Design approaches for dowel-type connections in CLT structures

and their verification

Andreas Ringhofer *)
Institute of Timber Engineering and Wood Technology
Graz University of Technology, Austria

Reinhard Brandner *)
Institute of Timber Engineering and Wood Technology
Graz University of Technology, Austria

Hans Joachim Blass

Karlsruhe Institute of Technology
Germany

*) Joint first authorship

This contribution has already been published in the proceedings of the

International Conference on Connections in Timber Engineering –
From Research to Standards, COST FP1402, Graz, Austria.

Summary
Within the last 20 years, cross laminated timber (CLT) has become one of the most
important building products in modern timber engineering. By the end of this decade
its annual worldwide production volume is expected to exceed 1,000,000 m³. There
is a strong interest of the industry, engineers, architects and contractors to implement
CLT in European product, design and execution standards. This is part of the cur-
rently ongoing revision of Eurocode 5, supported by COST Action FP1402. In this
context and in addition to the ULS and SLS verification of the panels themselves,
provisions regarding the design of connections in CLT composed by dowel-type fas-
teners are of utmost importance.
Within this contribution, we aim on collecting, discussing and validating related ap-
proaches for characteristic values in literature for connections and single dowel-type
fasteners in CLT. In addition, these models – especially dedicated to the withdrawal
and embedment strength of dowels, nails and self-tapping screws – are compared
with current regulations on dowel-type fasteners for solid timber and glulam as given
in Eurocode 5. These comparisons are made in order to identify a pending need of
modification of current Eurocode 5 equations and state-of-the-art regulations. Re-
garding the connection design, minimum spacing, edge and end distances as well as
additional geometrical conditions, regulations on the effective number of fasteners
in a group and minimum penetration depths are summarized. Finally, conclusions
with respect to the single fastener properties, withdrawal and embedment strength,

161
are made together with comments on regulations ensuring the integrity of CLT struc-
tures. Overall, we aim on presenting a compilation of the current state-of-the-art
knowledge on dowel-type fasteners in CLT as basis for implementing design provi-
sions for CLT in the new connections chapter of Eurocode 5.

1. Introduction
Cross laminated timber (CLT) is a planar, large dimension engineered timber prod-
uct, designed for structural purposes and capable of bearing loads in- and out-of-
plane. CLT, with dimensions tCLT ×wCLT × ℓCLT, is commonly composed of an uneven
number N of orthogonal layers (tℓ × wℓ) of finger-jointed laminations or wood-based
panels. Adjacent laminations within the same layer may also feature narrow face
bonding; without narrow face bonding gaps between the laminations may occur. The
orthogonal layers are typically bonded at their side face forming rigid composite el-
ements; flexible composites featuring layers connected by nails, staples or other fas-
teners are also on the market but not focused on within this contribution. Current
approvals of European CLT products allow gaps in width (wgap) up to 6 mm. Some
CLT products feature also laminations with one or more stress reliefs, usually
2.5 mm wide; see Fig. 1 and EN 16351 [1].

gap relief
tCLT
tℓ

wℓ wgap

Fig. 1. (left) Principal CLT layup and some definitions of geometry and execution;
(right) typical five-layer CLT element.

CLT has been used in manifold applications but primary as large dimension wall,
floor and roof element in single and multi-story buildings, halls and bridges. Alt-
hough only 20 years on the market, this product has been changing the timber engi-
neering sector in many ways, e.g.:
 by allowing architects thinking rather in planes and volumes than in lines;
 by supporting the timber engineering sector with a stand-alone structural
element enabling assembling of constructions amazingly fast, dry, clean and
with high precision;
 by providing all subsequent crafts conditions for easy and fast mounting and
manipulation, e.g. of additional layers (e.g. insulation as well as installation
layer and façade) and building services in general; and
 by offering end-users and investors highest building quality and a sustainable,
natural living and working environment.

162
CLT has been revolutionizing the timber engineering sector in analogy to the inven-
tion of particle boards in furniture industry which at that time initiated also big
changes. There are several analogies between CLT and particle boards e.g. the ho-
mogeneity of both products in comparison with the base material, i.e. a reduced var-
iation in physical / mechanical properties. Both products are planar, feature a
significantly reduced swelling and shrinkage in-plane and utilize base material (qual-
ity) which would be hard to use otherwise. Their industrial production processes are
relatively simple and the final products can be relatively easy manipulated and as-
sembled.
Apart from all these mutualities, differences in respect to scale, reliability and safety
requirements, service life and exposure shall be considered as well.
At the time the particle board entered the market also a significant change in furniture
design could be observed, again by thinking in planes and volumes (boxes) rather
than in lines and frames. New connection and fastener solutions, which were opti-
mized for particle boards, together with a high degree in standardizing products and
processes enabled extensive industrialization and economically prized furniture pro-
duction.
CLT is on the way to catch up also these two very important steps: standardization
(e.g. of layer thicknesses (tℓ = 20, 30, 40 mm), layups, base material quality and de-
sign approaches) and by developing a connection technique optimized for CLT; in
respect to the latter, there is still much room for further developments and improve-
ments. Aiming on versatile applicable connection solutions in CLT structures, a first
step could be differentiating into principal connection lines (see also Fig. 2).
Considering this and looking at structures as a whole, the possibilities in realizing
integral CLT structures also depend on the principle construction system. Single fam-
ily houses, residential, office and school buildings up to three to five stories are com-
monly erected as platform-frame systems (indirect vertical load transfer between
wall elements of vertically adjacent stories via soft floor elements).
Higher or heavily-loaded buildings are commonly designed as balloon-frame sys-
tems (direct vertical load transfer between wall elements of vertically adjacent sto-
ries).
Although CLT is currently used in superstructures hardly possible in timber one dec-
ade ago, a connection technique underlining the possibilities building with CLT is
still widely missing. In fact, current commonly applied fastener and connection so-
lutions, like angle brackets and hold-downs, are borrowed from light-weight timber
constructions; these connectors together with a wall-floor-wall connection in a typi-
cal platform-frame CLT structure are shown in Fig. 3. In optimizing angle brackets
for CLT structures a first step could be adapting the geometry to resist both, shear
and uplift forces, consolidating the tasks of current angle brackets (shear load trans-
mission) and hold-downs (transmission of uplift forces) in one connector (Flatscher
and Schickhofer [3]).

163
 (i) wall-to-foundation
(ii) wall-to-wall
(iii) floor-to-floor
(iv) wall-to-floor

Fig. 2. Definition of connections exemplarily for a CLT platform-frame structure

(Brandner et al. 2016 [2]; adapted).

Fig. 3. (a) wall-foundation connection with hold-down; (b) wall-floor-wall

connection with angle brackets and partially-threaded self-tapping screws; (c)
wall-wall edge-connection with partially-threaded self-tapping screws, (d) wall-
wall connection with double-threaded self-tapping screws; Schickhofer et al.
(2010) [4].

Apart from angle brackets, hold-downs and line connections realized by means of
partially, fully- or double-threaded self-tapping screws (see Fig. 3), for CLT struc-
tures also other connection solutions and first CLT system connectors are available;
e.g. solutions with (self-drilling) smooth dowels / tight-fitting bolts and inner metal
plates (e.g. Bernasconi [5]), the system connectors X-RAD (Polastri et al. [6], [7];
http://www.rothoblaas.com/products/fastening/x-rad 2017-07-07) or SHERPA CLT
Connector (Kraler et al. [8]; http://de.sherpa-connector.com/clt_connector 2017-07-
07), as well as special connection solutions, e.g. embedded steel tubes in combination
with glued- or screwed-in rods (Schickhofer et al. 2010 [4]); see Fig. 4.
All these connection solutions have in common a connection between metal elements
and CLT achieved via dowel-type fasteners, e.g. profiled nails (annular ringed shank
nails or helically threaded nails), fully-, partially- or double-threaded self-tapping
screws, smooth dowels or tight-fitting bolts as well as screwed- or glued-in rods,
which are either loaded axially, laterally or combined. Retailers like Rothoblaas,

164
Simpson Strong Tie, etc. support engineers with comprehensive (software) design
tools and tabularized characteristic performance-based connection properties; a de-
tailed verification of the single dowel-type fasteners themselves – fixing angle brack-
ets, hold-downs or system connectors to CLT – is thus not required. However, the
anchorage potential of single fasteners in CLT as described e.g. by the embedment
strength and the withdrawal strength are essential parameters for the development of
CLT connectors and for verification of line connections realized with self-tapping
screws or (self-drilling) dowels as well as individual connection solutions.

Fig. 4. (a) X-RAD system connector for balloon-frame CLT structures (Rothoblaas
© [9]); (b) SHERPA CLT-connector (Sherpa © [10]); (c) embedded steel tube and
glued-in rods (Schickhofer et al. 2010 [4]).

Connections decisively impact the behavior of CLT structures. Whereas in light-

weight timber structures wall and floor diaphragms behave more flexible, CLT dia-
phragms (walls and floors) are characterized by high stiffness and high resistance in
shear, tension and compression in-plane. Consequently, in CLT structures ductility
and energy dissipation have to be provided by adequate connection solutions, the
contribution from CLT itself is negligible. Considering the main requirements on
connections, as they are high resistance, high stiffness and high ductility, the signif-
icant difference in the behavior of light-frame and CLT diaphragms is necessitating
new ways of thinking in developing adequate connection solutions for CLT.
In view of ongoing harmonization and standardization processes and the upcoming
new version of the European Timber Design Code, Eurocode 5 (series of standards
EN 1995-x-x), it is intended to reorganize the chapter on connections by providing a
tool-box, supporting the engineer step-by-step with required basic characteristic val-
ues for different anchoring materials. A prerequisite therefore are generic approaches
instead of models limited to each individual timber product.
In view of that, the next chapters are organized by providing at first some general
notes on dowel-type fastener behavior in CLT; what is special and what can be
treated equally to solid timber and glulam. In that respect potential failure modes are
listed and discussed in conjunction with CLT. Following this, we summarize and
discuss the state-of-the-art of axially and laterally loaded dowel type fasteners an-
chored in CLT, emphasizing the withdrawal and embedment behavior, respectively.

165
Some suggestions for harmonization of current regulations on solid timber and glu-
lam with CLT are made. Finally, we summarize also current suggestions and regula-
tions on the connection design and requirements for securing structural integrity
before summary and concluding remarks are given.

2. CLT characteristics common with glulam and solid timber

Fig. 5. Principal CLT element: terms, dimensions, coordinate system.

CLT as well as glued laminated timber (glulam; GLT) are composed of solid timber
laminations, typically of board dimension. Similarities in design provisions between
solid timber (ST), CLT and GLT are expected, but the orthogonal layup of CLT de-
mands additional attention. At first, differentiation in the positioning of fasteners in
CLT side face (fastener axis oriented out-of-plane) and narrow face (fastener axis
oriented in-plane) is necessary; see Fig. 5 (we will see later in a proposed generic
approach for withdrawal properties that this differentiation would also make sense
for glulam and other laminated products). Furthermore, the influence of gaps and
potential stress reliefs has to be taken into account as placement of fasteners in gaps,
in particular if inserted in the narrow face and parallel to the grain, may lead to sig-
nificant losses in fastener performance. According to approvals of current European
CLT products, in the outer (top) layers gap widths up to 4 mm are allowed whereas
in the inner (core) layers the gap width is limited to 6 mm. The European product
standard for CLT, EN 16351 [1], limits gap widths generally to 6 mm. CLT produc-
ers aim on minimizing gap widths which is reflected already in Blaß and Uibel [11],
Uibel and Blaß [12, 13] where inner layers featured a mean and 95 %-quantile gap
width of 0.6 to 2.0 mm and 1.8 to 4.5 mm, respectively, whereas the 95 %-quantile
in the outer layers was only 1.0 to 1.6 mm.
Fasteners inserted via the CLT side face and loaded laterally behave rather ductile,
facilitating that all fasteners in a group of fasteners contribute to the load-bearing
capacity to their full extent, i.e. the effective number nef is equal to n, the number of
fasteners inserted; see also Fig. 6. This ductile behavior is due to the orthogonal layup
where transverse layers act as reinforcement and prevent early failures in tension
perpendicular to grain and block or row shear.

166
Apart from a generally observed minor influence of gaps (Blaß and Uibel [11] and
Uibel and Blaß [12, 13]), fasteners inserted via the CLT side face can be designed
similar to solid timber and glulam. Considering also that typical insertion angles are
(30°) 45° to 90°, the influence of the angle between load and grain orientation of
outer layers in case of laterally loaded fasteners, as well as the thread-grain angle
in case of axially loaded fasteners (see Fig. 7), is small and with some conservatism
even negligible; more on that later.

cross laminated timber

load [kN]

(CLT)
1 x smooth dowel d = 30 mm
glulam (GLT)
1 x smooth dowel d = 30 mm

0 10 20 30 40
displacement [mm]
Fig. 6. Load-displacement behaviour of laterally loaded smooth dowels in the CLT
side face (Schickhofer et al. 2010 [4]; adapted).

To reduce the influence of gaps on fastener performance, i.e. to reduce the probability
that the effective anchorage length of a fastener is solely placed in gaps, it is sug-
gested to penetrate a minimum of three layers (M = 3, with M as the number of pen-
etrated CLT layers); see Blaß and Uibel [11].
By anchoring fasteners in the CLT narrow face, the possibility of penetrating layers
with grain oriented parallel to the fastener axis has to be taken into account. This
circumstance influences also the probability that a fastener is inserted in a gap, i.e.
the influence of gaps on the fastener performance in the narrow face is higher than
in the side face. Furthermore, in a group of fasteners or even for one single fastener
different thread-grain angles are possible; see Fig. 8.
Laterally loaded fasteners in the CLT narrow face can either be loaded in-plane, i.e.
by being loaded parallel to the CLT side face, or out-of-plane, i.e. by being loaded
perpendicular to the CLT side face. In the latter case the possibility of tension per-
pendicular to grain failures, before reaching the capacity of fasteners, shall be taken
into account.
In general, when discussing the performance of a connection composed by laterally
or axially loaded dowel-type fasteners, the following potential failure mechanism
can be differentiated; see Table 1.

167
 Fax Fax
0° < α < 90°
0° ≤ α ≤ 90° 0° ≤ β < 90°
0° ≤ β < 90° Flat Flat

0° ≤ α ≤ 90°
0° < β ≤ 90°

Fax
Flat
Fig. 7. Definition of thread-grain angle and the angle between load and grain
orientation of outer layers , exemplarily for a screw inserted in side and narrow
face of CLT.

Fig. 8. (left) possible positions of dowel-type fasteners oriented perpendicular to

the grain of outer layers in the CLT narrow-face; (middle) group of axially loaded
self-tapping screws in the CLT narrow face, and (right) single screw featuring
different thread-grain angles (Brandner [14]; adapted); ST = solid timber.

Table 1. Summary of potential failure mechanism in case of axially or laterally

loaded single or group of fasteners
Loading Failure mode Failing material
Axial Withdrawal Timber
Head pull-through Timber
Fastener tension Steel
Tension perpendicular to grain (splitting) Timber
Block shear and row shear Timber
Lateral Embedment Timber
Yield in bending Steel
Block shear, row shear, plug shear Timber
Tension perpendicular to grain (splitting) Timber

168
As failures of the fastener itself (steel failures) do not depend on the applied timber
products we focus further on failure modes of CLT surrounding the fastener, in par-
ticular on the embedment and withdrawal capacity, as the related properties are of
major importance for the description of laterally and axially loaded fasteners, respec-
tively.
The density as indicating property for fastener performance in timber has to be dif-
ferentiated between side and narrow face insertion. For fasteners in the side face,
penetrating several layers, the characteristic density of CLT is proposed; see Blaß
and Uibel [11] and Uibel and Blaß [13] who defined the characteristic density of
CLT, based on laminations with strength class C24 and a characteristic density of
12,ℓ,k = 350 kg/m³ according to EN 338 [15], with 12,CLT,k = 400 kg/m³. Following
the proposal of PT SC5.T1 [16] for the new version of EN 1995-1-1 [17], for CLT
made of C24 or T14 laminations according to EN 338 [15] a value of 12,CLT,k =
385 kg/m³, based on the relationship 2,CLT,k = 1.1 ∙ 12,ℓ,k, analogously to glulam, is
proposed. For adjustment of models for withdrawal and embedment properties to
density, corresponding corrections are presented. As fasteners inserted via the CLT
narrow face anchor mostly only in one lamination, for calculation of embedment and
withdrawal capacity the characteristic density of the laminations themselves applies,
e.g. for C24 or T14 according to EN 338 [15]: 12,ℓ,k = 350 kg/m³.

3. Withdrawal strength of axially loaded dowel-type fasteners in CLT

3.1 Definitions, general comments and overview

Within this section, we focus on threaded or profiled dowel-type fasteners, enabling
a composite action with the surrounding timber material and thus an appropriate per-
formance when loaded in axial direction, i.e. self-tapping screws and profiled nails.
Note: glued-in rods or glued-in metal plates, which also show a reasonable perfor-
mance in case of axial loading, are not discussed in this paper. The composite action
between the axially loaded fastener and the timber component is expressed by the
withdrawal strength fax. In contrast to the withdrawal parameter, f1 = fax , the with-
drawal strength fax is defined as
Fax,max
fax  (1)
 ef  d  
with Fax,max as the fastener’s capacity in axial direction (equal to the withdrawal re-
sistance Fax,,Rk according to Eurocode 5 [17]), d as its relevant (nominal) diameter
and ℓef as its effective inserted thread length; the latter either including the fastener’s
tip length or not – as it is discussed later on.
Blaß and Uibel [11] (also published in Uibel and Blaß, [12, 13, 18]) were the first
who investigated the performance of dowel-type fasteners, situated in different posi-
tions in the side and narrow face of CLT. Their analyses comprised variations in
fastener type (dowels, self-tapping screws and nails), both angles and , different
layups of Central European CLT (also including solid wood-based panels,
169
N = 3 ÷ 7), randomly distributed gap widths wgap,mean up to 2.0 mm and different
nominal diameters d. With regard to single fastener properties, outcomes were used
for determining empirical regression functions as well as characteristic approaches
for withdrawal and embedment strength as basis for the design of dowel-type fasten-
ers inserted in CLT.
In respect to self-tapping screws and their axial load-bearing behavior in CLT, within
the last years further comprehensive studies were conducted at Graz University of
Technology, c.f. Ringhofer [19]. Amongst other topics, several experimental cam-
paigns are compiled dealing with a detailed analysis of selected parameters relevant
for the screws’ load-bearing behavior. This especially concerns the following param-
eters, the number of penetrated layers M (Reichelt [20]; Ringhofer et al. [21]), the
thread-grain angle and the gap configuration (gap type and width; see Grabner
[22]; Brandner [14]; Brandner et al. [23] and Silva et al. [24]). Based on a detailed
statistical analysis on available comprehensive data on screw withdrawal tests an
empirical, generic approach, presented in Ringhofer et al. [25]) and Ringhofer [19],
was established and successfully validated by means of independently derived data
sets.
In the following Sections 3.2 and 3.3 we summarize the main findings and recom-
mendations of both research programs dedicated to the withdrawal behavior of
dowel-type fasteners as carried out at KIT and Graz University of Technology. Fur-
thermore, we compare related models with those currently given in Eurocode 5 [17]
for determining the characteristic withdrawal strength particularly of axially loaded
self-tapping screws situated in solid timber or glulam.
Apart from the European research programs also the work done by Kennedy et al.
[26] is worth being mentioned since it represents one of the few non-European re-
search activities regarding the determination of mechanical properties of single fas-
teners situated in (the side face of) CLT elements.

3.2 Axially loaded self-tapping screws in CLT elements

The experimental program presented in Blaß and Uibel [11] and dedicated to the
withdrawal strength of axially loaded self-tapping single screws with outer thread
diameters d = {6, 8, 12} mm can be separated into two campaigns: one where the
focus was on CLT side face insertion and one where the screws were situated in the
CLT narrow face. The impact of gaps between laminations of one layer on the me-
chanical screw performance was investigated by four alternative possibilities of
screw insertion, covering the number of penetrated gaps Ngap = {0, 1, 2, 3} (CLT side
face) as well as the gap type (CLT narrow face, insertion in T- and butt-joints, c.f.
Fig. 8) while the mean gap width wgap,mean varied randomly between 0.2 and 2.0 mm.
For screws situated in the CLT side face, this was exclusively done for perpendicular-
to-grain insertion while for those situated in the CLT narrow face, both limits of
screw axis position to the single layer’s grain direction, = {0, 90}° were examined.
Specimens made of 3- and 5-layered CLT panels with five different layups were used
for 387 withdrawal tests. Subsequently, empirical regression functions for predicting

170
the withdrawal strength fax were determined based on the test results; the character-
istic approach is presented in Eq. (2):
0.75
9.02  d 0.2   ef0.1  
fax,k  and fax,k,corr  f ax,k   k  (2)
1.35  cos 2  sin 2  350 
and as primary insertion angle, e.g. for screws in CLT narrow face: = 0° (on a
conservative but general basis), and for screws in CLT side face = 90°. As given in
Eq. (2), for this approach a multiplicative power model is applied. Apart from , it
identifies the outer thread diameter, the timber density as well as the effective in-
serted thread length as influencing parameters. While influences from diameter and
density were adapted to the test results (both have a disproportional influence on fax
respectively), Blaß and Uibel [11] adopted the impact of ℓef on fax from the results of
a previous testing campaign conducted in solid timber, c.f. Blaß et al. [27] and
Eq. (4). Note: with regard to the definition of fax given in Eq. (1), it is worth men-
tioning that Blaß and Uibel [11] determined their experimental withdrawal strengths
with ℓef including the tip length, thus leading to a conservative interpretation of test
results.
Due to material homogenization, the smaller probability of gap insertion along ℓef as
well as the fact that parallel-to-grain insertion is impossible, the withdrawal strength
of screws situated in the CLT side face is of course significantly higher than of screws
situated in the CLT narrow face. This is expressed by the factor 1.35 in Eq. (2),
whose magnitude results from a data fit to those CLT narrow face test series with the
worst results for withdrawal strength (parallel-to-grain insertion in gaps).
The examinations carried out at Graz University of Technology base on similar pa-
rameter configuration as in Blaß and Uibel [11] but additionally comprised variations
in thread-grain angle within the limits {0, 90}° for CLT side and narrow face in-
sertion, as well as systematic examinations of the number of penetrated layers’ (and
their orientation) and the gap width’s (for the types shown in Fig. 8). Amongst others,
corresponding outcomes were applied for deriving a new generic approach for deter-
mining the withdrawal capacity of axially loaded self-tapping screws as presented in
Ringhofer et al. [25] and Ringhofer [19] and shown in Eqs. (3-5):
kρ
0.33  
f ax,k  kax,k  ksys,k  8.67  d and f ax,k,corr  f ax,k   k  (3)
 350 

1.00 for 45    90 1.00 ST

 
k ax,k  1  0.64  k gap,k , ksys,k  1.10 CLT, N  3 (4)
 0.64  k gap,k    for 0     45 1.13 GLT, N  5
 45 

0.90 CLT narrow face 1.10 0    90

kgap,k   , k   (5)
1.00 else 1.25  0.05  d   0

171
This new generic approach comprises the following features, which are different
from current approaches:
 Firstly, density correction by the power factor k is kept more flexible;
in-depth analyses identified thread-grain angle and outer thread diameter as
important influencing parameters.
 Secondly, for screws penetrating more than one layer when applied in lami-
nated timber products a significant homogenization was found. Apart from the
density, so far the only parameter indicating the anchorage material, a stochas-
tically determined system factor ksys is introduced, see Eq. (4). It allows
adjusting the withdrawal capacity related to the base material density, i.e. the
density of the laminations, according to the screw application, i.e. it increases
the withdrawal capacity the more layers are penetrated by the screw. This
circumstance neither can be covered by inserting the product density instead
of the base material density nor would this procedure be meaningful. This
commitment – aimed to cover screw application in laminated timber products
in general – reflects the generic character of this approach as opportunity to
decrease the number of models for different applications without a loss of
accuracy due to simplification.
 Thirdly, the systematic variation of gap width and type, especially in case of
CLT narrow face insertion, enabled the determination of a probabilistic model
quantitatively and explicitly describing the related negative impact on with-
drawal strength fax, c.f. Brandner [14]. Considering currently given limits with
regard to gaps and stress reliefs, c.f. Chapter 1, and further assumptions (e.g.
the variabilities of density and withdrawal strength), a corresponding simplifi-
cation in form of the multiplicative factor kgap is applied; see Eq. (5).
Besides the explained differences, the impact of both remaining influencing param-
eters, the outer thread diameter and thread-grain angle, is treated in a traditional way,
i.e. diameter adjustment by a negative power parameter and a bilinear model with
discontinuity at = 45° for considering the influence of the thread-grain angle. In
line with Blaß and Uibel [11], the placement of screws in the side face of CLT ele-
ments leads to significantly higher withdrawal strengths than that in their narrow
face. Deviating from Blaß and Uibel [11], Eq. (3) does not contain ℓef as influencing
parameter since no related impact on fax could be observed (Note: This is only valid
if tip length is not part of ℓef).

172
We now compare quantitatively the approaches given in Eq. (2) and Eq. (3), with
special focus on screw insertion in CLT elements, with the current model for deter-
mining the characteristic withdrawal strength according to Eurocode 5 [17]; see
Eq. (6), which is based on Blaß et al. [27] and 1212 withdrawal tests on axially loaded
screws inserted in solid timber at different outer thread diameters, effective insertion
lengths (counted by neglecting the tip length) and axis-to-grain angles.

18.0  d 0.5   ef0.1  kd  

0.80
 d
f ax,k  with f ax,k,corr  f ax,k   k  and k d  min  8 (6)
1.2  cos 2  sin 2  350   1
In fact, the approach in Eq. (6) is very similar to Eq. (2) (the impact of d is more
pronounced while the density is considered in a quite similar way). In comparison to
Eq. (4), the influence of the thread-grain angle is modelled according to Hankinson
[28] instead of a bi-linear approach, and there is the additional coefficient kd, which
decreases fax in case of d < 8 mm, which, together with the limitation to thread-grain
angles ≥ 30° in Eurocode 5 [17] is not part of Blaß et al. [27]. Since the product
CLT is not covered by the current version of Eurocode 5 [17], Eq. (6) does not spe-
cifically consider a related application. Nevertheless, selected parameter character-
istics enable a reasonable model comparison, which is illustrated in Fig. 9 to Fig. 11
in dependence of representative characteristic densities (which means single lamella
strength classes; for CLT side face insertion, the relationship 12,CLT,k = 1.1 ∙ 12,ℓ,k,
was applied) and outer thread diameters. Furthermore, ℓef was constantly set to 10 ∙ d,
as a related specification is necessary when applying Eq. (2) and Eq. (6).

C18 | T11 C24 | T14 C30 | T18

ρℓ,k = 320 kg/m³ ρℓ,k = 350 kg/m³ ρℓ,k = 380 kg/m³
ρCLT,k = 352 kg/m³ ρCLT,k = 385 kg/m³ ρCLT,k = 418 kg/m³
1.4
1.2 EC 5 [17]
fax,k,i / fax,k,ref [-]

1.0
0.8 Blaß and
0.6 Uibel [11]
0.4 Ringhofer
0.2 et al. [25]
0.0
6 8 10 12 6 8 10 12 6 8 10 12
d [mm]

Fig. 9. Comparison of characteristic withdrawal strength according to Eurocode 5

[17], Blaß and Uibel [11] and Ringhofer et al. [25] for varying outer thread
diameters of self-tapping screws related to that of Eurocode 5 [17] for d = 8 mm;
CLT side face and perpendicular-to-grain insertion (α = 90°).

173
 C18 | T11 C24 | T14 C30 | T18
ρℓ,k = 320 kg/m³ ρℓ,k = 350 kg/m³ ρℓ,k = 380 kg/m³
1.4
fax,k,i / fax,k,ref [-]
1.2 EC 5 [17]
1.0
0.8 Blaß and
0.6 Uibel [11]
0.4 Ringhofer
0.2 et al. [25]
0.0
6 8 10 12 6 8 10 12 6 8 10 12
d [mm]

Fig. 10. Comparison of characteristic withdrawal strength according to

Eurocode 5 [17], Blaß and Uibel [11] and Ringhofer et al. [25] for varying outer
thread diameters of self-tapping screws related to that of Eurocode 5 [17] for
d = 8 mm; CLT narrow face and perpendicular-to-grain insertion (α = 90°).
With regard to the comparison of characteristic withdrawal strengths for perpendic-
ular-to-grain insertions, once through several layers (CLT side face, Fig. 9) and once
into one layer (CLT narrow face, Fig. 10), two points are worth being discussed in
detail:
Firstly, even though input parameter treatment differs in the magnitude of the power
values to some extent, characteristic withdrawal strengths determined by Eq. (2) and
Eq. (6) result in a nearly equal value if practically relevant outer thread diameters
d = {8, 10, 12} mm are considered.
Secondly, the new approach presented in Eq. (3) leads to remarkably higher values
of fax,k if compared to the two others. Furthermore, this is independent from timber
density and outer thread diameter and – in case of Fig. 9 – it confirms the aforemen-
tioned statement of increased withdrawal strength due to the application of ksys ac-
cording to Eq. (4).
In Fig. 11, which displays the parallel-to-grain insertion in the CLT narrow face and
thus only comprises approaches which take the impact of gaps on fax into account,
the situation is different: here, the generic approach presented in Eq. (3) results in
significantly smaller withdrawal strengths than those determined by Eq. (2), irrespec-
tive of conducted parameter variations. This conservative estimation can be ex-
plained by a less pronounced relationship between density and withdrawal strength
in case of = 0° combined with a higher difference of fax between perpendicular-
and parallel-to-grain insertion, c.f. Eq. (4) and Eq. (5).

174
 C18 | T11 C24 | T14 C30 | T18
ρℓ,k = 320 kg/m³ ρℓ,k = 350 kg/m³ ρℓ,k = 380 kg/m³
1.4
1.2
fax,k,i / fax,k,ref [-]
Blaß and
1.0
Uibel [11]
0.8
0.6 Ringhofer
0.4 et al. [25]
0.2
0.0
6 8 10 12 6 8 10 12 6 8 10 12
d [mm]

Fig. 11. Comparison of characteristic withdrawal strength according to Blaß and

Uibel [11] and Ringhofer et al. [25] for varying outer thread diameters of self-
tapping screws related to that of Blaß and Uibel [11] for d = 8 mm; CLT narrow
face and parallel-to-grain insertion.

3.3 Axially loaded profiled nails in CLT elements

With regard to the performance of axially loaded, profiled nails situated in the side
and narrow face of CLT elements, examinations comparable to those for self-tapping
screws were not conducted at Graz University of Technology. Thus, the study pre-
sented in Blaß and Uibel [11] is the exclusive source for withdrawal properties of
nails in CLT discussed in this section. Since this program did not distinguish between
self-tapping screws and nails, it is referred to Section 3.2 for a related explanation.
The sole deviations to self-tapping screws are of course the smaller nominal diame-
ters d = {3.1, 4.0, 6.0} mm of the nails (loadbearing class III, according to DIN 1052,
1988) as well as the higher number of tests (523) carried out in three- and five-lay-
ered CLT panels with in total five different layups. The characteristic approach pro-
posed by Blaß and Uibel [11] again bases on an empirical regression model adjusted
to the test results and is shown in Eq. (7):
0.80
4  d 0.4  
f ax,k  with fax,k,corr  f ax,k   k  (7)
3.33  cos 2  sin 2  350 
and as primary insertion angle, e.g. for nails in CLT narrow face: = 0° (on a con-
servative but general basis), and for nails in CLT side face = 90°. For CLT featuring
gaps and/or stress reliefs, Blaß and Uibel [37] recommend to reduce the withdrawal
capacity according to Eq. (7) to 80 % for nails with d < 6 mm and to apply only nails
with d ≥ 4 mm and ℓef ≥ 8 ∙ d. Apart from the pre-factor 3.33, which is significantly
higher than that for self-tapping screws in Eq. (2) – possibly caused by smaller nom-
inal diameters while the average gap width was kept constant – the considered influ-
encing parameters as well as their treatment are similar to those given in Eq. (2).

175
Fig. 12 compares characteristic withdrawal strengths of profiled nails determined
according to Eq. (7) in dependence of the nominal diameter, the single layer’s char-
acteristic density as well as the position in the CLT panel (side vs. narrow face). It
again highlights the significant difference of withdrawal strength in dependence of
the nail location, which is not only caused by the aforementioned pre-factor but also
by applying a higher characteristic (product) density 12,CLT,k = 1.1 ∙ 12,ℓ,k, for CLT
side face if compared to narrow face insertion.

4.50
4.00 d = 3 mm
3.50 d = 4 mm
d = 5 mm
fax,k [N/mm²]

3.00
d = 6 mm
2.50
2.00 side face
1.50 narrow face
1.00
0.50
0.00
300 350 400 450 500 550
ρℓ,k [kg/m³]

Fig. 12. Characteristic withdrawal strength vs. characteristic density ℓ,k of

profiled nails: comparison between CLT side and narrow face insertion in
dependence of the nominal nail diameter; according to Blaß and Uibel [11].

In contrast to self-tapping screws or smooth shank nails, the currently valid version
of Eurocode 5 [17] refers to design values published in the manufacturers’ declara-
tions of performance (DoP) instead of providing a product-independent approach for
determining the characteristic withdrawal strength of profiled nails. This can be ex-
plained by significant differences in withdrawal strength of nails from different pro-
ducers, which are higher than the ones caused by a variation of common influencing
parameters such as the timber density or the nominal diameter [29]. In this recently
published source, it is reported that the related variability disabled the derivation of
a reasonable generic approach. With regard to the average withdrawal strength of
profiled nails, Sandhaas and Görlacher [29] determined a nonlinear, empirical re-
gression model for estimating this property, which is based on a comprehensive test
database available at KIT and given in Eq. (8), see:
fax,k  3.60  103  1.38
(8)
with as the density of solid timber as material applied for the tests. According to
Sandhaas and Görlacher [29], this approach has a rather limited predictive quality
due to a poor correlation between density and withdrawal strength. Nevertheless, it
represents the average declared withdrawal strength of profiled nails in solid timber
and shall be applied for a comparison with the one for CLT as published by Blaß and
Uibel [11], see Eq. (9).

176
 0.16  d 0.4  0.8
f ax  (9)
3.1  cos 2  sin 2
This comparison is subsequently illustrated in Fig. 13 in dependence of the nominal
nail diameter and the layer density at = 90° and identifies remarkably higher values
of fax determined by Eq. (8), especially for average timber densities of common soft-
wood strength classes (C24 and above) according to EN 338 [15].

8.00
7.00 d = 3 mm
6.00 d = 4 mm
fax [N/mm²]

5.00 d = 5 mm
4.00 d = 6 mm
3.00 Blaß and Uibel [11]
2.00 Sandhaas and
1.00 Görlacher [29]
0.00
300 350 400 450 500 550
ρℓ [kg/m³]

Fig. 13. Comparison of withdrawal strength of profiled nails estimated by Eq. (7)
with the approach published in Sandhaas and Görlacher [29] (Eq. (8)) in
dependence of nominal diameter and layer density; = 90°.

4. Embedment strength of laterally loaded dowel-type fasteners in CLT

4.1 General comments and overview

Blaß and Uibel [11] and Uibel and Blaß [12, 13, 18] were the first investigating lat-
erally-loaded dowel-type fasteners in CLT side and narrow faces of Central Euro-
pean CLT and solid wood-based panels. More recently, Kennedy et al. [26] report
on investigations conducted in North America on laterally-loaded threaded-fasteners
inserted in CLT side face.
In the following we summarize the main findings from the comprehensive report of
Blaß and Uibel [11] as introduced in Section 3.1 and compare the proposed relation-
ships with current design provisions for solid timber and glulam as given in Euro-
code 5 [17].
In general, the embedment strength represents a system property, the resistance of
timber against laterally loaded fasteners. In cases of fasteners penetrating several
layers in CLT side face or two or more different laminations in the CLT narrow face,
featuring different angles between fastener axis and load, this kind of system prop-
erty even comprises different contributions of layers and laminations as different
load-grain angles may be concerned.

177
4.2 Laterally loaded dowel-type fasteners in CLT side face
4.2.1 Smooth dowels and tight-fitting bolts
Blaß and Uibel [11] tested smooth dowels with diameters d = 8 to 24 mm in three-
and five-layer CLT elements, positioned apart or in gaps involving one to three lay-
ers, and loaded at 0°, 45° and 90° in respect to the outer layer’s grain orientation.
Thereby, a minor influence of the number of gaps on the embedment strength was
observed together with homogenized properties with increasing number of pene-
trated layers.
By means of regression analysis, for the characteristic embedment strength fh,k,CLT of
dowels inserted in CLT side face two models were found: the first, considering ex-
plicitly the CLT layup, and the second, more simplified approach, representing the
investigated CLT layups and tested configurations on an average basis, see Eq. (10).
1.2
32  1  0.015  d  
f h,k,CLT  with f h,k,CLT,corr  f h,k,CLT   k  (10)
1.1  sin 2  cos 2  400 
with as angle between load and grain of outer layers and k = 400 kg/m³. This re-
gression model is limited to the investigated parameters, the layup parameter of
tested panels, i.e. the sum of layer thicknesses oriented parallel to outer layers vs. the
sum of layer thicknesses oriented perpendicular to outer layers, tℓ,x,i / tℓ,y,i, which
was between 0.95 and 2.1, and the maximum layer thickness, which was max [tℓ,i] =
40 mm.
The influence of density on the embedment strength was found to be equal to solid
timber; see additional tests in Blaß and Uibel [11] as well as Blaß et al. [27]. An
adjustment of Eq. (10) to CLT,k = 385 kg/m³, as currently proposed by PT SC5.T1
[16], would lower the resistance by 4 %.
Fig. 14 compares the characteristic embedment strength according to Eq. (10) with
the current regulation for solid timber and glulam according to Eurocode 5 [17]; see
Eq. (11). In both equations, a similar relationship between dowel diameter and em-
bedment strength is observed. In contrast, the influence of the load-grain angle on
the embedment strength is found to be much smaller for dowels inserted in CLT side
face than in solid timber and glulam. This is because the embedment strength deter-
mined on dowels inserted in CLT side face comprises both, the influence of layers
oriented parallel and perpendicular to loading direction.
0.082  1  0.01 d  with ρ k =400 kg/m³ 32.8  1  0.01 d
f h,k,EC5  k
 f h,k,EC5  (11)
k90  sin 2  cos 2 softwood 1.35  0.015  d  sin 2  cos2

Furthermore, the characteristic embedment strength of dowels with smaller diame-

ters is closer to current regulations for dowels at = 0° whereas the characteristic
embedment strength of dowels with larger diameters is in-between current regula-
tions for dowels at 0°≤ ≤ 90°. This outcome may be explained in two ways: at first,
splitting failures, as being more frequent for dowels with smaller diameter and loaded

178
parallel to grain, are prevented in CLT by the orthogonal layup. Secondly, an influ-
ence of gaps on the embedment strength of dowels with smaller diameter could not
be observed; the smallest diameter tested was d = 8 mm.

45

d = 6 mm | 0° ≤ β ≤ 90°
d = 12 mm | 0° ≤ β ≤ 90°
385 kg/m³
40

ρCLT,12,k

d = 20 mm | 0° ≤ β ≤ 90°
35

30
fh,β,k [N/mm²]

25

20
EC 5 [17] | solid timber & GLT
15 d = 6 mm
d = 12 mm
10 d = 20 mm

5 Blaß & Uibel [11] | CLT

d = 6 mm | 0° < β ≤ 90°
0 d = 12 mm | 0° < β ≤ 90°
300 350 400 450 500 550 d = 20 mm | 0° < β ≤ 90°
ρCLT,12,k [kg/m³]

Fig. 14. Characteristic embedment strength vs. characteristic density of smooth

dowels and tight-fitting bolts: comparison between Eurocode 5 [17] (solid timber
and GLT) with Blaß and Uibel [11] and Uibel and Blaß [12], [13]; CLT side face).

Overall, regulation of embedment strength for smooth dowels and tight-fitting bolts
inserted in CLT side face is suggested equal to solid timber and glulam, with adjust-
ment factors for very small dowel diameters, accounting for a potential negative in-
fluence of gaps, and an adapted k90-factor (see Eq. (11)), taking into account the joint
action of layers featuring different load-grain angles in CLT.
4.2.2 Profiled nails and self-tapping screws
Blaß and Uibel [11] and Uibel and Blaß [12, 13] report also on embedment tests with
self-tapping screws and smooth nails in the side face of wood-based panels with layer
thicknesses tℓ,i ≤ 7 mm. The thin layers together with the reinforcing transverse lay-
ers lead to rather high characteristic embedment strengths fh,k,CLT in comparison with
fh,k,EC5 for glulam and solid timber according to Eurocode 5 [17], see
1.05
0.5  
fh,k,CLT  60  d with f h,k,CLTcorr  f h,k,CLT   k 
 400  (12)
with ρk =400 kg/m³
fh,k,EC5  0.082  k  d 0.3  f h,k,EC5  32.8  d 0.3

In CLT all nails and screws were inserted without predrilling. Although the embed-
ment tests were conducted with smooth nails, Eq. (12) for CLT is limited to profiled
nails, e.g. helically threaded nails and annular ringed shank nails.

179
As already observed in previous investigations, for dowel-type fasteners without
predrilling, no influence of load-grain angle on embedment strength is observed.
In their design proposal, Blaß and Uibel [11] and Uibel and Blaß [13] limit Eq. (12)
to layer thicknesses of tℓ,i ≤ 9 mm. Thus, for common CLT with standard layer thick-
nesses of tℓ,ref = 20, 30 and 40 mm this equation is of minor concern. In case of CLT
featuring layers with tℓ,i > 9 mm, Blaß and Uibel [11] suggest calculating the embed-
ment strength of laterally loaded profiled nails and self-tapping screws, inserted with-
out predrilling, equal to solid timber. This again allows harmonizing regulations for
solid timber, glulam and CLT.

4.3 Laterally loaded dowel-type fasteners in CLT narrow face

4.3.1 General comments
In contrast to CLT side face, in CLT narrow face positioning of fasteners parallel
and / or perpendicular to grain within and between single laminations as well as be-
tween layers is possible. Furthermore, gaps may have a more significant influence
on the embedment strength of fasteners positioned in narrow face than in side face.
Laterally loaded dowel-type fasteners positioned in the CLT narrow face can be
loaded in-plane, out-of-plane or a combination of both; see Fig. 15. If loaded out-of-
plane, tension perpendicular to grain stresses potentially causing early splitting and
delamination of layers have to be considered; minimum requirements on layer and
CLT thickness are presented later in Chapter 5. As the fasteners are inserted in or
between laminations, for calculation of the characteristic embedment strength the
characteristic density of the laminations, ℓ,k, applies.
Although Uibel and Blaß in Schickhofer et al. [4] recommend not using smooth dow-
els, tight-fitting bolts or profiled nails in the narrow face of CLT, neither for bearing
axial nor lateral loads, the models derived from testing are briefly presented.

F2

F1 F1 load out-of-plane
F2 load in-plane

Fig. 15. Laterally loaded dowel-type fasteners in CLT narrow face: principal load
directions.

180
4.3.2 Smooth dowels and tight-fitting bolts
For laterally loaded dowels in CLT narrow face the following relevant parameters
were identified (Blaß and Uibel [11]):
 angle between dowel axis and grain of the penetrated CLT layer;
 ratio between dowel diameter and thickness of the penetrated CLT layer,
d / tℓ,i;
 position of dowels relative to longitudinal and lateral gaps as well as
stress reliefs.
During testing, Blaß and Uibel [11] frequently observed splitting failures, due to ten-
sion stresses perpendicular to grain, in combination with rolling shear of the pene-
trated layer in case of dowels inserted parallel to grain. Low resistances were found
for dowels inserted perpendicular to grain, featuring a diameter d close to the pene-
trated layer thickness tℓ,i and loaded out-of-plane. This is because of the adjacent
layers with grain oriented parallel to dowel axis, which are activated in compression
perpendicular to grain.
From all tested configurations (dowels inserted parallel / perpendicular to grain, in /
close to gaps within / between layers) the lowest resistances were observed when
dowels with diameter d < tℓ,i were inserted parallel to grain and loaded in-plane.
Based on these outcomes and by means of regression analysis the following con-
servative but universally applicable approach, independent of the load-grain angle,
was defined:
0.91
 
f h,k,CLT  9  1  0.017  d with f h,k,CLT,corr  f h,k,CLT   k,  (13)
 350 
4.3.3 Profiled nails and self-tapping screws
By testing the embedment strength of dowel-type fasteners in the CLT narrow face,
smooth nails and self-tapping screws, inserted without predrilling, were used; see
Blaß and Uibel [11] and Uibel and Blaß [13, 18]. The embedment strength of screws
was determined for the threaded part.
By testing the same load configurations as for dowels, the lowest resistances were
again observed when nails or screws with diameters d < tℓ,i were inserted parallel to
grain and loaded in-plane. Based on these outcomes and by means of regression anal-
ysis the following conservative but universally applicable approach, independent of
the load-grain angle, was defined:
0.56
 
f h,k,CLT  20  d 0.5
with f h,k,CLT,corr  f h,k,CLT   k,  (14)
 350 
In contrast to Eurocode 5 [17] where the effective (core) diameter of screws, def, shall
be used, according to Eq. (14) the nominal (outer) diameter d applies for both, nails
and screws.

181
 def = 7.15 mm | 0° ≤ β ≤ 90°
45

def = 8.58 mm | 0° ≤ β ≤ 90°

350 kg/m³
40

ρℓ,12,k
35

30
fh,β,k [N/mm²]

25

20
EC 5 | solid timber & GLT
def = 4.29 mm | d = 6 mm
15
def = 5.72 mm | d = 8 mm
def = 7.15 mm | d = 10 mm
10
def = 8.58 mm | d = 12 mm
5 Blaß and Uibel [11] | CLT
d = 6 mm
0 d = 8 mm
300 350 400 450 500 550 d = 10 mm
ρCLT,12,k [kg/m³] d = 12 mm

Fig. 16. Characteristic embedment strength vs. characteristic density of self-

tapping screws: comparison between Eurocode 5 [17] (solid timber and GLT) with
Blaß and Uibel [11] (CLT narrow face); def = 1.1 ∙ (0.65 ∙ d) assumed; note: acc. to
Eurocode 5 [17] screws with def ≤ 6 mm are treated like nails (inserted without
predrilling; no influence of load-grain angle), with def > 6 mm as dowels (inserted
with predrilling; influence of load-grain angle).
In comparison to the embedment strength of nails and screws positioned in CLT side
face (see Eq. (12)), in CLT narrow face only 1/3 of the resistance can be utilized.
This is also reflected in Fig. 16 were a comparison between the approach of Blaß and
Uibel [11] and Uibel and Blaß [13] with that in Eurocode 5 [17] for screws inserted
in solid timber and glulam is presented.
Aiming on harmonizing the regulations for embedment strength of fasteners posi-
tioned in solid timber, glulam and CLT, for laterally loaded self-tapping screws (and
nails) in CLT narrow face an additional coefficient klat,CLT,NF, taking into account the
load configuration delivering the lowest embedment strength, and by additionally
neglecting the influence of the load-grain angle, as anchored in Eurocode 5 [17], is
proposed, see
f h,k,CLT,NF  klat,CLT,NF  fh,k,0,ST & GLT with klat,CLT,NF  0.25 (15)
with fh,k,CLT,NF and fh,k,0,ST & GLT as the characteristic embedment strength of laterally
loaded self-tapping screws in the narrow face of CLT and the characteristic embed-
ment strength in solid timber and glulam (GLT) at a load-grain angle of = 0°, re-
spectively.

182
5. Connection design

5.1 Introduction and overview

For an appropriate design of connections comprising groups of fasteners in CLT,
further specifics are required in addition to withdrawal and embedment strength,
which are necessary for determining the single fastener’s load-bearing capacity (ac-
cording to Eq. (10) for axial and e.g. according to Johansen [30] for lateral loading).
These additional specifics and verifications are often directly linked to connection
failure modes. One required verification is the resistance of the component’s net
cross-section. Another verification is the load-bearing capacity of the group of fas-
teners, which concerns their effective number, nef, as well as the minimum spacing
and the edge and end distances, ai. It is worth pointing out that both properties, nef
and ai, are closely related to each other, i.e. nef = n can only be achieved if the single
fastener in conjunction with the anchorage material provides enough ductility for
load redistribution and the minimum spacing (especially parallel to grain, a1) is ful-
filled, preventing unfavorable failure modes such as splitting parallel-to-grain or
block shear.
With regard to the state-of-the-art knowledge on both, nef and ai, many aspects are
already covered by the report of Blaß and Uibel [11]. This source contains a compre-
hensive study aiming on the lateral load-bearing behavior of connections composed
by dowels, nails and screws, situated in the side and narrow faces of CLT elements.
Further and more recent publications, focusing on predominately axially loaded con-
nections, are e.g. Mahlknecht and Brandner [31], Plüss and Brandner [33] and Brand-
ner [14]. The main outcomes and recommendations from these publications are
summarized and discussed in the following Sections 5.2 and 5.3.
Other publications which frequently examine dowel-type connections in CLT are
more related to their cyclic load-bearing behavior and not to nef and ai, e.g. Flatscher
et al. [34] and Gavric et al. [35].

5.2 Laterally loaded dowel-type connections in CLT

Blaß and Uibel [11] conducted a comprehensive test program on laterally loaded
connections in order to validate the applicability of their approaches for the charac-
teristic embedment strength of dowels, self-tapping screws and nails for design pur-
poses as well as for determining dowels and verifying nails and screws minimum
spacing for CLT; see Fig. 17.

183
 a3,(c,t) a1

ℓmin
a3,(c,t) a1

a2
a4,(c,t)

a3,(c,t) a1

tℓ,min
a4,(c,t)
side face

tCLT,min
a2
narrow face

a4,(c,t)
Fig. 17. Definition of minimum spacing, edge and end distances and further
geometrical boundary conditions, exemplarily for self-tapping screws; according to
Blaß and Uibel [11].
Apart from the applied fasteners and their position within the CLT element (side vs.
narrow face), Blaß and Uibel [11] also varied the type of connection (steel-to-timber
connections with inner and outer steel plate as well as CLT-CLT lap joints, both with
one or two shear planes), the fastener diameter (dowels: d = {8, 12, 16, 20, 24} mm;
screws: d = {8, 12} mm; profiled nails: d = {4, 6} mm), the number of fasteners in
the group, n, as well as their spacing, edge and end distances and further geometrical
boundary conditions, which are illustrated in Fig. 17.
The observations made by Blaß and Uibel [11] during testing connections of laterally
loaded dowel-type fasteners in the CLT side face are summarized briefly:
 Apart from some specimen which failed in the zone of load introduction,
other failure modes observed by testing connections with dowels were tension
failure in layers close to the shear plane, successive shear & rolling shear
failures and some block shear failures, whereby block shear failure emerged
without regularity.
 As the number of dowels placed parallel to the outer layer’s grain direction
did not influence the load-bearing capacity, Blaß and Uibel [11] concluded
that the effective number of fasteners nef can be set equal to their total
number n.

184
 The load-carrying capacity of connections with dowels was predicted by
means of the theory of Johansen [30] and the characteristic embedment
strength fh,k according to Eq. (10). This characteristic embedment strength fh,k
only implicitly represents an inhomogeneous stress distribution along the
fastener axis due to penetrated alternating orthogonal layers. Despite these cir-
cumstances and although the connection’s failure modes differed widely from
failure modes observed by testing single dowels, overall good to conservative
predictions of the load-carrying capacity of dowelled connections were made.
 Good validation for the embedment strength as input parameter in Johansen’s
[30] theory was also achieved for the majority of tested connections with self-
tapping screws and nails. This also concerned the withdrawal strength as basis
for the rope effect. The regulation of nef = n is also proposed for these fasten-
ers.
Similar to connections in CLT side face, proposals made by Blaß and Uibel [11] for
the CLT narrow face could also successfully be validated. Two important observa-
tions made during testing are summarized in brief:
 Connections with dowels situated in layers perpendicular to grain and loaded
parallel to grain failed by splitting already at small deformations. However,
due to the implicit conservatism in predictions according to Blaß and Uibel
[11] even these failure modes are covered but it is pointed out that the given
recommendations are only proven for tested CLT layups and configurations.
 Tests with laterally loaded self-tapping screws situated in the CLT narrow
face showed that for some configurations minimum spacing determined by
insertion tests in advance were too small. Consequently, Blaß and Uibel [11]
increased the corresponding values.
Minimum spacing, edge and end distances and further geometrical boundary condi-
tions, as defined in Fig. 17, as outcome of the comprehensive test campaigns and
proposal in Blaß and Uibel [11, 37] are summarized in Table 2 and Table 3.

185