Home

Exploratory Research on MEMS Technology for Air

1. 3 5 3 0 LMF by direct measurement A LMF by energy balance Linear LMF by energy balance 2 5 Linear LMF by direct measurement Se 2 0 LL 215 1 0 05 f 0 0 T T T 0 0 0 2 0 4 0 6 0 8 1 0 1 2 1 4 1 6 Mass Flowrate of Stream 2 g sec Figure 5 3 Comparison of LMF calculated from 1 direct measurement method and 2 energy balance method 31 Figure 5 3 also provides a calibration for estimating the systematic error in using the energy balance method for determining the time averaged liquid mass fraction The LMF as a function of ri for both methods can be described by the following two linear relationships LMFEj 2 145 m 5 4 LMFE oy 2 145 m 0186 5 5 Writing these two equations in terms of Mm and setting them equal will give the following equation for the actual LMF from the LMF determined by the energy balance method LMF 0 186 LMF LMF irecr EV 5 6 Eq 5 6 will be used in section 5 3 to calculate the liquid mass fraction present at the evaporator outlet for the characterizing evaporator exit flows experiment Now that the systematic error in the energy balance method for calculating LMF can be accounted for by using eq 5 6 it is possible to accurately measure the time averaged LMF present in a multi channel evaporator system In section 5 2 the outlet conditions of a multi channel evaporator fed by either
2. M 12 3 2 Theory of Operatloh epe tnte teet eh nne E eoo Dn Doe Fi sect esee Ca n IE EN sees uae Cane e Cu steers 12 3 3 Minimum Gap Serpentine RTD Design Equations cccsecccsccssseesseeeeeneeeeeeeesseesesneeenseeeneeees 13 3 4 Design of Serpentine Resistors for Constant Heat Flux eese 14 3 5 Sensor Fabrication 3 nece eene Ee 16 3 5 1 General description sce eet etre dE RO E E eR DERE a esas E EE 16 3 5 2 MicrofabriCatlori o em tet re Re emer E ate per e Ee tee de OR SR eue prev t 17 3 5 3 Packaging and installation in refrigerant piping sese eere 17 3 6 Sensor Calibration PE steers 18 3 6 1 4 wire resistance UE 18 3 62 In situ calibration technique sses renee eie e pee tee bh eie i aee ene 19 3 6 3 Companson Of tege ei Oo Ep e ORPHEUS ee e SERI UA 22 Chapter 4 Experimental Procedures eeeeeeeeeee eee eeeeeeeene nennen nnn 25 iv 44 Experimental Scope a a aao rao aa eaaa ae ee aaee aaa KE Eae Ee 25 4 2 System Start up and Operation csccccsecsseeeeseeeeeeeeeseeeeeseeeeneeeeeeeeeeseaesaseeeeeeeeeseeeseseaeenseeeesenanas 25 4 3 Procedures for Methods of Calculating LMF u 1cssccsseeccsseeeeeseeeseeseeseaneneneeeeeeeeseenseseenenseaenenees 26 4 3 1 Procedures and data collection erepti hot cer
3. gt 1 N k e Tvapor Tmix Te out Tsat p 9 10 o o 5 8 8 w 814 E 3 Sr S RE 4 4 2 2 0 t l 9 0 20 40 60 80 100 120 140 7 Time sec 16 16 14 o N Temperature C Tvapor Te out Tsat Ts 1 Time sec 20 40 60 80 Time sec N o Temperature C o oo e Tvapor 2 Te out Ts 1 Tsat Magnitude 8 Maximum of Te out Te out Ts 1 E requency Hz Time sec Magnitude Maximum of Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures Te out T vapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of T and Teow signals shown in b 54 a b c Run 5 AT sup 12 0 C LMF 1 06 Run 6 AT sup 11 9 C LMF 1 30 Temperature C oo Tvapor Tmix Te out Tsat Tmix Tsat e Tvapor Te out Temperature C oo 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Time sec Temperature C e Tvapor Te out EEN Tsat 160 Time sec Time s
4. Sensor Signals vi List of Figures Page Figure 2 1 Photo of experimental foot 3 Figure 2 2 Flow schematic of experimental facility showing refrigerant lines solid and water lines dashed 4 Figure 2 3 Photo of evaporators test section and instrumentation eene 5 Figure 2 4 Photo of the test section sete teen te toner ee eo Fera bevceseensassecdostsevensepdusebecdessusnveseevesetees 7 Fip re 2 5 Photo o instrumentation c ecc ertet Eed e eile o epu ese eege 9 Figure 2 6 Photo of the test rig highlighting the calorimeter laser and static muer 10 Figure 3 1 MEMS serpentine resistance Sensor 12 Figure 3 2 Dimensions of minimum gap Serpentine Tegistorg nennen nennen eene enne 15 Figure 3 3 Microfabricated sensor with 3 serpentine RT 17 Figure 3 4 MEMS serpentine resistor calibration by 4 wire resistance method 19 Figure 3 5 Sensor voltage vs current at two different vapor temperatures To and Teen 20 Figure 3 6 In situ calibration of MEMS sensors at two temperatures T 9 0 C top and T 14 3 C bottom NEE 23 Figure 3 7 Comparison of MEMS sensor calibration techniques eese ener 24 Figure 5 1 Schematic of evaporators static mixer and test section instrumentation used to determine LMF 30 Figure 5 2 P h diagram of entrained liquid evaporating in a superheated vapor stream 31 Figure 5 3 Comparison of LMF calculated from 1 direct measurement method and 2 e
5. After the laser section the flow encountered the MEMS resistance sensor The sensor was mounted between two glass plates and then installed into a flange in the test section piping A detailed discussion of the sensor design fabrication techniques and theory of operation is left for Chapter 4 After passing the MEMS sensor the flow entered a 5 long x 1 OD glass tube The glass tube allowed for evaporator exit flow visualization A thermocouple beaded type was inserted into the tube through a 4 inch diameter glass tube forming a tee The thermocouple was secured with a compression fitting using Teflon ferrules allowing the position of the thermocouple to be adjusted This was desirable because under certain flow conditions having a high liquid mass fraction splashing liquid would submerse the thermocouple causing drastic drops in the sensed evaporator exit temperature Next was the static mixer It is shown in Figure 2 6 along with the calorimeter The mixer was designed to stir the flow such that all of the entrained liquid was completely evaporated by the surrounding superheated vapor Provided liquid was present conditions at the exit of the mixer would be uniformly superheated at a temperature less than the original superheat temperature Pressure drop in the mixer was marginal and has been measured at no more than 3 kPa during conditions of interest The static mixer consisted of a 6 inch long by 2 4 inch diameter copper tube with a h
6. 0 Te out Tmix Tvapor Sei E e Tvapor S 67 Tmix o Te out E Tsat 44 Sa 0 20 40 60 80 100 120 140 e Tvapor Te out Tei Tsat Temperature C oo 0 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 50 40 g 30 Te out E Ts 1 s D S Frequency Hz Figure 5 6 Evaporator outlet temperature signal comparison at low LMF during TXV control MXV control and simulated maldistribution a during the entire run b at 40Hz sampling rate and c in the frequency domain OV a b oN e Case 1 TXV Control AT 8 3 C LMF 0 14 14 g E 87 e Tvapor E Tmix 9 6 Te out 5 Tsat Al LEE ee at 0 d d d 0 50 100 150 200 Time sec 14 Oo o 2 5 Tvapor e 64 e Te out e Ts 1 44 Tsat 24 0 d d d d d d d d d d d d 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 200 E Peak Ts 1 370 150 4 Te out o 3 Ts 1 E 100 4 D i Frequency Hz Temperature C Magnitude Temperature C 14 Case 2 MXV Control AT sup 7 7 C LMF 0 25 12 4 o Te out e Tvapor Tmi Tsat 14 Time sec 200 Tvapor gt Te out Ts 1 Tsat 200 Time
7. 150 Current i uA Figure 3 6 In situ calibration of MEMS sensors at two temperatures T 9 0 C top and T 14 3 C bottom 23 Tin situ C 0 5 10 15 20 T wire C Figure 3 7 Comparison of MEMS sensor calibration techniques 24 o Tsi o Ts2 Ts3 T4 wire Tin situ 25 30 35 Chapter 4 Experimental Procedures 4 1 Experimental Scope This chapter will familiarize the reader with the procedures for each experiment and the methods used to collect and process the data The following list describes the objectives of each experiment 1 Methods of Calculating LMF Develop a consistent method for quantifying the time averaged liquid mass fraction LMF at the exit of a plate evaporator 2 Correlating Instrument Signals to LMF Compare the signals of a thermocouple MEMS thin film resistance sensor and scattered laser light during the presence of small quantities of entrained liquid droplets 3 Characterizing Exit Flows of Plate Evaporators Measure the exit conditions of a 3 ton plate evaporator at various superheats to demonstrate when and how liquid mass fraction LMF is manifested at the evaporator exit 4 Comparing Thermocouples and MEMS Sensors Investigate using time domain and frequency domain analysis techniques the feasibility of using either a thermocouple or the MEMS sensor to detect and control LMF 5 MEMS Sensor Size vs Sensitivity to LMF
8. Investigate how the surface area of a MEMS thin film resistance sensor affects its ability to detect LMF 4 2 System Start up and Operation The reader is directed to Appendix B for the daily start up procedures of system These procedures were followed until a stable operating point was reached close to the test conditions Operating the refrigeration loop was sometimes difficult especially when using the thermostatic expansion valve because small perturbations of flowrates or temperatures would take up to 30 minutes to return to steady conditions However the expansion device TXV or MXV was typically set close to the desired superheat from the previous days operation When starting the system each day the inlet and outlet water temperatures were set first This was done by adjusting the hot water makeup flow from the condenser and by adjusting the water flowrate with the water pump bypass Then the desired refrigerant setpoints were achieved by making incremental adjustments to the refrigerant flowrate with the compressor bypass needle valves or by adjusting the TXV superheat setpoint or by adjusting the evaporator load water temperatures and flowrate This was a delicate and time consuming process and often required several hours to reach the desired conditions During start up it was important to monitor all of the flowrates pressures and temperatures on the data acquisition monitor in order to avoid any problems For example the evap
9. any refrigerant flow control scheme that reduces the superheat required for stable operation will improve evaporator performance A good control strategy would include sensors that can detect a small liquid fraction in a superheated vapor stream and can be easily incorporated into a feedback loop with the throttling device A further improvement to DX systems using plate evaporators would be to incorporate these sensors with a multi valve active feedback flow control strategy where refrigerant expansion can be independently controlled Such a strategy could be realized by using MEMS microelectromechanical systems flow control valves as the throttling device for each channel MEMS is an enabling technology that merges the capability to sense actuate and control the macro environment with micro scale systems The technology has emerged from the fabrication processes of semiconductor devices and it is finding applications in the areas of 1 fluid sensing 2 mass data storage 3 optics and imaging 4 miniature analytical instruments 5 biomedical sensors and many other areas MEMS has the potential to provide a small size low power consumption low mass low cost and highly functional solution to the problem of refrigerant flow distribution within multi channel evaporators Of course the same feedback control strategies could be accomplished with normal scale valves and the attendant sensors circuitry and controllers These systems would prove too
10. aol Ts 1 8 2 Frequency Hz Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures Te ous Tyapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of Ts and Te out Signals shown in b 56 a b c 16 Run 9 AT sup 11 9 C LMF 3 06 12 4 10 4 Temperature C oo e Tvapor Tmix Te out Tsat 20 40 60 80 100 120 140 Time sec Temperature C oo Tvapor Te out Ts 1 Tsat 50 Time sec 40 30 Magnitude Te out Ts 1 Frequency Hz Run 10 AT sup 10 2 C LMF 0 104 9 P Tmix 5 8T Te out S Tvapor 6 Q Tsat E o Le E H E 0 t d d d d d 0 20 40 60 80 100 120 140 160 Time sec 18 d o 8 Te e Te out 9 84 e Ts 1 5 Tvapor F 6i Tsat 4 PN 0 4 4 0 2 3 4 5 6 F7 8 9 10 11 12 Time sec 50 40 o Ei 22 Te out Ts 1 D G gt 20 Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures Te outs Tyapor MEMS sensor temperature T and saturation temperature
11. to 3 7 and iterate to determine the final values of the four key parameters g n t and w Table 3 1 summarizes the final design parameters of the three serpentine resistors that were made for this project It should be noted that the values in the table are design values only and should not be taken as exact quantities The actual size of the sensors is determined by the resolution of the microfabrication process so each sensor needs to be calibrated to determine precisely the reference resistance However in future calculations requiring sensor dimensions the values in the table will be used because they are within reasonable accuracy for our purposes Table 3 1 Design parameters for the serpentine RTD s sensor a um n g um w um I um As mm Raesign O q W m 1 975 10 25 75 9975 748 91 0 122 76k 2 483 5 25 76 5 2513 192 22 9 119 74 6k 3 189 2 25 82 403 033 3 67 111 69 5k 3 5 Sensor Fabrication 3 5 1 General description A thin rectangular silicon wafer is the foundation of the MEMS sensor On it are the three serpentine Ni resistors two thick gold current leads and 6 thin gold voltage leads Two of the voltage leads are unused and were added solely to provide redundancy in case a problem arose during fabrication The sensors are suspended in the middle of the flow stream by four narrow silicon bridges that have the full thickness of the wafer The bridges in earlier sensor generations were as thin as 40 um b
12. 5 1 shows a schematic of the main evaporator secondary evaporator test section with the static mixer and associated instrumentation Stream 1 is superheated vapor exiting the main evaporator with a mass flowrate m temperature Te ou and pressure Pe For all test conditions the main evaporator was run with 12 C superheat so it can be assumed with good accuracy that stream 1 is comprised solely of superheated vapor Stream 2 on the other hand is a low quality mixture at pressure P with mass flow rate rin The two streams mix in the static mixer and exit as a single phase superheated vapor at temperature T x It is assumed that heat loss to the environment is negligible and that all liquid droplets are evaporated before exiting the static mixer For the direct measurement method the definition of LMF eq 5 1 must be modified to give the LMF directly from knowledge of the flow rates m and M and stream 2 inlet quality a LMF 96 PES 2 x100 5 2 r 29 MEMS Sensor T glass Ban Tii Stream 1 Stream 2 A Static Mixer Main Evaporator Secondary Evaporator glass tube Expansion Valve Secondary Flowmeter Figure 5 1 Schematic of evaporators static mixer and test section instrumentation used to determine LMF Both mass flow rates are measured by the Micro Motion flowmeters However the quality x must be determined from the subcooled liquid temperature before expansion T
13. Frequency Hz Magnitude Temperature C Temperature C Run 16 AT sup 10 2 C LMF 1 78 e Tvapor Tmix Te out Tsat 14 60 80 100 120 140 Time sec N e Tvapor Te out Ts 1 Tsat 100 Time sec 80 4 o o I o Te out Ts 1 Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures T out Tyapor MEMS sensor temperature T and saturation temperature CT at 40 Hz sampling rate c Frequency spectra of T and Tout Signals shown in b 60 Run 17 AT sup 10 1 C LMF 2 20 o Tvapor 10 Tmix Te out 8 Tsat a Temperature C Time sec 14 10 e Tvapor o e Te out o 5 st Ts 1 8 Tsat b gl E o B 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 50 40 4 2 301 Te out 2 Ts 1 c Frequency Hz 140 Temperature C Magnitude Temperature C 14 Run 18 AT sup 10 3 C LMF 3 07 10 4 e Tvapor e Tmix Te out Test 160 Time sec e Tvapor Te out Ts 1 Tsat 50 Time sec 40 30 Te ou
14. Taass thermocouple was placed in the sheltered position so that the vapor temperature could be measured and the resulting temperature was corrected by 1 1 C to account for the thermocouple position 50 Temperature C Tglass 2 Tmix 50 60 70 80 90 100 Time sec Figure A 1 Calibration of the thermocouple in the glass tube to measure vapor temperature 51 Appendix B Procedures for Refrigeration System Startup Power ON Sequence 1 2 3 4 turn on desktop PC turn on HP Mainframe turn on DC power supply turn on power strip mounted on the wall DAQ Software Instructions 1 2 3 4 run HP Vee from shortcut on the desktop open file C HP VeePrograms JulyDAQ press the RUN arrow to begin data sampling data collection procedures a run Microsoft Excel and open file C MSOffice Templates hpveetemplate b enter number of data points desired in IF THEN ELSE box on the JulyDAQ screen sampling rate is about 1 5 Hz c make sure the To Excel block will write the correct number of data points ie rows and columns to the Excel spreadsheet d click on Collect Data checkbox to begin data collection e DONOT interrupt DAQ software during data collection f savethe Excel worksheet under a new filename where ever you like g repeat b e for new runs High Speed data collection a connect the wave generator to the ext trigger of the HP multimeter b set he wave gen
15. Vs1 the largest MEMS sensor voltage 2 3 4 Temperature of thermocouple in glass tube 3 6 Vs2 the middle MEMS sensor voltage and 7 Vs3 the smallest MEMS sensor voltage Each channel was scanned a total of 500 times at 40 Hz for a total of 12 5 seconds of data collection The HP multimeter was configured to advance on the external trigger connected to a wave generator putting out a square wave TTL signal The wave generator output frequency was dialed to exactly 40 Hz using a Fluke digital multimeter All high speed data collection was performed immediately after the 3 minutes of low speed data collection It can be reasonably assumed that process conditions did not vary significantly during the 60 seconds or so it took to switch to high speed data collection 27 4 5 2 Test envelope The test envelope for characterizing evaporator exit flows is shown in Table 4 2 For all cases the mass flowrate evaporator inlet temperature and subcooled refrigerant temperature were held constant Only the superheat temperature AT sup was varied from high to low superheat in order to describe how conditions at the evaporator outlet become unstable Table 4 2 Test envelope for characterizing evaporator exit flows 40 g s 2 C 0 2 C 12 C 10 C 8 C unstable 15 C 0 5 C adjust for desired evaporator load is 4 6 Procedures for Comparing Thermocouples and MEMS Sensors 4 6 1 Procedures and data collection An impor
16. a silicon wafer 250 um thick The serpentine resistors shown Figure 3 1 are very thin 1000 Angstroms and measure 0 0625 mm 0 25 mm and 1 0 mm in total surface area Since this will be the first generation of MEMS sensors it is advantageous to have three sensors on the same substrate in order to compare the effect of sensor size surface area on sensitivity to liquid droplets A constant DC current in the milli amp range passing through the sensors provides i R self heating so that a droplet will evaporate when it strikes the sensor The evaporation of droplets on the sensor surface causes the sensor temperature and resistance to decrease It is important that the thermal mass of the resistors is extremely small to give a very fast time response to refrigerant droplets impinging on the surface Experiments were conducted to measure the time varying voltage signal of each sensor while exposing the sensors to known liquid fractions in a superheated vapor The sensor output signal can be correlated in both time and frequency domains with thermocouple signals laser light intensity scattered by the entrained liquid droplets and the time averaged liquid mass fraction Figure 3 1 MEMS serpentine resistance sensors 3 2 Theory of Operation At high superheats or evenly distributed evaporator flows the sensor will work much like a hot wire anemometer where the self heating will cause the actual sensor temperature to be elevated above
17. a thermostatic expansion valve or a manual expansion valve are studied For these tests only the main evaporator in Figure 5 1 was used Consequently we could only measure the total refrigerant mass flow rate m which in this case is equal to the mass flow rate of liquid In plus the mass flow rate of superheated vapor m at the evaporator outlet The vapor total energy lost from the superheated vapor during evaporation of the entrained liquid in the static mixer is us D sapor h T h mix 5 7 This is the same as the left hand side of eq 5 3 with m replaced with m The enthalpy at the exit of the vapor static mixer hmix can easily be found from pressure and temperature measurements but the superheated vapor enthalpy h is a little more complicated because the superheated vapor temperature could not be directly measured by a thermocouple when liquid was present at the evaporator outlet In order to measure the vapor temperature the thermocouple was shielded from droplet impacts The experimental technique used to measure the vapor temperature is described in Appendix A The energy lost by the vapor is equal to the energy gained by the liquid during evaporation and mixing and can be written as Sm mh Dig hie ha 5 8 Eq 5 7 and eq 5 8 can be combined to solve for the liquid mass flow rate Mii which can then be substituted into eq 5 1 to solve for LMF Finally this LMF is substituted into eq 5 6 whi
18. bulky expensive and complex for controlling individual evaporator channels and thus would not be adopted by industry The purpose of this investigation was to measure the time varying signals of several sensors at the exit of a plate evaporator as superheat is decreased and correlate those signals with a time averaged liquid mass fraction LMF in the superheated vapor stream By investigating the sensor signals it was possible to understand the nature of the non equilibrium two phase flow exiting the plate evaporator The time averaged LMF of the evaporator exit flow stream was measured by using a passive flow mixer to agitate the two phase refrigerant mixture and evaporate any entrained liquid droplets with heat supplied by the surrounding vapor An energy balance of the inlet and outlet flows through the mixer provides the time averaged LMF The exit flows of three evaporator configurations were measured with a newly developed MEMS thin film resistance sensor with a laser and photodiodes for measuring light scattered by entrained droplets and with an exposed beaded thermocouple The evaporator configurations include 1 a plate evaporator fed by a thermostatic expansion valve 2 a plate evaporator fed by a manual valve and 3 a two evaporator configuration used to simulate maldistribution between parallel channels The MEMS sensor scattered laser light and thermocouple signals were recorded at both slow and fast rates in order to characterize
19. controlled Case 1 will show the performance of a real evaporator TXV system case 2 eliminates the influence of TXV dynamics and case 3 allows for comparison of instrument signals under controlled LMF conditions The graphs in Figure 5 6 Figure 5 7 and Figure 5 8 are representative of the conditions witnessed at the evaporator outlet for the three cases The figures depict typical evaporator behavior for each case as the outlet superheat temperature is reduced For practical operation it is always best to operate at the lowest possible superheat that allows only vapor to exit the evaporator There comes a point where further reductions in superheat for a given evaporator load leads to incomplete evaporation within the channels and entrained liquid appears at the outlet The liquid mass fraction LMF present during each test run therefore is a convenient basis for comparing the three cases Runs with relatively similar LMF for each case are grouped together in the figures Initially in Figure 5 6 the superheat is relatively high and not much liquid is present at the evaporator outlet For the purpose of comparison this is what will be called Jow LMF As superheat is reduced in Figure 5 7 and again in Figure 5 8 the amount of entrained liquid increases These two figures represent medium LMF and high LMF respectively Case 3 in Figure 5 6 represents the case of high superheat and almost no liquid at the exit During this test the
20. function of current i will be as shown in Figure 3 5 The deviation of Vi from linearity is due to the self heating of the sensor as the current is increased In order to approximate the initial slope of V i at i 0 the current source was varied from 2 mA while the voltage drop across each sensor was measured In this range the sensors do not exhibit any significant self heating So for very low current the 19 sensor temperature T will be equal to the vapor temperature T The initial slopes m and m at the two calibration temperatures can be written dV p a at T1 3 10 s lizo dV m di at T 3 11 From Ohm s Law V i R we can write sa Rn 3 12 aa gt eae di Tdi In the limit as i goes to zero di goes to Rs Combining this result with eq 3 8 gives 1 dV R 1 a T T 3 13 di S li 0 V at T V Sensor Current 1 Figure 3 5 Sensor voltage vs current at two different vapor temperatures T and T Then for the two calibration vapor temperatures T and T 2 the initial slopes are m R 1 a T 7 3 14 m R 1 a T T 3 15 20 where T gt T and T gt T 2 asi gt 0 Subtracting m from m m m R R o T T R R a T T 3 16 m m Bolt T 3 17 DL m RO TOT 3 18 To find the reference temperature To we can combine eq 3 14 and 3 15 R gt 3 19 E KE WEEK EE m
21. in the flange caused complete cracking of the silicon wafer Figure 3 3 Microfabricated sensor with 3 serpentine RTD s 17 3 6 Sensor Calibration 3 6 1 4 wire resistance technique The principle of operation of a serpentine resistor is the fact that the resistance to the flow of electricity is a function of the cooling due to the surrounding vapor stream and possibly entrained liquid The resistance as a function of sensor temperature T can be expressed as R R 1 0 T T 3 8 where R is the resistance at the reference temperature T 0 C and a sa is the temperature coefficient of resistivity Since is a constant material property in the range of temperatures of interest to refrigeration eq 3 8 reduces to the following linear relationship for R T R T R OR T 3 9 The above equation provides a basis for calibrating the sensor The reference resistance R is simply the y intercept of R T and ob is the slope The calibration is complete when amp and R have been determined for each of the three serpentine resistors An experiment for calibrating o and R for each serpentine resistor was carefully performed over several days The complete MEMS sensor and a calibrated type T beaded thermocouple were exposed to 3 different temperatures in order to develop the R vs T data The 3 temperatures were achieved by placing the sensor and thermocouple in 1 ambient conditions 2 a 3 cubic ft refrigera
22. m afm T m T m T m T 3 20 ae m T m Toa m m Jr 3 21 1 m T SS m T T 3 22 a m m Or 1 T C 3 23 Oo where m T ml C 3 24 m m Finally the fundamental serpentine resistor equation eq 3 7 can be combined with eq 3 18 and eq 3 23 to solve for the sensor temperature as a function of the measured current and voltage as follows R R 1 a T T 3 25 R 1 a T T 3 26 R oe Pee ea 3 27 T V LN T T Lat m T 3 28 sh Vol m m ool s2 i 1152 25541 21 The sensor temperature can be determined using eq 3 28 directly from the measured current i and voltage V The calibration provides the other four parameters mj m T and T 2 in the equation Figure 3 6 shows the calibration of all three sensors at the two temperatures T 9 0 C and T 14 3 C As expected self heating is negligible and the sensor voltage is proportional to current when the current is low less than 2 mA Linear least squares curve fits are shown for each sensor and the correlation coefficient is greater than 0 99 for all but one sensor indicating a strong linearity in the data 3 6 3 Comparison of results Sections 3 6 1 and 3 6 2 have outlined two independent calibration techniques for the MEMS sensors As a means of comparison the reader is referred to Figure 3 7 In the figure the calibration of each sensor is compared usin
23. of measurement is discussed separately in this section This includes the particular equipment used any calibration procedures performed and an assessment of measurement uncertainty 2 2 1 Pressure measurement Three absolute pressure transducers one differential pressure transmitter and one gage pressure transmitter were used to monitor refrigerant flow conditions The location of these sensors is indicated in Figure 2 6 A Sensotec 0 to 500 psia 0 to 3 450 kPa absolute pressure transducer was located in the subcooled refrigerant line before the expansion valve of the main evaporator The evaporating pressure and the static mixer exit pressure were measured by two Sensotec 0 to 200 psia 0 to 1 380 kPa absolute pressure transducers All three Sensotec transducers had a 0 to 5 VDC output and a reported accuracy of 0 1 of the full scale reading Since these transducers were newly purchased the factory calibration factors were used during data collection The main evaporator differential pressure was measured by a Setra 0 to 2 psid 0 to 13 8 kPa differential pressure transmitter It had a 4 to 20 mA output so a 250 Ohm resistor was used at the backplane of the multiplexer to convert the current output to the standard 0 to 5 VDC A Setra 0 to 250 psig 0 to 1 720 kPa gage pressure transmitter measured the calorimeter exit pressure It had a reported accuracy of 0 13 of the full scale reading at constant temperature Since this transmitter me
24. state 1 in Figure 5 2 and the pressure at the exit of the secondary evaporator P It is assumed that stream 2 undergoes a constant enthalpy expansion from T2sw tO a low quality mixture at the exit pressure Pe The quality x is thereby determined by P and the enthalpy of the subcooled liquid before expansion Figure 5 2 outlines this process on a P h diagram The LMF entering the test section can also be calculated by the energy balance method For this method it is assumed that states 4 and 6 in Figure 5 2 are completely defined by the independent measurements of temperature and pressure Also it is assumed that the two streams are completely mixed before exiting the static mixer with no heat loss to the surroundings State 2 is still at pressure P but this time the quality is assumed to be unknown An energy balance of the adiabatic mixing of superheated vapor and unknown quality gives m h h i 1 x h tn hni hi 5 3 The left hand side of the equation is the energy lost from the superheated vapor stream during the mixing process The right hand side accounts for the energy gained by the low quality mixture of stream 2 during evaporation and mixing The first term is the heat of vaporization required to boil liquid droplets to the saturation temperature and the second term is the energy necessary to heat saturated vapor to the final state at the static mixer exit The mass flowrates M and m are measured and all of the enthalpi
25. the centerline and the outer corners Keeping with the centerline approximation the active sensor length can be written as l nla w w n 1 g n 1 w 3 2 l na n 1 g 3 3 Combining this result with eq 3 1 and simplifying gives the active length in terms of the design variables g n and w l n w g g 3 4 3 4 Design of Serpentine Resistors for Constant Heat Flux For simplicity and to aid in comparing the effect that sensor size has on its ability to detect entrained liquid the three serpentine resistors were designed to have equal surface heat flux The resistance of the serpentine assuming constant temperature throughout the active sensor length is X EC cs R 3 5 where A s wt is the cross sectional area of the serpentine and x is the resistivity of the metal film The bulk value of resistivity for nickel given in the literature X 6 84 uQ cm has been used in this design 1 The heat flux per unit surface area of the sensor is a PF i S R S q A where i is the sensor current and A is the surface area which can be expressed in terms of the design variables g w andn A wl 2 w n w g g 3 7 The concept of the sensor is to be able to evaporate and detect individual liquid droplets Barnhart has employed P DPA laser diagnostics to measure average droplet diameters of 50 microns after the dryout point of horizontal tube evaporators 2 This serves as a good starting point for designi
26. the free stream temperature The extent to which the sensor temperature is higher will depend on the sensor current the temperature coefficient of the sensor amp C the convective heat transfer coefficient between the sensor and the free stream and the free stream temperature With the presence of liquid droplets the story becomes a little more 12 complicated Initially as small amounts of droplets strike the sensor they will be evaporated by the i R sensor heat As more and more liquid coats the sensor the sensor temperature is driven lower until the sensor is completely saturated and can no longer evaporate droplets before the next one strikes When the sensor is completely wetted there is a thin film of boiling liquid covering the surface so the sensor temperature will approach Te the excess temperature above the saturation temperature of the boiling liquid required to drive the boiling process Thus at any given time the sensor temperature can vary from an upper limit above the free stream vapor temperature to a lower limit of Te which is slightly greater than Tsa determined at the evaporator exit pressure What this means from a control viewpoint is that one sensor alone located in the evaporator outlet pipe can not provide an accurate measure of superheat especially when liquid is present due to maldistribution In which case it might be advantageous to use a second sensor in a location where it would be sheltered from droplet imp
27. the main evaporator outlet the glass tube and the static mixer outlet The thermocouple in the glass tube was used to measure the actual vapor temperature Tvapor during the test runs of case 1 and 2 in section 5 2 To do so it was positioned such that the thermocouple bead was in the recessed part of the glass tee and out of direct contact from droplet impacts Placing the thermocouple bead in the relatively stagnant gas volume instead of directly in the vapor flow does introduce some error which was calibrated in a simple experiment In order to quantify this error the system was run until a high degree of superheat AT sup 10 5 C remained stable with no indication of droplets The thermocouple in the glass tee was initially sheltered from the approaching vapor flow and was then placed in the center of the pipe cross section in direct contact with the approaching vapor flow The temperature of the glass tube thermocouple Tass and the static mixer outlet thermocouple Tmix were recorded for 100 seconds while the glass thermocouple position was adjusted and the resulting time traces are shown in Figure A 1 The figure indicates that when the thermocouple is sheltered from the approaching flow T sass will read 1 1 C higher than the true vapor temperature Tu and when the thermocouple is placed at the center of the pipe cross section Tase and Tmix are in agreement to within 0 1 C For all of the test conditions reported in section 5 2 the
28. the unsteady flow exiting the evaporator Results are presented in the time and frequency domain and indicate that as superheat is reduced entrained droplet mists emerge in regular intervals at the evaporator exit Three MEMS thin film resistance sensors were designed fabricated and calibrated against known liquid mass fractions for the purposes of this investigation This report documents the general theory of operation of the heated sensors the methodology for designing thin film serpentine resistors with minimum surface area the fabrication of the sensors on a thin silicon wafer and the calibration of the sensors The three MEMS sensors varied in size but all had the same surface heat flux They were driven with a small constant DC current to provide self heating in order to boil liquid droplets that would impact the sensor A comparison of the three sensor signals in the frequency domain reveals that the sensor size does not influence its ability to detect a small liquid mass fraction in a superheated vapor stream The performance of the MEMS sensor and the beaded thermocouple when exposed to small liquid mass fractions in a superheated vapor stream were compared This was done to assess the feasibility of using each instrument to control refrigerant flow to the evaporator by detecting and controlling LMF instead of controlling superheat as is the case of thermostatic expansion valves A unique comparison of the DC and AC signals as a function
29. 10 4 o8 E o 2 S 64 o Ei E o CA ot Tvapor Te out Ts 1 Tsat 0 d d d d d d d d d d d d 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 200 M 150 4 Peak Ts 1 540 o Ki 5 100 Te out 3 Ts 1 50 4 Frequency Hz Case 3 Simulated Maldistribution AT 8 6 C LMF 0 58 12 2 ZS E S o 2 E e 4 Tvapor Tmix Te out Tsat 2 ESP FN EF Mil NE Nia ed gO Nia ft fe 0 t t t t t t 0 20 40 60 80 100 120 140 Time sec 14 e 2 S o Q E ie Tvapor e Te out 41 Ts 1 Tsat 24 0 d d d d d d d d d d d d 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 200 er Maximum Ts 1 470 150 4 Te out 100 4 SE 2 Maximum of Te out 50 4 0 10 15 20 Frequency Hz Figure 5 8 Evaporator outlet temperature signal comparison at high LMF during TXV control MXV control and simulated maldistribution a during the entire run b at 40Hz sampling rate and c in the frequency domain 5 4 Comparison of thermocouple and MEMS sensor signals 5 4 1 Introduction to experiments The response of a beaded thermocouple and three MEMS sensors were formally compared in the time and frequency domains at various superheats and LMF in order to determine their potential feasibility for controlling plate evaporators These tests were conducted using the secondary evaporator to inject contr
30. 20 100 c Magnitude 80 Time sec Maximum of Te out Te out Ts 1 Frequency Hz Magnitude Temperature C Temperature C Run 24 AT sup 8 4 C LMF 1 47 e Tvapor o Te out Tsa Tmix 0 20 40 60 80 100 120 140 Time sec 14 124 o o Time sec 120 4 Te out Ts 1 Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures CT ou T vapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of T and Tout Signals shown in b 64 Run 25 AT sup 8 4 C LMF 2 03 o B e Tvapor Te out Tsat Tmix g 2 I a 5 Q 5 CZ 0 t t 0 20 40 60 80 100 120 140 Time sec 14 Gen Tvapor 10 4 Te out GO EEN o Tsat 5 8T b gl E o 0 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 100 80 Z 604 Te out 2 Ts 1 c 404 20 4 0 5 10 15 20 Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures Te ous Tyapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spe
31. 4 SCT Since amp is a constant material property this average value will be used for all three resistors in future calculations Bruun reports a typical value of 05 0 0038 C for platinum hot wire elements at room temperature 20 C 5 Our thin film serpentine resistors show a 16 improvement in temperature sensitivity compared to platinum hot wire anemometers 3 6 2 In situ calibration technique An in situ calibration of the sensors was also performed The in situ method has the advantage of accounting for any errors in the sensor temperature measurement that might be introduced by the data acquisition system This method provides a means of determining sensor temperature T directly from voltage and current measurements without relying upon explicit knowledge of o R and T To perform this calibration the sensor was installed in its mounting flange in the test section exactly as it was during normal operation This means using the same wiring to the current source the ammeter and the HP multimeter backplane The sensor was exposed to two different vapor temperatures by running the refrigeration system with the main evaporator at a high superheat ATsup 9 C to ensure no droplets were present Flow through the secondary evaporator was valved off during the calibration procedures For a sensor exposed to two different vapor temperatures T and T assuming the vapor velocity remains constant the sensor voltage V as a
32. 5 12 Effect of MEMS sensor surface area on signal frequency content for run 10 superheat 10 0 C LMF 0 32 46 Chapter 6 Conclusions and Recommendations 6 1 Conclusions The outlet flow of a 15 kW plate evaporator fed by a TXV was studied over a range of superheats from 12 C to 7 C in order to investigate how decreasing superheat affects droplet generation at the evaporator exit It was discovered that entrained refrigerant droplets appeared in the exiting superheated vapor stream at superheat temperatures as high as 9 C The liquid droplets seemed to come in discrete mists or clouds at regular intervals These droplet mists caused fluctuations in the thermocouple and MEMS sensor signals that resembled a sawtooth pattern with a period of about 7 seconds The presence of these droplet mists was also confirmed by measurements of scattered laser light intensity and by visual inspection through the glass tube in the test section Reducing superheat below 8 C caused the thermostatic expansion valve to go unstable otherwise known as hunting At the onset of unstable operation the time averaged LMF of entrained refrigerant droplets was measured to be 0 4 Droplet mist generation still occurred every 7 seconds during TXV hunting but the characteristic sawtooth temperature signal was superimposed over a slower mean temperature fluctuation with a period of nearly 3 minutes We have successfully fabricated and tested a prototype MEMS thin fi
33. F 1 76 amplitude variation is slightly higher There are also several large amplitude spikes on the order of 0 5 Volts These larger spikes are related to large clouds of droplets that were visually observed through the glass tube mounted in the test section Notice that during high LMF the thermocouple temperature drops rapidly after the passage of a droplet cloud but the MEMS sensor is unable to respond to the additional liquid LMF 0 28 LMF 1 76 14 0 14 0 g Te out Ts 1 12 A S 124 Tsat Light 3 5 E 5 ed 0 05 gt 1 2 o 9 g E S 2 5 8 E 8 o1 a o n a a E ES E o 2 a 4 Te ut Ts 1 9 3 Tsat Light 9 15 Qe S o 24 B 2 H 0 i 4 i 4 i i 4 i 4 i 4 t 0 2 0 4 4 4 i 4 i 4 i 4 0 2 0 1 2 3 4 5 6 7 8 9 10 td 12 H T Sige 4 sb 007 38 29 40 44 32 Time sec Time sec 160 160 120 4 120 4 o eo b EU Er 580 Maximum of Te out 5 80 bi K bi Frequency Hz Frequency Hz Figure 5 10 Thermocouple Te out MEMS sensor T and scattered laser light signals in the time domain a and frequency domain b for LMF 0 28 left and LMF 1 76 right 44 5 4 3 Frequency domain analysis Additional information about the performance of the thermocouple and MEMS sensor can be obtained by examining the frequency content of the signals The power spect
34. Tsa at 40 Hz sampling rate c Frequency spectra of T and Tout Signals shown in b 57 a b Run 11 AT sup 10 1 C LMF 0 28 20 14 S e Tvapor 9 6 Tmix 5 Te out Fat Tsat 2 EES T Oa OE aS A aS aaa 0 i i i i t 0 20 40 60 80 100 120 Time sec 14 12 10 9 T 2 sl E o Q E o E 44 o Tvapor e Te out fal Tsat 2 0 4 4 0 1 2 3 4 5 6 7 8 9 10 td 12 Time sec 100 m Maximum Ts 1 142 80 8 60 Te out Maximum of Te out D E 2 40 Frequency Hz 140 Run 12 AT sup 10 1 C LMF 0 45 p H ne E o a E e Tvapor Tmix E 44 Te out Tsat 2 KT I S 0 t t t 0 20 40 60 80 100 120 140 160 Time sec 14 12 o Tvapor Te out 10 a Tal Tsat e o 5 8 E o Q E o e 2 0 0 2 3 4 5 6 T 8 9 10 11 12 Time sec 100 80 604 Te out Ts 1 T D E 404 Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures CT out Tyapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of T and Te ou Signals shown in b 58 a b e Temperature C Magnit
35. University of Illinois at Urbana Champaign ACRE Air Conditioning and Refrigeration Center A National Science Foundation University Cooperative Research Center Exploratory Research on MEMS Technology for Air Conditioning and Heat Pumps T M Leicht P S Hrnjak and M A Shannon ACRC CR 44 January 2002 For additional information Air Conditioning and Refrigeration Center University of Illinois Mechanical amp Industrial Engineering Dept 1206 West Green Street Urbana IL 61801 217 333 3115 The Air Conditioning and Refrigeration Center was founded in 1988 with a grant from the estate of Richard W Kritzer the founder of Peerless of America Inc A State of Illinois Technology Challenge Grant helped build the laboratory facilities The ACRC receives continuing support from the Richard W Kritzer Endowment and the National Science Foundation The following organizations have also become sponsors of the Center Alcan Aluminum Corporation Amana Refrigeration Inc Arcelik A S Brazeway Inc Carrier Corporation Copeland Corporation Dacor Daikin Industries Ltd Delphi Harrison Thermal Systems General Motors Corporation Hill PHOENIX Honeywell Inc Hydro Aluminum Adrian Inc Ingersoll Rand Company Kelon Electrical Holdings Co Ltd Lennox International Inc LG Electronics Inc Modine Manufacturing Co Parker Hannifin Corporation Peerless of America Inc Samsung Electronics Co Ltd Te
36. a and frequency domain b for LMF 0 28 left and LMF 1 76 mehr 43 Figure 5 11 Frequency spectra of the scattered laser light signal at low LMF a and high LMF b low LMF 0 28 high LMF E 1 760 E 44 Vil Figure 5 12 Effect of MEMS sensor surface area on signal frequency content for run 10 superheat 10 0 C TIM S EE Figure A 1 Calibration of the thermocouple in the glass tube to measure vapor temperature viii List of Tables Page Table 3 1 Design parameters for the serpentine RTD s eee eee eeeceeseeeeceseceseceseceaecnaecaeecaeeeseseeeeseeeeeeeereseesees 16 Table 3 2 Summary of MEMS serpentine resistor calibration by 4 wire resistance method 18 Table 3 3 Summary of MEMS serpentine resistance sensor by the in situ method T 9 0 C and T 14 3 EE 22 Table 4 1 Test envelope for methods of calculating LAKE 26 Table 4 2 Test envelope for characterizing evaporator exit flows eese 28 Table 4 3 Test envelope for comparing thermocouples and MEMS sen gorg nene 28 1X Nomenclature Roman Symbols T TI Teout Tinix Tsat Trc Aire Ta Ts AT An V Vi V2 w maximum serpentine resistor size cross sectional area surface area of MEMS sensor MEMS sensor calibration constant gap between serpentines in MEMS sensor MEMS sensor current active length of serpentine resistor initial slope of V 1 initial slope of V2 i number of serpentines in MEMS sensor liquid ma
37. allest sensor By placing three MEMS sensors of varying size on the same substrate a direct comparison of the size effects can be made at identical flow conditions Figure 5 12 shows a comparison of the normalized frequency spectra of the time varying voltage signal from each sensor during conditions of high superheat AT 10 C and low liquid fraction LMF 0 32 These conditions allow a good comparison between the three sensors because they are most sensitive at lower liquid mass fractions Clearly the frequency content of each sensor is almost identical In some ways this is encouraging because it tells designers that an extremely small sensor requiring a minimum of current can detect the dynamics of the evaporator exit flow however it casts some doubt on the sensor s ability to detect individual droplets If the distribution of droplets through the pipe cross section can be assumed to be uniform then the larger sensor should naturally be impacted by more droplets than a smaller sensor The sensors were designed to have the same heat flux per unit area so a higher impact frequency was expected from the largest sensor The data does not show a higher impact frequency associated with increased sensor surface area 0 5 0 4 0 4 0 3 T R 107 03 Q A imm B i EI Q A 25mm 0 3 5 Rs 5 01 Q A 063m m 0 2 4 Magnitude 0 2 4 0 1 Frequency Hz Figure
38. ally decreases to reduce refrigerant flow to the evaporator and increase superheat It is interesting to examine the behavior of the evaporator outlet thermocouple temperature Te ou and the MEMS sensor temperature T at high sampling rates Graphs b show these temperatures sampled at 40 Hz These data were taken immediately after the slow speed data in graph a had been collected At low LMF Figure 5 6 Te out and T appear constant but as LMF increases there appears to be a definite temperature oscillation with a period nearly equal to 7 seconds This is supported by graphs a and b at medium LMF Figure 5 7 and high LMF Figure 5 8 The pattern of the MEMS sensor temperature is similar to a sawtooth curve The sawtooth pattern is a result of liquid impacting the sensor in bursts or clouds Initially a quantity of liquid strikes and rapidly cools the sensor Then the internal heat generation of the sensor causes the temperature to rise slowly until the next burst of liquid appears 7 seconds later This is of course an simplified view of the process Any amount of liquid occurring in between these droplet clouds will disturb the rise of the sensor temperature and this can be seen in graph b of Figure 5 8 Notice how the sawtooth pattern of T is quite jagged which is most likely due to smaller bunches of droplets distributed between the larger clouds that come every 7 seconds It is important to mention that the temperature oscillati
39. and 0 15 mm for the characteristic size of the three 15 sensors Two of the other key sensor dimensions g and t were determined by microfabrication constraints The gap g should be as small as possible to keep the sensor size small Our microfabrication process yielded a 25 um minimum gap Thickness t also needs to be small to reduce the sensor thermal mass A thin layer of nickel 1000 o A thick was placed on the silicon substrate to form the serpentine resistors With g and t fixed by microfabrication constraints and the approximate size a given for each sensor we only need to determine the serpentine width w of the serpentine In order to do this the nominal sensor resistance must be set such that when a small DC current is applied to the sensor the voltage drop is in the range of 1 to 5 VDC A majority of instruments and transmitters have outputs in this range so signal amplification would not be necessary and the signal could be read directly with a multimeter To keep power requirements of the sensor to a minimum the design current should be in the range of 1 to 25 mA Using Ohm s Law with a 25 mA current and 5 VDC voltage drop gives a sensor resistance of 100 Q Assuming this will be the design resistance for the largest resistor a 1 mm then the heat flux per unit area will be q 62 5 kW m when the current is 25 mA Given the target dimensions surface heat flux and design current the designer can apply eq 3 2
40. aporator at high superheat and running the other secondary evaporator with quality at the outlet These tests did not result in discrete droplet mist generation The droplet distribution appeared to be more uniform in time This was reflected in both the temperature histories and frequency spectra of the thermocouple and MEMS sensor signals Picking a representative test run say LMF 0 58 shows that there are significant frequencies in the MEMS sensor signal up to 6 Hz for the case of simulated maldistribution compared to significant frequencies only up to 2 Hz for the multi channel evaporator with TXV control The higher frequencies experienced during simulated maldistribution tests is indicative of more frequent 47 droplet impacts on the sensor The more frequently the sensor is impacted by droplets then the more the droplet mist will appear uniformly distributed in time Some remarks must also be made about the sensor design itself Actually three MEMS thin film resistance sensors of different size were made in the same silicon wafer The sensors ranged from 0 0625 mm to 1 mm in total surface area and were designed to have equal heat flux The best way to compare their performance is by their frequency spectra It has been shown that the frequency content of each sensor is nearly identical indicating that even the smallest sensor could be used to detect maldistribution in plate evaporators This is especially encouraging because it impl
41. aporator and high quality refrigerant from the secondary evaporator closely simulated the unsteady exit conditions of a plate evaporator The unsteady exit conditions of a TXV controlled plate evaporator were first explored in a series of tests using only the main evaporator The main evaporator shown in Figure 2 3 was fed by an ALCO series TCL thermostatic expansion valve that was modified for manual operation The temperature sensing bulb and diaphragm assembly were replaced with a micrometer handle attached directly to the valve stem cage assembly The micrometer handle allowed for precise control of evaporator feeding and eliminated hunting problems associated with conventional TXV s that add an additional layer of complexity to multi pass evaporator systems Operating at a fixed expansion valve position also permitted the investigation of evaporator dynamics independent of TXV dynamics over a wide range of superheat The main evaporator was a SWEP model B15x40 3 ton 10 5 kW capacity parallel plate heat exchanger It consisted of 19 refrigerant passages and 20 water passages operating in a counter flow configuration The plates had chevron style contours to enhance heat transfer Two phase refrigerant entered at the bottom of the evaporator evaporated vertically through the plates and exited as superheated vapor at the top of the evaporator The heat load to the evaporator was supplied by water from the water reservoir Thermocouples located i
42. aporator exit flows The MEMS sensor accurately detected refrigerant LMF as low as 1 5 in superheated evaporator exit flows iii Table of Contents Page e E e E iii WAST OT Oe UE Vii Nrogrl ataia ix Miele x Chapter 1 Introduction seccsesecctiiisdetiaetedectcetidata sa acectdecedensdetedeatdetidaaeadadensdenededeaweuianedenend 1 Chapter 2 Experimental Facility 0 cccccccsssssseeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeseeeeeeeeeeeseeeeeeenees 3 2 1 Parallel plate evaporator test facility 1ceeseeeeeeeeeeee eren ene nnmnnn nnna 3 2 1 T Refrigeration flow loop sc ite ete eee ep EES terne 3 25 2 Water flow loop nt eer peek o a n en re bo te e Here denotat 6 2 1 3 Test sectionz i seu mnm aeui nmm 6 KT E a E E E 8 2 2 Pressur measurement 5 cere em Ae eae ke nmn 8 2 2 2 Temperat re measurement eo p eo DROP RERO UB DOR Ue n UH 9 2 2 3 Refrigerant mass flow measurement seeseseeeeeeeeeneeenee nennen nenne enne rennen enne enne nenene 9 2 2 4 Power measureinent ue eden e eese ta eder dE Erden ct estesa e ris ente a Pepe 9 2 3 Easer Equipment 5 ee 10 2 4 Data Acquisition Hardware and Software esee eeeeeee enne nennt nnne nennt 10 Chapter 3 MEMS Resistance Sensors cessere 12 BAM OVOIVIOW
43. artz W L C M Burton and A J Jacobi Vapor liquid equilibria for R 22 R 134a R 125 and R 32 125 with a polyol ester lubricant measurements and departure from ideality ASHRAE Transactions v 102 367 374 1996 49 Appendix A Calibrating the Thermocouple in the Glass Tube to Measure Vapor Temperature In order to calculate the liquid mass fraction present at the evaporator outlet using the energy balance method described in section 5 1 it is necessary to know the vapor enthalpy state 4 in Figure 5 2 The vapor enthalpy is found using thermodynamic property relations with the evaporator outlet pressure and the vapor temperature In a real system refrigerant side maldistribution results in a non equilibrium mixture of entrained liquid within a superheated vapor making it difficult to determine the vapor temperature with a high degree of certainty A thermocouple inserted directly into the outlet pipe would be impacted by liquid droplets causing the thermocouple to indicate a temperature somewhere between the actual vapor temperature Tvapor and the liquid saturation temperature It was hypothesized that shielding the thermocouple from impacting droplets while still exposing it to the exiting vapor could give a reasonably accurate measure of the superheated vapor temperature Referring to the schematic of the evaporators and test section Figure 5 1 there are three thermocouples downstream of the main evaporator They are located at
44. asured gage pressure it was necessary to record the barometric pressure during each day of data collection The barometric pressure was measured with a Princo Instruments 20 to 32 inHg 67 7 to 108 kPa mercurial barometer It should be noted that the differential pressure transmitter and the gage pressure transmitter were both calibrated in situ against the Sensotec pressure transducers Voltmeters Oscilloscope gt Wave Generator Z Lock in Amplifiers Figure 2 5 Photo of instrumentation 2 2 2 Temperature measurement All of the temperatures throughout the refrigerant and water flow loops were measured with Omega type T ungrounded beaded thermocouple probes The beaded thermocouples were chosen as opposed to the shielded type so that the thermocouple junction would be directly exposed to the flow conditions and a faster response time could be attained This was especially important in the test section where fast unsteady processes occurred The thermocouples were calibrated along with the measuring system over a range of 0 to 30 C in an isothermal bath against NIST traceable thermometers The estimated uncertainty of the thermocouples is 0 1 C 2 2 3 Refrigerant mass flow measurement Two Micro Motion ELITE coriolis effect mass flow sensors were used to measure refrigerant mass flow rates The main evaporator flow was measured by a model CMF025 sensor with a 0 to 40 Ib min 0 to 300 g sec range while
45. ch as a thermostatic expansion valve TXV TXV s operate by limiting the refrigerant flow so that the superheat temperature at the exit is high enough to ensure stable operation In horizontal tube evaporators a superheat temperature of 5 C is usually enough for stability over a large range of evaporator loads but plate evaporators typically require superheats above 8 C for stable operation A high superheat degrades evaporator performance because it results in a large superheated zone in the evaporator where heat transfer coefficients are lower than in the two phase evaporating zone Ideally the evaporator should run with zero superheat for all evaporator loads to maximize the two phase heat transfer area of the evaporator This is impossible with DX systems using plate evaporators because distribution of the two phase refrigerant is complicated and current distribution schemes using orifice tubes are unreliable Multi channel evaporators also suffer from refrigerant side maldistribution caused by uneven loads within the evaporator Flow maldistribution can result in one or several channels being completely filled with two phase refrigerant while neighboring channels are dry at the exit Therefore in plate evaporators a high superheat is maintained to assure dry exit conditions and to prevent large quantities of liquid from carrying over and damaging the compressor Since high exit superheat temperatures degrade evaporator performance and capacity
46. ch corrects for the systematic error in the energy balance method 32 5 2 Correlating instrument signals to LMF Experiments were performed to correlate the output signals of a beaded thermocouple the MEMS resistance sensors and scattered laser light with the presence of liquid mass fraction For these experiments the secondary evaporator was used to inject liquid into the test section while the main evaporator was running at high superheat AT An 11 9 C These tests were conducted to ensure that temperature oscillations of the beaded thermocouple and MEMS sensor were caused by LMF as opposed to some other phenomenon like evaporator pressure fluctuations 14 g Te out Ts 1 o Light Temperature C Scattered Light Photodiode Voltage Time sec Figure 5 4 Evaporator outlet thermocouple temperature T ou MEMS sensor temperature T and scattered laser light photodiode voltage in the presence of small quantities of liquid LMF 0 41 For these tests an exposed beaded thermocouple the largest MEMS sensor R and the scattered laser light signals were sampled at 40 Hz for 12 5 seconds The thermocouple was located in the glass tube see schematic in Figure 5 1 such that the exposed bead was in the center of the pipe cross section In this way the thermocouple could be impacted by entrained refrigerant droplets while measuring T ow the evaporator outlet temperature Figure 5 4 shows the
47. ctra of Ts and Te out Signals shown in b 65
48. cumseh Products Company The Trane Company Valeo Inc Visteon Automotive Systems Wolverine Tube Inc York International Inc For additional information Air Conditioning amp Refrigeration Center Mechanical amp Industrial Engineering Dept University of Illinois 1206 West Green Street Urbana IL 61801 217 333 3115 Abstract This report presents an experimental investigation of refrigerant liquid mass fraction LMF in the exit flows of plate evaporators The objective is to identify a sensor that is capable of measuring small amounts of refrigerant liquid in the superheated vapor at evaporator exits This sensor should have the potential to be combined with an active control scheme that increases the fill factor of the evaporator while simultaneously reducing superheat temperature at the evaporator exit Four methods were used to detect refrigerant droplets in the superheated vapor stream exiting a plate evaporator 1 an energy balance calculation 2 a microfabricated thin film resistance sensor developed specifically for this project 3 an exposed beaded thermocouple and 4 photodiodes that detected laser light scattered by droplets The design fabrication calibration procedures and theory of operation of the MEMS thin film resistance sensor are also presented in this report Experimental results indicate that a MEMS thin film resistance sensor is more sensitive than a beaded thermocouple to LMF of non equilibrium ev
49. ec 16 127 Tvapor Te out Ts Tsat Temperature C oo 140 120 100 80 Magnitude 60 40 20 Frequency Hz 2 0 0 1 2 3 4 5 6 7 8 9 10 1 12 Time sec 140 120 Magnitude Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures Te ous Tyapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of T and Te ou Signals shown in b 55 Run 7 Run 8 AT sup 12 0 C LMF 0 AT sup 11 9 C LMF 2 27 16 16 121 e Tvapor Tmix 12 e Tvapor Tmix D Te out Tsat o Te out Tsat oa 107 9 9 5 a 2 S o o a 2 2 el o e 4 0 f f f H t t t t t t 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Time sec Time sec 16 16 Tvapor 127 ij Te out o e Tvapor 2 10 Tei v 107 Te out T gt S gl Tsat Rus Ts b zs E Tsat o o 5 E 2 124 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec Time sec 160 80 120 60 8 Te out 2 Te out Ts 1 2 W sot
50. elical copper sheet inside The helix provided enough mixing to allow LMF measurements as much as 5 To directly measure the quality of the evaporator exit fluid a calorimeter for measuring refrigerant entrained mass fraction completed the test section It should be noted that the calorimeter was not used in this capacity for the majority of test conditions performed in this study The calorimeter consisted of a 3 450 Watt Chromalox finned tubular heater inserted into 1 3 8 inch OD copper tube The heater was 48 inches long with 4 fins per inch The calorimeter was well insulated and was instrumented with thermocouples and pressure transducers at the entrance and exit Heater power could be conveniently adjusted by a variac and the power output was measured directly with a watt transducer The thermocouples pressure transducers and watt transducer are detailed in the next section The calorimeter served two purposes First when the system was running with relatively little liquid at the evaporator exit the heater was used to ensure that all of the liquid had been evaporated before entering the compressor Second the calorimeter could be used when the evaporator is fully wetted to measure exit quality 2 2 Instrumentation The refrigeration flow loop has been instrumented to measure refrigerant mass flow pressure temperature calorimeter heater power laser light intensity for Mie scattering experiments and MEMS sensor voltages Each type
51. epresent refrigerant piping and dashed lines represent water piping The compressor was a Copeland model ZR61K2 hermetically sealed scroll compressor The refrigerant was R22 Mineral oil circulated through the entire flow loop including the test section and was necessary to lubricate the compressor A common problem in these types of systems is oil accumulation in the evaporator when operating at high superheats This reduces evaporator capacity and could influence refrigerant side maldistribution i Condenser Y I prre Sa Micro Motion I ubcooler H20 Supply flowmeter rekt 4 Receiver i I 1 filt DLE pct 62e DEH 1 ed Y i T L p lt I sight glass 1 to h p Y cutoff metering P valve Micro Motion i i flowmeter Expansion X XD f Compressor Device i bypass V I to l p EE cutoff i I Ir Get EE i Se Secondary E it I Evaporator i OE oe a Se E d Ets Primary UN pag Ge X Be eect E coe pein A Evaporator 4 T i X bog l i pe D i H20 Res 4 jr i e Sr E I i i Dannnnnnnnnnn e i l 1 l e lass tube l co ake d Calorimeter Mixer 9 grain Q i HeNe Laser Water Pump MEMS Droplet Sensor Figure 2 2 Flow schematic of experimental facility showing refrigerant lines solid and water lines dashed Immediately on the discharge side of the compressor was a SWEP model B15x60 parallel plate condenser Condensers like this one are often used in unitar
52. erator to a square wave of magnitude 2 5V and adjust to the desired sampling frequency c open file C HPVee Programs FastDAQ d press start button on the left hand side of the window to collect data the multimeter block is set to sample 7 channels at 40Hz for 500 data points per channel you may have to adjust the aperture time if you choose another sampling frequency e save the data record under a new name f press the start button on the right hand side of the screen to extract the data this program will build the 7 arrays and also do a lot of post processing g return to JulyDAQ to monitor the system Running Refrigeration Loop refer to schematic 1 Initial Startup a make sure the following valves are OPEN ball valves at inlet amp outlet to test section manual TXV should still be set at last run condition b run DAQ software see above instructions c plug water pump into power strip d OPEN both water line ball valves on the wall e OPEN flowmeter bypass ball valve located next to the flowmeter this is necessary because the solenoid valve sticks at startup so we have to bypass it f turn ON compressor g CLOSE flowmeter bypass once system is stable Operating Procedure a adjust desired flowrate using manual TXV clockwise increases flowrate b adjust water temperature using hot condenser water gate valve very sensitive C use heater to keep suction line from frosting Shutdown a turn OFF compresso
53. eret ee nren trennen net 34 5 3 2 Case 1 TXV control of a plate evaporator eene nenne nennen nennen 34 5 3 3 Case 2 MSN control of a plate evaporator ener nennen nennen 35 5 3 4 Case 3 Controlled LMF with a two evaporator arrangemennt sese 36 5 3 5 Performance of the MEMS Senso scossssserssersesscosner e enne rennen nete ne tree tren nn en teen nennen 37 5 4 Comparison of thermocouple and MEMS sensor signals eres 41 5 4 1 Introduction to xperiments e neon tecti err ETETE ESEE E EES ES EEEE Er k eer iE 41 5 4 2 Time domain analysis iier terere ee eee re ET EE oer ie 42 25 4 3cRErequency domam analysis Le pee enit orte tent get eU REIR eee leslie cenpee te 44 5 4 4 Effect of MEMS sensor surface area on sensitivity to droplets sssssesesseesssreesrsrrerssrrreerrsreeresreerssene 44 Chapter 6 Conclusions and Recommendations esses 46 NREUOUIdl T c M 46 6 2 Recommendattons 47 REFER ENCES wiscasisinsescczncstessudventieeutateistecstenituasavanvivansacneastuas cnnuteaccaacteuvouentaneasteatensies 48 Appendix A Calibrating the Thermocouple in the Glass Tube to Measure Vapor TEMPS E 49 Appendix B Procedures for Refrigeration System Startup Appendix C Summary of Runs 1 25 for Comparing Thermocouple and MEMS
54. es can be determined from pressure and temperature measurements The exit quality of the secondary evaporator can now be solved for directly Once x is evaluated using eq 5 3 the LMF can be calculated using eq 5 2 30 Figure 5 2 P h diagram of entrained liquid evaporating in a superheated vapor stream A comparison of both methods of calculating LMF is shown in Figure 5 3 Both sets of data show a linear relationship of LMF to m the mass flow rate of stream 2 Error bars for each data point indicate the uncertainty in LMF due to instrument errors only The instrument error was calculated using the methods developed by Moffat 6 The true LMF at the inlet to the test section is most accurately represented by the direct measurement method of eq 5 2 This statement is based not just on the fact that the direct measurement method has less instrument error smaller error bars but because the energy balance method is based upon several simplifying assumptions First the energy balance method assumes no interaction between refrigerant and oil The oil present in the suction line measured to be 0 2 to 0 3 will absorb liquid droplets thereby preventing them from evaporating 7 This causes a negative departure of the calculated LMF from the true LMF Second the energy balance method assumes complete evaporation and mixing of the two streams before the static mixer outlet Third it assumes that the walls of the mixer are adiabatic
55. f the sensor The sensors used in this study are the minimum gap type so the gap g is the second necessary parameter in the design The minimum gap is a function of the resolution attainable during the photolithography process Masks made by the University of Illinois Office of Printing Services produced at best a 15 um gap These masks were made on transparencies from postscript files generated from original AutoCAD drawings of the sensors What is important to the designer is the overall size of the sensor This will influence the active sensor length the sensor s ability to detect small liquid droplets and the feasibility of packaging multiple sensors within plate heat exchangers The maximum serpentine RTD size is a n w g g nw n lg 3 1 Another key parameter in the sensor design is the active length The active length is defined as the total centerline length of the serpentine resistor For long RTD s with many serpentine passes n gt 10 or RTD s with a large aspect ratio w gt 100 the centerline approximation gives a reasonably accurate estimation of the true active 13 length However for RTD s with few serpentine passes in the case of our smallest sensor n 2 or small aspect ratios this approximation breaks down because electric current always flows in the path of least resistance In other words a larger percentage of the current will flow along the inner corners of the serpentine rather than along
56. flat brittle and much too delicate to be directly inserted into the refrigeration piping In an effort to protect the sensor from catastrophic fracture during service it has been mounted in a sandwich between two pieces of 14 thick plate glass A thin sheet metal plate having the same rectangular shape as the wafer is placed within the sandwich to help support the silicon bridges exposed to the refrigerant flow First a two part epoxy is spread evenly and extremely thin across one of the pieces of glass using a razor blade The sheet metal is placed on this piece of glass and allowed to dry Then another thin layer of epoxy is spread over the sheet metal and glass The sensor is carefully placed in exact alignment over the sheet metal support plate At the same time a thin layer of epoxy is laid over the second piece of glass and then placed in contact with the sensor The sandwich is clamped in a specially designed fixture and left to harden for 24 hours Once the epoxy has set the sensor is mounted within the test section in a copper pipe flange Two o rings in the flange provide a pressure tight seal against the glass Certainly other less fragile materials besides glass plate could have been used to create the sandwich However glass was selected because it is transparent which allows the entire wafer to be inspected after the epoxy has hardened This proved to be very useful since several sensors were rendered useless after uneven clamping
57. g 10 blades that chopped the beam at a user specified frequency The laser beam was directed through the test section pipe perpendicular to the flow direction A portion of the laser beam was scattered by entrained liquid droplets Some of the scattered light was collected by the photodiode located 3 inches above the flow centerline The unscattered portion of the laser beam was either absorbed by the refrigerant or passed through unaffected to the second photodiode located across from the laser The photodiodes output a voltage signal proportional to the collected light intensity and these signals were routed through coaxial cable to a pair of lock in amplifiers The lock ins can either output the unamplified photodiode signal through their monitor outputs or they can throw out unwanted frequencies and noise by comparing the photodiode signal to a reference signal If a reference signal is used which for our case was the chopper frequency then the lock in will filter out all frequencies in the input signal except that of the reference signal In this way the lock in is an extremely powerful tool for filtering noise from the photodiode signal 2 4 Data Acquisition Hardware and Software The data acquisition hardware included a Gateway Pentium P5 133 MHz personal computer PC connected via standard HP IB interface to a Hewlett Packard HP 1300A B size VXI Mainframe The mainframe housed a HP E1326B 5 V digit multimeter a HP E1345A 16 chan
58. g both the 4 wire resistance method and the in situ method A line representing T4 wire Tin situ is plotted The symbols on the graph indicate temperatures calculated using eq 3 1 with the 4 wire resistance calibration data in Table 3 2 and using eq 3 21 with the in situ calibration data in Table 3 3 The graph indicates a good agreement between the two calibrations in the temperature regions of interest to this study specifically 0 to 15 C This is in part due to the fact that the calibration temperatures used in the in situ method were 9 0 and 14 3 C Extrapolation of the resulting curve fits would naturally lead to higher deviations between the two calibration methods Table 3 3 Summary of MEMS serpentine resistance sensor by the in situ method T 9 0 C and T 14 3 C MEMS Roo Co sensor m A Q m AO Q Q C C di di s li 0 S iz0 Rai 111 3 113 9 0 491 217 9 Rs2 28 6 29 2 0 113 243 6 Rs3 5 18 5 30 0 023 219 8 22 250 200 Ss Vs4 0 1109 i 2 150 2 t R 0 999 gt 1 n 00 F Vs2 0 0140 is 2 g R 0 997 o gt S a t t 1000 500 1500 2000 2500 o L Vi 0 0028 is 100 R 0 885 150 Current i uA 300 250 Vsa 0 1139 is gt 200 P E a 150 o i 100 Vs 2 0 0292 i zm R 1 gt o Ka L 2 4 a gp 1000 500 500 1000 1500 2000 2500 50 Vs 0 0053 i 100 R 1
59. he relative amount of entrained liquid The thermocouple located directly at the evaporator outlet Te ou is also lower than the vapor temperature This thermocouple was suspended in the middle of the outlet pipe cross section and could easily be impacted by any liquid droplets that might be present in the flow stream The fact that Te out is lower than Typo is another indication that liquid droplets were present during this run because the droplets at temperature Tsat strike and cool the thermocouple below the temperature of the surrounding vapor In Figure 5 7 the superheat setpoint is reduced by 35 only a few tenths of a degree but the measured LMF almost doubles and the evaporator outlet temperature T ou begins to fluctuate by more than 2 C This condition can be seen in Figure 5 7 Consider for a moment graph a for this case which shows the outlet temperatures during the entire test run The Te ouw temperature oscillations seem to be superimposed over a longer time scale fluctuation on the order of 3 minutes This slow periodic fluctuation could be a consequence of the TX V responding to the high superheat in the outlet pipe by gradually opening and increasing the refrigerant flow to the evaporator The difference between case 1 and case 2 in Figure 5 8 indicates the temperature sensing bulb strapped to the outside of the outlet pipe can not react fast enough to the liquid rapidly emerging at the outlet and the valve opening eventu
60. heats 12 C a 10 C b and 8 6 C c Another interesting feature of the graphs is that the heated sensor temperature is often lower than the thermocouple At first this seems counterintuitive One would think that a heated resistor would always be hotter than a thermocouple simply because the self heating would promote evaporation of liquid droplets A possible explanation for the lower sensor temperature could be that the sensor surface was covered completely by a thin film of liquid or possibly oil If this were the case then the thin liquid film would boil on the surface of the sensor The sensor temperature would be driven down very close to the saturation temperature of the liquid because only a slight excess temperature is required for boiling In all three graphs the excess sensor temperature rapidly approaches T Tass 2 C In addition the surface area to mass ratio of the sensor is much larger than the thermocouple so a droplet has a greater cooling effect on the MEMS sensor than the thermocouple Therefore the sensor temperature is lower than the thermocouple when sufficient droplets are present 5 4 2 Time domain analysis Other comparisons between the beaded thermocouple and the MEMS sensors can be made from Figure 5 10 This figure shows the thermocouple temperature MEMS sensor temperature for the largest sensor Rj saturation temperature and scattered laser light photodiode voltage in the time domain at low liquid mas
61. ies that a very small device could be manufactured and even ganged together with several other sensors to measure the outlet conditions of each channel in a plate evaporator 6 2 Recommendations Probably the most successful outcome of this project is the production of a prototype MEMS thin film resistance sensor that can detect entrained liquid droplets in a superheated vapor stream Results from this study show that the sensor has the potential to detect the liquid mists entrained in the superheated vapor exiting plate evaporators A logical course of action would be to use the sensor to develop actuation and control strategies for multi valve operation The experimental facility is already equipped with two parallel evaporators and it would not be too difficult to install normal scale expansion valves and actuators that can be linked through control circuitry to the sensor Although the ultimate goal of this project is to match newly developed MEMS sensors with microfabricated valves actuators and control circuitry it would first be useful to validate the concept of multi valve control with normal scale components If multi valve operation on a macro scale can control refrigerant distribution in plate evaporators in a way that reduces superheat which it surely can then the focus should naturally shift towards developing and incorporating microtechnology into refrigeration systems We recognize the fact that the simulated maldistribution tests e
62. il in section 5 5 and 5 6 it is interesting to notice in Figure 5 6 Figure 5 7 and Figure 5 8 that the MEMS sensor is consistently more sensitive to LMF than the exposed beaded thermocouple This is encouraging because with a first prototype sensor we have a working device that is simple yet gives us enough information to measure LMF in a multi channel evaporator 38 6t a Temperature C o Case 1 TXV Control AT 8 9 C LMF 0 09 e Tvapor Tmix Te out Tsat RP DBRPRARBiCD E Be 0 fl 0 50 100 150 200 Time sec 16 b Temperature C Tei Tvapor gt Tvapor Te out Tsat 0 1 2 3 4 5 6 S 8 9 10 11 12 50 Time sec 30 1 c Magnitude Te out Ts 1 Frequency Hz Case 2 MXV Control AT sup 8 9 C LMF 0 08 Temperature C o Tvapor Tmix Te out Tsat D r EO POO 0 0 50 100 150 200 Time sec 16 14 Ts 1 Tvapor Oo o 2 E E 8 Tvapor 2 Te out Ts 1 2 67 Te out Tsat at 24 0 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 50 40 Magnitude Maximum of Tei Te out Ts 1 Maximum of Te out Frequency Hz Case 3 Simulated Maldistribution AT yup 8 6 C LMF
63. ingement in order to measure the vapor temperature and determine superheat 3 3 Minimum Gap Serpentine RTD Design Equations The smaller we can make the sensor while maintaining a high degree of functionality with regard to temperature accuracy and liquid detection capability the more economically viable the technology becomes It is important to remember that ultimately these sensors are to be ganged together so that the outlet state of each refrigerant channel can be measured In plate heat exchangers the refrigerant passes are typically not more than a few millimeters wide This puts a constraint on the allowable physical size of the sensor The RTD sensors designed and built for this study are a thin metal film that serpentines back and forth within a square outline The idea is to design the sensor with a minimum gap between serpentine lengths so that a long resistor can be compacted into the smallest possible surface area Figure 3 2 shows the general shape of the sensors used in this study with the important dimensions labeled symbolically There are six parameters shown in Figure 3 2 that define the sensor geometry and only four of those six are necessary to complete a design we choose g n t and w A further constraint is that the number of serpentine lengths n must be an integer Therefore it is common practice for the designer to select n as one of the four necessary parameters based upon the desired final size and active length o
64. is quite simple The refrigeration system was run so that the main evaporator sustained a stable 12 C superheat This was done to ensure that the evaporator exit contained no liquid The MEMS sensor current is was set at 25 mA for the duration of the experiments Lastly the secondary evaporator was used to induce liquid into the test section in the range of 0 lt LMF lt 4 It will be shown in section 6 2 that liquid mass fractions greater than 4 are well into the unstable region of plate evaporator operation and so they were not investigated 26 4 5 Procedures for Characterizing Evaporator Exit Flows 4 5 1 Procedures and data collection The goal of these experiments was to record the nature of multi channel evaporator exit flows over a range of superheats For these experiments the system was configured for three different cases 1 TXV control of the main evaporator 2 MXV control of the main evaporator and 3 MXV control of the main and secondary evaporators In cases 1 and 2 there was no flow through the secondary evaporator The refrigeration loop was run at normal operating conditions ie high superheat and then the evaporator superheat was decreased in each successive run The reduction of superheat was achieved by maintaining a constant inlet evaporator temperature Tein While at the same time lowering the water inlet temperature T The outlet evaporator temperature T 4 can never be higher than the water inlet tempe
65. lm resistance sensor that can detect entrained liquid droplets in a superheated vapor stream Performance of the sensor was investigated over a range of liquid mass fraction up to 4 and superheat temperatures from 8 C to 12 C Data collected over these conditions indicates that the sensor is best used when liquid fractions are below 1 Greater amounts of liquid tend to saturate the sensor and prevent it from boiling off the excess liquid The ability of the MEMS sensor to detect a small entrained LMF was compared to that of an exposed beaded thermocouple with encouraging results Although the thermocouple can indicate the presence of droplets over a wider range of LMF the heated MEMS sensor has a much greater sensitivity than the thermocouple Experiments using a plate evaporator with TXV control revealed that by the time TXV operation goes unstable entrained liquid mass fractions are typically 0 4 The sensor is most sensitive from 0 to 1 LMF and would therefore be a more attractive choice than a thermocouple for detecting entrained liquid as superheat is reduced The case of maldistribution between two parallel evaporator channels was investigated to see if outlet conditions would be similar to those seen in the 19 channel plate evaporator A two evaporator arrangement was used to simulate having a plate evaporator with only two refrigerant passes Maldistribution was created between the two evaporators by running one evaporator the 15 kW main ev
66. main evaporator was operating with enough superheat AT 8 6 C to ensure that no liquid was present at the outlet and the secondary evaporator flow was shut off As expected the three thermocouples located in the test section Teous Typo and Trix all indicate the same temperature This is because there is no liquid present in the exiting vapor stream The MEMS sensor temperature however is more than 2 C higher than the surrounding vapor This makes sense since there are no droplets present The heated sensor will always be hotter than the surrounding vapor unless colder liquid droplets at the saturation temperature Tsa strike and cool the sensor 5 3 2 Case 1 TXV control of a plate evaporator Case 1 in Figure 5 6 Figure 5 7 and Figure 5 8 shows the exit conditions of the main evaporator fed by a pressure equalized TXV In each figure the superheat is progressively reduced causing the LMF to increase Starting with Figure 5 6 the evaporator superheat is relatively stable at 8 9 C but there is already an indication of liquid at the outlet Notice that there is a measurable difference between the vapor temperature Tyapo and the static mixer outlet temperature Ti Since the pressure drop in the mixer was negligible this temperature difference can only be explained by the evaporation of entrained liquid by the surrounding vapor within the mixer Thus the magnitude of the temperature difference Tyapor Tmix gives an indication of t
67. mmediately at the entrance and exit of the refrigerant and water streams monitored process conditions Care was taken to position the exposed bead of the refrigerant exit thermocouple at the center of the exit pipe cross section Expansion 1 Secondary Y 8 Valve M a 74 Flowmeter Figure 2 3 Photo of evaporators test section and instrumentation The secondary evaporator shown in Figure 2 3 was somewhat unconventional because it was designed to add only small amounts of high quality refrigerant to the test section Flow rates through the secondary evaporator from 0 to 1 5 grams sec provided entrained liquid mass fractions LMF in the test section from 0 to 3 depending on main evaporator exit conditions Since the flow rate was so low and complete evaporation was not desired the secondary evaporator required a very small heat load Sufficient heat could be generated if and when needed by the ambient air such that a hot fluid heat source was not needed The evaporator was simply a 12 inch long by 1 4 inch diameter copper tube that had been brazed shut at one end Then several small holes were drilled in that end such that a uniform spray of high quality refrigerant could be injected into the test section along the streamwise direction A small needle valve with a micrometer handle served as the expansion device for the secondary evaporator Two thermocouples monitored the expansion process One measured the subcooled liquid temperat
68. mployed two evaporators which does not exactly replicate the process of refrigerant distribution at the inlet of a real plate evaporator fed by a TXV A more realistic approach would be to use a single expansion valve and a plate evaporator with only 2 refrigerant channels Maldistribution could be induced in the evaporator by unevenly heating one side with an external tape heater However the tests that were performed with the two evaporator arrangement give some insight into the type of outlet conditions that one could expect from a multi channel evaporator operating with multi valve control 48 REFERENCES CRC Handbook of Chemistry and Physics Cleveland OH CRC Press 58 ed 1977 Barnhart J S and J E Peters An experimental investigation of flow patterns in a horizontal tube evaporator ACRC TR 28 Air Conditioning and Refrigeration Center University of Illinois at Urbana Champaign 1992 Shannon M A T M Leicht P S Hrnjak N R Miller and F A Khan Thin film resistance sensor for measuring liquid mass fraction in super heated refrigerant International Journal of Sensors and Actuators 164 177 2001 Hewlett Packard Company HP E1326B E1411B 5 Digit Multimeter User s Manual 1996 Bruun H H Hot Wire Anemometry Principles and Signal Analysis Oxford New York 1995 Moffat R J Describing the uncertainties in experimental results Experimental Thermal and Fluid Science v 1 3 17 1988 M
69. n in b 62 a b 12 Temperature C Run 21 AT sup 8 6 C LMF 0 71 e Tvapor Tmix Te out Tsat 140 0 0 20 40 60 80 100 120 Time sec 14 e Tvapor Te out Ts 1 Tsat 104 e o 5 8T 5 co o E eo E 2 0 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 120 100 Magnitude 8 Maximum of Ts 1 Te out Ts 1 Frequency Hz Temperature C Magnitude 8 Run 22 AT sup 8 3 C LMF 1 04 Temperature C e Tvapor Tmix Te out Tsat 140 0 d 0 20 40 60 80 100 120 Time sec 14 124 o 120 Time sec 100 4 co o Maximum of Ts 1 Te out Ts 1 Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures T su Tyapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of T and Tout signals shown in b 63 a Temperature C Run 23 AT sup 8 2 C LMF 1 27 Tvapor e Te out Tsat e Tmix P 20 40 60 80 100 120 140 Time sec N o b Temperature C o 200 180 160 140 1
70. nal having components at 40 Hz The MEMS sensor was driven by a square wave current also at 40 Hz with a maximum amplitude of 20 mA supplied by a wave generator It is important to set the driving current frequency much higher than the sensor time constant so that the sensor sees the oscillating current as if it were constant Figure 5 5 shows these results The lock in output of each signal has been normalized their respective maximal value at the highest LMF What this figure indicates quite clearly is the sensitivity of each instrument to the amount of liquid in the test section The scattered laser light is proportional to LMF and is usable over a broad range of LMF The MEMS sensor however is extremely sensitive for liquid mass fractions less than 0 7 and then almost immediately becomes insensitive to additional liquid The results are encouraging because the high sensitivity of the MEMS sensor to LMF below 0 7 indicates that the sensor could potentially be very useful in detecting the onset of unstable superheat such as in cases 1 and 2 in In Figure 5 4 the smaller thermocouple temperature fluctuations indicate that it is much less sensitive to low LMF than the MEMS sensor An estimated initial slope of the thermocouple response to LMF has been added to Figure 5 5 showing less sensitivity at low LMF but an increase in useable range The MEMS sensor has an added degree of flexibility over a thermocouple because simply increa
71. nel digital oscilloscope capable of detecting frequencies to the GHz range The graphs in Figure 5 9 highlight the behavior of the test section temperatures over three different superheats as LMF was incresed These graphs show the superheat temperature AT An Tyapor Tsat the temperature of the thermocouple located in the glass tube T and the largest sensor temperature T all as a function of LMF A nice feature of the graphs is that they clearly indicate the sensitivity of the thermocouple and MEMS sensor to increasing amounts of liquid The bars on the thermocouple and sensor temperatures indicate the range of temperatures measured during the entire 12 5 seconds of data collection time The larger the error bars the more sensitive the instrument is to the presence of liquid droplets In all three graphs the MEMS sensor is most sensitive at LMF s below 1 which is exactly what the lock in data see section 5 2 revealed during the early stages of experimentation The thermocouple however is sensitive at higher LMF around 1 25 42 vapor e Tmix gi Al Te out 12 MM M e a T Tsat C 3 5 Tvapor Tmix vapor e Tmix Ts 1 Te out Ts 1 Te out T Tsat C 3 5 2 5 Figure 5 9 T Tsa vs liquid mass fraction behavior of thermocouples T mix Tom T vapor and MEMS sensor temperature T at three super
72. nel relay multiplexer a HP E1347A 16 channel 10 thermocouple relay multiplexer and a HP1353A 16 channel thermocouple FET multiplexer The multimeter and the three multiplexer boards were arranged in a scanning digital multimeter configuration The HP equipment was purchased because it provided 1 High measurement accuracy down to the micro volt range in particular for unamplified thermocouple voltages 2 High speed temperature measurements up to 100 K switches per second with the FET multiplexer and 3 Convenient measurement of DC voltage RMS AC voltage 2 wire resistance 4 wire resistance and temperature thermistors RTD s thermocouples All data acquisition programming was done using the HP VEE Visual Engineering Environment version 3 21 software program HP VEE is a general purpose high level iconic programming environment similar to National Instruments LabView The software provides features for instrument control data acquisition data processing data analysis and file management 11 Chapter 3 MEMS Resistance Sensors 3 1 Overview This chapter describes the efforts of this study to design manufacture calibrate and test a MEMS resistance sensor to be used in conjunction with micro valves for controlling parallel plate evaporators in refrigerators and heat pumps The MEMS sensor is actually three sensors in one It consists of three separate serpentine nickel Ni resistors that are evaporated on
73. nergy balance methodi usus cep epo OT DERE TO ENSURE RENT ERE EE 31 Figure 5 4 Evaporator outlet thermocouple temperature Tou MEMS sensor temperature T and scattered laser light photodiode voltage in the presence of small quantities of liquid LMF 0 41 6 32 Figure 5 5 Normalized lock in amplifier output of MEMS sensor and scattered laser light showing sensitivity to droplets eu ERR aoe aie da Se EE aes ON Sa EE Tea Dad go as eds ae a 33 Figure 5 6 Evaporator outlet temperature signal comparison at low LMF during TXV control MXV control and simulated maldistribution a during the entire run b at 40Hz sampling rate and c in the frequency Figure 5 7 Evaporator outlet temperature signal comparison at medium LMF during TXV control MSN control and simulated maldistribution a during the entire run b at 40Hz sampling rate and c in the frequency domain Ayes ae eee ROT EIU eiie 39 Figure 5 8 Evaporator outlet temperature signal comparison at high LMF during TXV control MXV control and simulated maldistribution a during the entire run b at 40Hz sampling rate and c in the frequency Figure 5 9 T Tsa vs liquid mass fraction behavior of thermocouples Ty T su Tyapor and MEMS sensor temperature T d at three superheats 12 C a 10 C b and 8 6 C c sse 42 Figure 5 10 Thermocouple Te ou MEMS sensor T and scattered laser light signals in the time domain
74. ng the MEMS sensors The power required to evaporate a 50 um diameter droplet of refrigerant R 22 is 0 017 mW As an upper bound assume the maximum droplet diameter is 500 um an order of magnitude larger The maximum power for evaporation would be 17 mW This assumes that a droplet will fully wet the sensor surface and be fully evaporated before the arrival of another droplet 14 ooo a maximum size of the RTD 8 gap l active resistor length n number of serpentines integer t thickness into the page w width of serpentine Figure 3 2 Dimensions of minimum gap serpentine resistors The allowable size of the sensor is dictated by plate evaporator geometry and the mechanisms of droplet evaporation Plate geometry provides an upper limit to the sensor size of around 1 mm If the sensor were any larger it would be impossible to install within the plate outlet Droplet evaporation provides the lower bound on size It is probably safe to assume that entrained droplets indeed have a characteristic dimension of 50 um A sensor that is smaller than the smallest characteristic dimension of a droplet would be unable to distinguish between small and large droplets because every droplet in the flow stream would be capable of fully wetting the sensor surface Thus the characteristic sensor dimension a of the three sensors was selected to span the range of 50 microns to 1 millimeter Specifically we tried to target 1 0 5
75. o notice between cases 1 2 and 3 is the similarity of the thermocouple time traces sampled at low speed 0 5 Hz shown in graphs a in the figures As LMF increases the magnitude of the evaporator outlet temperature fluctuations Te out continues to increase In Figure 5 7 and Figure 5 8 these fluctuations are about the same amplitude for the TXV control MXV control and controlled LMF tests The time scale of the Te ouw temperature fluctuations also seems to be relatively constant between the three cases There is one noticeable difference between case 3 and the behavior of a real evaporator in cases 1 and 2 When LMF is added via the secondary evaporator the evaporator outlet temperature T 4 is as much as 2 C lower than was measured during the TXV and MXV tests One possible explanation for this might be that there was more entrained liquid present during the case 3 tests but examination of the high speed 40 Hz data in graph b tells us something different Suppose that instead of the entrained liquid coming in discrete mists every 7 seconds the droplets are more evenly distributed If this were true the time trace of the MEMS sensor temperature T would not appear as a regular sawtooth pattern but would appear more random due to smaller droplet mists impinging on the sensor more frequently This is exactly what case 3 in Figure 5 7 and Figure 5 8 indicates The sensor and the exposed thermocouple are cooled more frequen
76. of LMF and superheat is presented for both instruments Results show that the MEMS sensor is more sensitive to LMF than the thermocouple Both instruments exhibit a saturation point beyond which they can no longer detect increases in LMF Chapter 2 Experimental Facility 2 1 Parallel plate evaporator test facility A new refrigeration system was built for the purposes of this study in the Laboratory for Plate Heat Exchangers at the University of Illinois in Urbana Champaign A photo of the test rig appears in Figure 2 1 It was designed to simulate operating conditions typical of water chillers of less than 60 tons 210 kW capacity using plate heat exchangers for evaporation The facility consists of three main parts 1 the refrigerant loop 2 the water loop 3 and the evaporator exit test section All of which will be discussed in this section Later in section 3 2 the facility instrumentation is discussed The data acquisition system is described in section 3 3 Figure 2 1 Photo of experimental facility 2 1 1 Refrigeration flow loop The refrigeration loop contained the four necessary elements of a vapor compression cycle refrigeration system compressor condenser expansion device and evaporator In addition there was a receiving tank for collecting high pressure liquid from the condenser a liquid subcooler and instrumentation for monitoring process conditions The system schematic is shown in Figure 2 2 where solid lines r
77. olled amounts of LMF into the test section The main evaporator was operated in the superheated range with a mass flow rate of 40 g s nominally In total 25 separate runs were conducted over 3 different superheats 12 C 10 C and 8 5 C and LMF from 0 to 3 5 A summary of each run including time histories and frequency spectra of test section temperatures can be found in Appendix B Data collected during each run consisted of 1 Global system parameters These measurements included flow rates pressures and temperatures necessary for energy balance calculations The data was collected at 0 5 Hz for 200 seconds 2 Fastsampling of test section instrumentation The raw scattered light photodiode voltage evaporator pressure MEMS sensor voltages and all test section thermocouples were recorded at 40 Hz for 12 5 seconds A sampling frequency of 40 Hz was selected because it provided the best opportunity to capture all significant frequencies present in the thermocouple MEMS sensor and scattered light signals During the facility shake down process Fourier analysis of the raw output of each instrument thermocouple MEMS sensors and scattered light revealed that any frequencies above 10 Hz did not make a significant contribution to the frequency spectra Therefore the sampling rate was selected as close as possible to the maximum expected frequency while still being above twice the Nyquist frequency The Fourier analysis was performed on a 4 chan
78. ons of both the thermocouple Te out and the sensor T 1 correlate identically to clouds of liquid that were large enough to be seen by the naked eye through the glass tube in the test section 5 3 3 Case 2 MXV control of a plate evaporator Case 2 is similar to case 1 except that this time a manual expansion valve was used instead of the TXV The results of case 2 are presented in Figure 5 6 Figure 5 7 and Figure 5 8 in order of increasing LMF The purpose of the manual expansion valve was to study the outlet conditions of a multi channel evaporator as superheat is reduced without the effects of TXV valve dynamics By using a MXV any phenomena witnessed at the evaporator outlet will only be a consequence of the physical mechanisms of evaporation among the parallel channels For all case 2 test runs the valve position was fixed and the system was allowed to come to a pseudo steady condition What this means is that the amplitude of the temperature oscillations of Te ou and T remained fairly constant and Tyapor Tmix and Tsa remained constant over the entire run The most important comparison between the TXV and MXV runs is that the sawtooth pattern of the MEMS sensor temperature occurs in both cases Again the period of these temperature oscillations is on the order of 7 seconds It can be inferred from the MXV runs that entrained liquid probably resulting from maldistribution between evaporator channels emerges at the outlet
79. orator inlet water temperature could not drop below 0 C because the plate evaporator could have been seriously damaged if the water froze inside Another parameter that required special attention was the compressor suction temperature There was no thermocouple in the loop to monitor this temperature so instead it was checked by hand to make sure the pipe was neither cold nor hot An adequate temperature was around room temperature The compressor suction pipe was fed from the calorimeter and the compressor bypass lines The calorimeter heater power was adjusted so that all liquid droplets have been evaporated and the vapor stream to the compressor was around room temperature Both of the needle valves on the bypass lines were also adjusted to provide a mixed compressor suction stream at room temperature 25 4 3 Procedures for Methods of Calculating LMF 4 3 1 Procedures and data collection Two methods of calculating LMF were tested 1 by an energy balance and 2 by direct flowrate measurements The two methods are compared in section 5 1 to determine which method most accurately measures LMF For this experiment the main and secondary evaporators were operating in parallel as shown in Figure 6 1 The main evaporator was set to a high superheat temperature while the secondary evaporator injected a small amount of saturated liquid into the test section There were a total of 14 test runs for this experiment 4 3 2 Test envelope For this e
80. piping in organized clouds every 7 seconds This is a significant finding because it indicates that mists of entrained liquid droplets will appear at regular and predictable intervals as a result of maldistribution within a multi 36 channel evaporator Furthermore it seems that this regular pattern remains constant even with increasing amounts of maldistribution or measured liquid mass fraction 5 3 4 Case 3 Controlled LMF with a two evaporator arrangement Case 3 in Figure 5 6 Figure 5 7 and Figure 5 8 show what can occur at the evaporator outlet when there is maldistribution between two parallel channels The two evaporator arrangement shown schematically in Figure 5 1 was adjusted such that the 3 ton main evaporator was running with high superheat at the outlet to ensure no droplets and the smaller secondary evaporator added a low quality mixture into the test section Manual expansion valves that were left at a fixed position during the test runs fed both evaporators This simulated a two channel evaporator in which maldistribution can be manually adjusted between the channels It should be noted that the test runs shown for case 3 in the figures were selected such that the average LMF and superheat during the run closely matched those of case 1 and 2 for comparison Specifically the low LMF data is from run 19 the medium LMF data is from run 11 and the high LMF data is from run 20 in Appendix B The first thing t
81. r b CLOSE test section ball valves c unplug water pump d CLOSE water line ball valves on the wall 52 a b c Appendix C Summary of Runs 1 25 for Comparing Thermocouple and MEMS Sensor Signals Run 1 AT sup 12 0 C LMF 0 12 4 2 10 o Tmix 8 8 Te out 2 e Tvapor E 6 Tsat o na 4L at 0 t t t 0 20 40 60 80 100 120 140 Time sec Temperature C E Ts 1 e Tvapor Te out Tsat 50 Time sec 40 30 Magnitude Te out Ts 1 Frequency Hz Temperature C Run 2 AT sup 11 9 C LMF 0 28 N d 2 104 g a S 8 Tvapor o e Tmix 5 6 Te out e Tsat 4 2d 0 t t t 0 20 40 60 80 100 120 140 Time sec 16 4T e Tvapor s Te out e Ts 1 Tsat Time sec 160 N o Te out Maximum of Te out Magnitude S B o Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures T su Tyapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of T and Te out Signals shown in b 53 a b c o Run3 AT sup 12 1 C LMF 0 39 Run 4 AT sup 12 0 C LMF 0 73
82. ra of both signals is shown in graphs b Figure 5 10 for both low and high LMF Fast Fourier Transform analysis was performed on the discrete temperature signals using the HP VEE software by first subtracting the mean of each signal from the original data set This was done to remove the DC component of the each signal thereby leaving only the oscillating portion of the temperature caused by both impingement of liquid droplets and saturation temperature fluctuation The first large spike at 0 16 Hz in both figures is due to the saturation temperature fluctuation All of the other frequencies can be attributed to the presence of liquid droplets In both figures the significant frequencies die out around 5 Hz Initially it was thought that the MEMS sensor would be able to detect individual droplet impacts at frequencies above 100 Hz This is clearly not the case especially at the higher LMF s when droplet clouds are observed visually in the glass tube 1 5 1 5 peaks at 0 25 Hz and 1 Hz 1 0 1 0 o o E z peak at 0 8 Hz t g ye g 2 0 5 0 5 0 0 4 0 0 WN Las ea 0 5 10 15 20 0 5 10 15 20 Frequency Hz Frequency Hz a b Figure 5 11 Frequency spectra of the scattered laser light signal at low LMF 2 and high LMF b low LMF 0 28 high LMF 1 76 Even the scattered laser light can not detect the passing of individual droplets The power spectra of the scattered laser light pho
83. rature because the evaporator is a counterflow heat exchanger Therefore it is possible to control superheat by simultaneously raising or lowering the water inlet temperature Twin and by adjusting the superheat setpoint of the TXV Data collection consisted of both low speed sampling of all instrumentation and high speed sampling of the test section instrumentation only The low speed data was used to measure system behavior over long periods of time and to calculate energy balances The high speed data was used to investigate the transient non equilibrium two phase flow conditions at the evaporator exit A total of 150 data sets of the 20 instrument channels were sampled and recorded at low speeds for a total of 200 seconds The HPVee data acquisition software determines the low speed data sampling rate which was a function of how many channels were multiplexed and how much data was updated on the computer screen HPVee could have been programmed to sample at a user defined rate however that was not necessary for the slow data collection The computer run time was recorded before each scan so that the exact time between multiplexer scans was known The majority of low speed data collection was at an average of 0 6 Hz High speed sampling of the test section instrumentation was taken at a user set 40 Hz Seven channels were sampled at this speed to measure Evaporator exit pressure Static mixer exit pressure Evaporator exit temperature
84. rom 12 C to 8 C Anything lower than 8 C superheat caused the main evaporator to spill liquid into the test section which would have biased our measurements of liquid mass fraction Table 4 3 Test envelope for comparing thermocouples and MEMS sensors 40 g s 2 C 0 2 C 12 C 10 C 8 C 0 to 2 adjust for desired evaporator load is 25 and 40 mA 28 Chapter 5 Experimental Results 5 1 Methods of Calculating LMF Maldistribution in plate evaporators is evident when liquid droplets are entrained in the superheated vapor stream exiting the evaporator One objective of this work is to develop a consistent method for measuring the liquid mass fraction LMF of entrained liquid Two methods for measuring LMF are compared they are 1 energy balance method and 2 direct measurement method The energy balance method is based on the reduction of vapor temperature at the static mixer due to the evaporation of droplets The direct measurement method calculates LMF by independently measuring the mass flowrate of superheated vapor and entrained liquid droplets LMF is defined here as the ratio of the mass flow rate of entrained liquid and the total mass flow rate LMF should not be confused with quality which describes vapor that is saturated Assuming liquid droplets and vapor have the same velocity at a given pipe cross section LMF can be formulated as VUEN LMF 96 x 100 5 1 Myiquia M vapor Figure
85. s fraction LMF 0 28 and high liquid mass fraction LMF 1 76 At low LMF the MEMS sensor is most sensitive to the presence of droplets This is evident by the large amplitude temperature fluctuations of the sensor compared to 43 relatively little temperature fluctuation by the thermocouple When large amounts of liquid are present the MEMS sensor is unable to detect the increasing amount of liquid but the thermocouple now is becoming more sensitive to liquid At the higher LMF the temperature fluctuations of the thermocouple increase and those of the MEMS sensor decrease It is important to notice that there are both slow and fast frequency components in the thermocouple and the MEMS sensor temperature traces The slow frequency component in both temperatures is exactly in phase with the saturation temperature displayed on the bottom of the figures Saturation temperature varies with evaporation pressure which oscillates due to the nature of the two phase evaporating flow in the evaporator Superimposed on top of the slow fluctuation is the high frequency content It is the high frequency oscillation that is a result of droplets impinging on the surface of the sensor and thermocouple The trace of the scattered laser light photodiode voltage in both figures gives a relative indication of the amount of liquid present during the run When LMF is only 0 28 the scattered light signal amplitude fluctuation is about 0 01 Volts but when LM
86. sec 150 4 100 4 50 4 Peak Ts 1 475 Te out Ts 1 Frequency Hz Case 3 Simulated Maldistribution AT yup 10 1 C LMF 0 28 14 12 _ 10 2 9 8 a Tvapor g Tmix 6T Te out o Tsat ES 4 DE DNE BS EE 0 i 0 20 40 60 80 100 120 140 Time sec 14 Temperature C Tvapor Te out Tsat Tei 200 Time sec 150 Magnitude S Te out Ts 1 Maximum of Te out Frequency Hz Figure 5 7 Evaporator outlet temperature signal comparison at medium LMF during TXV control MXV control and simulated maldistribution a during the entire run b at 40Hz sampling rate and c in the frequency domain Iv a b e Case 1 TXV Control AT 8 0 C LMF 0 38 4 Tsat Temperature C 0 50 100 150 200 Time sec e Tvapor Te out Tsat Temperature C Ts 1 Time sec 200 150 Peak Ts 1 476 Te out Ts 1 Magnitude 8 50 T Frequency Hz Case 2 MXV Control AT 7 22C LMF 0 32 12 2 g S E Tsat gt Te out O 4 e Tvapor Tmix e 24 0 d d d 0 50 100 150 200 Time sec 12
87. sing the driving current can extend its useable range 1 2 Initial slope indicates high sensitivity 1 0 T A A i a Ze NM i a A A 5 Theoretical nitial slope Ar E a a E P 5 ost f thermacouple oe 9 i 4 s Scattered laser light sensitive over T b d a broad range of LMF i A E 9 06 IA 3 H D iN i o H d E 047 fa P D o H KZ z i 2n 4 MEMS sensor eke scattered laser light 021 i m AE a a 0 0 E K i 0 0 1 0 2 0 3 0 4 0 Liquid Mass Fraction LMF 96 Figure 5 5 Normalized lock in amplifier output of MEMS sensor and scattered laser light showing sensitivity to droplets 5 3 Characterizing Evaporator Exit Flows 5 3 1 Introduction to experiments One goal of this study was to characterize the exit conditions of a plate evaporator over a range of superheat temperatures Normal superheat temperatures are high enough to ensure that no liquid leaves the 34 evaporator Reducing exit superheat too much can lead to maldistribution of the refrigerant amongst parallel channels in the evaporator causing entrained liquid droplets to be present in the exit flow This section describes the results of a series of experiments that were conducted using a 3 ton capacity PHE as an evaporator Three configurations were tested 1 PHE fed by a thermostatic expansion valve TXV 2 PHE fed by a manual expansion valve MXV and 3 a two evaporator arrangement in which LMF is manually
88. ss fraction MEMS sensor surface heat flux design resistance of MEMS sensor reference resistance at temperature T 0 C resistance of MEMS sensor thickness of serpentine in MEMS sensor Temperature sensor temperature using in situ calibration technique reference temperature MEMS sensor temperature MEMS sensor temperature large MEMS sensor temperature medium MEMS sensor temperature small refrigerant temperature at main evaporator outlet temperature at the static mixer outlet saturation temperature temperature measured by thermocouple in the glass tube sensor temperature using 4 wire calibration technique vapor temperature during sensor calibration vapor temperature during sensor calibration superheat temperature voltage drop across MEMS sensor sensor voltage at T sensor voltage at T width of serpentine in MEMS sensor Greek Symbols a X temperature coefficient of resistivity C resistivity LO cm Chapter 1 Introduction Large heat pumps refrigeration and air conditioning systems often use plate heat exchangers PHE as evaporators Plate evaporators consist of multiple refrigerant channels or plates in which the refrigerant evaporates vertically up each channel Typically the channels are connected in parallel but they can also be circuited in a serpentine fashion to achieve a desired performance level In direct expansion DX systems refrigerant is expanded through a single throttling device su
89. t Ts 1 Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures CT ou Tvapor MEMS sensor temperature T and saturation temperature CT at 40 Hz sampling rate c Frequency spectra of T and Tout Signals shown in b 61 Run 19 AT sup 8 6 C LMF 0 a Temperature C e Tvapor Tmix Te out Tsat 0 0 20 40 60 80 100 120 Time sec 16 e H P 2 T b 5 8f FERAS Tvapor co Te out 5561 Ts 1 m Tsat 44 2 0 0 1 2 3 4 5 6 7 8 9 10 11 12 Time sec 150 130 110 90 8 Te out Ts 1 amp 50 30 Frequency Hz 140 Temperature C Magnitude a o Temperature C Run 20 AT sup 8 6 C LMF 0 58 e Tvapor Tmix Te out Tsat 14 140 Time sec o o 150 130 o o Time sec Maximum Ts 1 470 Maximum of Te out Te out 10 6 Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures T su Tyapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of T and Tout Signals show
90. tant task of this project is to characterize the output signals of a beaded thermocouple and the MEMS resistance sensors over a range of superheats and LMF To accomplish this the main evaporator was outfitted with the manual expansion valve described in section 2 2 The manual valve allows the main evaporator to run at a fairly constant superheat without the transients induced by a thermostatic expansion valve The main evaporator exit superheat was kept high 12 C to ensure no entrained liquid was present A small amount of liquid was added to the test section inlet via the secondary evaporator and the LMF was calculated using the procedure outlined in section 5 2 Data collection for these experiments consisted of both low speed and high speed data sampling The same 20 channels measured in section 4 3 were again sampled at low frequency The same holds true for the 7 channels sampled at high frequency 40 Hz for 12 5 seconds During the high speed data sampling the MEMS sensor current was set at 25 and 40 mA so that different heating values could be compared 4 6 2 Test envelope The test envelope for these experiments is outlined in Table 4 3 Notice that the conditions are very similar to those in Table 4 2 and for good reason Itis important to investigate the behavior of the thermocouple and MEMS sensors over know liquid fractions and superheat conditions would be expected in a real system with TXV control The superheat was varied f
91. te reet re recreate chpnssdveveyeoepeneeee 26 4 32 TestenVelOpeiz iere He thee Dee tret EO E cre ete ioe Ucet tp Sl e Rte ts 26 4 4 Procedures for Correlating Instrument Signals to LMF cessere 26 4 4 1 Procedures and data collection sssssssssssesseeeeee eee enne enne ener nennen erri en nnns enne 26 A AND Test Envelopes ices eee oe bono IP emnt Dp p RO bI ders 26 4 5 Procedures for Characterizing Evaporator Exit Flows cernens 27 4 5 1 Procedures and data collection eet teen trente etse NEESS 27 4 52 Test enyelope s sesion ue eva ed umen usan Uu edente 28 4 6 Procedures for Comparing Thermocouples and MEMS Sensors eese 28 4 6 1 Procedures and data collection ete ettet rrt Leere d vine eese toes 28 4 62 Test envelopeiss ra I tete ocio teat pete tege ERR ies 28 Chapter 5 Experimental Results ceeeeeeeeeeeeeeeeeeeeeeee nennen nnn 29 5 1 Methods of Calculating LMF eeeeeeeeeeeeeeseseee seen eeenn nenne nennt nnn trans asina nennt anna as 29 5 2 Correlating instrument signals to LMF eeseeeeeeeeeeeees esee ennt nenne ennt nnn nn nnns 32 5 3 Characterizing Evaporator Exit Flows eeseeeeeeeeeeeeees seen entren nnne nain nnnm nnn tnnt nnns 34 5 3 1 Introduction to experiments sess nnne nennen retener enne tenete t
92. the secondary evaporator flow was measured by a model CMFO10 sensor with a 0 to 3 Ib min 0 to 23 g sec range Both flowmeters shown in Figure 2 3 were used in conjunction with their own model RTF9793 Field Mount Transmitter which outputs both flowrate and fluid density The transmitters can be seen in Figure 2 6 The flowmeters were factory calibrated and had a reported accuracy of 0 10 of F S zero stability flowrate x100 96 of the measured rate The zero stability of the CMFO025 sensor was 0 001 Ib min and for the CMFO10 it was 0 00015 Ib min 2 2 4 Power measurement The calorimeter heater power was measured with an Ohio Semitronics 0 to 4 KW 0 to 13 600 Btu hr watt transducer shown in Figure 2 6 It was calibrated in situ against two Fluke 4 5 digit multimeters one measuring voltage across the heater and the other wired as an ammeter measuring heater current The watt transducer has a reported accuracy of 0 04 of the full scale reading Transmitters for Flowmeters Figure 2 6 Photo of the test rig highlighting the calorimeter laser and static mixer 2 3 Laser Equipment Figure 2 5 and Figure 2 6 show the laser and related instrumentation used in the test section to identify the presence of entrained liquid droplets The hardware included a 2 0 mW Helium Neon laser and power supply a light chopper two photodiodes and two ThorLabs lock in amplifiers The light chopper was a 4 inch diameter windmill shaped disk havin
93. time traces of these three signals for very small amounts of liquid LMF 0 41 with high superheat AT 11 9 C The sensor was driven with a 25 mA DC current The scattered laser light signal is important because it directly correlates with the quantity of droplets present in the test section The photodiode voltage signal decreases as more droplets are present to scatter the beam Notice that the signal amplitudes of both the thermocouple and the MEMS sensor relate to the scattered light and thus droplets There is a slight time lag between the two temperatures and the laser signal of about 1 second This is because the laser gives an instantaneous measure of the liquid fraction whereas the thermocouple and sensor have thermal capacitance which delays their response to liquid Additionally the laser was located 6 inches upstream of the MEMS sensor and 14 inches upstream of the thermocouple so a further time delay can be expected for droplets to travel the distance between the laser and the other instruments 33 The MEMS sensor and scattered light signals were also measured with lock in amplifiers for liquid fractions up to 4 The lock in output is based upon the autocorrelation of the input signal to a reference signal For the scattered light signal the laser beam chopper frequency was used as the reference signal The chopper frequency was tuned to 40 Hz so the lock in output was proportional only to that portion of the photodiode sig
94. tly by droplet mists and are prevented from warming back up to the superheated vapor temperature Also notice that the Te ouw and T temperature fluctuations on a slower scale directly correlate with the saturation temperature fluctuations It seems that there are higher frequency temperature oscillations due to frequent cooling by droplet mists superimposed over a slower average temperature change related to the saturation temperature The frequency spectra in graphs c tell this story perhaps a little more clearly Consider the frequency spectrum for the three cases in Figure 5 8 A large spike at 0 16 Hz in all three cases dominates the thermocouple and MEMS sensor signals In other words the signals have a characteristic period of 6 25 seconds Recall that the temperature time data indicated that temperature oscillations had a period of about 7 seconds and that these oscillations coincided with droplet clouds witnessed in the glass tube The TXV control and MXV control cases only have significant frequencies below 2 Hz but the thermocouple and MEMS sensor signals contain higher frequencies during the controlled LMF tests These frequencies above 2 Hz indicate that indeed the 37 distribution of droplets was more uniform in case 3 than in case 1 or 2 which are dominated by discrete droplet clouds at 7 second intervals 5 3 5 Performance of the MEMS sensor Although the performance of the sensor will be discussed in greater deta
95. todiode voltage are shown in Figure 5 11 for LMF 0 28 and 1 76 At higher LMF there is clearly a stronger signal indicating more scattering due to liquid droplets There are more sophisticated techniques for measuring droplet size velocity and distribution such as phase Doppler particle analysis P DPA and laser Doppler velocimetry LDV These laser diagnostics techniques have seen widespread use in combustion research and aerosol sprays They require very expensive laser equipment that is extremely difficult to calibrate which is why those techniques were not used in this investigation 5 4 4 Effect of MEMS sensor surface area on sensitivity to droplets Entrained liquid droplets smaller than 50 um in diameter traveling at 10 m s have been measured in horizontal tube evaporators 2 It was suspected that the size of the sensor would directly affect its ability to detect such small droplets in the flow stream For example if a droplet is large enough to completely coat the small sensor upon impact it is sure to drop the sensor temperature very close to the liquid temperature However the same droplet impacting the large sensor would only cover about 746 of the total surface area and could not bring possibly lower the sensor temperature completely to the liquid temperature If the sensors are indeed sensitive enough to be 45 affected individual droplets then the larger sensor should prove less sensitive at lower LMF s than the sm
96. tor and 3 the freezer compartment of the refrigerator The sensor was exposed to each temperature for a period of 24 hours to ensure equilibrium conditions were reached Sensor resistances were measured with the HP 5 1 2 digital multimeter using a 4 wire technique To minimize any self heating of the resistors during the 4 wire measurements it is necessary to keep the source current to a minimum The effect of self heating can be minimized by selecting a higher resistance range on the multimeter since less current is applied However higher ranges yield lower resolution It was determined that the default setting of the multimeter 16384 Q range and 61 HA was adequate to prevent self heating and still provide 15mQ resolution see p 87 in 4 The results of the calibration are presented in Table 3 2 Table 3 2 Summary of MEMS serpentine resistor calibration by 4 wire resistance method MEMS nominal surface reference temperature sensor area resistance at coefficient of A mm2 To 0 C resistivity Ro Q a C 1 Rs1 0045 Rs2 0044 Rs3 0043 average 0044 18 120 CH R1 0 482 T 107 03 S R 0 999 amp SI Sen n Zen Zen 40 R2 n 427 51 U f ai R3 0 0214 T 5 01 H A n 0 Temperature C Figure 3 4 MEMS serpentine resistor calibration by 4 wire resistance method The average temperature coefficient of resistivity for the three resistors was found to be 0 004
97. ude Temperature C Run 13 AT sup 10 0 C LMF 0 82 e Tvapor Tmix Te out Tsat P Time sec N o e Tvapor Te out Ts 1 Tsat 120 Time sec 110 4 100 Qo o oo in i o Te out Ts 1 ms Maximum of Ts 1 Frequency Hz 14 Run 14 AT 10 1 C LMF 1 07 o Temperature C o 1 d e Tvapor Tmix Te out Tsat 14 140 Time sec o N Temperature C o e Tvapor Te out Ts 1 Test Time sec Magnitude Frequency Hz a Time trace of evaporator outlet and test section temperatures b Thermocouple temperatures T ou T vapor MEMS sensor temperature T and saturation temperature Tsa at 40 Hz sampling rate c Frequency spectra of T and Tout Signals shown in b 59 a b e Run 15 AT sup 10 2 C LMF 1 36 Temperature C e Tvapor Tmix Te out Tsat P Time sec 140 o N Temperature C o e Tvapor Te out Ts 1 Tsat 100 Time sec 80 o o Magnitude I o Te out Ts 1
98. ure just upstream of the valve and the other measured the saturation temperature within the evaporator tube 2 1 2 Water flow loop At the heart of the water loop shown in dashed lines in Figure 2 2 was the 15 Liter 4 gallon water mixing tank A Teel 1 2 hp centrifugal water pump pulled water from the bottom of the mixing tank and pumped it through the evaporator Chilled water from the evaporator then recirculated back to the mixing tank At the same time some of the hot condensing water was diverted into the mixing tank The hot condensing water and chilled water from the evaporator mixed in the tank to provide the appropriate inlet water temperature for the evaporator Water flow rate was controlled via a bypass line and throttling valve connected between the pump discharge and the mixing tank A drain hose located 20 cm above the bottom of the tank kept the water level in the mixing tank constant The water flow loop was flexible enough to provide a wide range of subcooled refrigerant temperatures from 4 to 25 C There were two ways the water loop could be configured in order to achieve this range First building cold water supply could be routed directly through the subcooler and on into the drain This provided a functional range of 12 to 25 C Second a small percentage of chilled water from the evaporator could be routed through the subcooler to reach the lower end of the range from 4 to 12 C For all of the runs conducted in this stud
99. ut that proved disastrous as the thin bridges were not strong enough to withstand the flow in the test section pipe These features are highlighted Figure 3 3 which shows the complete MEMS sensor after microfabrication Immediately behind the sensors the silicon has been etched back to a thickness of only 40 um This helps minimize heat conduction losses through the substrate and assures that most of the self heating is dissipated by convection to the surrounding vapor and conduction to liquid droplets boiling at the surface 16 3 5 2 Microfabrication For a detailed description of the microfabrication techniques used to produce the sensors the reader is directed to the work of Shannon et al 3 The specifics of microfabrication are relevant to this project only with regard to the constraints that fabrication techniques placed on the sensor design This study is not concerned with the impact and contribution that the sensor fabrication techniques have to the world of MEMS Rather this study views the MEMS sensor as a potential tool with which to measure and control maldistribution in multi channel evaporators It is the intent of this project to assess the feasibility of using these devices to realize a sensing and actuation strategy for multi valve flow control 3 5 3 Packaging and installation in refrigerant piping The prototype MEMS sensor is very large at least with respect to the typical scale of most MEMS devices It is also
100. xperiment the main evaporator superheat was held constant at ATsup 12 C The main evaporator flowrate was also held constant at m 40 g s while the secondary evaporator flowrate m vaired from 0 to 1 5 g s Data collection consisted of only low speed sampling A total of 150 data sets of the 20 instrument channels were sampled and recorded for a total of 200 seconds Table 4 1 Test envelope for methods of calculating LMF 40 g s Oto 1 5 g s 2 C 0 2 C LMF 0 to 3 5 adjust to maintain ATsup 12 C 4 4 Procedures for Correlating Instrument Signals to LMF 4 4 1 Procedures and data collection To correlate the thermocouple MEMS sensor and scattered light signals to the presence of droplets a known amount of liquid was injected into the test section Both qualitative and quantitative comparisons of the signals are made in section 5 2 The main evaporator was run at a high degree of superheat 12 C to ensure that no liquid was present at the evaporator exit and a small amount of liquid was added to the test section using the secondary evaporator The same seven test section instruments listed in section 4 3 were again scanned at high speeds In addition the largest MEMS sensor voltage and the scattered laser light photodiode voltage were measured using the lock in amplifiers 4 4 2 Test envelope The test envelope for correlating the thermocouple MEMS sensor and scattered light signals to liquid mass fraction
101. y the target subcooled temperature was 16 C so building water was always used in the subcooler 2 1 3 Test section The test section consisted of a laser section the MEMS resistance sensor a glass tube for flow visualization a static flow mixer a calorimeter and several thermocouples and pressure transducers for monitoring flow conditions The unsteady mixture of superheated vapor and entrained liquid droplets from the evaporators first passed through the laser section It consisted of a 2 0 mW Helium Neon laser a light chopper and two photodiodes The light chopper and other related laser instrumentation are discussed in section 3 3 The laser shown in Figure 2 6 was aimed through an optical window perpendicular to the flow Some of the beam was scattered by refrigerant droplets and the rest passed through the flow stream unaffected The unaffected laser light was collected by a photodiode positioned directly across the flow stream along the laser axis A portion of the scattered beam was collected by a second photodiode located above the flow centerline as seen in Figure 2 4 The entrained refrigerant droplet volume can be measured from these two photodiode signals Pressure Transducer Scattered Light Photodiode MEMS Sensor Circuitry Secondary Evaporator Thermocouple in glass tube Primary Evaporator MEMS Sensor between flange Figure 2 4 Photo of the test section
102. y A C systems Building cold water supply was used as the cold fluid in the condenser which operated in a counter flow arrangement A manually set water flow control valve adjusted condensing pressure A receiver was installed downstream of the condenser to collect the high pressure liquid before entering the liquid subcooler The subcooler was also a SWEP design plate heat exchanger model B8x20 A flexible albeit complicated water flow loop allowed for a wide range of achievable subcooled temperatures The water flow loop is discussed in more detail in section 3 1 2 Typically building cold water supply was circulated through the subcooler to provide a subcooled refrigerant temperature of 16 C before the expansion device Circulating cold water from the evaporator through the subcooler could attain colder subcooled temperatures Subcooled refrigerant then divided into two branches the main evaporator and the secondary evaporator This combination of the main and secondary evaporators constituted a parallel pass evaporator in which maldistribution could be induced by the user Manual expansion valves fed both evaporators Manual control allowed the flow through each expansion valve to be precisely controlled such that the main evaporator operated with high exit superheat while the exit of the secondary evaporator was in the quality region The exit streams reunited prior to the test section such that the combined flows superheated vapor from the main ev

Exploratory Research on MEMS Technology for Air

Contents

Download Pdf Manuals

Related Search

Related Contents