Preparing to search
In the literature various destructive and non-destructive methods for detecting compression wood are described. Some of these methods are summarised below:
|Destructive methods||Non-destructive methods|
Image analysis of transmitted images
X-ray diffraction (Silviscan, Woodtrax)
|Measurement of outer shape of logs
Shape measurements during cutting
The presence of compression wood can be suspected when stems are leaning or have a pronounced curvature. However even straight, vertical trees with circular boles can contain large amounts of compression wood.
Some authors have tried to classify trees according to their outer shape (Low 1964, Dyson 1969). Different authors have classified specific types of curvature differently, i.e. crook and sweep, see Timell (1986, Vol. II, s 755).
Several researchers have also tried to establish correlations between the deviation from the vertical and the extent of CW-formation (Low 1964, Shelbourne 1966).
According to Low there is no close correlation between outer shape and compression wood formation. Shelbourne found that severe compression wood content was correlated to stem straightness, but slight and moderate compression wood was not correlated.
According to Timell an increment borer could be used for detection of compression wood in standing trees, by making a visual assessment of the cores extracted.If torque drilling is measured this “torsiometer” can be used to evaluate the wood-specific-gravity and hence obtain an estimate of compression wood content.
The presence of compression wood can be suspected when logs have a pronounced curvature, are oval or have an eccentric pith position in the log ends (Koch et al 1990, Timell 1986). However, compression wood can also occur in perfectly concentric stems.
According to Lundgren (2000) the external shape of a trunk is a good indicator of the internal quality of the log and is therefore a cornerstone for quality assessment of logs. With new log scanners it is possible to obtain a detailed assessment of the true shape of the log. Variables such as unevenness, taper, ovality and straightness can be used in models for classification of logs into different quality grades (Lundgren 2000, Grace 1993a, Grace 1993b).
Gjerdrum and Warensjo (2001) developed shape parameters from 3D-data that could be used for automatic detection of logs with specific curvature (different crook types). The aim with the study was to characterise the geometric shape of these logs and see how they could be separated. According to the authors logs with sharp curvatures, that are prone to contain compression wood, could be detected by using these parameters in a model.
The method that was developed by Pillow (1941) is considered an accurate and convenient method especially for detection of mild forms of compression wood. The procedure is based on the observation that compression wood is opaque to transmitted light while normal wood is translucent, (helical cavities scatter the light). The method has been found to be especially useful for ascertaining the presence of compression wood within large areas of normal wood. The method has been widely applied since it was introduced, as can be seen from the authors cited below.
There are however some limitations for the method
The best contrast is observed in species with white sapwood such as Picea spp. or Abies balsamea
In species with highly coloured heartwood detection of compression wood is difficult. Fungal infection and other stains can also have the same effect.
Low and Shelbourne (1966) examined discs in transmitted light, delineated areas of compression wood, and measured them with a dot grid.
Burdon (1975) classified Pinus radiata discs in transmitted light into 6 different grades as described before.
According to the Tappi Standard (1955, 1972) for measurement of compression wood-content, 3-6 mm discs are cut and viewed in transmitted light. The compression wood areas are outlined with a pencil and then measured with a planimeter.
Another less time consuming method was developed by Andersson and Walter (1995).
The method uses image analysis techniques to divide an image of a disc viewed in transmitted light into different compression wood grades.
The repeatability was tested by letting two people classify the compression wood content in 10 Norway spruce discs. The classification was repeated four times during one week.. In total each disc was classified eight times. The mean and standard deviation as a percentage of the mean were calculated for mild-, severe- and total compression wood content. The result from the comparison indicated a good repeatability for the total compression wood content, a rather good repeatability for the mild compression wood content and a poor repeatability for the severe compression wood content. According to the authors this indicates that the operators had difficulty in discriminating between mild and severe compression wood. Therefore it is very important that operators are well trained and use the same colour display throughout the whole investigation.
A t-test showed that there were no significant differences between the two operators.
The repeatability of the angle from the pith to the gravity of the severe and mild compression wood was also tested. The result showed that the algorithm computes the angles well. However, a t-test showed a significant difference between each operators classification of the angle for severe compression wood. However, this could be explained by the large standard deviations in three of the discs that contained small amounts of compression wood.
According to the authors the method by using transmitted light increased the differences between the different wood types. The image analysis method is both faster and more objective compared to methods using a planimeter or counting points in a dot grid.
As already described in section 1 the reddish-brown colour is used by different authors as a means of detecting and classifying compression wood. The discussion about the colour fading while drying indicates a certain degree of uncertainty about the reliability of this feature when viewed in reflected light. According to Low (1964) the use of a green filter increases the contrast between normal wood and compression wood.
Westing recommends reflected light of wavelength of 480 nm for viewing compression wood.
Timell (1986) states:
“In doubtful cases examination has to be performed with a light microscope or a good stereomicroscope”
Another approach to improve the contrast between compression and normal wood is to use different staining methods. They are widely applied in conjunction with microscopic studies.
Different staining solutions that could be used include:
This method was developed by Trendelenburg and Meyer-Wegelin in 1955. Discs are cut in the green state and then dried. The greater longitudinal shrinkage of compression wood results in compression wood appearing more depressed on the surface of the discs than the adjacent normal tracheid.
Most methods for determining the density and proportion of latewood could be used for estimation of compression wood-content.
A scanning microphotomer uses visible light to measure ring width and percentage of latewood. The advantage of this method are convenience and low cost (Windendro).
Warped lumber often contains compression wood. However the main cause of distortion may be that the lumber contains juvenile wood or spiral grain.
Mild and moderate forms of compression wood are often difficult to detect in sawn timber.
Studies have been conducted at Luleå University of Technology with the aim of predicting moisture content and density in sawn wood of Scots pine using microwaves (Johansson 2001). The primary result from his work is a well-functioning apparatus for microwave scanning of wood in a laboratory environment. On of the major problems seems to be calibration, due to the richness in the microwave signals from the inhomogeneous wood material.
By using microwaves it is possible to measure electromagnetic properties of wood. Dry wood is a rather good electric insulator. With increased moisture content the ability to conduct electricity increase. The dielectric parameters for wood mostly depend on the density and the moisture content, but also on temperature, field frequency and the field orientation in relation to the grain of the angle. The polarisation ability in wood is very important. Dipole and interfacial polarisation play the main role at frequencies in the range from 105 – 1010 Hz .
The dipole polarisation is created by the orientation of the dipole molecules in the direction of an electric field. Interfacial polarisation arises from charges that are built up in interfaces between components like cell walls, interfibrillar channels and water. The degree of polarisation is influenced by the grain angle.
The scanner system that was developed consisted of three parts, the sensor system, the PC-system and the conveyer system. The sensor system consisted of a transmitting antenna, a receiving unit, a RF receiver, a multiplexer unit and a data processing unit.
The principle of measurement is electromagnetic transmission by a quasi plane through the wood.
To verify the results from the density measurements a medical CT-scanner was used.
The result show that microwaves are a good tool for predicting density and moisture content both as an average as well as a distribution prediction.
The very high accuracy of the density prediction indicates that it may be possible to predict strength and stiffness of wood using microwaves based on density predictions since density is one of the most important variables when predicting strength and stiffness of wood. Since compression wood tends to have a higher density than normal wood this method could probably also be used for prediction of CW-content in sawn boards.
In a Licentiate thesis from 1999 Nyström developed and tested four different image based methods for detection of compression wood.
The methods were:
|Spectral imaging||RGB-colour||X-ray||Tracheid effect|
Spectral imaging is a relatively simple method which has shown good results.
The use of RGB-colour scanning also shows promising results but with some variations depending on the moisture content as in sap and heartwood. Tracheid effect scanning shows very promising results on dry wood surfaces and X-ray imaging can be used to show compression wood of higher concentration inside dry timber (see table above).
According to Nordic Timber (Anon 1997) , the Nordic grading rules for sawn timber, compression wood should be considered during grading if the darker part covers more than 1/3 of the width of the year ring (annual increment) and has affected the shape of the board resulting in distortion.
The compression wood volume in the board is estimated by multiplying the compression wood area on the faces and thus constructing an imaginary box that encloses the compression wood within the board.
In a Doctoral thesis from 2001 Ohman states:
“If the exact position of compression wood within the sawn product could be determined before sawing all problems related to compression wood could be avoided. Examples of such techniques are X-rays and gamma-radiation, nuclear magnetic resonance and microwaves. X-ray techniques or gamma-radiation are both capable of depicting the internal density variation well, but the unknown amount and distribution of water makes it impossible to identify areas of compression wood with the level of accuracy needed.
Nuclear magnetic resonance is a method which measures the water content above the fibre saturation point and is therefore not capable of detecting compression wood.
Using microwaves it is possible to separate moisture content and wood density. In samples of the size of a log the damping of the signals is large and the ability to detect any density variations is small.”
In his studies Ohman found that by using the secondary features related to compression wood, such as magnitude of log sweep, ovality and amount of visible compression wood in log ends, he could achieve a rough indication of the amount of compression wood.
The relationship between the amount of compression wood in the sawn products and different log features were generally weak. Strongest correlation to the total amount of compression wood were the magnitude of log sweep and the amount of visible compression wood-content in the butt-end cut of the log. But in neither of the papers did the correlation coefficient exceed 0,60. Other features such as ovality in top and butt ends displacement of the pith and amount of juvenile wood demonstrated an even poorer correlation to the amount of compression wood.
The location of compression wood within a single plank can be predicted by the shape of the green plank. A convex shaped plank is a very strong indicator of the presence of compression wood and the larger the observed convex bow, the more compression wood can be expected.
According to one of the papers in his thesis “Modelling compression wood in sawn timber of Scots pine and Norway spruce” it is possible to separate logs with large amounts of compression wood from those with small amounts. The best prediction models (PLS – Partial Least Squares) of compression wood content in boards had an R2 value of 0.66 for Scots pine and 0.64 for Norway spruce. The variable that contributed the most was the size of the span that occurred between the centreboards directly after sawing (green-span). Compression wood-content in the butt end also contributed to the model.
The conclusion from the study was that these PLS-models had better predictability than the compression wood-log variable that is in use in Sweden today. However, this variable green-span is impossible to detect during log grading since it appears after sawing. Therefore this method can not be considered as a better method for log grading..
No correlation between compression wood content and twist was found.
According to his conclusions compression wood content in the butt end of the log is a rather poor indicator of the quality of the sawn products if only warp is considered.
Literature review performed by:
Swedish University of Agricultural Sciences
Departmentof Forest Management and Products
901 83 UmeÃ¥