WSMJ: Electromagnetic W~ve Interaction · content in solids and liquids, silllilar to hygrometry, which is a branch of metrology devoted to measurement of water vapor content in gases.

WSMJ: Electromagnetic W~ve Interaction

with Water and Moist Substances

ORGANIZER:

Andrzej Kraszewski

USDA-ARS

WSMJ

SUMMARIES

of the contributions to be presented at the Workshop on

Electromagnetic Wave Interaction with Water

and Moist Substances

held in conjunction with the IEEE Microwave Theory and Techniques Society

1993 International Microwave Symposium

Sponsored by the MTT-l1 Microwave Measurements Technical Committee

Atlanta, Georgia, June 14, 1993; 1 PM

Georgia World Congress Center, Room 217.

- 3 -

LIST OF CONTE TT

Page number

\Velcome to the \Vorkshop A. Kraszewski , USDA, ARS, Russell Research Center, Athens, GA .. 5

A. Physical Aspects of 11icrowave Aquametry

1. "Dielectric Relaxation Study of \\-rater/Oil Microemulsion Systems" T.K. Bose, R. Chahine, A. Ponton, R. Nozaki. Dept. of Physics, Dielectric Research Group, Universite du Quebec, TroiswRivieres, Quebec, Canada. . .... . ... 7

,~.\ 2. "Dielectric Properties of Water Dispersed and Confined in Different Systems" uJ"" P. Pissis*, A. Anagnostopoulou-Konsta, A. Kyritsis, R. Pelster, A. Enders, G. Nimtz.

c)r"" *Dept. of Physics, National Technical University, Zografou Campus, Athens, Greece. . ........ 11

,,1 3. "Free an? Bound v:r ater in Various M~trix Sys~ems. Stu~ied by Advanced Microwave Technique" V~ S. Mashlmo, N. MlUra. Dept. of PhysIcs, Tokal Umverslty, Kanagawa, Japan. . ....... 15

4. "Measurement of Bound and Free \Vater in Mixtures" A. Brandelik, G. Krafft. Kernforschungszentrum, Karlsruhe, Germany. . ......... 19

5. "Dielectric Mixture Theories in Permittivity Prediction: Effects of \Vater on Macroscopic Parameters" A. Sihvola. Electromagnetic Lab., Helsinki University of Technology, Espoo, Finland. ....... 21

6. "Effective Dielectric Constant of Moist Substances" S. Kobayashi. R&D Center, Sumitomo Metal Industries Ltd., Amagasaki, Japan. • ....••.. 25

7. "Hydrocarbon and 'Vater Estimation in Reservoirs Using Microwave Methods" T. Hac. c...shy, P.N. Sen, R. Taherian. Schlumberger-Doll Research, Ridgefield, CT. ...... ~. 27

B. Technical Aspects of ~1icrowave Aquametry

1. 11easuring Sensors and Circuits

8. "An Integrated Microwave Moisture Sensor" G.B. Gentili, G.F. Avitabile, F. Ceccuti, G.F. Manes. Dept. of Electronic Engng., Universita di Firenze, Florence, Italy. . ............................. 33

9. "Integrated Microwave Sensors for Agricultural and Industrial Applications" F. Volgyi. Dept. of Microwave Telecomm., Technical University of Budapest, Hungary. . ..... 39

10. "Development of Microstrip Sensor for Oil Palm Fruits" K.B. Khalid, Z.B. Abbas. Dept. of Physics, University Pertanian Malaysia, Serdang, Selangor, Malaysia. . ............ 45

11. "Material Characterization Using Microwave Open Reflection Resonator Sensor" R.J. King, J.C. Basuel, M.J. 'Verner, K.V. King. KDC Technology Corp., Livermore, CA ..... 53

12. "Microwave Resonant Sensors for Moisture Content Determination in Single Grain Kernels, Seeds and Nuts" A.W. Kraszewski, S.O. Nelson. USDA, ARS, Russell Research Center, Athens, GA ......... 55

- 4 -

13. "A New Moisture Content Measurement Method by a Dielectric Ring Resonator" S. Okamura, T. Masuda. Faculty of Engineering, Shizuoka University, Hamamatsu, Japan .... 59

14. "Design and Optimization of Electromagnetic Sensors for Dielectric Spectroscopy by Using .• ~ the (FD)2 TD Method" ~ G.B. Gentili, M. Leoncini, D. Bertolini, G. Salvatti, E. Tombari. Dept. Electronic Engng,

University of Florence & IFAM-CNR, Pisa, Italy. . ........ 63

15. "~fodeling of Open-Ended Coaxial Line Sensors for Determination of the Complex Permittivity of Materials at Microwave and MM-Wave Frequencies" C. Pournaropoulos, D. Misra. Dept. Electrical Engineering & Computer Science, University of Wisconsin, Milwaukee, WI. ....... 67 .

16. "Calculation of Sensitivity of Various Coaxial Probes Used in Microwave Permittivity Measurements" Y. Xu, R.G. Bosisio. POLY-GRAMES, Ecole Poly technique de Montreal, Quebec, :anada. . . .. 69

2. Measuring Methods and Experience

17. "Accurate Percent \Vater by Microwave Interaction Alone" C.W.E. \\'alker. Pacific Automation Instruments, Ltd., Richmond, B.C., Canada.

18. "Possibilities and Limitations of the Density-Independent Moisture Measurements with ~1icrowaves"

. ..... 75 .

K. Kupfer. Hochschule fur Architektur und Bauwesen, Weimar, Germany. . ........ 77

19. "On the Permittivity of Wood and the Measurement of Moisture and Mass Per Area in Veneer Sheets" E. Nyfors, P. Vainikainen, M. Fischer. Radio Laboratory, Helsinki University of Technology, Espoo, Finland. . .......... 79 '

20. "Experience with the Microwave Moisture Meter MICRO-MOIST" A. Klein. Laboratorium Prof. Dr. Berthold, Wildbad, Germany.

21. "Determination of Water Content in Oil Pipelines Using High Frequencies"

........... 81 .

S.K. Aggarwal, R.H. Johnston. Dept. Electrical Engng, University of Calgary, Alberta, Canada. . ......... 83

22. "Continuous Microwave Moisture Measurement for Particulate Materials and Fluids" V.N. Tran, Y.Shen. School of Science, Deakin University, Geelong, Victoria, Australia. • ..... 89 .

23. "Present Possibilities for Moisture Content Measurements by Microwaves" T.Lasri*, B. Dujardin, Y. Leroy, Y. Vincent, G. Mallick. *Universite des Sciences et Technologies de Lille, France. . . . . . . . . . .. 93 .

24. "Single and Multi-Frequency Phase Change Methods for Microwave Moisture Measurement" A. Robinson, M. Bialkowski. Dept. Electrical Engng, University of Queensland, Brisbane, Queensland, Australia. . .......... 95

25. "Bulk Moisture Measurement of Agro-Based Products by Estimating Phase Variation at Microwave Frequency" H. Singh, S. Shekhar, C.H. Shah. Central Electronics Engineering Research Institute, Pilani, Rajasthan, India. . ......... 97 '

- 5 -

Welcome to the Workshop!

This Summary of Workshop Papers has been assembled single-handedly by the

Organizer from materials received from the speakers. From among various sizes, forms

and styles I modified only the necessary items, condensed three full-length papers and

retyped several pages from poor fax copies. The subjects are also those proposed by the

speakers. The fact that there is more than one contribution on a given subject indicates

interest in that topic at various places around the world.

One of the principle aillls in organizing this meeting was to provide a broad forum

for exchange of information and ideas among all interested in the subject. The presence

of 25 speakers (more than 60 authors prepared the submitted manuscripts) from 12

countries, and the serious interest of several more who for various reasons were not able

to attend, indicates that there is an interest in microwave aquametry in many untries

of the world.

Some of you may not be familiar with the term, "microwave aquametry", and to

those I owe a few words of explanation. Aquametry deals with measurement of water

content in solids and liquids, silllilar to hygrometry, which is a branch of metrology

devoted to measurement of water vapor content in gases. The adjective microwave

indicates that measurements of moisture content are carried out by microwave

techniques. It has been demonstrated during the past 30 years that microwave

aquametry has its definite subjects and tasks (cognitive and practical objectives) and

particular research methods, as well as a characteristic instrument base. Microwave

aquametry investigates solids and liquids containing water by identifying their

properties in the electromagnetic fields. Microwave aquametry utilizes some well known

physical theories, for example the theory of dielectric mixtures and the theory of bound

water.

Besides these cognitive purposes microwave aquametry has strictly defined

practical objectives, namely, quantitative measurements of the water content in

materials important from the economical point of view. Since water appears in the

majority of materials encountered in nature, being introduced there purposefully during

specific processes or being present naturally, it is obvious that these objectives have far­

reaching economic importance. This fact notably influenced the progress in the past and

is evident in the present state of microwave aquametry.

- 6 -

All aspects of this defined branch of measuring science will be covered during the

workshop. The best world experts will discuss properties of moist materials and water

bound to various rnaterial structures, dielectric rnixture theories allowing prediction of

these properties from those of the constituents, development, theoretical principles and

practical applications of new lllicrowave sensors, as well as new methods that offer

density-independent moisture content Ineasurements in various environmen.ts.

Unfortunately, tilne for this rneetillg is very linlited, but I hope that the exchange

of information begun here in Atlanta \vill continue in the months to come and will

flourish in similar meetings in the future.

I want to thank my supervisor and friend, Dr. Stuart Nelson, for his help and

encouragement during the whole 9-1nonth preparation period, and Dr. Stephen Adam,

Chairman of the MTT-ll Microwave ~leasurement Technical Committee, for sponsor­

ing the vVorkshop.

I believe we all will have a fruitful, although short, meeting that you will enjoy

and remember.

Andrzej Kraszewski

WSMJ Workshop Organizer ..

- 7 -

DIELECTRIC RELAXATION STUDY OF W A TERIOIL

MICROEMULSION SYSTEMS

T.K.Bose, R. Chahine*, A Ponton and R. Nozaki

Department of Physics, * Department of Ingeneering Dielectric Research Group

i\BSTRACT

Universite du Quebec a Trois-Rivieres Trois-Rivieres, Quebec, Canada

Extensive dielectric measurements have been perfonned at 250

C on several water/oil microemulsion systems composed of water, oil, a surfactant and a co-surfactant (alcohol). Recent improvements in time-domain reflectrolnetry (TDR) have made it possible to make precise measurements of complex permittivity over a wide frequency range froln about 100 kHz to 20 GHz. Two processes, occurring in the low frequency region aroWld 1 :MHz and 100 :MHz seem to he of ionic origin. The relaxation observed in the high frequency range (1 GHz) could possibly be attributed to the presence of water and alcohol in the vicinity of the reverse lnicellar surface. The strength of the dielectric relaxation process in the very high frequency region aroWld (16 GHz) is dependent on the water content and is possibly related to pure water outside the interfacial layer.

- 8 -

SUl\tlMARY

During recent years considerable experimel1tal l -3 and theoretical4,5 effort has been devoted to the study of microemulsiol1s because of their practical importance and interesting physical and chemical properties. A microemulsion is composed of water, oil, a surfactant and often a co-surfactant. It is simply cOlnposed of "swollen micelles:" water droplets in oil or the inverse~ the water/oil interface is saturated by surfactant molecules. A microemulsion is fonned spontaneously with the right amount of compositions.

The sizes of the microemulsion droplets may vary from 10 to 100 A. Over the last few years, micellar systems have been studied by means of dielectric spectroscopy. The purpose of the present work is to report on electrical conductivity and complex pennittivity measurements in the frequency range 100 kHz - 20 GHz for two microemulsion systems, respectively composed of water, n-dodecane, I-pentanol, SDS (sodium dodecylsulfate) and water, toluene, 1 butanol, SDS. These systems have been chosen because their phase diagrams6 are well known and they contain the SaIne ionic surfactant (SDS).

The dielectric measurements have been carried out using time-domain reflectrometry (TDR) technique7

. Fig. 1 shows the basic concept of TDR. A step-like pulse produced by a pulse generator propagates through the coaxial line aIld is reflected from the sample section placed at the end of the line. The difference between the reflected and the incident pulses recorded in the time domain contain the information on the dielectric properties of the saInple. Since the dielectric response is recorded in the time-domain, one measurement covers a wide frequency range, sometimes over two decades. The measurement in the time domain is easily converted to the frequency domain by Fourier tran sform 8. From the measurement of the ionic microemulsion systems, it is clear that the TDR technique is applicable to the Ineasurement of the complex pennittivity of a sample with high de conductivity.

There is a definite advantage for studying the state of water in a microemulsion. As in living cells, water in a microemulsion exists both in the bound and free states. Another reason for studying water within a microemulsion of lower dielectric constant is the possibility of extending the Ineasurement range to higher frequencies. The cutoff frequency9 for the fundamental mode in a cylindrical wave guide is fmc (GHz) = 1.9 x I02/m-vE, where m is the diameter of the outer conductor in millimeters and E is the

- 9 -

dielectric constant of the substance filling the wave guide. For a coaxial cell design with 2 mm o.d., the cutoff frequency for pure water is around 10 GHz. We are, of course, assuming that the cylindrical wave guide cutoff frequency is valid in the fringing field space. Since the dielectric permittivity of the microernulsions studied is below 8 the high frequency limit is above 30 GHz.

m :>Tfng 1 10'h0::!

heac ~ V////A

pulse r(t)~ sample

gen. section

sampling oscilloscope CRT i computer ~

~ f GP-IB T

I peripherals 1

Figure 1

REFERENCES

1. M. Cazabat and D. Langevin, 1. Chern. Phys. 74, 3148 (1981). 2. M. Kotlarckyk, S.H. Chen, 1.S. Huang and M.W. Kim, Phys. Rev., 29A, 2057

(1984). 3. T.K. Bose, G. Delbos and M. Menibet, 1. Phys. Chern., 83, 867 (1989). 4. P.G. de Gennes and C. Taupin,1. Phys. Chern., 86, 2294 (1982). 5. W. Widorn, J. Chern. Phys.,,tl, 1030 (1984). 6. A.M. Bellocq, J. Biais, B. Clin, P. Lalanne and B. Lamanceau, J. Colloid Interface

Sci., 70, 524 (1979). 7. R. Nozaki and T.K. Bose, IEEE Trans. Instrum. Meas., 38,945 (1990). 8. R. Chahine and T.K. Bose, 1. Chern. Phys., 72, 808 (1980). 9. M. Merabet and T.K. Bose, 1. Phys. Chern., 92,6150 (1988).

Dr. Tapan K. Bose obtained B.Sc. (hons), M.Sc. and Ph.D. degrees in Physics from the University of Calcutta, India and Louvain, Belgium, in 1958, 1961 and 1965, respectively. He was a FOM research fellow at the University of Leiden (1965-67) and a research associate with Prof. Robert H. Cole at Brown University (1967-69). He joined the University of Quebec at Trois-Rivieres in 1969. He is now a full professor of Physics and director of the Dielectric Research Group. His research interest includes electromagnetic and thermodynamic properties of materials.

- 11 -

DIELECTRIC PROPERTIES OF WATER DISPERSED AND CONFINED 1N DIFFERENT SYSTEMS

P. Pissis1 , A. Anagnostopoulou-Konst~l, A. Kyritsis2 ,

R. Pelster3 , A. Enders3 and G. Nimtz3

1 Department of Physics, National Technical University of Athens, Zografou Campus, 15780 Athens, GREECE, 2University of Athens, Department of Physics, Panepis­timioupolis Zografou, 15771 Athens, GREECE, sUniversitat Kc5ln, IT Physikalisches Institut, Ziilpicher str. 77, SOOO Kc5ln 41, GERMANY

We report on detailed investigations of changes in the structure and the proper­

ties of water induced by dispersion and confinement in sm.a.ll volumes in three differ­

ent systems: di-myristoil-phosphatidylcholine (DMPC) bilayers, poly(hydroxyethyl

acrylate) (PHEA) hydrogels and butyl rubber containing hydrophilic components.

Different dielectric relaxation spectroscopy (DRS) techniques are used: broadband

ac techniques in the 5 Hz - 10 GHz frequency range and thermally stimulated depolar­

ization currents (TSDC) techniques in the 77 - 300 K temperature range. The latter

correspond to measuring dielectric losses VB. temperature at fixed frequencies of 10-2

- 10-' Hz and are characterized by ,high sensitivity, high resolving power and special

procedures for experimentally analyzing complex relaxation processes [1]. Additional

information on the structure and the properties of water and the matrix itself is ob­

tained. from differential sca'lning ca.lorimetry (DSC), sm.a.ll angle X-ray scattering

(SAXS) and dynamic and equilibrium water sorption and desorption measurements.

Several aspects of hydration and water confinement have been taken into account

in selecting the systems. DMPC bilayers are characterized by hydrophilic interactions

of water molecules with the head groups and one-dimensional confinement of water

in layers [2]. In PHEA hydrogels we have strong hydrophilic interactions of water

with polar side-chain groups and two-dimensional confinement of water in pores [3,4].

In the case of hydrophilic butyl rubber, finally, water diffuses into the rubber to the

hydrophilic components, which act as adsorption sites, and forms three-dimensionally

confined mesoscopic water droplets [5]. Details of the preparation of the samples, the

adjustment of water content, the experimental techniques used and the measurements

- 12 -

Information concerning the structure and the properties of water from DRS is

obtained from two different regions of the relaxation spectra:

(1) From the reorientation of water molecules themselves, which occurs in the MHz

- GHz frequency range in the liquid phase in ac measurements and at 110 - 150 K in the solid ice phase in TSDC measurements. The characteristics of this relaxation

(relaxation time, activation energy, entropy factor) reflect the influence of the sur­

rounding on the reorientating water molecules. In the case of subzero measurements,

in the solid ice phase, aspects" of freezing rome into play and have to be taken into

account in the interpretation of the results.

(2) From the in:fiuence of water on the relaxation and conductivity mechanisms ci. the matrix itself (main chain relaxation, side-chain relaxations, Maxwell-V/a.gael' and

space charge relaxations). It has been shown that this influence depends in a very sensitive way on the binding modes of water molecules (expressed e.g. in terms ci. tightly bound, loosely bound and free water molecules) and can be used to obtain

relevant information [6].

In the DMPC-water system the water content h was varied up to 0.50 w/w

(dry basis) corresponding to approximately N = 18 molecules of water per molecule

of DMPC. Water up to about he = 0.08, corresponding to Ne = 3.0 molecules cl. water per molecule of DMPC, was found to be molecularly distributed and tightly

(irrotationally) bound to the head groups: it does thus not contribute by reorientation

to the relaxation spectra and is unfreezable. These results agree quite well with

those obtained from equilibrium sorption isotherms. The rest of the water forms

clusters around the primary hydration sites and layers, and aystallizes to ice at low

temperatures. Its mobility is reduced compared to that ci. bulk (free water) with

the degree of reduction continuously decreasing with increasing h (loosely bound

water), whereas its polarizability does not cha.nge with h. There is no fraction c:l. (dielectrically) free water.

In the PHEA-water hydrogels the water content was varied up to about 0.5

w/w. In the frequency and temperature range of the dielectric relaxation of water

molecules, side-chain relaxations (, and 13,t» [3,4]) occur which are infiuenced by water. However, plots of the magnitude of these relaxations against h show cha.nges

of slope at about h = 0.30, suggesting a relaxation of loosely bound water molecules

(mobility restricted compared to that of bulk water) for h < 0.30. Support for this

interpretation comes from a detailed investigation of the shift of the glass transition

temperature Tg to lower temperatures with increasing h by DRS and DSC, which

becomes significantly less pronounced for h ~ 0.30. This result suggests that water up

to this value is molecularly distributed, whereas the excess above that forms clusters, ~·:"!_~"'r.eement with DSC an.d eonilibriUln lIDT'F)t.10T"' i~ot'hr!'i'ms rP.Sl'l1t,~_

- 13 -

In the butyl rubber-water system the results of SAXS, DSC and TSDC show

that water is dispersed in two sets of mesoscopic droplets with radii of the order of

magnitude of 2 run (SAXS). The relaxation of water rmlecules has been shifted to

higher frequencies compared to that of bulk water, contrary to the results described

above. This different behaviour is attributed to the hydrophobic environment of the

water droplets, resulting in a decrease of the mean number of hydrogen bonds of the

water molecules and, therefore, an increase of the rate of reorientation. The high sen­

sitivity of the TSDC measurements allows to show that there is ·no water molecularly

dispersed in agreement with the glass transition temperature of the rubber being not

shifted by water in the DSC and TSDC results.

Our results suggest that the reorientation mobility of water molecules is reduced

by a hydrophilic confinement and enhanced by a hydrophobic one, whereas the di­mensionality of the confinement has less, if any, significance. However, more systems

have to be investigated before generalizing these results.

REFERENCES

1. J. van 'Thrnhut, in: Topic3 in Applied PhY3ic3, Vol. 33: Elec tret", ed. G.M. Sessler (Springer, Berlin, 1980) p. 81.

2. P. Pissis, A. Enders and G. Nimtz, Chern. Phy~.(in press). 3. J.L. Gomez Ribelles, M. Monleon Pradac, A. KyritslS ana P. Pissis, to be pub-lished.· .

4. K. Pathmanathan and G.P. Johan, J. Polym. Sci., Polym. Phy"., Ed. 28, 675 (1990).

5. R. Perst~!J A. Kops, G. Nimtz, A. Enders, H. Kietzmann, P. Pissis, A. Kyritsis and D. Woermann, to be p:ublished.

6. P. Pissis, A. AnagnostopoUlou-Konsta, L. Apekis, D. Daoukaki- Diamanti and C. Christodoulides J. Non-Cry3talline Solid", 131-3, 1174 (1991).

Dr. Polycarpos Pissis was born on Cyprus in 1947. He received M.Sc. and Ph.D.

degrees in Physics from the University of Gottingen, Germany, in 1973 and 1977,

respectively. Since 1978 he is a Professor of Physics at the National Technical

University of Athens, Greece and a member of the Dielectrics Group of the Physics

Department. His research interest includes dielectric studies of materials, especially

hydrated biological materials and polymers.

- 15 -

Microwave dielectric study on bound water to biomaterials Satoru Mashimo and Nobuhiro Miura

Department of Physics, Tokai University Hiratsuka-shi, Kanagawa, Japan

Recently it has been shown by a microwave dielectric measurement that water has a local structure, often called as a cluster, of a cyclic hexagon or of a

. 1 2 pentamer.' If organic

compound like primary alcohol,l 2 acetone and p-dioxane is added

to water, plot of relaxation time against Xw shows a

critical change at an universal point Xw=0.83, where Xw is a

molar fraction of water. The point suggests existence of the hexagonal cluster. In the case

of water-p-dioxane,2 if Xw is

larger than 0.83, the normalized strength is increases linearly with xW.

The cluster of pure water which appear in this region are cycric and consist of six molecules. Compositional dependence of the normalized relaxation strength and that of the lorgarithmic relaxation time for water-p-dioxane mixture are shown in Fig.l.

80

70

50

1.5

en o 1.0

0.0 0.2 0.4 0.6 0.8 1.0

FlG.1. Gmphs orahe compositional dependcnce orthc norm"l~led rcla~a­lioll slrel1glh "mllhe logarillunic rcla~alion lime ror wu'c:r-p-dlo~"nc nll~· lures III 20 ·C.

Polysaccharide larger than maltotriose in aqueous solution

exhibits two relaxation peaks. 3 The high frequency relaxation one is due to the water molecules and the low frequency is due to overall rotation of the sugar molecules. However if glucose is added to water, it is observed that only a relaxation process is due

to cooperative orientation of glucose and water molecules. 3 It is shown that the hexagonal cluster can be replaced easily by a glucose molecule. Although the cluster can be replaced by the L-ascorbic acid molecule too, aqueous solution of L-ascorbic acid exibits two

relaxation peak. 3 It is suggested that water has a higher order structure and L-ascorbic acid brings a distorted structure of ice. Dielectric absorption and dispersion curves for glucose-water mixtures with various concentrations of glucose c(g/cc) at 25°C are shown in Fig.2. The absorption and dispersion curves for L-ascorbic acid in water solution with concentrations of 10,20, and 30wt% at 25°C are also shown 1n Fig.3.

In fact, it has been found that water has a higher order structure consisting of four to five local clusters. If organic

- 16 -

20wl%

60 '\ 80

" \J.) 40 \J.)

60

20 40

0

20 1.5

1.0 " \lJ

0')

~ 0.5

o

8 9 10 8 9 10

log f(Hz) log ((Hz)

FIG. 2. Dielectric absorption and dispcr<;ion curVl:S for glllco<;c-watcr FIG.3. Absorplion and dispcrsion curvcs for L·ascorbic acid ill walcr mixturcs with various concentrations of glucose c( g/cc) at 25 DC. Solulions with concentrations of 10, 20, and 30 wI. % al 25 DC.

compound with big hydrophobic part like tert-butanol and 2-propanol

is added to water, another change can be found at Xw=O.96. 4 This

point corresponds to the break point of the higher order structure. On the other hand, water has a further complicated structure in

biosystems. There exist a free water, nearly the same as pure water, and a bound water to biomolecules by hydrogen bonding. Relaxation time of the bound water is lOO-300times longer than that

of pure water.S

,6 As an example, dielectric absorption and

dispersion curves for dilute DNA aqueous solution are shown in 5 Fig.4. The normal B-type DNA has a lot of strongly bound water.

However if the water is replaced by some means, its structure changes to A-type or Z-type as is shown in Fig.S. Tropo-collagen and globule proteins also have the bound water. The relaxation strength of orientation of bound water depends on the water content

and vanishes for dried collagen. 6 This relaxation strength of

globule protein is proportion to the protein surface. 7

Structure of the bound water below oOe is quite interesting. In the case of DNA, the strongly bound water does not freeze, at least. until -100°C. and the weakly bound water freeze gradually as

temperature goes down and disappears at ~-60oe.8 On the other hand in the case of albumin. both the strongly and the weakly bound water freeze gradually from -20°C if temperature is

- 17 -

80

70 \.v

60

1.6

1.0 ~

w 0.4 en 0

- -Q2

-0.8 ...... -:; :,..- ........

. /' " ....... ./ ",,'" ......

/ ~ , , - 1.4

7 8 9 10 10 9 f( Hz)

Fig-li rc4. Die/eet ric dispersion ( (') nnd nl>sorpt ion (") curves for p()l.dele.de)· poly(d(;·dC) in \t,::l{cr with NaC/ (!i X III :~ !II Nne/, O.!i Jll~ J)NA/cm:') lit :WoC. Solid

cllrves ilrc c:l/cu/at cd from Eq. ( 1 ). Broken curve shows lhe hj~h·rrcqllency process and dolted curve shows the low· frequency process.

decreased. 9 Dielectric absorption and dispersion curves for albumin aqueous solution of 20wt% at -3S v C is shown in .Fig.S. The strongly bound water disappears at ~-130°C and the weakly bound water disappears at ~-120°C. It is suggested that the ice gives a damage to the protein surface, and the weakly bound water protects the protein from the ice of the free water. The fine structure of water depends on the kind of the biosystem. The dielectric measurement has revealed first the complicated structure.

A B z

Fig.S. The structure of A-, 13-. Z-DNi\.

T 1 I ,-10f-

tv 5>-:::::::::::~"'.". . - - - '. - - • - - - - - - - - - - - - • : ••• ~-:--:-=-::- =-::

Ol---l'---I------- ---1---·-/-

0- .................. ~, I'~"- -", ",,,.,. ,,~

" ," .~ 1,// >-:: · "",

, I ,. '., I -1 L-~ __ ~~~ __ ~ ____ ~ ____ ~~ __ ~~~

5 G 7 U Y IU log f(Hz)

Fig.G. Dielectric absorptIon and dispersion curves for alubumln aqueous solution of 20wt% at -35°C.

- 18

References

1. S. Mashimo, T. Crnehara, and H. Redlin, J. Chern. Phys. 95, 6257 (1991) .

2. s:- Mashimo, N. Miura, T. Umehara, S. Yagihara, and K. Higashi, J. Chern. Phys. 96, 6358(1992).

3. S. Mashimo, N. Miura, an~T. Urnehara, J. Chern. Phys. 97,6759(1992).

4. unpublished results 5. T. Umehara, S. Kuwabara, S. Mashimo, and S. Yagihara,

Biopolyrners 30, 649(1990). 6. N. Shinyashiki, N. Asaka, S. Mashlrno, S. Yagihara, and

N. Sasaki, Biopolymers 29, 1185(1990). 7-9. unpublished results --

Brief biographic data

Satoru Mashirno Nationality Japan Born 1946, 7, 23

1969 1971 1974 1974

1978'\19

1981 1984 1986 1991 1992

graduated at Waseda University (Tokyo), Physics B.Se graduated at Waseda University (Tokyo), Physics M.Se Dr.Sc (Ph.D), Physics, Waseda University Assistant professor of Tokai University (Tokyo) (Physics Department) Visiting Assistant professor of Brown University (U.S.A.) (Chemistry Department with Professor R. H. Cole) Associate professor of Tokai University, Physics Department Professor of Tokai University, Physics Department Visiting Professor of Brown University, Chemistry Department Vice-chairman of Department of Physics, Tokai University Chairman of Department of Physics, Tokai University

- 19 -

MEASUREMENT OF BOUND AND FREE WATER IN MIXTURES

Alexander BRANDELIK and Gerd KRAFFT Institut flir Meteorologie and Klimaforschung, Technologie­

Transfer Kernforschungszentrum Karlsruhe GmbH/ Universitat Karlsruhe

Postfach 3640 D - W - 75 Karlsruhe

Germany

Telefon:+49-7247-823913 Fax:+49-7247-825070

ABSTRACT

We report on the development of a new moisture measurement method. Our work was concentrated for soil moisture deter­mination, but also preliminary data are available on other pharma granulates. The new moisture model consists of blocks: recommendation by Birchak, semi empiric soil model by Dobson and Halikainen, our new definition of bound water and our new measurement method using the in-situ frost calibration. In the case of soil also swelling and shrinking are taken in account. The method do not require any sampling.·"for the purpose of calibration. The dielectric constants of the wet mixture will be measured at two different water cor tents in the natural and in the perfectly frozen stgtes. These four informations provide four unknown variables, namely the initial water content, the water content at the changed tage, the bound water and a new solid specific parameter. The deviations between measurements due to this model and the accurate expectations are less than 1% for as extreme soils as sand and clay. Investigationis started to link these soil moisture data on qround to the satellite data.

KEYWORDS

Moisture measurements, soil moisture, dielectric moisture probe, self-calibrating sensor, soil model, mixing rule and bound water.

- 21 -

DIELECTRIC MIXTURE THEORIES IN PERMITTIVITY PREDICTION: EFFECTS OF WATER ON MACROSCOPIC PARAMETERS

Ari H. Sihvola Helsinki University of Technology, Electromagnetics Laboratory

Otakaari 5 A, SF-02150 Espoo, Finland

This presentation tries to give an overview about dielectric mixing theories, more precisely about explicit rules that predict the macroscopic dielectric constant of a given heterogeneous material, focusing especially on mixtures containing water. Parameters in the mixing formula are the fractional vol­umes of the constituents and also the pel'mittivities of these components (multiphase mixtures consist of more than two components). What dis­tinguishes one mixing rule from another is the manner how it takes into account the microstructure of the mixture: the shapes of the various inclu­sions. This point is especially crucial when dealing with moist substances where water plays a central role.

As is well known, several mixing rules have been proposed in the literature for calculating the effective permittivity feff of substances encountered in industrial, geophysical, and materials science applications. From the ana­lytical electromagnetic point of view, the problem of solving the random medium problem is not easy due to the enormous amount of degrees of freedom in describing the boundary value problem. Therefore also differ­ent mixing rules can coexist and possess different areas of application with experimental confirmation.

The theoretical approach to the mixture problem normally starts with a sin­gle scatterer in a background medium. This scatterer, or inclusion, is most easily analysed if its shape is assumed to be ellipsoidal, or even spherical, due to the fact that the Laplace equation allows a constant field inside a shape like this. Therefore also the polarizability of the inclusion is expressed rather easily with the parameters of the problem. Polarizability is a crucial parameter in electromagnetic scattering analysis and other mixture-related treatments.

The oldest mixing rule for a two-component mixture with small spherical scatterers in a homogeneous background medium is the so called Maxwell­Garnett (MG) formula. Although this formula has been applied sometimes successfully in geophysical modeling, it, however, has been blamed for not being able to predict the properties of dense mixtures, nor high-dielectric­contrast composites.

But MG, although old, has not remained the only one: in the following, MG will be considered as a member of a larger family of mixing rules as presented in (Sihvola, IEEE Transactions Geosci. Remote Sens., Vol. 27, No.4, p. ~.()~_!l ~.~ 1a~O\ ! fUTf'\_nh~c.-,... "' ... ;v+~.-'"' ·,·"h4'>""'" 4>1._-"' 1...~.~>1..-<,.,,~;f ,......"...-t~ •• __ l. ....

- 22 -

permittivity €a and spherical inclusions of permittivity € occupy a volume fraction /, has effective permittivity €efl' obeying the following expression which could be termed as the generalized mixing rule:

€eff - €a € - €o ----------------- =/---------------€efl' + 2€o + v( €efl' - €a) € + 2€o + v( €efl' - eo)

Here, v is a dimensionless parameter that selects the member of the mixture­rule family. II = 0 gives Maxwell-Garnett form~a., The "coherent poten­tial" formula is the result of the choice II = 3. "Effective medium theory" emanates through II = 2. Also v = 1 gives a prospective mixing rule.

Layered inclusions can also be analyzed, and consequently mixtures con­taining inclusions that are <='rmed of multilayer component phases. This model is sometimes applicable to moist substances like, for example, wood, which has been the application of the works of Tinga et alii (Journal 0/ Applied Physics, Vol. 44, No.9, p. 3897-3902, 1973).

The generalized mixing formula above can be further generalized to treat also ellipsoidally shaped inclusions. Due to the increased order in the de­scription in the mixture, also the mixing rule depends on more parameters than the above one. For example, the case of a mixture with randomly ori­ented inclusions (which is also then isotropic, and the effective permittivity is a scalar) obeys the following rule:

/ ~ €a + Ni(€eff - €a) €eft' = fo + -3 ( f - Eo) L.J + N.-( _ )

i=l €a ,f €o

where N}, N 2 , N3 are the depolarization factors of the ellipsoids in the three orthogonal axis directions. €a has been termed "apparent permittivity" which takes into account the interaction between adjacent scatterers in a similar way as II in the earlier formula. Note the nonlinearity in feB of the formulae. For discussion, see (Sihvola and Kong, IEEE 7'ransactions Geosci. Remote Sens., Vol. 26, No.4, p. 420-429, 1988).

There are also other, more experiment ally-based mixing rules for the pre­diction. From these, let the following family of exponential mixing formulas be mentioned

where b is the exponent parameter. These formulae sum over all constitu­tents of the mixture (index n in the sum), and the symmetry with respect to all components is evident of this form. For more details, see (Sihvola and Alanen, IEEE Transactions Geosci. Remote Sens., Vol. 29, No.4, p. 679-687, 1990).

- 23 -

water has permittivity that is very high compared to air, and also compared to other constitutents of moist substances. This fact, the high dielectric contrast is a factor that distinguishes the predictions of various mixing rules from one another. Water is a problematic constituent in dielectric mixtures in other respects also. In fact, the inaccuracies in the predictions of mixing rules for wet materials in practice may not - in most cases - be a reflection of the inability of the mixing theory to describe nature, but rather the fact that the amount and the volume fraction of water phase in the mixture is not easily determined, nor is the shape of the microstructure of water inclusions. Also in the dielectric .properties of water themselves, the distinction of free and bound water has to be made, whiCh has a major effect on the mixing theories predictions, due, again, to the large dielectric contrast between these two types of water in substances.

One phenomenon that the analyses of water mixtures have been helping in shedding light to is so called percolation. Percolation could be defined to be a sudden change in some macroscopic parameter of a random medium as the volume fraction distribution exceeds a certain threshold value. In mixtures, this parameter most naturally is the permittivity, or in some other cases, the conductivity. Quasistatic mixing rules have been many times criticized of not jeing able to explain, or to predict percolation. However, the latest results (Sihvola, Helsinki University of Technology, Electromagnetics Labo­ratory Report Series, No. 124, August 1992) have proved that percolation is indeed inherently present in the mixing rules although in their derivation, the focus is not in that direction in the first place. Percolation comes visible in high-contrast mixtures; hence water is extremely interesting constituent in mixing theory analyses from this point of view also.

Dr. Ari Sihvola was born in Finland in 1957. He received the degrees of Diploma Engineer in 1981, Licentiate of Technology in 1984 and Doctor of Technology in 1987, all in electrical engineering, from the Helsinki University of Technology. He was a Visiting Engineer at the Massachusetts Institute of Technology and in 1990 worked as a Visiting Scientist at Pennsylvania State University, State College. He is Docent at the Electromagnetics Laboratory of the Helsinki University of Technology with interest in electromagnetic theory, remote sensing and radar applications.

- 25 -

EFFECTNE DIELECTRlC CONSTANT OF 1fOIST SUBSTANCE

Sumio Kobayashi

R&D Center, Sumitomo ~letal Industries, Ltd.'

1-8 Fuso-cho, Amagasaki, 660 Japan

Abstract. Moisture me.asurements in various substances by using microwave methods are based on

the fad that the effective die1ectric constant of the substance depends on the moisture. Typical models

for describing the relation between the effective dielectric constant and the moisture are the volumetric

mixing model and the index of refraction mixing mode1. The two models are compared with the

experimental results of microwave attenuation through moist limestone layer. Furthermore, the models

are discussed on the basis of numerical simulation.

Two simple models have been proposed for describing the effective dielectric constants of mixed

materials; they are the volumetric mixing model (VMM) and the index of refraction mixing model

(IRMM). In V!\fM, the effective dielectric constant of moist granular materialle is described by

(0)

~. 2.85CM L· 10.0 eM - «:) I •

CD 0 -Z

30 0 i= ~ ;:) Z ~. 7.5CM w

L- '.3.5~ t- 10 t-oct·

o 5 10 MO'STU~E (WET 9ASE).(%.)'-

• 0 MEASURED CALCULATED.

-II) o. z 0 t-~ :> z 1.&.1 t-... <

( b)

20 M • 0.05

• L • ,35cm

• •

10 •

o 10 20 30 40 50

TEM?£RATURE OF LIMESTONE

• MEAS t.,.: REO CALCULATED

Fig. 1. Comparison of the theoretical prediction based on IRMM with the experimental results. (a) DependE'nce of the microwave attenuation on the moisture and the operating frequency at 290 K, (b) dependence of attenuation on the temperature.

- 26 -

(1)

where f i (i = m, w, a) is the volume fraction of i-th constituent, (i (i = m, w, a) is the dielectric

constant of i-th constituent, and the subscript 711, wand a denote material, water and air, respectively.

On the other hand, the effective dielectric constant in IRMM is described by

(2)

The two models were compared with microwave attenuation measurement in moist limestone

(CaC03) layer. This material is scarcely soluble to water and has a low tan6. The VMM is found ot be

unsatisfactory for describing the experimental results, while the IRMM can predict the experimental

results: the frequency dependence and the temperature dependence of microwave attenuation, as shown

in Fig. 1.

The models are discussed on the basis of numerical simulation of microwave transmission in

mixed materials.

Dr. Sumio Kobayashi was born in Japan in 1944. He received B.S. degree in Electrical Engineering

In 1966, and Ph.D degree in applied physics in 1985 from the University of Tokyo. Since 1966 he is

with the Research Laboratories of Sumitomo Metal Industries, Ltd., where presently he is the Manager

of Ceramic Research Department. His professional interests include material processing and microwave

instrumentation for steel making industry.

- 27 -

Hydrocarboll alld "Tater Estilllatioll in Reser~Toirs USillg Microwa've Methods

Tartk J/. Habashy, Pnbifra Iv. Stn, and "I. Re:a Taherian

Schlumberger-Doll Research, Old Quarry Road, Ridgefield, CT 06877-4108

Abstract

The pore space of oilfield resen'oir rocks is filled with water and hydrocarbon, ,\Ve discuss

the inverse problem of estimating hydrocarbon content usi ng mlcrowave measurements of the

permittivity and conductivity. \\'e first re\'iew some borehole measurement tools and then

il1ustrate the mixing laws as well as the laboratory measurements which are used to interpret

the data. The currently a\'ailable commercial tools employ frequency ranging from 2 ~IHz

to 1.1 GHz. \\'e highlight features of a tool which operates at 1.1 GHz and it.s new antenna

configuration.

The modeling of the electromagnetic wave propagation in a borehole geometry, which is re­

quired to est imate permittivity and conductivity from tool measurements, is described. Next,

we discuss t he application of composite media models to extract the volume fraction of hydro­

carbon from the estimated permitth'ity and conductivity. Complexity of the rock geometry and

t he presence of mobile count.er-ions in clays render the petrophysical modeling a challenging

t ask. The laboratory data taken over a wide frequency range is used to test the \"alidity and

the limita t ions of the models.

I. Introduction

The purpose of this paper is to highlight some of the current methods of estimating hydro­

carbon and water in formations using microwave methods. ~licrowave methods measure the

complex permittivity of the formation, ((i..i:) = {'(w) - ~(w)lil..l,'{o. Here w is the angulnr fre­

quency, c(I..l,·) and ~(w) are the real part of t he permittivity and the conductivity, respecti\'dy.

(0 is the permitth'ity of "acuum. The rea] and imaginary parts of the complex pernutti"ity are

related by the well known Kramers-Kronig relation.

- 28 -

The electrical conductivity has always been the most important logging measurement in

estimating hydrocarbon content of reservoir. A common practice in borehole logging is to

estimate the water saturation Sw, which is the fraction of pores occupied by water, [roln the

measured electrical conouctiyity (J' of the rock and the conductivity of the saturating 'water

O'w. In practice, however, this method often fails because either the water conductivity is

unknown, or it varies rapidly with depth, or it is too snlall, or the interpretation scheme'S hased

on conductivity fail. It is highly desirable to have a method of estimating hydrocarbon which

does not depend on water conductivity. The dielectric measurement is such a possible a\"(>nue

(Gilmore et aI, 1987).

The permittivity of water is about 80 in the GHz range while that of rock matrix is bet ween

.5-9 and of hydrocarbon is about 2. In addition to this large contrast, a weak salinity oe'pen­

dence of t~ had been motivating reasons for using t' to estimate Sw' However, the complex

permittivity of water saturated rocks show a complicated frequency-dependent beha,·ior. In

the low frequency range (kHz and below), the permittivity can be as large as 108 due to the

presence of clays. In the mid megahertz range there is a strong influence of grain texture.

Both of these effects become smaller at higher frequencies in the GHz range. \Ve will discuss a

microwave tool which can measure the permittivity and which operates at 1.1 GHz.

II. The Electromagnetic Propagation Tool

The existing Electromagnetic Propagation Tool, EPT (mark of Schlumberger) is a linear

array consisting of two-transrrutter and two-receiver microwave slot antennas (Safinya et aI,

1987). Each ,lntenna is a cavity-backed slot antenna. The transmitters are excited by coaxial

lines with their inner conductors shorted to one end of the cavity. The receivers measure the

formation signal by the same set-up where the voltage or current is measured at the input of t he coax. These antennas are mounted on a metallic pad which is pushed against the borehole

wall. Each transmitter is turned on separately, while phase and amplitude measurements are

made at each receiver. In this way the phase shift and attenuation between the two receivers

is recorded for signals traveling in opposite directions. Averaging the two readings serves to

elinlinate imbalances in the two receivers, helps to correct for imperfections on the borehole

wall, and symmetrizes the tool response to thin beds. This mode of operation is known as the

bor~hole compensated mode.

Since these slot antennas have small sizes relative to the wavelength, they effectively behave

as point magnetic dipoles with moments perpendicular to the axis of the coaxial line. There are

two separate versions of the EPT tool which differ in the orientations of the magnetic dipoles

on t he tool pad. In one version, the antennas are mounted so that the dipole moments point

eno on to each other. This is called the endfire magnetic dipole array. In t he other version, the

antf'nnas are mounted so t hat the dipole moments are broadside to each other. This is referred " ,,\ 1\ \

- 29 -

B('cause of the high frequency of operation, these antennas are shallow rtnd have a d('p1 h of

in"C'stigation of about 6 inches. Thus, the EPT is only sensitive to the yut of the format.ion

which has been invaded by the borehole fluid.

From the phase shift and amplitude ratio nleasurements, one C2.n deri \'e an effect jye per­

mittiyity and conductiyity. These represent an averaged permittivity and conductiyity which

not only take into consideration the effect of the in"aded zone but also any standoff layer that

exist between the tool pad and the borehole wall. This standoff layer can be a mudcake layer

in front of a permeable sand zone or a layer of borehole mud in front of an impervious shale

zone.

Performing an endfire only or a broacside only measurement is not enough to pro"ide the

electrical parameters of the in\'aded zone. This is true, since anyone set of these measurements

prodde only two independent measurements which are not sufficient for the inversion of the

two electrical paramet.ers of the invaded zone and those of the standoff layer plus its thickness.

To overcome this problem, we ha,-e introduced a cross-dipole antenna t hat allows the simul­

taneous endfire and broadside measurements (see Figure 1) thus providing four independent

equations which can be combined with an independent measurement of the standoff thickness

(from an ultrasonic device that employs a pulse-echo technique) to invert for the four unknown

eject rical parameters (Habashy et aI, 1992 and 1993).

CI)

.: CI)

.... "'D "'DO c:i w

::,00< .••.. · .•.. ·.· .•. · •.•..• : .. : .•• ::.:·:.· •..• : •. · .•.• : ••.•. ·.:.· .•. ··.:.:.:;;::ffii; EE ..• :: ..•. :.::i:!.::i;:.i.! .•..• ::.i.i ... i.:.! ... ::i::: •. ·: .•.•. · ..•. • .. ·.i .. ·:· •.• ::.:.:.: .. ·:·.·.·.::' ..•.• : •. :.;::.: .• ::.~.! ... !i::!.:.:: •. ~ .•.• :· ... : .•...•. :.: .. ~.:.:.:.83:: ....•.•.. : .. ::.::·.: .• ·: .•.•. : •..•• :.··.;~~:~ .::;::::::;:::::::::::: :tr:::~~:::{\:~:~~~: :::: ::~:{\?~~:··~:;:~::)~::~~~:))tf\~~}~t~;.::t{·· :.:::~ ::t:~:i:.:: ... ::.::.:.:::::.:::.:.:;: ....

Figure 1. Schematic of the cross-dipole array.

Figure 2 shows the inyerted permittivity and conductivity by the endfire mode only, the

broadside only, and the cross-dipole mode as a function of the standoff thickness. It is clear that

the endfire only or the broadside only provides the formation permittivity and conducthojty with

gross errors beyond 0.2 inch standoff, whereast .cross-dipole provides an accurate est imate _ J ('\.... 1

- 30 -

50

a endfire

45 is broadside • inversion

>- 40 .... . -> ....

35 .... E '-Q)

30 a.

25

20 0.0 0.1 0.2 0.3 0.4 0.5

standoff (in. )

3.0

a endfire 2.5 is broadside

• inversion .......... E 2.0 .........

C/) .........

~ .... > 1.5 .... 0 ::J "'C c: 1.0 0 0

0.5 0.0 0.1 0.2 0.3 0.4 0.5

standoff (in. )

Figure 2. Inverted permittivity and conductivity.

-31 -

III. Interpretation

In interpreting the dielectric 1ogs, the water saturation is determined by a naiye formula

which is based on the assertion that the "comp1ex t ra,"el time" through a forma tion is an

a\"erage of the corresponding transit times through the components

where !h and tm are the perrruttiyities of hydrocarbon an.d t~e solid matrix which are generally

real. !w(w) = £w(w) - (j'U..( .• :)/i~·!o and (jw(""") includes any Joss due to dipolar rotation. The

above equation often works, however, it would imply that. the permittivity should be the same

if all quantities on. the RHS were the same. But experiments show that even when all the

quantities on the RHS are the same, the permitti,"ity can vary from rock to rock due to textural

differences. \Ve will not discuss the electrochemical effects in this paper, but general1y these

effects are small in the 1.1 GHz range.

~ficroscopic examination of rocks suggest that rocks with dispersion contain platey grains

with one small dimension, and two large din1f>nsions. These platey grains can act as thin

capacitors and thereby increase the permittivity of the rock. \Ve model the dielectric properties

of brine sat.urated rocks by assuming a bimodal distribution of grain shapes. A rock of porosity

¢ is assumed to have two types of grains; platey grains of ,"olume fraction p(! - ¢) and aspect

ratio 8, secondly, spherical grains of volume fraction (1 - p)(! - ¢). This model is found to

exp1ain the frequency dependent permittivity! and conductivity (j of several clay-free rocks

mf>asured in the laboratory o'-er a wide range of frequency and water conducth-ity. \Ve find

that 8 and p obtained from applying the model to t he experimental results at a single frequency

can reproduce experimentally measured values of ! and (! in the frequency range from 0.5 ~fHz

to 1.3 GHz (Baker et aI, 1985).

The volume fraction of solid IS built up iterath-ely. It involves computing the effective

medium properties of a mixture of insulating grains (of an infinitesimal amount) in a conducting

host; the resulting mediunl is used as a host, to which nlore grains are added. The resulting expression is integrated to obtain the fluid/solid v01ume fraction of interest. Hydrocarbon is

included in an analogous manner (Feng and Sen, 1985).

In principle, it is straightforward to generalize the iterated dilute limit technique to allow for

a spectrum of grain fractions with different aspect ratios. However, trying to solve the inverse

problem of finding t.he aspect ratio spectrum from dielectric measurements over a wide range

of frequency is difficult. This is because the problem is inherently ill-posed and requires high

accuracy in the experimental measurements and water properties. Also, grain based model

may not be so useful for predicting other physical properties of rocks such as fluid perrriea bility

which depend strongly on the pore geometry and !'ize rather than grain geometry. ~fodels for

dielectric dispersion based on pore geometry wil1 be of practicrlJ importa.nce.

- 32 -

References

Anderson, B., Safinya, K.A., Habashy, T.!\1., Da\'idson, A., and Gilmore, R.: "The Response

of t he Electromagnetic Propagation Tool to Bed Boundaries," SPE Journal on Formation Eval­

uation, pp. 458-464, December 1990.

Baker, P.L., Kenyon, \V.E., and Kester, J.!\f.: "EPT Interpretation esing a Textural

~fodel," Paper DD, SP\VLA 26th Logging Symposium, 1985.

Habashy, T.M., Safinya, K.A., and Beren, J.A.: "Apparatus and !\fethod of rsin~ Slot

Antenna Having Two Nonparallel E1ements,'" Fi1ed as a US patent and in foreign countries,

January, 1993, application serial number: 000,916.

Haba.shy, T.M., Safinya, K.A., and Beren, J.A.: "Electromagnetic Logging ~fethod and

Apparatus," Filed as a US patent, ~farch 9, 1992, application serial number: 848,576.

Habashy, T.M., Taherian, !\1.R., Dumont., A., and Beren, J.A.: "Electromagnetic Logging

Apparatus and ~fethod," Filed as a rs patent, :\farch 9, 1992, application serial number:

848,621.

Feng, S. and Sen, P.N.: "Geometrical :\lodel of Conductive and Dielectric Properties of

Partially Saturated Rocks," J. Appl. Phys. 58 (8), October 1985.

Gilmore, R., Clark, B., and Best, D.: "Enhanced Saturation Deterrrunation Using the EPT­

G Endfire Antenna Array," Transactions SP\VLA 28th Annual Logging Symposium, June 1987.

Safinya, K.A., Clark, B., Habashy, T.~f., Randall, C., and Perez-Falcon, A.: "Experimental

and Theoretical Study of the Electromagnetic Propagation Tool in Layered and Homogeneous

:\fedia," SPE Journal on Formation Evaluation, pp. 289-302, September 1987.

Reza Taherian received the B.Sc. degree in Chemistry from University of Tehran in 1971, the M.Sc. degree in Physical Chemistry from Tehran University of Technology in 1975, and the Ph.D. degree in Physical Chemistry from University of California Davis in 1981. He was a visiting scientist at SRI International from 1983-1985. From 1985 he has been a research scientist at Schlumberger-Doll Research. His current research interest includes Die1ectric Logging, Spontaneous Potential, Resistivity Logging, and e1ectrical properties of water saturated rocks. He is a member of IEEE, AGU, and SEG.

- 33 -

~~ INTEGRATED 1v1ICROWA VE MOISTURE SENSOR

Introduction

G.Biffi Gentili, G.F .. -\\itabile, F. Ceccuti and G.F. ~fanes Laboratorio di ~ficroelet1.ronica

Dipartimento di Ingegneria EJettronica Universiu' di Firenze

\;a S.W~3

50139 Firenze - ITALY

Local monitoring and moisture content control in surface layers is of great importance to the disciplines of agriculture, industrial applications and environment sensmg.

Mcro\vave dielectrometry represents a very promising method for moisture content estimation of porous and matrix systems.

Techniques based on the measurement of the complex reflection coefficient of an open-ended transmission line, perturbed by the material under test are well established [1) 2]. A coaxial line is mostly suited as the transmission line, while reflection measutement is usually perfonned on a network analyzer.

The same operational principle can be applied to moisture content measurements by properly modeling the complex dielectric pennittivity of substances as a function of \vater content [3]. A major drawback of this approach is represented by the length of the line connecting the sensing element to the network analyzer, which should be kept as short as possible in tenns of \\'avelengths to reduce parametric sensitivity. Furthermore, the need of the nehvork analyzer itself makes the test-set unsuitable for on-field applications.

An innovative moisture sensor based on the pre\-iously described principle is introduced and discussed in this paper~ the sensor exhibits the unique capability of simultaneously performing the required moisture measurement and of transmitting the associated infonnation.

The sensor is configured as an active patch antenna, where the sensing transmission line, being a part of a microwave oscillator feedback loop, directly transforms the reflection coefficient variation in a frequency off-set. The resulting \vavefonn, radiated by the patch can be then remotely read- out.

A simple, low-cost moisture sensor in thus realized, whose attractive features include operational flexibility, absence of artifacts and multipoint operation capability.

Active Integrated Antenna hasic operf!tion

The Active Integrated Antenna (AIA), already introduced [4], is a multilayer structure, based on a patch antenna coupled through a non-resonant slot to a pair of

- 34 -

feeding microstrips connected in a feedback loop to a transistor amplifier. With reference to Fig. 1, the patch acts as a radiating element disjoined by the

feed/oscillator nern'ork and as a resonant load for the feedback loop. The rrucrostrip pairs coupling the amplifier to the patch through the slot are tenninated in open circuit stubs, at the other side. The operating frequency is detennined both by the patch geometry and by the loop electrical length, \\'hile the open quarter-wave stubs should ideally represent a short circuit at the hvo disconnected ports.

The double feeding structure is attractive for a moisture sensor realization; any perturbation at the open ports, in fact, results in a corresponding variation of the feedback electricallength and, thus, in a transmitted frequency off- set.

The configuration is planar and inherently low-cost, being implemented on plastic substrates and a11o\\1ng standard bipolar transistor to be used. Operation up to X band can be easily achieved. .

Sensor operation principle

For operation as a moisture sensor, one of the open-circuit stubs of the original AIA configuration should be replaced by a sensing transmission-line element. Either a coaxial stub interfaced to the moistened medium through an open-end tennination or a planar transmission line (microstrip or coplanar \\raveguide) could be used for that purpose. In the latter case interaction \\1th the medium is performed by means of a thin dielectric superstrate, resulting in a local perturbation of the open structure fringing field.

According to Fig. 1 the two layer structure includes in the first layer the patch antenna electromagnetically coupled to the oscillator circuit via the non-resonant ~Iot. The second layer accommodates the oscillator circuit itself, along with the sensing element.

An equivalent lwnped-element circuit 9f the proposed configuration is presented in Fig. 2; the parallel resonant circuit represents the patch antenna coupled to the oscillator circuit through the slot. The sensing stub is modeled by a variable reflectance connected at the amplifier input port.

Sensor operation can be affected by medium and long term parametric variations, particularly temperature, resulting in frequency off-sets, which should be compensated fOf. For that purpose, the module can be arranged so that the amplifier input port is electronically switched alternatively betvveen the sensing element and a reference load. The resulting FSK \,.,aveform allows the measurement to be perfonned as difference between reference and off-set frequency.

Multipoint measurement can be also attained by frequency or time mUltiplexing of a number of individual battery po\vered sensors.

- 35 -

magnetic coupling ¢::>

dielectric subsuates

. open stUb

- 36 -

Sensor operation demonstration

Evidences of the above described operation principle has been given using both a circuit model and an AIA experimental set-up.

The sensor configuration is represented in Fig. 3, \\'here the sensing element appears as an open-ended coaxial line, to be interfaced to the material under rneasUrement~ the patch antenna and the slot are simply modeled by a parallel resonant circuit, coupled through an ideal transfonner to ports I and 2 and loading the micro\\rave amplifier, according to the schematic of Fig. 2.

The model \vas implemented on the Hewlett-Packard Micro\\rave Design Simulator (tvIDS) and the reference oscillation frequency, \\ith the coaxial stub replaced by a microstrip, was computed.

Using a 30x40 rnm patch on a CuClad substrate ( .79 mrn. tick, Er=2.54) a 2.194 GHz oscillation frequency resulted.

The configuration \\'as perturbed introducing in the simulation the reflection coefficient variation induced by interfacing the open-ended cable to a 0.6 porosity sand \\ith variable moisture content (We).

Theoretical data of reflection coefficient ,"'ere previously calculated according to the numerical analysis reported in [3], and used for the simulation, along with electrical and geometrical parameters of the coaxial stub.

Perturbation of the feedback-loop \\'as then evaluated and represented in Fig. 4, as a frequency deviation (LlF in :MHz) from the reference frequency and as a function of the moisture content, We.

An experimental evidence of the sensing capability of the proposed configuration is also given using the active antenna reported in [4].

The quarter-\\Tave microstrip stub connected to the amplifier input port is used as a sensing element interfaced to different material via a 0.8 nun thick polyethylene superstrate.

The reference frequency, i.e. \\ith no interfaced material, was 2.336 GHz. Experimental results are sho\\n in table 1 for different materials.

1\'f~'\ TERIAL U!\1)ER TEST

superstrate Sand Sand Distilled

(polyethylene) 0.6 porosity 0.6 porosity Water

(Wc=O) (Wc=O.6)

IMI 9 17 24 31

(MHz)

Table 1

--N

::I: ::E r........I

~ 0.)

rl: 0

~ c:: 0.)

=' c::r 0.) "-4

~

- 37 -

open-circuit stub

~ ~II-.-.-.-.-.~.p-.~~.~h.:a.-.-. I

I

t ------- ______ '

AlA

25----~----------~--------------.-----

20

Sand 15 (porosity-O.6)

10

5~--~----------~--~----~----~--~ 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Volumetric "rater Content [ Wc ]

- 38 -

Conclusion

A simple, low-cost moisture sensor based on an Active Integrated Antenna \\'here the sensing element is part of the feedback loop of a micro\\'ave oscillator, ha~ been discussed and demonstrated in this paper.

The moisture content can be directly eva1uated through the oscillator frequency off-set, thus avoiding the need of costly and cumbersome network ana1yzers to be used in the measurement test-set.

Other uruque features of the innovative configuration are remote read-out capability, multipoint operation and absence of artifacts. Invasive or non-invasive

operation can be easily obtained by sLmply changing the sensing element structure. S-band operation has been evidenced using a simple computer model; a

relationship between frequency off-set induced by the \\'ater content variation of a porous material has been derived, in qualitative agreement v.ith theoretical calcu1ations.

Sensitivity of the integrated sensor to different substances has been also demonstrated by using an S-Band AIA prototype.

Future \vork \\111 be devoted to the optimization of the micro\vave circuit and to an accurate characterization of the sensor trough laboratory and in-field measurements.

Reference

1- E. Tanabe and W.J. Joines: " A non-destructive method for measuring the complex pennettivity of dielectric materials at micro\vave frequencies using an open transmission line resonator". lEEE Trans. on Instr. and Meas., Vol IM-25, No.3. Sept. 1976.

2- J.R. Mosing, J.e.E. Bessons, M.Gex-Fabry and F.E. Gardiol: "Reflection of an open-ended coaxial line and application to nondestructive measurement of materials". IEEE Trans. on Instr. and Meas., Vol IM:-30, No.1. May 1981.

3-G.Biffi Gentili and F.Gori: "Open end coaxial probe for soil mixture measurement", International Microwave S)mposium, July 1987, Rio de Janeiro, Brasil.

4-G.F. Avitabile, S.Maci, G.Biffi Gentili, L.Roselli,G.F. Manes: "A two-port active coup1ed microstrip antenna". Electron. Letters. Vol.28, No.25, Dec.l992.

- 39 -

Integrated ~ficrowa\'f ~1oisture Sensors for Automatic Process Control

Ferenc Vo)gyi

Technic:l1 University of Budapest, Department of Microw;lve Telecommunication, H·) 1) I Budz.tpest. Goldmann ler 3, HUNGAR Y

ABSTRACT

The basic defin1 tions' and -;;Ia tions for calcQla­

tion of the main c~~racter1stics ~ the transmission,

ty~ micrOWAve moisture mea.Qrtment using two anten­

nas in the radiating near - field are 11ltr~Qced.

Making known the preliminary con.ideration. for the

desi;n,of integrated micro\o/ave moisture aensora, A

block 'l1Agra.m 11 diacuuaj. A lot of 1I1 .... ur.lII8ntJI

(tranam. and r.tQrn 10 •• , temperature effect.) _re

carri.d out, including the compensation of moisture . I

content at ~~llow dent field corn.

~7RODUCTlON 1 Initiation. of private enterpris.s are ""'portant •. 1

ir. econamic aitQation of H~9ary. 'rheae -f1rm. sh::lu14

:r.&l\ufacture-, or the farmera .houle! plant chea~r ~

mark. table go~s an~ crops, usingthe ben~fit' which

are given by m.:>dern technology. One of the mo.t 1m­port&nt technologie&l process i. drying corn up to

tile moisture cont.nt which is specified tor storage. F:a t.he al.ltomatic 0r"ration ct4ryin .. machines moiS­ture s.nsors are needed.

Our ver5Atile microwave moisture aenaors[ll were

Jroade by the uuge of ,microwave hybrid integrated

circuits (¥~lCJ on plastic aQbstrataa [3]. Nowadaya

we develored copper thickfilm MAIe-s on alumina ce­

razr.1c which ar. ch.a~r and have bett.r paramet4ra than goldfilm circuit., and these were integrated

with microatrip ant.nnas (active - array) CQlUyosing

the integrated micro~ave moisture .ensora.

Using the ·cla.sification of microwave aensor.

being used for moisture cont.nt meas~r&ment· in[2],

our instrwDent i.. transmission ty"e, a~rio:Uc,

open, built u~ with two antenna.. For .Qch an eq~­ipment -which La utiliz.d for agricultural ano! in­

dustrial automatic proc.s. control-high reliability

.tability, simplicity, awall-size and low co.t (i£

pos.ibl.) are the mAjor requirement ••

A lossy d1eleetr1c material with thickneu of t(em] attenuate. the pas.ing TEM WAVWI

1. -at- 0.91 WtgO.f.t [dB] /l/ where 4liI.-is t.he atten~Uon fa:tar in dB/cml f-h the

freql.lency in GHz, f.' 1a the reAl I'>o.&rt of the r- CO~la!

relative permittlv1ty.

t-f'-jEII. The po .... r diadpa tion w incUc. te~ b)' the ~ou fac­

tor, which 1& proportional to tn. tangent of the I lo.a angl ••

15/ The field atrenlJt.'l of the JlLic:roWAve field penetra­

ting into the louy J:>Ater1al decreaa .. exponentially.

':'he a~1n-je?th d.finej as be distance fr~ the sur­

face where the power density is reduced by a factor of 1/. (e- 2.7142) •

d - )./~fiWt'lt /61 The e~araettr1at1r 1r~f,apee of the ~lane wavo in a los.,y sl)aco.

%O·l77 [1-3 (trJf / hj (tS'O 12}Va, /7/ .... he l.!O~lcr ;HI$1;.;a" 11'1 volU!IIC V of bo=geneO\.\. ma-

"

'~":" ',\f.:::!:I :":: ..

, , ..

The objective. of this paper are 11 to lntr04~ca

the basic definition. and principles of the moiature ~ meaaurelil8nt using two antennaa, 2/to review our In- Aftar calc~latin9 a ie;-Cb&rac~'val~ •• refei-:'''

tegrat.d microwave moisture senaor family,l/to ahow I 11'1" to water 1D th. 0.4 -10 GSa frequ.ncy rang., 1t

same useful meaaur.ment results conn.cting to our

(near-field-, circuit atabi1ity and quality, ~~­

rature effect) examination ••

BASIC DEF1!IJITlONS AND PRINCIPLES

Auw::l.1ng a plane\of&ve prc?agating in toule .. fr.e­

-spac., the .. ttenuation betwcan a translf..it antenna

with ga1n Gt

aoJ receive~tenna with gain Gr placed

at a distance o( n a

At- 20.10g .1('01.1), -Gt -Gr [oiB 1 where ~ is the fr~e-s?ace wavel.ngth.

I /1/ I

! '), _ elf' .Hcr.ll ':f 30/ t r~n&] /2/ i

can bet .tated that. lIthe power cUaa1pat'" in water rap141y increa ... wlth frcq~ency, aA4 1. proportiO­

nal to f.£".tgl,which prod~t vad •• frOll 0.6 to 297,

the akln...septh varl •• fro. n.' to 0.12 em, thA at­tenuation fActor froll! 0.06 to '6.4 dB/c:..

2/ The free-as-a=e Attenuation et 10 Gaa between two antenna. having 5 dB gain and place.! at a distanc.

of 10 em, 1. 22.4 dB. With .. 11 - .b. alcro.trip

antenna. it 11 eAsy to "hi.v. 12 dl ,U.n,hence the

bue attenuation at calibration c:&n be expect'" to

be no more than •• 4 0. II ':'he water AI Iouy .... terial exh1b!t..I an 111.,adanc.

with approx.45 O~ 1' .. 1 part, therefore surface I'.f­

lection are ~xpected.

- 40 -

There are reflections from other ?art of 5&l!\ple -i too - depending on the geometry of sample. This w~s examined in t<41 using an a?proxir.oll te approach based on geometrical -optics. The results were: a ~nimum thickness and minimum transverse dimensions of the san.?le lire needed, de?ending on <the moisture content. To analyse these problems, we must have' knowledge of the exact solution of the electromagnetic field. between the antennas used in moisture content mea­surement, which is difficult.

substances. For tapered circular aperture, the nor­malized on-axis power ~ensity (fro~R.C.Hansen) is:

?~-.c.) =2U[oi.- ~il( s(n:X -t 12::~(1_t.O~::')] /10/

which is an oscillating function, and at-x-l, p-l. The peak power densi ty OCcurs a t about x • 0.1 and is nearly <42 (or 16.2 dH). The a.~totic value for a small distance i. 26.1 (14.2 dB). The on-axis power density at R • 2.L2/~ is,

~ • 311.P/64.L2 Ill/

ATTE'UATIO. OF ~ •

. - t F[S/:m~""", 0, A I .

------.---- --At moisture measurement using two antenna., the

receive antenna usually located in the radiating near-field (Fresnel- region) of the transmit aper­ture, and the expected attenuation at calibration (moist substance 18 out) .... ill be higher than in Eq.

11/, because Gr and Gt are the so called far-field gains (measured in the Fraunhofer resion). The limit of Fresnel- region i.,

,. .. ~/{2~/").1 R-.zt!-/). /91 .... here L-is the lArger dimension of the radiating

aperture.It i. useful to examine quantitatively the radiating near-field variation,especially along the axi., for such applications a. personnel radiation hazard. and short distance illumination of rooist

where P-is the radiatej power, L-is the diameter of the aperture.

The directivity 1, re:lucea and the main beam i. broadened in the near-field. The relative re~uct10n of gain is:

Sex}- G(x)/Go - p(X).x2 112/ where pex) is the normaliz~ on-axis power density - given in Eq. 10 -, Go-iS the gain at x-I. 'l'he beam broadening factor ia the sqUAre root of the

reciprocal of the directivity reduction factor,i.e.

"'C,,) =- -eC)l.l/-Bo = "/f~C~) & 1/",. f/{p(1-) Illi .... here .e.i. the beUIWidth at x -1.

SUrPosing that the radiated power is P - 10 mw, frequency f-10 GHz,diameter ofibe circular aperture L - 2.S em (tapered illwninat10n i. supposed) ,the calculated on-axis power den.ity (at the distance of 0,42 em from the aperture) using Eqs./9/-/121 ls. 'P - 10 mW/cm2, .... hich is t.he personnel safet.y level..

-'" -- - --,,-------

DES)GN OF INTEGRATED MICROWA \IE MOJSTL'RE SENSORS

Preliminary considerations.

11 It is useful to .elect the frequency ~the mois­ture sensor in the x- ban.1, because the ser.dtiv1ty of the microwave attenUAtion related to moisture c:on­tent (a dominant. 21 If we .hould perform the safety regulation. without any problems, no more~ 10 mW

• of radiated power i. suggested. 3/ USing the diag­ram of Fig.2, the m.asured aicrolAve attazluation of wet COrn having 28\ moisture content (at the frequ­ency of 10 GHz) is approx. 40 dB, 1f the thicme" of the sample 1. 4,9 cm. 4/ We must be apply a mi­

nimal input .ignal level of -40 dam, using a .imple Schottky detector, to achieve a sufficient signal­-noi.e ratio at the output of o\.U' equipment. SI The distance of 10 em between antenna. 1 •• uitable, for versatile application ••

From above considerations the free-srace attenu­ation will be no more than 10 dB, af'cal1bration, and

from Eq./ll comes that the mini=Al value of 11,2 dB of gain i. ne~e~ at such antennas. To get reliable

information about the moisture content of wet sub­stances a minimal value of 'illuminated eross sec­tion- al.o needed. It i. important from practical

I point of view that we mua-have only a limiteJ change

- 41 -

- - ---- _. of the received signal at the ca1ibration:-=0~r~a7t~t~he--------------------------------

I In our equipment, instead of the two - par&meter measurement of quast-dry substances, if~e range of (attenuation and phase) measurements which are accu-antennas has a little variation (be:ause of vibration t .. 11 i '-1 f roll e an~ we su t&~ e or laboratory conditions, we of dryer machine or e.g •• winging of textile). measure only the attenuation (it sim?lifies the re-

We sl.UIImarized -t.'le main s~ecific.tion ana our mea- "" i 1') d th r ~e ve an e errors are decreased by oth£r m~~~~. surea.ent results for such antennas in Table I (a.e I next paragraph). i"

Fiq.3

• , 4 , II ..... ~ ........ Ali,;. uc. »un:IIU,

Fig.4

§ l §

I , ~JU:~rNdY, Ilc.~ I:IIJ \, I I I

-\ i 'L:Cl IVIlle A. TE:l:lAS, I I ~~?tl., CUCV:.AA ,;\lHAJlr PI:£D , I

"- I I \~~~~:~ R"rA.l~ ~ I i K7 1\\1 t.:l:U;1.o ~ c~C::~'f .~o¥-~ ~t.D ~CA~S':a P

\~ ~ 1 \,!.'\ I I i\ I\, I I

~ I \ \./ ! I i I ; k :\. ! I

~" " I ~ ~~. ',! ;~.tYr i r ~'>.!\ ~ I I i :,... : \":~ I ,,~

~ \~ ~-.i. " I

1',~ __ . t'.... I

I ~~ ..... ~ .... § I

:~Sl!. >IfT£ lUI, : LOC, ID "\ i".: ~ I

~~~~~p~~~. f~JU:'O."~"" .~

:,.0 so;_ ;r ... ~ 1uo~1 -, I "'-I I " I ~

I 1 ~ t',i

--The-block diagram of the -Integrated ~o1ature Sen­sor for Grain- i. shown on Fig.l. Thetransm1tter and receiver are built tog~ther and are loeated outside of the sample .pace. A square-wave modulated (30kHz) systam wa. developed, thus a re~uced oscillator nOise and simplifi~ signal proce.sing onine receiver side can be achieved. The tranu.1tter osclllator(dielect-

:I. 1 .. 6 • • t& ~ I ric resonator osc1llat.cra DRO, working around 8 GHz. L\llCOE ar.M:Z:H ftAIl,M.'loQCIP AM]) JU:C:.AII:n:;r"" • (~ __ ~ with s1Jllple on-off JDOClulation.) is Integrated with the

I I I I',

Before dea1gning the aohture sensor. for cUffe- I III1n1ature circulator (reducing the load-pull effect) rent purposes,prelim1nLrY mea.urements[ll were con- ! and micro.trip antenna array.ln the receiver, a s1mp­ducted on lot. of granulated material. an.:! grain. le mcrostrip (planar) detector 1. u.ed, integrated

to obtain reference attenuation daterS). With given density and temperature of s.mpled material., the microwave ~ttenU4tion 1. characteristic to the mel.· ture content,~f the constant thickn ••• of .ample 1. maintained vith a proper clo.ed lensor Itructure [6].

Based on menured data in X -band, it can be· "noted that in the typical ao1sture range the characteristic i atteoUAtion value., 1- 50 d8 for corn, 2 - 20eS8 for i

I vheat, 2-20eS8 for leather, 0.5-12d8 for textile,etc

--Fig:2.hows the II\1crowav .. ·.-ttenuation of wet corn as a function of moisture content; the average density value taken from the literature and the loS. factor calculated from the measurament together with its regres.ion expre.,ion.

onto the recelver micro.tr1p antenna (thickness. 1,6 ~) protected by a plastic planar - radome, and are lo:ated In the p.sslng mo1.t - grain. ..cau.e of the high dynamic range requirem.nts,a 10garithm1c a~li­fier 1& used in the receiver. The Uo(M) output .1g­nal of the receiver 1. obtained after detection and ~_ a~llfication. Proce •• lng of calibration d.~, ~erature (.erv~ by the thermi.tor - temperature sensor) compensation are organized by a microproce.­'01'. The low variation of corn density In the sample 'pace 'surroun~ed by 40 x 40 =2 .quare Nul tUbe) was privl~ed by In a .. chanical way by directlng sheet •• In the practlce an o?timal control wa. fo~ based on 1 samrle / Nnute wlth IDOving averaging-

- 42 -

EXPERIMENTAL RESULTS

---_ .. 11 Integrated Moisture Sensor for &rain (MSG).

Our instruments (shown in Fig.l) \>Iere mounted at the input and output of a corn dryer with 80 ton of capacity and 15-20 tonlhour drying speed. A Burrows Model-700 ~iqital MOist. Counter was used~r calib­ration (see Fig.6). The built-in ~croproces80r se­lects automaticly the appropriate A -or B charocte­ristic, so the sensitivity at the lineAr range of H (16 - 29') will be as high as ~VI ~~. 1.119 V/'.For yellow dent field corn,a 8imple linear temperature compensation: M- -O.l4(T-22.S)was ~sea).~fficiently in the practice. ~he measured ~ncertainty of the moisture content deter~1nation was ~~. 0.41' (the same material was ~ea&ure:! twenty times) at M-23.7'. Using the nWTIE:rical coefii,ientu.nd the equation;or

one-parAmeter measurement in r6J,t~e cQlc~lateJ re­lative fluctuation of the material 3ensity 1n the :neasuring space was: IJf/f - 0.105, which is .. r4:tal val~e for this simple mechanical solution usin~ di~ recting sheets.

21 Quality and stability of mi~rowave circ~its used in our moisture sensor.

To characterize the quality and stability of our copper thickfilm microwave circ~its useJ in MSG,un­loaded q~ality factor (00 dashed line on Fig.7) and resonant frequency measurements of ring resonators were executed in a broad temperature ran~e (in com­

I

'I I' I

I I

1.2

\0

u >'"

s

'"' :E. -0

= , W

U

C .. , to3

0

>

:z At .. = 00

0 S 10 l:5 JO MOl 5 'f tI 11 E COli 'f E II If xC,}

Fig.6 The calibration curve of MSG for yellow dent field corn. Temperature ccmrenaation and sen­sitivity values are also 9iven.

.. . 0_

r--

~\

K-

-- -

l-Jftle1 " ... ..........1

'\ ...........

" --~ .. ~ ........

K" ~II

&= r-...... .........

........... +-=--r- I

~.-~-........

-~

~

, .. i .. .. ..

parison with our high telU?erature solperconduetor fAic- i .. rostrip resonators). At the temperature of 3~ K, --- - -- .- '-.- I'--- ~-t--r-~ · · · · i 00- 145 at the first resonant frequency f l -2.a6 GHZ" I I .. -and the l,requency-temperatur-. stability factor i. ~f' t -.. - - -

Y I ••••• Y •• • CaJ ~.~l_ -93 ppm/X. The attenuation constant for mic- -Fig.7 onl~d~--q-~11ty factor and resonant frequency rostrip line having Zo· SO Ohm characteristic impe- of copper th1ckf1lm ring resonator on al~na dance 1.: d.,-S.l dB 1m ant! the {requency dependence i. cl(1) - 2.97 it+ 0.017· f (GHZ). The lDeasured resulU , on-th.thirdr.;~-;' (around 9 G!!zf.i.-;-Qo·~~ ci.- 9.1 dB/m. These parameters are better, than the

meuurej values for gol~ thickfilm resonators real!-

ze~ on the .ace al~na substrate.

substrate aa a function of the temperature.

· o .. • .~~--------~~~~~~-T~~~---------1 :a .. II

• .. :ale

• •

In our MSG instruments small -Jimen.ion and low­profile micros trip antennas .... ere used 1ha.n radiating elements.To check the temperature dependance {o those resonant ArrAyS, A relatively narrow-ban! 4-element rectangular micro.trip antenna for the freq~ency of 4.4 GHz was Jr.ad*. The results 10 the temper.:. ture de­

pendent inp~t ret~n loss vers~s freq~~ncy ~asure­ment, ae on Fig.8.The relative req'.1ency-te.mf'p.rature

stability of the resonant array is: 86 ppm/ K. Be­CAuse our micros trip antennas used in MSG have 2.3 tin .. s higher relative ban::!width at the ir.q~ency of

"~ ~ l " •• QUI.Clf,. DlGIII

I-F19.8 Inp~t return loa.-of four element microstrip

antenna as a f~nctjon of the fr~uency. The pArameter of the curves i. the temperature. 8 GHz , this pr~.~!.e!l i_~yract1cAlly negligible.

- 43 -

31 Radiating near-field ex~)er·iillent •• al Transmission measure~nt (at calibration).

Different types ·of ar.tennas are used for microwave moisture measurement. Low transmission 10 •• (at CA­

libration,noiat substance is out),rnatchej input im­pedance, given value of -illuminated-cross section, and relatively smooth. loss -distance characteriatic are the main requir~~nt •• Because closed formulas

are known only for simple apertures (e.? for tape­red circular aperture:Eq./l/),we eXAmine~ this prob­lem eX?eriu~ntally.

Pyramidal horns are frequently used in the prac­tice. The received relative power of optimal pyram1-

dal horn (model I F. see Table 1) a. a function of the ranc;e between antenna. i. shown in Fig.3. At the

distance of R· 10 ern, the relative power iSI-6.4dB, the rapid variation near to thts dist~ee is:!1.2da I

Far-field gain is:Go·1S.6 dB,and from Eq./ll comes that the reduced qa1_~-.!.n_~~n.!ar-f1eld at R • 10 em 18 only 13.3 dB.

Tat.l. 1. Para.-tara aM ..... IU'a4 v.11M. o! the ta.tad JDCdel ••

Model Go • I 11 W'lI1nat.ed- A ·3 R- lOea I1G type dB a~.£pa '_?pro •• , 4m2 de; Act ~ (dB) dB A 7 •• D-lOcm, c1rc~l&,. 0.' " U.S 10.12 ~.20

• 6.1 D-l1.5=-, cue. 1.0 II 21.0 ~1.50 -0.15 C 12.6 7.2. 7.2 C1II aqu. 0.5 42 '.0:0.15 -0.70 D 12.0 7.5. 7.5ea aq~r. 0.6 .5 lG.' 10.60 -1.0 r; 13.0 12.5.10.;: rect.eD9 .1.2 S6/2. 1.6 to.n -2.1 r It.6 1l.la •• 7c::11 recto 1.0 U/l7.S 6 •• ! 1.20 -S.l

Go far-field galn in dB

A I -illuminated-surface /approx./ at ~x/-3dB/. taking into account the beam broadening# too.

"3 I 3-d! bell!lnddth,

P GIla

10.5 10.5 10.5

10.5

'.0 10.5

~ I the mea.ured uanam1adon lou ln dB, at Jl"'lOeii

AA I tranam. 10 •• variation nur the ~H.t.ance of 10clD At; zne •• ur~ gain reduction in the near field a;tlOcm -- --- ._---The exam1ned antenna model. are.

A: Circular primary feed, D1-21 mm,Do·48 mm,4qrooves S: Open circular waveClUide C-120, usinc; extreme

luge flange, D1- 17.5 am, Do. 60 =m. C: Circ. polarized m1crostrip ant, using 4-elements DI Linearly polarized m1cro.tr1p ant., w1~~ 4-elem. EI Circ. polarized !tJ.cro.tr1p antenna,with 8-eleJllenta F: Optimal pyr.am1dal horn, Ax ax L • llOx 84x 102 II1II • ,-

.. . , p)

I: t4 1;

Fig.9 Relative power level. in 5 dB/div coro~ressed vertical .cale of ~el. A and B, a. a func-tion of range between tran~it and r.e.i~. apertures.

Fig.4 shows the relative axial power den.iti •• ~ receive antenna. (model. A,D and C) a. a function ~ distance from transmit bop. The e)o;trapolat.d ·~.1n· of the loop from ll\easurements was about -15 dS.The •• curves are good agreement ",ith iheoreUcal4txan.1r.aU­ons, at the range of 6-14 em the models A an~ Care in the far-field region {l/R2 function),the model D at the range Of 4-1Oc:m 1. in the rajiaUng near-field (Fresnel) rec;ion.Transmission 10 •• (at~oi.ture mea­surement calibration) i. shown in Fig.S for ~delsl

A,C and E.A the distance otlO em, ~el C and E are in the l/R region and the attenuation is about only 9 dS. Model A fa in the 1/!\2re~ion, with the relati­ve level of the received power of -18.S dS.

The radiating properties of MI anter.n& (measureJ in the neAr-field) are chAnged eonsi1~rably taking the radiator into a metal k>x. Modeling this problem, an open eirc~lar waveg~iae with e~treme large flange (thi. i. the front wall ~the cylindric~metal box)

i·~~_~~4-~ __ '.~~~~

.'&' :!!. .. . .;:; .

• LI __ ~ __ L-~ ___ 4~1 __ ~_·_tL-~ __ ~ __ ~~~-L __ ~ __ ~ __

INIGI: IrNI:D ftAIWl, Ale uc. &PIImIIU'_.~~';:;;-Aa ____ _ .~

·a~~-~~~~--~~~-r-~~r--r-~~r--r--, •

..

= t ~

I ._ ...... _ ....... ~

--~--~ ~

c: i i E § ••

=0 L __ l-_,L--1_~4~-l __ ~ __ ~~~-L_~ __ ~~r--L~

II.1.II(;1 anwu:u t'IUoIIlIl. AW III:C • ..,.~. • (~). .

Received relative power level a. a function of range between transmit and receive aperture., using circularly polarized micro.tr1p antennas with four elen.nts. Input micro.trip line. are in the .ame position at Fig.10.

o;>poa1te position a~ ._ rllZ.ll.

- 44 -

'"" Uleasured. The-result is -shO..m in Fiq. 9 (curve fiB)

where a compres.ed v~rt1c4l scale (SJS/d1v)is us~~. A very large periodical variatjon ~the raceivea po- I

wer i. ol;.servable, while Lt has a smooth function at !nOde! A (curve: AA).

The above mentioned problem is avoidable, using circular polArization. We have ma.:!e four-elen,ent mic­rostrip antenna. with circular polArization (which is we 11 known from the Ji tera t ure). Loca Ung the in­P>lt micro.trip tine. into 5a:ne and opposite directiona the received relative power level. as a function of range between transmji' and ;'"~-::eive antenna. Are shown

in Figs.lO and 11. The ~~t results are given by the lAst configuration, because of the h1qh r~ae1ved level (-9 dB) an~ low variation (to.15 dB). It must be mentioned. that tn our Qriljinal solution the mic-rostrip power~litter network between radiating ele­ments had a quasi double symmetry (11.

I~~~~~~~~~~~=±~==~~~=b~ F1q.13-Input -return loss of

microstrip antenna as a function of the moist material from it.

CONCLUSIONS -------------------------------

.u-....::a ,. C1n Mft&l.AJ. rKa c= .. ftl. ~ t

.r--t4--i2---lr--74--;-~r__+--4'--~~'_~--~~~_=~~ ~_ The ~evel~pment a~d~idearread us. Ofinstrumenta-

...

.!or ;I

Ie i s

V , "

i lIS

L. F1S.12 Input return 10 •• of the tranamit

~cro.trip antenna loading it with different moist !LAterials at the distance of R(cmJ.

-- b/Inpu-t return 10 .. -meAsui.nt (at -moist su.bstAnce)

According to Eq./7/, the input return loas of the transmit antenna will deqrade, locating moiat auba­tance near to the radiaUng aperture. Thia effect was measured at the frequency of 2.45 GHz by the help of a motorized automatic input return los. measu.rement set~ using two type. of micro'trip antenna •• Fi9.12

.hows the input return loas of the trans&! t r"ctangu­lar microatrip ar.l.tenna bading \t with different moiat materials at the distance of It (em]. The nearest opti­mum place .. g- 4 cm, the minimal return loss RL-12.S dB, which ~ans VSWR- 1.6, reflection co.ff.lr~.24, mismatch loss MI.- 0.25 dB. Ul1ng a broad banded ~c­

rostrip antenna at the neArest optimum (alsO R- .. cm) RL- 15 dB, VSWR- 1.4, \I""l- 0.18, HI.- 0.14 dB. It is

interestinq, that sometimes the moist substance give. harder reflection, t.an the water tank itself. On the other hand, it u po. sible matching by the dry sample (Fiq.13, curve 2, at g • 5 cm), which nay be useful certain ca.e. (tow loa. reflecUnlj plate, or the wall of the dielectric .a~le holder).

tionior microwave moisture cor-tent meaaurement reac­hed a maturity. We have to take the nex .tepa coope­ration between agricultural-, microwave and field­

, engineera. One of the purposes of thi. ~per to sho" (exper1.mentally) the po .. ible reduction Of calibration errora &nd improvement of the .tability of equipment using inteqrated antennas and copper thickfilm micro­wave circuit •• It vas experienced that in terma of propagation 10 •• and near-field reflection., there~ an optimum size (or element number) of antenna, a~i­tably micro.trip &ntenna with circular polarization.

REFERENCES

[1] r.V~lqyia -Ver.atile MicrOWAve Moiature Senaor-Proc. of the SBMO' 89 Int.Microwave SYJDP. / !LUlU. Sao Paulo, Vol-II, pp.456-462, 24-27 3uly, 1989 Andrzej W.xraaze"akl1 -Microwave Aqu&metry-Needs ADd Per.pective.- IEEE 'l'rans.on Mft, Vol.3', No.5 pp. 828-835, May, 1"1. r.velqyl, L.3ach1movita end I.lCza6ki.-Oesl9D of Bybrid Integrated Microwave Circuit. on Pl.stic Substrate- IV .Nat.Conf. on Microwave Solid State Electronic., GdanSk/Poland, Proc. pp45-5l, 1977 3.M1adek, I.Seranl-Sample Geometry, Temperature and Denaity Factor~ in the K1crowave M8a.uremant of Moi sture- .:Journal of Microwave Power, Vol.lSN04 pp. 243-250, 1980. Swart O. Nel.on,-U aDd Microwave Dielectric Pro­pertie. of Agricultural Proc!l1ctl and their Appli-cations- Proc.of the Sa~·8' Int.MJcrowave / Brazil. Sao Paulo, Vol -I, pp. 29-35,

Syznp.

1989.

Andrzej Krasze"ski,-M1crovsve MQnitQriA~ of Koia­t~re Content in Grain - Further Consideration.-, Journai of K1crowave Power and ElectroNgneUc Energy, Vol.23, No.4, pp. 236-24', 1'88.

- 45 -

DEVELOPMENT OF 1fICROSTRlP SENSOR FOR OIL PALM FRUITS

Kaida b. Khalid and Zulkifly b. Abbas

Department of Physics, University Pertanian Malaysia, Serdang,

43400 Selangor, Daru} Ehsan, Malaysia.

Microwave aquametry or measurement of moisture content in liquid, semi-solid and solids by microwave technique is known to be accurate and rapid. For a smal1 sample size, it is conveniently measured by using a microstrip, since only a small part of the sample interacts with the line l . The application of this sensor for microwave aquametry in lossy liquid such as hevea rubber latex2 and determination of fat/water in fish have been successfully accomplished3• This paper describes the analysis and design of microstrip sensor for oil palm fruits. It was found that the amount of moisture content is higher at early stage of fruit development at 20 to 23 weeks after anthesis and this period is almost the same with the accumulation of oil in the mesocarp4. Therefore the close relationship between moisture content and oil content in mesocarp gives a possibility of using % Me/fresh mesocarp to gauge fruits ripeness. A functional relationship between insertion loss, I S21 I of the sensor and moisture content in mesocarp is developed and then compared with the experimental results. The performance and application of this sensor in setting up a har .... esting system for oil palm fruits has been discussed in detail by the authorss.

Design and Analysis The microstrip sensor and cross-section of sensing area are shown in Fig. 1. The sensor consists of three

parts: the coupling system (input/output) representing the transition between coaxial and stripIine, the 50 !l stripline section and semi-infinite double covered microstrip or sensing area. The substrate material is RT­Duroid with a relative dielectric constant of 10.7 and thickness of 1.27 mm. The microstrip section is protected by a thick layer of polymeric material with thickness of about 0.4 mm.

The dimensions of the sensing are.a were selected for a single fruit measurement, with volume of about 31.5 mm x 17 mm x 7.27 mm. The impedance of the microstrip line (in air) is about 61 !l. This impedance is matching to a 50-0 system at higher moisture content.

Fig. 1: A microstrip sensor for oil-palm fruits

- 46 -

In this analysis we consider the complex amplitude of an electromagnetic plane wave propagating through a medium which c.an be represented by the equation

E(d) = £(0) e -,d (1)

where £(0) is the complex amplitude of the wave at some reference plane and £(d) is the complex amplitude at a distance d and, is the propagation constant of the medium.

Figure 2: A signal flow-graph of the micro.trip .~nsor structure.

1 + r. -1 ~ e • 1 + r. .-J.t2 1 + r. -7 ~ • • 1 + ra.

r, -ra r. -r. r. -r. rb -ra.

1 - r, -7 ~ • • 1 - r. .-7.t2 1 - ra -7 ~ e I 1 - r. IlI,lIt trllllltiaa .trl,lul trll3'ltlOD .aro.trip trlll.1t101 .tr1,11111 tuUitUIl Coui,l-UripIiJlt HctiOi .tripline- uuhf aicro.trit- •• c:t1011 .tr1,liD.-

aiero.trip au, .tri,111. coad.l

. " 8 11 .-7.t1 8 21

. . 8~1 8;2 S 11 S 22 "

. s72 S12 .-7.t1

Figur. 3(a): A simplified signal flow-graph of Fig. 2

Figure 3(b): A simplifi.d signal flow-graph of rig. 3(a)

The reflection and transmission phenomena in the sensor structure can be represented by the signal flow graph as shown in Fig. 2. The signal flow graph in Fig. 2 can be simplified as in Fig. 3a and then 3b, by using Mason's Non-Touching Loop rules6• From Fig. 3b the insertion loss is given by

where the components of scattering matrix 5 21 , 5 21 , S22 and 5 11 are expressed by

where

5 21 = e -1i1(1

-"",I 5 21 = e '. 1(1

r 0)(1 - r m)/5"

ro)(l- r m)/5'

- r m(1 + r arm e - 21.'1 ) - f b(1 ~2= S

- r m(1 + fbf me - 21i1 ) - fb(l - r m)(l + r m) e - 21 ... ,11 5i'1 = S"

5' = 1 + f af me - 21.11

S" = 1 + fbfrne - 21.11

(2)

(3.1)

(3.2)

(3.3)

(3.4)

40

35 (/') (/')

o

0 30 ·c .-.J

o ~ 25 Q)

:.0

0 20 .-.J

C

B 15 V)

c o u 10 0 ·c

.-.J

0 5 cu Cl)

"0 0

20

.;.. 47 -

Frequency = 10.7 GHz

£ (water) = 61.9- j 29.8

C (oil) = 2.96-j 0.1 . £ (fibre) = 1.36-j 0.12

30 40 50 60 moisture content. ~

70 80

Fig. 4 Dierectric constant £' and diefectric ross £" in fresh mesocarp as functions of moisture content

75------~------~----~--

c 70

()

0 's/b -c 65 0 0.2 -0 Q)

0-E

0- 60 0.1 0

::; V)

°C 0.05 ~ 55 0

0.03 0 ~

0 0.02 .s:::. 050

0.01

45~-------~------+------+------~----~----~ 20

fig. 5

30 40 50 60 moisture content • "

70 80

Characteristic impedonc~ of loaded microstrip of various s/h ratios as functions of moisture content

- 48 '-20

slh

0.01

0.02

0.03

0.05

___ ~----- O. 1

=:::::::=----:..-__ ------:-- 0.2

5~-----.------~----~----~------~-----25 45, 55 65 75 85

moisture content ~

Fig. Effective dielectric constant of loaded microstrip against moisture content at various slh

21.0 20.0 19.0 18.0 17.0 16.0

en 15.0 -0 _ 14.0

13.0 ~ 12.0 0 11 .0 -'0.0 U 9.0 "- 8.0 ......., U 7.0 Q) 6.0 Q) 5.0 -0 4.0

3.0 2.0 1.0 0.0

--------------- /. - ~ -~~ ~~~ "'l~ .--'

... ---

• • I •

25.00

/ /'

V / /' . /' ,/

V ./ ./"" / / ./

",.

// ./ ~ ./ "/' ./ ;Ii' ,..,,-

~

// 7~ ~ /// ./ .

77' ./ ;7'

------..--

'f' ./ .------ .... ~ ,..- -----~

~ -. ~ -. -. -. T T T ••• . . . . 45.00 65.00

moisture content, ~

/' /

~ ./"

./ ~

~ ....-

-------

. . . .

• s/h • 0.01

0.02

0.03

0.05

0.1

0.2

85.00

Fig. 7: Dielectric l~ss of'loaded microstrip against moisture content at various slh

- 49 -

The reflection coefficient at the stripline - microstrip transition f m is given by

'

Zm-Zsl fm = Z +Z m S

(4)

where Z m and Z s (equal to 50 f2) are the characteristic impedance of the microstrip section and stripline section, respectively. The reflection coefficients at coaxial - stripline transitions fa (input) and fb (output) are determined with network analyser. These values are 0.003 and 0.004, respectively. 1m and IS are the complex propagation constants for microstrip section (sensing area) and stripline section, respectively, and are given by

and (5)

where Q is the attenuation constant, in this case due to the dielectric loss and {3 is the phase constant. The calculation of Z m' 1m and IS by using TEM analysis is given in detail by Khalid et aI7.

In order to calculate the above parameters a relation betwe'en the moisture content of the mesocarp and its permittivity must first be established. To achieve this the mesocarp will be considered as a mixture of oil, water and fibre. Using mixture theory the relative complex permittivity of the mesocarp can then be expressed as:

(6)

where V w' Viand V, are the volume fractions of water, fibre and oil, respectively, and (:" (j and (; are the corresponding complex permittivities. The volume fraction for fibre Viis considered to be constant volume of 16% 8, whereas the water content and oil content vary with ripeness. V, and V w can be written as

V, = 1 - V w - Vi (7) and

V _ _ M_(......;p i,-V--,;i:",,:,,+_P_'_--;P~1 V~i) w - Pw - Mpw + M P,

(8)

where M is the moisture content (wet basis), and Pi' P, and Pw are the specific densities of fibre, oil and water, respectively. These values are: 0.92, 0.93 and 1.00, respectively. Fig. 4 shows the variation of the dielectric constant, (', and the dielectric loss factor, (" with moisture content in the mesocarp. The variation of characteristic impedance, effective dielectric constant and dielectric loss factor of loaded microstrip section with the moisture content are given in Figs. 5, 6 and 7, respectively.

Results For this study six bunches of tenera variety from 11 years old oil palm were selected. The samples were

taken from the outer fruit around the equatorial of the bunch. The fruits were analyzed once a week from 12th week after anthesis until they were fully ripe and detached. The fresh mesocarp was sepa.rated from the nut and then cut into small pieces and these pieces were crumbled until they become an uniform semi-solid sample. The sample was press firmly into sensor to assure a perfect contact. The magnitude of the insertion loss, S21 I of the sensor was measured by using an automatic swept frequency Network Analyzer (HP 8720B). All measurements were carried out at 26°C. The actual moisture content of the sample was obtained by using the Karl Fischer method with an accuracy of ± 0.1 % of moisture.

Figure 8 shows the computed (Eqn. 2) and experimental results of the insertion loss I S21 I of the sensor at 10.7 GHz as a function of moisture content. The results show that the difference between computed and experimental data is less than 4.3%. The sensitivity of this sensor is about 0.6 dB/% moisture content.

Conclusions In this paper the analysis and design of microstrip sensor for oil palm fruits have been discussed. A close

agreement has been obtained between computed and experimental results of the insertion loss as a function of moisture content in fresh mesocarp. This agreement suggests that a realistic optimisation-based design approach with respect to the effects of the geometrical and electrical parameters of sensing area can be developed. Therefore, a sensor with higher sensitivity can be designed which will give a better prediction to the ripeness time of the oil palm fruits.

Acknowledgements This work was supported by IRPA Research Grant (4-07-05-021) coordinated by Ministry of Science,

Technology and Environment of Malaysia. The authors wish to thank Dr. Azis Ariffin of PORIM for helpful discussions during the course of this work, the management of the Farm Department, University Pertanian

42

40

38

36

.34

32

ro 30 "0

·28 en (/) ""6 .2~

c 24 0 - 22 ~ Q) (/)

20 c

18 -16

14

12

10 25

~

:

~ 'f'

/ /;

,I

• . ·-i

- 50 -

r: ':. • til . -/_1-~ -1'

/ . . . V.y

/ /~ I

/ . I . . ! I

I ! I V • . I / i

I / ~.

/ -V I . / .

/ . /-- .

/1 .. .

; I

40 55 70 85 moisture content • st

Fig. 8 . Comparison betv .. een meosured (.) and colcuroted-­ir.sertion loss os a function of moisture content

Malaysia for their excellent cooperation and technical staff of Applied Electromagnetic Lab., Physics Department UPM for their assistance in carrying out this project.

References 1. Kent, M. 1973. The Use of StripIine Configuration in Microwave Moisture Measurement II. J.Microwave

Power 8 (2): 189-194. 2. Khalid, K.B. 1988. The Application of ~ficrostrip Sensors for Determination of Moisture Content in

Hevea Rubber Latex. J.Microwave Power I.e EE, 21(1): 45-52. 3. Kent, M. 1990. Hand-Held Instrument for Fat/"'ater Determination in \\'hole Fish. Food Control, 1:

47-53. 4. Ariffin, A. 1984. Biochemical Aspects of Ripeness Standards. In: Proc. of the Symposium on Impact of

PoI1ination \\'eeviI on the Malaysian Oil Palm Industry.PORIM, K. Lampur. 5. KhaJid, K.B. and Abbas Z.B. 1992. A Microstrip Sensor-for Determination of Harvesting Time for Oil

Palm Fruits (Tenera: Elads Guineensis). J.Microwave Power &. EE, 27 (1): 3 - 10. 6. \Varner F.L. 1977. Microwave Attenuation Measurement. lEE ~fonograph Series 19, Peter Peregrinus

Ltd., Herts. 7. Khalid, K.B., Maclean T.S.M., Razaz M. and \Vebb P.\V. 1988. Analysis and Optimal Design of

l\ficrostrip Sensors. lEE Proc., 135 pt. H (3): 187-195. 8. Hartley, C.W.S. 1977. The Oil Palm. 2nd Ed., Longman Group Ltd., London; 22-223.

Dr. Kaida bin Khalid was born in Kemaman, Malaysia in 1952. He received B.Sc. degree in Physics from National University, M.Sc. in Solid-State Physics from University of London and Ph.D. degree in EE from University of Birmingham in 1986. He is an Associate Professor in the Depart.ment of Physics, University Pertanian Malaysia. His researrh interests include microwave aquametry and development of microwave

- 53 -

Material Characterization Using Microv..'ave Open Reflection Resonator Sensors

R. J. King*, ]. C. Basuel, M. J. \Ven1er and K. V. King KOC Technology Corporation

2011 Research Drive livermore, CA 94550

(510)449-4770

The f\1DA-1 000 sensor .is .an innovative, proprietaI)', microwave open reflection resonator. \\'hen the sensor makes physical contact ',"ith the ~1aterial Under Test (~1UT) its resonant frequency (fr) and normalized input resistance (ro) at resonance are influenced by the dielectric constant (E') and loss factor (E") of the ~fUT in a quantifiable way. Either of the measured electrical sets (fn r 0) or (E', E") are in turn relatable to numerous physical characteristics of the ~fUT. Since two independent electrical parameters are measured, it is possible to detennine two independent physical parameters simultaneously, e.g., the moisture content (mc) and the bone dry partial density (Pd ). In this way, density-compensated moisture measurements are made.

For automated industrial applications, (fn ro) are measured using a specially de\"eloped microwaye (narrow band) scalar network analyzer with on-board microprocessor for instrument control and for data acquisition, processing, display and output.

The sensors are application specific, depending on whether the l\fUT is a solid, a particulate or a liquid, on the range of (E', e ") on the material abrasiveness, on the desired spatial sampling resolution, and on mounting constraints in the on-line pr-:)cess. They can be made in a variety of shapes; flat or curved for flush-mountE"g to a pJane or curved surface, or cylindrical for immersion in ,a flow stream or for manual insertion into liquids and granular materials. For liquids and most solids, the sensors are about 3 cm in diameter, and for coarse particulates (e.g., wood chips), the sensors are about 18 cm long. The effective sensing depth into the l\IUT ranges from a few mm to several cm, chiefly detennined by the sensor size and resonant frequency which ranges from 0.5 to 8 GHz. Polarized sensors are used for measuring dielectric anisotropy, as when the l\ruT contains aligned fibers (e.g., wood and fiber reinforced composites). A thennocouple is embedded in each sensor to allow compensation for the temperature dependence of the l\1UT dielectric properties.

The sensors are calibrated to measure (e', £ ") using 'standard' dielectrics which may also be solids, powders or liquids. Best accuracy is achieved using liquid standards because their specific gravity is constant and they make consistent physical contact with the sensor. For industrial applications, the resolution (ability to distinguish small changes) is excellent, being of the order of one part in 103 dielectric units for liquids. Due to surface roughness of solids and particle size of granular materials, the accuracy and resolution are somewhat less for these materials. For mc measurements, the resolution is generally between 0.01% for liquids and 0.8% for very coarse particulates.

Typical Applications

The following figures show typical results for three different applications; wood flakes, oats and a KevJar fiber reinforced polyester composite.

- 54 -

50 ,

a - .00344 ,

a _ .00371 , ~

, c , , 0.15 ... ...... ~ '< ffl. 40 Pd ..... ~ C Fig. 1. Wood Flakes. ,t ...... . ... CD Correlation of me (wet ~ 30 Wood Flakes ,"\ 0' - 0.88 % ::::s , 0.10 ~ basis) and Pd as deter-ca ,.

(mc> 30%) ~

, ... mined by the MDA-IOOO .' me '< ~ 20 ,

sys tern and me by .. ,~

...... me by model CC w~ighing. Nominal Co) , • 0.05 .......

E 1 0 '" - - me fit () fr ... 450 MHz. Temp. :0: 24 ·C. ~ 0'.0.62 % 3(.)

.. ' (mc < 30%) x Pd by weighing , ~ ~

~ - Pd by model 0 0.00 0 1 0 20 30 40 50

me (% by weight)

20 0.6 C

0.5 ... Fig. 2. Oats. Correlation ...... '< ~ P , of me (wet basis) and Pd 0 ~ ~ , C .. 0.4 as determined by the ...... 1 5 , CD (1) Oats • ::::s MDA-I000 system and me , > , en ca .. 0.3 ::; by weighing. Nominal ~ ",t me '< fr - 1 GHz. Temp. - 18 ·C. ::1.

, , ....., '. 0.2 ...... Note the wide variation of 1 0 " • me by model CQ

Co) , me fit ....... Pd due to moisture. E . , .. a = 0.33 % ()

., x P d by weighing 0.1 3(.) , , ~ , - Pd by model ,

0.0 , 0 1 5 20

me (% by weight)

Fig. 3. KevIar fiber re-

1 2 inforced poIyes ter com-

0.4 posite. Correlation of , 1:.£.' and M." m with mc , ,

1 0 , (dry basis) as determined o ~. .. '

0.3 ,

by weighing. Here, c ' ~ .. ,

8 • g'" 6£" m is linearly m i'· ~ ~

proportional to the change 0.2 a,~ 6 ! . in the true e" of the MUT • af· .

aa ~ Thickness - 1.6mm. ,~ 4 Nominal fr - 4.5 GHz. ~ .

0.1 a, ,. , 2 , , , , ,

0.0 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0

me (% by weight)

- 55 -

Microwave Resonant Sensors for Moisture Content Determination

In Single Kernels and Seeds

Andrzej W. Kraszewski and Stuart O. Nelson

U.S. Department of Agriculture, Agricultural Research Service

Richard B. Russell Agricultural Research Center, Athens, Georgia 30613, U.S.A.

The moisture content of a material, M, expressed in percentage, wet basis, is defined as

M = mass of water. X 100 = water con~entrati?n x 100 mass of wet materIal wet material densIty (1)

Standard laboratory methods of moisture content determination are based directly on this definition.

The mass of water is removed from small samples of wet material (a few to several grams) by

evaporation or extraction, and by weighing the sample before and after drying the required information

is obtained. Too often it is assumed that grain and similar materials are homogeneous and uniform,

that water is evenly distributed throughout the lot of material, and differences in moisture content

determination come from the uncertainties in weighing. Such ideal materials are rarely found in nature,

and moisture variation in large-scale industrial production is often a serious problem.

The result of moisture content determination is always a single number, at most averaged for

several samples, and rarely accompanied by deviation or range

of values for those samples. The value of moisture content

determined by this method will be the same for all of the

moisture distributions illustrated in Fig. 1, in spite of

completely different practical steps (technological processes)

that need to be taken in each instance. This is especially

important for agricultural products, where the situation shown

in Fig. Ie (exist.ence of a mixture of two lots of different

moisture levels) can lead to spoilage of the whole lot of

material. To avoid such problems, as many samples of material

as possible, taken from various locations throughout the lot,

should be measured. For grain, seeds and similar materials,

testing of individual kernels or seeds would provide the desired

information.

~O\~t\)l~ c.o",:tG.~t

Fig. 1.

(a)

(b)

(c)

Single kernel moisture meters based on de conductivity or RF impedance measurements have been

developed, but we have found [1,2] that microwave resonant cavity measurements may provide an

interesting alternative, because they are fast, accurate and nondestructive. All kinds of microwave

- 56 -

resonators may be used for this purpose; however, a rectangular waveguide resonant cavity is one of the

simplest and easiest to build. Such a cavity is illustrated in Fig. 2. Principles of general perturbation

theory can be used to establish some general rules of resonant cavity applications for agricultural

products. In standard rectangular waveguide cavities, the object-to-cavity volume ratio should be <

10 - 3. In Fig. 3 the operating frequency 10 is related to the volume of the empty cavity operating in

the TElOS mode with le/lo = 0.65. Volumes of some kernels and seeds are indicated.

1 000000 .-----.----.-----..--~--r----___, STANDARD RECTANGULAR WAVEGUIDES

100000

1"1 10000 E u

W 1000 ~ :::> ....J 0

'00 >

10 WR-650 WR-284 WR-137 WR-90

0.5 30 FREQUENCY, GHz

Fig. 2. Fig. 3.

Two cavity parameters that are related to the dielectric properties of the object are the resonant

frequency shift D.F = 10 - I~, and transmission factor D.T = (~ LO - 1) = 10k . - 1; where 1 La

k = 20 (S210 - S218)' QL is the Q-factor of the cavity coupled to the external waveguides, S21 is the

voltage transmission coefficient of the cavity at resonance expressed in decibels and subscripts 0 and 8

refer to the empty cavity and the cavity containing the object, respectively. These cavity parameters

are related to t.he dielectric properties of the object [3]:

D.F = 2({' - 1) K 10 (~~)

D.T = 4{IIK2 QLO (~~),

(2)

(3)

where {* = {I - j{1I is the material permittivity, and K is a factor dependent upon object shape,

orientation, and permittivity. For example, K has a value of ,!L-2 for spherical objects. Equations (2) { + and (3) can be used for material permittivity measurements when the object is of well defined

dimensions and shape, and the values of 10, QLO and Vo are known for a given cavity. However, we

have found that because both cavity parameters, measured simultaneously for each object located at

the center of the cavity, are proportional to the volume of the object and its moisture content, their

ratio is independent of object size or volume: D.F {I -1 1 10

X = D.T = --r K 2QLO • (4)

This ratio is a definite function of the object permittivity (moisture content) for any given resonant

cavity. This has been shown for single soybean seeds [1], single peanut kernels [2] and single kernels of

- 57 -

wheat, where slight variations in object shape do not significantly affect the dependence of the ratio

upon moisture content, as shown in Fig. 4 and Fig. 5 for peanut and wheat kernels, respectively. \\Then

shape of the objects under test varies significantly, as for corn kernels, two separate measurements have

to be taken before and after the object is rotated in the cavity by 90°. Then the ratio of the averaged

parameters is a measure of the object moisture content.

35 40 ,...---....----..,...---,-----r----..

30 35 WHEAT I(ERNEl.S PEANUT KERNELS

f = .3.175 GHz f - 10.54 CHz

~ 25 ~ 30

~ L..

~ LA..

<j 20 <l 25 II )(

15 )( 20

10 15

5 10~---~--~----~---~--~ 2 4 6 8 10 12 14 10 12 14 16 18 20

MOISTURE CONTENT, " MOISTURE CONTENT, ~

Fig. 4. Fig. 5.

By applying the high sensitivity of microwave resonant sensors, a fast, accurate, and nondestructive

method of moisture content determination has been developed for grain kernels and seeds. An accuracy

.of 0.5% moisture content may be obtained for individual kernels and seeds without need for contact

between these objects and the measuring system, providing an effective tool for determining the

distribution of moisture content among kernels and detecting mixed lots of grain of different moisture

levels.

References

[1] A.W. Kraszewski, T.-S. You and 5.0. Nelson, Microwave resonator technique for moisture content

determination in single soybean seeds, IEEE Trans., vol. IM-38 (1) pp. 79-84, 1989.

[2] A.W. Kraszewski and 5.0. Nelson, Moisture content determination in single peanut kernels with a

microwave resonator, Peanut Science, 1993 (in press).

[3] A. W. Kraszewski and 5.0. Nelson, Observations on resonant cavity perturbation by dielectric

objects, IEEE Trans., vol. MTT-40 (1), pp. 151-155, 1992.

Dr. Stuart O. Nelson was born at Pilger, NE, in 1927. He earned .he B.S. and M.S. degrees in agricultural engineering and the M.A. degree in Physics from the University of Nebraska, and the Ph.D. degree from Iowa State University in 1972. Since 1954 he has been with the U.S. Department of Agriculture, Agricul­tural Research Service, currently at the R. B. Russell Agricultural Research Center in Athens, Georgia. His research interests include also a study of the dielectric properties of grain, seed, insects and minerals. He is a Fellow of the ASAE and the IMPI and a member of the National Academy of Engineering.

- 59 -

A New ftbisture Content Measurement Method

by A Dielectric Ring Resonator

Seichi Okamu ra and Takaaki Masuda

Dept. of Electronics Engineering, Shizuoka University

Hamamatsu, 432, JAPAN

ABSTRACT

This paper shows a new moisture content

measurement method for small or random size

material from the resonant frequency sh ift of a

dielectric ring resonator caused by the moisture

contained in the material which is set in the

hollow core of the ri ng.

The theoretical resonant frequency shift of the

resonator is obtained by using the mode matching

method. The correctness of the theory is confirmed

by close agreement between the calculated and

measured values for the disk-shaped and different

size samples whose dielectric constants are known.

These analysis show that the measurement of

resonant frequency shift per unit weight of a

sample is very effective for moisture content

measurements and this is shown in measurements on

the ca rdboard samples.

I mroOOCl'ICfi

Microwaves become important for the moisture

content measurements of industrial or agricu 1 tural

products such as paper, textile, green tee, etc.,

and are gradually replacing the conventional

measurement methods. In those measurements the

microwave attenuation or phase shift by the water

contained in such mater ials is read to dec ide the

moisture contents. For reliable measurements it is

necessary for the samples to have a larger cross­

sectional aria than the area of microwave

propagating region and have unifonn density.

So far, it has been difficult to measure the

moisture content of single and small size sample.

The wooden chips, for example, used in production

of paper have irregular shapes. Therefore their

moisture contents are measured by the conventional

drying method. The development of a n~ alternative

method is expected from various fields.

In this paper, we introduce a new moisture

measurement method of using a dielectric ring

resonator. A sample of a material to be measured is

set at the center of the resonator. It measures the

moisture contents from the resonant frequency of

the resonator. The resonant frequenc ies had not

been analyzed on such a resonator having samples

with different pennittivities from the

pennittivi ty of the ring.

~NANT FRBJJENCY ANALYSIS

The dielectric ring resonator for the new

moisture measurement method is shown in Fig.l(a) •

It is composed of a Teflon ring, whose relative

permittivity is 2.08, and two conducting plates.

The resonant mode i s ~1l' the size of the

dielectric ring is decided in accordance wi th the

resonant mode chart of:.:.electric rod resonators

[1]. The resonant frequency changes wi th the

pennittivity of the sample, which is influenced by

the moisture content of the material.

Figure 1 (b) shows the cross section of the

resonator used in the resonant frequency analysis.

The region I is a sample of a material to be

measured, the region IV is a dielectric ring, the

regi ons II , ill and V are air. The ou ter and inne r

radius of the dielectric ring and the radius of the

sauple are indicated a, b and c, ani the half height

of the dielectric ring and of the sample are

represented L and h, respectively. The shaded

portions show two metal plates attached to the ends

of dielectric ring.

In general, the electromagnetic field ~

sati sfies the fo llowing wave equat ion (1 ) on the

- 60 -

cylindrical coordinates with (r,B ,z) which is

derived from Maxwell's equations [2].

Since the relative pennittivity Ern of the

resonator is different in each regi on as (2), the

mode matching method of considering the higher

order resonance modes is used in the analysis to

obtain the resonant frequency.

1 (air) ( 2 )

[

El (sample)

Ern = 1 (air) ED (DielectricRing)

n=l n=2 n=3 n=4

1 (air) n-S

Hence, the magnetic field component of the

resonator is given in (3) as a swn of the component

of~, which is expressed by the dominant and

(a)

Oiele ctrc Ring

z

( b)

o =615.0mm 2a=199.9 2b= 40.0 2L= 56.3

Fig.! A dielectric ring resonator for moisture

content measurements. (a)The resonator used in

this paper. (b)The cross section of the resonator

when a sample is set in the center of the hollow

core.

higher order mode of the p-th TE1ll p (P= 1,2,3, ... ).

The subsc ript n is a number from 1 to 5 cor respond

to the region from I to V •

where w and Ito are the angular frequency and the

propagation constant of the electromagnetic wa.ve

in vacuwn. Rnp(r) and Znp( z) are radial and axial

func tions, respective ly, and

( 4 )

where {3 is the propagation constant in the z­np direction.

The other related electric and magnetic fields

are shown as follows:

( 5 )

[

Hnr= -4 t R~(r} Z~(Z) ko 1'-1

Hne- 0

(6)

where the mark of prime means the deri vative wi th

respect to the variable inside the parentheses.

From boundary conditions to these fields an

characteri stic equation is derived. The

theoretical resonant frequency can be obtained by

calculating the matrix deduced from the

characteristic equation. In this paper the 25x25

matr ix was used.

EVALUATIOO of CHARACTmISTIC ~UATI(J\I

To evaluate the characteristic equation we

compared the theoretical values wi th those

measured from dielectric samples with known

pennittivities. The measurement was perfonned with

- 61 -

r--- Directional ~

Sweep IGP-IB Computer

Coupler I---Oscillator -- (PC-9S0l)

I· I I i-

Scalar -0--

I I - ~ Weighing

I I c Network .....-oupler-l I I Coupler-2 Device

I I

o ielectric R" mg Resonator

Analyzer

Fig" 2 The block diagram of the resonant frequency

measurement system for a dielectric ring resonator

which is used as a new moisture content measurement system.

the system shown in the block diagram of Fig. 2. The

d ielect ric ring resonator is s timula ted by

microwaves from a sweep oscillator. The waves

detected with the coupler-2 are transmitted to the

scalar network analyzer and the resonant

frequencies are measured. The frequency is about

2120 MHz when no sample is inserted. The

t heoret ical resonance value is 2117.4 MHz.

The resonant frequency in experiments changes by

several megahertz due to the fluctuation of the

room temperature. Therefore, the difference of the

resonant frequencies before and after insertion of

a sample in the resonator is used as the resonant

frequency shi ft. In this way the effect of

temperature fluctuation is minimized. The values

3

~2 u.. <l

1

o

- Calculated

10 20 30 40 50 2h(mm)

Fig.3 Resonant frequency shi ft of ~ F as a

function of sample thickness.

of the shifL is negative in experiments, and in

Fig.3 its abso lute value is sho\o,'Tl as 6. F.

The data are obtained from samples of Teflon wi th

different diameters (2c) and thicJmesses (2h). The

solid lines represent the theoretical values based

on the characteristic eq..1ation and tre dark circles

indi ca te meas ured value s. I t can be seen tha. t Lh e

theoretical values agree well with the measured

ones. Therefore, the characteristic equation is

reliable to obtain the resonant frequencies of the

resonator. The discussion of the new moisture

cont ent measu rement me thad uti lizing a die lectri c

ring resonator are given in the following section.

~NANT ~NCY SHIFT and HOISTURt ~TENI'S

The relative pennittivi ty E of a moistened s sample is given by the Newton's role[3] as follows:

e. -e.o +11G (7 )

where E sO is the permittivity of the sample having

no moisture, '" is a coefficient for the

penni t ti vi ty depending on the sample dens i ty and G

is the moisture content at dry-basis.

As is shown in Eq.(7), a sample with different

moisture content can be replaced by a sample with

different pennittivity. Figure 4 shows the

theoretical resonant frequency shift of the

samples of different pennittivities as a function

of their thicknesses. I t is ev ident that the

resonant frequency shift AF is proportional to the

thickness when AF is less than about 3 MHz. As the

thickness is prop::>rtioral to the sample weight tone

shadd ex~t thlt the resonant frequency shi ft per

uni t we ight AFv is independent to the sample's

weight, that is, the quantity.

o 2 4 6 8 10 THICKNESS ( mm)

Fig.4 Theoretical resonant frequency shift of ~F

as a function of sample thickness.

- 62 -

Figure 5 shows the experimental results of

\' arious moisture contents taken from four

cardboard samples having di fferent thickness. In

t hi s exper iment it was pos sible to meas ure the

moisture contents up to 118 percent at dry basis

for various sample thicknesses. For high moisture

content measurements thin samples were used

whereas thick samples were used for low moisture

measurements. The data of all samples with

di fferent thicknesses are lined up as shown by the

dot ted line.

As is clearly indicated, one can measure the

moisture contents of cardboard from the

measurement of A Fv'

CCl'ICUJS 1005

A new method for lJDisture content mea9.lrement was

indicated. The theoretical resonant frequency and

the experimental results were analyzed and their

theoretical basis were given. There was a good

agreement between experimental results and

t heoret ical calculations. For the measurement of

other materials, it is necessary to examine the

resonant frequencies of the resonator with those

samples.

REFmEN(E)

[ 1] Y. Kobayashi and S. Tanaka, "Resonant Modes of

a Dielectric Rod Resonator Short-Circuited at

Both Ends by Parallel Conducting Plates", IEEE

Trans. Microwave Theory &. Tech., MTT-28 , 10,

pp. 10 77 -1085 I Oc t. 1980.

[2] D. Kajfez and P. Guillon, "Dielectric

Resonators" , pp. 72-73, Artech House, 1986.

[3] S. Oka and O. Tanaka, "Theory of Solid

Dielectric", pp.284-285, Iwanami, 1966.

01 -N J: ~

3.

2.0

~ 1.0 lJ...

<2

SAMPLE THICKNESS o 0.75 mm

[J 2.25

~ 3.75

• 5·30

o 20 40 60 80 100 120 MOISTURE CONTENT (-,. d.b.)

Fig.5 Resonant frequency shift per unit weight

~Fv as a function of mo isture content in four

cardboard samples (sample diameter = 35 nm).

- 63 -

DESIGN AND OPTIMIZATION OF ELECTROMAGNETIC SENSORS FOR DIELECTRIC SPECTROSCOPY BY USING THE (FD)2TD METHOD

G. Biffi Gentili, M. Leoncini Dep. of EI. Eng., Univ. of Florence, via S. Marta 3, 50139 Florence (I)

Abstract

D. Bertolini, G. Salvetti, E. Tombari IFAM-CNR, via del Giardino 7, 56127 Pisa (I)

A general method based on the Frequency-Dependent Finite­Difference Time-Domain solutions of Maxwell's equations is used to simulate Time Domain Reflectometry (TOR) measurements of the dielectric properties of dispersive/non dispersive materials.

A gaussian pulse is launched in a coaxial line which is terminated by a sample cell consisting of the electromagnetic sensor and the material under investigation. The reflected pulse is stored and the reflection coefficient determined.

The numerical approach proposed allows to get a deep insight in the analysis of the interaction between sensors and materials, so widening the range of application opportunities. In particular new sensors configurations for moist substances and matrix system containir.g water can be designed and optimized considering the dispersive dielectric behaviour of these substances.

Experimental TDR and frequency domain measurements have been carried out to validate the proposed approach and to test new sensors configurations.

Introduction The estimation of physical parameters of moist substances

and matrix systems trough microwave measurements represent a very stimulating topic in view of an extensive use of this approach in several fields as agriculture, building materials, cultural goods, industrial, biomedical and scientific application.

For each particular application a specific measuring circuit and sensor can be required, because of the different measuring environment and operational conditions.

The development of sensors for the indirect measure of the physical properties of moist substances and matrix systems in the aforementioned fields, is strongly conditioned on the availability of theoretical models of electromagnetic interaction between sensor and media under investigation.

Indeed the optimization of the sensors structure and the implementation of a straight calibration procedure can be more

- 64 -

easily performed trough a combined theoretical analysis and experimental characterization.

The response characteristic of the sensor embedded in the measuring environment is an observable quantity as the reflected (transmitted) signal in the time or frequency domain.

In the time domain the measurable quantity is a real, transient waveform as opposed to frequency domain where a complex steady­state waveform is measured, which expresses the resoonse to an alternating (AC) field at a specific frequency.

Measurement techniques have been ,developed for both domains, the translation from one to another being' possible via the Fourier transform.

Direct measurement in time domain, known as Time Domain Reflectometry (TOR) is a very effective method to perform high frequency dielectric spectroscopy, typically in the range 107 -1 010

Hz [1]. This broad band of frequencies is covered by a single measurement through the sampling in the time domain of a step­pulse reflected in a coaxial line terminated with a sample cell by the material which dielectric properties are under investigation. When the transfer function describing the electromagnetic sensor characteristics is known, it is then possible to extract the dielec~~ic properties of the material from the reflected pulse. In moist substances and porous media the water can be free and/or bound, exibiting in each phase a different dielectric response with the frequency. Therefore more insight about physical properties of such substances can be expected from a broad band dielectric characterization.

In this work the (FD)2TD method is used to simulate TOR measurements with the aim of designing new sensors configurations to fulfill a wide range of experimental requirements. The approach, first proposed by Yee in 1966 [2] for isotropic non dispersive media, is based on the finite difference approximation of Maxwell's equations both in space and time domain. More recently the formulation has been extended to include the frequency dependence of dispersive substances [3]. This last formulation known as Frequency-Dependent Finite-Difference Time-Domain ((FD)2TD), appears very appropriate because allows to take into account dielectric relaxation phenomena, provided the relaxation behaviour can be described by exponential time decay. The feed coaxial line and the sensor are modelled so as they are structured, while the dielectric properties of the material are assigned as input data. A gaussian electromagnetic pulse, superposition of TEM modes, is launched in the feed line at the input section, the refrected waveform is stored and the reffection coefficicent determined. This allows to simulate the sample cell response so having a deep insight of the sensor electromagnetic

- 65 -

r .............. R. F. Signal

Fig. 1a

Fig. 1b

Fig. 2

- 66 -

characte ri stics.

Modeling and experiments. I n order to test the performances of the method the

theoretical approach has been first applied to sensors commonly employed in dielectric spectroscopy. Different sensors configurations have been then theoretically analyzed and prototypes built and tested in laboratory.

In fig. 1 a,b the first two sensors configurations which have been simulated are schematically represented. Theoretically, the open-ended termination shown in fig. 1 a can be' considered as a complex capacitive load while the cell represented in fig. 1 b can be thought as an ideal section of an open circuit terminated coaxial line of lenght d and characteristic admittance Y filled with a sample of dielectric constant E [4]. The theoretical validation of the method has been carried out comparing the results obtained with the (FD)2TD simulation with those from analitical or numerical analysis which describes the load characteristics of the two sensors [5,6]. The analysis has been performed considering materials with a known dielectric behaviour, in a wide range of E

variations. The second part of the work concerns the (FD)2TD analysis and

the experimental verification of new sensors configurations which could be more suitable for the measurements of dielectric properties of material containing water. In fjg. 2 some of these configurations are shown. As appears from the figure a wide variety of sensors can be devised because of the indipendence of the proposed method from the geometrical structure of the sensor. A set of TDR and frequency domain experimental measurements, in a frequency range which extends up to 20 GHz, has been performed employing some sensors prototypes built in laboratory.

[1] R. H. Cole, "Time Domain RefJectometry,· Ann. Rev. Phys. Chern., 1977, vol. 28, pp. 283-300; [2] K. S. Yee, "Numerical solution of initial boundary value problems involving Maxwell's equations in isotropic media," IEEE Trans. Antennas Propagat., 1966, vol. AP-14, pp. 302-307; [3] R. Luebbers, F. P. Hunsberger, K. S. Kunz, R. B. Standler and M. Schneider, "A Frequency-Dependent Finite-Difference Time-Domain formulation for dispersive materials," IEEE Trans. on Electromagn. Compat., 1990, vol. EMC-32, pp. 222-227; [4] D. Bertolini, M. Cassettari, G. Salvetti, E. Tombari and S. Veronesi, "Time Domain Reflectometry to study the dielectric properties of liquids: some problems and solutions," Rev. Sci. Instrum., 1990, vol. 61, pp. 450-456. L5] E. C. Burdette, F. L. Cain and J. Seals, "In vivo probe measurement technique for determining dielectric properties at VHF through microwave frequencies," IEEE Trans. on Microwave Theory and Techniques, 1980, vol. MTT-28, pp. 414-426; [6] F. Gori and G. Biffi Gentili, "Open ended coaxial probe for soil moisture measurements," Internationat Microwave Symposium, 1987, July 27-30, Rio de Janeiro, Brasil

- 67 -

MODELING OF OPEN-ENDED COAXIAL LINE SENSORS FOR

DETERMINATION OF THE COMPLEX PERMITTIVITY OF MATERIALS AT

MICROW AVE AND MM WAVE FREQUENCIES

Ch:ris Pournaropoulos and Devendra Misra

Department of Electrical Engineering and Computer Science University of \Visconsin, Milwaukee, WI 53201

Non-destructive methods for measuring the complex permittivity of materials are desired in many areas of

science and engineering. An open-ended coaxial line sensor has attracted several researchers for this purpose

because of its applicability over a broau frequency range. In this method, the sensor is placed in contact with

material-under-test and its reflection coefficient (or admittance) is recorded. The permittivity of the material is

then calculated from a suitable model of the sensor. Some of these models are formulated empirically or on the

basis of static analysis of the coaxial opening and therefore, the error in the measurement increases at higher

frequencies (typically a few GHz) [1-3]. In another method reported to work at higher frequencies, nomograms

relating the measured reflection coefficient with the complex permittivity of material-under-teSt, are generated

at three different frequencies for a specific coaxial line [4]. Obviously, this technique has limited application. A

quasi-static admittance model and its modifications reported during past few years have been found satisfactory

in a frequency range of 1 to 20 GHz [5-7]. These models are formulated by approximating a variational

admittance expression for the coaxial opening [7]. In a recent study, we have used a rigorous formulation of the

aperture admittance which is evaluated numerically by the method of moments [8,9]. It is to be noted that this

rigorous formulation is not convenient for the characterization of materials in comparison with the variational

expression for the admittance. However, this study indicates that the variational formulation is fairly accurate

even in millimeter-wave band. Based on this formulation, we report in this paper a procedure for the electrical

characterization of materials in the frequency range of 1 to 40 GHz.

We have carried out a comparative study of various models over the frequency range of 1 to 40 GHz. In

one of the approximate models, it is assumed that the aperture admittance of coaxial line is a sum of two

terms, a static capacitance and a frequency dependent capacitance (model I). In this case, the first term is

linearly related with the complex permittivity of the terminating material while the second term has a

quadratic relation with it. The radiation from the coaxial opening is neglected in this case. In another

approach, the first term stays the same as before but the second term is replaced by a radiation conductance

(which is proportional to complex permittivity to the power of 2.5). In model III, all the three terms are

included. All of these models can be derived from the variational formulation mentioned above after suitable

- 68 -

approximations. These experimental and theoretical results are reported in this paper. Computational as well as

measurement aspects of the technique are discussed.

References

[1] T.W. Athey, M.A. Stuchly and S.S. Stuchly, "Measurement of radio frequency permittivity of biological

tissues with an open-ended coaxial line: Part I", IEEE Trans. Microwave Theory Techn., vol. MTT-30, pp. 82-

86, Jan. 1982.

[2] G.Gajda and S.S. Stuchly, "An equivalent circuit of an open-ended coaxial line", IEEE Trans. Instrum.

Meas., vol. IM-32, pp. 506-508, Dec. 1983.

[3] J.P. Grant, R.N. Clarke, G.T. Symm and N.M. Spyrou, "A critical study of the open-ended coaxial line

sensor technique for RF and microwave complex permittivity measurements", J .Phys.E: Sci.lnstrum., vol. 22,

pp. 757-770, 1989.

[4] J.R. Mosig. J.E. Besson, M. Gex-Fabry and F.E. Gardiol, "Reflection of an open-ended coaxial line and

application to nondestructive measurement of materials", IEEE Trans. Instrum. Meas., vol. IM-30, pp. 46-51,

March 1981.

[5] H. Zheng and C.E. Smith, "Permittivity measurements using a short open-ended coaxial line probe", IEEE

Microwave and Guided \~laves Letters, vol. 1, pp. 337-339, Nov. 1991.

[6] K. Staebell and D. Misra, "An experimental technique for in-vivo permittivity measurement of materials at

microwave frequencies", IEEE Trans. Microwave Theory Techn., vol. MTT-38, pp. 337-339, March 1990.

[7] D. Misra, M. Chabbra, B.R. Epstein, M.Mirotznik and K.R. Foster, "Noninvasive electrical characterization

of materials at microwave frequencies using an open-ended coaxial line: Test of an improved calibration

technique", IEEE Trans. Microwave Theory Techn., vol. MTT-38, pp. 8-14, Jan. 1990.

[8] S. Jenkins, A.W. Preece, T.E. Hodgetts, G.T. Symm, A.G.P. Warham and R.N. Clarke, "Comparison of

three numerical treatments for the open-ended coaxial line sensor", Electron. Letters, vol. 26, no. 4, pp. 234-

235, Feb. 1990.

[9] C. PoufLd.ropoulos and D. Misra, "A study on the coaxial aperture electromagnetic sensor and its appli­

cation in material -haracterization", IEEE Instrum. and Meas. Technology Conf., Irvine, CA, May 18-20, 1993.

Dr. Devendra Misra received B.Sc. and M.Sc. degrees in Electronic Engineering from Banaras Hindu

University, India, in 1971 and 1973, respectively, and Ph.D. in Electrical Engineering from Michigan State

University in 1984. He held research and teaching faculty positions in India during 1973-1980. Since August

1985 he has been with the University of Wisconsin - Milwaukee, where he is currently an Associate Professor in

Electrical Engineering Department. His interest include industrial and biomedical applications of electromag­

netic energy, microwave/mm-wave circuits, and numerical techniques in electromagnetics.

- 69 -

CALCULATION OF SENSITIVITY OF VARIOUS COAXIAL PROBES

USED IN MICROW A VE PERMITTIVITY MEASUREMENTS

Yansheng Xu and Renato G. Bosisio

Groupe de Recherches Avancees en Microondes et en Electronique Spatiale (POLY-GRAMES)

Departement de genie electrique et de genie informatique, Ecole Poly technique de Montreal,

C.P. 6079, succursale A, Montreal, Que., Canada H3C 3A 7.

The measurement ;~Jf microwave permittivity by using a coaxial line terminated by a circular waveguide or free space has been studied by many authors [1-13] and very wide band measurements have been achieved. However, the measurement sensitivity at low frequencies is often low and the accuracy become quite poor. In this paper we shall present a simple and effective method to improve the measurement sensitivity at low frequencies. Detailed numerical calculated data are presented which are very helpful in the selection of the probe parameters.

THEORY A simple and effective method to enhance the measurement sensitivity at low frequencies is to fill a

short section of the coaxial line inside the probe by the test sample material as shown in Fig. 1. The coaxial line in Fig. l(a), (b) and (c) is terminated by a circular waveguide (14] and in Fig. l(d) - by free space. The length of section "s" may be arbitrary, however, in our case we assume that the value of "s" equals one of three times the inner radius of the outer conductor of the coaxial line "a". It is clear that if "s" equals 0 then we have the ordinary probe described in the literature [1-14]. The coaxial line and circular waveguide operate in the principal modes and do not support propagation of higher order modes. It is also assumed that the outer radius of the inner conductor of the coaxial line "b" is small and amplitudes of higher order modes attenuate towards zero when they travel through distance "s". The calculation formula of the admittance at plane A for the cases shown in Fig. l(a), (b) and (c) is derived from a full wave analysis:

B A

(a) (b)

A

(e) (d)

Fig. 1. Some configurations of coaxial line probes for measuring the microwave permittivity

- 70 -

(1)

where Ya is the input a.dmittance (normalized) of the coaxial line a.t plane A, ! is the dielectric

constant of the test material and ko = 21r / .A,

where

(2)

The values of Xq can be shown to be the solution of the following equation which is a simplified

form of a more complex equation given in [2],

(3)

where .Aia (i=l, 2, ... ) are the ordered zeros of the Bessel function JO(Aia) and (na (n=1,2, ... )

are the ordered zeros of the mixed Bessel function JO((na)Yo((nb) - Jo((nb)}O((na) = 0 ,

and

Ai = 1

Ai = coth(,Bid)

for case Fig. 1 ( a)

for case Fig. 1 (b)

1 + (fO - f) e-2f3,d

A . - fo + f ,- ( ) . 1 _ fo - f e-2f3,d

fo + f for case Fig. 1 (c). It should be noted that A. and (n are real constants for fixed i and n.

(4)

(5)

(6)

The calculation formulas for static analysis are the same as equations (1) - (7) with the

substitution /3.-\ and (n-en, The calculation formula of the adlnittance at plane A for the

case shown in Fig. 1 (c) is:

- 71 -

Yo = ~kov: (oo[Jo(ra) - Jo(rbW dr . og b 10 I, (7)

where, = JI2 - k5f. The admittance at plane B may be obtained by using the formula as

follows:

Y _ c;-:/ Va + jtan(ks} - ViI tt-----..;...----:;....

1 + jVatan(ks) (8)

where k = k00' It should be pointed that here admittance Y is normalized to the coaxial line, filled by a material with dielectric constant It. .

NUMERICAL RESULTS In our calculation the normalized quantity d/a and koa are used to make the calculated curves universal

(that is, suitable for all cases with the same d/a and koa, and at the same time the values od a, d and ko or the frequency point may be different). The horizontal coordinate is selected to be the dielectric constant of the test sample ( because the slope of the calculated curves characterizes the measurement sensitivity. For simplicity we put alb = 2.303 and It = 1 in our calculations. Equations (1) - (8) show that in the static analysis Y T = Y/(j2koa) depends only on (a/b), (d/a), ks and (l/lt) for cases (a) - (c) and the frequency dependence exists only when l/(t changes with frequency. The numerical calculation of Y T for air-filled 50n termination, alb = 2.303 are shown in Fig. 2-5 for cases (b) and (c). Case (a) may be considered as a special case of (b) or (c) in which d-oo.

The data in Figs. 2 and 3 are calculated for cases koa = 0.001 and they may be considered as the static solution of. these cases. Both of them are obtained for ks = koa. They indicate that the admittance of the probe is increases when the thickness of the test material d/a decreases for case (b) and the admittance Y T

in case (c) decreases when d/a decreases. It can be seen also that Y T (which is proportional to the measure­ment sensitivity) increases significantly after adding section s when ( is small (increase up to 4 -10 times). Nevertheless, this factor of improvement becomes smaller when ( is large. This should not be a problem since the measurement sensitivity for high ( is generally much higher than for low ( case and we may further increase Y T for high ( case by using longer s if necessary. On the other hand, the resonance phenomena at high frequencies restrict the selection of small d and large s for case (b) - see Fig. 4 for koa = 0.3. ~feanwhile, the resonance effect does not take place for case (c) at high frequencies - see Fig. 5 for koa = 0.3. Therefore, it is of special interest to use this method of improvement for case (c). This method is also good for case (b) when the thickness d is not very small and we may use it to improve the measurement sensitivity without using too thin samples for measurement as suggested in [14]. The calculation results of case (d) - coaxial line terminated by a bulk of test material, are shown in Fig. 6 where the admittance of the probe Y is plotted against the dielectric constant of the test material (. It is clear that the measurement sensitivity is improved for all ( values of the test material. The improvement is more significant for low ( materials and can be enhanced further by using longer s.

In the case when ( is complex, the admittance Y becomes complex too. \Vhen E" < (', the condition satisfied in most practical cases, Y may be expanded into Taylor's series [14]:

y = y 1("=0 + aaY 1("=0 (-j fll) + .......

(=(, f (=(,

(9)

where f = f' - jf".

From Figs. 2, 3 and 5 it is clear that Y is close proportional to ( and hence the higher order differentials of Y with respect to ( are not large. It is e-nough to retain only one or two higher order terms in (9) and the calculation of complex Y may be performed without difficulty. The above calculation is performed with alb = 2.303, which corresponds to 50 n air-filled coaxial line. This results of calculation of 50 n coaxial line filled with dielectrics (with other ratios of a/b) show the same tendency of Y T'

- 72 -

40

:! 35

30 5 25

4 20 ~

~ 3 15

relative dielectric constant £, rel'alive dielectric constant c

Fig. 2. Case (b). Fig. 3. Case (c). Dependence of Y T on the dielectric const.ant of test material { with koa = 0.001 and ks = koa, alb = 2.303

200

150

100

50

r: 0

-50

-100

-150

-200 o ""'''''''''"c, 10 20 30 40 50 relbtive dielectric constant"£,

5

4

3

2

1

relative dielectric constant £,

CI.O~

- 73 -

CONCLUSIONS From the above numerical calculation results

we may conclude the following: 1. A simple and effective method of improvement

of the test sensitivity of microwave permittivity at low frequencies for coaxial probes is presented. The method is especially valuable for low ( test materials where the test sensitivity is usually quite low at low frequencies.

2. Some new coaxial probes based on. this method of improvement are demonstrated. Among them probe (c) is most suitable for wide bandwidth applications. Probe (b) shows also significant improvement for thick test samples with low l.

For probe (d) the improvement is obtained for all ( values of the test material and is most significant also for low microwave permittivity case.

0.15 r

~

0.1 f

>-

0.05

). For lossy materials the calculation may be performed by using Eq. (9) without difficulty.

, II!!' r It I!'! I!! J! t! t!! 11 "

REFERENCES

10 20 30 40 relative dielectric constant E.

Fig. 6. Dependence of Yon dielectric constant of the test material ( for case (d) with koa = 0.001 and alb = 2.303.

[1] N.E. Belhadj-Tahar and A. Fourrier-Lamer, "'Broadband analysis of a coaxial discontinuity used for dielectric measurements", IEEE Trans. Microwave Theory Techn., vol. MTT-34, pp. 346-350, 1986.

[2] N.E. Belhadj-Tahar, A. Fourrier-Lamer and H.Chanterac, "'Broadband simultaneous measurement of complex permittivity and permeability using a coaxial discontinuity", IEEE Trans. Microwave Theory Techn., vol. MTT-38, pp. 1-7, 1990.

[3] M.A. Stuchly and S.S. Stuchly, " Coaxial line reflection methods for measuring dielectric properties of biological substances at radio and microwave frequencies - a review", IEEE Trans. Instrum. Meas., vol. IM-29, pp. 176-183, 1980.

[4] B. Bianco, G.P. Drago, M. ~larchesi, C. Martini, G.S. Mela and S. Ridella, "Measurement of complex dielectric constant of human sera and erythrocytes", IEEE Trans. Instrum. Meas., vol. IM-28, pp. 290-295, 1979.

[5] T.\V. Athey, M.A. Stuchly and S.S. Stuchly, "Measurements of radio Frequency Permittivity of Biological Tissues with open-ended coaxial line: Part I", IEEE Trans. Microwave Theory Techn., vol. MTT-30, pp. 82-87, 1982.

[6] L.S. Anderson, G.B. Gajda and S.S. Stuchly, "Analysis of an open-ended coaxial line sensor in layered dielectrics", IEEE Trans. Instrum. :\feas., vol. IM-35, pp. 13-18, 1986.

[7] A. Kraszewski, M.A. Stuchly and S.S. Stuchly, "ANA calibration method for measurement of dielectric properties"', IEEE Trans. Instrum. Meas., vol. IM-32, pp. 385-387, 1983.

50

- 74 -

[8] A. Kraszewski, S.S. Stuchly, M.A. Stuchly and S.A. Symonds, "On the measurement accuracy of the tissue permittivity in-vivo", IEEE Trans. Instrum. Meas., vol. IM-32, pp.37-42, 1983.

[9] T.P. Marsland and S. Evans, "Dielectric measurements with an open-ended coaxial probe", Proc. lEE, vol. 134, pp. 341-349, 1987.

(10] J.R. Mosig, J.C.E. Besson, M.Gex-Fabry and F.E. Gardiol, "Reflection of an open-ended coaxial line and application to nondestructive measurements of materials", IEEE Trans. Instrum. Meas., vol. 1M-3D, pp. 46-51, 1981.

[11] D.K. Misra, "A quasi-static analysis of open-ended coaxial lines", IEEE Trans. Microwave Theory Techn., vol. MTT-35, pp. 925-928, 1987.

[12] D. Misra, M. Chabbra, B.R. Epstein, M. Mrotznik and K.R. Foster, "Noninvasive electrical characterization of materials at microwave frequencies using an open-ended coaxial line: Test of an improved calibration technique", IEEE Trans. Microwave Theory Techn., vol. MTT-38, pp. 8-13, 1990.

[13] K.F. Staebell and D. ~1isra, "An experimental technique for in vivo permittivity measurement of materials at microwave frequencies", IEEE Trans. Microwave Theory Techn., vol. MTT-38, pp. 337-339, 1990.

[14] Y. Xu and R.G. Bosisio, "Analysis of different coaxial discontinuities for microwave permittivity measurements", presented at CPEM'92 in Paris and accepted for publication in the special issue of IEEE Trans. on Instrumentation and Measurements.

Renato G. Bosisio was born in Italy in 1930. He received the B.Sc degree from McGill University, Montreal, PQ, Canada in 1951 and the M.S.E.E. degree from the University of Florida, Gainesville, in 1963. He has been engaged in microwave R&D work with various firms: Marconi and Varian in Canada, Sperry in the U.S., and English Electric in England. He is presently the Head of the Section d'Electromagnetisme et d'Hyperfrequences at Ecole Poly technique de Montreal, Montreal, Canada, where he teaches microwave theory and techniques. He is actively engaged in six-port technology, dielectric measurements and CAD of both active and passive microwave devices.

- 75 -

ACCURATE PERCENT WATER BY MICROWAVE INTERACTION ALONE

Charles W.E. Walker

PACIFIC AUTOMATION INSTRUMENTS LTD.,

11720 Voyageur Way, Unit 5, Richmond B.C., CANADA V6X 3G9

Microwaves provide an ideal means of measuring moisture in solids and liquids because of their

resonant interaction with water molecules, but there are three important complications which affect the

measurement and must be dealt with for accurate and consistent results.

The first important complication is that microwaves measure

mass H20/unit volume of material as measured

whereas moisture content in percentage is defined as

100 x mass H20/ unit ~ of material,

so that it is necessary either to assume a fixed value for the effer.-ive density which is defined as

mass of material/volume of material as measured

or to measure it.

It should be noted that the effective density 'g' of the material as it is measured is the true density

'p' of the material itself multiplied by its packing fraction 'I

g = pf.

Some people will claim strenuously that their material density is constant, which it may well be, but

the packing fraction, which is seldom constant, must not be overlooked.

One solution is to use a gamma gauge to measure effective density, but this means two

instruments with maybe two independent manufacturers. Also, gamma rays and microwaves have

drastically different wavelengths and do not 'see' identically the same sample of material and are

affected differently by the physical form and shape of the material as measured.

An ideal solution is for the water and the density to be measured by the same beam of

microwaves, which can be achieved by measuring both the phase change and the attenuation of the

signal in passing through the material. For the mathematical basis of this double measurement see

"Classical Electrodynamics" by Prof. J.D. Jackson, John Wiley & Sons, Inc., New York and London. ---------------~------~--------------~------~~----~

Phase change t/J = 211"...jli . t ~

1 2 L' ,( Power loss p= 2' E JI(2 C

in which t is the distance through the material, ,\ is the wavelength, JI is the frequency and E is the

electric field strength of the microwaves and (1 and (2 are the real and imaginary parts of the complex

- 76 -

dielectric constant of the material being measured.

The measured power attenuation in db, T], is then given in terms of the effective density g and

the water percentage W by:

and the measured phase change 1/J is given by:

log (to + 1) = g (10:~ W + K4W2)

in which K 1 to K 4 are const.ants. which are not simply interrelated. These are ,two linear equations in

x = 100 ~ Wand y = gW2 which may be solved by computer to give x and y and hence g and W.

The second important complication arises from other substances present in the measured material

which attenuate the microwaves or have a significant effect on the phase change. The most important

of these are ionic conductivity and free conductors such as uncombined metal atoms or Carbon. There

are very few cases where none of these are present. They can readily be allowed for in the microwave

measurement by using two microwave frequencies because, as shown by Maxwell's basic equations,

direct conductivity is not frequency dependent so that attenuation by the interfering conductors is the

same at both frequencies whereas the water absorption, being a resonant phenomenon is strongly

frequency dependent.

The ideal instrument for measuring water content in solid and liquid materials thus consists of

measuring simultaneously the attenuation at two microwave frequencies and the phase change at one of

them.

The third important complication is due to the fact that water will chemically combine with some

substances by Hydrogen bond and will surface bond to many solids with a bond strength greater than

the quantum energy of the microwaves. The result is that some of the water present is not measured by

the microwaves. This complication is best and simply taken care of by initial calibration tests.

Dr. Charles W.E. Walker graduated in Science in 1936 and received his M.S. degree in Nuclear

Physics in 1937 from the Cambridge University and Ph.D. degree in Solid State Physics from Simon

Fraser University in 1970. He started application of microwaves to industrial moisture measurements

(paper web on a papermaking machine) when with Beloit Corp., Beloit, Wisconsin in 1957. He owns

numerous patents on the subject and during the years applied microwaves to measure moisture in

several branches of industry (wood chips, plywood, concrete sand, foundry sand, coal, oil and Alberta

tar sands) in various countries around the world.

- 77 -

Possibilities and Limitations of Density-Independent i\loisture :\feasure­ments with l\ficrowaves K. Kupfer

Abstract

According to the literature a density compensation is possible for some materials within respective limits given by fonning the quotients of moisture and density dependent values of attenuation and phase shift. The complexity of water binding values, however, does not pennit to generalize these conditions. The range of validity must be checked anew for a density independent measurement in each particular case of application and thus also in the case of concrete aggregates. In conformity with practical tests the attenUation-phase shift diagram could be used as a mathematical model for a theoretical founded proof of independence on the densi~. With the diagram it \\'as possible to extend .the application range and the accuracy of a density-independent measurement.

Summary

Changes of bulk density are the main disturbances by using dielectric moisture measuring methods. The resulting measuring errors are bigger as those caused by interferences of salt portions at frequencies lower as 1 GHz and systematic errors of measuring device. Increasing of bulk density and layer thickness causes at the same moisture higher values of attenuation A and phase shift ct> (Fig.1a,b). It is simulated an elevation of moisture in the material. The calibrating curves of quotient Q=N4> does not exhibit such material changes (Fig.1e). By drawing the attenuation as a function of phase shift the calibrating curves, which are proportional to the not compressed (nc) and compressed (c) material, show linearity (Fig.2). The phasor leads with a deflnite angle 'V to the origin of the coordinate system by connecting of the same moisture values on both calibrating curves (nc) and (c). The tan '" is independent from bulk density and layer thickness but depends on the moisture in the material. It is defined by the quotient of attenuation and phase shift The A-ct>-diagram shows a model for a density independent moisture measurement With the diagram it was possibJe to extend the application range of a density independent moisture measurement to the moisture saturation range and to determine the periodicity of phase shift Fig.2 shows the A-cl>-diagram for gravel sand 0/2. Its calibrating curves are drawn in Fig.!. In this case it was possible to carry out density independent measurements in the range from 1 - 11 % moisture with an error s: 0,25 % moisture.

:,:,...----r----, -----,

Q ...1L. '"

~·P"'-'·'

c \IC

~ • t 1",1 I~-- --1 ~.:! 1-----i::IIAIiC:.----+------1

! c., ••• .,.&r • • ,lIaw

\IC QlIe tol' 'l.~ .• ~:., .. " ::.r'l c.,S -.i. Ie"" ~ . .n5 I ~I _ •

j A - c=- dio9WII

i ...... _111

A

•

•

----------------·---~i~

A • f (¢al III .!!r •• ,..- Weir

..... T.;. ~: ., ...

F",.. % .... tI.n ~..,t ~ C\n .. w9tft

difftnftt cI~ Ih .... ~ 1M

N"""'~. 0' cI..ny.

- 78 -

Coarse material with random grain fonns, grain sizes and grain distnbutions causes reflections and scatterings of electromagnetic field during the moisture measurement of aggregates for the concrete production. The values of attenuation increase at a coarser grain size. The slope of calibrating curve of the quotient is higher and the moisture range for a density independent moisture measurement is smaller (Fig.3). A possibility for a density independent moisture measurement a t a coarser grain size and a high moisture is drawn in Fig.4. By drawing all density values in the A-cI>-Diagramm the phasor is displaced to higher A-values. Considering only values with a middle up to rugh compression it is possible that the phasor Jeads to the origin of the coordinate system. Thats why a precompression has to be carried out to

enable a density independent measurement in the moisture saturation range. For gravel sand 0/8 it was possible to realize density independent moisture measurements in the range from 1,5 - 9,2 % moisture with an error !5i:: 0,25 % moisture (Fig.S). The material is homogenized by precompression. In addition it is posibble to extend the moisture range and the accuracy of a density independent measurement.

t 0 ...

9 • f (m) ....... II.

Go 10 ,., IS ~~ m ___ ~ Moishr. d.pendenf ~u!k-density-<1i!ferenc:e. at

= s~ .. sive Inc::-.cuing

I bl CcuDrcting curve. at a

I lulcuuive inc::-ecsing ot buUc density

~";"":'-'--~:;.5~-.-•• -~2S

¢-

1 2.5

Q .. f ( 1ft) I Q

9' ,o'Ql9let.,

~ roe

125

co~------~--%~--~'o~m---.--~:5

F"'tg. 3 hfll.ltnee of moishl".-dtp."d."t !:",ik-d."sify-CifferencH on tht quoti."t at various ~CIi'I size distri)utions

j 2.S

I Q 11 "m' ,- ""_t~ a.£!. ,..., .. 01.

raa

U~~----~~------~----~

Ql .0-"" ·eFl

a II C """ uc: CU. 0.1 11,'0" 0.;11 e ~ ~" !.J·te":mil

CO~----~~----~'O~--~%~~'5 m-

F"'sg. S Extension of rncisturt-rangt by pr.~ornpr.llion

The salts in aggregates are submitted solved pattern. In the microwave range polarization losses dominates compared with conductivity losses and influence the values of attenuation. Measuring in X­band \Vith a compressed gravel sand 0/2 it was shown that a maximum allowable salt content of 0,04 % does not influence on the measurements. Salt contents with higher concentrations in aggregates show the same results as compared with a temperature raising. In both cases the relaxation time is become smaller and the maximum of 2r" is

displaced to higher frequencies. The phasor in the A-<1'-diagram is shiftet to higher <l>-values. Dependend on frequency moist materials have different temperature coefficients. It is possible to minimize the temperature dependence at a certain temperature and moisture range by using a suitable measuring frequency. Measuring the moisture the temperature variation is specific for every material mixture. For compensation it is necessary to registrate the temperature coefficient The value of phase shift is relativly independent from salt content and temperature changes. In those cases there is the opportunity to applicate a phase measuring method.

References:

K. Kupfer: Detennination of moisture in aggregates for the concrete production by means of the microwave measuring t.ech!'jque.- Hochschule fur Architektur und Bauwesen Weim:ir~ Oiss. A (PhD),

- 79 -

ON THE PERMITTIVITY OF WOOD AND THE MEASUREMENT OF MOISTURE AND MASS PER AREA IN VENEER SHEETS

INTRODUCTION

Ebbe Nyfors, Matti Fischer, Pertti Vainikainen Radio Laboratory

Helsinki University of Technology

The permittivity of wood depends on the moisture, which makes the measurement of moisture by microwaves possible. The permittivity is, however, also affected by several other factors. The most important of these are the binding of water, temperature, density, ion conductivity, species of tree, moisture history of the wood, polarization, and frequency. To be able to measure the moisture accurately, the effect of these phenomena must be understood [1].

The binding of water causes the Debye-relaxation frequency to decrease. In wood the mean relaxation frequency fd is typically in the VHF or UHF range, whereas for free water it is higher than 9 GHz [2]. Because both the moisture and the temperature affect the mean strength of binding, both affect fd. The effect of the dry density is straight forward. The ion conductivity is important in the VHF range and below, causing E" to increase towards lower frequencies. Even some effect of the Maxwell-Wagner effect can be seen in £' at low frequencies. The species of tree has some extra effect on the permittivity not accounted for by the difference in density, but at least between Finnish birch trees and conifers the difference is small. The moisture history of the wood has a rather important effect. The permittivity of a sample of wood once dried and later remoistened is different from the permittivity of the same sample at the same moisture before drying. This hysteresis complicates research in the laboratory and the calibration of sensors. Because of the unisotropy of the structure of wood, the permittivity depends on the orientation of the electric field with respect to the grains. The maximum permittivity occurs with the electric field aligned with the grains. The effect of the phenomena above depends on the measurement frequency. Depending on whether the measurement frequency is higher or lower than fd, the effect offor example a change in temperature may cause the permittivity to either increase or decrease.

CHOICE OF FREQUENCY

The optimal frequency for a moisture sensor depends on the kind of sensor, what is expected from the sensor, and on the phenomena affecting the permittivity. Usually it is advisable to use a frequency rugh enough to avoid the ion conductivity. H inexpensive, non-contact meaSUl'ement of veneer sheets is demanded and the meaSUl'ement area should be tens of centimeters, a frequency just above the ion conductivity limit may be preferred.

MEASUREMENT OF MOISTURE AND DENSITY

Because the permittivity is a complex quantity (consists of a real and an imaginary part) two independent measurements can be performed with one sensor. That means e.g. measuring both the resonant frequency and the quality factor of a resonator or the phase shift and the attenuation of a transmission sensor. From the two measurements two quantities can be calculated, e.g. the moisture and the density or the mass per area. In many cases a simple way of calculating has been used. The value k = t':, I( c,. - 1) is a function of

- 80 -

moisture, but often almost independent of density [3]. From the moisture and e.g. the resonant frequency the dry density can be calculated. In wood the dependence ofk on moisture is, however, not monotonous. A maxjmum value is reached at about 10 % of moisture, but the dependence of k on moisture is also affected by the hysteresis mentioned above. Drying of the wood causes changes in the structure in such a way that the reintroduced water is distributed and bound in a different way. This effect depends on the wood species being e.g. stronger in spruce than birch. Experimental results will be given in the final paper.

SENSOR ARRAY FOR MOISTURE AND DENSITY MAPPING OF VENEER SHEETS

At the Radio Laboratory of the Helsinki UniversIty of Technology stripline resonator sensors have been developed for various applications. An array of sensors with two shaped center conductors, which are a quarter of a wavelength long, was developed for the fast mapping of moisture and dry mass per area of dried veneer sheets (Figure 1). The resonant frequency is 380 MHz and the resonant mode, which has the eletric field parallel to the veneer sheets, is used. The spot size is 10 em:x 30 em. The sensor is used for sorting the sheets according to moisture and density.

Figure 1. A stripline resonator sensor array for the fast mapping of the moisture and ma.ss per area of veneer sheets.

REFERENCES

[1] Nyfors, E., P. Vainikainen, "Industrial microwave sensors", Artech House, 1989, 350 p.

[2] Trapp, W., "Das Die1ectrische Verhalten von Hob und Zellulose im grossen Frequenz- und Temperaturbereich", Thesis, Technisthen Hoth schule, Braunschweig, 1954, SSp.

[3] Kent, M., "Complex permittivity of fish meal: A general discussion of temperature, density and moisture dependence", J. Microwave Power, Vol. 12, No 4, Dec. 1977, pp. 341-345.

- 81 -

EXPERIENCE WITH THE MICROWAVE MOISTURE METER

"MICRO-MOIST"

Albert Klein

Laboratorium Prof. Dr. Berthold, 7547 Wildbad, Germany

For more than three years, the microwave moisture meter "Micro-Moist" has been

on the market. It is based on a transmission measurement at approximately 3 GHz with

attenuation and phase shift measurements. The instrument allows density compensation

with an l,dditional gamma-ray transmission measurement and by a combination of

microwave attenuation and phase shift measurements. The instrument can be installed

at conveyor belts by using horn or spiral antennas. A standard chute is also available

which is used for non-pasty bulk materials such as wheat, dried lignite or artificial

fertilizer. Furthermore, there is a flowcell with a two-inch pipe designed for diary

products such as cream cheese or process cheese. Accuracies of 0.3 % moisture are

achieved with these materials with up to 85 % moisture content. A similar flowcell is

used in combination with a small screw conveyor, which can be installed at a transfer

station of two conveyor belts for realizing a bypass. The accuracy of measurements in

inhomogeneous materials such as clay lumps or the filter cake of, e.g., a coal slurry after

a vacuum-filter is much better.

Using these applications, the possibilities and limitations of microwave moisture

content measurements will be discussed. In this context the problems of disturbing

parameters will be explained, as for example, particle size, bound water, and salt

content. Last but not least, because of the long-time experience with development of

this instrument, economical ways to calibrate the instrument will be reported.

Dr. Albert Klein was born in Germany in 1946. His Ph.D. thesis in 1978 was on the

microwave properties of moist coal. He was a Research Engineer at Bergbau-Forschung

in Essen, Germany, from 1978 to 1986. Since 1986 he has been with Laboratorium Prof.

Dr. Berthold in Wildbad, as a manager of the microwave development.

- 83 -

DETERMINATION OF WATER CONTENT IN OIL PIPELINES

USING HIGH FREQUENCIES

S.K. Aggarwal

Institute of Paper Technology

University of Roorkee

Saharanpur, India

and R.H. Johnston

Departmen t 0 ineering

The University of Calgary

Calgary, Alberta, Canada

Monitoring of water in crude oil and in various petroleum products is required for a number of reasons.

If the water content in the fluid produced by oil wells can be determined, then enhanced oil recovery

techniques can be used to maximum advantage, as well as being of benefit to an oil company for production

control and accounting purposes. Also during the refining process, it is necessary to control the water level of

the crude oil since water containing various salts etc. may have adverse effects on the process equipment.

The difference in dielectric properties of water and oil is well known and understood and these properties

have been used by earlier researchers for determining water content in oil-water mixtures [1, 2] in mixtures

with water content. of up tp 10%.

The method described in this paper also makes use of the difference in the dielectric properties of water

and oil to determine the water content of oil. The attenuation and phase shift of an EM wave at 500 MHz

traveling through the fluid mixture are measured. A specially designed applicator carries the electromagnetic

wave in a coaxial transmission line likestructure through the flowing fluid. The attenuation and phase data

can be used to determine the water content and salinity at given temperature.

Measurement System

The measurement system required to perform the measurements is simple and is shown in Fig. 1. It

consists of a signal generator, which produces an EM wave at 500 MHz. The EM wave from the signal

generator travels to a signal splitter, which divides the signal into two separate paths. One of these paths

contains the applicator, and the signal, after being attenuated and phase shifted in the applicator, travels to

a vector voltmeter through a coaxial Tee. The other path contains a reference coaxial line (approximately

the same electric length as the main line containing the applicator). The vector voltmeter measures the

attenuation and phase shift of the EM wave traveling through the fluid mixture relative to the EM wave

traveling through the reference line.

- 84 -

COAX TEE AND TERMINATION

APPLICATOR J

MEASURED SIGNAL

SIGNAL SIGNAL

VECTOR GENERATOR VOLTMETER SPLITTER HP8654B 10.. REFERENCE

HP8405A

SIGNAL

VARIABLE LENGTH I COAXIAL

LINE COAX TEE AND TERMINATION

Fig. 1. Instrumentation used for measurement.

Applicator. The applicator is constructed from a forged steel cross which provides the basic structure that

is needed to withstand the pressure and the temperature of the wat.er/oil mixtures. The two outlets of the

cross that carry the mixture are unaltered, while the other two outlets are converted into a transmission line

structure to carry the electrical signal. Details of the applicator are shown in Fig. 2.

In actual two phase flow systems, it is well known that fluids may separate from each other. In a

pipeline carrying heavy oil and '\'ater, the oil can coat the pipe (probably due to a greater viscosity and

possibly a greater tendency to wet steel pipe than water). The oil will thus travel more slowly than the water

due to its greater viscosity and due to the location in the cross-sectional area that it occupies. The average

velocity of the water is greater than that of oil and the velocity differential will cause an error in translating

the fractional volume measurement to a fractional flow rate reading. If the water and oil are thoroughly

mixed in the form of small droplets, then two fluids should have the same average flow rates even when the

fluid velocity is not const.ant across the pipe cross section. Therefore, static fluid mixers are placed upstream

of the cross to reduce the above source of error. The ~nput and output pipes of the applicator have flanges

attached to facilitate its connection to a measurement or pipeline system.

Electrical Analysis

To complete the electrical analysis of the applicator it has been represented by a series of coaxial line

sections c.ascaded together. Exact knowledge of characteristic impedances, Zo' propagation constant 'Y and

lengths of respective sections is needed to calculate the voltage transmission coefficient for the coaxial line

applicator filled with watet/oil mixture. In order to find the values of discontinuities and propagation

constant in the section filled with the mixture (measurement chamber of the applicator), the relative

permittivity of the water and oil mixture must be known. A number of theories have been presented for the

permit "ity of dielectric mixtures of different geometrical configurations [4, 5] but not many theories are

available for predicting the permittivity of a mixture of liquid dielectrics when one of the components is

highly lossy. Ramu et al. [6, 7] have presented the following expression for calculating the complex

permittivity (; of liquid dielectric mixture and they report that it is accurate, even when the fractional 1,., ."

where

.' I , e =C' -J'CO -r f;..r ~r-

- 85 -

(k 1 k3 + k2 k 4 ) e; - (k2 k3 - kl k 4 ) e;' k} + kl

(k 2 k 3 - k 1 k 4 ) e; + (k 1 k 3 + k 2 k 4 ) e;' -J

k} + kl

k 1 = 2 e; (1 - V2 ) + ~ (1 + 2 v2 )

k 2. = 2 e~' (1 - V 2 ) + ~' (1 + 2 v 2 )

k 3 = e~ (2 + V2 ) + ~ (1 - V2 )

and k4 = e;' (2 + Y2 ) + e~' (1 - v2 )

(3)

(4)

(5)

(6)

(7)

where f} and t'{ represent the dielectric constant and the loss factor respectively, for the continuum fluid,

and t2 and t2 are the dielectric constant and the loss factor, respectively, for the dispersed fluid. v2

represents the fractional volume of the dispersed fluid in the continuum fluid.

Assuming that the oil is the continuum and the water is the dispersed fluid, respectively, Eqns (3) and

(7) can be modified to give

where

• enn = (k 1 k 3 + k 2 k 4 ) ero

k} + kl kl + kl

k 1 = 2 ero (1 -Y) + ero (1 + 2 v)

, , k 2 = ero (1 + 2 V)

k 3 = ero (2 + v) + e;o (1 - v)

and k 4 = e;o' (1 - v) .

(8)

(9)

(10)

(11)

(12)

By substituting values int.o Eqn.(8) and then calculating the propagation constant in the mixture, one may

calculate the phase shift of a wave traveling through a water/oil mixture as a function of the fractional

volume of the water cont.ent. The result is shown in Fig. 3, assuming that oil is the host material.

Results

Measurements were made with pure water/oil and salt water/oil mixtures. The concentration of sodium

chloride in salt water was 0.15 mol/I. A series of measurements were made at three temperatures, i.e. 20°C,

500e and 90oe. The results are shown for the measurements made at 50°C.

The phase shift data from four measurement runs using pure water and oil at 50°C are shown in Fig. 4.

The solid line represents a best fit including all runs. The data are well behaved at this temperature (and at

900e), At 200e shown a significantly larger scatter of ± 7%. Almost every point lies within ± 3% of the solid

ASBESTOS GASKET

Fig. 2. An exploded view of the applicator.

Dimensions are in mm.

-160....----------------,

U) LIJ

-140

-120

~ -100 C) LtJ o -.... -80 LL 3: U)

~ -60 c:t x Cl.

-40

----THEORETiCAL RESULTS • 100·4- 0% WATER • 0 ".-100 %

WATER • O%-IOO-JoWATER A 100 °4- 0% WATER

CO~~~-~~-~~~-~~-I-OO 20 40 60 80

WATER (". VOLUME)

Fig. 4. \Vithout salt

- 86 -

-160..-----------------,

-120

-40

o 20 40 WATER (".

Fig. 3. Theoretical phase shift of a wave in an oil/water

mixture, (* =79.95 - ;'2.2; fr*o=2.55; 1=34.22 mm. rw .'

.... !:: x U)

LIJ U)

c:t x Cl.

-160---------------.

-80

-60

-20

---- THEORETICAL RESULTS • 100%-0% WATER • 0%- 100% WATER • 0%-100% WATER ... 100 0;.- 0". WATER

20 40 60 eo WATER (% VOLUME)

Fig. 5. \Vith 0.15 mol/l of NaCl.

100

The phase shift in mixtures of water and oil at 50°C. The results are normalized with respect t.o 100% oil in the

applicator. Runs shown by cirdes and diamonds are performed together; runs shown with squares and triangles

- 87 -

line sketched in Fig. 4. The phase shift as a funct ion of salt water content at 500e is shown in Fig. 5. The

solid line represents the best fit of the points from four measurement runs. The measurement results with

salt water at 900e are good without significant ::-catter, but the results at 200e show a ± 2% scatter.

An illustration of the attenuation data for salt water/oil mixtures are plotted in Fig. 6. \\'hile the

deviation of the measured points from the solid line is substantial at 20 °e, the data are well behaved at 50°C

and 90°C. The attenuation for salt water is large compared to the attenuation in pure wat.er mixtures and

increases with temperature and salinity. The phase shift data for pure and for salt water mixtures at 500

e are plotted in Fig. i. It is also found that if a single phase shift curve is chosen for all data points in Fig. 7,

the error in measuring the water content from the phase data would be less than ± 5% in most cases.

IS ---- THEORETICAL RESULTS I 15 • 100%-0-/0 WATER I

14 • 0%-100% WATER A 0%-100'-0 WATER I

13 100%-0-/. WATER I • 12 I II I

J ~ 10 J '1:1

z Q f-<l ::::> z w f-f-<l

9 + 8

, 7

7 6 I 5 /

/ 4 . / 3 / 2

/ /

A / A

0 A It./ -I L--L __ ~~~~~~~~~~~~

o 10 20 :30 40 50 60 70 80 90 100 WATER (% VOLUME)

Fig. 6. The attenuation in mixtures of salt water

(0.15 molll of NaCI) and oil at 50°C.

Conc1usions

-160

-140

-120

en w ~ -100 CI w 0

... -80 ~

::x:: en w -60 en <l ::x:: a..

-40

-20

WITHOUT WITH SALT SALT

• 0 100%-0% WATER

• ¢ 0%-100% WATER

• c 0-4-100"0 WATER A ~ 00%-0% Wt,TER

to 20 30 40 50 60 70 80 90 100 WATER (% VOLUME)

Fig.i. The phase shift in mixtures of pure water and

oiJ, and salt water (0.15 mol/I) and oil at 50°C

A meter for measuring t.he wat.er content of pipeline carrying water and oil has been developed and

tested> Measurement of water and salt content of the mixture can be made by measuring the phase shift

and the attenuation of the EM wave passing t.hrough the mixt.ure of oil and water. If the salinity of the

mixture is not required, the phase shift data alone can be used to determine the water content of the mixture

with slightly reduced accuracy. It is to be noted that at al1 temperatures the curves are steep at the high

water content and the measured points lie very close to the solid lines, thus making the measurement

technique potentially accurate at high water levels. It is also apparent that the repeatability of the system is

especially good at high temperat.ures, which may be due to the absence of oil/water emulsions in the test

apparat.us. The applicator is compact and easy to integratf' ,.~:;th E'xistin". l)iDeJ:n,~ '''(LI:~'Tfi-:.n~,.

- 88 -

An important feature of the applicator is the apparent behaviour where oil acts as the continuous phase

even at low oil content. It is desirable to pick applicator conductor metals that ensure that this condition

always occurs. Thus the metallic parts of the applicator should be constructed of materials that are

oleophilic and hydrophobic.

REFERENCES

1. G.S.P.Castie and 1. Roberts, "A microwave instrument for the continuous monitoring of the water content of crude oil", Proceedings of the lEE, Vol. 62, No.1, pp. 103-108, January 1974.

2. D.A. Doughty, "Determination of water in oil emulsions by a microwave resonance procedure", Analytical Chemistry, Vol. 49, No.6, pp. 690-694, May 1977.

3. P.l Somolo, "Calculating coaxial transmission-line capacitances", IEEE Trans. on Microwave Theory and Techniques., p. 454, September 1963.

4. L.K.H. van Beek, "Dielectric behaviour of heterogeneous systems", Progress in Dielectrics 7, pp. 69-114, 1967.

5. W.R. Tinga, W.A.G. Voss, and D.F. Blossey, "Generalized approach to dielectric mixture theory", Journal of Applied Physics, Vol. 44, No.9, pp. 3897-3902, September 1973.

6. Y.N. Rao and T.S. Ramu, "Determination of the pennittivity and loss factor of mixture of liquid dielectrics", IEEE Trans. on Electrical Insulation, Vol. EI-7, No. 4, pp. 195-199, December 1972.

7. T.S. Ramu and Y.N. Rao, "On the evaluation and conductivity of mixtures of liquid dielectrics", IEEE Trans. on Electrical Insulation, Vol. EI-8, No.2, pp. 55-60, June 1 Q73.

- 89 -

Continuous mIcrowave moisture measurement for fluids and particulate materials.

V.Nguyen S FlREE Yang Shen

Deakin Ur., versity Victoria 3217 Australia

1. INTRODUCTION

There is a considerable interest in having a moisture meter incorporating a range of options for stand alone and remote measuring applications. There has been some moisture meters on the market using the microwave absorption characteristics of materials. In one standard system, microwave energy in the X-band is transmitted via a horn through the material to be detected by another horn. The meter is material specific ie it has to be calibrated with the material to be measured before it can be used. Recently, Meyer and Schilz [1], investigated the density related problem associated with the measuring technique and seemed to overcome it by measuring both amplitude and phase· of a microwave signal passing through a sample. The calibration problem, however, still remains. The major advantage of a microwave moisture sensor compared to a conventional method is its speed. In the design discussed below, a microwave signal at (4Gl:G) is used with a TEM cell as a contin uous moisture sensor to be tiseaa's a stand alone or a continuous meter and for local or remote measurements.

2. PRINCIPLE OF THE MOISTURE :METER

The meter is intended to be used in a network for the recording of moisture contents of liquids or particulate materials. For instance the grain authority may wish to obtain the mois!UIe levels of wheat in different silos scattered throughout a state or country. The meter sampler is a TEM cell which is open at the top and bottom to enable a material to flow through or which temporarily closed at the bottom to enable a batch measurement. The bottom plate is opened or closed by an user's command. The TEM cell has one movable side plate which is controlled automatically by the microprocessor via a stepper

- 90 -

motor. The plate is moved according to the detected signal to a position corresponding to the maximum. It should be appreciated that different materials have different dielectric properties, hence the moving plate enables the adjustment of characteristic impedance of the TEM cell.

The microwave signal is generated by a bipolar oscillator using surface mount technology and printeq circuit configuration. It is equipped with a power divider, one output feeding the cell and the other supplies a reference signal to the balanced mixer. A dual directional coupler is also available to measure the forward and reflected signal.

At the output of the TEM cell there is a further dual directional coupler to detect the transmitted and reflected signals. A sample of the transmitted signal is coupled to the mixer for calculating the phase variation through different samples.

3. :MICROPROCESSOR CONlROL

The microprocessor controls the operation of the moisture meter' as shown in Fig.l. It has ROM on board containing look-up tables of specific products to enable the computation of moisture contents. The microprocessor controls the bottom gate, the moving plate for tuning, data acquisition at the input and output of the TEM cell, prepare an estimate for the moisture content from a look-up table, output the result to different ports such as RS232, Centronics, LCD display and the built in printer for a local hard copy. The microprocessor and its peripherals are mounted on a PCB with a standard STD bus which includes a CPU, ROM, RAM, AID converter, timer and I/O ports. A key pad and an LCD display provide a local interface with the microprocessor. This option provides an on-site operator with the status of measurement, result, option prompts, date and time. When used in the stand alone mode, the commands are generated from the key pad. The built-in printer provides a hard copy of the measured result which includes sample type, operator entered sample code, date, time and moisture content.

In applications that do not require great accuracy, options such as plate tuning and/or the phase shift correction may be omitted.

The program controlling the meter is written in the assembly language and stored in EPROM.

- 91 -

Display

Keypad

Fig. 1. Block diagram of the microwave moisture meter

Q)

> (l)

0.8

0.6

~ 0.4

-0 Q) (J)

.~ 0.2

E L

o Z 0.0

10

.?:-5 :.0 o

.D o ...

a..

Reflection

iii i , i I I ( ii' iii i , Ii J iii iii i I a 20 40 60 80

Step positions of the plate Fig.2. Nomalised output and reflection from the TEM cell

O~MTnT~~~~TMTMTrrTrrTrr~nTMT~~Tn~nTnT~Tn

0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82

Normolised voltoge level

Fig.3. The variation of the output

- 92 -

The command set includes: a) Select material to be measured and carry out a measurement b) Print the result c) Transfer data to a remote host d) Switch between remote and stand-alone mode e) Set time and date f) Calibrate the meter for a specific product

4. RESULTS

Fig. 2 shows the normalised voltage level for the transmitted and reflected signal when the moving plate is shifted over its range. It can be seen that the measurement contains only small variations.

Fig. 3 shows a typical measurement of onion flakes. The variation of the normalised output is within ±2.5%.

5. CONCLUSION

A microwave moisture meter has been developed after the dielectric data of a range of products have been measured. The microprocessor provides a number of flexible options. The meter can be used in the stand-alone mode or in the remote mode for either a single sample or a continuous flow of material. The movable plate improve the measurement accuracy. The, phase shift option can also help but so far it has been found to be effective only over a narrow range of moisture content.

6. REFERENCES

[1] Meyer, W. and Schilz, W.M. 'Feasibility study of density independent moisture measurement with microwaves', IEEE Trans. MTT. Vo1.29, July 1981,pp.732-739.

7. ACKNOWLEDG.MENT

The authors wish to thank the Rice Growers Cooperation and Deakin ' University for the supporting grants.

- 93 -

PRESENT POSSIBILITIES FOR 1\10ISTURE l\lEASUREl\IENTS

BY 1\IICROWAVES

Ei~ B. Dujardin*, Y. Leroy*, Y. Vincent**, G. Mallick**

* Institut d'Electronique et de MicroeJectronique du Nord

Departement Hyperfrequences et Semiconducteurs - UMR CNRS N° 9929

Universite des Sciences et Technologies de Lille

59655 VILLENEUVE D'ASCQ CEDEX - FRANCE

** France Maintenance Automatisme - B.P. 210

54506 V ANDOEUVRE - FRANCE

INTRODUCTION

Recently, an increased interest has been shown in moisture

measurements on account of his great importance in a lot of industrial processes.

A few situations met in the industry have been experimented and have shown

that microwave measurement (amplitude/phase of the reflection/transmission

coefficient) is a serious candidate to the solution of these problems.

THEMEASURE~NTSANDTHESENSOR

Measurements at 2.45 GHz and 10 GHz have been made on the

following materials, with different moisture contents H (%).

- stone work sand (0 < H < 8)

- explosives (43 < H < 46)

- textiles (H < 80)

- barley (10 < H < 66)

- pharmaceutical products (H < 12)

- urban muds (70 < H < 82)

- shavings ofwood(21 < H < 46)

- fireproofmaterials (21 < H < 60)

- 94 -

Note that such methods can also been used for other non 1estructive

controls such as the localization of defects and the measurement ot .• e grain in

wood, density measurement of food materials, quality control of bricks, detection

of bubbles in plastic materials.

At the present time, a lot of laboratory studies have shown the

possibility to build measurement systems, devoted to these applications, much

more simple than automatic network analyzers [1, 2J ; and able to fulfill the low

cost requirements of microwave sensors. One of these investigation, based on an

interferometer associated to a calibrated impedance, described in [IJ, has been

tested on different materials such as sugar, sand and textile powders. A system,

working at 2.45 '-'Hz has shown to exhibit a reproductibility in the measurements

better than 10-2, and sometimes near of 10-3, in the case of dry materials (one-line

measurements on caster sugar).

Among the different antennas which are used, the rectangular

waveguide apertures can be modelized with respect to the amplitudes/phases of

the reflection/transmission coefficients. This method, based on the computation of

the near field radiated in homogeneous or multilayered materials [3] can also be

at the origin of in-situ permittivity measurements.

This work is supported by the Ministere de la Recherche et de

]'Espace. Decision d'aide n° 92 B 0274.

References:

\W-i Lasri, B. Dujardin, Y. Leroy "Microwave sensor for moisture measurements k in solid materials" lEE Proc.-H, Vol. 138, n° 5,1991, pp. 481-483.

[2] M. Craman, J. David, R. Crampagnes "Mesures de constantes dielectriques complexes it I'aide d'un dispositif autonome utilisant un calculateur compatible P.C". Revue de Physique Appliquee, Mai 1990, pp. 469-474.

[3] A. Mamouni, Y. Leroy, B. Rocquet, J.C. Van de Velde, Ph. Gelin "Computation of near field microwave radiometric signals: definition and experimental verification". IEEE Trans. M.T.T., Vol. 39, n° 1, 1991, pp.124-132.

D T . LASRI received the M.Sc. degree in Electrical Engineering in 1988 and the "thbe d'~ni\'~~" in 1992 from the tTniversite des Sciences et Techniques .de Lil1e, Fra~ce. ;He is p~esent1y an Assistant Professor in this University in the "Institut d'Electromqu.e et de Mlcroelectromque .du Nord" (UMR 9929 CNRS). His interest is in. th~ stud~' of short-r~n~e mlcrowave sensors, constructIon of devices a'1d modelintr of the elertrornav.: -ur sIgnals In k~::v .\·~l !als.

- 95 -

single and Multi-Frequency Phase Change Methods for Microwave Moisture Measurement

Ashley Robinson and Marek Bialkowski Department of Electrical and Computer Engineering

University of Queensland Queensland Australia 4072

Summary:

Previous work on microwave moisture measurement has shown that the moisture cont-ent of a material is proportional to the square of the relative permi tti vi ty. From this relationship the moisture content can be calculated from the phase change of a transmitted signal that has been normalised by the mass per area. For a single frequency, the problem with measuring the phase change is that it is confined between 0 and 2" which greatly limits the possible range of moisture contents measurable and/or the mass per area range. This problem can however be overcome by the continuous monitoring of the phase change of a single frequency signal or by using a multi­frequency system.

This paper will describe three methods of measuring the phase change without this confine.

The first of these methods is to simply measure the phase change of the transmitted signal at a single frequency and then continuously monitor the changes in phase as the moisture content and mass per area changes.

The second method is to measure the variation in the phase change over a frequency step or sweep. When the difference in the phase change is divided by the frequency shift, then a value proportional to the phase change is obtained.

The third method uses the periodic nature of the phase change as the frequency changes. Over a frequency sweep the measured phase has a sinusoidal nature with respect to the signal frequency. If the frequency of this sinusoid, as calculated using an FFT or other more accurate signal processing techniques, is divided by the range of the frequency sweep, then a value proportional to the phase change is obtained.

The three methods described are tested using simple computer simulations of calculating the relative permittivity and experimentally on beach sand.

To measure the phase change of the transmitted signal either a quadrature mixer or a well calibrated six port network analyser could be used. The paper will also present details on a calibration technique for a real time display six port which is used in moisture measurements.

- 97 -

Bulk Moisture Measurement of Agro-based Products by

Estimating Phase Variation at Microwave Frequency

H.Singh, S.Shekhar and C.H.Shah Central Electronics Engineering Research Institute

PILANI (Rajasthan) India.

Requirement of real-time continuous moisture meters for

agro-based products especially grains, fibres etc. are problems today. The

authors primary interest are in measurement of moisture of mill wet bagasse

coming out of canesugar mills. The monitoring and control of moisture in

bagasse is important from the point of its fuel value for the bagassed-based

furnaces as well as indicator of the parameters of sugarcane milling

performance. Another moisture measuring application of interest to the

authors is monitoring the moisture of grain being continuously loaded from

silos to trucks or being unloaded from trucks to silos. Both these

measurements require real-time continuous moisture measuring instruments.

The specific moisture range in the case of bagasse is from 40 to 50% and

that for the grain 1 to 10%. The accuracy requirement is within 2% of the

full range. A study of various techniques for moisture measurement has

shown that microwave methods stand out amongst them for the envisaged

continuous real time application [1]. Technique of microwave measurement

and its advantages have been described by many research works [2-4]. See

also cross refernces. But still no satisfactory solution for the above

- 98 -

envisaged application is available in view of the following reasons.

Due to variations in grain/fibre size, packing density, shape

and finish of surface, void amongst grains/fibres and continuous motions of

the material there is large variations in amplitude of the transmitted

microwave at the detector. The miss-matching of impedance of the

inhomogenious transmission media results in irregular reflection and

absorption when microwave passes through it. As the grain/fibre is in

movement, the amount of reflections and absorption become random. Thus

any measurement of the strength of the deteted microwave is unlikely to

indicate with some accuracy the moisture content in the transmitted

medium formed of grain/fibre. The authors are of the view that phase lag

measurement may give better indication of moisture compared to

attenuation in such cases.

Fig. 1 shows the proposed method for measuring phase lag

from three or more angles of the incident wave after passing through the

transmitted medium. Fig. 1 (a) schematically depicts the problem faced in

accurate measurement of attenuation due to grain size variations,surface

refl~ctions, impedance mismatch due to irregular void space and

inhomogenious transmission medium. Fig. 1 (b) shows an approximate

model to start with. The reflections occuring at interface of grain and air

may be internally absorbed also in the transmission medium but due to

frequent changes in the impedance mismatch the absorption is also

irregular. However the phase lag does not seems to be so much irregular.

I I I

- 99 -

I I I.

----~~~~~~'~~w~~-~P8~8~

SOURCE ~~i;L-----. COHERENT PHASE DETECTOR

SOURCE POINT 1

SOURCE POINT 2

SOURCE POINT 3

, , ,

,

,

I I I , I I

I I I

(a) VERTICAL SECTION

(b) APPROXIMA TE MODEL

• 0 , ,

, , ,

;

FUllY FILLED SPACE WITH GRAIN

VACANT SPACE WITH AIR

(OHERENT PHASE DETEOTOR

/ / ; ;

; / / ;

/ ; ; /

; /

(e) ELIMINATION OF EFFECT OF GRAIN SIZE AND SURFACE CONTOUR

FIG. 1 COHERENT PHASE - BASED MICROWA VE METHOD FOR MOIS TURE MEASUREMENT OF AGRI(UL rURAL

- 100 -

If one is able to measure phase lag from three or more angles

as shown in Fig. 1 (c) by not actually moving the source and delecter but by

taking continuous sample and passing it through a suitable mechanically

rotatable sample holder, one can get better estimate of moisture content

after processing the coherently detected phase lags/per unit length. A

compromise in design of the mechanical dimensions of the sample holder

depending upon grain size and choice of microwave frequency has to be

made. To minimise the effect of grain/fibre size on the measurement

accuracy one can choose higher frequency for larger size grains but to

measure phase lag accurately one has to choose lower frequency. It would

be noted that specific moisture measurement data proposed for the

envisaged applications would is only for quasi-equalibrium state of the

moisture.

References

1. C.H. Shah and H. Singh, "General Moisture and Humidity Measurement Techniques and on-line moisture measurement of Agro-based Products" Research Reports, CEERI/IEA/RR-23/Aug.1991, CEERI, PHani (Rajasthan) India.

2. A.W. Kraszewski, "MicrowaveAcquametry-NeedsandPerspectives· IEEE Trans. on MTT Vol. 39, No.5, pp 828-835, May,1991.

3. R.J. King, K.V. King and K. Woo "Microwave Moisture Measurment of Grains", IEEE Trans on 1M, Vol. 41, No.1, pp 111-115, Feb.1992.

4. S.O. Nelson, "Measurement and Application of Dielectric Properties of Agricultural Products" IEEE Trans on 1M, Vol. 41, No.1, pp 116-122, Feb. 1992.