Two-Phase Flow Pattern, Heat Transfer, and Pressure Drop in Microchannel Vaporization of CO 2

Transcription

1 Properties of saturated CO Two-Phase Flow Pattern, Heat Transfer, and Pressure Drop in Microchannel Vaporization of CO Density Saturated liquid Saturated vapour,,,5 Surface tension Jostein Pettersen / Armin Hafner ρ, kgm - σ, Nm -,4, 4, Norwegian University of Science and Technology Trondheim Norway - - 4, T, o C T, o C 4 Content Flow pattern recordings Background and purpose of study Eperimental method Flow visualization results Heat transfer and pressure drop data Dry-out issues Correlation of heat transfer and pressure drop Maldistribution Conclusions MOTIVATION: What are the consequences for the heat echanger design? Tube diameter. mm Digital video recorded at 4 frames/s (fps) Observation from an angle of o from above Play-back at fps Each test recorded at stable operating conditions T = o C q = kwm - G = 5 and kgm - s - 5 Transcritical CO cycle with internal heat echange =. By: Jostein Pettersen Mass flu G = 5 kgm - s - =. =.5 =.5 h =.

2 =. By: Jostein Pettersen =. By: Jostein Pettersen Mass flu G = 5 kgm - s - =.5 =. Mass flu G = kgm - s - =.5 =. =.4 % lubricant % lubricant 5 % lubricant =. =.5 By: Jostein Pettersen =. By: Jostein Pettersen Mass flu G = 5 kgm - s - =.5 Mass flu G = 55 kgm - s - % lubricant % lubricant 5 % lubricant =.5 % lubricant % lubricant 5 % lubricant =. % lubricant % lubricant 5 % lubricant % lubricant % lubricant 5 % lubricant =. By: Jostein Pettersen Flow regime maps Based on all observations at T= o C. Approimate transition lines Mass flu G = 5 kgm - s - =. =. % lubricant % lubricant 5 % lubricant jl, ms -,,5,4,,,,,,5,,5,,5,,,4,,, j v, ms - Stratified flow not observed (q > ) Generalised transition lines from literature did not fit observed regimes Transition to annular flow observed at very low superficial velocity Intermittent Annular Droplet Dispersed G, kgs - m Intermittent Annular Droplet Dispersed

3 ..5 Conclusions flow visualization Flow pattern observations with pure CO were dominated by intermittent flow at low vapour fractions, and wavy annular flow with entrainment of droplets at higher vapour fractions. Did not match predicted flow pattern maps. With lubricant Reniso 5E (,, 5%), almost every test shows a film flowing annulus along the tube wall. This film was assumed to consist mainly of oil. Thicker at higher oil concentrations. Intermittent flows are more dominant at smaller mass flues whereas only annular flow was observed at a higher mass flu of 55 kg/ms. Elongated vapour bubbles seemed to be coated with an oil film. This would tend to increase the surface tension and thus decrease the heat transfer coefficient. Observed flow patterns are quite similar with tests, made with pure CO In tests made with oil, no entrainment was found, which however does not necessarily mean that there was no entrained droplets in the flow. The material should be closer studied before final conclusions are made Local heat transfer data of Hihara and Tanaka () D = mm, T = 5 o C Geometry of heat transfer test tube Flow visualization test rig. Equivalent to tubes used in prototype heat echangers Etruded from aluminium alloy 5 flowchannels Length 54 mm Heated length 5.4 mm Tests conducted with horizontal tube 4 Observation tube ID =. mm Heat transfer test section design Test conditions Flow vaporization eperiments Water inlet/outlet Water flowchannel Parameter Symbol Unit Range Mass flu G kgm - s Heat flu q Wm - 5,, Evaporating temperature T oc - 5 Heat transfer tube CO manifold 5

4 Heat transfer test data Varying heat flu, T = o C Friction pressure drop test data Varying mass flu and temperature Heat transfer coefficient h, Wm - K - G = kgm - s - 5 G = kgm - s - 5 W/m 5 kw/m 5 kw/m 5 q = 5 5 q = 5 q = q = q = 5 q = 5,,,,,4,4,,,,,, Vapour fraction h, Wm - K - h, Wm - K - G = 5 kgm - s - 5 kw/m kw/m 5 q = q = (a) q = (b),,,4,,, Specific friction pressure drop dp/dz (kpa m - ) 5 4 G = G = T = o C q = kwm - G =,,,4,,, Vapour fraction dp/dz (kpa m - ) 4 T = T = T = (a) T = (b) q = kwm - G = kgm - s -,,,4,,, Heat transfer test data Varying mass flu, q = kwm - Pressure drop & heat transfer T = o C q = kwm - Heat transfer coefficient h, Wm - K - T = o C 5 5 G = kg/(m s) G = G =,,,4,,, Vapour fraction h, Wm - K - T = o C 5 G = 5 G = G = G = 5,,,4,,, dp/dz (kpa m - ) 5 kg/(m s) 4 kg/(m s) kg/(m s) G = G = G =,,,4,,, Vapour fraction h, Wm - K G = kg/(m s) G = G =,,,4,,, Note: Increased mass flu does not improve heat transfer but gives higher pressure drop Heat transfer test data Varying temperature Pressure drop & heat transfer T = o C q = kwm - Heat transfer coefficient h, Wm - K - T = T = T = (a) T = (b) T = 5 q = kwm - G = kgm - s -,,,4,,, Vapour fraction h, Wm - K - q = kwm - G = 5 kgm - s -,,,4,,, T = T = T = T = 5 Vapour fraction 4 4

5 Dryout vs Departure from Nucleate Boiling (DNB) Comparison of heat transfer model and eperimental data Dryout is CHF due to discontinuation of the liquid film on the tube wall, usually in annular flow in low/medium- flow due to disruption of liquid layer caused by surface wave instability in high- annular flow caused by dryup of the liquid layer on the heating wall due to entrainment and vaporization Dryout is not to be confused with DNB (Departure from Nucleate Boiling), which is a film boiling phenomenon DNB: Increased G gives higher turbulence, improved bubble transport and higher CHF Dryout: Increased G gives more entrainment and reduced CHF (lower cr ) 5 h, Wm - K q = q = Calc q = Calc q =,,,4,,, Earlier dryout than predicted may be due to flow oscillation in parallel flowchannels T = o C G = 5 kgm - s - h, Wm - K T = T = T = T = 5 Calc T = Calc T = Calc T = Calc T = 5 q = kwm - G = kgm - s -,,,4,,, Correlation of heat transfer data Mechanism Model Cooper (4) Nucleate boiling Gorenflo () Convective evaporation Kattan et al. () a M odel for com bining Asym ptotic m odel b, nucleate and convective with n = he at tra nsfer Kon kov (5) data scaled from H Dryout inception O to C O using Ahm ad () c Post-dryout heat transfer Shah and Siddiqui ().. 4 k l a: h ce. Re l Pr l n b: n / h ( h n nb ) ( h ce ) c: Using scaling factor based on W eber-reynolds number or Barnett num ber Best combination: Cooper (4) nucleate boiling correlation, and Barnett number version of Ahmad () scaling factor Friction pressure drop dp/dz calculated, (kpa m - ) 5 4 Compared to Friedel () +% 4 5 dp/dz measured, (kpa m - ) -% dp/dz calculated, (kpa m - ) Compared to Lombardi and Carsana () CESNEF- +% dp/dz measured, (kpa m - ) Mean deviation:.%.4% Average deviation: -.4% -.% Underprediction at low mass flu and low temperature -% Correlation of heat transfer data Calculated mean heat transfer coefficients based on points between measured inlet and outlet vapour fraction h calculated, Wm - K h measured, Wm - K - +% -% Average deviation +.% Mean deviation 5.% Overprediction when model predicts too high cr Underperdiction due to Cooper (4) nucleate boiling correlation Change to Gorenflo () correlation gave significant overprediction, and average/mean deviation of appro. 5% Convective contribution may be too high Conclusion evaporation inside micro channels Intermittent and annular flow regimes dominate Droplet (entrained) flow important at high mass flu Reduced and irregular film thickness may give dryout and reduced heat transfer even at moderate vapour fraction Observed two-phase flow regime transitions were not predicted well by eisting models and generalized flow charts Etreme variation in measured CO heat transfer coefficient Nucleate boiling dominates prior to dryout Non-equilibrium effects in post-dryout heat transfer Combination of models for nucleate boiling, convective evaporation, dryout, and postdryout heat transfer is needed 5

6 Conclusion (evap. cont.) Temperature s Influence of lubricant may be significant and has to be investigated Design advice for microchannel evaporators (T > o C): Use low mass flu heat transfer remains unchanged, pressure drop is reduced, and critical vapour fraction is increased Use liquid (over)feeding assures nucleate boiling, reduces flow distribution problems, avoids dryout/post-dryout heat transfer Temperatures are measured at ten locations unequally distributed over the length of the test section. Thermocouples (Type E) are made of. mm constantan/chrome wire, soldered together and embedded in.5 mm deep and mm long slits (Fig. below). Electrically insulating, thermally conductive epoy and firmly pressed indium were used to ensure optimum contact between thermocouples and the tube wall, and to avoid the disturbance of heat transfer between heating cables. heating cables thermally conductive epoy section thermocouples mm diameter aluminium tube Tube wall temperature at various heat flues on different points of temperature Test Facility Condenser Receiver Pump 4 Flow Meter 5 Preheater Section Test Section wall temp. tw [ C] Q = W/m 4. Thermocouple Position mean.. wall temp. tw [ C] Q = 5 W/m The design of the test rig is based upon technology and eperience developed in our laboratory for eperimental research on heat transfer and pressure drop, by: Measured heat transfer coefficient, t sat = - C; G = kg/m ; Q = W/m.. using a fully instrumented test section consisting of two 5 mm long horizontal smooth tubes, connected by a vertical bend (r = 5 mm), to represent the typical part of an evaporator. The inner diameter of the tube is mm and the wall thickness is,5 mm.. using a preheater to secure fully developed flow of defined vapour fraction into the measuring section, to secure stable s at different points along the evaporating process.. using electrical heating to provide a known specified heat flu in the test section, so that local heat transfer coefficients may be found by measuring the temperature difference between the tube wall and the evaporating refrigerant at locations of known pressure and vapour fraction. [W/m K] 4 initial data..4.. [W/m K] 4 after calibration..4..

7 Measured heat transfer coefficient [W/m K] error interval with a random error in the temperature of +/-. K. (t sat = - C; G = kg/m ; Q = W/m ). [W/m K] Influence of varying saturation temperature t sat = -5 C t sat = - C t sat = 5 C (tsat= -5 C; G = kg/m s; Q = W/m ) maesurement [W/m K] (tsat= - C; G = kg/m s; Q = W/m ) [W/m K] (tsat= 5 C; G = kg/m s; Q = W/m ) C, completely new variation, where the heat transfer is decreasing constantly; from a very high value of around 4. W/m K at vapour fraction., to a value around. W/m K near to.. Influence of varying heat flu (heat transfer coefficient) Influence of varying heat flu (pressure drop) [W/m K] Q = kw/m Q = kw/m Q = kw/m (tsat= - C; G = 4 kg/m s; Q = W/m ) [W/m K] (tsat= - C; G = 4 kg/m s; Q = W/m ) maesurement [W/m K] (tsat= - C; G = 4 kg/m s; Q = W/m ) Q = kw/m Q = kw/m Q = kw/m (tsat= - C; G = 4 kg/m s; Q = W/m ) 5 4 dp dp dp dp..4.. (tsat= - C; G = 4 kg/m s; Q = W/m ) 5 4 dp dp dp dp..4.. (tsat= - C; G = 4 kg/m s; Q = W/m ) 5 4 dp dp dp dp..4.. vapour fraction increasing heat flu at a high mass velocity (4 kg/m s) is changes the pattern from a convective to a nucleate boiling regime [W/m K] Influence of varying mass flu (tsat= - C; G = kg/m s; Q = W/m ) 4 G = kg/m s G = kg/m s G = 4 kg/m s..4.. [W/m K] (t sat= - C; G = kg/m s; Q = W/m ) [W/m K] (tsat= - C; G = 4 kg/m s; Q = W/m ) nucleate boiling has a strong effect at lower mass flues, so that only a small increase in heat transfer is gained from increasing the mass velocity from to kg/m s. maesurement Influence of varying mass flu 5 4 G = kg/m s G = kg/m s G = 4 kg/m s (tsat= - C; G = kg/m s; Q = W/m )..4.. dp dp dp dp 5 4 (t sat= - C; G = kg/m s; Q = W/m )..4.. dp dp dp dp (tsat= - C; G = 4 kg/m s; Q = W/m ) dp dp dp dp

8 .. Influence of varying saturation temperature t sat = -5 C t sat = - C t sat = 5 C (tsat= -5 C; G = kg/m s; Q = W/m )..4.. vapour fraction dp dp dp dp (tsat= - C; G = kg/m s; Q = W/m )..4.. dp dp dp dp (tsat= 5 C; G = kg/m s; Q = W/m )..4.. vapour fraction dp dp dp dp Heat transfer at various mass flues [kg/(m s)] The influence of the mass flu on the heat transfer coefficient of CO is significant. If the mass flu is increased from to kg/(m s) the calculation as well as the show a rise in the heat transfer coefficient of about 4%, and a rise of about % if the mass flow is increased to kg/(m s). The peak in the graph indicates the pseudocritical temperature. At bar the maimum c p is located at 4. C. Measured pressure drop and the Fuchs-Correlation (tsat= - C; G = kg/m s; Q = W/m ) dp dp dp dp Correlation: Fuchs Fuchs, P. H., 5, Trykkfall og varmeovergang ved strømning av fordampende væske i horisontale rør og bend, Dr. Lic. grad, Department of Refrigeration, NTH, Trondheim, Norway. 5 4 (tsat= - C; G = 4 kg/m s; Q = W/m ) dp dp dp dp Correlation: Fuchs Influence of heat flu to the heat transfer coefficient ( or kw/m ) The influence of varying heat flu on the heat transfer coefficient was rather small. In this test series, the heat flu was either or kw/m. The heat transfer coefficient varied less than to % as a result of varying heat flu. The measured values show a high accuracy compared with the calculated results. From Gnielinski's correlation one can find an eplanation of the (negligible) influence of the heat flu on the heat transfer coefficient; The influence is taken into account by means of the correction factor (Pr/Prw).. Since the wall temperature changes with the heat flu, the Prandtl number at the wall temperature (Prw) changes as well. However, due to the small eponent of. the correction factor is almost for the test conditions applied. Gascooling inside the microchannel tube 5 circular ports/channels with. mm inner diameter Influence of the pressure on the heat transfer coefficient (- bar) Since the heat transfer coefficient reaches a maimum at the pseudocritical temperature, this maimum will change with varying pressure. As may be observed, the ma-imum heat transfer coefficient increases when the pressure approaches the critical pressure of CO (. bar). For a mass flu of kg/(m s) and a heat flu of kw/m, the peak heat transfer coefficient is about, W/(m K) for a pressure of bar, and about, W/(m K) at bar. This corresponds to a more than % rise when the pressure is decreased from to bar. An increase of the pressure from to bar leads to a drop of the maimum heat transfer coefficient of about %. A more or less negligible influence of the pressure on the heat transfer coefficient can be seen in case of temperatures "far away" from the pseudocritical value, i.e. at rather low or high temperatures. This is visualised by the asymptotic approach of the curves below 5 C and above C. Again, the deviation between and calculation was small.

9 Measurements versus correlations "simple Gnielinski" means Gnielinski s correlation for the Nusselt number using the Haaland () friction factor and neglecting the influence of the wall temperature (see above). The other variant of the original Gnielinski correlation was "Gnielinski (VDI)" where the Filonenko (54) friction factor was used. l is the tube length, and f is the pressure drop factor. Influence of the pressure [-- bar] on the pressure drop ( l tube = 5 mm) Since the pressure has a large influence on the pseudocritical temperature, the pressuredrop characteristics change with the absolute value of the pressure. If the pressure changes from to bar and from to bar, the mean pressure drop changes about +% and -%, respectively, in the temperature range to C. The corresponding change of the heat transfer coefficient was + and -%, respectively. is the tube roughness. Influence of the mass flu [-- kg/(m s)] on the pressure drop ( l tube = 5 mm) For the single-phase pressure drop: VDI (heat atlas) recommends the implicit form for the friction factor according to Colebrook & White: As epected the mass flow rate influences the pressure drop considerably. In case of an increase of the mass flu from to kg/(m s) and from to kg/(m s), the measured pressure drop increased by about % and 4%, respectively. As stated before, the heat transfer coefficient increased by about 4% and %, respectively, at the same conditions. Summary gascooling with micro channels In a transcritical cycle, the refrigerant is cooled down at a supercritical pressure. In this region the influence of the critical point on the properties is large. This fact leads to conditions in CO equipment differing considerably from in systems using conventional refrigerants. The eperimental results confirm that CO offers high heat transfer coefficients at supercritical pressures. A comparison between eperimental data and common correlations showed an acceptable correspondence. Especially the Nusselt number based on Gnielinski's correlation in combination with Haaland s friction factor correlated the eperimental heat transfer data well. The epected pressure drop of CO in a refrigerant cooler is rather high, but due to the high pressure level, the effect on temperature loss is moderate. A comparison between eperimental data and calculated results according to the Colebrook & White correlation showed a satisfactory agreement. Influence of the heat flu [ or kw/m ] on the pressure drop ( l tube = 5 mm) Npressure drop. In general, the influence of varying temperature on pressure drop changes significantly close to the pseudocritical temperature. The steep gradient is caused by the rapid change in CO -density in this region, Flow maldistribution (Results from Vist, 4) Distribution of flow into parallel flowhannels between tubes in a manifold between parallel ports in a tube Non-uniform distribution of singlephase flow Separation of two-phase flow in manifold Reduced performance Uneven frosting R-4a, G = / kgm - s - (manifold/tubes), in =. Vapour fraction [-],,,,4, inlet =.5 inlet =.4 inlet =. inlet =. 4 Tube # 54

10 mm mm 4mm 4mm Eperimental test rig Conclusions maldistribution P E T m p m T Water condenser Water condenser m p T T p p E p Outlet manifold Inlet manifold T T E m Eperimental s: Severe maldistribution of both phases: Vapour entering the first tubes and liquid entering the last tubes The ID mm manifold showed improved distribution Literature T-junction correlations: Good predictions of mm manifold data Larger deviations compared to mm manifold data Low manifold mass flu: correlation between branch vapour fraction and manifold vapour mass flu High manifold mass flu: constant liquid take-off fraction The tested manifold can be analysed as a series of T-junctions 55 5 Test section Eamples of heat echanger designs 5 Manifolds, round tube and multiport tube (MPE) ECO Group - Tube-in-fin mm Refrigerant Inlet mm Refrigerant Inlet Consept : Tube OD. mm Fin spacing mm Vertical tube spacing 5 mm Horizontal tube spacing.5 Tube configuration Staggered Consept (Chages compared to ): Fin spacing.4....mm Horizontal tube spacing.5 mm 5

11 ..5 BEHR Industrietechnik - Tube-in-fin Plate in Shell Eample: Tube OD Fin spacing Vertical tube spacing Horisontal tube spacing Tube configuration.44 mm mm 5 mm 5 mm In-line Capacity 5 to kw/unit Design Temperature C to + C Design Pressure Standard design, 5 and 4 bar up to bar 4 Multiport Etruded Tubes - MPE Advantages: Compact Low cost for larger series Disadvantages: Defrosting and water shedding may be a challenge Tube-in-shell Advantages: All capacities Several manufacturerers Disadvantages: Large volumes Heavy. Components for MPE heat echangers: Hydro Aluminium 5 Tube-in-tube Printed Circuits Printed-circuit CO or water, low pressure side CO High pressure side Capacity: According to design Design Temperature: Cryo to + C Design Pressure up to 5 bar

12 Why internal heat echanger? Ensures superheated gas to the compressor To avoid liquid slugging Increased compressor efficiencies Increased cooling capacities at high ambient temperatures Reduced optimal high pressure Increased energy efficiency at high ambient temperatures Different types of internal heat echangers (IHX) have been developed Tube-in-tube Star in tube Brazed MPE tube type Soldered tube-by-tube Internal Heat Echange (Mobile A/C) From U of I - Pega Hrnjak Idling 4 o C ambient Application of an IHX (water heat pump C C) t [ C] 5 Condenser 4 Throttle 5 bar Compressor t [ C] 4 5 Condenser Throttle Compressor IHX 5 Evaporator 4 Evaporator Accu 5 Accu h [kj/kg] h [kj/kg] 5 bar IHX may damp the influence of high-side pressure on COP & increases capacity Eamples of Internal heat echanger designs Etruded tube counterflow type Coiled counterflow type (Sakakibara et al., Eur. Pat. Appl.)