Finding the Right Natural Refrigerant System for Industrial Refrigeration

Inexpensive and highly efficient, ammonia (NH₃) has for decades dominated  the  industrial refrigeration landscape. But the refrigerant’s position has evolved in recent years.

For one thing, ammonia’s unavoidable toxicity has led to an industry-wide effort to develop systems that use far less charge than traditional liquid-overfeed systems. This has resulted in the proliferation of low-charge technologies, from central DX and packaged units to systems that combine ammonia and CO₂. More recently, transcritical CO₂ systems, which have been used predominantly in food retail, have begun to emerge in industrial facilities as an option that removes ammonia altogether.

So now two natural refrigerants, ammonia and CO₂, or a combination of both, are competing for space in cold storage and industrial plants. Any of them is superior to synthetic refrigerants like R22 or HFCs, but which of the natural options is best for a given application?

One of the more popular industrial options to emerge in recent years is the NewTon ammonia/CO₂ secondary packaged system, produced by Japanese OEM Mayekawa. Mayekawa said last year that it expected to sell 330 sets of NewTons during its current fiscal year, which runs from April 2019 to March 2020. (The number of sets is equal to the number of compressors used; some NewTon units contain more than one compressor.) This would bring the total number of sets sold past 2,000 (95% in Japan) since the system was introduced in 2008.

The first Mayekawa NewTon system in Australia was installed and commissioned last July in a new cold-storage distribution center in Western Sydney; the work was handled by Tri Tech Refrigeration Australia, Sydney, Australia. The project encompassed five NewTons: three medium-temperature NewTon-C units supplying 708kW (201.3TR) of cooling capacity; and two low-temperature NewTon R6000 units providing 198kW (56.3TR) of capacity. The combined cold-storage area is 44,094m³ (1,557,165ft³).

Mack Hajjar, Projects Engineer for Tri Tech, noted last October that “based on energy usage data obtained so far, overall typical compressor power consumption is considerably lower than a conventional two-stage pumped [liquid-overfeed] ammonia plant servicing an industrial cold store of a similar size.”

Hajjar’s energy assessment was challenged by Stefan Jensen, Managing Director, Scantec Refrigeration Technologies, Brisbane, Australia, who designed and constructed what he calls the first stick- built centralized low-charge ammonia system “of the modern era” in 2012-2013, in Perth, Australia. That system proved to be 67% more efficient than a comparable R404A air-cooled system in the same area belonging to the same end user, he said.

In a letter to Accelerate Magazine (January 2020), Jensen cited a paper “Comparing Energy Consumption and Life Cycle Costs of Industrial Size Refrigeration Systems,” authored by T. Lund, M. Skovrup and M. Holst of Danfoss, and presented in August 2019. “In all three jurisdictions investigated (Rome, Frankfurt and Oslo), the R744/R717 (CO2/NH3) cascade system is less energy efficient than the equivalent dual-stage R717 system with liquid overfeed.” For example, the Danfoss study found COPs of 4.35, 4.15 and 3.60 for two-stage liquid-overfeed ammonia systems for the highest loads examined (900kW/256TR low temperature; 2,700kW/768TR medium temperature) in Oslo, Frankfurt and Rome, respectively; by contrast the comparable ammonia/CO2 cascade systems had COPs of 3.92, 3.76 and 3.35.

Jensen’s favorite energy metric is specific energy consumption (SEC) in kWh/m³*a (m³ for refrigerated space, a for annum). He pointed out that according to a paper authored by Mayekawa researchers H. Asano and N. Mugabi, the best SEC achieved by NewTon systems is approximately 35kWh/m³ for a distribution center with a refrigerated volume of 44,000m³(1,553,845ft³). The paper, “Actual Energy Conservations by Using NH3/CO2 Refrigeration System,” was published in the proceedings of the 3rd International Conference of Saving Energy in Refrigeration and Air Conditioning (ICSERA), held in 2013.

Jensen compared the SEC for the NewTon with that of a low-charge DX two-stage ammonia system. According to ICR 2019 paper #466, “Real Energy Efficiency of DX NH3 versus HFC,” authored by Jensen, a centralized DX NH3 system delivers an SEC of 20-24 kWh/m³*a when servicing a distribution center with a refrigerated volume of 44,000 m³, compared to the 35kWh/m³*a for the same volume cited by the Mayekawa researchers.

Jensen doesn’t believe in ammonia-inventory reduction at all costs. “The art is to reduce the ammonia inventory and at the same time improve energy efficiency compared with a stick-built ammonia liquid overfeed system and transcritical CO2,” he said. “This is an art that isn’t mastered by many, but [ammonia’s] transport and thermodynamic properties make this a task that is nevertheless achievable.”

Energy-efficiency improvement “is the path to long-term survival of the ammonia industry as a whole,” Jensen said. “Nothing will match the energy performance of an ammonia refrigeration system done well. The industry has to start doing these systems well or lose significant market share.”

But he acknowledges that some operators are willing to pay an energy penalty to keep ammonia out of the refrigerated space.

To do an ammonia system well, Jensen said, equipment suppliers need to under- stand “how ammonia behaves during evaporation in tubes and what causes that behavior.” In addition, contractors need to know “what determines the optimum operating envelope for evaporators, and specify the correct design parameters to suppliers.”

Another important design feature needed to maximize energy efficiency in low-charge ammonia refrigeration systems is to minimize or eliminate the presence of liquid in suction lines during all operating conditions. “This is exactly what a centralized, low-charge ammonia system does,” Jensen said.

Jensen pointed to the potential risk asso- ciated with ammonia/CO₂ cascade heat exchangers: If the ammonia and CO₂ mix, they form ammonium carbamate, which seriously impacts operation. In one case Jensen cited, a system was shut down for four months, and repair costs were half the original cost of the system. In the U.S., “insurance costs for NH₃/CO₂ cascade are escalating and driving users towards centralized low-charge ammonia.”

Supporting Energy Claims for NewTon

Tri Tech’s Hajjar agrees that a two-stage DX ammonia system is, in principle, more efficient than its ammonia/CO₂ cascade counterpart.

Ammonia DX systems have a number of advantages, Hajjar noted, including no liquid-ammonia pumping requirement, high-efficiency, speed-controlled reciprocating compressors, no high- or low-side heat exchangers, no wet-return losses, no suction superheat and no (or less) hot-gas relief load if defrost drainers are used.

An end user who uses reticulated subcritical CO₂ and/or water-cooled condensing with cooling towers, like a NewTon system, is “deliberately sacrificing energy efficiency for increased safety and in some cases [reduced] cost,” said Hajjar. “In our [NewTon] application, the question was not whether to use a secondary refrigerant, but which one and how – DX or pumped ammonia were definitely out of the question.”

He stands by his previous comments that the NewTon’s energy consumption is considerably lower than that of a conventional two- stage pumped-ammonia liquid-overfeed system when comparing similar-size facilities.

But that estimate was low because in spring/summer the power consump- tion would be higher, he acknowledged. “Realistically, power consumption would be closer to 1.5GWh p.a.” This translates to 34.09kWh/m³*a for a 44,000m³ (1,553,845ft³). facility, which is consistent with what Mayekawa researchers Mugabi and Asano predicted – 35 kWh/ m³*a for NewTon in a 44,000m³ space.

By contrast, if a site like the one in Australia with the first NewTon units were to instead use conventional two-stage pumped ammonia, “I would expect energy usage of 3GWh (68kWh/m³*a),” said Hajjar. He referred to figure 8 in Jensen’s paper, “Real energy efficiency of DX NH3 versus HFC,” where “the bulk of conventional two-stage pumped ammonia systems consume 40-70 kWh/m³*a.”

Hajjar also noted that SEC, as a performance metric, “doesn’t normalize for proportion of low-temperature/medium-temperature, door usage, geographic location, facility layout, etc. Realistically it is more of a general guide for efficiency comparisons.”

It should also be noted, said Haggar, that the NewTon ammonia system is not a conventional cascade system, because it uses CO₂ as a secondary fluid only. “It relies on a more efficient ammonia vapor-compression cycle for refrigeration, instead of compressing and reticulating CO₂ in a vapor-compression cycle on the low stage.”

“So while the NewTon NH₃/CO₂ secondary system sacrifices COP due to their hot- and cold-side heat exchangers, done principally to make the plant safer, there are many features that compensate for this, making the NewTon system very efficient, nonetheless,” continued Hajjar. “As a result, it is important to consider the configuration of ammonia/CO₂ cascades when making comparisons, and so the COP comparisons made in the Danfoss study, for example, may not always apply.”

The NewTon’s other energy-enhancing-features, he said, are as follows:

• The medium- temperature and low-temperature NewTon systems are independent. Therefore the effective medium-temperature suction pressure and low- temperature interstage pressures are independent, and can be therefore optimized for each respec- tive system. By contrast, in a two-stage system, the medium-temperature suction and low-temperature interstage must be common, and under certain circum- stances not optimized for either stage, or optimized for one stage but not the other.

• The medium-temperature system uses a single-stage economized cycle. The low-temperature system uses a two-stage cycle with both its respective high- and low-stage cycles economized. This is inherently more efficient than a single-stage medium-temperature cycle providing intercooling for a single-stage low-temperature cycle.

• The NewTon system does not use hot-gas defrost. A conventional pumped ammonia system uses hot-gas defrost, which relieves a considerable amount of hot gas to suction, thereby increasing energy consumption; with some even relieving to the low-temperature stage, making the  system  less  efficient.  It is possible to use defrost drainers to mitigate hot-gas relief loads and return only liquid to the medium-temperature vessel, but “conventional” pumped-ammonia systems don’t always do this.

• The NewTons have permanent magnet motors, which are high-efficiency motors.

• NewTon systems don’t have a tendency to draw air and moisture in at negative pressures like conventional low-tempera- ture ammonia open-drive compressor systems, because NewTons don’t have shaft seals. Over time, energy penalties due to trapped non-condensables and water accumulation, which appear in conventional ammonia systems, may not manifest themselves in NewTon plants.

CO₂  vs. low-charge ammonia

While some operators are reducing ammonia charge in low- charge or ammonia/CO₂ systems, a growing number of industrial facilities are opting for a “no-charge” ammonia system, that is, transcritical CO₂. According to Terry Chapp, Principal, Collaborative Solutions, Terry Chapp & Associates, CO₂ systems are better suited for industrial systems with smaller capacity – from 50TR (176kW) to 300TR (1,055kW)  –  than ammonia systems, which can cover up to 2,500TR (8,792kW). Jensen acknowl- edged that for smaller warehouses – under 5,000 -10,000m³ (176,573 – 353,147ft³) in refrigerated volume – the additional cost of a centralized low-charge ammonia system compared to a transcritical CO₂ system would require “too long a return on investment.”

CO₂ systems are still regarded as working better in relatively cooler climates, though optimized systems have improved efficiencies in warmer climates.

One CO₂ operator in the U.S. is Henningsen Cold Storage, based in Hillsboro, Oregon (U.S), which, in June 2018, installed its first transcritical CO2 system – one of the first industrial transcritical systems in the U.S. – at an 111.000ft² (10,312m²) facility in Grandview, Washington (U.S.). It supplies 187TR (658kW) of refrigeration capacity, 157 (552) for a freezer and 30 (106) for a dock area.

Henningsen, which runs 12 cold-storage warehouses, mostly in the Pacific Northwest part of the U.S., is one of the most efficient operators in the North American refrigerated warehouse industry. Its facilities, including ammonia and CO₂ systems, average 0.569 kWh/ ft³*a (20.3kWh/m³*a), compared to  an industry average of 1.13 (40.4). (Henningsen is measuring the energy consumption of the entire facility, not just the refrigerated space.) Under the direction of Pete Lepschat, director of engineering, Henningsen has built several low-charge central ammonia systems that use substantially less ammonia than the company’s traditional plant.

In contrast to Jensen’s DX ammonia systems, Henningsen uses a “minimalist overfeed model,” which Lepschat believes saves more money. At the company’s first warehouse in Salem, Oregon, built in 2013-2014, he installed Evapco evaporator coils with a 1.2-1 overfeed ratio (for every 1.2 part of liquid ammonia, one part evaporates), compared to the 3-1 ratio at a traditional Henningsen facility.

Another technique Henningsen employs to cut ammonia charge is to use what Lepschat calls a “pump-down system” in lieu of a high-pressure receiver. The system features a small vessel that maintains a low-level of ammonia  and is continuously feeding the gas to a low-pressure receiver via a subcooler.

In its second  low-charge  warehouse in Salem, built in 2017,  Henningsen was able to reduce SEC to 0.288kWh/ ft³*a (10.3kWh/m³*a), the lowest in the company, and one of the lowest in the cold-storage industry. This became the standard for all future facilities, including the first CO₂ warehouse in Grandview. (Henningsen converted another warehouse, in Scranton, Pennsylvania (U.S.), to transcritical CO₂ in 2018.)

The installed cost of the Grandview CO₂ system was $534,000 less than the Salem ammonia system; between $200,000 and $400,000 was saved by not building a machine room, which also cut 5-6 weeks in installation time. The reliability of the CO₂ system and low- charge-ammonia standard has been the same. But on an energy basis, the Grandview CO₂ system, which uses an adiabatic gas cooler, had much higher energy costs than the standard facility’s system, though that was before several energy-efficiency measures were added. Those included: dock dehumidification, evaporator/condenser fan control, hot-gas defrost, gas cooler optimization, compressor VFDs and a glycol pump VFD. In the subsequent commissioning of the system, the low-temperature suction setpoint was changed from -25°F (-32°C) to -16°F (-27°C). In addition, evaporator and gas cooler fan speeds were changed from fixed to modulating.

In Lepschat’s most recent data, which covered January 2019 through January 2020, the Grandview CO₂ system has an SEC of 0.384kWh/ft3*yr (13.7kWh/m³*a), well below the average for Henningsen’s facilities (which is itself considerably below the industry average) but still 25% above the standard Salem number. But Henningsen is working with system manufacturer Carnot Refrigeration to improve rack efficiency, Lepschat noted. In addition, he aims to improve his fan control systems. With those changes, “I think we will either match Salem [on energy consumption] or actually do better.”

In other areas, the CO₂ system is saving Henningsen money compared to the standard low-charge ammonia plant in Salem. For example, the CO₂ system uses far less water in its adiabatic gas cooler, which also reduces the sewer bill, and it requires no water treatment. In addition, an ammonia facility requires a full-time engineer to keep an eye on it, while a CO₂ plant does not. Maintenance on the CO₂ system takes place just twice annually, and generally involves minor upgrades “like changing the oil on your car,” said Lepschat. And a CO₂ facility can dispense with U.S. federal regulatory updates for PSM (process safety management) and RMP (risk management plan) that are needed at an ammonia warehouse; Lepschat estimates he saved “six figures” at the CO₂ facility by not having to develop an initial PSM/RMP plan.

In the event that energy parity with Salem ‘s standard is approached or matched, the savings on other factors would result in operating costs at the Grandview facility that are up to 20% better than Salem’s, said Lepschat.

Another factor favoring transcritical CO₂ is heat recovery. “CO₂ has a pretty big advantage in this area,” said Lepschat. This is because the sensible and latent portions of the discharge gas are almost the inverse of those for NH3, that is, 70% of the rejected heat is sensible and 30% latent for CO₂, as opposed to 30% sensible & 70% latent for NH₃, he explained. That means a CO₂ system is better for generating hot water and space heating than an ammonia system.

At the Grandview facility, heat reclaim is used to warm the floor under the freezer and to minimize defrost in the freezer by getting rid of moisture in the dock area. The system defrosts with hot gas.

Another consideration is the  number of compressors used. At Grandview, Henningsen employs 15 semi-hermetic reciprocating compressors in its CO₂ system. “So if one goes down, we have 14 to back it up,” Lepschat said, and replacement costs are not high. “Redundancy is built in.” By contrast, an ammonia system may use one large screw compressor, with one expensive replacement unit.

Jensen said that some CO₂ systems may have better annualized energy performance than large-scale centralized ammonia systems with liquid overfeed because these CO₂ systems employ multiple small reciprocating compressors and have very little or no liquid in the suction lines.

Given everything he knows now, Lepschat was asked which refrigerant he would lean toward for future plants, ammonia or CO₂? “I never say never, but as of today I see no advantage to building an ammonia refrigerated facility.”