A 9 kW duty that takes JP-8 from 17.8 °C to 4.4 °C at 25.7 L/min needs only about 2 kg/min of LN2, but that LN2 cannot be allowed to touch the fuel-side wall. The working design puts a low-temperature intermediate fluid between the LN2 and the JP-8, then controls that fluid's supply temperature above the fuel's freeze and cloud point.
Which quick fixes fail on a direct LN2-to-JP-8 exchanger?
Run LN2 straight against the fuel. LN2 boils at 77 K (about −196 °C). Even the warm end of the design gas temperature, 173 K, is about −100 °C. The fuel's wall film freezes long before the bulk reaches 4.4 °C, and the exchanger plugs. Trace water in the fuel adds a second failure: ice forms at 0 °C, so any coolant that may meet wet fuel needs a minimum inlet temperature above 0 °C.
Raise pressure and velocity to force the heat transfer. This helps liquid-to-liquid duty on chilled water or brine. Water-side velocities near 60 ft/s have been run on hydraulic oil coolers, but the price was extreme noise and high-alloy stainless to resist erosion. It does nothing for a boiling cryogen, and it does not lower the fuel-side wall temperature.
Switch to chilled water or −5 °C brine. The heat capacity is high and the freezing risk drops, but where chilled water is not available it is off the table. If brine is acceptable, chilling it by sparging LN2 into it and running it through a compact titanium exchanger is a legitimate variant. Check the brine's own freeze point against the LN2 contact temperature.
Add more surface on the LN2 side. Boiling LN2 at a large wall-to-fluid temperature difference forms a vapor layer that cuts the heat transfer coefficient sharply. More area lowers the heat flux per square metre but does not remove the film, and the wall stays cryogenic.
Split into a pressurized-LN2 exchanger followed by a cold-N2-gas exchanger. Holding the first stage at a pressure that suppresses boiling and flashing into a second stage lets each geometry be optimized. It still presents a wall far below the fuel's freeze point in at least one stage, so it is only usable if the fuel-facing surface sees a tempered fluid.
Why does a direct exchanger both freeze the fuel and transfer heat poorly?
The fuel-side wall temperature comes from the series resistance between the two fluids. With bulk fuel at Tfuel and LN2 at TLN2:
Here h_fuel is the fuel-side film coefficient, t/k is the wall resistance, and h_boil is the LN2-side boiling coefficient. Fuel gets viscous as it chills, so h_fuel drops and the wall runs colder, toward the LN2 side. Meanwhile h_boil is depressed by the gas film, which makes the boiling side the bottleneck. You get the worst combination: the average duty is starved by film boiling, while local wall spots at the fuel's inlet-side surfaces drop below the freeze or cloud point and build a frozen layer that insulates further.
An intermediate fluid fixes the first problem by capping the coldest temperature the fuel wall can see at the coolant supply temperature. It moves the film-boiling problem into a chiller where fouling and freezing can be handled separately, and it gives the loop thermal mass that damps the LN2 valve.
What does the 9 kW duty actually demand?
Check the numbers before sizing anything. 46 lb/min of JP-8 is about 20.9 kg/min, or 0.348 kg/s. The 25.7 L/min flow then implies a density near 0.81 kg/L. A 13.4 K temperature drop at 9 kW implies a specific heat of about 1.93 kJ/kg·K (derived from the stated numbers, 9 / (0.348 × 13.4)). Read the actual density and specific heat at the two temperatures from the fuel data sheet and rerun the heat balance; the specific heat of the real batch sets the real duty.
On the LN2 side, 90% transfer efficiency puts the required cooling at 10 kW. The stated 0.033 kg/s (2 kg/min) corresponds to roughly 300 kJ/kg of usable cooling per kilogram of LN2, which matches handbook latent heat of about 199 kJ/kg plus about 100 kJ/kg sensible heat from 77 K to 173 K. Roughly a third of the cooling comes from warming the gas, so a design that dumps the gas at 77 K wastes it. Continuous operation consumes about 120 kg of LN2 per hour.
The vent stream is about 1.7 m³/min of nitrogen at ambient conditions (assuming 1.165 kg/m³ at 20 °C and 1 atm). Vent it outdoors and put oxygen monitoring in any space where the exchanger, chiller, or vent line can leak.
How do I set the coolant floor temperature and choose the intermediate fluid?
The coolant supply temperature sets the whole design. The cold-end approach for a counterflow exchanger is the JP-8 outlet temperature minus the coolant inlet temperature:
- Coolant supply at 0 °C: 4.4 K approach at the cold end, so a large area or high
U. - Coolant supply at −10 °C: 14.4 K approach, a smaller exchanger.
Two cases decide the floor:
- If water traces in the fuel are possible, hold the coolant supply above 0 °C. Accept the tight approach and the extra area.
- If the fuel is confirmed dry, read the freezing point and cloud point from the fuel certificate for the batch and set the floor at the higher of the two plus a margin you can defend. A lower floor cuts exchanger size but leaves less room for a control upset.
Two intermediate-fluid options fit:
-
Low-temperature liquid coolant. Duratherm XLT-120 is a low-temperature heat-transfer fluid that has been used for this exact duty. Read its viscosity-versus-temperature curve and minimum operating temperature from the datasheet. Pump head and coolant-side
hboth change as the fluid chills. - Refrigerant such as R134a. Evaporating at a set pressure holds the fuel-side surface near the saturation temperature for that pressure, so the pressure setpoint limits the wall temperature. Read saturation pressure against temperature from the refrigerant property tables. R134a freezes far above 77 K, so the condensing side against LN2 needs the same freeze protection as any other LN2 wall.
How do I build the loop and exchangers?
- Set the coolant floor from the fuel certificate and the water-contamination decision above, and record it as the low-limit setpoint.
- Size the fuel-to-coolant exchanger in counterflow:
A = Q / (U · F · LMTD). Use the fuel's real viscosity at the cold end forh_fuel, and keep the fuel side turbulent where the pressure drop allows. - Size the LN2 chiller on the film-boiling coefficient, not the nucleate-boiling coefficient, because the wall sits far above 77 K. Look at cryogenic pool-boiling data (NASA technical reports cover it) before assuming a value.
- Prevent the coolant from freezing on the LN2 surface. Circulate the coolant at velocity across the LN2 coil or jacket, and confirm the coolant's freeze point against the coldest LN2 wall temperature.
- Recover the sensible heat of the N2 gas (77 K to 173 K) in a gas-to-coolant stage if the gas and coolant temperatures allow it, and vent outdoors.
- Wire the interlocks (low coolant supply temperature, low coolant flow, low JP-8 flow) to close the LN2 valve.
Which signals control the loop, and what does a wrong value look like?
Use cascade control: the JP-8 outlet temperature at 4.4 °C is the outer loop, the coolant supply temperature is the inner loop, and the LN2 control valve is the final element. The transport delay from the LN2 valve through the chiller, the coolant piping, and the exchanger to the JP-8 outlet sensor is the dead time. Measure that delay with a step in the valve before touching any gain. Feed forward the fuel duty from measured flow and inlet temperature: Q = m_dot · cp · (T_in − T_out_target).
| Signal | Source | Wrong-value symptom |
|---|---|---|
| JP-8 outlet temperature (target 4.4 °C) | RTD downstream of the exchanger | A sensor in a stagnant pocket reads warm, so the controller overcools and the coolant supply chases toward the freeze floor. |
| JP-8 flow (25.7 L/min) | Flow meter on the fuel line | Reads high: feedforward over-calls duty and pulls the coolant colder. Reads low: the outlet runs warm. |
| JP-8 inlet temperature (17.8 °C) | RTD upstream of the exchanger | Drift gives a steady feedforward error that the outer loop must trim. |
| Coolant supply temperature | RTD at the fuel exchanger coolant inlet | Reads warm: the LN2 valve overdrives and the actual supply undershoots the floor. |
| Coolant flow | Flow meter or pump speed feedback | Flow falls as viscosity rises at low temperature, the outlet warms, the LN2 valve opens further, and the coolant gets colder: a runaway toward freezing. |
| LN2 valve position / consumption | Control valve feedback, tank weight or flow meter | Saturated open with cold vent gas: chiller undersized or film-boiling-limited. |
| Vent gas temperature (design 77 K to 173 K) | Cryogenic-rated sensor in the N2 exhaust | Reads near 77 K: LN2 leaves without fully warming, so cooling is wasted. |
| JP-8 side pressure drop | Differential pressure transmitter | Climbs at constant flow: wax or ice is building on the wall. |
Tell wiring or sensing faults from tuning problems by trending the coolant supply temperature against the LN2 valve position. A supply that oscillates with the valve at a period matching the dead time is a tuning problem. A supply that sits off setpoint with the valve pinned at one end is a capacity, sensor, or wiring problem, and no gain change fixes it.
How do I prove the design before running a full-duty fuel test?
- Circulate the coolant with no fuel flow. Step the LN2 valve and record the coolant supply temperature response, the dead time, and whether the low-limit interlock trips cleanly at the floor.
- Run water or a fuel substitute at fuel-side flow with the coolant held at the top of its range. Confirm the JP-8 side heat balance from measured flow and temperatures against the 9 kW target before dropping the coolant setpoint.
- Compare the two energy balances: the fuel-side
m_dot · cp · ΔTagainst the coolant-side balance and against LN2 consumption (tank weight or metered flow). LN2 use near 0.033 kg/s at 9 kW confirms the 90% efficiency assumption; a much higher use points to loss, film boiling, or vent gas leaving cold. - Lower the coolant setpoint in steps toward the floor while watching the JP-8 side differential pressure. A rising differential pressure at constant flow is the first sign of solid formation. Stop and raise the setpoint when it appears.
- Trip-test each interlock: cut coolant flow, cut JP-8 flow, and force a low coolant temperature, and confirm the LN2 valve closes each time.
FAQ
Can I cool JP-8 directly with liquid nitrogen in a shell-and-tube exchanger?
No. The LN2-side wall sits far below the fuel's freeze and cloud point, so the fuel freezes on the surface and the exchanger plugs. Trace water in the fuel adds ice at 0 °C, which is why a coolant supply above 0 °C is advised where wet fuel is possible.
Does boiling LN2 give a high heat transfer coefficient?
Not when the wall is far warmer than 77 K. A vapor film forms and cuts the coefficient sharply, so size the LN2 side on film-boiling data and read cryogenic pool-boiling curves before assuming a value.
Can I use R134a or a low-temperature heat-transfer fluid as the intermediate loop?
Yes. An LN2-to-JP-8 exchanger has been built with Duratherm XLT-120 as the intermediate fluid, and R134a is a workable refrigerant option. Read the fluid's freeze point, viscosity at the coldest temperature, and saturation data from its datasheet or property tables before committing to the pump and chiller.
Can I stop tuning and escalate to official support?
Stop when the coolant supply temperature sits off setpoint with the LN2 valve pinned, when the JP-8 side differential pressure climbs at constant flow, or when the two energy balances disagree by more than your measurement uncertainty. Take the trends and heat balance to the exchanger builder, the LN2 supplier, and the coolant manufacturer's technical support, and get the fuel freeze and cloud point confirmed in writing before you retest.