Improving thermal energy efficiency in a Norwegian dairy utilising integrated CO2 refrigeration systems: performance data and energy efficiency improvement possibilities

Improving thermal energy efficiency in a Norwegian dairy utilising integrated CO 2 refrigeration systems: performance data and energy efficiency improvement possibilities Lukas K¨ OSTER a,b,* , Jan BENGSCH a,b,* , Kristina Norne WIDELL a a SINTEF Ocean, Dep. Fisheries and New Biomarine Industry, 7465, Trondheim, Norway b NTNU, Department of Energy and Process Engineering, Kolbjørn Hejes vei 1B, 7491 Trondheim, Norway ARTICLE INFO Keywords: CO 2 Refrigeration Heat pumps Dairy Dairy processing Natural refrigerants ABSTRACT Dairies are considered very energy intensive due to the high demand for thermal energy at different temperature levels. This study investigates the thermal energy demand of the largest organic dairy in central Norway, which utilises a fully integrated CO 2 refrigeration system to cover their cooling and hot water demand and an electric steam boiler for pasteurisation, sterilisation and cleaning in place (CIP). The specific energy flows of 2024 were analysed monthly and thermal demand profiles created. The specific energy consumptions were found to be between 27.5 Wh/l to 43.0 Wh/l for cooling and 34.3 Wh/l and 48.4 Wh/l for hot water heating. The electric steam boiler accounted for 36.4% of the plant’s electricity consumption in 2024. Therewith, the study aims to present real world data of the (thermal) energy consumption of a dairy and thereby lay ground for further system developments, which are conceptualised in this paper. 1. Introduction According to (Mehta et al., 2024), Norwegian greenhouse gas emissions (Scope 1 +2) associated with energy use in the food value chain accounted for 4.07 MtCO 2 e in 2019: around 13 % of this was attributable to food and beverage production. According to (Egas et al., 2021), the dairy supply chain is estimated to be responsible for 3–4 % of the global anthropogenic GHG emissions. The OECD-FAO Agricultural Outlook (OECD and Food, 2022) forecasts the milk production to increase 4.6 % in the European Union from 2019–2021 to 2031, while the production in India and Pakistan is forecasted to increase by 42 % and 44 %, respectively. The food processing industry, especially dairies, has seen a significant increase in energy and water consumption in recent years, driven by stricter hygiene and cleaning standards (Ladha-Sabur et al., 2019). The high energy demand is due to the need for thermal energy at different temperature levels (Briam et al., 2015), which is historically associated with high greenhouse gas emissions (GHG) (Selvnes et al., 2023). With Norway and the European Union aiming to be climate neutral by 2050, measures to reduce GHG emission must be taken (European Commission, 2020). (Üçtu˘ g, 2019) compared 31 life cycle assessment studies for the production of dairy products and found that raw milk production had the highest share in environmental impact. For the impact factor “energy use” in milk processing, the share was found close to raw milk production. Generally, storage and use of dairy products were found to have minor effects on the environmental footprint. More than half of the recommendations were about milk processing and 10 of 31 studies suggested the use of more energy efficient equipment. (Üçtu˘ g, 2019) Traditionally, separate systems are used to cover thermal demands at different temperature levels: compression refrigeration systems for cooling and fossil fuel based systems for heating (Selvnes et al., 2022). Hence, the decarbonisation of process heat and the increased integration of heat recovery from refrigeration systems are promising solutions for reducing the environmental impact (Selvnes et al., 2022); (Ahrens et al., 2021); (Schlemminger et al., 2022). Detailed measurements of the main thermal energy consumers are essential to understand both the cooling and process heat requirements in the dairy industry throughout the production day, as well as seasonal fluctuations. Such measurement data can be applied in various ways, including the optimisation of energy efficiency within dairies, the development of accurate load profiles for system simulations, and numerous other applications. However, there is a clear gap in the availability of open-access, real-world data on thermal energy consumption in dairies and the wider food industry. This information is * Corresponding authors. E-mail addresses: [email protected] (L. K¨ OSTER), [email protected] (J. BENGSCH), [email protected] (K.N. WIDELL). Contents lists available at ScienceDirect International Journal of Refrigeration journal homepage: www.elsevier.com/locate/ijrefrig https://doi.org/10.1016/j.ijrefrig.2025.10.001 Received 19 May 2025; Received in revised form 2 September 2025; Accepted 1 October 2025 International Journal of Refrigeration 180 (2025) 564–571 Available online 2 October 2025 0140-7007/© 2025 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). crucial for designing efficient decarbonisation strategies and for the validation of simulation models. (Jesper et al., 2021); (Sandhaas et al., 2022); (Miserocchi et al., 2024) Therefore, the aim of this study is to showcase the energy consumption and thermal demand patterns of an entirely fossil fuel free dairy that uses natural refrigerants only. The dairy is located in southern Norway and is the largest organic dairy in Norway with a production volume of 20.5 million liter of processed milk in 2024. Energy efficiency measures have been ongoing since 2015 and work has been documented by (Selvnes et al., 2023); (Selvnes et al., 2022); (Bengsch et al., 2024); (Bengsch et al., 2023); (K¨ oster et al., 2024). This study is the culmination of previous efforts, with sensors being installed through several projects. Through integrating CO 2 refrigeration systems with heat recovery, introducing an electric steam boiler and other measures, the dairy was able to reduce its GHG emissions by 41 % from 2018 to 2024, while increasing the volume of processed milk by 56 %. Over the same period, this led to a 49 % reduction in the specific energy consumption per litre of milk processed to 118 Wh/l. At first, the system configuration and methods are introduced. Subsequently, the specific thermal energy consumption and key performance indicators of the thermal energy system are presented. Following, the daily thermal energy patterns of the cooling demand, hot water demand and steam demand during production days are evaluated. Furthermore, improvement opportunities for the thermal energy system are showcased, depending on the temperature level. The underlying data of the study is made available under the DOI: 10.5281/ zenodo.15111145. 2. System description The thermal energy system of the dairy consists of two main systems: a refrigeration system to meet cooling and hot water demand, and a steam boiler to supply high-temperature thermal energy for processes such as pasteurisation and sterilisation. Five CO 2 refrigeration systems are installed in parallel with staggered capacity, as shown in Fig. 1. In total, the nominal installed cooling capacity is 350 kW, however, the CO 2 unit 1 & 2 are prototypes from 2015 and are expected to go out of service. Propylene glycol is used as a secondary fluid for CO 2 units 2, 3 & 5 on the cold side with an integrated sensible energy storage of 5 m 3 and a capacity of 27 kWh. The temperature of the supplied glycol is −3.5 to −5 ◦C. Therefrom, the cold thermal demand of several cooling rooms and cold storages is covered. The process cooling consumer uses water as a cooling medium and a 9 m 3 sensible energy storage with a capacity of 15 kWh is installed. In June 2024, the CO 2 unit 4 was directly integrated into the return line of the process cooling consumers and CO 2 unit 1 & 4 supplied chilled water to the thermal energy storage. Beforehand, the CO 2 unit 4 was integrated into the glycol loop, parallel to CO 2 unit 2, 3 & 5. The setpoint temperature of the supplied process water is approx. 1 ◦C. For redundancy, the process cooling consumer loop is connected through a heat exchanger to the glycol cooling loop. The heat sink of the CO 2 units is a hydronic system, which produces hot water up to 65 ◦C for covering hot water demands. Only CO 2 unit 5 utilises an ambient gas cooler in parallel to the water-cooled gas cooler. The hydronic system is of open type and utilises cold tap water, with inlet temperatures of 6.5 ◦C on average. Hot water from the CO 2 refrigeration systems is stored at approximately 65 ◦C in sensible hot water storages of two sets of 12 ×400 l storage tanks. The hot water system supplies several hot water and CIP consumers, as indicated in Nomenclature AC Air conditioning avg average CIP Cleaning in Place COP Coefficient of performance DH District heating QEnergy (J) ˙ QHeat (W) GC Glycol GHG Greenhouse gas emissions HtCR Heating to cooling ratio HTHP High temperature heat pump HW Hot water ˙ mmass flow rate (kg/s) MP Measurement point pPressure (Pa) PC Process cooling cpSpecific heat capacity (kJ/kg-K) T Temperature (K) Fig. 1. Simplified P&ID of the thermal energy system of the dairy including the installed measurement points. L. K¨ OSTER et al. International Journal of Refrigeration 180 (2025) 564–571 565 Fig. 1, and supplies feedwater to the electric boiler. CIP truck hall (HW05) is a CIP station located in the milk delivery hall for cleaning of the milk delivery trucks and CIP process (HW-07) is a CIP station used for cleaning process equipment, such as the pasteurisers. The measurement point of other hot water consumers (HW-06) includes consumers for rinsing of processing equipment, butter production and other uses. The electric boiler has a set capacity of 500 kW el and maintains 6 bar in the steam system during the day and 4 bar overnight. The electric boiler covers heating demands of pasteurisation, sterilisation and CIP at temperatures between 65 ◦C to 100 ◦C. In addition, the dairy is coupled to the district heating grid for covering their AC and space heating demands of the office and production buildings. 3. Material and methods In Fig. 1, the measurement points of the energy flow and mass flow meters are marked and their corresponding sensor types defined in Table 1. The term energy flow meter refers to a measurement device, which measures temperatures and mass flow for a given fluid and calculates the heat flow rate based on Eq. (1). Data was collected at oneminute intervals and a Savitzky-Golay Filter was applied with a 15-minute interval and a third degree polynomial to the calculated heat loads. Thermodynamic properties of the secondary fluids are evaluated using the Coolprop library based on (Bell et al., 2014). Data of the electric boiler was collected from the electricity provider in an hourly interval. Cooling and thermal loads of the heat sink of the CO 2 machines were calculated after Eq. (1) for each measurement point (MP), based on (Incropera et al., 2017). ˙ QMP =˙ mMP⋅cp,fluid(Tavg,MP,p)⋅ΔTMP (1) Until October 2024, city water was utilised to cool yoghurt after incubation, whereas after October 2024, yoghurt cooling was carried out through integration into the cold process water loop (as shown in Fig. 1). Therefore, until October 2024, the cooling demand of yoghurt was calculated based on the monthly production, a constant specific heat capacity of cp=3.3 kJ/kgK and a temperature difference of 25 K based on Eq. (1). Direct process water cooling by the CO 2 unit 1 & 4 was calculated as following: Since the cooling load on CO 2 unit 1 was not measured by the installed sensors, it was assumed that it operated at a cooling capacity of 35 kW per production day for 8 h from 07:00 – 15:00, based on information of the plant operator. The cooling load of CO 2 unit 4 was calculated by an energy balance from the condenser (HW-02) and electricity consumption. The measurement equipment of the hot water consumers (HW-05, HW-06, HW-07) measured mass flow only, therefore, the energy flow was estimated based on available information from the plant operator regarding the supply temperature. The supply temperature to the consumers was estimated at 65 ◦C and the city water inlet temperature at 6.5 ◦C on average and Eq. (2) calculates the hot water consumer thermal demands, based on (Incropera et al., 2017). ˙ QMP =˙ mMP⋅cp,fluid(Tavg,p)⋅(Tsupply −Tcity water)(2) The heating to cooling ratio (HtCR) was defined according to Eq. (3) and is introduced as a new parameter: HtCR =(Qheating consumption +Qsteam demand ) Qcooling consumption (3) Therewith, the HtCR described the ratio between the heating demand to the cooling demand and indicates whether the dairy can be in its current configuration self-sufficient in terms of its thermal energy demand. The HtCR showcased whether the dairy can, by for example the usage of a heat pump, utilise its excess heat from refrigeration systems, to cover heating demands of other processes without the need for external heat sources. The combined Coefficient of Performance (COP comb. ) per CO 2 machine is defined as: COPcomb.=COPheating +COPcooling =2⋅ ˙ Qgas cooler ˙ Wcompressor −1 (4) and the cooling COP cooling is defined as: COPcooling = ˙ Qgas cooler − ˙ Wcompressor ˙ Wcompressor (5) 4. Results and discussion In this section, the energy flows and demand patterns of the dairy in 2024 are presented and analysed. Section 4.1 showcases the specific monthly-aggregated thermal energy consumption for different consumers. Furthermore, seasonal influences are presented by comparing daily energy demand patterns for a summer and winter month. Since there were changes in the thermal system during the year and the production strategy was changed from "make to order" to a mix of "stock-based production" and "make to order", the months of August (summer) and December (winter) were compared. This means that the same boundary conditions apply to the production method and the setup of the thermal energy system. This comparison is intended to clarify the seasonal fluctuations based on the influence of the ambient temperature on the thermal loads. It is only necessary to keep in mind that before October yoghurt was cooled with tap water (applies to the summer month of August under consideration) and this was then switched to the cold process water loop (applies to the winter month of December). As the thermal load profiles of working days and weekends and bank holidays differ greatly, only working days are shown in the comparison of summer months (August) and winter months (December) in Sections 4.2, 4.3 and 4.4. Table 1 Overview over utilised sensors and indicated accuracy. All sensors measure supply and return temperature to the consumer, as well as the mass flow. Hot water sensors HW-05, HW-06 and HW-07 measured mass flow only. Measurement points – subsystems Abbreviation Medium Sensor-type Indicated accuracy Process Cooling PC-01, PC-02 Water Badger M1000 ±0.3 % Glycol cooling loop – cooling rooms, glycol – ice water heat exchanger, heat load to CO 2 units GC-01, GC02, GC-03, GC04 Propylene glycol (35 %) KROHNE Optisonic 3400 ±0.3 % +2 mm/s Hot water – CO 2 unit 2, 3, & 5, HW CIP process, HW CIP truck hall, HW production HW-01, HW02, HW-03, HW-04, HW05, HW-06, HW-07 Water Kamstrup Ultraflow 54/ 24 ±(2+ 0.02⋅qp/q)% But smaller than ±5 % District heating – AHU 1, AHU 2, truck hall, ventilation H-01, DH-02, DH-03, DH04 Water Kamstrup Ultraflow 24 ±(2+ 0.02⋅qp/q)% But smaller than ±5 % All Temperature sensors All energy flow meters Water/ Propylene glycol (35 %) Kastrup Temperature Sensor 63 (PT500) ±0.08 K L. K¨ OSTER et al. International Journal of Refrigeration 180 (2025) 564–571 566 4.1. Thermal energy demand – plant level The monthly thermal energy consumption is shown in Fig. 2 as a specific energy consumption per liter processed milk. The thermal consumers are grouped depending on their type of supply (cooling, hot water and steam) and respective utilisation. Additionally, the consumed water and processed milk per month are shown. Throughout the year, the specific energy consumption of the cold storage cooling & AC (GC03) ranges between 7.5 Wh/l in January and 18.7 Wh/l in May 2024. Generally, a higher cooling demand of the cold storages was observed during the summer months May – September. Seasonal variations are further investigated in Section 4.2. Process cooling, consisting of the indicated column “glycol process cooling” (GC-02) and the “direct water process cooling” varied between 18.0 Wh/l in April and 25.0 Wh/l in July 2024. The consumer “direct water process cooling” accounted for the cooling demand on CO 2 unit 1 & 4. For the hot water consumer, the HW CIP process (HW-07) in the first seven month of the year had a thermal consumption between 14.4 and 18.1 Wh/l in February to July 2024. In July 2024, the production plan was changed to a hybrid production plan, consisting of a “stock-based production” and a “make-to-order production”. As a result, the HW CIP process demand decreased from 13.8 Wh/l in August to 12.3 Wh/l in December 2024. Larger batch sizes and therefore a reduced intermediate rinsing is found as the reason. The specific energy consumption of CIP in the truck hall (HW-05) was found to be between 6.7 Wh/l in March 2024 and 9.6 Wh/l in May 2024. CIP in the truck hall is dependent on the number of milk deliveries per day. The other hot water consumers (HW-06) were found between 9.7 Wh/l and 11.5 Wh/l in February and January 2024, respectively. The specific thermal energy consumption of water flowing to the accumulation tank and therefrom either to the drain or being utilised for preCIP activities (“HW overproduction / preCIP”, HW-04), ranged from January to July between 10.1 Wh/l in April and 13.9 Wh/l in March 2024, with an outlier in May 2024 of 21.1 Wh/l. In May 2024, due to public holidays, the production was shifted temporarily from five to four production days per week at a similar amount of processed milk compared to March 2024. For this reason, the production was carried out in larger batch sizes, which led to reduced CIP and rinsing activities. Additionally, a high average ambient temperature of 15.3 ◦C in May, with a consequently higher cooling load led in total to a higher overproduction of hot water from heat recovery in May 2024. The specific steam boiler electricity demand is shown as two consumptions: A baseload and a steam demand. The baseload was defined as the load present outside of production hours, specifically between 00:00 – 03:00. The mean value was found at 35.8 kW el, the median at 31.0 kW el and the first and third quantile at 29.0 kW el and 33.5 kW el, respectively. The baseload was due to heat losses to the ambient from the steam system. The specific steam baseload ranges between 13.3 Wh/ l and 17.4 Wh/l in October and May 2024, respectively. On average, the specific steam baseload is responsible for 36.4 % of the steam boiler electricity consumption. Overall, the steam boiler energy consumption was found to consume 35 – 40 % of the plants electricity consumption. Furthermore, CO 2 unit 3, 4 & 5 consumed between 8 – 18 % of the plants electricity consumption. The HtCR was found between 1.9 in August 2024 and 2.9 in February 2024 and on average, the HtCR of 2024 was 2.2. Hence, the dairy was, in its current operation, reliant on additional, external energy sources to cover the heating demand, which was covered through direct electricity to heat conversion in the el. steam boiler. As indicated in Fig. 1, district heating was used for space heating of the dairy including the office buildings. It consumed up to 29 % of the total heating demand (excluding steam) in January 2024. In the summer from May to August 2024, the district heating demand reduced down to <5 % of the total heating demand. District heating is excluded from Fig. 2, since it is not a specific thermal energy consumption per liter of milk. 4.2. Cold side From Fig. 2, a seasonal dependency could be linked to the cooling load. To confirm this, the daily thermal energy demand pattern of August (summer) and December (winter) in 2024 are compared in this section of the cooling load (GC-01 & Direct process water cooling). These months were selected, since the only system change within this period was the integration of yoghurt cooling into the consumer process cooling loop. Fig. 3 shows the combined cooling load of the system, including process cooling and cooling rooms and AC for production days. The illustration is a half violin plot, which contains minutely values for the Fig. 2. Specific (thermal) energy consumption per month of the dairy plant as well as the monthly processed milk and monthly water consumption. L. K¨ OSTER et al. International Journal of Refrigeration 180 (2025) 564–571 567 working days of a month. The violin plot is based on a gaussian kernel density estimation and the density estimation is plotted per hour of the day and highlights the variability and tendency of the cooling demand. A higher bandwidth indicates a higher probability of values at a given cooling demand. When comparing the cooling demand between August and December 2024 in Fig. 3, one can see clearly that both follow a similar trend. The demand rises with the start of the production day between 04:00 to 05:00, has a main production time between 07:00 and 14:00 o’clock and a small peak in the evening between 18:00 and 21:00. The smaller peak in the evening is due to milk delivery. While the average cooling demand in August during night times (23:00 – 04:00) is between 39 kW to 56 kW, the average demand in December is between 21 kW to 41 kW. During the working hours, the difference between the average cooling load of the summer and winter months becomes with up to 52 kW even greater. When comparing the differences in the average thermal energy consumption during production days, August is consuming with 2898 kWh nearly 25 % more energy than December with a value of 2302 kWh. To investigate the reason for higher cooling demands during summer, Fig. 4 shows the cooling demand for the cooling rooms and AC only, combined with the ambient temperature on the 2nd y-axis. The average ambient temperature was found at 15.5 ◦C in August and 0.1 ◦C in December. It is visible that the hourly average of the cooling demand and the ambient temperature in August are following the same trend. The average cooling demand in December 2024 is quite constant, with values between 11 kW to 30 kW. Higher values were observed during production hours between 08:00 to 17:00. During August 2024, the average cooling demand of the cooling rooms and AC fluctuated between 26 kW and 62 kW per hour. Based on Fig. 2, the specific thermal energy consumption of cooling rooms and AC per month was found at 40.2 Wh/l and 27.8 Wh/l for August and December 2024, respectively. Furthermore, the daily average thermal energy consumption of the cooling rooms and AC was 1098.9 kWh in August and 519 kWh in December 2024. This underlines the dependency on the ambient temperature. Besides ambient temperature, the cooling demand of the cooling rooms is dependent on the quantity and type of stored products. As stated in Section 2, the thermal energy system was partly rearranged in June 2024. This was done to change the 80 kW CO 2 unit 4 from cooling glycol to direct cold process water cooling as suggested by (Selvnes et al., 2022). Hence, the evaporation temperature could be lifted, which in return increased the cooling COP by 0.83 from June to July 2024, as it can be seen from Table 2. When comparing the average cooling COP of the first six months of 2024 to the second six months, the COP increased by over 40 %. The combined heating and cooling COP of the 80 kW unit 3 was on average 4.3. Before the integration of the 80 kW unit 4 into the ice water circuit, the combined COP was found at 4.5 and after the integration at 5.94. It should be noted that the COPs for the pressure ratio (23 bar to 89 bars before conversion) in both units are very low. This is due to a very large temperature approach between Fig. 3. Seasonal influence on the combined process cooling, cooling rooms and AC cooling demand. A comparison between August (summer) and December (winter) in 2024 displayed for all working days in the month as a half violin plot and an average cooling demand over the time of the day. Fig. 4. Seasonal influence on the cooling demand of the cooling rooms and AC together with the average hourly ambient temperature. A comparison between August (summer) and December (winter) in 2024 displayed for all working days in the month as a half violin plot and an average cooling demand over the time of the day. L. K¨ OSTER et al. International Journal of Refrigeration 180 (2025) 564–571 568 water inlet and CO 2 outlet temperature from the gas cooler of about 20 K. This was due to a blockage caused by limescale deposits, at least for the CO 2 unit 3, which has now been rectified so that the current COPcooling is about 3. Improvement opportunities cold side According to the theoretical study done by (Selvnes et al., 2022) on this dairy, increasing the evaporation temperature for different consumers by e.g. moving away from glycol as the secondary fluid for AC and the cooling rooms can cut the annual electrical energy consumption for cooling by over 20 %. As seen in Table 2, the integration of the CO 2 unit 4 into the consumer process cooling circuit increased the COP cooling by around 40 %. Since food quality and safety is one of the main goals of a dairy, it is important that all improvement solutions ensure a quick cool down of the milk below 4 ◦C to slow down the bacteria growth (Murphy et al., 2013). With high volumes of milk being delivered or increased production, there might not be enough cooling and heating capacity available. In that case, an opportunity is to install a thermal energy storage, either on the hot or cold side. Thereby, an increase in the utilisation of electricity for a combined production of both heating and cooling, and therefore energy efficiency, can be achieved by decoupling supply and demand (Ahrens et al., 2021). Ice banks used as a cold thermal energy storage are common in a lot of dairies and provide a reduction in the necessary installed peak refrigeration capacity (Christensen and Kauffeld, 1998); (Finer et al., 1993); (Kumar et al., 2001); (Song et al., 2024). However, their working principle, with ice building up on coils to store thermal energy, leads to a reduced evaporation temperatures when the ice thickness is increasing, and therefore lower COPs (Finer et al., 1993); (Mađeri´ c et al., 2022). A reduced evaporation temperature can also be seen when looking at the scraped surface generator for ice slurry production, which has according to Kauffeld and Gund (Kauffeld and Gund, 2019) a 7 - 10 K lower evaporation temperature than the so called supercooling ice slurry production method. With the supercooling method the evaporation temperature stays nearly constant during slurry production, as shown in (B´ ed´ ecarrats et al., 2010), which results in a higher energy efficiency than the other cold thermal energy storage solutions discussed. 4.3. Hot water side The thermal demand of the hot water consumers (HW CIP process - HW07, HW CIP truck hall -HW-05, HW other consumers - HW-06) is shown in Fig. 5 during production days for August 2024 against December 2024. As it is not currently measured how large the utilised proportion of ‘HW overproduction/ preCIP’ is, it is excluded from Fig. 5. Throughout the night-time (24:00 – 4:00), the average heating demand is found to be between 5 – 35 kW during August 2024 and 9 – 30 kW in December 2024. At the start of the production day at 5:00, the average heating demand increased to 147 kW and 158 kW for August and December 2024, respectively. CIP and pre-washing activities prior to production are found to be responsible for the increase in heating demand. The distribution of the kernel density at 5:00 indicates a widespread distribution of data and upon further investigation, it can be concluded that there is no fixed, but a day-by-day thermal energy demand pattern at the start of the production. Throughout working hours, the average hot water consumer thermal demand increases and peaks at 16:00 with 202 kW in August 2024 and with 192 kW at 17:00 in December 2024. During 12:00 – 19:00, the distribution of hot water consumer heating demand is widespread, with peaks higher than 600 kW, but with the highest kernel density estimation of heating demands below 100 kW. Hence, the distribution indicates a dependency on the produced products and their respective heating and cleaning demands. From 17:00, the average heating demand decreases until nighttime. A comparison between August and December 2024 shows that the daily heating demand pattern of the hot water consumers closely align. Hence, the seasonal influence is minor, and the thermal energy pattern and consumption is dependent on the produced products. That can also be seen when comparing the average energy consumption during production days for hot water with 2506 kWh in August and 2753 kWh in December 2024. The specific monthly energy consumption of the hot water heating demand of 32.7 Wh/l in August and 31.1 Wh/l in December 2024 underlines these findings. Improvement opportunities hot water side As the HW consumers are CIP processes, manual rinsing and other consumers, it is difficult to assess to what extent staff training or Table 2 Monthly cooling COPs of the 80 kW CO 2 units 3 & 4 in 2024. COP cooling Month CO 2 Unit 3 CO 2 Unit 4 January 1.85 1.77 February 1.84 1.77 March 1.82 1.77 April 1.72 1.73 May 1.65 1.82 June 1.39 1.62 July 1.65 2.45 August 1.60 2.51 September 1.57 2.49 October 1.63 2.46 November 1.61 2.43 December 1.53 2.48 Fig. 5. Seasonal influence on the hot water consumer thermal demand. A comparison between August (summer) and December (winter) in 2024 displayed for all working days in the month as a half violin plot over the time of the day. L. K¨ OSTER et al. International Journal of Refrigeration 180 (2025) 564–571 569 automation can reduce HW consumption and is beyond the scope of this paper. It was shown that a change in the production schedule resulting in larger batch sizes can reduce the need for CIP and therefore water consumption. Although it is known that on some days, no HW goes directly into the drain as overproduction, there are also days, especially in summer, when more HW is available than required. Therefore, the possibility of a larger HW storage or a control system that dissipates the excess heat via the aircooled gas cooler of the CO 2 unit 5, when the HW storage is full, should be investigated. 4.4. Steam side From the thermal energy demand on the plant level in Fig. 2, it could be seen that more than 50 % of the total heating demand was provided by steam in 2024 (under the assumption of a electricity to heat conversion of one). Furthermore, 35 – 40 % of the total electricity energy demand is consumed by the electric boiler. Therefore, the steam demand during working days and the seasonal influence on it are shown in Fig. 6. It presents the comparison between the steam demand in August (summer) and December (winter) for all measured values (half violin plot) and the hourly average. During night hours, the thermal energy consumption of the steam boiler averages for both cases between 33 kW and 37 kW. Between 03:00 and 05:00, the pressure in the boiler is increased from 4 to 6 bar to its operational pressure to start with sterilisation and disinfection of processing equipment from 05:00 to 06:00. This leads to the highest peak in the hourly average steam demand. In August 2024, this peak occurs at 06:00 with an average value of 246 kW and a maximum value of 328 kW. In December 2024, the steam boiler started up on average one hour earlier, with an average and maximum steam demand of 232 kW and 334 kW, respectively. During the production day, the steam demand stayed constantly high, with average values around 200 kW, where the maximum measured value in August occurred at 16:00 with 427 kW and in December at 13:00 with 449 kW. After that, the steam demand is continuously decreasing until it reaches its baseload of 35.8 kW on average, outside of production hours. The average and maximal daily thermal energy consumption during production days for the steam demand in August 2024 was 2979 kWh and 3929 kWh, respectively. For December 2024, the average and maximal energy consumption was slightly higher with 3318 kWh and 3983 kWh. Improvement opportunities for thermal demands with temperatures larger than 70 ◦C As presented in Section 4.1, the steam boiler is responsible for 35 – 40 % of the dairy’s electricity consumption. In combination with the thermal efficiency of the steam boiler of close to 1, there is a strong case for energy efficiency improvements. Under the premise that heat sources are available, (high) temperature heat pumps (HTHP) provide efficient solutions to replace the steam boiler. Possible heat sources are excess hot water from overproduction of the CO2 systems and greywater after CIP. Development in the HTHP sector is highly dynamic and innovative. The comprehensive report by (Zühlsdorf, 2023) collected manufacturers data on their HTHP products, including temperature and technical readiness level (TRL) levels and maximum supply temperature. There, it shows that especially for a maximum supply temperature of below 110 ◦C, systems at high TRL level are commercially available. (Zühlsdorf, 2023) (Ahrens et al., 2021) developed an integrated energy system for a dairy for covering the thermal cooling and heating demands at various temperature levels. The paper reported an overall system COPs of 4.1, for a limited period of one week. During the investigated period, the ammonia/water heat pump achieved a COP of 5.9 for heat source temperatures of 67/60 ◦C and heat sink temperatures of 73/95 ◦C. (Ahrens et al., 2021) The choice of system and energy efficiency is highly dependent on the available heat source, and therefore, generalised statements are difficult to make. However, similar applications have shown heating COPs of 2.2 to 2.9 (Zühlsdorf, 2023), and therefore, the integration of a HTHP is regarded as a strong case for energy efficiency and helps to cut GHG emissions further, especially when switching from fossil fuel steam generation to HTHP. 5. Conclusion This study presented the energy consumption of a Norwegian dairy plant for the year 2024. The thermal energy system consisted of parallel integrated CO 2 refrigeration systems, which cover the cooling and hot water demands. A steam boiler with a capacity of 500 kW el covered heating demands for temperatures above 65 ◦C, namely pasteurisation, sterilisation and cleaning in place (CIP). The specific cooling energy consumption was found between 27.5 Wh/l in April and 43.0 Wh/l in July 2024, while the specific hot water heating energy consumption varied between 34.3 Wh/l in December 2024 and 56.0 Wh/l in July 2024. The specific steam consumption was found at 38.3 Wh/l in October 2024 and 48.4 Wh/l in July 2024. A baseload of the steam boiler was present at a mean value of 35.8 kW el due to heat losses. Overall, the steam boiler was responsible for 36.4 % of the plants electricity consumption in 2024. Daily thermal energy patterns (working days) were obtained for the cooling demand, hot water demand and steam demand and the distinctive patterns analysed for August and Fig. 6. Seasonal influence on the steam demand from the electric steam boiler. A comparison between August (summer) and December (winter) in 2024 displayed for all working days in the month as a half violin plot over the time of the day. L. K¨ OSTER et al. International Journal of Refrigeration 180 (2025) 564–571 570 December 2024. The data on which the study is based is available in Zenodo (see DOI: 10.5281/zenodo.15111145) to fill the gap in realworld data. Based on the overall specific energy consumption and the daily energy patterns, the potential for improvements in the dairy was shown. The implementation of a high temperature heat pump was identified as a viable case for a reduction in energy consumption, under the premise that a heat source is available. Furthermore, the integration of a cold thermal energy storage as a system to mitigate peak cooling demands was found as an effective system improvement, which can in addition provide greater flexibility to the plant operators. Funding sources The activity described in this manuscript has been performed within the project ENOUGH. ENOUGH has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 101036588. Data statement Data utilised within the presented figures will be made available in the Zenodo ENOUGH community under the DOI: 10.5281/ zenodo.15111145. CRediT authorship contribution statement Lukas K¨ OSTER: Writing – original draft, Methodology, Data curation, Conceptualization. Jan BENGSCH: Writing – original draft, Methodology, Data curation, Conceptualization. Kristina Norne WIDELL: Writing – review & editing, Project administration, Funding acquisition. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. The author is an Editorial Board Member/Editor-in-Chief/Associate Editor/Guest Editor for this journal and was not involved in the editorial review or the decision to publish this article. References Ahrens, M.U., Foslie, S.S., Moen, O.M., Bantle, M., Eikevik, T.M., 2021. Integrated high temperature heat pumps and thermal storage tanks for combined heating and cooling in the industry. Appl. Therm. Eng. 189. https://doi.org/10.1016/j. applthermaleng.2021.116731. May. B´ ed´ ecarrats, J.-P., David, T., Castaing-Lasvignottes, J., 2010. Ice slurry production using supercooling phenomenon. Int. J. Refrig. 33 (1), 196–204. https://doi.org/10.1016/ j.ijrefrig.2009.08.012. Jan. Bell.I.H. , J. Wronski, S. Quoilin, and V. Lemort, ‘Pure and pseudo-pure fluid thermophysical property evaluation and the open-source thermophysical property library CoolProp’, vol. 53, no. 6, pp. 2498–2508, 2014, doi: 10.1021/ie4033999. Bengsch.J. , L. K¨ oster, E.S. Svendsen, H. Selvnes, and K.N. Widell, ‘Sizing and optimization of a cold thermal energy storage (CTES) for a dairy: a case study’, Jun. 2024. doi: 10.18462/iir.iccc2024.1060. Bengsch.J. , et al., ‘Integriertes CO2k¨ alteanlagensystem einer Molkerei: energieflussanalyse und potenzial für einen kalten thermischen energiespeicher’, 2023. Briam, R., Walker, M.E., Masanet, E., 2015. A comparison of product-based energy intensity metrics for cheese and whey processing. J. Food Eng. 151, 25–33. https:// doi.org/10.1016/j.jfoodeng.2014.11.011. Apr. Christensen.K.G., and M. Kauffeld, ‘Ice slurry accumulation’, 1998. Egas, D., Pons´ a, S., Llenas, L., Col´ on, J., 2021. Towards energy-efficient small dairy production systems: an environmental and economic assessment. Sustain. Prod. Consum. 28, 39–51. https://doi.org/10.1016/j.spc.2021.03.021. Oct. European Commission, ‘Stepping up Europe’s 2030 climate ambition - investing in a climate-neutral future for the benefit of our people’, 2020. Accessed: Jan. 16, 2023. [Online]. Available: https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CE LEX:52020DC0562&from=EN. Finer, S.I., Cleland, A.C., Lovatt, S.J., 1993. Simple mathematical model for predicting the transient behaviour of an ice-bank system. Int. J. Refrig. 16 (5), 312–320. https://doi.org/10.1016/0140-7007(93)90003-Q. Jan. Incropera, F.P., Dewitt, D.P., Bergman, Lavine, A.S., 2017. Incropera’s principles of heat and mass transfer, 8th, global edition edns. John Wiley & Sons. Jesper, M., Pag, F., Vajen, K., Jordan, U., 2021. Annual industrial and commercial heat load profiles: modeling based on k-means clustering and regression analysis. Energy Convers. Manag. 10, 100085. https://doi.org/10.1016/j.ecmx.2021.100085. Jun. Kauffeld, M., Gund, S., 2019. Ice slurry – history, current technologies and future developments’. Int. J. Refrig. 99, 264–271. https://doi.org/10.1016/j. ijrefrig.2019.01.010. Mar. K¨ oster.L. , J. Bengsch, E.S. Svendsen, K.N. Widell, and S. Jenssen, ‘Increasing the energy efficiency of a dairy by implementing a high temperature heat pump – a theoretical evaluation.’, Jun. 2024, doi: 10.18462/IIR.ICCC2024.1061. Kumar, P.R., Balaji, S.R., Lal, D.M., Kalanidhi, A., 2001. Energy optimisation studies in a dairy industry ice bank tank. Int. J. Ambient Energy 22 (4), 181–188. https://doi. org/10.1080/01430750.2001.9674857. Oct. Ladha-Sabur, A., Bakalis, S., Fryer, P.J., Lopez-Quiroga, E., 2019. Mapping energy consumption in food manufacturing. Trends Food Sci. Technol. 86, 270–280. https://doi.org/10.1016/j.tifs.2019.02.034. Apr. Mađeri´ c, D., Pavkovi´ c, B., Delaˇ c, B., ˇ Carija, Z., 2022. Impact of the pipe row spacing on the capacity of ice bank formed in a volume-limited water bath. Therm. Sci. Eng. Prog. 30, 101254. https://doi.org/10.1016/j.tsep.2022.101254. May. Mehta, S., et al., 2024. A comparative assessment of food chain emissions from Norway, UK, Germany and Italy. In: Presented at the 8th International Conference on Sustainability and the Cold Chain, Tokyo, Japan. https://doi.org/10.18462/iir. iccc2024.1100. Jun. Miserocchi, L., Franco, A., Testi, D., 2024. A novel approach to energy management in the dairy industry using performance indicators and load profiles: application to a cheese dairy plant in Tuscany, Italy’. Energy 310, 133240. https://doi.org/10.1016/ j.energy.2024.133240. Nov. Murphy, M.D., Upton, J., O’Mahony, M.J., 2013. Rapid milk cooling control with varying water and energy consumption. Biosyst. Eng. 116 (1), 15–22. https://doi.org/ 10.1016/j.biosystemseng.2013.05.007. Sep. OECD and Food, 2022. Agriculture organization of the united nations, OECD-FAO agricultural outlook 2022-2031. OECD-FAO Agricultural Outlook. OECD. https://doi. org/10.1787/f1b0b29c-en. Sandhaas, A., Kim, H., Hartmann, N., 2022. Methodology for generating synthetic load profiles for different industry types. Energies 15 (10), 3683. https://doi.org/ 10.3390/en15103683. May. Schlemminger,C. , M. Bantle, S. Jenssen, M. Dallai, and M. Dorin, ‘Industrial high temperature heat pump for simultaneous production of ice-water and process-heat’, 2022, doi: 10.18462/iir.gl2022.166. Selvnes, H., et al., 2023. Integrated CO 2 refrigeration and heat pump system for a dairy plant: energy analysis and potential for cold thermal energy storage. In: Proceedings of the 26th IIR International Congress of Refrigeration, Paris, France: IIR. Augd.o.i: 1.0.18462/iir.10.18462/iir.icr.2023.0435. Selvnes, H., et al., 2022. Integrated CO2 refrigeration and heat pump systems for dairies. In: 15th IIR-Gustav Lorentzen Conference on Natural Refrigerants (GL2022). Proceedings. Trondheim, Norway, June 13-15th 2022., Trondheim, Norway: IIR. https://doi.org/10.18462/iir.gl2022.53. Jun. Song, H., Verdin, P.G., Zhang, J., 2024. Research developments and applications of ice slurry. Energies 17 (20), 5213. https://doi.org/10.3390/en17205213. Oct. Üçtu˘ g, F.G., 2019. The environmental life cycle assessment of dairy products. Food Eng. Rev. 11 (2), 104–121. https://doi.org/10.1007/s12393-019-9187-4. Jun. Zühlsdorf, B., 2023. High-Temperature Heat Pumps: Task 1–Technologies. Annex 58 About High-Temperature Heat Pump. Heat Pump Centre c/o RISE – Research Institutes of Sweden. L. K¨ OSTER et al. International Journal of Refrigeration 180 (2025) 564–571 571