A Novel Needle Probe Based on High-Speed Complex Permittivity Measurements for Investigation of Dynamic Fluid Flows

A Novel Needle Probe Based on High-Speed Complex Permittivity Measurements for Investigation of Dynamic Fluid Flows
of 8
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 56, NO. 4, AUGUST 2007 1249 A Novel Needle Probe Based on High-SpeedComplex Permittivity Measurements forInvestigation of Dynamic Fluid Flows Marco Jose Da Silva,  Student Member, IEEE  , Eckhard Schleicher, and Uwe Hampel  Abstract —For the investigation of multiphase or multicompo-nent flows, which are of interest, for instance, in oil extractionand processing or in chemical engineering, there are only fewsuitable measuring techniques. For this reason, we have developeda high-speed complex permittivity needle probe. Such probes areable to distinguish the different phases or components of a flow bymeasuringthecomplexvalueoftheelectricalpermittivityatahighdata rate (up to 20000 samples/s). The performance of the system,as well as its ability to differentiate organic substances, has beenanalyzed. A time-resolved experiment in an oil–water–gas flow, aswell as a two-substance mixing experiment in a stirred tank, ispresented.  Index Terms —Complex permittivity measurement, concentra-tion measurement, heterogeneous mixture, multicomponent flow,multiphase flow, needle probe. I. I NTRODUCTION M ULTIPHASE and multicomponent flows can be foundin many industrial application areas, for instance, inchemical engineering, oil extraction and processing, food anddrink processing, and pharmaceutical industry. For many rea-sons, the high-speed phase or component distribution mea-surement of such flows is of industrial and scientific interest.Thus, in chemical engineering and oil processing, the moni-toring of increasingly complex industrial processes is required.Typical objectives are assessment product quality, preventionof undesired or hazardous process conditions, monitoring of educt transport, and efficiency of component mixing or productextraction. Furthermore, in the scientific community, fast mea-suring techniques are gaining increasing importance for the ex-perimental investigation of multiphase flows and the validationof computational fluid dynamics codes.Measurement with local needle probes is a well-establishedmethod for the investigation of two-phase flows. Needle probescan identify the phase, which is momentarily present at thelocation of the needle tip. Different measuring techniques havebeen employed for the needle probes in the past. Most sen-sors employ conductivity, capacitance, optical, and temperature Manuscript received June 30, 2006; revised May 3, 2007. The work of M. J. Da Silva was supported by the “Coordenação de Aperfeiçoamento dePessoal de Nível Superior” (CAPES), Brazil. Part of this work was presented atthe 2006 IMTC Conference (Paper IM-6100).The authors are with the Institute of Safety Research, ForschungszentrumDresden-Rossendorf e.V., 01314 Dresden, Germany (e-mail: versions of one or more of the figures in this paper are available onlineat Object Identifier 10.1109/TIM.2007.900419 measurement techniques [1]–[3]. Each of these techniques haslimitations regarding the range of substances that it is able tomeasure [3], [4]. Conductivity or resistivity probes can onlydifferentiate an electrically conducting phase from a noncon-ducting one. Capacitances probes encounter problems whenpermittivities of the two phases are similar. The same difficultyoccurs with optical probes and low refractive index differences.Temperature probes require sufficient temperature differencesin the substances, which are to be discriminated. Therefore,these probes have been utilized almost exclusively in two-phaseflow measurements.In the last years, some efforts have been made by applyingdual modality measuring techniques for enhancing the rangeof substances in the investigation of heterogeneous mixtureflow, in which two different sensing techniques are used todistinguishthedifferentflowconstituents.Prasser etal. [5]havedeveloped a conductivity probe with integrated microthermo-couple, which is able to synchronously measure temperatureand conductivity and which has been applied to investigatea steam–air–water flow. An optical needle probe combiningreflectance and fluorescence methods has been reported in [6]for the investigation of oil–water–gas flows. A dual fiber opticalneedle probe was described in [7], also for the investigation of oil–water–gas flows. Reflectance signals of a cleaved fiber wereused for gas/liquid discrimination and reflectance signals of anoblique probe for oil/not-oil discrimination. In [8], a sensorthat uses parallel optical and conductivity measurements for theinvestigation of three-phase slugflow incapillaries isdescribed.Many multiphase flow meters that are currently used in the oiland gas industry are based on dual modality measurements,e.g., monoenergetic gamma ray absorption combined with re-sistance or capacitance measurements or dual energy gammaray absorption measurements (for a review, see [9] and [10]).The use of complex electrical impedance/admittance measure-ments, i.e., the measurement of both real and imaginary parts(or amplitude and phase), has been reported for multiphase flowsensors in [11] and [12]. Here, an integral and nonintrusiveapproach was used for water-cut measurements. Furthermore,in [13], the development of a sensor combining ultrasonic andelectrical measurement for the investigation of heterogeneousmixtures is described.The combination of conductivity with dielectric constantmeasurements could enhance the applicability of needle probesystems. This means that instead of measuring in just one line(real or imaginary axis) in the complex plane of permittivities,one may use the full complex plane to characterize the fluids. 0018-9456/$25.00 © 2007 IEEE  1250 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 56, NO. 4, AUGUST 2007 Fig. 1. Schematic diagram of needle probe and photographs of both manufactured probes. The electrodes are of 1 and 0.9 mm diameter for probes “one” and“two,” respectively. Complex permittivity measurements are a standard tool in an-alytical chemistry [14], in which measuring times on the orderof seconds to hundred of seconds are common to investigatesometimes entirely unknown substances. On the other hand,in process diagnostics, measuring times in the range of mi-croseconds to milliseconds are required in order to investigatedynamic processes, e.g., mixing of substances and multiphaseflow. However, the substances involved (and, consequently,their electrical properties) are known  a priori . Commercialmeasuring instruments like  RLC   meters or impedance analyzerscan only achieve a few measurements per second and are thusnot suitable for high-speed measurements, which are requiredin the investigation of dynamic fluid flow.This paper describes a novel needle probe measuring systembased on high-speed complex permittivity measurements. Thisneedle probe can be employed for the investigation of dynamicprocesses in multiphase or multicomponent flow. The systemperformance in measuring the complex permittivity is evalu-ated. First results for the three-phase flow and the mixing of two substances are presented.II. N EEDLE  P ROBE  S YSTEM The system consists of a needle probe and proper electroniccircuitry that is responsible to generate and measure the sine-wave signals that are required to determine the complex per-mittivity. The probe itself is built in a double coaxial geometry(Fig. 1). Two stainless-steel electrodes (i.e., the excitation andmeasuring electrodes) are isolated from the common groundelectrode. Both coaxial structures are soldered together orplaced inside a third stainless-steel tube. We have manufacturedtwo needle probes with different sizes. Fig. 1 shows both probesand their dimensions. The electrical capacitance of the free pairof electrodes (sensitive area) is approximately 0.06 and 0.1 pFin air for the probes named “one” and “two,” respectively.Theprincipleofoperationoftheneedleprobesystemisgivenas follows. The electronics applies a sinusoidally alternatingvoltage to the excitation electrode and determines the complexvalue of current (i.e., amplitude and phase) flowing from theexcitation electrode toward the measuring electrode. The ad-mittance  Y   is calculated by Ohm’s law, i.e., the ratio betweencurrent  I   and voltage  V    (for details, see the next section).The measured admittance is linked with the electrical con-ductivity  κ  and relative permittivity  ε r  of the fluids at theprobe tip by [14] Y   =  k g ( κ +  jωε 0 ε r ) .  (1)The geometry factor  k g  reflects the ratio of the cross-sectionalarea of the sampled fluid volume and the length or distancebetween the electrodes. This way, the introduction of differentfluids at the probe tip will produce different measured admit-tance values, allowing these substances to be distinguished.From (1), we note that a large area/length ratio is desired toproduce large admittance values that are easier to detect by theelectronics. There exist, hence, two possibilities to optimize thegeometry. The probe tip can either be designed with a large areaor the distance between the electrodes can be decreased. Never-theless, there are limits for these arrangements. A larger probetip (free electrodes) and, thus, a larger area would produce ahigher disturbance in the flow, which isundesired. Furthermore,a smaller distance between the electrodes can make the probeinsensitive to given flow conditions; for instance, air bubblesmight not be detected due to dewetting problems at the probetip. Therefore, we constructed two probes. Probe “one” is betterappropriate for a bubbly flow due its arrow-type tip, which canbetter pierce bubbles and its larger electrode distance. Probe“two” was optimized for the investigation of a mixing processin a stirred tank, which is better described in Section IV.Evidently, other scaled-up or -down variations of the probecan be designed in order to adapt to other applications orconstraints.  A. Electronics A block diagram of the electronic circuitry can be seenin Fig. 2. A direct digital synthesizer (DDS) IC (AD9835,Analog Devices) generates the excitation sine-wave voltage.Frequencies from 100 kHz up to 1 MHz can be generatedand are user selectable via a control program in the PC. Thesine-wave excitation voltage amplitude can be adjusted by apotentiometer, and its maximum value is limited by the supplyvoltage of the output buffer, which is  ± 5  V (bipolar supply).  DA SILVA  et al. : NOVEL NEEDLE PROBE BASED ON HIGH-SPEED COMPLEX PERMITTIVITY MEASUREMENTS 1251 Fig. 2. Block diagram of electronic circuitry. Typical currents flowing between the electrodes are in themicroampere range.The excitation voltage  V   i  is supplied to the excitation elec-trode by means of a coaxial cable. The measuring electrodeis connected to a transimpedance amplifier also by means of a coaxial cable. The output voltage of the transimpedanceamplifier isdirectlyproportional tothecurrent I  M   flowing fromthe excitation to the reference electrode, i.e., V   o  =  I  M  Y  f  (2)where  Y  f   is the admittance in the feedback loop of the op-erational amplifier (OpAmp). Since the unknown measuringadmittance  Y  M   of the substance at the sensitive tip is groundedby the OpAmp’s virtual ground, the current  I  M   is the productof the input voltage  V   i  and this admittance. Hence V   o  =  I  M  Y  f  =  V   i  · Y  M   ·  1 Y  f  .  (3)The output voltage  V   o  is then input to the amplitude andphase detector block. The IC AD8302 (Analog Devices) real-izes the function of measuring amplitude and phase. This ICdetermines the amplitude ratio and the phase difference from V   o  and a reference signal V   R . The AD8302 measures phase dif-ferences in two quadrants (e.g., from 0 ◦  to 180 ◦ ) and achievesthe best accuracy in the phase measurement only between phasedifferences of 20 ◦  and 160 ◦  (nominal accuracy  = 0 . 5 ◦ ) [15].Since the phase difference caused by the needle probe wasexpected to be as small as a few degrees for nonconductingsubstances, a phase shifter was required to allow the AD8302to work with highest accuracy. The phase shifter was easilyimplemented by means of a second DDS, which generatesa reference voltage  V   R  with known amplitude and digitallyprogrammable phase. We usually generated a 90 ◦  phase delay,thus placing the operation point of the AD8302 in the centerof its measuring range. For this reason, both DDS generatorsare phase synchronized by sharing the same clock source. Abig advantage of a DDS-based phase shifter is the fact that thephase shift is independent of frequency. The two DC output sig-nals of the AD8302, which carry the information on amplitudeand phase difference, are fed into the A/D converter (12 bits,20000 samples/s per channel) of the universal serial bus (USB)data acquisition (DAQ) module PMD-1208FS (Measurementand Computing). The USB DAQ module is connected to aPC through a USB 2.0 port. The digital port of the USBDAQ module is used to program the DDS sine generators.Furthermore, a control and measurement software in the PCdetermines the frequency and phase lag of each DDS, controlsthe A/D conversion sequence, and realizes the calculation of theadmittance from the measured amplitude and phase values.  B. Calibration The measuring admittance  Y  M   represents the sum of thefluid admittance  Y  x  and a residual admittance  Y  R  (e.g., straycapacitance, resistance, and inductance of cables), i.e., Y  M   =  Y  x  + Y  R .  (4)This fact makes a calibration routine necessary. From (3) and(4), we obtain V   o  =  − Y  x  + Y  R Y  f  V   i  =  Y  x − V   i Y  f  −  Y  R Y  f  V   i .  (5)As can be seen in (5), the fluid admittance  Y  x  is linearlydependent on the output voltage  V   o  in the form Y  x  =  α + β   · V   o .  (6)Therefore, a two-point calibration is sufficient to determinethe variables α  and β  . For this purpose, two known admittances( Y  1  and  Y  2 ) are connected to the probe, and  V   o  for bothconditions (i.e.,  V   ′ o  and  V   ′′ o  ) is measured. The values for  α  and β   are then determined as follows: α  =  Y  2  · V   ′ o  − Y  1  · V   ′′ o V   ′ o  − V   ′′ o (7) β   =  Y  1  − Y  2 V   ′ o  − V   ′′ o .  (8)Fluids are better characterized by their complex relativepermittivity, rather than their impedance or admittance. Thecomplex relative permittivity  ε ∗ r  is defined as [16] ε ∗ r  =  ε ′  +  j  · ε ′′  =  ε r  +  κ jε 0 ω  (9)where  ε ′  and  ε ′′  are the real and imaginary parts of   ε ∗ r , re-spectively,  ε r  is the relative permittivity or dielectric constant( ε r  =  ε ′ ), κ is the conductivity, ε 0  is the permittivity of vacuum(8.85 pF/m),  ω  is the angular frequency ( ω  = 2 πf  , where  f   is  1252 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 56, NO. 4, AUGUST 2007 the excitation frequency), and  j  = √ − 1 . If two known liquidswith known  ε r  and  κ  are used for the calibration procedureas described above, the system will also be able to determinethe complex relative permittivity  ε ∗ r  from measurements of   V   o ,since admittance  Y   and complex permittivity are equivalent.Another form to write (1) is [14] Y   =  ε ∗ r  · ε 0  ·  jω  · k g .  (10)This way, instead of using  Y  1  and  Y  2  in (7) and (8), one can use ε ∗ r 1  and  ε ∗ r 2  to calibrate the needle probe. Equivalent to (6) ε ∗ r  =  µ + ν   · V   o  (11)applies, whereby (7) and (8) are modified to determine  µ  and  ν  as follows: µ  =  ε ∗ r 2  · V   ′ o  − ε ∗ r 1  · V   ′′ o V   ′ o  − V   ′′ o (12) ν   =  ε ∗ r 1  − ε ∗ r 2 V   ′ o  − V   ′′ o .  (13)With the help of the above formulas, it is now possible tocalibrate a complex permittivity needle probe at two materialswith known electrical properties, e.g., air and water. In the casewhen one is interested only in differentiating the phases orcomponents of a fluid flow, e.g., gas phase detection in a bubblyflow, the calibration procedure is no longer necessary, since themeasured complex voltage  V   o  already carries this informationin its complex value. Each component or phase is associatedwith a point in the complex plane. Thus, this value can be evalu-ated directly to determine each phase or component of the flow.III. S YSTEM  P ERFORMANCE  E VALUATION  A. Drift and Noise System outputs, i.e., amplitude and phase measurements,were evaluated for drift and random noise. A 100-k  Ω  resistorwas connected to the needle probe electrodes, and the outputswere monitored for 1 h at a maximum sampling rate of 20 kS/s.A 1-min moving-average filter was used to reduce randomnoise for determination of the drift. System drift was definedas the maximum deviation of the outputs over the period. Themeasured drift was 34.13 mV and 0.637 ◦ . Random noise wasestimated as the standard deviation of the signals over a timeequal to 1 min. Measured random noise was 14.42 mV and0.272 ◦ . Applying error propagation rules for (11), one canestimate the driftand random noise for the complex permittivityparameter as follows: E  ( ε ∗ r ) =  ν   · E  ( V   o )  (14)where  E  ( ε ∗ r )  represents the error in the complex permittivity, E  ( V   o )  is the error in the measured output voltage, and  ν   isdefined in (13). Drift and random noise were then calculatedfrom (14) by using typical calibration values for  ν  . Drift valueswere 0.4272 and 0.0023  µ S/cm for permittivity and conductiv-ity, respectively. Furthermore, random noise value was 0.1810for permittivity and 0.0009  µ S/cm for conductivity. Fig. 3. Comparison in the capacitance measurements of an  RCL  meter andthe needle probe system. The dashed lines represent the 5% deviation from theideal line.Fig. 4. Comparison in the resistance measurements of an  RCL  meter and theneedle probe system. The dashed lines represent the 5% deviation from theideal line.  B. Admittance Measurement  The system has been extensively tested by measuring dif-ferent resistors, capacitors, and parallel combinations thereof.For this purpose, a set of SMD components (size 1206) werebrought in contact with the probe electrodes by means of aconnector. Resistor and capacitor values were chosen to matchthe expected measuring values of common fluids. Componentsin the range 100 k  Ω  to 1 M Ω  (seven values) and up to 15 pF (sixvalues) were used. A reference value of each component hasbeen first measured by an  RLC   meter (Fluke PM6304,  ± 0.4%nominal accuracy). All reference values lay in the nominal10% tolerance range of the components. First, the system wascalibrated with two resistors (with nominal values equal to100 k  Ω  and 1 M Ω ), and the excitation frequency was set to200 kHz. The sampling frequency was 20 kS/s. The feedback admittance Y  f   employed was100k  Ω inparallelwith10pF.Thevalues measured by the system were then compared with thereference values from the  RLC   meter. Diagrams with the resultsfor resistance and capacitance measurements can be seen inFigs. 3 and 4, respectively. All results fall within a 5% deviation(dashed lines in Figs. 3 and 4). Values shown in the figures arethe mean value for ten measurements.  DA SILVA  et al. : NOVEL NEEDLE PROBE BASED ON HIGH-SPEED COMPLEX PERMITTIVITY MEASUREMENTS 1253 Fig. 5. Results of conductivity measurement by the needle probe comparedwith reference values. The dashed lines represent the 7% deviation from theideal line.Fig. 6. Results of relative permittivity measurement by the needle probecompared with reference values. The dashed lines represent the 7% deviationfrom the ideal line. The mean value of the relative errors for all 42 measurementsis 2.24% for resistance and 2.27% for capacitance. Maximumrelative errors of 5.11% and 3.98% were obtained in theresistance and capacitance measurements, respectively. Thisexperiment shows the system accuracy in the range  | Y  |  = 0 . 1 –21.3 mS and angles from 0 ◦  to 86.9 ◦  at a fixed frequency.For the case where other measuring ranges are required, thefeedback admittance  Y  f   can be easily adjustable by selectingappropriate values of capacitors and resistors, or the excitationfrequency can be adapted to achieve the required measuringrange. C. Complex Permittivity Measurement  The system has also been tested in measuring differentliquids and air. The liquids were put in a metal cylinder of 10 cm height and 10 cm diameter, and the probe was locatedin the center. First, the system was calibrated by measuringair (empty probe) and deionized water. The excitation fre-quency was 200 kHz, and the sampling frequency was 20 kS/s.The measurements were carried out with both needle probedesigns. Figs. 5 and 6 show the results of measured properties TABLE IR EFERENCE  R ELATIVE  P ERMITTIVITY  ε r  AND  E LECTRICAL C ONDUCTIVITY  κ  OF  D IFFERENT  L IQUIDS AND  A IR (mean value of ten measurements) compared with referencedata. These figures also show a 7% deviation line (dashed).Reference relative permittivity was taken from [17] and cal-culated for the ambient temperature of the samples. Referenceconductivity and temperature of the samples were measuredby the conductivity meter Cond 330i (WTW GmbH, Germany,0.5% accuracy). The reference values are shown in Table I.The maximum relative error compared with the referencevalues was 7.0% for permittivity and 4.9% for conductivity.Both constructed needle probes present similar results, whichshows that the calibration routine fully compensates the differ-ent geometry factors (due to the different dimensions) of eachprobe.IV. E XPERIMENTAL  R ESULTS  A. Three-Phase Flow Measurement  In order to examine the capability of the present system tomeasure dynamic phenomena, it has been employed to measureair and water bubbles flowing in gasoline. The system was,again, calibrated by measuring air and deionized water. Theexcitation frequency was 200 kHz, and the sampling frequencywas 20 kHz. Probe “one” was then located inside an acrylicglass column (rectangular section, 20  ×  10 mm) filled withgasoline. The flow of bubbles was synchronously recorded witha high-speed video camera (MotionPro HS Series, Redlake)in order to evaluate the needle probe data. First, air bubbleswere generated at the bottom of the column. Fig. 7 shows thetemporal evolution of relative permittivity and conductivity reg-istered by the needle probe for the measurement of a single airbubble (approximately 200 mm 3 ), as well as three single-shotimages of the ascending bubble. The relative permittivity valueis lowered by the passage of the bubble, but the conductivitysignal remains constant, as expected, since gasoline and air arenonconducting. As a second experiment, water bubbles with aconductivity of 2  µ S/cm were dropped into the column. Fig. 8shows the results. Again, three single-shot images show thewater bubble (approximately 100 mm 3 ) descending in gaso-line, and the diagram shows the temporal evolution of relativepermittivity and conductivity. In this experiment, both signalsincrease, since water is conductive and has a larger relativepermittivity  ( ε r  ≈  80)  than gasoline  ( ε r  ≈  2) .A further experiment has been carried out to demonstratethe capability of the system in the investigation of three-phase flows. A few air bubbles and then a few water bubbles ( κ  = 2  µ S/cm )  were sequentially generated in gasoline. The
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks