The Prediction of The Flexural Lumber Properties From Standing South African-Grown Pinus Patula Trees

Thevariationandpredictionofstructurraltimberp
propertieso
ofstandingPinuspatulatrees
usingnondestructivemethods
by
CoenraadBrandW
Wessels
DisserttationpresentedforthedegreeofDoctorofScieenceinForesstry(WoodP
ProductsScieence)in
University
theFacultyofAgriSciences,atStellenboschU
Superviso
or:Prof.TRyypstraand
Cosupeervisor:Dr.FFSMalan
April2014
Stellenbosch University http://scholar.sun.ac.za
DECLARATION
Bysubmittingthisdissertationelectronically,Ideclarethattheentiretyoftheworkcontained
thereinismyown,originalwork,thatIamthesoleauthorthereof(savetotheextentexplicitly
otherwisestated),thatreproductionandpublicationthereofbyStellenboschUniversitywillnot
infringeanythirdpartyrightsandthatIhavenotpreviouslyinitsentiretyorinpartsubmitteditfor
obtaininganyqualification.
Thisdissertationincludesthreeoriginalpaperspublishedinpeerreviewedjournalsandtwo
unpublishedpaperscurrentlyinpreparationforsubmissiontoanaccreditedscientificjournal.The
developmentandwritingofthepapers(publishedandunpublished)weretheprincipalresponsibility
ofmyselfand,foreachofthecaseswherethisisnotthecase,adeclarationisincludedinthe
dissertationindicatingthenatureandextentofthecontributionsofcoauthors.
Date:31January2014
Copyright2014StellenboschUniversity
Allrightsreserved
Abstract
Pinuspatulaisthemostintensivelyplantedconiferinthetropicsandsubtropics.InSouthAfrica
Pinuspatulaplantationsarethemainsawlogresourceforstructurallumberproduction.Improved
intensivesilviculturalpracticesandtreebreedinghaveresultedinmarkedincreasesintherateof
growth.Toreapthefinancialbenefitsofthefastergrowth,plantationmanagersaremoreandmore
inclinedtoreducerotationages,whichinevitablyresultsintheproductionofhigherproportionsof
juvenilewoodatfinalharvest,andlumberwhichoftendoesnotmeettheminimumrequirements
forstiffnessforstructurallumber.Knowledgeofthevariationandtheaccuratepredictionofthe
mechanicalpropertiesofthetimberofstandingtreescanhavevariousbenefitsforgrowersand
processorsoftrees.Itcanbeusedfortreeallocationtodifferentprocessingfacilities,forprocessing
productionplanning,andtoassisttreebreederstoscreenandselectforsuperiorbreedingmaterial.
Theobjectivesofthisstudywere(1),toexaminethewithinandbetweentreevariationinwood
propertiesofyoungSouthAfricangrownPinuspatulatreesknowntohaveimportantimpactsonthe
suitabilityofsawnlumberforstructuralpurposesand(2),todevelopempiricalpredictionmodelsfor
theflexurallumberpropertiesfromstandingPinuspatula,basedonvariablesthatcouldbeassessed
nondestructivelyfromstandingtrees.
Samplematerialwasobtainedfrom170trees(1620yearsold)establishedin17compartments
alongtheMpumalangaescarpmentofSouthAfrica.Alargenumberofvariableswhichcouldbe
obtainednondestructivelyfromthetreeswhiletheywerestillstanding,weremeasured.Thetrees
weresubsequentlyfelledandtwologs,2.1minlength,wereextractedfromeachtreeattwoheight
positions.The340logswereprocessedinto1402piecesoflumberforfurthermeasurementsand
destructivetesting.
Resultsshowedthatthemeanmodulusofelasticitymeasuredonedge(MOEedge)wasfarbelowthe
limitssetforstructuralgradesoftwoodtimberinSouthAfrica.Allthedesirablepropertiesfor
structurallumberimprovedwithdistancefromthepithwiththeexceptionofthe5thpercentilevalue
formodulusofrupture(MOR),whichwashigheratthepiththanfortheboardsprocessedadjacent
tothepith.Boardsprocessedfromthelowerpartofthestemweresuperiorinmostofthe
importantpropertiescomparedtothosehigherupinthestem.
SeparatemultipleregressionmodelsforpredictingtheaveragedynamicMOE(MOEdyn)ofindividual
boards,treesandcompartmentsweredeveloped.Themodelsmanagedtoexplain68%,60%and
95%ofthevariationinMOEdynrespectively.ThemodelsdevelopedforMORexplained40%and42%
ofvariabilityatboardandtreelevelrespectively.Atcompartmentlevel,80%ofthevariationinthe
5thpercentileMORvaluecouldbeexplainedbythemodel.Sensitivityanalysesshowedthatsite
indexatbaseageof10years,acoustictimeofflight,wooddensityandringwidthwerethemost
influentialvariablesintheMOEmodels.Themodelsindicatedthattreeslendernessduringearly
growthseemstoplayamajorroleindeterminingthedynamicMOEandMORoflumber.Thisisin
agreementwithEulersbucklingtheoryandthebendingstresstheory.
Microfibrilangle(MFA)anddensityweremeasuredonradialstripstakenfromasubsampleoftrees
withtheSilviscan3technology.ThemeanmicrofibrilangleperyearringinPinuspatulavaried
between7oand29o.IngeneralMFAdecreasedwithdistancefromthepithandheightaboveground
level.Amultipleregressionmodelincludingmicrofibrilangle,densityandringwidthexplained71%
ofthevariationinthedynamicMOEofboards.Sensitivityanalysisonthemodelshowedthat
microfibrilangleanddensityhadroughlyequalinfluencesonpredictingtheMOEdynofPinuspatula
boards.
Opsomming
Pinuspatulaisdiemeesaangeplantenaaldhoutspesieindietropieseensubtropieseareasvandie
wreld.DitisdiegrootstebronvansaagblokkevirdieproduksievanstrukturelehoutinSA.
Intensieweboskultuurpraktykeenboomtelinghetgeleitotnmerkbareverhogingindiegroeitempo
vandiespesie.Plantasiebestuurdersisgevolglikgeneigomrotasieouderdommeteverlaag,watlei
totngroterpersentasiejeughoutwatnieaandieminimumstyfheidvereistesvanstrukturelehout
voldoennie.Kennisvandievariasieendieakkuratevoorspellingvandiemeganieseeienskappevan
staandebomekanvoordeleinhouvirbeidedieverbouersenverwerkersvanbome.Ditkan
byvoorbeeldvanhulpweesmetdietoewysingvanbomeaanverwerkingsfasiliteite,vir
produksiebeplanning,environdersteuningmetdiekeusevanteelmateriaalvirboomtelers.
Diedoelwittevanhierdiestudiewas(1),omdiebinneentussenboomvariasieindie
houteienskappe,watnbepalendeinvloedhetopdiegeskiktheidvanjongSuidAfrikaansePinus
patulabomevirstrukturelehoutproduksie,teondersoeken(2),omempiriesemodellevirdie
voorspellingvandiebuigeienskappevanplanketeontwikkel,gebaseeropveranderlikeswatnie
destruktiefopstaandePinuspatulabomegeevalueeris.
Monstersvirdiestudieisverkryvanaf170bome(1620jaaroud),geplantin17vakkeopdie
MpumalangaplatorandvanSuidAfrika.nGrootaantalveranderlikesisniedestruktiefgemeetop
diestaandebomewaarnadiebomegevelisentweesaagblokke,2.1minlengte,isoptweehoogte
posisiesuitelkeboomverwyder.Die340blokkeisverwerktot1402plankevirverderemetingsen
destruktiewetoetse.
Resultatehetgetoondatdiegemiddeldemodulusvanelastisiteitgemeetopdiedwarskant
(MOEedge)aansienliklaerwasaswatvereiswordvirstrukturelegraadhoutinSuidAfrika.Aldie
gewensteeienskappehettoegeneemmetafstandvanafdiemurgbehalwedie5depersentiel
breekmodulus(MOR),wathorwasvirmurgplankeasviraangrensendeplanke.Plankeafkomstig
vandielaerdelevandiestamhetoordiealgemeenbetereienskappegehadasplankeafkomstigvan
diehordele.
Veelvuldigeregressiemodellekon68%,60%en95%vandievariasieindiegemiddeldedinamiese
MOE(MOEdyn)opdievlakvanonderskeidelikindividueleplanke,bomeenvakkeverklaar.Die
modellevirMORkon40%en42%vandievariasieoponderskeidelikplankenboomvlakverklaar.
Diemodelvir5depersentielMORvanvakkekon80%vandievariasieverklaar.n
Sensitiwiteitsanalisehetaangetoondatgroeiplekindeksopouderdom10,akoestiesevlugtyd,
digtheidenjaarringwydtediebelangriksteveranderlikeswaswatMOEdynbenvloedhet.Diemodelle
hetaangetoondatdieslankheidvanbometydensvroegroeivermoedeliknbelangrikeinvloedop
dieMOEdynenMORvanplankehet.DitisinooreenstemmingmetEulerseknikteorieendie
buigsterkteteorie.
DiemikrofibrilhoekendigtheidvannsteekproefvandiebomeisgemeetmetdieSilviscan3
apparaat.Diegemiddeldemikrofibrilhoekperjaarringhettussen7oen29ovarieer.Hierdievariasie
washoofsaaklikafhanklikvanboomhoogteenaantaljaarringevanafdiemurg.nVeelvuldige
regressiemodelwatmikrofibrilhoek,digtheidenjaarringwydteinsluit,kon71%vandievariasiein
MOEdynverklaar.nSensitiwiteitsanaliseopdiemodelhetaangetoondatmikrofibrilhoekendigtheid
ongeveerewebelangrikwaswatbetrefhulleinvloedopdievoorspeldeMOEdynvanPinuspatula
planke.
Acknowledgements
IwanttothankGod,thecreatoroftrees,forprovidingmewiththisopportunityforsuchan
interestingandrewardingstudy.
Thankstodr.FrancoisMalanandprof.TimRypstraforyoursupervision.Especiallyforyour
meticulouseditingofthepapers.Ihavelearnedmuchduringthisprocess.
Foryourhelpinmeasuring,testingandbreakingofsamplesIwanttothankWilmourHendrikse,
GeorgeDowse,JustinErasmus,DeswinTitus,ChristiaanSmitandDerikLerm.
ThankyoutoBarryMullerandJosLouwofNMMUwhoinvitedmetomakeuseofthePinuspatula
samplematerialwhichwasinitiallydestinedforanotherstudy.Alsoforyouandyourstudentshard
workduringthetwoweeksoffellingandremovaloftreesoneofthegoodmemoriesofthisstudy!
ThankstoSawmillingSAandtheNRFsThripprogrammeforfundingthisstudy.
KLFprovidedthetreesandsomefunding,YorkTimbersthesawmillprocessingfacility,Hans
MerenskyTimbertransportedthesawnlumbertoStellenbosch,andCapePineprovidedgrading
equipment.Thesecontributionsaresincerelyappreciated.
Finally,IdedicatethisthesistoZerilda,JurieandMalanmyskadubome.
Tableofcontents
Abstract.................................................................................................................................................2
Opsomming...........................................................................................................................................3
Acknowledgements...............................................................................................................................4
Chapter1.Introduction........................................................................................................................6
Chapter2.WesselsCB,MalanFS,RypstraT.2011.Areviewofmeasurementmethodsusedon
standingtreesforthepredictionofsomemechanicalpropertiesoftimber.EurJForest
Res130(6):881893............................................................................................................9
Chapter3.DowseGP,WesselsCB.2013.ThestructuralgradingofyoungSouthAfricangrownPinus
patulasawntimber.SouthernForests75(1):717...........................................................31
Chapter4.WesselsCB,MalanFS,NelDG,RypstraT.Inpress.Variationinstrength,stiffnessand
relatedwoodpropertiesinyoungSouthAfricangrownPinuspatula.SouthernForests76.
............................................................................................................................................51
Chapter5.WesselsCB,SeifertT,LouwJH,MalanFS,RypstraT.Unpublished.Thepredictionofthe
flexurallumberpropertiesfromstandingSouthAfricangrownPinuspatulatrees..........68
Chapter6.WesselsCB,MalanFS,KiddM,RypstraT.Unpublished.Thevariationofmicrofibrilangle
inSouthAfricangrownPinuspatulaanditsinfluenceonthestiffnessofstructural
lumber.................................................................................................................................93
Chapter7.Summaryofresearchresults...........................................................................................105
References.........................................................................................................................................107
AppendixA:Declarationsofcandidateandcoauthors....................................................................109
Chapter1.Introduction
Plantedforestsarerapidlyexpandingonaglobalscaleatabout5millionhaperyearandcurrently
accountsforabout7%ofthetotalafforestedareaworldwide(FAO,2013).In1980therewere18
millionhaofplantedforests,comparedto187millionhain1990and264millionhain2012(Carleet
al.,2002;FAO,2013).CarleandHolmgren(2008)estimatedthatin2005abouttwothirdsofthe
globalindustrialwoodsupplyoriginatedfromcommercialplantations.
Pinuspatulaisthemostintensivelyplantedconiferinthetropicsandsubtropics.Itisestimatedthat
morethanonemillionhectaresareplantedwiththisspecieswithabouthalfofthatinAfrica
(Wright,1994).PinuspatulaisalsoplantedintheAndeancountriesofSouthAmericawithpotential
toincreasetheareaunderthisspeciesinthehighaltitudeareasinBrazil(HodgeandDvorak,2012).
InSouthAfricaitisthemostimportantcommercialplantationsoftwoodresourcewithatotalof
338923haplantedwithPinuspatulatrees(DAFF,2009).TheMpumalangaescarpmentisthelargest
sawloggrowingareainSouthAfricawithPinuspatulathemainspeciesplanted.
SouthAfricawasoneofthefirstcountriestoestablishplantationforestryonalargescale,startingin
thelatenineteenthcentury.By1960theforestryareahadincreasedtoabout1millionha(Owen
andVanderZel,2000).Duetoashortageofsuitablelandavailableforafforestation,aswellas
competitionfromagricultureandwatercatchment,theareaunderforestplantationsinSouthAfrica
hassincestabilised.Tomeetthecountrysgrowingneedsforwoodthisresultedinincreasing
emphasesintheforestandwoodprocessingindustriesonimprovedvolumeproductionperunit
areathroughimprovedsilviculturalpracticesandgeneticimprovement,aswellasimprovedwood
productyieldandquality.
However,theincreasedsizeofthecorewoodzone,andthebiggerproportionofcorewoodthat
resultswhenrotationsagesareshortenedtoreapthefinancialbenefitsofthefastergrowth,has
becomeawoodqualityfactorofgrowingconcernworldwide(Cown,2006;Malan,2010).Cown
(2006)statesthatresearchersaroundtheworldhaveconfirmedthataggressivesilvicultural
regimeshavecausedasignificantreductioninmechanicalpropertiesofplantationgrownpines.
StudiesinSouthAfricahaveshownsharpreductionsinsomeofthemechanicalpropertiesofpine
lumberprocessedfrommaterialharvestedatayoungerage,astreesreachmerchantablesizemuch
earlierduetofastergrowthrates(Burdzik,2004;Wesselsetal.,2011).Whilethefinancial
importanceofincreasedvolumeproductionofplantationsisundisputed,itisincreasinglyimportant
thatforestmanagersandresearcherstakeintoconsiderationtheadverseeffectsoftheiractionsand
effortsonendproductquality.
Morethan70%ofthesolidsawnlumberproducedinSAissoldasstructuralorbuildingtimber
(CrickmayandAssociates,2011),awoodproductcategorywhichhastocomplytoverystrict
strengthandstiffnessrequirements.Giventhechallengescausedbyanincreasingproportionof
juvenilewoodinthetimberresource,thereisagrowingneedtobetterunderstandthevariationin
themechanicalpropertiesofthelumberfromplantationgrowntrees.Thereisalsoaneedfornon
destructivemethodscapableofaccuratelypredictingthesepropertiesfromstandingtrees.Amongst
others,suchinformationcanbeusedtoassistindecisionsrelatedtotheallocationoftreesto
differentprocessingfacilities(Mathesonetal.2002;Cown2006;Wangetal.2007),forprocessing
productionplanning(Uusitalo1997;Wesselsetal.2006),tostudytheeffectofsiteandsilviculture
factorsonwoodquality(Wang2000b;Grabianowskietal.2004;Wangetal.2005)andtoassisttree
breederstoscreenandselectforsuperiorbreedingmaterial(Ivkovietal.2009;Lindstrmetal.
2002;Launayetal.2002).
Theobjectivesofthisstudywere:
ToexaminethewithinandbetweentreevariationinwoodpropertiesofyoungSouth
AfricangrownPinuspatulatreesknowntohaveimportantimpactsonthesuitabilityofsawn
lumberforstructuralpurposes;
Todevelopempiricalpredictionmodelsfortheflexuralpropertiesoflumberproducedfrom
youngPinuspatulatrees,basedonrelatedvariablesthatcouldbeassessednon
destructivelyonstandingtrees.
Structureofthedissertation
Thisdissertationconsistsofanintroduction(Chapter1),followedbythreepublishedandtwo
unpublishedpapers,eachaddressingaspecifictopicwithinthescopeofthisstudy(Chapters2to6).
Thetwounpublishedpapersarecurrentlyinpreparationforsubmissiontoanaccreditedscientific
journal.Chapter7containsafullsummaryofalltheresearchresults.
AppendixAcontainssigneddeclarationsbythecandidateandcoauthorsregardingthenatureand
extentofthecontributionsofthedifferentauthors.
Thestudywasperformedonsamplematerialobtainedfrom170Pinuspatulatrees(1620yearsold
atthetimeofsampling)establishedin17compartmentsontheMpumalangaescarpmentofSouth
Africa.Abriefsummaryofeachpaperscontributiontowardsmeetingtheobjectivesofthisstudy,
arepresentedbelow.
Chapter2
WesselsCB,MalanFS,RypstraT.2011.Areviewofmeasurementmethodsusedonstandingtrees
forthepredictionofsomemechanicalpropertiesoftimber.EurJForestRes130(6):881893.
Thispaperreviewedthecurrentliteratureonnewandexistingnondestructiveorlimited
destructivepropertymeasurementmethodsonstandingtreesthatcanassistwiththe
predictionof,inmostcases,themodulusofelasticityandmodulusofruptureoftimber.
Chapter3
DowseGP,WesselsCB.2013.ThestructuralgradingofyoungSouthAfricangrownPinuspatulasawn
timber.SouthernForests:AJournalofForestScience75(1):717.
InthispapertheefficiencyofthecurrentvisualandmechanicalgradingrulesonyoungPinus
patulasawnlumberwereassessedandthepotentialofsomeindicatorpropertiestobeused
asstructuralgradingparameterswereevaluated.
Intermsoftheobjectivesofthisdissertation,themostimportantresultsreportedinthis
paperwerethecorrelationsbetweenbasicwoodandlumberpropertiesandthestiffness
andstrengthoflumber.
Chapter4
WesselsCB,MalanFS,NelDG,RypstraT.Inpress.Variationinstrength,stiffnessandrelatedwood
propertiesinyoungSouthAfricangrownPinuspatula.SouthernForests:AJournalofForestScience
76.
Thispaperdescribedthewithinandbetweentreevariationinmodulusofrupture,modulus
ofelasticity,density,andtwistoflumberfromthePinuspatulatrees.
Apartfromthestrengthandstiffnessvariation,thevariationintherelationshipbetween
strengthandstiffness,whichisimportantforefficientstructuralgrading,wasalso
investigated.
Chapter5
WesselsCB,SeifertT,LouwJH,MalanFS,RypstraT.Unpublished.Thepredictionoftheflexural
lumberpropertiesfromstandingSouthAfricangrownPinuspatulatrees.
Thepaperdescribedempiricallybasedmodelsforpredictingtheflexuralpropertiesofthe
woodproducedbythePinuspatulatrees.
Modelswerebasedonthepropertiesofstandingtreesandtheireffectivenesswas
evaluatedonboard,treeandcompartmentlevels.
Chapter6
WesselsCB,MalanFS,KiddM,RypstraT.Unpublished.ThevariationofmicrofibrilangleinSouth
AfricangrownPinuspatulaanditsinfluenceonthestiffnessofstructurallumber.
Thisstudywasperformedonasubsampleof30trees(outof170trees),henceitformsa
separatestudyandwasnotincludedasapartofChapter5.Thereasonforthesmaller
samplewasthecostassociatedwithmeasuringmicrofibrilangleanddensityontheSilviscan
3apparatus.
Thepaperdescribedthevariationinmicrofibrilangleandthedeterminationofthe
relationshipbetweenmicrofibrilangleandthedynamicmodulusofelasticityofsawnPinus
patulalumber.
TheformatofthepapersappearinginChapters26wereasrequiredforthejournalstheyappeared
inorwillbesubmittedforreviewhencethedifferencesinformatbetweenchapters.
Chapter2.
Areviewofmeasurementmethodsusedonstandingtreesforthepredictionofsome
mechanicalpropertiesoftimber
Publishedin:EuropeanJournalofForestResearch(2011)130:881893
C.B.Wessels
UniversityofStellenbosch
PrivateBagX1,
Matieland,7602
SouthAfrica
Tel:+27218083319
Fax:+27218083603
email:cbw@sun.ac.za
F.S.Malan
KomatilandForests
POBox574,
Sabie,1260
SouthAfrica
T.Rypstra
PrivateBagX1,
Matieland,7602
SouthAfrica
Abstract
Theaccuratepredictionofthemechanicalpropertiesthatcanbeexpectedfromtimberfrom
standingtreeshasmanybenefitsforthegrowersandprocessorsoftrees.Itincludessupportintree
breedingselection,treeprocessingallocationdecisions,siteandsilviculturalresearchandprocessing
productionplanning.Anumberofmethodshavebeendevelopedoverthelastfewdecadeswith
significantinterestintherecentpastinespeciallyacousticmethods,nearinfraredspectroscopy
methodsandtheAustralianmultipropertymeasurementsystemknownasSilviscan.Thispaper
reviewsthecurrentliteratureonnewandexistingnondestructiveorlimiteddestructiveproperty
measurementmethodsonstandingtreesthatcanassistwiththepredictionof,inmostcases,the
modulusofelasticityandmodulusofruptureoftimber.
Keywords:nondestructive;mechanicalproperties;standingtrees;review
Introduction
Themechanicalpropertiesofsawntimberplayanimportantroleinmanyapplications.Itis
particularlyrelevantforstructuraltimberthatistimberusedinloadbearingstructures.InSouth
Africaroughly70%ofallsawntimberisclassifiedasstructuralorbuildingtimber(Crickmayand
Associates2009).Forgrowersandprocessorsoftreesintendedforstructuraltimberproduction,the
accuratepredictionofthemechanicalpropertiesofthetimberofstandingtreeshavevarious
benefits.Amongstothers,suchinformationcanbeusedtoassistindecisionsrelatedtothe
allocationoftreestodifferentprocessingfacilities(Mathesonetal.2002;Cown2006;Wangetal.
2007),forprocessingproductionplanning(Uusitalo1997;Wesselsetal.2006),tostudytheeffectof
9
siteandsilviculturefactorsonwoodquality(Wang2000b;Grabianowskietal.2004;Wangetal.
2005)andtoassisttreebreederstoscreenandselectforsuperiorbreedingmaterial(Ivkovietal.
2009;Lindstrmetal.2002;Launayetal.2002).
Therearethreeimportantissuesinvolvedinpredictingthemechanicalpropertiesofproductsfrom
standingtrees:
1.Onemustbeabletoidentifyandproveameaningfulrelationshipbetweenmeasureable
propertiesofatreeandthemechanicalpropertiesoftimberfromthattree.
2.Onemustbeabletomeasuretherelevantpropertiesinalimiteddestructiveorpreferablyanon
destructiveway.
3.Inordertomakewholetreeandstandbasedpropertypredictionsfromlimitedsamplingpointsin
atreeandstand,knowledgeisrequiredontherelationshipbetweenpropertiesatthesampling
pointsinatreeandstandandthepropertyvariationthroughtherestofthetreeandstand.
Thispaperwillfocusmainlyonthesecondissueviz.measurementmethodswhichcanbeusedinthe
predictionofthemechanicalpropertiesoftimberofstandingtrees.Literatureonexistingandnew
methodsofmeasurementwillbereviewedanddiscussed.
Thereviewwillfocusspecificallyonthepropertiesrelevanttostructuraltimberperformance,
namelylongitudinalstiffnessandthesixultimatestressvaluesoftimberproductsusedintimber
designcodes.Includedaremeasurementofunderlyingorbasicpropertiesthatinfluencestructural
timberperformance,suchasdensityandmicrofibrilangle.Althoughotherpropertiesmayalsoplaya
roleinthevarioustimberstrengthgradingsystemsi.e.discolourationanddeformation,only
mechanicalpropertiesareconsideredhere.Measurementofrandomlyoccurringdefectsintree
stemsthathaveaninfluenceonmechanicalpropertieslikeresincracks,decayandreactionwoodis
notconsideredinthisreview.
Propertiesinfluencingthemechanicalbehaviouroftimber
Themechanicalbehaviouroftimberisaresultofthebasicphysical,anatomicalandchemical
characteristicsoftherawmaterial.Inmanycasesconcomitantpropertiestotheactual
characteristicsofinterestaremeasuredandusedtoevaluateorpredictthelikelyperformanceof
productsfromastandingtree.Anunderstandingoftherelationshipbetweenthebasic
characteristicsoftimberandthemechanicalbehaviourarerequiredinordertodiscussmostofthe
measurementmethodsusedforpredictionpurposes.Therelationshipsbetweenvarious
concomitantpropertiestostrengthandstiffnesshavebeenwelldocumentedinresearchon
structuralgradingsystems.
Knots,annualringwidth,modulusofelasticity(MOE)anddensityofsawntimberformthebasisof
manystructuraltimbergradingsystemsandaregenerallyacceptedaspropertiesthatcanbeusedto
predictthestrengthandstiffnessoftimber.Johansson(2003)comparedthecoefficientof
determination(r)obtainedinsixdifferentstudiesbetweenknotproperties,annualringwidth,
densityandMOEasindependentvariablesandsawntimberstrength(tensileandbending)as
dependentvariableofNorwayspruce(Johanssenetal.1992;Hoffmeyer1984;Hoffmeyer1990;
Lackner1988;Glosetal.1982,Johansson1976).InallcasesMOEwasfoundtobeabetterpredictor
oftensileorbendingstrengthoftimber(0.53r0.74)thantheotherthreepropertiesused
individually(0.16r0.44).Usingcombinationsofthesepropertiestopredictstrengthoftimber
piecesincreasedthecoefficientofdeterminationsomewhat.Whereknotdataiscombinedwith
MOEforpredictionofstrengthofNorwayspruceandSouthernPinetimber,thecoefficientof
determinationincreasedby0.1to0.17comparedtousingonlyMOE(Johanssonetal.1992;
Johanssonetal.1998;Orosz1969).
10
Microfibrilanglehasbeenshowninmanystudiestobeasinfluentialasdensity,andsometimeseven
moreso,especiallyinjuvenilewood,forthepredictionofstiffnessofclearwoodsamples.Caveand
Walker(1994)arguedthatmicrofibrilangleistheonlypropertythatcanexplainthelargevariation
inMOEofPinusradiatafromthepithoutwards.EvansandIlic(2001)foundthatdensityalone
accountedfor70%ofthevariationintheMOEofEucalyptusdelegatensisclearwoodsamplesand
microfibrilanglealoneaccountedfor86%ofvariation.Microfibrilangleanddensitytogether
accountedfor96%ofvariationinMOE.Megrawetal.(1999)foundthatdensityandmicrofibrilangle
togetherexplained93%ofvariationofMOEinsmallclearwoodsamplesofPinustaeda.Therelative
importanceofeachpropertyinexplainingvariationinMOEchangewithlocationinthetreeatthe
baseofthestemmicrofibrilanglehadalargerinfluencethandensityonMOEvalues,withthe
oppositetobetrueat5mstemheight.Thestrengthandstabilitypropertiesofjuvenilewoodhave
alsobeenfoundtobeaffectedmorebydifferencesinmicrofibrilanglethanwooddensity,whilein
maturewooddensitytendstoplayamoredominantrole(CaveandWalker1994).Duetothe
cumbersomemethodsusedinthepasttomeasurethemicrofibrilangle,anumberofworkers
exploredtherelationshipbetweentracheidlengthandthemicrofibrilangleandderivedformulasto
predictmicrofibrilanglefromtracheidlength(Preston1948;WardropandDadswell1950;Echols
1955;andSmith1959,ascitedinCaveandWalker1994).InSmithsstudy,tracheidlength
accountedfor58%ofthevariationinmicrofibrilangle,whereasinEcholsstudy91%ofvariability
wasaccountedforbydifferencesinmicrofibrilangle(asquotedinHuangetal.2003).
Spiralgrainhasaninfluenceonsomemechanicalpropertiesofwood.Aswoodisanorthotropic
material,itsstrengthpropertiesaresignificantlydifferentinthethreemutuallyperpendicular
directionsoraxes.Therefore,whenthegrainorientationisnotparalleltothatofthestem,the
actualstrengthvalueswillbeacombinationofparalleltograinandperpendiculartograin
characteristics.Spiralgrainintreestemsisoneofthemaincausesofgraindeviationinsawn
products,ascuttingnormallytakesplaceparalleltothestemaxis.ForpropertiessuchasMOE
(longitudinal),bendingstrengthandtensilestrengthparalleltograin,graindeviationresultsin
strengthreductions.
Wilson(1921)testedthestrengthpropertiesofclearpiecesofWhiteash,SitkaspruceandDouglas
firwithvariouslevelsofgraindeviation.Hefoundanaveragebendingstrengthlossof11%witha
grainangleof3.8degrees,19%withagrainangleof5.7degrees,and45%withagrainangleof11.3
degrees.Dinwoodie(2000)statedthatlongitudinaltensionstrengthisaffectedmoreseverelyby
graindeviationthanlongitudinalbendingstrengthandcompressionstrength.Hankinson(1921)
developedanequationtocalculatethestrengthoftimberatanygrainanglewhentheparallelto
grainandperpendiculartograinstrengthvaluesareknown.Thenegativeeffectofgraindeviation
onwoodstrengthiswidelyacceptedand,asaresult,restrictionsareincludedinmanystructural
timbergradingstandardsi.e.EN518(1995),SANS17832(2005)andAS2858(2003).
Manyrandomlyoccurringdefectsinthetreestemalsoplayaroleinthemechanicalperformanceof
timber.Restrictionsontheoccurrenceofthesedefectsareusuallycontainedinstructuralgrading
rulesandcanalsoincluderesinsplits,biologicaldecay,ringsplitsandreactionwood.Measurement
andpredictionoftheserandomdefectsinstandingtreesisnotdiscussedinthispaper.
Mechanicaltreestiffnessmeasurement
SeveralworkersdevelopedmechanicaltreebendingapparatitodeterminetheMOEofstanding
trees(VafaiandFarshad1979;KoizumiandUeda1986;Launayetal.2000;Launayetal.2002).The
basicmechanisminallthesesystemsistheapplicationofabendingmomentonthetreetrunkand
measurementofdeflectionataspecificpoint(Fig1).ThelongitudinalMOEofthetrunkcanbe
calculatedusingtheresultsfromthesetests.Launayetal.(2000)arguedthatthisapproachgives
betterestimatesofthemechanicalpropertiesoftimberproductsfromtreesthanlocalised
11
measurementsofbasicproperties,suchasdensity,sinceitcoversaverticalrangeinthestemand
includetheeffectofbranchesorknots.Launayetal.(2000)foundamoderatecorrelationcoefficient
(r=0.54)betweenthetreeMOEandaverageboardMOEofDouglasfirandLarchtrees.These
methods,however,arefairlytimeconsumingasonly2050treescanbemeasuredperday
dependingontheapparatusbeingused(Launayetal.2000).
a
F
F
a
Mz
Mz
Fa
Fa
Fig.1Principleoftwobendingtestsonstandingtrees(adaptedfromLaunayetal.2002)
Branchandknotassessmentmethods
Visualtreegradingmethodsweredevelopedinthepasttoevaluatethepotentialofatreeto
producespecificproducts.Branchingandknotcharacteristicsareusuallythemaincriteriafor
evaluationinsuchsystems.Schroederetal.(1968),asreportedinClarkandMcAlister(1998),
developedtreegradingrulesforSouthernPinebasedontheevaluationonthenumberofclearfaces
inthebottom4.9mofthestem.Biologicalandmechanicaldamage,sweepandcrookwerealso
considered.Equationsweredevelopedtopredictstructuraltimbergrades,giventhetreegrade,age
anddiameter.AsimilartreegradingsystemwasdevelopedbyBrisbinandSonderman(1971)for
EasternWhitePineintheUnitedStates.AccordingtoClarkandMcAlister(1998)theSouthernPine
treegradingsystemiseffectiveingradingoldermatureSouthernPinetreesbutnotsoeffectivein
gradingyoungertrees.Asaconsequencenewtreegradingrulesweredevelopedandtestedon
PinustaedaandPinusechinatatreesofagesvaryingfrom22to73years.Themainevaluation
criteriaincludedbranchproperties(number,size,live/dead)forspecificheightsections,diameter
atbreastheight,age,straightness,seams,cankersanddecay.Thehighestgradetrees(Grade1)
producedrelativelyhigherproportionsofthebestvisuallygradedstructuralgrades(No.1and
better)thanthelowestgradetrees(Grade3).PrestemonandBuengiorno(2000)developedamodel
topredictSouthernPinetreegradesandsubsequentlytimbergradesdirectlyfromtreeandstand
levelvariables(diameter,height,standbasalarea,sitequality,ownership).However,theyconcluded
thatthemodelhadlowexplanatorypower.
InastudybyBier(1985)onNewZealandgrownradiatapine,treesandlogsweregradedaccording
tobranchindex,whichistheaverageofthelargestbranchdiameterfromeachlogquadrantas
wellasthebasicdensityofthelogs.Boardssawnfromtheselogsweredestructivelytestedin
bendingandtherelationshipbetweenbendingpropertiesandthelogvariableswasdetermined.Itis
interestingtonotethatthebasicdensityofalogshowedastrongercorrelationwithboththe
averageMOEandmodulusofrupture(MOR)ofboardsfromthatlogthanwiththebranchindex.
However,thebranchindexofthelogshowedbettercorrelationwiththeminimumMOEandMOR
obtainedfromboardsfromalogthanbasicdensity.Thisisasignificantfindingsincetheweakest
andloweststiffnesstimberiswhatdeterminesmechanicalgradestressesandnotaveragevalues
(Madsen1992).InanotherstudybyBier(1986)itwasshownthatforlowbranchindexlogsand
treesnostructuralgradingforindividualboardsisrequiredsinceallthetimberfromsuchlogswillbe
12
abovetheminimumNewZealandstressgrade.Thedisadvantageofsuchanapproachisthatno
higherstressgradeproductscanberecoveredfromatree.
Uusitalo(1997)statedthatknotpropertiesinScotspinearebyfarthemostimportantquality
indicatorsofthetimber.HedevelopedamodeltoassessthequalityofstandingScotspinewhich
couldbeusedinsawmillproductionplanning.Thevariablesmeasuredonasampleoftreesfrom
eachstandincludeddiameter,heighttofirstdeadbranch,crownheightandtreeheight.However,
thesecharacteristicswerenotdirectlycomparedwiththemechanicalpropertiesoftimber.
Inintensivelymanagedsoftwoodplantations,pruningofthelowersectionofthestemproduces
cleartimberintheoutersectionsofthebottomlog.Thispracticeisaimedmainlyatproducinghigh
valueappearancegradetimberbutwillalsohaveaneffectonthemechanicalpropertiesoftimber,
sinceitchangestheknotproperties.Incertainmarketconditionssawmillersmaychoosetoconvert
prunedlogstostructuraltimber.InNewZealand,Park(1989and1994)developedaprunedlog
indexwhichcanbeusedtoevaluateacompartmentoftreesintermsofthepotentialtoproduce
cleargradetimber.Alimiteddestructivesampleisrequiredtoevaluatethesizeoftheknottydefect
coreofacompartmentoftrees.Asimilarlimiteddestructiveevaluationsystemwasdevelopedin
SouthAfricaandcanbeusedforsawmillproductionplanningandforthepredictionoftimber
gradestoexpectfromaspecificcompartment(Wesselsetal.2006).Neitherofthesestudies
exploredtherelationshipbetweenthedefectcoresizeandmechanicalpropertiesoftimberfrom
theprunedsectionofthestem.
Methodsformeasuringspiralgrain
Measurementofspiralgrainoftreesfrompithtobark(oracrosstheradius)isrelativelysimpleto
performondisksectionstakenfromatreestembutunfortunatelyitinvolvesdestructivesampling
(Brazier1965;Kromhout1966).Anumberofmethodsweredevelopedtomeasurethegrainangles
fromincrementcores,whichislessdestructive.However,precautionsneedtobetakentoavoidor
minimisetwistingofthecoresduringboring.Itisalsoimportantthattheintreeorientationofa
coreismarkedaccuratelytoensurethatthemeasurementstakenonthecorereflectgrain
orientationatthepointofsamplingprecisely.
Noskowiak(1968)provedthatincrementcoresfrom4.5mmborerswereunsuitableforgrainangle
measurementbecausetorsionalstressesduringboringexceededtheelasticlimitofthecore,causing
thecorestobecomepermanentlytwisted.Coresfrom12mmborers,however,didnotdeform
permanently,providedthattheborerwasturneduniformly.Noskowiaksmethodrequiresmounting
thesamplesinaluminiumcubes,extrudingthecore,andmicroscopicexaminationofcoresunder
incidentlight.Harris(1984)proposedamethodwherebarkwindowsarecutinoppositesidesofthe
stemandthegrainanglemeasuredonthestemsurfacewithineachwindow.Anincrementcoreis
thenremovedacrosstheentirediameterfromonewindowtothenext.Grainanglescanbe
measuredatdifferentgrowthringsrelativetoaplanedsurfaceonthecore.Bytakingreadingsfora
growthringonoppositesidesofthepithandgettinganaverageforeachgrowthring,theinfluence
ofacorewhichisnotremovedatrightanglesinrelationtothestemaxis,iseliminated.Angle
readingscanalsobecorrectedusingthereadingstakenatthecoresurfacesonthestandingtree.
Buksnowitzetal.(2008)evaluatedtheuseofXraydiffractometryformeasuringgrainangleover
radialsamplesofNorwayspruce,usingtheSilviscansystem.Agoodrelationshipbetween
goniometricmeasurements(similarmethodtothatofBrazier1965)andXraydiffractometrywas
found(r=0.87).TheSilviscanapparatuswasdesignedtoaccommodateincrementcores.
GindlandTeischinger(2002)usedpartialleastsquareanalysisofvisibleandnearinfrared
reflectancespectraonLarchblockstopredictthegrainangle.Coefficientsofdeterminationof0.77
13
and0.80wereobtainedrespectively.Althoughsandedwoodblockswereusedinthisstudyitwas
assumedthatasandedsurfaceonanincrementcorewillyieldsimilarresults.
Amethodbasedonthefactthatthedielectricconstantofwoodisabout1.5timeshigherparallelto
grainthanacrossthegrainisoftenusedtopredictgrainangleinsawnboards(McLauchlanetal.
1973;Samson1984;McDonaldandBendtsen1986;Samson1988;Samsonetal.1993).However,
noneofthesestudiesmentionthepotentialofusingthismethodforsmallsamplesorincrement
cores.
AnapparatusreportedtobeeffectiveandfastistheSpiralometer,adevicemakinguseofthe
tracheideffect(Brashawetal.2009).Alaserisfocussedonarotatingincrementcoreandmaximum
transmissionoccurswhenthelaserbeamandlongitudinalaxistracheidsarealigned.Thedeviceis
designedtotakeagrainanglereadingevery1mm.
Anumberofstudiesevaluatedtheuseofmicrowavesforthedetectionofgrainangle(Jamesetal.
1985;Ghodgoankaretal.2000,KaestnerandBth2005;SchajerandOrhan2006).Noneofthe
studiesdiscussedtheapplicabilityofthesemethodsonsmallsamplesorincrementcores.Onestudy
(Jamesetal.1985)examinedtheuseofpolarizationangleofatransmittedmicrowaveforpredicting
ofgrainangle.Itisimportant,however,thatthetestsamplemustbeofsufficientthickness(not
quantified)toensureaccuratepredictionofthegrainangle.Thissuggeststhatthemethodmightnot
besuitabletomeasuregrainorientationinincrementcores.
Whenaccuratemodelsforgrainanglefrompithtobarkexist,onlygrainanglemeasurementunder
thebarkisrequired.Twofairlysimplemethodstoobtainsuchmeasurementsaredescribedby
Hallingbck(2010).Inthefirstoneasectionofbarkmustberemovedandtheexposedcambium
scribedsothatthegrainangleonthecambiumcanbemeasuredmechanically.Alternatively,asmall
wedgecanbepushedintotheoutermostannualringsofthewood.Asitispushedinitisforcedto
alignparalleltothetracheidcells,sothegrainanglecanbesimplyquantifiedbymeasuringthe
inclinationanglebetweentheorientationofthewedgeandthatofthestemaxis.
Densityandgrowthringassessmentmethods
Densityhaslongbeenconsideredasoneofthemostimportantwoodproperties,ifnotthemost
important,intermsofitseffectonthequalityofsolidwoodproducts.ZobelandVanBuijtenen
(1989)concludedtherefore,specificgravitylargelydeterminesthevalueandutilityofwoodand
overshadowstheimportanceofotherwoodproperties.Althoughthisviewischallengedbysome
(CaveandWalker1994),manystructuralgradingstudiesconfirmedthatdensityplaysanimportant
roleindeterminingsomemechanicalpropertiesofwood(Johansson2003).Seenattheanatomical
level,densityisthecombinedresultofanumberofcharacteristicsincludingcellwallthickness,cell
diameter,growthringwidth,amountofrayandvesselelements,andtheratioofearlywoodto
latewoodinagrowthring.Strengthandstiffnesspredictionmethodsusingthewidthofgrowthrings
are,therefore,alsoincludedasadensitydependentpredictionmethod.
Fromthe1960stheuseofincrementborerstocollectsamplesfromstandingtreesfordensity
determinationbecamemorecommon(Cown2006).Thereareseveralmethodstodeterminedensity
fromincrementcoresamples.Gravimetricmethodsfordensitydeterminationmakeuseofthe
accuratedeterminationofthevolumeandmassofthesampleataspecificmoisturecontent(Yao
1968).Arelativelyeasyandpopularwayofdeterminingthedensityofsmallsamples,likethosefrom
incrementcores,isthemaximummoisturecontentmethoddevelopedbyDianaSmith(1954).This
methodusesthemassofasamplewhensaturatedwithwater,itsovendrymassandtheabsolute
densityofwoodcellwallmaterial(1,53g/cm3),todeterminethebasicdensityofsmallsections.In
somewoodspecies,extractives,whichisnotpartofthewoodsubstanceandwhichcontribute
14
nothingtowoodstrength,needtobeextractedfromthewoodtoensuremorereliabledensity
valuesforpredictivepurposes(Tsoumis1991).
Severalindirectmeasurementmethodshavebeendevelopedinthepasttopredictdensityand
growthringpropertiesfromincrementcores.Spectroscopicstudiesorradiationdensitometryisone
ofthemostwidelyusedmethodstomeasurethedensityofwoodfromincrementcores.Oneofthe
earliestmethodsinvolvedtheuseofXrayradiographsofwoodsamples(Polge1966).Usinga
calibrationwedge,opticaldensitiesareconvertedintowooddensity.Latermethods,alsoreferredto
asdirectradiationdensitometry,measuretheamountofradiationabsorbedbywoodtocalculate
thelinearattenuationcoefficientanddensity.Differentsourcesofradiationareusedforthesei.e.X
rays,betaraysandgammarays(Polge1978;CownandClement1983;MalanandMarais1992;
Divosetal.1996).ArecentdevelopmentindensitometrywasthedevelopmentoftheSilviscan
system,whichisamultipropertymeasuringdevicewhichcanuseincrementcoresandperformX
raydensitometry,Xraydiffractionformicrofibrilmeasurementandimageanalysesofmicroscopic
imagesofcells(Evansetal.1998).Microwaveshavealsobeenusedinnumerousstudiesfordensity
determination,mainlywithtimbergradingstudies(SchajerandOrhan2006;Tiurietal.1980;James
etal.1985;Martinetal.1987;LeicesterandSeath1996).
Densitometryallowsforahighresolutionmeasurementofdensitysothatbothinterringandintra
ringdensityvariationanalysiscanbeperformed(Fig2).Themethodlendsitselftoautomationand
therelativelyfastmeasurementoflargenumbersofsamples(Polge1978).Intermsofstandingtrees
themostlaboriousandtimeconsumingpartofthismethodisthecollectionofincrementcoresfrom
thetreesandtheaccuratepreparationofthesamplesfordensitometry.
1000
900
Density (kg/m )
800
700
600
500
400
300
200
100
0
0
10
20
30
40
50
60
70
80
90 100 110 120 130 140 150 160 170
Distance from pith (mm)
Fig.2Adensityprofilemeasuredwithagammaraydensitometer
RecentlyportableXraycomputedtomography(CT)systemsweredevelopedforthescanningof
standingtrees(Habermehletal.1999).Thesesystemsusegammaraysasasourceofradiationand
usethescanneddatatoreconstructathreedimensionalinternaldensityimageofastandingtree
trunk.Applicationsofsuchsystemsincludeinspectionofparkandstreettrees(Habermehletal.
1999),growthringassessment(Onoeetal.1984),andmoisturecontentstudiesoflivingtrees
(Tognettietal.1996).NoliteraturewerefoundwheremobileCTscannerswereusedforpredicting
mechanicalpropertiesofstandingtrees,althoughthedensitydatawillobviouslybeusefulforsuch
purposes.
Anindirectdensitymeasurementmethodthatusesincrementcoresisthelasersandblasting
method(Lesnino1994).Radialincrementcoresaresandblastedandthedepthprofilemeasured
withalasermeasurementtechnique.Acorrelation(r)of0.86betweendensitymeasuredwitha
stereogravimetricmethodandthelasersandblastingtechniquewasobtained.Thebigadvantageof
thismethod,accordingtotheauthor,isthatsamplesrequirelesspreparationthanstereogravimetric
andradiationdensitometrymethodsandmeasurementscanbeautomatedeasily.
15
Inordertoobtainmorerapidresults,someindirectdensitymeasurementmethodsontreestems
havebeendevelopedwherenoincrementcoresarerequired.ThePilodynwoodtesterwas
developedinSwitzerlandandinjectsaspringloadedstrikerpinintothestemofatree(Cown1978).
Thepenetrationdepthcanbeusedasanindirectmeasureofthedensityoftheoutersectionofthe
stem.TherelationshipbetweendensityresultsoftheouterportionsofatreestemandPilodyn
readingswasfoundtoberelativelystrongwithacorrelationcoefficientof0.97forradiatapineof
differentages(Cown1978).InanotherstudyCownetal.(1998)reviewedsixteenotherstudies
wherethePilodyntesterwasusedandcomparedtoouterwoodbasicdensitiesofstandingtrees.
TheresultsofthesestudiesintermsofthecorrelationcoefficientsbetweendensityandPilodyn
readingsvariedbetween0.47and0.96.AccordingtoCownetal.(1998)thegeneralconsensusis
thatthePilodynmethodisnotsufficientlyaccurateforthedeterminationofdensityofanindividual
treebutcanbeverycosteffectiveandaccurateforcomparinggroupsoftrees,suchasingenetic
trials.
AnotherindirectmethodusedtopredictdensitymakesuseoftheTorsiometerwhichmeasuresthe
torquewhenanincrementboreristurnedintoatree.InCownsstudy(1978)thePilodynmethod
wasalsocomparedtoTorsiometerresults.ApartfromthefactthataTorsiometerwasslowerthan
thePilodyn,thecorrelationwithdensitywasalsoworse.Amodernandimprovedversionofthe
Torsiometerisaresistancedrill,knownastheResistograph,whichmeasuresvariationsindrilling
resistanceofathinneedlewhendrivenintoatreeataconstantforce(Rinnet.al.1996).Variations
inpowerconsumptionismeasuredelectronicallyandisdirectlyrelatedtovariationsindensity.
ChantreandRozenberg(1997),asreportedinIsikandLi(2003),evaluatedthistoolandalthough
theyreportedasignificantcorrelationwiththemeanradialdensityofDouglasFirstrips,theirresults
showedthatitcannotbeusedforaccuratelypredictingdensityvariationwithinatrunk.IsikandLi
(2003)founditaneffectivetoolfordensityselectionoftreesinaPinustaedatreeimprovement
program.
Nearinfraredspectroscopyisanindirectmethodfordensitydeterminationthathasbeenstudiedby
severalresearchersandthatcanbeusedonincrementcoresfromstandingtrees.Schimlecketal.
(2001)obtainedacoefficientofdeterminationof0.93betweendensityandNIRdataforthe
calibrationsetofEucalyptusdelegatensissamples.Viaetal.(2003)obtainedcoefficientsof
determinationexceeding0.71whenpredictingthedensityofPinuspalustrissamplesusingvarious
statisticalmodels.Similarresultswereobtainedinotherstudies(HoffmeyerandPedersen1995;
Haukssonetal.2001;SchimleckandEvans2004).Kochetal.(1998)foundthatdensitycanalsobe
predictedusingthefarinfraredspectroscopymethodwhich,accordingtotheauthors,hassafety
advantagesoverothertechniques.
Wimmer(1995),reportedamethodofestimatingwooddensitybymicroscopicallymeasuring
anatomicalcellfeatures.Acorrelationcoefficientof0.85wasobtainedbetweenestimateddensity
valuesandmeasuredvalues.
Microfibrilanglemeasurementmethods
Microfibrilangleofwoodreferstotheorientationofcellulosemicrofibrilsinthesecondarycellwall
inrelationtothelongitudinalaxisofthecell.Mosttechniquesformeasuringthemicrofibrilangle
canuseincrementcoresforthesemeasurements.
Huangetal.(1998)andBarnettandBonham(2004)providedreviewsonthemeasurement
techniquesofmicrofibrilangle.Thereisarangeoftechniques,includingthefollowing(takenfrom
BarnettandBonham,2004):Polarizationmicroscopywasthefirsttechniqueformeasuring
microfibrilangle.Theproblemoflightpassingthroughtwowallsofatracheidorfibre,inwhichthe
16
fibrilsformoppositesidesofthespiral,wassolvedindifferentwaysbyPreston(1934),Page(1969)
andDonaldson(1991).Anotheropticaltechniquethathasbeenappliedmakesuseofthefactthat
naturalorinducedchecksandsplitsinthecellwallfollowsthedirectionofthemicrofibrilsintheS2
layer.Theangleofthesechecksandsplitsnormallyservesasareasonableaccuratemeasureofthe
averageMFA(Preston1947;Huang1995).Similartechniquesutiliseiodineinfiltrationofthecell
wall,whichcrystallizesinthechecksorsplitsbetweenthemicrofibrils,andassistwithmeasurement
ofcheckorientation(BaileyandVestal1937;SenftandBendtsen1985).Theorientationoftheoval
apertureofcrossfieldpitshasalsobeenusedtomeasuremicrofibrilangle(Pillowetal.1953;
Donaldson1998).Themicrofibrilanglecanalsobedeterminedbyusingdecaycavitiescausedby
softrotfungi,whichalignthemselvesalongthemicrofibrilsintheS2layer(Anagnostetal.,2000).
SmallangleandwideangleXrayscatteringareothertechniquesthathavebeenusedformany
years(Wardrop1952;Meylan1967).Techniquesusedinrecentyears,anddiscussedbyHuangetal.
(1998),includethefollowing:MicroRamanspectroscopy(Pleasantsetal.1997),transmission
ellipsometry(Yeetal.1994)andconfocalmicroscopy(VerbelenandStickers1995;Batcheloretal.
1997).Huangetal.(1998)concludedthatmeasuringpitaperturesisthesimplestmethodforusein
field,whilstiodinestaining,complementedwithultrasonicchecking,issuitableformoreaccurate
measurement.
Themethodofchoice,accordingtoBarnettandBonham(2004)remainsXraydiffraction.The
advantagesofXraydiffractionoverothermethodshavebeendiscussedbyCave(1997a,b)and
includethespeedofdeterminationandaccuracyofXraydiffractionaswellasthefactthatthe
averagemicrofibrilanglecanbedeterminedforalargesampleoftracheids.Silviscan2usesXray
diffractiontoobtainmicrofibrilanglesdirectlyfromincrementcores(Evansetal.1998).
Recently,nearinfraredspectroscopy(NIR)wasusedinseveralstudiestodeterminemicrofibril
anglesofwoodspecimens(SchimleckandEvans2004;Kelleyetal.2004b;ZbonakandBush2006).
AllthesestudiesindicatedrelativelyhighcorrelationcoefficientsbetweentheNIRpredicted
microfibrilangleandthemeasuredangleforthecalibrationdatasets(r>0.8).Thecorrelation
coefficients,however,forindependenttestorpredictiondatasetswereconsiderablylower(r<0.6).
BarnettandBonham(2004)concludedthatitisaxiomaticamongthoseworkinginthefieldof
microfibrilmeasurementthattheresultsobtainedforasinglesamplecanvarywidelyaccordingto
themethodused.
Acousticmeasurementmethods
Jayne(1959)proposedthehypothesisthattheenergystorageanddissipationpropertiesofwood
arecontrolledbythesamemechanismsthatdeterminestaticbehaviourofthewood.Asa
consequenceusefulrelationshipsbetweentheacousticandvibrationalproperties,andstatic
elasticityandstrengthareattainable.Forinstance,astrongrelationshipexistsbetweenthe
microfibrilangle,animportantdeterminantofwoodstiffness,andtheacousticvelocityinwood.
Wangetal.(2007)reportedacoefficientofdetermination(R)valueof0.855betweenacoustic
velocityandmicrofibrilangleofradiatapineandEvansandIlic(2001)anRvalueof0.86.
Thereareanumberoftechniquesthatcanbeusedtomeasuretheseenergystorageanddissipation
propertiesoftimberbutthepossibilitiesforstandingtreesaremorelimited.Themethodofchoice
instandingtreesisthemeasurementofstresswavespeedthroughthestemofatree.Inatypical
setup(Fig3)atransmitterprobeandreceiverprobeisinsertedinthesapwoodofatreestemanda
stresswaveisinducedbytappingthetransmitterprobewithalighthammer(WangandRoss2002).
Thetimeofflight(TOF)ofthewaveismeasuredbetweenthetransmitterandreceiverprobe.The
TOFcanbeusedtocalculatethewavespeedandthedynamicmodulusofelasticity(MOEd)ineither
thelongitudinalorradialdirectionofthestem.TheMOEdiscalculateddirectlyfromthewavespeed
17
andthegreendensityoftimber,whichnormallyrelateswelltotothestaticMOE.Wielingaetal.
(2009)foundthatgreenwooddensityofPinusradiatavarieswithinsuchasmallrangethatitcanbe
consideredaconstantvalueinthecaseofstandingtrees,meaningthatMOEdisalmostdirectly
relatedtovelocity.
Fig.3Atypicalstresswavemeasurementsetup
Wangetal.(2007)statedthattheprecisionofacoustictechnologyhasbeenimprovedtothepoint
wheretreequalityandintrinsicwoodpropertiescanbepredictedandcorrelatedtostructural
performanceofthefinalproducts.WangandRoss(2002)reviewedanumberofstudiesthat
exploredtherelationshipbetweenstandingtreestresswaveproperties(radialandlongitudinal)and
somemechanicalpropertiesoflogs,sawntimberandclearwoodfromthesetrees.Correlation
coefficientsbetweenstandingtreeacousticmeasuresandthestaticmodulusofelasticity(MOEs)of
defectcontainingtimberfromthesetreesvariedfrom0.33to0.64(IkedaandArima2000;
Mathesonetal.2002;Ishigurietal.2006).InstudiesrelatingMOEdofstandingtreeswithMOEsand
MORofclearpiecesofwoodfromthosetreestheauthorsobtainedcorrelationcoefficientsof0.63
0.69forMOEsand0.360.65forMOR(Wangetal.2000a;Wangetal.2001;Ivkovietal.2009).
OtherstudiesexploredtherelationshipbetweenstandingtreeacousticmeasurementsandtheMOE
ofclearwoodortimberdetermineddynamicallyorwithtransversevibrationmethods(Huang2000;
Wuetal.2000;Wangetal.2005;Grabianowskietal.2006).
Ilic(2003)proposedamethodofusingimpactinducedresonancevibrationsonsmalllongitudinal
beamsof20x2x150mmthatcanbeobtainedfromtheouterwoodofstandingtrees.Resultsshowed
anexcellentcorrelation(r=0.98softwoodsand0.97hardwoods)betweenthedynamicMOEdof
thesesmallbeamsanddynamicMOEdofstandardclearspecimens.Inadifferentstudy,thedynamic
MOEdofradiatapineofthesesmalllongitudinalbeamswerefoundtobegoodpredictorsofboth
thestaticMOEs(r=0.71)andMOR(r=0.61)ofclearsamplesfromtreesofoneofthestudysites.
However,onanothersiteMOEdprovedtobeonlyapoorpredictorofMOEs(r=0.31)andMOR
(r=0.03)(Ivkovietal.2009).Arapidmethodforremovingsuchbeamsfromastandingtreewasnot
discussedinthesearticles.
Inaslightlydifferentapproachforpredictingstandingtreewoodstiffness,Bucur(1983)determined
ultrasonicwavespeedsthrough5mmincrementcorestoobtainthreestiffnessmoduliandthree
shearmodulifortheincrementcores.Theseultrasonicallydeterminedmoduliwerecomparedto
staticallydeterminedclearwoodmodulifromthesametrees.Moderatecorrelationsbetweenthe
variousultrasonicallydeterminedmoduliandstaticallydeterminedmoduliwereobtainedi.e.
correlationbetweendynamiclongitudinalstiffnessofanincrementcoreandstaticlongitudinal
stiffnessofastandardclearspecimenwas0.67.
18
Ultrasonicimagingortomographyisatechniqueusedtoreconstructthematerialunderinspection
fromglobalwavepropagationdata(Bucur,2003).Thistechniquecanalsobeusedonstandingtrees.
AswithCTscanningacrosssectionalpropertyimagecanbeobtainedwhichcanbeexpanded,with
enough crosssectional images, into a threedimensional internal image of the material. Ultrasonic
imaging on standing trees is mostly used for decay studies (i.e. Comino et al. 2000). According to
Bucur(2003)theelasticconstantsofsawnlumberwascalculatedusingultrasonicimaginginastudy
byChazelasetal(1988).Itisassumedthatthesamemightbepossibleforstandingtreesmaking
ultrasonic imaging a potential method for predicting the mechanical properties of timber from
standingtrees.
Nearinfraredspectroscopy
Nearinfrared(NIR)spectroscopyinvolvesthestudyoftheinteractionofelectromagneticradiationin
thenearinfraredregion,whichistheinfraredregionclosesttothevisibleregionwithmaterials(Fig
4).TheNIRregionextendsfrom780to2500nminwhichthespectramaybecharacterizedbythe
assignmentoftheabsorptionbandstoovertonesandcombinationsoffundamentalvibrations
associatedwithCH,OHandNHbonds.AsNIRspectroscopicresultsontheirownareoflimited
use,multivariateanalysistechniquesneedtobeemployedtoanalysethesignificanceof
relationshipsbetweenNIRmeasurementsandotherpropertiesofinterestinwood.SinceNIR
spectroscopyisrelativelyfast,cheapandeasytoperform,itisanidealmethodtopredictrelated
woodpropertieswhicharetimeconsumingandcostlytoobtain(Soetal.2004).
0.0001nm
0.01nm
Gammarays
Xrays
10nm
Ultra
violet
1000nm
Infrared
Visiblelight
0.01cm
1cm
1m
100m
Radiowaves
RadarTVFMAM
Fig.4Theelectromagneticspectrum
NIRdatacanbeobtainedfromincrementcoresfromstandingtrees.Theanalysisnormallyinvolves
calibrationusingapartofthesamples(calibrationset)toquantifytherelationshipbetweentheNIR
resultsandthepropertyofinterest.Theremainingsamples,thetestorpredictionset,areusedto
testthereliabilityoftherelationshipthatwasobtained(Schimlecketal.2001).
VariousworkersinvestigatedthepotentialofNIRspectroscopytopredictmechanicalpropertieslike
MOEandMORofsmallclearwoodsamples.Schimlecketal.(2001)foundcoefficientsof
determinationof0.90and0.77fortherelationshipsbetweenNIRdataandMOEandMOR
respectivelyforclearEucalyptisdelegatensissamplesofthecalibrationset.Similarresultswere
obtainedinotherstudies(GindlandTeischinger2001;ThummandMeder2001;Kelleyetal.2004a;
Kelleyetal.2004b).Viaetal.(2003)foundwholetreeregressioncoefficientsofdetermination
largerthan0.84forbothMOEandMORofclearPinuspalustrissamplesemployingdifferent
statisticalanalysistechniques.Theyfound,however,thatwhenonlypithwoodwasconsidered,the
predictiveabilityofthemodelsdecreasedsignificantly.HoffmeyerandPedersen(1995)investigated
theabilityoftheNIRmethodtopredicttheMORoffullsizeddefectcontainingtimberbutthe
coefficientofdeterminationforthepredictionsetwasfairlyweak(r=0.29).
Discussionsandconclusions
Itisclearfromtheexistingworkonpropertymeasurementmethodsthattheobjectiveofpredicting
mechanicalpropertiesfromstandingtreeswillplayanimportantroleinselectingthemost
appropriatemethodofstudy.Forinstance,severalauthorsmentionedthatacertainmethodmaybe
19
sufficientforgenetictrials,butnotforindividualtreepredictions(e.g.Cownetal.1998;IsikandLi
2003).Inordertomakepredictionsofthemechanicalpropertiesofdefectcontainingtimberfrom
standingtrees,aspeciesandresourcespecificapproachwillhavetobefollowed.Foraspecific
forestresource,agoodstartingpointshouldbeaninvestigationintotherelativestrengthof
correlationsbetweenpropertiesmeasurableonsawnboardsandthemechanicalpropertiesofthese
boards(e.g.Johansson2003),whichwillgiveanindicationofstandingtreepropertiesthatneedto
beconsidered.
Animportantaspecttokeepinmindwhentryingtopredictthemechanicalpropertiesofstructural
timberandspecificallytherecoveryofstructuralgradesfromastandingtreeresource,isthatthe
characteristicstrengthsordesignstrengthsofagradeisdeterminedbythe5thpercentilestrength
values(Fig5)inotherwordstheweakportionofthestrengthdistributioncurve(Madsen1992).
Thismeansthatitisessentialforanypropertyandmethodusedinsuchastudytobeanaccurate
predictoroftheweakportionofthestrengthdistributioncurve.Withafewexceptions(e.g.Bier
1985;Viaetal.2003)moststudiesreferredtoabove,evaluatedmethodsorpropertiesthatcanbe
usedforstrengthpredictionpurposesusingtherelationshipofthatspecificpropertywithstrength
overthefullstrengthdistributioncurve.Forstiffness,however,designcodesoftenuseboth5th
percentileandaverageMOEvaluesmakingpredictionsofthefullstiffnessdistributionnecessary
(i.e.CSAO8601,2001).
16
th
5 percentile
Percentageofpieces
14
12
10
8
6
4
2
0
10
15
20
25
30
35
40
45
50
55
ModulusofRupture(MPa)
60
65
70
Fig.5AtypicalhistogramoftheMORoftimberwiththe5thpercentilevalueindicated
Althoughtheuseofclearanddefectfreetimberinstructuralapplicationsisverylimited,byfarthe
mostofthestudiesusedclearanddefectfreetimbertoevaluatestandingtreeproperty
measurementmethods.Itisobviousthatthedefectfreestrengthandstiffnesswillalsohavean
effectonthepropertiesofdefectcontainingtimberpieces,butinsomecasesdefectsmight
overshadowtheimportanceofcertainproperties,especiallyintheimportantweakportionofthe
strengthdistributioncurve.Anareathathasreceivedlimitedattentionfromscientistsinrecent
yearsisthatofknotandbranchassessmentmethods.Sinceknotsplayanimportantroleintimber
mechanicalproperties(Johanssen2003),andespeciallyintheweakportionofthestrength
distributioncurve(seeforinstancethestudyofBier1985),considerationofthispropertyin
predictivestudiesmightbeadvantageousforsomespecies.
Theefficiencyofstandingtreepropertymeasurementmethodshavebeenevaluated,almost
exclusivelyintheliteraturereviewed,intermsofbendingstrength(MOR)andlongitudinalstiffness
(MOE).Designcodesgenerallyspecifysixcharacteristicstrengthvaluesaswellaslongitudinal
stiffnessvaluesforstructuraldesignapplications.Theassumptioninproofgradingormechanical
stressgrading,whereonlyonevaluelikeMOEisoftenmeasured,isthatagoodrelationshipexists
betweenthedifferentstrengthandstiffnessvalues.Foranytreeresourcethisassumptionwillhave
tobevalidated.
20
Insomecasesacombinationofmethodsandpropertiesmightberequiredtoobtaintherequired
accuracyofprediction.Atechniquewhichcanbeusedtodeterminewhichisthebestmeasurement
methodforaspecificforestresource,istoperformapathanalysis(Downesetal.2002andIvkovi
etal.2009).Resultsarerepresentedinpathdiagramstohelptheresearchertodecidewhatmethod
orcombinationofmethodsisthebestforaspecificstudy(Fig6).
Fig.6PathanalysisforMOEofradiatapine.Correlationcoefficientsindicatedforeacheffect.DEN=density,
RW=ringwidth,MfA=microfibrilangle,SLG=spiralgrain(fromIvkovicetal.2009).
Althoughitisdifficulttocomparestandingtreepropertymeasurementmethodswithoutspecific
projectorstudyobjectivesinmind,onecandrawsomegeneralconclusionsfromtheliterature
surveyed.Anumberofmethodshavebeendevelopedoverthelastfewdecades.Morerecentlythe
focuswasprimarilyonacousticandNIRspectroscopymethods,aswellasonmultiproperty
measurementsystemssuchasSilviscan.Eachmethodhasspecificapplicationsandadvantages,but
theyallrequirearelativelylowlabourinput,resultinginrelativelylowmanpowercost.Acoustic
methods,suchasthestresswavetimer,arethefastestandlowestcostmethodsasasingleoperator
canobtainareadingwithinamatterofseconds.Variousstudiesfoundsignificantcorrelations
betweenacousticstresswavepropertiesandtheMOEandMORvaluesofclearwoodandtimber.
NIRspectroscopymethodshavetheadvantageofbeingabletopredictmanydifferentbasic
propertieslikedensity,microfibrilangle,spiralgrainaswellasMOEandMORofclearwoodfrom
incrementcores.However,thisapproachrequiresaprocessofcalibrationasitisnotadirect
measurementmethodofanyproperty.Silviscanusesanautomatedmeasurementprocessof
incrementcorestoobtaindensity,microfibrilangleandcelldimensionalproperties.Spiralgrain
anglecanalsobeobtainedfromSilviscanresults.Thesebasicpropertieshavebeenshowntohavea
significantrelationshipwithsomemechanicalpropertiesofwood.
21
References
AnagnostSE,MarkRE,HannaRB(2000)Utilisationofsoftrotcavityorientationforthe
determinationofmicrofibrilangle.WoodFiberSci32:8187.
AS2858(2003)AustralianStandard.Timbersoftwoodvisualstressgradingrulesforstructural
purposes.
BaileyIW,VestalMR(1937)Theorientationofcelluloseinthesecondarywalloftrachearycells.Jof
theArnoldArboretum18:185195.
BarnettJR,BonhamVA(2004)Cellulosemicrofibrilangleinthecellwallofwoodfibres.BiolReviews
79:461472.
BatchelorWJ,ConnAB,ParkerIH(1997)Measuringthefibrilanglesoffibresusingconfocal
microscopy.Appita50:377380.
BierH(1985)Bendingpropertiesofstructuraltimberfroma28yearoldstandofNewZealandPinus
radiata.NewZealandJofForSci15(2):23350.
BierH(1986)Logqualityandthestrengthandstiffnessofstructuraltimber.NewZealandJofForSci
16(2):176186.
BrazierJD(1965)Anassessmentoftheincidenceandsignificanceofspiralgraininyoungconifer
trees.ForProdJ15:308312.
BrashawBK,BucurV,DivosF,GoncalvesR,LuJ,MederR,PellerinRF,PotterS,RossRJ,WangX,Yin
Y(2009)Nondestructivetestingandevaluationofwood:Aworldwideresearchupdate.ForProdJ
59(3):714.
BrisbinRL,SondermanDL(1971)Treegradesforeasternwhitepine.USDAFor.Serv.Res.Pap.NE
214.
BucurV(1983)Anultrasonicmethodformeasuringtheelasticconstantsofwoodincrementcores
boredfromlivingtrees.Ultrasonics,May1983:116126.
BucurV(2003)Nondestructivecharacterizationandimagingofwood.Springer354p.
BuksnowitzC,MllerU,EvansR,TeischingerA,GrabnerM(2008)ThepotentialofSilviScansXray
diffractometrymethodfortherapidassessmentofspiralgraininsoftwood,evaluatedby
goniometricmeasurements.WoodSciTechn42:95102.
CaveID,WalkerJCF(1994)Stiffnessofwoodinfastgrownplantationsoftwoods:theinfluenceof
microfibrilangle.ForProdJ44(5):4349.
CaveID(1997a)TheoryofXraymeasurementofmicrofibrillangleinwood.Part1.Theconditionfor
reflectionXraydiffractionbymaterialswithfibretypesymmetry.WoodSciTechnol31:143152.
CaveID(1997b)TheoryofXraymeasurementofmicrofibrilangleinwood.Part2.Thediffraction
diagramXraydiffractionbymaterialswithfibretypesymmetry.WoodSciTechnol31:225234.
22
ChantreG,RozenbergP(1997)Candrillresistanceprofiles(Resistograph)leadtowithinprofileand
withinringdensityparametersinDouglasfirwood?InZhang,S.Y.,Gosselin,R.,andG.Chauret(eds),
ProceedingsofIUFROWoodQualityWorkshop:Timbermanagementtowardwoodqualityandend
productvalue,Quebec,1822August1997.pp4147.
ChazelasJL,VergneA,BucurV(1988)Analysedelavariationdespropritesphysiqueetmchanique
localesduboisautourdesnoeuds.(Woodlocalpropertiesvariationaroundknots).Actesdu
ColloqueComportementMcaniqueduBois.GSRhologieduBois,Bordeaux,France,pp344347.
ClarkA,McAlisterRH(1998)Visualtreegradingsystemsforestimatinglumberyieldsinyoungand
maturesouthernpine.ForProdJ48(10):5967.
CominoE,SoccoV,MartinisR,NicolottiG,SambuelliL(2000)Ultrasonictomographyforwooddecay
diagnosis.InBackhaussGF,BaderH,IdezakE(eds)IntSympPlantHealthinUrbanHorticulture.Mitt.
BundesanstLandForstwirtschaft,Braunschweig,p279.
CownDJ(1978)Comparisonofthepilodynandtorsiometermethodsfortherapidassessmentof
wooddensityinlivingtrees.NewZealandJForSci8(3):38491.
CownDJ,ClementBC(1983)AwooddensitometerusingdirectscanningwithXrays.WoodSci
Technol17:9199.
CownDJ,McConchieM,McConchieDL(1998)DevelopmentsinPilodynassessmentoftreestems
andlogs.In:ProceedingsoftheEleventhInternationalSymposiumonNondestructiveTestingof
Wood,September911,MadisonWisconsin.
CownDJ(2006)Woodqualityinstandingtimberevolutionofassessmentmethodsinplantations.
In:Kurjatko,S.,Kdela,J.andR.Lagaa(eds.)2006.Proceedingsofthe5thIUFROSymposiumWood
StructureandProperties06,September36,SliaSielnica,Slovakia.Organisedjointlybythe
FacultyofWoodSciencesandTechnologyoftheTechnicalUniversityofZvolenandtheIUFRO
Division5ForestProducts5.01.00.
CrickmayandAssociates(2009)SouthAfricanLumberIndex:November2009.
CSAO8601(2001)CanadianStandardsAssociation.EngineeringDesigninWood.
DinwoodieJM(2000)Timber:Itsnatureandbehaviour.E&FNSpon,LondonandNewYork.257pp.
DivosF,SzegediS,RaicsP(1996)Localdensitometryofwoodbygammabackscattering.HolzRoh
Werkst54:279281.
DonaldsonLA(1991)Theuseofpitaperturesaswindowstomeasuremicrofibrilangleinchemical
pulpfibres.WoodFiberSci23:290295.
DonaldsonLA(1998)Betweentracheidvariabilityofmicrofibrilanglesinradiatapine.InB.G.
Butterfield(ed),1998,ProceedingsoftheIAWA/IUFROInternationalworkshoponthesignificance
ofmicrofibrilangletowoodquality,pp.206224,Westport,UniversityofCanterburyPress,
Canterbury,NewZealand.
23
DownesGM,NyakuengamaJG,EvansR,NorthwayR,BlakemoreP,DicksonRL,LausbergM(2002)
Relationshipbetweenwooddensity,microfibrilangleandstiffnessinthinnedandfertilizedPinus
radiata.IAWAJournal23(3),253265.
EcholsRM(1955)Linearrelationoffibrilangletotracheidlength,andgeneticcontroloftracheid
lengthinslashpine.TropicalWoods102:1122.
EN518(1995)EuropeanStandard.Structuraltimber.Grading.Requirementsforvisualstrength
gradingstandards.
EvansR,HughesM,MenzD(1998)MicrofibrilanglevariationbyscanningXraydiffractometry.
AppitaJournal51:2733.
EvansR,IlicJ(2001)Rapidpredictionofwoodstiffnessfrommicrofibrilangleanddensity.ForProdJ
51(3):5357.
GhodgaonkarDK,MajidWMBWA,HusinHB(2000)MicrowavenondestructivetestingofMalaysian
timberforgradingapplications.WorldConferenceonTimberEngineering,WhistlerResort,British
Columbia,Canada,July31August3.
GindlW,TeischingerA(2001)Therelationshipbetweennearinfraredspectraofradialwood
surfacesandwoodmechanicalproperties.J.NearInfraredSpectroscopy9:255261.
GindlW,TeischingerA(2002)ThepotentialofVisandNIRspectroscopyforthenondestructive
evaluationofgrainangleinwood.WoodFiberSci34(4):651656.
GlosP,HeimeshoffB(1982)Mglichkeitenundgrenzenderfestigkeitssortierungvonbrettlamellen
frdenholzleimbau.InIngenieurholzbauinFoschungundPraxis(EhlbeckundSteck).Bruderverlag,
Karlsruhe.(InGerman)
GrabianowskiM,ManleyB,WalkerJCF(2004)Impactofstockingandexposureonouterwood
acousticpropertiesofPinusRadiatainEyrewellForest.NewZealandJFor:Aug2004.
GrabianowskiM,ManleyB,WalkerJCF(2006)Acousticmeasurementsonstandingtrees,logsand
greenlumber.WoodSciTechnol40:205216.
HabermehlA,RidderHW,SeidlP(1999)Computerizedtomographicsystemsastoolsfordiagnosing
urbantreehealth.InLemattreM,LemattreP,LemaireF(ed),1999,ProcIntSymponUrbanTree
Health,ActaHorticulture496.
HallingbckH(2010)GeneticImprovementofShapeStabilityinNorwaySpruceandScotsPineSawn
Timber.DoctoralThesis,SwedishUniversityofAgriculturalSciences,
Uppsala.
HankinsonRL(1921)Investigationofcrushingstrengthofspruceatvaryinganglesofgrain.AirForce
InformationCircularNo.259,U.S.AirService.
HarrisJM(1984)Nondestructiveassessmentofspiralgraininstandingtrees.NewZealandJForSci
14(3):39599.
24
HaukssonJB,BergqvistG,BergstenU,SjstrmM,EdlundU(2001)Predictionofbasicwood
propertiesforNorwayspruce.InterpretationofNearInfraredSpectroscopydatausingpartialleast
squaresregression.WoodSciTechnol35:474485.
HoffmeyerP(1984)Omkonstruktionstraesstyrkeochstyrkesortiering.ISkovteknologi.Ethistoriskt
ogperspektiviskstrejtog.DanskSkovforening.(InDanish).
HoffmeyerP(1990)Failureofwoodasinfluencedbymoistureanddurationofload.Doctoralthesis,
StateUniversityofNewYork,CollegeofEnvironmentalScienceandForestry,Syracuse,NewYork.
HoffmeyerP,PedersenJG(1995)EvaluationofdensityandstrengthofNorwaysprucewoodbynear
infraredspectroscopy.HolzRohWerkst53:165170.
HuangC.L(1995)Revealingfibrilangleinwoodsectionsbyultrasonictreatment.WoodFiberSci27:
4954.
HuangCL,KutschaNP,LeafGJ,MegrawRA(1998)Comparisonofmicrofibrilanglemeasurement
techniques.InB.G.Butterfield(ed),1998,MicrofibrilAngleinWood:ProceedingsoftheIAWA/
IUFROInternationalworkshoponthesignificanceofmicrofibrilangletowoodquality,pp.177205,
Westport,UniversityofCanterburyPress,Canterbury,NewZealand.
Huang,C.L.,2000)Predictinglumberstiffnessofstandingtrees.InDivos,F.(ed),2000,Proceedings,
12thInternationalsymposiumonnondestructivetestingofwood,September1315;Universityof
WesternHungary,Sopron:173180.
HuangCL,LindstrmH,NakadaR,RalstonJ(2003)Cellwallstructureandwoodproperties
determinedbyacousticsaselectivereview.HolzRohWerkst61:321335.
IkedaK,ArimaT(2000)Qualityevaluationofstandingtreesbyastresswavepropagationmethod
anditsapplication.II.Evaluationofsugistandsandapplicationtoproductionofsugistructural
squaretimber.MokuzaiGakkaishi.46(3):189196.(InJapanese).
IlicJ(2003)DynamicMOEof55speciesusingsmallwoodbeams.HolzRohWerkst61:167172.
IshiguriF,KawashimaM,IizukaK,YokotaS,YoshizawaN(2006)Relationshipbetweenstresswave
velocityofstandingtreeandwoodqualityin27yearoldHinoki(ChamaecyparisobtusaEndl).JSoc
MatSci,Japan55(6):576582.
IsikF,LiB(2003)Rapidassessmentofwooddensityoflivetreesusingtheresistographforselection
intreeimprovementprograms.CanJForRes33:24262435.
IvkoviM,GapareWG,AbarquezA,IlicJ,PowellMB,WuHX(2009)Predictionofwoodstiffness,
strength,andshrinkageinjuvenilewoodofradiatapine.WoodSciTechnol43:237257.
JamesWL,YenYH,KingRJ(1985)Amicrowavemethodformeasuringmoisturecontent,density,
andgrainangleofwood.USDAForestProductsLaboratory,ResearchnoteFPL0250.
JayneBA(1959)Indicesofquality:Vibrationalpropertiesofwood.ForProdJ9(11):413416.
JohanssonCJ(1976)Tensilestrengthofglulamlaminations.ChalmersUniversityofTechnology,
SteelandTimberStructures,InternalreportnoS76:18(InSwedish).
25
JohanssonCJ,BrundinJ,GruberR(1992)StressgradingofSwedishandGermantimber.A
comparisonofmachinestressgradingandthreevisualgradingsystems.SwedishNationalTesting
andResearchInstitute,SPReport1998:38.
JohanssonCJ,BostrmL,BrunerL,HoffmeyerP,HolmquistC,SolliKH(1998)Laminationsforglued
laminatedtimberEstablishmentofstrengthclassesforvisualstrengthgradesandmachinesettings
forglulamlaminationsofNordicorigin.SwedishNationalTestingandResearchInstitute,SPREPORT
1998:38.
JohanssonCJ(2003)Gradingoftimberwithrespecttomechanicalproperties.InThelandersson,S.,
andJ.L.Larsen.(eds).2003.TimberEngineering.JohnWileyandSonsLtd.
KaestnerAP,BthLB(2005)Microwavepolarimetrytomographyofwood.IEEESensorsJournal
5(2):209215.
KelleySK,RialsTG,GroomLH,SoCH(2004a)Useofnearinfraredspectroscopytopredictthe
mechanicalpropertiesofsixsoftwoods.Holzforschung58:252260.
KelleySK,RialsTG,SnellR,GroomLH,SluiterA(2004b)Useofnearinfraredspectroscopyto
measurethechemicalandmechanicalpropertiesofsolidwood.WoodSciTechnol38:257276.
KochM,HunscheS,SchuacherP,NussMC,FeldmannJ,FrommJ(1998)THzimaging:anewmethod
fordensitymappingofwood.WoodSciTechnol32:421427.
KoizumiA,UedaK(1986)Estimationofthemechanicalpropertiesofstandingtreesbybendingtests
I.MokuzaiGakkashi32(9):669676(InJapanese).
KromhoutCP(1966)Treesamplingjigforwoodqualitytests.ForestryinSouthAfrica6:107112.
LacknerR,FoslieM(1988)GranfraVestlandetStyrkeochsortierung.NorwegianInstituteofWood
Technology,Report74.(InNorwegian).
LaunayJ,RozenbergP,PaquesL,DewitteJM(2000)Anewexperimentaldeviceforrapid
measurementofthetrunkequivalentmodulusofelasticityonstandingtrees.AnnForSci57:351
359.
LaunayJ,IvkovichM,PaquesL,BastienC,HigelinP,RozenbergP(2002)Rapidmeasurementoftrunk
MOEonstandingtreesusingRigidimeter.AnnForSci59:465469.
LeicesterRH,SeathCA(1996)Applicationofmicrowavescannersforstressgrading.In:4th
InternationalWoodEngineeringConference,NewOrleans,2:435440.
LesninoG(1994)Thelasersandblastingmethod:Anewmethodforthequalitativeannualring
analysisofconifers.WoodSciTechnol.28:159171.
LindstrmH,HarrisP,NakadaR(2002)Methodsformeasuringstiffnessofyoungtrees.HolzRoh
Werkst60:165174.
MadsenB(1992)Structuralbehaviouroftimber.TimberEngineeringLtd,NorthVancouver,Canada.
26
MalanFS,MaraisPG(1992)Somenotesonthedirectgammaraydensitometryofwood.
Holzforschung46(2):9197.
MartinP,ColletR,BarthelemyP,RoussyG(1987)Evaluationofwoodcharacteristics:Internal
scanningofmaterialbymicrowaves.WoodSciTechnol.21(4):361371.
MathesonAC,DicksonRL,SpencerDJ,JoeB,IlicJ(2002)AcousticsegregationofPinusradiatalogs
accordingtostiffness.AnnForSci59:471477.
McDonaldKA,BendtsenBA(1986)Measuringlocalizedslopeofgrainbyelectricalcapacitance.For
ProdJ36(10):7578.
McLauchlanTA,NortonJA,KusecDJ(1973)Slopeofgrainindicator.ForProdJ23(5):5055.
MegrawR,BremerD,LeafG,RoersJ(1999)StiffnessinLoblollypineasafunctionofringposition
andheight,anditsrelationshiptomicrofibrilangleandspecificgravity.InThirdWorkshop,
Connectionbetweensilvicultureandwoodqualitythroughmodellingapproachesandsimulation
software,IUFROWPS5.0104,LaLondeLesMaures,France,Sept512,1999.
MeylanBA(1967)MeasurementofmicrofibrilangleinPinusradiatabyXraydiffraction.ForProdJ
15:5158.
NoskowiakAF(1968)Spiralgrainpatternsfromincrementcores.ForProdJ18:5760.
OnoeM,TsaoJW,TamadaH,NakamuraH,KogureJ,KawamuraH,YoshimatsuM(1984)Computed
tomographyformeasuringtheannualringsonalivetree.NuclearInstrumentsandMethodsin
PhysicsResearch221:213220
OroszI(1969)Modulusofelasticityandbendingstrengthratioasindicatorsoftensilestrengthof
lumber.JofMaterials,4(4):842864.
PageDH(1969)Amethodfordeterminingthefibrillarangleinwoodtracheids.JofMicroscopy
90:137143.
ParkJC(1989)Applicationsoftheseesawsimulatorandprunedlogindextoprunedresource
evaluationsAcasestudy.NewZealandJForSci19(1):6882.
ParkJC(1994)EvaluatingprunedsawlogqualityandassessingsawmillrecoveriesinNewZealand.
ForProdJ44(4):4352.
PillowMY,TerrelBZ,HillerCH(1953)Patternsofvariationinfibrilanglesinloblollypine.USDAFPL
reportD1935.
PleasantsSW,BatchelorWJ,ParkerIH(1997)Measuringthefibrilangleofbleachedfibresusing
microramanspectroscopy.51stAppitaAnnualGeneralConference,Melbourne,Australia.2:545
549.
PolgeH(1966)Etablssementdescourbesdevariationdeladensitduboisparexploration
densitomtriquederadiographiesdchantillonsprlevslatariresurdesarbresvivants.AnnSci
ForestXXIII,1206.(InFrench).
27
PolgeH(1978)Fifteenyearsofwoodradiationdensitometry.WoodScTechnol.12:187196.
PrestemonJP,BuongiornoJ(2000)Determinantsoftreequalityandlumbervalueinnaturaluneven
agedsouthernpinestands.CanJForRes30:211219.
PrestonRD(1934)Theorganisationofthecellwalloftheconifertracheid.PhilisophicalTransactions
oftheRoyalSociety,SeriesB224:174191.
PrestonRD(1947)ThefinestructureofthewalloftheconifertracheidII.Opticalpropertiesof
dissectedwallsinPinusinsignes.ProceedingsoftheRoyalSocietyB134:202218.
PrestonRD(1948)Thefinestructureofthewalloftheconifertracheid:IV.Dimensional
relationships.BiochimicaetBiophysicaActa2:37083.
RinnF,ScheingruberFH,SchrE(1996)ResistographandXraydensitychartsofwoodcomparative
evaluationofdrillresistanceprofilesandXraydensitychartsofdifferentwoodspecies.
Holzforschung50(4):303311.
SamsonM(1984)Measuringgeneralslopeofgrainwiththeslopeofgrainindicator.ForProdJ
34(7/8):2732.
SamsonM(1988)Transversescanningforautomaticdetectionofgeneralslopeofgraininlumber.
ForProdJ38(7/8):3338.
SamsonM,TremblayC,LanglaisPA(1993)Measuringslopeofgrainbyelectricalcapacitanceat
moisturecontentsabovefibersaturation.ForProdJ43(2):5860.
SANS17832(2005)SouthAfricanNationalStandard.Sawnsoftwoodtimber.Part2:Stressgraded
structuraltimberandtimberforframewallconstruction.
SchajerGS,OrhanFB(2006)Measurementofwoodgrainangle,moisturecontentanddensityusing
microwaves.HolzRohWerkst64:483490.
SchimleckLR,EvansR,IlicJ(2001)EstimationofEucalyptusdelegatensiswoodpropertiesbynear
infraredspectroscopy.CanJForRes31(10):16711675.
SchimleckRL,EvansR(2004)EstimationofPinusradiatatracheidmorphologicalcharacteristicsby
nearinfraredspectroscopy.Holzforschung58:6673.
SchroederJG,CampbelRA,RodenbachRC(1968)Southernpinetreegradesforyardandstructural
lumber.ResearchpaperSE40.SouthernResearchstation,Asheville,N.C.15pp.
SenftJ,BendtsenBA(1985)Measuringmicrofibrillaranglesusinglightmicroscopy.WoodFibreSci
17:564567.
SmithDM(1954)Maximummoisturecontentmethodfordeterminingspecificgravityofsmallwood
samples.Report2014U.S.DepartmentofAgriculture.
SmithWJ(1959)TracheidlengthandmicellarangleinHooppine(AraucariacunninghamiiAit.)
theirvariation,relationships,anduseasindicatorsinparenttreeselection.QueenslandForest
ServiceResearchNotesNo.8,Brisbane,Australia.
28
SoCL,ViaBK,GroomLH,SchimleckLR,ShupeTF,KelleySS,RialsTG(2004)Nearinfrared
spectroscopyintheforestproductsindustry.ForProdJ54(3):616.
ThummA,MederR(2001)Stiffnesspredictionofradiatapineclearwoodtestpiecesusingnear
infraredspectroscopy.J.NearInfraredSpectroscopy9:117122.
TiuriM,JokelaK,HeikkilaS(1980)Microwaveinstrumentforaccuratemoistureanddensity
measurementoftimber.JMicrowavePower15:251254.
TognettiR,RaschiA,BeresC,FenyvesiA,RidderHW(1996)Comparisonofsapflow,cavitationand
waterstatusofQuercuspetraeaandQuercuscerristreeswithspecialreferencetocomputer
tomography.PlantCellEnviron19:928938.
TsoumisG(1991)Scienceandtechnologyofwood.Structure,properties,utilization.VanNostrand
Reinhold,NewYork.494pp.
UusitaloJ(1997)Preharvestmeasurementofpinestandsforsawingproductionplanning.Acta
ForestaliaFennica259.56p.
VafaiA,FarshadM(1979)Modulusofelasticityofwoodinstandingtrees.WoodScience12(2):93
97.
VerbelenJP,StickensD(1995)Invivodeterminationoffibrilorientationinplantcellwallswith
polarisation.JofMicroscopy177:16.
ViaBK,ShupeTF,GroomLH,StineM,SoCL(2003)Multivariatemodellingofdensity,strengthand
stiffnessfromnearinfraredspectraformature,juvenileandpithwoodoflongleafpine(Pinus
palustris).NearInfraredSpectroscopy11:365378.
WielingaB,RaymondCA,JamesR,MathesonAC(2009)EffectofgreendensityvaluesonPinus
Radiatastiffnessestimationusingastresswavetechnique.NewZealandJournalofForestryScience
39:7179.
WimmerR(1995)Intraannualcellularcharacteristicsandtheirimplicationsformodellingsoftwood
density.WoodFiberSci27(4):413420.
WangSY,LinCJ,ChiuCM(2005)EvaluationofwoodqualityofTaiwaniatreesgrownwithdifferent
thinningandpruningtreatmentsusingultrasonicwavetesting.WoodFiberSci37(2):192200.
WangX,RossRJ,McClellanM,BarbourRJ,EricksonJR,ForsmanJW,McGinnisGD(2000a)Strength
andstiffnessassessmentofstandingtreesusinganondestructivestresswavetechnique.Research
paperFPLRP585.U.S.DepartmentofAgriculture,ForestProductsLaboratory,Madison,WI.11p.
WangX,RossRJ,McClellanM(2000b)Strengthandstiffnessassessmentofstandingtreesusinga
nondestructivestresswavetechnique.ResearchpaperFPLRP600.U.S.DepartmentofAgriculture,
ForestProductsLaboratory,Madison,WI.9p.
WangX,RossRJ,McClellanM,BarbourRJ,EricksonJR,ForsmanJW,McGinnisGD(2001)
Nondestructiveevaluationofstandingtreeswithastresswavemethod.WoodFiberSci33(4):522
533.
29
WangX,RossRJ(2002)Nondestructiveevaluationofgreenmaterialsrecentresearchand
developmentactivities.In:Pellerin,R.F.,andR.J.Ross(eds.).2002.Nondestructiveevaluationof
wood.ForestProductsSociety,Madison.
WangX,CarterP,RossRJ,BrashawBK(2007)Acousticassessmentofwoodqualityofrawforest
materialsapathtoincreasedprofitability.ForProdJ57(5):614.
WardropAB,DadswellHE(1950)Thenatureofreactionwood:II.Thecellwallorganisationof
compressionwoodtracheids.AusJofSciRes3(1):113.
WardropAB(1952)ThelowanglescatteringofXraysbyconifertracheids.TextileResJ22:288291.
WesselsCB,PriceCS,TurnerP,DellMP(2006)Integratingharvestingandsawmilloperationsusing
anoptimizedsawmillproductionplanningsystem.InAckerman,P.A.,Langin,D.W.,andM.C.
Antonides(eds),2006,ProceedingsoftheInternationalPrecisionForestrySymposium,Stellenbosch
University,SouthAfrica,510March2006.ISBN0797211217.
WilsonTRC(1921)Theeffectofspiralgrainonthestrengthofwood.JofForestry19:740747.
WuSY,GormanTG,WagnerFG(2000)Effectofslopeaspectandscanningintensityonthe
correlationbetweenstresswavespeedsinDouglasfirtreesandlumberMOE.Presentedatthe54th
annualmeetingoftheForestProductsSociety;2000June1821;SouthLakeTahoe,Nevada.
YaoJ(1968)Modifiedmercuryimmersionmethodindeterminingspecificgravityofsmall,irregular
specimens.ForProdJ18(2):5659.
YeC,SundstromMO,RemesK(1994)Microscopictransmissionellipsometrymeasurementofthe
fibreangleandrelativephaseretardationofsingle,intactwoodpulpfibres.AppliedOptics33:6626
6637.
ZbonakA,BushT(2006)Applicationofnearinfraredspectroscopyinpredictionofmicrofibrilangle
of14yearoldpinuspatula.In:Kurjatko,S.,Kudela,J.,andR.Lagana(eds).WoodStructureand
Properties06.ArboraPublishers,Zvolen,Slovakia.
ZobelBJ,VanBuijtenenJP(1989)Woodvariation.Itscausesandcontrol.Springer,BerlinHeidelberg
NewYork.363p.
30
Chapter3.
ThestructuralgradingofyoungSouthAfricangrownPinuspatulasawntimber
Publishedin:SouthernForests(2013)75:717
G.P.Dowse
C.B.Wessels*
PrivateBagX1,
Matieland,7602
email:cbw@sun.ac.za
Abstract
Inthisstudysawntimberfrom1620yearoldPinuspatulatreeswereobtainedfromawidevariety
of sites along the Mpumalanga escarpment in South Africa. The objectives of this study were to
assess the efficiency of the current visual and mechanical grading rules on this timber and to
evaluatethepotentialofsomeindicatorpropertiestobeusedasstructuralgradingparameterson
thisresource.Alargenumberofnondestructivemeasurementswereperformedonallthesamples
(i.e. Xray density scanning, acoustic frequency measurement, yearring measurements, stiffness
measurement)beforeitwasdestructivelytestedinbendingandtension.TheyoungP.patulatimber
tested in this study had good bending strength (MOR) properties with higher characteristic grade
values than required. The timber, however, had low stiffness and did not comply with the SANS
101631 requirements for mean MOEedge for any of the structural grades. Dynamic MOE (MOEdyn),
calculatedfromacousticfrequencytestsonthetimber,wasfoundtobethebestsinglepredictorof
MOEedge, MOR and tension strength. Multiple regression analysis showed that a combination of
MOEdyn, density and knot parameters can be used to improve the predictability of some of the
strengthandstiffnesscharacteristicsofthetimber.
Keywords:structuralgrading,nondestructiveevaluation,youngPinuspatula,
Introduction
About34%oftheareaundercommercialforestryplantationsinSouthAfricaismanagedforsawlog
production. Pinus patula is the most important South African commercial plantation softwood
resourcewithatotalof338923haplantedwiththisspecies(DAFF,2009).OfthetotalSouthAfrican
sawmillproductoutput,nearly70percentisclassifiedasstructuralorbuildingtimber(Crickmayand
Associates,2009).StructurallygradedtimberinSouthAfricaisbeingusedmainlyinrooftrussesof
residentialhouseswherethesafetyoftheinhabitants,buildersandmaintenanceworkersisatstake.
TreebreedingprogrammesinSouthAfricaemphasiseahighgrowthrate.Togetherwithareduction
inrotationagesofsawlogtreesthisresultsinlogswitharelativelyhighproportionofjuvenilewood.
Itis awellestablished factthatthepropertiesofjuvenilewoodcandifferquitedramaticallyfrom
thatofmaturewood(Kennedy,1995;ZobelandSprague,1998).Thecurrentstructuralgradingrules
andthegradedesignpropertiesofSouthAfricanpinewasestablishedintheearly1980s(Knuffel,
1983, 1984) when the composition of the sawlog resource was probably different in terms of the
relativejuvenilewoodcontentoflogs.Thequestioniswhetherthecurrentstructuralgradingrules
arestilleffectiveinseparatingthisresourceaccordingtoitsactualstrengthandstiffnessproperties.
The objectives of this study were to assess the efficiency of the current visual and mechanical
grading rules on young Pinus patula sawn timber and to evaluate the potential of some indicator
propertiesthatcouldbeusedasstructuralgradingparametersonthisresource.
31
Forstructuralgradingsystemsitisimportantthattheindicatorpropertiesusedtogradethetimber
have a strong relationship with timber strength and stiffness. There are many studies where the
strength of this relationship has been determined for different properties and species and only a
selectedfewwillbediscussedhere.
Table 1 summarises the results of six studies on the species Picea abies (as summarised by
Johansson,2003).FromTable2itcanbeobservedthatmodulusofelasticity(MOE)isthesinglebest
predictorofbothbendingandtensionstrength.WhereknotdataisaddedtoMOE,the R2values
improved.Knot,density,andringwidthdataare,ontheirown,poorpredictorsofbothbendingand
tensile strength. When locations of knots are taken into account, prediction values can improve
slightly (Johansson, 2003). Knuffel (1984) found from laboratory data that visual grading of South
Africanpinecouldbeimprovedbyplacingmoreemphasisonknotdatathanondensity.
Table1:Degreesofdetermination(R )fromvariousinvestigationsoftherelationshipbetweenstrengthand
otherpropertiesofNorwaysprucetimber(Piceaabies)thesourcesoftheinvestigationarenumberedfrom1
to6(Johansson,2003)
Characteristicsthatcanbe
DegreeofdeterminationR2
measurednondestructively
MOR
Tension
Source
[1]
[2]
[3]
[4]
[1]
[5]
[6]
Knots
Annualringwidth
0.27
0.21
0.2
0.27
0.16
0.2
0.25
0.44
0.36
0.36
0.42
0.33
0.30
0.28
Density
0.16
0.3
0.16
0.4
0.38
0.29
0.38
MOE,bending
0.72
0.53
0.55
0.56
0.70
0.69
0.58
MOE,flatwise,shortspan
0.74
Knots+annualringwidth
0.37
Knots+density
0.38
Knots+MOE
0.42
0.39
0.49
0.38
0.55
0.61
0.64
0.73
0.58
0.64
0.70
0.76
[1].Johanssonetal.(1992),[2].Hoffmeyer(1984),[3].Hoffmeyer(1990),
[4].LacknerandFoslie(1988),[5].Glosetal.(1982),[6]Johanssen(1976)
0.78
Table2:Degreesofdetermination(R2)forMORpredictionsbydifferentindicatorproperties(Glos,2004).
Characteristicsthatcanbemeasurednon
DegreeofdeterminationR2
destructively
Knots
0.15 0.35
Density
0.20 0.40
Frequency,ultrasonicspeed
0.30 0.55
MOE
0.40 0.65
Knots&density
0.40 0.60
Knots&MOE
0.55 0.75
Knots&density&frequency
0.55 0.80
Glos (2004) quoted similar results to that of Table 1 but his results were not speciesspecific and
acoustic resonance frequency measurements were included (Table 2). Of interest in Glos data is
that a combination of knot properties, density and acoustic resonance frequency results, can
increase thedegreeof determination forpredictingstrengthto0.80. Modern gradingsystemsare
abletomeasureallthesepropertieswithinasinglesystem.
32
The results reported by Gaunt (1999) are very relevant to this study since comparisons of the
relationshipbetweenMOEandMORweredonebetweenoldcropandyoung19yearoldradiata
pine from New Zealand. Flatwise stiffness values (MOEp) were related to MOR as well as MOEedge
valuesforthetwosamples.TheR2valuesfoundareshowninTable3.Theauthorconcludesthatthe
relationship between MOR and MOE for radiata pine in New Zealand has deteriorated to such an
extentasaresultofshorterrotation agesthat itmakesmachine stressgradingbasedonstiffness
often unreliable and inefficient in terms of grade recoveries. A weakness of this study is that the
differencesbetweenoldcropand19yearoldradiataresultscannotbeascribedsolelytothecrop
agedifferences,asthe19yearoldcropwastestedatthepositionoflowestMOE,whiletheoldcrop
wastestedwithrandomdefectplacement.
Table3:ComparisonofoldandnewcropR2valuesforNewZealandradiatapine(Gaunt,1999)
Oldcrop
19yearoldcrop
MOEpvs.MOR
52.4
15.03
MOEpvs.MOEedge
68.37
47.87
Ishengoma et al. (1995) found that the basic density at different heights of the juvenile wood of
Pinus patula varies between 353376 kg/m3 and for mature wood between 433495 kg/m3. The
meanMOEofclearsectionsofjuvenilewoodwas4793MPaversus8939MPaformaturewood.
Burdzik(2004)testedstructuraltimberfromfoursawmillsinlowdensityregionsinSouthAfrica,
and found that only one sawmills timber made the grade requirements for bending and tensile
strength.Noneofthesawmillsgradedtimbermadetherequirementformeanmodulusofelasticity
(MOE).Itwasnotmentionedwhetheryoungtreeresourceswereusedatthesesawmills.
MethodsandMaterials
Thelogsusedforthisstudywereobtainedfrom17differentcompartmentsof1620yearoldPinus
patulatreesintheMpumalangaescarpment,SouthAfrica(seeTable4).Thecompartmentsvaried
between810and1930mabovesealevel,hadameanannualrainfallofbetween840and1640mm
and a mean annual temperature of between 13.7 and 19.4C. The sample consisted of 170 trees
(tentreespercompartment)whichwereprocessedfirstlyinto340logsandfinallyinto1402boards.
Two 2.1mlong logs were removed from each tree one from the pruned section of the stem at
2.3mheight andone from the unpruned section at7mheight. The logs were processed ata local
sawmillusingframesaws and a cantsawingpattern (Figure1). Only boards from thecant section
were used for this study. In practice this is often the case in South Africa where the side or wing
boards are sawn to industrial grade dimensions and only the cant contains structural grade
dimensions. The secondary breakdown saw used a curvesawing device so that the grain of the
boardswaspredominantlyparalleltothelongitudinaldirection.Theboardswerekilndriedusinga
mediumtemperatureschedule.
33
Table4.Generaldata
aforeachcom
mpartmentandthemeand
diameteratbreeastheight(D
DBH)andheig
ghtofthe
mpartment.
tensampleetreespercom
ComparttmentID
A(E66)
B(E28a)
C(G21)
D(D1)
E(D88)
F(E55a)
G(E36c)
H(E22)
I(E5)
J(E15)
K(E35)
L(C22)
M(E3)
N(D74)
O(A1a)
P(D11)
R(J20)
Mean
Plantation
A
Age
(yyrs)
Mean
DBH
(cm)
Nelshoogte
Nelshoogte
Nelshoogte
Uittsoek
Uittsoek
Uittsoek
Uittsoek
Uittsoek
Berrlin
Berrlin
Berrlin
Blyyde
Mo
orgenzon
Mo
orgenzon
Mo
orgenzon
Wilgeboom
Wilgeboom
17
19
16
17
17
20
19
17
19
19
16
20
17
19
16
18
19
18
36
33.8
26.2
32.7
30.2
32.3
31.9
27.6
36.5
37.4
29.1
34
31.4
26.9
27.8
29.4
33.4
31.6
Mean
height
(m)
20.9
21.8
18.4
22.3
18.4
20.1
23.0
20.8
23.8
23.8
18.0
27.0
20.6
16.4
19.0
22.8
24.0
21.2
SiteIndex10
(m)
14.3
14.5
15.5
15.7
15.3
14.6
16.8
16.5
16.7
17.6
16.5
18.5
13.5
9.6
13.6
19.6
16.8
15.6
Mean
annual
precipitation
(mm)
106
61
103
36
105
57
94
44
94
42
115
51
90
02
84
40
128
84
108
82
100
06
115
56
101
15
99
97
86
62
124
42
129
99
105
52
Mean
n
annuaal
temperatture
(C)
16.0
16.1
16.1
17.4
17.3
13.7
14.0
14.2
16.1
15.9
17.2
16.1
14.3
16.2
15.1
19.4
18.5
16.1
Figure1:Theecantsawingpatternused
d
Xraydeensityscanniingandknottparameters
After drrying, the boards
b
weree scanned with
w
a comm
mercial Gold
deneye702 Xray scann
ner from
Microtec.Figure2sshowsanexaampleofasscannedboard.TheGold
deneyescannerhasanu
umberof
pticalcamerras, a laser scanning
s
devvice andan Xray
X
densityy scanning device
d
to
sensors includingop
ofknotparametersinclu
udingknotlo
ocationandq
quality,boarrddimension
ns,board
measureeanumbero
warp, p
presence of pith, density, cracks, resin pockeets and barkk. Only kno
ot location and size
measureementsfrom
mthescanneranddensityvariationp
perboardweereusedforfurtheranallysis.The
density calibration of this macchine was not done as specified by the manu
ufacturer and it was
34
decided tousethe manuallydeeterminedab

bsolute denssityvalues.TThevariation
nindensity withina
oard(Xdev)measuredb
bythescanne
erwasusedforfurtheraanalysis.
singlebo
F
Figure2:Exam
mpleofscanneedboardshow
wingopticala
andXrayview
ws
mthexraysccanner,theffollowingkno
otparameteerswere
Byusinggtheknotinfformationgaatheredfrom
determined:
isthenu
umberofkno
otsonasingglepieceoftimber;
#Knots
isthem
meanarea(m
mm2)covered
dbyindividuaalknotsonaa
Knotaveerage(knotaavg.)
particularpieceofsawntimber.Xrayscanningevaluateedeach
nwhenaboardislayingflaton
boardfrromthevertticaldirection
itswidth
hface.Theaareameasure
ementissim
milartothekn
not
covered
dareaonew
willseeifyoulookfromaboveataknotina
transparentboard.TTakenoteth
hatthisisdiffferenttotheeknot
esearchstudies;
arearattio(KAR)ofteenusedinre
Knotareea/board
istheto
otalarea(asdetectedbytheXraysccanner)ofthepiece
oftimbe
erthatcontaainsknotmaterial(mm2);
Knotmaaximum(kno
otmax)
isthearrea(asdetecctedbytheX
Xrayscanner)coveredbyythe
biggestknotonapieceoftimbe
er(mm2);
KSC
meterdevelo
opedforthissstudy.Itistthe
isacalcculatedparam
producttofknotsizeeandthestre
essindexofaspecificknot.The
stressin
ndexisaratioofthebendingstressaataspecifickknotto
themaxximumbendingstressinapieceoftimber.Them
maximum
bendinggstresswillb
bePL/bh2witththetestseetupused(see
Figure3
3fordefinitio
onsofvariab
bles).Thestressindexwillbe
higheraattheupperandloweredgesofabo
oardandlenggthwise
intheceentrethird(mm2);
KParf,KP
Parc
aretwoknotparameterscalculaatedbytheM
Microtecscanner
onsandcalcu
ulationsforttheseparam
meters
software.Explanatio
otavailablettous.
wereno
35
Figure3
3:Thestressd
distributioniinaboardw
wasusedtoca
alculateastrressindexattaspecifickn
not.
Annualringmeasurements
Theend
dsoftheboaardsweresaandedtopro
ovideclearlyyvisibleannualrings.Th
henumber ofannual
o
ringson
neachboard
dwas counteedand num
mbered from thepith outwards.Thecambial agee(mean,
maximumandminim
mum)oftheewoodwithineachboardwasthereeforeknown
n.Thecambialageis
ntfromthep
pith.Alined
drawnfromtthepithtoth
hebarkandperpendiculartothe
theagebyringcoun
usingadialccalliper.The linewas
growth rings(Figuree4)wasuseedtodoringgwidthmeassurementsu
hboard.Meaasurementsw
wereroundeedtothe
drawntoincludetheemaximum numberofrringsineach
m,andmean
nringwidthw
wascalculateed.
nearest0.1mm.Forreachboardtheminimum,maximum
Figure4:Annualringwidthmeassurementsta
akenperpend
diculartoann
nualrings
Acousticcresonancefrequency
The aco
oustic reson
nance frequeency of eacch board was measured using thee AGrader Portable
softwaree from Falco
on Engineering. The boaards were placed on two
o polystyrene supports, 450 mm
from eaach end, to provide cleaarance betw
ween the boaards and thee table (Figu
ure 5), as well
w as to
prevent frequency channelling through thee table. A hit was then made with a wooden hammer
d. The sound
d caused by the resonan
nce of the wave
w
was
which seends a vibraation through the board
recorded
d with a microphone on the oppossite end of the
t board. Each
E
board was tested twice to
ensure tthe correct reading. During testing a 5% triggeer level was used with a
a 5512 Hz frrequency
range. The
T dynamicc MOE was then
t
determ
mined from the frequenccy and the density,
d
measured by
useofm
massandvolu
ume,withth
hefollowingfformula:
MOEdyn=(2lf)2where:
w
isthedynamicmod
dulusofelastticity,inMPaa;
MOEdyn
d
of th
he test speciimen at the moisture co
ontent at th
he time of teesting, in
is the density
kg/m3;
l
engthofthetestspecimeeninmeterstotheclose
estmm;and
isthele
f
isthefrrequencyoftthetestspeccimen,inHertz.
36
Figure5:FFrequencytesstingsetup
nicalgrading
Mechan
Staticflaatwisemech
hanicalgradingonalltheesamplesweereperformedusingaTTRUgraderaaccording
toSANSS10149(2002
2)seeFigu
ure6.Them
modulusofellasticitycalculatedfrom thisgradinggprocess
was refeerred to as MOEflat. During the mecchanical grad
ding there were
w
about 14
1 boards th
hat failed
underth
hetestload.Theseboard
dscouldnotbeusedforfurthertesting.
Figure6:Flatwisem
mechanicalgra
adingoftestm
material
Visualgrading
Visual ggrading of all of the bo
oards was do
one by a SA
ANAS (South
h African Naational Accreeditation
System)accredited professionalvisualgradeerfromthe independentcompanySSATAS(South
hAfrican
AuditingSerrvices).Theb
boardswereegradedaccordingtoSA
ANS17832((2005)visualgrading
TimberA
rules.Fo
oranalyticalpurposestheeboardswereallocatedagradeconsideringalltthedefects(aswould
be the ccase in a mill setup). The reason for downgradee of each bo
oard was alsso noted i.e. density,
warp(tw
wist,springaandbow)an
nddefects(knots,mechaanicaldamaggeetc.).Duriingvisualgraadingfor
this study, a separaate grade allocation waas documen
nted for eacch board in 3 categoriees. These
a
to density, kn
nots (sizes and distributtion) and
categoriies included the grades allocated according
otherdeefectsincludingwane,warp,resinan
ndmechanicaaldamage.
Warp,m
moisturecon
ntentanddensitymeasu
urements
Warp(tw
wist,bow,spring)ofeacchboardwaasmanually measuredacccordingto SANS17831
1(2004).
Themoiistureconten
ntwasmeassuredwitha resistancem
moistureme
eter.Density wascalculattedfrom
theweigghtanddimensionaldattaandwherrerequiredinthedataaanalysisproccessitwasccorrected
37
formoisturecontent.Notethatthewarpmeasurementwasnotpartofthevisualgradingprocess
andwasperformedbyadifferentoperator.
Destructivetests
The sample material was divided into two groups based on the board position and a random
allocationfunction.Pithboardsweremarked0,thetwoboardsoneachsideofthepithboards1,
thetwoboardsfurtherremovedfromthepithboards2etc.Arandomfunctionwasusedtoallocate
theboardsfromthesamepositioninalog(i.e.thetwo1boards)intothetwodifferentgroups.One
group was tested in bending and the other in tension. During the sawmilling and drying process
someboardswerelostandafewboardswerebrokenduringmechanicalgradingsothatatotalof
1345boardsoutof1402boardsweredestructivelytested.
Bending and tension tests were performed according to SANS 6122 (1994) with a few exceptions
whichwillbeexplainedbelow.Duetotestequipmentlimitations,thebendingtestspanwas1950
mmwhichisslightlylessthantherequired18:1spantodepthratio.Thetestswereperformedata
rateofdeformationof14mm/minwhichwashigherthanprescribed.ResearchfromMadsen(1992)
suggeststhatthetestingspeedfortimberonlystartstoplayarolewithmaterialstrongerthan35
MPa, which is stronger than normal commercial timber and far higher than the characteristic
strength of any South African pine grade. Of the total bending sample of 699 only 683 tests were
completed due to samples that twisted out of the test setup. Another 9 sample results were
discarded due to numbering errors leaving a total sample size of 674. Eight of the boards tested
were of slightly shorter lengths than 1950 mm, and the calculations of the strength and stiffness
propertieswereadjustedaccordingly.
Duetotestmachinelimitations,specimensforthetensiletestshadtobecuttoalengthof1060mm
tofitintothetestingsetupwithaspanof880mmwhichislowerthanthe9xdepthprescriptionin
SANS 6122 (1994). The test load was limited to 100 kN which is the maximum value that can be
testedwiththeloadcell.Atestspeedof15mm/minwasused,whichisverymoderatecomparedto
other literature (Madsen, 1992; Rayda, 2003) but higher than that required by SANS 6122 (1994).
Thetensiontestsonthestrongerportionofthetimberprovedproblematicasthesurfaceareaof
the machine grips in contact with the timber was too small. This caused the strong test pieces to
breakprematurelyatthegrips.Themaximumloadcapacityoftheloadcell,whichwas100kN,was
alsonothighenoughtotestthestrongestpiecestofailure.Thetensiontestresultsmight,therefore,
be somewhat lower for some of the stronger pieces of timber than what it would have been if a
better test setup was possible. Of the total tension sample of 623 only 611 tests were completed
duetosampleswhichslipped.
Bending strength, tension strength, MOE and 5th percentile values were calculated according to
SANS6122(1994).Themodulusofelasticitycalculatedfromthedestructivetestswasreferredtoas
MOEedge.
Statisticalevaluation
For single correlation analysis, simple Pearson correlations were performed on all the variables in
order to achieve a correlation matrix which includes all the possible indicator properties. Cooks
distance was used to determine some outliers from the testing process. Simple scatterplots were
drawnupforselectedvariablesusingaLeastSquaresapproachwithalinearmodelfitted(Y=a+
bX).
Whenperformingmultipleregressions,asummaryofbestsubsets,MallowsCp(Mallows,1973)and
forward stepwise regression was done in order to quantify colinearity between the different
variables and to decide which variables were significant enough to add to the multiple regression
38
statistics.TheMallowsCpnumberassessesthefitofaregressionmodelandattempttoaddressthe
issueofoverfitting.
Resultsanddiscussion
The results for the nondestructively evaluated (NDE) properties from the timber in this study are
depicted in Table 5. The table also shows the yield percentage of structural grades if grading was
basedononlythespecificpropertyevaluated.
Table5:NDEpropertiesofthe1345sawn38x114x2100mmPinuspatulaboards
Gradeyieldaccordingtospecific
property(%)
XXX
S5
S7
S10
Mean
Standard
deviation
Maximum
Minimum
Density*(m)
(kg/m3)
428
41.2
640
332
2.23
49.4
36.8
11.6
MOEflat(MPa)
5320
1934
14490
1961
49.6
35.8
12.6
2.4
MOEdyn(MPa)
8732
2295
16937
4251
Annualring
width(mm)
10.3
4.73
41.8
0.60
Bow(mm)
2.24
2.6
45.0
0.24
99.8
Spring(mm)
2.62
2.11
23.0
0.16
99.8
Twist(mm)
13.3
8.13
45.0
56.9
43.1
Moisture
content(%)
8.72
0.96
12.0
4.00
Property
*Densitywasdeterminedatthekilndriedmoisturecontentof
eachboardandcorrectedasperSANS17831(2004)
Densityhasalwaysbeenconsideredanimportanttimberpropertyinthepredictionofsometimber
qualityandstrengthproperties.Basedondensityalone2.23%oftheboardswererejected,49.4%
fellintotheS5class(360kg/m3),36.8%intotheS7class(425kg/m3)and11.6%intotheS10class
(475 kg/m3). The fact that very few boards were rejected for structural use due to density is
probablybecausethisresourcewouldalreadyhavebeenimprovedthroughtreebreedingselection
based on density. Density was still an important grade determining property for this resource,
allowing less than half of the timber to be in the S7 and S10 grades according to the current
structuralgradingrules.
The absolute values for MOEdyn, which is the dynamic modulus of elasticity calculated from the
naturalfrequencyanddensity,wasnotreallyimportantonitsownbutwasusedaspredictorofthe
other strength and stiffness properties. It was, however, noted that the values were higher than
MOEflat, and MOEedge. MOEdyn is an elastic property influenced by the total material makeup of a
board, while both MOEflat and MOEedge are primarily influenced by localised weak points in areas
wherestressesarehighestduringabendingtest.
Thevaluesforbow,springandtwistareusedinstructuralgradingduetothepracticalrequirements
forrelativelystraighttimberwhenmanufacturingproductsfromit.Itdoesnotdirectlyinfluencethe
timberstrengthandstiffnessproperties.Itcanbenoticedthatthevaluesforbowandspringwere
muchlowerthantheallowablebowof21mmandallowablespringof15mm(SANS17832,2005).
39
However,themeantwistwas13.36mmwhichwasabovetheallowabletwistof10mmaccordingto
SANS17832(2005).Basedontwistalone,56.9%oftheboardswererejected.
Destructivetestresults
ResultsfromthedestructivetestsaccordingtotheSANS1783(2005)visualandSANS10149(2002)
mechanicalgradeallocationoftimber(includingwarpandothernonstructuralrequirementsofthe
gradingrules)areshowninTable6.ThetablealsocontainsthecharacteristicstressvaluesforSouth
Africanpine(SANS101631,2003).
40

Table6:Timberstrengthandstiffnesspropertiesdeterminedbymeansofdestructivetests.VisualgradinginaccordancetoSANS1783(2005)andmechanicalgradingin
accordancetoSANS10149(2002).The5thpercentileMOEvalueswereobtainedfromthedraftSANS101632document.
Visualgrading
Property
MOR
Tension
SANS
visual
grade
Mechanicalgrading
Mean
5th
percentile
Yield
Yieldexcl.
warped
timber
(MPa)
(MPa)
(%)
(%)
Mean
5th
percentile
Yield
Yieldexcl.
warped
timber
(MPa)
(MPa)
(%)
(%)
Grade
potential
(%)
SANS101631
requirement
All
674
30.1
15.1
674
30.1
15.1
XXX
345
28.0
14.7
51.2%
17.1%
455
26.9
13.9
67.5%
50.0%
1.0%
S5
241
28.8
14.5
35.8%
59.2%
167
33.6
17.9
24.8%
35.2%
5.2%
11.5(5th)
S7
67
39.8
17.7
9.9%
17.2%
44
46.3
17.7
6.5%
12.5%
25.0%
15.8(5th)
S10
21
49.5
27.2
3.1%
6.5%
51.5
1.2%
2.4%
68.8%
23.3(5th)
All
605
12
7.8
611
12.0
7.8
XXX
286
10.9
6.6
47.3%
16.8%
392
10.5
6.9
64.2%
48.8%
2.9%
S5
228
11.3
7.2
37.7%
59.5%
155
13.1
7.9
25.4%
33.1%
43.7%
6.7(5th)
S7
67
16.5
9.7
11.0%
16.6%
58
17.7
10.1
9.5%
15.1%
52.9%
10(5th)
S10
24
18.4
11.6
4.0%
7.1%
20.0
1.0%
3.1%
0.5%
13.3(5th)
All
674
5888
3835
674
5888
3835
XXX
345
5566
3795
51.2%
17.1%
455
5451
3713
67.5%
50.0%
24.4%
S5
241
5750
3786
35.8%
59.2%
167
6419
4415
24.8%
35.2%
27.1%
4630
(5th)
7800
(mean)
S7
67
7381
5126
9.9%
17.2%
44
7935
5052
6.5%
12.5%
28.4%
5700
(5th)
9600
(mean)
S10
21
8081
5031
3.1%
6.5%
8732
1.2%
2.4%
20.1%
7130
(5th)
12000
(mean)
MOEedge
41
Table7a:Pearsonscorrelationcoefficient(r)betweendestructiveandnondestructivelytestedvalues.Shadedareashasnosignificantcorrelation(p>0.05).
Coefficientofvariation
MOEedge
MOEedge
MOR
Tension
MOEdyn
MOEflat
Ringmax
Ringmin
Ringmean
MOR
0.695
Tension
MOEdyn
0.797
0.701
0.800
MOEflat
0.760
0.659
0.696
0.866
m(Densitymanual)
0.593
0.404
0.580
0.668
0.599
s(Densityscanner)
0.419
0.371
0.519
0.573
0.536
0.880
Ringmax(Maximumringwidth)
0.502
0.402
0.480
0.601
0.490
0.398
0.315
Ringmin(Minimumringwidth)
0.418
0.337
0.512
0.513
0.413
0.441
0.385
0.617
Ringmean(Meanringwidth)
0.523
0.429
0.562
0.653
0.537
0.486
0.414
0.854
0.906
0.443
0.333
0.474
0.524
0.502
0.247
0.199
0.513
0.328
0.479
Xdev
0.320
0.181
0.373
0.340
0.374
0.189
0.275
0.340
0.168
0.297
Knotavg(Avg.knotarea)
0.318
0.368
0.378
0.345
0.338
0.210
0.195
0.190
0.136
0.200
Knotmax(Max.knotarea)
0.438
0.428
0.486
0.460
0.437
0.187
0.207
0.338
0.211
0.324
KnotArea/Board
0.507
0.454
0.516
0.570
0.508
0.238
0.227
0.418
0.258
0.393
#Knots
0.304
0.147
0.258
0.294
0.250
0.040
0.053
0.296
0.150
0.244
KSC(Knotstresscalculation)
0.293
0.420
0.272
0.256
0.248
0.069
0.081
0.073
0.016
0.050
KParf(Knotparameterfullboard)
0.373
0.391
0.371
0.391
0.390
0.062
0.032
0.268
0.194
0.275
KParc(Knotparametercentrethird)
0.326
0.367
0.296
0.316
0.354
0.069
0.038
0.243
0.143
0.232
Boardposition
42
Table7b:Pearsonscorrelationcoefficient(r)betweendestructiveandnondestructivelytestedvalues.Shadedareashasnosignificantcorrelation(p>0.05)
Board
KnotArea/
Coefficientofvariation
position
Xdev
Knotavg
Knotmax
Board
#Knots
KSC
KParf
KParc
Xdev
0.592
1
Knotavg(avg.knotarea)
0.191
0.019
Knotmax(Max.knotarea)
0.461
0.421
0.587
KnotArea/Board
0.562
0.401
0.519
0.700
#Knots
0.485
0.577
0.380
0.169
0.477
KSC(Knotstresscalculation)
KParf(Knotparameterfull
board)
KParc(Knotparametercentre
third)
0.159
0.025
0.273
0.242
0.397
0.122
0.381
0.095
0.427
0.508
0.488
0.102
0.205
0.274
0.121
0.360
0.412
0.413
0.112
0.291
0.715
43
Bendingstrength(MOR)results
Thegradepotentialcolumnshowsthat93.8%ofallthetimberpiecescouldpotentiallybeGradeS7
or S10 based on MOR values alone. The 5th percentile values for both visual and mechanical
structuralgradesweremuchhigherthanrequiredbySANS101631(2003).Evenifthefullungraded
sample was considered, the 5th percentile value was still 15.1 MPa which is fairly close to the
requiredvalueforgradeS7timber.ItisclearthatthisyoungPinuspatularesourcehasverygood
bending strength properties. Downgrading to the XXX grade was mainly due to warp and the 5th
percentile values for the XXX grade of both mechanical and visual grading were well above that
requiredforGradeS5.
The5thpercentilevalueforbendingstrengthisusuallystronglyinfluencedbyknotrelatedproperties
suchasthelocalgraindeviationaroundaknot.Forayoungresourcethebranchesinatreewillbe
smaller than for mature trees which will also result in smaller knots. It is hypothesized that the
bending strength of the mature resource used to establish the grade properties for South African
pine might have been worse because of the larger knots that would have been present in such a
resourceandsubsequentlargergraindeviationsaroundtheseknots.Eventhoughthevisualgrading
processwilllimitthesizeofknotsallowedinaboardthegraindeviationcausedbysuchaknotmight
onoccasionstillbepresentinaboardevenwhenonlyasectionoftheknotgoesthroughtheboard.
Tensionstrengthresults
Mostoftheboardsthatweretensiontestedfailedatthegrips,whichwouldnegativelyaffectthe
strengthvalues.Normallythesevalueswouldhavebeendiscarded,butsincegripfailuresincluded
most of our sample it was decided that they should be included. The tension grips made use of a
wedgeeffecttoincreasetheclampingforceatthegripsasthetensionloadincreased.Thewoodat
thegripscompressedandfailurewasusuallyduetoacombinationofthetensionstressandthegrip
compression.Eventhoughtheabsolutefailurevaluesofmostofthesampleswouldhavebeenlower
thanthosewhichcouldhavebeenachievedwithnogripfailures,thetestresultsstillprovideduseful
information.Mostimportantly,withnogripfailures,valueswouldhavebeenhighersotheresults
providedaconservativepictureoftensionstrength.Thetensionvaluesshould,therefore,inreality
bebetterforthehighergrades,sincethiswaswheregripfailuresandupperloadlimitationsmostly
occurred.Forvisualgradingthe5thpercentilevaluesforGradesS7andS10werebelowthatrequired
from SANS 101631 (2003). This would most probably be due to the test setup limitations which
often caused grip failures for higher strength material. The 5th percentile value for Grade S7
mechanicallygradedtimberwashigherthanrequired.Despitethelimitationsofthetestingmethod,
thegradepotentialcolumnshowsthat97.1%ofallthesamplesfellwithineithergradeS5,S7orS10.
Althoughonecannotdrawfirmconclusionsfromtheseresultsduetotestsetuplimitations,itseems
asifthetensionstrengthofthisresourcemightbeadequate.
Modulusofelasticity
ThemeanMOEedgeforthefullsample(5888MPa)wasabout25%lowerthanrequiredforGradeS5
timber.Table6showsthatthemeanMOEedgevaluesforallthegradeswerewellbelowthatrequired
bySANS101631forbothvisualandmechanicalgrades.Althoughthegradepotentialcolumnshows
thatonly24.4%ofthetimberwouldhavebeenrejectedbasedonMOEedgeitshouldbenotedthat
for this column only the 5th percentile MOEedge requirement listed in the current draft version of
SANS 101631 were considered and not the mean MOEedge requirement per grade (which is not
possibleinthiscontext).FromtheseresultsitisclearthatthisyoungPinuspatularesourcehadvery
poorstiffnessproperties.
Mechanicalgradingwas,forobviousreasons,betteratseparatingtimberaccordingtostiffness.The
mean MOEedge of mechanically graded rejects (XXX) was lower than that of visually graded timber
44
whereas for the structural grades it was higher. The variation of MOEedge values is important for
structural grading of timber since there is a mean MOEedge requirement in SANS 101631 (2003)
unlikestrengthpropertieswhereonlythe5thpercentilevalueisusedtodescribeaproperty.Itcan
bededucedthatthisyoungPinuspatularesourcehadmuchlowerstiffnessthantheresourcewhich
was used to develop the visual and mechanical grading rules for South African pine in the past.
Hencethecurrentgradingruleswerenoteffectiveinmeetingtherequirementsforstiffnessofthe
various structural grades. The reason for the low stiffness was most probably the larger average
microfibrilanglesassociatedwithjuvenilewood.
Visualandmechanicalgradeyields
Thegradeyieldofthisresourceaccordingtovisualgradingstandardswasquitepoorwithroughly
onlyhalfthetimbersuitableforstructuralgradesinboththebendingandtensionsample.Mostof
the downgrading with visual grading was due to the high twist values in the timber (Note: the
measurements for the warp values and grade recoveries in the previous Table 5 were performed
separately from the visual grading by a different operator therefore the fact that according to
Table 5 there was 56.9% of the timber rejected when only twist was taken into account whereas
Table6showsthatthetotalrejectyieldinboththebendingandtensionsamplewerelessthanthat
value). When warp was not taken into account in visual grading (in the Yield excluding warped
timbercolumn),only17.1%ofthebendingsampleand16.8%ofthetensionsamplewererejected.
Better drying practices can lower warp in timber, and, with modern finger jointing operations the
problem of twist can be reduced by simply crosscutting twisted pieces and joining them again
althoughthere are some recoverylosses andsignificant extracosts involved insuchanoperation.
Withmechanicalgradingtheyieldsweresignificantlypoorerthanwithvisualgradingand67.5%of
thetimberwasnotsuitableforstructuralpurposes.Evenwhenwarpwasnottakenintoaccount,the
yieldwasstillonly50%forstructuralgrades.
Aconcernwiththistimberresourceisthatdespitethelowgradeyieldsformechanicalgradingitstill
did not fulfil the current MOEedge requirements for structural South African pine (SANS 101631,
2003).Inpracticeit meansthatif bordervalues of indicating properties for mechanical andvisual
gradingareincreasedsothatMOEedgerequirementsaremet,thegradeyieldswillbeevenlower.At
the same time the bending strength 5th percentile values for structural grades were much higher
than required. This indicates that the relationship between various properties such as the ratio of
MOR:MOEforthisresourceisverydifferenttowhatitwaswhenthecharacteristicvaluesforSouth
Africanpinewasdeterminedthreedecadesago.
Defectplacementwithingradetesting
ThecurrentSouthAfricaningradetestingstandard(SANS6122,1994)prescribebiasedplacement
oftheworstdefectindestructivetestingofasample.The2.1mloglengthseffectivelymeantthat
therewaslittleoptionwithdefectplacementandthatdefectplacementcanbeviewedasrandom
placementoftheworstdefectforbendingtests.Fortensiletestsampleswithatestingspanof880
mmthetestresultscanbeviewedaslimitedbiasedtestingsincethe2.1mboardlengthwasmuch
shorter than is normal for South African structural products. The Australian and New Zealand in
gradetestingstandards(AS/NZS4063,1992)andtheISO13910(2005)standardprescriberandom
placement of defects. Madsen (1992) is of the view that random placement of the sample is the
more correct method for ingrade testing since that is what occurs during construction of timber
structures the worst defect is not always, and never deliberately, placed in the highest stressed
area. The advantage of random placement of defects is that higher 5th percentile MOR values are
obtained. A change to the current SANS 6122 standard should be considered although other
factorsshouldalsobetakenintoaccountsuchastherequiredsamplesizeswhichmightbehigher
forrandomplacementofdefects.
45
Correlationsbetweenmeasuredproperties
Allcorrelationcoefficients(r)betweenthemeasuredpropertiesareshowninTables6aandb.
MOEedgevs.MOR
Oneofthemostwidelyusedrelationshipsintimbergradingisthatbetweenstiffnessandbending
strength. Gaunt (1999) found that younger P. radiata material has a much lower MOE vs. MOR
correlationthanoldermaterial.Alowcorrelationmakesthedeterminationofefficientgradingrules
more difficult, as timber with high stiffness does not necessarily have a high MOR. However, the
coefficientofdeterminationobtainedinthisstudy(R2 =48.2%)waswithintheexpectedrange(see
Table2andFigure7).
MOR : MOEedge
MOR = -2.3337+0.0055*x
80
MOE Edge:MOR (Mpa): y = -2.3337 + 0.0055*x;

r = 0.6949, p = 0.0000; r2 = 0.4828
70
60
MOR (MPa)
50
40
30
20
10
0
0
2000
4000
6000
8000
10000
12000
14000
MOEedge (MPa)
Figure7:ScatterplotofMORagainstMOEedge
Densityvs.MOR,MOEedgeandtension
ThecorrelationsofMOEedge,MORandtensionwithdensity(Table7a)werecomparablewithresults
foundinotherstudieswhereR2valuesofbetween2030%wereobtained(Glos,2004).
MOEedgevs.MOEflat
A surprising result was the low correlation between MOEedge and MOEflat (r = 0.76) from Table 7a.
The low correlation might be explained by the fact that the MOEflat value was determined at the
weakestpointofthetimberandonflat,probablyincludingabigknotpercentage,asopposedtothe
randomtestsetupforthedeterminationofMOEedge.
46
MOEdynvs.MOEedge/MOR/Tension
A very encouraging NDE property measured was that of MOEdyn. MOEdyn was the best single
predictor for MOEedge, MOR and tension (Table 7a). The correlation values were even better,
althoughnotbymuch,thanthecorrelationwithMOEflatwhichwascalculatedfromflatwisebending
ontheweakestpointofthesamples.ThehighercorrelationvaluesofMOEedgeandMORwithMOEdyn
comparedtoMOEflatshowspromisewhentakingintoconsiderationthatMOEdynmeasurementsare
probablyalowercostmethodathighproductionspeedscomparedtothemeasurementofMOEflat.
Otherproperties
The other variables that were deemed to be statistically influential were: annual ring widths (ring
max, ringmin and ring avg.),thepositionofthetimberpiece within thetree (position)andknots
(knotavg.knotmax,knotarea/board,KSC,KParfandKParc).Itwasnoticedthattheringaverage,
which is the mean width of the annual rings on each board, had a better correlation with both
MOEedgeandtensionthantheindividualknotparameters.Thiswasnotsuchasurprisingresultfor
MOEedge,astheknotsdonotplaysuchabigroleaswouldbeexpectedforMORandtension.Itwas,
however,surprisingthat thesingleknot parametersdidnotplaya bigger role inthe predictionof
tensionstrength.Thismighthavebeenduetothelimitationsinthetensiontestingdiscussedearlier.
ThecorrelationbetweenMORandringwidthaveragewasalsoonlyslightlylowerwhencompared
toMORvs.knotareaperboard(Tables6aandb).ThebestpredictorforMOEedge,MORandtension,
betweenthedifferentknotmeasurementstaken,wasthetotalknotareaperboard.Thisseemsto
beabetterpredictorthanthemaximumknotsizeinaboard.Theknotstresscalculationandknot
parameters, which was formulated by combining knot sizes and positions, did not show good
correlations.
Usingcombinationsofpropertiestopredictstrengthandstiffness
Multiple regression analysis was used to determine how well combinations of NDE properties can
predictthestrengthandstiffnessoftheyoungP.patulatimber.Acombinationofthebestsubsets,
MallowsCpandforwardstepwiseregressionmethodswereusedtoidentifypropertiestoincludein
predictionmodels. Table 8depicts the degreesofdeterminationbetween thedestructivelytested
properties and combinations of indicator properties. These were the indicator properties with the
highestinfluenceonthecharacteristicstrengthpropertiesoftimber.Bymeansofstatisticalanalysis
ofthecolinearityonlythepropertiesthatweredeterminedtobeinfluentialinthedestructivetest
results were displayed in the table. Note that the rvalues for single properties might be slightly
differenttothatinthecorrelationtable(Tables6aandb) duetohigher samplenumberspossible
whenonlyindividualcorrelationswereconsidered.
Table8:Coefficientsofdeterminationvalues(percentages)betweendestructivetestsandbothsingleand
combinedindicatorproperties.Pearsonscorrelationusedforsinglepredictorsandforwardstepwiseregression
formultiplepredictors
MOEedge
R2
MOR
R2
TENSION
R2
MOEdyn
63.57
46.55
62.93
MOEflat
57.82
47.32
53.69
Predictors
m(Densitymanual)
35.10
23.83
33.71
KParf,KSC,Knotmax
23.21
33.67
31.14
m,KParf,KSC
43.49
42.53
45.91
MOEdyn,m,KParf,KSC
62.59
54.26
64.38
Knot parameters, comprising of knot size and calculated parameter values (KParf, KSC, Knot max),
showedadegreeofdeterminationof33.67%forMOR.Thisisintheupperrangeofvaluesfoundby
47
Glos(2004).Variouscombinationsoftimberpropertieswerestatisticallyevaluatedandalsotested
forcolinearityasexplainedinthemethodsandmaterialssectionofthisreport.CombiningMOEdyn
with density and knots increased the predictability of MOR by almost 7% over the use of MOEdyn
alone.
The best prediction values for MOEedge, MOR and tension were obtained by combining knot
parameters,densityandMOEdyn.Statisticalanalysisshowedthatthedensitycontributesverylittleto
this prediction value and that the addition of the knot parameters to MOEdyn in the prediction of
MOR has a roughly 6% increase compared to MOEdyn on its own. MOEdyn is, however, the best
individualpredictorofMOEedge,MORandtension.
When comparing the R2 values in this study with that obtained by Glos (2004), it shows that the
predictabilityofMORusingcombinationsofpropertiesforthisresourceisquitelow.Glosmentions
R2valuesofbetween5580%andwithacombinationofpropertiesthebestR2valueobtainedwas
54.26%. The predictability of tension strength (despite the limitations in our test setup) was
relativelygoodat64.38%.
Conclusionsandrecommendations
TheyoungP.patulatimbertestedinthis studyhadgoodbendingstrength(MOR) propertieswith
higher characteristic grade values than required. The tension strength values were similar to the
requirementsasspecifiedintheSANS101631(2003)documentforallthestructuralgrades.There
were,however,somelimitationsinourtensiletestsetupwhichcausedtheresultstobelowerthan
it should have been if a correcttest setup was possible. The timber had low stiffness and did not
comply with the SANS 101631 (2003) requirements for mean MOEedge for any of the structural
grades.
Of all the individual nondestructive predictors of strength and stiffness, the dynamic modulus of
elasticity (MOEdyn) showed the best correlation with MOR, MOEedge and tension strength. When
combinations of NDE properties were used to predict stiffness and strength the degree of
determinationcouldbeincreasedinsomecasesoveronlyusingasinglepredictorsuchasMOEdynor
MOEflat.
It is recommended that ingrade testing programs be conducted at sawmills which use a

predominantly young Pinus patula resource. If not compliant to the current strength and stiffness
requirements,newgradingmethodssuchasacousticgradingshouldbeconsidered.
Acknowledgements
The following organisations and individuals are gratefully acknowledged for their help with this
study: KLF for supplying the trees for the project and funding some tasks (Francois Malan). York
Timber for the sawing, drying and Xray scanning of the timber (Deon Breytenbach, Leslie Glass,
Bruce Waller). Jos Louw and Barry Muller from NMMU, Saasveld for allowing us to use the test
materialandforcooperationinthisproject.CapeSawmillsfortheuseoftheirmechanicalgrader
and help with the setup of the test (Lewis Silberbauer). Professor Daan Nel for his advice on the
statistical analysis. Peter Muller for advice on structural grading. Martin Bacher from Microtec for
adviceandsupportwiththeXrayscanninganalysis.SawmillingSAforsponsoringthisproject.
48
References
AS4063/NZS.(1992).AustralianandNewZealandStandard.TimberstructuralproductsStrength
andstiffnessevaluation.
Burdzik,W.(2004).GradeverificationofSouthAfricanpine.SouthernAfricanForestryJournal202:
2127.
CrickmayandAssociates.(2009).SouthAfricanLumberIndex,Feb2009.
DAFF.2009.ReportoncommercialtimberresourcesandprimaryroundwoodprocessinginSouth
Africa2008/9.DepartmentofAgriculture,ForestryandFisheries,RSA.
Gaunt,D.J.(1999).Machinestressgradingrevisited.NZTimberDesignJournal1(8):1018.
Glos,P.,Heimeshoff,B.(1982).MglichkeitenundGrenzenderFestigkeitssortierungvon
BrettlamellenFrdenHolzleimbau.In:IngenieurholzbauinFoschungundPraxis(EhlbeckundSteck).
Bruderverlag,Karlsruhe.
Glos,P.(2004).Newgradingmethods.ProceedingsofCOSTE29Symposium,FlorenceOctober27
29.CNRIvalsa,SanMicheleallAdige,Italy,p18.
Hoffmeyer,P.(1984).Omkonstruktionstraesstyrkeochstyrkesortering.ISkovteknologi.Ethistoriskt
ogperspektiviskstrejtog.DanskSkovforening.
Hoffmeyer,P.1990.Failureofwoodasinfluencedbymoistureanddurationofload.DoctoralThesis,
StateUniversityofNewYork,CollegeofEnvironmentalScienceandForestry,Syracuse,NewYork.
IshengomaRC,GillahPR,IddiS.1995.Basicdensity,tracheidlengthandstrengthofjuvenileand
maturewoodofPinuspatulagrowninTanzania.SouthAfricanForestryJournal172:1923.
ISO 13910. (2005). Structural timber Characteristic values of strengthgraded timber Sampling,
fullsizetestingandevaluation.
Johansson,CJ.(1976)Tensilestrengthofglulamlaminations.ChalmersUniversityofTechnology,
SteelandTimberStructures,Internalreportno.S76:18.
Johansson,CJ,Brundin,J.,andGruberR.(1992).StressgradingofSwedishandGermantimber.A
comparisonofmachinestressgradingandthreevisualgradingsystems.SwedishNationalTesting
andResearchInstitute.SPReport1998:38.
Johansson,CJ.(2003).Gradingoftimberwithrespecttomechanicalproperties.In:Thelandersson,
S.,Larsen,H.J.(eds)(2003).TimberEngineering.JohnWiley&SonsLtd.445pp.
Kennedy,R.W.(1995).Coniferouswoodqualityinthefuture:concernsandstrategies.WoodSci.
Technol.29:321338
Knuffel,W.E.(1983).IngradetestingofSouthAfricanpinevisualgrades.Hout240,CSIR.
Knuffel, W. E. (1984). An improved visual and mechanical stress grading system for South African
pinetimber.Hout318,CSIR.
49
Lackner,R.andFoslie,M.(1988).GranfraVestlandenStyrkeochsortering.NorwegianInstituteof
WoodTechnology,Report74.
Madsen, B. (1992). Structural behaviour of timber. Timber Engineering Ltd, North Vancouver,
Canada.405pp.
Mallows,C.L.(1973).SomeCommentsonCp.Technometrics15(4):661675.
Rayda, R. R. (2003). Nondestructive evaluation of lumber from cull and suppressed growth trees.
Laramie,Wyoming:M.S.,DepartmentofCivilandArchitecturalEngineering.
SANS 10163 (2003). South African national standard. The structural use of timber Part 1: Limit
statesdesign.Edition2.3.
SANS 10149 (2002). South African national standard. The mechanical stress grading of softwood.
Edition1.2.
SANS 17831 (2004). South African national standard. Sawn softwood timber, Part 1: General
requirements.
SANS,17832(2005).SouthAfricannationalstandard.SawnsoftwoodtimberPart2:Stressgraded
SANS, 6122(1994).SouthAfricannational standard.Qualificationtestingofsolid structuraltimber

and laminated structural timber (glulam) for verifying timber grading systems in accordance to a
givenstandard.
Zobel,B.J.,Sprague,J.R.(1998).Juvenilewoodinforesttrees.Springer,Heidelberg,300pp.
50
Chapter4.
Variationinstrength,stiffnessandrelatedwoodpropertiesinyoungSouthAfricangrown
Pinuspatula
Inpress:AcceptedtobepublishedinSouthernForests(2014)76
C.B.Wesselsa*,F.S.Malanc,D.G.Nelb,T.Rypstraa
a
DepartmentofForestandWoodScience
PrivateBagX1,
Matieland7602,SouthAfrica
Tel:+27218083319
Fax:+27218083603
email:cbw@sun.ac.za
b
CentreforStatisticalConsultation
PrivateBagX1,
Matieland7602,SouthAfrica
c
KomatilandForests
POBox574,
Sabie1260,SouthAfrica
Abstract
Theobjectiveofthisstudywastoexaminethevariationinandintercorrelationamongwood
propertiesdeterminingthesuitabilityof1620yearsoldSouthAfricangrownPinuspatulatreesfor
structuraltimber.Atotalof1112sawnboardsfrom340logs,170treesand17different
compartmentswereexamined.Sawlogsweretakenfromtwoheightlevelsfromeachtree.The
meanmodulusofelasticitymeasuredonedge(MOEedge)wasfarbelow,andthemeantwisthigher
thanthelimitssetforstructuralgradesoftwoodtimberinSouthAfrica.Allthedesirableproperties
forstructuraltimberimprovedwithdistancefromthepithwiththeexceptionofthe5thpercentile
valueformodulusofrupture(MOR)whichwashigheratthepiththanfortheboardsprocessed
adjacenttothepith.Boardsprocessedfromthelowerpartofthestemweresuperiorinmostofthe
importantpropertiescomparedtothepropertieshigherupinthestem.Thecorrelationbetween
thedynamicmodulusofelasticity(MOEdyn)andMORofboardsprocessedfromthelogstakenhigher
upinthestemwasmuchweakerthaninthecaseoftheboardsprocessedfromthelogtakenfrom
thelowerpartinthestem,suggestingthatindirect(nondestructive)predictionofMORdecreased
inreliabilitywithincreaseinheightintrees.Arelativelystrongnegativecorrelationwasfound
betweenthemeangrowthringwidthsofthepithboardsandthemeanMOEdynvaluesper
compartment,suggestingthatslowerinitialgrowthinacompartmentresultedinincreasedmean
stiffnessofboardsfromthatcompartment.
Keywords
Pinuspatula,structuraltimber,modulusofelasticity,modulusofrupture,twist
51
Introduction
Pinuspatulaisthemostintensivelyutilisedconiferinthetropicsandsubtropics,whereitiswidely
plantedasanexotic.Itisestimatedthatmorethanonemillionhectaresareplantedtothisspecies
withabouthalfofthatinAfrica(Wright1994).InSouthAfricanitisthemostimportantcommercial
plantationsoftwoodresourcewithatotalof338923haplantedwithPinuspatulatrees(DAFF
2009).Asubstantialportionofthatareaismanagedforsawlogproduction.Ofthetotalsolidsawn
timberproductioninSouthAfricamorethan70%issoldasstructuralorbuildingtimber(Crickmay
andAssociates2011).Thistimberismainlyusedinrooftrussesandotherbuildingcomponents
wherethesafetyoftheinhabitants,buildersandmaintenanceworkersisatstake.Itfollowsthatthe
mechanicalpropertiesofPinuspatulatimberareofmajorimportancetothesawmillingindustryin
SouthAfrica.
Pineplantationsmanagedforsawlogproductionweretraditionallyharvestedatabout30yearsof
age.Inthelastfewyears,however,therotationageforsawtimberplantationsinSouthAfricahas
beenreducedbymanygrowersandtheaveragerotationagehasfallenfromabout30yearsin1994
tolessthan23yearsin2004(Crickmayetal.2005).Thereasonforthisreductioninrotationagecan
largelybeattributedtotheincreaseingrowthrateoftreesthroughtreebreedingandimproved
silviculturalpractices,tomeettheincreaseddemandforsawntimberandthegrowingneedto
enhancetheproductivityofourexistingforestryresource.Unfortunatelytheshorteningofrotation
agetoreapthefinancialbenefitsofthefastergrowth,hasadirectimpactonwoodqualityasit
causesanincreasedproportionofjuvenilewoodintherawmaterialenteringtheprocessing
industry.Juvenilewoodisknowntobestructurallydifferentandconsiderablymorevariablein
virtuallyallwoodpropertiescomparedtomaturewood,resultingalsoinhighervariabilityinits
performanceandbehaviouralcharacteristics(Kennedy1995;Malan2010;ZobelandSprague1998;
XuandWalker2004).
Forstructuraltimberproducersandusers,twoaspectsareimportant.Inthefirstinstancethe
absolutevaluesofthestrengthandstiffnesspropertiesofsawntimberfromaspecifictimber
resourceshouldbesuchthatanacceptablepercentageoftimbermeetstherequirementsforthe
variousstructuralgradesassetoutinSANS101631(2003).Secondly,theabilitytopredictthese
strengthpropertiesnondestructivelyisalsoveryimportantforaspecificwoodresourcesinceit
ensuresreliablestrengthandstiffnessproperties(aftergrading)forusersofstructuraltimber.
Burdzik(2004)performedmechanicaltestsonvisuallygradedtimberthatpassedthelowestSouth
Africanstructuralgrade(S5)fromfoursawmillsfromlowwooddensityareasinSouthAfrica.It
wasfoundthatnoneofthesawmillsproducedgradedtimberthatconformedtothestrengthor
stiffnessrequirementsofSANS101631(2003).Theresultthussuggestedthatinsomeareasin
SouthAfrica,therelationshipbetweenthevisualgradingcriteriaandstrengthandstiffness
propertiesofSouthAfricanpinewasnotthesameasinthepastwhenthecurrentlyappliedgrading
rulesweredeveloped.Burdzik(2004)providednoinformationonthepinespeciesandage
distributionofthelogsprocessedbythesesawmills,butspeculatedthatanincreasedproportionof
juvenilewoodmayhavebeenthemainreasonforthelowstrengthandstiffnessexperiencedby
thesemills.
Stiffness,orMOE,isoneofthebestandmostwidelyusednondestructivepredictorsoftimber
strength(Johansson,2003).InNewZealand,Gaunt(1999)investigatedtherelationshipbetween
MOEandbendingstrengthofjuvenileandmaturePinusradiata.Heconcludedthatthisrelationship
wassoweakforjuvenilePinusradiatathatmachinestressgradingusingthisrelationship,became
unreliableandinefficient.
52
Anunderstandingofthewithinandbetweentree,andbetweensitevariationinstrengthand
stiffnessofSouthAfricanpine,andtheabilitytoquantifyandpredictthesevariationpatterns,
especiallyinthecorewoodsectionoftrees,wouldprovidesawloggrowersandprocessorswitha
reliableinformationbaseforthedevelopmentofmuchneededprocessingdecisionsupporttools.It
canbeexpectedthatinSouthAfricatheneedforsuchtoolswillgrowinimportancewiththe
continuouschangesinthequalityoftheSouthAfricanpinetimberresource.
Theobjectiveofthisstudywastoexaminethewithinandbetweentreevariationinwood
propertiesofyoungSouthAfricangrownPinuspatulatreesknowntohaveimportantimpactsonthe
suitabilityofsawntimberforstructuralpurposes.Apartfromthestrengthandstiffnessvariation,
thevariationintherelationshipbetweenstrengthandstiffness,whichisimportantforefficient
structuralgrading,wasalsoinvestigated.Althoughwarpdoesnothaveadirecteffectonwood
strengthandstiffness,itwasneverthelessincludedinthestudyassawntimberneedstobe
relativelystraighttobeutilisableinastructure.
Materialsandmethods
Treesampling
Sawlogswereobtainedfrom17differentPinuspatulacompartmentsintheMpumalangaforest
areaofSouthAfrica.Thecompartmentsvariedinagefrom16to20years,altitudesfrom810to
1930mabovesealevel,meanannualrainfallfrom840to1299mmandameanannualtemperature
rangeof13.7to19.4C.Siteindicesatbaseyear10rangedfrom9.6to19.6(Table1).
Compartment
identification
A(E66)
B(E28a)
C(G21)
D(D1)
E(D88)
F(E55a)
G(E36c)
H(E22)
I(E5)
J(E15)
K(E35)
L(C22)
M(E3)
N(D74)
O(A1a)
P(D11)
R(J20)
Mean
Plantation
Age
(yrs)
Nelshoogte
Nelshoogte
Nelshoogte
Uitsoek
Uitsoek
Uitsoek
Uitsoek
Uitsoek
Berlin
Berlin
Berlin
Blyde
Morgenzon
Morgenzon
Morgenzon
Wilgeboom
Wilgeboom
17
19
16
17
17
20
19
17
19
19
16
20
17
19
16
18
19
18
Mean
DBH
(cm)
36
33.8
26.2
32.7
30.2
32.3
31.9
27.6
36.5
37.4
29.1
34
31.4
26.9
27.8
29.4
33.4
31.6
Mean
height
(m)
20.9
21.8
18.4
22.3
18.4
20.1
23.0
20.8
23.8
23.8
18.0
27.0
20.6
16.4
19.0
22.8
24.0
21.2
SiteIndex10
(m)
14.3
14.5
15.5
15.7
15.3
14.6
16.8
16.5
16.7
17.6
16.5
18.5
13.5
9.6
13.6
19.6
16.8
15.6
Mean
annual
precipitation
(mm)
1061
1036
1057
944
942
1151
902
840
1284
1082
1006
1156
1015
997
862
1242
1299
1052
Mean
annual
temperature
(C)
16.0
16.1
16.1
17.4
17.3
13.7
14.0
14.2
16.1
15.9
17.2
16.1
14.3
16.2
15.1
19.4
18.5
16.1
Table1Generaldataforeachcompartmentandthemeandiameteratbreastheight(DBH)andheightofthe
tensampletreespercompartment.
Astratifiedsamplingprocedureintermsoftreediameterswasfollowedsothatthesampletrees
representedtheproductivetimbervolumeavailablefromthecompartments.Ineachcompartment
onetreewasrandomlyselectedfromthefirstquartile(smalldiameter),twotreesfromthesecond
53
quartile,threetreesfromthethirdquartileandfourtreesfromthefourthquartile(largediameters),
givingatotaloftensampletreespercompartment,thus170treesfortheentireinvestigation.
Boardpreparation
Two2.1mlonglogswereremovedfromeachtreethebottomlogfromtheprunedsectionofthe
stemat2.3mthickendheightandthetoplogfromtheunprunedsectionat7mthickendheight
(Figure1).Thelogswereprocessedatalocalsawmillusingframesawsandacantsawingpattern.
Thecrosssectionaldimensionofalltheboardswas38x114mm.Onlyboardsfromthecantsection
wereusedforthisstudyasitprovidedtestmaterialthatrepresentedthefullradiusofeachlog.
Boardswereconsecutivelynumberedfromthepithtowardstheoutside,startingwith0toindicate
boardscontainingpithtissue,followedbythenumbers1and2(Figure1).Whentwoboardsfrom
thesamelogcontainedpithtissue(whichwasoftenthecase),bothweremarkedas0.
Boardswerekilndriedusingamediumtemperaturescheduletoatargetmoisturecontentof12%.
Bottom log
Toplog
Board2
Board1
Board0
Board0
Board1
Board2
0m
2.3m
Board2
Board1
Board0
Board1
Board2
4.4m
7m
9.1m
Fig.1Positionoflogsandboardssampledfromthetreestem.Notethatboardposition0signifiesboardsthat
containedpith.Sometimestherewasonlyonepithcontainingboardinalog(seetoplogabove)and
sometimestwopithcontainingboards(seebottomlogabove)
Boardevaluation
DriedboardswerescannedwithacommercialGoldeneye702XrayscannerfromMicrotec.This
scannerisfittedwithanumberofsensorsincludingopticalcameras,alaserscanningdeviceandan
Xraydensityscanningdevice.Forthepurposeofthisstudyonlyknotlocationandsizedatafromthe
scannerwereusedforfurtheranalyses.
Thefollowingknotcharacteristicswerecalculatedandanalysed:
Totalknotarea(mm2):Thetotalareaoftheboardthatcontainedknots.Xrayscanning
evaluatedeachboardfromtheverticaldirectionwhenaboardwaslayingflatonits
widthface.Theareameasurementissimilartotheknotcoveredareaseenwhenlooked
fromaboveataknotinatransparentboard.
Maximumknotarea(mm2):Theareacoveredbythelargestknotorcombinationof
knotsonapieceoftimber.
Individualboarddensitiesweredeterminedbyweighingandcalculatingthevolumeofeachboard
fromdimensionaldata,correctedformoisturecontentaccordingtoSANS17831(2004).The
moisturecontentwasmeasuredwitharesistancemoisturemeter.Warp(twist,bow,spring)ofeach
boardwasmeasuredaccordingtoSANS17831(2004).
Thenumberofannualringsoneachboardwascountedandnumberedfromthepithoutwards.The
cambialageoryearringnumber(mean,maximumandminimum)ofthewoodwithineachboard
wasthereforeknown.Thecambialageistheagebyringcountfromthepith.Ringwidth,fromstart
ofearlywoodtoendoflatewood,weremeasuredtothenearest0.1mm(Figure2).Foreachboard
theminimum,maximum,andmeanringwidthsweredetermined.
54
Fig.2Rad
dialringwidth
hmeasuremeentstakenperrpendicularto
oannualrings
oardwasmeeasuredusingtheAGrad
derPortable
Theacousticresonancefrequenccyofeachbo
ng.Twomeasurementsw
weretakenp
perboardtoensurecorreectness.
softwareefromFalconEngineerin
Duringttestinga5%ttriggerlevelwasusedwitha5512Hzfrequencyrange.Thed
dynamicMOEwas
determinedfromtheresonancefrequencyaandthedenssitywiththefollowingformula:
MO
OEdyn=(2llf)2
wheere:
micmodulusofelasticity,,inMPa;
MOEdyn isthedynam
isthedensityofthetestspecimenattthemoisturecontentatthetimeofftesting,inkkg/m3;
hofthetestsspecimeninmeterstoth
heclosestmm
m;and
isthelength
l
istheresonaancefrequen
ncyofthetestspecimen,,inHertz.
f
psbyrandom
mlyallocatinggthetwobo
oardsateach
hradial
Thetesttboardsweredividedinttotwogroup
positiontoadifferen
ntgroup.Boardsinoneggroupwereeearmarkedforbendingttestsandtheeboards
ortensionteestsparallelttograin.Whereonlyone
eboardwasaavailableforra
intheotthergroupfo
specificposition,itw
wasallocated
dtothebend
dingtestsam
mple.
ntestswereperformedaapproximatelyaccordingtothepresccriptionsoutlinedin
Bendinggandtension
SANS61
122(1994).D
Duetoequipmentlimitattionsthetesttspanusedfforcarryingo
outthebend
ding
testswaas1950mm,whichwassslightlyshortterthantherequired18:1spantodeepthratio.Thetests
werepeerformedataarateofdefo
ormationof14mm/min.
699boardsavvailableforb
bendingtests,683weresuccessfullytestedassomeofthebo
oards
Ofthe6
twistedduringtheteestprocess,renderingth
hemunsuitab
bleforfurtheertesting.Th
hetestresulttsof
ngerrors.
another9boardsweereeliminateedfromthedatasetduetonumberin
n,thespecim
mensfortenssiletestingh
hadtobecutttoalengthof1060
Duetottestequipmeentlimitation
mmtoffitintotheteestingmachin
ne.Thisresu
ultedinatesttspanof880
0mm,which
hisshorterth
hanthe
ntodepthraatioprescribedbySANS6
6122(1994)..Atestloadof100kN,w
whichisthe
9:1span
maximumloadcapacityofthelo
oadcell,andatestspeed
dof15mm/m
minwereuseed.
estrongerbo
oardsproved
dproblematiccassomeoffthestrongeerboardsbro
oke
Tensiontestsonthe
urelyattheggrips.Themaaximumload
dcapacityoftheloadcell(100kN)wasalsonoth
high
prematu
enoughtotestthestrongestboaardstofailurre.Thetensio
ontestresulltsmighttheereforehavebeen
orstrongerb
boardsifabeettertestsettuphadbeenpossible.O
Ofthetotaltensionsamp
pleof
higherfo
623boards,611werretesteddueetosamplesswhichslipped.
werecalculattedaccordingtoSANS61
122
Bendinggstrength,teensionstrenggthandMOEEedgevaluesw
(1994).
55
InthefinalanalysisallboardswithmorewanethanallowedinSANS17832(2005)werealso
discardedleavingatotalof1112boards.
Statisticalanalysis
Inthestatisticalanalysis,threewaycrossclassificationANOVAsweredoneusingcompartment,log
position,andboardpositionasfactors.TheeffectofthesefactorswereanalysedonMOR,MOE,
tensionstrength,densityandtwist.Ifthethreewayinteractionwassignificantitwasinvestigated
withBonferronimultiplecomparisons.Otherwisethesignificanttwowayinteractionswere
interpreted.Whereafactorwasnotincludedinanysignificantinteractions,themaineffectwas
interpreted.IfenoughdatawasnotavailableforathreewayANOVA,successivetwowayANOVAs
weredone.
Resultsanddiscussion
ThevariationinthemostimportantpropertiesoffullsizedboardswithinatreecanbeseeninTable
2.
Board0
Description
No.ofboards(n)
Yearringno.
Yearringwidth
(mm)
3
Density(kg/m )
Bow(mm)
Spring(mm)
Twist(mm)
Totalknotarea
2
(mm )
Max.knotarea
2
(mm )
Mean
95%Conf.Int
Mean
95%Conf.Int
Mean
95%Conf.Int
Mean
95%Conf.Int
Mean
95%Conf.Int
Mean
95%Conf.Int
Mean
95%Conf.Int
Mean
95%Conf.Int
Mean
Bottom
log
Toplog
Board1
Bottom
log
Toplog
232
203
253
193
3.0
3.2
5.5
5.4
2.83.2
3.03.4
5.25.7
5.25.7
13.1
12.2
10.4
10.4
12.613.6 11.812.7 10.010.9
9.910.9
417.4
411.5
426.6
420.0
413421
407416
422431
415425
2.0
2.0
2.4
2.4
1.72.3
1.82.3
1.92.8
2.02.8
2.8
2.6
2.5
2.6
2.53.2
2.32.9
2.32.8
2.32.8
17.4
18.5
12.1
11.7
16.318.5 17.419.7 11.313.1 10.712.7
16513
15024
11728
11830
15731
14332
11095
11213
17296
15716
12362
12447
4505
4522
2983
3272
42544755 42354810 27923175 30603485
7170
7372
8650
8457
MOEdyn(MPa)
MOEedge(MPa)
MOR(MPa)
Tensilestrength
(MPa)
2
95%Conf.Int
Mean
95%Conf.Int
n
5thPercentile
Mean
95%Conf.Int
n
5thPercentile
Mean
95%Conf.Int
n
R value:MOEdynvsMOR
(linearregression,allsignificantat
0.05level)
69827358 72167528 84098892 82218693

5204
5041
5948
5873
50265381 48725209 56986199 56686078
124
112
134
106
15.77
15.58
14.70
13.11
27.58
24.84
30.16
27.86
26.029.1 23.526.2 28.232.1 26.029.7
124
112
134
106
6.65
6.57
7.12
7.05
9.15
8.80
11.63
11.22
8.79.6
8.59.4 10.912.3 10.512.0
108
91
119
95
0.41
0.34
0.62
0.33
0.36
0.51
Allbottomlogboards:0.64
Alltoplogboards:0.36
Allboards
Board2
Bottomlog
117
8.1
7.78.5
7.9
7.48.3
450.2
443458
2.0
1.62.3
2.4
2.12.7
8.6
7.59.7
6614
Toplog
61
7.9
7.58.4
8.4
7.89.1
436.1
428445
2.0
1.62.4
2.2
1.82.6
8.9
7.210.6
8725
57367492 78359614
2094
2812
18692319 26023023
10859
9695
9261
1044611272
10130
7153
6358
67337574 58836834
58
33
22.04
15.16
41.44
30.79
38.044.9 26.934.6
58
33
7.93
6.54
16.40
14.57
15.217.6 12.716.5
59
29
0.59
0.29
0.54
1112
4.9
10.8
424.4
2.2
2.6
13.7
12622
3548
8422
5755
586
15.03
29.52
588
6.90
11.40
524
0.56
Table2Measureddataperboardposition.Dataforthefewboard3and4positionswerenotincluded
individuallybutincludedintheAllboardscolumn.
56
Modulusofelasticity
Stiffnesswasmeasuredstaticallyontheboardsedge(MOEedge)aswellaswithanacousticmethod
(MOEdyn).TheMOEedgemeasurementisthemethodgenerallyusedtodeterminethecharacteristic
anddesignMOEvaluesfortimber.ThemeanMOEedgeofthesamplewas5755MPa(Table2)which
wasfarbelowtheminimumof7800MParequiredforSouthAfricanstructuralgradesawntimber
(SANS101631,2003).ThisresultsuggeststhatingeneralthemeanMOEedgeorstiffnessofP.patula
grownintheforestareasampledisabout26%belowtherequirementforstructuralsawntimber,
whichshouldbeofamajorconcern.ThemeanMOEedgevaluesincreasedsharplyfromthepith
boardsofthebottomlog(5204MPa)tothesecondboards(7153MPa)ofthebottomlog.Studiesby
Malanetal.(1997)andMalan(2001)showedthatthefirst12to15yearsofgrowthmoreorless
representsthecorewoodzone(sometimesreferredtoasthejuvenilewoodzone)ofthisspeciesin
termsofwooddensity.Thesecondboardshadameanyearringnumberof8.1and7.9forthe
bottomandtoplogsrespectivelyandwouldthusstillbeinthejuvenileregion.Thejuvenilewood
regionischaracterisedbyrapidlychangingwoodpropertieswhichstabiliseasthewoodapproaches
maturity(ZobelandvanBuijtenen1989).Itcan,therefore,beexpectedthatboardsfromthemature
regionformedlatershouldhavehigherMOEedgevalues.Inadifferentstudyitwasfoundthatthe
meanMOEedgevaluesfromsawntimberfromsawmillswhichusearesourceconsistingofmainly
matureP.patulalogsharvestedataround28yearswerefoundtobeintheregionof7800MPa
(CraffordandWessels2011).
Currentspecificationsrequiresawntimbertoconformtoeithervisualormechanicalgrading
specificationsinordertoqualifyasstructuralproducts.Actualmechanicaltestingisnotrequiredon
acontinuousbasisintermsofcurrentspecifications(SANS178312004andSANS101492011)
whichmeansthatsawmillsprocessinglogsfromlocallygrownpineareoftenunawareofthefact
thatthestructuralsawntimbertheyproducemightnotconformtothestiffnessrequirementsofthe
respectivegrades.
Thedynamicmodulusofelasticity(MOEdyn)isarguablyabetterindicationofthemeanstiffnessofa
pieceoftimbercomparedtothestaticmodulusofelasticitydeterminedwhentestedontheedge
(MOEedge),asthefullvolumeofmaterialinfluencesMOEdynwhereasMOEedgeisinfluencedmoreby
thelocalstiffnessofthematerialatthehighlystressedareasofaspecifictestsetup.AsMOEdynwas
determinedonalltheavailableboards,contrarytoMOEedge,whichwasdeterminedonlyonboards
earmarkedforbendingtests,itprovidedabettermeasureforquantifyingtheextentandpatternsof
variationinstiffnessamongindividuallogs,treesandcompartments.
Athreewayanalysisofvariationindicatedthatcompartment,logpositionandboardposition
significantlyinfluencedmeanMOEdyn.Asignificantinteractionexistedbetweenlogpositionand
boardposition(Table3,Figures3and4).Somecompartmentswithtoofewdatapointshadtobe
omittedtomakethethreewayANOVApossible.Thepercentagevariationexplainedbyeach
componentisalsolistedinTable3anditisnoticeablethatboardpositionwasthemostinfluential
componentexplaining26.5%ofthevariationinMOEdyn.Compartmentexplainedabout8.0%ofthe
variation.
57
Sourceofvariation
SS
Degrees
offreedom
MS
Variance
components
(%)
Compartment
2.62E+08
12
2.18E+07
9.05
0.0000
8.0%
Logposition
2.56E+07
2.56E+07
10.61
0.0012
0.8%
Boardposition
8.69E+08
4.35E+08
180.06
0.0000
26.5%
Compartment*Logposition
3.66E+07
12
3.05E+06
1.26
0.2359
1.1%
Compartment*Boardposition
8.53E+07
24
3.56E+06
1.47
0.0675
2.6%
Logposition*Boardposition
3.87E+07
1.93E+07
8.01
0.0004
1.2%
Compartment*Logposition*Boardposition
7.34E+07
24
3.06E+06
1.27
0.1768
2.2%
Error
1.89E+09
784
2.41E+06
57.6%
Table3AthreewayanalysisofvarianceforMOEdyn(MPa).Significantfactorsatthe5%levelareshaded.The
lastcolumngivesthepercentagevariationthatisexplainedbyeachcomponent.
Sincecompartmentwasnotinvolvedinanysignificantinteractions,themaineffectcouldbe
interpreted.ThemeanMOEdynvaluespercompartmentvariedbetweenabout7500and10000MPa
(Figure3).ItisinterestingtonotethatthemeanMOEdynvaluesofcompartmentsfromthesame
estateandwhichwererelativelyclosetoeachotheroftendifferedsignificantlyi.e.compartmentsA,
B,andCwerefromthesameestate(Nelshoogte)aswellascompartmentsD,E,FandG(Uitsoek).
Compartment; LS Means
Current effect: F(12, 784)=9.0500, p=.00000
Effective hypothesis decomposition
Vertical bars denote 0.95 confidence intervals
11000
10500
10000
MOEdyn (MPa)
9500
9000
8500
8000
7500
7000
6500
A
Compartment
Fig.3Themeansand95%confidenceintervalsofMOEdynforboardsfromdifferentcompartments
TheinfluenceofboardpositionandlogpositiononMOEdyncanbeseeninFigure4.Asexpected
MOEdynincreasedsharplyfromthepithboardstothesecondboards.ThemeanMOEdynofpith
boardsofthebottomlogsdidnotdiffersignificantlyfromthepithboardsofthetoplogs.Thiswas
alsothecaseforfirstboards.Forsecondboards,however,themeanMOEdynwassignificantlyhigher
forboardsfromthebottomlogthanfromtoplogboards.Thiscouldprobablybeascribedtothe
effectofpruningon2ndboards.Allthecompartmentsreceivedpruningupto3mataround7years
andpruningto5mheightataround9years.Thesecondboardsofthebottomlogswhichhada
meancambialage(yearringnumber)of8.1wouldthusbepartiallyknotfree.Thisissupportedby
thefactthatboththemeantotalknotareaandthemeanmaximumknotareaforsecondboardsof
thetoplogswassignificantlydifferentthanforsecondboardsofthebottomlogs(Table2).
58
Logpo
os*BoardPos; LS Means
Current ef f ect:
e
F(2, 784)=8.01
136, p=.00036
Ef f ectiv e hy pothesis deco
omposition
Vertical bars denote
d
0.95 conf id
dence interv als
12000
11500
11000
10500
MOEdyn (MPa)
10000
9500
9000
8500
8000
7500
7000
6500
6000
Top
B
Bottom
Log position
Board
B
0
Board
B
1
Board
B
2
Fig.4Theemeansand9
95%confidenceintervalsofMOEdynofth
hedifferentboardpositionsfromthetop
pand
bottomlo
ogs
ofthemeanMOEdynvalue
esandthem
meanringwid
dthofpithboardsofeacch
Alinearregressiono
edagoodreelationshipoffR2=0.76(FFigure5).Wh
henalinearrregressionoffthe
comparttmentshowe
meanrin
ngwidthsofthepithboaardsandmeaanMOEdynofalltheboardspercomp
partmentwaas
perform
medtheR2vaaluewasstill0.50.There
easonforthissrelationship
pismostpro
obablythat
comparttmentswithslowinitialggrowthproducepithboaardscontainingmoreand
dolderyearrings.
Olderyeearringswillhavehigherrdensityand
dlowermicro
ofibrilanglesswhichisasssociatedwith
hhigher
stiffnesss.Additionallly,theseoldeeryearringssinthepithb
boardsareusuallyalsolo
ocatedattheeedges
ofaboardwheretheeirinfluenceeintermsofstiffnessonedgeofthefullboardw
willbemaximised.
uppressiono
ofearlygrow
wthmightbeonestrategyytoincreaseethe
Thesereesultssuggesstthatthesu
meanM
MOEedgeandM
MOEdynofbo
oardsfromth
heyoungPin
nuspatulareesource.Also
onoticethatfaster
growthaatalaterstaagesuchasth
hatexperien
ncedbycomp
partmentFd
didnotseem
mtoinfluenceeMOEdyn
negativeely.Thiscom
mpartmenthaadthesloweestinitialgrowthbutath
harvestinghaadanaboveaverage
meanDB
BHandstillh
hadoneofth
hehighestm
meanMOEdynofallthecompartmentss.
Fig.5Asscatterplotand
dregressionlineofthemeanMOEdynan
ndmeangrow
wthringwidthssofpithboard
ds(board
0)ofthebottomlogsffromdifferenttcompartmen
nts.Individualcompartmen
ntscanbeidentifiedbytheeletter
nexttoeeachplot
59
TheMOEedgeoftheyoungPinuspatulatimberwaslowerthanrequiredforstructuralgrades.Ifthe
SouthAfricansawmillingindustryplanstousethisresourceforstructuralproductsitshouldfocuson
waystoimprovethisproperty.RecentresearchinNorthAmerica(P.taeda)andNewZealand(P.
radiata)showedthathighinitialplantingdensityledtohigherstiffnesswood(Rothetal.2007,
Waghornetal.2007,Laserreetal.2005,Laserreetal.2009).Stiffnesswasshown,intheseand
otherstudies(i.e.Wattetal.,2006),toberelatedtotreeslendernesswhichisafunctionofplanting
density.Thismightbeoneoptionworthinvestigating.Highinitialplantingdensityisalsoassociated
withsuppressionofinitialgrowthwhichwasshowninthisstudytoincreasethemeanMOEdynof
boards.Takenotethatalthoughindividualtreegrowthmightbesuppressedbyhighplanting
density,theperhectarevolumeyieldsareoftenhigherduetothehigherstemcounts.
Modulusofrupture(bendingstrength)
TheabsolutevaluesofMORforthisyoungresourceweresurprisinglyhigh.Thevalueofmost
interestintermsofstructuraltimberisthe5thpercentilevalueorcharacteristicbendingstrength
whichisusedfordesigningstructuresfromaspecificgradeofwood.Foralltheboardstestedinthis
study,the5thpercentilevalueasdeterminedaccordingtoSANS6122(1994),was15.03MPa(see
Table2)whichiswellabovetherequired5thpercentilevalueforthelowestandmostcommonly
usedstructuralgradeinSA(S5,11.5MPa)andveryclosetotherequirementforthenextstructural
grade(S7,15.8MPa).
Contrarytoexpectation,the5thpercentileMORvaluesforthepithboardswerehigherthanthatof
thefirstboardsforboththebottomandtoplog(Table2).The5thpercentileMORvalueofthepith
boardsofthetoplogswasevenhigherthanthatofthesecondboardsofthetoplog.Thiswasan
interestingresultwhichdidnotconformtotheexpectedtrendofimprovedpropertiesforboards
furtherawayfromthepith.ThemeanMORvalues,however,performedasexpectedwherethe
meanMORforsecondboardswashigherthanthatoffirstboardswhichwasagainhigherthanthat
ofpithboards.The5thpercentilevaluefor2ndboardsimproveddramaticallyforbottomlogboards
butremainfairlylowfortoplogboards.Thiswasmostprobablyduetothefactthatthesecond
boardofthebottomlogalreadyformedpartoftheprunedsectionwheretheeffectofknotswas
removed.
BendingstrengthorMORisusuallyrelatedtotheknotpropertiesofboards.Themeanmaximum
knotareasofpithboardsandmeantotalknotareaswerehigherthanforthefirstboards(Table2).
Othervariablesthatareusuallycloselyrelatedtothebendingstrengthoftimberincludedensityand
MOEwhichwerebothlowerinthepithboardsthanthefirstboards.Apossibleexplanationforthe
high5thpercentileMORvaluesofpithboardsmightbethatthemaximumknotareameasuredby
theXrayscanningdeviceforapithboardwasoftenacombinationofafewsmallknotsfromasingle
branchwhorlresultinginahighmaximumknotareameasurement.Becausetheseweremostly
smallknots,graindeviationaroundtheknots,whichistheactualcauseofweaknessinbending(see
Walker,1993),wouldnotbeverypronounced.Forthefirstboardswheresingleknotsareusually
bigger,amorepronouncedgraindeviationmightresultinverylowbendingvaluesforafewboards
butonlywheretheknotwassituatedclosetoahighstressarea.Thismayexplainwhysome
boardsfurtherfromthepithwerecharacterisedbyhighermeanMORvaluesbutwithlower5th
percentileMORvalues.Inotherwords,theweaktailsectionoftheMORhistogramfor1stand2nd
boardsweremorepronouncedthanforpithboardseventhoughthemeanvalueswerequitehigh.
Whenvisualgradingwasperformedonalltheboards,andboardsthatdidnotmakethestructural
gradeswereomittedfromthesample,thesameresultwasobtained.The5thpercentilevalueofthe
1stboardswasstilllowerthanthatofthepithboards.However,inthiscasethevalueswerecloserto
oneanotherthanwhennogradingwasperformed.Thismeansthatvisualgradingwasnot
60
completelysuccessfulineliminatingthecauseofthelower5thpercentilebendingstrengthvalues
obtainedin1stboards.
TherewasnotenoughdataforathreewayANOVAandthereforesuccessivetwowayANOVAs
weredone.TheresultsshowedthatthemeanMORdifferedsignificantlybetweencompartments,
logpositionandboardposition.Therewasasignificantinteractionbetweenboardpositionandlog
positionatthe0.05level.ForthepithandfirstboardstheMORdidnotdiffersignificantlybetween
thetopandbottomlogs(Figure6).However,forthesecondboardalargeandsignificantdifference
betweenthemeanMORvaluesofthetopandbottomlogwasfound.Thiswasmostlikelyduetothe
influenceofpruningonthesecondboardsofthebottomlog.
Logpos*BoardPos; LS Means
Current ef f ect: F(2, 561)=5.8854, p=.00295
Ef f ectiv e hy pothesis decomposition
Vertical bars denote 0.95 conf idence interv als
50
45
MOR (MPa)
40
35
30
25
20
Board 0
Board 1
Board Position
Board 2
Top log
Bottom log
Fig.6Themeansand95%confidenceintervalsofMORofthedifferentboardpositionsfromthetopand
bottomlogs
Tensilestrength
Aswithbendingstrength,themostimportantvaluedeterminingthesuitabilityofsawntimberfor
structuralpurposesisthe5thpercentiletensilestrength.The5thpercentiletensilestrengthforallthe
testboardscombinedwas6.9MPa(seeTable2)whichwasslightlyhigherthanthe5thpercentile
valueof6.7MParequiredforgradeS5timber.Itshouldbetakenintoconsiderationthatthetensile
testsetupwassuchthatmostpiecesfailedprematurelyduetocompressionbytheclampingeffect
atthegrips.Thisprobleminfluencedthestrongerpiecesmorethanweakerpieceswhichusually
failedataknotandhadabiggerimpactonthemeantensilestrengththanthe5thpercentilevalue.
Consideringtheyoungageofthisresourceaswellastheproblemwithgripfailures,thetensile
strengthseemstobeacceptable.
Similartothebendingstrength,the5thpercentiletensilestrengthvaluesshowedadifferenttrend
whencomparedtomeantensilestrengthvalues.The5thpercentiletensilevaluesoftheboards
increasedonlymoderatelyfromthepithboardstowardsthe2ndboardswhereasthemeanvalues
increasedfairlysharply(Table2).The5thpercentiletensilestrengthvalueofthe2ndboardfromthe
bottomlog(7.93MPa)wasforinstanceonly19%higherthanthevalueforthepithboardofthe
samelogwhereasthemeanvalue(16.40MPa)wasnearly80%higher.The5thpercentilevalueofthe
toplogs2ndboardwaslowerthanthatofthepithboard.Thiswasmostprobablyduetothemore
severegraindeviationwhichwasassociatedwiththelargerknotswhichcharacterisedtheweakest
61
boardsoriginatingfromthetoplogs.The2ndboardsofthebottomloghadamuchhigher5th
percentilevalue.Theseboardsoriginatedfromtheprunedsectionswheretheboardswereless
affectedbythepresenceofknots.Analysisofvariationindicatedthatthelogpositionandboard
positionsignificantlyinfluencedthemeantensilestrengthandthattherewasasignificant
interactionbetweenboardpositionandcompartment.
Warp(bow,spring,andtwist)
Theaveragebowandspringobservedinalltheboardsexaminedwereonly2.2mmand2.6mm
(Table2)whichwasconsiderablylessthantheallowablelimitsforstructuralgradetimbersetbythe
SANS17832(2005)of15mmand21mmrespectively.Bowandspringwereclearlynotaproblemin
thisresource.Themeansofthesepropertiesfordifferentboardandlogpositionswereverysimilar.
Theamongboardvariabilityinbothpropertiesdecreasedwithdistancefromthepith.
Twistintheboardsexaminedaveraged13.7mm/m,withwelloverhalftheboardsnotpassingthe
minimumrequirementof10mm/mfor38x114x2100mmsizestructuraltimber(SANS17832,
2005).Athreewayanalysisofvarianceindicatedthatcompartmentandboardpositionsignificantly
influencedtheleveloftwist(Table4).Somecompartmentswithtoofewdatapointshadtobe
omittedtomakethethreewayANOVApossible.Thepercentagevariationexplainedbyeach
componentisalsolistedinTable4andboardpositionwasthemostinfluentialcomponent
explaining18.5%ofthevariationintwist.Twistwassignificantlydifferentbetweenboardsfrom
differentradialpositions,withthemosttwistoccurringinpithboards,followedbythe1standand
2ndboards(Figure7,Table2).Thiswasmostlikelyduetothehigherlevelsofspiralgrainthatusually
characterisesthecentrecoreofthisspecies(GerischerandKromhout1964;Malan2010).
Compartment
Logposition
SS
Degrees
offreedom
MS
1302.6
12
108.5
2.208
0.0100
Variance
components
(%)
2.52
0.01
0.0
0.000
0.9914
0.00
Boardposition
9573.4
4786.7
97.371
0.0000
18.52
Compartment*Logposition
459.5
12
38.3
0.779
0.6728
0.89
Compartment*Boardposition
1552.1
24
64.7
1.316
0.1431
3.00
206.2
103.1
2.097
0.1235
0.40
Compartment*Logposition*Boardposition
1037.7
24
43.2
0.879
0.6316
2.01
Error
37557.8
764
49.2
72.66
Table4.Athreewayanalysisofvariancefortwist(mm).Significantfactorsatthe5%levelare
shaded.Thelastcolumngivesthepercentagevariationthatisexplainedbyeachcomponent.
62
BoardPos; LS Means
Current ef f ect: F(2, 764)=97.371, p=0.0000
Ef f ectiv e hy pothesis decomposition
Vertical bars denote 0.95 conf idence interv als
22
20
18
Twist (mm)
16
14
12
10
8
6
4
Board 0
Board 1
Board Position
Board 2
Fig.7Themeansand95%confidenceintervalsoftwistforboardsfromthedifferentpositionsinalog
Density
Densityasanindividualpropertyislessimportantforstructuraltimberthanstrengthandstiffness.
However,densityhasapositivecorrelationwiththedifferentstrengthandstiffnesscharacteristics
whichareofimportanceinstructuraltimberandthereforemoststructuraltimbercodesalsorequire
minimumdensityvaluesforthevariousstructuralgrades.Also,thewithdrawalresistanceofsome
structuraljoiningmethods,likenailplates,dependsonthedensityoftimber.
ThedensityrequirementforstructuraltimberinSouthAfricaisaminimumdensityof360kg/m3for
gradeS5and425kg/m3forgradeS7at12%moisturecontent(SANS178322005).Onaveragethe
densityofthematerialwassufficienttomeettheminimumrequirementforS5gradestructural
timber.Manyboardsoriginatingfromradialpositionsfurtherawayfromthepithinfactmetthe
minimumrequirementforS7gradestructuraltimber.
Athreewayanalysisofvarianceindicatedthatcompartment,logpositionandboardposition
significantlyinfluencedthedensity(Table5).Somecompartmentswithtoofewdatapointshadto
beomittedtomakethethreewayANOVApossible.Thepercentagevariationexplainedbyeach
componentisalsolistedinTable5.Compartmentwasthemostinfluentialcomponentexplaining
10.5%ofthevariationindensity.Itwasinterestingtonotethatdensitywastheonlypropertywhere
compartmentwasmoreinfluentialthanboardposition.
63
SS
Degrees
offfreedom
MS
Compartmeent
118025
12
9835
5
9.11
0.0
000000
Variiance
compo
onents
(%
%)
10
0.51
Logposition
11452
11452
10.61
0.0
001171
1..02
Boardposittion
78505
39253
36.38
0.0
000000
6..99
Compartmeent*Logposition
18886
12
1574
4
1.46
0.1
134486
1..68
Compartmeent*Boardpositio
on
21056
24
877
0.81
0.7
722443
1..88
n
1972
986
0.91
0.4
401388
0..18
Compartmeent*Logposition*Boardposition
20598
24
858
0.80
0.7
745407
1..83
Error
852483
790
1079
9
75
5.91
Table5.A
Athreewayaanalysisofvarrianceforden
nsity(kg/m ).SSignificantfacctorsatthe5%
%levelareshaaded.The
lastcolum
mngivesthep
percentagevaariationthatissexplainedbyyeachcompo
onent.
Coefficieentofdeterm
mination(R2)betweenM
MOEdynandM
MOR
Thecoeffficientofdeetermination
nofboardsfrromthetoplogwasconssistentlymuchsmallerth
hanfor
boardsffromthebotttomlog(Tab
ble2).TheR2valueforalllthetoploggboardscom
mbinedwas0
0.36
compareedtoanR2vvalueof0.64forallthebottomlogbo
oardscombined(Figure8
8).Whatwasvery
interestiingwasthattherelationshipremainedweakfortoplogboarrdsevenwheenonlythe1
1stor2nd
2
boardsw
wereconsideeredindividu
ually(R =0.3
33and0.29).Thediffere
enceintheM
MOEdynvs.MO
OR
nd
2
coefficieentofdeterm
minationbetweenthe2 boardsoftthebottomaandtoplogs(R =0.59an
nd0.29)
canposssiblybeexplainedbytheesignificantd
differencesinthetotalknotareaand
dmaximumkknot
areasrespectivelyoffthesetwoggroups.Thesignificantlylargerknotssinthetoplo
og2ndboards
he1st
possiblyycausedalessspredictablleMORvaluee.ThereasonforthediffferenceinR2valueforth
2
boardso
ofthetopan
ndbottomlo
ogs(R =0.62
2and0.33)w
waslessclearrsincethekn
notareavalu
uesdid
notdiffeersignificanttlyfortheseboards.
Fig.8Scaatterplotswith
hlinearregresssionlinesoftheMOEdynvss.MORrelatio
onshipofboardsfromtheb
bottom
logs(left)andtoplogss(right)
nofMOEdynvvs.MORforp
pithboardsw
wassmallertthanfortheother
Thecoeffficientofdeetermination
boards((R2=0.34to0.41)butfellwithintheexpectedrange(Glos20
004).Thisrelationshipwaasnot
nearlyaspoorastheeMOEvs.M
MORrelationsshipreported
dbyGaunt(1999)forjuvvenileradiatapine
fromNeewZealand.
ultsindicated
dthattherelationshipbe
etweenMOEEdynandMOR
Rforpithboardsandtop
plogs
Theresu
wasfairlyweakandincaseswheerethesesecctionsofatrreeareproceessedseparaately,structu
ural
gradingbasedonthisrelationshipwillbelesssefficient.
Radial,llongitudinallandbetweeencompartm
mentvariatio
oninproperrties
Asexpecctednearlyaallpropertiessdeterminin
ngthesuitabilityofsawntimberforsstructuralpurposes
improveedwithincreeasingdistancefromthepith.Theonlynotableexxceptionwassthe5thperccentile
valueforMORwhich
hwashigherrforthepith
hboardsthan
nforthe1stb
boards(Table2).Thiswaasquite
64
aninterestingandunexpectedresultaseventhemeanMORvalueforthe1stboardswas
significantlyhigherthanthemeanvalueforpithboards.Boardpositionwasbyfarthemost
influentialcomponentinexplainingthevariationinMOEdynandtwist(Tables3and4).
LogpositionhadasignificanteffectontheMOEdyn,MORandtwistalthoughitsinfluencewas
relativelysmallcomparedtoboardposition(Tables3and5).Differencesbetweentopandbottom
logboardsbecamemorepronouncedinthe2ndboardswherepruninginfluencedsomeofthe
propertiesofthebottomlogboards(Figure4and6).Themajordifferenceintopandbottomlog
boardswastherelationshipbetweenMOEdynandMOR.TheR2valuesofboardsfromthetoplogs
weremuchlowerthanthatofboardsfromthebottomlogswhichwillresultinlessefficient
structuralgradingoftoplogboards(Table2andFigure8).
CompartmenthadasignificantinfluenceontheimportantpropertiesofMOEdyn,MOR,twistand
density.Forallofthesepropertiestheboardpositionwasmuchmoreinfluentialthanthe
compartmentexceptfordensity(Tables3,4,and5)wherecompartmentexplained10.5%ofthe
variance.
Conclusions
Basedontheresultsfromthisstudy,thefollowingconclusionsaremade:
ThemeanMOEedgeofthisyoungPinuspatularesourcewasfarbelowthatrequiredfor
structuralgradetimberinSouthAfrica.
The5thpercentilebendingstrength(MOR)wasmuchhigherandthe5thpercentile
tensilestrengthslightlyhigherthanthatrequiredforthelowestSAstructuralgrade.
ThemeantwistofthesamplewashigherthanallowedforstructuraltimberinSouth
Africa.Themeanbowandspringwerefarbelowthemaximumlimitallowed.
Mostofthedesirablepropertiesforstructuraltimberwereimprovingforboardsfurther
removedfromthepith.Theonlyexceptionwasthe5thpercentilevalueforMORwhich
washigherforthepithboardsthanforthe1stboards.Boardpositionwasbyfarthe
mostinfluentialcomponentinexplainingthevariationinMOEdynandtwist.
LogpositionhadasignificanteffectontheMOEdyn,MORandtwist.TheR2value
betweenMOEdynandMORofboardsfromthetoplogsweremuchlowerthanthatof
boardsfromthebottomlogmakingthenondestructivepredictionofMORusingMOEdyn
lessefficientfortoplogboards.
Anegativeandsignificantcorrelationexistedbetweenthemeangrowthringwidthsof
thepithboardsandthemeanMOEdynvaluespercompartment.Inotherwordswhenthe
initialgrowthofacompartmentwasslow,themeanMOEdynvaluewashigh.
Theresultsfromthisstudycanbeusedforvariouspurposesbutwillbeespeciallyapplicablefor
processingdecisionsupportinstructuraltimberprocessingorganisations.Furtherstudiesshould
focusonthemethodsavailablethatcouldpossiblyimprovethestiffnessoftheyoungPinuspatula
resourcesuchashigherplantingdensities,treebreedingselectionstrategies,andothersilvicultural
interventions.Methodstoreducetwistinboardsfromthisresourceshouldalsoreceiveattention.
Acknowledgements
Thefollowingorganisationsandindividualsaregratefullyacknowledgedfortheirhelpwiththis
study:SawmillingSouthAfricaandtheNRFsTHRIPprogrammeforsponsoringthisproject.
KomatilandForestsforsupplyingthetreesfortheprojectandfundingsometasks.YorkTimbersfor
thesawing,dryingandXrayscanningofthetimber.GeorgeDowseandWilmourHendriksefor
laboratorytesting.JosLouwandBarryMullerfromNMMU,Saasveldforcooperationinthisproject.
65
References
BurdzikW.2004.GradeverificationofSApine.SouthernAfricanForestryJournal202:2127.
CraffordPH,WesselsCB.2011.TheflexuralpropertiesandstructuralgradingofSAPine.Reportto
SawmillingSA.Copyobtainablefromcbw@sun.ac.za.
CrickmayDG,LeBrasseurJ,StubbingsJA,DaughertyAE.2005.Studyofthesupplyanddemandof
industrialroundwoodinSouthAfrica.ReportbyCrickmayandAssociates.Copyobtainablefrom
mandy@crickmay.co.za.
CrickmayandAssociates.2011.LumberIndexforOctober2011.Copyobtainablefrom
GauntDJ.1999.Machinestressgradingrevisited.NZTimberDesignJournal1(8):1018.
GerischerGFR,KromhoutCP.1964.NotesonbreastheightspiralityindominanttreesofPinus
patula,Pinustaeda,andPinuselliottii,withspecialreferencetotreebreeding.ForestryinSouth
Africa5:8197.
GlossP.2004.Newgradingmethods.ProceedingsofCOSTE29Symposium,FlorenceOctober2729.
CNRIvalsa,SanMicheleallAdige,Italy,p18.
JohanssonCJ.2003.In:Thelandersson,S.,Larsen,H.J.(eds)(2003).TimberEngineering.JohnWiley
&SonsLtd.445pp.
KennedyRW.1995.Coniferouswoodqualityinthefuture:concernsandstrategies.WoodSci.
Technol.29,321338
LasserreJP,MasonEG,WattMS,MooreJR.2009.Influenceofinitialplantingspacingandgenotype
onmicrofibrilangle,wooddensity,fibrepropertiesandmodulusofelasticityinPinusradiataD.Don
corewood.ForestEcologyandManagement258:19241931.
LasserreJP,MasonEG,WattMS.2005.TheeffectsofgenotypeandspacingonPinusradiataD.Don
corewoodstiffnessinan11yearoldexperiment.ForestEcologyandManagement205:375383.
MalanFS,RetiefRJ,MaleJR.1997.Theinfluenceofplantingespacementonthewooddensityand
pulpingpropertiesofPinuspatula.SouthAfricanForestryJournal180:2332
MalanFS.2001.WooddensitypatternsinSouthAfricanpinewithspecialreferencetotheeffectof
abnormalcompressionwoodinP.taeda.UnpublishedInternalCompanyReport02/2001,South
AfricanForestryCompanyLtd.
MalanFS.2010.CorewoodinSouthAfricanpine:necessityandopportunitiesforimprovement.
SouthernForests:aJournalofForestScience,72(2):99105.
66
RothBE,LiX,HuberDA,PeterGF.2007.Effectsofmanagementintensity,genetics,andplanting
densityonwoodstiffnessinaplantationofjuvenileloblollypineinthesoutheasternUSA.Forest
EcologyandManagement246:155162.
SANS10149.2002.SouthAfricanNationalStandard.Themechanicalstressgradingofsoftwood.
SANS101631.2003.SouthAfricanNationalStandard.ThestructuraluseoftimberPart1:Limit
statesdesign.
SANS17831.2004.SouthAfricanNationalStandard.Sawnsoftwoodtimber,Part1:General
requirements.
SANS17832.2005.SouthAfricanNationalStandard.SawnsoftwoodtimberPart2:Stressgraded
SANS6122.1994.SouthAfricanNationalStandard.Qualificationtestingofsolidstructuraltimber
andlaminatedstructuraltimber(glulam)forverifyingtimbergradingsystemsinaccordancetoa
givenstandard.
WaghornMJ,MasonEG,WattMS.2007.Influenceofinitialstanddensityandgenotypeon
longitudinalvariationinmodulusofelasticityfor17yearoldPinusradiata.ForestEcologyand
Management252:6772.
WalkerJCF.1993.PrimaryWoodProcessingPrinciplesandPractice.ChapmanHall,London.
WattMS,MooreJR,FaonJP,DownesGM,ClintonPW,CokerG,DavisMR,SimcockR,ParfittRL,
DandoJ,MasonEG,BownHE.2006.Modellingtheinfluenceofstandstructural,edaphicand
climaticinfluencesonjuvenilePinusradiatadynamicmodulusofelasticity.ForestEcol.Manag.229
(13):136144.
WrightJA.1994.UtilizationofPinuspatula:Anannotatedbibliography.OxfordForestryInstitute
OccasionalPaperno.45.
XuP,WalkerJCF.2004.Stiffnessgradientsinradiatapinetrees.WoodScienceandTechnology
38(1):19.
ZobelBJ,SpragueJR.1998.Juvenilewoodinforesttrees.SpringerVerlagBerlinHeidelberg,300pp.
ZobelBJ,VanBuijtenenJP.1989.Woodvariation.Itscausesandcontrol.Springer,Heidelberg,
363pp.
67
Chapter5.
Unpublished
ThepredictionoftheflexurallumberpropertiesfromstandingSouthAfricangrownPinus
patulatrees
C.B.Wesselsa*,F.S.Malanb,T.Seiferta,J.H.Louwc,T.Rypstraa
a
PrivateBagX1,Matieland7602,SouthAfrica
Tel:+27218083319
Fax:+27218083603
email:cbw@sun.ac.za
b
PostgraduateForestProgramme,
UniversityofPretoria
PrivateBagX20,Hatfield0028,SouthAfrica
c
SchoolofNaturalResourceManagement
NelsonMandelaMetropolitanUniversity
PortElizabeth6031,SouthAfrica
Abstract
Pinuspatulaisthemostintensivelyplantedconiferinthetropicsandsubtropics.InSouthAfrica
Pinuspatulaplantationsarethemainsawlogresourceforstructurallumberproduction.Improved
intensivesilviculturalpracticesandtreebreedinghaveresultedinmarkedincreasesintherateof
growth.Toreapthefinancialbenefitsofthefastergrowth,plantationmanagersaremoreandmore
inclinedtoreducerotationages,whichinevitablyresultsintheproductionofhigherproportionsof
juvenilewoodwhichoftenyieldlumberwhichdoesnotmeettheminimumrequirementsinstiffness
forstructurallumberwhenharvested.Thepurposeofthisstudywastodevelopempiricallybased
modelsforpredictingtheflexuralpropertiesofthewoodproducedfromrelativelyyoungPinus
patulatrees.Modelswerebasedonthepropertiesofstandingtreesandtheireffectivenesswas
evaluatedatboard,treeandcompartmentlevels.Modelsofthiskindarebecomingincreasingly
importanttoserveastoolsinunderstandingandmanagingtheeffectsofshorterrotationageson
thequalityofthewoodproduced.Samplematerialwasobtainedfrom170Pinuspatulatrees(1620
yearsoldatthetimeofsampling)establishedin17compartmentsontheMpumalangaescarpment
ofSouthAfricawhichrepresentedanumberofdiversesiteconditions.Alargenumberofvariables
whichcouldbeobtainednondestructivelyfromthetreeswhiletheywerestillstandingwere
measured.Thetreesweresubsequentlyfelledand340logs(2pertree)extractedfromthetreesand
processedinto1402boardsforfurthermeasurementsanddestructivetesting.Multipleregression
modelsweredevelopedwhichmanagedtoexplain68%,60%and95%ofthevariationinthe
dynamicmodulusofelasticity(MOE)onindividualboards,treesandcompartmentslevels
respectively.Themodelsdevelopedformodulusofruptured(MOR)explained40%and42%of
variabilityatboardandtreelevelrespectively.Atcompartmentlevel,80%ofthevariationinthe5th
percentileMORvaluecouldbeexplainedbythemodel.Sensitivityanalysesshowedthatsiteindexat
baseageof10years,acoustictimeofflight,wooddensityandringwidthwereinfluentialvariables
intheMOEmodels.ThemodeldevelopedforpredictingMORatcompartmentlevelincludedsite
indexatbaseage10years,branchangle,branchspacingandringwidthasinfluentialvariables.The
68
modelsindicatedthattreeslendernessduringearlygrowthseemstoplayamajorrolein
determiningthedynamicMOEandMORoflumber.ThisisinagreementwithEulersbucklingtheory
andthebendingstresstheory.
Keywords:modulusofelasticity,modulusofrupture,Pinuspatula,lumber
1. Introduction
Plantedforestsarerapidlyexpandingonaglobalscaleatabout5millionhaperyearandcurrently
accountforabout7%ofthetotalafforestedareaworldwide(FAO,2013).In1980therewere18
millionhaofplantedforests,comparedto187millionhain1990and264millionhain2012(Carleet
al.,2002;FAO,2013).CarleandHolmgren(2008)estimatedthatin2005globallyabouttwothirdsof
theindustrialtimberoriginatedfromcommercialplantations.
Pinuspatulaisthemostintensivelyplantedconiferinthetropicsandsubtropics.Itisestimatedthat
morethanonemillionhectaresareplantedwiththisspecies;abouthalfofthatinAfrica(Wright,
1994).PinuspatulaisalsoplantedintheAndeancountriesofSouthAmericawithpotentialto
increasetheareaunderthisspeciesinthehighaltitudeareasinBrazil(HodgeandDvorak,2012).In
SouthAfricaitisthemostimportantcommercialplantationsoftwoodresourcewithatotalof
338923haplantedwithPinuspatulatrees(DAFF,2009).TheMpumalangaescarpmentisthelargest
sawloggrowingareainSouthAfricawithPinuspatulathemainspeciesbeingplanted.
SouthAfricawasoneofthefirstcountriestoestablishplantationforestryonalargescale,startingin
thelatenineteenthcentury.By1960theforestryareahadincreasedtoabout1millionha(Owen
andVanderZel,2000).Duetoashortageofsuitablelandavailableforafforestation,aswellas
competitionfromagricultureandwatercatchment,theareaunderforestplantationsinSouthAfrica
hassincestabilised.Tomeetthecountrysgrowingneedsforwoodthisresultedinincreased
emphasesintheforestryandwoodprocessingindustriesonhighervolumeproductionperunitarea
throughimprovedsilviculturalpracticesandgeneticimprovement,aswellasimprovedwood
productyieldandquality.
However,theincreasedsizeofthecorewoodzone,andthebiggerproportionofcorewoodthat
resultswhenrotationagesareshortenedtoreapthefinancialbenefitsofthefastergrowth,has
becomeawoodqualityfactorofgrowingconcernworldwide(ZobelandSprague,1998;Cown,2006;
Malan,2010).Cown(2006)statethatresearchersaroundtheworldhaveconfirmedthataggressive
silviculturalregimeshavecausedasignificantreductioninmechanicalpropertiesofplantation
grownpines.
StudiesinSouthAfricahaveshownsharpreductionsinsomeofthemechanicalpropertiesofpine
lumberprocessedfrommaterialharvestedatayoungerage,astreesreachmerchantablesizemuch
earlierduetofastergrowthrates(Burdzik,2004;DowseandWessels,2013;Wesselsetal.,2011a).
StudiesbyDowseandWessels(2013)haveshownthatthemeanmodulusofelasticity(MOE)of
plantationgrownsoftwoodlumberharvestedbeforetheageof20yearscanbemorethan25%
belowtherequirementoftheloweststructurallumbergradeinSouthAfrica,whichwillhavea
significanteffectonrevenue.Whilethefinancialimportanceofincreasedvolumeproductionof
plantationsisundisputed,itisincreasinglyimportantthatforestmanagersandresearcherstakeinto
considerationtheadverseeffectsoftheiractionsonendproductquality.
Morethan70%ofthesolidsawnlumberproducedinSAissoldasstructuralorbuildingtimber
(CrickmayandAssociates,2011),awoodproductcategorywhichhastocomplytoverystrict
69
strengthandstiffnessrequirementsGiventhechallengescausedbyanincreasingproportionof
juvenilewoodinthetimberresource,thereisagrowingneedfornondestructivemethods,capable
ofaccuratelypredictingthemechanicalpropertiesofthelumberfromstandingtrees
Modelstoaccuratelypredictmechanicalpropertiescanserveausefulroleinmanagingthe
challengesoffastgrowingsoftwoodplantationsandshorterrotationages.Treelevelpredictionscan
assisttreebreederstoscreenandselectforsuperiorbreedingmaterial(Launayetal.,2002;
Lindstrmetal.,2002;Ivkovietal.,2009),whileatsawnboardandcompartmentlevels,predictions
canbeusedtoassistindecisionsrelatedtotheallocationoftreestodifferentprocessingfacilities,
especiallywherestructurallumberisanoption(Mathesonetal.,2002;Cown,2006;Wangetal.,
2007).Modelscanalsobeusedtoassistinprocessingproductionplanning(Uusitalo,1997;Wessels
etal.,2006)andtostudytheeffectsofsiteandsilviculturefactorsonthemechanicalpropertiesof
wood(Wangetal.,2000;Grabianowskietal.,2004;Wangetal.,2005).
Thepurposeofthisstudywastodevelopempiricallybasedpredictionmodelsfortheflexurallumber
propertiesfromstandingPinuspatulaselectedfromanumberofdiverseforestrysitesonthe
MpumalangaescarpmentinSouthAfrica.Theintentionwastoevaluatevariousinputvariablesin
thesemodelsfromdatathatcouldbeobtainednondestructivelyfromstandingtrees.Thisstudyis,
totheauthorsbestknowledge,thefirstoneofthisnatureperformedonPinuspatulaandtheonly
oneforanyspecieswheresuitablecompartmentlevelmodelsweredevelopedtopredicttheMOE
andMORofitslumber.
2. Background
Structuralengineersandotherdesignersoftimberconstructionsusesixdifferentstrengthsanda
stiffnessvalueinthedesignofastructure.Sinceapieceoflumbercanonlybedestructivelytestedin
onestrengthmode,thequestionarises,whichofthestrengthpropertiesarethemostimportantin
termsofenduserequirements.InastudybyPetersonandWessels(2011)ithasbeenfoundthat
bendingstrengthormodulusofrupture(MOR)andstiffnessormodulusofelasticity(MOE)werethe
twomostimportantdesignpropertiesforresidentialrooftrussconstructioninSouthAfrica.Since
morethanhalfofallSouthAfricassawnlumberisutilizedinroofconstructions(pers.comm.Roy
Southey,SawmillingSouthAfrica,Feb2013),themostappropriateevaluationmethodforlumber
destinedforstructuralusewillthereforebethebendingtestfromwhichboththeMOEandMOR
canbederived.
Thecharacteristicstrengthandstiffnessvaluesusedindesigningtimberstructuresaredetermined
bytestinglargenumbersoffullsizedstructuralgradelumbermembersaprocessreferredtoasin
gradetesting.Inthepastthesepropertieshavebeendeterminedonsmalldefectfreewood
specimensbutithasbeenshownthatthefracturebehaviourinclearwoodcomparedtodefect
containinglumberisverydifferent(Madsen,1992).Althoughclearwoodtestingismoreconvenient
withfewersourcesofvariation,itseldomgivesarealisticindicationofthestrengthandstiffness
characteristicsoffullsized,defectcontaininglumber.
TheMOEofwoodisawellresearchedtopicandisknowntodependonanumberofbasicwood
properties.EvansandIlic(2001)foundthatdensityaloneaccountedfor70%ofthevariationinthe
MOEofclearEucalyptusdelegatensiswoodsampleswhilemicrofibrilanglealoneaccountedfor86%
ofthevariation.Thecombinedeffectofthesepropertiesaccountedfor96%ofvariationinMOE.
Megrawetal.(1999)foundthatdensityandmicrofibrilangletogetherexplained93%ofvariationof
MOEinsmallclearwoodsamplesofPinustaeda.Thereisalsoastrongrelationshipbetweenthe
acousticvelocityinthelongitudinaldirectionandthemicrofibrilangleofwood.Wangetal.(2007)
70
andEvansandIlic(2001)reportedcoefficientsofdetermination(R)of0.855and0.86respectively
(forP.radiata).Forfullsizedspecimenstherelationshipsweremuchweaker.Dowse(2010)found
thatthedensityoffullsizedPinuspatulalumberexplained30%ofthevariationinMOEandwhen
knotpropertieswereadded,thepercentageincreasedto36%.Acousticorvibrationalmethods
performedmuchbetterwiththedefectcontaininglumber.Forinstance,PellerinandRoss(2002)
reportedanumberofstudieswheretheMOEcouldbepredictedwithacoefficientofdetermination
ofmorethan90%.
ThebendingstrengthorMORoflumber,aswithMOE,dependsonseveralwoodproperties.The
resultsofanumberofstudiesonfullsizedlumberhavebeensummarisedfromJohansson(2003)
andGlos(2004)inTable1.FromthefairlylowR2valuesitisclearthatMORisacomplexproperty
thatcannotbepredictedeasily.Althoughfailuresarealmostalwaysassociatedwithgraindistortion
causedbyknots,themeasurableknotpropertiessuchasknotsizeanddistributiondidnotexplain
variationinMORverywell.MOEhasbeenfoundtobethebestsinglepropertyforexplainingthe
variationinMOR.Acombinationofacoustic,densityandknotpropertiesexplainedupto80%ofthe
variationinMOR.
Table1.TherangeofcoefficientofdeterminationvaluesbetweenMORandotherlumberproperties
onindividualboardsfromvariousstudies(compiledfromJohansson,2003,andGlos,2004).
Properties
Coefficientof
determinationrange
(R2)
Density
0.16 0.40
Knotproperties
0.150.35
Annualringwidth
0.200.44
MOE
0.40 0.72
Acousticorvibrationalproperties
0.30 0.55
Knotsanddensitycombined
0.380.60
MOEandknotscombined
0.580.73
Acoustic,knotsanddensitycombined 0.55 0.80
Manymethodshavebeendevelopedinthepasttodeterminesomeofthepropertieslistedabove
fromstandingtreesusingnondestructivemethods.Thesemethodshavebeenextensivelyreviewed
byWesselsetal.(2011b).
Animportantaspecttokeepinmindwhentryingtopredictthemechanicalpropertiesofstructural
lumberisthatthecharacteristicstrengthofagradeisdeterminedbythe5thpercentilestrength
value(Figure1),inotherwordsbyresultsrepresentingtheweakportionofthestrengthdistribution
curve(Madsen,1992).Itisthereforeessentialthatanypropertyandmethodusedinapredictive
studyhastobeanaccuratepredictoroftheweakportionofthestrengthdistributioncurve.For
stiffness,however,designcodesoftenusethemeanMOEvalues,necessitatingpredictionsofthe
fullstiffnessdistribution(i.e.SANS101631,2003andCSAO8601,2001).
71
16
th
5 percentile
Percentageofpieces
14
12
10
8
6
4
2
0
10
15
20
25
30
35
40
45
50
55
60
65
70
ModulusofRupture(MPa)
Figure1.AtypicalhistogramoftheMORoflumberwiththe5thpercentilevalueindicated
InSouthAfricathemeanrotationageofpineplantationsgrownforsawlogproductionhasreduced
from14.1yearsin1983to11.3yearsin2003(CrickmayandAssociates,2004).Thissuggestsa
reductioninmeanharvestingagefromabout28yearsin1983toabout23yearsin2003resultingin
increasedproportionsofjuvenilewood(orcorewood)whenharvested.
3.
Materialsandmethods
3.1
Descriptionofthestudyareaandsamplecompartments
ThestudyareaislocatedalongtheMpumalangaescarpment,SouthAfrica,stretchingfrom2348S
to2549Sandfrom3002Eto3059E.Theareaisgeologicallycomplexwithlargevariationinsoil
characteristics,altitude,precipitationandtemperature.Atotalof17samplecompartmentswere
selectedacrossthearea.Thesamplecompartmentsvariedinagefrom16to20years,situatedat
altitudesvaryingfrom810mto1930mabovesealevel,meanannualprecipitationsvaryingfrom
840mmto1299mmandmeanannualtemperatureswhichrangedfrom13.7Cto19.4C.Site
indicesatbaseage10years(SI10)rangedfrom9.6to19.6(Table2).Thecompartmentsreceived
thenormalcommercialmanagementtreatmentsofweeding,thinningandpruning.
Theworkdescribedinthispaperisoneofseveralstudiesperformedonthesameexperimental
material.PreviousstudiesincludethosereportedbyLouwandScholes(2002,2003and2006)onthe
influenceofsitefactorsonnitrogenmineralizationinforestsoilsoftheMpumalangaescarpment
area,thedevelopmentofamethodtopredicttheknottydefectcore(Munalula,2010)andastudy
whichevaluatedthestructuralgradingparametersforthisparticularresource(DowseandWessels,
2013).Someresultsfromthementionedstudieswereusedasinputsintheresearchdescribedin
thispaper.Acomprehensivedescriptionoftheenvironmentalvariables,geology,soiland
productivityofthesitescanbeviewedinLouwandScholes(2002,2003and2006).
Atotalof126environmental,soil,leafanalysis,andproductivityvariablesweremeasuredor
calculatedforeachcompartment.Thesevariableswerealsoconsideredforthedevelopmentof
predictivemodelsdescribedinthisstudy.Forbrevityssakeonlyvariablesthatwerefoundto
contributesignificantlytothemodelsaredescribedintheResultssection.
72
Table2.Generaldataforeachsamplecompartment,themeandiameteratbreastheight(DBH)and
themeanheightofthetensampletrees.
Sample
compartment
Plantation
Age
(yrs)
A(E66)
B(E28a)
C(G21)
D(D1)
E(D88)
F(E55a)
G(E36c)
H(E22)
I(E5)
J(E15)
K(E35)
L(C22)
M(E3)
N(D74)
O(A1a)
P(D11)
R(J20)
Mean
Nelshoogte
Nelshoogte
Nelshoogte
Uitsoek
Uitsoek
Uitsoek
Uitsoek
Uitsoek
Berlin
Berlin
Berlin
Blyde
Morgenzon
Morgenzon
Morgenzon
Wilgeboom
Wilgeboom
17
19
16
17
17
20
19
17
19
19
16
20
17
19
16
18
19
18
Mean
DBH
(cm)
36
33.8
26.2
32.7
30.2
32.3
31.9
27.6
36.5
37.4
29.1
34
31.4
26.9
27.8
29.4
33.4
31.6
Mean
height
(m)
SiteIndex
atage10
(m)
20.9
21.8
18.4
22.3
18.4
20.1
23.0
20.8
23.8
23.8
18.0
27.0
20.6
16.4
19.0
22.8
24.0
21.2
14.3
14.5
15.5
15.7
15.3
14.6
16.8
16.5
16.7
17.6
16.5
18.5
13.5
9.6
13.6
19.6
16.8
15.6
Mean
Mean
annual
annual
precipitation temperature
(mm)
(C)
1061
16.0
1036
16.1
1057
16.1
944
17.4
942
17.3
1151
13.7
902
14.0
840
14.2
1284
16.1
1082
15.9
1006
17.2
1156
16.1
1015
14.3
997
16.2
862
15.1
1242
19.4
1299
18.5
1052
16.1
3.2Treemeasurements
Ineachofthe17samplecompartments,astratifiedsamplingprocedurebasedontreediameterwas
followedsothatthesampletreesrepresentedtheproductivetimbervolumeavailablefromeach
compartment.Onetreewasrandomlyselectedineachcompartmentfromthefirstquartile(small
diameter),twotreesfromthesecondquartile,threetreesfromthethirdquartileandfourtrees
fromthefourthquartile(largediameters),givingatotaloftensampletreesperplot,thus170trees
fortheentireinvestigation.
The Fakopp TreeSonic microsecond timer was used to calculate the speed of an acoustic
longitudinalstresswaveatbreastheightofeachofthestandingsampletrees.Thisdevicemeasured
thetimeofflightofastresswaveinducedbyahammertapbetweentwoprobes,hammeredinto
theouter10to15mmofthestemonemeterapartaroundbreastheight.Soundvelocityisoften
usedinstudiesasanindirectindicatorofthestiffnessoftheouterwoodintrees.Theuseofacoustic
technologyinwoodstudieshasbeenextensivelyreviewedbyWangandRoss(2002).
Afterfelling,theheightofeachtreeandtheheighttothefirstbranchwhorlweremeasured.Asall
treeshadbeenpruned,branchesgenerallyonlystartedataheightof7maboveground.Thenumber
ofbranchwhorls,maximumbranchdiameterandthebranchangleofonerandomlyselectedbranch
weremeasuredforeverytwometersectionofthetrunkuptoaheightof19meters(Figure2).
Adiscwasremovedfromthestematthebreastheightlocationandlaterusedtoperformringwidth
measurements on the crosssection in order to determine the annual growth rate of each tree
(Munalula,2010).Availablepruningrecordswereusedtodeterminethemaximumknottyordefect
coresizeofallthelogsoriginatingfromtheprunedsectionofeachtree(Figure2).Forcomparative
73
purposes the maximum defect core size of each log was expressed as a percentage of the log
diameter.
defectcore
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
0m
branch angle
yearrings
branchdiameter
1.5m
3m
5m
7m
Figure2.Defectcorereconstructionusingringwidthdataandpruningrecords.Branchproperties
measuredarealsoshown(adaptedfromMunalula,2010).
Two2.1mlonglogswereremovedfromeachtree;onefromtheprunedsectionofthestemat2.3m
heightandonefromtheunprunedsectionat7mheight(Figure3),whichyielded340sawlogs.
3.3Boardmeasurements
Thelogswereprocessedatalocalsawmillintoboardsofcrosssectionaldimensionsof40x120mm,
usingframesawsandacantsawingpattern(Figure3).Onlyboardsprocessedfromthecantwere
usedforthisstudysincetheseboardsrepresentedthefulldiameterofeachlog.Asthesecondary
breakdown saw was fitted with a curvesawing device, the grain direction of the boards was
predominantlyparalleltothelongitudinalaxisofthelog.Atotalof1402boardswereproduced.The
boardswerekilndriedtoatargetmoisturecontentof12%
usingamediumtemperatureschedule.
Boards were numbered based on their position from the pith. Boards containing pith tissue were
marked 0, the two boards on the outer side of the pith boards were numbered 1, the next two
boardswerenumbered2,andsoon.Sometimestherewasonlyonepithcontainingboardinalog
andsometimestwopithcontainingboards(Figure3).
Afterdryingboarddensitieswerecalculatedfromthemassanddimensionaldataofeachboardand
correctedformoisturecontentwhennecessary.Moisturecontentwasmeasuredwitharesistance
moisturemeter.
The ends of the boards were sanded to improve the visibility of the annual rings. The number of
annualringsoneachboardwascountedandnumberedfromthepithoutwards.Thisdatawasused
toestimatethecambialages (mean,maximum andminimum)ofthewoodwhichcomprisedeach
board.Thecambialagewasbasedonringcountsfromthepith.
74
Ringwidthsweremeasuredperpendicularlytothegrowthringboundariesfromthepithtothebark,
using a digital calliper. Measurements were rounded to the nearest 0.1 mm. For each board the
minimum,maximumandmeanringwidthswerecalculated.
Bottom log
Toplog
Board2
Board1
Board0
Board0
Board1
Board2
0m
2.3m
Board2
Board1
Board0
Board1
Board2
4.4m
7m
9.1m
Figure3.Positionoflogsandthenumberingofboardsprocessedfromthecants.
The acoustic resonance frequency of each board was measured using the AGrader Portable
software from Falcon Engineering (http://www.falconengineering.co.nz). The dynamic MOE was
determinedfromthefrequencyandthedensityusingthefollowingrelationship:
MOEdyn=(2lf)2.Equation1
where:
MOEdyn =
Dynamicmodulusofelasticity(MPa),

=
Density(kg/m3),
l
=
Lengthofthetestspecimen(m),
f
=
Frequencyofthetestspecimen(Hz).
Thesamplematerialwasdividedintotwogroupsbasedonboardpositionandarandomallocation
function.Therandomfunctionwasusedtoallocatetheboardsfromthesamepositioninalog(i.e.
thetwonumber1boards)intothetwodifferentgroups.Onegroupwastestedinbendingandthe
otherintension.Tensiontestresultswillnotbediscussedinthispaperasmostboardsfailedatthe
grips which might cause unreliable results. A total of 57 boards had to be discarded due to
breakageswhichoccurredduringprocessingaswellasduetonumberingerrors,whichreducedthe
numberofboardsavailablefordestructivestrengthtestingto1345.
Bending tests wereperformed incompliancewith South African Bureau ofStandards specification

SANS 6122 (2008). Of the 699 boards that were subjected to bending tests 674 yielded useful
results.Sixteencouldnotbetestedsuccessfullyduetoexcessivewarp,whilethetestresultsofnine
boards had to be discarded due to numbering errors. The MOR and MOE for each board were
calculated from the bending tests. The stiffness calculated from this test is referred to as the
MOEstaticasopposedtoMOEdyn,whichwasdeterminedfromacousticmeasurements.
MOEdyn was taken as the dependent variable in developing predictive models for wood stiffness
ratherthanthestaticMOEstatic.MOEdynwasexpectedtogiveabettermeasureofthemeanstiffness
ofa pieceoflumber as itreflectsthe stiffnessoftheentireboardmass,whereas MOEstaticatbest
givesameasurementofthelocalstiffnessofthematerialatthehighlystressedareasofaspecific
test setup. Contrary to MOEstatic,, which was determined on a subsample of boards, MOEdyn
assessmentswereperformedonallavailableboards,whichmadeitpossibletostudytheextentand
patternsofvariationinstiffnessamonglogs,treesandcompartmentsinfarmoredetail.
75
Itshouldbe noted thata number ofproperties were measured onbothboards and/or fromdiscs
fromthetrees.Inthisstudyitwasdeemedpreferabletoremovethelogprocessingstepasasource
oferrorandratheruse,wherepossible,measurementsconductedontheboards.Forinstance,tree
ring widths were measured on both discs from trees and on the individual boards after sawmill
processing. In this case the tree ring widths measured on the boards were used in developing
predictivemodels.Inpractice,toobtainthetreeringwidthsofa standingtree anincrementcore
willhavetobeobtained,ringwidthsmeasured,andindividualtreeringswillhavetoberelatedto
boards from different positions in the stem. By using measurables from boards, inaccuracies in
relatingtreepropertiesatspecificpositionstoboardpropertiesareavoided.(Note:Inthispaperthe
extraction of increment cores from trees were considered nondestructive although, strictly
speaking,itisaminimalinvasivemethod).
3.4Statisticalanalysis
Threedifferentsamplelevelswereusedinthisstudyviz.individualboards(n=1345),trees(n=170),
andcompartments(n=17).Somevariablesweremeasuredonboards,someontreesandsomefor
compartments.Wherevariablesweremeasuredonindividualboards,themeanvalueofaspecific
variableforalltheboardsfromatreewasusedasthetreelevelvalue.Similarly,themeanvalueofa
variableforallthetreesinacompartmentwasusedasthecompartmentlevelvalue.
SimplePearsoncorrelationswereperformedbetweenall143variables.Mostofthevariables
consideredwereenvironmental,soil,leafanalysis,andproductivityvariablesasdescribedinLouw
andScholes(2002,2003,and2006).Toreducethenumberofvariablestoconsiderinthemultiple
regressionanalysis,afactoranalysiswasperformedandtogetherwiththeresultsofthecorrelation
analysis,somevariableswereremovedfromthedatasetusedintheregressionanalysis.Multiple
regressionanalysiswasperformedusingthebestsubsetsinStatistica(www.statsoft.com)to
developpredictivemodels.Mallow'sCpvaluewasusedasthecriterionforchoosingthebestsubset
ofpredictoreffects.Thismeasureofthequalityoffitaddressestheissueofoverfitting.Ittendsto
belessdependentthantheR2valueonthenumberofeffectsinthemodel,andhence,ittendsto
findthebestsubsetthatincludesonlytheimportantpredictorsoftherespectivedependentvariable
andthushelpsestablishingparsimoniousmodels.Ordinarymultipleregressionwaspreferredabove
othermethodssuchasmixedmodelsduetothetechniquesavailabletoselectindependent
variablesfromalargenumberofpossibilities.PredictivemodelsweredevelopedfortheMOEdynand
MORofindividualboards,treesandcompartments.ForindividualboardstheMOEdynandMORwere
usedasdependentvariables.FortreesthemeanMOEdynandmeanMORvalueoftheboardsfroma
treewereusedasthedependentvariables.ForcompartmentsthemeanMOEdynandthe5th
percentileMORvalue(MOR5perc)oftheboardsfromacompartmentwereusedasthedependent
variables.
Sensitivityanalyseswereperformedonthemodelstodeterminetheinfluenceofvarying
independentvariables,oneatatime,onthedependentvariables(Pannel,1997).
4. Results
4.1Correlationanalyses
Thevariablesthatappearinthecorrelationmatrixandinsomeofthepredictivemodelsdeveloped
shownlateroncanbeseeninTable3.
76
ResultsofPearsoncorrelationanalysesforselectedvariablesareshowninTable4.Onlyvariables
whichenteredtheregressionmodelswereincludedinthetable.Correlationcoefficientswithregard
toallthreesamplinglevels,whereapplicable,arepresentedineachcell.Thevaluesinthefirstrow
arethecorrelationcoefficientswhereboardswerethestatistical(experimental)unit,thecorrelation
coefficientsinthesecondrowarebasedontreevaluesandthoseinthelastrowarebasedon
compartments.Forvariablesmeasuredattheboardlevel,themeanvalueforalltheboardsfroma
treewasusedfortreelevelcorrelationsandmodels.Similarly,forvariablesmeasuredatthetree
level,themeanvalueforallthetreesfromacompartmentwasusedforcompartmentlevel
correlationsandmodels.
Table3.Measuredandderivedvariablesusedinpredictivemodels.
Level
Variable
Unit
Description
Board
MOEdyn
MPa
thedynamicmodulusofelasticityforaboard
MOR
MPa
themodulusofruptureforaboard
MOR5perc
MPa
the5thpercentileMORvalueforallboardsfroma
compartment
LogPos
m
themidpointlogheightinthetreefromwhichaboard
wasprocessed
BoardPos
theradialpositionofaboardwith0beingapithboard
(seeFigure3)
RingWidth
mm
themeantreeringwidthinaboard
Density
kg/m3
thedensityofaboardat12%moisturecontent
Tree
DBH
cm
thediameterat1.3mheightofatree
TOF
s
theacousticFakopptimeofflightreadingforatree
betweentwoprobes1mapart
BranchDia
mm
themaximumbranchdiameterinthebottomtwo
metersoftheunprunedsectionofthestemofatree
BranchAngle
degrees
theanglebetweenthebranchandstemofarandomly
selectedbranchinthebottomtwometersofthe
unprunedsectionofthestemofatree
BranchSpacing
thenumberofbranchwhorlsinthebottomtwo
metersoftheunprunedpartofthestemofatree
DefCor
theratioofthemaximumdefectcorediametertolog
diameterforatree
Compartment Topheight
m
themeanheightofthefourlargestdiametertrees
sampledpercompartment
Age
years
theageoftreesinacompartment
SI10
m
thesiteindexordominantheightatindexage10ofa
compartment
MAP
mm
themeanannualprecipitationforacompartment
MayP
mm
themeanprecipitationduringMayforacompartment
MAT
degreesC themeanannualtemperatureforacompartment
JulMinT
degreesC theminimumtemperatureduringJulyfora
compartment
NK
theratioofN:Kforacompartmentdeterminedfroma
leafanalysis8yearspriortofellingforthisstudy
77
Table4.Pearsoncorrelationcoefficientsbetweenselectedboard,tree,andcompartmentvariables.Thetopvalueineachcellistheboardcorrelationvalue,thesecond
valuetreecorrelations,andthebottomvaluecompartmentcorrelations.Statisticallysignificantcorrelationsat0.1,0.05and0.01probabilitylevelsareindicatedbythe
symbols*,**and***respectively.
1.MOEdyn
2.MOR
3.MOR5perc(comp)
4.LogPos(board)
5.BoardPos(board)
6.RingWidth
7.Density
1
1
1
1
,76***
,68***
,89***
,74***
,13***
,55***
,62***
,42***
,71***
,67***
,65***
,87***
1
1
1
,77***
,21***
,37***
,43***
,39***
,57**
,48***
,36***
,80***
,70***
1
,13***
,14***
1
,47***
,30***
,55**
8.Topheight
,06*
,08**
,16*
9.Age
,11***
,24***
10.SI10
11.DBH
12.TOF
13.BranchSpacing
14.BranchDia
15.BranchAngle
16.DefCor
17.MAP
18.MayP
,11**
,28***
10
11
12
13
1
1
1
,46***
,40***
,56**
,11***
,23***
,52**
1
1
1
,22***
,15*
1
1
1
,19***
,20**
,10***
,13***
,24***
1
1
1
,09***
,09***
,08**
,17*
,08*
,16*
,18***
,32***
,52**
,30***
,23***
,49***
,49**
,20***
,24***
,49**
,09***
,15*
,19***
,27***
,45*
,08***
,17**
,51**
,09***
,19**
,18***
,38***
,09***
,06*
,14*
,07**
,07**
,12***
,14***
,28***
,49**
,15***
,30***
,53**
19.MAT
,16***
,27***
,55**
,17***
,28***
,56**
,08*
20.JulMinT
,06*
,09**
21.NK
,19***
,30***
,62***
,14***
,23***
,45*
,11***
,15*
,08**
,08**
,07**
,12***
,23***
,57**
,22***
,28***
,63***
,23***
,30***
,61***
,09***
,11***
,19**
,12***
,15*
,20***
,42***
,79***
,20***
,26***
1
1
1
,59***
,52***
,54**
,70***
,72***
,71***
,35***
,32***
,62***
,19***
,22**
,47*
,40***
,40***
,76***
,23***
,25***
,47*
,18***
,17**
,46*
,22***
,18**
,43***
,41***
,41*
,41***
,38***
,09***
,38***
,45***
,29***
,21**
,45*
,12***
,16***
,14***
,16*
,25***
,24***
,48**
,19***
,15*
,53***
,52***
,52**
,56***
,56***
,58**
,07**
,16***
,19**
,32***
,31***
,54**
,23***
,23***
,43*
1
1
1
,28***
,30***
,21***
,19**
,43*
,24***
,24***
,21***
,23***
,06*
,41***
,40***
,38***
,35***
,29***
,24***
,336***
,351***
,36***
,45***
,42*
,20***
,22***
,45*
,16***
,37***
,31***
,11***
,31***
,26***
,46*
,23***
,18**
,17***
,15*
,38***
,34***
,1***
1
1
1
,19***
,17**
,23***
,26***
,45*
,19***
,18**
,14***
,17**
1
1
1
,06*
,42*
,30***
,34***
,46*
,22***
,21**
14
1
1
1
,25***
,23***
,63***
,11***
15
16
17
18
19
20
21
1
1
1
1
1
1
,95***
,94***
,95***
,50***
,49***
,47*
,57***
,55***
,55**
1
1
1
,42***
,40***
1
1
1
,84***
,85***
,84***
,31***
,29***
1
1
1
,11***
,06*
,26***
,25***
,42*
,09***
,18***
,18**
,06*
,21***
,25***
,42*
1
1
1
,49***
,47***
,47*
,51***
,49***
,50**
,09***
,20***
,19**
,14***
,18***
,16*
,19***
,20**
,43***
,40***
,42*
,06*
1
1
1
78
4.2PredictivemodelsforMOEdynandMOR
MultipleregressionmodelsweredevelopedtopredicttheMOEdynandMORforindividualboards,
treesandcompartments.Thenumberofinputvariableswasverylargeandbyinspectingthe
Pearsoncorrelationcoefficientsanddoingafactoranalysis,variableswhichwereconsideredless
influentialwereexcludedfromtheregressionanalyses.
Asmentionedpreviously,the5thpercentileMORvalueofaboardgradeisusedinthedesignof
structures.Sincetherewerenotenoughboardsfromindividualtreestodeterminea5thpercentile
valuethemeanMORvalueforeachtreewasusedinstead.Treesthatyieldedfewerthan5boardsin
total,orfewerthan2boardsperlogsuitablefortesting,werediscarded.Asaresultthedataofonly
142treesoutof170treescouldbeconsideredforthetreelevelanalysis.
ForeachcompartmentthemeanMOEdynvaluesforalltheboardsfromthatcompartmentwere
calculatedandusedasthedependentvariableinthepredictivemodel.The5thpercentileMORvalue
ofalltheboardsfromacompartmentwasusedastheotherdependentvariable(MOR5perc).Dueto
thelimitednumberofcompartments(n=17)amaximumoffiveindependentvariableswereallowed
forthecompartmentlevelmodelstoavoidoverparameterisation.
Thecoefficientofdetermination(R2)andtheparametersforthemodelsdevelopedarepresentedin
Table5.Figures4to7showthepredictedvs.observedvaluesfortheMOEdynandMOR5percmodels.
Table5.Parametersandcoefficientsofdetermination(R2)ofthepredictivemodelsdevelopedfor
MOEdynandMORonindividualboard,treeandcompartmentlevels.Allmodelsweresignificantat
the0.001probabilitylevel.Parametersmarkedwith*,**,***weresignificantatthe0.1,0.05and
0.001probabilitylevels,respectively.
Coefficientof
determination(R2)
Parameters:
Intercept
LogPos
BoardPos
RingWidth
Density
SI10
TOF
BranchDia
BranchAngle
BranchSpacing
DefCor
MayP
MAT
JulMinT
NK
Topheight(m)
Age(years)
Board
MOEdyn
Tree
MOR
MOEdyn
Compartment
MOR
0.682
0.402
0.600
0.421
1870.6
5.08
0.72***
2.83***
0.69***
0.08***
1.89***
0.07**
0.13***
290.9
8.16
90.2**
21.4***
150.3***
11.0**
11.5**
1.04**
0.05**
1.03***
803.4***
152.4***
24.4***
145.3***
11.5***
12.6***
15.1***
2679.5***
104.8***
77.0**
13.26**
0.49**
734.2***
1.34***
1.66**
3544.7***
91.5**
282.2**
249.5**
895.1***
MOEdyn
0.952
*
6296.0
***
315.4
18.2***
43.2*
18.3**
0.15***
9.09**
2.98**
MOR5perc
0.798
1.69***
0.64**
0.28***
0.90**
81.2**
80
Dependent variable: MOE dyn per board (MPa)

18000
16000
Predicted Values
14000
12000
10000
8000
6000
4000
2000
2000
4000
6000
8000
10000
12000
14000
16000
18000
Observed Values
Figure4.Thepredictedvs.observedvaluesfortheboardlevelMOEdynand95%confidencelimits
(seemodelparametersinTable5).Thebrokenlineindicatesthe1:1relationship.
Dependent v ariable: Mean MOEdyn per tree (MPa)

12000
11500
11000
10500
Predicted Values
10000
9500
9000
8500
8000
7500
7000
6500
6000
5500
6000
7000
8000
9000
10000
11000
12000
Observ ed Values
Figure5.Thepredictedvs.observedvaluesforthetreelevelMOEdynmodeland95%confidence
limits(seemodelparametersinTable5).Thebrokenlineindicatesthe1:1relationship.
81
Dependent variable: Mean MOE dyn per compartment (MPa)

9600
9400
9200
9000
Predicted Values
8800
8600
8400
8200
8000
7800
7600
7400
7200
7200 7400 7600 7800 8000 8200 8400 8600 8800 9000 9200 9400 9600
Observed Values
Figure6.Thepredictedvs.observedvaluesforthecompartmentlevelMOEdynand95%confidence
limits(seemodelparametersinTable5).Thebrokenlineindicatesthe1:1relationship.
Dependent variable: MOR5perc per compartment (MPa)

19
18
17
Predicted Values
16
15
14
13
12
11
10
9
9
10
11
12
13
14
15
16
17
18
19
Observed Values
Figure7.Thepredictedvs.observedvaluesforthecompartmentlevelMOR5percmodeland95%
confidencelimits(seemodelparametersinTable5).Thebrokenlineindicatesthe1:1relationship.
4.3Sensitivityanalyses
Table6showstheresultsofsensitivityanalysesperformedonthemodelsdevelopedforMOEdynand
MOR.Fortheboardandtreelevelmodelseachindependentvariableinthemodelswasvariedfrom
the5thpercentileobservedvaluetothe95thpercentileobservedvalue.Forthecompartmentlevel
82
modelstheindependentvariableswerevariedfromtheminimumvaluetothemaximumvalue.The
effectofthesechangesonthedependentvariablewasexpressedastheindependentvariables
influence(%).Theinfluenceistherelativeeffectthatthechangeintheindividualvariablehas,
comparedtothetotalrangeinMOEdynandMORvalues.Forexample,whenRingWidthintheboard
levelMOEdynmodelwaschangedfromthe5thpercentileobservedvaluetothe95thpercentile
observedvalue,andalltheotherparameterswerekeptconstantattheirmeanobservedvalues,the
changeinMOEdynwas13.7%ofthetotalchangeinMOEdynpossible(Table6).
Table6.ResultsofsensitivityanalysesonthepredictivemodelsfromTable5.Independentvariables
oftreeandboardlevelmodelswerechangedfromthe5thtothe95thpercentileobservedvalueand
forcompartmentlevelmodelsitwaschangedfromtheminimumtomaximumobservedvalues.The
influence(%)istherelativeeffectthatthechangeintheindividualvariablehascomparedtothe
rangeinMOEdynandMOR.
Independentvariables
LogPos
BoardPos
RingWidth
Density
SI10
TOF
BranchDia
BranchAngle
BranchSpacing
DefCor
MayP
MAT
JulMinT
NK
Topheight
Age
Influence(%)
Tree
Board
MOEdyn
12.4%
13.7%
22.0%
6.8%
4.4%
4.2%
4.1%
8.6%
8.7%
3.4%
MOR
4.3%
7.1%
10.2%
12.2%
14.5%
4.6%
7.1%
7.0%
6.6%
11.7%
18.1%
8.4%
MOEdyn
4.5%
15.0%
12.3%
4.6%
4.1%
12.3%
8.0%
13.0%
11.2%
14.8%
MOR
20.2%
14.5%
32.7%
21.5%
12.2%
Compartment
MOEdyn
MOR5perc
32.2%
30.0%
12.5%
15.9%
19.1%
28.2%
30.6%
20.2%
20.9%
9.4%
5. Discussion
ThepredictivemodelsdevelopedforMOEdynhadmoderatecoefficientsofdeterminationatsawn
boardaswellastreelevel(R2=0.682andR2=0.600respectively,seeTable5)butahighcoefficient
ofdeterminationatcompartmentlevel(R2=0.952).Comparedtootherstudieswherethemodulus
ofelasticityoflumberatboardlevelwaspredictedfromstandingtrees,theR2valuefoundinthis
studywasrelativelyhigh.Forinstance,IkedaandArima(2000)foundanR2valueof0.410inthecase
ofmatureSugitrees,Wagneretal(2003)andChestnut(2009)foundR2valuesof0.591and0.174
respectivelyinstudiesonDouglasfirtreesandBier(1985)foundanR2valueof0.480onradiatapine
trees(allthesewereboardlevelpredictions).
TreelevelmodelsforpredictingMOEandMORdevelopedbyLiuetal.(2007)for90100yearold
blacksprucetrees,usingbothstandandtreecharacteristicsasinputs,showedR2valuesof0.65and
83
0.68forMOEandMORrespectively.Huang(2000)developedatreelevelmodelforloblollypineand
obtainedanR2valueof0.51andLaunayetal.(2000)foundanR2valueof0.29forDouglasfirand
larchtrees.MostofthestudiesonlumberdidnotconsidertheMORofthelumber.Bier(1985)
obtainedR2valuesofbetween0.30and0.37predictingthemeanandminimumMORoflumber
fromP.radiatatrees.TheauthorsarenotawareofstudieswhereanMOEmodelforlumberwas
developedatacompartmentlevel.
ThehighcoefficientofdeterminationforthecompartmentlevelmodelforMOEdyn(R2=0.952)was
surprisingwhencomparedtotheboardandtreelevelmodelsofthisandotherstudies,wherethe
R2valuesweregenerallylowerthan0.7.Asthiswasthefirstcompartmentlevelstudyofthistypeof
whichweareawareof,wecannotcompareitwithotherresults.ThehighR2valuewasprobablydue
tothewithintreeandwithincompartmentvariabilityinpropertiesthatgetaveragedforallthe
boardsfromacompartment.Similarly,individualprocessingdecisionswhichinfluenced,for
instance,thepositionofknotswithinaboardwillalsolargelybecancelledatacompartmentlevel
duetothisaveragingeffect.
AlthoughasignificantcorrelationexistsbetweenthevaluesofboththeboardandtreelevelMOEdyn
models,itdeviatessubstantiallyfromthe1:1relationship(Figures4and5).Thisisanindicationofa
strongtendencytooverpredictandunderpredictatlowandhighMOEdynvaluesrespectively.
InthemultipleregressionmodelsforMOEdyn,theindependentvariablesDensity,RingWidth,SI10,
andTOFappearinthemodelsatallthreesamplinglevels(Table5).Densitywasthebestsingle
independentvariableforpredictingMOEdynwithaPearsoncorrelationofr=0.67forboards,r=0.65
fortreesandr=0.87forcompartments(Table4).Forboards,thecorrelationvaluecorresponds
roughlytothosefoundinvariousstructuralgradingstudiesreviewedextensivelybyJohansson
(2003).ThesensitivityanalysisoftheMOEdynmodelsshowedthatDensitywasthemostinfluential
parameteratboardandtreelevelandthesecondmostinfluentialatthecompartmentlevel(Table
6).
Densityhaslongbeenconsideredasoneofthemostimportantwoodproperties,ifnotthemost
important,intermsofitseffectonthequalityofsolidwoodproducts.Forinstance,intheir
extensivereviewZobelandVanBuijtenen(1989)concludedtherefore,specificgravitylargely
determinesthevalueandutilityofwoodandovershadowstheimportanceofotherwood
properties.Thisviewhaslaterbeenchallengedbymanyauthorsi.e.Cherryetal.(2008)who
questionedwhethertreebreedersshouldselectsolelyforwooddensitybecause(i)thetimeof
flightoracousticvelocityisabetterpredictorofstiffnessthandensity,(ii)wooddensityhasa
negativecorrelationwithgrowth,and(iii)densityisexpensivetomeasureonincrementcores.Cave
andWalker(1994)arguedthatonlymicrofibrilangle,andnotdensity,canexplainthelargeincrease
ofstiffnessoverthefirst30yearsofgrowthoffastgrownplantationsoftwoods.Sincemicrofibril
anglewasnotmeasuredinthisstudyitwasnotpossibletoassessitsinfluenceonwoodstiffness(or
MOE)relativetothatofdensity.
Inthecaseofallthreemodellevels,acousticassessments(TOF)carriedoutontheouterwoodofthe
standingtreesweremuchlessinfluentialthanDensity.Thisislargelyduetothefactthatvelocity
assessmentsonstandingtreesonlyserveasameasureoftheouterwoodstiffnessofthetreestem.
ThelesserinfluenceofTOFwasconfirmedbytheresultsofthesensitivityanalysesontheMOEdyn
models,whichshowedthattheinfluenceofTOFwasverysmallattheboardandtreelevel(<5%)
andonlymoderatelyinfluentialatcompartmentlevel(Table6).
RingWidthshowedaninverserelationshipwithMOEdynwithcorrelationsof0.62atboardlevel,0.42
attreeleveland0.71atcompartmentlevel(Table4),suggestingthatMOEdyndecreaseswith
84
increasin
nggrowthraate.Atcompartmentleve
elRingWidth
hwasthemo
ostinfluentiaalvariable(TTable6),
followed
dbyDensity..Attreeleveeltheinfluen
nceofRingW
Widthinthem
modelwasno
oticeablylow
wer.As
withDen
nsity,RingW
Widthcanbem
measuredon
nincrementcoresfromsstandingtree
es.Itiscorreelatedto
theDBH
Hoftrees,altthoughitwasslightlysurrprisingtoseeethemoderratetolowccorrelationof0.49
between
nRingWidthandDBHatatreelevel.Thesumofalltreeringwidthsatap
particularheeight
levelinaastemwillin
nfactbetheunderbarkd
diameteroftthestemattthatpoint.TThepoorcorrrelation
mightbeeduetothefactthatsom
metreeringsoccurinup
ptoinasmanyasthreeb
boardsfromalog
whileotthers,suchastheouterrrings,mightn
notbepreseentinanyoftheboards.RingWidthw
willthus
beweightedtoward
dstheyoungeeryearringsspresentinsseveralboard
ds.Othercau
usesfortheweak
ompartmenttswereofslightlydiffere
entagesandalso
correlationmightinccludethefacctthattheco
mlogsfromttwodifferentheightsinaatree.
thatboaardswerereccoveredfrom
weremoderatteandsignifficantnegativvecorrelatio
onsbetweenDensityandRingWidthaatall
Therew
threelevvels(Table4
4).Inconifersslargeringw
widthsareofftenassociatedwithsmaallerlatewood
percentaageandsubssequentlyallowerdensitty.However,sinceRingW
WidthandDeensitybothappearin
themod
delsitclearlyysuggeststhattheeffecttofringwidtthonMOEdynnisnotpurellyduetoitseffect
ondensity.
older,morematureringsswillbe
Italsoneedstobeemphasizedtthatsmallerringwidthseensurethato
utatorcloseetothepith.Sincemicroffibrilanglesusuallydecreasessharplyfrom
presentinboardscu
ncreases,suchboardswillbecharactterisedbysm
mallermicroffibril
thepithoutwards,aanddensityin
boardswillth
hustendtob
bestiffer.Thering
anglesandhigherdeensity(CaveandWalker,,1994)andb
wasalsorelatedtotreesh
hape,whichmayaffectm
mechanicalp
properties.Th
heroleoftreeeshape
widthw
isdiscusssedlater.
Thesiteindexatyeaar10(SI10)cconsistentlyaappearsinallthemodelssforbothMOEdynandMOR.This
nedintermssofthe
suggestssanindirecteffectoftreeeheightonMOEandMOR,whichcaanbeexplain
resistanceoftreesto
obucklingan
ndbendingffailure,asillu
ustratedinFigure8.
Figure8.B
Basicmodelsforatreelo
oadedbyselffweight(leftt)andwindload(right).
oadedbyitsownmassbeehavesinassimilarwayaasarod,fixed
datoneend
d,undercom
mpressive
Atreelo
loading(Figure8).U
UsingtheEuleerformulafo
orbucklingo
ofarodfixed
datoneend,theMOEreequired
ngfailureyieldsEquation
n2(derivedffromEulerfo
ormula,Hibb
beler,2005):
towithsstandbucklin
.Equation2
hoftherod(m),Pisthecompressiveeloadontheerod(N),andristherad
diusof
whereListhelength
therod(m).
85
oadingcanbemodelledaasacantileveeredrodsub
bjectedtoap
pointload(FFigure8).
Atreeunderwindlo
usingthebendingstressformulasho
owsthattherodrequiressthefollowin
ng
Basicstaaticstheoryu
bendinggstrengthorMORtoavo
oidabendinggfailure(derrivedfrombeendingstresssformula,Hibbeler,
2005):
...Eq
quation3
mulaitisclearthattherattioL2/r4deteerminestheMOErequireedforbuckling
FromtheEulerform
dingstressfo
ormulashow
wsthattheL//r3ratiodete
erminestheMORrequireedto
resistance.Thebend
ndwindload
d.Treeswithincreasedsllenderness(i.e.treesthaataretallinrrelationtoth
heir
withstan
diameteer)thereforerequirewoo
odthatshigherinMOEaandMORinordertoincrreasetheir
resistancetobucklin
ngunderitso
ownweightorbreakageduetoexcessivewindlo
oads.Therad
diusor
othepowerfourandthrreerespectivvelyinthetw
woformulas,,isalsorelatively
diameteer,whichisto
moreinffluentialthantheheightofthetree.Apositiveco
orrelationbeetweentreeslendernessand
MOEhasbeenfound
dinseveralsstudiesonottherspecies(i.e.Lasserreeetal.,2005
5;Wattetal..,2006;
Rothetal.,2007;Lassserreetal.2009).
OEdyn,andwaas
InthissttudystemsleendernessattfellingagewasveryweeaklycorrelattedwithMO
thereforrenotincludedinanyofthemodels.However,b
bothRingWid
dth,whichisrelatedtoth
he
diameteerofatreeandSI10,whichisrelatedtotreeheigght,appearin
nallthemod
delsforMOEEdynand
MOR.In
nfact,thesevvariablesapp
pearinallmodels,stresssingtheimpo
ortantinfluencetheybotthhave
onwood
dstrengthan
ndstiffness.AsRingWidtthhasanegaativeparame
etervalueinthemodelsandSI10
positiveparameters,thesetwovvariableswill(combined))functionsim
milartoasleendernessratioin
dels.Thebigdifferencew
withtheslendernessratio
oatfellingaggeisthefacttthatSI10w
wasthe
themod
dominan
ntheightataage10,whileeRingWidth isweightedtowardsgro
owthduringttheearlieryears.It
appearsasiftheinclusionandin
nfluenceofR
RingWidthan
ndSI10intheemodelsforrMOEdynand
dMOR
might,aatleastpartiaally,duetoittseffectonsstemformorslendernesssduringearliergrowthinthe
trees.Assonewouldexpectfrom
mtheEulerfo
ormulaandtthebendingsstressformu
ula,RingWidtth,
whichisrelatedtoraadius,wasm
moreinfluenttialinthemo
odelsthanSII10,whichisrelatedtoleength
(Table6).
on(2003)haasfoundindestructivebe
endingandttensiontestss,thatfailureewasalmostt
Johansso
exclusiveelycausedb
byknots.Inth
hisstudyfaillurewasalso
ousuallyinitiatedaround
dknots.Insp
piteof
themarkedinfluencceofknotsonMOR,theccorrelationsbetweenMO
ORandMOR
R5percwithbranching
anchDia,Bra
anchAngle,BranchSpacin
ngandDefCo
orwereeitheernotstatistically
variablessuchasBra
owever,bran
nchvariablessneverthelesssappearedina
significantorweaklyysignificant((Table4).Ho
DefCorwasp
presentinallltheboardaandtreeleveelmodels.BrranchAngleaand
numberofmodels.D
MOR5percmod
delwherethesensitivitystudy
BranchSSpacingwereepresentintthecompartmentlevelM
showsth
hattogetherritaccountsforabout41
1%influenceinthemodeel(Table6).IItwassomew
what
surprisin
ngthatBrancchDiawasno
otincludedinthelatterm
modelasitaappearedinaalltheotherMOR
modelsandwillobviouslyberelatedtoknottsize.Ataco
ompartmentlevelitmeansthatthevvisible
diametersattthebottom2mofthecrowndidnotaddtotheMOR5percmo
odelwhere
branchd
BranchA
AngleandBra
anchSpacing
gwerealread
dyincluded.
werelesspro
ominentinth
heboard
Therearreanumberofpossiblerreasonswhyybranchcharracteristicsw
andtreeelevelmode
elsthanexpeected.Inthefirstplaceth
heinfluenceofaknotonthebendingg
strength
hofaboardiisstronglydeependenton
nthelocation
noftheknott.Forinstancce,theeffectof
knotssittuatedatornearhighsttressedareas,suchasatthebottomedgeandclo
osetotheceentre
(lengthw
wise)ofaboaard,isknowntobefarm
morepronouncedonthebendingstreengthofabo
oard
thankno
otssituatedelsewherein
ntheboard.Asknotlocaationwasaraandomvariaableitmightdilute
86
theeffectofthemeasuredbranchcharacteristics.Sincethebottomsectionsofthetreeswere
pruned,measurementofbranchingcharacteristicscouldonlybeperformedonthesecondlogs
whichrepresentedthe79mheightsectionofthetrees.Thebranchingcharacteristicsofthebottom
logwerethereforenotdirectlyreflectedbyanyofthevariablesexceptforDefCor.
Asexpected,branchingcharacteristicswerenotveryinfluentialintheMOEdynmodelswhereknots
playarelativelylessimportantrole.
Theratioofnitrogentopotassium(NK),whichwasdeterminedforeachstandbyleafanalysiseight
yearspriortosamplingforthisstudy,appearsinanumberofmodels,andhasamoderateinfluence
onMOEdynandMORassuggestedbytheoutcomeofthesensitivityanalysis.Thisvariablewasalso
foundtocorrelatepositivelywithMOEdynandMORatboard,treeandcompartmentlevels(Table4).
Nitrogenisrequiredfortreegrowthasitisaconstituentofproteins,nucleicacidsandseveralother
importantsubstances,whilepotassiumisessentialforcelldivisionanddevelopment.Itisvery
mobileandsoluble,principallybeingusedinyoungtissues(Acheetal.,2010;Barreletetal.,2006).
ThefactthattheNKparameterwaspositiveinallthemodelsmaysuggestthattreesgrownin
compartmentscharacterisedbyacombinationofhighnitrogenandlowpotassiumlevelstendto
producewoodwhichhashigherMOEdynandMOR.
LouwandScholes(2003),inastudyusingthesamefoliardata,concludedthattoolittleNmightbea
growthlimitingnutrientonthesesitesandageclassesforPinuspatula.Theyobservedlarge
variationinthedifferentfoliarelementconcentrationsbetweenseasons.HigherconcentrationsofN
werefoundintheneedlesduringtheactivegrowthseason,whilehigherconcentrationsofKwere
observedduringthedormantseason.
Barreletetal.(2006),analysingseasonalprofilesofKinNorwaysprucewood,foundthatKseemsto
accumulatemainlyinthelatewood,inotherwords,mainlyduringdormantseasonalgrowth.Acheet
al.(2010),inareviewoftheeffectofpotassiumonwoodformationinpoplar,concludedthat
potassiumpeaksinthecambialregionduringtheactivegrowthperiodoftreeswhereitisessential
forcelldivisionandelongation.
Inthisstudy,itcanbearguedthatiftheeffectofNKonwoodformationwaspurelyafunctionofthe
rateofgrowth,itwouldnothaveaddedtotheMOEdynandMORmodelswhereRingWidthwas
alreadyincluded.DuetothestrongseasonalvariationthatexistsforespeciallyKindifferentpartsof
trees,itcanbehypothesized,thattheNKratiomightberelatedtotheearlywoodtolatewood
formationswitchinthewood.ThiseffectpossiblyexplainswhybothNKandRingWidthsignificantly
contributedtothemodels.
Earlywoodandlatewoodwerenotassessedquantitativelyinthisstudybutavisual
inspection/assessmentandcomparisonofdiscsfromthecompartmentwiththehighestNKratio
versusthecompartmentwiththelowestNKratiowasdone.Theannualringsofthediscstakenfrom
thecompartmentcharacterisedbyahighNKratio(compartmentF)clearlyexhibitedlargerlatewood
percentagesthanthosefromthelowNKcompartment(compartmentL).Giventhatlatewoodhas
higherstiffnessthanearlywood(Wattetal.2006),theeffectwouldbeahigheroverallstiffnessof
thewood,whichwouldsupportourhypothesis.Amoredetailedstudywouldberequiredto
thoroughlytestthishypothesis.
ApartfromtheinsightgainedbydevelopingmodelsfortheflexuralpropertiesofPinuspatulagrown
inSouthAfrica,themodelsitselfhavepracticalsignificance.Thetreelevelmodelsmightbeusefulin
treeimprovementprogrammesforselectingtreeswithimprovedstrengthandstiffness.Experience
87
inindustry,supportedbyrecentstudiesbyDowseandWessels(2013),haveshownthatlowMOEof
sawnlumberisbecomingmoreandmoreproblematicinSouthAfricanstructuraltimber.Thehigh
predictivepowershownbythecompartmentlevelmodel,whichexplained95%ofvariabilityin
MOEdyn,provedthatthismodelcouldbeextremelyusefulinpractice,asitwouldenablethe
identificationofcompartmentsthatareexpectedtoyieldasignificantproportionofstructuralgrade
lumberintheiroutputspriortoharvesting.Compartmentsthatdonotfallintothiscategorycanbe
earmarkedeitherforotherpurposes,suchasindustrialorappearancegradelumber,wherestrength
andstiffnessareoflessimportance,orthedecisioncanbemadetoallowthecompartmentstogrow
olderinordertoincreasetheproportionoflumbersuitableforstructuralpurposesatfinalharvest.
Therelativelystronginfluenceoftreeslendernessduringearlygrowthmightprovideanideal
opportunitytoimprovewoodstiffnessduringjuvenilegrowththroughsilvicultureofourcurrent
geneticallyadvancedpines.Forinstance,plantingatcloserspacingwillresultinthedevelopmentof
moreslenderandlessexposedtrees,producingwoodthatmightbeofhigherstiffness.This
tendencyhasalreadybeenconfirmedbystudiesonotherspecies(Lasserreetal.,2005;Wattetal.,
2006;Rothetal.,2007;Lasserreetal.2009).
Furtherstudiesinvolvingsamplestakenfromsuitablespacingtrials,arethereforestrongly
recommended,aspositiveresultswillprovidearelativelysimple,easytoapply,silvicultural
technique,capableofenhancingtheformationofcorewoodwithincreasedstiffness.Findinga
shorttermsolutiontowardsincreasedcorewoodstiffnessisofutmostimportance,asselective
breedingforincreasedcorewoodstiffnessinsawtimber,whichiscurrentlyactivelypursuedinmany
breedingprogrammesinSouthAfrica,hasquiteamorelongtermhorizon.
6. Conclusions
ItwaspossibletodevelopmultipleregressionmodelstopredictMOEdynandMORofdefect
containinglumberataboard,treeandcompartmentlevelforplantationgrownPinuspatulafrom
theMpumalangaescarpment,SouthAfrica.Themodelsdevelopedwerecapableofexplaining68%,
60%and95%ofthevariabilityinMOEdynatindividualboard,treeandcompartmentlevel
respectively.Thebestmodelsdevelopedexplained40%and42%ofvariabilityinMORataboard
andtreelevelrespectively.Atcompartmentlevelthebestmodelexplained80%ofthevariabilityin
MOR5perc.
Resultsofsensitivityanalysesshowedthatsiteindexatbaseageof10years,acoustictimeofflight,
wooddensityandringwidthwereinfluentialvariablesintheMOEdynmodels.IntheMOR5percmodel
atcompartmentlevelsiteindexatbaseage10years,branchangle,branchspacingandringwidth
wereinfluentialvariables.
Themodelsdevelopedalsosuggestthattreeslendernessduringearlygrowthmayplayanimportant
roleindeterminingtheMOEdynandMORofboardsoriginatingfromthecorewoodzone,whichis
consistentwiththeEulerbucklingandthebendingstresstheories.
88
References
Ache,P.,Fromm,J.,HedrichR.,2010.Potassiumdependentwoodformationinpoplar:seasonal
aspectsandenvironmentallimitations.PlantBiology12:259267.
Barrelet,T.,Ulrich,A.,Rennenberg,H.,Krhenbhl.,2006.Seasonalprofilesofsulphur,phosphorus,
andpotassiuminNorwaysprucewood.PlantBiology8:462469.
Bier,H.,1985.Bendingpropertiesofstructuraltimberfroma28yearoldstandofNewZealand
Pinusradiata.NewZealandJ.ofForSci.15(2):23350.
Burdzik,W.,2004.GradeverificationofSApine.S.Afr.ForestryJ202:2127.
Carle,J.,Holmgren,P.,2008.Woodfromplantedforests.Aglobaloutlook20052030.ForestProd.J.
58(12):618.
Carle,J.,Vuorinen,P.,DelLungo,A.,2002.Statusandtrendsinglobalforestplantation
development.ForestProd.J.52(7):213.
Cave,I.D.,Walker,J.C.F.,1994.Stiffnessofwoodinfastgrownplantationsoftwoods:theinfluence
ofmicrofibrilangle.ForestProd.J.44(5):4349.
Cherry,M.L.,Vikram,V.,Briggs,D.,Cress,D.W.,Howe,G.T.,2008.Geneticvariationindirectand
indirectmeasuresofwoodstiffnessincoastalDouglasfir.Can.J.ofForestRes.38(9):24762486.
Chestnut,I.M.,2009.NondestructiveassessmentofstandingDouglasfirtreesandlogstoestimate
lumberquality.MScthesis,UniversityofIdaho.
Cown,D.J.,2006.Woodqualityinstandingtimberevolutionofassessmentmethodsinplantations.
In:Kurjatko,S.,Kdela,J.andR.Lagaa(eds.)2006.Proceedingsofthe5thIUFROSymposium
WoodStructureandProperties06,September36,SliaSielnica,Slovakia.Organisedjointlyby
theFacultyofWoodSciencesandTechnologyoftheTechnicalUniversityofZvolenandtheIUFRO
CrickmayandAssociates.,2004.Supplyanddemandstudyofsoftwoodsawlogandsawtimberin
SouthAfrica.Copyobtainablefrommandy@crickmay.co.za.
CrickmayandAssociates.,2011.SouthAfricanLumberIndexforOctober2011.Copyobtainablefrom
CSAO8601.,2001.EngineeringDesigninWood:CanadianStandardsAssociation.
DAFF.,2009.ReportoncommercialtimberresourcesandprimaryroundwoodprocessinginSouth
Dowse,G.P.,2010.SelectedmechanicalpropertiesandthestructuralgradingofyoungPinuspatula
sawntimber.MScForthesis.DepartmentofForestandWoodScience,StellenboschUniversity.
Dowse,G.P.,Wessels,C.B.,2013.ThestructuralgradingofyoungSouthAfricangrownPinuspatula
sawntimber.SouthernForests75(1):717.
89
Evans,R.,Ilic,J.,2001.Rapidpredictionofwoodstiffnessfrommicrofibrilangleanddensity.Forest
Prod.J.51(3):5357.
FAO.,2013.Plantedforests.http://www.fao.org/forestry/plantedforests/en/.Accessed
26/06/2013.
Glos,P.,2004.Newgradingmethods.ProceedingsofCOSTE29Symposioum,FlorenceOctober27
29.CNRIvalsa,SanMicheleallAdige,Italy,p18.
Grabianowski,M.,Manley,B.,Walker,J.C.F.,2004.Impactofstockingandexposureonouterwood
acousticpropertiesofPinusRadiatainEyrewellForest.NZJ.Forestry,August2004.
Hibbeler,R.C.,2005.Mechanicsofmaterials.SIsecondedition.PearsonPrenticeHall.871pp.
Hodge,G.R.,Dvorak,W.S.,2012.GrowthpotentialandgeneticparametersoffourMesoamerican
pinesplantedintheSouthernHemisphere.SouthernForests74(1):2749.
Huang,CH.,2000.Predictinglumberstiffnessofstandingtrees.In:Divos,F.(ed).2000.Proceedings
of12thInternationalsymposiumonnondestructivetestingofwood.September1315.Universityof
WesternHungary,Sopron,Hungary:173180.
Ikeda,K.,Arima,T.,2000.Qualityevaluationofstandingtreesbystresswavepropagationmethod
anditsapplicationII:Evaluationofsugistandsandapplicationtoproductionofsugi(Cryptomeria
japonicaD.Don)structuralsquaredsawntimber.JournaloftheJapanWoodResearchSociety46(3):
189196.
Ivkovi,M.,Gapare,W.G.,Abarquez,A.,Ilic,J.,Powell,M.B.,Wu,H.X.,2009.Predictionofwood
stiffness,strength,andshrinkageinjuvenilewoodofradiatapine.WoodSci.Technol.43:237257.
Johansson,CJ.,2003.Gradingoftimberwithrespecttomechanicalproperties.InThelandersson,S.,
andJ.L.Larsen.(eds).2003.TimberEngineering.JohnWileyandSonsLtd.446pp.
Lasserre,JP.,Mason,E.G.,Watt,M.S.,2005.TheeffectsofgenotypeandspacingonPinusradiataD.
Doncorewoodstiffnessinan11yearoldexperiment.ForestEcol.Manag.205:375383.
Lasserre, JP., Mason, E.G., Watt, M.S., Moore, J.R., 2009. Influence of initial planting spacing and
genotype on microfibril angle, wood density, fibre properties and modulus of elasticity in Pinus
radiataD.Doncorewood.ForestEcol.Manag.258:19241931.
Launay,J.,Rozenberg,P.,Paques,L.,Dewitte,JM.,2000.Anewexperimentaldeviceforrapid
measurementofthetrunkequivalentmodulusofelasticityonstandingtrees.Ann.For.Sci.57:351
359.
Launay,J.,Ivkovich,M.,Paques,L.,Bastien,C.,Higelin,P.,Rozenberg,P.,2002.Rapidmeasurement
oftrunkMOEonstandingtreesusingRigidimeter.Ann.For.Sci.59:465469.
Lindstrm,H.,Harris,P.,Nakada,R.,2002.Methodsformeasuringstiffnessofyoungtrees.Holzals
RohWerkst.60:165174.
Liu,C.,Zhang,S.Y.,Cloutier,A.,Rycabel,T.,2007.Modelinglumberbendingstiffnessandstrengthin
naturalblacksprucestandusingstandandtreecharacteristics.ForestEcol.Manag.242:648655.
90
Louw,J.H.,Scholes,M.,2002.Theinfluenceofsitefactorsonnitrogenmineralizationinforestsoils
oftheMpumalangaescarpmentarea:SouthAfrica.SouthAfricanFor.J.193:4763
Louw,J.H.,Scholes,M.,2003.FoliarnutrientlevelsasindicatorsofsitequalityforPinuspatulainthe
Mpumalangaescarpmentarea.SouthAfricanFor.J.193:4763
Louw,J.H.,Scholes,M.,2006.SiteindexfunctionsusingsitedescriptorsforPinuspatulaplantations
inSouthAfrica.ForestEcol.Manag.225:94103.
Madsen,B.,1992.Structuralbehaviouroftimber.TimberEngineeringLtd,NorthVancouver,Canada.
405pp.
Malan,F.S.2010.CorewoodinSouthAfricanpine:necessityandopportunitiesforimprovement.
SouthernForests72(2):99105.
Matheson,A.C.,Dickson,R.L.,Spencer,D.J.,Joe,B.,Ilic,J.,2002.AcousticsegregationofPinus
radiatalogsaccordingtostiffness.Ann.For.Sci.59:471477.
Megraw,R.,Bremer,D.,Leaf,G.,Roers,J.,1999.StiffnessinLoblollypineasafunctionofring
positionandheight,anditsrelationshiptomicrofibrilangleandspecificgravity.InThirdWorkshop,
Connectionbetweensilvicultureandwoodqualitythroughmodellingapproachesandsimulation
software,IUFROWPS5.0104,LaLondeLesMaures,France,Sept512,1999.
Munalula,F.,2010.Amethodforthenondestructivedeterminationoftheknottycoresizesof
standingPinuspatulatrees,basedonringwidthassessmentsatbreastheightandthepruning
history.MSc(WoodProductsScience)thesis.DepartmentofForestandWoodScience,Stellenbosch
University.
Owen,D.L.,VanderZel,D.W.,2000.Trees,forestsandplantationsinSouthernAfrica.In:Vander
Zel,D.L.(ed).SouthAfricanForestryHandbook.SouthernAfricanInstituteofForestry.
Pannell,D.J.,1997.Sensitivityanalysisofnormativeeconomicmodels:Theoreticalframeworkand
practicalstrategies,AgriculturalEconomics16:139152.
Pellerin,R.F.,Ross,R.J.,2002.Transversevibrationandlongitudinalstresswavenondestructive
evaluationmethods.In:Pellerin,R.F.,andR.J.Ross(eds.).2002.Nondestructiveevaluationofwood.
ForestProductsSociety,Madison.210pp
Petersen,NO.,Wessels,C.B.,2011.Timberproperties androoftrussdesign.ReporttoSawmilling
SouthAfrica.CopyobtainablefromRoySoutheyatsoutheys@iafrica.com.
Roth,B.E.,Li,X.,Huber,D.A.,Peter,G.F.,2007.Effectsofmanagementintensity,genetics,and
plantingdensityonwoodstiffnessinaplantationofjuvenileloblollypineinthesoutheasternUSA.
ForestEcol.Manag.246:155162.
statesdesign.Edition2.3.
SANS6122.2008.Qualificationtestingofsolidstructuraltimberandlaminatedstructuraltimber
(glulam)forverifyingtimbergradingsystemsinaccordancetoagivenstandard.Edition2.1.
91
Uusitalo,J.,1997.Preharvestmeasurementofpinestandsforsawingproductionplanning.Acta
ForestaliaFennica259.p56.
Wagner,F.G.,Gorman,T.M.,Wu,SY.,2003.Assessmentofintensivestresswavescanningof
DouglasfirtreesforpredictinglumberMOE.ForestProd.J.53(3):3639.
Wang,X.,Ross,R.J.,2002.Nondestructiveevaluationofgreenmaterialsrecentresearchand
developmentactivities.In:Pellerin,R.F.,andR.J.Ross(eds.).2002.Nondestructiveevaluationof
wood.ForestProductsSociety,Madison.210pp.
Wang,X.,Ross,R.J.,McClellan,M.,2000.Strengthandstiffnessassessmentofstandingtreesusinga
ForestProductsLaboratory,Madison,p9.
Wang,SY.,Lin,CJ.,Chiu,CM.,2005.EvaluationofwoodqualityofTaiwaniatreesgrownwith
differentthinningandpruningtreatmentsusingultrasonicwavetesting.WoodFiberSci.37(2):192
200.
Wang,X.,Carter,P.,Ross,R.J.,Brashaw,B.K.,2007.Acousticassessmentofwoodqualityofraw
forestmaterialsapathtoincreasedprofitability.ForestProd.J.57(5):614
Watt,M.S.,Moore,J.R.,Faon,J.P.,Downes,G.M.,Clinton,P.W.,Coker,G.,Davis,M.R.,Simcock,R.,
Parfitt,R.L.,Dando,J.,Mason,E.G.,Bown,H.E.,2006.Modellingtheinfluenceofstandstructural,
edaphicandclimaticinfluencesonjuvenilePinusradiatadynamicmodulusofelasticity.ForestEcol.
Manag.229(13):136144.
Wang,X.,Ross,R.J.,McClellan,M.2000.Strengthandstiffnessassessmentofstandingtreesusinga
Wright,J.A.,1994.UtilizationofPinuspatula:Anannotatedbibliography.OxfordForestryInstitute
Wessels,C.B.,Dowse,G.P.,Smit,H.C.,2011a.TheflexuralpropertiesofyoungPinuselliottiixPinus
caribaeavar.hondurensistimberfromtheSouthernCape,andtheirpredictionfromacoustic
measurements.SouthernForests73(3&4):137147.
Wessels,C.B.,Malan,F.S.,Rypstra,T.,2011b.Areviewofmeasurementmethodsusedonstanding
treesforthepredictionofsomemechanicalpropertiesoftimber.Eur.J.ForestRes.130(6):881893.
Wessels,C.B.,Price,C.S.,Turner,P.,Dell,M.P.,2006.Integratingharvestingandsawmilloperations
usinganoptimizedsawmillproductionplanningsystem.In:Ackerman,P.A.,Langin,D.W.,Antonides,
M.C.(eds)ProceedingsoftheInternationalPrecisionForestrySymposium,StellenboschUniversity,
SouthAfrica,510March2006.ISBN0797211217
Zobel,B.J.,Sprague,J.R.,1998.Juvenilewoodinforesttrees.SpringerVerlagBerlinHeidelberg,
300pp.
Zobel,B.J.,VanBuijtenen,J.P.,1989.Woodvariation.Itscausesandcontrol.Springer,Berlin
HeidelbergNewYork.363p.
92
Chapter6.
Unpublished
ThevariationofmicrofibrilangleinSouthAfricangrownPinuspatulaanditsinfluenceon
thestiffnessofstructurallumber
C.B.Wesselsa*,F.S.Malanb,M.Kiddc,T.Rypstraa
a
Tel:+27218083319
Fax:+27218083603
email:cbw@sun.ac.za
b
PostgraduateForestProgramme,
UniversityofPretoria
PrivateBagX20,Hatfield0028,SouthAfrica
c
CentreforStatisticalConsultation
Abstract
ReductionintherotationagesofsoftwoodsawlogplantationsinSouthAfricaiscausingincreased
proportionsoflowstiffnesssawnlumberatfinalharvest.Ithasbeenshownforsomespeciesthat
themicrofibrilangle(MFA)oftheS2layeroftracheidsisstronglyrelatedtothemodulusofelasticity
(MOE)ofwood,evenmoresothanwooddensity,especiallyinwoodformedduringjuvenilegrowth.
TheobjectivesofthisstudyweretodescribethevariationinMFAinyoungPinuspatulatreesandto
determinetherelationshipbetweenMFAandthedynamicMOEofsawnPinuspatulalumber.Thirty
1620yearoldtreesfromsixcompartmentsfromtheMpumalangaescarpmentwereprocessedinto
discsandlumber.MFA,densityandringwidthweremeasuredattwoheightlevelsusingSilviscan3.
TheaverageannualringMFAvariedbetween7oand29o;thepatternofvariationdependingmainly
onheightlevelandtheringnumberfromthepith.TheMFAinP.patulafollowedthesamewithin
treevariationtrendsasinNewZealandgrownPinusradiatabuttheaverageMFAwaslowerin
absolutetermsanddifferencesbetweenheightlevelswerelesspronounced.MFAanddensity
exhibitedhighlysignificantPearsoncorrelationsof0.73and0.70respectivelywithboarddynamic
MOE.Amultipleregressionmodel,whichincludedMFA,densityandringwidth,explained71%of
thevariationinthedynamicMOEofboards.AsensitivityanalysisonthemodelshowedthatMFA
anddensityhadapproximatelysimilarinfluencesonpredictingthedynamicMOEofPinuspatula
boards.
Introduction
Approximately70%ofthesawntimberproducedinSouthAfricaissoldasbuildingorstructural
gradelumber(CrickmayandAssociates2011)andismainlyusedinrooftrussmanufacture.For
softwoodsawmillprocessorsitisthereforeimportantthatahighpercentageofthelumber
producedconformstothestructuralgraderequirements.Stiffnessandbendingstrengtharethetwo
mostimportantmechanicalpropertiesinlumberusedforresidentialrooftrussconstructionsin
SouthAfrica(PetersenandWessels,2011).Stiffnessorthemodulusofelasticity(MOE)determine
93
theresistancetodeflectionwhensubjectedtobendingstressaswellasthecompressiveloadthat
slendermemberscanwithstand.
Asaresultofacceleratedgrowthduetotheeffectsoftreebreedingandimprovedsilvicultural
practices,themeanharvestingageoftreesforsawlogproductioninSouthAfricandroppedfrom
14.1yearsin1983to11.3yearsin2003(CrickmayandAssociates,2004).Thissuggestsamean
harvestingagereductionfromabout28yearsin1983toabout23yearsin2003.Theyounger
harvestingagesresultedinincreasedproportionsofjuvenilewoodand,asaresult,asignificant
reductioninaverageMOEofstructurallumber,aswellasincreasedproportionsoflumbernot
conformingtotheminimumstrengthandstiffnessrequirementsforstructurallumber(Burdzik,
2004;Wesselsetal.2011;DowseandWessels,2013).
InSouthAfricaPinuspatulaisthemostimportantcommercialplantationsoftwoodresourcewitha
totalof338923haplantedwiththisspecies(DAFF,2009).TheMpumalangaescarpmentisthe
largestsawloggrowingareainSouthAfricawithPinuspatulathemainspeciesbeingplanted.
StudiesbyDowseandWessels(2013)andWesselsetal(Inpress)showedthatthemeanlumber
stiffnessof1620yearoldPinuspatulafromtheMpumalangaescarpmentwasabout25%lower
thanrequiredforthelowestSANSstructuralgrade.
Manystudiesemphasizedtheinfluenceofmicrofibrilangle(MFA)oftheS2layeroftracheidsonthe
MOEofwood(i.e.Cave1968and1969;CaveandWalker,1994;Megrawetal1999;EvansandIlic,
2001;Downesetal,2002;EvansandKibblewhite2002).CaveandWalker(1994)arguedthat
microfibrilangleistheonlypropertythatcanexplainthelargevariationinMOEinPinusradiata
fromthepithoutwardsandthatMFAisaprincipalpredictoroftimberquality,withdensity
behavingasanauxiliaryvariable.Moststudieswereperformedonsmallclearwoodspecimensbut
onlyafewofthesestudiesattemptedtorelateaverageMFAtothestiffnessofsawnlumber(i.e.
Downesetal,2002;Vikrametal,2011).
InspeciessuchasPinusradiataandPinustaedathevariationinMFAhasbeenstudiedextensively
andiswellunderstood(Donaldson1992and1996;Cownetal,1999;Megrawetal,1999;Xuand
Walker2004;Burdonetal,2004;Isiketal,2008).InPinusradiatathegeneraltrendofvariationin
MFAisarapiddecreasefromthepithuptoring10,followedbyatransitionphase,andthenlittle
changeafterring20(Figure1).MFAdecreasesfromthebasetotheupperpartsofthestem.Similar
trendswerefoundinPinustaeda(Burdonetal,2004;Megraw1985;Megrawetal1998and1999).
Density(kg/m3)
MFA(o)
50
500
480
40
460
440
0m
30
1.4m
20
Upperstem
10
420
400
380
360
340
Inner
corewood
Outer
corewood
5
Transition Outerwood
10
Ringsfrompith
15
20
Figure1.TypicalvariationofdensityandmicrofibrilangleinPinusradiata(adaptedfromBurdonet
al,2004).ThebrokenlinesindicateMFAandthesolidlinedensity.
DespitetheimportanceofMOE,anditsknowncloserelationshipwithMFA,verylittleattentionhas
beengiventothewithinandbetweentreedifferencesinthispropertyintheSouthAfrican
softwoodindustry.TheonlystudyconductedtodateinSouthAfricainvolvedtheuseofnear
94
infraredspectroscopytechnology(ZbonakandBush,2006).Inthisstudycalibrationmodelswere
developedtopredicttheMFAatbreastheightin14yearoldPinuspatulatreesfromKwaZuluNatal,
butnoattemptwasmadetodescribethevariabilityalongtheradiusoralongthelongitudinalaxisof
thetree,ortorelateMFAtotheMOEofthewood.
Thepurposeofthisstudywastwofold:Firstly,toexaminethevariationinMFAindifferentPinus
patulagrowthsitesradiallyatdifferentheightsandsecondly,todeterminetherelationshipbetween
MFAandtheMOEofsawnPinuspatulalumber.
Materialsandmethods
SampletreeswereselectedfromsixcompartmentsalongtheMpumalangaescarpment,South
Africa.DetailsofthesitessampledaregiveninTable1.Theplotsreceivedthenormalcommercial
managementtreatmentsofweeding,thinningandpruning.
Astratifiedsamplingprocedureintermsoftreediameterswasfollowedsothatthesampletrees
representedtheproductivetimbervolumeavailablefromthecompartments.Tentreeswere
selectedfromeachcompartment.Onetreewasrandomlyselectedfromthefirstquartile(small
diameter),twotreesfromthesecondquartile,threetreesfromthethirdquartileandfourtrees
fromthefourthquartile(largediameters).Thesesampletreeswerealsousedbyotherresearchers
forpreviousstudiesrelatedtostrengthandstiffnessvariation,structuralgradingandpulping
properties(Kipuputwaetal,2010;DowseandWessels,2013;Wesselsetal,Inpress).
Fiveofthetentreeswererandomlyselectedforthepurposesofthisstudy.
Table1.Samplesitesdetailsandsitemeans.
Sampleplot
identification
A(E66)
F(E55a)
G(E36c)
K(E35)
N(D74)
R(J20)
Plantation
Age
(yrs)
Nelshoogte
Uitsoek
Uitsoek
Berlin
Morgenzon
Wilgeboom
17
20
19
16
19
19
Mean
DBH
(cm)
36.0
34.8
32.0
30.2
24.6
33.4
Mean
height
(m)
20.5
20.9
23.1
18.0
16.6
23.7
Site
Indexat
age10
(m)
14.3
14.6
16.8
16.5
9.6
16.8
Mean
annual
precipitation
(mm)
1061
1151
902
1006
997
1299
Mean
annual
temperature
(C)
16.0
13.7
14.0
17.2
16.2
18.5
Fromeachtreesamplediscswereremovedatbreastheight(1.3m)andat6mheightlevels(Figure
2).Asawlogof2.2minlengthwasremovedbetweenthetwodiscs.Allsawlogswereprocessedinto
boardsof40x120mm(greendimensions)andkilndriedtoatargetmoisturecontentof12%.
ThestaticMOEontheflatsideofeachboard(MOEflat)wasdeterminedaccordingtoSANS10149
(2002).TheacousticresonancefrequencyofeachboardwasmeasuredusingtheAGraderPortable
softwarefromFalconEngineering.ThisvaluewasusedtocalculatethedynamicMOE(MOEdyn)for
eachboard:
MOEdyn=(2lf)2
where:
MOEdynDynamicmodulusofelasticity,inMPa;
Densityofthetestspecimenatthemoisturecontentatthetimeoftesting,inkg/m3;
l
Lengthofthetestspecimeninmeterstotheclosestmm;and
f
Resonancefrequencyofthetestspecimen,inHertz.
95
Sample
disc
Sample
disc
Pulping
log
0m
1.3m
2.3m
120mm
40mm
Sawlog
4.5m
6m
Treeheight
Figure2.Samplediscs,sawlogpositionandthesawingpatternused.
MFAandwooddensitymeasurementswereconductedonradialstripstakenfromboththebreast
heightand6mdiscs,usingSilviscan3(Evansetal,1999),atresolutionsof2mmand0.025mm
respectively.TheMFAanddensityvalueswererecordedonanannualringbasisusingtheradial
densityprofiletodemarcatedifferentyearrings.
Yearringsonbothendsofeachoftheboardsweredatedbyreconstructingindividuallogsand
countingtheringsfromtheoutsidewherethebarksidesurfacelayerwasstillvisibleontheboards.
IfthestemsurfacewasnotvisibletheringwidthpatternsfromtheSilviscanoutputwasusedtodate
individualringsonboardends.
TheMFAanddensityofspecificringsweredeterminedasthemeanfromthetwodiscs.Forevery
boardameandensity(DensitySS)andmeanMFAwascalculatedbasedontheyearringspresentin
theboard.Ameanringwidth(RingWidthSS)wascalculatedforeachboardfromtheSilviscanring
widthdataofbothheights.Meanringageperboard(RingAge)wascalculatedbasedontheageof
treeringsonbothendsofeachboard.Theringagestartedatonefortheouterringnexttothebark
andincreasedtowardsthepith.
AmixedmodelrepeatedmeasuresANOVAwasperformedusingStatisticasoftware
(www.statsoft.com)totestthestatisticalsignificanceoftheeffectofcompartment,ringageand
heightasmainfactors,andtheirinteractions,onMFAanddensity.Multipleregressionmodelswere
developedtopredictindividualboardMOEdynfromtheMFA,wooddensityandringwidth.Stepwise
forward,backwardandbestsubsetsselectioncriteriawereusedtoidentifythosefactorsthat
contributedsignificantlytothemodels.Sensitivityanalyseswereperformedonthemodelto
determinetherelativeinfluenceofvaryingtheindependentvariables,oneatatime,onMOEdyn
(Pannel,1997).
Results
InthemixedmodelrepeatedmeasuresANOVA,treeswereusedasarandomeffectandtheother
factorsasfixedeffects.ResultsfortheANOVAonMFAshowedthattherewasasignificantthree
wayinteractionbetweencompartment,heightandringsfromthepith(Table2).Duetothe
96
complexityinvolvedonlythetwowayinteractionswereinterpreted.Therewasahighlysignificant
interactionbetweenringsandheightandamoderatelysignificantinteractionbetween
compartmentandringsfromthepith.ThevariationinMFAperyearringfromthepithandforeach
heightlevelisillustratedinFigure3.Takenotethatonlyrings0to11wereconsideredintheANOVA
assomedisksat6mheighthad11yearringsonly.
Densityperyearringwassignificantlyinfluencedbyallthetwowayinteractions(Table2).Figure4
showsthevariationindensityasafunctionofyearringsandheight.
Table2.ANOVAtableforMFAanddensityoftheyearringswithcompartment,heightandrings
frompithasfactors.
Sourceofvariation
MFA
Numerator Denominator
DF
DF
Compartment
5
24
Height
1
24
Ringsfrompith
11
264
Compartment*Height
5
24
Compartment*Ringsfrompith
55
264
Height*Ringsfrompith
11
258
Compartment*Height*Ringsfrompith
55
258
Density
F
1.9079
19.8627
249.4173
0.1922
1.5550
5.0686
1.6693
0.130341 4.24746 0.006586

0.000165 28.86990 0.000016
0.000000 81.70313 0.000000
0.962580 3.26880 0.021636
0.012291 5.27747 0.000000
0.000000 2.36930 0.008361
0.004488 1.28312 0.103536
Least Squares Means

35
30
MFA (o)
25
20
15
10
0
0
8
Rings from pith
10
12
14
16
1.3m disk
6m disk
Figure3.VariationinMFAatdifferentheightsandringsfromthepith.
97
Least Squares Means

700
650
Density (kg/m3)
600
550
500
450
400
350
300
0
10
Rings from pith
12
14
16
1.3m disk
6m disk
Figure4.Variationindensityatdifferentheightsandringsfromthepith.
Pearsoncorrelationsbetweenthevariouspropertiesmeasuredandcalculatedforindividualboards
arepresentedinTable3.AmultipleregressionmodelwasdevelopedforMOEdynusingthecalculated
boardpropertiesfromSilviscanmeasurementsasinputs.MFAandDensitySSwerehighlysignificant
(p<0.001)andRingWidthSSwasasignificantparameter(p<0.05)seeTable4.Themodelcould
explain71%ofvariationinMOEdyn.Agraphshowingtheobservedvs.predictedvaluesisshownin
Figure5.ThemodelslightlyunderpredictsforhighMOEdynvaluesandoverpredictsforlowMOEdyn
values.
Asensitivityanalysiswasperformedontheregressionmodel(Table4).Themean,5thpercentileand
95thpercentilewasdeterminedfromtheobservedvaluesforeachofMFA,DensitySSand
RingWidthSS.Inthemodelaspecificvariablewaschangedfromits5thpercentilevaluetothe95th
percentilevaluewhiletheothervariableswerekeptconstantattheirmeanobservedvalues.The
changeinMOEdynfromthemodelasaresultwasexpressedinabsoluteterms(MOEdyn)andasa
percentageinfluence(Influence)ofeachvariableinthemodel(Table4).Forexample,bychanging
MFAfromits5thpercentileobservedvalueof10.1otoits95thpercentileobservedvalueof24.2oin
theregressionmodelthepredictedMOEdyndecreasedby2884MPa.
Table3.Pearsoncorrelationsbetweenvariouspropertiesmeasuredorcalculatedforindividual
boards(*,**,***significantat0.05,0.01,and0.001probabilitylevels,respectively).
MOEdynMOEflat DensitySS MFA RingAge RingWidthSS Densityboard
MOEdyn
1.0000 .90*** .70*** .73*** .49*** .71***
.69***
MOEflat
1.0000 .68*** .64*** .50*** .59***

.68***
DensitySS
1.0000 .47*** .36*** .58***

.90***
MFA
1.0000 .61*** .69***

.39***
RingAge
1.0000 .63***
.23*
RingWidthSS
1.0000
.51***
Densityboard
1.0000
98
Table4.AmultipleregressionmodelfortheMOEdynofPinuspatulalumber(R2=0.71).Sensitivity
analysisresultsshowtherelativeinfluenceofeachindependentvariableinthemodel.Parameters
markedwith*,**,***weresignificantatthe0.05,0.01and0.001probabilitylevels,respectively.
Regressionmodel
Parameters
Intercept 6404.8***
DensitySS 16.3***
MFA
203.8***
RingWidthSS151.2*
Mean
452
17.5
10.9
Sensitivityanalysis
th
th
95
MOEdyn
5
percentile percentile
3
3
377kg/m
553kg/m 2868MPa
o
o
10.1
24.2
2884MPa
6.1mm
15.6mm
1441MPa
Influence
(%)
39.9%
40.1%
20.0%
Observed Values vs. Predicted

Y = 2506.926+0.7069*x; 0.95 Conf.Int.
Dependent variable: MOEdyn
16000
15000
14000
Predicted Values
13000
12000
11000
10000
9000
8000
7000
6000
5000
4000
4000
6000
8000
10000
12000
14000
16000
Observed Values
Figure5.Predictedvs.observedvaluesofamultipleregressionmodelforMOEdyn(Table4).The
modeland0.95confidenceintervalsandthe1:1lineareindicated.
DiscussionandConclusions
ThemeanMFAperyearringinPinuspatulavariedbetween7oand29oalongthefirst16yearrings
fromthepith.ThereweresignificantdifferencesinMFAwithinthetreedependingontheheightin
thetreeandringnumberfromthepith(Figure3).AverageMFAvariedfromabout10oatthe16th
yearringat1.3mheightabovegroundleveltonearly30oatthepith.(Figure3).At6mheightabove
groundleveltheaverageMFAwasabout2o5olowerthanatbreastheightexceptatthepithwhere
theaverageMFAwasapproximatelythesameatthetwoheightlevels.
ComparedtoNewZealandgrownPinusradiatatheMFAofPinuspatulaat1.3mwasroughly10o
smalleratsimilaryearringsfromthepith(Figures1and3).Thedifferencesatthedifferentheight
levelswerelesspronouncedcomparedtoNZPinusradiata,showingdifferencesinMFAofupto10o
betweenbreastheightandupperstemcomparedto2o5oforPinuspatula.
Densitywassignificantlydifferentatvariousrings,heightsandcompartments.Itvariedfromjust
over400kg/m3perringnexttothepithtoaround600kg/m3atring16(Figure4).Thetrendsare
similartothatofPinusradiataalthoughinabsolutetermsPinuspatuladensitywasroughly100
kg/m3higheratsimilarringpositions(Figure1).
TheMOEdynmeasuredoneachboardwascloselyrelatedtoMOEflat(r=0.9),themostcommon
methodusedformachinegradingoflumber(Table3).DensitySS,ascalculatedfromSilviscan
measurementsalsohadastrongcorrelationwiththedensityofthefullboard(r=0.9)despitethe
99
factthatnoattemptwasmadetocorrecttheDensitySScalculationfortherelativevolumeofeach
ringinaboard.Duetopitheccentricity,logsthatwerenotcenteredcorrectlyintheprimaryand
secondarybreakdownsaws,thetaperofthelog,andthecurvatureofrings,theactualvolumeofa
ringinaboardwastoocomplextodetermine.ForthisstudytheMFAanddensityofeachyearring
thatappearsinaboardcountedthesameweightintermsofitscontributiontothecalculatedmean
MFAanddensity(DensitySS)ofthatboard.
TheboardpropertiescalculatedfromSilviscanmeasurementsMFA,DensitySSandRingWidthSShad
similarandhighlysignificantcorrelationcoefficientswithMOEdynof0.73,0.70and0.71respectively
(Table3).TheRingAgealsohadasignificantcorrelationwithMOEdynof0.49.
Themultipleregressionmodel(Table4)whichincludedMFA,DensitySSandRingWidthSSexplained
71%ofthevariationintheMOEdynoftheboards.SensitivityanalysisonthemodelshowedthatMFA
wasthemostinfluentialvariableinthemodel(40.1%),followedbyDensitySS(39.9%)and
RingWidthSS(20.0%).BoththePearsoncorrelationsandthesensitivityanalysisontheregression
modelthussuggestthatmicrofibrilangleanddensityareofroughlysimilarimportanceinexplaining
theMOEdynoftheboards.Whenseparatingboardsaccordingtotheirdistancefromthepith,the
PearsoncorrelationcoefficientsshowthatdensityandMOEdynwereslightlybettercorrelatedfor
boardsclosertothepiththanMFAandMOEdyn(notshowninResultssection).Thisiscontraryto
resultsobtainedfromstudiesonsmallclearsamplesofPinusradiatawhereMFAwasfoundmore
influentialincorewood(i.eCaveandWalker1994;WalkerandButterfield1996;DicksonandWalker
1997).Itmustbementionedthatthefewresultsavailablefromstudiesonfullsizedlumberseemto
partiallycontradictthoseonsmallclearspecimens.InastudyonPinusradiataDownesetal(2002)
foundthatbothdensityandMFAwerecorrelatedtoboardMOEforoneofthesitesconsidered.In
thecaseofanothersitetheeffectofdensityonMOEwasmoredominantthantheeffectofMFA.
Vikrametal(2011)foundthatdensityhadagreaterdirecteffectonthestaticMOEofDouglasfir
lumberthanMFA.
ThenegativecorrelationofRingWidthSSwithMOEdyn(Table3)andthefactthatRingWidthSSalso
enteredthemultipleregressionmodelasanegativeparameter(Table4)isnoteworthy.The
sensitivityanalysissuggeststhatRingWidthSSisabouthalfasinfluentialasDensitySSandMFAin
determiningMOEdyn.ThiswasmostprobablypartlyasaresultoftheeffectofRingWidthSSonthe
slendernessoftrees.Severalstudiesfoundastrongrelationshipbetweentreeslendernessandthe
stiffnessofitswood(i.e.Lasserreetal.,2005;Wattetal.,2006;Rothetal.,2007;Lasserreetal.
2009).
Resultstodatesuggestthattheestablishmentoftreesathigherplantingdensitieswiththeobjective
toencouragethedevelopmentofmoreslendertrees,mightbeausefulpracticalplantation
managementtooltoincreasetheproductionofstifferandstrongerwood,especiallyofthewood
producedduringjuvenilegrowth.Higherplantingdensitiesalsoresultinlessexposuretowindit
hasbeenshownthatwindexposureresultsinlessstiffwood(Bascunetal.,2006).Whilethis
approachwillmostcertainlyimpairdiametergrowth,thefinancialgainsthatwillresultfromthe
improvedqualityofthewoodproducedmightsurpassthenegativeeffects(ifany)causedbythe
reducedgrowthrate.
Higherplantingdensityandtheresultantreductioningrowthratemaynotnecessarilyhavean
adverseeffectonvolumeproductionperunitareaandmayevenincreaseit,dependingonspecies
andsiteconditions(VanLaar,1978).
Microfibrilangle,togetherwithdensity,areclearlyimportantpropertiesintermsoftheirinfluence
onthestiffnessofsawnlumberfromyoungPinuspatula.ThepotentialuseofMFAdatabytree
100
growersandprocessorsdependsontheapplication.Someauthorssuggestedthattreebreeders
refrainfrommeasuringandselectingforMFAduetothecostandthatmuchcheaperacousticstress
wavemethods,anddensityratherbeusedbecauseoftheirexcellentabilitytopredictMOE(Vikram
etal2011).However,inanotherstudybytheauthoronthesamesamplematerialitwasfoundthat
acousticvelocityassessmentsperformedonstandingtreesexplainedconsiderablylessofthe
variationinMOEdynthanMFA(Wesselsetal,Inpreparation).
MeasurementofMFAforPinuspatulacanbeparticularlyusefulinstudieswheretheeffectoftime
dependenttreatmentssuchasthinningorfertilisingonwoodpropertiesneedstobeevaluated.
SinceMFAcanbemeasuredonindividualyearrings,theeffectofatreatmentonMFAcanbe
evaluatedveryeffectivelyoveragrowthperiod.
GiventheproblemsassociatedwithreducedrotationagesonthestiffnessofplantationgrownPinus
patulainSouthAfrica,theuseofMFAinformationofthisspeciesinbothresearchandindustrial
studiescanbebeneficialinmanagingtheseproblems.
101
References
BascunA,MooreJR,WalkerJCF.2006.Variationsinthedynamicmodulusofelasticitywith
proximitytothestandedgeinradiatapinestandsontheCanterburyPlains,NewZealand.NZJournal
ofForestry11/2006.
BurdzikW.2004.GradeverificationofSApine.S.Afr.ForestryJ202:2127.
BurdonRD,KibblewhiteRP,WalkerJCF.MegrawRA,EvansR,CownDJ.2004.JuvenileVersus
MatureWood:ANewConcept,OrthogonaltoCorewoodVersusOuterwood,withSpecialReference
toPinusradiataandP.taeda.ForestScience50(4):399415.
CaveID.1968.Theanisotropicelasticityoftheplantcellwall.WoodScienceandTechnology2:268
278.
CaveID.1969.ThelongitudinalmodulusofPinusradiata.WoodScienceandTechnology3:4048.
CaveID,WalkerJCF.1994.Stiffnessofwoodinfastgrownplantationsoftwoods:theinfluenceof
microfibrilangle.ForProdJ44(5):4349
CownDJ,HerbertJ,BallRD.1999.ModellingPinusradiatalumbercharacteristicsPart1.Mechanical
propertiesofsmallclears.NZJForestScience29:203213.
CrickmayandAssociates.2004.Supplyanddemandstudyofsoftwoodsawlogandsawtimberin
SouthAfrica.Copyobtainablefrommandy@crickmay.co.za.
CrickmayandAssociates.2011.SouthAfricanLumberIndexforOctober2011.Copyobtainablefrom
DicksonRL,WalkerJCF.1997.Pines:growingcommoditiesordesignertrees.Commonwealth
ForestryReview76:273279.
DonaldsonLA.1992.WithinandbetweentreevariationinmicrofibrilangleinPinusradiata.New
ZealandJ.ForestryScience22:7786.
DonaldsonLA.1996.EffectofphysiologicalageandsiteinmicrofibrilangleinPinusradiata.IAWAJ.
17:421429.
DownesGM,NyakuengamaJG,EvansR,NorthwayR,BlakemoreP,DicksonRL,LausbergM.2002.
Relationshipbetweenwooddensity,microfibrilangleandstiffnessinthinnedandfertilisedPinus
radiataD.Don.IAWAJ.23:253265.
DowseGP,WesselsCB.2013.ThestructuralgradingofyoungSouthAfricangrownPinuspatulasawn
timber.SouthernForests75(1):717.
EvansR,HughesM,MenzD.1999.MicrofibrilanglevariationbyscanningXraydiffractometry.
AppitaJ.52(5):363367.
102
EvansR,BookerR,KibblewhiteRP.2001.Variationofmicrofibrilangle,densityandstiffnessinfifty
radiatapinetrees.In:Proceedingsof55thannualgeneralAppitaconference,Hobart,Tasmania,
Australia.
EvansR.IlicJ.2001.Rapidpredictionofwoodstiffnessfrommicrofibrilangleanddensity.Forest
Prod.J.51(3):5357.
EvansR,KibblewhiteRP.2002.Controllingwoodstiffnessinplantationsoftwoods.In:BeallFC(ed.)
2002.Proceedingsof13thinternationalsymposiumonnondestructivetestingofwood,Universityof
California,Berkeley,1921August2002,p6774.
IsikF,GumpertzM,LiB,GoldfarbB,SunX.2008.Analysisofcellulosemicrofibrilangleusingalinear
mixedmodelinPinustaedaclones.CanadianJForestResearch38:16761689.
KipuputwaC,GrzeskowiakV,LouwJH.2010.Theuseofnearinfraredscanningforthepredictionof
pulpyieldandchemicalpropertiesofPinuspatulaintheMpumalangaescarpmentareaofSouth
Africa.SouthernForests72(34):181189.
Lasserre,JP.,Mason,E.G.,Watt,M.S.,2005.TheeffectsofgenotypeandspacingonPinusradiataD.
Doncorewoodstiffnessinan11yearoldexperiment.ForestEcol.Manag.205:375383.
Lasserre,JP.,Mason,E.G.,Watt,M.S.,Moore,J.R.,2009.Influenceofinitialplantingspacingand
genotypeonmicrofibrilangle,wooddensity,fibrepropertiesandmodulusofelasticityinPinus
radiataD.Doncorewood.ForestEcol.Manag.258:19241931.
Megraw,RA.1985.Woodqualityfactorsinloblollypine.TAPPIpress,Atlanta,GA.88p.
MegrawR,LeafG,BremerD.1998.Longitudinalshrinkageandmicrofibrilangleonloblollypine.In:
ButterfieldBG(ed.).1998.ProceedingsofIAWA/IUFROinternationalworkshoponthesignificanceof
microfibrilangletowoodquality,Westport,NewZealand,p2761.
MegrawR,BremerD,LeafG,RoersJ.1999.Stiffnessinloblollypineasafunctionofringpositionand
height,anditsrelationshiptomicrofibrilangleandspecificgravity.In:NepveuG(ed.)1999.
ProceedingsofIUFROWPS5.0104thirdworkshoponconnectionbetweensilvicultureandwood
qualitythroughmodellingapproachesandsimulationsoftware.
PannellDJ.1997.Sensitivityanalysisofnormativeeconomicmodels:Theoreticalframeworkand
practicalstrategies,AgriculturalEconomics16:139152.
Petersen NO, Wessels CB. 2011. Timber properties and roof truss design. Report to Sawmilling
SouthAfrica.CopyobtainablefromRoySoutheyatsoutheys@iafrica.com.
Roth,B.E.,Li,X.,Huber,D.A.,Peter,G.F.,2007.Effectsofmanagementintensity,genetics,and
plantingdensityonwoodstiffnessinaplantationofjuvenileloblollypineinthesoutheasternUSA.
ForestEcol.Manag.246:155162.
SANS10149.2002.SouthAfricanNationalStandard.Themechanicalstressgradingofsoftwood.
Edition1.2.
VanLaarA.1978.ThegrowthofunthinnedPinuspatulainrelationtospacing.SouthAfricanForestry
Journal107:311.
103
VikramV,CherryML,BriggsD,CressDW,EvansR,HoweGT.2011.StiffnessofDouglasfirlumber:
effectsofwoodpropertiesandgenetics.CanadianJ.ofForestResearch41:11601173.
WalkerJCF,ButterfieldBG.1996.Theimportanceofmicrofibrilanglefortheprocessingindustries.
NewZealandJForestry40(4):3440.
Watt,M.S.,Moore,J.R.,Faon,J.P.,Downes,G.M.,Clinton,P.W.,Coker,G.,Davis,M.R.,Simcock,R.,
Parfitt,R.L.,Dando,J.,Mason,E.G.,Bown,H.E.,2006.Modellingtheinfluenceofstandstructural,
edaphicandclimaticinfluencesonjuvenilePinusradiatadynamicmodulusofelasticity.ForestEcol.
Manag.229(13):136144.
WesselsCB,DowseGP,SmitHC.2011.TheflexuralpropertiesofyoungPinuselliottiixPinus
WesselsCB,MalanFS,NelDG,RypstraT.Inpress.Variationinstrength,stiffnessandrelatedwood
propertiesinyoungSouthAfricangrownPinuspatula.SouthernForests76.
WesselsCB,MalanFS,LouwJH,EvansR,RypstraT.Inpreparation.Thepredictionoftheflexural
propertiesoflumberfromstandingSouthAfricangrownPinuspatulatrees.
ZbonakA,BushT.2006.Applicationofnearinfraredspectroscopyinpredictionofmicrofibrilangle
ofan14yearoldPinuspatulastand.In:KurjatkoS,KudelaJ,LaganaR(eds).2006.Proceedingsof
WoodStructureandProperties06,ZvolenSlovakia.ArboraPublishers,pp.175180.
XuP,WalkerJCF.2004.Stiffnessgradientsinradiatapinetrees.WoodSci.Technol.38:19.
104
Chapter7.Summaryofresearchresults
Themainfindingsofthefullinvestigationasreportedinthepapers(Chapters2to6)are
summarisedbelow.
Thesawnlumberfromthe1620yearoldPinuspatulatreeshadlowstiffnessandthemean
MOEedgeofthevisualandmechanicalstructuralgradesdidnotcomplywiththeSANS10163
1(2003)requirements.Thisindicatesthattherelationshipbetweenvariousindicating
propertiesforgradingforthisresourceisverydifferenttowhatitwaswhenthe
characteristicvaluesforSouthAfricanpinewasdeterminedthreedecadesago.Asa
consequencegradingrulesandmethodswillhavetobeadaptedforthisresource.
The5thpercentilebendingstrengthforthisresourcewasmuchhigherandthe5thpercentile
tensilestrengthslightlyhigherthanthatrequiredforthelowestSAstructuralgrade.
Mostofthedesirablepropertiesforstructuraltimberimprovedwithdistancefromthepith.
Theonlyexceptionwasthe5thpercentilevalueforMORwhichwashigherforthepith
boardsthanforthe1stboards.
LogpositionorheightabovegroundlevelhadasignificanteffectontheMOEdyn,MORand
twistinboards.TheR2valuefortherelationshipbetweenMOEdynandMORofboardsfrom
thetoplogsweremuchlowerthanthatofboardsfromthebottomlog,suggestingthatthe
predictionofMORfromthenondestructiveassessmentofMOEdynwouldtendtobeless
efficientforlogboardsoriginatingfromhigherupinthetree.
PredictivemultipleregressionmodelsforMOEdynandMORoflumberweredevelopedata
board,treeandcompartmentlevel.Modelsthatweredevelopedwerecapableofexplaining
68%,60%and95%ofthevariabilityinMOEdynatindividualboard,treeandcompartment
levelrespectively.Thebestmodelsdevelopedexplained40%and42%ofvariabilityinMOR
ataboardandtreelevelrespectively.Atcompartmentlevelthebestmodelexplained80%
ofthevariabilityinMOR5perc.
Resultsofsensitivityanalysesshowedthatsiteindexatbaseage10,timeofflightacoustic
measurements,densityandringwidthwerethemostinfluentialvariablesinthemodels
developedforpredictingMOEdyn.InthemodelforpredictingtheMOR5percatcompartment
levelsiteindexatbaseage10,thebranchangle,branchspacingandringwidthwerethe
mostinfluentialvariables.
Themodelsdevelopedalsosuggestthattreeslendernessduringearlygrowthmighthave
playedanimportantroleindeterminingtheMOEdynandMORofboardsoriginatingfromthe
corewoodzone,whichisconsistentwiththeEulerbucklingtheoryandthebendingstress
theory.
Meanyearringmicrofibrilangle(MFA)variedbetween7oand29ointhefirst16yearrings
fromthepith.IngeneralMFAdecreasedwithdistancefromthepithandheightabove
groundlevel.ThemicrofibrilangleofPinuspatulafollowedthesamewithintreetrendsas
Pinusradiatabutwaslowerinabsolutetermsinthefirstspeciesanddifferencesbetween
heightlevelswerelesspronounced.
MicrofibrilangleanddensityhadhighlysignificantPearsoncorrelationsof0.73and0.70
respectivelywithboardMOEdyn.Amultipleregressionmodelwhichincludedmicrofibril
angle,densityandringwidthexplained71%ofthevariationintheMOEdynofboards.
Sensitivityanalysisonthemodelshowedthatmicrofibrilangleanddensityhadroughly
similarinfluencesonthepredictedMOEdynofPinuspatulaboards.
Theresultsfromthisstudycanbeusedbytreebreeders,growersandprocessorsofPinuspatula
treesaswellasresearchersworkingonthisspeciestohelpmanagetheproblemsassociatedwith
fastergrowthandlowerrotationages.Thepredictivemodelscanassistwithallocationoflogsto
processingfacilities,decisionsrelatedtoharvestingage,andselectionofsuperiorbreedingmaterial
intreeimprovementprograms.
105
Outcomesofthisstudy
Sincethestartofthisstudy,severaloftherecommendationswhichresultedfromtheoutcomeof
thestudiesdescribedinChapters2to6havebeenadoptedbyindustryorhavepromptedfurther
investigations.Ingradetestingprogramshavebeenconductedatsawmillswherebothmatureand
predominantlyyoungPinuspatulaareprocessed(CraffordandWessels,2011;Wesselsand
Froneman,2012).Theseresults,combinedwiththeresultsofthestudyoutlinedinChapter3
resultedinthedevelopmentofanewSouthAfricanNationalStandards(SANS)gradingquality
controlcodefortheSouthAfricansawmillingindustry.Thefirstdraftofthenewgradingquality
controlcodehasalreadybeenacceptedbySawmillingSAandiscurrentlyintheprocessofbeing
finalisedbySANS.
Thisstudyalsoinitiatedacollaborativeresearchprojectfundedbyindustrywiththeobjectiveto
investigatenewforestmanagementregimesforspecificSouthAfricanpinespecies.Thisprojectwill
focusprimarilyontheeffectsofhigherinitialplantingdensities,adjustedthinningregimesand
changesinrotationage.Plantingatcloserspacingwillresultinthedevelopmentofmoreslender
andlessexposedtrees,producingwoodthatmightbeofhigherstiffness.
Resultsinthisstudysuggestedthatarelationshipmightexistbetweenthenitrogentopotassium
ratiodeterminedfromleafanalysis,andthelatewoodpercentageofwood.Sincelatewoodhasa
higherstiffnessthanearlywood,thegrosseffectofanincreasedlatewoodproportioncouldbean
increaseinthestiffnessofthewoodingeneral.Unfortunatelythisaspecthasnotbeenconsideredin
thisstudy.Furtherstudiesarethereforestronglyrecommended.Abetterunderstandingofthe
factorsthataffectthetransitionfromearlywoodtolatewoodformationwillbeavaluable
contribution.
106
References
Burdzik,W.,2004.GradeverificationofSApine.S.Afr.ForestryJ202:2127.
Carle,J.,Holmgren,P.,2008.Woodfromplantedforests.Aglobaloutlook20052030.ForestProd.J.
58(12):618.
Carle,J.,Vuorinen,P.,DelLungo,A.,2002.Statusandtrendsinglobalforestplantation
development.ForestProd.J.52(7):213.
Cown,D.J.,2006.Woodqualityinstandingtimberevolutionofassessmentmethodsinplantations.
In:Kurjatko,S.,Kdela,J.andR.Lagaa(eds.)2006.Proceedingsofthe5thIUFROSymposium
WoodStructureandProperties06,September36,SliaSielnica,Slovakia.Organisedjointlyby
theFacultyofWoodSciencesandTechnologyoftheTechnicalUniversityofZvolenandtheIUFRO
Crafford,P.L.,Wessels,C.B.(2011).TheflexuralpropertiesandstructuralgradingofSAPine.Report
toSawmillingSA.DepartmentofForestandWoodScience,StellenboschUniversity.Copyobtainable
fromRoySouthey(southeys@iafrica.com).
CrickmayandAssociates.,2011.SouthAfricanLumberIndexforOctober2011.Copyobtainablefrom
DAFF.,2009.ReportoncommercialtimberresourcesandprimaryroundwoodprocessinginSouth
FAO.,2013.Plantedforests.http://www.fao.org/forestry/plantedforests/en/.Accessed
26/06/2013.
Grabianowski.M.,Manley,B.,Walker,J.C.F.2004.Impactofstockingandexposureonouterwood
acousticpropertiesofPinusRadiatainEyrewellForest.NewZealandJFor:Aug2004.
Hodge,G.R.,Dvorak,W.S.,2012.GrowthpotentialandgeneticparametersoffourMesoamerican
pinesplantedintheSouthernHemisphere.SouthernForests74(1):2749.
Ivkovi,M.,Gapare,W.G.,Abarquez,A.,Ilic,J.,Powell,M.B.,Wu,H.X.2009.Predictionofwood
stiffness,strength,andshrinkageinjuvenilewoodofradiatapine.WoodSciTechnol43:237257.
Launay,J.,Ivkovich,M.,Paques,L.,Bastien,C.,Higelin,P.,Rozenberg,P.2002.Rapidmeasurement
oftrunkMOEonstandingtreesusingRigidimeter.AnnForSci59:465469.
Lindstrm,H.,Harris,P.,Nakada,R.2002.Methodsformeasuringstiffnessofyoungtrees.HolzRoh
Werkst60:165174.
Malan,F.S.2010.CorewoodinSouthAfricanpine:necessityandopportunitiesforimprovement.
SouthernForests72(2):99105.
Matheson,A.C.,Dickson,R.L.,Spencer,D.J.,Ilic,J.2002.AcousticsegregationofPinusradiatalogs
accordingtostiffness.AnnForSci59:471477.
107
Owen,D.L.,VanderZel,D.W.,2000.Trees,forestsandplantationsinSouthernAfrica.In:Vander
Zel,D.L.(ed).SouthAfricanForestryHandbook.SouthernAfricanInstituteofForestry.
statesdesign.
Uusitalo,J.1997.Preharvestmeasurementofpinestandsforsawingproductionplanning.Acta
ForestaliaFennica259.56p.
Wang,SY.,Lin,CJ.,Chiu,CM.2005.EvaluationofwoodqualityofTaiwaniatreesgrownwith
differentthinningandpruningtreatmentsusingultrasonicwavetesting.WoodFiberSci37(2):192
200.
Wang,X.,Carter,P.,Ross,R.J.,Brashaw,B.K.2007.Acousticassessmentofwoodqualityofraw
forestmaterialsapathtoincreasedprofitability.ForProdJ57(5):614.
Wang,X.,Ross,R.J.,McClellan,M.2000.Strengthandstiffnessassessmentofstandingtreesusinga
Wessels,C.B.,Price,C.S.,Turner,P.,Dell,M.P.2006.Integratingharvestingandsawmilloperations
usinganoptimizedsawmillproductionplanningsystem.InAckerman,P.A.,Langin,D.W.,andM.C.
Antonides(eds),2006,ProceedingsoftheInternationalPrecisionForestrySymposium,Stellenbosch
University,SouthAfrica,510March2006.ISBN0797211217.
Wessels,C.B.,Dowse,G.P.,Smit,H.C.,2011.TheflexuralpropertiesofyoungPinuselliottiixPinus
Wessels,C.B.,Froneman,G.M.,2012.ThestiffnessandbendingstrengthofyoungSAPine.Reportto
SawmillingSA.DepartmentofForestandWoodScience,StellenboschUniversity.Copyobtainable
fromRoySouthey(southeys@iafrica.com).
Wright,J.A.,1994.UtilizationofPinuspatula:Anannotatedbibliography.OxfordForestryInstitute
108
Append
dixA:Declaarationsofccandidateaandcoauth
hors
109
110
111
112
113

The Prediction of The Flexural Lumber Properties From Standing South African-Grown Pinus Patula Trees

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

The Prediction of The Flexural Lumber Properties From Standing South African-Grown Pinus Patula Trees

Загружено:

Авторское право:

Доступные форматы

Thevariationandpredictionofstructurraltimberp

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

90 100 110 120 130 140 150 160 170

Distance from pith (mm)

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

decided tousethe manuallydeeterminedab

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

MOE Edge:MOR (Mpa): y = -2.3337 + 0.0055*x;

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

It is recommended that ingrade testing programs be conducted at sawmills which use a

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

SANS, 6122(1994).SouthAfricannational standard.Qualificationtestingofsolid structuraltimber

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

69827358 72167528 84098892 82218693

Stellenbosch University http://scholar.sun.ac.za

Stellenbosch University http://scholar.sun.ac.za

1.0000 .68* .64* .50* .59*

1.0000 .47* .36* .58***

1.0000 .61* .69*