Delft University of Technology
Large-scale transportation and storage of wood pellets
Investigation of the change in physical properties
Gilvari, Hamid; van Battum, Coen H.H.; van Dijk, Simon A.; de Jong, Wiebren; Schott, Dingena L.
DOI
10.1016/j.partic.2020.12.006
Publication date
2021
Document Version
Final published version
Published in
Particuology
Citation (APA)
Gilvari, H., van Battum, C. H. H., van Dijk, S. A., de Jong, W., & Schott, D. L. (2021). Large-scale
transportation and storage of wood pellets: Investigation of the change in physical properties. Particuology,
57, 146-156. https://doi.org/10.1016/j.partic.2020.12.006
Important note
To cite this publication, please use the final published version (if applicable).
Please check the document version above.
Copyright
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons. Takedown policy
Please contact us and provide details if you believe this document breaches copyrights. We will remove access to the work immediately and investigate your claim.
This work is downloaded from Delft University of Technology.
ContentslistsavailableatScienceDirect
Particuology
j ou rn a l h om ep ag e :w w w . e l s e v i e r . c o m / l oc a t e / p a r t i c
Large-scale
transportation
and
storage
of
wood
pellets:
Investigation
of
the
change
in
physical
properties
Hamid
Gilvari
a,c,∗,
Coen
H.H.
van
Battum
a,
Simon
A.
van
Dijk
b,
Wiebren
de
Jong
c,
Dingena
L.
Schott
aaSectionofTransportEngineeringandLogistics,DepartmentofMaritimeandTransportTechnology,FacultyofMechanical,MaritimeandMaterials
Engineering,DelftUniversityofTechnology,TheNetherlands
bChemicalServices,UniperBeneluxNV,TheNetherlands
cSectionofLargeScaleEnergyStorage,DepartmentofProcess&Energy,FacultyofMechanical,MaritimeandMaterialsEngineering,DelftUniversityof
Technology,TheNetherlands
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:Received15October2020 Receivedinrevisedform 27November2020 Accepted8December2020 Availableonline21January2021 Keywords:
Woodpellets
Large-scaletransportation Durability
Finesanddust Breakage
Mechanicaldegradation
a
b
s
t
r
a
c
t
Thechangeinphysicalpropertiesofwoodpellets,withafocusonparticlesizedistributionsduetopellet breakageandattrition,wasstudiedinalarge-scale(∼450ton/h)transportationsystem.Criticallocations withahighprobabilityofbreakagethroughthewholetransportationsystemwerechosenandsampled tostudytheeffectoftransportationsystemdesignandoperationonthemechanicalpropertiesofpellets. Bulkdensity,mechanicaldurability,moisturecontent,andparticlesizedistributionofpelletswere char-acterizedforeachsample.Analysisofvarianceshowedthatthereweresignificantdifferencesbetween thepercentagesofsmallparticles(<5.6mm)inthesamplestakenatdifferentlocations,especiallyatone withaverticalfreefallof7.8m.Onaverage,thisrelativelylongdropincreasedtheproportionof parti-cles<5.6mminthesamplesfrom8.73%to14.09%,andthatofparticles<3.15mmfrom4.82%to9.01%. Moreover,themeasurementsshowedawidedeviationinthemechanicaldurabilityvalues,betweena minimumof90.8%andamaximumof98.7%,whichwerenotcorrelatedtothesamplingpointsbutrelated topelletproperties.Itcanbeconcludedthatpellettransportationsystemsrequiremorededicateddesign strategiestopreventbreakageandattrition.
©2021ChineseSocietyofParticuologyandInstituteofProcessEngineering,ChineseAcademyof Sciences.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http:// creativecommons.org/licenses/by/4.0/).
Introduction
Worldwideuseofwoodpelletsasarenewableenergycarrier
hasincreased sharply, from∼12millionmetric tonsin 2008to
56millionmetrictonsin2018(Calderón,Gauthier,&Jossart,2019).
Morethan27millionmetrictonsofwoodpelletswereconsumed
inEuropein2018,morethan45%ofthemforindustrialpurposes.
However,notallpelletsconsumedinEuropeareproducedby
Euro-peancountries. Approximatelyone-thirdareimportedfromthe
UnitedStatesandCanada(Calderónetal.,2019).
Duetoitsinherentlyhighmoisturecontentanditslowbulk
andenergydensities,biomassisusuallydensifiedtoimprovethese
properties;aprocessthatisadvantageousfortransportationand
handling steps(Tumuluru, Wright,Hess,& Kenney,2011).
Pel-letsarenormallyproducedunderhightemperatureandpressure
∗ Correspondingauthor.
E-mailaddress:h.gilvari@tudelft.nl(H.Gilvari).
conditionsinaso-calledpelletizationprocess,whichinvolvesthe
elasticandplasticdeformationofparticlesandthesofteningof
nat-uralbinderssuchasstarch,protein,lignin,fat,andfiberstohelp
agglomerateparticles(deSouzaetal.,2020;Kaliyan&VanceMorey,
2009;Mani,Tabil,&Sokhansanj,2008;Nanou,Huijgen,Carbo,&
Kiel,2018).
Thefragilenatureofpelletscausestheirattritionandbreakage
throughouttheentirelogisticchain(Oveisietal.,2013).Asaresult,
averagepelletlengthmaydecreaseandtheamountofgenerated
smallparticlescanincrease(Boac,Casada,&Maghirang,2008).This
hasconsequencesforpellettransportation,handling,andstorage,
suchasanelevatedriskofdustexplosion,fire,segregation,
arch-ingandequipmentfouling,healthissuesforpeopleinhalingthe
dust,andlosinganotableportionofthematerial(Aarseth,2004;
Ilic,Williams,Farnish,Webb,&Liu,2018;Ramírez-Gómez,2016).
Moreover,pelletbreakageand thegenerationofsmallparticles
underminethepelletizationeffortbydecreasingthebulkdensity
(Sjöström&Blomqvist,2014).
https://doi.org/10.1016/j.partic.2020.12.006
1674-2001/©2021ChineseSocietyofParticuologyandInstituteofProcessEngineering,ChineseAcademyofSciences.PublishedbyElsevierB.V.Thisisanopenaccessarticle undertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).
Theabilityofbiomasspelletstoremainintactduringloading,
unloading,feeding,andtransportiscalled“mechanicaldurability”
(ISO,2014).Fromthedefinitionitself,themechanical durability
cangiveanindicationoftheextenttowhichthematerialcankeep
itsinitialshape.Inpractice,however,durabilityischaracterized
bymeasuringtheamountofpre-definedsmall-sizeparticles
cre-atedinlaboratorytests.Thissmallsizemaybedifferentinsomeof
theliterature,butasizeofbetween3and5mmisusually
consid-ered(Grahametal.,2016;ISO,2015).Accordingtotheliterature,
mechanical durabilitydependsonthecharacteristicsoftheraw
material;specifically,factorssuchasparticlesizeand moisture
content,pelletizationprocessparameters,andstorageconditions
(Gilvari,deJong,&Schott,2019;Shuietal.,2020;Stelteetal.,2012;
Whittaker&Shield,2017).
Undoubtedly,theextentofpelletbreakageandattrition very
muchdependsonthetypeofpellet;thehigherthepellet’sstrength,
theloweritsbreakageandattritionratesare.However,pelletsmay
undergomechanicaldegradationduetohighimpactsand
compres-sionforcescausedby,forinstance,longdrops,increasednumber
of handlingsteps,andpoorequipmentdesign (Ilic etal., 2018;
Ramírez-Gómez,2016).
Different breakage mechanisms have been proposed for
biomasspelletsduringtransportationandhandling,namely
break-age into two or more pellets while practically keeping the
cylindricalshapes,attritionofpelletsurfacesandends,and
crush-ingofthewholepellet(Boacetal.,2008;Thomas,1998).
Inindustry,variousequipmentisusedforthetransportation,
handling,andstorageofpellets.Ofthese,grabs,beltconveyors,
pneumatic conveyors, pipe conveyors,hoppers, transferchutes,
silos, and bins are the most common (Dafnomilis, Lodewijks,
Junginger,&Schott,2018;Larsson,Lestander,Crompton,Melin,&
Sokhansanj,2012).Eachofthesetypesofequipmentmaydegrade
thepelletsmechanically,duetoattrition,compression,orimpact.
Recently,researchhasbeenunderwaytocharacterizethebreakage
andattritionofpellets(Boacetal.,2008;Gilvari,deJong,&Schott,
2020; Murtala,Zigan, Michael, &Torbjörn,2020; Ochkin-König,
Heinrich,&Dosta,2018).Forinstance,researchonpneumatic
con-veyorshasshownthatpelletattritionincreaseswiththeincreasein
particlevelocity,whichisinducedbyincreasingtheairinlet
veloc-ityanddecreasingthemassflowofpellets.Moreover,ashorter
bendradiusincreasesthebreakageofpellets(Aarseth,2004;Jägers,
Wirtz,Scherer,&Behr,2020).
Itisknownfrompreviousstudiesthatthemagnitudeofimpact
force, number of handling steps, and amount of bulk material
directlyinfluencethebreakageandattritionofpellets(Boacetal.,
2008;Kotzur,Berry,Bradley,Dias,&Silva,2016;Oveisietal.,2013)
investigatedtheeffectofdropheight,pelletmass,andrepeated
handlingstepsonthedegradationofwoodpelletsduringfreefall
while theyplaced thepelletsin abag madeofsynthetic
mate-rials. Theyalsostudiedtheeffectof pelletmassper bagand of
repeatedhandlingsteps,concludingthatgreaterdropheightsand
anelevatednumberofhandlingstepsincreasetheamountoffines.
Moreover,pelletmassshowsalinearcorrelationwiththe
gener-atedmassoffinesaslongastheinitialmassiskeptlowerthan
1kg.
Looking atcorn-basedanimalfeedpellets,Boacetal.(2008)
studiedtheeffectofthenumberofhandlingstepsonthequantity
ofbrokenpellets(particlespassingasievewitha5.6mmmesh)
anddustparticles(<0.125mm)generated.Thefeedpellets
con-tained13.2%moistureandhadanominaldiameterof6.4mmand
adurabilityof92.9%,accordingtothetumblingboxmethod.Their
testsetupincludedabucketelevatorwithaheightof54.9m,which
cycledthepelletsfromthebottomofstoragebinstotheirtopand
fromafirstbintoasecondone,thenviceversa,withanaverage
flowrateof59.4metrictonsperhour.Theyrepeatedthedischarge
andloadingofthebinseighttimesandobservedanotableincrease
intheproportionofparticles<5.6mm,from17.5%to50.2%.The
averagedustgenerationwas0.069%pertransfer.
Asshownabove,existingliteratureregardingthemechanical
strength,breakagebehavior,andgenerationoffinesanddustof
biomasspelletsislimitedmainlytolaboratoryorpilot-scale
stud-ies.The objective of this paper, therefore, is to quantify small
particlesatdifferentpositionsinalarge-scale“realworld”
trans-portationsystem(∼450ton/h).Thisshouldallowustodetermine
themajortransportstepsinwhichpelletbreakageandattrition
occurand one can useit as a benchmarkfor investigating the
changesinpelletpropertiesinanyotherpellettransportsystem.In
addition,thechangesinotherpelletproperties—suchas
mechani-caldurability,bulkdensity,andmoisturecontent—duetomultiple
transportationstepsarestudied.For thatpurpose,apellet-fired
powerplantintheNetherlandswaschosenasacasestudyto
inves-tigatethebreakageandattritionofwoodpellets.Thetransportation
systeminthispowerplantconsistsofagrabunloadingsystem,belt
conveyors,andstorageinasilo.
Materialsandmethods
Thematerialforthiscasestudyoriginatedwithananonymous
companyintheUSA.Awiderangeofwoodyfeedstockwasused
toproducethepellets,includingboth softand hard woods.No
informationaboutthedensificationprocesswasdisclosed.After
theirproductionand localstorage inthe USA,thepelletswere
transportedinbulktoalocalportandloadedintoanocean-going
vesselwithacapacityof28,000metrictons,thenshippedacross
theAtlanticOceantotheportofAntwerp,Belgium.Heretheywere
transshippedtobargesholding∼2,500metrictonseach.Twelve
bargeswerethusrequiredfor theentiretransatlantic cargo.No
informationwasdisclosedregardingchangestotheparticlesize
distributionofthepelletsuptothisstage.Thetwelvebarges
pro-ceededtotheportofRotterdam,wheretheyberthednexttothe
enduser’splantatarateofonebargeperday.Thecargocomprised
amixtureofsevendifferenttypesofwoodpellets,whichdiffered
bycolor,diameter,andlengthasshowninTable1.
Physicalpropertiesofthepellets,includingtheirmoisture
con-tent, bulk density, diameter,length, and mechanical durability,
weremeasuredasfollows.Moisturecontentwasmeasured
accord-ingtoENstandard14774-2(EN,2009),using300gofpelletsthat
wereplacedinanovenat105◦Cfor24h.Themoisturecontent
wasthencalculatedbasedonthedifferencebetweenthemassof
thepelletsbeforeandafterthetest.Bulkdensitywasmeasured
accordingtoENstandard15103(EN,2010)bymeansofa5Lsteel
cylinderusing thetapmethod.Thebulk densitywascalculated
fromthemassofpelletsdividedbythevolumeofthecylinder.
Pel-letdiametersweremeasuredaccordingtoENstandard16127(EN,
2012)bymeansofadigitalcaliper.Twenty-eightpelletswere
mea-suredinthisway:fourperpellettype,chosenatrandom(Table1).
PelletlengthdistributionsweremeasuredaccordingtoEN
stan-dard16127(EN,2012),usinganin-houseimageprocessingtool;
detailsofthiscanbefoundinourpreviouspublication(Gilvari,De
Jong,&Schott,2020).Mechanicaldurabilitywasmeasured
accord-ingtoISOstandard17831-1(ISO,2015),whichisacommonglobal
methodfordurabilitymeasurementofbiomasspelletsandisused
extensivelybyresearchersinthisfield(Brunerová,Müller, ˇSleger,
Ambarita,&Valáˇsek,2018;Dyjakon&Noszczyk,2019;Gilvari,Cutz
etal.,2020;Larsson&Samuelsson,2017;Stelteetal.,2012;Thrän
etal.,2016).Themechanicaldurabilitywasdeterminedby
plac-ing500±10gofsievedpellets(round-holedsievewithascreen
sizeof3.15mm)ina tumblingcan.Thiswasrotatedata speed
of50rpmfor10min.Afterthetest,thesamplewassievedagain
usingthesamesieve,thenweighed.Itsmechanicaldurabilitywas
Fig.1.A2-dimensionaloverviewofthetransportationsystem.Thenumbersshowthesamplinglocations.AdetailedoverviewoftheautomaticsamplerisshowninFig.2. Thesizeoftheequipmentmaynotrepresenttherealcase.
Table1
DifferenttypesofwoodpelletsshippedfromtheUSAtotheportofRotterdam.
Row Type1 Type2 Type3 Type4 Type5 Type6 Type7
1
2
3
4
sievebythemassoftheinitialsamplemultipliedby100.The mois-turecontent,bulkdensity,andmechanicaldurabilityexperiments wererepeatedtwice;thereportedvaluesarethusthemeanvalues ofduplicatemeasurements.
To determinethechangeinparticlesizedistributions ofthe samples,threedifferentsieveswithscreensizesof5.6mm(square holes),3.15mm(roundholes),and1mm(squareholes)wereused. Thesievewiththesmallestscreen sizedeterminedtheamount
of dust (< 1mm),the medium one determined the amount of fines(1mm<fines<3.15mm),andthelargestonewaschosento determinetheamountoflumps(3.15mm<lumps<5.6mm).All particleslongerthan5.6mmwereconsideredtobe“whole pel-lets”.Hereafter,allparticlessmallerthan3.15mmarereferredto as“finesanddust”,andallthosesmallerthan5.6mmas“small particles”.
Tocharacterizetheclassifiedparticles,everysamplewasfirst weighedandthensievedmanuallyusingthethreescreensizes.The massesofdust,fines,andlumpswerethenweighedandrecorded. ThepercentageofeachcategorywascalculatedusingEq.(1):
Pi=
mi
mt ×
100% (1)
wherePiisthepercentageofcategoryiinasample,miisthemass
ofcategoryiinthesample,andmtisthetotalmassofthesample.
Thepercentageof“wholepellets”wascalculatedbysubtractingthe
percentageofsmallparticles(<5.6mm)from100.
Samplinglocationsandmethods
OncethepelletshadarrivedinRotterdam,theywereunloaded
byaclamshellgrabintoahopper(seeFig.1).Thisfedtheincoming
materialontoacoveredbeltconveyor(hereafterreferredtoasthe
firstconveyor).Thegrab’scapacityis10metrictonsandtheheight
ofthehopperis4m.Adustfilterisinstalledjustafterthe
hop-pertocollectairbornedustbeforethematerialisloadedontothe
firstconveyor.Thistransportedthepelletsonanupwardinclineto
thetopofatransfertowerwithaheightof24.0m,ataspeedof
1.7m/s.Thelengthandwidthofthefirstconveyorare150.0and
1.4m,respectively.It,therefore,hasanangleofinclinationof9.2◦.
Atthetopofthetransfertower,thepelletsweredroppedfromthe
firstconveyorontoasecondonebymeansofafreefallwitha
ver-ticaldropof7.8m.Betweenthesetwoconveyorsisachainbucket
sampler,whichcollectedsamplesataconsistentcross-sectionof
thematerial’sstreamduringitsfreefall.Thisisanautomatic
sam-pler,alreadyinstalledinthesystemforon-sitesamplingpurposes.
Thesamplesittookweretransferredtoasmalldosingconveyor
(length3.0m,width0.2m),whichinturnfedarotarytubedivider
asshowninFig.2.Theroleoftherotarytubedividerwastodivide
asmallportionofthesamplesbyrotatingthematerialsinsucha
waythat,eventually,2kgofsampleswerecollectedevery11min.
Thesecondconveyorhasalengthof500mandtransferredthe
pelletsataspeedof2.6m/stothetopofthesilo,intowhichthey
weredroppedviaafreefall.Wheneverrequired,pelletsare
dis-chargedfromthebottomof thesilo forfurtherprocessing.The
samplesforthisstudyweretakenonthreeconsecutivenon-rainy
daysinthewinterof2019.Everyday,abargecarrying2500metric
tonsofpelletswasunloadedintothesystem.Sixdifferentlocations
throughouttheentiretransportationsystem,fromthebargetothe
silo,wereidentifiedforthesamplingusedinthisstudy.Samples
weretakenatdifferentincrementsfromdifferentsampling
loca-tions,andallwereanalyzedintermsofparticlesizedistribution,
mechanicaldurability,bulkdensity,andmoisturecontent.A
sum-maryofthesamplinglocations,daysofsampling,andnumberof
samplestakenateachlocationisgiveninTable2.Intotal,77
sam-pleswerecollectedforthisstudy.Allwereplacedinsealedplastic
Fig.2.Detailedoverviewoftheautomaticmechanicalsamplersysteminthe trans-fertower.
bagstopreventmoistureuptakeandanyfurtherchangestopellet
propertiesduetovaryingenvironmentalconditions.Thesamples
werethentransportedcarefullytoalaboratoryattheDelft
Univer-sityofTechnologyforfurtheranalysis.Thesamplinglocationsare
explainedindetailbelowandtheyareshowninFig.1.
Samplinglocation 1wassituated insidethebarge.Samples
werecollectedfromtheuppersurfaceofthepelletcargoat
dif-ferenttimeintervals.Atthislocation,thephysicalpropertiesofthe
woodpelletscanbecharacterizedjustbeforetheyentertheend
user’stransportationsystem.Asmentionedbefore,thepelletswere
unloadedfromthebargeusingaclamshellgrabwithacapacityof
10metrictons.Eachgrabcycletookapproximatelyoneminute,so
emptyingtheentirebarge(∼2500metrictons)tookmorethan4h.
Toensurethatthesamplesrepresentthepelletpropertiesat
dif-ferentlayersinsidethebarge,samplesweretakenevery30min.
Ninewerecollectedonthefirstdayandeightonthesecond,using
aplasticscooptogatherapproximately6kgofmaterialpersample.
Sampling location 2 was situated in an extraempty barge
(hereafter referred to as the second barge)positioned next to
thepellet-loadedbarge.Notethatnosuchsecondbargeisused
innormalplantoperations.However,thissamplinglocationwas
introducedinordertoinvestigatetheeffectonthepelletproperties
ofthemechanicalforcesexertedduringgrabbing.Atthislocation,
onegrabofthematerialwasdischargedeveryhourtocollectfive
individualgrabsintotal.Toobtainarepresentativesamplefrom
everydischargedgrab,thegrabcontentsweredumpedontoaflat
surfaceinthesecondbarge,creatingapileofwoodpellets.
Sam-plingwasthenperformedaccordingtoEN/TSstandard 14778-1
(EN/TS,2005)bytakingninesamplesofapproximately0.5kgeach
fromdifferentlocationswithinthepile(Fig.3),usingaplasticscoop.
Fig.3. Schematicrepresentationofthepileofpelletsinthesecondbargeandofthesamplestakenit(cyanmarkers).Notethattherealpiledeviatedfromaperfectcone shape.
Table2
Summaryofsamplinglocations,daysofsampling,andnumberofincrementsperlocation.
Samplinglocation Locationnumber Dayofsampling Numberofincrements
Barge 1 12 89 Secondbarge 2 1 5 Firstconveyor 3 1 8 2 9 Rotarydivider 4 3 17 Secondconveyor 5 12 89 Silo 6 3 4 Total – – 77
Fig.4.Isolationofthesamplesontheconveyorusingthestopped-beltmethod.
Theseninesampleswerecollectedfromthetop(onesample), mid-dle(threesamples),andbottomofthepile(fivesamples),andwere combinedtocreateonesinglesample.Samplingatthislocationwas undertakenonlyonthefirstday.
Samplinglocation3wassituatedinsidethetransfertowerat
theendofthefirstconveyor,whichbringsthepelletsfromthe hop-pertothetransfertower.Thissamplinglocationissituatedjust beforethepelletsweredischargedontothesecondconveyor(see
Fig.1).Thislocationwasusedtoestimatetheeffectonthematerial’s
propertiesofgrabbingandofaverticalfreefallof4m(thehopper’s
height).Notethatineachgrabbingcycle,thepelletsweredumped
atthecenterofthehopper,andsotheeffectonthemofhopperwall
impactscanbeconsiderednegligible.Thissamplinglocationis
use-fulasabenchmarktocheckthereliabilityofthepropertiesofthe
samplesdividedbytheautomaticsampleratlocation4.Atthis
loca-tion,samplesweretakenusingthestopped-beltmethodprovided
forinEN/TSstandard14778-1(EN/TS,2005).Inotherwords,the
entiretransportationsystemwashaltedatthesamplingtimesand
thesamplesweretakenfromthebeltconveyorafterbeingisolated
fromtheothermaterialonitbymeansoftwosteelplates(Fig.4).
Consequently,allthematerialonacross-sectionoftheconveyor
beltwascollectedandlabeledascomingfromsamplinglocation
Fig.5. Thesiloandsamplingports.
3attherelevanttime.Eachsamplecontainedapproximately6kg
ofmaterial.Aswithlocation1,samplingatthislocationwas
per-formedondaysoneandtwo.Moreover,allthesamplesherewere
collectedatthesametimeasthosefromlocation1(thebarge).In
all,17sampleswerethuscollectedatthislocation.
Samplinglocation4wasapartoftheautomaticmechanical
samplerwithinthetransfertower.Samplestakenatthislocation
canbeusedtoinvestigatethereliabilityofthematerialproperties
providedbytherotarydividerwithregardtotheotherlocations
inthetransportationsystem,e.g.attheendofthefirstconveyor.
Eachsamplewascollectedautomatically,fillingaplasticbagwith
approximately2kgofmaterialwithin11min.Sampleswereonly
takenatthislocationonthethirdday,in17increments.
Sampling location 5was situated atthe end of thesecond
conveyor,justbeforethematerialwasdischarged intothesilo.
Samplingatthis locationcandemonstrate theeffectofa 7.8m
verticalfreefall(betweenthetwoconveyors)onpellet
Table3
Pelletpropertiesatdifferentlocationsondifferentdaysofsampling.Numbersin parenthesesshowthestandarddeviations.
Property Location Dayofsampling Averagevalue
Moisturecontent(%) All All 5.9(0.3)
Bulkdensity(kg/m3) All All 640(25)
Mechanicaldurability(%) All All 97.6(1.3)
Diameter(mm) 3,4 2,3 7.11(0.59) Lengthdistributiona (mm) 1, 3, 4 1, 2, 3 50%<13.40 99%<32.70 100%<42.10
aBasedonthepelletlengthdistributionsofsamplesshowninFig.7.
wasassumedtobenegligible.Thestopped-beltmethodwiththe
sametypeofsteel-plateisolatorsasatsamplinglocation3(Fig.4)
wasusedhere,too.Samplesweretakenatthislocationondays
oneandtwoonly,eachweighingapproximately6kg.Onaverage,
thesewerecollectedthreeminutesaftereachofthoseatlocation
3,sincethatwashowlongittookforthepelletstobetransferred
fromtheretolocation5onthesecondconveyor.
Sampling location6wassituatedatthestoragesilo(witha
heightof31mandadiameterof9m)andwasusedonthethirdday.
Foursampleshavebeencollectedfromthefourhatchesinstalled
atthewallbecausesamplinginsidethesilowasimpossibleatthe
timeoftheexperiments.Thesehatcheswerelocatedattwolevels,
andtwosamplesweretakenateachlevel(seeFig.5).
Resultsanddiscussion
Mechanicaldurability,bulkdensity,andmoisturecontent
Fig.6showsthepelletpropertieswithregardtothesampling
locationsanddays.Analysisofvariance(ANOVA)ofall77samples
revealedthattherewasnobiasintheresultsforpelletpropertiesat
differentlocationsorondifferentdays(˛=0.05,i.e.95%confidence
interval).FromFig.6,itcanbeseenthatmechanical durability,
bulkdensity,andmoisturecontentareindependentofthedayof
samplingand ofthesamplinglocation.Moreover,therewasno
correlationbetweenbulkdensityandmoisturecontent.However,
themechanicaldurabilityvariedbetween90.8%and98.7%inall
locations,andthisexceededthe2%repeatabilitylimitsetbythe
standard(ISO,2015).Theaveragevaluesofpelletproperties
mea-suredatdifferentlocationsonthreedifferentdaysofsamplingcan
befoundinTable3.Thistableimpliesthatalthoughalarge
devi-Fig.7. Pelletlengthdistributionsofdifferentbatchesofpelletsusedfor mechani-caldurability(DU)test(legend:digitsfromlefttoright=samplingday-sampling location-numberofincrement-repetitionnumber).
Fig.8.Size-classifiedparticlesaftersieving.
ationindurability,onlya fewsampleswidelydivergefromthe
averagevalueandthattheaveragedurabilitycanbeconsideredas
reliabledatainourstudy.
Nocorrelationwasfoundbetweenthemechanicaldurability
ofthepelletsandthenumberofhandlingstepstheyunderwent.
Nevertheless,therearetwoexplanationsforthevariationwefound
inmechanicaldurability.First,inourpreviousstudy(Gilvari,De
Jongetal.,2020)weshowedthatpelletlengthdistributions(PLDs)
canbeacontributingfactorinthemeasuredmechanicaldurability
Fig.9.Sievinganalysisatlocations1(firstbarge),3(endofthefirstconveyor),and5(endofthesecondconveyor).Thesamplenumberscorrespondtothesamplingorder.
value,whichincreaseswithgreaterpelletlengths.Tostudythis
effect,thePLDsoftensamples(fiveoutof77samples,withtwo
replicationseach)weredeterminedbeforedurabilitytests(Fig.7)
usingourself-developedin-houseimageprocessingtool(Gilvari,
DeJongetal.,2020).AscanbeseeninFig.7,asamplecontainingthe
highestquantityofsmallparticles(<5.6mm)showedthelowest
mechanical durabilityvalue.Take,forexample,samples2-3-2-1
and2-3-2-2inFig.7.Thesesamplesweretworandomlyselected
500gfromtheinitial6kgsamplethatwastakenonday2atlocation
3asthethirdincrement.However,theirdurabilityvaluesdifferby
4.7%.Thisrelativelyhighdeviationcanconfirmtheeffectoflength
distributions.Second,astheshareofeachpellettypeineachsample
wasnotcountedhere,itissuspectedthatotherfactorssuchasthe
shareofpellettypemaycauseabiasintheresults.
Particlesizedistributions
Fig.8showsasampleimageoftheclassifiedparticles.Inthis,it
canbeseenthatallparticleslongerthan5.6mmarewholepellets
retainingtheircylindricalshape.
Asexplainedinsection2,samplesatlocations1and3were
col-lectedatthesametimeandeachdiscretesampleatlocation5was
collectedonaveragethreeminutesaftereachcollectionatlocations
1and3.Therefore,itislikelythatallcorrespondingsamples
origi-natedfromthesamelocationinthebarge.Hence,theeffectonthe
particlesizedistributionsattheselocationsofdifferent
transporta-tionunitscanbeinvestigatedwithregardtotheiroriginalparticle
size.
Fig.9showstheresultsofsievinganalysisatlocations1,3,and
Fig.10.Percentagesofsmallparticles(<5.6mm)atdifferentlocationsonthefirstandseconddaysofsampling.
Table4
Averageshareofcategorizedparticlesatlocations1,3,and5onthefirstandseconddaysofsampling.
Location Samplingday %ofsmallparticles(<5.6mm) %offinesanddust(<3.15mm) %ofdust(<1mm)
1(firstbarge) 1 10.28 6.11 2.50 2 10.40 5.93 2.27 3(endoffirst conveyor) 1 8.61 4.91 1.95 2 8.87 4.73 1.74 5(endofsecond conveyor) 1 14.24 9.05 3.81 2 13.93 8.98 3.77
sampling.Accordingtotheseresults,oneachdaythebargescontain anotableamountofsmallparticlesatdifferentsamplingtimesprior tounloading.Thisalsoshowsthatdifferentlayersofthebarge’s cargo(fromtoptobottom)containdifferentparticlesize distribu-tions.However,noevidencewasfoundregardingtheaccumulation ofsmallparticlesinonespecificlayeroratonespecificlocationin thebarge.Theresults(Table4)showthat,onaverage,theshareof
eachsizecategoryondayoneisconsistentwiththeresultsonday
two.
Thestatisticalanalysisshowsasignificantdifferencebetween
theamountoffinesanddustinthesamplescollectedat
differ-entlocations(˛=0.05,i.e.95%confidenceinterval).Themeasured
p-valueforone-wayANOVAanalysisbetweenthesampling
loca-tionsandtheamountoffinesanddust(<3.15mm)waszero,which
meansthattherewasasignificantdifferenceintheaveragevalues.
Onaverage,theproportionoffinesanddustwas6.02%atlocation
1(firstbarge),4.82%atlocation3(endofthefirstconveyor),and
9.01%atlocation5(endofthesecondconveyor)onthefirstand
seconddays.
Asimilartrendtothatforfinesanddust(<3.15mm)isobserved
forallsmallparticles(<5.6mm)aswell,sincetheANOVA
analy-sisshowssignificantdifferencesbetweenthepercentagesofthese
by samplinglocation.The averageproportion ofsmall particles
(< 5.6mm)ondaysoneand twowas10.33%atlocation1(first
barge),8.73%atlocation3(endofthefirstconveyor),and14.09%at
location5(endofthesecondconveyor).Thisindicatesthata
signif-icantquantityofsmallparticles(<5.6mm)isgeneratedthroughout
theentiretransportationsystem,fromthebargetotheendofthe
secondconveyorjustbeforethematerialentersthesilo.
Oneinterestingresultisthereductioninthenumberofsmall
particles(<5.6mm)afterthepelletsaregrabbedanddischarged
ontothefirstconveyor,i.e.betweenlocations1and3(Fig.10).The
proportionofsmallparticlesdecreasedrelativelybetweenthese
twolocationsbyupto16.24%onthefirstdayand14.71%onthe
secondday.Therearetwoexplanationsforthis.Firstly,anumber
ofsmallparticlesescapedfromthegrabintotheairwhilethegrab
wasloadinginthebargeandshiftingupwardstowardsthe
hop-per.Secondly,thedustfiltersatthebeginningofthefirstconveyor
removedsomedust;however,theaccumulateddustinthosefilters
wasnotmeasuredinthisstudy.Ontheotherhand,bycomparison
withlocation3(endofthefirstconveyor),theaverageproportion
ofsmallparticles(<5.6mm)atlocation5(endofthesecond
con-veyor)increasedfrom8.61%to14.24%onthefirstdayandfrom
8.87%to13.93%onthesecondday,showingarelativeincreaseof
65.74%and57.04%,respectively.Assumingthattheeffectof
con-veyorvibrationsonthegenerationofsmallparticles(<5.6mm)is
negligible,alloftheseextraonesmusthavebeengenerateddue
totheverticalfreefallof7.8matthetransferpointbetweenthe
twoconveyors.Boacetal.(Boacetal.(2008))observedanaverage
increaseof3.83%inthenumberofparticlessmallerthan5.6mmin
onetransportationcycleoffeedpelletsinabucketelevatorandsilo
feedingsystemforpelletswiththedurabilityof92.9%anda
nom-inalpelletdiameterof6.4mm.Inthisstudy,however,anaverage
5.36%increaseinthenumberofparticlessmallerthan5.6mmwas
observedsolelyduetoafreefallof7.8m.Thisisprobablyrelated
tothecompressionandimpactforcesassociatedwiththeamount
ofmaterialbeingtransported,whichwasaround450ton/hinthe
presentstudyand54.9ton/hintheworkofBoacetal.(Boacetal.
(2008)).
Theresultsshowthattheaverageproportionofsmallparticles
atlocation2(secondbarge)was6.56%,witha1.51%standard
devia-tion.Thiswaslowerthanthefigureforlocation1(10.34%).During
grabbingfromthefirstbargetothesecondone,itwasobserved
Fig.11.Shareofsmallparticlesatlocations2,4,and6.
leakage.Becauseofthis,themeasuredsmallparticlesarenot
rep-resentativeforthewholeparticlesandtheeffectofgrabbingonthe
generationofsmallparticlesremainsunclear.
Fig.11showsthatthepercentageofsmallparticlesatlocation
4 (automaticsampler)onthethirddaywas6.05%,witha
stan-dard deviationof 2.04%.Asdiscussed earlierinthissection,the
resultsfromthefirsttwodaysshowahighrepeatabilityinterms
ofgeneratingsmallparticlesateachsamplinglocation.Therefore,
theshareofsmallparticlesateachlocationondaysoneandtwo
canbepostulatedtobesimilartothatonthethirdday.Basedon
thisassumption,theresultsforthefirstandseconddaysatlocation
Fig.12.Correlationsbetweenclassifiedsmallparticles.
thirddayatlocation4(automaticsampler).Theaverage
propor-tionofsmallparticlesatlocation3onthefirstandseconddayswas
8.73%,whileatlocation4itwas6.05%.Thus,onaverage,thenumber
ofsmallparticlesdecreasedbetweenthesetwolocationsby2.68%,
showingthattheautomaticsampler(location4)doesnotproduce
similarparticlesizedistributionswhencomparedtolocation3.A
possiblereasonforthisisthesegregationofbiggerparticlesdue
totherotationofparticlesintherotarydivider,whichseparates
largerparticlesfromthesmalleronesinsuchawaythatthelatter
arerejectedasresidue(seeFig.2).
Theaveragepercentageofsmallparticles(<5.6mm)atlocation
6(silo)isshowninFig.11.Onaverage,thesemadeup4.00%of
thesamplestakenfromthesilohatches.Comparingtheproportion
ofsmallparticlesherewiththatatlocation5(endofthesecond
conveyor),itcanbeconcludedthatthewholepelletsaccumulated
atthesilowalls.Thiscanbeexplainedbypercolationsegregation,
wherebythesmallerparticlesarecapturedinthevoidsandbigger
onesmovedowntheslopetowardsthewalls.
Thomas(Thomas(1998))showedthatpelletfragmentationand
attritionarethetwomajormechanismsforphysicaldegradation
ofpellets.Fig.12showstheshareoflumps,fines,anddustinall
thesamplestakenfromdifferentlocationsandincludesasurface
fitthatisthebestfitforallthesamples.Althoughthebreakage
mechanismwasnotstudiedinthiswork,Fig.12showsthatthe
sharesofdust,fines,andlumpsatdifferentlocationsandon
differ-entsamplingdaysarecorrelatedtooneanotherwithanR2value
of0.866.Inotherwords,inallsamples,thequantityofeach
size-classifiedparticlewasincreasedbyincreasingtheamountofother
size-classifiedparticles.
Conclusions
Inthisstudy,wehaveinvestigatedthechangeinthephysical
propertiesofcommercialwoodpelletswithafocusonbreakage
and attrition behaviorduring large-scale(∼450ton/h)transport
andstorageinapellet-firedpowerplantintheNetherlands.Our
mainconclusionisthattransferringthewoodpelletsviaafreefall
fromaheightof7.8m(thetransferpointbetweentwo
convey-ors) increasestheproportionofsmallparticles(<5.6mm)from
8.74%to14.09%,onaverage,andtheamountoffinesanddustfrom
4.82%to9.01%forpelletswithanaveragemechanicaldurabilityof
97.6%(basedonISOstandard17831-1).Thisemphasizesthe
impor-tanceofequipmentdesignandoperationwithrespecttomaterial
degradation.Themechanicaldurabilityofthesamplestakenat
dif-ferentlocationsdifferedbyupto7.9%,whilenocorrelationwas
observedbetweenmoisturecontent,bulkdensity,sampling
loca-tion,ordayofsamplingandmechanicaldurability.Thissuggests
thatotherproperties,mostprobablypelletlengthdistribution,play
asignificantroleinthemeasuredmechanicaldurabilityvalue.
Declarationofinterests
Theauthorsdeclarethattheyhavenoknowncompeting
finan-cialinterestsorpersonalrelationshipsthatcouldhaveappearedto
influencetheworkreportedinthispaper.
Acknowledgements
The authors would like to thank Jurgen Visser and Ronald
KoomansofUniperBeneluxfortheircarefulplanningand
coor-dination,aswellasforsharingtheirsamplingexpertise.Wethank
EovarvanDijk,KoenJongste,andSjoerdSmidofPeterson
Rotter-damforcoordinatingandexecutingthesamplingonsite.Wealso
thankSjaakvanHurck(ESI/EurosiloBV),FrankvanderStoep(EBS),
LeoLokker(EMO),MarkBouwmeester,andEmielvanDorp(RWE)
forusefuldiscussionsinpreparingthemeasurements.Wethank
TjibbeBronandJorisBerendschot(DelftUniversityofTechnology)
fortheirsupportinanalyzingthesamples.
ThisstudyhasreceivedfundingfromtheTopConsortiumfor
KnowledgeandInnovationfortheBiobasedEconomy(TKI-BBE),
undergrantnumberBBE-1713(Biomassapellets:Degradatie
tij-denstransportenhandling).
References
Aarseth,K.A.(2004).Attritionoffeedpelletsduringpneumaticconveying:The influenceofvelocityandbendradius.BiosystEng,89,197–213.
Boac,J.M.,Casada,M.E.,&Maghirang,R.G.(2008).Feedpelletandcorndurability andbreakageduringrepeatedelevatorhandling.AppliedEngineeringin Agriculture,24,637–644.
Brunerová,A.,Müller,M., ˇSleger,V.,Ambarita,H.,&Valáˇsek,P.(2018).Bio-Pellet FuelfromOilPalmEmptyFruitBunches(EFB):UsingEuropeanStandardsfor QualityTesting.Sustainability,10,4443.
Calderón,C.,Gauthier,G.,&Jossart,J.-M.(2019).Bioenergyeuropestatisticalreport 2019-keyfinding.
Dafnomilis,I.,Lodewijks,G.,Junginger,M.,&Schott,D.L.(2018).Evaluationof woodpellethandlinginimportterminals.Biomass&Bioenergy,117,10–23.
deSouza,H.J.P.L.,Arantes,M.D.C.,Vidaurre,G.B.,Andrade,C.R.,AdeCO,C.,de Souza,D.P.L.,etal.(2020).Pelletizationofeucalyptuswoodandcoffee growingwastes:Strategiesforbiomassvalorizationandsustainablebioenergy production.RenewableEnergy,149,128–140.
Dyjakon,A.,&Noszczyk,T.(2019).Theinfluenceoffreezingtemperaturestorage onthemechanicaldurabilityofcommercialpelletsfrombiomass.Energies,12, 2627.
EN14774-2:2009.SolidBiofuels-DeterminationofMoistureContent-OvenDry Method-Part2:TotalMoisture-SimplifiedMethod,EuropeanCommitteefor Standardizationn.d.
EN15103.Solidbiofuels-Determinationofbulkdensity.2010,European CommitteeforStandardizationn.d.
EN16127.SolidBiofuels-Determinationoflengthanddiameterofpellets.2012, EuropeanCommitteeforStandardization.n.d.
EN/TS14778-1:2005.SolidBiofuels-Sampling-Part1:Methodsforsampling, EuropeanCommitteeforStandardizationn.d.
Gilvari,H.,deJong,W.,&Schott,D.L.(2019).Qualityparametersrelevantfor densificationofbio-materials:Measuringmethodsandaffectingfactors-A review.Biomass&Bioenergy,120,117–134.
Gilvari,H.,Cutz,L.,Tiringer,U.,Mol,A.,deJong,W.,&Schott,D.L.(2020).The effectofenvironmentalconditionsonthedegradationbehaviorofbiomass pellets.Polymers,12,970.
Gilvari,H.,deJong,W.,&Schott,D.L.(2020).Breakagebehaviorofbiomasspellets: Anexperimentalandnumericalstudy.CompPartMech.,https://doi.org/10. 1007/s40571-020-00352-3
Gilvari,H.,DeJong,W.,&Schott,D.L.(2020).Theeffectofbiomasspelletlength, testconditionsandtorrefactiononmechanicaldurabilitycharacteristics accordingtoISOstandard17831-1.Energies,13,3000.
Graham,S.,Ogunfayo,I.,Hall,M.R.,Snape,C.,Quick,W.,Weatherstone,S.,etal. (2016).Changesinmechanicalpropertiesofwoodpelletsduringartificial degradationinalaboratoryenvironment.FuelProcessTechnol.
Ilic,D.,Williams,K.,Farnish,R.,Webb,E.,&Liu,G.(2018).Onthechallengesfacing thehandlingofsolidbiomassfeedstocks.Biofuels,BioprodBiorefining,12, 187–202.
ISO16559:2014.Solidbiofuels-Terminology,definitionsanddescriptionsn.d. ISO17831-1:2015.Solidbiofuels-Determinationofmechanicaldurabilityof
pelletsandbriquettes-Part1:Pellets.,n.d.
Jägers,J.,Wirtz,S.,Scherer,V.,&Behr,M.(2020).Experimentalanalysisofwood pelletdegradationduringpneumaticconveyingprocesses.PowderTechnology, 359,282–291.
Kaliyan,N.,&VanceMorey,R.(2009).Factorsaffectingstrengthanddurabilityof densifiedbiomassproducts.Biomass&Bioenergy,33,337–359.
Kotzur,B.,Berry,R.,Bradley,M.,Dias,G.C.,&Silva,A.C.(2016).Influenceofpellet lengthonbreakagebyimpact.InProc.12thint.Conf.Bulkmater.storage,Handl. Transp.
Larsson,S.H.,&Samuelsson,R.(2017).PredictionofISO17831-1:2015 mechanicalbiofuelpelletdurabilityfromsinglepelletcharacterization.Fuel ProcessTechnol,163,8–15.
Larsson,S.H.,Lestander,T.A.,Crompton,D.,Melin,S.,&Sokhansanj,S.(2012).
Temperaturepatternsinlargescalewoodpelletsilostorage.AppliedEnergy, 92,322–327.
Mani,S.,Tabil,L.G.,&Sokhansanj,S.(2008).Effectsofcompressiveforce,particle sizeandmoisturecontentonmechanicalpropertiesofbiomasspelletsfrom grasses.Biomass&Bioenergy,30,648–654.
Murtala,M.A.,Zigan,S.,Michael,S.A.B.,&Torbjörn,A.L.(2020).Fuelpellet breakageinpneumatictransportanddurabilitytests.RenewableEnergy,157, 911–919.
Nanou,P.,Huijgen,W.J.J.,Carbo,M.C.,&Kiel,J.H.A.(2018).Theroleofligninin thedensificationoftorrefiedwoodinrelationtothefinalproductproperties. Biomass&Bioenergy,111,248–262.
Ochkin-König,O.,Heinrich,S.,&Dosta,M.(2018).Investigationofbreakageand attritionbehaviorofwoodpellets.In9thint.Conf.Conveyinghandl.Part. Solids-CHoPS.
Oveisi,E.,Lau,A.,Sokhansanj,S.,Lim,C.J.,Bi,X.,Larsson,S.H.,etal.(2013).
Breakagebehaviorofwoodpelletsduetofreefall.PowderTechnology,235, 493–499.
Ramírez-Gómez,Á.(2016).Researchneedsonbiomasscharacterizationtoprevent handlingproblemsandhazardsinindustry.PartSciTechnol,34,432–441.
Shui,T.,Khatri,V.,Chae,M.,Sokhansanj,S.,Choi,P.,&Bressler,D.C.(2020).
Developmentofatorrefiedwoodpelletbinderfromthecross-linkingbetween specifiedriskmaterials-derivedpeptidesandepoxidizedpoly(vinylalcohol). RenewableEnergy,162,71–80.
Sjöström,J.,&Blomqvist,P.(2014).Directmeasurementsofthermalpropertiesof woodpellets:Elevatedtemperatures,finefractionsandmoisturecontent.Fuel, 134,460–466.
Stelte,W.,Sanadi,A.R.,Shang,L.,Holm,J.K.,Ahrenfeldt,J.,&Henriksen,U.B. (2012).Recentdevelopmentsinbiomasspelletization–Areview.BioResources, 7,4451–4490.
Thomas,M.(1998).Physicalqualityofpelletedfeed:Afeedmodelstudy.
Thrän,D.,Witt,J.,Schaubach,K.,Kiel,J.,Carbo,M.,Maier,J.,etal.(2016).Moving torrefactiontowardsmarketintroduction–Technicalimprovementsand economic-environmentalassessmentalongtheoveralltorrefactionsupply chainthroughtheSECTORproject.Biomass&Bioenergy,89,184–200.
Tumuluru,J.S.,Wright,C.T.,Hess,J.R.,&Kenney,K.L.(2011).Areviewofbiomass densificationsystemstodevelopuniformfeedstockcommoditiesfor bioenergyapplication.Biofuels,BioprodBiorefining,5,683–707.
Whittaker,C.,&Shield,I.(2017).Factorsaffectingwood,energygrassandstraw pelletdurability–Areview.RenewableandSustainableEnergyReviews,71, 1–11.