activity data on which we have based our calculations are provided in Appendix 3. We also intend to post the main results on the Cranfield web site. These may be supplemented with updated results if work in other projects, especially the Defra-funded IS0222, indicates that changes are appropriate in the light of new data or understanding. This will mainly apply to pre-farm gate activities. If resources permit, more pre-farm gate background information on methods, data and calculations will be posted. In the absence of sufficient detail, please contact the authors.

Comparisons are based on the average (or at least representative) position for the relevant industry based on available data. The criticism may therefore be made that these comparisons have been made with historic rather than current information. The most up-to-date systems will typically have significantly greater productivity that will more than off-set any additional inputs. For example, young high density orchards in the UK produce significantly greater yields which will reduce their GHG impact. As indicated above, LCA needs constant updating and we have given indications of where current trends in production appear likely to lead.

Shipping

The energy requirement needed for the long-distance transport of produce, and the consequent GHG emissions, is a term which appears to have received relatively little attention. This may be because long-distance transport by ship is very energy efficient, with estimates of between 10 and 70 g CO2/t/km, compared with estimates of 20-120 and 80-250 g CO2/t/km for rail and road respectively (Marintek, 2008). Nevertheless, the range quoted above is relatively large and over the distances from the southern hemisphere to the UK of up to 20,000 km, usage of inappropriate emission factors (EFs) could lead to significant errors in estimates of total PEU and GWP. In addition, the produce under consideration is transported under refrigeration, whether cooled or chilled, and this adds an additional burden to the energy requirement which is not covered by the standard energy use and emission data. Estimates of this additional burden range from 10 to 50% of basic energy use. We therefore included the Norwegian Marine Technology Research Institute (Marintek) in our consortium in order to provide updated estimates of energy requirements for refrigerated transport by sea.

All of the produce studied in this project are transported to the UK in 'reefer' ships (i.e. refrigerated, with the produce kept in containers rather than in a bulk hold). The basic energy requirement of such ships was estimated to be 0.24 MJ/t/km. This requirement increases to 0.27 MJ/t/km when transporting frozen produce and 0.29 MJ/t/km for the transport of chilled. Based on these data a default EF for CO2 emissions of 0.0178 kg/t/km was used in the estimates of GHG emissions from shipping.

Tomatoes

Introduction

The main difference between UK and Spanish tomato production is that in the UK there is a much greater requirement for heating which consumes much more gas and electricity. All fully-commercial UK production is under glass, only hobby and semi- commercial systems will use plastic housing. The situation is reversed in Spain with only a few sites with glass. The bulk of production is in polyethylene-covered houses of varying sophistication together with some outdoor summer field production (although this crop will rarely be imported to the UK). A range of greenhouse designs are used in both the UK and Spain and the more modern and sophisticated systems tend to be used for the more demanding high-value speciality crops (e.g. some of the sensitive on-vine products) in both countries. Nearly all conventional UK production is based on hydroponic systems using either substrates (of which there are a range) or nutrient film systems, whereas Spanish systems tend to be more varied and soil is still widely used as the growth medium for conventional commercial crops. Nutrient solutions are recirculated in many UK systems and similar practices are used in some Spanish operations, however run-to-waste systems are more common in Spain.

Spain has traditionally dominated off-season supply to the UK, while Northern European production (primarily UK and Holland and more recently Poland) has traditionally dominated the April-October period. However, recent investment in modern structures with sophisticated heating and lighting systems means that growers in the UK and Holland can now supply tomatoes year-round, although the analysis presented relates to the principal April-October production system. Winter production in Northern Europe is still limited and the bulk of UK supply at this time of year comes from Spain, the Canaries, Morocco, Italy and Israel. The seasonality of different tomato products in the UK varies with year-round supply of some products such as loose cherry from Spain and the other Mediterranean countries and more defined winter seasons for lines such as classic round (Spanish/Canary supply from October through to March with joint supply at each end of this period). The fruit from Spain and other countries will clearly involve longer transport and the associated transport energy intensity for these tomatoes entering UK food system will be greater than for comparable UK product.