THESIS FOR DEGREE OF DOCTOR OF PHILOSOPHY

Environmental Improvements of the Post-Farm Dairy Chain: Production Management by Systems Analysis Methods JOHANNA BERLIN

Department of Energy and Environment CHALMERS UNIVERSITY OF TECHNOLOGY Göteborg, Sweden, 2005

Environmental Improvements of the Post-Farm Dairy Chain: Production Management by Systems Analysis Methods JOHANNA BERLIN © JOHANNA BERLIN, 2005 ISBN 91-7291-655-9 Doktorsavhandlingar vid Chalmers tekniska högskola Ny serie nr 2337 ISSN 0346-718X ESA Report 2005:6 ISSN 1404-8167 Department of Energy and Environment Division of Environmental Systems Analysis Chalmers University of Technology SE-412 96 Göteborg Sweden Tel: +46 31 772 10 00 http://www.esa.chalmers.se Telephone to author: +46 31 335 56 00 E-mail to author: [email protected]

Chalmers Reproservice Göteborg, 2005

Environmental Improvements of the Post-Farm Dairy Chain: Production Management by Systems Analysis Methods JOHANNA BERLIN Department of Energy and Environment Division of Environmental Systems Analysis Chalmers University of Technology

Abstract The production of dairy products is becoming more centralised at the same time as the number of different products is steadily increasing. In this thesis, the environmental impact of such ongoing development trends in the post-farm dairy chain was evaluated and improvements were suggested. Methods for production management and environmental systems analysis (life cycle assessment, material flow analysis and substance flow analysis) were combined and used in the evaluations. A first assessment of potential future developments in the dairy chain showed that the least preferable scenario from an environmental point of view was the one most similar to trends in the dairy chain of today. Subsequent investigations revealed the same result. Large dairy units with long distance transports lead to a higher environmental impact than small dairy units. On the other hand, small dairy units are those for which the environmental impact is affected the most by the rising variety of cultured products. The changed consumption patterns towards more cultured products and cheese, instead of milk, cause an increased environmental impact with regard to the cultured products, whereas for cheese no clear effect was found. To enable counteraction of negative environmental effects of increased product variety, a method to sequence the production of cultured dairy products with as little environmental impact as possible was developed. The method combines production management methods and environmental systems analysis. A heuristic solution to the sequencing problem was developed and, to the extent possible, validated with an optimisation. The method was used in a case study which revealed not only the importance of a waste minimised sequence but also that of a low production frequency. Life cycle assessment was combined with an actor analysis to examine the potential of the actors in the post-farm chain (dairy, retailer and consumer) to decrease the environmental impact of dairy products. Cutting down waste of product proved to be an effective way to reduce environmental consequences. Saving energy and improving transport patterns gave in general smaller reductions. Choosing organic products decreased most environmental categories at the expense of increased eutrophication. Keywords: environmental systems analysis, life cycle assessment, LCA, environment, dairy, dairy products, cheese, yoghurt, production scheduling, actor analysis

III

List of Publications This thesis is based on work reported in the following five appended publications which are referred to by their Roman numerals in the text. Paper I

Sonesson, U., and Berlin, J. (2003). Environmental Impact of Future Milk Supply Chains in Sweden: A Scenario Study. Journal of Cleaner Production 11:253-266.

Paper II

Berlin, J. (2002). Environmental Life Cycle Assessment (LCA) of Swedish Semi-Hard Cheese. International Dairy Journal 12: 939-953.

Paper III

Berlin, J., and Sonesson, U. (2005). A Life Cycle Based Method to Minimise Environmental Impact of Dairy Production through Product Sequencing. Journal of Cleaner Production, In print.

Paper IV

Berlin, J., Sonesson, U., and Tillman, A.-M. (2005). Minimising Environmental Impact by Sequencing Cultured Dairy Products: Two Case Studies. Submitted manuscript (March 2005).

Paper V

Berlin, J., Sonesson, U., and Tillman, A.-M. (2005). An Actor Analysis of the Environmental Improvement Potentials in the PostFarm Milk Chain Using Life Cycle Assessment. Submitted manuscript (August 2005).

Reprints of Papers I - III are printed by the kind permission of Elsevier Science Ltd.

V

Other publications by the author Sonesson, U., and Thuresson*, J. (2001). Mjölkkedjans miljöpåverkan – en miljösystemanalys av framtidscenarier av försörjningskedjan för mejeriprodukter (Environmental Impact of the Milk Chain: An Environmental Systems Analysis of the Supply Chain for Dairy Products, in Swedish). SIK Report 2001 No 681, SIK: The Swedish Institute for Food and Biotechnology, Göteborg, Sweden. * Author’s surname prior to Berlin.

Berlin, J. (2001). Life Cycle Inventory (LCI) of Semi-Hard Cheese. SIK Report 2001 No 692, SIK: The Swedish Institute for Food and Biotechnology, Göteborg, Sweden. Berlin, J. (2003). Life cycle assessment (LCA): An introduction. In: Environmentally-friendly food processing. Edited by: Mattsson, B. and Sonesson, U., Woodhead Publishing Limited, Cambridge, England. Berlin, J. (2005). Tänk på miljön – Ät upp maten! (Think about the environment: Eat up your food!, in Swedish). In: Mat för Livet – om framtidens livsmedel (Food for Life: food for the future, in Swedish), The Royal Swedish Academy of Agriculture and Forestry (KSLA), Stockholm, Sweden.

VI

The Research Program FOOD 21 The thesis work was carried out within the FOOD 21 research program. FOOD 21 is an interdisciplinary research program funded by the Foundation for Strategic Environmental Research (MISTRA). Natural and social scientists cooperate in analysing the sustainability of agricultural food production from farm to fork. The long-term goal of the program is to define optimal conditions and to develop systems and technologies for a sustainable food chain that offers the consumers high quality products.

VII

Acknowledgements Many people and organisations have assisted me during my years as a doctoral student and I am very grateful to all. My research was carried out in the context of the FOOD 21 research program and financed by the Swedish Foundation for Strategic Environmental Research (MISTRA) to which I wish to express my appreciation. I am indebted to my supervisors, Professor Anne-Marie Tillman and Dr. Ulf Sonesson, a most complementary team. Anne-Marie guided me skilfully into research, environmental systems analysis, publishing articles and gave me good advice. Ulf was a source of inspiration for my research, gave his valuable opinions and was always available for discussions. It is also a pleasure to acknowledge Dr. Berit Mattson and Professor Thomas Nybrant for making good suggestions and giving support along the way and to Professor Hans Lignert, Professor Tomas Olsson, and Dr. Karin Östergren for valuable comments and helpful suggestions on improvement of the thesis. Lora Sharp McQueen is thanked for revising the language of the thesis. Several people contributed data and their knowledge to the papers appended in this thesis. I am grateful to all, in particular Christel Cederberg, C. Cederberg AB; Allan Nilsson and Jörgen Karlsson, Arla Foods Falkenberg; Inger Larsson, Arla Foods; Mustafa Aoufi and Anna-Lena Östensson, Arla Foods Östgötamejeriet; Urban Sterner, Skånemejerier; and Marcus Henningsson, Flora Vita AB, for sharing their expertise in farming and dairy processing. Furthermore, for invaluable assistance with information and data, I am grateful to my colleagues in the Process and Environmental Engineering group at SIK,

VIII

especially Friederike Ziegler, Britta Nilsson, Katarina Lorentzon, Anna Flysjö, Jennifer Davis, Anna Fritzon, Tomas Angerwall and Eva Olsson. Thanks go to all of the people at the Division of Environmental Systems Analysis at Chalmers for always welcoming me as one of the team although my work place has been at SIK. I am deeply thankful to my loving and caring parents, Anita and Roger, for their support and babysitting. My parents in law, Agneta and Bengt, are thanked for all warm concern and babysitting. I am grateful to my sister, Josefin, her husband, Jan, and their lovely children, Amanda and Linnéa, as well as my sister, Ida, and her boyfriend, Anders, for always supporting me. Finally, I thank my beloved husband and daughter, Henrik and Viola, who are the most important persons of my life. Henrik has encouraged me and given me all the support and love I could wish. From Viola and her infectious laughter, I have learned the meaning of life.

Göteborg, August 2005 Johanna Berlin

IX

Table of Contents Abstract .......................................................................................................................III List of Publications...................................................................................................... V Other publications by the author .......................................................................VI The Research Program FOOD 21.......................................................................... VII Acknowledgements ................................................................................................ VIII Table of Contents ........................................................................................................ X 1

Introduction ............................................................................................................ 1 1.1 The Food Market and the Dairy Industry.................................................. 2 1.2 Aim and Objectives....................................................................................... 4 1.3 Appended Papers .......................................................................................... 5

2

The Environmental Perspective of the Dairy Chain ......................................... 8 2.1 Life Cycle Assessment of Dairy Products.................................................. 8 2.2 Types of Environmental Impact Related to the Dairy Chain ................. 9 2.3 Environmental Improvements................................................................... 10 2.4 Environmental Improvements from an Actor’s Perspective................. 13

3

Methods................................................................................................................. 16 3.1 Environmental Systems Analysis Tools ................................................... 16 3.2 Scenario Techniques in LCA..................................................................... 21 3.3 Operational Analysis .................................................................................. 23

4

The Environmental Consequences of Current Trends and Options for Improvement ........................................................................................................ 27 4.1 Environmental Impact of Future Milk Supply Chains in Sweden: A Scenario Study ............................................................................................. 27 4.2 Life Cycle Assessment of Semi-Hard Cheese ......................................... 29 4.3 Minimising Environmental Impact by Sequencing Cultured Dairy Products: Two Case Studies ....................................................................... 30

X

4.4 An Actor Analysis of the Environmental Improvement Potentials in the Post-Farm Milk Chain Using Life Cycle Assessment ...................... 34 4.5 Discussion..................................................................................................... 36 5

Methodological Contributions to Environmental Systems Analysis ............ 41 5.1 A Life Cycle Based Method to Minimise Environmental Impact of Dairy Production through Product Sequencing ...................................... 41 5.2 An Actor Analysis of the Environmental Improvement Potentials in the Post-Farm Milk Chain using Life Cycle Assessment ....................... 43 5.3 Discussion..................................................................................................... 44

6

Conclusions........................................................................................................... 46

7

Future Work ......................................................................................................... 47

References ................................................................................................................... 49

XI

1 Introduction A tasty meal that smells good and looks appetizing to enjoy with a loved family and dear friends is a pleasure of life. While food gives us nutrients, proteins, fats and carbohydrates, it also acts as a source of delight, both for taste-buds and on a social plane. Before we can enjoy a meal, the food has to be prepared, it is purchased from a retailer, processed by an industry, and the raw materials are produced by agriculture. Different modes of transportation have moved the food from one location to another. These activities affect the environment by the use of resources and by emissions to air, water and soil. For example, the energy used in the life cycle of the food chain, agriculture to consumption, was estimated to be approximately 17% of the total energy use in Sweden (Uhlin, 1997). Of this total, agriculture accounted for 15 - 18%, industry 17 - 20%, distribution 20 29% and consumption 38 - 45%. Agriculture stands for approximately 50% of all eutrophication emissions in Sweden, whereas the reminder originates mainly from sewage and transport (SEPA, 1997a, 1997b, 1997c). To the greenhouse gases, the food system contributes around 28% (calculation based on SEPA, 2004 and Uhlin, 1997). The Swedish population consumes seven million tonnes of food each year, the largest part being dairy products, which constitute 25% of the total food intake (SEPA, 1999). The consumption of milk in Sweden is high compared with the average EU value (111.5 kg versus 76.6 kg), but for cheese the consumption is just below average, 17.40 kg versus 18.18 kg (Swedish Board of Agriculture, 2004a). From dairy products, Swedes receive 14% of their energy intake, 25% of their protein, and as much as 66% of their calcium (Swedish Board of Agriculture, 2004b). The importance of dairy products in the Swedish diet is also shown by the Swedish National Food Administration recommendation of a

1

daily consumption of half a litre of milk or a corresponding amount of other dairy products. Before the objective of this thesis is stated, the current trends in the food market and the dairy industry are described. The interrelation of the appended papers is given as the last part of the introduction.

1.1 The Food Market and the Dairy Industry The world market for food products is becoming more integrated and globalised. Integration is expanding with positive economic development, rising population and greater urbanisation (Swedish Dairy Association, 2000). This globalisation affects activities in the life cycles of foods, i.e. agriculture, manufacture, retailing, consumption and the transports involved. Heavy competition between dairy manufacturers, the requirements of product development, and the production of a wider variety of products are the forces working to cause manufacturers to cooperate and merge. This results in larger manufacturing companies with an international market similar to that of retailers. Several mergers and purchases of companies have taken place in the dairy industry. For example, in northern Europe, most dairy companies have merged into regional or national companies, and some even into international companies (Swedish Dairy Association, 2000). Most dairies in Europe are owned by farmers’ co-operatives. There is a trend in the dairy industry towards production in a few large specialised dairies. This specialisation can mean that one dairy produces mainly consumer milk, a second cultured products and a third cheese. This leads to more and longer transports from the dairy to the retailer. Fewer dairies and the dairy farmers’ similar movement towards fewer farms also imply longer distance of transports from the farm to the dairy. 2

The milk chain is a pushing system which has implications for the product portfolio: the milk produced at the farms must be processed promptly into products at the dairy. Since changing the volume of milk production at a dairy farm is a slow process, it is not possible to adjust the amount of incoming milk to rapidly changing market requirements, nor can milk be stored for long periods of time. As the volume of incoming milk to the dairy cannot easily be adjusted, the mix of outgoing products is changed instead, i.e. when the market requirements change, the dairy industry must shift to other products that meet the new requirements. This drives an increasing diversity of products. The companies also contribute to diversity by releasing new products to generate a greater demand for their output. There has been a trend in the past 20 years towards increased consumption of cultured products and cheese, with a decrease in consumption of drinking milk products and butter. In Sweden, cultured milk and cream and yoghurt showed an increase of 18% from 1985 to 2001. The increase in cheese consumption was 13% from 1985 to 2001, particularly soft cheese. Reductions noted in consumer milk consumption were 26% and in butter consumption 37% from 1986 to 2001 (Swedish Board of Agriculture, 2004b).

3

1.2 Aim and Objectives The aim of this work is two-fold. The first aim is to increase knowledge of the environmental impact of the post-farm dairy chain and to assess potential improvements. The second aim is to contribute to development of methodology for environmental systems analysis. Specific objectives related to the first aim are: ▬ assessment of the environmental consequences of ongoing changes in society, which influence the dairy product chain, and ▬ generation of improvement options and assessment of their environmental consequences. Specific objectives related to methodology are: ▬ to develop a method to design an environmentally preferable, or even best, sequence for products that are produced consecutively with the same equipment, and ▬ to develop a method of identifying the activity, for each actor in the postfarm chain, which offers the greatest environmental improvement in a life cycle perspective. In the work covered by the five papers, the goals dealt with are as follows: • to develop potential future scenarios for the milk supply chain; • to identify the key environmental issues for the milk supply chain; • to acquire environmental data for the life cycle of cheese; • to determine the key environmental issue in the life cycle of cheese; • to design a sequence, for a given set of products, which minimises milk waste in a multi-product manufacture; • to investigate the environmental role of the frequency of each product in a sequence; and • to identify the actions, by the dairy, retailer and household, which offer the most environmental improvement in a life cycle perspective. 4

1.3 Appended Papers The thesis is based on five papers, the work for which was conducted from a top down perspective according to the systems analysis approach. Figure 1 shows how the papers are related to each other. Environmental Impact from Future Milk Supply Chains in Sweden: A Scenario Study (Paper I)

Environmental Life Cycle Assessment (LCA) of Swedish Semi-Hard Cheese (Paper II)

A Life Cycle Based Dairy Model to Minimise Environmental Impact by Product Sequencing (Paper III)

Minimising Environmental Impact by Sequencing Cultured Dairy Products: Two Case Studies (Paper IV)

An Actor Analysis of the Environmental Improvement Potentials in the Post-Farm Milk Chain Using Life Cycle Assessment (Paper V)

Figure 1. Interrelation of the five appended papers.

Paper I, a scenario study, gives an overview of the environmental impact of the dairy sector today and of potential changes in the sector according to present trends. During the work with the paper, stakeholders including representatives from a dairy company, a major food retail company, a dairy equipment supplier, an environmental consultant, and researchers working in related areas were interviewed. After an initial round of visits to all those involved, the research group and the persons interviewed attended a seminar at which the scenarios for study were broadly sketched; thereafter, a more detailed description of each scenario was prepared. The preliminary results from simulations using these scenarios were then presented to the same group of people at a second seminar where we made modifications. Finally, the scenarios were given their final form 5

which included: variation of products, both small and large scale dairies, differences in modes of transport, different kinds of retailers, and a diversity of packaging volumes. A model based on material flow and substance flow accounting, together with the impact assessment part of life cycle assessment (LCA), was constructed to assess the scenarios. The knowledge gained about the dairy chain and current trends then constituted the basis for selecting specified subjects to be further investigated in the subsequent work. A lack of good quality, published, environmental data on cheese production and the fact that a major quantity of the milk produced will end up as cheese (approximately 10 kg milk is required to make 1 kg of cheese) together with the ongoing increase in cheese consumption (13% from 1985 - 2001, Swedish Board of Agriculture, 2004b) were the reasons for carrying out a life cycle assessment of cheese. That dairy products have become more and more diversified during recent years is a fact. Product diversity was hard to include in the scenario study (Paper I) as there was no information to be found about its environmental implications. When diversity was discussed in the reference group, it emerged that it has been dealt with from economic and technical viewpoints, but there were few studies from an environmental standpoint. Hence, research was initiated on designing a method that would minimise environmental impact by product sequencing at the dairy (Paper III). The production of cultured products is mostly affected by the greater product diversity when compared with drinking milk and cheese as their batch volumes are much larger. A method for sequencing cultured products was developed. A heuristic solution, which was designed intuitively and based on production rules, was worked out for yoghurt production. To determine whether the heuristic solution gave the best possible sequence from an environmental perspective, an optimisation solution was also made. This detailed scheduling model was successfully included in a life cycle assessment of

6

the production schedule at two dairies, Paper IV. The sequenced products were yoghurt, sour cream, cold sauce and crème fraiche, all with multiple flavours. During the work with the case studies, the role of frequency of each product to be sequenced attracted attention. Technical scenarios with differing frequencies were evaluated with life cycle assessment methodology in order to improve the environmental impact. Improvement assessment was also the topic of Paper V, but this time from an actor perspective. A literature search did not reveal any publication dealing with the measures taken by actors in the post-farm dairy chain from a life cycle perspective. Hence, Paper V searched for the potential action, by the dairy, retailer and household, that offers the most environmental improvement of the product life cycle. The actions investigated were improved energy efficiency, better transport patterns, reduced milk and product losses and organic labelling. The three products considered were milk, yoghurt and cheese. Literature studies and interviews with stakeholders to estimate improvement potentials were used in combination with LCA methodology.

7

2 The Environmental Perspective of the Dairy Chain A life cycle environmental perspective is applied in this work. First, previously published studies of the dairy chain using life cycle assessment methodology are given here. Next, the types of environmental impact related to the dairy chain are discussed, followed by a section about how reductions of these impacts are treated in the literature. The last part examines how actors can contribute to environmental improvements.

2.1 Life Cycle Assessment of Dairy Products As with most food items, milk products originate from agriculture. The dairy farm produces the milk, and it is collected by a truck which delivers it to the dairy. At the dairy the milk is processed into a variety of dairy products and packaged for the consumer. After that they are delivered to the retailer where the products are displayed for consumers on a refrigerator shelf or in a cold room. A dairy item purchased by a consumer is transported to the household and stored in the refrigerator before the final consumption. Each of these activities in the milk chain causes environmental impact. The impacts of dairy products have been identified and evaluated in several studies using life cycle assessment (LCA) methodology. Nilsson and Lorentzon (1999) studied the environmental consequences of processing milk. Høgaas Eide (2002b) made an LCA of milk in which three milk-producing dairies were investigated. A screening LCA of milk powder was undertaken by Blonk et al. (1997). Lorentzon et al. (1997) studied the environmental effects of coffee cream, from processing at the dairy to the purchaser. An LCA of the production of cultured milk was made by Grøtan (1996), while butter was the subject of the LCA pilot study at an Italian dairy company by Masoni et al. (1998). A soft cheese, a Camembert, was examined in an LCA by Bernhard and Moos (1998). Although the system boundaries differed in these publications, a consistent finding of those which included a farming component was that agriculture had by far the 8

greatest environmental impact for most parameters. The ranking of the contributions of the other life cycle phases to the environmental impact is not as clear. The answer differs depending on the product and the environmental impact category considered. However, the dairy, the production of packaging, and the transport between retailer and household seem to make a higher contribution than the other transports, the retailer and the household.

2.2 Types of Environmental Impact Related to the Dairy Chain Agriculture contributes to global warming by emissions of methane and nitrous oxide and to a lesser extent through emission of CO2 originating from the use of fossil fuels. Livestock is the source of most of the methane emissions partly because of ruminants’ enteric fermentation and partly due to manure management with methane production under anaerobic conditions. Nitrous oxide is released as a result of nitrification and de-nitrification processes in the soil, as well as nitrogen transformation in manure. Emissions of nitrogen pollutants are also the source of both eutrophication and acidification. The release of ammonia is linked to the farmyard manure. Ammonia is not acidifying in a chemical sense, but it has a strong acidifying effect as a result of nitrification in the soil. Nitrate leaches from the arable land. Dairy farming makes heavy demands of the land for example by soil erosion and compaction. On the other hand, grazing ruminants also preserve valuable biotopes. Concerning the use of resources, phosphorus should be highlighted, since it is used not only as fertiliser but also as a mineral feed additive (Cederberg, 2002). During processing at the dairy, separation, homogenisation and pasteurization use most energy (Høgaas Eide and Ohlsson, 1998, Nilsson and Lorentzon, 1999). The cleaning operations have also been identified as a major source of environmental impact (Lorentzon et al., 1997). Water, cleaning agents (commonly used are nitric acid and sodium hydroxide) and energy are required. The Cleaning in Place (CIP) system is usually used in dairies. This means that 9

rinsing water and cleaning agents are circulated through tanks, pipes and process lines without dismantling the equipment. Effluents consist of milk residues and water containing the cleaning agents. How much of the effluent reaches the environment depends on the sewage treatment. The production of the package as well as the waste management of packaging is considered critical to the environmental impact, especially for products with a low degree of processing. For dairy products, consumer milk is the least processed, next are cultured products, and the most processed is cheese. The manufacture of packaging as well as its distribution was shown to have a considerable impact on energy use (17% of the total life cycle) and global warming potential (18%) in a study of milk by Høgaas Eide (2002a). The package was a one litre paperboard carton. The waste management of the carton was the main contributor to eco-toxicity (59% of the total life cycle). The design of the package is also highly relevant for the product loss in the consumer phase (Johansson, 2002). At the retailer and in the household, the electricity needed to keep the products cold causes environmental impact. The environmental impact depends on the energy mix used for producing the electricity. The Swedish average electricity mix is made up of approximately 45 % each of hydroelectric- and nuclear power, the remainder being produced from oil and combined heat and power plants using bio fuel (Swedish Energy Agency, 2004). Hydroelectric power affects biodiversity and landscape aesthetics and nuclear power cause emissions of radioactive substances and radioactive waste.

2.3 Environmental Improvements Within the food sector measures to decrease environmental impact are continuously applied. Examples include improved plant nutrient balances on farms, more energy efficient processes, reduction of material in packaging, 10

better logistic solutions, and environmental requirements for procurement. Action for environmental improvement is traditionally taken at each part of the food chain. Measures to reduce environmental load in the agricultural phase are important, as all LCAs of dairy products have identified farming as the dominant contributor to the environmental impacts (see Section 2.1). Different types of farming practices have been investigated in several studies. A review of the environmental impact assessment of conventional and organic milk production was made by de Boer (2003). Three studies of the differing agriculture practices in northern Europe, i.e. south western Sweden, the Netherlands and southern Germany, were reviewed. Cederberg and Mattsson (2000) assessed organic and conventional farming practices in Sweden, and the Dutch study was carried out by Iepema and Pijnenburg (2001). Both of these found that organic milk production is a way to reduce pesticide use and mineral surplus. However, this type of farming requires more grazing land than conventional farming. For Sweden this was positive in that it promotes the domestic goals of preserving biodiversity. However in the Netherlands, land is a scarce resource, which makes greater land use a negative impact. Haas et al. (2001) who published the German study came to the conclusion that, by renouncing mineral nitrogen fertilizer, organic farming reduced the energy use and the global warming potential in comparison with conventional farming. All three studies were based on comparisons of experimental farms. The conclusion by de Boer (2003) was that differences between production systems require assessments of a large number of farms for each production system. Another study of quantified improvement action in agriculture, made by Hospido et al. (2003), dealt with Galician (Spanish) milk production. The actions investigated were the reduction of milk losses during milking, a changed feed composition with more maize and less silage, and the implementation of

11

treatment systems for water and air emissions. Moreover, a measure introduced to improve dairy herds is the reduction of mastitis. Mastitis is an inflammatory reaction of udder tissue to bacterial, chemical, thermal or mechanical injury. The environmental impact of mastitis was assessed with LCA methodology by Hospido and Sonesson (2005). A standard scenario for the incidence of mastitis (present-day reality in Galicia) was compared with an improved one. A reduction in the incidence of mastitis implies that the same amount of milk would be produced in a shorter period of time or by fewer cows, hence the impact on the environment would be lower. For example, the greenhouse gas emissions from the agriculture sector of Spain could be reduced by 0.56% if measures were taken to reduce the incidence of mastitis. At the dairy, measures to lower the environmental impact include using less energy, cleaning agents and water, decreasing the wastage of milk and other products, thus raising the yield. A systems analysis of the energy used at two dairies was published by Karlsson et al. (2004). Each step of the process and each item of equipment was assessed. Specific actions to save energy were identified and quantified with time perspectives from one to more than five years; they concluded that the energy use could be decreased by 8% if all of the actions were realised. Høgaas Eide et al. (2003) assessed four Cleaning in Place (CIP) methods using life cycle assessment methodology. Production of cleaning agents, transport, actual cleaning at the dairy and waste management of the containers were included. The actual cleaning at the dairy was the most important part of the life cycle. The CIP methods, enzyme-based cleaning and one-phase alkaline cleaning, both using small volumes and low temperatures, were found to be the best alternatives for the impacts of energy use, global warming, acidification, eutrophication and photo-oxidant formation. A wasteoptimised product scheduling could decrease wastage of milk and products and in that way increase the yield. A methodology for incorporating ecological considerations into the optimization of design and scheduling of batch processes

12

was conducted by Stefanis et al. (1997); this included a case study of a cheesemaking dairy. The optimizations were based on both process economics and environmental impact. In a theoretical study, Grau et al. (1995) introduced optimization of process-sequence dependent changeover waste (as an environmental issue) in product scheduling of a batch production unit. Høgaas Eide (2002a) reviewed eleven LCAs of milk packaging and concluded that a light-weight recyclable package, with properties that did not increase the loss of product, was the best milk packaging alternative. Improvement options for the retailer and in the household include more energy efficient cold storage and other action to decrease product losses. An analysis of the ways to reduce the energy requirements of refrigerator cabinets at the retailer was made by Axell (2002). A test of 119 household refrigerators was made by Sonesson et al. (2003). The difference was as much as 11.27 MJ/(per litre and year) between the refrigerator using the most energy and the one using the least. Studies of improvement actions to reduce the losses of dairy products in the household are rare. There are different kinds of losses in the household, such as product left in the container when the consumer considers it is empty, losses during preparation, food left on the plate or in the glass, or thrown in the bin or the sink when it has spoiled or passed shelf life. Each of these losses could be decreased. The importance of the design of the packaging for food losses was tested by Johansson (2002). In a comparison of packages for yoghurt, the cup shape had 3.4% yoghurt left when considered empty, whereas the gable top had 8.7%. Losses of dairy products after mealtimes and storage were studied by Sonesson et al. (2005) but they did not include improvement actions.

2.4 Environmental Improvements from an Actor’s Perspective A range of actions is available to decrease the environmental impact. The best choice for each actor may not be the same for all who are involved in the same consumption chain because their domain of influence differs. For example, 13

although the consumer is not able to influence the production schedule at the dairy, there is still a choice between organic or conventional milk during purchasing. Actions taken by one person may lead to substantial improvements in phases of the life cycle controlled by other people. Environmental studies of the food chain from an actor perspective are rare in the literature. Jungbluth et al. (2000) published one of the few and pointed out the consumer as the actor with the greatest potential to influence the food chain in environmental matters. They found that the consumer had the widest range of choice to make decisions that reduce the environmental impact. The method applied was reviewing published LCA studies. Five single aspects of decisions were identified: type of agricultural practice, origin, packaging material, type of preservation and consumption. These decisions were then assessed in simplified LCAs of meat and vegetables. The most important options for a reduction of environmental impact were found to be refusal of air-transported products, a preference for organic foods and a reduction in meat consumption. Another important decision made by the consumer is the choice between foods. An environmental comparison of four meals was made by Carlsson-Kanyama (1998). Although the study did not include an actor analysis, it is relevant for this topic in that it is a study of decisions made by one actor in the food chain. This comparison concluded that a meal composed of tomatoes, rice and pork has nine times higher impact on climate change emissions than a meal made from potatoes, carrots and dry peas. A subsequent investigation by CarlssonKanyama et al. (2003) shows that contrasting diets with similar energy content can vary in energy input by a factor of four, from 13 to 51 MJ. They also concluded that the least energy consuming diets, which are far from the Swedish average, are not in line with current trends. In another study of the influence of decisions taken by one actor in the food chain, Lindgren and Elmquist (2005) examined the environmental and

14

economic impacts of decision making at an arable farm. Variations in prices, subsidies, the farmer’s attitudes about environmental concerns, and the farmer’s skill in making production allocation choices were studied. With regard to economic performance, either organic farming or a conventional cultivation with a large amount of pesticides and fertilizers offered the most profit. The former benefited from higher subsides and selling prices, the latter from large yields. Regarding environmental impacts, the result depended on the impact category studied. To obtain a low contribution to eutrophication and acidification, conventional farming was preferred, due to ammonia emissions during slurry spreading. Nevertheless, to reduce global warming, the organic alternative was preferable because no mineral fertilizer was used.

15

3 Methods This thesis is based on five papers, the work for which was conducted by various methods. The aim of each study determined the choice of methods and system boundaries used during modelling. The methods used were: environmental systems analysis methods of various types, mostly life cycle based; a scenario technique; quantitative problem techniques; interviews and seminars with actors in the dairy chain; and visits to dairies. A general description of the methods follows with specific comments about how they were used in the work.

3.1 Environmental Systems Analysis Tools Environmental systems analysis includes several tools which can be categorised as flow models, monetary models, procedural methods and risk assessment. Examples of tools based on physical flows are life cycle assessment, material flow accounting and substance flow accounting. Cost-benefit analysis facilitates assessing total costs, including environmental costs, and benefits from a planned project. In design for the environment, a wide range of procedural methods are used, in an effort to include the environmental dimensions into the design process. Risk assessment is a broad term that covers several types of assessments, to deal with human health or environmental aspects. The risk can also vary from diffuse to specific and can be associated with customary usage or accidents. Frameworks for comparing differing approaches was devised by Baumann and Cowell (1999); they were further developed in Wrisberg et al. (2002) and Finnveden and Moberg (2005). Baumann and Cowell (1999) emphasize the importance of practical integration of existing approaches for a variety of applications, rather than developing new tools, and Wrisberg et al. (2002) agreed. Furthermore, Finnveden and Moberg (2005) came to the conclusion that, depending on the objects the tools focus on, different tools cannot easily replace each other. In the following the tools used in the papers

16

included in this thesis are briefly outlined; life cycle assessment, material flow accounting and substance flow accounting. 3.1.1

Life Cycle Assessment

Life cycle assessment (LCA) is used in all of the appended papers. In Paper I, the impact assessment element of LCA methodology was chosen for the interpretation of the result. Paper II is a descriptive LCA of cheese. An LCA study is either descriptive or change-oriented. A descriptive (attributional or accounting) study describes a system as it actually is. A change-orientated (consequential or effect-oriented) study analyses the consequence of a choice. In Paper III, a sequencing method is worked out and it is shown that it may be linked to LCA. To evaluate the scenarios of different frequencies of production in a sequence, LCA was linked to the sequencing method (Paper IV). The work in Paper V started with descriptive LCAs carried out for milk, yoghurt and cheese. Then these LCAs were changed to show the estimated improvement measure of each actor. A comparison of the changed LCAs revealed the action that made the greatest difference from an environmental point of view for each actor in the chain. A brief description of LCA in general with specific examples from the papers follows. Life cycle assessment is a tool for evaluating the environmental impact associated with a product, process or activity during its life cycle. This is accomplished by identifying and quantitatively or qualitatively describing its requirements for energy and materials, and the emissions and waste released to the environment. The life cycle is included in the assessment, which means that the product under study is followed from the initial extraction and processing of raw materials through manufacturing, distribution, and use, to final disposal, including the transports involved. Besides identifying and quantifying the environmental impact of the product or activity, LCA also identifies what activities in the product life cycle contribute the most to this impact. An LCA is 17

an ISO standardised tool (ISO, 1997, 1998, 2000a, 2000b) and included in the standard is a working procedure, illustrated in Figure 2 and described below.

Goal & Scope Definition

Inventory Analysis Interpretation Impact Assessment Classification Characterisation Normalisation Weighting

Figure 2. Working procedure for an LCA. The unbroken line indicates the order of procedural steps and the dotted lines show iteration. (Baumann and Tillman, 2004, and ISO, 1997)

An LCA starts with an explicit statement of the goal and scope of the study, the functional unit and allocation methods used, the system boundaries, the assumptions and limitations, and the impact categories chosen. The functional unit is quantitative and corresponds to a reference flow to which all flows in the LCA are related. Allocation is the method used to partition the environmental load of a process when several products or functions share the same process. The allocations used in the papers are mostly based on economic partitioning, that is, on the value of the items produced as reflected in their relative prices or the gross sales value. Economic allocation is commonly used in relation to food products. Various system boundaries can be chosen depending on the purpose of the study. When the whole system is assessed, from resource exploration to the waste management, the study is designated a cradle to grave study. Sometimes only parts of the life cycle are of interest: in a cradle to gate analysis, resource exploration and production are included but not use and waste 18

management. Another example is gate to gate analysis in which neither the resource exploration nor waste management are included; sometimes the use phase is omitted as well. The goals and scope, excluding the allocation, used in the appended papers, are summarised and listed in Table 1. In the inventory analysis a flow model of the technical system is constructed using data on inputs and outputs, i.e. resources, energy requirements, emissions to air and water, and waste generation for all activities within the system boundaries. The inventory analysis is followed by impact assessment, in which the data are interpreted in terms of their environmental impact. In the classification stage, the inventory parameters are sorted and assigned to specific impact categories. The next step is characterisation, where inventory parameters are multiplied by equivalency factors for each impact category. Thereafter all parameters included in each impact category are added to obtain the total for that category. Examples of environmental impact categories are acidification, eutrophication and global warming. For many LCAs, the analysis is concluded by a characterisation, which is the case for all LCAs included in the appended papers. However, some analyses involve the further step of normalisation, in which the results of the impact categories are compared with the total impact in the geographical region relevant for the study. The size of the region depends on the nature of the impact; some impacts are global while others are regional or even local. During weighting, the kinds of environmental impacts are weighted against each other to find an overall value for the total environmental impact. For the purpose of the papers in this thesis, normalisation and weighting were regarded as unnecessary.

19

Table 1. An overview of goals and scope in the LCA approach in the appended papers. Goal

Functional unit

System

Environmental

boundary

impact

Milk chain

Assess the environmental

The total

Gate to grave:

Energy, global

Paper I

impact of future supply

amount of milk

from incoming

warming,

chains for dairy products.

from the farms

dairy transport

eutrophication,

in a region

to household

acidification and

consumption

POCP Resources,

LCA of

Acquire data and identify

1 kg semi-hard

Cradle to gate:

Cheese

key issues in the life cycle

cheese at the

from dairy farm energy,

Paper II

of cheese.

consumer table

to consumer

eutrophication,

table

acidification, global warming, POCP, eco and human toxicities

Sequence

Construct a sequence

One processing

Cradle to gate:

model

model to minimise milk

sequence of

from dairy farm acidification,

Paper III

waste. Obtain the impact

yoghurt

to dairy

global warming

of a production sequence.

products

delivery gate

and POCP

Cultured

Obtain the impact by

One processing

Cradle to gate:

Eutrophication,

product

production of a sequence

sequence of

from dairy farm acidification,

sequencing

at two dairies. Investigate

yoghurt, sour

to dairy

global warming

Paper IV

the environmental

cream, cold

delivery gate

and POCP

implications of the

sauce and

Eutrophication,

frequency of each product crème fraiche produced. Most

Identify the actions taken

• 1 kg milk

Cradle to grave: Energy,

effective

by the dairy, retailer and

consumed

from dairy farm eutrophication,

actor action

household, that give the

• 1 kg yoghurt

to household

global warming

Paper V

most environmental

consumed

consumption

and POCP

improvements in a life

• 1 kg cheese

cycle perspective.

consumed

POCP is photochemical ozone creation potentials

20

3.1.2

Material Flow Accounting and Substance Flow Accounting

Material flow accounting (MFA) describes all in- and outflows and accumulation of a material, substance or element in a geographic area for a given period of time (Udo de Haes et al. 1997). Depending on the type of material studied, a further distinction of MFA is often applied. Bulk-material flow analysis studies flows of materials, such as wood, iron or plastics, in a given region. Flows of substances such as nitrogen compounds and single elements such as cadmium or lead within a region are traced in a substance flow accounting (SFA) (Udo de Haes et al. 1997). Van der Voet et al. (1995) state that one of the aims of SFA is to obtain an overview of the economic and environmental flows in a specific geographical region. Cederberg (1999), for example, quantified the flows of nitrogen, phosphorus and potassium connected with the production and consumption of food in a Swedish district. Both MFA and SFA were used in Paper I, together with the impact assessment element of LCA. The flow under study was milk in a region of central Sweden. Together with the accounting for the resources and energy consumption, this is the part that is based on MFA. However, accounting for the in- and outflows of all the emissions that occurred in each step of the milk product system as well as taking into account the protein content can be considered in an SFA. A life cycle perspective was used in the study.

3.2 Scenario Techniques in LCA In the context of LCA the SETAC Europe working group on scenario development defined a scenario as a description of a possible future situation relevant for specific LCA applications, based on specific assumptions about the future and, when relevant, a description of a path from the present to the future (Weidema et al., 2004, Pesonen et al., 2000). They distinguished three types of scenario application in LCA: technology, environment, and valuation. In this thesis most of the scenarios used can be categorised as the technology type, which concern the technosphere or more specifically the product system, except 21

for those used in Paper I. They were based on changes in societal developments and consumer behaviour, which in turn led to modified product systems. According to van der Voet et al. (1995), changes in society, which affect substance flow, can be linked to an SFA study. Examples of this are the SFA studies connected to scenario analysis, presented by Sonesson et al. (1997) and Björklund et al. (2000). Both of these dealt with the environmental consequences of scenarios for waste management in Swedish cities. 3.2.1

What-if and Cornerstone Scenarios

According to Weidema et al. (2004) and Pesonen et al. (2000) there are two principal approaches to scenario development in LCA studies: what-if scenarios and cornerstone scenarios. The what-if approach is the most widely used of the two and has a shorter time perspective. The environmental impact of specific changes is compared or tested. What-if scenarios often result in quantitative comparisons of the options selected. In a cornerstone approach scenarios are chosen to give an overall view of the subject of study. They mark the outer limits of possible developments in order to ensure that the differences between them stand out clearly and to facilitate the identification of key differences. What-if scenarios were used both in Paper IV (Cultured product sequencing) and Paper V (Actors’ action selection). In Paper IV case studies were carried out for several technical scenarios. Three scenarios for each dairy, with differing frequencies of the products, were devised for the sequencing. To vary the product frequency means to vary how often a given product was made weekly. The frequencies of the products chosen for the scenarios were: the frequency currently used for each product (2 - 5 times per week); twice a week for each product; and, in the last scenario, 1 - 2 times per week for each one according to their shelf life. The scenarios were designated: Reference, Goal and Future. The Reference scenario reflected the current situation. The Goal scenario was 22

assumed to be achievable for most dairies within a reasonable time. The Future scenario was believed to be attainable in the future. In Paper V improvement measures were environmentally assessed and compared. Each action taken by each actor can be viewed as a what-if scenario, although it was not termed so in the paper. The actions investigated were improved energy efficiency, better transport patterns, reduced milk and product losses and organic labelling. These changes could be made by the dairy, the retailer and the household, although each actor could undertake only some of them. The time perspective was five years from now. The scenario technique used in Paper I (the milk chain study) mirrored possible developments in the milk supply chain. The scenarios were defined with ideas from the reference group discussions (see Section 1.3) but with simplifications to make them feasible. The selections were made to give an overall view of the milk chain and thus represent the most extreme developments in society and consumer behaviour, similar to the cornerstone approach. The reference scenario reflects the milk chain as it was structured in 1999. A version called Large Scale assumes a shift towards larger units within both industry and among retailers. The Splendid Times version resembles the Large Scale but with the difference that there has been greater economic growth. An economic recession with a decrease in the use of cars was included in the Harsh Times version. The last scenario simulated was the Green IT Wave which assumes a less materialistic lifestyle. Each variation, except the reference scenario, had a time perspective of 20 years in the future.

3.3 Operational Analysis Operational analysis is a group of quantitative techniques for solving systems analysis problems. Included in the category are optimisation, linear programming, dynamic programming, and queue theory (Gustafsson et al., 23

1982). In Paper III a heuristic procedure (an intuitively designed procedure) was worked out to achieve the environmentally preferred production sequence of yoghurt products. The heuristic solution was validated with an optimised one. Both of the solutions are described below. 3.3.1

The Heuristic Solution for a Product Sequence

Heuristic procedures are intuitively designed and can give a good approximate solution. Although they cannot be guaranteed to give an optimal solution they are often used for very large problems (Hillier and Lieberman, 1995). The heuristic procedures used in Paper III were based on rules used in a yoghurt producing dairy. More specifically, the rules were based on the characteristics of each product, which determined the choice of technique selected for a product change. The techniques were cleaning, rinsing and the pushing principle. Cleaning causes the most waste, use of cleaning agents and water, followed by rinsing which uses only water; the pushing principle causes the least waste and does not use cleaning agents or water. Therefore the best schedule uses the pushing principle the most, while rinsing and cleaning are used as seldom as possible. The first step was to make a matrix of all of the products and list their individual characteristics (base, presence of rhubarb and vanilla, allergenic potential and intensity in colour). Then the sorting procedure starts with grouping according to the yoghurt base. Within each base, products containing rhubarb and vanilla were placed last. Products with allergic substances were next to the last. Finally, the rest of the products within the base group were sorted by increasing colour, pale ones first and the dark ones last. This sorting procedure is a solution to the production schedule. However, there is no guarantee that this is the optimal solution, because the sorting is done according to processing rules used during manufacturing, and there is not a test of all possible solutions to find the best one for the schedule. 24

3.3.2

Finding a Waste Minimised Sequence for a “Travelling Salesman Problem”

Finding the optimal product sequencing solution involves searching through all possible combinations of the manufacturing order of the set of products to be sequenced. Our problem had similarities to the “travelling salesman problem” (TSP): given a set of N cities, find the shortest route connecting them all, with no city visited twice (Sedgewick, 1988). For our problem, the cities were interpreted as yoghurt types and the routes connecting them were weighted according to the waste volume caused by a product change. The waste volume was determined by the product change technique (cleaning, rinsing, pushing principle), which in turn was governed by the processing rules. This problem formulation gave rise to a weighted, directed TSP. Moreover, the TSP graph was complete since there is a route between any two products in the graph. A large TSP is insoluble in practice, as the number of solutions that must be checked grows in proportion to the faculty of the cities involved (N! = 1 · 2 · 3 ··· N). The optimisation presented in Paper III was made to validate the heuristic solution for the waste minimised sequence, for as large a number of products as possible within a reasonable time. For more information about TSP, see Sedgewick (1988). For the optimisation in Paper III the problem was: Given a mix of products, find the production sequence that causes the least waste. The waste that occurs during a product change depends on the properties of the two products. The problem has as many as N! solutions for the schedule, where N is the total number of products in the sequence. First an exhaustive search was made to check all possible solutions for the scheduling of products. The result of this was not satisfactory; as it took 89 minutes to check the number of solutions (12! = 479 001 600) for 12 products on a standard laptop. To enable scheduling more products, it was decided to reduce the number of solutions checked without sacrificing optimality. The sum of the waste, x, for the first sequence was 25

calculated. For the following searches, it was fruitless to continue any sequence for which the summed waste was greater than x, therefore these ones were removed. After this 14 products could be sequenced with minimum waste in 140 minutes. To improve the optimisation even more, a method known as branch-and-bound was chosen (Sedgewick, 1988). For a given partial product sequence, a lower bound of the total waste of all product sequences, which started with that particular partial product sequence, was computed. If the waste of the best sequence found so far was less than this bound, then all of those sequences could be disregarded and did not need to be searched. The algorithm was applied recursively until all possibilities were searched. In this way the workload was significantly reduced, since the bound could be computed very efficiently. The branch-and-bound technique reduced the number of solutions dramatically. By using both of the techniques described to limit the full searches, we were able to make a schedule of 21 products within a reasonable time (30 minutes). With 21 products 5.1 · 1019 sequences were tested. The best of those was the solution to our problem. Note that the algorithm is still guaranteed to find the waste minimised sequence (Sedgewick, 1988).

26

4 The Environmental Consequences of Current Trends and Options for Improvement Current trends in society and industry have consequences for the milk chain and its environmental impact. Assessment of these is one of the aims of this work and was dealt with in three of the appended papers (I, II and IV). When information gathered reveals a trend of negative impact, it is time to find a way to improve the situation. Improvement possibilities, another objective of the thesis, were dealt with in all the appended papers, in particular Papers IV and V. This chapter gives an overview of the findings from each of the papers. A discussion about the findings in relation to trends and improvement options concludes the section.

4.1 Environmental Impact of Future Milk Supply Chains in Sweden: A Scenario Study The first paper aimed to form an overview of some potential developments of the supply chain for dairy products in a specific region of Sweden, and the effects these would have on of the environment. The milk supply chain under study is located in the central part of Sweden, roughly an east-west line 100 km north of Stockholm and southward, excluding the southernmost region, Skåne. The milk flow in this region was investigated with the tools MFA and SFA and, to some extent, LCA (impact assessment). The life cycle perspective used in the study included resource use for transporting whole milk from farms to dairies, processing in dairies, distribution to retailers, retail stores, transport to households and finally storage in homes. The environmental impact of energy production, manufacture of packaging material, waste management and sewage treatment caused by the milk chain was also integrated. The agriculture was not included. The major dairy products included were drinking milk, cream, butter, cultured products (e.g. yoghurt) and cheese.

27

Five scenarios were worked out to mirror possible developments in the milk supply chain, see Section 3.2.1. The total volume of milk from the farms in the area was constant in all scenarios; the same amount of milk was also leaving the system but in different combinations of products and losses. The most preferable scenario from an environmental view for most impact categories was the one designated Harsh Times, while the least preferable scenario was Splendid Times. These two were the extremes of the economic growth in society. In Harsh Times the price was the most important factor, and in Splendid Times service was essential. This was shown by the products consumed, the kind of retailers used, how to get to the retailer, and the size and material used for packaging. In Harsh times drinking milk is the product most consumed. Electronic shopping is frequently used or the neighbourhood retailer, as fewer people can afford a car. The products are produced in large cardboard packages to reduce the price of the products. In Splendid Times the amount of cheese and cultured products consumed were increased at the expense of drinking milk. Travelling by car to distant retailers or specialised shops is the usual way of purchasing food. Many small bottles and packages of products with different flavours are preferred. The bottles and packages are made of polyethylene terephthalate (PET) and high-density polyethylene (HDPE). Measures could be taken to improve all five scenarios. The production of packaging materials, the waste management and the transports had the greatest impact on the environment and resource use in all five. For transports it was the part between retailers and households that contributed the most. Consequently, improvement action in these areas would decrease the environmental impact. The industrial part was important when considering resource use, such as the net energy turnover, but had a minor impact on the effect categories included.

28

4.2 Life Cycle Assessment of Semi-Hard Cheese The purpose of the cheese study was to identify the environmental consequences of Swedish cheese production and the most environmentally important activities within its life cycle. LCA fulfilled the requirements of this objective and was found to be the appropriate method for the study. The focus of the investigation was on the cheesemaking dairy. One of the most popular semi-hard cheeses in Sweden was selected, Hushållsost. The system studied covered the extraction and production of the ingredients required for cheesemaking, as well as retailers, households, waste management and the transports involved. The main outcome was that milk production at the farm was the activity in the life cycle that contributed most to the environmental impact categories included. The result agrees with other LCAs of dairy products (see Section 2.1). The agricultural activities accounted for as much as 93% to 99%, depending on impact category, of the total life cycle contribution. The contribution from the cheesemaking dairy was 0.5% to 4% depending on impact category. Apart from the agriculture and the cheesemaking dairy, the retailers and the production of plastic were also contributors to the environmental impact of Hushållsost. To make substantial improvements in the environmental performance of cheese production, it is necessary to address the activities that contribute the most to the environmental impact. Improving farming practices with the environment in mind could substantially raise the performance of the system studied. However, farming is beyond the scope of this thesis, and therefore I refer to Cederberg (1998) and Cederberg and Bergström (1999) who suggest possible ways of improving milk production. From the dairy’s perspective, an important improvement would be to decrease the amount of milk required to produce cheese, and in that way reduce the 29

environmental impact from the agriculture, as less milk would need to be produced. Identifying and minimising the losses of milk during the production of cheese would lower the consumption of milk without affecting the final product. There are methods to increase the yield of cheese during the cheesemaking process, but they all have the drawback of affecting the quality of the cheese. These methods include increasing the water and salt content of the cheese, and retaining more whey protein in the curd (Bertelsen et al., 1983, Johansson, 2001). Raising the protein content of the incoming milk also gives a better yield of cheese (Johansson, 2001). The protein content of milk depends on such factors as the breed of cattle and the fodder used. However, the choice of fodder will affect the environmental impact of farming; hence, with the information available, it was not possible to predict how changing this might affect the outcome of the entire system. Consumers, too, could contribute by minimising wastage of cheese in the household and reducing car transportation from the retailer to the household.

4.3 Minimising Environmental Impact by Sequencing Cultured Dairy Products: Two Case Studies The diversity of cultured milk products available continues to rise. In Europe the dairy sector holds the top position in terms of innovative markets in the food sector (Innovaction, 2003). From the dairy perspective, it is mostly the production of cultured products that is affected by increased diversity. The reasons are that it is the cultured products which are available in greatest variety and also that cultured milk cannot be recycled into the process again. The waste from cultured milk is either used as animal fodder or, when the water content is high in the milk residues, it goes to the sewage treatment plant. Loss of product (also called waste) occurs during each change of product; most of these take place just before the product is packaged. When product diversity rises, the 30

number of product changes increases and, consequently, a rise in waste of product occurs. The amount of waste, as well as use of cleaning agents, water and energy requirements, depends on the products involved in the change. Therefore the production schedule, where the processing order is decided, is a key activity for reduction of the rising environmental burden caused by diversity. A model was designed to generate the best sequence of products, from an environmental point of view, which causes the least waste possible while a constant total volume is produced (Paper III). Two case studies using the sequencing model were reported in Paper IV. It was found that the dairies do have options to counteract the environmental effects associated with their production sequences, for example to use a waste minimised order in the production planning for each single day of production. A second option was the frequency of production of each product. By examination of production schedules on a weekly basis, it was found that the same type of product was produced as many as five times in the same week. Therefore, scenarios with variations in the frequency of these types were assessed, using the model (Paper III), and analysed with LCA methodology. The frequency with which products are processed was found to have a significant influence on the amount of waste generated. The results clearly showed that a decrease in frequency of production per product reduced the waste generated by the sequence. When the frequency was changed from 2 - 5 times per product and week to twice weekly (from Reference scenario to Goal scenario, see Section 3.2.1), the waste was decreased from 11 715 kg to 8 698 kg for Dairy A and from 12 194 kg to 9 301 kg for Dairy B per week. (The dairies investigated were designated Dairy A and Dairy B.) On a yearly basis, the reduction of waste would be the amount corresponding to approximately 3.5 days of production for Dairy A and 4 days for Dairy B. It was found to be

31

possible to reduce the product frequency even further, and this was done in the Future scenario (see Section 3.2.1); this one had a product frequency of 1 - 2 times per product depending on its shelf life. A comparison of the waste generated each week in the Reference scenario with that in the Future scenario showed the waste was decreased from 11 715 kg to 5 900 kg for Dairy A and from 12 194 kg to 7 100 kg for Dairy B per week. On a yearly basis, the reduction of waste would be the amount corresponding to approximately 7 days of production at Dairy A and 7.5 days at Dairy B. Less waste not only reduces environmental impact but also makes economic savings possible. Dairies can reduce cost not only by decreasing waste but also by reducing working time. With a changed product frequency, it was possible to decrease the number of production days by one day per week with the Future A scenario at Dairy A. At Dairy B the packaging was reduced by one day per week for one of the machines with the Goal B scenario; packaging was reduced by another day per week at each of two machines with the Future B scenario. By changing the product frequency, leading to fewer product changes, substantial environmental improvement was achieved with reduced product waste, energy savings, and a decrease in use of cleaning agents and water. In a life cycle perspective, a decrease of the impact categories of 1.3% was achieved with the Goal A sequence, and 2.5% with the Future A sequence at Dairy A; this was 1.5% with Goal B and 2.6% with Future B at Dairy B. By changing the perspective to the dairy where the actual improvement could take place, it was revealed that the reduction in environmental impact, for the Goal and Future scenarios in comparison with the Reference scenario, would be even greater than the dairies’ own contribution for some categories (Table 2). This result is due to the extreme dominance of agriculture in the life cycle environmental impact of dairy products, see Section 2.1 and Paper II.

32

Table 1. The life cycle impact reduction (including agriculture) for the scenarios Goal and Future (due to fewer product changes than in the Reference scenario), in relation to each dairy’s environmental impact for the Reference scenario.

Dairy A Goal A Future A Dairy B Goal B Future B

GW

EP

AC

POCP

(% decrease)

(% decrease)

(% decrease)

(% decrease)

33

310

99

22

63

600

190

43

38

280

110

25

68

490

200

45

GW: Global Warming (100 year perspective) EP: Eutrophication AC: Acidification POCP: Photochemical ozone creation potentials.

While a decrease of the product waste generated by any sequence would probably be an environmental improvement, this depends on the dairy’s response to alternatives and the consumers´ response to the dairy’s choice. The dairy has the option to reduce the milk volume processed while maintaining the same volume of products for sale as before. This offers the environmental improvements described above. The other option is that the dairy could choose to raise the volume of products for sale instead. For such a strategy to be successful the consumers must increase their intake of cultured milk products, which would mean lowering their intake of other food items. Comparing the environmental impact of the other food items with that of the cultured milk products would show whether the environment would become better or worse. Although production scheduling is not a major issue in environmental work in dairies today, the clear improvement potential this study shows may change this. That the product frequency in the production schedule has an impact on waste is common knowledge within the dairy industry, but that it was shown to be a parameter of such large magnitude is something new. No study known has been conducted with processing data to test the role of product frequency before. 33

Another advantage of working with production scheduling is that these environmental improvements do not involve any equipment investments.

4.4 An Actor Analysis of the Environmental Improvement Potentials in the Post-Farm Milk Chain Using Life Cycle Assessment The challenge in working with environmental improvements is to select the action offering the most substantial progress. However, not all actions are open to all actors in a product chain. The aim of Paper V was to identify which one of the measures, i.e. waste reduction, increased transport efficiency, energy savings or the choice of organic labelled products, offers the greatest improvement potential for the dairy, the retailer or the household. The products assessed were milk, cheese, and yoghurt. The systems under study for the three products were: agriculture, dairy processing, retailer, households and all connected transports. By collecting data on possible improvements from the post-farm actors themselves and from literature, and by using these data to recalculate published LCAs on the selected dairy products, the aim was met. The result is presented for the three actors individually. For the dairy, no action stands out as superior to others. It can improve its energy use and transport system, while also decreasing the wastage. The choice of purchasing organic products was not considered to be within the power of the dairy. Of the three product types, the greatest improvement potential is for yoghurt (2% reduction of the life cycle impact), where all three actions lead to less global warming and energy use, and decreased wastage also diminishes the eutrophication and photochemical ozone creation potentials (POCP). For drinking milk dairies, improving transportation is the most efficient action (reduction of 2%). For cheesemaking dairies, actions that reduce wastage improve all impact categories (by 1%), while greater energy efficiency is visible only as decreased use of energy (by 1%), not as reductions of the other impact categories.

34

The improvement potentials in the retail sector have a limited effect on the life cycle environmental impact of dairy products. The retailer can reduce its own wastage and energy use. By adopting more energy efficient refrigerators, the total life cycle energy use can be decreased by 1%. It is the actions of households that offer the largest improvement potential. Waste minimisation by the consumer would clearly lead to significant environmental improvement, since all impact categories are reduced in this instance (in the range of 2% to 9%). Choosing organic products has a major positive effect on all three products with regard to energy use (improvement potential in the range of 8 to 14%) and global warming (2%). However, there is a risk of increasing some impact categories. Choosing organic milk raises the eutrophication greatly, more than 20%. Also POCP shows a slight rise for cheese (1%). The reason for the high contribution to eutrophication for organic milk products, a result also reported by Cederberg and Mattsson (2000), is the high nitrate loss per kilo of milk produced at an organic farm. This high nitrate loss has two explanations. First, even though the nitrate loss per hectare is lower in organic farming than in conventional farming, the yields are also lower, which means a higher nitrate loss per kilogram yield. Second, the two farming practices differ in the choice of concentrate feed. Peas, which have a rather high nitrate leakage in relation to yield, are commonly used in the organic feed. The conventional farm purchases concentrate feed with a lower nitrate discharge per kg feed (Cederberg and Mattsson, 2000). When the possible improvement potentials by the three actors are compared, the household has by far the greatest improvement potential, followed by the dairy and then the retail sector. This is because households are less efficient today, causing large losses; they are still using inefficient home transport and cold storage. The dairy industry can still make improvements but, since both the processing and transport are efficient today, the potential for further efficiency

35

is lower in percentage. For example, the waste from the dairy processing of drinking milk is often reused in the process for yoghurt production. It should be kept in mind that even if each improvement might seem small in relative numbers, the dominant part of the environmental impact originates from a part of the system not directly affected by the actors studied, namely agriculture. Despite this, some of the improvement actions studied, potentially undertaken by actors other than agricultural ones, lead to a substantial decrease in the life cycle environmental impact, for example a decrease of 14% was achieved in the category energy use by choosing organic cheese.

4.5 Discussion By comparing the milk supply chain scenarios in Paper I with the ongoing trends in the food market and the dairy industry (Section 1.1), it is possible to gain an understanding of the direction of the environmental impacts affected. The scenario that correlates the most closely with the trend description in Section 1.1 is Splendid Times, based on a positive economic development of society. Furthermore, the effect of globalisation is shown by production in a few large specialised dairies. Other similarities are that cultured products and cheese in small packages are preferred to drinking milk. In spite of the trend, the diversity of products was not identified as an issue in itself in Paper I. Unfortunately, the current scenario contributed the most to the environmental impact of the five assessed; if product diversity had been included, it would have raised the impact even higher (Paper IV). To extend the assessment of the trends, the scale of milk processing, changed consumption pattern, and diversity of products are more thoroughly discussed. By using the scenarios in Paper I, it was possible to evaluate the differences in environmental impact in relation to the scale of dairy production. Although the number of dairies differed in the scenarios, the amount of milk processed was 36

the same. The extreme scenarios with a comparable product portfolio were Harsh Times, with the fewest dairies, and Green IT-wave, with more than four times as many. The environmental impact from processing at the dairy, the transportation from farms to the dairies, and the deliveries from the dairies to the retailers are likely to be affected by the scale of dairy processing. The impacts from these activities are summed in Table 3, for both of the scenarios.

Table 3. The differences in environmental impact and energy use caused by the scale of milk processing. Scenario Harsh Times produced in a few large dairies and Green IT-wave in several smaller dairies. NET

GW (tonnes

EP (tonnes O2 AC (kmol H+ POCP

(GJ/year)

CO2 eq./year)

eq./year)

eq./year)

(tonnes ethene eq./year)

Harsh Times Green IT Wave

2.2

0.95

5600

14

25

2.5

0.95

4500

12

20

Net: Net energy turnover, i.e. the amount of energy purchased GW: Global Warming (100 year perspective) EP: Eutrophication AC: Acidification POCP: Photochemical ozone creation potentials eq: equivalents

When compared with small scale production, large scale dairy production was not environmentally better to the extent that it could offset the increased impact of the longer transport distance, see Table 3. This result may be influenced by a slight difference in the product portfolios, and certainly by the fact that rising product diversity was not reflected in these scenarios. The growing diversity of products has consequences during processing, such as a corresponding rise in product changes, which in turn raises the waste of product, hence the environmental impact, Paper IV. The environmentally preferred 37

production has as few product changes as possible and when a change is required the techniques using least resources should be used. During the work with Papers III and IV, a difference between small dairy companies and large ones of the effects of product diversity appeared. Small companies with high product diversity produce only a small volume of each. This implies a large amount of waste during production of a small volume. It showed that the amount of waste depending on the product change was sometimes larger than the amount of product obtained. However, in larger dairy units, where the produced volume is higher, the proportion between waste and product was more favourable. It was also found that for the large dairy units studied, the product order was good and could not be improved in the sense of waste minimisation. Moreover, we found both in large and small dairy units a great potential for improvement by reducing the number of times each product was produced, Paper IV. A high frequency of production of the same type of product during the same week and also sometimes twice during the same day was common. The latter was due to late orders from the sales department. This lack of communication between production and sales management has environmental and economic consequences. The highlighting of the importance of production scheduling and the product frequency for the environmental impact, with consequent economic saving potential, has received a very positive response from the dairy industry. An increased consumption of cheese and cultured products at the expense of drinking milk may change the environmental impact of the dairy chain. In a qualitative comparison of cheese and milk it was found that cheese is processed more at the dairy and it is also stored for quite some time to mature, Paper II. On the other hand, the proportion that should be compared is ten litres of milk to one kg of cheese, as that is the amount of milk required to produce the cheese. This has consequences for the packages, as also ten packages are required for the milk and just one for the cheese, although the type of packaging

38

differs. All transportation after the dairy is also tenfold in volume for the milk. Furthermore, the loss of drinking milk is higher in the household than the cheese losses (Paper V). From the comparison of cheese and milk, no straightforward answer can be given about a change in environmental impact. However, a qualitative comparison of yoghurt and milk may be done. Yoghurt is processed more at the dairy and the product diversity is higher, Paper IV. Large product losses in the household are also a drawback for yoghurt, Paper V. This comparison indicates that a change in consumption from milk to yoghurt has negative consequences on the environment. Nevertheless, a comparative LCA for cheese and milk and for yoghurt and milk are required to obtain definite answers. To conclude, the trends in society and industry have negative consequences for the environmental impact of the dairy chain. First the scenario of Splendid Times showed that the environmental impact is increasing with the ongoing trends of today. Then the further assessment of some of the trends generated a similar result. Large dairy units with long incoming and outgoing transports have a higher impact than small dairy units. On the other hand, the great increase in environmental impact caused by the rising variety of cultured products was found to affect small dairy units the most. A negative environmental impact would probably occur, if yoghurt were consumed instead of milk. However, as an exception to a rule, the change in consumption from milk to cheese is not likely to change the environmental impact. Everything can be improved, including unwanted environmental effects. Although within the dairy chain, agriculture is the actor which causes the largest part of the environmental impact of dairy products, Paper V revealed that the post-farm actors do have significant options to improve the environmental impact. For some environmental categories, the improvement potential for an individual actor was even greater than its own contribution to the category. For

39

example, when the dairy’s reduction of wastage in producing yoghurt was assessed, a decrease of eutrophication by 2.4 times more than the dairy’s own contribution was possible. Another result revealed by this work is the identification of waste of product as a major environmental issue in a life cycle perspective. For the actors in the post-farm dairy chain, the waste of milk and product has not previously been considered an environmental issue except from the perspective of waste management. However, when the impacts of the common environmental activities such as transport efficiency, energy savings and organic production were compared with minimisation of product waste, the latter was as effective as the other actions; for several environmental categories, it was shown to be the preferred action (Paper V). Since the waste of product is a parameter that affects not only the part of the life cycle where it arises, but also the earlier parts of the chain, it is of significant importance to include the life cycle from the cradle in an assessment. For milk products this implies that the system boundary should include the agriculture and further steps in the life cycle before the product waste occurs. This work deals with milk products, but product waste minimisation is likely to be a major important environmental issue for all products with a high environmental impact in the early part of the life cycle, e.g. meat, aluminium packaging and paper products.

40

5 Methodological Contributions to Environmental Systems Analysis A problem solving approach was applied in this work. First, the problem was identified, then available methodologies were explored. When a suitable method was lacking, it was necessary to devise one. Two of the Papers, III and V, contribute to methodology which is described below and followed by a discussion.

5.1 A Life Cycle Based Method to Minimise Environmental Impact of Dairy Production through Product Sequencing The rising number of dairy products affects their environmental impact in a life cycle perspective. During dairy processing, the production schedule is affected by more frequent product changes, hence also cleaning operations. This causes more milk waste, use of cleaning agents, water and energy, which all contribute to the environmental impact in a life cycle context. To counteract this increasing environmental impact, a method was developed to schedule a large number of products in a way that would cause the least possible environmental impact. During the search for a suitable method, only two studies were found that took the environment into account during production scheduling: a theoretical study of process-sequence dependent changeover waste, conducted by Grau et al. (1995), and a methodology for incorporating ecological considerations into the optimisation scheduling of a cheesemaking dairy (Stefanis et al., 1997). The latter included a case-study of a cheesemaking dairy, but included only two products. Consequently, to solve the problem of sequencing a great number of products, a new method was required. The goal of Paper III was to find a practical method to calculate a sequence of a great number of cultured products, which is optimal or close to optimal, from a 41

waste minimisation perspective. Furthermore, to show the full environmental implications of the waste minimisation, the sequencing model should be possible to connect to life cycle assessment (LCA) methodology. An inter-disciplinary approach was chosen for the method, making use of both production scheduling and environmental systems analysis. To find the sequence of products that is optimal from an environmental perspective, the target function (also called optimisation criterion) must be carefully selected. Studies of life cycle assessment literature gave us the function. Life cycle assessments were made for several dairy products, see Section 2.1, and a consistent finding was that agriculture had the greatest environmental impact. Consequently, it was concluded, for the remaining parts of the life cycle of dairy products (the dairy, retailer, consumer household, waste treatment and also all connected transports), that the action which would offer the best outcome from an environmental perspective was the minimisation of milk waste. For the sequencing problem, minimisation of milk waste would also lead to a minimised use of cleaning agents and water for cleaning during a product change in the sequence, depending on the techniques used for product changes. Therefore, the choice of target function fell on milk waste. The preferred method would be to find an optimised solution through mathematical optimisation. Nevertheless, this problem has similarities to the “travelling-salesman-problem” (see Section 3.3.2), which means that to find the optimal solution involves searching through a vast number of potential solutions. In practice, this is possible for only a limited number of products (Sedgewick, 1988). Hence, a heuristic method was developed, which was able to handle a large number of products, see Section 3.3.1. A drawback to a heuristic solution is that it cannot be guaranteed that it is also optimal (Hillier and Lieberman, 1995). Therefore, it was decided to validate the result of the heuristic method with a sequence achieved through optimisation, with as large a number of products as could be handled with the optimisation method within a 42

reasonable time. By using techniques to limit the full searches during optimisation (see Section 3.3.2), 21 products could be scheduled within a reasonable time (30 minutes). The algorithm was still guaranteed to find the weight minimised sequence (Sedgewick, 1988). A sequence was simulated with both the heuristic and the optimised solutions and then the results were compared. Several starting orders were tested, and the results from both of the methods always gave the same sequence. Accordingly, we can state that the heuristic method gives the optimal sequence from an environmental perspective (that is a waste minimised solution) up to 21 products. This implies that the method used will also find optimal solutions for all production sequences including fewer products, since all possible combinations are tested in the algorithm. There is a strong reason to believe, although it is difficult to prove mathematically, that the heuristic solution for a sequence including more than 21 products will also be the optimal one, as there is no known impediment in the sequence that relates to the number of products. This production scheduling method gives the sequence, for a given set of products, which causes a presumably minimum amount of milk waste and, consequently, a low use of cleaning agents and water. Paper III also successfully demonstrated that the full environmental consequences of production according to the sequence could be assessed with LCA methodology.

5.2 An Actor Analysis of the Environmental Improvement Potentials in the Post-Farm Milk Chain using Life Cycle Assessment To reduce unwanted environmental consequences, a range of actions is available. The problem is to know what measures offer the most substantial improvement for each actor. A quantified assessment of the improvement potential of the actors in the post-farm chain was not found in the literature. Therefore, a method needed to be devised. The method of life cycle assessment 43

was chosen for the analysis, which was combined with the identification and quantification of environmental improvement potentials available to the actors along the post-farm milk chain. First, ordinary LCAs were undertaken for the products under study (milk, yoghurt and cheese) to find a reference for the environmental impact of each product. Then, to select the kind of improvement action to examine, a literature study was undertaken together with interviews of representatives from the dairy industry. The most significant measures were found to be increased energy efficiency, improved transport patterns, reduced product losses and purchase of organic products. It was decided that neither product nor package changes were to be considered. Literature study and interviews were used again, this time for estimation of the improvement potential by each actor along the chain. Then, the estimated values of the dairies were verified by their representatives. Thereafter, the LCAs were recalculated using the improvement values for each measure. To find the greatest potential, the environmental impact of the modified LCAs and the original ones were compared for each product. To identify the most efficient action for the dairy, the retailer and the household, separately, the potential actions of each of them were analysed one at a time. The method used in this study rests on the availability of reliable LCA data from the production system, combined with accurate estimates of the improvement potentials. Also necessary is the life cycle perspective to describe the full effect of a potential improvement. This is particularly important for measures dealing with waste reduction.

5.3 Discussion The combination of environmental systems analysis with production scheduling, as was used in Paper III, is a new approach to execute product sequencing for the dairy industry. To choose an environmental systems analysis approach (i.e. 44

LCA) when searching for the environmental target function, used for the sequencing, was found to be the best way of finding the parameter to minimise for the most environmental improvement. For sequence solving, operational analysis was used (a systems analysis methodology), which was then combined with LCA for evaluation of the environmental impact of the nearly waste minimised sequence. This approach of finding the production schedule that contributes the least to the environmental impact, while still producing the same amount of products, was appropriate for analysis of production with existing process equipment. Life cycle assessment studies are often interpreted with dominance and contribution analyses, i.e. what life cycle phases and particular environmental loads (emissions and resource consumptions) contribute the most to the over-all results. In LCA there are seldom interpretations of the sphere of influence of the various actors along the product chain. Paper V demonstrated the feasibility of such an approach, by showing that the life cycle environmental implications of improvement potentials may be quantified on an actor basis. Moreover, a life cycle approach turned out to be crucial, since measures taken by an actor had consequences in other parts of the product life cycle. Both of the methods developed can be used on other product life cycles. The production scheduling model can be used as it is for any dairy, producing cultured products. Furthermore, the model can be applied to any batch production sequence, although the processing rules must be changed to suit the products under study. The actor analysis methodology suggested can be used as it is for any life cycle.

45

6 Conclusions Research for this thesis was guided by a two-fold aim: to increase the knowledge of the environmental impact of the post-farm dairy chain, while assessing potential improvement, and to contribute to development of methodology for environmental systems analysis. This double aim was met in the five appended papers together. The most important findings are listed. • The future scenario closest to current trends, as well as the assessments of production in large units, increased consumption of yoghurt, and the rising number of cultured products, all highlight an increase in the environmental impact of the dairy chain. • The agriculture part is the most dominant contributor to the environmental impact in the LCA of cheese; the most important improvement in a life cycle perspective for the cheesemaking dairy is to decrease its waste of milk and cheese. • A method of constructing the environmentally preferred, life cycle based, production sequence was developed for cultured dairy products. Using the method will result in both environmental and economic savings for the dairy. • Reduction of the frequency of production of each cultured product lowers the environmental impact. • An actor analysis method was devised to help the actors to find the measure that makes most substantial environmental improvement in a life cycle perspective. • Waste of product was revealed to be a key issue for environmental consequences in a life cycle perspective, and reducing the waste offers a substantial decrease in the impact.

46

7 Future Work The research done in this thesis has provoked other interesting questions. As waste of product turned out to be such an important environmental issue for the dairy chain, although it had not previously been seen as one, I have several suggestions about product waste. First, since the consumer has a large influence on the environmental impact, which involves great losses of product, and data about household losses is very poor, a thorough assessment of consumer behaviour is needed. Second, more accurate information about losses of milk and products at the other parts of the post-farm chain, such as the retailer and dairy, is also needed. Third, an assessment of milk waste sources from the milking in the phase of agriculture would be useful. Fourth, the scope can be broadened to the whole food chain, and to cover the losses for Sweden. Almost certainly, the environmental impact from the losses makes up a large part of the total impact of the food sector. This thesis has revealed that increasing the diversity of cultured dairy products raises the environmental impact, both in a life cycle perspective and at the dairy unit. This trend can be assumed to continue, by observing the dairy shelves at German or British retailers, since Sweden still has fewer products; this is likely to be an even more important issue in the future. The consequences would be greatest for small dairy units, with rising losses of product leading to an increase in environmental impact as well as economic losses. When considering small dairy units compared with larger ones, the issue of the increasing diversity in dairy products should be included. Therefore, a suggestion is to update and improve the scenarios of Paper I with respect to the consequences of an increasing number of dairy products. A specific suggestion for improvement of the dairy process is to reduce the milk waste in dairies by production scheduling, and to combine it with better

47

monitoring of the losses. Henningsson (2005) showed how to reduce losses by increasing the control of them with an optical instrument, conductivity meter and density meter. This would decrease the amount of waste during each product change independently of changing techniques used. Combining the waste minimised sequence with Henningsson’s monitoring methods would be very interesting and has great potential indeed. The final suggestion is to use the actor analysis method for other food chains. I will conclude this thesis with a recommendation for environmental improvement: Let us keep in mind that each food item has an environmental history. If a food product is wasted instead of consumed, all of the environmental impact that it has caused during its life cycle has happened for nothing. So think about the environment, and eat up your food!

48

References Axell, M. (2002). Vertical display cabinets in supermarkets –Energy efficiency and the influence of air flows. Doctoral Thesis, Department of Building Services Engineering, Chalmers University of Technology, Göteborg, Sweden. Baumann, H., and Cowell, S. J. (1999). An evaluation framework for conceptual and analytical approaches used in environmental management. Greener Management International, Journal of Corporate Environmental Strategy and Profile: 109-122. Baumann, H., and Tillmann, A.-M. (2004). The Hitch Hiker´s Guide to LCA. An orientation in life cycle assessment methodology and application. Studentlitteratur, Lund, Sweden. Bernhard, S., and Moos, T. (1998). Ökobilanz des Camemberts (Ecobalance of Camembert, in German). Fachverein Arbeit und Umwelt, Zurich, Switzerland. Bertelsen, E., Sven-Nilsson, E., and Willart, S. (1983). Från mjölk till mejeriprodukter –Driftskontroll (From milk to dairy products: Operation management, in Swedish). Livsmedelsbranschernas yrkesnämnd : Brevskolan, Stockholm, Sweden. Björklund, A., Bjuggren, C., Dalemo, M., and Sonesson, U. (2000). Planning biodegradable waste handling in Stockholm, Journal of Industrial Ecology 3: 4358.

49

Blonk, H., Lafleur, M., and van Zeijts, H. (1997). Towards an environmental information infrastucture for the Dutch food industry: Exploring the environmental information conversion of five food commodities, Appendix 3: Screening LCA on milk powder. IVAM Environmental Research & Centre for Agriculture and Environment, Amsterdam, the Netherlands. Carlsson-Kanyama, A. (1998). Climate change and dietry choices: How can emissions of greenhouse gases from food consumption be reduced. Food Policy, 23(3-4): 277-293. Carlsson-Kanyama, A., Pipping Ekström, M., and Shanahan, H. (2003). Food and life cycle energy inputs: consequences of diet and ways to increase efficiency. Ecological Economics 44: 293-307. Cederberg, C. (1998). Life Cycle Assessment of Milk Production: A Comparison of Conventional and Organic Farming. SIK Report 1998 No 643, SIK: The Swedish Institute for Food and Biotechnology, Göteborg, Sweden. Cederberg, C. (1999). Flows of plant nutrients in food production and consumption in a Swedish district. Ambio 28: 502-508. Cederberg, C., and Bergström, S. (1999). Näringsflöden och markanvändning i ekologisk och konventionell mjölkproduktion i sydsverige (The flow of nutrients and land use in ecological and conventional milk production in the south of Sweden, in Swedish). Hushållningssällskapet i Halland, Halmstad, Sweden. Cederberg, C. and Mattsson, B. (2000). Life cycle assessment of milk production; A comparison of conventional and organic farming. Journal of Cleaner Production 8: 49-60.

50

Cederberg, C. (2002). Life Cycle Assessment (LCA) of Animal Production. Doctoral Thesis, Department of Applied Environmnetal Science, Göteborg University, Göteborg, Sweden. de Boer, I. J. M. (2003). Environmental impact assessment of conventional and organic milk production. Livestock Production Science 80: 69-77. Finnveden, G., and Moberg, Å. (2005). Environmental systems analysis tools: An overview. Journal of Cleaner Production 13: 1165-1173. Grau, R., Espuna, A., and Puigjaner, L. (1995). Environmental Considerations in Batch Production Scheduling. Computers and Chemical Engineering 19 Suppl.: 651-56. Grøtan, M. (1996). Livsløpsanalyse ved produksjon av kulturmelk (Life cycle assessment of production of cultured milk, in Norwegian). Norges Landbrukshøgskole, Institutt for Næringsmiddelfag, Ås, Norway. Gustafsson, L., Lanshammar, H., and Sandblad, B. (1982). System och Modell En introduktion till systemanalysen (System and Model: An introduction to Systems Analysis, in Swedish). Studentlitteratur, Lund, Sweden. Haas, G., Wetterich, F., and Köpke, U. (2001). Comparing intensive, extensive and organic grassland farming in southern Germany by process life cycle assessment. Agriculture, Ecosystems and Environment 83: 43-53. Henningsson, M. (2005). Loss Minimisation in Dynamic Food Processes, Doctoral Thesis, Division of Food Engineering, Department of Food Technology, Engineering and Nutrition, Lund Institute of Technology, Lund University, Lund, Sweden.

51

Hillier, F. and Lieberman, G. (1995). Introduction to Operations Research. McGraw-Hill Book Co, Singapore. Hospido, A., Moreira, M. T., and Feijoo, G. (2003). Simplified life cycle assessment of galician milk production. International Dairy Journal 13: 783-796. Hospido, A., and Sonesson, U. (2005). The environmental impact of mastitis: A case study of dairy herds, Science of the Total Environment 343(1-3): 71-82. Høgaas Eide, M., and Ohlsson, T. (1998). A Comparison of Two Different Approaches to Inventory Analysis of Dairies. International Journal of LCA 3(4) 209-215. Høgaas Eide, M. (2002a). Life Cycle Assessment (LCA) of Industrial Milk Production. Doctoral Thesis, Department of Food Science, Chalmers University of Technology, Göteborg, Sweden. Høgaas Eide, M. (2002b). Life Cycle Assessment (LCA) of Industrial Milk Production. International Journal of LCA 7(2): 115-126. Høgaas Eide, M., Holmleid, J. P., and Mattsson, B. (2003). Life Cycle Assessment (LCA) of Cleaning-in Place Processes in Dairies. Lebensmittel Wissenschaft und Technologie 36: 303-314. Iepema, G., and Pijnenburg, J. (2001). Conventional versus Organic Dairy Farming. A Comparison of Three Experimental Farms on Environmental Impact, Animal Health and Animal Welfare. MSc Thesis, Animal Production Systems Group, Wageningen University, Wageningen, The Netherlands.

52

Innovaction BV. (2003). In: CIAA Confederation of the food and drink industries of the EU. Data and trends of the EU food and drink industry. CIAA Confederation of the food and drink industries of the EU, Brussels, Belgium. ISO. (1997). Environmental Management: Life Cycle Assessment: Principle and Framework, EN ISO 14040:1997, International Organization for Standardization, Geneva, Switzerland. ISO. (1998). Environmental Management: Life Cycle Assessment: Goal and Scope Definition and Inventory Analysis, EN ISO 14041:1998, International Organization for Standardization, Geneva, Switzerland. ISO. (2000a). Environmental Management: Life Cycle Assessment: Life Cycle Impact Assessment, EN ISO 14042:2000, International Organization for Standardization, Geneva, Switzerland. ISO. (2000b). Environmental Management: Life Cycle Assessment: Life Cycle Interpretation, EN ISO 14043:2000, International Organization for Standardization, Geneva, Switzerland. Johansson, B. (2002). Förpackningens betydelse för produktförluster i hemmet. Spill uppmätt med hjälp av testgrupp. (The influence of the package for product losses in the household. Losses measured by a test panel, in Swedish), Packforsk, Stockholm, Sweden. Johansson, R. (2001). Personal communication. Arla Foods Ost AB Falkenbergsmejeriet. SE-311 44 Falkenberg, Sweden. Jungbluth, N., Tietje, O., and Scholz, R. (2000). Food Purchases: Impacts from the Consumers´ Point of View Investigated with a Modular LCA, International Journal of LCA 5(3): 134-142. 53

Karlsson, M., Rohdin, P., Karlsson, F., and Moshfegh, B. (2004). Energikonsekvenser av strukturerat energieffektivitetstänkande för Arla Foods (EKSET) (Energy consequences by a structured energy efficiency thinking for Arla Foods, in Swedish), Department of Energy Systems, Linköping University of Technology, Linköping, Sweden. Lindgren, U., and Elmquist, H. (2005). Environmental and economic impacts of decision-making at an arable farm: An integrative modelling approach. Ambio 34 (4-5): 393-401. Lorentzon, K., Olsson, P., Reimers, V., and Stadig, M. (1997). Uthållig livsmedelsproduktion - En energi- och miljöstudie med inriktning mot kyl-, frysoch helkonservbehandling (Sustainable food production: A study of energy requirements and environmental consequences by the treatment of refrigeration, freezing and preservation, in Swedish), SIK: The Swedish Institute for Food and Biotechnology, Göteborg, Sweden. Masoni, P., Sára, B., Scartozzi, D., Tarantini, M., and Raggi, A. (1998). A Life Cycle Assessment Pilot Study in an Italian Dairy Company. Paper presented at: 3rd Int. Conf. on Ecobalace, Tsukuba, Japan. Nilsson, B., and Lorentzon, K. (1999). Energi- och miljöpåverkan inom mejerier (The Energy requirements and their related Environmental Impact from Milk Producing Dairies, in Swedish). SIK: The Swedish Institute for Food and Biotechnology, Göteborg, Sweden.

54

Pesonen, H.-L., Ekvall, T., Fleisher, G., Huppes, G., Jahn, C., Klos, Z. S., Rebitzer, G., Sonnemann, G. W., Tintinelli, A., Weidema, B. P., and Wenzel, H. (2000). Framework for scenario development in LCA, International Journal of LCA 5(1): 21–30. Sedgewick, R. (1988). Algorithms. 2nd edition. Addison-Wesley Publishing Company. Reading, Massachusetts, USA. SEPA. (1997a). Fosfor: livsnödvändigt, ändligt och ett miljöproblem (Phosphorus: Essential, Limited and an Environmental Problem, in Swedish). Report 4730, Swedish Environmental Protection Agency, Stockholm, Sweden. SEPA. (1997b). Kväve från land till hav: huvudrapport (Nitrogen from Land to Sea: Main report, in Swedish). Report 4735. Swedish Environmental Protection Agency, Stockholm, Sweden. SEPA. (1997c). Kväveläckage från svensk åkermark : beräkningar av normalutlakning och möjliga åtgärder : underlagsrapport (Nitrogen Leakage from Swedish Agricultural Land., in Swedish). Report 4741. Swedish Environmental Protection Agency, Stockholm, Sweden. SEPA. (1999). Svensk produktion med miljön i fokus (Swedish production focused on the environment, in Swedish). Swedish Environmental Protection Agency, Stockholm, Sweden. SEPA. (2004). Sweden´s National Inventory Report 2004: Submitted under the United Nations Framework Convention on Climate Change. Swedish Environmental Protection Agency, Stockholm, Sweden.

55

Sonesson, U., Dalemo, M., Mingarini, K., and Jönsson, H. (1997). ORWARE: A simulation model for organic waste handling systems. Part 2: Case study and simulation results. Resources, Conservation and Recycling 21: 39-54. Sonesson, U., Janestad, H., and Raaholt, B. (2003). Energy for Preparation and Storing of Food. SIK Report 2003 No 709. SIK: The Swedish Institute for Food and Biotechnology, Göteborg, Sweden. Sonesson, U. Antesson, F., Davis, J., and Sjödén, P.-O. (2005), Home Transports and Wastage: Environmentally Relevant Household Activities in the Life Cycle of Food. Ambio 34 (4-5): 371-375. Stefanis, S. K., Livingston, A. G., and Pistikopoulos, E. N. (1997). Environmental impact considerations in the optimal design and scheduling of batch processes. Computers and Chemical Engineering 21(10): 1073-94. Swedish Board of Agriculture. (2004a). Marknadsöversikt – animalier, Rapport 2003:24, (A review of the market: The yield of meat and dairy products, in Swedish) Jönköping, Sweden. Swedish Board of Agriculture. (2004b). Konsumtion av livsmedel och dess näringsinnehåll –Uppgifter t.o.m. år 2002, Rapport 2004:7 (The consumption and nutrient contentof food, in Swedish) Jönköping, Sweden. Swedish Dairy Association. (2000). Vision 2010 Mjölk (Vision 2010 Milk, in Swedish) Swedish Dairy Association, Stockholm, Sweden. Swedish Energy Agency. (2004). Energy in Sweden: Facts and Figures 2004. Swedish Energy Agency, Eskilstuna, Sweden.

56

Udo de Haes, E., van der Voet, E., and Kleijn, R. (1997). Substance flow analysis (SFA): An analytical tool for integrated chain management: In: From Paradigm to Practice of Sustainability, pp. 32-42. Paper presented at: ConAccount workshop, 21-23 January 1997, Leiden, the Netherlands. Uhlin, H.-E. (1997). Energiflöden i livsmedelskedjan (The energy flow of the food chain, in Swedish), Vol. 4732, Swedish Environmental Protection Agency, Stockholm, Sweden. van der Voet, E., Kleijn, R., van Oers, L., Heijungs, R., Huele, R., and Mulder, P. (1995). Substance flows through the economy and environment of a region – Part 1: Systems definition. Environmental Science and Pollution Research 2: 8996. Weidema, B. P., Ekvall, T., Pesonen, H.-L., Rebitzer, G., Sonnemann, G.W., and Speilmann M. (2004). Scenarios in life-cycle assessment. Society of Environmental Toxicology and Chemistry (SETAC), Pensacola, FL, USA. Wrisberg, N., Udo de Haes, H. A., Triebswetter, U., Eder, P., and Clift, R. (2002). Analytical tools for environmental design and management in a system perspective. Kluwer Academic Publishers, Dordrecht, the Netherlands.

57