Drying and Re-conditioning of
Pre-twisted Boards.
Laboratory and Industrial Tests.
Report on WP 2.4 and WP 2.7 of the STRAIGHT project.
SP Building Technology and Mechanics SP REPORT 2005:14
SP Swedish National T
Drying and Re-conditioning of
Pre-twisted Boards.
Laboratory and Industrial Tests.
Abstract
Twist is currently one of the main quality problems affecting softwood timber for construction. Several methods to minimise this problem were investigated in an EU-project. The results for one method, drying boards in a pre-twisted position, are presented in this paper. It has generally been found that boards (Norway spruce) can be deformed towards straightness using this method. The tendency of a board to twist during drying is primarily dependent on the grain angle of the board. Laboratory tests have shown how much a board should be pre-twisted during drying in order to achieve a straight result. In practice, almost all studs tend to twist in the same direction. Pre-twisting the entire kiln stack equally much, gives a better average result than the conventional process. By measuring the grain angle for each board (“tracheid effect” with a laser beam) boards can be grouped and then treated “individually”.
The force exerted by a board on its holders during drying in a pre-twisted position has been measured. The force decreases considerably during the heating phase, remains rather constant during the drying phase and decreases during cooling. When the variation in torsional stiffness as a function of temperature and moisture content is accounted for, it is found that a creep deformation of the board occurs both during the drying phase and the cooling phase. Deformation due to changes in temperature seems therefore to take place in addition to mechano-sorptive creep. This has been proved by heat treating boards (wet or dry) wrapped in plastic to avoid changes in moisture content. Boards that still show excess twist after drying can therefore be “corrected” using a heat treatment process. It was also observed that boards exhibit a memory effect in this context. The wood material seems to remember processes which the board was exposed to at earlier stages. This memory fades over time.
The results obtained in the laboratory have been verified by a series of industrial tests, including the effect of pre-twist on an entire kiln stack. Recommendations for industrial implementation and an economic assessment are finally presented.
Key words: Twist, drying, pre-twist, grain angle, creep, torsional stiffness, memory effect, heat treatment, Norway spruce
SP Sveriges Provnings- och SP Swedish National Testing and Forskningsinstitut Research Institute
SP Rapport 2005:14 SP Report 2005:14 ISBN 91-85303-45-3 ISBN 91-85303-45-3 ISSN 0284-5172 Borås 2005 Postal address: Box 857,
SE-501 15 BORÅS, Sweden
Telephone: +46 33 16 50 00
Telex: 36252 Testing S
Telefax: +46 33 13 55 02
Contents
Abstract 2 Preface 4 Sammanfattning 5 1 Introduction 7 2 Laboratory tests 7 2.1 Preliminary investigations 7 2.2 Experimental set-up 92.3 Preliminary drying tests 9
2.3.1 Freely moving boards 9
2.3.2 Boards dried pre-twisted 10
2.4 The pre-twist deformation mechanism 12
2.5 Predicting the influence of pre-twist 20
2.6 Small scale tests 22
2.7 Torsional stiffness as a function of temperature and MC 26
2.8 Analysis of the twist deformation mechanisms 29
2.9 Conclusions from the laboratory tests 35
3 Industrial tests 36
3.1 Drying tests 36
3.1.1 First test 36
3.1.2 Second test 38
3.2 Heat treatment tests 41
3.2.1 First test 41
3.2.2 Second test 43
4 Conclusions and recommendations 46
5 Economic assessment 47
Preface
The work presented in this report relates to the European shared-cost project STRAIGHT – “Measures for improving quality and shape stability of sawn softwood timber during drying and under service conditions” within the EU’s 5th framework programme, contract number QLK5-2001-00276. The report summarizes the results of two work packages, WP 2.4 “Twisted pack drying” and WP 2.7 “New conditioning techniques”.
The measurements presented in this report have been made at AB Trätek, Swedish Institute for Wood Technology Research, which in 2004 was merged into SP Swedish National Testing and Research Institute, in Stockholm, Sweden and at a number of Swedish sawmills.
Sammanfattning
Skevhet är för närvarande ett av de viktigaste kvalitetsproblemen vad gäller sågat barrträdsvirke för byggnadsändamål. I ett EU-projekt har flera metoder undersökts för minimering av detta problem. Resultat med en sådan metod – torkning av reglar i ett motskevat läge – presenteras i denna rapport. Det har generellt framkommit att
granbräder kan bli deformerade i riktning mot rakhet med denna metod. Tendensen hos en bräda att bli skev under torkningen beror primärt av fibervinkeln. Baserat på
laboratorietester har det bestämts hur mycket en bräda bör motskevas under torkning för ett rakt resultat. I praktiken tenderar nästan alla reglar att skeva åt samma håll. Genom att motskeva hela stapeln lika mycket i torken, så kan ett bättre medeltal uppnås än med en konventionell process. Genom att mäta fibervinkeln för varje bräda (enligt “trakeid-effekten” med en laserstråle) så kan grupper formas vilka ges en “individuell” behandling.
Kraften som utövas av en bräda på dess infästningar under torkning i motskevat läge har mätts upp. Kraften minskar väsentligt under uppvärmningsfasen, hålls ganska konstant under torkningsfasen och minskar igen under nedkylningsfasen. När variationen i vridstyvhet som funktion av temperatur och fuktkvot beaktas, så finner man att en krypdeformation sker både under torkningsfasen och under nedkylningen. Utöver mekano-sorptiv krypning, verkar en deformation utlöst av en temperaturändring att ske. Detta har bevisats genom en värmebehandling av bräder (våta eller torra) insvepta i plast för att förhindra en förändring av fuktkvoten. Bräder som fortfarande efter torkningen är kraftigt skeva kan alltså “korrigeras” med en värmebehandlingsprocess.
Det har även observerats att bräder uppvisar en speciell minneseffekt i detta
sammanhang. Trämaterialet verkar minnas processer som påverkat brädan i tidigare stadier. Detta minne avklingar med tiden.
Resultat erhållna i laboratorieskala har verifierats i en serie industriförsök inklusive effekten av motskevning av en hel torkstapel. Slutligen ges några rekommendationer för en industriell implementering samt presenteras en ekonomisk lönsamhetsbedömning.
1
Introduction
Twist is at present one of the main quality problems for construction softwood timber. The EU funded project STRAIGHT has investigated different methods to minimise this problem. Results obtained with one method, drying and thermal treatment of boards in a pre-twisted position, are presented in this report. This report covers the work packages WP 2.4 and WP 2.7 of the STRAIGHT project.
It is well known that boards in the top layers of a kiln stack will be more twisted after drying than boards further down the stack, if there is no top load on the stack. The reason is that boards in the lower part of the stack are kept straight during the drying process by the weight of the timber above, while boards at the top are more or less free to move. This raises the question whether boards kept pre-twisted during the process, i.e. twisted in the opposite direction to their inherent twist direction, could be used as a method for
minimising twist. This idea is the subject of the research work that is reported in the following article.
2
Laboratory tests
The research work started with a laboratory scale experimental investigation. The aim was to find out whether the basic idea was correct, and if it is, to determine the amount of pre-twist needed in different cases. The aim was to also obtain a better understanding of the phenomena involved in the deformation of the board towards straightness.
In the second stage, the results from the laboratory work were tested and verified at an industrial scale. The results from the industrial tests are presented in section 3.
2.1
Preliminary investigations
Preliminary tests were carried out to establish the order of magnitude of the forces needed to pre-twist the boards and the amount of pre-twist required making boards straight after drying. The results were used to design the equipment used in the main testing series and to select the cases that would be tested.
The relationship between board torsional moment and the torsional angle is important. Figure 1 presents this relationship for a number of 48 x 100 mm2 green Norway spruce boards. The average graph is given in Figure 2.
0 5 0 10 0 15 0 20 0 25 0 30 0 35 0 40 0 45 0 0 5 10 15 20 25 3 0 Twist, degrees/meter
Figure 1. Relationship between twist and corresponding torsional moment (Nm) for a number of 48 x 100 mm2 green Norway spruce boards.
0 50 100 150 200 250 300 350 400 0 5 10 15 20 25 30 Twist, degrees/m Torsional moment, Nm
Figure 2. Relationship between torsional moment and twist for 48 x 100 mm2 green Norway spruce
The above figures show that a linear relationship exists up to around 15o/m. A comparison of board densities in Figure 1 clearly indicates that higher density boards require higher moments to achieve the same torsion. The relationship is of course also dependent on moisture content and on temperature, which will be discussed later in the report.
The relationship between torsional moment and twist angle for the linear (elastic) part of the graph in Figure 2 can be expressed as
K
G
l
M
⋅
⋅
=
ρ
(1))
b
/
a
(
F
3
b
a
K
3⋅
⋅
=
a≥b (2) where ρ = Torsional angle (radians)M = Torsional moment (Nm)
l = Length of twisted part of the board (m) G = Modulus of torsion (MPa = N/mm2)
K = Factor depending on the cross sectional area and shape (mm4) a = Board width (mm)
b = Board thickness (mm)
F(a/b) = Factor for rectangular cross sections, found in textbooks on the subject (-). (a/b = 2 gives F(a/b) = 0,686).
From Figure 2 and from additional measurements for green Norway spruce made at room temperature, G is found to be around 500 MPa.
2.2
Experimental set-up
The main test series was performed in a laboratory dryer in which up to 4 boards can be dried in a twisted position. A few boards which are free to move can be dried at the same time. The twisted boards are fastened in a specially designed steel frame which is inserted and locked into the dryer. The boards have to be around 1,5 m long and both ends are fastened by bolts in holders so that the effective twisted length is 1,37 m. The holder in one end is fixed and the other can be turned around an axis (in the board direction) and fixed in the position required. The torsional moment needed to twist the board into that position could be measured.
The drying schedules used in the tests were quite normal with a gradually increasing wet bulb depression. It should, however, be mentioned that the schedule structure was based on “constant dry bulb, decreasing wet bulb temperature” and not on “constant wet bulb, increasing dry bulb temperature”, which still is more common in practice.
2.3
Preliminary drying tests
To check the general validity of the pre-twist approach, two boards were dried using different drying schedules. The pre-twist used in both cases was 17,5o/1,37 m. The grain angle and the final MC were 0,5o, 15,8% and 3,6o, 17,8% respectively. After drying and immediately after removal from the steel frame, the remaining twist (from 17,5o) was14o and 11o respectively. During the subsequent first hours, the remaining twist decreased 1o but remained after this unchanged for several days. This preliminary result indicated that a considerable induced (counter)-deformation could be achieved during the drying process. This deformation also appeared to be reasonably stable. These results clearly justified a more thorough investigation of the pre-twist approach.
2.3.1
Freely moving boards
It is necessary to have a reference when determining the amount of pre-twist required, i.e. the inherent tendency of a board to twist, as a function of relevant board properties. The most logical reference is the amount of twist a board which is completely free to twist and
move experiences. It should be noted that drying in a fixed straight position (as in a normal kiln stack) could be considered as drying with zero degrees pre-twist. This inherent tendency to twist was investigated by drying 16 spruce boards of
dimensions 47 x 100 mm2 and lengths of around 1,5 m. Properties measured were grain angle (by the scratching method on the outer board surface) final MC, density, distance from the pith and twist both before and after drying. Regression analysis showed a significant relationship between grain angle and twist after drying, but no other
correlations. This is surprising as both the final MC (shrinkage) and the distance from the pith should, according to theory (Stevens et al, 1960) also exert an influence. This may be due to the small variance of these variables and the limited number of boards. The
correlation found can be expressed as
Θ ⋅ = ρ 1,220 (R2 = 0,81) (3) where ρ = Twist (o/m) Θ = Grain angle (o ). The result is also presented in Figure 3.
-2 0 2 4 6 8 10 -2 0 2 4 6 8
Grain angle, degrees
Twist, degrees/m
Figure 3. Twist after drying as a function of the grain angle for boards that are free to move.
It should be noted that Eq.(3) predicts the amount of twist for boards positioned at the top of a kiln stack where there is no top loading.
2.3.2
Boards dried pre-twisted
A series of drying tests carried out with pre-twisted boards were performed in the above described drier. The results for 18 of these boards will be presented here. The properties measured in the previous section were measured for each board. The pre-twist angle used was of course also recorded. When the results were analysed, only the pre-twist angle and the grain angle were found to significantly influence post drying twist. As previously, no correlation with final MC or the distance from the pith was found and this is perhaps due to the same reason mentioned above. It should also be mentioned that the structure (and temperature levels) of the drying schedules used, were almost the same in all cases.
The correlation found based on these tests can be expressed as p
812
,
0
510
,
0
⋅
Θ
−
⋅
ρ
=
ρ
(R2 = 0,89) (4) where ρp is the pre-twist used (o/m) which was kept constant during the drying process.The result is illustrated by Figure 4, which compares measured and predicted twist after drying. -12 -10 -8 -6 -4 -2 0 -12 -10 -8 -6 -4 -2 0
Predicted twist, degrees/m
Measured twist, degrees/m
Figure 4. Measured and predicted (Eq.(4)) twist after drying for tests with pre-twisted boards.
As shown, the pre-twist was in all cases too high, i.e. the boards were made to twist in the opposite direction than the “normal”. This quite clearly shows that it is possible to “deform” a board by a pre-twist operation, so that it becomes much straighter than it would have been without pre-twisting. This applies to the period shortly after the drying process is completed. Long-term behaviour has to be separately investigated.
Figure 5 presents Eq.(4) in graphical form and also includes the behaviour of boards that are free to move (Eq.(3)).
-6 -4 -2 0 2 4 6 -2 0 2 4 6 8
Grain angle, degrees
Twist after drying, degrees/m
0 degr./m 2 degr./m 4 degr./m 6 degr./m 8 degr./m free
Figure 5. Twist after drying as a function of grain angle and pre-twist. The figure shows that forcing the boards straight during drying, as occurs in the lower part of a normal kiln stack (zero degree pre-twist) reduces post drying twist by almost 60% when compared with boards that are free to move.
If the target is no twist shortly after drying, then Eq.(4) can be solved to give the pre-twist needed as a function of the grain angle:
Θ
⋅
=
ρ
p0
,
628
(5)These results apply for final MC targets (around 15%) sawing pattern (studs close to the pith) and drying schedules that are reasonably close to those used in the test series. The situation may be different if considerable deviations occur and the results should
therefore be considered to only be indicative. The new findings reported in the following sections also have an important influence on the results.
2.4
The pre-twist deformation mechanism
Most of the tests described above were carried out using a fixed pre-twist. The force exerted by the board on its holders was not measured during the drying process, except for the initial force needed to twist the board before the drying process, which was measured in some cases. However, during the final part of the test series, this force (the torsional moment) was continuously measured during the process. Due to short and therefore not very well defined lever lengths and other practical problems, the overall calibration of the measurement procedure was inaccurate. Absolute values are therefore not reliable. However, relative changes for a board in a given test should reflect the real situation well.
Figure 6 presents the development of the forces exerted by four boards in the first drying test. Measurements were made using the devices mentioned. The aim was to find out at which stage of the drying process the pre-twist induced deformations actually occurred. The result was surprising and will be discussed in more detail, as it strongly influenced further investigation of the pre-twist approach.
0 0,2 0,4 0,6 0,8 1 1,2 02-02-19 12:00 02-02-20 00:00 02-02-20 12:00 02-02-21 00:00 02-02-21 12:00 02-02-22 00:00 02-02-22 12:00 kN Load cell 1 Load cell 2 Load cell 3 Load cell 4
Figure 6. Force exerted by four different pre-twisted boards on their holders during drying.
At the start of the drying process, high force levels occur which decrease rapidly during the heating period (around 6 hours). There are two possible and easily accessible explanations why this decrease occurs. Firstly, as the wood temperature increases it becomes softer and the force needed to keep the board pre-twisted decreases. Secondly, a creep deformation may occur. However, as the wood MC is almost constant during this heating phase, mechano-sorptive creep is unlikely. When the real drying process starts at the end of the heating period, there is almost no change in force. This is surprising, as the conditions for mechano-sorptive creep (external force + moisture content change) are clearly fulfilled. During this phase, the wood MC decreases, which makes the wood stiffer. There may therefore be a balance between the change in stiffness and a creep process. Finally, an increased force level would be expected during the cooling period, as the wood becomes stiffer with decreasing temperature. However, the force on the
contrary decreases, indicating a different type of creep behaviour than mechano-sorptive creep, as the MC should have remained almost unchanged during this phase.
In summary, this simple test strongly indicated that
a) Mechano-sorptive creep as a deformation mechanism in the pre-twist process appears to be less important than expected.
b) Deformations instead appears to occur during changes of temperature, i.e., during the cooling and possibly also during the heating phases. These findings required further investigation and changes were therefore made to the original research plan.
The first check used boards wrapped in plastic film, so that there would be almost no change in MC. This therefore represents a “heat treatment” of the boards and not a drying process. In addition, two of the boards were green (not dried) and two had been
previously dried (dried in the previous test). All boards were equally pre-twisted, i.e., the dried boards were now subjected to an additional twist deformation.
As shown in Figure 7, the temperature was first increased to around 70oC and then kept constant at this temperature for 3 hours. After that the temperature was increased to around 80oC. The cooling phase was started after further 3 hours at 80oC. The wood
temperature was simultaneously measured. A time lag is of course seen in these graphs. Due to this time lag, the change in temperature from 70 to 80oC is not very distinct.
0 10 20 30 40 50 60 70 80 90 100 0,00 3,00 6,00 9,00 12,00 15,00 18,00 Time, h
Temperature and force
Wet board temp. Dry bulb temp. Dry board temp. Wet bulb temp Load cell 1 Load cell 2 Load cell 3 Load cell 4
Figure 7. Forces exerted by two wet and two dry boards and air and wood temperatures during the drying process.
An examination of the force curves of Figure 7 shows a similar pattern to that shown in Figure 6. In both cases, around half of the original force on board holders has disappeared during the heating phase. However, in the Figure 7 case, mechano-sorptive creep is excluded and the changes seem to be due to a temperature change (in addition to the softening of the wood at higher temperatures). A slight additional change may perhaps be detected during the step from 70 to 80oC. During the cooling phase, a further decrease in force is initially recorded with a subsequent recovery in one case. After this heat
treatment is completed, the MC and wood temperature are essentially the same as before the treatment. However, the force exerted by each board on the holders has decreased by 60-70%. This means that around the same amount of pre-twist has been realized as a deformation of the board. This deformation has occurred at least during the cooling phase, but perhaps also during the heating phase. Temperature change therefore seems to be an important factor.
It is also remarkable that both the green (above FSP) and the dried (below FSP) boards exhibit this behaviour. There is no obvious difference in the shape of the curves of Figure 7. A deformation towards straightness could therefore be achieved before, during or after the drying process, by a special treatment of the board in a twisted state.
The first test of this section was repeated with green boards, but with pre-twist angles closer to the optimal angle for post drying straightness. The results are presented in Figure 8.
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 Time, h Tem p
erature and force
Dry bulb Wet bulb Load cell 1 Load cell 2 Load cell 3 Load cell 4
Figure 8. Forces exerted by four drying boards and air temperatures.
Figure 8 shows that the development of torsional force is in three cases very similar to the pattern seen in the first test (Figure 6). The fourth board shows a quite different curve, indicating a small realized deformation after drying. However, when this board was measured, only a small final twist was found which was at the same level as the three other boards. This shows that the load recorded by this load cell was incorrect, and illustrates well the practical problems related to these kinds of measurements. It has to be remembered that the load cells used in these tests are loaded for a considerable period of time and are exposed to a hot and very humid environment. It should also be mentioned that the ends of the boards are slightly deformed by holder clamping forces, which in addition to wood shrinkage cause small movements that are superimposed on the ideal behaviour. The 1,4 meter distance between the holders is the same as that in a real kiln stack, with a sticker spacing as wide as 1,4 meter.
A great deal of calibration work was carried out and related steps were taken to improve the reliability of the continuous load and other measurements during the process. This is not reported here. However, we can summarise this work by stating that especially the torsional force measurement is not fully reliable and should only be considered to be relative and indicative. The direct measurement of the twist after drying (or heat treatment) is of course simple and reliable. However, the amount of realized counter-deformation as a function of processing time is therefore not fully reliable. This makes determining active deformation mechanisms difficult to establish in detail.
The following test used four green boards, two of which were dried in a straight state (zero pre-twist) and two being pre-twisted according to their measured grain angle. The drying schedule now includes a short increase of temperature to 105oC at the end of the drying cycle. The result is presented in Figure 9.
-20 0 20 40 60 80 100 0 10 20 30 40 50 60 70 Time, h
Temperature and force
Dry bulb Wet bulb Load cell 1 Load cell 2 Load cell 3 Load cell 4
Figure 9. Forces exerted by four drying boards, two pre-twisted and two fixed in a straight position.
Due to the small grain angles in this test, both the imposed and the gradually developing torsional forces are moderate. The zero-point of the torsional force scale is therefore not well defined. The two pre-twisted boards do however also show a decrease in the force during the heating phase. The two other boards had an initial MC close to the FSP. These boards therefore started to develop twist (torsional moment) early, which was then suppressed by the change in temperature effect. This is probably what is observed in the initial part of these curves.
The torsional force curves clearly show the responses to the temperature peaking at the end of the process. However a common trend is difficult to find. The reason for this may be that force levels are low and responses are obscured by secondary effects.
The next test was performed using two green boards and two dried boards, all wrapped in plastic film to avoid changes in moisture content during treatment. The heat treatment consisted of a rapid increase to 105oC in around 2 hours, maintenance of the temperature at this level for 4 hours before the cooling phase was started. The results are presented in Figure 10.
-25 0 25 50 75 100 125 150 175 200 225 0 1 2 3 4 5 6 7 8 9 10 11 Time, h Tem p
erature and force
Dry bulb Wet bulb Load cell 1 Load cell 2 Load cell 3 Load cell 4
Figure 10. Forces exerted by two green boards and two dried boards, all wrapped in plastic film and subjected to a short period heat treatment.
During the heating phase, a substantial decrease in measured load was recorded. A further decrease was recorded during the cooling phase. Due to the high temperatures and partly high load levels, these changes are clear and distinct. The results therefore clearly indicate that imposed twist and a simultaneous change in wood temperature creates a deformation in the direction of the external force. As the wood becomes softer at higher temperatures, it is unclear how much of the deformation occurs during the heating phase and how much during the cooling phase.
The next is a special test carried out to establish the importance of pure viscoelastic deformation during the pre-twist process. Four green boards were wrapped in plastic film, pre-twisted and kept at room temperature (constant) for 88 hours. The results for three of the boards are shown in Figure 11. The results for the fourth board were lost due to technical problems.
0 20 40 60 80 100 120 140 160 180 200 0 10 20 30 40 50 60 70 80 90 Time, h Tem p
erature and force
Load cell 1 Load cell 2 Load cell 3 Temperature
Figure 11. Forces exerted by three boards in a constant climate.
A slight decrease in torsional force is seen at the beginning of the test. However, it is quite clear that viscoelastic behaviour is of minor importance, at least at these
temperatures. This is a strong indicator that the deformations achieved in previous tests cannot be attributed to pure viscoelastic properties of wood.
The above viscoelastic test was continued using a two-cycle heat treatment. The temperature was first raised to 100oC and maintained at this level for around 3 hours. After a cooling phase, the temperature was again increased to 100oC and maintained at this level for a further 3 hours, before final cooling was started. The development of torsional forces is shown in Figure 12.
0 20 40 60 80 100 120 140 160 180 0 5 10 15 20 Time, h Tem p
erature and force
Dry bulb Load cell 1 Load cell 2 Load cell 3 Wood temp 3 Wood temp 4
Figure 12. Forces exerted by boards wrapped in plastic and subjected to a two-cycle heat treatment.
The load cell curve pattern is similar to earlier results during the first heat treatment cycle. The reduction in force during heating and cooling seems not to have been influenced by the preceding twisting at constant room temperature. The graph shows a delay in the response of the force with respect to the change in temperature. This is probably due to a deviation in time registrations between different data acquisition systems.
However, what is remarkable is that the second heating phase did not cause any significant additional changes in the force exerted by the boards. A change due to softening of the wood at higher temperatures was at least expected. Either softening is balanced by a recovery of the previous twist deformation or the stiffness properties of the boards have changed during the process. The second cooling phase, however, produced a new decrease in the torsional force similar to the decrease experienced during the first cooling phase, but perhaps not as strong.
After the second heat treatment cycle, the boards were released from the steel frame and deformations were measured. The boards (again wrapped in plastic film) were then put back into the frame and pre-twisted to around 6o/m from the free position, i.e. pre-twisted more than in the previous part of the test. Then a third heat treatment was carried out, temperature being raised to 100oC. The results are shown in Figure 13.
0 20 40 60 80 100 0 2 4 6 8 10 12 Time, h
Temperature and force
Dry bulb Load cell 1 Load cell 2 Load cell 3 Wood temp 3 Wood temp 4
Figure 13. Forces exerted by boards wrapped in plastic and subjected to a third heat treatment cycle with increased pre-twist.
The third heating phase caused a decrease in the force exerted by the boards which was similar to the first heating phase (although perhaps with a slightly lower efficiency) but different to the second heating phase. The increase in the pre-twist angle seems to have made further deformation possible. A further decrease in force was recorded during the third cooling phase, similar to that experienced in the first cooling phase but a little stronger than the decrease in the second cooling phase.
Finally, the plastic film was removed from the boards. The boards were measured and placed in the dryer and dried. Two of the boards were dried in a twisted position, but twisted in the opposite direction to earlier treatment. The other two boards were free to move during
the drying process
. The results are presented in Figure 14. Due to a failure in the steam supply system, the wet bulb temperature dropped considerably after around30 hours. This caused some checking of the boards (freely moving boards). Due to shrinkage, one of the boards came loose and dropped out of the holder, probably close to the point where the force was almost zero (load cell 1, lower curve in Figure 14). The final part of this load curve does thus not reflect the real twist of that board.
-10 10 30 50 70 90 110 130 150 0 10 20 30 40 50 60 70 Time, h
Temperature and force
Dry bulb Wet bulb Load cell 1 Load cell 2
Figure 14. Forces exerted by boards during drying and drying schedule realized. The load curves in Figure 14 show the same general behaviour recorded in earlier
measurements. The inherent tendency of the board is to twist and as the pre-twist acted in the same direction, the load dropped to zero and even lower. Whether a memory effect is involved in this behaviour cannot be stated without further analysis.
2.5
Predicting the influence of pre-twist
The results from the main test series described above, together with the results from the earlier preliminary investigation, can here be summarised in a few simple equations. Analyses show that twist deformation from drying processes and twist deformation from heat treatment processes (without drying) must be separated.
For drying processes the following equation is found
p 0
1
−
ρ
=
0
,
650
⋅
Θ
−
0
,
872
⋅
ρ
ρ
(R2 = 0,956) (6)where
ρ0 , ρ1 are twist (o/m) before and after drying
Θ is grain angle (o
) ρp is pre-twist used (
o
/m)
Eq.(6) is reasonably close to Eq.(4), which was based on the behaviour of green boards that were straight before they were dried. No statistically significant correlation was found with density, final MC or temperature change. The reason may be that the variation of these parameters was too small to define dependence.
-12 -9 -6 -3 0 3 6 9 12 -12 -9 -6 -3 0 3 6 9 12
Predicted twist, degrees/m
Measured twist, degrees/m
Figure 15. Measured and predicted (Eq.(5)) twist after drying for tests with pre-twisted boards.
The pre-twist required to compensate for a given grain angle is deduced from Eq.(6) as
Θ
⋅
=
ρ
p0
,
745
(7)which is more than shown by the previous results (Eq.(5)).
For heat treatment processes, tests with boards wrapped in plastic film, the following equation is found
p 0
1
=
ρ
−
0
,
695
⋅
ρ
ρ
(R2 = 0,88) (8) It should be noted that there is no statistically significant correlation with the grain angle in this case. This is quite logical. There is no correlation with density or MC in this case either. It should also be noted that the coefficient in Eq.(8) has a lower value than the corresponding coefficient in Eq.(6). This perhaps indicates that mechano-sorptive creep has an influence on the twist deformation in drying processes.However, the constant temperature test (Figure 11) clearly shows that the factor 0,695 in Eq.(8) in reality is dependent on the magnitude of wood temperature change. If this is accounted for, then the following equation is obtained
T
00867
,
0
p 0 1=
ρ
−
⋅
ρ
⋅
∆
ρ
(R2 = 0,86) (9) where ∆T is the change in wood temperature (oC
)
during heat treatment. A comparison of predicted (Eq.(9)) and measured twist deformation during heat treatment is given in Figure 16.0 2 4 6 8 10 0 2 4 6 8 10
Predicted twist, degrees/m
Measured twist, degrees/m
Figure 16. Measured and predicted (Eq.(9)) twist deformation after heat treatment. The above results apply to 50 x 100 mm2 Norway spruce boards close to the pith, in the temperature range 20-105oC and, for drying processes, with a final MC of about 14-15%.
2.6
Small scale tests
An additional test series was carried out to obtain a more diversified picture of the mechanisms operating during heat treatment of a twisted piece of wood. Only very small pieces of wood were used in this test series. Length was 335 mm, width was around 30 mm and thickness was 4 – 6 mm.
In the main test series described above, the pre-twist angle was kept constant and torsional moment varied during the process. In these additional tests, torsional moment was kept constant and the twist angle was varied. The wood sample was fastened at its midpoint and levers were attached to both ends of the sample. Weights were further attached to the levers in such a way that they could easily be removed and re-attached. Twisting acted in opposite directions so that the force at the (fixed) sample midpoint was almost zero. The downward movements of the diagonally opposite corners of the samples were measured and the sum of these values was taken as a (relative) measure of twist. The intention was to support the wood sample in such a way that only pure twisting occurred without any simultaneous bending deformation. This was, however, not completely successful. The “twist” values are therefore partly affected by simultaneous bending. If bending deformation is proportional to the twist deformation, which is a plausible assumption, then the measured values are still a useful relative measure of twist. The wood samples in these tests were wrapped in plastic film to avoid changes in
moisture content. The MC of the dried samples remained in the 8-12% range during testing. The entire testing device was placed in a small heating chamber, where the temperature could be regulated. Due to the small sample size, it can be assumed that the wood temperature is very close to the surrounding air temperature in the chamber.
The first small-scale test is shown in Figure 17, which shows twist (as defined above) as a function of temperature. The test starts in the lower left corner at about 28oC with an unloaded sample (twist = 0). When the sample is loaded there is a step increase in twist.
This is therefore a pure elastic response. The temperature was then gradually increased in the chamber to around 80oC, after which it was allowed to cool. Twist was measured a few times during this process. The last point is recorded after unloading and the remaining twist therefore represents the twist deformation achieved during the process.
0 1 2 3 4 5 6 20 30 40 50 60 70 80 90 Temperature, C Twist
Figure 17. Twist development during heat treatment of a small sample.
Increased twist was expected during the heating period as the wood becomes softer. Twist was also expected to follow the same curve backwards during the cooling phase
However, twist actually increased during cooling. This shows that twist deformation occurred at least during the cooling phase and possibly also during the heating phase. The time span for the complete treatment was 106 minutes.
The second test was performed in the same way, and the results are shown in Figure 18.
0 2 4 6 8 10 12 14 16 18 20 20 30 40 50 60 70 80 90 Temperature, C Twist
The process again starts in the lower left corner with attachment of the weights. In this test, the point of unloading can be clearly seen in the top left corner. The time span from loading to unloading was 104 minutes. The time between the two last points (no torsion applied) was around 16 hours and shows that the twist deformation achieved during heat treatment was fairly stable. The results in general are very similar to those seen in Figure 17.
The third test was performed in the same way as the two previous tests, but with the following extension. Each time twist was measured, an additional measurement was made with the weights temporarily removed. This procedure, which only took a few seconds, allowed the amount of twist deformation realised at each stage of the process to be recorded. In addition, at temperature levels 40, 60 and 80oC, the temperature was kept constant for around 15 minutes and twist measurements were performed at both the beginning and the end of these constant temperature periods. Due to practical difficulties, it was not possible to achieve stable temperatures during these 15 minutes periods. The results of this test are presented in Figure 19.
0 1 2 3 4 5 6 7 8 15 25 35 45 55 65 75 85 Temperature, C Twist
Figure 19. Twist and twist deformation during heat treatment.
The upper curve in Figure 19 is derived using the same measurement procedure as was used in Figures 17 and 18 and is similar to these earlier results. The lower curve, however, shows achieved twist deformation, i.e. twist measured with the torsional
moment temporarily removed. This lower curve indicates that ¾ of the wood sample twist deformation occurred during the heating period and only ¼ during the cooling phase. Furthermore, the pairs of points at 40, 60 and 80oC show that there was almost no deformation during these 15 minutes constant temperature periods. The time span from first loading to final unloading in this test was 203 minutes, including five 15 minutes “temperature stops”. The time span between the last two points of the lower curve is 15 hours. Deformation surprisingly increased during this long final period with no load, which may be due to measurement inaccuracy.
The fourth test was performed with a green wood sample. The temperature was raised to around 100oC without any torsional load acting on the wood sample. The sample was loaded at this temperature and the temperature was then gradually reduced. Twist was measured at specific temperature levels during this cooling phase and twist deformation
was measured by temporarily removing torsional load as described above. The results are presented in Figure 20.
The fifth test was performed in the same way as the fourth, but using a dried wood sample. The results are presented in Figure 21.
0 1 2 3 4 5 6 20 40 60 80 100 120 Temperature, C Twist
Figure 20. Development of twist and twist deformation during cooling of a green sample.
0 1 2 3 4 5 6 20 40 60 80 100 120 Temperature, C Twist
Figure 21. Development of twist and twist deformation during cooling of a dried sample. The upper curves in Figures 20 and 21 represent total twist (loaded) and the lower curves represent twist deformation (without load). Both show that substantial twist deformation was achieved and the magnitude seems to be higher than during the cooling stage of the heat treatment processes illustrated in Figures 17-19. Twist deformation appears to be more pronounced for the green wood sample than for the dried sample. The time span in these tests from first loading to the end of the cooling phase was 2 – 3 hours. The time between the last two points in the diagrams was however 3 and 5 days respectively.
Samples were continuously loaded during this period. As shown, only a small additional twist occurred during this long period of constant conditions.
The sixth small-scale test was performed in the same way as the third test, but with an improved method for avoiding bending of the wood. The results are presented in Figure 22. 0 0,5 1 1,5 2 2,5 15 25 35 45 55 65 75 85 Temperature, C Twist
Figure 22. Development of twist and twist deformation during heat treatment. The upper curve in Figure 22 represents total twist and the lower curve twist deformation. Twist deformation development in this case is more pronounced during the cooling phase than during the heating phase. This differs from the results from the third test (Figure 19). One reason may be that bending has been reduced, giving a better measurement of pure twist.
In conclusion, the results of the small-scale tests can be summarised as follows. - A change in temperature is essential for the development of substantial twist
deformation.
- Twist deformation occurs during cooling. However, the small-scale tests have not definitely proved that deformation also occurs during heating. This is due to problems with separating twist and bending deformation.
- Viscoelastic creep seems to be of minor importance in this context.
- As all samples used in the small-scale tests were wrapped in plastic film, the importance of mechano-sorptive creep could not be analysed.
It should be noted that these tests were carried out with a constant torsional moment, while the torsion angle varied during the process.
2.7
Torsional stiffness as a function of temperature
and MC
Changes in torsional force, as measured in the above described tests with drying of pre-twisted boards, are caused by two processes. Firstly, creep behaviour, which is the interesting part in this context, and secondly by a change in torsional stiffness due to changes in wood temperature and MC. Data on torsional stiffness is required if we are to be able to separate these two effects. There appears to be only limited information on this
in the literature, particularly on temperature dependence. It was therefore decided to perform a series of tests to determine these properties.
The measurements were performed in the same laboratory dryer using the steel frame for the boards used in the pre-twist tests described above. Four green 50 x 100 mm2 Norway spruce boards were wrapped in plastic film (to avoid changes in MC and to slow down temperature changes) and fastened in the steel frame. Weights were added to a lever system applying a known torsional moment was to each board. The resulting torsion was measured. Two different loads were used, around 114 Nm and around 173 Nm. This was first carried out at room temperature (25oC). The temperature in the dryer was then raised to 50oC and maintained at this level until the boards had reached this temperature
throughout. The dryer was then opened and the same measurement procedure was quickly carried out. This procedure was also repeated at 80oC. The boards were then dried to about 14% and then again wrapped in plastic film. The second part of the test was a full repetition of the first part, this time using dried boards. The boards were free to move and not subjected to external forces during the whole process, except for the short torsion angle measurement procedure.
It should be mentioned that the wood was very soft at 80oC, particularly the green boards. Creep was observed to start immediately the load was applied. The level of torsion increased within a few seconds and it was therefore not easy to measure the initial torsion angle accurately. At higher temperatures and for green wood, the results are therefore not very reliable. However, each board behaved similarly throughout the test.
Average results for the green boards are presented in Figure 23. The average results for dried boards are presented in Figure 24. Identical scales are used in Figures 23 and 24 so showing the difference in stiffness between wet and dry wood. Figure 25 illustrates the relative dependence on temperature. As previously mentioned, green wood torsional stiffness is considerably reduced at higher temperatures.
0 20 40 60 80 100 120 140 160 180 200 0 5 10 15 20 Torsion, degrees/m Torsional moment, Nm 25 C 50 C 80 C
Figure 23. Relationship between torsional moment and twist for green 50 x 100 mm2 spruce.
0 20 40 60 80 100 120 140 160 180 200 0 5 10 15 20 Torsion, degrees/m Torsional moment, Nm 25 C 50 C 80 C
Figure 24. Relationship between torsional moment and twist for dried 50 x 100 mm2 spruce. 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 25 35 45 55 65 75 85 Temperature, C Relative stiffness Green Dried
Figure 25. Influence of temperature on relative stiffness.
More detailed analysis of the results shows that neither green nor dried boards exhibit a significant dependence on wood density. This is surprising, but is maybe due to the limited number of boards used. Dried boards do however exhibit a significant dependence on wood MC, which is as expected. This also indicates that green boards could be
incorporated into a general correlation if these boards are assigned an MC equivalent to the FSP (free water is assumed not to influence stiffness). The final results are presented in Figure 26.
0,03 0,04 0,05 0,06 0,07 0,08 0,09 0,1 0,11 25 30 35 40 45 50 55 60 65 70 75 80 Temperature, C Torsion/Moment, degrees/Nm2 Green 18% 16% 14% 12%
Figure 26. Ratio between torsional moment and the resulting torsion as a function of temperature and MC for 50 x 100 mm2 Norway spruce.
Figure 26 shows the ratio between torsion (o/m) and the torsional moment (Nm) for 50 x 100 Norway spruce below the proportionality limit, as a function of temperature and MC. This result can now be used to separate creep behaviour from the pre-twist test results.
2.8
Analysis of the twist deformation mechanisms
Two different simultaneous processes cause the behaviour recorded in the tests. Firstly, a change in wood stiffness due to changes in wood temperature and MC. Secondly, a (more or less permanent) twist deformation (creep), which is the subject of this investigation. In the small-scale tests it was possible to separate these two by temporarily removing the external forces acting on the sample. In the main tests and with full size boards, this procedure was not possible. However, by predicting the change in stiffness according to Figure 26, it should be possible to separate the two in the main test series. This method for separating twist deformation from the total response has been applied to some of these tests as described below.
The first test, shown in Figure 6, is analysed in the following way. The wood temperature was not measured in this test. MC as a function of time is also not known. Temperature and MC changes were therefore estimated by using the drying simulation model
TORKSIM. Torsional stiffness could then be estimated from the correlation presented in Figure 26. If it is assumed that board response is pure elastic, then the expected relative change in the load cell output can be calculated as a function of time. Deviation of this curve from the measured curve indicates that twist deformation has taken place. For simplicity, only the average load curve has been analysed and not the curves for each board. The result is presented in Figure 27. The measured curve in Figure 27 corresponds to the average of the curves in Figure 6.
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 0 10 20 30 40 50 60 70 80 Time, h Torsional moment Measured Calculated
Figure 27. Comparison of measured and calculated loads in the first test.
The calculated curve follows the measured curve during the heating period. However, the curves clearly separate as the boards start to dry. This indicates that no twist deformation occurred during the heating phase, that mechano-sorptive deformation occurred during the drying phase and that further deformation occurred during the cooling phase. It should be mentioned that MC related stiffness change was calculated based on the average board MC. Surface MC in reality drops below FSP earlier than average MC. If this had been taken into account, the curves would had diverged earlier and perhaps almost
immediately after the heating phase ended.
The second test (Figure 7) used two green and two dried boards, and all were wrapped in plastic to avoid changes in MC. Board temperatures were measured and stiffness change could therefore be directly calculated. The average results for the two wet green boards are shown in Figure 28.
0 10 20 30 40 50 60 70 80 0,00 3,00 6,00 9,00 12,00 15,00 18,00 Time, h
Temperature and moment
Wet board temp. Measured Calcuilated
Figure 28. Comparison of measured and calculated load development for wet boards in the second test.
The figure shows that the measured curve follows the calculated curve reasonably well until the cooling phase starts. Twist deformation therefore only occurs during the cooling phase, and this is in line with the previous case as there is now no change in MC.
The same was now calculated for the two dried boards and the results are shown in Figure 29. 0 10 20 30 40 50 60 70 80 90 100 0,00 3,00 6,00 9,00 12,00 15,00 18,00 Time, h
Temperature and moment
Dry board temp. Measured Calculated
Figure 29. Comparison of measured and calculated load development for dried boards in the second test.
It seems that twist deformation also occurs for dried boards during the heating phase, particularly during the first 6 hours. Additional deformation also occurs during the cooling phase.
The third test (Figure 8) was analysed in the same way as the first test, i.e. the wood temperature and MC were estimated using TORKSIM software. The results are presented in Figure 30. 0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Time, h
Temperature and moment
Wood temp. Calculated Measured
Figure 30. Comparison of measured and calculated loads for drying boards in the third test.
The result is quite similar to the behaviour recorded in Figure 27. The comment relating to the starting point for curve divergence also applies here.
Next the fifth test (Figure 10) was analysed. All boards in this test were wrapped in plastic film. Two boards were green and two had previously been dried. The results for the two green boards are given in Figure 31.
0 20 40 60 80 100 0 1 2 3 4 5 6 7 8 9 10 11 Time, h
Temperature and moment
Dry bulb Measured Calculated
Figure 31. Comparison of measured and calculated loads for green boards during heat treatment.
Air temperature rises above 100C in this test. However, it is assumed that maximum wood temperature is 100oC for these wet boards. There appears to be a shift in time registration between the temperature and load curves (about 0,5 hours). This may be due to the difference between the load curves during the first two hours (a “negative” twist deformation is highly improbable). The results are otherwise similar to Figure 28, i.e. twist deformation occurs only during the cooling phase.
The results for the two dried boards are given in Figure 32.
0 20 40 60 80 100 120 140 160 180 200 220 0 1 2 3 4 5 6 7 8 9 10 11 Time, h
Temperature and moment
Dry bulb Measured Calculated
Figure 32. Comparison of measured and calculated loads for dried boards in heat treatment.
The load curves follow each other during the heating phase but then diverge. This differs from the result shown in Figure 29 where the curves start to separate immediately, although the heat treatment processes are quite similar. Twist deformation again occurs during the cooling phase in Figure 32.
The test, which consisted of several stages (Figures 11-13) was then analysed. The most interesting stage is the two-cycle heat treatment process presented in Figure 12. The results for this stage are given in Figure 33.
0 20 40 60 80 100 120 140 160 0 5 10 15 20 Time, h
Temperature and moment
Wood temp. Measured Calculated
Figure 33. Comparison of measured and calculated loads for two-cycle heat treatment of green boards.
The behaviour during the first heat treatment cycle is similar to that seen in Figures 28 and 31, i.e. no twist deformation until the cooling phase starts. The behaviour recorded during the second cycle is however very surprising. The twist deformation that occurred during the cooling phase of the first cycle appears to almost fully recover during the heating phase of the second cycle. Changes in the measured load curve during the second cycle are therefore small compared to the expected curve. After the second cycle, the sample had almost the same state as after the first cycle.
This test was continued with a third heat treatment. The boards were released from the pre-twisted state, measured and put back into the steel frame with an increased pre-twist. This third cycle is presented in Figure 34
0 10 20 30 40 50 60 70 80 90 100 0 2 4 6 8 10 12 Time, h
Temperature and moment
Wood temp. Measured Calculated
Figure 34. Comparison of measured and calculated loads for additional heat treatment of green boards.
In Figure 34 we can again see a recovery of twist deformation during the heating phase, i.e. measured load decreases less than the change expected from the stiffness decrease due to the rising temperatures. This can be described as being a memory effect. During the cooling phase, this recovery is reverted and more twist deformation is created. The total result is therefore an increase in deformation. We can therefore say there is no memory effect during the cooling process.
After the previous test, the boards were released from the steel frame, the plastic film removed and the boards measured. The boards were then put back into the frame and dried, but now with a pre-twist in the opposite direction. The aim of this is to restore the excessive twist deformation created in the earlier stages. The analysis of this test was carried out as previously, with board temperature and MC predicted by TORKSIM software. The results are presented in Figure 35.
-20 0 20 40 60 80 100 120 140 160 180 200 220 240 0 10 20 30 40 50 60 70 Time, h
Temperature and moment
Dry bulb Measured Calculated
Figure 35. Comparison of measured and calculated loads for the final stage. In Figure 35 (which relates to Figure 14) the measured and calculated load curves follow each other closely during the heating period, i.e. no additional twist deformation is created. No memory effect is therefore seen in this part of the process. However, when drying starts, the curves diverge very rapidly which is probably due to a memory effect. It should be noted that all pre-twist is realised (unlike previous tests) and the load drops to zero and even reverses. This therefore indicates a memory effect also during the drying stage.
Due to technical problems (as mentioned at the end of chapter 4) the final MC was much lower than the target in this test. The final part of the calculated load curve is therefore based on MC values that are outside the area of validity for the stiffness correlation given in Figure 26. This load curve is therefore not reliable after 50-55 hours of drying time and is the reason why decreasing stiffness is wrongly predicted during the cooling phase.
2.9
Conclusions from the laboratory tests
It has been experimentally proved that pre-twisting processes can be used to produce boards with improved straightness. Deformation towards straightness appears to be fairly stable in a constant climate, except for a slight reversal immediately after release.
Correlations for the level of pre-twist required have been established for simple drying and heat treatment cycles.
Before this investigation started, it was assumed that the changes in shape were due to the mechano-sorptive creep behaviour of wood. It was also assumed that deformation must be linked to a simultaneous drying process. This assumption was not correct. Tests revealed that heat treatment (change of temperature) without drying produced a similar result. Deformation in the required direction can therefore also be achieved through the heat treatment of either green or dried boards. It is not limited to a drying process. This is particularly important for boards that show remaining excess twist after the drying process, as these can be “corrected” by a final heat treatment procedure.
These new findings required some changes to the experimental approach. It was of course important to determine the importance of a moisture change (mechano-sorptive creep) in relation to a temperature change (“mechano-thermal” creep) on deformation. Separation of these mechanisms is not easy. One reason for this is that board torsional stiffness is strongly dependent on temperature and MC. This dependence was therefore determined in a separate experimental series.
The results indicate that there is no significant deformation of pre-twisted boards during the heating up phase (constant MC). During the drying phase (constant temperature) mechano-sorptive creep is important. During the cooling phase (constant MC) creep induced by temperature change is important. Where there are simultaneous changes in MC and temperature, a combined effect is achieved in the pre-twist procedure. It was also observed that wood exhibits a certain degree memory effect. The wood material appears to remember processes that acted on the board in earlier stages. This memory fades over time. More experimental work is needed to understand these
phenomena better. Japanese researchers (Iida et al 2002a, Iida et al 2002b, Ishimaru 2003, Takahashi et al 2004) have reported similar behaviours.
3
Industrial tests
3.1
Drying tests
3.1.1
First test
The first industrial test consisted of both LT drying (low temperature drying at around 60oC) and HT drying (high temperature drying at around 115oC). 54 boards were dried in a pre-twisted position in both LT and HT drying tests and were compared with 54 boards which had been dried without pre-twisting. The centre piece of 3 ex log, 41 x 147 mm2 Norway spruce was used and dried in the bottom package of the kiln stack. The wedges used for pre-twisting had an inclination of 3:50 and in opposite directions at both ends of the stack (horizontal in the centre). This corresponds to a pre-twist of 1,19 o/m. The results are summarized in Table 1 and Figure 36.
Table 1. Results from industrial drying tests with normal and pre-twisted boards at both low and high temperature. Values represent average ± standard deviation.
Test Grain angle
degrees Final MC % Twist mm/100mm/2m Twist o /m Accepted boards LT reference 2.13±1.17 13.9±0.9 4.29±2.93 1.23±0.84 30/54 = 56% LT pre-twist 1.81±1.38 13.5±0.9 0.56±2.64 0.16±0.76 45/54 = 83% HT reference 1.98±1.18 14.2±1.2 3.08±2.43 0.88±0.70 36/54 = 67% HT pre-twist 1.49±1.38 14.0±1.2 -0.45±2.08 -0.13±0.60 52/54 = 96% -1 -0,5 0 0,5 1 1,5 2 2,5 -2 -1 0 1 2 3 4 5 6
Grain angle, degrees
T w is t, d e g rees/ m LT reference LT pre-twist HT reference HT pre-twist
Figure 36. Relationship between grain angle and twist after drying in industrial tests. It is clear that the pre-twist method has decreased average twist to a level quite close to zero in both the LT and HT tests. Unfortunately, as shown in Table 1, the average grain angle was not the same in all four tests. Direct comparison is therefore not fully justified. Figure 36 therefore is a more correct reflection of the results. However, by using the dependence seen in Figure 36, the results can be adjusted to an average grain angle of 1.85o (Table 2). There is no statistically significant dependence on final MC in this material.
Table 2. Results from Table 1 adjusted to a common grain angle.
Test Grain angle
degrees Twist mm/100mm/2m Twist degrees/m Accepted fraction, % LT reference 1.85 3.91 1.12 51 LT pre-twist 1.85 0.59 0.17 86 HT reference 1.85 2.93 0.84 67 HT pre-twist 1.85 -0.21 -0.06 94
Table 2 indicates that high drying temperatures give lower final twist without pre-twist. The same effect is seen in tests with pre-twist, as the wedges used had an inclination that was slightly too low in the LT test, but slightly too high in the HT test. The last column of Table 1 gives the number of boards with a twist less than ±4mm/100mm/2m. In Table 2, this has been calculated from the mean value and standard deviation assuming a normal (Gauss) distribution. This may in some cases be a more reliable method. The proportion of approved boards has increased considerably as a result of using the pre-twist method.
This is due to two changes. The average twist is closer to zero and the standard deviation has slightly decreased. This seems to generally be true.
3.1.2
Second test
A second industrial test was carried out at a different sawmill using Norway spruce 38 x 125 mm2 with a 2 ex log sawing pattern. A conventional drying schedule was used with the wet bulb temperature at 70oC and the dry bulb temperature increasing from 78 to 87oC. The average final MC was 13,2%. The aim of this test was to find out how far up the kiln stack the twisted basement acted. In addition the pre-twist method as such was of course investigated. The stack configuration used in the kiln is presented in Figure 37.
D C4 X4 A4 o o B4 r X3 C3 B3 A3 X2 C2 B2 A2 X1 C1 B1 A1
Figure 37. Stack configuration in the test. The locations of the wedges are shown. An automatic deformation measuring device (Finscan) is installed at the sawmill. The device can also measure twist. The original idea was to measure the twist after drying, for all boards in stacks A, B, C and X. It was however subsequently discovered that the direction of twist (+ or -) was not recorded by this particular device. Where average twist for the package was close to zero, recorded values had little value. It was therefore decided that a limited number of boards should be manually measured (including grain angle) to enable some results to be obtained from this test. The two bottom layers (11 +11 boards) from each package of kiln stacks A, B and C where selected for measurement. Practical problems meant that the “C4” package was taken from a different stack than the original C-stack and the B4 package was smaller than normal.
If we first consider the C-stack. These were not pre-twisted and therefore represent a normal drying procedure. It was discovered that twist after drying is significantly depending on grain angle but not dependent on the position in the stack (height from the basement). The initial weight (around 460 kg) of the 21 board layers above the samples taken from package C4 therefore seems to be sufficient to keep all sample boards straight during drying. The results are shown in Figure 38. The average grain angle for all 88 boards was 0,33 degrees and average twist after drying was 0,62 degrees/m. The grain angle average for this kiln stack was remarkably low and explains why no dependence on position in the stack was found.
-3 -2 -1 0 1 2 3 4 5 6 -4 -3 -2 -1 0 1 2 3 4
Grain angle, degrees
Twis t, de gre e s /m
Figure 38. Relationship between grain angle and twist after drying, for boards taken from a kiln stack without pre-twist.
We next consider the A-stack, which was pre-twisted with wedges inclined 1:5,79 at both ends of the stack. This is equivalent to a pre-twist angle of 3,50o/m. The amount of pre-twist selected was high to clearly test how far up the stack the effect reaches. When pre-twist after drying was analysed, it was found that twist is significantly dependent on both grain angle and board position in the stack. If the position in the stack is measured in terms of the initial weight of the boards above that point, then it theoretically appears correct to assume that this influence is exponentially dependent on this variable. Using this variable, the relationship can be illustrated by Figure 39. The average grain angle for all samples from the A-stack (73 boards) was 1,66 degrees and average twist after drying was -0,37o/m. The grain angle for the A-stack was therefore different from the grain angle of the C-stack. -5 -4 -3 -2 -1 0 1 2 3 4 -4 -2 0 2 4 6 8
Grain angle, degrees
Tw ist a ft e r dry ing, deg rees/m Samples 500 kg/m2 1000 kg/m2 1500 kg/m2 2000 kg/m2
Figure 39. Twist after drying as a function of grain angle and board location in the pre-twisted A-stack (expressed as initial top load in that location).
Figure 39 clearly shows that the effect of the inclined basement is decreasing upwards. If the line for 500 kg/m2 (which approximately corresponds to the weight of the top package A4) is compared with the line in Figure 38, then these two lines appear to be almost identical. This indicates that the boards in package A4 remained straight (without pre-twist) during drying, i.e. the pre-twist applied by the bottom wedges did not reach
package A4. The packages A3-A1 show a gradually increasing influence of the wedges. However, only the bottom package A1 is strongly influenced. This means that boards prone to twist, selected for example based on a grain angle measurement, should be placed in the bottom part of the kiln stack. As previously mentioned, the wedges used in this test had inclinations which were too steep. If a lower inclination had been used, then the 500 kg/m2 line in Figure 39 would probably have remained unchanged, while the remaining lines would probably have moved closer together. This would have given a better overall result.
The B-stack is finally considered. In stack B, the inclined basement is countered by wedges in the opposite direction between packages B2 and B3. In principle, only
packages B1 and B2 are pre-twisted and packages B3 and B4 are not. Wedge inclinations are the same as for the A-stack. When the post drying twist was analysed, it was
discovered that twist in the upper part (B3, B4) was significantly different to the twist in the lower part (B1, B2). A significant dependence on grain angle was found as expected within these two parts. However, no dependence was found on the position in the stack. This result is illustrated by Figure 40. The average grain angle of the samples from packages B3, B4 was 1,18 degrees and the average for packages B1, B2 was 0,91 degrees. The average twist after drying was 1,63 and -2,01 degrees/m respectively. The average for the entire B-stack was 1,04 degrees and -0,21 degrees/m.
-4 -3 -2 -1 0 1 2 3 4 5 -3 -2 -1 0 1 2 3 4 5
Grain angle, degrees
Twis t a ft e r dry in g , de gre e s /m Package 3-4 Package 1-2 Package 3-4 Package 1-2
Figure 40. Twist after drying as a function of grain angle for the two parts of stack B. The upper line in Figure 40 is slightly steeper than the line in Figure 38 or the uppermost line in Figure 39. However, the position in the diagram is approximately the same. This indicates that the drying behaviour in packages B3 and B4 best corresponds to drying without pre-twist. There were fewer boards in the B-stack (Figure 37) and when this is taken into consideration, the lower line in Figure 40 agrees well with the lower lines in Figure 39. This shows (as expected) that the process in the bottom packages corresponds to drying with pre-twist, obviously with a more uniform pre-twist angle in the vertical direction than in the A-stack. As the results show, the wedges used had an inclination far greater than the optimal. If more optimal inclinations had been used, the result would probably have been more correct. However, the results from the stacks A and B show that boards prone to twist should be placed in the bottom part of the stack and that there are two different ways to apply the pre-twist. One way is by only an inclined basement, where effective pre-twist gradually decreases upwards in the stack. A second way is where pre-twist is interrupted by additional counteracting wedges, which gives a more
