harmaceutical, biotech-nology and specialtychemical companies are

challenging the heattransfer community to

provide solutions that enablecritical processes to operate at

extremely cold temperatures. In thepast, it was adequate to operate at

temperatures as low as -80˚F (-62.2˚C).Now industry continues to push for coldertemperatures. Low-temperature heattransfer fluid manufacturers and heattransfer companies are being asked to pro-vide systems that can run reliably at -148˚Fto -184˚F (-100˚C to -120˚C).

Why such low temperatures? For cer-tain chemical reactions the rule of thumbis that the reaction time is increased by afactor of two for each 18˚F (10˚C) reduc-tion in operating temperature. If the tem-perature is too high, the reaction time isvery quick, adversely impacting qualityand repeatability of results.

A number of design considerationsmust be taken into account when operat-ing at these extreme conditions. Thisarticle reviews the outcome of recentresearch of heat exchanger design andheat transfer fluid performance for low-temperature operation. It defines practi-cal low-temperature operation of thevarious heat transfer fluids for a giventype of heat exchanger. The performancecharacteristics of the different fluids are

discussed, as is the performance of heatexchangers as heat transfer fluids beginto freeze within them.

Common low-temperature applica-tions in a pharmaceutical plant are reac-tor jacket cooling and vacuum freezedrying (lyophilization). A heat transfersystem that can provide consistent heattransfer fluid temperature is essential forproduct quality and repeatable resultsfrom batch to batch.

If a heat transfer fluid begins to freezewhen exposed to cold operating temper-atures, then heat transfer is less efficient.This result leads to temperature increasesin the freeze dryer or reactor, compro-mising product quality.

A heat exchanger performance test-ing package was constructed to evaluatethe operational characteristics of differ-ent types of heat exchangers and low-temperature heat transfer fluids. The testsetup was fully computerized to capturekey operating variables while anexchanger was operating. Each type ofexchanger was operated with four differ-ent low-temperature heat transfer fluidswith identical mass flow rates and tem-peratures to allow for comparison underidentical operating conditions.

During the research, several helicallycoiled heat exchangers were constructedwith thermocouples attached at differentlocations on the heat transfer surface. Theheat transfer surface temperature is

markedly different when soliddeposits are present. A dataacquisition system collectedtemperature measurementswhile the exchanger was inoperation. These measure-ments allowed monitoringwhen heat transfer surfacesdeveloped solid deposits.

The exchangers handledprogressively colder heattransfer fluid; liquid nitro-gen was used as the coolantin each heat exchanger. Thisarrangement simulatedreactor jacket cooling serv-ice, which typically uses liq-uid nitrogen as the coolant.Variables affecting per-formance were systemati-cally varied to determinethe practical operatingranges for each heat trans-fer fluid.

Methanol, SylthermXLT (Dow), DynaleneMV (Dynalene) and HFE7000 (3M) were the fluidsanalyzed. Most of theanalysis was done withheliflow heat exchangers,although brazed plate heatexchangers were com-pared as well. The testinginvolved varying heat

Plants must evaluate a number of design factors whenoperating heat exchangers at ultra-low temperatures

How LowCan You Go?

January 2003 | 35www.chemicalprocessing.com

H E A T T R A N S F E R /P R O C E S S C O O L I N G

By Jim Lines

transfer fluid flow rates and inlet tem-peratures, as well as liquid nitrogenflow rates and operating pressures.

Heat transferfluid propertiesA good heat transfer fluid for low-temperature service must have a lowfreeze-point temperature, low viscos-ity and low thermal diffusivity.Depending on the operating range ofthe temperature control system, itmight need to be capable of operatingsafely at hot temperatures. Table 1 com-pares the fluid properties of the fourheat transfer fluids tested at -130˚F (-90˚C).

A generalized heat transfer correla-tion for the heat transfer fluid thatdefines how fluid properties impactheat transfer is expressed by:

Where:Re = Reynolds numberPr = Prandtl numberDh = hydraulic diameterC = constanta = positive exponent that is less than 1.0b = positive exponent that is less than 1.0

Key observations from the general-ized correlation include:• A high density, specific heat and thermal

conductivity are good for heat transfer.• A low viscosity is good for heat transfer.

It is important to keep the boundarylayer thin, with efficient heat transferthrough the boundary layer. The firstbracketed expression is the Reynoldsnumber, which is an indication of thethickness of the fluid boundary layernear the heat transfer surface. A highReynolds number is important.

The second bracketed expression isthe Prandtl number, which affects thetemperature gradient through theboundary layer. It is an indication of therate by which heat is given up by theheat transfer fluid to the coolant.

Table 2 shows just how different the

Reynolds and Prandtl numbers can be inthe case of 15,000 pounds per hour(lb/hr.) (6,800 kilograms per hour[kg/hr.]) of heat transfer fluid at -130˚F(-90˚C) in a heliflow heat exchanger. Theexchanger has 12.3 square feet (sq. ft.)(1.14 square meters [sq. m.]) of heattransfer surface.

Testing showed that an evaluation ofthe freeze point of a heat transfer fluid isinsufficient to determine the suitabilityof a fluid for a given application. Thefreeze point, along with the fluid proper-ties, the heat exchanger design and fluidvelocity, plays a part in inhibiting theonset of freezing. Just because the tem-perature of the fluid leaving a heatexchanger is well above the freeze pointof that fluid does not necessarily meanthe fluid will not freeze inside the heatexchanger. This consideration is key.

How each fluid performedFreeze-point temperature alone is notan indicator of whether or not a fluidwill freeze in a heat exchanger. Fluidproperties, velocity and heat exchangerdesign play important roles as well. Fig.1 and Fig. 2 compare the four test fluids

under identical operating conditions.Thermocouples attached to the heat

transfer surface indicated whether or notthe fluid was freezing onto the heattransfer surface. Note how dramaticallydifferent the thermocouple-measuredtemperature was when freezing occured.Once solid deposits are present, they actas an insulator and drive the surface tem-perature to much colder levels. The dif-ference is more than 100˚F (56˚C)between unfrozen and frozen heat trans-fer surfaces for the conditions tested.

Even if the outlet temperature fromthe heat exchanger is well above freez-ing, deposit buildup can occur inside aheat exchanger. This effect can be insidi-ous, as runaway freeze up can sneak upon the control system if it is not prop-erly configured.

The performance graph for SylthermXLT shows that one of the thermo-couples indicates the presence offrozen deposits even when the out-let fluid temperature is -90˚F (-68˚C). Syltherm XLT freezes wellbelow -90˚F, at -168˚F.

As the Syltherm XLT is pro-gressively cooled to approximately -100˚F inlet and -110˚F outlet, anotherregion in the heat exchanger experiencesa condition of freezing and defrosting.When the fluid is cooled further to -110˚F inlet, that region freezes entirely.

The freezing and defrosting condition

36 | January 2003 www.chemicalprocessing.com

H E A T T R A N S F E R /P R O C E S S C O O L I N G

Heat transfer coefficient = C x ReaPrb xthermal conductivity

Dh

Heat transfer coefficient = C x( )( )( )specific heat x viscositythermal conductivity

density x velocity x Dhviscosity

thermal conductivityDh

a b

Table 1. Heat Transfer Fluid Properties

Methanol Syltherm XLT Dynalene MV HFE 7000

Conventional units

Reported freeze point -143.5˚F -168˚F <-200˚F <-188.5˚F

Specific gravity 0.867 0.97 0.93 1.732

Specific heat, Btu/pounds ˚F 0.515 0.368 0.338 0.224

Thermal conductivity, 0.147 0.0795 0.094 0.0576Btu/hour feet ˚F

Viscosity, cP 6.1 33.7 20 4.9

SI units

Reported freeze point -97.5˚C -111˚C <-129˚C <-122.5˚C

Density, kg/cu. m. 867 968.3 930 1732

Specific heat, kilojoules 2.15 1.541 1.41 0.946per kilogram ˚Kelvin

Thermal conductivity, 0.2544 0.1324 0.159 0.0974W/meter ˚Kelvin

Viscosity, milliPascals seconds 6.1 33.7 20 4.9

January 2003 | 37

is imperceptible by a control system mon-itoring heat transfer fluid outlet tempera-ture. The freezing/defrosting conditionoccurs because as the ice begins to form,local velocity increases, increasing thelocal heat transfer coefficient and chang-ing the temperature distribution near theice surface. A warmer condition is createdat the deposit surface, melting the deposit.Not until the fluid temperature is loweredis the temperature differential sufficient toovercome the defrosting condition.

Take note of how different thethermocouple-measured temperature isbetween an unfrozen and frozen condi-tion. The thermocouple measures a tem-perature at the heat transfer surface ofapproximately -150˚F when deposits arenot present.

When a deposit is formed, it repre-sents a step change in temperature. Thetemperature drops very quickly toapproximately -275˚F. The solid is aninsulator. Warm heat transfer fluid can-not readily conduct heat through thedeposit. Therefore, the surface tempera-ture under the deposit approaches theliquid nitrogen coolant temperature.

Dynalene MV, a fluid that has areported freeze point of less than -200˚F,actually froze at a temperature compara-ble to Syltherm XLT’s freeze point of-168˚F. It is not clear why this occurred;therefore, further testing/analysis isneeded. Nevertheless, Dynalene MV iscomparable to Syltherm XLT in terms ofa practical operating temperature with-out freezing, even though it is reportedto have a lower freeze point.

HFE 7000 was the best fluid tested forthe purposes of the test. It could reach thecoldest temperature without loss of per-formance as a result of freezing. Further-more, HFE 7000 has a very low viscosity,even at cold temperatures, which keepspressure drop across the heat exchangerlow. For the conditions of the test repre-sented in the graph, HFE 7000 could runat an inlet of approximately -140˚F. Thefluid underwent freezing and defrosting;however, the unit performed well.

Methanol worked well for tempera-tures above -130˚F. Because it is a singlecomponent, it had the most stablethermocouple-measured temperatures.Dynalene MV, Syltherm XLT and HFE

7000 are mixtures. They are composed ofa number of different fluids, with eachhaving varied freeze points. Methanolcannot reach the cold temperatures thatHFE 7000 can; however, it does performwell at temperatures above -130˚F.

Monitoring heatexchanger performanceFreezing can occur inside a heat

exchanger even if the heat transferfluid temperature out of the exchangerindicates the temperature is well abovefreezing. A control system that meas-ures heat transfer fluid outlet tempera-ture is unreliable. It is not a good wayto avoid freezing. Runaway freezeup islikely to occur when this type of con-trol is used.

The ideal way to monitor heat

www.chemicalprocessing.com

H E A T T R A N S F E R /P R O C E S S C O O L I N G

Thermocouple 1 (F)Te

mpe

ratu

re, ˚

F

Runtime, minutes0

-350

-300

-250

-200

-150

-100

-50

0

20 40 60 80 100 140 160 180120

Thermocouple 2 (F)Thermocouple 3 (F)Thermocouple 4 (F)Thermocouple 5 (F)Syltherm XLT inlet (F)Syltherm XLT outlet (F)LN2 temp in (F)N2 temp out (F)

Figure 1. 15,000-pph Syltherm XLT/300 pph LN2 (59 psia)96C6C-16S Heliflow - Heat rejection 36,000 to 40,000 Btu/hr.

Tem

pera

ture

, ˚F

-350

-300

-250

-200

-150

-100

-50

0

Runtime, minutes0 20 40 60 80 100 140 160 180 200120

Thermocouple 1 (F)Thermocouple 2 (F)Thermocouple 3 (F)Thermocouple 4 (F)Thermocouple 5 (F)HFE 7000 in (F)HFE 7000 out (F)23-LN2 temp in (F)24-N2 temp out (F)

Figure 2. 15,000-pph HFE7000/500 pph LN2 (59 psia)96C6C-16S Heliflow - Heat rejection 60,000 to 70,000 Btu/hr.

exchanger performance is via ther-mocouples attached to the heat trans-fer surface; however, not all heatexchangers lend themselves to such asetup. When a thermocouple senses adramatic drop in temperature, say100˚F or more, freezing already isoccurring at that location. Liquidnitrogen flow rate can be loweredmomentarily until the deposit is drivenoff by the change in temperature gradi-ent that will result.

Lower liquid nitrogen flow ratereduces the heat transfer coefficient onthe coolant side, warming the heattransfer surface. This type of proactivecontrol is excellent and far better thanother reactive methods. It does notimpact heat rejection by the exchanger.Normally, one or two minutes of reducednitrogen flow are all that is needed.

Another option is to measuregaseous nitrogen outlet temperaturefrom the heat exchanger. Normally, liq-uid nitrogen enters the exchanger and isvaporized and superheated to within25˚F to 40˚F of the heat transfer fluidtemperature. As deposits form, theexchanger becomes less efficient. Not asmuch heat is given up to the liquidnitrogen. The nitrogen outlet tempera-ture becomes colder as the amount ofsuperheating is reduced.

In the HFE 7000 temperature vs.time graph, note how the gaseousnitrogen temperature diverges fromthe heat transfer fluid temperature. Asthe unit develops progressively morefrozen deposits, the nitrogen outlettemperature becomes colder andcolder. This type of control will work;however, it is a reactive rather than aproactive control.

A third control option is to measureheat transfer fluid pressure drop. As theheat transfer fluid solidifies onto theheat transfer surfaces, flow passagesbecome restricted and pressure dropacross the exchanger increases. Thisapproach is harder to control becausepressure drop does increase as operatingtemperature becomes colder. As a fluid isprogressively cooled, its viscosityincreases. Certain fluids have steep vis-cosity vs. temperature curves in thecolder operating range. The increase in

pressure drop could be the result ofincreased viscosity, not freezing.

Practical operating rangesEach fluid has a particular practicaloperating limit in coiled tube heatexchangers. A coiled tube exchangeroffers the lowest practical operatingtemperature. Shell-and-tube exchangershave warmer practical limits as a resultof maldistribution of flow on the shell-side and localized areas of low velocity.Freeze occurring in these regions willcompromise exchanger performance.

Brazed plate heat exchangers alsowill have warmer limits because thethermal efficiency of a plate heatexchanger maximizes the heat transferfluid and liquid nitrogen heat transfer

coefficient — which normally is what isdesired — however, here it causes pre-mature freezing. Also, the narrow pas-sages in a plate heat exchanger make itextremely susceptible to runawayfreezeup. Under this condition, per-formance is lost very quickly, and theunit freezes solid. No heat transferoccurs, and a reaction or freeze-dryingbatch must be discarded.

What influences performanceOnce a fluid is selected, steps can betaken to maximize performance andimprove cold temperature operation.For example, heat transfer fluid velocityis extremely important. A higher veloc-ity is better than a lower one. Fig. 3shows when liquid nitrogen flow is 300

38 | January 2003 www.chemicalprocessing.com

H E A T T R A N S F E R /P R O C E S S C O O L I N G

Table 3. Practical Operating Limits in Coiled Heat Exchangers

Fluid Practical temperature limit,˚F (˚C)

Methanol -110 to -130˚F (-79 to -90˚C)

Syltherm XLT -115 to -125˚F (-82 to -87˚C)

Dynalene MV -115 to -130˚F (-82 to -90˚C)

HFE 7000 -145 to -165˚F (-98 to - 109˚C)

Table 2. Heat Transfer Fluid Comparison

Methanol Syltherm XLT Dynalene MV HFE 7000

Fluid velocity ft./sec. (m./sec.) 8.4 (2.6) 7.5 (2.3) 7.8 (2.4) 4.2 (1.3)

Reynolds number 3,160 571 960 3,900

Prandtl number 52 377 174 46

Met

hano

l tem

pera

ture

s, ˚F

Methanol flow rates, lb/hr.

-140

-130

-120

-110

-100

-90

25,000

500 pph nitrogen400 pph nitrogen300 pph nitrogen200 pph nitrogen

20,00015,000

7,500 lb/hr.= 4 ft./sec.

10,0005,000

Figure 3. Methanol Practical Operating Range –Heliflow Heat Exchanger Nitrogen Pressure at 60 psia

January 2003 | 39

pounds per hour (pph) for a heliflowheat exchanger, at 4 feet per second(ft./sec.) (7,500 pph) methanol could becooled to -115˚F without freezing occur-ring. At 6 ft./sec. (11,000 pph). methanolat -121˚F was acceptable. For 8 ft./sec.(15,000 pph) and 12 ft./sec. (21,000pph), methanol temperatures of -126˚Fand -130˚F, respectively, are achievable.

Another possibility is to increase thenitrogen supply pressure. The vaporiza-tion temperature of nitrogen variesgreatly with pressure.

As operating pressure increases, theboiling temperature increases. A higherboiling temperature will help to keepthe heat transfer surface warmer and,consequently, reduce the possibility offreezing the heat transfer fluid. Nitrogenpressure of 150 psia vs. 50 psia can lowerthe practical operating temperature of aheat transfer fluid by 10˚F to 15˚F.

A third possibility is try to mini-mize liquid nitrogen usage by maxi-mizing the amount of superheat. Onceliquid nitrogen is vaporized, the gas isheated (superheated) in the exchanger.The closer the nitrogen gas outlet tem-perature is to the heat transfer fluidtemperature, typically, the colder theheat exchanger is able to operate. Alower nitrogen flow will reduce thenitrogen heat transfer coefficient andcause the heat transfer surface to bewarmer. This approach helps to avoidfreezing.

From the Methanol Practical Oper-ating Range graph, at 15,000 pph ofmethanol (8 ft./sec.), 200 pph of nitro-gen allowed the temperature of the heattransfer fluid to reach -130˚F withoutfreezing. For 300 pph nitrogen, it is -125°F. At 400 pph and 500 pph nitro-gen flow, -120°F and -115°F, respec-tively, were the coldest temperatureswithout incurring freezing.

ConclusionLow-temperature applications in whichheat transfer fluid temperatures are below-100˚F (-73˚C), are challenging. Plantsmust consider more than the heat transferfluid freeze point. The physical propertiesof the fluid, the type of heat exchangerand nitrogen supply conditions play keyroles in satisfactory performance. Cold

operation is possible; however, specialcare with heat transfer fluid selection, thecontrol system to monitor freezing andheat exchange design are important. CP

Lines is vice president of marketing for Gra-ham Corp., where he is responsible for appli-cation engineering, computer engineering,estimating, research and development and

marketing activities. He can be reached at(585) 343-2216; fax: (585) 343-1097; e-mail: jlines@graham-mfg.com.

www.chemicalprocessing.com

Reader interest reviewPlease circle the appropriate number on the reader service card to indicate

the level of interest in this article.

High 534 Medium 535 Low 536

H E A T T R A N S F E R /P R O C E S S C O O L I N G

Tem

pera

ture

, ˚F

-300

-350

-250

-200

-150

-100

-50

0

Runtime, minutes0 20 40 60 80 100 140 160 180 200120

Thermocouple 1 (F)Thermocouple 2 (F)Thermocouple 3 (F)Thermocouple 4 (F)Thermocouple 5 (F)Dynalene MV in (F)Dynalene MV out (F)23-LN2 temp in (F)24-N2 temp out (F)

Figure 4. 15,000-pph Dynelene MV/300 LN2 (59 psia)96C6C-16S Heliflow - Heat rejection 36,000 to 40,000 Btu/hr.

Tem

pera

ture

, ˚F

-300

-350

-250

-200

-150

-100

-50

0

Runtime, minutes0 20 40 60 80 100 140 160 180 200120

Thermocouple 1 (F)Thermocouple 2 (F)Thermocouple 3 (F)Thermocouple 4 (F)Thermocouple 5 (F)MeOH in (F)MeOH out (F)LN2 temp in (F)N2 temp out (F)

Figure 5. 15,000-pph MeOH/300 LN2 (59 psia)96C6C-16S Heliflow - Heat rejection 38,000 Btu/hr.