Stress grading of Swedish and German timber. A comparison of machine stress grading and three visual grading systems.

Jan Brundin

Rolf Gruber

Stress Grading of Swedish and

German Timber

A comparison of machine stress grading and

three visual grading systems

TRÄTEK

Swedish Institute for Wood

Technology Research

FMPA - Otto Graf Institut

Forschungs- und

Material-prufungsanstalt

Baden-Wijrttemberg

Swedish National Testing and Research Institute

Building Technology

Jan Brundin

Rolf Gruber

Stress Grading of Swedish and

German Timber

A comparison of machine stress grading and

three visual grading systems

Detta digitala dokument skapades med anslag från

Stiftelsen Nils och Dorthi

Troedssons forskningsfond

SP

Swedish National Testing and Research Institute t • #

Building Technology * r ^ x V ^

SP REPORT 1992:23

iSPB

The report describes a project in which spruce timber (Picea abies) from

Germany and Sweden was graded visually and by machine, and subsequendy

tested to failiu-e. Grading was carried out according to the Nordic T-rules, the

German DIN 4074 and the ECE rules. The timber was also passed through two

types of grading machines, a Computermatic and a Cook Bolinder.

From bending tests on 58x120 mm timber it appears that the three visual grading

systems correspond well. The second highest grade i. e. the Nordic T2, the

German DE^ 4074 SIO and the ECE S8 had very similar 5th percentile bending

strength and mean modulus of elasticity values. These grades also had the highest

yield values (42 - 63%). The Swedish machine stress grade T30M showed

strength and modulus of elasticity values close to those of the highest visual

grades, but the yield of T30M was more than twice that of the highest visual

grades.

For 34x145 mm timber tested in tension, good agreement was only found between

the Nordic T2 and the German DIN 4074 SIO grade.

The relation between strength and the flatwise modulus of elasticity based on data

from the grading machines was virtually identical for German and Swedish

timber.

Combination of stif&iess data from the grading machine with edge knot size

in-creases the accuracy of the grading considerably.

Keywords: timber, visual stress grading, machine stress grading, tensile

strength, bending strength

SP

SP RAPPORT 1992:23

ISBN 91-7848-341-7

ISSN 0284-5172

Borås 1992

Swedish National Testing and

Research Institute

SP REPORT 1992:23

Postal address:

Box 857, S-501 15 BORÅS

Sweden

Telephone +46 33 16 50 00

Telex 36252 Testing S

Telefax+ 46 33 13 55 02

Abstract 2

Preface 5

1 Introduction 6

1.1 Stress grading in Germany and Sweden 6

1.2 Purpose of the investigation 7

1.3 Notations 7

2 Test material 9

2.1 Origin and quality 9

2.2 Handling and conditioning 10

3 Grading and measurements 11

3.1 Machine stress grading 11

3.2 Visual grading 14

3.2.1 General 14

3.2.2 D I N 4074 15

3.2.3 The ECE-rules 16

3.3.4 The Nordic T-rules 17

4 Testing and evaluation 18

4.1 Bending test (58x 120 mm timber) 18

4.2 Tension test (34x145 mm timber) 18

4.3 Determination of moisture content, oven dry density 19

and ring width

5 Results 20

5.1 General 20

5.2 Grading yield 20

5.3 Density, knot data, ring width and moismre content 22

5.4 Strength and modulus of elasticity 25

5.4.1 Statistical evaluation 25

5.4.2 Bending test (58x120 mm timber) 25

5.4.3 Tension test (34x145 mm timber) 30

6 Relation between strength and grading parameters 36

6.1 Visual grading 36

6.2 Machine stress grading 37

6.3 Combined visual and machine stress grading 40

6.4 Combined visual and density grading 41

7 Relation between different types of M O E 42

7.1 Edgewise M O E vs machine M O E 42

7.2 Tensile M O E vs machine M O E 44

7.3 Comparison of machine M O E s 45

8.2 Comparison of grading systems 47

8.2.1 General 47

8.2.2 58x120 mm timber-bending test 47

8.2.3 34x145 mm timber-tension test 48

8.3 Relation between strength and different grading parameters 48

9 References 51

Appendixes 53

1 Test data from all specimen 53

2 Descriptive statistics 84

This project has been carried out in cooperation between the Swedish Institute for

Wood Technology Research (TRÄTEK), Forschungs- und

Materialpriifungs-anstalt Baden-Wiirttemberg (FMPA) and the Swedish National Testing and

Research Institute (SP).

The project was planned at a meeting in Borås with the following people present:

Prof Peter Glos,

Borimir Radovic, Dip. eng.,FMPA

Rolf Gruber, Dip. eng. F H , F M P A

Jan Bnindin, Msc, TRÄTEK

Carl-Johan Johansson, Lic. eng. SP

Visual grading and testing took place at SP in Borås. Knot measurements were

carried out by KjeU Sjöberg, TRÄTEK, Jan Brundin, TRÄTEK and Bertil

Stenman, SP. Grading according to the German standard was done by Rolf

Gruber and Erich Niethammer, FMPA.

All the timber was machine stress graded at Anderssons Sågverk in Borgstena

(Computermatic) and at SP (Cook-Bolinder).

The bending and the tension tests were carried out by Bertil Stenman, Thomas

Claesson, Bertil Johansson and Kjell Pettersson from SP.

Transformation of knot projections into digital form, calculation of different knot

measures and grading according to the ECE-rules and the Nordic T-rules were

done at the Technical University of Denmark imder the supervision of Preben

Hoffmeyer.

Jiirgen König from TRÄTEK produced the MOE-distributions in figure 3.2.

Peter Glos and Borimir Radovic also read the manuscript and gave many valuable

comments that were considered in the final version of the report.

The project was financed by contributions from SIND (Swedish Industry Board),

TRÄTEK and Vereinigung Deutscher Sägewerksverbände e .V.

Stuttgart Stockholm Borås

1.1 Stress grading in Germany and Sweden

The amount of stress graded timber in Sweden is about 400 000

mVyear

(1990).

Most of it, about 250 000 mVyear, of which 150 000 m' is exported to the United

Kingdom, is machine stress graded.

The total Gennan production of stress graded timber is not known. The annual

production for glued laminated timber (glulam) alone is, however, over 300 000

m^. In Germany, machine stress grading has not yet been introduced on a large

scale. Only one stress grading machine is in operation. The machine, in which the

gamma radiation - absorption principle is used, is called IsoGreComat and is in

operation in a glulam factory.

In Sweden stress graded timber is mainly used in nail plate trusses, floor beams

and glued laminated timber. In Germany stress graded timber is mainly used for

all types of roof structures; not many floor beams are used.

Despite the fact that machine stress grading is quite common in Sweden it is at

present not used in glued laminated timber.

In Sweden, two standards for visual grading exist, the T-mles [7] for ordinary

structural timber and the L-rules [13], which contain a special grading system for

glued laminated timber. The T-rules were introduced 1951 and similar rales have

later been introduced in the other Nordic coimtries. These rules have recenfly

been modified to become a Nordic T-rale system [6], with the grades T 3 , T 2 , T l

and TO.

The Swedish T-rules contain three grades, namely T30, T24 and T18, which

correspond to the Nordic T3, T 2 and T l respectively.

The first grading machine in Sweden was approved in 1974. B y June 1991 the

number had grown to 34. In 1978 came special approval rules for machine stress

grading of stractural timber [14]. The grades are denoted T30M, T24M and

T18M.

In Germany, visual grading is carried out according to D I N 4074, first published

in 1939. A n important step was taken when the revised version of D I N 4074 was

issued in September 1989 [3]. Now the standard also includes rales for machine

stress grading.

D I N 4074 and the Nordic T-rules together, with a couple of other national

grading standards, will be adopted as European standards.

Next year the standardization work in C E N will result in a European standard for

machine stress grading. This is likely to have a major effect on the development

of different techniques of grading timber.

Setting values are required to operate grading machines. These values have to be

The export of structural timber, both visually and machine stress graded, from

Sweden to Germany has to date been limited to built-in components in

prefab-ricated houses and to laminations for glulam. On several occasions the question

has been raised of how the Swedish grades correspond to the German ones. Lack

of data has made it difficult to answer that question, so rough assumptions have

had to be made.

1.2 Purpose of the investigation

The purpose of the present investigation has been to compare three different

visual grading systems, namely the Nordic T-mles, the German D I N 4074 rules

and the European, ECE-rules [4]. It is, for example, of interest to establish

whether the German SIO grade corresponds to the Nordic T 2 grade (equal to the

Swedish T24 grade), as has been assumed in connection with export of Swedish

wood frame houses to Germany.

Another purpose has been to study the strength - stiffiiess relationship for both

Swedish and German spmce timber, to find out if the same basis for machine

setting values may be applied in both countries.

Finally, the aim has been to study whether increased precision can be gained by

combining machine grading with some kind of knot measure.

Note, that it has not been the purpose to compare the quality of Swedish and

German timber. The material used in the investigation can not be regarded as

representative of the timber produced in the both countries.

1.3 Notations

C O V coefficient of variation

C R A T I O sum of knots on the face and the edges divided by (w+2 t)

Ecomp

modulus of elasticity in flatwise bending based on load-deflection

data from a Computermatic machine

^cook

modulus of elasticity in flatwise bending based on load-deflection

data from a Cook Bolinder machine

E„ edgewise modulus of elasticity

E , modulus of elasticity in tension

F force acting on the timber in the grading machine

G shear modulus

K A R

L

M K A R

M O E

MOR

N R A T I O

T K A R

W R A T I O

h

n

r

t

u

w

5

Po,

knot area ratio

span

area of the projection of all knots in the quarter of the width

divided by w t/4

modulus of elasticity

modulus of ruptine - is used synonymous for bending strength

knot size on the edge divided by t

area of the projection of all knots in a cluster divided by w-t

knot size on the face divided by w

bending strength

tensile strength

depth of the cross section

depth correction factor

number of specimens

coefficient of correlation

thickness of the timber

moisture content

width of the timber

deflection

density with mass determined at a moistine content of 0% and

volume at 12% moisture content

2.1 Origin and quality

The test material came from three different saw miUs, two in Sweden and one in

southem Germany, see map in figure 2.1. The species was Picea abies. The

timber was planed and dried to a moisture content of about 12 %. Information

about dimensions, number of pieces and quality is given in table 2.1.

Värmland

Moelven

Valåsen A

Is

_{\ Västergötland}

Limmareds Skogar AB

Stockholm

Göteborg

Stuttgart

N • Miinchen

Heidenheim/ Ulm

Firma Sturm Holzverarbeitung

Table 2.1 Dimensions, number of pieces and quality of test material.

Origin

Saw mill

Thick- Width Length No of Quality

ness pieces

(mm) (mm) (m)

HeidenheimAJlm

Värmland

Heidenheim/Ulm

Värmland

Southern

Västergötland

Firma Sturm, 58

Holzverarbeitung

Moelven

Valåsen A B

58

Firma Sturm, 34

Holzverarbeitung

Moelven

Valåsen A B

Limmareds

Skogar A B

34

34

120 6.0 120 S7,S10')

and S13

120 4.2 135 Saw

falling

145 5.0-6.0 154 S7,S10')

andS13

145 4.5 42 Saw

falling

145 5.1-5.4 49 Saw

falling

1) Strength classes in DIN 4074.

The dimension 58x120 mm is common in structural applications in Germany, and

34x145 mm is used in glued laminated timber in both Germany and Sweden.

When the German timber was graded an effort was made to obtain as equal

numbers of S13, SIO and S7 as possible. This was however not achieved and

what is known is only, that the quality is better than S7 and that S13 in proportion

is likely to be greater than what is normally to be expected.

The 58x120 mm timber was sawn 2 ex log, whereas it is not quite clear how the

34x134 mm timber was sawn. A considerable part of it was probably sawn 3 ex or

even 4 ex log.

2.2 Handling and conditioning

On arrival at SP in Borås, the moisture content of the timber was measured. The

values were between 11 and 14 %. To achieve as uniform as possible moismre

content in the test material, it was placed under cover in the laboratory.

3 Grading and measurements

3.1 Machine stress grading

The test material was passed through two types of grading machines, first a Cook Bolinder machine and then a Computermatic.

The Cook Bolinder machine measures the force required to obtain a given deflec-tion, see figure 3.1. To account for initial deflections, each piece of timber is sent through the machine twice so that the piece is bent in two directions, after which the mean force value from the two passages is calculated. The machine used in this project was a laboratory version which enabled readings of the force at every 10 mm. No grading was performed with the Cook Bolinder. It was only used to determine the bending stiffness along the timber accurately, to locate the weakest point, see section 3.2.1.

In a normal production machine the mean force at every 100 mm is compared with set values for each grade. The lowest force value determines the grade to which a piece of timber belongs.

In this project, the force was recorded at every 10 mm except for a length of 480 mm at the ends of the timber. A l l the data were stored on P C discs to be used later at the visual grading. The E^„„^-value was calculated according to the following formula:

-cook = F L 3 / ( 5 - 4 8 1 ) (3.1)

where F is the force, L is the distance between the support rollers (900 mm), 6 is the preset deflection and I is the moment of inertia.

Figure 3.2 gives two examples of M O E distribution along a piece of timber. Loadcell

Deflection roller

Timber

10000

5000

*

_{•1 j}

\ J

_{\ ^}

/

S p e c i m e n No. c 7

L e n g t h 4 . 5 1 m

o .5 1 1.5 2 2 . 5 3 3 . 5 4 4 . 5

[m]

15000

10000

5000

S p e c i m e n No. c 1

L e n g t h 4 . 5 1 m

O .5 1 1.5 2 2 . 5 3 3 . 5

[m]

4 4 . 5

Figure 3.2 M O E distributions along two pieces of timber obtained by means

of the Cook Bolinder machine.

The grading in the Computermatic machine took place at a sawmill. This machine type is the most common one in Sweden, with a number of 32 of a total of 34 grading machines in operation. The principle of this machine type is described in figure 3.3. During the passage through the machine a constant load (F) acts on the timber. The deflection (5) is recorded at every 152 mm. The highest deflection value along the timber determines the grade. Owing to the construction of the machine, 600 - 700 mm of the timber ends are not measured. The distance be-tween the support rollers ( L ) is 914 mm. The load is chosen so that the bending stress is always 13.8 MPa. The modulus of elasticity (E^„^p) may be calculated according to equation (3.1) in the same way as E^^^^.

In the grading machine the highest deflection value along the timber is compared with maximum allowable values for the grades.

The allowable deflection values are based on the relation between the bending strength and the M O E in bending (E^). This relation was determined for Swedish pine and spruce timber in the mid 1970s and is assumed to be the same for both species [1]. For timber thicknesses from 34 to 63 mm the following E^-values are valid for the three machine grades:

T30M E ^ > 11580 MPa T24M E „ > 9250-"-T18M _{m —} 7610

-"-The deflection limits used in the grading machines are derived using the relation between the edgewise M O E (E^) and the flatwise M O E (E^^^p), equation 3.2 below, and equation 3.1 above.

E . = 1 . 2 1 - E _ p H - 8 4 0 (3.2)

Transducer B

Transducer A Measuremeni of bow

Loadroller

Air pressure cylinder

Figure 3.3 Principle of the Computermatic machine. Initial deflection is measured with the "shape arm", and is subtracted from the deflection at the load roller.

3.2 Visual grading

3.2.1 General

The Ep„„k-distribution for each piece of timber, see figure 3.2, was used as a basis

for the visual grading. The point with the lowest E^„„,^-value was located. With

few exception this corresponded to knots or top breaks. The knots were measured

and the projection of the knots on the timber cross section was recorded on a

millimeter paper as is shown in two examples in figvue 3.4. I f the location of the

lowest E^„„,^ - value was too close to the timber end, the second or the third lowest

value was choosen to enable this point to fall between the loads in the bending

test respectively between the grips in the tension test.

The ring width was measured on the density specimens that were taken close to

the location of the failure after testing, see section 4.3.

By means of a computer programme developed at the Technical University of

Denmark [5] the grades according to the E C E - mles [4] and the Nordic T-mles

[6] were determined. Later a correction with respect to the ring width was done.

Grading according to D I N 4074 [3] was carried out separately, but referred to the

same location along the timber as the previous grading.

3.2.2 D I N 4074

Measurement and calculation of the knot ratio A for single knots in square sawn timber

Measurement and calculation of the knot ratio A for single knots in boards

Measurement and calculation of the knot ratio A for a group of knots in boards

Splay knots (a7) are not taken into account when edge knots are < 1/2 (S/7), < 1/3 (SIO), < 1/5 (S13) of the thickness

Figure 3.5 Measurement of single knots and group of knots according to D I N

4074 [3].

Table 3.1 Requirements on knots, ring width and pith according to

D I N 4074 [3].

S7

Single knot (square A < 3/5

sawn timber)

Single knot (boards) A < 1/2

Group of knots (boards) A < 2/3

Ring width Unlim.

average

SIO

A<2/5

S13

A < 1/5

A < 1/3 A < 1/5

A < 1/2 A < 1/3

< 6 mm < 4 mm

3.2.3 TheECE-rules

17^^'^^W - ^ - 4 ^

^ =

k . - - - ^

—

a group o1 knots in a piece and their projectton on a cross-sectional plane.

(b) Front view of projection plane, showing projection ol knots (hatched)

Figure 3.6 The ECE grading principle [4].

Table 3.2 Requirements on knots and ring width according to the

ECE-mles [4].

SIO

S8

S6

either

<2r

either

21

Margin KAR

<l/5

<l/2

>l/2

<l/2

>l/2

Total KAR

<l/5

<l/3

<l/5

<l/2

<l/3

Ring width

< 6 mm

< 6 mm

< 10 mm

average

3.2.4 The Nordic T-rules

Face knot=df

Edge knot=d|j

Knot size=di+d2+d3

Figure 3.7 Measurement of single knots and groups of knots according to the

Nordic T-rules [6].

Table 3.3 Requirements on knots and ring width according to the Nordic

T-mles [6].

T3

T2

T l

TO

Edge knots

Face knots

Group of knots

Ring width

average

1/3 of thick. 1/2 of thick. 4/5 of thick. 1/1 of thickness

1/6 of width 1/4 of width 2/5 of width 1/2 of width

Maximum sum of knots equal to greatest allowable face +

greatest allowable edge knot.

4. Testing and evaluation

4.1 Bending test (58x120 mm timber)

Testing and evaluation was carried out according to ISO 8375 - Timber

struc-tures. Solid Timber in Structural Sizes; Determination of some Physical and

Mechanical Properties [10]. This means, for instance, that the span was 18 times

the nominal depth of the cross-section, i. e. 2160 mm. Load was applied at the

third points of the span. The loading rate (constant deformation rate) was adjusted

so that failure occured within 3 to 7 minutes. The curvature was measured in the

constant moment zone in the way shown in figure 4.1.

The worst defect, section 3.2.1, was placed in the zone where the curvature was to

be measured. The specimen was also oriented so that the "worst" edge was on the

tension side to obtain strength values on the safe side.

DEFLECTION TRANSDUCER

I 6 0 0

V ^ 0 ^ 720 7 2 0 ,

o

Figure 4.1 Test arrangement for measuring modulus of mpture (f^) and

modulus of elasticity in bending (E^)

4.2 Tension test (34x145 mm timber)

Testing and evaluation were performed according to ISO 8375 [10]. The distance

between the grips was 9 times the nominal width of the timber, i. e. 1305 mm.

Wedge type grips were used, which allowed no rotation of the timber ends. The

loading rate (constant deformation rate) was adjusted so that failure occured

within 3 to 7 minutes.

The elongation between two points spaced 4 times the width of the timber was

measured on both faces, see figure 4.2. The worst defect, see section 3.2.1, was

placed between these points.

WEDGE GRIPS

TRANSDUCER

580

1305

WEDGE GRIPS

Figure 4.2 Test arrangement for measuring tensile strength (f,) and modulus of

elasticity in tension (E,)

4.3 Determination of moisture content, density and ring

width

After testing, a disc of the cross section was cut out near the location of faUure.

On this disc, the moisture content (u) was determined according to ISO 3130 [8].

The density (po,,2) was determined according to ISO 3131 [9]. On the same disc

the average and maximum ring width was measured, as described in figure 4.3.

i m m

25 mm

RW=1 in mUlimeters/number of rings

Results

5.1 General

Grading yield is presented in section 5.2, density, knot data, ring width and

moisture content in section 5.3 and strength and modulus of elasticity in section

5.4.

Results from every specimen can be found in appendix 1, descriptive statistics in

appendix 2 and correlations in appendix 3.

5.2 Grading yield

In the tables below, the results from the visual and the machine stress grading are

presented. It should be observed that the grading has been performed under

conditions that deviate somewhat from the normal grading procedure. As is

described in section 3.2 the weakest point along the timber was located by means

of the bending stiffiiess measurement carried out in the Cook Bolinder machine.

The grade at that point was then determined with the different grading systems.

It is likely that the result would have been different i f the grading according to

one rule had been carried out independent of the others. For example, the worst

defect in a piece of timber according to the ECE-rules is probably not the same as

according to DIN 4074.

Table 5.1 Grading yield for 58x 120 mm timber.

Grading system

Yield (%)

Nordic T-rules

T3

T2

Tl

TO

reject

18

44

34

4

0

Machine

T30M

T24M

T18M

reject

Computermatic

50

42

8

0

DIN 4074

S13

SIO

S7

reject

13

60

27

0

ECE-rules

SIO

S8

S6

reject

Table 5.2 Grading yield for 34x145 mm timber

Grading system

Yield ( %)

Nordic T-rules

T3

T2

Tl TO reject

5

19

49 26 1

Machine

T30M

T24M

T18M

reject

Computermatic

72

24

3

1

DIN 4074

S13

570

57

reject

30

50

18

2

ECE-rules

SIO

S8

56

reject

8

43

23

26

The values in table 5.1 and 5.2 may be compared with results from an earlier

investigation of Swedish spruce and pine timber [11], see table 5.3. An equal

number of machine stress graded T30M and T24M timber was visually graded

according to the ECE-rules and the Swedish T-rules. The dimensions were

47x150 and 50x125 mm and the timber came from five different sawmills. The

same tendencies as in the present investigation were observed, namely that the

visual grading gives a very low yield for the highest grades.

Table 5.3 Results from an earlier investigation [11] concerning visual

grad-ing of machine graded timber from five Swedish saw mills. Spmce

and pine timber with the dimensions 47x150 and 50x125 mm. The

Swedish T-mles are in all essentials equal to the Nordic T-mles.

Grading system

Yield ( %)

Machine

T30M

T24M

Computermatic

50

50

Swedish T-mles

T30

T24

T18

reject

18

33

18

31

ECE-rules

SIO

55

56

reject

5.3 Density, knot data, ring width and moisture content

Mean and coefficient of variation (COV) values are given in tables 5.4 and 5.5.

The following can be noted:

For the bending specimens (58x120 mm) there is only a minor difference

in density between the visual grades, whereas, as expected, there is a

considerable density difference between the machine grades.

For the tension specimens (34x145 mm), however, there is a pronounced

difference in density also for the visual grades, which is somewhat

sur-prising.

The machine estimates knot size fairly well. For the bending specimens

mean TKAR for T30M is 0.22 and for T18M 0.31.

Judging from density and ring width values the overall quality of the

timber is quite good. For instance the maximum average ring width

values are 5.2 mm for the bending specimens and 5.1 mm for tension

specimens. In the German DIN 4074 class 510, the ECE class 88 and the

Nordic class T2 up to 6 mm is allowed.

600' 550+ 500+ 300+ 20O H 1 1 H r - 0 . 5 6 Y—27.6*X+473 450+

4

R i n g w i d t h - a v e r a g e (mm) R i n g w i d t h - a v e r a g e (mm)

Figure 5.1 Relation between density (p^ ,2)and average ring width for a)

bending (58x120 mm) and b) tension (34x145 mm) specimens.

Table 5.4 Density, knot size, ring width and moismre content of the bending

test specimen (58x120 mm). TKAR is the area of the projection of

all knots in a cross section divided by the cross section area of the

timber.

Grade Density, Po ,2

Knot size,

TKAR

Ring width

average, RW

Moisture

content, u

Mean GOV

Mean GOV

Mean GOV

Mean GOV

(kg/m3

_)(%)

_(%)

(mm) (%)

(%)

(%)

Nordic T-rules

T3 415

8

0.15

36

2.2

25

12.8

6

T2 405

8

0.23

23

2.3

26

12.9

6

T l 404

9

0.32

22

2.7

33

13.1

6

TO 403

18

0.43

24

3.3

20

12.8

6

Machine Computermatic

T30M 423 7

0.22

38

2.1

27

12.9

6

T24M 393 8

0.28

33

2.7

29

13.0

6

T18M 383

15

0.31

39

2.6

34

12.8

6

DIN 4074

S B 412

8

0.19

42

2.2

21

12.9

7

SIO 407

9

0.24

35

2.3

27

12.9

6

S7 404

11

0.32

31

2.9

32

13.0

6

ECE-rules

SIO 410

9

0.13

34

2.2

20

12.7

6

S8 406

8

0.23

24

2.4

30

12.9

6

S6 409

8

0.35

15

2.6

38

13.1

6

reject 398

14

0.43

24

2.8

25

13.0

7

All German timber

398

8

0.24

37

2.6

29

12.4

3

All Swedish timber

Table 5.5 Density, knot size, ring width and moisture content of the tension

specimens (34x145 mm). TKAR is the area of the projection of

all knots in a cross section divided by the cross section area of the

timber.

Grade Density,p(, ,2

Knot size.

Ring width

Moisture

TKAR

average, RW

content, u

Mean COV

Mean COV

Mean COV

Mean COV

(kg/m3)(%)

_(%)

_{(mm) (%)}

_(%)

_(%)

Nordic T-rules

T3 446

10

0.16

50

2.0

44

11.7

3

T2 419

10

0.22

40

2.2

28

11.8

4

T l 404

10

0.30

28

2.7

33

11.8

4

TO 397

7

0.36

30

2.8

32

11.9

4

Machine Computermatic

T30M 418 9

0.27

36

2.3

32

11.8

4

T24M 382 7

0.34

34

3.3

24

11.9

4

T18M 375 3

0.38

34

3.7

24

11.9

3

DIN 4074

S13 432

9

0.21

37

2.0

36

11.8

4

SIO 405

9

0.30

29

2.6

31

11.8

4

57 384

5

0.38

29

3.3

23

12.0

4

ECE-rules

SIO 431

12

0.14

37

1.9

44

11.7

5

S8 411

10

0.23

25

2.5

34

11.8

4

S6 406

8

0.32

19

2.6

30

11.9

4

reject 396

7

0.42

16

3.0

27

11.9

4

All German timber

400

8

0.31

34

2.8

32

11.9

4

All Swedish timber

5.4 Strength and modulus of elasticity

5.4.1 Statistical evaluation

For each grade three characteristic strength values, 5th percentiles, have been

calculated. The first value is based on the Gaussian distribution. Non-central

t-distribution has been used to estimate the 5th percentile at 75 % confidence level

[2]. This method gives a conservative 5th percentile value, especially when the

distribution is as skewed as for the tension test, see figures 5.8 to 5.10.

The second value is calculated by using a non-parametric approach. The lower 5th

percentile ranked test value has been determined. No adjustment has been made to

account for the higher variation, that is normally to be expected, for this method.

The values obtained in this way are slightly above the non-central t-distribution

values for the bending tests and much above for the tension tests.

The third value is the non-parametric 5th percentile value after multiplication

with a depth factor. The formula has been chosen according to the proposal in

prEN 384 [16] which will become European standard.

kd.p^ = (h/150)»-2 (5.1)

where h is the depth of the cross-section.

5.4.2 Bending test (58x120 mm timber)

As can be seen in table 5.6 the corresponding grades of the visual systems have

very similar mean strengths and MOE values. The fiist grade, i. e. T3, S13 and

SIO respectively, have 5th percentiles (corrected non-parametric), of 35.1 to 38.5

MPa. For the second grade T2, SIO and S8, where the yield was highest, the

agreement is even better. The bending strength is 28.5 to 29.4 MPa.

The first machine stress grade, T30M, corresponds well with T3, S13 and SIO but

T24M and T18M have somewhat lower MOE-values than the their visual

counterparts. T18M also has a considerably lower strength value than T l , S7 and

S6. It is quite natural that the MOE-values differ more for the machine grades

than for the visual grades, as the machine grading is based on the stiffness of the

timber.

A comparison can be made with results from the investigation [11] mentioned in

section 5.2. For 50x125 mm and 47x150 mm spruce timber {Picea abies) mixed

with 25 % pine (Pinus silvestris) the strength and MOE data below were obtained.

These are somewhat lower than the corresponding values in table 5.6:

Bending strength MOE

(5th percentile) (mean)

T30M 34.1 13500

T24M 25.0 11200

For the two lower ECE-grades the results are confusing. The strength of 88 is

lower than that of 86. The mean MOE of 88 is, however, as expected higher than

for 86. According to [5] similar results have been observed in an earlier

investiga-tion.

There was a considerable number of rejects from the ECE-grading. Therefore, the

bending strength and the MOE values of these have also been calculated. The

levels are comparable with those of T18M.

The required values for the Nordic grades and the machine grades are the

characteristic strength properties given in the Swedish building code N R l [15].

Grade

Bending strength

(MPa)

MOE

(MPa)

T30M, T3(=T30)

T24M, T2(=T24)

T18M,T1(=T18)

T0(=K12)

30

24

18

12

12000

10500

9000

8000

For all of the grades above, the 5th percentile strength values and the MOE mean

values exceed the code values by 5 to 60 %.

N o r d i c T2 DIN 4074 SIO

MOR (MPa)

Mean value=47.0 MPa

5th percent.=28.7 MPa

MOR (MPa)

Mean value=47.5MPa

5th percent.=29.4 MPa

Figure 5.2 Distribution of bending strength values (58x120 mm timber) for

the Nordic T2 grade and the DIN 4074 grade 810. The values have

not been corrected for depth.

Machine T30M Machine T24M

>1

O

(U

u

MOR (MPa)

Mean value=53.3 MPa

5th percent.=36.7 MPa

U

US

tr

u

MOR (MPa)

Mean value=40.4 MPa

5th percent.=29.5 MPa

80

Figure 5.3 Distribution of bending strength values (58x120 mm timber) for

T30M and T24M. The values have not been corrected for depth.

ECE S8 ECE S 6

>1

O

O)

p

u

MOR (MPa)

Mean value=46.9 MPa

5th percent.=29.8 MPa

80

MOR (MPa)

Mean value=41.4 MPa

5thpercent.=31.2MPa

Figure 5.4 Distribution of bending strength values (58x120 mm timber) for

EGE grade S8 and S6. The values have not been corrected for

depth.

Table 5.6 Edgewise bending strength and modulus of elasticity

Grade

Bending strength, f^

MOE, E„

Mean COV

5 t h p e r c e n t i l e

Mean COV

distrb. param. for depth

(MPa) (%)

(MPa) (MPa) (MPa)

(MPa) (%)

Nordic T-rules

T3

55.4

19

36.2 40.3

38.5

14550 15

T2

47.0 23

28.4 30.0 28.7

13050 18

T l

40.3

20

25.7 29.0 27.7

11680 21

TO

34.0 29

12.7

9650 21

Machine Computermatic

T30M

53.3

19

35.7 36.7

35.1

14540 14

T24M

40.4

10

27.1

29.5

28.2

11460 13

T18

33.2 21

19.9 19.9

19.0

9460 15

DIN 4074

S13

55.0

18

36.5

38.6 36.9

14430 15

SIO

47.5

23

28.5 30.8 29.4

13190 18

S7

39.0 24

22.2 26.1

25.0

11300 19

ECE-rules

SIO

54.8

22

31.8 36.7 35.1

14650 17

S8

46.9 23

27.9 29.8 28.5

12920 18

S6

41.4

17

28.1

31.2 29.8

11940 16

reject

35.4 24

19.2

19.9

19.0

10300 20

All German timber

46.4 24

12830 19

All Swedish timber

45.9 26

12790 21

The relation between bending strength and MOE is shown in figiue 5.5. A

com-parison can be made with results from two earlier investigations, [1] and [11]. In

[1], which gave the basis for the setting values for the grading machines in

Sweden, the relation between bending strength and MOE was found to be

f„ = 0.00383 E„ - 2.4

(5.2)

This relation was calculated from results of approximately 2200 bending tests

with an equal number of spruce and pine timber samples and also with equal

numbers of the dimensions 38x150, 50x100, 50x150 and 50x200 mm.

In [11], spmce and pine timber firom five saw mills in southern Sweden was

investigated. The bending strength - MOE relation was

f „ = 0.00378 E „ - 3 . 8

(5.3)

Both relations are virtually identical with the one presented in figure 5.5. All three

relations are compared in figure 5.7.

As can be seen in figure 5.6, there is only a small difference between German and

Swedish timber as far as the relation between bending strength and MOE is

con-cerned.

(13

CM

O

/t

/

/

/

/

MOE edge (MPa)

25000

Figure 5.5 Relation between edgewise bending strength and MOE for

German and Swedish 58x120 mm spmce timber.

fö lOO 90+ 8 0' 70+ 60+ 50+ 4 0 O

s

MOE e d g e ( M P a )

a

/ t

MOE e d g e ( M P a )

b

Figure 5.6 Relation between bending strength and MOE for 58x120 mm

timber from a) HeidenheinVUlm and from b) Värmland.

100

O

50+

5000 10000 15000 20000

MOE edge (MPa)

25000

0.00384*X-3.2 p r e s e n t i n v e s t i g a t i o n , 58x120 0.00383*X-2.4 38x150, 50x100, 50x150, 50x200 [ 1 ] 0.00378*X-3.8 f i v e s a w m i l l s , 50x125, 47x150 [ 1 1 ]

Figure 5.7 Bending strength - MOE relations from different investigations.

5.4,3 Tension test (34x145 mm timber)

Owing to the small number of specimens, non-parametric 5th percentile values

have not been calculated for theT3 grade, the T18M machine grade and the ECE

810 grade. Good agreement is only found between the Nordic T2 grade and the

DIN 4074 810 grade with corrected 5th percentile values of 19.0 and 18.4 MPa

respectively. The strength distributions show pronounced skewness, see figure 5.8

to 5.10. This explains why the 5th percentile based normal distribution is greater

than when the non-parametric approach is used.

The required 5th percentile values for the Nordic grades and the machine grades

are the characteristic strength properties given in the Swedish building code N R l

[15]. These are

Tensile strength MOE

(MPa) (MPa)

T30M, T3(=T30)

T24M, T2(=T24)

T18M,T1(=T18)

T0(=K12)

20

16

11

8

12000

10500

9000

8000

These values are all exceeded by the 5th percentiles of the Nordic visual grades.

The machine grade values do not, however, reach the code level. The reason for

this is probably incorrect setting values, and will be discussed later in section 7.1.

As in the bending test the EGE grade S8 has lower strength than S6. One reason

for this is the extremely high coefficient of variation of the S8 strength values.

Because of the high yield of rejects, strength and MOE values for these are also

presented.

30--o

25--(D

U

i n

T e n s . s t r . ( M P a )

T e n s . s t r . (MPa)

Mean value =29.0 MPa

5th percent. =13.5 MPa

Mean value=35.1 MPa

5thpercent.=19.1 MPa

Figure 5.8 Distribution of tensile strength values (34x145 mm timber) of the

Nordic grade T l and the machine stress grade T30M. The values

have not been corrected for depth.

D I N 4 0 7 4 S 1 3 D I N 4 0 7 4 S I O

T e n s . s t r . (MPa) T e n s . s t r . (MPa)

Mean value=43.8 MPa

5th percent.=29.0 MPa

Mean value=28.2 MPa

5thpercent.=18.5MPa

Figure 5.9 Distribution of tensile strength values (34x145 mm timber) for the

DIN 4074 grades S13 and SIO. The values have not been corrected

for depth.

E C E S 8 E C E S 6

T e n s . s t r . (MPa) T e n s . s t r . (MPa)

Mean value=33.0 MPa

5th percent.=15.6 MPa

Mean value=31.5 MPa

5th percent.=18.5 MPa

Figure 5.10 Distribution of tensile strength values (34x145 mm timber) for the

ECE grades S8 and S6. The values have not been corrected for

depth.

Table 5.7 Tensile strength and modulus of elasticity

Grade

Tensile strength, f(

MOE,

Mean GOV

5 t h p e r c e n t i 1 e

Mean GOV

distrib. param. for depth

(MPa) (%)

(MPa) (MPa) (MPa)

(MPa) (%)

Nordic T-rules

T3

47.2 22

25.9

16430 13

T2

38.6 32

16.1

19.1

19.0

13580 19

T l

29.0 37

10.1

13.5

13.4

11970 20

TO

23.3

35

8.5

12.3

12.2

10950 21

Machine Computermatic

T30M

35.1

33

12.8

19.2

19.1

13330 17

T24M

20.0 24

11.1

12.3

12.2

9740 12

T18M

12.8 32

—

7444 22

DIN 4074

S13

43.8

26

23.6 29.0 28.8

14680 14

SIO

28.2 26

15.4

18.5

18.4

11980 16

S7

18.4 31

8.1

9.2

9.1

9530 17

ECE-rules

SIO

42.9 29

18.9 —

—

14670 20

S8

33.0 38

11.0

15.6

15.5

12750 20

S6

31.5

30

14.3

18.5

18.4

12420 18

reject

20.9 31

9.3

8.9

8.8

10450 21

All German timber

28.5 42

11850 24

All Swedish timber

33.9 37

12850 17

The relation between tensile strength and MOE is shown in figure 5.11. The

coefficient of correlation is 0.83, which is a comparatively high value. In a

Swedish investigation from 1975 [12] the following relation betweeen tensile

strength and MOE was found for 33x155 mm spruce timber:

f, = 0.0048 E, - 20

(5.4)

The coefficient of correlation was 0.75. For low MOE-values i e 5000 to 10000

MPa this relation corresponds well with the regression line in figure 5.11.

As can be seen in figure 5.12, somewhat different regression lines are obtained

depending on the origin of the timber.

One point is specially marked out in figure 5.12 a. The failure in this case

occur-red at a compression wrinkle which was probably caused at felling or by wind

load on the standing tree.

fö

CM

-P

Cn

0)

u

-p

O)

H

-H

0)

/

/

+

MOE t e n s i o n (MPa)

Figure 5.11 Relation between tensile strength and MOE for 34x145 mm

German and Swedish timber.

100 fö S 80- 70-•4-1 Ö l 60-G 0) 50-+J to 40-O) 30-• H W C 20-<D EH 1 0-0 s t d dev of r e g r - 5 . 5 y-0.00372*X-15.5 r - 0 . 8 9 n - 1 4 8 5000 10000 15000 20000 25000

MOE t e n s i o n (MPa)

a

/t

4 '

/ A

A-A

1

MOE t e n s i o n (MPa)

b

2

U

Ä A

V A . A

//A _

^4.*^A

/ A

MOE t e n s i o n (MPa)

c

Figure 5.12 Tensile strength - MOE relations for 34x145 mm spruce timber

from a) Heidenheim/Ulm, b) Värmland and c) Västergötland.

6. Relation between strength and grading

parameters

6.1 Visual grading

In table 6.1, the relation between strength and the visual grading parameters, knot

size and ring width are presented. The different knot measures are

CRATIO: Sum of knots on the face and the edges divided by (w+2 t),

where w is the width of the board and t is the thickness.

WRATIO: Knot size on the face divided by w.

NRATIO: Knot size on the edge divided by t.

TKAR: Area of projection of all knots in a cluster divided by

w t .

MKAR: Area of projection of all knots in the outer quarter of the

width divided by wt/4.

Of these the CRATIO and TKAR show the best correlation with both the tensile

and bending strength. A considerable improvement is obtained by adding the ring

width (RW) to the regression equation.

100-90+ 8C" 20+ 10+ Y—60.9*X+61.4 s t d dev of regr-9.1 r - 0 . 5 1 n=250

A

AA

TKAR

a

T K A R

Figure 6.1 Relation between strength and knot size (TKAR) for a) 58x120 and

b) 34x145 mm timber.

Table 6.1 Relation (Y=Co +C, •X1+C2 X2) between strength and different

visual grading parameters.

Y X,

_Co

c,

C2

r

Std. dev. of

regression

Bending test of58x120 mm timber

f CRATIO

m

61.5

-69.1

0.53

9.7

f„ WRATIO

50.1 -28.2

0.20

11.2

f„ NRATIO

m

56.0 -28.0

0.24

10.5

f„ TKAR

61.4

-60.9

0.51

9.8

f„ MKAR

59.0

-37.5

0.50

9.9

f„ CRATIO RW(mean)

70.3

-56.5 -4.9

0.62

9.0

f„ CRATIO RW(max)

70.5

-57.4 -2.9

0.61

9.1

f„ TKAR

RW(mean)

70.6

-49.5 -5.0

0.61

9.1

f„, TKAR

RW(max)

71.5

-51.2 -3.1

0.61

9.1

Tension test of34x145 mm timber

f^ CRATIO

49.9

-82.2

0.61

9.7

f, WRATIO

34.7

-37.6

0.26

11.7

f, NRATIO

40.9

-18.2

0.42

11.0

f j TKAR

49.5

-65.5

0.59

9.9

f, MKAR

42.8

-28.2

0.45

10.9

f, CRATIO RW(mean)

58.8

-59.2 -5.6

0.71

8.7

f, CRATIO RW(max)

61.1 -60.3 -4.1

0.71

8.6

f, TKAR RW(mean)

59.0 -46.5 -5.8

0.70

8.8

f, TKAR RW(max)

62.0

-48.5 -4.4

0.71

8.6

6.2 Machine stress grading

Based on the force and deflection data from the grading machines, flatwise

MOE-values (E^Q^p and E^^^,;^) have been calculated. The relations between these and

strength are presented in table 6.2 and in figure 6.2 and 6.3. With both the

Com-putermatic (E^^mp) the Cook-Bolinder (E^^^,,) machines, the tensile and the

bending strength are estimated with considerably higher accuracy than when

using knot size and ring width, see table 6.1 and 6.2. For example the r-value for

the ft-E^„^p relation is 0.81, whereas the corresponding value for the

As can be seen in figure 6.2, the f^-E^^^p and f,-E^„„p relations are very similar

for German and Swedish timber. This implies that it is possible to use the same

setting values for the grading machines in both countries.

There is a marked difference between the regression lines for E^omp

^cook-reason for this could be the upper load limit in the Gook-Bolinder machine. To

avoid damage to the timber, the load is prevented from exceeding a certain limit.

This means that specimens with flatwise MOE values exceeding a certain limit

will be underestimated by the machine. This limit seems to be about 12500 MPa

for the 58x120 mm specimens and about 14000 MPa for the 34x145 mm

speci-mens. The difference between E^^^p and

^ c o o k

discussed in greater detail in

section 7.3.

5 0 0 0 1 0 0 0 0 15000 2 0 0 0 0 25000

MOE comp (MPa)

A H e i d e n h e i m / U l m O V ä r m l a n d 0.00474»x+3.0 H e i d e n h e i m / U l m - - 0.00567«x-4.5 Värmland Q) 20+

A

MOE comp (MPa)

H e i d e n h e i m / U l m V ä s t e r g ö t l a n d / V ä r m l a n d

0 . 0 0 4 2 1 * X - 1 4 . 3 H e i d e n h e i m / U l m

0 . 0 0 4 3 8 * X - 1 4 . 3 V ä s t e r g o t l a n d / V ä r m l a n d

Figure 6.2 Relation between strength and grading machine MOE (Ej.„^p),

a) 58x120 mm timber - bending test

Table 6.2 Relation (Y=Go+ G, X,) between strength and machine MOE

Y X, Go G, r stddevof

regression

Bending test of58x120 mm timber

comp

7

-"cook

- 1.0 0.00523

-23.0 0.00666

Tension test of34x145 mm timber

'comp ^cook

-14.8 0.00433

-32.8 0.00546

0.70 8.3

0.74 7.8

0.81 7.3

0.80 7.6

O

2

MOE c o o k ( M P a )

a

S

60--c

50--U

MOE c o o k ( M P a )

b

Figure 6.3 Relation between strength and grading machine MOE (E^„„^)

a) 58x120 mm timber - bending tests

6.3 Combined visual and machine stress grading

One disadvantage of the machine grading is that the effect of a knot on the

flat-wise bending stifftiess is nearly the same irrespective of whether the knot is in the

middle of the face or at the edge. At the same time, it is known that edge knots

normally initiate failure both in tension and bending tests. It is, therefore of

interest to study whether the estimation of the strength by the grading machine is

improved when information about knots is taken into account.

As can be seen in table 6.3 the r-value is increased when the CRATIO, NRATIO,

TKAR and MKAR are added. For bending strength, the r-value increases from

0.70 to 0.75 when TKAR is included and for tensile strength, r increases from

0.81 to 0.84. CRATIO and TKAR are both measures of the total size of the knots

in a cross-section. However, NRATIO, which is the size of the largest knot on the

two edges, gives approximately the same increase of r, despite the very low

corre-lation between the strength and the NRATIO, see table 6.1. The same applies to

MKAR which is the size of the knots in the outer quarter of the width.

This means that an important improvement of the mechanical stress grading could

be achieved by simply adding a device that detects knots on the timber edges.

Table 6.3 Relation (Y=Co-(-C, X,-t-C2 X2) between strength and machine and

visual grading parameters.

Y X, X j C,

C2

Co r std dev of

regression

Bending test of58x120 mm timber

^comp

0.00523

-1.0

0.70

8.3

CRATIO

0.00437

-39.1 15.3 0.75 7.5

NRATIO

0.00482

-18.1

9.0 0.75

7.6

TKAR

0.00443

-35.2 15.0 0.75 7.5

MKAR

0.00453

-24.3 13.6 0.77 7.3

Tension test of34x145 mm timber

f,

0.00433

-14.8 0.81 7.3

f,

CRATIO

0.00357

-30.5

0.1 0.83 6.8

f,

NRATIO

0.00393

-9.5

-5.3 0.84 6.7

f,

TKAR

0.00359

-29.9

1.5 0.84 6.6

MKAR

0.00390

-14.3

-4.2 0.84 6.7

6.4 Combined visual and density grading

Combining density and different knot measures gives a fairly good prediction of

strength, but not as good as the machine MOE alone, which can be concluded

from comparing tables 6.2 and 6.4.

This implies that grading machines based on the combination of density and knot

measurements will not be able to estimate strength as accurately as the bending

type machines, especially as the density and knot measurements are likely to

contain larger errors in a machine than in the present laboratory investigation.

Density seems to be a better predictor of tensile strength than of bending strength.

The r-value for the

relation is almost twice the r-value for the

relation.

Table 6.4 Relation

between strength and

combined visual and density grading parameters.

X,

Co C,

r std dev of

regression

Bending test of58x120 mm timber

Po,12 - 1 0 . 6 0 . 1 3 9 0 . 4 5 10.3

Po,12

CRATIO

Po,12

TKAR

Po,12

NRATIO

Po,12

MKAR

Po,12

CRATIO NRATIO

Po,12

TKAR

MKAR

Tension test of34x145 mm timber

f. Po,12 -52.1 0 . 2 0 3 0 . 6 2 9.8

ft Po,12

CRATIO

f. Po,12

TKAR

f. Po,12

NRATIO

f. Po,12

MKAR

ft Po,12

CRATIO NRATIO