Operating charge in a flooded evaporator is not a fixed value; it is a variable inventory controlled by liquid level, refrigerant density, oil fraction, and the transient distribution of liquid in the low-side pipework. On a horizontal shell-and-tube ammonia evaporator with 25.4 mm outside diameter low-fin tubes, 1,067 mm shell diameter, and a wetted shell-side volume of 1.42 m³, a level setpoint of 60% of shell diameter corresponds to a nominal liquid refrigerant volume of 0.85 m³ only when the contained fluid is pure R-717 at the evaporating temperature. Field measurements using a guided-wave radar level transmitter with a predicted dielectric constant of 16.5 for liquid ammonia and 2.2 for a mineral oil-rich bottom layer show that the apparent level can become a composite value when the oil film exceeds 25 mm, because the pulse reflects from the oil-rich interface rather than from the lean ammonia upper layer. The resulting charge calculation, performed from the level volume curve of the vessel and the liquid density at −35°C of 683 kg/m³ for ammonia and 890 kg/m³ for naphthenic mineral oil, overstates the active refrigerant mass by 8% to 18% depending on the oil accumulation depth. Charge reporting under these conditions must therefore distinguish between total low-side mass and equivalent pure-refrigerant mass, especially where inventory thresholds under EN 378-2:2016 or ISO 5149-1:2014/Amd 1:2020 are used to determine refrigeration safety class compliance. A flooded evaporator in ammonia service should therefore be audited with a level instrument that contains an interface function, not with a simple float switch alone, because the float can remain in the refrigerant-rich bulk while the oil-rich lower layer displaces the active charge volume and reduces wetted-surface contact near the bottom of the shell.

What Conditions Cause Oil Return to Fail Even With a Correctly Sized Low-Pressure Receiver?

Oil return failure in a flooded evaporator is less a problem of bulk oil transport than of phase separation at the surface of the liquid and the velocity at the liquid drains. In a pumped liquid overfeed ammonia circuit with a circulation rate of 1.5:1 to 2.0:1, the mass fraction of lubricant discharged from the oil separator is typically 3 mg/kg to 10 mg/kg when a coalescing separator with a 0.3 µm glass-fibre element is maintained with a differential pressure below 35 kPa. The oil that reaches the flooded evaporator is not evenly distributed inside the shell. Because R-717 and naphthenic mineral oil exhibit an upper critical solution temperature below typical cold-storage evaporating temperatures, the oil phase separates rapidly after the refrigerant flashes, accumulating in low-velocity zones along the bottom tube rows and near the pass partitions. If the liquid recirculation pump is taking suction from a bottom nozzle positioned 150 mm above the shell bottom, oil removal depends on drag from the flowing liquid and on the height of the settled oil layer; at an oil layer thickness below 5 mm, the horizontal convective velocity at the interface may be less than 0.04 m/s, which is insufficient to entrain a naphthenic oil with kinematic viscosity of 68 mm²/s at 40°C measured in accordance with ISO 3104:2023. A drain pot with a capacity of 12 L to 18 L per 1 MW of evaporator capacity, fitted with an electric oil drain valve operated on a 90 s opening cycle every 4 h, is the minimum practical starting point, but published field data for oil return rates from specific tube bundle geometries is limited, and commissioning verification is commonly performed by measuring the mass of drained oil per 100 operating hours against the expected separator carryover. Where the evaporator shell contains an integrally baffled liquid distribution section, the separation velocity is reduced below the oil droplet terminal velocity, causing oil droplets smaller than 80 µm to remain suspended in the boundary layer and to migrate toward the bottom tubes rather than returning to the low-pressure receiver.

On a production-scale cold-storage installation equipped with a 315 kW reciprocating compressor and a horizontal flooded air-cooling evaporator, the low-side charge calculated from the refrigerant level at 58% shell fill was 96 kg of R-717. The initial oil return audit drained 3.4 kg of oil from the evaporator low-point pot during the first 120 h of operation, but the compressor oil separator residual mass concentration remained below 7 mg/kg, indicating that the evacuation was recovering accumulated oil rather than a steady-state leak. Repeated level sensor checks after oil removal showed the apparent guided-wave level shifted by approximately 11 mm, which changed the computed charge by 6 kg. This magnitude is relevant for machinery-room release calculations, where the low-side inventory under ASHRAE 15-2022 and local fire-code calculations may be used to define refrigerant quantity limits. The oil drain sequence was then changed to a 2 s blowdown followed by 60 s settling before the level measurement was accepted by the controller, preventing the level loop from responding to a mixed oil-refrigerant interface. When the drain pot was isolated and the oil sample was tested, the moisture content was 62 mg/kg by ASTM E1064, which exceeded the plant acceptance limit of 50 mg/kg and required a nitrogen-purge of the pot before returning any oil to the compressor crankcase. This procedure prevented an otherwise invisible interaction between water, ammonia, and the oil additives from forming a viscous emulsion at the liquid level interface, which would have increased the apparent oil volume and further disturbed the radar level signal.

When Hot Gas Defrost Returns a High-Viscosity Oil Slab to the Low-Side Receiver

Hot gas defrost on flooded evaporator coils introduces a transient charge redirection that is frequently absent from static charge calculators. During a defrost cycle at +10°C coil surface temperature, the shell-side liquid inventory is displaced by hot gas and returns to the low-pressure receiver through the wet suction line, carrying oil slugs that have become mobile only because of the temporary viscosity reduction. The viscosity of a naphthenic mineral oil with ISO VG 68 is approximately 68 mm²/s at 40°C, but at the flooded evaporator operating temperature of −35°C it may exceed 600 mm²/s to 1,200 mm²/s depending on the pour point measured by ASTM D97-17b. As the coil warms, the oil layer transitions from a bound film to a mobile liquid, and if the return line is oversized at 100 mm to 150 mm diameter, the downward-sloping horizontal run may carry a stratified layer at 0.3 m/s to 0.5 m/s. If the line instead rises vertically after the evaporator, the hot gas velocity must exceed 6 m/s to 10 m/s to carry entrained oil droplets of 100 µm to 300 µm back to the receiver, based on Stokes settling calculations. Oil that remains in the coil re-distributes on the next cooling cycle and produces a measurable reduction in the overall heat transfer coefficient of 7% to 15% after four consecutive defrost cycles on a finned-tube flood coil, according to published fouling-factor studies on ammonia evaporators. The charge audit must therefore treat defrost as a dynamic oil-return event, not as a zero-mass-flow period, and the low-pressure receiver level controller must be configured with a defrost-triggered measurement delay of at least 120 s to avoid false charge readings caused by returning oil and refrigerant foam.

An oil return system using a differential-pressure-driven return line from the low-pressure receiver to a heated oil rectifier can reduce the effective oil inventory without relying solely on gravitational drain from the evaporator. In one configuration, the low-pressure receiver is fitted with a bottom oil pot of 20 L capacity, heated by a 1.5 kW electric trace heater controlled to maintain 35°C to 45°C. The pot is isolated by two 25 mm ball valves and connected to the suction side of a screw compressor through a 10 mm capillary line that provides 200 kPa to 350 kPa differential pressure for oil transport. The oil-rectifier vessel separates residual ammonia from the oil by heating the mixture to 45°C and allowing the ammonia vapour to vent to the compressor suction through a 12 mm line with a check valve cracking pressure of 20 kPa. Recovered oil is tested for water content by ASTM E1064 or equivalent Karl Fischer method, and returned to the compressor crankcase only when moisture is below 50 mg/kg. This arrangement prevents the oil logged in the evaporator from affecting the level controller because the low-point pot becomes the dominant oil sink. The trace heater on the pot must be interlocked with a high-temperature cutout at 55°C, because higher surface temperatures on residual oil can liberate refrigerant vapour too quickly and create a liquid-level surge in the receiver that mimics a sudden charge increase.

Refrigerant–Oil Solubility Data and Charge Adjustment in Low-Charge Packages

For halocarbon refrigerants, oil return behaviour differs fundamentally from ammonia because the lubricant and refrigerant form a homogeneous liquid phase above the miscibility boundary. An R-134a flooded evaporator using a polyol ester lubricant with ISO VG 32 will typically operate with a single liquid phase down to −40°C, and the oil leaves the evaporator by solution in the returning liquid rather than by mechanical entrainment. In comparison, R-22 with mineral oil is partially miscible at −30°C, and the oil-rich phase can remain in the evaporator shell if the liquid charge is insufficient to sweep it through the suction header. Charge calculations for halocarbon flooded evaporators therefore require a solubility-weighted liquid density and an effective viscosity that reflects the dissolved refrigerant. The effective viscosity of a polyol ester with 32 mm²/s base grade at 40°C can fall below 9 mm²/s when saturated with R-134a at −20°C, which improves oil return but also reduces the hydrodynamic sealing of the liquid passage in the evaporator. Published refrigerant-oil solubility curves from lubricant manufacturers generally report the critical point of phase separation as a function of temperature and pressure, and the charge mass is adjusted by adding 5% to 12% to the calculated liquid mass to account for the oil-rich recirculating fraction. Where a package is designed for 0.50 kg/kW refrigerant charge, an unaccounted oil-rich inventory of 8% can shift the apparent charge by 0.04 kg/kW, which is enough to affect the low-pressure control stability during pull-down from +20°C to −35°C. The correct adjustment requires a mass balance around the low-pressure receiver, using the separate liquid density of each phase and the measured oil volume at the receiver bottom, rather than a single homogeneous-density assumption.

On a two-stage ammonia system serving a 650 kW spiral freezer, the flooded low-stage evaporator was configured with a high-pressure liquid feed from a 150 mm pumped main and a suction return to a horizontal low-pressure receiver of 2.4 m length and 1.2 m diameter. The oil return audit revealed that the low-point drain pot in the evaporator collected only 0.6 L of oil in 24 h at full load, while the receiver pot collected 1.8 L, confirming that oil was being carried beyond the evaporator and depositing in the receiver. The liquid level transmitter on the receiver showed a slowly increasing apparent level of 18 mm over 72 h despite no change in refrigerant charge, consistent with an accumulated oil-rich bottom layer of 34 mm depth. After the receiver was drained, the measured charge inventory decreased by 23 kg, and the level setpoint was reduced by 12 mm to maintain the same usable liquid seal above the pump suction. Operators implemented a daily oil drain schedule with a 60 s automated drain sequence, a 30 s settling time, and a level verification delay of 15 s to avoid false low-level trips. The plant then operated for 500 h without compressor oil makeup exceeding 1.9 L per 1,000 operating hours, and the suction superheat at the compressor remained stable within 0.5 K of the 1.0 K setpoint.

Measuring Oil Return Efficiency Across a Variable Condenser Pressure Range with a Coriolis Mass Flow Network

Oil return efficiency is not a single percentage; it must be measured against the compressor oil loss rate under the same operating conditions. A production-scale ammonia low-charge package with a 450 kW screw compressor, a water-cooled condenser operating from 25°C to 42°C, and a flooded plate-and-shell evaporator with 38 stainless-steel plates was instrumented with a Coriolis mass flow meter on the oil drain line, a differential pressure transmitter across the oil separator, and a level sensor on the low-pressure receiver. The separator residual mass concentration was inferred from the compressor oil consumption and the compressor discharge flow calculated from an orifice plate with a 25.4 mm bore and a differential pressure of 1.2 kPa to 2.8 kPa. Over a 200 h test period, the drained oil mass was 2.1 kg, while the compressor oil makeup was 2.6 kg, yielding an apparent oil return effectiveness of 81% during periods with condenser pressure at 1,200 kPa and evaporator pressure at 93 kPa. At condenser pressure below 950 kPa, the oil drain flow decreased by 35%, and the oil separator differential pressure dropped below 15 kPa, which indicated reduced gas flow to compress the oil pot. These field results were used to reset the oil drain valve timer from 120 s to 180 s at low condenser pressure, and to require a minimum condenser pressure of 1,050 kPa before automatic oil return is enabled. Published data for this exact plate evaporator and oil separator combination is limited; the values are therefore plant-specific and are not generalized beyond the tested compressor speed range of 40 Hz to 60 Hz.