w

thogonalhe purposestructedenticalfrommedia.

l sensorsains toAuto-

ISATRANSACTIONS®

ISA Transactions 41~2002! 13–18

Lensed optical fiber sensors for on-line measurement of flo

Sallehuddin Ibrahim,a,* Robert Garnett Green,b Ken Dutton,b Ruzairi Abdul Rahima

aFaculty of Electrical Engineering, Universiti Teknologi Malaysia, Johor Bahru, 81310, Johor, Malaysia

bSchool of Engineering, Sheffield Hallam University, Sheffield, S1 1WB, U.K.

~Received 6 September 2000; accepted 5 February 2001!

Abstract

This paper describes a system using lensed optical fiber sensors that are arranged in the form of two orprojections. The sensors are placed around a process vessel for upstream and downstream measurements. Tof the system is for on-line monitoring of particles and droplets being conveyed by a fluid. The lenses were conusing a custom heating fixture. The fixture enables the lenses to be constructed with similar radii resulting in idcharacteristics with minimum differences in transmitted intensity and emission angle. By collimating radiationtwo halogen bulbs, radiation can be obtained by the sensors with radiation intensity related to the nature of theEach sensor interrogates a finite section of the measurement section. Each sensor provides a view. Paralleprovide a projection. Signal processing is carried out on the measured data in the time and frequency dominvestigate the latent information present in the flow signals. © 2002 ISA—The Instrumentation, Systems, andmation Society.

Keywords: Lens; Optical fiber; Sensor; Measurement; Flow

fordi-isntn isill

hisn-ofenen-vitym-

r is

elden

-

ectin-er-d be-ndnniced

ofes,beys-ed.

1. Introduction

There are several ways of classifying sensorstomographic imaging but generally they can bevided into hard-field and soft-field sensors. Thenables the identification of several importacharacteristics of the sensors. Such classificatioimportant in order to assess whether the flow whave an influence on the field of the sensor. Twill have an effect on the selection of the recostruction algorithm and on the overall designthe image reconstruction system, with consequeffects on the system cost. A soft-field sensor gerates a nonhomogenous field, and the sensitidistribution inside the field changes as the paraeter distribution changes. An optical fiber senso

*Corresponding author.

0019-0578/2002/$ - see front matter © 2002 ISA—The Instru

t

categorized as a hard-field sensor where the fidistribution is independent of the medium betwethe source and the detector@1#. As such, the ma-terial in the flow is assumed only to vary the intensity of the received signal.

Using the conventional~nonradiation! sensingmethods, the operational environment may affthe transducer’s performance and result in ancorrect measurement. In the case of optical-fibbased sensor systems, these effects are reducecause fibers have well-documented stability aimmunity to external interference. The utilizatioof optical fiber sensors also enables the electropart of the flowmeter to be completely separatfrom the measurement head and thus problemselectrical/flame proof safety do not arise. Besidthe use of optical fibers allows more sensors toplaced around the process than many other stems and thus a higher resolution can be achiev

mentation, Systems, and Automation Society.

14 Sallehuddin Ibrahim et al. / ISA Transactions 41 (2002) 13–18

Fig. 1. A side view of the fiber optic model.

s-anb-ors

ngt-

ndthe-b-hethets,s-in

Ar-mas a

edown-do

ct-of

es

ltoa-al

u-the

toof

bem-ersins

the ac-blethens-iusis

ei-

nt

Optical fiber sensors can provide a crossectional image of material distribution. Suchimage is formed by the reconstruction of data otained from an array of sensors. These sensmonitor the emitted radiation~light! for fluctua-tions caused by particles or droplets interferiwith the beam when passing through it. Collimaing the radiated beam from a light source apassing it through a flow regime ensures thatintensity of radiation~light! detected on the opposite side is related to the distribution and the asorption coefficients of the different phases in tpath of the beam. Since different phases inflow usually have different absorption coefficienso the intensity of the transmitted radiation is asociated with the proportion of different phasesthe beam path and hence to the flow regime.ranging several such beams can generate infortion about a flow field across an area, such across-section diameter of pipe@2#.

The main advantages of this optical-fiber-bassystem are that each light beam averages the flinformation within its cross section and the sesors are mounted outside the flow pipe so theynot invade the flow, thus providing a noncontaing method of measurement. The performancethe system is independent of temperature, prsure, and viscosity of the fluid.

2. Modeling the optical fiber lens

In order to optimize the utilization of the opticafiber for the tomographic system, it is importantmodel the optical fiber lens. As such, an investigtion is carried out on the way both ends of optic

-

-

fibers are terminated since the termination inflences the acceptance and emission angles oflight energy transmitted by the fiber. In orderavoid the loss of the beam intensity, the beamlight must diverge as little as possible. This canachieved by using optical lenses. Besides, it is iportant to design the lenses such that all fibhave smooth surfaces and similar radii resultingidentical characteristics with minimum differencein transmitted intensity and emission angle.

The optical fiber lens model with lenses at boends is depicted in Fig. 1. Each lens can havdifferent value of radius. The model takes into acount that a large divergent beam is undesirabecause it decreases the energy centered inbeam and can cause overlapping of adjacent tramitted beams at the receivers. The optimum radto provide a lens for a particular applicationdetermined from the model.

From the model, it can be shown that

u05sin21S h

R2D2sin21F n1

nmsinH sin21S h

R2D

2sin21S h

R1D1sin21S n0 sinu i

n1D J G

in whicha5the angle between the incident light and thhorizontal axis~convergence/divergence of incdent radiation!.u i5the incident angle.b5angleof transmitted light~emergent ray!.u05divergentangle or angle between the incidelight and the horizontal axis.

15Sallehuddin Ibrahim et al. / ISA Transactions 41 (2002) 13–18

Fig. 2. A diagram of the aluminum block and polishing rotor.

t

k

r

er

-an

uchen-toes

op-kr-

lesa-ted5oft-edl-

0.er

ed.us-

ns

to-

l fi-diisyowes

he-

n-s aro-eson-otro-ll-

s.uc-

n05refractive index of the medium at the fronend of the optical fiber.n15refractiveindex of the optical fiber core.n25refractiveindex of the fiber cladding.nm5refractive index of the medium at the bacend of the optical fiber.h5radiusof the fiber opticcore50.5 mm.R15radiusof the lens at the front end of the fibeoptic.R25radiusof the lens at the back end of the fiboptic.

3. Preparation of the optical fiber

A flow imaging system using optical fiber sensors must have a good resolution to generateimage that accurately represents the flow. As sit is important to ensure that the lenses have idtical characteristics and as such it is importantstandardize the method in preparing the lensBefore the lens is constructed, the end of thetical fiber is polished. A simple aluminum bloc~Fig. 2! is used to hold the fibers tightly in a vetical manner during the polishing process.

There are 20 holes in the block and this enab10 fibers to be inserted in the block simultneously. The fibers are inserted into holes locain the block in such a way that approximatelymm of the fiber end protrudes from the bottomthe block. Then the top aluminum block is tighened to ensure that the fibers are positiontightly. The block containing the fiber is then po

.

ished with a range of grit papers down to 120The polishing process is completed when the fibend is flat and the rough surfaces are removThe ends of the fibers are then cleaned with indtrial methylated spirit and dried.

A custom heating fixture is used to form the leon the prepared flat surface~Fig. 3!. The tempera-ture range of the fixture can be varied from 90150 °C. It uses a feedback control loop to maintain the temperature within60.4 °C of the setvalue. The lenses were designed such that albers will have smooth surfaces and similar ra~1.19 mm! resulting in identical characteristicwith minimum differences in transmitted intensitand emission angle. This method provides a lcost approach to the production of optical lenson plastic optical fibers.

4. Measurement system

The lensed optical fibers are arranged in tconfiguration shown in Fig. 4. The system employs two orthogonal projections with each cosisting of 16 sensors. Each sensor interrogatefinite section of the measurement section and pvides a view. The fibers are mounted into holmachined into the measurement section in a nintrusive manner, which means that the flow is naffected by their presence. Optical energy is pjected into the measurement volume using a wecollimated beam from two powerful projectorThe measurement section is of modular constr

16 Sallehuddin Ibrahim et al. / ISA Transactions 41 (2002) 13–18

Fig. 3. Lens making fixture.

90ren-e.

ob-ouu-

e,ed.den-

esye-heno-ht-tobyat

tion and comprises a series of perspex blocksmm square with an 80 mm diameter central boso that when bolted together they provide a cotinuous 80 mm diameter internal flow passagThe square shape of the section allows opticalservation and measurement to be made withoptical distortion—as would be the case if a circlar pipe was used.

The receiver fiber is coupled to a photodiodenabling the received light level to be measurThe sensor is operated in photoconductive mowhich produces a voltage proportional to the i

t

,

tensity of light incident on it. As an object passthrough the light beam, the level of light energfalling on the receiver fiber is reduced and a rduction in the output of the diode is observed. Tvoltage is conditioned such that, when there isflow, there is maximum light intensity at the receiver and a zero voltage is indicated. With higflow rates of particles or droplets, a maximum ouput of 5 V is generated. The main effect taken inconsideration is attenuation of optical energyparticles or droplets cutting the light path such ththe effects of diffraction are neglected.

Fig. 4. The setup of the tomographic system.

ntsar--em

he

n

eicles

lledtedc-rti-c-ere-n-orn

eec-betin

er-ndo-

foritwnedeoristlyle, ists

po-of

17Sallehuddin Ibrahim et al. / ISA Transactions 41 (2002) 13–18

Results obtained from the 32 measuremewere transferred into the computer and a lineback-projection~LBP! algorithm was used to estimate the solids distribution that would providthese measurements. In this calculation, the coputed sensitivity maps were combined with tmeasured data~sensors’ readings! to obtain an im-age of the particle concentration distributiowithin the conveyor.

5. Results

A laboratory scale flow rig was used for thmeasurements. Silica sand with a mean partsize of 600 nm was fed vertically downwardthrough the measurement section at a controfeed rate. Several flow rates were investigaranging from 40 to 320 g/s. A mechanical obstrution was inserted into the conveyor to enable aficial flow regimes to be developed. This obstrution restricted the solids to half or quarter of thconveyor cross section. The optical sensorsponse to solids flow is shown in Fig. 5. The cocentration profiles at the flow rate of 200 g/s ffull flow, half-flow and quarter-flow are shown iFigs. 6~a!–~c!.

For the full flow, the particles should ideally buniformly distributed across the measurement stion. Fig. 6~a! shows that the particles appear tofairly uniformly distributed. In the half-flow resul@Fig. 6~b!#, the particles are mostly concentratedthe central part of the pipe. The result for quartflow shows the particles occupying the right-haside of the image, which agrees well with the psition of the obstruction.

Fig. 5. Optical sensor reponse to solids flow.

-

6. Conclusion

Lensed optical fiber sensors are suitablemonitoring flowing materials and they exhibhigh linearity when used to measure solids’ florates. Optical fibers enable sensors to be desigwith a very wide bandwidth, thus enabling thmeasurement of high-speed flowing particlesdroplets. The light extinction method used in thproject, whereby the detector is aligned direcinto the light source and the size of the particshadow is measured as it crosses the light pathsuitable for measurement of particles or dropleequal to and greater than 100mm. Optical sensorsare intended for use where the conveyed comnent ratio is less than 10% v/v. In this type

Fig. 6. Concentration profile for~a! full-flow at 200 g/s,~b!half-flow at 200 g/s, and~c! quarter-flow at 200 g/s.

llthe

y:th-

G.,lesra-

-

-

-

cesrs i

u-inion

a

ldipalis aHe

uchla-

me

andIEE

atandiousnent

d ofging

-mn-.

en-itiandE.cess

18 Sallehuddin Ibrahim et al. / ISA Transactions 41 (2002) 13–18

conveying, the material being monitored is wedispersed and experience shows that many ofsensor outputs are not registering particles.

References

@1# Williams, R.A. and Beck, M.S., Process TomographPrinciples, Techniques and Applications. ButterworHeinemann, 1995.

@2# Dugdale, P., Green, R.G., Hartley, A.J., Jackson, R.and Landauro, J., Characterisation of single bubbby an optical tomographic system. Process Tomogphy. 81–92~1994!.

Sallehuddin Ibrahim wasborn in Johor, Malaysia on4 September 1968. He received the Bachelor of Engi-neering ~Electronics! degreefrom Queen Mary and West-field College, University ofLondon, UK, in 1990, MSc inInstrumentation from the Uni-versity of Manchester, UK, in1992 and Ph.D. in Process Tomography from Sheffield Hal-lam University, UK, in 2000.He has been a lecturer in Universiti Teknologi Malaysia

since 1992. His current research interests are in the area of protomography and sensors. He has published more than 30 papeprocess tomography, metrology, and education.

sn

Dr. Ken Dutton was born in Sheffield in September 1952. He gradated from Sheffield City Polytechnic, UK, in 1975 with a BScControl Engineering. He then worked for the British Steel Corporatin process control R&D for 10 years, during which time he gainedPh.D. in control systems~on flatness control in multiroll mills!, fromSheffield Hallam, University, UK. In 1985 he came back to SheffieHallam University as a Senior Lecturer and was promoted to PrincLecturer in Control System Analysis and Design since 1987. HeChartered Engineer, and a Fellow of the IEE and of the InstMC.wrote a book entitledThe Art of Control Engineeringpublished byAddison Wesley Longman in 1997. His research interests cover mof control engineering, including: system analysis, modeling, simution, multivariable systems, and fuzzy logic control.

Emeritus Professor Robert Garnett Green was born on 30 May1937. He received the BSc Physics~Hons! from the University ofNewcastle, UK, in 1959, MSc and Ph.D. in Control Engineering frothe University of Bradford, UK, in 1974 and 1981, respectively. Hhas about 38 years of industrial experience in the field of controlinstrumentation. He is a Chartered Engineer, and a Fellow of theand of the InstMC. He has been a Professor of InstrumentationSheffield Hallam University since June 1993. He has authoredcoauthored various papers on the design and application of vartransducers, especially capacitance transducers, for two compoflow measurement. His current research interests are in the fielmulticomponent flow measurement systems and tomographic imausing optical sensors.

Dr. Ruzairi Hj. Abdul Rahim received the BEng. Degree with honours in Electronic System And Control Engineering in 1992 froSheffield City Polytechnic, UK. He received his Ph.D, in Instrumetation Engineering from Sheffield Hallam University, UK in 1996Currently he is the Head of Department of the Control and Instrumtation Panel in the Faculty of Electrical Engineering, UniversTeknologi Malaysia. He is a senior member of the InstrumentationControl Society of Malaysia, ICSM and an ordinary member of IEEHis current research interests are process tomography and promeasurement.