Anda di halaman 1dari 10

Matthys, E. F. Heat Transfer to Non-Newtonian Fluids The Engineering Handbook. Ed. Richard C.

Dorf Boca Raton: CRC Press LLC, 2000

1998 by CRC PRESS LLC

54
Heat Transfer to Non-Newtonian Fluids
54.1 The Fluids 54.2 Friction 54.3 Heat Transfer
Laminar Regime Turbulent Regime

54.4 Instrumentation and Equipment

Eric F. Matthys
University of California, Santa Barbara

Most fluids used in industrial applications are non-Newtonian. The very name tells us that we lump in this category all fluids except one type, the Newtonian ones. Newtonian fluids are those that obey Newton's law relating shear stress and shear rate with a simple material property (the viscosity) dependent on basic thermodynamics variables such as temperature and pressure, but independent of flow parameters such as shear rate and time. These fluids constitute therefore only one category of fluids: the "simpler" ones. Non-Newtonian fluids are then defined as being all the other ones! One might therefore reasonably assume that the state of knowledge on non-Newtonian fluid heat transfer is much more extensive than that of Newtonian fluid heat transfer, but that is not the case. A major reason why this is so is that it is usually much more difficult to study non-Newtonian fluids because of their complex nature and complex interactions with the flow field. As a result, the majority of the information on non-Newtonian fluids is very empirical and consists primarily of simple friction or heat transfer correlations where the constants are determined by best-fit of experimental data. In general, many engineers dealing with fluid mechanics or heat transfer think automatically only of Newtonian fluids because these are usually the only fluids covered in such undergraduate engineering classes. The average engineer has therefore typically never been exposed to non-Newtonian fluids before his or her first encounter with them on the job. This has important consequences as far as engineering practice, because serious mistakes can easily be made if one is not aware of the bizarre behavior of some of these fluids. Practically speaking, when the fluid is not a very simple or one-component liquid or gas such as water, oil, air, nitrogen, and so on, there is a very good chance that it may exhibit some non-Newtonian properties. Fluids such as suspensions of particles and fibers, slurries, polymer solutions and melts, surfactants, paints, foodstuffs, body fluids, soaps, inks, organic materials, adhesives, etc., may all exhibit non-Newtonian behavior. It behooves the engineer then to be aware of the potential problems and to decide on an informed basis whether to use simpler (but perhaps leading to large errors) Newtonian fluid correlations or to use the more complex information (maybe) available on

1998 by CRC PRESS LLC

non-Newtonian fluids. Given the breadth and complexity of the field, the nonspecialist reader will be better servedin this author's opinionby a chapter that focuses on giving a general idea of the basic issues and difficulties involved and on providing some useful references. Mentioning only a few equations is indeed more likely to lead to inadvertent inappropriate usage than to be very helpful. The reader is instead referred to two excellent and readily available handbook articles which provide numerous correlations: one by Irvine and Karni [1987] for purely viscous fluids and one by Cho and Hartnett [1985], which emphasizes more viscoelastic fluids. Some other references will also be given hereafter.

54.1 The Fluids


It should be noted first that one needs only to address the issue of moving fluids here, because the very definition of a non-Newtonian substance implies that we are dealing with a fluid undergoing flow. Accordingly, only convective heat transfer needs to be discussed here, because both conduction and radiation are normally unaffected and require only that appropriate material properties such as thermal conductivity and emissivity be known, as in the case of Newtonian fluids. Another point to consider is that the definition of a non-Newtonian fluid covers only the shear viscosity of the fluid, but many non-Newtonian fluids will also show a complex extensional flow behavior. Some non-Newtonian fluids, perhaps the most difficult to deal with, are also viscoelastic. The first step in trying to predict the heat transfer behavior of a non-Newtonian fluid is therefore to determine its nature and type. For simplicity, one often classifies these fluids as purely viscous versus viscoelastic, the former lacking the elastic characteristics of the latter. Their viscosity can either increase or decrease both with shear rate and with time, and some fluids may also exhibit a yield stress. Viscoelastic fluids in general exhibit memory effects, but some may show particularly large time-dependent effects in addition to shear raterelated variations (e.g., some surfactant solutions). In general, the heat transfer to purely viscous fluids can be quantified by relatively simple modified Newtonian fluid correlations, and with some information on the viscosity one can then predict reasonably well the heat transfer and friction. These fluids have been studied early on, and many flow configurations have been investigated. Reviews and handbooks, even somewhat older ones, may provide most of the state-of-the-art information. For viscoelastic fluids, on the other hand, and especially in turbulent flow, much less information is available; and this is an area of active current research. Some of these fluids exhibit a fascinating propertythat of reducing greatly the friction and heat transfer under turbulent flow conditions. These phenomena are called drag and heat transfer reductions. For these fluids, using a Newtonian-like approach to predict heat transfer may well result in very large errors. This field is evolving rapidly and much of the current information may need to be found in technical journals. It is relatively easy to determine whether a fluid might be viscoelastic with some simple qualitative experiments such as recoil, finger-dip and fiber-pull, rod climbing, die swell, tubeless siphon, etc. One could also pretty much guess that most fluids that are not viscoelastic but are

1998 by CRC PRESS LLC

nevertheless mixtures, solutions, or suspensions may likely be purely viscous non-Newtonian fluids. Beyond such simple classification, however, some data on the viscous or viscoelastic nature of the fluid will be necessary to predict heat transfer or friction quantitatively. This information has to be acquired from some rheology experiments, although rough estimates can at times be found in the literature. One should be especially cautioned, however, against using viscosity models beyond their range of applicability, as is often done for the ubiquitous power-law model, for example. It cannot be emphasized enough for the engineer who is not yet familiar with non-Newtonian fluids that one should be particularly wary of relying on one's experience with Newtonian fluids or one's intuition when tackling the more complex non-Newtonian fluids. The latter's behavior may indeed be very surprising. For example, a few parts per million of a polymer in solution in water could well reduce the heat transfer by a factor of 10. Accordingly, the reader is strongly advised to consult some of the additional material referred to hereafter to develop enough understanding of the issues to avoid such problems.

54.2 Friction
The reader is referred to Chapter 35 for fluid mechanics information on these fluids. An approach similaralthough simplerto that discussed here for heat transfer can usually be followed for friction. It is important to note, however, that the usual strong coupling between friction and heat transfer for Newtonian fluidswhich enables one to predict heat transfer if one knows friction and vice versa for these fluidsmay not necessarily hold for non-Newtonian fluids, another source of trouble and error.

54.3 Heat Transfer


As mentioned earlier, one is usually reduced to using simple empirical correlations to predict the heat transfer for these fluids. In addition to the handbooks listed above, the reader is also referred to a number of reviews for additional information: Metzner [1965], Skelland [1967], Dimant and Poreh [1976], Shenoy and Mashelkar [1982], Hartnett and Kostic [1989], and Matthys [1991]. As in the case of Newtonian fluids, one may use the Nusselt number to quantify the convective heat transfer coefficient. In the case of a purely viscous non-Newtonian fluid, the Nusselt number will usually be a function of modified Reynolds and Prandtl numbers. For viscoelastic non-Newtonian fluids, it will be in addition a function of another nondimensional number that may be related to the elasticity of the fluid through a relaxation time (e.g., a Weissenberg number) or to the extensional viscosity of the fluid or some other parameter. It is crucial to distinguish between the several types of Reynolds (and corresponding Prandtl) numbers that are used for the prediction of friction and heat transfer for these fluids. (Note that such a distinction is not always made clearly in the literature, and one should proceed cautiously.) Often generalized Reynolds and Prandtl numbers are used for laminar flow, whereas apparent numbers may be more suitable for turbulent flow. (For applied engineering work, one may also often use solvent-based numbers.) This issue is beyond the scope of this article, and good discussions of these parameters

1998 by CRC PRESS LLC

can be found in the reviews listed above. Since the main effect of the non-Newtonian character of the fluid on the heat transfer results from interactions between fluid and flow field through large variations in viscosity or elasticity (sometimes over several orders of magnitude), the variations in other thermophysical parameters (e.g., thermal conductivity and specific heat capacity) are often of much lesser importance. One may then often assume that these properties vary only with temperature in a given manner for a specific non-Newtonian fluid or even that they are those of the Newtonian solvent (e.g., for low-concentration solutions). Naturally, for highly concentrated suspensions of fibers or particles, for example, one has to use appropriate properties for the actual fluid, and there are procedures available to predict such properties based on the concentration of the suspended material. For more complex or lesser-known fluids, one may well be forced to measure these properties directly. It is useful to remember that for viscoelastic fluids, in particular, the large reductions in heat transfer seen in turbulent flow are not related to modifications of the "static" thermophysical properties, but rather to fluid/turbulence interactions, a dynamic process. This explains why these fluids do not normally show such reductions in heat transfer in the laminar regime. The distinction between laminar and turbulent flows is therefore even more crucial to make here than for Newtonian fluids. Note that the transition between the two regimes takes place under relatively similar conditions for both Newtonian and non-Newtonian fluids, and that the usual Newtonian fluid criteria for transition are often used.

Laminar Regime
Heat transfer in the laminar regime can be fairly readily predicted for non-Newtonian fluids because of the simple nature of the flow and the absence of significant elastic effects. A good early discussion on this issue for some fluids is provided in Skelland [1967]. Generally, one simply modifies the Nusselt number for a Newtonian fluid with a coefficient that involves a parameter reflecting the change in velocity profile (e.g., a power-law exponent n) and possibly the aspect ratio if it is internal flow in a noncircular duct [Irvine and Karni, 1987; Hartnett and Kostic, 1989]. For heat transfer in the entrance region, one has to introduce another parameter such as a Graetz number to account for the temperature profile development (as in the case of Newtonian fluids). Also as in the case of Newtonian fluids, the laminar thermal entrance region can be fairly long for high Reynolds numbers. Overall, the change in Nusselt number introduced by the non-Newtonian nature of the fluid is often relatively moderate in the laminar regime, and, if nothing else is available, the use of a Newtonian value may not be too far off in first approximation for fluids with a moderate power-law exponent, for example. For external flow, simple empirical corrections to Newtonian expressions can also be used [e.g., Irvine and Karni, 1987]. For free and mixed convection, relatively little information is available, and the reader is referred in particular to the review by Shenoy and Mashelkar [1982].

Turbulent Regime
For this regime it becomes necessary to pay particular attention to the distinction between purely viscous and viscoelastic fluids. Heat transfer to many purely viscous fluids can be adequately

1998 by CRC PRESS LLC

predicted by the use of simple relations involving the Nusselt number and apparent Reynolds and Prandtl numbers or by taking advantage of the analogy between friction and heat transfer for these fluids [Metzner, 1965; Cho and Hartnett, 1985]. The thermal entrance region for these fluids is generally similar to that of Newtonian fluids and very short (e.g., about 20 to 50 diameters for a tube). For viscoelastic fluids, on the other hand, the situation is very different. Some suspensions of very long fibers may exhibit such a nature, as will polymer solutions and many complex fluids. Dilute solutions of a polymer or surfactant in tube flow, in particular, are of great interest. These fluids may indeed exhibit dramatic friction and heat transfer reductions with respect to the solvent alone at the same Reynolds number. This effect is very large, even with very small (e.g., subpercent level) traces of polymer or surfactant additives. The level of drag and heat transfer reductions at a given Reynolds number will generally increase with concentration of drag-reducing additive in the solvent. Interestingly, there is a minimum below which the drag and heat transfer can no longer be reduced, the so-called drag and heat transfer reduction asymptotes [e.g., Matthys, 1991]. One advantageous feature of this asymptotic regime is that the concentration no longer plays a role there in determining the friction and heat transfer, which also means that the Nusselt number is no longer a function of an additional parameter besides the Reynolds and Prandtl numbers. Good correlations exist for this regimein fact, surprisingly robust ones. The asymptotes are apparently the same for all drag-reducing fluids. In the region between Newtonian and asymptotic behavior, the friction and heat transfer are functions of the concentration (i.e., viscoelasticity) of the fluid as well as of the diameter of the pipe. At this time, there is no universal technique that allows us to predict the heat transfer in this region based on simple material properties measurement. The diameter effect in particular is very challenging, and, even though some scaling methods have been proposed, there has not been enough experimental data made available to establish unambiguously the limits of their applicability. Note also that for a given concentration and diameter, the level of friction and heat transfer reduction in this regime will still depend on the Reynolds number (i.e., shear stress or shear rate), and a fluid may exhibit Newtonian behavior at low Reynolds number but then progressively revert to asymptotic behavior at high Reynolds number. In some cases there may be an onset shear stress that has to be exceeded before any drag or heat transfer reduction behavior is observed. Interestingly, the heat transfer is usually reduced proportionally more than the friction for drag-reducing fluids, (e.g., 10 times versus 5 times for a typical asymptotic solution). For polymer solutions, however, the heat transfer reduction will also decrease faster than the drag reduction as the polymer is degraded mechanically. It should also be noted that the entrance lengths for these fluids are much longer than for Newtonian fluids. For polymer solutions in tubes, for example, the thermal entrance length may easily be several hundred diameters long (compared to 20 or so for Newtonian or purely viscous fluids). The hydrodynamic entrance length, on the other hand, may reach "only" 100 diameters. These apparent discrepancies between heat transfer and friction are often attributed to some "uncoupling" between the two and a breakdown of the classic Newtonian Reynolds or Colburn analogies. The very long thermal entrance lengths, in particular, should be kept in mind when designing or analyzing heat exchangers for these fluids, as it is likely that the

1998 by CRC PRESS LLC

heat transfer may never reach fully developed conditions in many exchangers. Expressions exist that give the heat transfer in the entrance region as a function of distance [e.g., Matthys, 1991]. In appropriate nondimensional terms, the entrance region heat transfer development is independent of the Reynolds number in many cases. Interestingly, and contrary to the common belief that the entrance lengths are uncoupled for polymer solutions, recent studies have shown a very good coupling between the hydrodynamic and thermal developments for drag-reducing surfactant solutions [Gasljevic and Matthys, 1994]. Another important issue in studying or using viscoelastic fluids is that of degradation. Indeed, the polymeric fluids, for example, being macromolecular in nature, may be very susceptible to mechanical degradation. This means thatwhen subjected to flow in tubes and especially through pumps and filterssome macromolecules will be permanently broken and that the drag and heat transfer reductions will then be lost, partially at first and completely later on. This can be often seen as an increase in friction or heat transfer at high Reynolds number. Such degradation can also be caused by thermal or chemical processes. A consequence is that polymer solutions are not well suited for recirculating flows. Surfactant solutions, on the other hand, are much less susceptible to permanent degradation, although the drag and heat transfer reductions can be eliminated completely (but reversibly) under high shear stress (e.g., at high Reynolds number or in hydraulic components). They also possess the remarkable ability to reconstitute very rapidly after being subjected to high shear (e.g., in a centrifugal pump). In that respect they are very attractive and, even though little understood at this time, will undoubtedly become used more in the future. Many applications of such fluids are possible that would capitalize on their remarkable drag-reducing properties. We are studying presently, for example, the use of surfactant solutions as energy conservation agents in hydronic heating and cooling systems, a very promising application indeed.

54.4 Instrumentation and Equipment


Much of the equipment encountered or needed when dealing with non-Newtonian fluids is similar to that used with Newtonian ones, although, of course, some additional equipment may be needed for the quantification of the non-Newtonian nature of the fluid. Typically, this latter work requires the use of rheometers, perhaps over a wide range of temperatures. Numerous such devices exist and are discussed in books on rheology. As far as basic instrumentation, a few words of caution are in order. Pressure measurements, for instance, can be complicated by viscoelastic pressure hole errors. These can be minimized by paying particular attention to the shape and uniformity of the tap holes or by using differential pressure measurements whenever possible. Flow measurements may be particularly troublesome, and one should not rely automatically on flow meters designed for watersuch as orifice meters, turbine meters, etc.which can give large errors for non-Newtonian fluids unless a careful calibration has been conducted for the specific fluid and flow conditions. Positive displacement devices are more suitable. Ultrasonic flow meters may also be used depending on their built-in velocity profile calibration. Velocity measurements with typical LDV systems should be readily possible, but the use of hot-wire anemometry would probably lead to large errors in many cases if appropriate corrections are not introduced.

1998 by CRC PRESS LLC

For temperature measurements, thermocouples, RTDs, thermistors, thermometers, and so on are all generally suitable under steady state conditions. Note that the flow field may be affected, however, and that heat transfer may also be reduced, which may mean, for example, that a longer time might be needed before a steady state measurement is achieved. Indeed, unsteady measurements may be more difficult. When drag-reducing flows are involved, the temperature gradients in a tube, say, may also be much larger than for Newtonian fluids at a given heat flux because of the reduced convective heat transfer coefficients. More extensive mixing may then be necessary for bulk temperature measurements. Note also thatas discussed abovethe thermal entrance length may be very long for these fluids and that fully developed conditions may never be reached in practical situations. The performance of some heat exchangers may be significantly impacted by the fluids, especially drag-reducing fluids [e.g., Gasljevic and Matthys, 1993], and caution should be exercised there as well. Mixing devices may also perform differently with non-Newtonian fluids. Centrifugal pumps in most cases will work appropriately, even with improved efficiency at times, but may also degrade the fluid if excessive shear stress is applied. Pump and valve flow curves do not appear to be changed dramatically in most cases. Naturally, issues of erosion, corrosion, fouling, disposal, etc. should also be investigated for the fluid of interest.

Defining Terms
Asymptotic regime: The regime of maximum drag and heat transfer reductions that can be achieved by drag-reducing fluids. Degradation: A modification imparted to the fluid by mechanical, thermal, or chemical effects that changes its properties (e.g., a reduction in viscosity or drag-reducing capability). Heat transfer (and drag) reduction: A sometimes dramatic decrease in heat transfer (and friction) in turbulent flow due to the viscoelastic nature of the fluid. Non-Newtonian fluid: A fluid that does not obey Newton's law of simple proportionality between shear stress and shear rate. Its viscosity is then also a function of shear rate, time, etc. Nusselt number: A nondimensional number involving the convective heat transfer coefficient. It is a measure of the fluid capability to transfer heat to a surface under flow. Polymer: A material constituted of molecules made of repeating units (monomers). Prandtl number: A nondimensional number comparing the diffusivity of momentum and heat. It reflects the relative ease of transport of momentum (friction) versus energy at the molecular level. Purely viscous non-Newtonian fluid: A non-Newtonian fluid that does not exhibit elasticity. Reynolds number: A nondimensional number involving the flow velocity and the fluid viscosity. It is a measure of the flow rate and level of turbulence in the flow.

1998 by CRC PRESS LLC

Rheology: The study of flow and deformation of matter (generally associated with measurements of viscosity and other material properties for non-Newtonian fluids). Surfactant: A material leading to modified (e.g., reduced) surface tension effects because of dual hydrophilic and hydrophobic nature. Viscoelastic non-Newtonian fluid: A fluid thatin addition to variations of viscosity with shear ratealso exhibits an elastic character (i.e., has a memory, shows recoil, etc.). Weissenberg number: A nondimensional number reflecting the viscoelasticity of a fluid through its relaxation time (similar to the Deborah number).

References
Cho, Y. I. and Hartnett, J. P. 1985. Non-Newtonian fluids. In Handbook of Heat Transfer Applications, ed. W. M. Rohsenow, J. P. Hartnett, and E. N. Ganic, pp. 2.12.50. McGraw-Hill, New York. Dimant, Y. and Poreh, M. 1976. Heat transfer in flows with drag reduction. In Advances in Heat Transfer, vol. 12, pp. 77113. Academic Press, New York. Gasljevic, K. and Matthys, E. F. 1993. Effect of drag-reducing surfactant additives on heat exchangers. In Developments in Non-Newtonian Flows, ed. D. Siginer, vol. AMD-175, pp. 101108. ASME, Washington, DC. Gasljevic, K. and Matthys, E. F. 1994. Hydrodynamic and thermal field development in the pipe entry region for turbulent flow of drag-reducing surfactant solutions. In Developments in Non-Newtonian Flows II, ed. D. Siginer. Vol. FED-206, pp. 5161. ASME, Washington, DC. Hartnett, J. P. and Kostic, M. 1989. Heat transfer to Newtonian and non-Newtonian fluids in rectangular ducts. In Advances in Heat Transfer, vol. 19, pp. 247356. Academic Press, New York. Irvine, T. F. and Karni, J. 1987. Non-Newtonian fluid flow and heat transfer. In Handbook of Single-Phase Convective Heat Transfer, ed. S. Kakac, R. K. Shah, and W. Aung, pp. 20.120.57. John Wiley & Sons, New York. Matthys, E. F. 1991. Heat transfer, drag reduction, and fluid characterization for turbulent flow of polymer solutions: Recent results and research needs. J. Non-Newtonian Fluid Mech. 38:313342. Metzner, A. B. 1965. Heat transfer in non-Newtonian fluids. In Advances in Heat Transfer, vol. 2, pp. 357397. Academic Press, New York. Shenoy, A. V. and Mashelkar, A. R. 1982. Thermal convection in non-Newtonian fluids. In Advances in Heat Transfer, vol. 15, pp. 143225. Academic Press, New York. Skelland, A. H. P. 1967. Non-Newtonian Flow and Heat Transfer, John Wiley & Sons, New York.

1998 by CRC PRESS LLC

Further Information
The reader is also referred to journals such as International Journal of Heat and Mass Transfer, Journal of Heat Transfer, Journal of Non-Newtonian Fluid Mechanics, and Journal of Rheology. Symposia covering this subject are held at conferences by ASME, AIChE, and the Society of Rheology. Numerous excellent research papers in this field have also been published recently byamong othersthe groups led by Y. Cho, A. Ghajar, J. Hartnett, T. Irvine, R. Sabersky, A. Steiff, H. Usui, and J. Zakin. The reader is also welcome to contact the author for additional information.

1998 by CRC PRESS LLC

Anda mungkin juga menyukai