The Carbon Footprint of Fruit Storage: A Case Study of the Energy and Emission Intensity of Cold Stores

Abstract

Despite their importance in all transportation chains, logistical sites—and in particular refrigerated facilities—are the weakest link in current emissions literature. This is largely due to a lack of quantitative research that focuses on these facilities. This article is the first of its kind to assess the emissions of eight refrigerated facilities that handle and store fresh fruit. In 2020, the analyzed facilities moved a total of 646,572 pallets of fresh fruit and emitted 32,225 t of CO₂e. Five of the largest facilities were responsible for handling 18.83% of the total fresh fruit exported from South Africa during 2020. The results revealed that storing and handling a pallet of fruit in a large-scale commercial cold store requires 7.62 kWh of electricity per day. Storing and handling fresh fruit is carbon intensive since each pallet stored translates to 7.52 kg CO₂e d⁻¹. However, other factors such as the seasonality and volume of fruit handled, facility characteristics and the availability of solar electricity systems, among others, all have a significant impact on the emissions value of the facility and on the emission intensity per pallet moved through the facility.

1. Introduction

International trade was at an all-time high of USD28.5 trillion for 2021—an increase of 13% compared to pre-pandemic levels [1]. The global economy has evidently recovered to a large extent after the COVID-19 pandemic. Alongside this recovery is an automatic increase in freight logistics since the exchange of goods still represented 78.2% of the total global trade (USD22.3 trillion) during 2021 [1]. Freight logistics has been and will always be a vital part of the world economy because all supply chains depend on the movement of goods [2]. If GDP growth remains coupled with international freight flow, the International Transport Forums (ITF) project that by 2050, freight transport alone will have grown by a factor of 2.6 to 345 trillion t-km compared to 2015 [3].

In 2019, the transport sector as a whole was responsible for 8.2 Gt of CO₂ or 27% of all greenhouse gas (GHG) emissions [4]. After the industrial and building sectors, the transportation industry is the sector responsible for most global emissions. This is confirmed by Wang and Ge [5] from the World Resource Institute (WRI), who estimate that the transport sector as a whole emitted 24% of all emissions during 2016. Focusing on freight logistics specifically, road freight, ocean freight, rail freight, air freight, and logistics sites collectively emit approximately 5 to 5.5% of the total global GHG emissions [6,7]. Other authors, such as Rüdiger et al. [8], estimate that up to 7% of Germany’s total GHG inventory for 2015 is due to freight logistics. Note the difference between freight transport (physical movement of cargo by various modes) and logistics, which includes both freight transport and activities at logistical sites such as warehouses, terminals, and ports. Higgins et al. [9] emphasize the importance and fundamental role of logistical sites in international, national and local transportation systems for the successful movement of goods. As international trade continues to grow, so does the demand for logistical facilities to handle, store or transship the increasing volume of goods [10]. In addition, the expected growth in the transport sector [3] and events such as the COVID-19 pandemic, which disrupted the global supply chain to such an extent that global transit times increased by approximately 25% [11], means that even more logistical sites will be used and required in future.

In terms of the emissions of logistical sites, the World Economic Forum (WEF) estimates that logistical buildings contribute 13% of the total freight sector’s emissions—equivalent to the air freight sector and the rail freight sector’s emissions combined [6]. The extent of emissions from logistical facilities is confirmed by numerous sources in the literature. Authors such as Ries et al. [12] estimate that up to 20% of all freight logistics emissions in the USA could be as a result of warehouses alone. Rüdiger et al. [13] state that emissions from warehousing and transshipment facilities could be a quarter of the total emissions of the freight logistics sector. Rüdiger et al. [8] also estimate that nearly 15% of emissions from freight are a direct result of activities at logistical sites.

Although the importance of analyzing the emissions of logistical sites is evident, only limited quantitative research has been done in the field. The most notable is the emission intensity factors (emission intensity factors state the amount of emissions emitted per logistical unit when performing a specific activity and are measured as kg CO₂e per logistical metric such as t-km, TEU-km, t-moved, pallet-d, etc.) developed by Dobers, Perotti and Fossa [14] for the Fraunhofer Institute for Material Flow and Logistics (IML), which estimate the carbon intensity of moving goods through different types of logistical sites. See Section 1.2 for an exposition of available emission intensity factors for logistical sites. These emission intensity factors [14] are also used and recommended by the Global Logistics Emissions Council (GLEC) Framework [2]. The scarcity of any form of emissions data for facilities or sites at which goods are handled, stored, or transshipped is confirmed by McKinnon [15]. Several other authors [2,12,14,16,17,18] stress the importance and the valuable contribution of future research that is required for logistical facilities in this regard.

Various factors such as the design, technology, equipment, internal operations, and type of commodity influence a logistical facility’s energy usage. Authors such as Lewczuk, Kłodawski, and Gepner [19] analyzed these factors to estimate a facility’s energy usage and emissions. This type of analysis requires a bottom-up approach whereby the impact of the individual factors is assessed to determine the facility’s total emissions. However, the article’s focus is not to investigate how factors influence the emissions but rather to increase the visibility of the scale of emissions at these facilities. For this, a top-down approach is preferable since it assures that emissions are not over- or underestimated. A top-down approach makes fewer assumptions and inevitably ensures that the actual energy consumption and associated emissions are apportioned to the goods that moved through the facility. This article is the first known top-down analysis to determine the carbon footprint of fresh fruit storage and handling in cold stores.

1.1. Research Aims

This article aims to assess and quantify the following items by using data from several refrigerated logistical facilities that handle and store fresh fruit as a commodity:

• Determine the total emissions of each analyzed facility (t CO₂e);
• Calculate the average electricity consumption rate (kWh pallet-d⁻¹) of moving and storing a pallet in a facility for a day;
• Determine an average emission intensity factor (kg CO₂e pallet-d⁻¹) for each facility to estimate the carbon intensity of fresh-fruit handling and storage at the facility.

1.2. Literature Review

1.2.1. Definition of Cold Stores

This article analyses warehouses and transshipment sites that handle two special classes of refrigerated goods, classified as fresh (4 °C to 7 °C) and sensitive (0 °C to 2 °C) [20]. The facilities analyzed in this article are purpose-built for the handling and storage of fresh fruit, but vegetables and dairy-related products are other examples of goods that fall in this “chilled” temperature range. However, the storage of fresh fruit is different from that of dairy and vegetables since some fresh fruit also requires a controlled atmosphere where the gaseous concentrations of oxygen, ozone, ethylene, water vapor, nitrogen, and carbon dioxide in the circulated air are regulated [21].

In the remainder of this article, warehouses and transshipment sites that specialize in the handling and storage of fresh fruit are referred to as cold stores. Du Plessis, Van Eeden and Goedhals-Gerber [22] refer to these cold stores as inland fruit facilities.

1.2.2. Calculating Emissions

The quantification of emissions is seemingly a straightforward process based on the actual or estimated amount of energy consumed by a facility. This energy consumption is then multiplied by an emission factor (emission factors are used to convert the amount of fuel and energy used to emissions and are measured as kg CO₂e per quantity such as ℓ, kWh, kg) to translate the energy consumed to emissions. In addition to energy, the effect of fugitive emissions is incorporated by assessing the refill values of refrigerants. The total emissions value (energy and refrigerants) is then apportioned among all the goods that have moved through a facility to estimate the emissions of one logistical unit. Refer to Section 3 for the equations required to quantify and apportion emissions.

The most suitable methodology to quantify emissions of a logistical facility is suggested by Dobers et al. [20]. This methodology aims to provide users with a generic step-by-step guide to assess and audit the carbon emissions of any type of logistical building for the purpose of calculating the emissions of a logistical chain. However, Dobers et al. [20] state that adjustments may be made to account for site-specific characteristics, such as the type and nature of the goods moved, to provide more meaningful results. The latter is, however, easier said than done, as explained in Section 2.1.

The methodology developed by Dobers et al. [20] is built on the emissions accounting principles of the internationally recognized GHG Protocol [23]. In addition, the methodology is endorsed by the GLEC Framework [2] and used as their methodology of choice for logistical sites. The analysis is aligned with a lifecycle approach to assess the total fuel lifecycle emissions, i.e., the well-to-wheel (WTW) emissions of all fuels. The assessment analyzes all relevant GHGs and then expresses the results in terms of carbon dioxide equivalent (CO₂e). Simply stated, CO₂e indicates the global-warming potential that a given substance has when compared to carbon dioxide.

1.2.3. The Allocation Problem and Dwell Days

The goods that move through a facility differ greatly due to differences in their weight and size, their varying refrigeration requirements and subsequent energy consumption, the time duration of storage, and handling requirements. This complicates how fuel, refrigerants, emissions or even costs are divided among goods. All the above-mentioned give rise to what is known as the allocation problem: how to ensure that emissions are apportioned in a fair and equitable manner among all goods.

In particular, the most important factor influencing the allocation in a refrigerated facility is the duration of storage. Subsequently, to apportion the emissions of a refrigerated facility fairly and accurately, the duration of storage, referred to as dwell days, must be accounted for. It is important to understand the effect dwell days have on the allocation of emissions. If the dwell days of a cold store were to be an average value of x days, it is assumed that all goods spend x days on average in the facility. When fuel, refrigerants or emissions are apportioned among the goods that have moved through the facility, the total values are divided by the product of dwell days and the number of goods moved through the facility (see Equations (3)–(5)). This means that all results in an assessment are a function of both the dwell days of goods in the facility and the throughput of the facility. The authors are of the opinion that this is the fairest and most accurate method of incorporating time as a variable in the assessment. This is contrary to current literature by authors such as Dobers et al. [20], Dobers and Perotti et al. [14], Iriarte et al. [24] and Smart Freight Centre [2], which does not account for dwell days at all.

1.2.4. Existing Emission Intensity Factors

The most comprehensive set of GHG emission intensity factors in literature was developed by Dobers and Perotti et al. [14] and entailed the assessment of 168 logistical sites in Europe. The analysis and results were conducted and reported on an annual basis and, therefore, disregard seasonality. All logistical sites were classified into three categories: facilities that transship ambient, chilled or mixed goods; facilities that store and transship ambient, chilled or mixed goods; and facilities that store ambient, chilled or mixed goods. It is interesting to note that the sample size of chilled facilities was a mere 18 sites in total and no reference was made to the type of refrigerated product, i.e., fresh, sensitive, pharmaceutical, or frozen. Furthermore, Dobers and Perotti et al. [14] mention that an average weight of 450 kg was assumed per pallet for ten facilities; however, they do not suggest a factor to convert between weight and pallets. The results of the assessment are shown in

Table 1. Emission intensity factors for logistical sites from literature.

Type of Logistical Site	Ambient (kg CO2e t−1)	Mixed (kg CO2e t−1)	Chilled (kg CO2e t−1)
Transshipment	3.4	3.8	11.1
Storage and transshipment	1.7	12.3	7.3
Warehouse (storage)	1.9	8.9	8.2

Adapted with permission from Dobers and Perotti et al. [14].

.

1.2.5. A South African Perspective

The growth in the global fresh-fruit export industry is unprecedented. Authors such as Du Plessis et al. [18] state that the industry will grow by 6% per annum on average. South Africa is no exception to this growth trajectory. The fresh-fruit export industry has grown by 11% from 2020 to 2021; this after a 10% growth during the previous year [25]. In addition, South Africa is the world’s second-largest citrus exporter and the leading exporter of fresh fruit by volume in the southern hemisphere [26]. This rapid expansion, coupled with the scale of the industry, undoubtedly creates challenges.

Another perspective unique to South Africa is the source and availability of fuels such as electricity, LNG, and diesel. No distribution pipelines for LNG or diesel to the cold stores exist in South Africa. Eskom Holdings SOC Ltd. is the only power utility in South Africa and depends on coal for electricity generation. According to Eskom [27], 90.4% of all the electricity in 2020 was generated by coal-fired plants by burning 108.6 million t of coal. This makes South Africa’s electricity the most carbon-intensive or “dirty” electricity in the world compared to peers [28]. Despite the environmental impact, coal will remain the primary source of energy in years to come due to the affordability compared to alternative energy sources, the economic contribution of the coal industry to the economy, and the role of employment [28].

Finally, South Africa is different from other countries due to the extreme vulnerability of the national electricity grid. Eskom’s generation capacity is often not sufficient to meet the electricity demand, which is then lowered by removing regions from the national grid on a rotational basis. This concept is referred to as “loadshedding” and is, unfortunately, a common phenomenon in South Africa. During loadshedding, businesses and some households use diesel generators to supply electricity. This requires a significant financial investment and also leads to increased operational costs since electricity generated by diesel generators is more expensive than grid electricity.

1.3. The Role and Function of Cold Stores

Apart from connecting different elements in a transport chain to balance the efficient flow of goods, cold stores also perform several other functions:

• Storage of goods for short durations (less than 24 h) or extended periods (several days or months);
• Acting as a buffer in the fresh-fruit supply chain to ensure a constant supply to markets;
• Providing a link between various transportation modes such as road, air, rail and deep-sea transport;
• Temperature control and/or re-cooling of fruit to the optimal storage temperature;
• Cold sterilization of fruit to kill any pests, microbes or fungi in or on the fruit;
• Allowing producers or production regions to consolidate fruit loads to use transport modes more efficiently;
• Providing a link between a change in the functional unit of transport, e.g., from pallets to refrigerated (reefer) containers;
• Providing an inspection site for fruit to ensure compliance with phytosanitary requirements in the country of import.

Note that no repacking of fruit (except after inspection of samples), order picking, sorting, relabeling, or customization is performed at cold stores. The pallets of fruit, therefore, do not undergo any type of transformation when moving through a cold store and can be seen as a unit for the purpose of handling and emissions calculation.

2. Research Methodology

This research article uses the Fraunhofer Institute’s Guide for greenhouse gas emissions accounting at logistical sites [20] as the base methodology. Section 2 deals with the key methodological aspects of the study. These include the general principles of assessment, which are discussed in Section 2.1. Section 2.2 explains how the base methodology was adjusted and which assumptions were made in the study. In Section 2.3 the boundaries of assessment are defined, while Section 2.4 elaborates on the data collection phase. Finally, Section 2.5 discusses how the collected data were assessed and prepared for calculation. The calculation phase is discussed in Section 3 while the analysis, interpretation and reporting of results are presented in Section 4 and Section 5. Figure 1 provides a visual representation of the methodology used in the study.

2.1. General Principles of Assessment

In terms of the sources of emissions, two types of energy are used in cold stores, namely electricity and diesel. No LNG is used at any of the facilities. Electricity is used by the refrigeration plant, general lighting, offices, etc., while the diesel is used for back-up generators and/or handling equipment at the facility. In addition, to comply with the standards and methodology set out by Dobers et al. [20] and the WBCSD and WRI [23], a third source of emissions is also included: the leakage of refrigerants. Since all the facilities are temperature-controlled, refrigerant leakage must be included in the assessment.

In terms of the various organizational scopes defined in the GHG Protocol: Corporate Accounting and Reporting Standard [23], the following cold-store emissions are analyzed in the assessment:

• Scope 1 emissions: The emissions due to the burning of fuels, i.e., tank-to-wheel (TTW) in assets owned or controlled by the reporting company. Also included in this scope are the emissions due to refrigerant leakage from equipment at the reporting company’s plant.
• Scope 2 emissions: The indirect emissions due to purchased electricity by the reporting company.
• Scope 3, category 3 emissions: The indirect upstream emissions due to the extraction, production, and transportation of fuel and energy-related products that are not included in Scope 1 or Scope 2 emissions. This includes the upstream emissions of purchased fuels, i.e., well-to-tank (WTT) and the upstream emissions of purchased electricity, taking transmission and distribution losses into account. For a detailed discussion of calculating a reporting company’s category 3 emissions, refer to the WBCSD and WRI’s Technical Guidance for Calculating Scope 3 Emissions [29].

The results of this article will state the total operational lifecycle emissions of a cold store.

2.2. Technicalities of Assessment

This section discusses the adjustments made to the Dobers et al. [20] methodology, as well as the assumptions made to perform the study.

2.2.1. Adjustments to Methodology

Dobers et al. [20] state that the guidance provided is generic and may be adjusted to account for both site- and product-specific characteristics. In light of the latter, the authors deemed it appropriate to (1) adjust the methodology in terms of the functional unit of analysis, (2) incorporate a time-based element in the analysis, and (3) account for emissions on a monthly and yearly basis. These three aspects are discussed below.

• In this assessment, the unit of analysis is pallets instead of the conventional tonne that is generally used in the emissions realm. According to Dobers et al. [20], the use of pallets as units of analysis is acceptable as long as these units are used throughout the assessment and allow for comparison of emissions over years. This means the resultant unit for emission intensity factors is kg CO₂e pallet⁻¹ instead of kg CO₂e t⁻¹. The reason for this adjustment is justified: all cold stores, fruit exporters, producers and stakeholders refer to pallets as the metric or functional unit after fruit has been packed and palletized and distribution commences. In addition, the pallet weight is often not captured by cold stores since this adds an added level of complexity to business operations and does not contribute any significant value. Using the consignment-based indicator (pallets) instead of the weight-based indicator (tonnes) still provides a consistent metric that is already in use. If the conversion between pallets and weight is required, the conversion factor of 1 030 kg pallet⁻¹ can be used.
• The time duration of storage or dwell days directly affects the utilization percentage of the facility (i.e., how well the available capacity of the cold store is used). This, in turn, affects the emissions allocated to each pallet that moves through the facility. It is deemed essential to incorporate the time duration of storage in the assessment to provide an accurate and true representation of both the electricity consumption rate and emission intensity of storing a pallet. The time aspect is incorporated by either collecting the average dwell days of pallets from cold stores or by obtaining the average utilization percentage of the cold store. If neither of the two variables is available, a time-based element cannot be incorporated into the assessment. Refer to Section 3, Equation (2) for the conversion between dwell days and average utilization.
• Dobers et al. [20] suggest that emissions should be accounted for on an annual basis to remove seasonal effects. The authors are, however, of the opinion that the seasonal effect of different harvest seasons for different fruits must be assessed. Different fruit types, each with different refrigeration requirements and throughput volumes, are moved through a cold store during different months of the year. In addition, the effect of ambient temperature during different months of the year is also incorporated by performing the analysis on a monthly basis. However, to comply with Dobers et al. [20], the assessment and results of some facilities are done for both a monthly and an annual period.

2.2.2. Assumptions Required for Assessment

To analyze any cold store at a site or facility level, the following ten assumptions are made:

• The weight difference between fruit types and packaging configurations is negligible. This means all pallets of fruit have the same weight.
• All pallets have the same number of dwell days in a facility, independent of the type of fruit and the destination market.
• The energy consumed due to cold sterilization is apportioned to all pallets that move through the facility. The effect of fruit destined for different markets is therefore negligible.
• Pallets in the facility are all handled or processed in a comparable fashion, from offloading to storage to the eventual loading for shipment to the destination market.
• The energy effect of some fruits arriving at optimal storage temperature and others arriving above target temperature is allocated evenly among all pallets that move through the cold store.
• All fruit types and packaging configurations have the same cooling characteristics.
• The type of packaging has little to no effect on the energy efficiency and use of the cold store.
• The effect of fruit loss in any of the analyzed cold stores is ignored.
• Only the operational emissions or use-phase emissions of equipment and infrastructure are assessed.
• Leakage of refrigerants occurs in the same period that refilling with refrigerants occurs. Furthermore, the quantity of refrigerants refilled during a year is apportioned evenly between all months of the year.

Intuitively, some of the ten assumptions may seem more realistic or appropriate than others. However, until more detailed data collection occurs at cold stores, these assumptions are required to perform the assessment for any cold store.

2.3. Boundaries of Analysis

The operational control approach prescribed by Dobers et al. [20] was used in this study. This means the following activities are included: inbound and outbound handling of pallets at the cold store; handling required for the storage and the retrieval for outbound transport; refrigeration of the fruit during storage or transshipment; internal yard logistics required during everyday operation; lighting of the facility; IT equipment used in the facility; and all office equipment and space used for administrative purposes.

At the analyzed cold stores, battery-powered and diesel forklifts are used as handling equipment to handle all pallets. Also included in the scope, as previously mentioned, were the fugitive emissions due to refrigerant leakage from the refrigeration plant or equipment. Subsequently, all GHG emissions that are emitted within the geographical location (the yard and the actual building) of the cold store and from equipment or infrastructure owned or controlled by the cold store were covered in this assessment.

Aligned with Dobers et al. [20], the following activities are deemed to be out of scope and were omitted from the assessment: the inbound and outbound transport to and from the facility by road transportation, employee commuting, business travel, and the manufacturing and dismantling of any asset (building or equipment).

2.4. Data Collection

The data required for this article have been classified into two groups, namely consumption data collected from the various cold stores and emission factors retrieved from the literature.

2.4.1. Cold-Store Consumption Data

The type of data collected is intrinsically linked to the goal of the assessment. As the desired emission intensity values become more refined, more comprehensive data capturing and collection are required. However, this is often not available due to the data not being captured for each activity category (order picking, storage, repacking, or other value-adding activities). The benefit of such an added level of complexity is, however, debatable. Nevertheless, since all pallets of fruit are deemed to be similar and processed in the same fashion, none of the above-mentioned activity categories applied to the analyzed cold stores. This means that an average emission intensity factor for the entire cold store can be calculated based on overall data values for the entire site. This high-level energy consumption data is almost always readily available—as was the case in this study—but is often difficult to obtain due to its sensitive nature.

In order to calculate an average emission intensity factor at site level, all the data listed in

Table 2. Consumption data collected from the various cold stores.

Electricity	Diesel	Refrigerant Data	Logistical Data
Total amount of grid electricity (kWh) used	Total amount of diesel (ℓ) used at the facility by generators and/or equipment	Types of refrigerants used at the facility	Pallet capacity of facility
Total amount of electricity generated (kWh) by solar panels		Refrigerant capacity of all systems (kg)	Total throughput at the facility (number of pallets)
		Total refill values of refrigerants (kg)	Average dwell days or utilization percentage of the facility

was collected for several years, as available (ranging from one to six years). It is essential that all the listed data values are collected. If a single data value is absent, it compromises the assessment as prescribed by Dobers et al. [20], which, in turn, affects the integrity of the results.

The data in Table 2 was collected from facilities via email after an initial data-collection meeting and data-sharing agreement had been put in place. This was preceded by a site visit to the cold stores and various meetings to discuss the data requirements. The data was supplied to the researchers either in Excel, PDF, or text format.

Due to the seasonal nature of fresh fruit, it was requested that the data values listed in Table 2 be provided per month. If the data were not available per month, the cold store supplied the values for every year.

2.4.2. Emission Factors (EF)

Following the identification of sources of emissions, suitable emission factors (EF) were selected to convert the amount of fuel used or refrigerants emitted to emissions volumes. For this, several reputable sources, as advised by Dobers et al. [20], Smart Freight Centre [2] and the WBCSD and WRI [23], were retrieved and used.

In terms of an EF for electricity, Eskom’s 2021 Integrated Report was used to provide the most recent EF. According to Eskom [27], a total of 205,635 GWh of electricity was sold during 2020 and 201,624,115 t of CO₂e were emitted in the same year. This resulted in an EF of 0.98 kg CO₂e kWh⁻¹ for electricity supplied to a South African consumer through the national electricity grid. This calculated value excludes losses due to transmission and distribution (technical losses), losses due to theft (non-technical losses), own internal usage by Eskom, and wheeling (electricity generated by small, independent power producers). As the calculated EF is slightly higher than the suggested 2017/18 value of 0.97 kg CO₂e kWh⁻¹ [28], the most recent and conservative value of 0.98 kg CO₂e kWh⁻¹ is used in this article.

No EF for South African diesel fuel could be found in the literature or obtained from local fuel suppliers. Subsequently, the EF for diesel fuel (WTW of 3.24 kg CO₂e ℓ⁻¹) was retrieved from the European Standard EN 16258 [30] and used in the assessment. This is according to the Smart Freight Centre’s [2] recommendation that, in the likely event that no country-specific value exists, the EF for European fuels should be used. Note that the value of 3.24 kg CO₂e ℓ⁻¹ is for all grades of diesel fuel since there is no differentiation for diesel between different fuel grades such as 500 or 50 parts per million (ppm).

To quantify the emissions of refrigerants, the global-warming potential of all refrigerants is assessed on a 100-year time horizon. The factors were retrieved from the Intergovernmental Panel for Climate Change’s 4th Assessment Report [31], since this report provides the most comprehensive list of EF. These are also the values used by the Smart Freight Centre [2] in the GLEC framework. (Although the 6th Assessment Report is available, it may not be quoted, cited, or distributed as of yet. The 5th Assessment Report does not provide a comprehensive list of all refrigerants; therefore, to avoid confusion, the 4th Assessment Report was used throughout this study.)

2.5. Data Assessment and Preparation Phase

After the data from the various cold stores were received, it was assessed for completeness and correctness. The data values were then captured and stored in a single Excel document. If any of the data values mentioned in Table 2 were missing or regarded as outliers, the representative at the cold store was contacted to provide the value or confirm that the specific data values were correct.

Due to different cold stores having different business practices, the data were not captured by all the cold stores in a consistent manner. The data preparation phase was therefore essential to eliminate inconsistencies and ensure that consistent data would be available for calculation purposes.

3. Calculations

To calculate (1) the total emissions of a cold store, (2) the rate of electricity consumption, and (3) the emission intensity of a pallet, the formulae listed below were used. Some formulae are based on Dobers et al. [20] but have been adjusted, as explained in Section 2.2. Calculations were performed on a monthly and a yearly basis, depending on the completeness of the data collected. The formulae used to calculate the results stated in Section 4 are listed in the order of use in the remainder of this section.

3.1. Total Site Emissions

To quantify the total site emissions (research aim 1) of a cold store for a given period, Equation (1) was used:

$$Emissio{n}_{Total}={{\displaystyle \sum }}_{i=1}^n\left(Quantit{y}_i·E{F}_i\right)$$

(1)

where:

• Emission_Total = the total GHG emissions of the cold store for a period (kg CO₂e)
• n = the total number of fuels (diesel, electricity or refrigerant) consumed
• Quantity_i = the amount (ℓ, kWh or kg) of i used
• i = the type of fuel (diesel, electricity) or refrigerant used
• EF_i = the emission factor for i (diesel, electricity or refrigerant) in (kg CO₂e per unit).

3.2. Dwell Days or Utilisation

A time-based allocation was preferred to fairly distribute emissions among pallets with short- and long-storage dwell times. To incorporate a time-based element in the analysis, either the dwell days or utilization data for the facility was collected. Since the dwell days are a variable in the remainder of the calculations, the utilization percentage was converted to dwell days using Equation (2). Note that Days_Period in Equation (2) is the total number of days in a period for which the data has been supplied. Days_Period for monthly assessments can therefore be either 28, 29, 30, or 31 days, whereas Days_Period for annual assessments is always 365 days.

$$D_{dwell}=\frac{Capacit{y}_{Facility}·Day{s}_{Period}·Utili{z}_{Ave}}{N_{Pallets}}$$

(2)

where:

• D_dwell = the average number of dwell days a pallet spends in the facility (days)
• Capacity_Facility = the total pallet capacity of the facility (pallets)
• Days_Period = the number of days in the analyzed period (days)
• Utiliz_Ave = the average utilization of the facility in the time period (%)
• N_Pallets = the number of pallets moved during the analyzed time period (pallets)

3.3. Electricity Consumption Rate

To calculate the electricity consumption rate of a pallet (research aim 2), Equation (3) was used:

$$Electricit{y}_{Pallet}=\frac{Q_{electricity}}{D_{dwell}·N_{Pallets}}$$

(3)

where:

• Electricity_Pallet = the amount of electricity used per pallet-day (kWh pallet-d⁻¹)
• Q_electricity = the total amount of electricity used by the cold store (kWh)
• D_dwell = the average dwell days a pallet spends in the facility (days) (Equation (2))
• N_Pallets = the number of pallets moved during the time period (pallets)

3.4. Diesel Consumption Rate

The diesel consumption rate per pallet was calculated by using Equation (4):

$$Diese{l}_{Pallet}=\frac{Q_{diesel}}{D_{dwell}·N_{Pallets}}$$

(4)

where:

• Electricity_Pallet = the amount of diesel used per pallet-day (ℓ pallet-d⁻¹)
• Q_diesel = the total amount of diesel used by the cold store (ℓ)
• D_dwell = the average dwell days a pallet spends in the facility (days) (Equation (2))
• N_Pallets = the number of pallets moved during the time period (pallets)

3.5. Emission Intensity Factor

To calculate the emission intensity factor (research aim 3) of a pallet, Equation (5) was used:

$$EI{F}_{Pallet}=\frac{Emissio{n}_{Total}}{D_{dwell}·N_{Pallets}}$$

(5)

where:

• EIF_Pallet = the average emission intensity factor for a pallet-day (kg CO₂e pallet-d⁻¹)
• Emission_Total = the total GHG emissions of the cold store (kg CO₂e)
• D_dwell = the average number of dwell days a pallet spends in the facility (days)
• N_Pallets = the number of pallets moved during the time period (pallets)

3.6. Average Values

Equation (6) can be used to calculate the average or arithmetic mean of monthly or yearly values. For this study, this equation was used to calculate the average rate of electricity consumption and average emission intensity factor for the various months and years.

$$Averag{e}_{period}=\frac{{\sum }_{i=1}^n{X}_i}{n}$$

(6)

where:

• Average_Period = the average value for the period assessed
• X_i = the individual data values to be averaged
• n = the number of data values to be summed

Finally, to calculate the weighted average dwell days in Section 4.2, Equation (7) was used. The weight (w_i) applied to different dwell-day values indicates the importance of the specific value; for example, if 19 of the 21 data points had an average of 6.72 dwell days, w_i would be 19/21 for the value of 6.72 days.

$$Weighte{d}_{Ave}=\frac{{\sum }_{i=1}^n{w}_i·X_i}{{\sum }_{i=1}^n{w}_i}$$

(7)

where:

• Weighted_Ave = the weighted average value
• n = the number of terms to be averaged in the calculation
• w_i = the weight applied to each X_i value
• X_i = the data value to be averaged

4. Results

The results of the analysis are given in three different sections. Section 4.1 states the annual results for a small seasonal cold store that only operates a few months per year. In Section 4.2, both the monthly and the annual results of several large commercial cold stores that operate year-round are given. Section 4.3 indicates the monthly and annual results for two cold stores that are regarded as exceptions to the status quo. These results are included to illustrate the extent of possible results.

4.1. Seasonal Cold Stores

The results for a small seasonal cold store with a pallet capacity of 345 pallets are given in

Table 3. Annual results for a small seasonal cold store.

	Dwell Days
	1	2	3	4	5	6	7
Electricity consumption rate (kWh pallet-d−1) 1	57.49	28.75	19.16	14.37	11.50	9.58	8.21
Diesel consumption rate (ℓ pallet-d−1)	0.20	0.10	0.07	0.05	0.04	0.03	0.03
Emission intensity factor (kg CO2e pallet-d−1)	56.99	28.50	19.00	14.25	11.40	9.50	8.14
Utilization percentage of facility	8.95%	17.89%	26.84%	35.79%	44.73%	53.68%	62.63%

¹ If the unit should be changed to kWh pallet⁻¹, multiply the dwell day number by the value in the column.

. This cold store (Facility A) emitted 448.41 t of CO₂e over a three-year analysis period. It is deemed seasonal since it only operates from December to March during the harvest season of the production region. After the harvest season, the facility is closed since there are no other products or produce in the region that require cold storage. The results in Table 3 are based on three years of data, during which 9 428 pallets of fruit were moved through the facility. Because the cold store is closed for a large portion of the year (April to November), the assessment was performed on an annual basis instead of specifying results for the different months. Subsequently, it is assumed that the results in Table 3 are representative of any working month of the year.

From the cold store’s data and a discussion with the facility manager, it was evident that the actual dwell days varied considerably. This variation in dwell days is due to several logistical reasons beyond the scope of this article. Nevertheless, as a consequence of the erratic dwell days, the calculations were performed for a varying number of dwell days ranging from one to seven. This range represents all possible dwell days of fruit moving through this facility.

The electricity consumption rate in Table 3 states the total amount of electricity required to handle and store a pallet in the facility for a single day, given that all other pallets in the facility have the same dwell period (as stated in each column). Likewise, the diesel consumption rate, with units of ℓ pallet-d⁻¹, states the total amount of diesel fuel “consumed” by a pallet during a single day in the cold store. The amount of diesel used at a facility is dependent on power outages (loadshedding); thus the diesel consumption varied considerably during the three analyzed years. The authors therefore advise that the diesel consumption results not be analyzed in any further depth or detail, such as evaluating trends or correlations. The emission intensity factor states the actual weight, in kg, of GHG emitted by a pallet in a day. Finally, the utilization of the facility is indicated as a percentage to provide the reader with a gauge of the logistical performance of the cold store. Note that no fugitive emissions are stated in Table 3 since no refilling of refrigerants occurred at the facility during the three-year analysis period.

4.2. Large-Scale Commercial Facilities

4.2.1. Monthly Assessment Results

The three large commercial cold stores (Facility B, C, and D) assessed in this section have a very large pallet capacity and handle significant volumes of fresh fruit. Based on the 2020 data of the three analyzed facilities, a total of 360,000 pallets of fruit were handled and stored in the three facilities in this year. In the same period, a total of 260 million cartons of fresh fruit were exported from South Africa [32]. If approximated (assume that a pallet consists of 80 cartons of fruit), the 260 million cartons translate to 3.25 million pallets of fresh fruit. This means the three analyzed facilities in this section were responsible for 11.07% of the total fresh fruit exported from South Africa during 2020.

The smallest facility analyzed in this section (Facility B) has a pallet capacity of 5000 pallets, with all three facilities having a combined pallet capacity of 16,100 pallets. During the analysis period, the assessed facilities collectively handled and stored a total of 1.73 million pallets of fruit. All three facilities handle various types of fruit (citrus, table grapes, pome fruit, stone fruit, subtropical fruit, etc.) from several production regions across the country and operate 12 months per year. Each average monthly value displayed in

Table 4. Monthly results for large commercial cold stores.

	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Electricity consumption rate (kWh pallet-d−1) 1	6.62	8.87	8.01	10.92	7.92	9.16	7.21	8.35	11.41	19.77	10.94	5.90
Diesel consumption rate (ℓ pallet-d−1)	0.02	0.02	0.02	0.02	0.02	0.01	0.01	0.01	0.02	0.05	0.05	0.05
Hypothetical emission intensity factor— assuming no solar (kg CO2e pallet-d−1)	6.52	8.74	7.91	10.74	7.81	9.00	7.09	8.20	11.22	19.49	10.82	5.92
Actual emission intensity factor—with solar (kg CO2e pallet-d−1)	4.03	6.78	6.55	6.60	6.05	4.91	5.64	5.67	6.06	13.49	7.02	3.22
Utilization percentage of facility	49.74%	42.63%	40.70%	37.23%	49.83%	47.60%	53.62%	46.02%	40.49%	14.64%	15.86%	41.73%

¹ If the unit should be changed to kWh pallet⁻¹, multiply the values in the table by 6.72.

is based on approximately 16 data points for that specific month. All results are calculated for the actual recorded dwell days of 6.72 d that fruit spent in the facility.

In terms of the total emissions, Facility B emitted 28,262 t of CO₂e in order to move 931,320 pallets of fruit during the 83 months analyzed. Facility C emitted 32,283 t of CO₂e during an 88-month period, during which it moved 539,187 pallets. For Facility D, data were only available for 23 months. During this period, the facility moved 260,042 pallets, which emitted 10,364 t of CO₂e.

The electricity consumption rate in Table 4, with units of kWh pallet-d⁻¹, states the total amount of electricity required by a pallet during a single day, regardless of the source of electricity. Since all three of the analyzed facilities have solar plants installed, a large proportion of the required electrical energy was generated and supplied by solar. This renewable energy reduces the use of grid electricity and leads to a lower emission intensity factor. The authors deemed it appropriate to state a dual emission intensity factor (kg CO₂e pallet-d⁻¹): a higher hypothetical value, which assumes that no solar energy is used, and an actual emission intensity factor, which incorporates the use of solar power. This differentiation was done to allow benchmarking with other cold stores that do not have solar systems installed. The diesel consumption rate, with units of ℓ pallet-d⁻¹, is once again dependent on power outages, and should therefore not be analyzed further. Also, note that no refrigerant leaks were reported, and no refrigerant gas was refilled at any of the three facilities. However, if a leak were to occur, the emissions contribution would still be negligible since Ammonia (NH3), which has a global-warming potential of zero, is used as refrigerant at all three cold stores.

Figure 2 provides a visual representation of the results given in Table 4. From Figure 2, it is evident that the actual emission intensity factor is significantly lower than the hypothetical emission intensity factor. On average, the solar system reduced the dependence on grid electricity by 24.5% across all months. In particular, solar electricity reduced the amount of grid electricity consumed by 45.9% and 45.7% in September and December, respectively. Note that the emission intensity factor is the highest during October and November since a small volume of fruit moves through the facilities during these months. This means the total emissions during these months are apportioned to a smaller number of pallets, which leads to the higher value.

Figure 3 displays the total megawatt-hours (MWh) generated by each facility since the installation of the solar systems. Each facility has a different installed capacity, hence the difference in generated values between facilities. From Figure 3, a general trend is apparent: during the summer months (October to April), the solar system generates more electricity than during the winter months (typically May to August). In June, the least amount of electricity is generated by the solar plants, presumably due to weather-related conditions during the South African winter.

4.2.2. Annual Assessment Results

Apart from assessing and reporting the results on a monthly basis, an assessment was also done of annual data. For this, Facility B, C, and D as well as two other facilities, E and F, were analyzed on a yearly basis. Facilities E and F have a combined pallet capacity of 10,000 pallets, which allows for comparison to Facilities B, C, and D in terms of capacity. This means all the analyzed facilities have a combined capacity of 26,100 pallets. Both Facility E and F, however, only supplied data for a one-year period as no other historical data were available for either facility. During the one-year period, Facility E had a throughput of 143,000 pallets and emitted 7012 t of CO₂e emissions. During the same period, Facility F moved 109,000 pallets of fruit and emitted 6190 t of CO₂e. All five facilities (B, C, D, E, and F) collectively moved 1.98 million pallets of fruit and emitted 84,114 t of CO₂e during the period of analysis.

Based on the collected data from facilities, 612,000 pallets of fruit moved through the analyzed facilities in 2020 alone. This means the analyzed facilities in this section were responsible for 18.8% of the total fresh fruit exported from South Africa during 2020.

Table 5. Annual results for large cold stores.

	Average Annual Value
Electricity consumption rate (kWh pallet-d−1) 1	7.62
Diesel consumption rate (ℓ pallet-d−1)	0.014
Hypothetical emission intensity factor— assuming no solar (kg CO2e pallet-d−1)	7.52
Actual emission intensity factor— with solar (kg CO2e pallet-d−1)	5.72
Utilization percentage of facility	40.25%

¹ If the unit should be changed to kWh pallet⁻¹, simply multiply the value in the table by 6.76.

states the average values of the five analyzed facilities. The averages are based on 21 data points, with each data point representing one year’s data for a cold store. In terms of the dwell days, the weighted average dwell days of the 21 data points were used in all calculations. In total, 19 of the 21 data points had an average of 6.72 dwell days, while the remaining two had a dwell period of 6.89 d and 7.51 d, respectively. The results in Table 5 have been calculated for a weighted average dwell days value of 6.76 d.

If the results of Table 5 are to be used in a simulation study (agent-based modelling or discrete event modelling), Appendix A should be consulted. In Appendix A, a statistical distribution is fitted to the annual results of Table 5.

Due to the installation of solar plants at three of the five cold stores at the beginning of 2018, a total of 9.45 GWh of electricity was generated over a three-year period. This substitution of 9.45 GWh of “dirty” South African grid electricity by clean renewable energy led to a cumulative emissions saving of 9260 t of CO₂e, as shown in Figure 4. If the solar plants had been absent, the 9260 t of CO₂e would have been emitted into the atmosphere. It is interesting to note that the emissions trajectory of the analyzed facilities changed dramatically in 2020 due to the effect of the Covid-19 pandemic. It is, however, yet to be seen if this emissions reduction will be recouped in years to come.

Using a combination of solar and grid electricity has a major impact on the emission intensity factor. From Figure 5, it is apparent that the actual emission intensity factor decreased by nearly 2 kg CO₂e pallet-d⁻¹ from 2018 onwards owing to the installation of solar systems. Solar systems are, therefore, a relevant decarbonization strategy for cold-store facilities. All facilities have the potential space and capability to expand their solar systems. The practicality of becoming completely self-sufficient with solar electricity is, however, debatable. Power generation only occurs during daytime, meaning that electricity has to be stored for night-time consumption. None of the analyzed facilities store electrical energy as batteries are extremely expensive and have a limited lifespan. In addition, the required battery capacity to store this amount of energy would be significant and impractical. Furthermore, due to reduced generation in winter months, the required capacity would need to be increased exponentially to remain self-sufficient, which would lead to overproduction during the summer months.

4.3. Cold Stores Regarded as Exceptions

The two cold stores analyzed in this section, Facilities G and H, are regarded as exceptions. These cold stores were built several decades ago and have not been upgraded or equipped with newer technology as is the case for Facilities A to F. The use of old refrigeration technology inevitably led to reduced energy efficiency and high emissions values. In particular, the potential impact that refrigerant leakage can have on the emissions of a facility was evident at these facilities. (It should be noted that both cold stores have in the meantime been closed due to age and inefficiency.) The analysis and results of Facility G and H are only stated on a monthly basis to emphasize the seasonal movement of cargo through the facilities.

Facility G had a pallet capacity of approximately 1000 pallet locations and handled 406,905 pallets over a period of 5 years and 7 months. The facility emitted 9600 t of CO₂e over the analysis period of 67 months. The results of a monthly analysis for facility G are indicated in

Table 6. Monthly results for exception cold store G.

	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Electricity consumption rate (kWh pallet-d−1) 1	20.51	34.27	12.92	8.74	8.10	7.74	7.75	10.56	13.83	47.12	32.12	9.63
Emission intensity factor (kg CO2e pallet-d−1)	20.10	33.58	12.67	8.57	7.93	7.59	7.59	10.35	13.55	46.18	31.48	9.44
Utilization percentage of facility	30.22%	18.16%	37.98%	57.83%	55.13%	52.24%	52.12%	43.07%	33.79%	9.41%	13.77%	31.07%

¹ If the unit should be changed to kWh pallet⁻¹, simply multiply the values in the table by 2.

. Note that no diesel fuel data was reported as no generators were present at the facility due to the exemption of this facility from loadshedding. Also, no fugitive emissions were reported since no refilling of NH3 refrigerant had occurred at the facility in the 67 months. For facility G, the results in Table 6 were calculated for a dwell-days period of two days.

The second exception cold store, Facility H, is a typical example of the potential effect of fugitive emissions. This cold store has a pallet capacity of approximately 3000 pallets and emitted 24,839 t of CO₂e during an eight-year period, during which the facility moved 347,059 pallets. In total, 33% or 8197 t of this facility’s total annual emissions are due to the leakage of 2.1 t of refrigerant (R-507 and R-22). These two refrigerants have a very high global-warming potential and the leakage of even a small amount of refrigerant leads to a significant increase in emissions.

The results for a monthly assessment of Facility H are shown in

Table 7. Monthly results for exception cold store H.

	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
Electricity consumption rate (kWh pallet-d−1) 1	24.54	25.82	17.79	12.73	5.40	4.07	3.02	2.96	4.11	23.89	24.46	33.24
Fugitive emissions (kg CO2e pallet-d−1)	75.47	64.31	58.43	44.14	9.34	5.22	3.12	2.17	7.13	36.22	43.03	49.88
Emission intensity factor (kg CO2e pallet-d−1)	99.52	89.61	75.87	56.62	14.64	9.21	6.08	5.07	11.16	59.63	67.01	82.46
Utilization percentage of facility	9.45%	8.33%	8.76%	19.34%	46.42%	61.40%	84.24%	96.12%	51.67%	13.37%	15.21%	13.98%

¹ If the unit should be changed to kWh pallet⁻¹, simply multiply the values in the table by 8.

. The results were calculated for a dwell-day period of eight days. Apart from the high electricity consumption rate during some months, the fugitive emissions in Table 7 are responsible for between 43% and 78% of the total emission intensity factor for August and April respectively. Fugitive emissions are, therefore, a significant contributor to the emission intensity factor across all months. Once again, note that no diesel consumption is reported for the same reason as was the case with Facility G.

5. Discussion

The discussion of the results is grouped into three parts: Section 5.1 examines the impact of various sources of emissions; Section 5.2 recommends the most appropriate results for future research; and Section 5.3 provides a reflection on the assessment as a whole.

5.1. Sources of Emissions

Electricity is the largest contributor to the total emissions of all the analyzed facilities. All eight facilities depend on carbon-intensive grid electricity from Eskom. When benchmarked against international peers that also use predominantly coal in the energy mix, Eskom has the highest emission factor (kg CO₂e kWh⁻¹) of all utilities [28]. The installation of solar systems at cold-storage facilities as an alternative to grid electricity will become increasingly important in future. Solar energy not only reduces the emissions of a cold store, but also makes the facilities less vulnerable to loadshedding and power outages. Despite only having generation capability during daytime, solar energy will become important to ensure the sustainability of cold stores. These facilities are also easier to decarbonize compared to transport vehicles. Future emissions reduction efforts in the distribution chain should be focused on cold-storage facilities.

Fugitive emissions and the use of old refrigeration technology can have a significant impact on the total emissions and emission intensity factor of a cold store, as was shown in Section 4.3. The choice of refrigerant used at refrigerated facilities is extremely important and requires careful consideration. In future, refrigeration systems that use refrigerants with a low or zero global-warming potential, such as NH3, N2 or water, should be used. Furthermore, the use of new refrigeration technology is essential to improve the energy efficiency of a facility. The leakage of refrigerants leads to higher electricity consumption and hence to higher emissions. Facilities normally only realize that a leakage has occurred once they see a spike in electricity consumption.

All facilities use relatively little or no diesel fuel to run back-up generators in the case of loadshedding. When the effects of electricity and diesel on emissions are compared, it is evident that diesel has been responsible for less than 1% of the total emissions of any cold store. This may, however, change if more diesel has to be used to run generators to supply electricity.

5.2. Results

It is recommended that the results of the large-scale commercial facilities be used in all future research or analysis (Section 4.2) of carbon emissions for fruit exports. These results best reflect the current cold-storage industry in South Africa, which is responsible for the export of fresh fruit. Most of the fruit exported from South Africa moves through one of the large-scale commercial facilities. Hence, it is better to use the results of these large-scale facilities.

Due to the seasonal nature of values, emission calculations on a shipment level should preferably be based on the monthly values in Table 4. However, if the month is unknown, the annual values in Table 5 must be used.

Based on the annual results that were presented in Section 4.2, storing a pallet of any type of fruit for a day in cold storage requires 7.62 kWh of electricity. If cold stores have no solar system installed, a hypothetical emissions value of 7.52 kg CO₂e pallet-d⁻¹ is possible. However, due to installed solar systems, the actual emission intensity factor declines to 5.72 kg CO₂e pallet-d⁻¹. This factor will differ for each cold store depending on the capacity of solar systems available, if any.

5.3. Reflection on Assessment

This section lists some important considerations that affected the assessment process and associated results. These include:

• Sensitive nature of organizational data: Due to the sensitive nature of organizational data, it is often difficult or impossible to obtain high-quality primary data. In the majority of cases, cold-store managers or other members of senior management are the only individuals who have access to the required data. These individuals are reluctant to share data for privacy reasons or due to a lack of understanding of the potential value of an assessment. In terms of the quality of data, some facilities do not record or keep records of all the required data for an emissions assessment. In some instances, as was the case for the diesel consumption at one of the cold stores in this study, the quantity of fuel used per month was not recorded and had to be derived from fuel invoices. This lack of quality data hampers emissions assessments since missing data cannot be substituted from literature, even if such data were available. Better capturing of the appropriate data elements mentioned in Section 2.4 is, therefore, required for more accurate assessments.
• A lack of a standard methodology for cold stores as logistical sites: There are still considerable research challenges with assessing, quantifying and gauging GHG emissions at a facility level, even when using the methodology developed by [20]. This lack of a prescriptive standard leaves room for erroneous assumptions and leads to both inaccurate and incomparable results.
• Allocation of emissions among logistical units: The process of fairly allocating the total emissions among all goods that moved through the cold store is a major problem (the allocation problem). The allocation of refrigeration energy should not only be fair but also practical to apply. Some shipments of fresh fruit are stored for a short duration and others for extended periods. The dwell days will certainly have an impact on the emissions value of a specific shipment. In addition to a time aspect, pallets of fruit have different configurations (prepared for standard or high-cube containers) or sizes, which lead to different weights.
• Dwell days and utilization: The number of dwell days and the utilization percentage of a facility have a major impact on the assessment results. Allocating emissions fairly to pallets that spend a couple of days in the cold store should be different to pallets that do so for weeks. The consumption rate of electricity (kWh pallet-d⁻¹) and the emission intensity factor (kg CO₂e pallet-d⁻¹) are inversely related to the number of dwell days. If a time-based element is incorporated in an assessment, the dwell days must be accurate.
• Availability of emission factors: The accurate quantification of emissions is dependent on the availability of applicable emission factors, as stated in Section 2.5. Emission factors for fuels such as diesel are not available for South Africa. Since the feedstock and production process of fossil fuels differ from one geographical location to another, the emission factor could potentially be much higher in South Africa, particularly since South Africa’s fuel is partly derived from coal. If newer emission factors for South African electricity or diesel becomes available, the energy consumption in Table 3 can be multiplied by the new factors to provide a timelier hypothetical emission intensity factor.

6. Conclusions

This article is the first to use a top-down approach to assess and apportion the emissions of a cold store that handles and stores fresh fruit. The results revealed that cold stores that handle and store fresh fruit emit significant amounts of GHG. In total, the eight cold stores analyzed in this article collectively emitted 119,001 t of CO₂e while moving 2.74 million pallets of fresh fruit. This large emissions value is due to a combination of the high electrical energy required for refrigeration, the carbon intensity of South Africa’s electricity and, finally, the sheer volume of fruit that moved through the analyzed facilities. Solar systems are, however, a good decarbonizing strategy to reduce the overall emissions of a cold store. To realize the ambitious United Nations Sustainable Development Goals and to meet the GHG reduction targets of the Paris Agreement, more action than the installation of solar panels is required.

The analysis of cold-store data revealed that storing and handling a pallet of fruit for a day in a large commercial cold store requires 7.62 kWh of electricity. This value is dependent on the volume and type of fruit that moves through the cold store. If the cold store has no solar system installed, a hypothetical emission intensity factor of 7.52 kg CO₂e pallet-d⁻¹ is possible. These results are valuable since they make it possible to quantify the emissions when a shipment of fresh fruit is moved through a cold store. However, emissions assessments for logistical facilities, including this study, are hampered by the absence of prescriptive standards.

Finally, it is recommended that a representative organization be established to represent and further the interest of cold stores or refrigerated facilities. This organization should prescribe a standard methodology for different types of cold stores to assist and enable organizations to collect, assess and prepare data and accurately calculate meaningful results. Furthermore, a representative organization could—and should—become an anonymous repository for all results in order to generate benchmarks for the cold-store industry.