Is the Era of Ammonia Liquid Overfeed Ending?

Stefan Jensen
In four years, it will be the centenary of the liquid overfeed patent issued to YORK Corporation. The liquid overfeed plant came into widespread use around the middle of the last century. This coincided with an upsurge in the consumption of frozen foods that led to the construction of very large freezing plants that warranted the practical introduction of the liquid overfeed concept.

During the 1950s and 1960s, energy efficiency and sustainability were not exactly hot topics. In developed countries, the emphasis then was on converting economies from being agriculturally based towards becoming industrialized and the lifting of middle-class living standards.

The refrigeration industry greats conducted extensive research within the field of ammonia refrigeration. Current industry handbooks still refer to academic research papers authored in the 1950s to the 1980s and these handbooks remain in widespread use within the ammonia refrigeration industry today.

Large ammonia inventories have been under increasing pressure almost throughout the entire world for the last two decades or so. There are various reasons for this, and these mostly relate to the regulatory environments of the individual jurisdictions.

Part of the industry response both in the United States but also in Europe has been the introduction of lower ammonia overfeed rates. Does this prevent elevated pressure drops in the suction line networks of liquid overfeed systems? Mathematically, it does. Commonly used correlations for two phase pressure drops indicate this is the case.

If a plant is designed for an average overfeed rate of 1.5 to 1 at full load, what happens when that plant operates at 20% load? Unless the overfeed ratios of individual evaporators are kept constant irrespective of load, then the average overfeed rate at 20% load will become 7.5 to 1.

This means that to realize any real energy performance benefits of low overfeed rates for a refrigerating plant, it becomes necessary to apply almost identical control efforts for individual evaporators as one would apply in a dry expansion system.

Refrigerant injection into evaporators of dry expansion NH3 plant is frequently controlled by motorized expansion valves employing the refrigerant wetness (quality) at the evaporator exit as a control signal. How often is this a design feature of practical low overfeed rate installations? Not very often at all – the costs are clearly a deterrent.

What happens if the refrigerant mass flow through an overfeed evaporator is regulated as a function of load? This is not a simple question. To answer that it is necessary to first understand what operating envelope the evaporator in question was designed for at the outset. In practical liquid overfeed plant design, determination of the optimal evaporator operating envelope is rarely considered in depth. This is because getting it a little wrong does not appear to cause any great problems during commissioning and indeed during operation. It gets cold.

Getting the optimization of the evaporator operating envelope wrong with a dry expansion evaporator can, however, cause significant evaporator performance deficiencies and other issues affecting plant components both upstream and downstream of the evaporator.

Within the context described above, low overfeed evaporators are not that different to dry expansion evaporators. The question is whether this is considered adequately during the design of a low overfeed plant. Are evaporator suppliers provided with the range of conditions that the evaporator will be subjected to throughout its working life or are suppliers being provided with one operating point for the selection/design?

In the last decade, there has been mounting evidence that the liquid overfeed concept prevents ammonia from being the best it can be in terms of energy efficiency. Some of this evidence is illustrated in Figure 1, which shows examples of specific energy consumption values in kWh.m-3.year-1 for mixed refrigerated warehouses as a function of refrigerated volume in m³.

Here it is important to note the relatively close cluster formed by the green dots as opposed to the significant spread between all the remaining dots. The refrigerating plants represented by the green dots all have one thing in common – there is no high-density, liquefied refrigerant present in the suction line network.

All other plants visualized are of the liquid overfeed type. Indeed, the plants represented by the yellow dots were all constructed between 1999 and 2013 for one major logistics operator in Australia, all included the latest energy efficiency measures of that era, and all have been subjected to extensive fine-tuning by the plant owner.

Other evidence is visualized in Figure 2. This illustrates two conceptually identical dual stage, central NH3 refrigerating plants both belonging to the same owner, both situated in the same geographic area, both managed by the same personnel, and both performing identical functions. Both plants employ reciprocating compressors and variable frequency drives throughout.

The smaller plant was constructed in 2010. The larger plant was constructed in 2018 to replace the older plant that had been outgrown by the transport company in question. The only differences are that one plant is dry expansion (DX) and the other employs liquid overfeed, the DX plant evaporators are specifically designed for that purpose and the DX plant condenser is marginally more oversized than the condenser of the liquid overfeed plant. The energy performance records in both cases cover minimum one year – for the older plant several years.

The “best practice” in this context is a polynomial regression analysis of the recorded SEC-values for a range of centralized, low charge NH3 refrigerating plant servicing mixed refrigerated warehouses across Australia. The best practice graph enables the comparison of SEC values for dissimilar refrigerated volumes.

The recorded energy performance penalty associated with the presence of high density, liquefied refrigerant in the suction network is in this practical example around (1-0.97/1.4)*100 ≈ 31%. Put another way, the liquid overfeed plant consumes 1.4/0.97=1.44 times more energy per unit refrigerated volume.

This difference is not explained in full by the minor conceptual differences between the two refrigerating plants. Rather, it is likely that the bulk of the energy performance difference is caused by the differences in refrigerant feed methods.

At the 2017 GCCA Expo in Chicago, the presentation “Low Charge ADX Ammonia” by Watters and Nelson highlighted similar energy performance improvements for centralized, DX NH3 refrigerating plant versus liquid overfeed. The improvement range presented was 18% to 38%, but there were conceptual differences in the practical plant comparisons made to enable postulation of this range.

Figure 3 compares the energy performance of a new 2,100,000 ft³ mixed warehouse with the energy performances of many North American warehouses employing the plant concepts as marked (red dots). The warehouse represented by the stars is serviced by a dual stage NH3 DX plant that also provides refrigeration capacity to blast freeze 300 metric tonnes of meat in cartons per week.

Again, a similar pattern is visible. Elimination of the liquefied refrigerant from the suction line network appears to deliver significant energy performance benefits.

The blue star represents actual electricity consumption records for the first four months of 2021. These are the warmer months in Australia. The green star represents storage only without blast freezing. This is a calculated correction based on the amount of product frozen.

Does this mean the end of an era for the liquid recirculation concept? To answer this question, it is important to quantify the energy performance penalty associated with mixing relatively high density liquefied refrigerant into the suction network of a large, expansive centralized ammonia refrigerating plant.

Based on the practical observation illustrated in Figure 2, it is postulated here that the energy performance penalty caused by liquid overfeed can be as high as 30%.

The facilities that the comparison in Figure 2 is based on are singlestory buildings with ceiling suspended induced draught coolers and valve stations and NH3 pipelines in the ceiling space. At each evaporator outlet is, therefore, a wet riser. Although this may not be ideal for liquid overfeed, it represents a very, very common design.

There is little doubt that had these wet risers been avoidable, the energy performance penalty recorded could have been less or perhaps even nonexistent. However, as most practitioners would know, this is not how things are in practice where the wishes and desires of other project stakeholders often override those of the refrigeration plant designer.

In 2019, an attempt was made by Nitschke to quantify the energy performance penalty of liquid overfeed compared with dry expansion through mathematical modeling. The model used employed the latest Yashar correlation for wet riser pressure drop estimates. This correlation is also a feature of the new IIAR Piping Handbook.

As Nitschke showed in his 2019 Ohrid paper, the modeling failed to provide accurate results below system load percentages of 40% for the plant in question. The Yashar correlation and probably all other such correlations are not valid for the flow reversal scenario in wet risers. This is probably one of the reasons for the spread in liquid overfeed SEC-values in Figure 1.

Plant oversizing is rampant throughout the refrigeration industry. This is not always the fault of designers. Often this is a response to the design brief supplied by plant owners who attempt to plan for future growth. The result, however, is often that plants spend most of their operating lives in part load and wet risers therefore rarely emerge from the flow reversal scenarios.

The symptoms of these things as far as liquid overfeed plants are concerned are well known by most practitioners – liquid management problems, brining, excessive energy consumption, pump cavitation, and excessive ammonia inventories to name a few.

Consider a person standing on a chair with a half-inch garden hose in the mouth and the other end of the garden hose just above the ground. It is very easy breathing through the hose. With the end of the hose in a bucket of water, breathing becomes impossible. The density ratio of air and water is almost identical to the liquid/vapor density ratio of ammonia at -31F, yet these are the working conditions that millions of ammonia boosters are asked to accommodate daily across the world.

Suction line networks of large, expansive liquid overfeed plants can be exceedingly complex. These can connect dozens – at times hundreds of evaporators through pipelines, elbows, tee’s, isolation valves, regulating valves, orifices, risers, traps, and droppers. Combine this suction network with evaporators designed to reflect a wide array of rules of thumb relating to bottom feed, top feed, vertical headers, horizontal headers, counter flow, parallel flow, circuit orifices, and the territory quickly becomes one characterized by more unknown unknowns than known unknowns.

By continuing the employment of liquid overfeed as a concept, ammonia refrigerant is prevented from delivering the best energy performance that it is capable of. The answer to the question posed by the title of this article is therefore in the affirmative. The long era of liquid overfeed is ending. It must ensure that ammonia refrigeration is the best it can be and is able to compete in terms of energy efficiency with other natural refrigerant-based solutions.

The technologies required for eliminating liquefied refrigerant from suction lines are available. Unless the ammonia refrigeration industry embraces these technologies, it will miss out on a significant proportion of the refrigerant conversion business that will be a direct result of the global HFC phase-down and – in time – the global HFO phasedown.

Watters, R.; Nelson, B.I. “Low Charge ADX Ammonia”, Proceedings Global Cold Chain Expo, June 13-15, 2017, Chicago, IL.

Nitschke, T.; Jensen, S.S. “Thermodynamic Modelling of Liquid Overfeed and Dry Expansion Feed Central NH3 Refrigeration Plants to Determine Differences in Energy Performance”, Proceedings 8th IIF/IIR Conference Ammonia and CO2 Refrigeration Technologies, Ohrid, North Macedonia 2019.