THE INSTITUTE OF REFRIGERATION

Advance Proof. Private to members Copyright 2007 The Institute of Refrigeration No publication or reprinting without authority THE INSTITUTE OF REFRIGERATION CARBON DIOXIDE FOR SUPERMARKETS by Campbell, A. 1,2, Missenden, J.F. 2,. and Maidment, G.G. 2 1 Tesco Stores Ltd. Cirrus Building B Shire Park Welwyn Garden City Al7 1AB 2 London South Bank University, 103 Borough Road, London, SE1 OAA, 02078157626, maidmegg@lsbu.ac.uk (Session 2006-2007) To be presented before the Institute of Refrigeration at the Institute of Marine Engineering, Science and Technology, 80 Coleman Street, London EC1 on 5th April 2007 at 5.45pm INTRODUCTION The effects of refrigerants on the environment are well documented. Supermarkets contribute both directly and indirectly to global warming. Directly, greenhouse gas emissions occur through the leakage of HFC refrigerants used in refrigeration systems for refrigeration of food. These refrigerants have very high global warming potential with GWP s in excess of 1000. However, supermarkets also indirectly produce CO 2 as they are large consumers of electricity, consuming as much as 15 000 GWh (54 000TJ) per annum and approximately 40% of this is consumed by refrigeration equipment. Over the last 20 years, legislation has prohibited the use of many ozone-depleting refrigerants including CFCs and HCFCs, however, the use of HFC refrigerants is still legal and commonplace. In recent years natural refrigerants have been proposed as an environmentally friendly solution for the refrigeration industry. These refrigerants which include ammonia, hydrocarbons and carbon dioxide, do not contribute to ozone depletion and have low global warming potentials. Carbon dioxide offers a long-term solution suitable for many applications in refrigeration and heating, from domestic applications utilizing heat pumps, to providing hot water and heating to commercial applications for supermarket refrigeration. The technology is future proof requiring no further retrofitting to the next refrigerant solution promoted commercially. To investigate the performance of R-744 refrigeration, a prototype system was constructed and tested to demonstrate proof of principle. This Proc. Inst. R. 2006-07. 7-1

73,6 +31 Deg.C Supercritical Critical point Liquid -10 o C 26 Bar Bar 5,2 Solid Solid -Liquid -28 o C 11Bar Liquid - vapour - 56,6 Deg.C Triple point Solid - Vapour Vapour - 78,4 Deg.C Enthalpy (kj/kg) Figure 1. Simplified P-H Diagram [1]. paper reports on a practical investigation of R-744 refrigeration in a commercial application and describes an investigation of its design, commissioning and energy performance. Also discussed are the practical experiences gained from building a R-744 system and the benefits of using such a natural refrigerant. The main characteristics of R-744 as a refrigerant are that the critical and triple points are relatively low at 31 o C (73 Bar) and -56.6 o C (5.2 Bar) respectively. In order to remain below the critical temperature of 31 o C it is necessary to use a cascade system to provide a heat sink to allow the CO 2 to condense. The system restrictions in relation to temperature and pressure can be seen below in the simplified P-H diagram R-744 operates at a far higher pressure than standard refrigerants although this is not excessively high compared to similar engineering applications. Figure 2 demonstrates the differences in operating pressures of standard refrigerants and R-744. It can be seen from the graph that while R-744 operates at higher pressures than conventional refrigerants, the pressures relative to those say of automobile power steering systems (115 Bar) are low although power steering utilizes liquid rather than a gas. Power Steering 115 Bar Figure 2. Comparison of refrigerant pressures. Proc. Inst. R. 2006-07. 7-2

1 2 R-404a COMPRESSOR 5 6 R-744 COMPRESSOR R-1270 CYCLE 8 7 3 PHE PULSE VESSEL R-744 H.T CYCLE- 7 o C R-744 PUMP 4 R-744 L.T. CYCLE Figure 3. A Typical cascade supermarket refrigeration system using R1270/R-744. The reasons for the renewed interests in CO 2 as a viable natural refrigerant is the advantages associated with the properties of CO 2 over other natural refrigerants. CO 2 has a high volumetric refrigeration capacity, as a result of its high vapour density when compared to other refrigerants. As a consequence, refrigeration compressors using CO 2 are 6-8 times smaller than those of R22 systems in displacement terms. Further benefits of high vapour density are realised when applied to pipe-work suction pipe work and valves can be reduced by several sizes. This will result in a significant reduction in material used, not only in pipe and component sizes but also pipe gauge. CO 2 also possess improved heat transfer properties allowing the evaporator to operate at a higher temperature than competitive R404a evaporators. Evaporators may work at least 2K higher than conventional R404a evaporators. These improved heat transfer properties result in increased capacity in standard heat exchangers. Application of R-744 in supermarkets The excellent safety and environmental properties associated with R-744 enable its systems to offer real advantages to the retail food industry. Its non-toxic and non-flammable nature is important because large refrigerant charges are used in supermarkets and much of the refrigerant runs in pipes through the customer area. Since the Montréal protocol there has been much discussion about the use of natural refrigerants using ammonia and hydrocarbons in supermarkets. Although these refrigerants offer environmental benefits they can t be used in large volumes in such public spaces. However a number of systems have been installed and tested using a central ammonia / hydrocarbon chiller and a secondary distribution system using Tyfoxit, propylene glycol and other heat transfer fluids. The disadvantages with these fluids is that the safe, non toxic, non flammable fluids are expensive in terms of revenue and capital to implement for a variety of reasons at low temperature. These include high viscosity, low heat transfer and poor thermophysical properties, as well as practical complications through corrosion and material compatibility. The use of CO 2 as a distribution fluid was highlighted by Pearson to the Institute in 1993[2]. The use of CO 2 in conjunction with a hydrocarbon or ammonia primary chiller presents the opportunity to refrigerate retail food areas with a low carbon, safe and low GWP combination of refrigerants. Low GWP is important because of the potential refrigerant leakage rates that are a characteristic of the complexity and size of these systems. This is shown in Figure 3, which shows a typical cascade CO2 system which has been reported used in supermarkets [3]. This can be seen to comprise of medium- and low-temperature sections serving chilled and frozen food chillers and freezers respectively. Proc. Inst. R. 2006-07. 7-3

Figure 4. PH chart showing operation of Cascade Refrigeration System. The ph chart in figure 4 shows the cascade subcritical cycle using a primary refrigerant together with a cascade CO 2 system. A number of these systems have now been tested and it is reported that energy savings of between 11% -15 [4] have been achieved compared with a typical R-404a reference cycle. In terms of carbon the savings will be greater since direct emissions will be several orders of magnitudes lower due to the low GWP of the natural refrigerants. Alternatively CO 2 has been used in a transcritical cycle in retail applications [5]. A number of systems have been reported and the conceptual schematic and PH chart are shown in figures 5 and 6. It can be seen that the high pressure condition is above the critical point and as such heat rejection occurs without change of phase in the gas cooler. One key advantage of this cycle is the availability of high temperature energy at the compressor PULSE VESSEL Figure 5. A schematic of a transcritical CO 2 refrigeration system. Proc. Inst. R. 2006-07. 7-4

Figure 6: A PH chart of a transcritical CO2 refrigeration system. discharge which offers potential for high grade heat recovery. One further advantage of the system is it s simplicity with a single refrigerant and compression cycle. However it can be seen that the intermediate gas is returned to the compressor suction and this re-expansion at low pressure presents an avoidable and significant reduction in COP. pulse vessel, which acts as liquid receiver and separation vessel. The high temperature R-744 evaporator is fed from the pulse vessel via a A Pilot Commercial System The R-744 systems that have been proposed and installed in supermarkets are generally industrial in nature and this is shown in figure 7. The consequence of this higher specification is capital cost. It has been estimated, that compared with conventional R-404a systems capital costs are in the region of 20% higher [6]. As a consequence initial work carried out has been to develop a commercial CO 2 system that is more economic, but retains safety and integrity features. This pilot system is shown in Figure 8. The pilot system was designed as a dual temperature cascade arrangement with refrigerant R-404A as the heat rejecting/ condensing medium for the R-744 two-stage cycle. This arrangement enables the R- 404a system to be a small compact unit, with minimal charge and leakage risk. The R-404A cycle evaporates at -15 o C and condenses at 36 o C and uses two scroll compressors. The evaporator is a plate heat exchanger, which acts as the condenser for the R-744 cycle. It condenses the discharge gas from the R-744 compressor serving the LT system and the flash gas from the wet suction on the high temperature (HT) R-744 evaporator at. The R-404A evaporator was sized to remove the heat from both the LT and HT R-744 cooling systems. At the HT condition the R-744 is condensed and held in the Figure 7. Danish Supermarket Pack Installed at ISO Roskilde Denmark. Figure 8. Pilot R-744 system. Proc. Inst. R. 2006-07. 7-5

Figure 9. CAD drawing of pilot commercial system. refrigerant pump, which feeds the refrigerant at a 4:1 circulation ratio. The R-744 at -10 o C acts as a volatile secondary refrigerant in respect to sensible and latent cooling making the evaporator more efficient. The refrigerant returns to the pulse vessel from the evaporator as a saturated liquid via the wet suction where it is re-condensed. The flash gas leaves the vessel and enters the condenser to liquefy and then return to the vessel at -10 o C. The low temperature liquid R-744 also leaves the pulse vessel at -10 o C and feeds its evaporator where it is expanded to boil at -28 o C, returning via the dry suction to the R-744 compressor, thus completing the cycle. A detailed as built drawing can be seen in figure 9 complete with control system. Performance of the Pilot Commercial System The performance of the pilot system was calculated and measured. The table below shows the calculations based on the performance expected by the manufacturers and the results gained from the readings on the test rig. The COP is the standard form, where COP = Q cooling / P input and the COP of the system is COSP = Q e M.T. + Q e L.T. P 404A + P cond fans + P pump +P CO2 Where Q e M.T. is the medium temperature load Q e L.T. is the low temperature load P 404a is the power input to the R404a compressors P cond fans is the power input to the condenser fans P pump is the power in to the refrigerant pump P CO2 is the power in put to the CO 2 compressor Using the various manufacturers software to establish the expected power input and performance parameters the design output is shown in table 1. This also shows the measured and computed Proc. Inst. R. 2006-07. 7-6

CoolPack CYCLE ANALYSIS : COMBINATION OF ONE-STAGE CYCLES > TWO-STAGE CASCADE SYSTEM, DX EVAPORATORS LOG(p),h-DIAGRAM T 4,HT : 33.6 [ C] T 2,HT : 55.7 [ C] T 5,HT : 33.6 [ C] Q 4-5,HT : 0.00 [kw] 5 HT 4HT Q C,HT : 40.5 [kw] m HT : 0.2624 [kg/s] T C,HT : 36.0 [ C] 2 HT 3 HT SUB-DIAGRAM WINDOWS T 4,LT : -12.0 [ C] T 5,LT : -12.0 [ C] Q 4-5,LT : 0.00 [kw] W HT : 12.35 [kw] T 2,LT : 33.4 [ C] 5 LT 4 LT Q C,LT : 11.6 [kw] T C,LT : -10.0 [ C] 2 LT Q E,HT : 29.6 [kw] T E,HT : -15.0 [ C] 7 HT 3 LT 1999-2001 Department of Mechanical Engineering Technical University of Denmark Version 1.46 TOOL C.10 6 HT X 6,HT : 0.40 [kg/kg] m LT : 0.03679 [kg/s] Q E,LT : 10.0 [kw] T E,LT :-28.0[ C] 6 LT X 6,LT : 0.12 [kg/kg] REFRIGERANT HT : R404A COP HT :2.399 COP* HT : 2.374 REFRIGERANT LT : R744 COP LT :6.349 COP* LT : 6.443 T 8,HT : -6.0 [ C] 8 HT 1 HT T 1,HT : -11.0 [ C] 7 LT W LT : 1.575 [kw] 9 T 8,LT : -20.0 [ C] 8 LT 1 HT T 1,LT : -20.0 [ C] η CARNOT,HT : 0.47 η CARNOT,LT : 0.47 Figure 10. Cycle Prediction for Low Temperature System from Cool Pack Software. ACTUAL design High Side High Side Copeland ZB45KCE-TFD x 2 Copeland ZB45KCE-TFD x 2 T c 36 C T c 36 C T e -15 C T e -15 C Q e 17.8kW Q e 15.0kW P 404A 7.0kW P 404A 8.0kW P cond fans 0.5kW P cond fans 1.7kW THR 24.8kW THR 23.0kW COP 404A 2.39 COP 404A 1.5 M.T. M.T. T e -10 C T e -10 C Q e 4.7kW Q e 4.7kW P pump 0.8kW P pump 1.0kW COP CO2 MT 5.7 COP CO2 MT 4.7 L.T. L.T. Bitzer 2KC-3.2 K Bitzer 2KC-3.2 K Tc -10 C Tc -10 C T e -28 C T e -28 C Q e 10.0kW Q e 10.0kW P CO2 1.8kW P CO2 1.5kW THR 11.8 THR 11.5 COP CO2 LT 5.6 COP CO2 LT 6.7 SYSTEM SYSTEM P total 10.1kW P total 12.2kW COP Total 1.5 COP Total 1.2 Table 1. Pilot system s recorded and design performance. Proc. Inst. R. 2006-07. 7-7

performance for the actual system under similar conditions. It can be seen from the comparison that there is reasonable agreement between design and measured performance. Discrepancy between the two is due to transient measurement conditions and manufactures tolerances. The calculated comparative COP for a similar R-404a system was below 1 and the relative improvement in performance of the CO 2 system is similar to that reported by others [4] This was achieved using a more commercial style of pack. Case Studies To reduce the carbon footprint of their stores, of which circa 40% is due to refrigeration, Tesco commissioned C0 2 installations at Swansea, Wick and Shrewsbury. The store at Swansea was located in Llansamlet in an urban environment and comprised 60,000ft 2 / 5574 m 2 of nett retail area, with 354 kw of chilled food and 57 kw of frozen food. The systems consisted of one sub-critical system serving 19.5 m (14.98kW) of frozen food cabinets and one transcritical system serving 22.5 m (38.5kW) of dairy and produce chilled food cabinets. The motivation for both schemes was to test their operational and energy performance in a live application, with a view to making a generic corporate system specification for future stores. The sub-critical system was chosen specifically as an energy efficient and low carbon alternative to conventional a R-404A single stage system. The transcritical system was chosen as a low carbon alternative to a HFC system. Although transcritical systems are normally designed to operate more efficiently with a heat reclaim facility, this system was installed without this, primarily to avoid the additional complexity associated with the control of heating and cooling output. This feature and its impact are subject to further study. Consequently from an energy perspective the transcritical system may be optimized for significant performance improvements. In addition the store also had four R-404a (2 x HT,2 x LT) single stage systems serving the remainder of the chilled and frozen food. Schematics for the sub-critical and transcritical systems are shown in figures 11 and 12. The sub-critical system in figure 11 consists of a two R-290 H.T. close coupled air cooled liquid chillers each with 2kg of refrigerant charge. Each pack includes brazed plate heat exchanger evaporators, R- 22 charged Danfoss TEV s, Bitzer semi hermetic reciprocating compressors and brazed plate heat exchanger condensers, connected to a dry air cooler. This configuration was chosen to optimise system output for refrigerant charge. The low temperature R-744 circuit for the DX frozen cabinets consists of Bitzer Octagon compressors specifically produced for sub-critical applications. The evaporator output was controlled by AKV electronic evaporation devices linked to a pressure transducer rather than Figure 11. Cascade system installed at Swansea. Proc. Inst. R. 2006-07. 7-8

PULSE VESSEL Figure 12. Transcritical System installed at Swansea. temperature probes. This is a critical part of the system performance, inherently complicated by the properties of C0 2 causing conventional expansion control to be unsuitable. The transcritical system consisted of two Dorin semi hermetic reciprocating compressors running with 2 pole motors at 2980rpm. The system was configured to run in trans and sub-critical operation depending on the ambient temperature and therefore condensing temperature. When the ambient is below 22 o C the system runs as a standard sub-critical refrigeration system when the ambient reaches 22 o C the system is driven into transcritical mode through the use of an automatic control valve. System Performance The systems have now been operating for one year, without failure. System performance has been compared against like for like HFC systems installed as a control. Figure 13 demonstrates a better COP COP R290/R-744 VER R-404a COP 1.5 1.45 1.4 1.35 1.3 1.25 1.2 1.15 1.1 1.05 Jun-06 Jul-06 Aug-06 Sep-06 R-404a R-744 Figure 13: Comparative measured COP of Cascade R-744 cycle and control R-404a. Proc. Inst. R. 2006-07. 7-9

for the R-744 cascade L.T. system against the HFC control. Observation from site showed this to be partly due to the reduced temperature lift associated with the improved heat transfer within the R-744 heat exchanger, and also reduced system pressure drop. It was observed on start up that the increase heat transfer resulted in ice build up on the food on the CO 2 cabinets only. This was rectified by raising the evap temp which consequently raised the COP. The data taken from the transcritical system needed further clarification and could not be reported at this point. Operational Factors Other observations relate to the operational characteristics associated with the systems. The compressors selected for the transcritical system used 2 pole motors which were particularly noisy, however subsequent versions have overcome this limitation. Also standard in store maintenance procedures have been successfully followed with all systems. Specifically and not previously covered in the literature is the in store cleaning of C0 2 cabinets using hot water at 60 C. This was successfully achieved because all shut off valves had been eliminated within the cabinet vicinity allowing safe expansion of the C0 2 back to the compressors. Overall the installation was simply and efficiently executed within normal Tesco construction time frames. Capital was greater for both systems, however we expect with economies of scale there to be no significant difference in system cost between R-744 and R-404a technology. Carbon savings In order to evaluate the potential carbon saving associated with direct and indirect emissions, system performance was evaluated based upon the Shrewsbury store. Comprising of 342 kw of H.T. and 69kW of L.T. load, this store will utilise 2 cascade systems for the whole store, using R-1270 as the primary refrigerant and R-744 as both the L.T. DX refrigerant and the H.T. secondary volatile refrigerant. When the store is complete the refrigeration system will be the largest supermarket natural refrigerant solution installed in the UK to date. In order to justify this system a TEWI investigation to compare the systems with a standard reference using typical industry performance parameters was carried out. The results of this investigation are shown in table 2. The key headline from this is the significant carbon reduction associated with the R-744 scheme which is due primarily to the avoidance of HFC emissions with a high GWP factor of 3750. Future Developments The potential for transcritical systems has not as yet been fully explored using monitored data. Also there is considerable potential for use of high grade waste heat to drive an integrated sorption cooling system and other heat demand applications. A hybrid system using either series compressors or intermediate feed would produce better COP for transcritical applications comprising of DX L.T. and flooded H.T. as can be seen from figure 14. Further work and modelling will be carried out using this system and the research will include investigations into using the rejected heat to provide heating and System and period of assessment Standard R-404a system TEWI 1 year Cascade system TEWI 1 year Difference between systems over 1 year CO2 1310.8 TONNES OF CO2 536.7 TONNES OF CO2 774.2 TONNES OF CO2 Standard R-404a system TEWI 10 year Cascade system TEWI 10 year Difference between systems over 10 year 13108.4 TONNES OF CO2 5366.8 TONNES OF CO2 7741.5 TONNES OF CO2 Based on 61% run time Table 2. Results of TEWI comparison of standard versus Natural cascade systems. Proc. Inst. R. 2006-07. 7-10

115 BAR AMBIENT ABOVE 31 oc GAS COOLER EXPANSION INTERMEDIATE PRESSURE 26 BAR PULSE VESSEL COMPRESSOR EXPANSION L.T. EVAPORATOR Figure 14. Future Transcritical Dual Temperature Pack. cooling for the store using LTHW and sorption cooling. Results will be reported at a later date. CONCLUSION The investigation has clearly demonstrated there are considerable energy savings to be made by using a R- 744 system. The calculations demonstrate that the R- 744 system reduces substantially CO 2 emissions when compared with traditional systems. The practical experience gained from building and commissioning R-744 systems has shown that there are no insurmountable problems with using R-744 as a refrigerant. The problems associated with the procurement of equipment will diminish with growth in the use of R-744. The benefits of R-744 are: Smaller compressors Smaller pipe work and valves Smaller heat exchangers Greater energy efficiency Environmentally more friendly Reduction in green house gas emissions Cheaper refrigerant With the imminent arrival of the F-gas regulations, the running costs of systems can be further reduced by the avoidance of inspections, with the use of R- 744. There is no reason to delay the design of new plants using R-744 as the components and technology are now available. The only hindrance to wider use is the unfamiliarity of designers with these types of systems and available trained staff to service and install these systems. References [1] Danfoss. CO 2 Refrigerant for industrial Refrigeration, Danfoss 2002 Arhus [2] Pearson S.F., Development of improved secondary refrigerants, Institute of Refrigeration 1993 [3] Van Riessen G., NH 3 /CO 2 Supermarket refrigeration systems with CO2 in the cooling and freezing section, 6 th IIR Gustav Lorentzen Natural Working Fluids Conference. Glasgow. 2004 [4] Pachai A.C., Experiences with CO 2 as a refrigerant in supermarkets, 6 th IIR Gustav Lorentzen Natural Working Fluids Conference. Glasgow. 2004 [5] Heinbokel B., Gernemann A., One year experiences with the first CO2-DX system for medium and low temperature refrigeration at Swiss megastore, C-dig meeting 22/23.09.05 Zurich [6] Girotto S., et al, Commercial refrigeration system using CO2 as the refrigerant, International journal of refrigeration 27 2004 717-723 Pearson A., Carbon Dioxide Short Course 6 th IIR Gustav Lorentzen Natural Working Fluids Conference. Glasgow. 2004 Acknowledgements The authors would like to thank Tesco Stores Limited for their permission to publish this paper. Proc. Inst. R. 2006-07. 7-11