Measurement Techniques to Monitor and ControlFluidization Quality in Fluidized Bed Dryers: AReview
description
Transcript of Measurement Techniques to Monitor and ControlFluidization Quality in Fluidized Bed Dryers: AReview
-
Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=ldrt20
Download by: [Diego Arroyave] Date: 10 January 2016, At: 21:06
Drying TechnologyAn International Journal
ISSN: 0737-3937 (Print) 1532-2300 (Online) Journal homepage: http://www.tandfonline.com/loi/ldrt20
Measurement Techniques to Monitor and ControlFluidization Quality in Fluidized Bed Dryers: AReview
Mortaza Aghbashlo , Rahmat Sotudeh-Gharebagh , Reza Zarghami , Arun S.Mujumdar & Navid Mostoufi
To cite this article: Mortaza Aghbashlo , Rahmat Sotudeh-Gharebagh , Reza Zarghami , ArunS. Mujumdar & Navid Mostoufi (2014) Measurement Techniques to Monitor and ControlFluidization Quality in Fluidized Bed Dryers: A Review, Drying Technology, 32:9, 1005-1051, DOI:10.1080/07373937.2014.899250
To link to this article: http://dx.doi.org/10.1080/07373937.2014.899250
Published online: 22 May 2014.
Submit your article to this journal
Article views: 239
View related articles
View Crossmark data
Citing articles: 7 View citing articles
-
Review Article
Measurement Techniques to Monitor and ControlFluidization Quality in Fluidized Bed Dryers: A Review
Mortaza Aghbashlo,1 Rahmat Sotudeh-Gharebagh,2 Reza Zarghami,2
Arun S. Mujumdar,3 and Navid Mostou21Department of Agricultural Machinery Engineering, Faculty of Agricultural Engineering andTechnology, University of Tehran, Karaj, Iran2Multiphase Systems Research Lab., Oil and Gas Processing Centre of Excellence, School ofChemical Engineering, College of Engineering, University of Tehran, Tehran, Iran3Department of Food Engineering, King Mongkuts University of Technology Thonburi,Bangkok, Thailand
Fluidized bed dryers (FBD) are commonly employed in manyindustries to dry particulate solids. FBDs provide good solids mixing,high rates of heat and mass transfer, and relative ease of materialhandling. For efcient operation, it is important to be able to monitorand control the uidization regime, particle size distribution (PSD),moisture content, and bulk density as well as product chemicalproperties. This review provides an overview of the trends in theapplication of different experimental techniques to monitor andcontrol the hydrodynamic conditions of FBDs which inuence theparticle physiochemical properties. This review covers a wide rangeof measurement techniques, including infrared moisture sensor(IR), near infrared (NIR) spectroscopy, analysis of pressure uctua-tions, optical imaging techniques, acoustic emission (AE), electricalcapacitance tomography (ECT), spatial lter velocimetry (SFV),Raman spectroscopy, focused beam reectance measurement(FBRM), microwave resonance technology (MRT), triboelectricprobes, positron emission particle tracking (PEPT), and some noveltechniques for monitoring and control of FBDs. The present reviewsummarizes the use of the diverse techniques and outlines their meritsand limitations. Prospects for future research in this area are alsoidentied. The measurement techniques can be used for research,development, and operation of uidized bed equipment used innon-drying applications as well.
Keywords Bed hydrodynamics; Control; ECT; IR; Moisturecontent; NIR; Particle size distribution (PSD); PEPT;Raman spectroscopy
INTRODUCTION
As widely used drying systems, FBDs have foundmany applications in almost all agricultural, biochemical,chemical, pharmaceutical, food, ceramics, polymer,dyestuff, and other process industries. FBDs can be utilizedas an efcient dehydration method, not only for moistparticulate and granular products with susceptibility to ui-dization, but also for removing moisture from suspensions,solutions, dilute pastes, or sludges in a bed of inert parti-cles.[1,2] Particulate materials are commonly dried by hotair or superheated steam to a desired level of moisture con-tent. Fluidization provides high rates of heat and masstransfer, improves uniformity of temperature prole acrossthe bed, facilitates material handling and solids mixing, per-mits processing of temperature-sensitive solids, and offershigh thermal efciency of drying process. Generally, FBDscan be used for drying of particulate materials, agglomer-ates, granules, coatings, and layers. In uidized bed granu-lation, coating, and agglomeration, the complexity becomeseven more serious because of a series of transient intercon-nected phenomena; i.e., binding liquid spraying, particleagglomeration, and wall deposition. Thus, these processesare very strict and even impossible to estimate, monitor,and control during the operation. Moreover, scale-up ofill-dened processes, such as drying, is well-known to be aproblematic practice of industry and can be expensivewhen it goes wrong. Traditionally, FBDs are inspected bysimple methods with easily measurable variables, such astemperature and humidity of the exhaust air, bed axialand average temperatures, and variation of pressure dropand temperature throughout the bed. However, thesemeasurements provide little insight into the complex hydro-dynamic phenomena such as incipient deuidization and
Correspondence: Rahmat Sotudeh-Gharebagh, MultiphaseSystems Research Lab., Oil and Gas Processing Centre of Excel-lence, School of Chemical Engineering, College of Engineering,University of Tehran, P.O. Box 11155-4563, Tehran, Iran; E-mail:[email protected]
Color versions of one or more of the gures in the article canbe found online at www.tandfonline.com/ldrt.
Drying Technology, 32: 10051051, 2014
Copyright # 2014 Taylor & Francis Group, LLCISSN: 0737-3937 print=1532-2300 online
DOI: 10.1080/07373937.2014.899250
1005
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
physicochemical properties of particles and are, therefore,ineffective and not useful for real-time monitoring andgood closed-loop process control.[3] Stability of uidizationin FBDs plays an important role in the quality of thenished product by uniform distribution of the ow ofthe drying media, enhancing heat and mass transfer ratesand preventing collapse of the bed. Moisture content, bulkdensity, and PSD are other critical parameters, signicantlyaffecting the stability of uidization and the quality ofend-products. In some cases, knowledge of the chemicalproperties of materials being dried is also very critical. Onthe other hand, to develop an automated FBD system forspecied end-product quality and monitoring, real-timemeasurement techniques are often integrated with mechan-ical and instrumental facilities to avoid in-process manualmanipulation.
According to the Food and Drug Administration(FDA), Process Analytical Technology (PAT) can bedivided into three categories, including at-line, on-line,and in-line analyzers,[4] as shown in Fig. 1. PAT has beenused to dene a systematic approach for real-time measure-ments to design, analyze, scale up, and control manufactur-ing processes through the monitoring of critical quality andperformance properties for primary and in-process materi-als. At-line analyzers measure the required properties bytaking a sample, isolating it from the environment, and ana-lyzing it in close proximity to the process stream. On-lineanalyzers determine the desired characteristics of materialsby directing the sample from the process to measurementdevice and returning it to the process stream in most situa-tions. It should be mentioned that in on-line mode the dry-ing uid is commonly circulated in the measuring loop tokeep constant the temperature of the sample. In-line analy-zers (intrusive or non-intrusive) are quick measuring devicesor probes to record the sensing data without removing asample by direct placing of them into the process stream.[5]
In the complex and dynamic environment associatedwith FBDs, it becomes necessary to explore technologicaland scientic solutions for preventing deuidization andensuring mean quality parameters of the particles duringthe process. Furthermore, application of monitoring andcontrol techniques plays an important role in several majorareas in research and industry to optimize and automate theprocess, to assure high reproducibility in the end-productquality, to enhance security aspects of the process, tominimize the number of failures in batches, and to lowerexpenditures of energy and human resources.[3] The aimof this article is to review and update the usage of varioustechniques and instruments available for monitoring andcontrolling of uidization and physiochemical characteris-tics of particles. In particular, this review covers applica-tions of an infrared moisture sensor (IR), near infraredspectroscopy (NIR), pressure uctuations, optical imagingtechniques, acoustic emission (AE), electrical capacitancetomography (ECT), spatial lter velocimetry (SFV),Raman spectroscopy, focused beam reectance measure-ment (FBRM), microwave resonance technology (MRT),triboelectric probes, positron emission particle tracking(PEPT), and some other miscellaneous and innovative stra-tegies for control.
van Ommen and Mudde[6] have reported the use ofdifferent techniques for measuring the voidage distributionin uidized beds. Burggraeve et al.[4] have reviewed the analy-tical techniques for monitoring and control of uidized bedgranulators for pharmaceutical application. da Silva et al.[3]
discussed tools for monitoring and control of coating andgranulation processes in uidized beds. Published reviewshave also looked at the application of measurement techni-ques in uidized beds,[7] process control methods and scale-upof pharmaceutical wet granulation processes containinguidized bed granulating,[8] control engineering in drying tech-nology consisting of FBDs,[9] and near infrared spectroscopyand chemometrics in pharmaceutical technologies includingFBDs.[10] However, this review differs from previous onesby including all aspects of FBDs utilized for drying itself inuidized beds as well as drying during uidized bed coating,granulation, and agglomeration processes and presenting themost employed techniques for process monitoring, control,and automation. In addition, advantages and disadvantagesof each technique are mentioned and recommendationsand perspectives are provided for future works.
The diagnostic techniques discussed in this review arealso applicable to other gas-solids contacting devices suchas modied uidized beds, spouted beds, circulating uidbeds, vibrating uid beds, pneumatic conveyors, and vari-ous designs of uidized bed reactors as well. An interestingfeature of FBDs is that particle wetness, particularly atthe surface, can affect uidization quality signicantly.For very high wetness it is possible to cause deuidization,which affects the performance adversely.FIG. 1. Schematic of in-line, online, and at-line product measurements.
1006 AGHBASHLO ET AL.
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
MONITORING TECHNIQUES
Conventional Techniques
Traditionally, monitoring particle moisture and processend-points, the point at which a drying is completed,in FBDs has been carried out based on measurement ofprocess variables such as outlet air temperature, outlet airhumidity, inlet and outlet temperature difference, and othersimple measurements of miscellaneous parameters accord-ing to the dryer characteristics. Particle size measurementand tests of the chemical properties of the product areperformed by acquiring samples during the processingand subsequent analysis using off-line techniques. Fluidiza-tion quality was usually identied by visual observation andevaluation of global hydrodynamic parameters by applyingempirical models. The most important studies regarding theapplication of conventional techniques for monitoring ofFBDs are listed in Table 1.
Alden et al.[11] applied the temperature difference techni-ques using simple psychometric calculation to control theprocess end-point in FBD. The end-point of the granulationprocess was successfully identied and controlled by a com-puter program. Watano et al.[12] successfully examined theuctuations of power consumption in an agitated uidizedbed granulator to monitor granule growth and to determinethe process end-point by computing the coefcient ofvariation of the utilized power. In another investigation,Watano et al.[13] developed a fuzzy controller for control-ling the bed height, which was successively measured byan ultrasonic sensor during uidized bed granulation. Thecontroller effectively prevented deuidization and channel-ing by controlling the bed height. Sivashanmugam andSundaram[14] developed two different empirical models forpredicting the pressure drop for both dilute and dense phaseow regimes in FBD for ragi drying with an acceptabledeviation, and presented a correction coefcient for thedense slugging ow regime. In a similar study, they dis-tinguished the mixed ow behavior up to a certain particleReynolds number from the residence time distributionstudies.[15] El-Nans et al.[16] measured minimum spoutingvelocity as a function of moisture content for drying ofsludge and found a decrease in the minimum spoutingvelocity by progressing of the drying process. Temple andvan Boxtel[17,18] used simple measurements for continuousand batch tea FBD based on wet material feed rate anddirect feedback of the moisture content and intermediateexhaust temperature to design a full automated controlsystem. In addition, this research group has published sev-eral papers about design and application of a full automatecontrol system based on different methods of controllertuning,[19] a custom-built electronic data logger applied tocombination of experimental ndings with modeling andsimulation results,[20] and an algorithm-employed transientexhaust air temperature.[21] These strategies appropriately
controlled the drying process better in most cases comparedto the manual control.
Larsen et al.[22] proposed a control strategy based onin-process thermodynamics calculation using an alternativethermodynamic factor according to enthalpies of actual andadiabatic vented drying air during uidized bed coating.Two separate control loops were suggested in order tomaximize the coating spray rate and keep the process inmass and energy balance. An inner control loop controlledthe product temperature by the inlet air temperature and anouter control loop controlled the relative outlet airhumidity and the degree of consumption of the potentialvaporization energy by the spray rate. Devahastin et al.[23]
correlated empirically the size of shrimp, bed height, andnozzle diameter to the hydrodynamic characteristics of ajet spouted bed of shrimp using the Buckingham p theorem.The temperature difference method has shown an appropri-ate accuracy level in approximating the drying end-point,in which the uidization activity signicantly affecteddetection of the process end-point.[24] Yuzgec et al.[25] satis-factorily employed a model-predictive controller based onthe dynamic recurrent neural networks to predict moisturecontent and product activity during bakers yeast dryingusing data calculated from heat and mass equationsthrough dried granules. This methodology gave simulationresults in the short time required for real-time control appli-cations. In a similar work, Koni et al.[26] optimized dryingconditions to maximize product quality while minimizingthe energy consumption by proposing a recurrent neuralnetwork-based algorithm for developing quality andprocess models which were solved by a genetic algorithm.Matero et al.[27] satisfactorily used multi-way methods witha few process variables to recognize successful batchgranulations from unsuccessful runs and achieved usefulinformation that can improve comprehension of the FBD.
Traditionally, easy-to-use simple techniques were used tomonitor FBDs. More recently, due to the dynamic andcomplex nature of the FBD process, these simple techniquesare found to be unsuitable for some industrial applicationsbecause of their great operator-dependency and often poorrepeatability. In some cases, these techniques can modifythe internal ow of the uidized bed, leading to interferencewith the actual process measurements. On the other hand,heat and mass balance equations commonly used to deter-mine the end-point and estimate the moisture content ofparticles in FBD with respect to inlet and outlet drying airtemperatures. However, drying of particles stronglydepends on the humidity of the inlet-air, a small changeof which may result in a considerable error in the predictedvalue. Moreover, mechanistic models may fail due to sim-plifying assumptions and non-consistency between dryingbatches.[28] Monitoring of FBDs by traditional techniquesis time-consuming and sometimes causes rejection of the
MONITORING OF FLUIDIZATION QUALITY IN DRYERS 1007
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
TABLE1
MostimportantstudiesregardingtheapplicationconventionalforFBDsmonitoring
Author(s)
Measurement
technique(s)
Application(s)
Dryer
type(s)
Process
target(s)
Rem
arks
Alden
etal.[11]
Tem
perature
difference
technique
Tocontroland
instrumenttheFBD
Laboratory-scaleFBD
DryingofAllopurinoland
lactose
mixture
granules
obtained
with
ProvidoneK30in
water
Thedeveloped
program
basedon
thermodynamicsconcept
successfullydetected
end-pointofgranulation
process.
Watanoet
al.[12]
Power
consumption
forgranule
agitationalongwith
coefcientof
variation
Toanalyze
the
granulationprocess
andprogress
of
particles
Top-sprayagitated
uidized
bed
granulator
Granulationofmixture
oflactose
andcorn
starchwith
hydroxypropylcellulos
Apracticalmethodforthe
determinationofan
optimum
operational
end-pointin
thetumbling
uidized
bed
granulation
wasdeveloped.
Watanoet
al.[13]
Intelligentcontrol
basedonultrasonic
heightmeasurement
Developingafuzzylogic
controller
forbed
heightcontrolling
duringuidized
bed
granulating
Lab-scaletop-spray
agitateduidized
bed
granulator
Lactose
andcornstarch
granulatingwith
hydroxypropylcellulose
Fuzzylogiccontroller
favorably
maintained
the
bed
heightatthe
predetermined
valuefrom
initialto
nalstageof
uidized
bed
granulation.
Sivashanmugam
and
Sundaram
[14]
Pressure
drop
measurement
Topropose
theem
pirical
modelsforcalculating
pressure
dropat
differentowregimes
Laboratory-scale
annularcirculating
FBD
DryingofRagi
Twoseparate
modelswere
suggestedforcomputing
pressure
dropfordense
sluggingowanddilute
phase
owregimes
with
theaveragedeviationof
10%.
Sivashanmugam
and
Sundaram
[15]
Residence
time
distributionusing
pulseinputoftracer
Todeterminetheow
pattern
Experimentalannular
circulatingFBD
DryingofRagiparticle
Thetheoreticalmean
residence
timeusing
one-dimensionaltanksin
series
anddispersion
modelagreed
wellwith
theexperimentaldata.
El-Nanset
al.[16]
Measuringminimum
spoutingvelocity
Tocharacterizethebed
hydrodynamicsand
mass
transfer
coefcient
Lab-scalespoutedFBD
Dryingofsludge
Increasingthemoisture
contentofsludgeparticles
ledto
anincrem
entin
minimum
spouting
velocity.
Tem
pleandvan
Boxtel[1
7]
Conventionalcontrol
Toidentify
someofthe
limitationsona
controller
inan
industrialFBD
based
onmodelingresults
Industrial-scaleFBD
Tea
drying
Wet
product
feed
rate
should
beprecisely
controlled
tomaintain
moisture
dischargefrom
dryer
atconstantvalue.
1008
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
Tem
pleandvan
Boxtel[1
8]
Conventionalcontrol
Tocontrolandmonitor
thedried
product
moisturecontentbased
ontwofeedback
controllers
ContinuousFBD
Tea
drying
Inferentialcontrolbasedon
outlet
airtemperature
successfullycontrolled
moisture
ofoutlet
product.
Tem
pleet
al.[19]
Conventionalcontrol
Tuningofcontrollers
usingdifferentstrategy
toprecise
controlling
ofnalproduct
moisture
ContinuousFBD
Tea
drying
Theintegratedqualitative
measure
andintegral
squarederrorwasthebest
choicein
controller
tuningofFBD
under
differentoperating
condition.
Tem
pleet
al.[20]
Custom-m
ade
electronicdata
logger
and
controller
Tomonitorandcontrol
teadryer
ExperimentalFBD
Tea
drying
Thedeveloped
system
was
satisfactorily
installed
inthemajority
ofrelated
factories
anddrying
process
wascontrolled
betterthanmanual
controlling.
Tem
pleet
al.[21]
Conventionalcontrol
Todetectthedrying
end-pointusinginlet
andexhaust
temperature
measurement
Lab-scaleFBD
Developinganalgorithm
forautomaticend-point
determination
Theend-pointwas
successfullydetermined
bydeveloped
algorithm.
Larsen
etal.[22]
Conventionalcontrol
Tocontrolthedrying
process
basedon
thermodynamics
concept
Lab-scaleuidized
bed
coaterwithboth
top
andbottom
spraying
techniques
Sugarspheres
and
microcrystalline
cellulose
pelletscoating
withEudragit1NE
30D,Eudragit1RS
30D,andAquacoat
ECD1lm
polymers
Thethermodynamicsmodel
wassuccessfully
incorporatedinto
anew
process
controlstrategy
bycalculatingthedegree
ofutilizationofthe
potentialevaporation
energyoftheventedair
andtherelativehumidity
ofexhausted
dryingair.
Devahastin
etal.[23]
Recodingthe
minimum
spouting
velocity
and
maximum
and
steadyspouting
pressure
drops
Todeterminethebed
hydrodynamics
Experimentaljet
spoutedFBD
Shrimpdrying
Threeem
piricalcorrelations
weredeveloped
for
Reynoldsnumber,
maximum
pressure
drop
andspoutingpressure
dropbasedonthe
Buckingham
pmethodas
functionshrimpand
dryer
characteristics.
(Continued
)
1009
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
TABLE1
Continued
Author(s)
Measurement
technique(s)
Application(s)
Dryer
type(s)
Process
target(s)
Rem
arks
Lipsanen
etal.[24]
Tem
perature
difference
method
Todetectthedrying
end-point
Bench-scaleuidized
bed
granulator
Granulationofmixture
of
ibuprofenanda-lactose
monohydrate
by
aqueoussolutionof
polyvinylpyrrolidone
Thetemperature
difference
techniquesatisfactorily
identied
drying
end-pointatdifferent
humidityofdryingair.
Yuzgec
etal.[25]
Articialintelligent
controlbasedon
simpleprocess
measurement
Topredictthemoisture
contentandproduct
activityusinga
dynamic
neural-network-based
model-predictive
controlstructure
solved
bygenetic
algorithm
Industrial-scaleFBD
Bakersyeastdrying
Thepresentedmethodology
successfully
approximatedthe
moisture
contentand
product
activityand
accordingly
anintelligent
controlsystem
suggested
basedonsimulation
results.
Koniet
al.[26]
Intelligentcontrol
basedon
neural-network-
basedmodelsand
modied
genetic
algorithm
using
process
variables
Todetermineand
controltheoptimal
conditionsto
maximize
product
quality
while
minimizingenergy
consumption
Large-scalebatchFBD
Dryingofbakersyeast
Theoptimalcontrol
algorithm
byapplyingthe
recurrentneuralnetwork
modelsforestablishing
thequality
andprocess
models,solved
througha
modied
genetic
algorithm,wassuggested
topromote
the
perform
ance
ofthedrying
processofbakersyeastin
batchuidized
bed.
Matero
etal.[27]
Multi-waymodels
withprocess
variablesincluding
mass
temperature,
inletairtemperature
andoutlet
air
temperature
Todistinguishthe
successfulbatch
granulationsfrom
unsuccessfulruns
Bench-scaletop-spray
uidized
bed
granulator
Granulationof
hydrophobic
pharm
aceutical
ingredient,hydrophilic
excipientin
amonohydrate
form
and
polymericexcipientwith
polyvinylpyrrolidoneas
binder
liquid
Theparallelfactoranalysis
PARAFAC2method
provided
agood
separationbetweenthe
successfuland
unsuccessfulbatches
comparedwiththe
PARAFACmethod.
1010
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
product if the target condition is not met. Conventionaltechniques frequently provide time-averaged informationabout the bed hydrodynamics and particle properties.
Infrared (IR) and Near Infrared (NIR) Spectroscopy
IR spectroscopy applies the infrared zone ofelectromagnetic radiation with longer wavelengths thanthose of the visible light, varying from 700nm to 1mm,and a frequency span of approximately 300GHz up to430THz. The IR technique has been commonly utilizedfor measuring the moisture in FBDs and is based on certainIR wavelengths which are absorbed due to the hydrogen ofwater molecules as they pass through the particulatematerial (in the wave number range 1600 and 1700 cm1).NIR spectroscopy is a technique covering the transitionfrom the visible spectral range to the mid-infrared regionof the electromagnetic spectrum in the wavelength rangeof 8002500 nm (wave number range 12,5004000 cm1),mainly indicating the vibrations of CH, OH, SH andNH bands. Absorbance in the NIR region results frommolecular overtone and combination vibrations of the fun-damental mid-infrared bands. Thus, the particle size andmoisture content can be tracked effectively using NIR spec-troscopy due to the sensitivity of absorbance in this region tovariations in the moisture content, particle size, and chemi-cal state.[3] Pasikatan et al.[29] have illustrated the basicprinciples of NIR spectroscopy relevant to particle sizemeasurement and its dependency on sample preparations,technique of presentation, reference methods, calibrationdevelopment, and validation. Roggo et al.[30] reviewed phar-maceutical applications of NIR spectroscopy chemometricsin three different subsections, including qualitative analysesand classications, regression methods and quantitativeapplications, and online applications. The chemometricmethods used for analysis of NIR spectra were categorizedinto three main approaches: mathematical pretreatments,classication methods, and regression methods. Their studyalso contained an overview of NIR spectroscopy applicationto uidized bed granulation and coating to monitor onlinethe moisture content, chemical compound content,end-point detection, and coating thickness. De Beer et al.[5]
comprehensively reviewed the use of NIR spectroscopy forthe in-process monitoring of pharmaceutical productionprocesses with special emphasis on pharmaceutics and dos-age forms. A review of the literature, a summary of whichis given in Table 2, shows that a considerable number ofresearchers have used NIR for monitoring of physico-chemical properties of the product within the FBDs.
Watano et al.[31,32] employed an IR moisture sensor tomonitor moisture content during a uidized bed granu-lation process and established a fully automated systemby means of a moisture feedback controller and an adaptivefuzzy controller for controlling the moisture, respectively. Again scheduling based on drying capacity was introduced
into a fuzzy control system, which effectively controlled themoisture content at various inlet air temperatures. Further,Watano et al.[33] investigated IR absorption with variouspowder characteristics, such as water-absorbing potentialand granule size, to create a relationship between granulewater content and the IR absorbance spectra. Generally,the relationship between moisture content and the IR absor-bance should be identied before granulation and drying foraccurate monitoring of the moisture using an IR probe. Inanother study, the same research group reported that therelationship between granule moisture content and the absor-bance of IR spectra is not inuenced by the air ow rate ofthe purge air for preventing particle adhesion, uidizing airvelocity, agitator rotational speed and spray mist size, whileit is affected signicantly by the uidizing air temperatureand the liquid ow rate at extremely low agitating speed.[34]
Kirsch and Drennen[35] successfully employed NIR spec-troscopy in at-line mode to estimate the amount of polymercoat to tablet core with a maximum standard error of 1.07%.Inline NIR spectroscopy was also successfully applied forpredicting water uptake and size of particles during the ui-dized bed drying stage of granulating.[36] A good correlationwas also presented between NIR wavelengths and granulesize.[37] Rantanen et al.[38] proved that the moisture contentof granules during spraying and drying phases can beapproximated using the NIR spectrum with a standard errorof 0.2%. Morris et al.[39] employed a fast drying method byusing a higher inlet air temperature to accelerate the dryingprogress without increasing the bed temperature beyondsafe limits towards the end of uidized bed drying of Ibupro-fen granules along with real-time monitoring of moistureusing NIR spectroscopy. The NIR absorbance showed asimilar trend with the moisture ratio of the product duringthe drying process. In a similar study, Wildfong et al.[44]
showed that NIR spectroscopy can successfully detect theend-point of the granulation process and determine themoisture content of the product quite well.
Rantanen et al.[40] employed a four-wavelength NIR sen-sor for monitoring uidized bed granulation=drying andutilized three of the recorded spectra for moisture contentdetermination. In similar works, the same research groupproved that the measurement of water content duringgranulation can be accurately carried out by NIR spec-troscopy at a wavelength of around 1940 nm.[41,42] Theyalso studied the effects of different parameters on a four-wavelength NIR probe and compared ANN and PLSregression for the prediction of the particles moisturecontent. It was reported that ANN is more suitable forapproximation of the water content over PLS regression.[43]
Rasanen et al.[45] applied NIR spectroscopy to measurethe moisture content of several materials during drying inmultichamber microscale FBD equipment. Moisture con-tent and drying phase were effectively tracked through threewavelengths of the NIR region. However, inlet air humidity
MONITORING OF FLUIDIZATION QUALITY IN DRYERS 1011
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
TABLE2
ApplicationofNIR
spectroscopyformonitoringofFBDs
Author(s)
Technique(s)
Application(s)
Dryer
type(s)
Process
target(s)
Rem
arks
Watanoet
al.[31]IR
moisture
sensor
withfeedback
moisture
controller
Tocontrolthegranules
moisture
Experimentaltop-spray
uidized
bed
granulator
Lactose
andcornstarch
granulatingwith
Hydroxypropylcellulose
Thegranulesmoisture
content
wassuccessfullymeasured
via
IRmoisture
sensorand
thegranulegrowth
wasfully
controlled
usingamoisture
feedback
controller.
Watanoet
al.[32]Smartcontrol
Real-timemonitoring
andcontrollingof
moisture
contentusing
IRmoisture
sensor
alongwithadaptive
fuzzycontroller
Experimentalagitated
uidized
bed
granulator
Mixture
oflactose
and
cornstarchgranulation
with
Hydroxypropylcellulose
Granulemoisture
contentwas
effectivelycontrolled
at
variousdryingair
temperaturesbyintroducing
again
schedulingto
adaptivefuzzysystem
based
ondryingcapacity.
Watanoet
al.[33]IR
moisture
sensor
Toinvestigate
the
relationship
between
granulemoisture
contentandIR
absorption
Lab-scaletop-spray
agitateduidized
bed
granulator
Granulationoflactose
and
cornstarchwith
Hydroxypropylcellulose
Therelationship
between
granulemoisture
content
andabsorbance
ofIR
spectrawasprofoundly
affectedbythe
water-absorbingpotentialof
powder.However,granule
size
could
affectthis
relationshipduringdryingof
wet
granules.
Watanoet
al.[34]IR
moisture
sensor
Toinvestigate
theeffect
ofoperatingcondition
ontheaccuracy
of
moisture
measurement
byIR
moisture
sensor
Lab-scaletop-spray
agitateduidized
bed
granulator
Mixture
oflactose
and
cornstarchgranulating
with
Hydroxypropylcellulose
Theoperationalconditiondid
notaffecttherelationship,
exceptatextrem
elylow
dampeningspeed,whilethe
relationship
wasinuenced
bytheuidizationair
temperature
andliquid
ow
rate.
Kirschand
Drennen
[35]
NIR
spectroscopy
Todeterminethe
amountofpolymer
coatapplied
totablet
cores
Experimentaluidized
bed
coater
Coatingofplacebo
containinglactose,
microcrystallinecalluses,
andmagnesium
stearate
withethylcellulose
and
hydroxypropylcellulose
asbindingsolution
TheNIR
spectroscopy
satisfactorily
predictedthe
amountoflm
applied
for
coating.
1012
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
Frakeet
al.[36]
NIR
spectroscopy
Totrack
thegranule
moisture
contentand
particlesize
variations
Bench-scaletopspray
uidized
bed
granulator
Granulationofmagnesium
carbonate
with
Polyvinylpyrrolidoneand
Hydroxypropyl
methylcellulose
NIR
spectroscopyem
ployed
formonitoringparticle
growth
andmoisture
contentwithacceptable
accuracy
duringuidized
bed
granulating.
Rantanen
and
Yliruusi[37]
NIR
spectroscopy
Tomeasure
thePSD
Bench-scaleuidized
bed
granulator
Granulationof
microcrystallinecellulose
NIR
spectroscopywavelengths
wereaccurately
correlated
withthegranuleparticle
size.
Rantanen
etal.[38]
MultichannelNIR
techniquewiththree
statistical
parameters
Tomonitorthemoisture
content
Bench-scaleuidized
bed
granulator
Granulationofthree
differentform
ulations
containing
microcrystallinecellulose,
lactose
monohydrate,
maizestarch,mannitol,
verapamilhydrochloride
withpolyvinlpyrrolidone
asbinder
Theem
ployed
multichannel
NIR
techniquesatisfactorily
estimatedthemoisture
contentofgranulesduring
sprayinganddryingphases.
Morriset
al.[39]
NIR
spectroscopy
Tomonitorthemoisture
content
ExperimentalFBD
Ibuprofen-starch
granulationusing
Polyvinylpyrrolidoneas
thebinder
solution
NIR
spectroscopysuccessfully
tracked
themoisture
variationduringuidized
bed
granulation.
Rantanen
etal.[40]
Four-wavelength
NIR
sensor
Tomeasure
themoisture
content
Laboratory
uidized
bed
granulator
Granulationofglass
ballotiniand
microcrystallinecellulose
withpoly[1-(2-oxo-
1-pyrrolidinyl)ethylene]
withgelatinasbending
solution
Thegranulewatercontent
could
betrustworthyand
quicklymeasuredusingonly
afewNIR
wavelengths
aroundthewaterband.
Rantanen
etal.[41]
MultichannelNIR
sensorin
conjunctionwith
PCA
Toselect
theNIR
wavelengthsfor
measuringmoisture
content
Bench-scaleuidized
bed
granulator
Granulationoftheophylline
anhydrate
andsilicied
microcrystallinecellulose
with
polyvinylpyrrolidone
solutionasbinder
NIR
spectraanalyzedwith
PCA
accurately
detectedthe
granulesmoisture
content
duringprocessing.
Rantanen
etal.[42]
NIR
spectroscopy
Tomonitorthemoisture
content
Bench-scaletop-spray
uidized
bed
granulator
Productionofgranule
containingmannitol,
pregelatinized
starch,and
polyvinylpyrrolidone
solution
Theapplied
NIR
spectroscopy
set-upwithamultichannel
detectorwasapowerful
techniqueformoisture
measurementduring
granulation.
(Continued
)
1013
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
TABLE2
Continued
Author(s)
Technique(s)
Application(s)
Dryer
type(s)
Process
target(s)
Rem
arks
Rantanen
etal.[43]
Four-wavelength
NIR
sensorin
combinationwith
partialleastsquare
(PLS)andarticial
neuralnetwork
(ANN)
Topredictthegranule
moisture
content
Bench-scaleuidized
bed
granulator
Granulationofanhydrous
theophyllinewith
Polyvinylpyrrolidoneas
bendingsolution
TheANN
wasfoundto
have
more
estimativepower
with
theindependenttestdata
thanthePLS.
Wildfong
etal.[44]
NIR
spectroscopy
Tomonitorthemoisture
contentvariationand
todetectend-pointin
onlinemode
ExperimentalFBD
Granulationof
Ibuprofen-starchwith
Polyvinylpyrrolidone
solutionasbinder
Moisture
contentofgranules
andprocess
end-pointwas
precisely
estimatedbyNIR
spectroscopy.
Rasanen
etal.[45]NIR
spectroscopy
Tomeasure
moisture
contentandto
detect
dryingphase
in-line
mode
Multichamber
microscaleFBD
Dryingofdisodium
hydrogen
phosphates
withthreedifferentlevels
ofhydrate
waterandwet
theophyllinegranules
NIR
spectroscopywas
successfultechniquein
in-linedeterminationof
productmoisture
anddrying
phase.
Daviset
al.[46]
NIR
spectroscopy
withstandard
norm
alvariate
Tomonitorthe
polymorphic
transform
ationsof
glycineduringdrying
phase
Lab-scaleFBD
Granulationofmixture
of
c-glycineand
microcrystallinecellulose
withwaterasthe
granulatingliquid
NIR
spectroscopywas
successfullyapplied
toidentify
theaandcform
sof
glycineduringgranulation
anddrying.
Green
etal.[47]
NIR
spectroscopyin
conjunctionwith
PLS
Tomonitorthemoisture
contentandto
investigate
theeffects
ofsamplingonmethod
precision
Pilot-scaleFBD
Dryingofgranules
containingdifferentratios
ofthesamemajor
excipients(lactose
monohydrate
and
microcrystallinecellulose)
anddifferentbulk
drug
andother
minor
excipients
Usefulinform
ationwas
obtained
regardingthe
impact
ofsamplingon
inaccuracy
ofin-lineNIR
spectroscopymethod.
PaulFindlay
etal.[48]
NIR
spectroscopy
Tomonitorthemoisture
contentandparticle
size
fordetectingthe
end-pointof
granulationprocess
Experimentaltop-spray
uidized
bed
granulator
Granulationofmixture
of
acetaminophen
or
ibuprofenwithlactose
monohydrate
and
microcrystallinecellulose
bypovidoneasbending
solution
Thegranulationend-pointwas
effectivelydetectedusing
calibratedNIR
spectroscopy
bymoisture
contentand
particlesize
measurement.
1014
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
Nieuwmeyer
etal.[49]
NIR
spectroscopyin
conjunctionwith
principle
componentanalysis
(PCA)andPLS
Toidentify
themoisture
contentandparticle
size
ofgranules
Laboratory-scaleFBD
Granulationoflactose
usingde-mineralized
water
TheNIR
spectrum
was
precisely
correlatedwiththe
granulemoisture
content
andparticlesize.
Romer
etal.[50]
NIR
spectroscopy
withPCA
Tomonitorthephase
transform
ations
MicroscaleFBD
Dryingofpelletscontaining
erythromycindihydrate
andmicrocrystalline
cellulose
TheNIR
spectroscopy
detectederythromycin
dihydrate
transform
ations
toitsisomorphicdehydrate
form
inthepelletsat
temperaturesabove45 C
.Lee
etal.[51]
NIR
spectroscopyin
conjunctionwith
averagingand
clusteringofspectra
Tocontrolthecoating
thickness
Custom-fabricated
uidized
bed
coater
Coatingofmixture
ofve
pharm
aceutical
ingredients,including
microcrystallinecellulose,
lactose
monohydrate,
d-m
annitol,magnesium
stearate,andcorn
starch,
usingbinder
solutionsof
polyethyleneglycoland
hydroxypropyl
methylcellulose
Thecoatingthicknesswas
controlled
aswellas3%
deviatedfrom
theactual
thickness.
Alcala`
etal.[52]
NIR
spectroscopy
withPLSandPCA
Topredictthemoisture
content,particlesize
distributionandbulk
density
Industrial-scaleuidized
bed
granulator
Granulatingof
microcrystallinecellulose
withmaizestarch
solutionasbinder
Thepharm
aceuticalgranule
properties
such
asmoisture
content,bulk
density,and
particlesize
wereexcellently
monitoredusingtheNIR
spectroscopy.
Mark
etal.[53]
NIR
spectroscopy
withmultivariate
calibrations
Todeterminetheideal
end-pointbytaking
into
accountproduct
quality,watercontent
andresidualsolvent
ExperimentalFBD
Dryingofantibiotic
Anautomatedmonitoring
system
basedonthe
continuousassessm
entof
NIR
spectroscopydata
was
developed
forauidized-bed
dryingprocess
ofa
pharm
aceutical
interm
ediate.
Peinadoet
al.[28]NIR
spectroscopyand
experimental
moisture
determinationalong
withPLS
Tospecifythedrying
end-point
Fullcommercial-scale
FBD
Dryingofhydrochloride
saltcontaining
micro-crystalline
cellulose,sodium
starch
glycolate
andpovidone
Thepresentedmethodology
waseffectivelyapplied
for
in-linedeterminationof
productmoisture
anddrying
end-point.
(Continued
)
1015
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
TABLE2
Continued
Author(s)
Technique(s)
Application(s)
Dryer
type(s)
Process
target(s)
Rem
arks
Lee
etal.[54]
NIR
spectroscopyin
conjunctionwith
PLS
Toquantify
thecoating
thickness
Custom-m
ade
experimentaluidized
bed
coater
Coatingofmixture
of
microcrystallinecellulose,
lactose
monohydrate
and
polyvinylpyrrolidone
withpolyethyleneglycol
andhydroxypropyl
methylcellulose
ascoating
materials
Thecoatingthicknessasahigh
quality
end-point
designationcould
be
precisely
estimatedvia
in-lineNIR
spectroscopy
measurement.
Hartungetal.[55]NIR
spectroscopy
Tomonitorthemoisture
content
Experimentaluidized
bed
granulator
GranulatingofEnalapril
maleate
withlactose
monohydrate,maize
starch,andsodium
hydrogen
carbonate
NIR
spectroscopywas
successfulin
monitoringof
granulesmoisture
content.
Konaet
al.[56]
IntegratingNIR
spectroscopywith
humidityand
temperature
data
loggersalongwith
PLSandPCA
Process
understanding
andfaultdiagnosing
Experimentaltop-spray
uidized
bed
granulator
Fexofenadine
hydrochloride,
microcrystallinecellulose
andlactose
monohydrate
blendsgranulationwith
polyvinylpyrolidoneas
bendingsolution
Process
understandingand
faultdiagnosingsuccessfully
carriedoutbycouplingNIR
spectroscopy,humidity,and
temperature
data
inconjunctionwith
multivariatebatchmodeling.
Heiglet
al.[57]
On-andoff-lineNIR
spectroscopyalong
withPLS
Topredicttheresidual
moisture
content
Small-scalecold-m
odel
FBD
Dryingofdibasiccalcium
phosphate
anhydrous
On-andoff-lineNIR
spectroscopywassuccessful
inpredictionofmoisture
contentwhen
comparedwith
actualmoisture
content
data.However,onlineNIR
spectroscopyledto
more
precise
resultsthanoff-line
NIR
spectroscopy.
Hayashiet
al.[58]NIR
spectroscopy
withPLS
Toevaluate
thewater
contentofgranules
andto
estimate
the
constantdryingrate.
Lab-scaleFBD
Dryingofextruded
Riboavin
granules
obtained
bycombination
oflactose
astheller,
potato
starchasthe
disintegratingagent,and
hydroxypropylcellulose
asthebindingagent
Moisture
contentofgranules
waspredictedaccurately
usingthePLSmodeland
accordingly
constantdrying
rate
wasestimated.
1016
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
had a profound effect on the apparent absorbance of waterand subsequent NIR measurement results. Davis et al.[46]
quantied the polymorphic transformations of c-glycine toa-glycine during the drying phase of a wet granulation usingNIR spectroscopy. The results were qualitatively conrmedusing X-ray diffraction of the powder. Green et al.[47] carriedout different experiments in FBDs at scales of 65, 300, and600L using different sampling schemes (dynamics, owing,and stationary solids) to better understand and improvethe accuracy of theNIR spectroscopy technique. The processheterogeneity had an undesirable role in the measurementaccuracy. In another investigation, the NIR spectroscopygauge was calibrated for simultaneous real-time monitoringof particle size and moisture content as well as detecting thegranulation end-point.[48] Their results were in good agree-ment with ofine analytical measurements for determiningthe end-point. Nieuwmeyer et al.[49] discriminated variousstages of the drying process using NIR spectroscopy with asmall relative error compared to Karl-Fischer analysis. Theparticle size was also determined with a small predictionerror between nes and granules. NIR spectroscopy andX-ray diffractometry detected the modication of erythro-mycin dehydrate solid state to its isomorphic dehydrate formthrough FBD based on the moisture content of the pellets attemperatures greater than 45C.[50]
Lee et al.[51] introduced NIR spectra averaging and clus-tering procedures to establish a proper dynamic calibrationmodel for the measurement of coating thickness. The spec-tra averaging method for a small number of spectra provedto be a reasonably good dynamic calibration model with ahigh correlation coefcient. However, the PCA-based clus-tering technique was proposed for a large number of NIRspectra. Alcala` et al.[52] utilized PCA and PLS as qualitativeand quantitative methods for analyzing NIR spectra tomonitor the uidized bed granulation process of pharma-ceutical materials. A good correlation was found betweenthe absorbed NIR spectra and granule properties. Inanother investigation, the absorbed NIR spectra were ana-lyzed by the multivariate statistical method for evaluatingthe quality, water content, and residual solvent of antibio-tics on FBD. The results were conrmed by ofine measure-ments such as high performance liquid chromatography(HPLC), Karl Fischer back-titration method, and gas chro-matography (GC).[53]
Peinado et al.[28] developed a PLS model using NIR spec-tra and experimental moisture measurement for real-timedetermination of the end-point in a FBD. Conrmationtests revealed that the employed strategy was useful in inlineend-point specication. Lee et al.[54] developed excellentcorrelations between the coating thickness of pellets pre-dicted by inline NIR monitoring in conjunction with PLSand two ofine methods, including confocal laser scanningmicroscopy and laser diffraction particle size analysis,during uidized bed coating. Hartung et al.[55] found a good
relationship between water content determined by NIRspectroscopy and Karl-Fischer titration for Enalaprilmaleate formulation during uidized bed granulating.Kona et al.[56] integrated real-time product moisturecontent detection obtained using NIR spectroscopy withhumidity and temperature of the drying bed and establishedstatistical process monitoring charts (SPMC) by simul-taneous application of PLS and PCA techniques foruidized bed granulation and drying. NIR spectroscopy,along with humidity and temperature data loggers, appearsto be promising for effective process control and fault deter-mination. Heigl et al.[57] compared the PLS modeling ofonline NIR spectra with PLS modeling of ofine NIR spec-tra according to a reference method (loss-on-drying) formonitoring the moisture content of dibasic calcium phos-phate anhydrous on FBD. The results of ofine NIRspectroscopy indicated that the amount of withdrawn sam-ple and sampling time interval can cause bias in the moist-ure content predictions. Hayashi et al.[58] continuouslypredicted the water content of granules with negligible errorby employing a PLS model based on the NIR spectra andloss-on-drying measurements and accordingly dis-tinguished the constant drying regime.
NIR spectroscopy is a real-time, fast, safe, reliable, andnon-intrusive technique which requires no sample prep-aration, little or no modication to the existing facility,and minimal or no analyst intervention.[28,39] However, thistechnique is formulation-specic and requires calibrationbased on reference methods. In this technique, the measur-ing window sometimes becomes coated with the wet pow-der, making a moisture measurement impossible or false.However, the fouling problem of the window of the NIRmeasurement can be generally solved by the use of a suitableair supply system and specialized interfacing in order togain reliable data during the drying process.[4] The equip-ment cost is relatively high and multiple sensors must beinstalled to monitor local moisture content or particle sizein a uidized bed.[59] This technique is insensitive to impu-rities and only the surface moisture of the material can bedetermined due to the short wavelength. Absorption spectrastrongly depend on the particles distribution within thebed, the powder density of the solid material,[60] and sampletemperature. NIR spectroscopy needs black-box multi-variate calibration techniques and sophisticated softwarefor interpretation of very diffuse and non-specic spectraldata.[61] The accuracy of the measurement is confoundedby the alteration in the chemical composition of the solidduring processing and the O-H band interferes with otherbands of interest.[62] On the other hand, scattering andabsorptive attributes of solid product differed because ofmodications in the color and surface structure of the par-ticle. Location of the NIR within the bed sensor is a criticalparameter in the measurements and it is useless for monitor-ing of bed hydrodynamic behavior or assuring of spatial
MONITORING OF FLUIDIZATION QUALITY IN DRYERS 1017
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
resolution. It is worth mentioning that the NIR imaging sys-tem (hyperspectral imaging spectrometer) has been developedwithin the last few years with the ability to simultaneouslyrecord spectral and spatial information of particles. Thisnew system can be a useful tool in the future for real-timemonitoring of particle physiochemical attributes and bedhydrodynamics. The moisture content affects the measure-ment in monitoring the particle size,[63] and it gives only stat-istical characteristics without giving actual information aboutparticle size.[64] This technique, especially the IR moisturesensor, is not actually non-intrusive due to its direct effecton the heat and mass transfer coefcient.
Pressure Fluctuations
Pressure uctuations in a uidized bed are mainly relatedto the formation, rise, and collapse or eruption of bubbles,clusters, and agglomerates.[65] However, transient pressureuctuations are quite complex and dynamic phenomenawhose exact origin is not yet entirely understood.[66] Inthe case of FBDs, variation of the water content of particlesinuences bubble characteristics. Therefore, monitoringand proper analysis of pressure uctuations can lead to amore detailed and deeper insight into the process andpossibly generate new ideas for improvement. Real-timecondition monitoring of uidization, particle size, andmoisture content can be obtained by simple measurementof pressure uctuations. Nevertheless, interpretation andunderstanding of pressure signals is complicated due totheir intrinsically non-local nature.[66] It should be men-tioned that pressure uctuations measurement is not a noveltechnique for monitoring of FBDs. However, various inno-vative methods, including statistical, frequency domain(fast Fourier and wavelet), nonlinear (fractal, chaos), andrecurrent plot analyses, have been developed and utilizedin recent years for interpretation of pressure uctuationsof uidized beds.[6771] Recently, several researchers haveattempted to review papers on measurement and analysisof pressure uctuations in uidized beds. Bi[72] criticallyreviewed the complex pressure uctuations phenomenonin gassolid uidized beds. Sasic et al.[73] reviewed bothmodeling and experimental techniques for investigatingthe uid-dynamic behavior of gassolid uidized beds usingpressure signals. Van Ommen and Mudde[6] focused onmeasurement techniques employed for elucidating the voi-dage distribution in gas-solid uidized beds. Van Ommenet al.[66] provided a critical review of time-series analysistechniques applied for interpreting the pressure signals inuidized beds. Table 3 summarizes some of the recentresearch and the most important results obtained usingpressure uctuations monitoring and analysis of FBDs.
Li et al.[74] found that the addition of smaller particlesincreases the frequency and amplitude of pressure uctua-tions and improves the gassolids contact of soybean inFBD. Chaplin et al.[75] studied the inuence of inlet air
temperature, initial mass of the wet bed, and pressure sensorposition on bed pressure uctuations as well as bedmass andPSD by the S-statistic. Chaos theory was applied to interpretand monitor the bed behavior for measuring the productstate and moisture content.[75] Hydrodynamic changes ofthe FBD specied by the S-statistic were profoundly inu-enced by the product moisture content. The S-statistic wassensitive to the PSD only in the dry bed and superior overthe frequency and amplitude analyses in identifying thehydrodynamic changes of the FBD. Chaplin et al.[76] contin-ued their studies by implementing S-statistic analysis ofpressure uctuations to a lab-scale FBD for online compari-son of the S-statistic with entrainment, bed temperature, andoutlet air temperature. The employed methodology was ableto give an early warning of the undesirable hydrodynamicstate by selecting an appropriate reference state. In a similarwork, the S-statistic of the pressure uctuations was notsensitive to the changes in particle size during uidizedbed granulating of Mannitol.[77] However, two differentregions in the initial stages of granulation and the nalstages of drying and granules moisture content weredetected by this method. Lopes et al.[78] compared visualobservations along with the statistical and spectral analysesof data obtained from online pressure uctuations measure-ments during spouted bed coating. It was shown that stat-istical analysis is an adequate technique for identicationof the spout instability while the dominant frequency wasnot suitable for distinguishing between the uidizationregimes perceived during coating.
Wormsbecker et al.[79,80] investigated the effect of vesselgeometry and uidization regime on hydrodynamics duringdrying of placebo pharmaceutical granules. They foundtransitions from high to low frequency in pressure uctua-tions of the conical bed and an increment in the bubblingfrequency of the cylindrical bed resulting from differentparticle circulations patterns, both prior to and after uidi-zation of the granule. They also reported a multiple bub-bling regime and a coalescence-dominated regime in theconstant rate and falling rate periods, respectively. The sim-ple bed pressure drop and temperature measurements wereable to detect the end-point of the granulation process.[81]
Karimi et al.[65] decomposed the raw pressure uctuationsinto 10 sub-signals by wavelet transform and successfullydeveloped a linear relationship based on the seventh sub-signal, corresponding to the macrostructure (large bubbles)and the supercial air velocity to predict the moisture con-tent of wet rice through FBD. Three various approachesof signal processing, used to pressure uctuation measure-ments, including dominant frequency analysis, narrow-band standard deviation analysis, and attractor compari-son, were compared to extract quantitative informationabout the granule size in a uidized bed granulator.[82]
The standard deviation of the narrow band ltered signalwas satisfactorily applied to monitor particle size and
1018 AGHBASHLO ET AL.
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
TABLE3
Someoftherecentresearchandthemostimportantresultsobtained
usingpressure
uctuationsmonitoringandanalysisofFBDs
Author(s)
Technique(s)
Application(s)
Dryer
type(s)
Process
target(s)
Rem
arks
Liet
al.[74]
Pressure
uctuations
withstandard
deviationand
power
spectra
Tostudytheuidization
velocities,mixing
mechanisms,and
uidizationquality
Lab-scaleFBD
Dryingofsoybeanseeds
Profoundimprovem
entwas
obtained
ingassolids
contactingbyintroducing
smallparticlesinto
thebed
oflargeparticles.
Chaplinet
al.[75]Pressure
uctuationin
conjunctionwith
chaosanalysis
(S-statistic)
Totrack
thebed
hydrodynamic
Experimentalconical
lab-scaleFBDs
Dryingofwet
placebo
granulecontaining
lactose
monohydrate
and
microcrystallinecellulose
asller,croscarm
ellose
sodium
asdisintegrant,
hydroxypropyl
methylcellulose
asbinder
andUSPwaterassolvent
S-statisticwassuperiorin
identifyingtheuidized
bed
state
over
thestandard
deviationordominant
frequency
techniques.
Chaplinet
al.[76]Pressure
uctuation
withS-statistic
Tostudythebed
hydrodynamic
Lab-scaleFBD
Dryingofgranule
consistinglactose
monohydrate
(ller),
microcrystallinecellulose
(ller),croscarm
ellose
sodium
(disintegrant),
hydroxypropyl
methylcellulose
(binder),
andwater(solvent)
Main
hydrodynamic
variationswerespecied
by
theS-statisticanalysisof
high-frequency
pressure
uctuationdata.
Chaplinet
al.[77]Pressure
uctuation
analyzedby
S-statistic
Tomonitorthemoisture
content
Experimentaltop-spray
uidized
bed
granulator
Mannitolgranulatingwith
hydroxypropylcellulose
asbindingsolutionand
waterassolvent
Granulemoisture
changes
within
thegranulator
monitoredbyusingthe
applied
techniquewithout
theneedforthedirect
measurementofmoisture.
Lopes
etal.[78]
Pressure
uctuations
alongwiththe
statisticandspectral
analysis
Tomonitorthechanges
thatoccurred
during
particlecoatingin
aspoutedbed
byusing
real-timepressure
uctuation
measurement
Lab-scalecone-
cylindricalspoutFBD
ABSandpolystyrene
particlescoatingby
EudragitL30-D
551
basedpolymeric
suspension
Statisticalmethod
satisfactorily
differentiated
thespoutuidized
bed
instabilitybasedonpressure
uctuationdata.
(Continued
)
1019
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
TABLE3
Continued
Author(s)
Technique(s)
Application(s)
Dryer
type(s)
Process
target(s)
Rem
arks
Worm
sbecker
etal.[79]
Pressure
uctuation
analysisin
conjunctionwith
averagecycle
frequency
and
standard
deviation
analyses
Tounderstandthebed
hydrodynamic
behavior
ExperimentalFBD
Dryingofwet
placebo
granulecontaining
lactose
monohydrate
and
microcrystallinecellulose
asller,croscarm
ellose
sodium
asdisintegrant,
hydroxypropyl
methylcellulose
asbinder,
andUSPwaterassolvent
Thepotentialofusingpressure
uctuationmeasurementsto
monitorandcontrol
uidizationstate
was
dem
onstrated.
Worm
sbecker
etal.[80]
Pressure
uctuation
analysiswithboth
timedomain
(standard
deviation
andaveragecycle
frequency)and
frequency
domain
(dominant
frequency
and
power
spectra)
analyses
Tostudytheinuence
of
vesselgeometry
onthe
hydrodynamic
behavior
Cylindricalandconical
laboratory-scalebatch
FBDs
Dryingofgranule
containingcroscarm
ellose
sodium
(disintegrant),
lactose
monohydrate
(ller),microcrystalline
cellulose
(ller),
hydroxypropyl
methylcellulose
(binder).
andwater(solvent)
Thedominantfrequency
decreasedduringdryingin
theconicalbed,whileit
increasedduringdryingin
thecylindricalbed.
Royet
al.[81]
Pressure
dropand
temperature
measurement
Todetectthegranulation
end-point
Experimentaltop-spray
uidized
bed
granulator
Granulationofurea
Thebed
pressure
dropand
temperature
were
successfullyauthenticated
foridentifyingtheend-point
ofuidized
bed
granulation.
Karimiet
al.[65]
Pressure
uctuations
withtime-domain
andfrequency
domain
analyses
Tomonitorthemoisture
contentandbed
hydrodynamic
ExperimentalFBD
Dryingofwettedrice
particles
Theoriginalpressure
uctuationsignalwas
decomposedinto
10
subsignalswhichweremore
sensitiveto
moisture
variationsthanother
investigatedparameters.
DeMartin
etal.[82]
Pressure
uctuation
withthreedifferent
techniques
ofsignal
processing,
includingdominant
frequency
analysis,
narrow-band
Tomonitorthemoisture
contentandparticle
size
duringuidized
bed
granulation
Lab-scaleuidized
bed
granulator
Granulationofmixture
of
lactose
monohydrate
and
starchwith
hydroxypropylcellulose
asbindingsolutionand
waterassolvent
Theparticlesize
andwater
contentwerecorrelatedto
thedata
derived
withthe
pressure
uctuationsensor
bynarrow-bandstandard
deviationsignalprocessing
technique.
1020
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
standard
deviation
analysis,and
attractor
comparisontool
Silvaet
al.[83]
Pressure
uctuationin
conjunctionwith
Gaussianspectral
pressure
distributionand
controlofthe
airowrate
andthe
coatingsuspension
owrate
usingPI
controllers
Tomonitorandcontrol
thedeuidization
phenomenonin
auidized
bed
coating
process
basedon
pressure
uctuation
data
Experimentaluidized
bed
coater
Microcrystallinecellulose
coating
Gaussianspectralanalysisof
pressure
uctuationdata
indicatedhighpotentialfor
applicationsin
uidized
bed
coatingprocesses,whilethe
PIcontroller
successfully
maintained
uidization
dynamicin
stablestate
basedonpresentedanalysis.
Prata
etal.[84]
Pressure
uctuation
measurementfor
process
control
Topreventthe
agglomerationby
pausingtheliquid
injection
Lab-scaleuidized
bed
coaterequipped
with
two-uid
nozzle
Microcellulose
beads
coatingwithgum
Arabic
Thedeveloped
controller
effectivelycontrolled
uidized
bed
coating,while
granulesagglomerationwas
avoided
bystoppingthe
liquid
sprayingbasedonbed
pressure
measurements.
Donget
al.[85]
Pressure
uctuations
withstatistical
methods
Toquantify
thebed
hydrodynamics
ExperimentalFBD
Dryingofspentliquor
mixed
withcorn
branfor
theproductionofyeast
Thebed
hydrodynamic
behaviorwasreasonably
qualied
bymeasuringthe
localbed
pressure
uctuations,exhaustair
temperature,andrelative
humidity.
1021
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
moisture content during uidized bed drying and granulationprocesses. Silva et al.[83] developed PI and PID controllers forcontrolling drying air ow rate and spraying liquid to avoidthe deuidization phenomenon during the uidized bed coat-ing, according to the Gaussian spectral analysis of pressureuctuations. The PI controller showed a good performanceover the PID controller and subsequently the stable uidiza-tion conditions were obtained using the PI control system.Prata et al.[84] designed a single-input=single-output controlstrategy based on the bed pressure measurement to preventagglomeration of granules by manipulating liquid sprayingduring uidized bed coating. Agglomeration was effectivelyavoided and homogeneous coating and separated particleswere obtained by the developed control system. Recently,Dong et al.[85] found that a small variation in the averagepressure drop and the standard deviation can be used forearly warning of the transition of the bed state from channel-ing to uniform uidization.
Pressure uctuation measurement is sensitive, accurate,fast, robust, relatively cheap, virtually non-intrusive, andrelatively easy to implement in lab-, pilot-, and industrial-scale units, even under harsh conditions. This techniquecan be considered to be a truly non-intrusive technique ifthe pressure transducer is ush-mounted at the vessel wallor if differential pressure measurements are applied[75];thus, distortion of the ow around the point of measure-ment is avoided.[66] Otherwise, this technique can modifythe local hydrodynamics of the uidized bed and mightnot be a reliable means of measurement.[86] This techniqueprovides information only about global or time-averagedhydrodynamic behavior of the bed. Therefore, it is uselessfor monitoring the local uidization phenomena inside thebed[87] and ascertaining the location in the bed in whichvariations in the dynamic behavior are taking place duringdrying.[88] On the other hand, signal processing is an essen-tial tool for extracting information from the recoded press-ure uctuations related with the particle physical propertiesand the bed hydrodynamics. Unfortunately, few techniquesare available for successful and satisfactory processing ofpressure signals to monitor physical properties of particlesbeing processed, such as signal energy, average cycle time,dominant frequency, and attractor comparison tools.Pressure measurement needs intrusive pressure taps andthe pressure transducer needs to be placed inside the pro-cess itself for industrial and experimental applications. Thistechnique does not provide detailed information about bedheight of uidization media and clear knowledge duringprocessing of very ne particles. Furthermore, to preventfouling of the pressure transducers with ne wet powder,continuous back-ushing with costly pressurized air or amechanical scraper is necessary, which in turn diminishesthe sensitivity of the probe.[89] Pressure uctuations analy-sis provides statistical information without resolvingspecic particle sizes.[64] In addition, identication of the
source of uctuations amongst many simultaneouslyoccurring phenomena is very difcult due to the extremelycomplex local ow structure through the uidized bed.[86]
Optical Imaging Technique
Optical imaging is one of the earliest techniquesemployed in the uidized bed drying process for real-timedetermination of physical properties of the granule,including granule size, PSD, and shape,[4] as well as theuidization regime. This transforms an image captured bya charge coupled device (CCD) camera into the digital formand performs some pretreatments on it, such as ltration,noise reduction, and pattern recognition, in order to obtaina rened image or to extract useful insights from it.[3] Imageacquisition, preprocessing, segmentation, extraction, andrepresentation of the characteristic parameters are the mainsteps in image processing analysis.[90] Recently, particleimage velocimetry (PIV) has been introduced for determin-ing velocity elds by capturing two images shortly aftereach other and computing the distance individual particlestraveled within this time. From the specied time intervaland the recorded movement, the instantaneous velocityvector eld can be identied in a cross-section of a ow.As illustrated in Fig. 2, the PIV apparatus contains aCCD camera, a strobe or laser with an optical arrangementto conne the physical region illuminated, a synchronizer toperform as an external trigger for adjusting the CCDcamera and laser and the seeding particles. There areremarkable applications of the optical imaging techniquein the literature to characterize uidized bed hydrodyn-amics as well as physical properties of particles, which aresummarized in Table 4.
The rst application of image processing in FBD goesback to 1995, when Watano and Miyanami[91] developeda system based on the image processing technique for onlinemonitoring of PSD and shape of granules in uidized bedgranulation. The particle imaging probe was able to predictthe granule shape precisely, being equivalent to the accu-racy of the ofine sieve method. They extended their inves-tigations to more precise controlling of FBD using an imageprocessing system supplemented with fuzzy rules,[92] a fuzzylogic controller based on image processing technique,[93]
and an adaptive feedback fuzzy controller based on aparticle image probe and an image processing system.[94]
Saadevandi and Turton[95] found that particle velocityand bed voidage are functions of both axial and radialpositions using the data obtained by the video imagingtechnique. However, the spray rates did not inuence theparticle velocity and voidage measurements. Particle trajec-tories at different particle loading, jet air velocity, andposition of the Wurster tube were investigated using theuorescent technique by Karlsson et al.[96] Trajectories ofparticles were successfully tracked, even with unexpectedparticleparticle collision and particlewall collision.
1022 AGHBASHLO ET AL.
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
Narvanen et al.[97] employed a 3D image analysis techniquefor determination of particle size within a uidized bedgranulator. Three basic colors, red-green-blue (RGB), wereused to lighten a at granule bed surface from threedifferent orientations. The PSD obtained by the image pro-cessing method satisfactorily corresponded to those of thesieve analysis. Good agreement was observed between theresults of confocal laser scanning microscopy (CLSM)and the chemical analysis in determining the coating thick-ness of microparticles through the uidized bed coater.[98]
It is worth mentioning that the CLSM is a technique forcapturing high-resolution optical images with depth selec-tivity. The key property of the confocal microscopy is itscapability to capture blur-free images of thick specimensat different depths, a process known as optical sectioning.Images are acquired locally and reconstructed with anappropriate computer program, permitting 3D reconstruc-tions of topologically intricate objects. Mozina et al.[99]
employed digital visual imaging to assess the sphericaldiameter, coating thickness, and undesirable agglomerationof pellets as well as classication and analysis of pelletswithin the uidized bed coater. Accuracy, precision, stab-ility, and speed of the developed technique were conrmedby experimental results. Feasibility of image analysis withdifferent feature selection approaches for excludingirrelevant and redundant information was investigated tomonitor the coating thickness of pharmaceutical pelletsduring uidized coating.[100] A strong correlation wasfound between image features and process parameters.Wang et al.[101] applied PIV to visualize and classify annu-lar bed ow patterns in the bottom spray uidized bedcoater and accordingly detected three types of ow patternswithin the drying bed. The ow patterns were considerablyinuenced by the coating uniformity, which was determined
by color coating and subsequent tristimulus colorimetry ofinline samples. Liew et al.[102] quantied particle recircula-tion through the partition column using a high-speed-imaging-based visiometric process analyzer and ensemblecorrelation PIV. Their results were in good agreement withthe data obtained using an image tracking method.
The optical imagining technique is a real-time, non-intrusive, rapid, low-cost, efcient, repeatable, accurate,high-resolution, consistent, and objective inspection toolbased on image analysis. Image processing provides reliableinformation not only on the PSD, but also direct infor-mation on particle sizes. The PIV offers several advantages,such as simplicity of the experimental set-up and ease ofscale-up procedures, with applicability for non-intrusivelyobtaining a complete velocity vector eld.[102] However,the PIV can only detect close-wall velocity vector eldsand its measurement may be negatively inuenced by walleffects.[103] The CLSM has also several advantages, includ-ing high-resolution blur-free images, easy visualization of3D structures by stereo pairs, capability of controllingdepth of eld, elimination of background knowledge awayfrom the focal surface, and the capability of collectingsequential optical proles from coarse samples.[104] How-ever, image processing necessitates large computationalefforts for data processing due to the very large amountsof data generated when compared with other monitoringtechniques. Such a disadvantage would be diminished withfurther progress in computer technology. On the otherhand, the wavelength of light, the desired eld of inspection,and the pixel density of the digital camera conne the resol-ution of the optical imaging. Non-uniform illuminating,variable solid density in the bed, imperfections or dirt onthe bed wall, background light, and photo-bleaching cannegatively inuence the light intensity and following image
FIG. 2. Schematic diagram of typical PIV apparatus and its measurement principle.
MONITORING OF FLUIDIZATION QUALITY IN DRYERS 1023
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
TABLE4
ApplicationofimageprocessingformonitoringofFBDs
Author(s)
Technique(s)
Application(s)
Dryer
type(s)
Process
target(s)
Rem
arks
Watanoand
Miyanami[9
1]
Imageprocessing
Tomonitorthegranule
size
distributionand
shape
Laboratory-scaletop-
sprayagitateduidized
bed
granulator
Granulationoflactose
and
cornstarchwith
Hydroxypropylcellulose
Thegranulesize
obtained
usingimageprocessing
techniquewasagreed
well
withdata
determined
by
sieveanalysis.
Watanoet
al.[92]Im
ageprocessingin
combinationwith
fuzzylogic
Tocontrolthegranule
growth
Lab-scaletop-spray
agitateduidized
bed
granulator
Granulationofmixture
of
lactose
andcornstarch
granulatingwith
Hydroxypropylcellulose
solution
Thedeveloped
system
accurately
controlled
the
granulegrowth
atvarious
operatingconditionsand
samplesattributes.
Watanoand
Miyanami[9
3]
Intelligentcontrol
Tocontrolthegranule
size
byem
ploying
imageprocessing
techniquein
conjunctionwith
fuzzylogic
Dryer
type:Lab-scale
agitateduidized
bed
granulator
Mixture
oflactose
and
cornstarchgranulation
Particlesize
obtained
using
imageprocessingtechnique
precisely
agreed
withsieve
analysisandsubsequently
theuidized
bed
granulator
accurately
controlled
by
fuzzylogiccontroller.
Watano[94]
Imageprocessing
techniquewith
fuzzylogic
controller
Tomeasure
and
controlthe
granulegrowth
Lab-scaleagitated
top-sprayuidized
bed
granulator
Granulationof
pharm
aceuticalpowders
composedoflactose
and
cornstarchwithasolid
binder
byspraying
puried
water
Granulegrowth
inuidized
bed
granulationwasdirectly
andcontinuouslymonitored
usinganimageprocessing
system
andalsoan
automatedcontrolsystem
of
granulegrowth
basedona
fuzzycontrolsystem
was
presented.
Saadevandiand
Turton[95]
Video
imaging
technique
Tomeasure
the
velocity
ofparticle
andvoidageofbed
Experimental
bottom-sprayuidized
bed
coater
Coatingofglass
particleas
modelmaterial
Hydrodynamicsofuidized
particlespassingthroughthe
liquidsprayinasemicircular
uidized
bed
coatingdevice
andtheeffectsoftheliquid
sprayonparticlevelocity
andvoidageprolesin
the
sprayingzonewasexactly
surveyed.
1024
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
Karlssonetal.[96]High-speedvideo
cameraequipped
withanopticallter
withaUVlampto
excite
the
uorescence-m
arked
particles
Tofollowthetrajectory
ofsingleparticlesin
thefountain
region
Laboratory-scale
spoutedbed
coater
withdrafttube
Dryingofsandparticles
coatedwith
ethyl-cellulose
Auorescenttechniquewas
satisfactorily
applied
for
detailed
studiesofthe
trajectoriesofparticlesinthe
fountain
region.
Narvanen
etal.[97]
3Dtopographicimage
processing
Onlinemonitoringof
particlesize
distributionduring
uidized
bed
granulation
Bench-scaleuidized
bed
granulator
Theophyllin
anhydrate
and
a-lactose
monohydrate
granulatingwith7.5%
aqueoussolutionof
polyvinylpyrrolidone
Particlesize
measurementwas
satisfactorily
perform
edusing3D
imageprocessing
techniqueandtheresults
corresponded
quitewellto
those
ofoff-linesieve
analysis.
Depypere
etal.[98]
Confocallaser
scanning
microscopy
Todeterminethelm
coatingthicknessand
subsequentthickness
heterogeneity
Laboratory-scale
uidized
bed
withboth
thetop-sprayand
bottom-spray
conguration
Coatingofglass
beads,microbeadswith
sodium
caseinate,and
gelatinASF=A
aqueous
solutions
Theapplied
methodologywas
ableto
predictmicrocapsule
coatingthicknessdownto
11.5mm
.
Mozinaet
al.[99]
Digitalvisual
imaging
Todeterminethesize
andshapeofpellets,to
detecttheadverse
agglomerationof
pellets,andto
classify
thepellets
Full-scaleuidized
bed
coater
Pelletcoatingwithsolution
consistingthe
hypromellose
phthalate
anddibutylsebacate
inmixture
ofacetoneand
ethanol
Pelletsize
andcoating
thicknesswereeffectively
determined.
Kucheryavski
etal.[100]
Imageprocessing
withtwodifferent
feature
selection
approaches
Tomonitorthecoating
thicknessin
at-line
mode
Pilot-scaleuidized
bed
coater
Coatingofnonpareil
sugar=starchpellets
withacetaminophen
(Paracetamol),
Acryl-EZE1and
Opadry
Red
Theanglemeasure
approach
resultswerefoundto
be
more
subtleandconsistent
thanwavelet
decomposition
inmonitoringcoating
thickness.
Wanget
al.[101]
PIV
Tostudytheannularbed
owpatternsandits
effect
oncoating
uniform
ity
Lab-scalebottom
spray
uidized
bed
coater
Nonpareilscoatingwith
Hydroxypropyl
methylcellulose
Coatuniform
itywas
profoundly
affectedby
annularbed
owpatterns
identied
byPIV
.Liewet
al.[102]
High-speedimaging
coupledwith
ensemblePIV
Tomonitortheparticle
recirculationwithinthe
partitioncolumnand
toobtain
theparticle
displacement
probabilitydensity
function
Experimentalbottom
sprayuidized-bed
coater
Coatingofsugarpellets
with
hydroxypropylmethylcel-
lulose
Theparticledisplacement
probabilitydensity
function
wasconsistentwithresults
ofanimagetracking
method.
1025
Dow
nloa
ded
by [D
iego A
rroya
ve] a
t 21:0
6 10 J
anua
ry 20
16
-
attributes. Most imaging techniques are limited totranslucent media and two-dimensional images, apart fromconfocal laser scanning imaging, which is suitable forthree-dimensional imaging.[97] Thus, the optical imagingtechnique is not usable for large-scale FBDs in which opa-que metal chambers are used. Solids moisture content canaffect the reective attributes of the particles, particularlyfree surface water, and the captured image quality andcharacteristics.
Acoustic Emission (AE)
Many years earlier, human hearing was utilized tomonitor a drying process and the process end-point wasdetermined by listening to the acoustic signals emitted fromthe dryer.[63] An acoustic monitoring technique can beapplied in both active and passive modes. In the activeacoustics mode, an acoustic wave is transmitted into th