Available online at www.sciencedirect.com
Journal of Cleaner Production 16 (2008) 483e498 www.elsevier.com/locate/jclepro
Minimising environmental impact by sequencing cultured dairy products: two case studies Johanna Berlin a,b,*, Ulf Sonesson b b
a Environmental Systems Analysis, Chalmers University of Technology, SE-412 96 Go¨teborg, Sweden SIK AB, The Swedish Institute for Food and Biotechnology, P.O. Box 5401, SE-402 29 Go¨teborg, Sweden
Received 31 March 2005; accepted 5 October 2006 Available online 18 December 2006
Abstract The increased production of cultured milk products has environmental consequences. To counteract the environmental impact from the dairy industry, it is important to process the products in a sequence designed to minimise waste. In a previous study a model was constructed to minimise the waste caused by a sequence for a given set of products and to calculate the environmental impact of a waste minimised sequence. This study applies successfully the model in case studies at two dairies. The number of products to be sequenced varied: Dairy A had 34 products and Dairy B had 16. The sequenced products were yoghurt, sour cream, cold sauce and cre`me fraiche, all with multiple flavours. The difference in number of products to be sequenced offered the opportunity to use both of the two model sequencing solutions: the heuristic and the optimised. The role of frequency of each product to be sequenced was investigated. Scenarios with differing frequencies were used in the case studies. The result showed clearly that the waste caused by a sequence decreased when product frequency was reduced. From a life cycle perspective, the environmental impact of processing cultured milk products can be greatly reduced by adopting sequences with fewer changes of product. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Production scheduling; Product frequency; Waste minimisation; LCA; Yoghurt
1. Introduction 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 [1]. From a life cycle perspective, diversity affects the environmental impact of dairy products in different ways. At the dairy company level, decisions are taken on issues such as number and location of dairies and product portfolio. Decisions of this kind can affect transportation patterns both to
* Corresponding author. SIK AB, The Swedish Institute for Food and Biotechnology, P.O. Box 5401, SE-402 29 Go¨teborg, Sweden. Tel.: þ46 31 3355600; fax: þ46 31 833782. E-mail address:
[email protected] (J. Berlin). 0959-6526/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jclepro.2006.10.001
and from dairies. Production scheduling is a key activity at the dairy. It influences a wide range of issues with environmental implications, such as waste of product, need for cleaning of production equipment and waste of packaging. More kinds of products require more space in retailers’ refrigerated shelves, with increasing energy consumption. The product waste at the retailer can also be affected by the rising number of products to be stored. Consumers, finally, may generate more waste as they buy a wider variety of products in smaller packages. With a broader spectrum of products in the fridge, more products may be wasted because they are not used in time. Waste also occurs because some product is always left in the container; smaller containers lead to an increase in packaging waste. Several dairy product studies have been made from an environmental life cycle perspective: milk [2], coffee cream [3]
484
J. Berlin, U. Sonesson / Journal of Cleaner Production 16 (2008) 483e498
soured milk [4], butter [5], milk powder [6], soft cheese [7] and semi-hard cheese [8]. Although the system boundaries differed in these studies, a consistent finding in all of those that included farming was that agriculture had the greatest environmental impact. From the system perspective of the milk chain, the clear dominance of agriculture implies that the most important action that may be undertaken by the following parts of the supply chain, to decrease the environmental burden, is to minimise milk losses. For the dairies’ direct impact, Høgaas Eide [9] calculated that cleaning operations accounted for 80% of eutrophication impacts and 30% of the energy use. To decrease the dairies’ direct impact, less cleaning operations and milk losses seem to be the best option. For the sequencing problem, minimisation of milk waste would also give a minimised use of cleaning agents and use of water for cleaning during a product change in the sequence, depending on the techniques used for product changes. This paper concentrates on how adverse environmental effects may be counteracted at the dairy production unit level. Hence, it does not address all environmental implications of increased product diversity. At the dairy production level, product sequencing attracts attention, since the amount of waste generated by the sequence and the number of cleaning operations depends on the order selected. For cultured products the same equipment is used for several varieties. Consequently, for a wider variety of products more product changes are required in the sequence. During a change, action has to be taken to prevent mixing of the products involved. The order affects the amount of waste generated during a change, since the characteristics of the two products involved determine the technique used for the change. In turn, the technique dictates the volume of waste, water and cleaning agents. Therefore, the environmental impact of producing a set of products is closely linked to the production sequence. Although production scheduling has received much attention in the literature, few papers include environmental considerations. Various models, algorithms and optimisation tools have been tried for solving production scheduling problems in the food sector. A heuristic simulation model, based on a network representation, was devised for the bacon industry by Shah et al. [10]. Their purpose was to determine the most efficient production schedule by finding the best way of selecting the sequence of items processed in the same equipment when two or more menu items are competing for priority. Another heuristic simulation model based on operations research applied to a theoretical convenience food system was developed by Guley and Stinson [11]. Their purpose and heuristic rules are of the same kind as in the study by Shah et al. [10]. To enable improvement of production scheduling, from the perspective of fulfilling customer needs, a model based on finite capacity planning was used for low fat spreads and cider [12]. Alternative designs for milk powder production were analysed with both a process design tool and an economic analysis
one [13]. A similarity between most scheduling studies is that they use either a cost based criterion or a system performance one. Only two studies were found that took the environment into account. A methodology for incorporating ecological considerations into the optimisation of design and scheduling of batch processes was conducted by Stefanis et al. [14]; this included a case study of a cheese-making dairy. The optimizations were based on both process economics and environmental impact. Only two cheeses were included, consequently the relation between the production of a great number of products and the environment was not investigated in that study. The second study introduces minimisation of process-sequence dependent changeover waste (as an environmental issue) in product scheduling of a theoretical batch production unit [15]. Although the procedure optimizes an objective function that accounts for the amount of product changeover waste, it does not include any other explicit environmental parameters or categories. Studies of production scheduling that include environmental considerations are thus very limited, in the dairy industry as well as in other food industries. Therefore, we constructed a model that could assist dairy production units to schedule their production of multiple cultured products with the lowest possible environmental impact [16]. A heuristic solution was used to predict best possible product sequence from this point of view. To determine whether the heuristic solution actually gave the best possible sequence from an environmental perspective, an optimisation was also made. A life cycle perspective is required to obtain information about the total environmental impact of a sequence. Therefore, this detailed scheduling model was also included in a life cycle assessment (LCA) of the production schedule. The first option to decrease the sequence related waste and environmental impact is to sequence each day’s production in a waste minimised manner. Since most waste occurs at cleaning operations, these will be minimised at the same time. A second option is to also consider the frequency at which each product is produced, i.e. once a day, twice a week, etc. Dairy products are often produced on a daily basis or two to three times a week, despite their relatively long shelf life of up to 28 days. This is a result of market demands and probably also lack of insight of the environmental and economic consequences of high product frequency. The objective of this study is to make a thorough test of the previously constructed model [16] and to calculate the environmental impact associated with the waste minimised sequences. Since the number of products is crucial to the choice of solution used for the waste minimised sequences, the amount of waste generated and number of cleaning operations needed, case studies of two dairies were carried out. The first one handled a great number of products; to sequence them, the heuristic solution of the model was required. The second dairy handled fewer products, which gave us the opportunity to apply the optimising solution that was used to
J. Berlin, U. Sonesson / Journal of Cleaner Production 16 (2008) 483e498
verify the heuristic one. The contrasting dairies chosen facilitate a discussion about the generality of the result. The present study also investigates the role of the frequency of each product in a sequence. Scenarios with variations in the frequencies of the products were sequenced and analysed with LCA methods.
2. Method To sequence a given set of products for production can be done with alternative solutions. In an earlier study by Berlin et al. [16], a model was developed to find a waste minimised sequence and to calculate its environmental impact with LCA methods. The present study makes a thorough test of that model by using it for case studies at two dairies. Contrasting dairies were chosen for the case studies, to test the generality of the model. Dissimilarities between the case studies were: the characteristics, the volume and number of products processed in the same equipment. The two dairies also had different production conditions, which established the processing rules that had to be met. The dairies are hereafter referred to as Dairy A and Dairy B. This section describes why two solutions to the sequencing problem were chosen, the scenarios in general and the LCA method applied to assess the environmental impact of a production according to the schedule. Next, Section 3 on modelling describes the sequencing solutions found for each dairy and the specific process data used in the scenarios.
2.1. Sequencing of products Products cannot be made in just any order, as processing rules must be followed. These rules depend on the characteristics of the products and the conditions at a specific dairy. For each change of product in a sequence, waste is generated. There are numerous possible sequences that follow the processing rules for a set of products. We wanted to find a waste minimised solution for the sequences. To achieve this, two kinds of solutions were applied to sequence the products in the two dairies, a heuristic solution and an optimised one. Both solutions are described in detail by Berlin et al. [16]. The heuristic solution can handle sequences including a great number of products. The optimised solution has a numerical limit, 21, for sequencing. The limitation depends on the type of problem. To find the optimal product sequencing solution involves searching through all possible combinations of the manufacturing order of a set of products. The number of solutions for a given schedule is as many as N!, where N is the total number of products in the sequence. Dairy A was producing 34 products, hence the heuristic solution of the sequence was chosen. For Dairy B the optimised solution could be used, as only 16 products were involved. Both solutions for the sequencing model are briefly described in Section 3.
485
2.2. Scenarios of product frequencies Case studies were carried out for different technical scenarios. Three scenarios for each dairy, with varying frequencies of the products, were devised for the sequencing. The time span used was one week of production. To vary the product frequency, i.e. how often a given product is made weekly, we gathered information on its contribution to the amount of waste arising from the sequences. We also deepened our understanding of the environmental improvement potential of changing the product frequency. The frequencies of the products chosen for the scenarios were: the frequency currently used (two to five times per week) for each product; 2 times per week for each product; and, in the last scenario, one to two times per week for each one according to their shelf life. The scenarios were designated: Reference, Goal and Future. They were named for the probability of the dairy’s capacity to change their production rules. The Reference scenario reflected current situation. The Goal scenario was assumed to be achievable for most dairies within a reasonable time. The Future scenario was assumed to be attainable in the future; however, apart from changes at the dairy, the retail management would also be involved. The scenarios are further described in Section 3.
2.3. Life cycle assessment of the production according to a sequence To assess the life cycle environmental impact of the production according to the sequences for each scenario, the methodology of life cycle assessment was used. Berlin et al. [16] had already successfully performed an LCA in combination with the sequencing model, and the same approach was used in this study. The sequencing model finds the waste minimised sequence for each scenario. This sequence includes a list of products and the techniques used for their changes, see Appendix. To make a production according to the sequence causes impact on the environment as each activity affected by the sequence gives rise to emissions to water and air and uses natural resources for extraction and processing of energy and ingredients. The activities affected are the system under study and is illustrated in Fig. 1. They are also further described in Section 2.3.1. The functional unit is the volume of products produced in the sequence on a weekly basis. The functional unit for Dairy A was 219 000 kg and for Dairy B it was 182 000 kg. The goals for the data quality were: average data that reflect the types of technologies used, originate from the right geographic area (Sweden) and are not too old. For some data it was impossible to fulfil the requirements, then the time scale was widened and the geographic area was broaden to Norway and Europe. As can be seen in Table 1 the age of data varies but the geographic origin is acceptable. Even so, the data age of some activities is rather old, although the data of the most important activity from an environmental perspective, Milk production/Dairy farming, are recent. Therefore, the data age in a total was
J. Berlin, U. Sonesson / Journal of Cleaner Production 16 (2008) 483e498
486
Heat and Electricity Production
Milk Production
Water Treatment
Trp Dairy
Sewage Treatment
Cleaning agent Production Trp Trp
Product Treatment
Product Packaging
Fodder
Product
Fig. 1. The production schedule in a life cycle perspective. All affected activities were modelled (Trp e transportation).
considered to be good enough for the purpose of the study. The environmental impact was calculated according to life cycle inventory (LCI) methodology [17e19]. The environmental impact categories considered were: eutrophication, acidification, global warming (100 years) and photochemical ozone creation potential (POCP). The categories specified include the key parameters for the environmental impact of food production identified by Mattsson [20]. The key inventory parameters are nitrous oxide, methane, ammonia and energyrelated emissions. The contribution to selected environmental categories was calculated according to life cycle impact assessment (LCIA) methodology [17,19,21,22]. The inventory data sources as well as sources for the used environmental impact categories equivalency factors can be found in Berlin et al. [16]. 2.3.1. System boundaries Within the system under study all activities affected by the production were included, such as the choice of techniques selected for product changes, and the treatment of waste (Fig. 1). Products were changed just before the filling equipment. The filling of containers is designated product packaging in Fig. 1. Before the cultured product was packaged, it was treated for both hygienic and product purposes (product treatment in Fig. 1). The product treatment includes milk
Table 1 The age and geographic origin of data for the main activities Activity
Data age
Geographic origin
Milk production/dairy farming Dairy processing Water treatment Nitric acid production Sodium hydroxide production Sewage treatment
2003 1997 1993 1997
