CIB-W18/46-7-4

INTERNATIONAL COUNCIL FOR RESEARCH AND INNOVATION

IN BUILDING AND CONSTRUCTION

WORKING COMMISSION W18 - TIMBER STRUCTURES

WITHDRAWAL STRENGTH OF SELF-TAPPING SCREWS IN HARDWOODS

U Hübner

Association of the Austrian Wood Industries

AUSTRIA

MEETING FORTY SIX

VANCOUVER
CANADA
AUGUST 2013
 Withdrawal strength of self-tapping screws in hardwoods

Ulrich Hübnera

Abstract The axial withdrawal resistance of self-tapping screws of the diameters 6 mm (Z-9.1-
435, 2009), 8 mm (Z-9.1-656, 2007), 10 and 12 mm (Z-9.1-519, 2011) was tested according to
ON EN 1382 (1999) in glue laminated timer (GLT) made of European ash (Fraxinus excelsior
L., n = 2 657) with angles α between the fiber direction and the screw axis of 0◦ , 15◦ , . . . , 90◦ .
Screws with 4 mm diameter (Z-9.1-435, 2009) and threaded bars  20 mm (Z-9.1-777, 2010)
according to DIN 7998 (1975) were pulled out parallel and perpendicular to the grain to
determine the influence of the diameter over a large range. The influence of the screw tip
and the screw embedment length were analyzed. Self-tapping screws of diameters 8, 10 and
12 mm were also pulled out of European beech (Fagus sylvatica L., n = 371) and Black locust
(Robinia pseudoacacia L., n = 300) parallelly and perpendicularly to the grain.
If the moisture content rises 1 % the withdrawal resistance of screws decrease 2.7 % parallel to
the grain and 2.4 % perpendicular to the grain direction. The lower withdrawal resistance of
the tip should be considered with lef = lnom − 1.1 d, where lnom is the inserted nominal length
of the screw. The characteristic withdrawal resistance of self-tapping screws in European
hardwoods should be calculated for the regular thread with a bilinear design model with a
constant resistance from 30◦ to 90◦ according to R ax,k = fax,k π d lef with f ax,k = 7 · 10−4 · ρ1.6
k ·
d−0.34 and a linear reduction of 30 % to α = 0◦ . Additionally the thread of partial-threaded
screws should be embedded at least 2 d under the surface of the timber for 0◦ ≤ α ≤ 30◦ .
If the summation of characteristic withdrawal resistances was calculated according the new
design model for angels of 30◦ , 45◦ , . . . , 90◦ , the diameters 6, 8, 10 and 12 mm, the effective
lengths of 4 d, 5 d, 6 d and the characteristic density 672 kg/m3 the result exceeds the solution
according to ON EN 1995-1-1 (2009) by a factor of 1.79

The design of withdrawal resistance of self-

tapping screws according to ON EN 1995-1-1
1 Introduction (2009) is based on tests with spruce. Eckel-
man (1975), Jablonkay (1999) and Schneider
GLT made of birch (Griesser, 2012), chestnut (1999) gave an indication of the capacity of
(MPA Stuttgart, 2012), European beech (Z- screws in hardwoods but they used old fash-
9.1-679, 2011), European ash (Bogusch, 2011) ioned screws, only one diameter and species
and oak (Z-9.1-704, 2012) but also laminated respectivelly. Thus it was necessary to anal-
veneer lumber (LVL) made of European beech yses a wide range of screw diameters, angles
(Pollmeier, 2012) was developed. Effective between screw axis and fiber direction with
connections should use the impressive perfor- one species (ring porous European ash) and to
mance of these materials. compare the results with two other species rep-
resenting different wood anatomies and den-
a
sities (diffuse porous European beech, ring
Association of the Austrian Wood Industries
huebner@holzbauindustrie.at
porous Black locust) in order to create a re-
T +43 (0)1 712 26 01 23 liable data base for a new design model.

1
Table 2.1: Ratio of the core diameter and the pitch to the outer diameter for self-tapping screws with
a German technical approval (see DIBt, 2012)
core to pitch to characteristic tension

 strength
4 R t,u,k / πd22 N/mm2

outer diameter [−] outer diameter [−]
diameter 6 8 10 12 6 8 10 12 6 8 10 12
# of approvals 35 41 29 14 35 41 29 14 35 40 29 16
Minimum 0.60 0.63 0.60 0.57 0.37 0.42 0.39 0.36 588 554 567 623
Mean 0.64 0.65 0.63 0.60 0.58 0.62 0.57 0.49 878 862 816 763
Maximum 0.72 0.68 0.69 0.71 0.82 0.81 0.75 0.55 1 110 1 477 1 027 977

2 Method ( 20 mm) was 140 mm. Pilot tests showed

the need for reinforcement normal to the rod
axis with four or eight screws  8/160 mm to
The beech and ash logs were harvested in the prevent splitting (see Figure 3.5)). The den-
Austrian region Bucklige Welt in the growth sity and the moisture content of the hardwood
area Östliche Randalpen. The logs of Black around the screw axis were determined on
locust came from Zalaegerzeg/Hungary. The cuboids 4 d×4 d×specimen thickness to mini-
ash and Black locust boards (27 mm) were mize the influence of density variations within
glued with MUF Kauramin 683 and hardener the specimens. The moisture content of the
Kauramin 688 to form the GLT. The ash spec- specimens was analyzed according to ON EN
imens for 15◦ ≤ α ≤ 75◦ were sawn from 13 183-1 (2004).
six GLT. The beech scantlings had a section
of 60 mm × 120 mm. All specimens were pre- The withdrawal resistance was tested accord-
drilled with diameters equal to the root diam- ing to ON EN 1382 (1999) with the diameter
eters round down to 0.5 mm. In general the of the free surface around the screw of 4 d The
specimens were stored in a climate chamber influence of radial or tangential screw axis to
with (20±2) ◦ C and (65 ± 5) % relative humid- the annual growth rings was neglected follow-
ity. The influence of moisture contents from 5 ing Cockrell (1933).
to 24 % on the withdrawal strength was inves- The influence of the screw tip and the em-
tigated parallelly and perpendicularly to the bedded length were also analyzed. The sam-
grain direction. ples contained in general 60 specimens and for
The geometry of the regular thread of the used the comparison of mean values 40. The den-
screws of 4 mm, 6 mm, 8 mm, 10 and 12 mm sity of the samples were representative for the
are repesentative for most of the screws with basic population corresponding to the third
a German technical approval. The Table 2.1 method proposed by Leijten, Köhler and Joris-
shows the mean but also minimum and maxi- sen (2006, p. 9).
mum ratios of the core diameter and the pitch
of the outer diameter.
3 Analysis of the test results
The regular spacing and edge distance was 5 d.
The thickness of the specimens was limited by
the tension resistance of the screws and rods. 3.1 Moisture content and density
For the diameters 4 to 10 mm the thickness of
the specimen was lef 6 d. Screws with  12 mm The values for the moisture content are listed
were tested with lef 4 d and the rods with lef 8 d in Table 3.1. The mean moisture content of
The minimum width of specimen for the rods the regular specimens is equal to the mois-

2
  
Table 3.1: Moisture content of the regular samples Table 3.2: Density ρ12 kg/m3 of the samples
Species n min. mean max. CoV Species n min. mean max. CoV
[−] [%] [%] [%] [%] [−] [%]
Ash 2 657 8.20 10.75 14.80 3.33 Ash 2 620 555 746 918 6.11
Beech 371 10.30 11.02 11.80 1.93 Beech 371 582 719 851 6.09
Black locust 300 10.00 10.85 14.90 4.33 Black locust 300 609 750 884 6.84
All 3 328 8.20 10.80 14.90 3.41 All 3 291 555 744 918 6.29

ture content of ash in a climate of (20 ± 1) ◦ C thread. Specimens of the same thickness were
and (65 ± 5) % relative humidity and desorp- tested with the tip flush to the surface and
tion of moisture. Adsorption would lead to outside the specimen. The sample references
umean = 13.9 % as tested with 43 specimens. in Table 3.3 show ES for European ash, the
screw diameter, the angle between screw axis
Two samples (nq = 37, n⊥ = 40) made of
and grain direction and the thickness of the
ash were conditioned to a mean moisture con-
specimen.
tent (umean,q = 31 %, umean,⊥ = 27 %) under
the fiber saturation point of 33 % according
to Popper and Niemz (2009) to analyses the Table 3.3: Samples with screw tip outside or inside
influence of the moisture content on the with- the specimens
drawal strength parallel and perpendicular to tip outside tip inside xi
the grain direction. The two paired samples ES08_00_48 ES08_00_48S 1.28
had a mean moisture content of 11 %. The ES08_90_32 ES08_90_32S 2 .02
equations (1) and (2) represent the regression ES08_90_48 ES08_90_48S 0.98
lines for the withdrawal resistance versus the ES08_90_64 ES08_90_64S 0 .76
moisture content. If the moisture content rises ES10_00_60 ES10_00_60S 1.29
1 % the withdrawal resistance of screws paral- ES10_90_60 ES10_90_60S 0.98
lel to the grain direction decreases 2.7 % and ES12_00_60 ES12_00_60S 1.00
perpendicular to the grain 2.4 % relative to the ES12_90_60 ES12_90_60S 1.10
withdrawal resistance at the reference mois- Mean x̄ 1.11
ture content of uref = 12 %. This relationship
was also used to adjust the withdrawal resis-
tance of European beech and Black locust to The mean value x̄ = 1.11 results if the out-
uref . liers 2.02 and 0.76 in Table 3.3 are excluded
and equation (3) was derived to calculate the
R ax,mean,0 = −0.453 u + 22.2 (1) effective embedment length lef based on the
nominal embedment length lnom .
R ax,mean,90 = −0.438 u + 23.8 (2)
lef = lnom − 1.11 d (3)
The density was adjusted to uref according to
ON EN 384 (2010, p. 13) and Table 3.2 gives Equation (3) corresponds to the regulation
an overview. lef = lnom − d in SIA 265 (2003, p. 66), ON EN
1995-1-1 (2006, p. 83), Pirnbacher and Schick-
hofer (2009, p. 35), Eckelman (1975, p. 35)
3.2 Influence of the screw tip the rule to subtract the tip length according
to Newlin and Gahagan (1938, p. 6). CEN
It is obvious that the withdrawal strength of TC124 decided after a discussion over sev-
the screw tip is lower than that of the regular eral years with the statement: “For screws

3
it is clearly defined that the point length is 24

standard. withdr. strength fax,std [N/mm²]

22.8
included in the threaded length, both in EN 22
20.9
1995-1 and EN 14592. It is decided to have the 20
same rules for calculation of the withdrawal ca- 18
pacity of screws and nails by using the profiled 16 ●

length” (Ravasse, 2011, p. 7). “. . . all scientists 14 ●

●
●
agrees that the truth is that the tip should be 12 ●
●
● ●
●
disregarded” (Munch-Andersen, 2013). The 10 ● ●
● 8.2
challenge is a coherent system of test and de- 8 6.6
sign standard with logical rules for all dowel- 6
type connectors. 4
2
● ash 0° ● beech 0° ● Black locust 0°
The withdrawal test according to ON EN 1382 ash 90° beech 90° Black locust 90°
(2012) should use specimens with a thickness 0
relative to the diameter of the dowel type fas- 0 4 8 12 16 20 24 ∞
tener because of the influence of the transverse diameter d [mm]
strain near the surface, which can be seen as
a systematic error. The penetration depth Figure 3.1: Standardized withdrawal strength ver-
ld should only include the regular thread and sus diameter
not the whole profiled part including the tip
and eventually the special threads of a reamer
shaft. ing values.

fax,0,mean = 27.5 d−0.378 (4)

3.3 Influence of the diameter
fax,90,mean = 27.3 d−0.291 (5)
Screws with diameters of 4, 6, 8, 10 and 12 mm fax,0,mean = 8.25 + 14.6 e −0.152 d
(6)
and rods with a thread of  20 mm accord- −0.0624 d
ing to DIN 7998 (1975) were pulled out paral- fax,90,mean = 6.63 + 14.3 e (7)
lelly and perpendicularly to the grain to deter-
mine the influence of the diameter on the with- Büeler (2011, p. 22) published fv,mean =
drawal strength of European ash over a large 6.7 N/mm2 (CoV = 16.7 %) for five tests
range. Self-tapping screws of diameters 8, 10 with glulam beams made of European ash
and 12 mm with the same technical approvals (W 140 mm, H 480 or 600 mm, L 1 380 to 2 420
were also pulled out of European beech and mm). The rupture area due to withdrawal
Black locust parallelly and perpendicularly to tests parallel to the grain have to be around
the grain. the screw and not necessarily at the weakest
area of the section. This could be the expla-
Figure 3.1 shows the normalized (ρmean , uref )
nation for limd→∞ fax,0,nrm = 8.2 N/mm2 >
mean values of the withdrawal strength paral-
6.7 N/mm2 . The gray and white filled circles
lel and perpendicular to the grain for all diam-
for the other two species follow the curves for
eters and wood species. Two different models
ash. The exception ROB12_90_60 proves the
were used for the regression analysis of the
rule.
values for ash. The gray curves represent the
equations (4) and (5) and the black curves the Finally the regression curve (8) based on a
equations (6) and (7). Power functions are of- power function was calculated for the 5th per-
ten used to express size effects, but the loga- centiles of the withdrawal strength of all sam-
rithmic functions leads to more realistic limit- ples for α = 0◦ and α = 90◦ .

4
 3.5 From regression analysis to design
model
fax,05 = 23.3 d−0.340 (8)
The first regression analysis was based on the
This exponent of 1 − 0.340 = 0.660 describes
design model of ON EN 1995-1-1 (2009) for
the influence of the screw diameter on the
the withdrawal resistance and equation (9)
withdrawal resistance. Frese and Blaß (2009,
shows the logarithmized version. This was nec-
p. 10) published the exponent 1 − 0.3423 =
essary to obtain normal distributed residues
0.656 for self-tapping screws in spruce and Ra-
with the same variation over the whole diam-
jak and Eckelman (1993, p. 29) 1 − 0.355 =
eter range.
0.645 for MDF-panels. Thus the influence of
the diameter of d 0.66 on the withdrawal re-
ln R ax,mean = ln A + B ln lef + C ln ρ12 + . . .
sistance is reliable for MDF, soft- and hard-
. . . D ln d − ln sin2 α + E cos2 α


woods. The design model for the withdrawal
resistance in ON EN 1995-1-1 (2009) emanates (9)
from d 0.5 , published in the fundamental work ln Rax,mean = ln A + B ln lef + C ln ρ12 + . . .
of Bejtka (2005, p. 21) for screws with diam- (
eters from 6 to 12 mm. The wider range of ln (1 − E (30◦ − α)) 0◦ ≤ α < 30◦
D ln d −
diameters from 4 to 20 mm allows a more pre- 0 30◦ ≤ α ≤ 90◦
cise regression analysis. (10)

The standardization of the withdrawal

3.4 Withdrawal resistance strength of European beech, European
ash and Black locust was calculated for
Figure 3.2 on p. 6 shows the boxplots of the the reference values is ρmean = 744 kg/m3 ,
withdrawal strength at uref of all regular sam- dstd = 10 mm and lef = 60 mm with the coeffi-
ples with European ash. The withdrawal cients B = 0.972, C = 1.604 and D = 0.666
strength decreases with increasing diameter as of equation (9).
described in Section 3.3. The black bars of the
medians for screws parallel are always lower
than for screws perpendicular to the grain di- Table 3.5: Parameters for equation (10)
rection. The inter quartile range for α = 90◦ is parameter estimate standard error
in general wider than for 30◦ ≤ α ≤ 75◦ . Ta-
A −12.9 0.230
ble 3.4 indicates the coefficients of variation B 1.08 0.020 8
for all samples including extreme values. C 1.58 0.034 6
D 0.568 0.021 8
E 0.00576 0.000 136
Table 3.4: Coefficients of variation for all samples
of withdrawal strength fax,12 for Euro-
pean ash
The boxplot for the normalized withdrawal
dia- angle screw axis/fiber direction strength of all tested ash samples are shown in
meter 0◦ 15◦ 30◦ 45◦ 60◦ 75◦ 90◦
Figure 3.3 for different angles between screw
4 13.8 – – – – – 14.7 axis and grain direction. The white squares
6 13.0 13.9 10.7 9.2 9.4 10.1 11.8
indicate the 5th percentiles and the dotted
8 19.4 16.3 9.8 11.0 9.8 8.1 14.3
10 18.5 15.6 11.1 10.9 9.2 7.7 12.2 curve connects the 5th percentiles for α = 0◦
12 17.1 17.1 12.6 13.3 8.7 8.4 14.6 and α = 90◦ with a Hankinson-function. The
20 15.2 – – – – – 8.8 white squares for 15◦ ≤ α ≤ 75◦ are always
above the dotted line on the (very) safe side.

5
 withdrawal strength fax,12 [N/mm²]

25 25
diameter [mm]
4 6 8 10 12 20
20 ●
20
●

●
●
15 15
●
●

10 ●
●
10
●
●

5 5

n 62 62 63 59 60 61 62 61 62 60 61 61 61 61 61 66 62 62 60 60 62 60 60 61 61 60 60 61 60 59 58 61 n
ES04_00_24
ES04_90_24
ES06_00_36
ES06_15_36
ES06_30_36
ES06_45_36
ES06_60_36
ES06_75_36
ES06_90_36
ES08_00_48
ES08_15_48
ES08_30_48
ES08_45_48
ES08_60_48
ES08_75_48
ES08_90_48
ES10_00_60
ES10_15_60
ES10_30_60
ES10_45_60
ES10_60_60
ES10_75_60
ES10_90_60
ES12_00_60
ES12_15_60
ES12_30_60
ES12_45_60
ES12_60_60
ES12_75_60
ES12_90_60
ES20_00_160
ES20_90_160
Figure 3.2: Boxplots of the withdrawal strength for ash
stand. withdrawal strength fax,std [N/mm²]

●
●
●
18 ●
●
●
●
●
●
˜−5.2 med (|ˆ
i − ˜|) ≤ ˆi ≤ ˜+5.2 med (|ˆ
i − ˜|)
●

16 ●
●
●
●
●
●
(11)

14 Equation (11) was choosen due to its robust-

ness according to Hampel (1985, p. 98) to elim-
12 inate 24 outliers of 2 621 datasets based on the
● ●
●
residues ˆi and the mean residue ˜.
10
R ax,mean = 2.39 · 10−3 lef ρ1.6
mean d
0.66
...
8 (
1 − 0.006 (30◦ − α) 0◦ ≤ α < 30◦
...
n 640 242 241 242 246 242 742 1 30◦ ≤ α ≤ 90◦
(12)
0 15 30 45 60 75 90
R ax,05 = 2.2 · 10−3 lef ρ1.6
05 d
angle screw axis/fibre directionα [°]
0.66
...
(
1 − 0.006 (30◦ − α) 0◦ ≤ α < 30◦
Figure 3.3: Boxplots of standardised withdrawal . . .
strength 1 30◦ ≤ α ≤ 90◦
(13)

The estimates and their standard errors of the

parameters in equation (10) are listed in Ta-
The break of slope for the bilinear curve is ble 3.5. For practical reasons the parameter
always the 5th percentile for α = 30◦ . The B was set to 1.0 and the other parameters
slope of the left part results from the tangent were calculated again. Then C was rounded
to the 5th percentile for α = 15◦ . The sum to 1.6. The regression analysis was repeated
of the differences between the 5th percentiles until all parameters were rounded and fixed.
and the bilinear curve is smaller than between R ax,mean,i according to equation (12) and the
the bilinear curve and the dotted line. The test results are plotted in Figure 3.4 and form
following regression analysis was thus based an ideal type scatter band. The orange circles
on the bilinear equation (10). for Black locust and the red circles for beech

6
fit to the circles for ash. The different wood ues were compared to all the measured values
anatomies can be sufficiently represented by and for 32 of 44 samples all test values includ-
the densities. ing the extreme values exceed the estimates.
In nine samples 98 % passed, in two sampels
●
●●●
●
theoretical withdrawal resistance [kN]

● ●● ●●
European beech 97 % and in one sample the limit of 95 % was ●● ●
●●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
100 ●●
●●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●
●

Black locust reached. Only 16 out of 2 621 vaues or 0.6 % ●●●

●●
●●
●
●
●
●
●●
●●●
●●●
● ● ●
were on the unsafe side. ●

50
●●●
●
●
●
●
●
●●
●
●
●
●
●
●●
●
●
●
●●
●
●●
●●●
●
●
●●
●
●
●
●
● Pirnbacher and Schickhofer (2012, p. 132)
●●●
●●
●
●
●●
●●
●
●●
●
●●
●
●●
●
●
●●●
●
●
●
●
●●
●●●
● ●●●
●●
●●●
●
●●
●
●
●●
●
●
●●
●
●
●
● ●

●●●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
stated concerning the long term behavior of
●●●●
●●●
●
●●
●
●
●●
●
●
●
●
●●
●
●●
●
●
●●
●●
●●
●
●
●●
●
●●
●
●●
●
●
●●
●●
●
●● ●
●
●●
●
●●●
●
●
●●
●
●●
●
●●
●
●●
●●
●
●●
●
●
●
●●
●
●
●●
●
●●
●
●
●
20 ●●●
●
●
●●
●
●
●
●
●●●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●
●●
●
●
●
●●
●●
screws parallel to the grain: “when the applied
●●●
●●
●●
●●
●●
●●
●
●
●●
●
●
●●
●
●
●
●
●
●
●
●●
●
●
●●●
●●
●
●
●
●●
●
●●●
●●
● ●
●●●●●
●●
●
●●
●
●
●●
●
●
●
●●
●
●●
●
●
●●
●●●
●
●●
●●●
●
● ●
●
●●
●●
●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●●
● ●
●
●
●●●●●
●
●
●
●
●
●
●● load does not exceed a threshold at about 73 %
●●●
●
●
●
●●
●
●●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●●●
●
●
●●●
●●
●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●●●
●● of the ultimate strength their behavior is cov-
10 ●●
●●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●●
●
●
●●
●
●●
●
●
● ●●●
●●
●
●
●
●
●
●
●
●
●
●
●
●
● ●● ●
●
●
●
●
●
●
●
●
●
●●
●
●
●● ered by the application of kmod as currently
●● ●●●
● Diameter European ash
●●●● ●●
●●●
●
●●
●
●
●
●
●
●
●
●●
●●
●
●
present in the EC5”. SIA 265 (2012, p. 71) re-
●●●●●●●●
●
●●
●●
●●● 4 mm 10 mm
5 ●●
●
●
●
●
●
●
●
●
●●●
●●
●
●
●
●●
●
●
●●
●
●●●
●
●●
●
●●
●
●●
●●
quires a minimum effective thread length of
●●●
●● ●
●
● ●
●
●● ●
●
6 mm 12 mm
●
● ● 8 mm 20 mm 100 mm or 8 d respectively for screws parallel
to the grain direction. A similar rule was for-
5 10 20 50 100 mulated in SIA 265 (2003). Uibel and Blaß
empirical withdrawal resistance [kN] (2013, p. 132) reported about long term tests
in spruce with screws axially loaded with 70 %
Figure 3.4: Tested versus theoretical withdrawal
resstance according to equation (12)
of the design resistance. 19 of the 48 screws
(40 %) failed during the test of almost five
years. Equation (13) is valid for short term
The adjusted coefficient of determination was loads with a load duration of (300 ± 120) s.
2 = 0.966. The normal distributed residues
radj Pirnbacher (2011, p. 81) reported a positive
had an estimated standard deviation of σ̂ = effect of the embedment of the thread of
0.1073. Thus the expected value for the with- partially-threaded screws on the long term be-
drawal resistance has a coefficient of variation havior. Summarizing these publications it is
of 10.77 %. The relation of the 5th percentile necessary to reduce the withdrawal resistance
to the mean of the expected withdrawal re- for screws parallel to grain with the factor 0.01
sistance is constant for all parameter combi- instead of 0.006 in equation (13) and addition-
nations according to the design model and ally use an embedment length of lemb = 2 d.
equal to 0.823. The 5th percentile of the den-
sity of the dataset is 672 kg/m3 and the mean R ax,k = 2.2 · 10−3 lef ρ1.6
k d
0.66
...
3
(
value 743 kg/m . If the characteristic density 1−0.01 (30◦ − α) 0◦ ≤ α < 30◦ ∩ lemb = 2 d
is introduced in equation (12) a reduction to . . .
1 30◦ ≤ α ≤ 90◦
90.5 % follows. The parameter exp(A) has to
be multiplied with 1−(0.905 − 0.823) = 0.917 (14)
to get the necessary 82.2 % of the expected
If all the characteristic withdrawal resistances
withdrawal resistance equal to the estimated
are calculated according to equation (14) for
5th percentile of equation (13).
angles of 30◦ , 45◦ , . . . , 90◦ , the diameters 6,
The 5th percentile of the density of each 8, 10 and 12 mm, the effective lengths of
species was calculated and introduced in equa- 4 d, 5 d, 6 d and the characteristic density
tion (13) to calculate with each dataset of 672 kg/m3 the results exceed the solution ac-
lef,i and αi the estimated 5th percentile of the cording to ON EN 1995-1-1 (2009) by the fac-
withdrawal resistance. These theoretical val- tor 1.79. A second advantage is that the new

7
model covers a wider range of diameters from
4 to 20 mm and angles between screw axis and
grain direction from 0◦ to 90◦ .
Engineers like mechanically logical design
models where the parameters with their cor-
responding units give reasonable results. In
equation (15) fax,k is the withdrawal strength
in N/mm2 not the withdrawal parameter ac-
cording to ON EN 1995-1-1 (2009).

R ax,k = fax,k π d lef (15)

The description of fax,k should cover com-

plete different rupture mechanisms. The with-
drawal strength parallel to the grain is the
shear strength around the outer diameter of
the screw with an influence of the screw tip
and the transverse strain near the surface of
the timber. The withdrawal resistance perpen-
dicular to grain is much more complicated to Figure 3.5: Cut in half specimen after withdrawal
test (ES20_90_160_54,  20 mm,
describe. The rupture is a mix of shear cracks
α = 90◦ )
parallel to grain due to the bending of the
fibers, rolling shear, compression and tension
perpendicular to grain (see Figure 3.5). Addi- for European ash according to Hübner, Rasser
tional size effects have to be considered. The and Schickhofer (2010) as the black dashed
equation (14) summarizes the different param- curve. The blue curve was plotted according
eters and effects. The alternative is to com- to ON EN 1995-1-1 (2009). The slope of the
bine equations (15) and (16). blue curve in Figure 4.1 is flat due to the ρ0.5
k
for softwoods. The design model of SIA 265
f ax,k = 7 · 10−4 · ρ1.6
k ·d
−0.34
... (2003) is no longer valid but the red curve was
( plotted to show the former performance com-
1−0.01 (30◦ − α) 0◦ ≤ α < 30◦ ∩ lemb = 2 d
... pared to ON EN 1995-1-1 (2009). DIN 1052
1 30◦ ≤ α ≤ 90◦ (2008) was replaced by the Eurocode. The
(16) light green horizontal lines limitation accord-
ing to DIN 1052 (2008) are the result of the
restriction to ρ k = 500 kg/m3 for softwoods
4 Comparison with other for the three load bearing classes for screws.
The steep slope due to ρ2k in the former design
design models model is remarkable. The German technical
approval Z-9.1-519 (2012) was the first which
Schneider (1999) published a design model allowed the application of modern screws in
for the withdrawal resistance of self-tapping hardwoods. The dark green curves are the
screws (anchor for window frames  7.5 mm) closest to the black curve for equation (14).
in European beech. The orange curve and Figure 4.1 shows that the design models for
circle in Figure 4.1 a and 4.2 a respectivelly soft- and hardwoods should consider the dif-
represent this model. Equation (14) was plot- ferent influences of the density. The influence
ted as the black curve and the design model of the diameter is the same for soft- and hard-

8
 characteristic withdrawal resistance [kN]

characteristic withdrawal resistance [kN]

characteristic withdrawal resistance [kN]

50 DIN 1052 TK 1−3 Schneider (1999) 50 50 Rt.u.k
Z−9.1−519 Hübner (2010)
45 Rax(LH)
45 45
EN 1995−1−1
40 SIA 265 40 40
35 35 Rt.u.k 35
30 30 30 ● 29.4
● 166 %
29 163 %
25 25 25
Rt.u.k 22 160 %
●
● ● 21.6 122 %
20 20 21.6 157 % 20
● 17.8 100 %
● 15.5 154 % ● 16.2 118 %
15 15.1 150 % 15 ● 13.8 100 % 15
●
11.3 112 %
10 ● 10.1 100 % 10 10
5 5 5
ρmin ρ05 ρmax ρmin ρ05 ρmax ρmin ρ05 ρmax
0 0 0
400 500 600 700 800 900 1000 400 500 600 700 800 900 1000 400 500 600 700 800 900 1000
characteristic density [kg/m³] characteristic density [kg/m³] characteristic density [kg/m³]

(a) diameter 8 mm (b) diameter 10 mm (c) diameter 12 mm

Figure 4.1: Withdrawal resistance for screw diameters (lef = 7 d, α = 90◦ )

20 40 40
char. withdrawal resistance Rax,k [kN]

char. withdrawal resistance Rax,k [kN]

char. withdrawal resistance Rax,k [kN]

18 35 35
16
30 30
14 13.1 127 % ● 27.8 184 %
●
●
● 12.6 122 % 25 25
12 12.3 119 %
10 ● 10.4 100 % 20 ● 20 156 % 20 ● 19.7 131 %
●
●
8 ● 16 125 % ●
15.1 100 %
15 ●
12.8 100 % 15
● ●
6 12.6 99 % 12.6 84 %
10 10
4 ● Schneider (1999)
2 DIN 1052 TK 1−3 SIA 265 (2003) 5 5
Z−9.1−519 Hübner (2010)
0 EN 1995−1−1 Rax(LH) 0 0
4 6 8 10 12 14 16 18 20 4 6 8 10 12 14 16 18 20 4 6 8 10 12 14 16 18 20
screw diameter d [mm] screw diameter d [mm] screw diameter d [mm]

(a) characteristic density 500 kg/m3 (b) charact. density 650 kg/m3 (c) characteristic density 800 kg/m3

Figure 4.2: Withdrawal resistance for different characteristic densities (lef = 60 mm, α = 90◦ )

woods. The curves in Figure 4.2 are more par- technical approvals for screws the spacing per-
allel with the exception of the blue curve for pendicular to the grain direction can be re-
ON EN 1995-1-1 (2009). The slope for diame- duced to a2 = 2.5 d if a1 · a2 = 25 d2 .
ters d ≥ 8 mm is too flat but for d < 8 mm it
Plieschounig (2010) recommends a1 ≥ 7 d and
is too steep.
a2 ≥ 3,5 d with a1 · a2 = 24.5 d2 for screws per-
pendicular to the grain. Parallel to the grain
direction a quadratic grid with a = 5 d and
5 Further research for α = 45◦ seam a1 ≥ 6 d and a2 ≥ 4 d with
a1 · a2 = 24 d2 logically. For the time being
these are theoretical considerations and more
The new design model allows reaching the ten- research like published by Mahlknecht (2011)
sile resistance of the screws with shorter effec- has to be carried out.
tive lengths. The question is how to increase
the transferable load. One answer could be
screws with a greater ratio of core to outer
diameter. Another answer are optimized min-
imum spacings’s and edge/end distances. In

9
6 Acknowledgment [DIN 1052 2008] DIN 1052: Entwurf, Berech-
nung und Bemessung von Holzbauwerken – Allge-
meine Bemessungsregeln und Bemessungsregeln für
This research work Was part of the COMET- den Hochbau. Dezember 2008
K-project 1.2.3_hardwood_connections of the
[DIN 7998 1975] DIN 7998: Gewinde und Schraube-
competence centre holz.bau forschungs gmbh nenden für Holzschrauben. Februar 1975
and was carried out in collaboration with the
Institute for Timber Engineering and Wood [Eckelman 1975] Eckelman, Carl A.: Screwholding
Technology at the Graz University of Tech- performance in hardwoods and particleboard. In:
Forest Products Journal 25 (1975), No. 6, p. 30–35
nology and industrial partners HAAS Fertig-
bau GmbH & Co.KG and SPAX International [Frese and Blaß 2009] Frese, Matthias ; Blaß,
GmbH & Co. KG. The withdrawal tests Hans Joachim: Models for the calculation of the
withdrawal capacity of self-tapping screws. In: Pro-
were executed with the aid of Bernd Heis-
ceedings of CIB-W18. Dübendorf, Schweiz, August
senberger, Markus Kummer, Manuel Lögl and 2009. – Paper 42-7-3
Mathias Rasser. The research work was fi-
nanced through the Austrian State Ministry of [Griesser 2012] Griesser, Doris: Fichte und Birke
als tragende Elemente. In: Der Standard (2012),
Economic Affairs and Employment, the Min-
7. März, p. 17
istry of Transport, Innovation and Technol-
ogy as well as by the Steirischen Wirtschafts- [Hampel 1985] Hampel, Frank R.: The breakdown
förderungsgesellschaft mbH and the federal points of the mean combined with some rejection
rules. In: Technometrics 27 (1985), No. 2, p. 95–
state Steiermark. 107

[Hübner et al. 2010] Hübner, Ulrich ; Rasser,

Mathias ; Schickhofer, Gerhard: Withdrawal re-
References sistance of screws in common ash (Fraxinus Excel-
sior L.). In: Proceedings of 11th World Conference
[Bejtka 2005] Bejtka, Ireneusz: Karlsruher on Timber Engineering (WCTE) vol. 1. Riva del
Berichte zum Ingenieurholzbau. vol. 2: Verstärkun- Garde, Italy, 21. Juni 2010. – Paper 481
gen von Bauteilen aus Holz mit Vollgewinde-
schrauben. Karlsruhe : Universitätsverlag Karl- [Jablonkay 1999] Jablonkay, Peter: Schrauben
sruhe, Januar 2005. – URL http://digbib.ubka. auf Ausziehen, ETH Zürich, Forstwissenschaften,
uni-karlsruhe.de/volltexte/1000003354. – date Abteilung VI, diploma thesis, 1999
of access: 12. Juli 2010. – ISBN 3-937300-54-6
[Leijten et al. 2006] Leijten, Ad J. M. ; Köhler,
[Büeler 2011] Büeler, Michael: Schubverhalten Jochen ; Jorissen, André J. M.: Timber density
von Brettschichtholzträgern aus Buche und Esche. restrictions for timber connection tests according to
Zürich, Institut für Baustatik und Konstruktion – EN 28 970/ISO 8970. In: Proceedings of CIB W18.
Stahl-, Holz- und Verbundbau, ETH Zürich, Mas- Florence, Italy, August 2006. – Paper 39-21-1
terarbeit, 4. Juli 2011
[Mahlknecht 2011] Mahlknecht, Ursula: Unter-
[Bogusch 2011] Bogusch, Walter: Eschenholz im suchung von rechtwinklig zur Faser eingebrachten,
Tragwerkbau. In: Newsletter Forum Holzbau 51 axial beanspruchten Schraubengruppen im Voll-
(2011), März holz und Brettsperrholz, Institut für Holzbau und
Holztechnologie der TU Graz, diploma thesis, 2011
[Cockrell 1933] Cockrell, Robert A.: A study of
the screw-holding properties of wood / New York [MPA Stuttgart 2012] MPA Stuttgart: EC-
State College of Forestry. Syracuse, New York, 1933. Certificate of Conformity 0672 for GLT made of
– Technical Publication No. 44 sweet chestnut. Sierolam S. A., Los Cuentos s/n
Argüelles, 33188 Siero, Spain. 31. August 2012. –
[DIBt 2012] DIBt: Verzeichnis der allgemeinen certificate number 0672-CPD-114.21.53
bauaufsichtlichen Zulassungen – Zulassungsbereich:
Holzbau und Holzwerkstoffe. Berlin : Deutsche In- [Munch-Andersen 2013] Munch-Andersen, Jør-
stitut für Bautechnik (DIBt), 22. Februar 2012. – gen: Re: AW: Effective length l_ef for screws. E-
research report Mail. 14. Juni 2013

10
[Newlin and Gahagan 1938] Newlin, J. A. ; Gaha- [Popper and Niemz 2009] Popper, Rudolf ;
gan, J. M.: Lag-screw joints: Their behavior and Niemz, Peter: Wasserdampfsorptionsverhalten aus-
design / United States Departement of Agriculture. gewählter heimischer und überseeischer Holzarten.
Washington, D. C., Januar 1938. – Technical bul- In: Bauphysik 31 (2009), No. 2, p. 117–121
letin No. 597
[Rajak and Eckelman 1993] Rajak, Zaini Ithnin Bin
[ON EN 13 183-1 2004] ON EN 13 183-1: Haji Abdul ; Eckelman, Carl A.: Edge and face
Feuchtegehalt eines Stückes Schnittholz – Teil 1: withdrawal strength of large screws in particleboard
Bestimmung durch Darrverfahren. 1. Februar 2004 and medium density fiberboard. In: Forest Products
Journal 43 (1993), No. 4, p. 25–30
[ON EN 1382 1999] ON EN 1382: Holzbauw-
erke – Prüfverfahren – Ausziehtragfähigkeit von [Ravasse 2011] Ravasse, François: Minutes meeting
Holzverbindungsmitteln. 1. November 1999 Paris REV 2011 on April 19 and 20. 27. Juli 2011.
– N 1079
[ON EN 1382 2012] ON EN 1382: Timber Struc-
tures – Test methods – Withdrawal capacity of tim- [Schneider 1999] Schneider, Patric: Auszugsfes-
ber fasteners. Mai 2012 tigkeit von EJOT-Rahmenschrauben / Eidgenös-
sische Technische Hochschule. Zürich, 29. Oktober
[ON EN 1995-1-1 2006] ON EN 1995-1-1: Eu-
1999. – research report
rocode 5: Bemessung und Konstruktion von
Holzbauten – Teil 1-1: Allgemeines – Allgemeine [SIA 265 2003] SIA 265: Holzbau. 1. März 2003
Regeln und Regeln für den Hochbau. November 2006
[SIA 265 2012] SIA 265: Holzbau. 1. Januar 2012
[ON EN 1995-1-1 2009] ON EN 1995-1-1: Eu-
rocode 5: Bemessung und Konstruktion von [Uibel and Blaß 2013] Uibel, Thomas ; Blaß,
Holzbauten – Teil 1-1: Allgemeines – Allgemeine Hans J.: Joints with dowel type fasteners in CLT
Regeln und Regeln für den Hochbau. 1. Juli 2009 structures. In: Harris, R. (eds.): Focus Solid Tim-
ber Solutions - European Conference on Cross Lam-
[ON EN 384 2010] ON EN 384: Structural timber
inated Timber (CLT). Graz : TU Graz, 21./22. Mai
– Determination of characteristic values of mechan-
2013, p. 119–134. – ISBN 1-85790-181-9
ical properties and density. 15. Mai 2010

[Pirnbacher 2011] Pirnbacher, Gernot: [Z-9.1-435 2009] Schmid Schrauben Hain-

Screws_Long_Time: Zeitabhängige Entwicklung feld GmbH: Holzbauschrauben Star Drive,
der Traglast und Kriechverhalten von selbst- Rapid, Rapid Perfekt und Schmid SP als
bohrenden Holzschrauben / holz.bau forschungs Holzverbindungsmittel. Berlin : Deutsches Institut
gmbh. Graz, Oktober 2011. – Vorläufiger interner für Bautechnik, 16. Dezember 2009. – Allgemeine
Forschungsbericht bauaufsichtliche Zulassung Z-9.1-435

[Pirnbacher and Schickhofer 2009] Pirnbacher, [Z-9.1-519 2011] SPAX International GmbH &
Gernot ; Schickhofer, Gerhard: Base parameters Co. KG: Spax-S Schrauben mit Vollgewinde als
of self-tapping screws. In: Proceedings of CIB-W18. Holz-Verbindungsmittel. Berlin : Deutsches Insti-
Dübendorf, Schweiz, August 2009. – Paper 42-7-1 tut für Bautechnik, 11. März 2011. – Allgemeine
bauaufsichtliche Zulassung Z-9.1-519
[Pirnbacher and Schickhofer 2012] Pirnbacher,
Gernot ; Schickhofer, Gerhard: Zeitabhängige [Z-9.1-519 2012] SPAX International GmbH &
Entwicklung der Traglast und des Kriechverhal- Co. KG: Spax-S Schrauben mit Vollgewinde als
tens von axial beanspruchten, selbstbohrenden Holz-Verbindungsmittel. Berlin : Deutsches Insti-
Holzschrauben / holz.bau forschung gmbh. 2012. tut für Bautechnik, 31. Januar 2012. – Allgemeine
– research report bauaufsichtliche Zulassung Z-9.1-519

[Plieschounig 2010] Plieschounig, Stefan: [Z-9.1-656 2007] Schmid Schrauben Hainfeld

Ausziehverhalten axial beanspruchter Schrauben- GmbH: Holzbauschrauben Star Drive Vollgewinde
gruppen. Graz, Institut für Holzbau und Holztech- und Star Drive 2 Gewinde als Holzverbindungsmit-
nologie der TU Graz, diploma thesis, November tel. Berlin : Deutsches Institut für Bautechnik,
2010 25. Juni 2007. – Allgemeine bauaufsichtliche Zulas-
sung Z-9.1-656
[Pollmeier 2012] Pollmeier Massivholz GmbH
& Co. KG: Furnierschichtholz aus Buche: [Z-9.1-679 2011] Studiengemeinschaft Holzle-
Datenblatt/Statische Kennwerte. Creuzburg, imbau e. V.: BS-Holz aus Buche und BS-Holz
21. Dezember 2012. – research report Buche-Hybridträger. Berlin : Deutsches Institut

11
 für Bautechnik, 7. Juni 2011. – Allgemeine bauauf-
sichtliche Zulassung Z-9.1-679

[Z-9.1-704 2012] Elaborados y Fabricados

Gámiz S. A.: VIGAM Brettschichtholz aus Eiche.
Berlin : Deutsches Institut für Bautechnik, 04. Mai
2012. – Allgemeine bauaufsichtliche Zulassung Z-
9.1-704

[Z-9.1-777 2010] SFS intec GmbH:

Gewindestangen mit Holzgewinde als
Holzverbindungsmittel. Berlin : Deutsches In-
stitut für Bautechnik, 30. November 2010. –
Allgemeine bauaufsichtliche Zulassung Z-9.1-777

12