Timber grading and scanning Digest 492di

gest

Since the 1980s there has been remarkabledevelopment of advanced timber grading,scanning and sorting technology (Szymani, 1999).This technology ranges from the use of X-rays forinternal imaging of logs and boards, tomicrowaves for detecting sloping grain. Opticalboard scanners use a variety of methods based onimage analysis techniques, including the lasertracheid effect. In some cases these techniques areused in conjunction with machine strengthgraders, which work by determining stiffness bybending.

Log sorting and processing

To convert logs into sawn lumber efficiently, theshape and size of logs must be known. Thisinformation can be used to control log orientationprior to sawing and to optimise cutting patterns,enabling maximum volumetric yield to beobtained while at the same time minimising thenumber of boards rejected owing to wane. Logshape scanners have developed from relativelysimple shadow-type scanners (using alternatingarrays of light emitting diodes and photo-receivers) which determine log outline anddiameter, to multiple camera-laser scanners whichcan accurately and quickly determine the precisethree-dimensional shape of the log. In the lattertype of scanner, images of laser lines projectedonto the log surface are processed allowing shapecalculation by triangulation (Figure 1 on page 2).

Chris Holland and Tim Reynolds

BRE Centre for Timber Technology and Construction

3D log shape data processed as two

overlapping surfaces

Timber is an immensely useful but naturally variable material. Wood can contain features, such as knots and sloping grain, that may not be suitable for certain end uses. Dimensional defects and distortion can also affect the use of timber. To use timber reliably for structural purposes, it is important that the strength properties of any member fall within certain limits. Machine grading is a form of non-destructive testing that allows timber to be sorted into strength classes, also enabling timber unsuitable for construction to be rejected. Timber for non-structural uses, such as furniture or flooring, may also be sorted to meet certain appearance grades.

This Digest details advances in grading and scanning technology, for both logs andsawn timber, and changes to structural timber grading due to European harmonisation.

The shape of logs can be used to indicate thelikely quality of the timber. It is readily possible torecord log variables such as sweep (log curvature)and taper, recognise butt logs (those cut from thebase of a tree), assess out-of-roundness and othershape features associated with lower qualitytimber. Abrupt changes in log curvature, forexample, indicate disturbed grain which maycause downgrade in sawn timber. Branchswellings may also be correlated to the size ofknots. Compression wood in logs may beindicated by their shape and form (Warensjo,2003). However, compression wood can also formin straight, round trees. Some types of internaldefect, including severe cross-grain resulting fromstem deviation in early tree life, cannot be readilyindicated by log shape.

Laser triangulation can also be used to measurethe width and thickness of boards together withprofile checking. Stroboscopic directionalillumination can detect wane.

Internal imaging of logs

Computed axial tomography (CAT) is the processof imaging parallel to the length axis of an object(Figure 2). Images can be combined to produce athree-dimensional model of the internal structure.CAT methods can be based on the use ofmicrowaves, acoustics, nuclear magneticresonance and X-rays. X-ray CAT scanners showthe greatest potential for internal imaging of logsand have been used both experimentally and inindustrial trials (Nystrom and Johansson, 1985).By rotating the X-ray source-and-detector arrayaround the log or rotating the log itself within afixed detector system, a three-dimensionalrepresentation of the internal structure can beobtained.

The quality of the lumber cut from a log is afunction of the number, size and type of internaldefects; for example branches or knots. Recentwork at BRE has highlighted the particularinfluence on machine grading stiffness of certaintypes of knot towards the base of Sitka sprucetrees; it has also highlighted the influence ofparameters that can only be directly measured bymicroscopic examination (eg microfibril anglewhich is quite poorly indicated by variables suchas rate of growth and density).

It is possible to rotate a log so that, onconversion into planks, branches form less criticalknots; however this procedure would do little tosolve the problem of highly disturbed grain whichoccurs at large whorls. Species such as Sitkaspruce tend to have a relatively high number ofquite small branches that are both difficult todetect and to avoid. Clearly logs that haveunusually large branches or other abnormalgrowth characteristics are worth diverting awayfrom structural timber production. Other types ofinternal defect in logs include rot and splits, thesebeing particularly relevant in hardwoodprocessing. Internal imaging of logs isundoubtedly an impressive and rapidly developingtechnology that will become more commonly usedin the future as the need for sorting logs increasesand the equipment becomes more affordable.

2

Figure 1 3D log shape scanner based on laser sheet-of-light triangulation(Image courtesy of Microtec)

Figure 2 Cross-sectional image ofa softwood logproduced by X-rayCAT scanner(Image courtesy ofMicrotec)

Strength grading of timber

The process of strength grading is indicative butdoes not give the actual strength of a piece oftimber – this can only be achieved by testing thepiece to destruction. The aim of strength grading isto assess the timber against a predetermined set ofcriteria so that it can be attributed to a strengthclass or grade stress. This helps to determinevalues for structural performance that can be usedin safe design. Safety is of vital importance forstructural applications, but using forest resourcesefficiently is also important. Grading is therefore amix of maintaining suitable safety standards whilemaximising yield.

Strength grading – of whatever type – is basedon statistical distributions of not only strength, butalso stiffness and density. These data are gatheredfrom exhaustive testing of the available supply foreach species that is to be strength graded. In thecase of UK-grown Sitka spruce, for example,machine settings are based on samples suppliedfrom all the major growing regions of the UK. Itwould not be appropriate, therefore, to grade Sitkaspruce grown in another country, using UKmachine settings, without verifying the validity ofthese values.

In essence, strength grading of timber breaksdown into two distinct types.

● Visual grading of softwoods and hardwoodswhere the timber is assessed against a set ofparameters that describe a visual grade. Thevisual grade can then be attributed to a strengthclass or set of grade stresses to facilitate use byengineers.

● Machine strength grading where the timber isgraded against settings derived from a settingsmodel and the timber is automatically attributedto a strength class. Machine grading is onlycarried out on softwoods.

Visual gradingVisual grading is the longest established methodof strength grading, going back to 1952 with theintroduction of CP 112. CP 112 covered bothsoftwoods and hardwoods as well as the code ofpractice for structural use, later to becomeBS 5268-2.

Softwood gradingThere is no agreed European visual standard forsoftwood grading; all visual grading standards areNational Documents. However, since 1995,BS EN 518 has been in place. This gives guidanceon what a national visual grading standard shouldinclude. On harmonisation of the Europeanstructural codes, BS EN 518 will be withdrawnand replaced with a new version which willinclude information contained in prEN 14081.

The current method used in the UK is a knot arearatio (KAR) which is based on test results and hastwo possible visual grades, GS and SS (BS 4978).GS is the General Structural grade and is the lowerof the two grades; SS is Special Structural and thehigher grade. KAR grading is based on giving adifferent weighting to knots that are present indifferent parts of the cross-section of the timber.The cross-section of the timber is split betweentwo outer margins and the central region, knots inthe marginal portions having greater weightingthan knots in the central portion.

The parameters relating to these grades apply toall timber species that have been approved forgrading. However the same visual grade can beattributed to different strength classes dependingon the timber species, reflecting thediscrimination and robustness of the gradingsystem.

In terms of structural performance, the resultsof visual grading are conservative and the stressesattributed to a visual grade are lower than can beachieved by machine grading the same timber. Theprocess is slow compared to machine grading andeach practitioner has to be trained, tested andmonitored by a third party certification bodyapproved by the UK Timber Grading Committee(UKTGC).

Hardwood gradingHardwood grading in the UK is carried out underBS 5756 and predominantly covers tropicalhardwoods. In recent years oak and sweet chestnuthave been added. Unlike softwoods, hardwoodvisual grading is based on the ratio of the size of aknot compared to the dimensions of the face of thetimber it appears on. KAR grading has provedproblematic to implement with hardwoods as hasmachine grading using bending-type machines.

3

The future for visual gradingCompared to machine grading, visual grading isslow when carried out to the full rigours of thestandards. Greater speed and precision arepossible using modern scanning techniques; thekey to this lies in the development of softwarecapable of determining correctly, from outputdata, the features present within the timber.

The nature of this development would be a formof machine-enhanced visual grading, working tocurrent grading standards. The burden of proofrequired for suitability to grade would be todemonstrate that this process grades as well as, orbetter than, a trained visual grader.

Hardwoods are an obvious starting point due tothe mode of grading; that is, the mapping ofsurface knots and comparing them to the face inwhich they appear. However, additional hardwarewould be needed to carry out rate of growth andslope of grain determinations.

Softwood grading using new scanningtechnologies to BS 4798 would be morecomplicated as this would require determinationof the pith position in order to calculate KAR. Likehardwoods, slope of grain and rate of growthwould also need to be assessed.

Rate of growthVisual grading standards for hardwoods andsoftwoods have specific requirements for rate ofgrowth determination, although this variable initself is quite a poor indicator of wood quality.Rate of growth can be determined by an operatoror by using image analysis techniques. Recentwork at BRE has centred on developing softwarecapable of calculating rate of growth and alsodetermining the position of pith, even where thismay lie outside of the piece of timber beingassessed (Figure 3). Difficulties can arise,however, where knots appear at the board ends orthe growth pattern is irregular. Pith position hasbeen found to be a good indicator of distortion inthe form of twist (see ‘Timber distortion’ on page 11).

Machine gradingOperating principles of machine grading arecovered in BRE Digest 476. The basis on whichmachine grading is applied is statisticaldistributions for strength, stiffness and density. Allgrading machines adopt this approach (or onesimilar); for each type of machine there will be a‘settings model’, which is an algorithm or set ofalgorithms that, for any given machine setting,provide a set of data that can predict the 5thpercentile for strength, and associated meanstiffness and density. All three parameters shouldmeet, or exceed, the requirements of the strengthclass for a particular size of timber to beacceptable for grading.

Deriving the settings for a single grade (egstrength class C16 and reject) is quite simple, butwhere grading in combination (eg C16/C24/reject),the process becomes more complicated since thesets of distributions relevant to the two grades canoverlap. If the settings are not arranged so thatthere is separation between the grades, this canlead to inaccuracies. Using statistical distributionsalso explains why, for a particular strength class,different indicating parameters can be obtained.This is because the higher grade takes the stiffermaterial leaving the lower grade deficient instiffness at the C16, 5th percentile level. Toachieve adequate stiffness the machine setting istherefore raised so that once again all therequirements of the strength class are met,including stiffness. Putting it another way, the C16indicating parameter when grading C16/reject canbe relatively low, provided that none of the highermaterial is removed by attempting to grade at C24level. This may also have implications for logsorting as well as tree selection.

Strength graded timber for general use relies onload sharing: the stronger and stiffer piecessupporting the less strong and less stiff timber.Take away the stronger and stiffer material, andimpoverished samples are left. This is why thepractice of visual grading to select the betterpieces of timber and then machine grading whatremains to increase yield is not permitted.Regrading material that has been rejected at moredemanding machine grade settings in theexpectation that it will pass at lower settings is alsonot permitted.

Rejects result from determining thecharacteristic strength; that is, the strength levelfor which 95% of the sample is in excess of thisvalue. The characteristic strength is determined inorder to make the most efficient but safe use of theforest resource. By implication 5% of the timberwill fall below the safe use level when thecharacteristic strength is close to the gradestrength of the batch of timber concerned.

4

Figure 3 Screenshot of BRE developed softwareshowing rate of growth and pith prediction. The whitedot is the predicted pith position

Changes to machine grading through European harmonisationMachine grading is currently controlled byBS EN 519 and implemented under third partycertification. However, work is currentlyunderway on the drafting and approval ofharmonised structural codes, one of which isprEN 14081-1 to 4, and these will become themain standards for rectangular strength gradedtimber. Many of the parts of this draft standard arealready in the public domain for comment. Thedrafts make some significant changes to thecurrent standards.

● Method of determining machine approvalcriteriaThe main change in this area is the removal ofthe requirement to achieve a particularcoefficient of determination between themachine indication parameter and measuredstrength. In BS EN 519 there was a requirementfor a coefficient of determination of at least 0.47between machine indicating parameter andmeasured strength for each species to begraded. This gave a good indication of thepotential for the machine and indicated howwell the machine might predict strength. Whatit did not indicate was how well the model usedto derive the machine settings worked.Therefore, it was possible to have a machinethat was theoretically capable of giving anaccurate prediction of strength yet had a poormodel that resulted in inaccurate grading. Thenew method is to use a statistical approachcalled the total global matrix. In this approachthe whole grading process is evaluated andbased on how well the grade criteria are met.This judges the grading model far better thanthe previous method, so it is possible that amachine with a poor coefficient ofdetermination can be made to grade safely andappropriately to strength classes (though theconsequence might well be a very poor yield).This allows the use for certain applications ofgrading technologies that would have beenpreviously excluded.

● Derivation of machine settings (prEN 14081-4)The new approach adopted for derivingmachine settings is that they will be setnationally for the timber produced within thatcountry. Traditionally, machine settings havebeen derived in the UK by BRE and issued tothe third party certification bodies by theUKTGC. This applied to all timber that enteredthe UK regardless of species or country oforigin. On harmonisation, each countrysupplying strength graded timber will have toproduce machine settings for that timber. Asimplified method of determining the settingsrequired for particular sizes is set out for eachcountry in Part 4 of the code. Therefore, sawmills will be able to determine the setting theyneed for the timber they are grading based onthe country of origin and type of machine. Thesettings will be BS EN 338 machine settingsand not BS 5268 settings.

● Third party certificationHarmonisation of the European standardsmight end the mandatory requirement for thirdparty certification of the grading process fortimber graded in or entering the UK. There areimportant actions being taken to get this matterrectified and for mandatory third partycertification to be maintained within the UK.This is to be resolved by the UKTGC in the near future

5

Methods of machine grading

Within the UK, structural softwood timber hasbeen machine graded for over 30 years, almostexclusively by what is termed bending-type, orcontact, machines. These machines use one of twobasic approaches. The first is where the machinehas a set deflection for a size of timber andmeasures the force required to produce thedeflection. The second applies a set load for aparticular size of timber and measures thedeflection produced. In essence both methodsmeasure stiffness and this is related to measuredstrength. Three bending-type machines areapproved for use in the UK: the Computermatic,the Cook-Bolinder and the Raute Timgrader. Thebenefits of these types of machine for productionpurposes are that they are robust and show areasonable degree of accuracy in the prediction of strength.

A large number of species in a wide range ofsizes can be graded using bending-type machines.The main drawbacks are limited speed ofoperation, that they do not grade the full length ofthe board, and that their operating alignment isout-of-phase with normal mill production.Although bending-type machines do not directlymeasure the variables that are known to influencestrength such as knots or sloping grain, they havethe advantage that they directly measure theimportant criteria of stiffness, albeit the flatwisebending stiffness rather than the edgewise bending stiffness.

Current and potential new technologiesThe new technologies that are already in UK andEuropean sawmills are● stress wave grading● X-ray grading● neural network systems.

Other technologies being developed are:● microwave grading● combination machines (Figure 4).

Stress wave and dynamic gradingThe operating principle of stress wave grading isrelatively simple: the measurement of the speed ofa stress wave propagating through a timber board.This function is related to stiffness which is finallyrelated to measured strength. The stress wave istypically generated by a small impact hammer asthe timber passes through the grading machine.The plank ends are usually trimmed to give aconsistent impact target. A microphone at theopposite end of the board picks up the transmittedwave while a laser is used to determine the lengthof the board. Recording the stress wave by a non-contact laser is also possible. Either the transittime of the stress wave or the fundamentalfrequency of the induced oscillation is used in thecalculation of an indicating parameter. Since themass of the board is not measured, the indicating

6

Figure 4 Combined stress grader and timber scanner. Each board face is scanned by a scanner head comprising a single camera and a laser line projector mounted in a protective box(Image courtesy of Ersson)

parameter is an indirect measure of the stiffness ofthe board. The main benefit of these machines isthat they are arranged in line with the normalsawmill function saving space and conveyorequipment. One drawback in the technique is thatit is sensitive to moisture content. A seconddrawback is that the machine indicating parameteris not related directly to strength but to stiffness,which is then related to strength.

X-ray gradingX-ray graders assess the density of the timber,recording knot sizes and positions together withclear wood density. The source of the X-rays iseither a high voltage tube or radioactive isotope,with detection being performed by a scintillatorarray. An algorithm is used to relate the receiveddata to a machine indicating parameter which, inturn, is related to measured strength. In thelaboratory, X-ray graders have proved highlyeffective at strength grading but in the sawmillsexperience has shown that they tend to produceslightly lower yields than bending machines.

Before their full potential came to berecognised, initial work was carried out with X-ray graders as devices that could be added to thefront end of conventional bending-type machines.As non-contact machines they can grade at speedsup to 350 m/min, thus increasing productivity. Thewhole length of the board is graded and there is noneed for visual over-rides. It is these advantagesthat make X-ray machines attractive to sawmillerswho may be prepared to suffer a small drop inyield to get greater productivity.

The main drawback of these machines is thatthey are moisture content sensitive as moisturereduces the transmittance of the X-rays, hencemaking the material appear denser than it is. Forthis reason a moisture content meter may beplaced in line with the X-ray grader. Compressionwood, which tends to be denser than normal woodbut actually less stiff (Reynolds and Moore, 2004),may also adversely affect such equipment. Theformation of severe compression wood is alsoassociated with stem deviation in trees, and,therefore, sloping grain.

X-ray type machines have been approved tostrength grade European redwood/whitewood, andUK-grown Sitka spruce.

Neural network systemsAn artificial neural network (ANN) is a computer-based information processing system. In the caseof an ANN designed to grade timber, the systemacquires information from the boards beingassessed (the precise form of this data is notcritical) and relates this to the ultimateperformance. The key element is the novelstructure of the information processing system –composed of a large number of highlyinterconnected processing elements – whicheffectively learns by example. ANNs can gradeeither to an identified strength for each piece oftimber, as current machines do, or by patternmatching and attributing the timber to a strengthclass based on the knot distribution and otherfeatures. These machines can grade with greatspeed, depending on the means of capturing the data.

At least one neural network machine has been approved for grading of Europeanredwood/whitewood for the glulam industry in Austria.

Microwave gradingLike X-rays, microwaves were first investigated asan additional method for enhancing the accuracyof bending-type machines. The principle of theuse microwaves is to determine the slope of grainand to indicate the presence of knots by theirinfluence on the attenuation and phase of polarisedsignals. Measuring slope of grain alone is unlikelyto provide sufficient data to be able to predict thestrength grade of timber with sufficient accuracy,and the most likely use is in combination withother machines. However, combination machineshave drawbacks as well as benefits, and these arediscussed in the next section. Microwavemachines, like X-ray machines, are sensitive to themoisture content of the timber.

7

Combination machinesA number of grading machines combine direct orindirect stiffness measurement of boards (byactual bending or by dynamic methods) togetherwith knot detection using X-rays or sloping graindetection using microwaves. Other variants arealso possible, including using optical knotdetection on the surface of boards combined withstiffness or density measurement. The basicprinciple is that the greater the number of gradingtechnologies measuring different parameters ofstrength that can be brought together into a singlegrading operation, the more accurate theprediction of strength will be. This isfundamentally true, but there are drawbacks.Increasing the number of technologies in a singlegrading operation can result in diminishing returnswhile costs increase markedly. That is to say, thebest prediction of strength (as established by thecoefficient of determination) may be achieved bythe first method of grading – any additionalmethod results in only a marginal improvement tothe coefficient of determination and the increase inimprovement is progressively less with eachadditional method. Therefore, for combinationgrading to be effective, careful selection of thegrading technologies is needed. However, byincreasing the accuracy of a grading machine thedevice may be made more reliable. A gradingmachine which is poor at predicting strengthbecause it is not measuring certain types of defector misinterpreting variables has to operate in anover-conservative manner. Therefore, byincreasing accuracy, both yield and reliability canbe increased. Adding optical detectors to bendingtype machines which do not measure stiffness overthe entire length of the board can obviate the needfor an operator to be present to perform visualover-rides.

Machine control and output control

Most grading operations are machine controlledwhere the machine is the only effective measure ofdetermining strength and there is a system in placeto regulate that grading process. This processworks well where there is a need for flexibility, forgrading several different sizes within one shift, orfor regularly changing the species being graded.

Where there are long production runs of asingle dimension and of single species, outputcontrol has advantages. This method is favoured inthe USA. The basic principle is to grade the timberagainst a machine setting or indicating parameterbut to check the output of the grading processperiodically throughout the course of a productionrun by proof loading a number of pieces of gradedtimber. If the timber is found to be safe with regardto the proof load then the process is ‘in control’,but should the results of the proof test indicate thegrading is unsafe then corrections based onstatistical parameters must be applied.

8

Optical board scanners

Optical board scanners use a variety of cameratypes and optical detectors, together with infraredlaser dot and laser line illumination (Figure 5). Inmost cases the boards are fed through the scannerlongitudinally on a conveyor. Basic scanners canuse an array of photo-detectors to detect darkfeatures such as knots. Modern advanced scannersuse several interacting measurements to detectdefects such as knots, splits and resin pockets – onall four faces of the board simultaneously – as wellas dimensional defects.

High-resolution line cameras are used to detectdefects that are darker than the surrounding wood– such as black knots – and for dimensionalchecking. Line cameras capture the light reflectedfrom the board in a one-dimensional array, asopposed to a two-dimensional grid such as atelevision image or bitmap (Figure 6). A two-dimensional image suitable for further analysis ofvariables (eg knot size) is obtained by processingthe camera output having calibrated the system forthe board size. The board length is obtained byrelating the data acquired from the cameras to thespeed of the conveyor. The light source isfluorescent tubes (ie visible light). Colour camerasare also used for detecting differences in colour,facilitating the sorting of timber for appearance.

Data from all the various sensing devices isused to build up a false-colour image of the board(Figure 7) on which defect zones can be identified.

9

Figure 5Sophisticatedboard scanner withmultiple cameraand laser arrays(Image courtesy ofInnovativ Vision)

Figure 6 Line camera output. The width of the board is indicated by thereceived light response (red line); knots on the board face are indicated by thedrop in received light (yellow line)

Figure 7 False-colour image of board showing knot zones

Tracheid effect

When a laser beam is projected onto timber thelight is scattered along the grain – thisphenomenon being termed the tracheid effect. Theangle of the observed halo around the projectedlaser dot corresponds to the grain angle on theboard (Figure 8). This information can be used todetect disturbed grain around sound knots thatmight otherwise be difficult to detect. Arrays oflaser dots projected across the whole width of thesurface of the board can be analysed to obtaininformation on the size of the knots.

Another method based on the laser tracheid effectis to capture data from cameras viewing a laserline projected across the width of the board.Again, this form of illumination highlights lightcoloured, sound knots. The laser light is usuallyinfrared, detected by infrared sensitive cameras.Sophisticated board scanners can also use the ratioof the laser dot width to length to indicate thepresence of other types of defect such ascompression wood.

Optical board scanners may be used in a widevariety of applications such as in the flooring andfurniture industries, and in the fabrication ofglulam and finger-jointed timber. Scanners cancontrol cross-cutting equipment to removedefects, and also sorting equipment that can turnor orientate pieces of timber so that in finalproducts (eg furniture or cladding) defects arehidden. The exposed edges of products such aswindow frames can be made so that they are freeof knots, as can fixing points for screws or nails.By scanning boards prior to grading, they can betrimmed or cross-cut so that features likely tocause downgrade are removed or their effectminimised.

10

Figure 8 Tracheid effect. The laser halo indicates the timber grain angle

Timber distortion

Forms of distortion are bow, spring and twist(Figure 9).

Distortion in the form of twist is known to becaused by spiral grain (Stevens, 1961; Brazier,1965). Although the magnitude of spiral grain inUK-grown Sitka spruce, for example, varies(typically from around 3–6°), the position of thepiece of timber with respect to the pith or centre ofthe log is of equal influence. Pieces cut from thecentre of the log tend to twist more (Johansson,2002). Sorting timber purely on the basis of slopeof grain may therefore not be the entire solution.Methods of automatic laser measurement of slopeof grain have been examined by Nystrom (2002),and equipment using these techniques is currentlyundergoing industrial trials in Sweden for sortingat log, canted log and sawn timber stages. Arelatively simple algorithm is used to determinegrain angle from images of laser halos; butcomplications arise in boards that have

high numbers of knots and a wandering pith.Cooper and Maun (2004) describe a method of re-engineering boards to reduce twist.

Bow and spring in boards is caused byunbalanced longitudinal shrinkage. Both juvenilewood (Thornqvist, 1993) and compression wooddiffer from normal or mature wood because of the significant longitudinal shrinkage. Animbalance of juvenile wood or compression woodfrom one side or face of a board to another cantherefore cause distortion. Timber which is fastgrown tends to have a high proportion of juvenilewood. Compression wood is a type of reactionwood that affects conifers that have been partiallyblown over, trees on the windward side of exposed plantations, the lower parts of treesgrowing on slopes, and wood below heavybranches (Desch and Dinwoodie, 1996; see alsowww.forestry.gov.uk/compressionwood). Bothcompression wood and slope of grain can beindicated by the laser tracheid effect (Figures 8and 10).

Determining the amount of compression woodwithin a board, and its propensity to distort, basedpurely on surface imaging is problematic. It is notpossible to determine the extent of compressionwood within a board without knowledge of thegrowth ring structure. Some forms of scanner maybe capable, by viewing the board ends, ofdetermining both the proportion and position ofjuvenile wood in a board together with thecompression wood content on each of the faces.However, it may be more appropriate to adopt atimber drying and handling process that providesmaximum restraint, therefore limiting distortionrather than adopting a sorting strategy. For sometypes of glue laminated products or re-engineeredtimber, sorting using the above scanningtechniques will be particularly useful.

11

Figure 9 Types of timber distortion

Figure 10 Laser halo in compression wood (left) and normal wood (right)

L

L

L

Bow

Spring

Twist

References

Cooper G and Maun K. 2004. STRAIGHT – measures for improving quality and shape stability of sawnsoftwood timber during drying. Forestry Wood Chain Conference, Edinburgh, Sept 2004 (unpublished)

Desch H E and Dinwoodie J M. 1996. Timber – structure, properties, conversion and use (7th edn), London:MacMillan Press.

Johansson M. 2002. Moisture-induced distortion in Norway spruce timber – experiments and models (doctoralthesis). Göteborg (Sweden): Chalmers University of Technology.

Nystrom J. 2002. Automatic measurement of compression wood and spiral grain for the prediction ofdistortion in sawn wood products (doctoral thesis). Luleå (Sweden): Luleå University of Technology, 2002.

Nystrom J and Hagman O. 1999. Real-time spectral classification of compression wood in Picea abies,Journal of Wood Science (1999), 45 30–37.

Nystrom J and Johansson L G. 1985. TINA – a system for measuring dimensions and quality of logs. Internalseminar on internal defect scanning of logs, Oslo, 20–21 June 1985.

Reynolds T N and Moore G. 2004. The effect of compression wood on timber quality. World Conference onTimber Engineering, Lahti (Finland), June 14–17 2004.

Szymani R (ed). Scanning technology and process optimization – advances in the wood industry. SanFrancisco: Miller Freeman Books.

Thornqvist T. 1993. Juvenile wood in coniferous trees. Document D13:1993. Stockholm: Swedish Council forBuilding Research.

Warensjo M. 2003. Compression wood in Scots pine and Norway spruce – distribution in relation to externalgeometry and the impact on dimensional stability in sawn wood (doctoral thesis). Umeå: Swedish University ofAgricultural Sciences.

BRE Digests445 Advances in timber grading476 Guide to machine strength grading of timber

British StandardsBS 4978:1996 Specification for visual strength grading of softwoodBS 5268-2:2002 Structural use of timber. Code of practice for permissible stress design, materials and

workmanshipBS 5756:1997 Specification for visual strength grading of hardwoodBS EN 338:2003 Structural timber. Strength classesBS EN 518:1995 Structural timber. Grading. Requirements for visual strength grading standardsBS EN 519:1995 Structural timber. Grading. Requirements for machine strength graded timber and grading

machines

Draft European StandardsprEN 14081-1 Timber structures. Strength graded structural timber with rectangular cross section. Part 1:

General requirementsprEN 14081-2 Timber structures. Strength graded structural timber with rectangular cross section. Part 2:

Machine grading. Additional requirements for initial type testingprEN 14081-3 Timber structures. Strength graded structural timber with rectangular cross section. Part 3:

Machine grading. Additional requirements for factory production controlprEN 14081-4 Timber structures. Strength graded structural timber with rectangular cross section. Part 4:

Machine grading. Grading machine settings for machine controlled systems

12

BRE is committed to providingimpartial and authoritative informationon all aspects of the built environmentfor clients, designers, contractors,engineers, manufacturers, occupants,etc. We make every effort to ensurethe accuracy and quality of informationand guidance when it is first published.However, we can take no responsibilityfor the subsequent use of thisinformation, nor for any errors oromissions it may contain.

BRE is the UK’s leading centre ofexpertise on building and construction,and the prevention and control of fire.Contact BRE for information about itsservices, or for technical advice, at:BRE, Garston, Watford WD25 9XXTel: 01923 664000Fax: 01923 664098email: enquiries@bre.co.ukWebsite: www.bre.co.uk

Details of BRE publications are availablefrom:www.brebookshop.comorIHS Rapidoc (BRE Bookshop)Willoughby RoadBracknell RG12 8DWTel: 01344 404407Fax: 01344 714440email: brebookshop@ihsrapidoc.com

Published by BRE Bookshop

Requests to copy any part of thispublication should be made to:BRE Bookshop, Building Research Establishment,Watford, Herts WD25 9XXTel: 01923 664761Fax: 01923 662477email: brebookshop@emap.com

© Copyright BRE 2005March 2005ISBN 1 86081 748 3

www.bre.co.uk

Acknowledgments

The work for this Digest was funded by the Forestry Commission.Technical support was also given under a related DTI-sponsoredPartners in Innovation project on timber scanning technology.