Paper

lable at ScienceDirect

Food Control 86 (2018) 332e341

Contents lists avai

Food Control

journal homepage: www.elsevier.com/locate/foodcont

Temperature performance and food shelf-life accuracy in cold food supply chains e Insights from multiple field studies

M. G€oransson a, *, F. Nilsson a, Å. Jevinger b

a Packaging Logistics, Department of Design Sciences, Lund University, Box 118, 221 00 Lund, Sweden b Department of Computer Science and Media Technology, Malm€o University, Box 50500, 202 50 Malm€o, Sweden

a r t i c l e i n f o

Article history: Received 6 July 2017 Received in revised form 27 September 2017 Accepted 23 October 2017 Available online 1 November 2017

Keywords: Cold chain Dynamic shelf life Food waste Food quality Retail displays Temperature monitoring

* Corresponding author. E-mail addresses: [email protected] (M.

plog.lth.se (F. Nilsson), [email protected] (Å. Jeving

https://doi.org/10.1016/j.foodcont.2017.10.029 0956-7135/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

A challenge in perishable food industry today is variable and unknown food quality caused by different temperature conditions. This sometimes leads to unreliable printed shelf lives (best before dates) and food waste. Hence, temperature monitoring and control along cold food supply chains (FSCs) are essential for maintaining food quality and safety of perishable food products. This paper evaluates the temperature performance of cold food supply chains in relation to dynamically predicted shelf life and printed shelf life. Based on an in-depth study of actual temperature conditions of food products collected from field tests made in Swedish FSCs (from production to retail cold storage and retail displays), complete FSC scenarios were created. The results showed a significant difference in product shelf life between the most and least efficient FSCs, and between dynamically predicted and printed shelf life. Overall, the distribution from production to retail represents an efficient part of the FSC, in contrast to retail display storage. This study emphasizes the importance of a full-time temperature monitoring system to confirm food quality. A temperature monitoring system can be used to enable dynamic shelf life prediction, increase FSC transparency, and support food producers to proactively improve printed shelf lives.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

There are many challenges associated with providing food safely and with high quality to the market. This is especially true for perishable food products with short shelf lives that need temperature-controlled food supply chains (FSC), e.g. fresh fish and processed meat products (Aung & Chang, 2014). Quality, timeliness, and safety are central for all FSC actors. However, these demands can be hard to live up to as the margins in the food industry are low (Dani, 2015; G€obel, Langen, Blumenthal, Teitscheid, & Ritter, 2015) at the same time as there is an increased concern of food waste.

Food waste is one of the largest environmental burdens in to- day's society (Food and Agriculture Organization of the United Nations, 2013) with great economic and social implications. Ac- cording to Gustavsson, Cederberg, Sonesson, Van Otterdijk, and Meybeck (2011, pp. 1e38) one third of all edible food for human consumption is wasted globally and one of the UNs goals for

G€oransson), fredrik.nilsson@ er).

sustainable development is to decrease global food waste by 50% by 2030 (Report of the Secretary-General, 2016). A great deal of food waste can be related to poor temperature conditions and the lack of safe production (processing hygiene and initial status of raw in- gredients). Temperature has been pointed out as one of the most influential factors affecting the quality of chilled food (Jol, Kassianenko, Wszol, & Oggel, 2006). Studies report that mis- managed temperature in the logistics of perishable food can cause up to 35% of loss of products (Z€oller, Wachtel, Knapp, & Steinmetz, 2013) as well as problems with product returns, financial losses, increased operational costs, and relationship problems among supply chain actors (Beulens, Broens, Folstar, & Hofstede, 2005; Raab, Petersen, & Kreyenschmidt, 2011). Another identified cause of food waste is printed shelf life (best before date) (Fox & Fimeche, 2013). Lindbom, Gustavsson, and Sundstr€om (2014) estimate that over 50% of all food waste in the industry derives from expired best before dates. Furthermore, 2/3 of household food waste is still fit for consumption (Ventour, 2008). Recent research (DYNAHMAT, 2016; Jevinger, G€oransson, & Båth, 2014) concludes that an increased alignment between printed shelf life and actual product shelf life has the potential to reduce food waste in both FSCs and in

M. G€oransson et al. / Food Control 86 (2018) 332e341 333

households. Eriksson, Strid, and Hansson (2016) conclude that “there are great opportunities for reducing food waste by decreasing temperatures in supermarkets …” and highlight the possibility of extended shelf life from lower temperature storage to be a central factor.

A common praxis in FSCs is that the major part of a product's shelf life should remain when the product reaches retail. In Swe- den, the praxis states that for most of the perishable food products, 1/6 of the shelf life is devoted to the FSC from production, through wholesale and distribution (Jensen, Båth, & Lindberg, 2013). Hence, FSC actors are required to distribute products time-effectively with intact cold chains to avoid unnecessary food waste, retain food quality (Dani, 2015) and stay competitive.

Kuo and Chen (2010) and Hafliðason, �Olafsd�ottir, Bogason, and Stef�ansson (2012) stress the advantages of monitoring tempera- ture throughout the entire cold FSC, to ensure food quality in the supply chain. They suggest a combination of FSC temperature monitoring and food shelf life prediction models. Today, tempera- ture monitoring is common inside vehicles and warehouses; however, temperature sensors are often placed close to the cooling unit, which means that product temperatures cannot be obtained (Grunow & Piramuthu, 2013). G€oransson, Jevinger, and Nilsson (2018) states that accurate product temperatures can be obtained by monitoring temperature at a pallet or preferably a secondary packaging level. Furthermore, temperature monitoring in FSCs enables transparency, which Beulens et al. (2005) conclude to be a critical factor for both food safety and consumer trust.

In this paper, we address the challenges of food waste and quality control by evaluating temperature performance in a num- ber of cold FSCs in relation to the product shelf life of perishable products. Product shelf life was divided into two categories; printed shelf life and dynamically predicted shelf life (DPSL), for further performance evaluation. Based on an in-depth study of actual temperature data of food products collected from field tests made in Swedish FSCs (from production to retail cold storage and retail displays), different scenarios were created. The scenarios represent the most and least efficient complete FSCs (from production to consumer purchase) and illustrates food quality and shelf life dif- ferences within the same food segment; information that rarely reach FSC managers and customers. With insights gained from these scenarios, this paper provides recommendations for, and the further development of, food waste mitigation and food quality management in FSCs.

2. Cold food supply chain management and monitoring

A number of factors such as legislation, industry praxis, product characteristics, supply chain setups, global food trade and re- lationships between FSC actors influence the management of cold FSCs. Quality control of FSC operations is vital and FSC actors are required to assure storage and handling temperatures, e.g. take temperature samples of received goods and monitor cooling units. While temperature monitoring inside vehicles and warehouses is common, monitoring on pallet level or even closer to the products (i.e., primary or secondary packaging) is not common (Grunow & Piramuthu, 2013). Moureh and Flick (2004) report temperature differences of up to 12 �C inside trucks. Measuring only surround- ing temperature rather than close to the actual products may cause entire batches of food to be falsely rejected, or accepted, based on temperatures affecting only a part of the load. Raab et al. (2008) reported that temperature mappings within a poultry supply chain showed temperature fluctuations between �5 �C and 15 �C within a vehicle at different locations during unloading, while the

actual product temperature only changed slightly. Additionally, pickup and delivery activities, especially during summertime, have been reported to result in broken cold chains (Carullo, Corbellini, Parvis, & Vallan, 2009). In summary, the lack of accurate temper- ature data related to the food products, showing whether the temperature has been correct or abused, causes unnecessary food waste since FSC actors do not want to risk customer confidence and health.

Nonetheless, while the movement of food products influences the product shelf life, the storage and handling of perishables in retail outlets have a great impact on the quality and safety of chilled food products, as 5/6 of the shelf life is devoted to retail displays and households (Claro, Neto, & de Oliveira Claro, 2013). Derens- Bertheau, Osswald, Laguerre, and Alvarez (2015) and Derens, Palagos, and Guilpart (2006) are examples of the few found studies that have examined the product temperature from pro- duction until storage at home. Derens-Bertheau et al. (2015) conclude that the transportation after purchase and the house- hold refrigerators are the most sensitive links related to tempera- ture, followed by the retail displays. Lund�en et al. (2014) report, from a study of retail displays in 32 retail outlets in Finland, that temperature violations were observed for 50% of the products they monitored (where temperatures exceeded the maximum recom- mended storage temperature by 3 �C, or for more than 30 min exceeded the recommendation by 1 �C). Willocx, Hendrick, and Tobback (1994) found similar results, where temperature varia- tions and abuse were reported for 30% of the products studied in Belgian refrigerated retail display cases. In another study, Kou, Luo, Ingram, Yan, and Jurick (2015) examined retail displays for fresh- cut leafy green vegetables. They conclude that “The effect of ambient temperatures and the relatively large temperature varia- tion between samples located on the front rows and those at the back rows appear to be the major technical challenges hindering the compliance of FDA Food Code without freezing the products” (Kou et al., 2015 p.691). Another study also examined different positions of refrigerated retail displays; it reported that 97% of the high temperature abuse was located on the front row (Evans, Scarcelli, & Swain, 2007).

There has been a growing body of literature addressing the problems of cold chains, e.g., Hafliðason et al. (2012); Kuo and Chen (2010); and Abad et al. (2009). Several of the articles report different technological solutions using RFID tags (Grunow & Piramuthu, 2013; Ruiz-Garcia & Lunadei, 2011), wireless sensor networks (Carullo et al., 2009) and new concepts such as dynamic shelf life (G€oransson & Nilsson, 2013; Jevinger et al., 2014; Tromp, Rijgersberg, Pereira da Silva, & Bartels, 2012; Wang & Yue, 2017). A common conclusion in the literature is that continuous temper- ature monitoring provides potential benefits for supply chain management with improved quality control, transparency, and less waste. Hsiao and Huang (2016, p.187) add that “Interorganizational time-temperature sharing could enhance food safety and quality, and further enhance the competitive advantage of food supply chains as a whole.” However, inter-organizational information ex- change of this data together with other product characteristics are not being applied effectively in FSC management (Eden et al., 2010, pp. 115e129). This is mainly due to managerial challenges, as in- formation sharing is primarily a matter of trust and not technology or data, as sophisticated technological solutions already exist (Giguere & Householder, 2012; Raab et al., 2011). Nonetheless, both the FSC and sustainability literature emphasize alignment, coop- eration, and information sharing as essential in order to improve overall efficiency, minimize waste, increase food quality, and improve FSC sustainability factors (Aung & Chang, 2014; Dani,

M. G€oransson et al. / Food Control 86 (2018) 332e341334

2015). One field of information used in cold FSC is the printed shelf life.

Printed shelf life is a term that includes a number of sub defi- nitions, used by the food industry, to indicate quality and give guidance to the customer and consumer, e.g., best before date, expiry date, use by date (Hasselstr€om & Wallentin, 2015). Printed shelf life is also used for FSC planning and control; together with product-specific information, it is used in FSC distribution and for traceability (Hasselstr€om & Wallentin, 2015). Food producers state the printed shelf life of their products, but have no control over how the products are treated downstream the FSC (Jevinger et al., 2014). Consequently, the printed shelf life is based on the maximum recommended storage temperature, e.g., þ4 �C for fresh fish and þ8 �C for smoked ham in Sweden (Swedish National Food Agency, 2002). Printed shelf life is argued to be one of the major reasons for food waste both in the supply chain and ultimately in households (Lyndhurst, 2011; Rahelu, 2009). Perishables distrib- uted in a well-functioning FSC, that keeps handling and storage temperatures on or below the recommended temperature, have in most cases a longer shelf life than the printed shelf life states (G€oransson & Nilsson, 2013). This causes FSC actors to throw away perfectly good food in each stage of the cold chain, as do consumers who strictly follow the printed shelf life (Ventour, 2008). Moreover, an extended shelf life for perishable products has been proven to reduce food waste at the retailer's premises. In a Nordic research project (Møller et al., 2016), the shelf life of minced meat was prolonged from 4 to 8 days using modified atmosphere packaging (MAP), and implemented in 629 retail outlets of a large retail chain in Norway. The results show a reduction of waste in the retail outlets from 6 to 11.5% food waste of minced meat down to 1.8e3% with the prolonged shelf life.

3. Methodology

This paper studies cold FSCs of two products: 200 g of sliced smoked ham and 400 g of fresh cod, both packed in modified at- mosphere (MAP). The studied FSCs include food producers, logistics service providers, wholesalers, distributors, and retailers, all located in the southern and middle parts of Sweden. In total, 25 field tests were carried out between 2014 and 2016 to determine temperature along the FSCs. The field tests were focussed on two different parts of the FSCs (see Fig. 1): from food production to cold storage in retail (Test 1: Smoked ham and Test 2: Fresh cod), and retail display storage (Test 3: Smoked ham and Test 4: Fresh cod). In the first part of the FSCs (Test 1 and Test 2), time and temperature sensors were placed inside the secondary packages (in this case, plastic returnable crates) containing the food products. In order to get temperature data as close to product temperatures as possible, all sensors were placed in direct contact with the primary packages. For validity reasons, some of the sensors partly followed the same FSC. Furthermore, some of the sensors following the same FSC were placed in different secondary packages on the same pallet. This enabled a comparison analysis of temperature variations and associated product shelf lives within a pallet. In the second part of the FSCs (Test 3 and Test 4), sensors were placed at the innermost

Fig. 1. Supply chain scope of

(close to the cooling unit) and outermost positions in retail displays to catch temperature variations within each display.

3.1. Field tests: production to retail cold storage

Test 1 was performed in ham FSCs (Field tests 1.1e1.8) and Test 2 in cod FSCs (Field tests 2.1e2.8) from production until retail storage. Field tests 1.1e1.5 were conducted at the same occasion with sen- sors placed in different secondary packages on the same pallet from production to wholesale. At the wholesaler, the pallet was split and reassembled for distribution to five different retail stores. Field tests 1.6e1.8 were conducted in separate FSCs at different occasions. Field tests 2.1e2.2 were conducted in the same FSC from produc- tion to retail cold storage, but the sensors were placed in different secondary packages on the same pallet (in the middle and upper corner of the pallet). Field tests 2.3e2.6 were conducted in the same way as Field tests 1.1e1.5 i.e. the same flow from production to wholesale and thereafter split for distribution to different retail stores with sensors in each secondary package. Field tests 2.7e2.8 were conducted in separate FSCs at different occasions.

All sensors were placed inside the secondary packages directly after the food products were packed, and followed the FSC until the primary packages were placed in retail displays.

3.2. Field test: retail display

Test 3 was performed in a ham retail display (Field tests 3.1e3.5) and Test 4 was performed in cod retail displays (Field tests 4.1e4.4). All temperature sensors included in Test 3 were placed in retail displays with an open front to the customer. During nighttime, the displays were covered with curtains to decrease energy consump- tion and create a uniform temperature in the refrigerators. Field tests 3.1e3.2 were conducted at different locations in the same retail display (Fig. 2). These temperature sensors were placed in the front of the display, directly behind the price tags, in order to monitor the temperatures of the outermost food products. Field tests 3.3e3.4 were performed in the same manner as Field test 3.1e3.2, but in a different retail store. The same applies to Field test 3.5, however, the sensor of Test 3.5 was placed in the back of the refrigeration retail display (close to the cooling unit) in order to monitor the temperature of the innermost food products.

Field tests 4.1e4.2 were conducted in a retail display with another design than the above described. This display had a compartment close to the floor, similar to a smaller open refriger- ation box. The sensor of Field test 4.1 was placed at the bottom of the refrigeration box, close to the back wall; and the sensor of Field test 4.2 was placed on top of the outermost food package, locating the sensor in the middle of the box. Field tests 4.3e4.4 were con- ducted in the same type of retail display as used in Test 3 (shown in Fig. 2). The sensors were placed in the front, directly behind the price tags in the same retail display (similar to how Field tests 3.1e3.2 were performed).

the studies in this paper.

Fig. 2. Example of retail displays and placement of temperature sensors in Test 3.

M. G€oransson et al. / Food Control 86 (2018) 332e341 335

3.3. Shelf-life prediction models

The temperature data from the field tests were used in product specific microbiological prediction models to predict the actual food quality and shelf life (i.e. DPSL) at any given time in the FSC. The prediction models were also used to establish a static shelf life (SSL) for the respective food product. The SSL is constructed under the assumption that the food products are stored in their maximum recommended storage temperature in Sweden, i.e. 8 �C for sliced smoked ham and 4 �C fresh cod. The SSL differs from the printed shelf life as the food producers often reduce the printed shelf life with one or several days to gain a safety margin and assure good food quality to consumers. The SSL is more compatible with DPSL than printed shelf life, as they are constructed from the same model without safety margins. Therefore, in this paper, SSL is used as a tool to compare and evaluating the DPSLs gained from the field test data.

Previous studies within Dynahmat (DYNAHMAT, 2016) have tested the suitability and validity of several microbiological

1 MAP sliced smoked ham (aw ¼ 0.977e0.983, NaCl ¼ 2.0%, pH ¼ 6.26e8.48). Storage studies at 8 �C were performed with product samples taken directly from the production line just before the packing process. Moreover, the initial bacterial load of lactic acid bacteria varied between <1 log cfu/g and 1.88 log cfu/g, with a mean value of 1.1 log cfu/g. Mataragas et al. (2006) initial bacterial load of 1.21 log cfu/g, was chosen as initial bacterial load in the prediction model applied in this paper due to the wide spread of the results and the mean value proximity 1.21 log cfu/g (for further details see G€oransson et al. (2018)). The maximum bacterial load of lactic acid bacteria, identified by Mataragas et al. (2006), was also chosen. They found the threshold level of 8,62 log CFU/g in processed ham to be considered as the end of its shelf life (DYNAHMAT, 2016).

2 MAP fresh cod. Storage studies at 8 �C were performed with product samples taken directly from the production line just before the packing process. Moreover, no or very low levels of Photobacterium phosphoreum were found in the initial bacterial load samples (<1 cfu/25 g) (DYNAHMAT, 2016). These results is in line with Dalgaard, Mejlholm, Christiansen, and Huss (1997). The initial concentration of P. phosphoreum applied in this paper is, in accordance to the maximum initial concentration found in the storage studies, �1.4 log cfu/g (¼1 cfu/25 g). End of shelf life for MAP fresh cod (50% CO2 and 50% N2) was evaluated to 7.9 ± 0.4 log cfu/g, corresponding to Dalgaard, Mejlholm, & Huss (1997).

prediction models for the MAP sliced smoked ham and the MAP fresh cod used in the studied field tests. The prediction models by Mataragas, Drosinos, Vaidanis, and Metaxopoulos (2006),1 of MAP smoked ham, and Dalgaard, Mejlholm, & Huss (1997),2 of MAP fresh cod, were chosen for further use in the Dynahmat project, as well as in this paper, as they had the best bias and accuracy factor in the storage studies3 (DYNAHMAT, 2016).

Mataragas et al. (2006) base their prediction model on the growth of lactic acid bacteria, which are the specific spoilage or- ganisms in MAP smoked ham. The printed shelf life for the smoked ham, set by the food producer, is 25 days; and the SSL is in this paper established, using Mataragas et al. (2006) prediction model, to 26.2 days.

Dalgaard, Mejlholm, & Huss (1997) base their prediction model on the growth of Photobacterium phosphoreum, which is the specific spoilage organism in MAP fresh cod. The cod has a printed shelf life of 8 days set by the food producer; and a SSL of 7.8 days established using Dalgaard, Mejlholm, & Huss (1997). The storage studies within DYNAHMAT (2016) showed that the Dalgaard model slightly over predicts the microbial growth resulting in diminu- tively shorter shelf lives produced by the model.

During the field tests, Bluetooth Low Energy sensors (nRF51822, Nordic Semiconductor) and RFID sensors (RT0005, CAEN RFID) were used for measuring the temperature in the FSCs (DYNAHMAT, 2016). The temperature measurements were continuously broad- casted by the sensors, and collected by Sony Xperia mobile phones every 10 min. The mobile phones transmitted the sensor data, including time stamps, to a web server where it was stored in a relational database. Actors involved in the project could in real- time follow the temperature data and the corresponding dynami- cally predicted shelf lives, through a website. The data could then be retrieved via a PHP-based application, and imported to Excel. The calibrations of all sensors were controlled in a thermal incubator.

3.4. Field test analysis and food supply chain scenarios

The time and temperature data from the field tests were used in the microbiological prediction growth models for ham and fish, which provided DPSLs and SSLs of the food at any given time during the test period. The DPSLs and SSLs at specific times are expressed as loss of DPSL and SSL from the start of the field test. The loss of DPSL and SSL during Tests 1 and 2 was determined at the end of retail cold storage. The loss of DPSL and SSL during Tests 3 and 4 was determined at the end of retail display storage.4 Test 3 was performed for 11.6e17.6 days. Multiple time durations for the end of retail display period were chosen to represent the possible var- iations in time a product is normally placed on a retail shelf before it is purchased by a consumer. The end of retail display period was determined at 1, 5 and 10 days for the ham products. The periods of time was decided after discussions with retail managers, which revealed that most sliced smoked ham products were sold within 10 days. Test 4 was performed for approximately 5 days and the end of retail display period was determined at 1, 3 and 5 days for the cod products. Due to the cod's short shelf life, it is normally sold until

3 The two prediction models have been modified to include the thermal inertia of the primary package and the headspace in the primary package. G€oransson et al. (2018) describe the modifications of the models in detail.

4 The initial microbial concentration of lactic acid bacteria in Test 3.1e3.5 was set to 2.39 log cfu/gram, which corresponds to the bacteria level after 1/6 of 25 days (maximum recommended time from production to retail) with a static temperature of 8 �C. The initial microbial concentration of P. phosphoreum in Test 4.1e4.4 was set to 0.35 log cfu/gram, which corresponds to the bacteria level after 1/6 of 8 days with a static temperature of 4 �C.

M. G€oransson et al. / Food Control 86 (2018) 332e341336

the end of printed shelf life. The results of the shelf-life predictions are presented in terms of

loss of DPSL, loss of SSL, and the difference between loss of DPSL and loss of SSL. The loss of DPSL emphasizes the quality of the food but also the efficiency of the FSC operations in terms of time and temperature management. Some of the FSCs included in the field tests were longer (distance) than others. Thus, it is not always possible to increase time efficiency to match the shorter FSCs. It is thereby also relevant to look at the difference between lost DPSL and lost SSL to reduce the influence of the time factor. However, to eliminate the time factor entirely, only FSCs with the same length (time) should be compared.

The field test data from production to retail cold storage was combined with field test data from the retail displays to create FSC scenarios (FSCS) of entire FSCs of ham (Tests 1 þ 3) and cod (Tests 2 þ 4). The most and least efficient field tests of each category (Tests 1e4) were chosen to create 4 FSCS for each food product. These FSCS represent the most efficient and the least efficient FSCS in terms of loss of DPSL. The FSCS visualize the effects the collected empirical data has on product shelf life throughout the entire FSC. The efficiency of the generated FSCSs were evaluated in terms of lost DPSL and the difference between DPSL and SSL.

4. Results and analysis

4.1. Production to retail cold storage

4.1.1. Smoked ham Fig. 3 shows a variation in temperature for Test 1 between �2 �C

and 10.2 �C, where the vast amount of the temperature data are well below the maximum recommended storage temperature for smoked ham (8 �C). Further, it can be seen that the initial tem- perature measurements are above the maximum recommended temperature. This is due to the environmental temperature in the packaging area of the producer (12 �C). However, the ham is sliced

Fig. 3. Measured temperatures obtained in Field tests 1.1e1.8. The static temperature of 8 �C is the maximum recommended storage temperature for smoked ham in the Swedish FSCs. The product temperature does not correspond to the measured tem- perature in the beginning (initial surrounding temperature), as product temperature at packing is 4 �C.

Table 1 Loss of shelf life of smoked ham in Test 1. Most and least efficient FSCs are marked.

Field test Distribution time [days]

Loss of DPSL [%]

1.1 1.8 1.7 1.2 1.9 2.0 1.3 1.8 1.4 1.4 1.9 1.6 1.5 2.0 2.6 1.6 0.8 1.6 1.7 1.1 0.2 1.8 0.9 1.4

at a temperature of 0 �C and reach the temperature of approxi- mately 4 �C at the time of packing. This temperature deviation is included in the prediction model to match the product tempera- ture. In addition, Field test 1.4 registered temperatures below zero during a few hours. The food products did not freeze; however, their organoleptic quality may have been affected.

Table 1 shows the loss of DPSL and SSL of smoked ham at the end of Field tests 1.1e1.8, as well as the difference between the loss of DPSL and loss of SSL. The loss of DPSL, from the initial 26.2 days to the remaining shelf life in the retail cold storage, is lowest for Field test 1.7, which makes this field test the most efficient in relation to time (short delivery time) and temperature (low temperatures) management. In contrast, Field test 1.5 is the least efficient, though it can be noted that it is still remarkably better than the limit of 1/6 (16.7%), from production to retail.

A comparison of the results within each category shows that Field test 1.4 was the most efficient field test, while Field test 1.6 was the least efficient field test (column 5). All field test in Test 1 had positive results in difference between lost DPSL and SSL, indicating that all field tests are relatively temperature efficient in relation to the maximum recommended storage temperature of 8 �C. Please note that the difference between lost DPSL and lost SSL reduces the influence of the time factor.

4.1.2. Fresh cod Fig. 4 shows a variation in temperature between �2 �C and

14.7 �C, for Test 2. Notably, the temperature data in Field tests 2.3e2.6 exceeded the maximum recommended storage tempera- ture for fresh cod (4 �C) over long periods. Furthermore, the products in Field tests 2.3e2.6 were below zero for a short period of time. The reason for this is that the products were placed in a freezer before they reached wholesale, in order to reduce food quality loss and pass the temperature sampling inspection at the

Loss of SSL [%]

Difference, lost DPSL and SSL [days] [%]

6.7 1.3 (5.0%) 7.3 1.4 (5.3%) 6.7 1.4 (5.3%) 7.3 1.5 (5.8%) 7.5 1.3 (5.0%) 3.2 0.4 (1.6%) 4.1 1.0 (3.9%) 3.6 0.6 (2.1%)

Fig. 4. Measured temperatures obtained in Field tests 2.1e2.8. The static temperature of 4 �C is the maximum recommended storage temperature for fresh cod in Swedish FSCs. The product temperature does not correspond to the measured temperature in the beginning (initial surrounding temperature), as product temperature at packing is 4 �C.

Fig. 5. Measured temperatures obtained in Field tests 3.1e3.5. The static temperature of 8 �C is the maximum recommended storage temperature for smoked ham in Swedish FSCs.

M. G€oransson et al. / Food Control 86 (2018) 332e341 337

wholesaler. Had the wholesaler known the full temperature history, the products would probably not have been accepted. The products in Field tests 2.1-2-2 were also exposed to below-zero tempera- tures, see Fig. 4. These products were placed outside, on the re- tailer's loading bay during a cold winter's day.

Table 2 shows the loss of DPSL and SSL, as well as the difference between these, of fresh cod at the end of Field tests 2.1e2.8. The loss of DPSL from the initial 7.8 days is lowest for Field test 2.2, where 11.8% was lost. Field test 2.2 is therefore the most efficient in rela- tion to time and temperature management, whereas Field test 2.6 is the least efficient, losing almost 1/3 (30.3%). This is well beyond the limit of 1/6 (16.7%), from production to retail. Due to the short distribution time and shelf life, small deviations in time or tem- perature had a substantial impact on the results of lost shelf life. Only Field tests 2.1 and 2.2 have lower loss of DPSL than 16.7%. These results indicate poor time management for some of the ac- tors in the FSC. It is important to note that retail cold storage, before retail display, is not included in the 1/6 number. However, our field tests include data from when the products were placed in a cold storage for a few hours. This had only negligible effect on the results though.

Table 2 shows that Field test 2.8 had the highest efficiency (0.3 days) in terms of difference between loss of DPSL and SSL; however, it was slightly longer than the other field tests. Field test 2.6 was the least efficient field test with a difference between lost DPSL and SSL of �0.2 days. The negative figure implies increased FSC tempera- tures and too long lead times, which can have severe consequences on food safety, especially for highly perishable food products. This means that the printed shelf life may be violated (depending on the safety margin). The results presented in Fig. 4 and Table 2 clearly indicate inappropriate FSC handling in Field tests 2.3e2.6. Conse- quently, the involved companies made necessary operational and management changes after having received these results. Further- more, the results show small, but tangible, differences between the field tests included in the same FSC, due to final distribution to different retail outlets (Field tests 2.1e2.2 and Field tests 2.3e2.6). This can have cumulative effects depending on the time and tem- perature management in the rest of the FSC and in households.

4.2. Retail display

4.2.1. Smoked ham Fig. 5 shows a variation in temperature between 3.2 �C and

14.2 �C, for Test 3. The temperature data collected in Field test 3.5, representing the innermost temperature in the retail display, is well below the recommended maximum storage temperature for smoked ham. Of the sensors placed at the front of the retail display, Field test 3.1 and Field test 3.4 show temperatures that are mainly below the maximum recommended storage temperature. The sensors of Field test 3.1 and Field test 3.2 are located fairly close to each other, in the same retail display. However, the collected

Table 2 Loss of shelf life of MAP fresh cod in Test 2. Most and least efficient FSCs are marked.

Field test Distribution time [days]

Loss of DPSL [%]

2.1 0.9 12.4 2.2 0.9 11.8 2.3 2.2 29.2 2.4 2.2 24.8 2.5 2.2 29.6 2.6 2.2 30.3 2.7 1.8 23.4 2.8 2.5 28.9

temperature data differs by several degrees. This temperature dif- ference can be explained by the airflow pattern in the refrigerator. The sensor of Field test 3.1 is located at the same height as the cooling unit and is provided with a free forced airflow. The sensor of Field test 3.2 is placed slightly below the cooling unit, where the forced airflow is partly blocked. The same phenomenon seems to occur in the retail display where the sensors for Field tests 3.3e3.4 are placed. The results from Field test 3.2 and Field test 3.3 indicate poor design of the retail display causing insufficient airflow patterns.

Table 3 shows the most efficient and least efficient field tests in terms of loss of DPSL and difference between DPSL and SSL, in Test 3. Due to limited space, we have excluded the results from the other field test s in Test 3. Field test 3.5 represents the most temperature- efficient retail display included in the field tests. After 10 days, less than 20% of the DPSL was lost, which corresponds to a difference of 4.9 days. However, it is not common that food products have the same location in a fridge during their entire time in the retail display. Field test 3.3 is the least efficient field test, losing 54.8% of its DPSL after 10 days. Smoked ham products stored in these tem- perature conditions will have a shorter DPSL than the SSL. This reduced food quality also indicates a reduced food safety.

4.2.2. Fresh cod Fig. 6 shows a variation in temperature between 0.2 �C and

7.4 �C, for Test 4. The temperature data collected in Field tests 4.3e4.4, representing the products placed in the front of a retail display is below the maximum recommended storage temperature for fresh cod. Field tests 4.1e4.2 represent products in the inner and middle part of a retail display. Fig. 6 clearly shows how the cooling unit turns on and off causing temperature variations. Moreover, the retail display design used in the field tests seem to only obtain accurate product temperatures close to the cooling units, where the fridge's own temperature sensor is placed. This seems to be the case for many retail display designs used on the market today.

Loss of SSL [%]

Difference, lost DPSL and SSL [days] [%]

12.6 0.0 (0.2%) 12.6 0.1 (0.8%) 28.3 �0.1 (�0.8%) 28.3 0.2 (3.0%) 28.3 �0.1 (�1.2%) 28.3 �0.2 (�2.0%) 24.3 0.1 (1.0%) 32.3 0.3 (3.4%)

Table 3 Loss of DPSL of smoked ham in Field tests 3.3 and 3.5. Generated DPSLs and SSLs are presented after 1, 5 and 10 days of retail display. Due to limited space, only the most and least efficient field tests are presented.

Field test Time in retail display [days]

Loss of DPSL [%]

Loss of SSL [%]

Difference, lost DPSL and SSL [days] [%]

3.3 1 6.2 3.8 �0.6 (�2.4%) 5 29.4 19.1 �2.7 (�10.3%) 10 54.8 38.1 �4.4 (�16.7%)

3.5 1 1.9 3.8 0.5 (1.9%) 5 9.7 19.1 2.5 (18.7%) 10 19.4 38.1 4.9 (28.1%)

Fig. 6. Measured temperatures obtained in Field tests 4.1e4.4. The static temperature of 4 �C is the maximum recommended storage temperature for fresh cod in Swedish FSCs.

M. G€oransson et al. / Food Control 86 (2018) 332e341338

Table 4 shows the most efficient and least efficient field tests in terms of loss of DPSL and difference between DPSL and SSL, in Test 4. Field test 4.3 is the most temperature-efficient retail display; after 5 days, only 42.7% of the DPSL was lost, which corresponds to a positive difference of 25% and 1.7 days. Field test 4.1 is the least efficient field test, losing 67.9% of its DPSL after 5 days. Fresh cod products stored in these temperature conditions will have a slightly shorter DPSL than the SSL.

Table 4 Loss of DPSL of fresh cod in Field test 4.1 and Test 4.3. Generated DPSLs and SSLs are presen efficient field tests are presented.

Field test Time in retail display [days]

Loss of DPSL [%]

4.1 1 13.9 3 40.9 5 67.9

4.3 1 8.3 3 26.1 5 42.7

Table 5 Loss of shelf life of MAP smoked ham in different FSCs. Results from each scenario are p

FSC scenario Time FSCS [days]

Loss of DPSL [%]

FSC scenario 1 (Field tests 1.7 þ 3.5)

2.1 2.1 6.1 9.8 11.1 19.6

FSC scenario 2 (Field tests 1.5 þ 3.3)

3.0 8.6 7.0 31.7 12.0 57.1

FSC scenario 3 (Field tests 1.6 þ 3.3)

1.8 7.5 5.8 30.6 10.8 56.0

FSC scenario 4 (Field tests 1.4 þ 3.5)

2.9 3.5 6.9 11.3 11.9 21.0

4.3. Food supply chain scenarios

In order to obtain a complete picture of the entire FSC for ham and cod, i.e. from production until being picked by consumers in retail stores, the field tests were combined to generate FSC sce- narios. By combining the most efficient with the least efficient field tests, the effects on product quality could be visualized and eval- uated. The results are presented below, followed by a discussion of their implications.

4.3.1. Smoked ham Table 5 shows distinct differences between the most and least

efficient FSC scenarios illustrated in Fig. 7 (Field tests 1.7 þ 3.5 and Field tests 1.5 þ 3.3 in relation to loss of DPSL and Field tests 1.6 þ 3.3 and Field tests 1.4 þ 3.5 in relation to the difference be- tween loss of DPSL and SSL). The differences mainly derive from the major variations in the refrigeration capacity of retail displays with different technology and models.

Table 5 shows the generated loss of DPSL for FSC scenarios 1e4. The two highlighted losses of DPSLs, 2.1% after 2.1 days in FSC scenario 1 and 57.1% after 12.0 days in FSC scenario 2, show the greatest difference in the data collected, when combined into scenarios. The results imply that a retail outlet may promote food

ted after 1, 3 and 5 days of retail display. Due to limited space, only the most and least

Loss of SSL [%]

Difference, lost DPSL and SSL [days] [%]

12.8 �0.1 (�1.3%) 38.5 �0.2 (�2.9%) 64.1 �0.3 (�4.6%) 12.8 0.4 (5.4%) 38.5 1.0 (14.9%) 64.1 1.7 (25.7%)

resented at 1, 5 and 10 days of retail display.

Loss of SSL [%]

Difference, lost DPSL and SSL [days] [%]

7.9 1.5 (5.9%) 23.2 3.5 (13.3%) 42.2 5.9 (22.7%) 11.3 0.7 (2.7%) 26.6 �1.4 (�5.2%) 45.6 �3.0 (�11.5%) 7.0 �0.1 (�0.5%) 22.3 �2.2 (�8.4%) 41.3 �3.9 (�14.7%) 11.2 2.0 (7.7%) 26.4 4.0 (15.2%) 45.5 6.4 (24.5%)

Fig. 7. Generated scenarios of FSC temperature curves based on determined temper- atures in real FSCs from production to retail and in retail displays. The static temper- ature of 8 �C is the maximum recommended storage temperature for smoked ham in Swedish FSCs.

M. G€oransson et al. / Food Control 86 (2018) 332e341 339

products with up to 55% difference in lost shelf life, mainly depending on retail display time and the refrigeration quality of the retail display. Furthermore, the generated FSC scenarios of smoked ham resulted in a maximum difference between the values of the lost DPSL and SSL difference of 10.3 days (see FSC scenarios 3 and 4 at 10 days of retail display storage, in Table 5). The length of FSC scenario 3 and 4 also differs. This means that if the consumers select products based on the longest printed shelf life, they may get a product with much lower food quality (as in FSC scenario 3), than if they had chosen a product with a one-day shorter printed shelf life (as in FSC scenario 4). However, the generalizability of these conclusions is limited since the time factor plays a part in the result.

Fig. 8. Generated scenarios of FSC temperature curves based on determined temper- atures in real FSCs from production to retail and in retail displays. The static temper- ature of 4 �C is the maximum recommended storage temperature for fresh cod in Swedish FSCs.

Table 6 Loss of shelf life of MAP fresh cod in different FSCs. Results from each scenario are prese

FSC scenario Time FSCS [days]

Loss of DPSL [%]

FSC scenario 5 (Field tests 2.2 þ 4.3)

1.9 19.7 3.9 37.4 5.9 54.0

FSC scenario 6 (Field tests 2.8 þ 4.3)

3.5 36.8 5.5 54.6 7.5 71.2

FSC scenario 7 (Field tests 2.6 þ 4.1)

3.2 43.8 5.2 70.8 7.2 100.0

4.3.2. Fresh cod For fresh cod, Field tests 2.2 þ 4.3 and Field tests 2.6 þ 4.1

formed the most and least efficient scenarios in relation to loss of DPSL, while Filed tests 2.8 þ 4.3 and Field tests 2.6 þ 4.1 formed the most and least scenarios in relation to difference between DPSL and SSL. Fig. 8 shows that these scenarios have temperatures either mainly below or mainly above the recommended maximum stor- age temperature. As for smoked ham, temperature variations are mainly due to the major differences in quality of retail refrigeration.

In Table 6, the generated loss of DPSL for the above scenarios is shown. The two highlighted losses of DPSL, 19.7% after 1.9 days in FSC scenario 5 and 100% after 7.2 days in FSC scenario 7, express the greatest difference in the data collected, when combined into FSC scenarios. These scenarios imply that retail outlets may promote and sell food products with up to 80.3% difference in lost shelf life. These results indicate that the least efficient scenario for fresh cod does not measure up to required FSC operations standard. Conse- quently, consumers may be exposed to unsafe food products that may lead to food waste or food-borne illnesses. A comparison be- tween the results from scenario 6 and scenario 7, which are about the same length, 7.5 days, shows that the maximum difference between the values of the lost DPSL and SSL difference is 2.5 days. This difference is about 1/3 of the entire product shelf life. As a result, the improvement potential in the overall quality of the Swedish FSC of fresh cod and similar highly perishable products needs to be addressed.

5. Discussion

Managing cold FSCs under the conditions of the field tests and scenarios is challenging in a number of ways. Producers base their predictions on worst-case scenarios, whereas wholesalers and re- tailers base their product acceptance on momentary measurements of temperature and printed shelf life data, often causing unnec- essary waste.

In our studies, all actors involved have been able to monitor the actual temperature of products in real time. With accurate and reliable information of temperature history, a number of opportu- nities exist which are also reported in literature (Kuo & Chen, 2010). For instance, the combination of a system that can detect broken cold FSCs and collect data on actual temperature impact could enable producers to mark their products with more accurate printed shelf lives. Furthermore, the companies involved in our study have confirmed that access to actual temperature data would have positive effects on both waste reduction and food safety. The increased transparency among the FSC actors could also minimize unnecessary waste in the handover of products and provide, for most products, longer shelf lives due to well-performing FSCs. However, other researchers (e.g., Aung & Chang, 2014) have raised the challenges concerning the FSC actors' willingness to share data and information with suppliers and customers. The FSC is today

nted at 1, 3 and 5 days of retail display.

Loss of SSL [%]

Difference, lost DPSL and SSL [days] [%]

24.9 0.4 (5.3%) 50.6 1.0 (13.2%) 76.2 1.7 (22.2%) 44.7 0.6 (7.9%) 70.3 1.2 (15.8%) 96.0 1.9 (24.8%) 40.6 �0.3 (�3.2%) 66.2 �0.4 (�4.6%) 91.9 �0.6 (�8.1%)

M. G€oransson et al. / Food Control 86 (2018) 332e341340

suboptimal based on both the lack of sufficient temperature monitoring systems and the lack of shared information from such systems. If food waste and other inefficiencies in FSCs are to be effectively handled, new ways to collaborate and use information collectively are required. Finally, cold chain malpractices such as temperature abuse during parts of the FSC, leading to decreased food quality and unguaranteed food safety, are not discovered if the temperature is correct at the time of delivery or if it is not measured in transit points. As an example, the products corresponding to Field tests 2.3e2.6 would have been refused by wholesale and retail if continuous temperature monitoring had been applied. Hence, early warning systems could detect such malpractices and enable improvements for both each actor and the whole supply chains as well as new solutions for increased consumer confidence and safety.

There is a significant difference in how long a product is placed in a certain position in retail displays, e.g., at the front or at the back and the temperature it is exposed to. In discussions with retail staff, it is estimated that for volume products such as the smoked ham in our study, turnover time is rather short (5e10 days). During this time, products are moved from the back to the front in the displays. For low-volume products, e.g., organic alternatives, the turnover time is longer. In view of this, there are a number of factors to consider when determining the printed shelf life of a product, and today's cold chains do not cover them.

As a final point, while both in terms of timeliness and temper- ature levels, most of the FSCs studied are performing very well from production until retail storage, the temperature challenges are found in the retail displays, which is also confirmed by several other researchers (e.g. Lund�en et al., 2014; Willocx et al., 1994; Kou et al., 2015). Derens-Bertheau et al. (2015) report that 88% of the temperature measures in retail outlets are in the range of 0e4 �C for a sliced ham product. While these results are significantly better than our results, it still means that 12% of the products are exposed to higher temperatures than recommended. Consequently, our re- sults point out the need to include the retail display as many products are kept in the displays during a significant part of their shelf life, to get a complete picture of the factors and innovations that affect food waste and safety.

6. Conclusions

This paper sets out to evaluate the temperature performance of cold FSCs in relation to product shelf life. From the results, a number of conclusions can be drawn. The cold FSCs measured provide high performance levels of both time and temperature from production until retail storage. However, one under- performing cold FSC, from production to retail storage, was found. This emphasizes the importance of continuous monitoring to find temperature abuse that is not visible in today's praxis. While temperature abuse can occur in the FSC, we have found that the main challenges of cold FSCs are found in the retail outlets when products are placed and stored in retail displays. It can also be concluded that the retail display part of the cold FSC is missing from most previous FSC studies, which is remarkable, since food prod- ucts are placed in fluctuating temperatures for long periods in retail outlets.

In line with previous research, the results indicate a number of opportunities presented by more accurate and reliable monitoring of temperature in FSCs. These include systems to handle temper- ature deviations and alarms quickly and effectively, at the same time as producers can base their shelf life predictions on data ob- tained from their supply chains. Further research is encouraged in the area of dynamic shelf life in order to provide the food industry with guidance to reduce food waste and make cold chains more

efficient and safe. More studies are also needed that includes the retail stage of the cold chain i.e. from production until retail display.

Acknowledgements

This work was supported by the Swedish innovation Agency - Vinnova [grant number 2013-02803]. We would like to thank the involved personnel from the case companies.

References

Abad, E., Palacio, F., Nuin, M., Gonz�alez de Z�arate, A., Juarros, A., G�omez, J. M., et al. (2009). RFID smart tag for traceability and cold chain monitoring of Foods: Demonstration in an intercontinental fresh fish logistic chain. Journal of Food Engineering, 93(4), 394e399.

Aung, M., & Chang, Y. (2014). Temperature management for the quality assurance of a perishable food supply chain. Food Control, 40, 198e207.

Beulens, A. J. M., Broens, D. F., Folstar, P., & Hofstede, G. J. (2005). Food safety and transparency in food chains and networks, relationships and challenges. Food Control, 16(6), 481e486.

Carullo, A., Corbellini, S., Parvis, M., & Vallan, A. (2009). A wireless sensor network for cold-chain monitoring. IEEE Transactions on Instrumentation and Measure- ment, 58(5), 1405e1411.

Claro, P., Neto, S. A. L., & de Oliveira Claro, P. B. (2013). Sustainability drivers in food retail. Journal of Retail Consumer Service, 20(3), 365e371.

Dalgaard, P., Mejlholm, O., Christiansen, T. J., & Huss, H. (1997). Importance of Photobacterium phosphoreum in relation to spoilage of modified atmosphere- packed fish products. Letters in Applied Microbiology, 24(5), 373e378.

Dalgaard, P., Mejlholm, O., & Huss, H. (1997). Application of an iterative approach for development of a microbial model predicting the shelf-life of packed fish. In- ternational Journal of Food Microbiol, 38(2), 169e179.

Dani, S. (2015). Food supply chain management and logistics: From farm to fork. London: Kogean page.

Derens-Bertheau, E., Osswald, V., Laguerre, O., & Alvarez, G. (2015). Cold chain of chilled food in France. International Journal of Refrigeration, 52, 161e167.

Derens, E., Palagos, B., & Guilpart, J. (2006). The cold chain of chilled products under supervision in France. In 13th world congress of food science & technology 2006 (p. 823), 823.

DYNAHMAT. (2016). Report: Insights and learnings from the challenge-driven inno- vation project DynahMat. Web site http://dynahmat.com [2017-06-12].

Eden, M., Raab, V., Kreyenschmidt, J., Hafliðason, T., Olafsd�ottir, G., & Bogason, S. (2010). Continuous temperature monitoring along the chilled food supply chain. Food chain integrity: A holistic approach to food traceability, safety, quality and authenticity. Cambridge: Woodhead.

Eriksson, M., Strid, I., & Hansson, P. A. (2016). Food waste reduction in super- marketseNet costs and benefits of reduced storage temperature. Resources, Conservation and Recycling, 107, 73e81.

Evans, J. A., Scarcelli, S., & Swain, M. (2007). Temperature and energy performance of refrigerated retail display and commercial catering cabinets under test conditions. International Journal of Refrigeration, 30(3), 398e408.

Food and Agriculture Organization of the United Nations. (2013). Food wastage footprint: Impacts on natural resources: Summary report. FAO.

Fox, T., & Fimeche, C. (2013). Global food: Waste not, want not. London: Institute of Mechanical Engineers. Jan.

Giguere, M., & Householder, B. (2012). Supply chain visibility: More trust than technology. Supply chain Management Review, 20e25.

G€obel, C., Langen, N., Blumenthal, A., Teitscheid, P., & Ritter, G. (2015). Cutting food waste through cooperation along the food supply chain. Sustainability, 7, 1429e1445.

G€oransson, M., Jevinger, Å., & Nilsson, J. (2018). Shelf-life variations in pallet unit loads during perishable food supply chain distribution. Food Contol, 84, 552e560.

G€oransson, M., & Nilsson, F. (2013). The role of biosensors in future food supply chain. In Proceedings of the 25th conference of the nordic logistics research network (NOFOMA) (Gothenburg, Sweden).

Grunow, M., & Piramuthu, S. (2013). RFID in highly perishable food supply chains e remaining shelf life to supplant expiry date? International Journal of Production Economics, 146(2), 717e727.

Gustavsson, J., Cederberg, C., Sonesson, U., Van Otterdijk, R., & Meybeck, A. (2011). Global food losses and food waste. Rome: FAO.

Hafliðason, T., �Olafsd�ottir, G., Bogason, S., & Stef�ansson, G. (2012). Criteria for temperature alerts in cod supply chains. International Journal of Physical Dis- tribution & Logistics Management, 42(4), 355e371.

Hasselstr€om, L., & Wallentin, E. (2015). Minskat matavfallemilj€onytta och kost- nadsbesparingar. Naturvårdsverket. Retrieved Februari, 15, 2016.

Hsiao, H. I., & Huang, K.-L. (2016). Time-temperature transparency in the cold chain. Food Control, 64, 181e188.

Jensen, S., Båth, K., & Lindberg, U. (2013). Vilken effekt skulle s€ankt temperatur i kylkedjan få på matsvinnet? Rapport, 6596, 11e48.

Jevinger, Å., G€oransson, M., & Båth, K. (2014). A field test study on a dynamic shelf life service for perishables. In Proceedings of the 26th conference of the nordic

M. G€oransson et al. / Food Control 86 (2018) 332e341 341

logistics research network (NOFOMA) (Copenhagen, Denmark). Jol, S., Kassianenko, A., Wszol, K., & Oggel, J. (2006). Issues in time and temperature

abuse of refrigerated foods. Food Safety, 11(6), 30e35. Kou, L., Luo, Y., Ingram, D. T., Yan, S., & Jurick, W., II (2015). Open-refrigerated retail

display case temperature profile and its impact on product quality and micro- biota of stored baby spinach. Food Control, 47, 686e692.

Kuo, J.-C., & Chen, M.-C. (2010). Developing an advanced multi-temperature joint distribution system. Food Control, 21(4), 559e566.

Lindbom, I., Gustavsson, J., & Sundstr€om, B. (2014). Minskat svinn i livsme- delskedjaneett helhetsgrepp. SR866 Slutrapport, Sweden.

Lund�en, J., Vanhanen, V., Myllym€aki, T., Laamanen, E., Kotilainen, K., & Hemminki, K. (2014). Temperature control efficacy of retail refrigeration equipment. Food Control, 45, 109e114.

Lyndhurst, B. (2011). Consumer insight: Date labels and storage guidance. Banbury, UK: WRAP.

Mataragas, M., Drosinos, E. H., Vaidanis, A., & Metaxopoulos, I. (2006). Development of a predictive model for spoilage of cooked cured meat products and its vali- dation under constant and dynamic temperature storage conditions. Journal of Food Science, 71(6), 157e167.

Møller, H., Hagtvedt, T., Lødrup, N., Andersen, J. K., Madsen, P. L., Werge, M., et al. (2016). Food waste and date labelling: Issues affecting the durability. Nordic Council of Ministers.

Moureh, J., & Flick, D. (2004). Airflow pattern and temperature distribution in a typical refrigerated truck configuration loaded with pallets. International Jour- nal of Refrigeration, 27(5), 464e474.

Raab, V., Bruckner, S., Beierle, E., Kampmann, Y., Petersen, B., & Kreyenschmidt, J. (2008). Generic model for the prediction of remaining shelf life in support of cold chain management in pork and poultry supply chains. Journal on Chain and Network Science, 8(1), 59e73.

Raab, V., Petersen, B., & Kreyenschmidt, J. (2011). Temperature monitoring in meat supply chains. British Food Journal, 113(10), 1267e1289.

Rahelu, K. (2009). Date labeling on food. Nutrition Bulletin, 34(4), 388e390. Report of the Secretary-General. (2016). Progress towards the sustainable develop-

ment goals (Economic and environmental questions: Sustainable development No. E/2016/75*). New York: Economic and Social Council, United Nations.

Ruiz-Garcia, L., & Lunadei, L. (2011). The role of RFID in Agriculture: Applications, limitations and challenges. Computers and Electronics in Agriculture, 79(1), 42e50.

Swedish National Food Agency. (2002). Livsmedelsverkets f€orfattningssamling. LIVSFS 2002:51, x15 https://www.livsmedelsverket.se/globalassets/om-oss/ lagstiftning/nummerordninge-upphord-lagstiftning/2002/livsfs-2002-51- hantering.pdf.

Tromp, S.-O., Rijgersberg, H., Pereira da Silva, F., & Bartels, P. (2012). Retail benefits of dynamic expiry datesdsimulating opportunity losses due to product loss, discount policy and out of stock. International Journal of Production Economics, 139(1), 14e21.

Ventour, L. (2008). The food we waste. London, UK: K. WRAP. Wang, J., & Yue, H. (2017). Food safety pre-warning system based on data mining for

a sustainable food supply chain. Food Control, 73(B), 223e229. Willocx, F., Hendrick, M., & Tobback, P. (1994). A preliminary survey into the tem-

perature conditions and residence time distribution of minimally processed MAP vegetables in Belgian retail display cabinets. International Journal of Refrigeration, 17(7), 436e444.

Z€oller, S., Wachtel, M., Knapp, F., & Steinmetz, R. (2013). Going all the way e detecting and transmitting events with wireless sensor networks in logistics. In Eight IEEE workshop on practical issues in building sensor network applications (pp. 39e47).