CCC ANNUAL REPORT UIUC, August 12, Heat Transfer During Spray Cooling Using Steady Experiments. * Xiaoxu Zhou * Brian G.

Transcription

1 CCC ANNUAL REPORT UIUC, August 12, 21 Heat Transfer During Spray Cooling Using Steady Experiments * Xiaoxu Zhou * Brian G. Thomas Alberto Hernandez Humberto Castillejos Andres Acosta * University of Illinois at Urbana-Champaign CINVESTAV, Unidad Saltillo, Mexico Project Objectives Quantify spray water cooling heat transfer rate by interpretating steady experiments with computational modeling Improve secondary-cooling zone temperature-prediction in continuous-casting of steel CON1D, CONONLINE Better control of steel quality Uniformly-distributed temperature in slab University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 2

2 Schematic of Experiment Setup Induction heating is used to heat up platinum sample Power controller is used to maintain sample temperature at a specific value University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 3 Setup Details Spray Nozzle Side View TC Wire Cylinder Plastic Cover X Ceramic Body Z Platinum Sample Y Box and Sample Top View Quartz Glass Copper Coil Top View Side View TC Wires Plastic cover and quartz glass are used to keep spray water from ceramic body, only exposing front surface of platinum sample to the water Induction Copper Coil University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 4

3 Steady Spray Cooling Experiments Experiment procedure: Pt sample is induction-heated with sample TC temperature (Ts) controlled in stages of 1, 2,, 11, 12, 11,, 2, 1 o C. Each Ts stage takes 8 min. Heating Cooling Two types of experiments: wet (spray is on) dry (spray is off) Sample locations during spraying: University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 5 Typical Ts and I tot Measurements from Wet Experiment Water Flow Rate=4.6lpm, Air Flow Rate=14lpm,Nozzle Centered (June 26, 9) Total current measurement approaches steady state at the end of the 8 min for each stage 25 s 2 s Sample temperature Ts=3 C University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 6

4 Typical Ts and I tot Measurements from Dry Experiments Water Flow Rate= lpm, Air Flow Rate= lpm,nozzle Centered (July-31, 9) Sample Temperature, o C Total Current, A Each sample temperature stage takes 5 min. Steady total current stays almost the same for heating and cooling, no hysteresis presented. University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 7 Induction Heating Modeling Electromagnetics equation Heat conduction equation Induced current density, J ind 1 2 T A = jωσ ( T ) A + J ext ρcp = ( k( T ) T ) + Q µ t A ω Magnetic potential angular frequency µ permeability ε σ Total current density, J tot permittivity electrical conductivity j Complex unit square root of -1 Q (Heat Source) Two equations are fully coupled via T and A Material Type Non-conductive (air, ceramic, water) J tot 2 /(2σ(T)) Conductive, with J ext (copper coil) J ind 2 /(2σ(T)) Conductive, without J ext (platinum) Commercial package COMSOL is chosen to do simulation University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 8

5 Unit: mm 2D Axi-symmetric model Modeling Domains and Boundary Conditions Axial symmetry A θ = 6 12 Magnetics Domain 4 h=1w/m 2 K R=13.5 Heat Transfer Domain h=1w/m 2 K Ceramic Body A θ = Axial symmetry Copper Coil Induction coil shape is complicated It consists of 1.5 loops Model needs calibration z Quartz Glass Window Platinum Sample θ r A θ = Thermocouple Ri=5.5 h_cw.5 4 h_spray h_front 1 University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 9 spray Materials Properties Table. 1 Electrical and Thermal Material Properties* Specific Heat Density Electrical Conductivity Thermal Conductivity (J/kg K) (kg/m 3 ) (1/m ohm) (W/m K) Copper /(1+.39(T-2)) 4 (25 o C~15 o C) Platinum /(1+.38(T-2)) T T 2 Water Ceramic ??? Ceramic thermal conductivity is missing, which requires model calibration too. Table. 2 Temperature-dependent Platinum Emissivity* Temperature, o C Emissivity * University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 1

6 Temperatu ure, oc Model Calibration by Dry Experiment to Decide R i and k ceramic (T) Calibration to decide R i TC Measurement Ri=5.8mm Ri=5.5mm Ri=5.mm Measured Current Time, s R i =5.8, 5.5 and 5. mm chosen to do transient simulation for dry experiment from 7 o C to 8 o C. R i =5.5 mm gives the best match of temperature with transient measurement R i =5.5mm used in calibrating model to get k ceramic k cermaic (T) is obtained by matching each TC measurement from 1 o C to 12 o C in steady state dry experiment simulation. 1. Comparison between TC measurement and prediction from transient dry experiment simulation. 2. good match; R i =5.5mm and k ceramic (T) are good to use. Measured Curre ent, A C Thermal Condu uctivity, (W/mK) Calibration to obtain ceramic thermal conductivity k ceramic (T) Temperature, o C Time, s University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 11 Temperature, o C Measurment Prediction Side Experiment for Validating Calibrated Ceramic Thermal Conductivity Cylindrical ceramic body resting on a thin metallic sheet Sheet is heated up by a Bunsen burner Top surface is exposed to natural convection in the laboratory environment The lateral surface is isolated Steady temperature profile along the axis is measured by 9 TCs TC Location, mm TC Location mm University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 12

7 4 Side Experiment Modeling and Validation Results Note: K ceramic (T) used in simulations Temperature,C e,c test2 test3 Modeling Distance,m 35 test4 6 test5 3 test1 Modeling 5 Modeling Distance,m Distance,m University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 13 Temperature,C Temperature,C Example Case Modeling to Extract HTC Methodology: steady-state simulation adjust h_spray to match temperature prediction with TC measurement Conditions Ts=7 o C total current = A h_spray = 71 W/m 2 K h_cw = 2.96e4 W/m 2 K h_front = 55 W/m 2 K nozzle water flow rate = 4.6lpm air flow rate = 14lpm position: X=mm, Y= mm (centered), Z=mm University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 14

8 Magnetic Potential Distribution and Total Current Distribution Magnetic potential distribution Total Current distribution Magnetic potential radiates away from the region of high current flow which caused it Magnetic potential can only penetrate into the sample at a very small distance ~.4mm (skin depth) Induced current is stronger at the right side of the sample, while total current is stronger at the left side of the coil. University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 15 Heat Source Distribution and Temperature Distribution Heat Source distribution Temperature distribution Heat Generated, W Sample Coil Heat Out, W Spray Coil Window Natural convection Platinum Sample Radius, mm Total Total University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 16 Temperature, o C C D

9 ray Heat Transfer Coefficient, kw/m /m^2k Spr Nozzle Centered Spray Heat Transfer Coefficients Y=mm(Nozzle Centered) WaterFlow Rate=4.6lpm, Heating WaterFlow Rate=4.6lpm, Cooling WaterFlow Rate=3.5lpm, Heating WaterFlow Rate=3.5lpm, Cooling WaterFlow Rate=2.5lpm, Heating WaterFlow Rate=2.5lpm, Cooling Sample Surface Temperature, C Increasing water flow rate increases heat transfer coefficient. During heating, HTC peaks around 15~2 C, then decreases as sample surface increases During cooling, HTC keeps increasing gradually. Hysteresis is shown in heat transfer coefficient curves. HTC is independent of temp at temp >85 C University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 17 Nozzle Centered--Spray Heat Flux 5. Y=mm(Nozzle Centered) Spray Heat Flux, MW/m^ WaterFlow Rate=4.6lpm,Heating WaterFlow Rate=4.6lpm,Cooling WaterFlow Rate=3.5lpm,Heating WaterFlow Rate=3.5lpm,Cooling WaterFlow Rate=2.5lpm,Heating WaterFlow Rate=2.5lpm,Cooling Transient_2.2lpm Transient_1.761lpm Transient_3.321lpm Transient_3.34lpm Sample SurfaceTemperature, C Increasing water flow rate increases spray heat flux. Spray heat flux also shows hysteresis Leidenfrost temperature is around 85 o C --Transient results by Sami Vapalahti, etc, Spray Heat Transfer Research at CINVESTAC, Steady measurement gives higher heat flux than transient measurement P26, CCC report, 27 University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 18

10 Mechanism of Hysteresis Water Layer sample Steam Layer sample Spray Droplet Water Layer Fig. 1 Fig. 2 Heating process: spray droplet impinges on water layer, touches surface, boils, and takes heat away. (Fig. 1) High heat removal keeps surface cold. Cooling process: At high sample surface temperature (>~86oC), a stable steam layer forms on the sample surface. (Fig. 2) This steam layer acts as a barrier to heat transfer and decreases heat removal. Low rate of heat removal sustains the air gap to low temperatures before droplets finally can penetrate through Result: difference in heat transfer at intermediate temperatures according to history (heating or cooling) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 19 t, kw/m 2 K Spray Heat Transfer Coefficient Nozzle Water Flow Rate=4.6lpm --Spray Heat Transfer Coefficients WaterFlowRate=4.6lpm Y=mm,Heating Y=mm,Cooling Y=9mm,Heating Y=9mm,Cooling Y=18mm,Heating Y=18mm,Cooling Sample Surface Temperature, C HTC,W/m 2 K C 3 C 5 C 7 C 9 C 11 C Water Flux Rate Hysteresis exists for different location from spray centerline. Moving further away from spray centerline decreases HTC. Difficult to correlate water flow rate footprint measurements with HTC. More details of spray dynamics needed (droplet distribution, size, velocity, etc, --collaboration work at CINVESTAV, Mexico) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou Y Direction,mm Water Flux Rate, l/m 2 s

11 Nozzle Water Flow Rate=4.6lpm --Spray Heat Flux 5. WaterFlowRate=4.6lpm Spray Heat Flux, MW/m Y=mm,Heating Y=mm,Cooling Y=9mm,Heating Y=9mm,Cooling Y=18mm,Heating Y=18mm,Cooling Sample Surface Temperature, C Moving further away from spray centerline decreases spray heat flux Hysteresis is shown in heat flux curves University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 21 Nozzle Water Flow Rate=2.5lpm --Spray Heat Transfer Coefficients Spray Heat Transfer Coefficient, kw/m^2k WaterFlowRate=2.5lpm Y=mm,Heating Y=mm,Cooling Y=9mm,Heating Y=9mm,Cooling Y=18mm,Heating Y=18mm,Cooling Sample Surface Temperatrue, C Location closer to the center gives higher heat transfer coefficient. Hysteresis is shown in spray heat transfer coefficient curves University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 22

12 Nozzle Water Flow Rate=2.5lpm --Spray Heat Flux 3.5 WaterFlowRate=2.5lpm 3. Spray Heat Flux, kw/m^ Y=mm,Heating Y=mm,Cooling Y=9mm,Heating Y=9mm,Cooling Y=18mm,Heating Y=18mm,Cooling Sample Surface Temperature, C Location closer to the center gives higher spray heat flux Hysteresis is shown in heat flux curves University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 23 Conclusions for Experiments and Modeling This report presents a new technique combining experimental measurements with computational modeling to obtain local heat fluxes of water spray cooling a platinum sample for various nozzle operating conditions. Apparatus (induction heating) was modeled for 23 sample temperatures in each heating-cooling cycle. Spray heat transfer coefficient and heat flux curves for 3 different nozzle operating condition and 3 different positions from the spray centerline are shown in the report. Heat transfer coefficient varies from 2W/m 2 K to 35W/m 2 K. Heat flux varies from.5mw/m 2 to 6MW/m 2. Total power loss to the ambient air and the front quartz glass window is relatively low (less than 8% of total heat generated). The fraction of power going to the spray is around 45~55%. University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 24

13 Conclusions for Practical Results Both spray heat transfer coefficient and spray heat flux show hysteresis which likely is related to formation of vapor layer on the sample surface. The Leidenfrost temperature (minimum heat flux) is around 84~86 o C for platinum. Heat transfer coefficient around surface temperature 1~6 o C for heating is much larger (2~5%) than that during cooling, for all operating conditions studied. Increasing water flow rate increases spray heat transfer coefficients and spray heat fluxes by 1~6%, for the same nozzle position. Moving further away (9mm, 18mm) from the spray centerline decreases spray heat transfer coefficients and fluxes by 1~7%, for the same nozzle operating condition. The decrease is more gradual than the drop in water flow would suggest. Heat transfer coefficient decreases as the sample surface temperature goes from 2 o C to 12 o C, while increasing from 12 to 1 o C during cooling. University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 25 Plant Observations and Measurements Compared with Prediction by CON1D Xiaoxu Zhou Brian G. Thomas Department of Mechanical Science and Engineering University of Illinois at Urbana-Champaign University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 26

14 Plant Measurements at Nucor (Whale and Pyrometer Measurements) Three cases of whale / no whale (Nucor Dec 9, 23) Case 1 Case 2 Case 3 Casting Speed (ipm) Spray Pattern (less water ) Plant Observation No Whale No Whale C. Experimental data* Whale Two cases of pyrometer measurements (Case 4-1, low spray; Case 4-2, high spray) (Nucor Jan 13, 26) A. Caster operation conditions Parameter Time Casting Speed Spray Pattern Caster Pouring Temp Value Jan. 13, 26, 16:1-16:4 hrs ipm (3.61 m/min) Pattern 4 (low spray), 7 ( high spray) South o C B. Pyrometers locations Pyrometer Distance from Meniscus, mm *Amar s CCC annual meeting 26 University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 27 Simulation Details Note: Wedged-hat spray heat transfer coefficients are used * Spray widths come from Sami s experimental footprint measurements* Measured mold heat flux from the Nucor plant is used Superheat=(no superheat flux, but initial temperature starts from pouring temperature) spray width length (m) (m) Calibrated spray HTC functions Spray HTC tuning parameters z1 z2 z3 h1 h2 h *Sami Vapalahti, 26 CCC Meeting Report: Delavan Nozzle Characterization at CINVESTAV University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 28

15 Whale Formation Prediction by CON1D Case 3 Whale Scale Surface Temp Shell Thickness Case 2 No Whale Scale Surface Temp Shell Thickness University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 29 Whale Formation Prediction by CON1D Scale Surface Temp Shell Thickness Case 1 No Whale University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 3

16 Strand Surface Temperature Prediction by CON1D and Comparison with Measurements (black dots) Case 4-1 (low spray) T=12 o C T=8 o C Scale Surface Temp Shell Thickness Measurements Case 4-2 (high spray) T=13 o C T=9 o C Scale Surface Temp Shell Thickness Measurements University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 31 Discussion Good match with observation of whale formation. Good match with last three pyrometers. Bad match with the first two pyrometers, may due to dense steam. Possible method to make simulation match better: Include a scale layer in CON1D scale layer is always observed in casting, scale layer decreases heat transfer rate, increases metallurgical length, scale layer surface temperature should be lower than steel surface temperature, lower scale surface temperature and longer metallurgical length is favorable in the current simulations. scale effect is studied in the next talk. University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 32

17 Spray HTC Comparison Spray HTC, W/m^2K 4 Avg_11.43 L/m^2s (Mold3.5-9) L/m^2s L/m^2s Avg_3.71 L/m^2s ( ) 3.32 L/m^2s Avg_1.79 L/m^2s ( ) CON1D output Avg_.56 L/m^2s (W ) 2.37 L/m^2s.26 L/m^2s Avg_.21 L/m^2s (W ) Avg_.9 L/m^2s (W ) Avg_.18 L/m^2s (W19822) 4.6lpm 95C 4.6lpm 15C 4.6lpm 115C 2.5lpm 95C 2.5lpm 15C 2.5lpm 115C 4.6lpm_Nozaki Prediction 2.5lpm_Nozaki Prediction (W19822) Lab Measurements Based on lab measurement of water impact density L/m^2s Measured spray heat transfer coefficients (HTC) from CINVESTAV are compared with those used by CON1D in case 3. Most measured spray HTCs lie in the range of what used by in case 3, except HTCs from 4.6lpm experiments at the spray centerline and one HTC from 4.6lpm at the surface temperature of 95 o C. High spray HTCs for 4.6lpm at the spray centerline is mainly caused by the high impact water density (2.18L/m 2 s). The shape of measured HTCs follow the shape of what are used in CON1D, except the ones from 4.6lpm at 95 o C. Besides impact water density, such parameters as spray water pressure, air pressure, nozzle type and nozzle standing distance also make difference between measured spray HTCs and those used in CON1D. Spray HTCs predicted by Nozaki equation lie in the range of what is used in CON1D University of Illinois at Urbana-Champaign Spray Width, mm Metals Processing Simulation Lab Xiaoxu Zhou 33 Spray Heat Flux Comparison L/m^2s Spray HTC, W/m^2K L/m^2s 11.6 L/m^2s 2.37 L/m^2s L/m^2s.26 L/m^2s Avg_11.43 L/m^2s (Mold3.5-9) Avg_3.71 L/m^2s ( ) Avg_1.79 L/m^2s ( ) Avg_.56 L/m^2s (W ) Avg_.21 L/m^2s (W ) Avg_.9 L/m^2s (W ) Avg_.18 L/m^2s (W19822) 4.6lpm 95C 4.6lpm 15C 4.6lpm 115C 2.5lpm 95C 2.5lpm 15C (W19822) Lab Measurements 2.5lpm 115C 4.6lpm_Nozaki 1 C 2.5lpm_Nozaki 1 C CON1D output Based on lab measurement of impact water density Spray Width, mm University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 34

18 Summary Calibrated spray HTC functions in CON1D are able to match whale observations and the last three pyrometers measurements, but not do well in matching the first two pyrometer measurements. Most measured spray HTCs lie in the range of what are used by in case 3, except HTCs from 4.6lpm experiments at the spray centerline and one HTC from 4.6lpm at the surface temperature of 95 o C. The shape of measured HTCs follow the shape of what used in CON1D, except the ones from 4.6lpm at 95 o C. University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 35 Acknowledgements Continuous Casting Consortium Members (ABB, Arcelor-Mittal, Baosteel, Corus, LWB Refractories, Nucor Steel, Nippon Steel, Postech, Posco, ANSYS-Fluent.) National Science Foundation GOALI DMI (Online) University of Illinois at Urbana-Champaign Metals Processing Simulation Lab Xiaoxu Zhou 36