Tree

Physiology

9, 383-399

0 1991 Heron Publishing-Victoria,

Canada

Dynamics of swaying of Picea sitchensis

R. MILNE
Institute of Terrestrial Ecology, Bush Estate, Penicuik, Midlothian EH26 OQB, Scotland Received October 1. 1990
Summary

Six 26.year-old Sitka spruce (Picea sitchensis (Bong.) Cam) trees growing in a Scottish plantation were swayed manually to determine their mechanical dynamics. The natural frequency of sway of the intact trees (mean height 14.2 m and mean stem diameter 14.5 cm at 1.3 m) was on average 0.35 Hz. The variation of this frequency with tree size was found to be well described by engineering mechanics theory. In particular, shape parameters could be defined for both intact and branchless trees, which, along with stem size, density and elasticity, could predict the natural sway frequency using a simple formula. The damping of sway was found to consist of three components, (1) interference of branches with those of neighbors, (2) aerodynamic drag on foliage, and (3) damping in the stem. For the sample of six trees, which spanned the diameter range at the experimental site, the importance of these three components to overall damping was in the ratio 5/4/l for the median sized tree. Interference between neighbors depended on the distance to neighbors, as well as on the sizes of the chosen tree and its neighbors. Aerodynamic damping was larger for larger trees and the energy lost to this force was similar in magnitude to that calculated to be lost using drag coefficients from published wind tunnel and other studies. The amount of damping from the stem was linearly related to stem diameter.

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

Introduction In British commercial conifer plantations, damage caused by normal winter gales is a major cost factor and limit to growth (Miller 1986). The U.K. Forestry Commission have developed a Windthrow Hazard Classification (Booth 1977) which estimates the risk of wind damage for a planting site. This classification system has proved useful to the forest manager and has recently been improved (Miller 1985), but the link between weather and soil conditions and the resulting risk of windthrow is largely empirical. A description of tree movement and stresses based on physical principles would be a useful further improvement. The basic statics of a tree bending in the wind have been investigated by Petty and Swain (1985), Blackbum et al. (1988), and Milne and Blackburn (1989) among others. Elastic instability, buckling under the trees own weight, or snow loading, and the influence of mechanics on tree growth have been described by McMahon and Kronauer (11976), King and Loucks (1978) and King (1986). In windy climates, the horizontal forces from the wind are those that are most likely to cause damage and the commonest form of damage is overturning where the root plate is torn from the ground. The wind is gusting in nature and several authors have emphasized the importance of dynamic effects in determining movement and stress in conifer trees (Mayhead 1973, Holbo et al. 1980, Papesch 1984, Milne 1986, Mayer 1987, Blackbum et al. 1988). These dynamic effects are likely to increase the bending of stems,

384

MILNE

and hence the load on root systems, when the pattern of wind gusts becomes coupled to the natural swaying frequency of the trees. The swaying of trees during wind gusts is random within limits imposed by wind turbulence and the natural elastic properties of the stem and canopy. To quantify the dynamic properties of the tree and the stresses imposed on the roots from windstorm data, it would therefore be necessary to investigate the randomness of the wind as well as the response of the tree. An alternative to studying tree movement in the wind is to sway the tree manually and to record the natural sway period and the rate of decay after cessation of swaying (Mayhead et al. 1975, Gardiner 1989). These two parameters can be used to describe the dynamics of motion for any known applied force using standard mechanical engineering methods (Thomson 1988). These dynamic parameters of the tree, treated as a distributed mechanical-elastic system, might also be calculated from the shape, mass, and elastic properties of the various components of the tree, i.e., stem, branches, and needles. Obviously the detailed motion of individual branches would be difficult to describe in this way. If the swaying of the stem, however, could be described adequately from limited stem and crown dimensions, then methods developed in building design (e.g., Simiu and Scanlan 1978) might be used to estimate the risk of damaging stresses occurring in trees grown under different conditions (e.g., spacing). Here a program of manual swaying of 23-year-old Sitka spruce (Picea sitchensis (Bong.) Carr) is described. The site was an even-aged, unthinned, commercial plantation (spacing density 3800 trees per hectare in the vicinity of the experimental area) at Moffat Forest, southwestern Scotland (5517 N, elevation 360 m). The natural sway frequency of the trees and the factors affecting the damping of motion were estimated from the recorded movement of each tree. The measured sway frequencies were compared with estimates of natural frequency from tree form and mass distribution using standard mechanical theory. The parameters quantifying the damping of motion were investigated in terms of tree biomass distribution and published drag coefficients. Methods In accordance with the usual practise in U.K. commercial forestry, the trees investigated had been established on the upturned soil from a deep tine plough which produces parallel ditches and mounds some 0.5 m in depth. This leads to significant asymmetry in the root structure of the tree, but it is not known whether there are any significant effects on the stem or foliage. Trees were chosen so that the range of diameters at breast height (i.e., 1.3 m from the ground) was similar to the range of diameters in the neighboring O.l-ha block and the stems and canopies were not obviously asymmetrical. Each tree chosen was swayed manually and the movement of the stem at 6 m from the ground along and across the ploughing furrow was recorded by means of displacement sensors. Sensors were attached to posts in the ground at distances of 4 and 6 m from the stem in two directions at right angles. One direction was always across the ploughingfurrow, whereas the other was along the

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

SWAYING OF PICEA

SlTCHENSIS

385

furrow, in whichever was the most convenient direction. Thin nylon line connected the sensors to the tree stem. The layout of the sensors is illustrated in Figure 1. The sensor movements were recorded at 10 readings per second on a data logger. To estimate the reduction in height of the tree stem at 6 m as well as the horizontal displacement, two sensors were used in each direction. Each sway test was carried out by attaching a rope to the tree 9 m from the ground and pulling several (usually three) times to build up an oscillation and then releasing the rope so that the tree swayed naturally to rest. Each tree was swayed, in calm weather, (1) as it normally stood in the forest, i.e., with branch interference with neighbors, (2) with branch interference removed by pruning or tying back of neighbors, and (3) with all of its branches removed, i.e., the stem only. Each sway test, which consisted of between 10 and 30 natural, unforced, sway cycles of the stem after the initial displacement, was repeated four times across the furrow and four times along the furrow to include information on any asymmetry of dynamic response. Visual observation suggested that the tree moved in the plane of the initial pulling force and bent from the ground without any higher modes (i.e., no S-shaped bending at a greater frequency) being excited. Near the end of the swaying, the pattern of movement became more variable because of interference with neighbors or asymmetry in tree form. The motion of a damped elastic system, with no imposed forces, under these simplifying assumptions

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

14 13 12 11 10 9y Height (m) 8 7 6 Method of tree sway measuring

5 4 3 2/
1

Figure 1. Position of displacement sensors and lines attached to tree stem used in measurements of dynamic sway of Sitka spruce trees. The dashed line shows position of rope used to start tree swaying.

386

MILNE

is given by

M~+C~+KX=D, dt2 dt
where X is the displacement (in Mode-Generalized Coordinates) of the Mode-Generalized Mass, M, K is a Mode-Generalized elastic constant relating displacements to any imposed forces, and C is a viscous damping coefficient relating damping force to the velocity, again in Mode-Generalized form. The Generalized Coordinates take account of the distributed nature of the tree (see, e.g., Thomson 1988 for details). It is assumed that the tree can be described as an assemblage of vertical l-m sections and oscillates in only one mode. The mass of the stem and foliage is assumed to be located at the center line of the stem, which is acceptable because the crowns were not obviously asymmetrical. All branch and needle weights act on the tree only through the forces applied to the stem. The shape of the oscillatory mode, i.e., the horizontal displacement of the individual sections, can be described by the static bending curve. Mode-Generalization seeks to provide a description of the system with one displacement variable, where the kinetic, elastic, and damping energies of the Generalized system are equal to those in the distributed system. The displacements (xi) of the individual sections of the distributed system are related to the Generalized Displacement (X) by xi = l.tiX, where pi is the mode shape and is taken as the displacement of an individual section relative to the displacement of the section with the largest movement, here the tip of the stem. This leads to the Generalized Mass being given by M = Cyfmi where mi is the mass of an individual section. Each of the sections in such a system oscillate (i.e., sway) near the fundamental natural frequency fn = =/23r, but the oscillations decay at a rate determined by the damping coefficient C. The value of this damping also affects the frequency of sway to a small extent. If C has the value C, = 47cMfn, the damping is termed critical and the oscillations will die out in one cycle. If C < Cc, then the oscillations will decay over several cycles and 5 = C/C, is termed the damping ratio. The parameters fn and 5 were determined for each sway from the recordings by measuring respectively the time between peaks (T) and the reduction in amplitude of consecutive peaks above the mean displacement. For small values of L,fn = l/T and 5 follows a logarithmic decay described by

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

where a, and a,,~ are the amplitudes of consecutive positive peaks (Thomson 1988). For the branch contact tests, because the sway damped quickly, only the first two or three cycles after removing the rope were used for this calculation. For the no branch contact tests four to five cycles were used and for the no branch tests six to ten cycles were used. Within these selected sections of each test, the average logarithmic decrement and hence the damping ratio (5, Equation 2) was calculated.

SWAYING

OF PICEA

SITCHENSIS

387

Individual logarithmic decrements between successive cycles were within + 10% of this average. Between the sway tests, with and without branch interference, each tree was bent using a winch and cable attached 9 m up the stem, and held at a maximum of 1 m horizontal displacement at 9 m. While the tree was in this position the bent shape of the stem was recorded by means of measuring lines between points at 2, 3, 4,5,6,7, 8 and 9 m along the stem and the 4 and 6 m reference points out from the stem, used for the across-furrow sway tests (Figure 2). The tree was held in this position for about 30 min to allow these measurements to be made. The force applied and the horizontal displacements were then used to estimate the elasticity of the stem (Youngs modulus) using a numerical model based on a mechanical description of the tree as l-m vertical sections of stem and foliage (branches plus needles) of measured mass (Milne and Blackbum 1989). The elasticity of the stem was estimated by an iterative method where the assumed elasticity of the stem in the model was altered until the displacement calculated by the model matched (to within 1 cm) the displacement measured at the attachment point of the winch cable. For this elasticity, it was found that the stem displacements, both above and below the cable, as estimated by the model, were within 20 cm of those measured. Given the difficulties of measuring these distances in the forest this was considered to be acceptable. The mode shape for each tree was estimated from this static bending model using point loading and distributed mass, as in the determination of elasticity. This will introduce only small errors to the theoretical calculations of natural frequency relevant to wind loading because most needle area was above the point of application of the force from the rope. It is also known that the exact form of the mode shape is not, from

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

14r 13 12

2 IMethod Youngs of measuring

IO 9
8 Height (m)

modulus

6
5 4 t

6m

4m

Figure 2. Layout of lines used to measure position of bent Sitka spruce stem was bent using cable and winch pulling across the furrow.

stem to determine

eksticity.

The

388

MILNE

theoretical considerations, a strong influence on the energy in a dynamic system (see e.g., Thomson 1988). After the sway test with no interference between neighbors, all the branches were removed from the tree and the swaying of the stem alone was recorded. The branches cut from each tree were weighed to give the vertical distribution of foliage fresh mass in l-m vertical sections. Subsamples of branches and needles were taken and estimates of the ratio of needle mass to projected area were determined using a television camera and image analysis computer system. These mass to area ratios were then used to estimate the needle area in l-m vertical sections of each tree. After swaying was completed, each tree stem was felled and cut into l-m sections. The fresh mass and the diameters at each end, and in the middle, of each stem section were measured. Results Stem elasticity General information on the sample trees is shown in Table 1 and information on elasticity and fresh density is given in Table 5 (see below). Dry density of Sitka spruce was not measured for the sample trees, but is in the range of 40-45% of fresh density. Youngs modulus of elasticity varied from 6.5 to 9.0 GPa, but its value was not correlated with tree size for fresh density. Height, stem diameter, foliage mass, and needle area were inter-correlated, as expected. Tree dynamics The natural swaying characteristics of the Sitka spruce trees are illustrated in Figure 3 where sample recordings of sway, across the furrow, are presented for Tree 11 under conditions (1) with neighboring branch contact, (2) with no branch contact, and (3) with branches removed. These sample recordings are typical of those obtained, and for a given tree and damping situation, were highly reproducible. Differences in natural frequency between whole and branchless trees due to loss of branch mass were readily seen, as were differences in damping due to the contrasting
Table I. Characteristics Diameter breast height (cm) I 2 12 10 11 5 10.6 11.5 14.0 15.5 16.9 18.3 13.6 13.0 13.8 14.3 14.8 15.5 61.5 78.7 88.1 158.7 160.9 190.8 of six 26-year-old Height (m) Stem mass (kg) Sitka spruce Total foliage mass (kg) 15.5 28.4 19.6 55.8 46.2 64.6 trees at Moffat Live foliage mass 0%) 7.2 19.0 11.9 38.9 38.2 42.6 7.7 17.4 15.7 38.4 41.2 47.8 Forest, southwestern Scotland.

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

Tree no.

Needle area b*)

Mean neighbor distance (ml 1.6 2.1 1.6 1.8 1.6 2.0

SWAYING

OF PICEA

SITCHENSIS

389

-2

0.2

Branch contact

-0.2

1 0

I 10 Time

I 20 (seconds)

I 30

I 40

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

0.3 z z iz : % .Lo cl 0.2 0.1 0.0 -0.1 -0.2 -0.3 L 0 / 10 Time I 20 (seconds) I 30 I 40

0.4 0.3 z z E 0 cl .-E: D 0.2 0.1 0.0 -0.1 -0.2 -0.3 -0.4 t 1 0 / 10 Time 20 (seconds) I 30 -I I 40

Figure 3. Example of relative stem movements, measured at 10 readings per second, of a Sitka spruce tree (Tree 11) swaying to rest in Moffat Forest, southwestern Scotland after manual displacement. (a) With no removal of neighbors, (b) with interference of neighbors removed, and (c) all branches removed from sample tree.

effects of branches under the different test conditions. The natural frequencies of sway and the damping ratios for each of the six trees are shown in Tables 2 and 3, respectively. Each entry in Tables 2 and 3 is an average of four sway tests and individual test values were within 5% of the mean for the four tests conducted on that tree for a given branch damping. Sways usually had characteristics agreeing to within

390

MILNE

Table 2. Measured natural sway frequencies (Hertz) of six intact Sitka spruce trees at Moffat Forest, southwestern Scotland for different damping conditions and sway direction. Each value is a mean of four sway tests. Individual sway test results were averages over several cycles (see text) and were within 5% of these means. Tree no. Across furrow Branch contact 7 2 12 10 11 5 0.28 0.30 0.40 0.42 0.40 0.39 No contact 0.25 0.29 0.39 0.31 0.40 0.37 No branches 0.40 0.48 0.63 0.69 0.63 0.63 Along furrow Branch contact 0.21 0.33 0.40 0.37 0.40 0.37 No contact 0.24 0.30 0.40 0.40 0.40 0.37 No branches 0.4-O 0.50 0.63 0.67 0.67 0.63

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

Table 3. Measured damping ratios of six Sitka spruce trees at Moffat Forest, southwestern Scotland for different damping conditions and sway direction. Each value is a mean of four sway tests. Individual sway test values were calculated from the average logarithmic decrement of amplitude over several cycles (see text) and were within 5% of these means. Tree no. Across furrow Branch contact I 2 12 10 11 5 0.134 0.115 0.215 0.109 0.118 0.103 No contact 0.066 0.066 0.060 0.062 0.064 0.056 No branches 0.026 0.034 0.021 0.015 0.024 0.06 Along furrow Branch contact 0.079 0.105 0.163 0.123 0.115 0.094 No contact 0.066 0.079 0.056 0.057 0.067 0.073 No branches 0.025 0.032 0.022 0.017 0.025 0.014

5% across and along the ploughing furrow. Larger differences were found in damping because of branch contact across and along the furrow (e.g., the damping ratio of Tree 7 was 0.134 across the furrow but 0.079 along the furrow), but these were caused by asymmetries in either the canopy of the swayed tree or its neighbors. There was variation from tree to tree in both natural frequency and damping ratio, but this variation was limited in extent. Discussion Stem elasticity The Youngs moduli were in the range 6.5 to 9.0 GPa (see Table 5) and were larger than those found for Sitka spruce by Milne and Blackbum (1989) (2.0-6.4 GPa) in the same forest. Cannel1 and Morgan (1987) measured Youngs moduli of 3.9-6..7 GPa in three Sitka spruce stems and concluded that the modulus was lower for living

SWAYING

OF PICEA

SITCHENSIS

391

trees than for green sawn timber. The results presented here agree more closely with those of Gardiner ( 1989) who measured the elasticity of mature Sitka spruce by both bending and swaying and found a range of values from 3.8 to 8.4 GPa. The low values found by Milne and Blackbum (1989) may have been due to rotation of the tree roots, because in that study bending was recorded at soil breakage in overtuming tests, which may not have occurred until after significant soil or root damage. In this study minimum bending of the stem was applied, to eliminate rotation of the tree root plate. Silviculture and soil conditions may also influence the elasticity of the stem, but there has been no systematic study of such effects. Further investigation of such effects on stem elasticity is needed, because this parameter is of considerable importance in determining the dynamics of bending. Such investigations would also be of relevance to the study of the structural strength of the stem, which will determine stem snap both from the wind and from snow loads. Tree dynamics Mean values of the natural frequency (f;l) and damping ratio (5) for each of the six Sitka spruce trees swayed were calculated by combining the sways along and across the furrow. These values are given in Table 4 for the three damping conditions of (1) branch contact, (2) no contact, and (3) no branches. The measured natural frequencies were compared with theoretical estimates using two methods: (1) application of the principle of conservation of energy using a method attributed to Lord Rayleigh, (2) an extension of this method described by Blevins (1979). Comparison of measured natural frequencies with Rayleigh estimates The natural frequency of a multi-element system, like a tree, can be estimated by the Rayleigh method. This method uses the principle of conservation of energy which predicts that, as the tree sways, energy is transferred from the potential form to the kinetic form but the total energy remains constant, if there is no damping or applied force. Thus the maximum kinetic energy, which occurs as the tree sways through the rest position, must equal the change in gravitational potential energy, which occurs between the

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

Table 4. Measured natural sway frequency (Hertz) and damping ratio for six Sitka spruce trees at Moffat Forest, southwestern Scotland for different damping conditions. The mean of the results from Tables 2 and 3 for across the furrow and along the furrow sway tests are presented here. Tree no. Frequency Branch contact 0.28 0.32 0.40 0.40 0.40 0.38 No contact 0.25 0.30 0.40 0.39 0.40 0.37 No branches 0.40 0.49 0.63 0.68 0.65 0.63 Damping Branch contact 0.107 0.110 0.189 0.116 0.117 0.099 ratio No contact 0.066 0.073 0.058 0.060 0.066 0.065 No branches 0.026 0.033 0.022 0.016 0.025 0.015

7 2 12 10 11 5

392

MILNE

rest condition and the limit of displacement. The potential energy is stored in elastic form in the deformed stem and hence, for a tree considered as an assemblage of l-m vertical sections, the kinetic (KE), potential (PE), and elastic (EE) energies can be calculated as Kl!&,=XOS &$mi = C 0.5 ni Bi , (3) (4)

Change in PE = EI&,

where the subscript i refers to a vertical section of the tree (in this case l-m long), Xi is its maximum horizontal displacement, mi is the mass in the section, Tli is the bending moment in the stem, and 8i is the angular displacement of the stem, both at the maximum bent position. Equating the energies allows estimation of co,,which is common to all sections, hence C 0.5TJi 8i =Z0.5XfT7Zi7 (5)

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

where on = 27rfn. This method was applied using the measured stem and foliage masses of each of the six Sitka spruce trees; the displacements and bending moments used were those calculated by the static bending model applied to the whole tree or the stem alone and using the previously estimated elasticity of the stem. The natural frequencies for the whole tree and for the stem alone calculated in this way showed good agreement with the measured frequencies from the sway tests (Figures 4 and 5). The natural frequencies from the branch-contact sways tended to be higher than those from the no-contact sways, but both were close to the frequency calculated by the Rayleigh method for that tree. In the case of the stems alone, the measured frequencies tended to be greater than those by the Rayleigh method. The measured sway frequency is only equal to the theoretical natural frequency of a system when the damping ratio of the system is small (Thomson 1988) and, although for the range of damping observed this should not be important, non-linear interaction of the variable may contribute to the differences between the presented predicted and observed values. Estimates of Blevins h from measured natural frequencies It has been shown that the natural frequencies of slender, tapered beams can be expressed by the general formula (Blevins 1979) f = A2 * 2xh*

[1
El0 PAO

(6)

wheref, is a natural frequency, lo (= n: r4/4) and A0 are the area moment of inertia and the area of cross section at the base of the beam, respectively, h is the height, E is the

SWAYING

OF PKEA
0.5

SlTCHENSIS

393

0.4
8 0

0.3 V 0.2
0 /

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

0.1

0.0 0.0 0.1 Rayleigh

0.2
frequency

0.3
(Hz)

0.4

0.5

Figure 4. Natural frequencies of sway of whole Sitka spruce trees. Measured by manual swaying in Moffat Forest, southwestern Scotland, (a) including contact with branches of neighbors, y-axis (O), (b) no contact with neighbors, y-axis (v), and (c) calculated by the Rayleigh method, x-axis. The line indicates where the frequencies would lie if the two methods agreed.

1 .o

0.8

0.6

0.0

0.2
Rayleigh

0.4
frequency

0.6
(Hz)

0.8

1 .o

Figure 5. Natural frequencies of sway of the branchless stems of Sitka spruce trees. Measuredby manual swaying in Moffat Forest, southwestern Scotland, y-axis, and calculated by the Rayleigh method, x-axis. The line indicates where the frequencies would lie if the two methods agreed.

394

MILNE

modulus of elasticity, and p is the material density. The dimensionless parameter h is a function of the shape, mass distribution, orientation (vertical or horizontal), nature of fixing (clamped, hinged, etc.), mode of bending, and taper of the beam. This parameter could be calculated from first principles for a beam with added masses, but such values are not normally tabulated in the literature. Although a tree with a canopy is a much more complicated structure than a simple, uniform, tapered beam, there is sufficient similarity to investigate whether a useful 3Lcould exist for trees. Because the value of 3Lfor a tree is difficult to calculate from a description of the shape, etc., it was calculated for each tree (Table 5) from Equation 6 for both the intact and branchless conditions, using the measured dimensions of the stem, measured sway frequency, and the elasticities estimated in the static bending tests. The density of the stem at its base was calculated from the dimensions and the weight of the bottom l-m stem section of each tree. The data in Table 5 show that individual values of astern,and more interestingly ht ree,are within 5% of the mean for the six trees, and could therefore be useful in investigating the variation of sway frequency at other forest sites. If the variation of h,,, with silviculture can be simply described, then natural frequency could be predicted using measurements of density and stem size on selected trees with a species mean (or with more work a site mean) for modulus of elasticity.
Components of damping Equation 1 assumes viscous damping (i.e., damping force proportional to velocity) for the decay of tree sway, which implies a constant logarithmic decrement of sway amplitude. In the analysis of damping in the sway tests, the average logarithmic decrement of sway amplitude over the selected block of cycles was thus taken as an estimate of this constant. The proportionality constant or damping coefficient (C) between velocity and force is C = {C, = <4rcMf, where < is the damping ratio given in Table 4, M is the Generalized Mass and f is the (measured) natural frequency. The total damping force 7(fb) on a single tree can be considered in terms ,of three components:

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

Fd=Cv=(Cb+Ca+Cs)v,

(7)

Table 5. Parameters for summary description of natural sway frequency of Sitka spruce trees at Moffat Forest. E (GPa) is Youngs modulus of elasticity, p (kg m-) is fresh density of wood at stem base and ht,, and h,,, are dimensionless parameters (See Equation 5) for the intact and branchless conditions. Tree no. E Wa) 6.5 1.5 8.0 9.0 6.5 1.6 P (kg m-) 960 1010 910 1050 850 910 2.03 2.00 2.13 2.09 2.17 2.06 2.08 2.60 2.55 2.70 2.80 2.72 2.68 2.67

I 2 12 10 11 5 Mean

SWAYING

OF PICEA

SITCHENSIS

395

where Cb refers to inter-branch damping between neighbors, C, refers to aerodynamic drag damping of branches and needles in the air, and C, refers to damping in the stem and roots. The velocity V = dX/dt is in terms of Generalized Coordinates, which account for the spatial distribution of the tree. In the sway test with branch contact, all three components would be found, whereas in the sway with no branch contact, only (C, + C,) would apply. Only C, operates in the swaying of the stem alone. Hence the difference between the damping coefficients found in the branchcontact and no-contact tests will give an estimate for Cb, whereas the difference between the no-contact and no-branch tests will give an estimate for C,. Thus for each tree
Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

ca = Snc43Mfnc - cs

where the subscripts t and s refer to the whole tree and the stem alone and the subscripts nb, nc and bc refer to the sway tests with no branches, with no branch contact, and with contact between branches of neighbors, respectively. The results of these calculations are given in Table 6 for each tree. In studies of the effect of wind on trees and canopies, drag coefficients are found to depend on leaf area, hence in Table 6 the values of C, and Cb relative to needle area are given for each tree. The inter-branch drag scaled in this way was particularly large for Tree 12, but there was also variation among the other five trees. Inter-branch drag will depend on the amount of foliage on neighbors, and the distance to the neighbors, both of which will be influenced by local spacing density. The amount of needles on a tree generally increases with stem diameter. The influence of neighbor distance on inter-branch drag will be significant where branches overlap but not otherwise. This non-linear effect might be simply described by assuming it is inversely proportional to the
Table 6. Mode-Generalized Moffat Forest, southwestern aerodynamic parameter, Tree no. 7 2 12 10 11 5 Mean damping coefficients (N s m-) for the sway of six Sitka spruce trees at Scotland. L is tree leaf area (m*), Cb is inter-branch damping, C, is a spacing competition

drag damping and C, is stem damping. S = dnllz (cm m-2) (see text), indicating branch interference with neighbors, is also shown.
cb

C, 1.21 2.46 2.27 6.06 7.09 6.97

CT 0.68 1.12 0.94 1.56 2.80 1.55

CblL

Cd

1.53 2.17 7.24 7.48 7.64 4.8 I

0.20 0.12 0.46 0.19 0.19 0.10 0.21

0.16 0.14 0.14 0.16 0.17 0.15 0.15

3.6 3.2 5.1 2.8 4.0 2.5

396

MILNE

square of inter-tree distance. To describe the overall effect of the neighbors, the diameters (d,,) of, and the distances (In) to, the nearest neighbors (usually 8) of each of the six sample trees were measured, and a neighbor interference parameter (S) equal to the average of d,/l$ was calculated (Table 6). This parameter showed that the neighbors of Tree 12 were significantly different from the neighbors of the other sample trees. The mean inter-branch drag, excluding Tree 12, was calculated to be 0.16, whereas the average value for the aerodynamic drag was 0.15, indicating that branch contact would be as important as aerodynamic effects in damping tree sway during storms. In simple terms, the branches in a closely planted forest interlock and limit tree movement thus reducing the risk of wind damage. The interference parameter, S, which measures inter-neighbor effects, may be of use in describing damping in forests with different planting densities. Landsberg and Thorn (197 1) and Landsberg and Jarvis (1973) have discussed wind drag on Sitka spruce shoots in a wind tunnel and on layers of canopy in a pole-stage forest. They found that, both in shoots and canopy layers, there were sheltering effects which reduced drag particularly at higher wind speeds. They produced formulae for the change in drag with wind speed for shoots and canopy layers which showed that drag force was proportional to needle area and to wind velocity raised to a power between 1.5 and 2. To compare these descriptions of drag with the aerodynamic damping force found in the sway studies, the energy lost in one cycle of tree sway was calculated from both the p&lished wind drag data and the model suggested by the swaying. From the wind drag data this energy was calculated by assuming that as the branched moved through the air they would experience a drag equivalent to that when the wind flowed over stationary branches. The drag formulation of Landsberg and Jarvis (1973) was applied to each l-m layer of tree canopy measured in the sway studies and knowing the frequency (i.e., that measured) with which all layers moved through a sinusoidal cycle of sway, the work done by the damping force could be calculated for a given sway amplitude. The energy lost to aerodynamic damping in the Mode-Generalized description of tree sway (which assumes viscous damping) is given by ~~c~C&X~,,,~~ where X,,, is maximum displacement of the stem tip during a sway. These two different estimates of energy lost to damping in one sway were calculated for a range of typical amplitudes of sway and proved to be of similar magnitude (for example see Figure 6 for Tree 5) but with the Mode-Generalized (viscous damping) formulation giving larger values for smaller sway amplitudes. The 0 to 2.5 m range of sway amplitudes used here will have equivalent maximum branch velocities of 0 to 6 m s-l, which indicates that, in the low, wind-speed range, the sway studies predict more drag than the wind drag studies. Theoretical calculations by Landsberg and Thorn (1971) and the measurements of Landsberg and Jarvis (1973) both suggest a rise in drag at low velocities (< 1.5 m s-), but their limited experimental studies could not provide an accurate quantitative assessment of its extent. Damping of sway motion due to the structure of the stem, although the least important of the three components measured, was significant, particularly for small trees (Table 6 and Figure 7). The stem damping component, C,, was correlated (R =

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

SWAYING

OF PICEA

SITCHENSIS

397

5oor------

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

3 Sway amplitude (m)

Figure 6. Energy lost in aerodynamic damping per cycle of sway in Sitka spruce foliage (Tree 5 in sway studies at Moffat Forest), as a function of sway amplitude. The two methods used involve: (a) drag coefficients for canopy layers published by Landsberg and Jarvis (1973) (0) with measured sway frequency and mass distribution; and (b) Mode-Generalized dynamic model (with viscous damping) from sway studies (0).

20

16

v) t F .E ::

12

0.10

0.12 Diameter

0.14 at 1.3

0.16 m (m)

0.18

0.20

Figure 7. Components of damping in Sitka spruce expressed as a function of stem diameter at 1.3 m. The components of the total damping coefficient due to (1) branch interference. cross-hatched: (2) aerodynamic drag, open: and (3) stem damping, solid, are shown.

398

MILNE

0.71) with stem diameter. Hoag et al. (1971) investigated the internal damping properties of the branches of almond trees and found a damping ratio of 0.01 to 0.016 compared to 0.015 to 0.033 found here for Sitka spruce (Table 4). They did not study the effect of branch diameter on damping but found that, below a water content of 30%, the damping decreased significantly. Conclusions The dynamics of movement of plantation-grown Sitka spruce can be described using simple engineering mechanics, given a description of the shape and mass distribution of the tree. The natural frequency of movement can be predicted using Rayleighs energy method, or simple parameterizations for a given tree form. Damping of stem sway has three components, interference between the branches of neighbors, aerodynamic damping of foliage, and damping in the stem and roots. In a median tree for the sample studied (stem diameter 15.5 cm, height 14.3 m, plantation density 3800 per hectare), these three components were in the ratio 5/4/l in importance to total damping. The aerodynamic and stem damping components were both larger for larger trees and could be reasonably predicted from foliage and stem characteristics. However, the damping component due to branch interference depended on spacing density, but could not be studied fully in the single, unthinned, uniformly planted, site studied. Simple mechanics should therefore form a useful basis for improved methods of predicting the risk of wind damage to forests growing under different spacing and thinning regimes, where data are available on the shape and mass distribution of trees grown under these different conditions.
Acknowledgments
The help of Robbie Wilson at the Institute of Terrestrial Ecology is gratefully acknowledged. Brian Manzie of Heriot-Watt University spent time on an undergraduate project, designing and constructing the mountings for the sway sensors. Partial funding for this work was provided by the Commission of the European Communities under their Materials Research Programme, Contract No. MA-0061. UK(BA).

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

References
Blackburn, P., J.A. Petty and K.F. Miller. 1988. An assessment of the static and dynamic factors involved in windthrow. Forestry 61:29-43. Blevins, R.D. 1979. Formulas for natural frequency and mode shape. Van Nostrand Rheinhold, New York. Booth, T.C. 1977. Windthrow hazard classification. Forestry Commission Research Information Note 22/77/SILN, Forestry Commission, Edinburgh. Cannell, M.G.R. and J. Morgan. 1987. Youngs modulus of sections of living branches and tree trunks. Tree Physiol. 3:355-364. Gardiner, B.A. 1989. Mechanical characteristics of Sitka spruce. Forestry Commission Occasional Paper No. 24, Forestry Commission, Edinburgh. Hoag, D.L., R.B. Fridley and J.R. Hutchinson. 1971. Experimental measurement of internal and external damping properties of tree limbs. Trans. ASAE 16:20-28.

SWAYING OF PICEA SITCHENSIS

399

Holbo, H.R., T.C. Corbett and PJ. Horton. 1980. Aeromechanical behaviour of selected Douglas-fir. Agric. Meteorol. 21:81-91. King, D.A. 1986. Tree form, height growth, and susceptibility to wind damage in Acer saccharum. Ecology 67:980-990. King, D.A. and O.L. Loucks. 1978. The theory of tree bole and branch form. Rad. Environ. Biophys., 15:141-165. Landsberg, J.J. and P.G. Jarvis. 1973. A numerical investigation of the momentum balance of spruce forest. J. Appl. Ecol. 1064-655. Landsberg, J.J. and A. Thorn. 1971. Aerodynamic properties of a plant of complex structure. Quart. J. Roy. Meteorol. Sot. 97565-570. Mayer, Hi 1987. Wind-induced tree sways. Trees 1:195-206. Mayhead, G.J. 1973. Sway periods of forest trees. Scottish For. 27: 19-23. Mayhead, G.J., J.B.H. Gardiner and D.W. Durrant. 1975. Physical properties of conifers in relation to plantation stability. Unpublished report of Forestry Commission Research and Development Division, Edinburgh. McMahon, T.A. and R.E. Kronauer. 1976. Tree structures: deducing the principle of mechanical design. J. Theor. Biol. 59:443-466. Miller, K.F. 1985. Windthrow hazard classification. Forestry Commission Leaflet No. 85, HMSO, London. Miller, K.F. 1986. Recent aeromechanical research in forest plantations. In Minimising Wind Damage in Coniferous Stands, Proc. Workshop Lovenholm, Denmark 1986. Commission European Communities, Brussels, pp 7-11. Milne, R. 1986. Methods of modelling tree stem bending under wind loading. In Minimising Wind Damage in Coniferous Stands, Proc. Workshop Lovenholm, Denmark 1986. Commission European Communities, Brussels, pp 12-16. Milne, R. and Blackbum, P. 1989. The elasticity and vertical distribution of stress within stems of Picea sitchensis. Tree Physiol. 5: 195-205. Papesch, A.J.G. 1984. Wind and its effects on (Canterbury) forests. Ph.D. Thesis, Univ. Canterbury, Christchurch, New Zealand. Petty, J.A. and C. Swain. 1985. Factors influencing stem breakage of conifers in high winds. Forestry 58:75-84. Simiu, E. and R.H. Scanlan. 1978. Wind effects on structures: an introduction to wind engineering. John Wiley, New York. Thomson, W.T. 1988. Theory of vibration with applications. Allen and Unwin, London.

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013

Downloaded from http://treephys.oxfordjournals.org/ by guest on June 2, 2013