Development of Wooden Portal Frame Structures
with Improved Columns
Dr. Masahiro Noguchi
Post Doctoral Fellow
Tokyo Institute of Technology, Yokohama, Kanagawa, Japan
Prof. dr. Kohei Komatsu
Research Institute for Sustainable Humanoshere, Kyoto University, Uji, Kyoto, Japan
This paper proposes two semi-rigid timber frames with a more effective structural
performance are compared and a structural design method is derived. Usually the joints are
located at the intersection of the beam and the column. The first frame type applies two types
of joints. A high strength capacity rigid glued joint is used to replace the traditional
beam-to-column joint while a second ductile semi-rigid joint is positioned at the area with low
bending moments.. The beam pieces that run between the column and the semi-rigd joints are
so well fixed to the column that they form one integral part. In the second frame type the
horizontal beam between the columns is extended beyond the location of the semi-rigid joints
of the first frame. This creates a large overlapping area where mechanical fasteners such as
bolts are generously spaced. Due to the large fastener spacing the stiffness is enhanced as well
as the strength.
In Japan, the lifetime cycles of most housing are around 20-30 years. This might be
considered as wastes of resources and energies from the global environment perspective. Most
governing factor is durability due to old fashioned use of the house. As a solution of this
problem, it is thought to build houses with skeleton structures which allow free partition of
spaces by future owners. To develop the skeleton structure effectively, multi-story frame with
span of 6 to 10m are required. For this reason, we pay our attentions on the multi-story
wooden portal frame structures.
The approaches of many researches in the past 1-8 were that the structural performances were
improved by improving only the moment transmitting connections. However, it can be
thought that more parameter influences: as example, location of the moment connection,
member, and so on.
In this article, two types of wooden portal frame structures were proposed. Both having
vertical columns added a short horizontal member with glue joints as shown the shadow area
in Fig. 1 (a) and (b). These vertical members were defined as “improved columns”. The aim
of this article was to show the structural advantages of this type of improved columns.
Material and methods
First type is the structures changed the location of moment transmitting ductile connection
with improved columns (Type E), in Fig. 1 (a). There is no ideal rigid beam to column joint
having stronger than member, regarded as rigid, ductile. Thus, in order to improving the
structural performance of semi-rigid structure, it could be thought better to make the position
move to where bending moment is small.
Second type is the structure whose panel zones were extended with the improved column
(Type S), in Fig 1 (b). The panel zone, in this article, was defined as the overlapping area
where column and beam met. The structural performances of moment resisting joints were
always limited in the height of both members geometrically. If the height at joint part is
increased, the larger moment resisting joint can be made. In bolted cross lapped joint,
mechanical properties were governed by the mechanical property of single bolted joint and
the adjacent bolt space. This result in that beam to column moment resisting joint having
higher capacity on stiffness and strength could be expected, extending the panel zone.
Finally, the portal frame with traditional bolted cross lapped joints is shown in Fig. 1 (c) as the
control type (Type C).
Nine portal frame specimens were built, three types × three replications. Each column
member was 3000 × 200 × 120 mm, pairs of beam members were 3000 × 200 × 60 mm. All
specimens were made of Sugi (Japanese Cedar, Cryptomeria japonica) glulam having JAS
(Japanese Agricultural Standard) strength grade of E65 – f 220 (MOE = 6500 MPa and
MOR= 22 MPa). The average moisture content was 11 %. All specimens were two story
miniature semi-rigid frame structures. Each leg joint was shown in Fig. 2 (a).
(a) Type E (b) Type S (c) Type C
Fig 1 Specimen
(a) Leg joint (b) Knee joint of Type C (c) Knee joint of Type S
Fig 2 J oint detail
Preparation of the improved columns
Two rectangle holes were made in each column and drilled eight circular holes as shown in
Fig. 3 (a). Each rectangle hole was cross-section of 200 ×30 mm, depth of 160 mm. Each
circular hole have diameter of 18 mm, length of 100 mm. Tenon member was also made, as
shown in Fig. 3 (b). Central tenon was the width of 29.5 mm, the depth of 200 mm, and the
tenon length of 155 mm. Similar to the mortise, each tenon have eight circular hole having
diameter of 18 mm, length of 100 mm were drilled in longitudinal direction.
The tenon and the slender steel rods of diameter 16 mm were driven into the rectangle holes
and circular holes respectively, and they were fixed with epoxy resin adhesive using
sledgehammer. They finally they were formed a F-shaped assemblage were completed. We
confirmed the adhesive injection by observing overflow of adhesive from holes.
The insert length of the steel rod in each member was set to 100 mm. The time to cure was set
at least two weeks.
(a) Mortise (b) Tenon
Fig 3 Construction of T-shaped member for the improved column
Assemble of portal frame specimens
Three types of portal frame specimens were assembled with improved columns, pairs of
beams and short bases using bolts as shown in Fig. 1. The clearance between bolts and
pre-drilled holes were 1.5 mm, hole diameter was 12 mm and bolt diameter was 10.5 mm.
Figures. 2 (b) and (c) show the bolt arrangement. Bolt arrangement of Type E was
geometrically the same as that of Type C.
Measurements and test procedure
The portal frame specimens were subjected to cyclic loading by applying a horizontal lateral
force at the top of the specimens, as illustrated in Fig. 1, (c). Cyclic loading tests were carried
out based on the protocol shown in Table 1. Story drift θdrift was calculated by the Eq. (1).
where δa : Displacement at roof beam (mm)
δb : Displacement at column base (mm), h : Distance between device for δa and that forδb.
Table 1 Load protocol
Displacement Angle R (rad)
The number of cycles
Results and Discussions
In Type C, failure did not occurred up to the end of the stroke length of hydraulic actuater.
Bolted moment transmitting joints were yielded and then worked as the plastic hinges, which
made collapse mechanism of the structures. Similarly, Type E specimens gave no failure up to
the end of stroke. However collapse mechanism was different from Type C. Fig. 4 shows the
typical failure mode of Type E specimens. As can be seen in Fig. 4, the failure was occurred
not at the regions where vertical and horizontal member met, but the regions where the bolted
moment transmitting joints were located. Keep this failure mode in mind to the latter
discussions. In Type S, the sprit failure occurred at the outer bolt hole in moment joint as
illustrated in Fig. 2 (c). But final fatal reduction of load was due to the sprit of bolt hole in
timber beam (see Fig. 2 (c)).
Fig. 4 Failure mode of Type E
Shear force - story drift curve
Typical shear force - story drift curves for the three different types of portal frames specimens
are shown in Fig. 5. From Fig. 5, it is obvious that type E and S specimens have remarkable
advantages on structural performance to the control type, especially stiffness. Therefore, the
portal frames proposed in this article obviously have high possibility for rational wooden
portal frame structures. In latter chapters, the detail features were discussed.
Fig 5 Shear force-story drift relationship
Story drift (rad)
Shear force (kN)
Table 2 shows the test results with respect to the initial stiffness determined by both visual
readings (visual method) and method proposed by Japan Housing and Wood Technology
Center (HOWTEC) 9 for the three different types of portal frames. The main difference
between visual and HOWTEC methods are whether initial slip value was contain or not. The
stiffness determined by visual method did not contain initial slip, while those by HOWTEC
method contain initial slip. As can be seen in Table 2, the stiffness of Type E and S are around
1.7 and 3.5 times as large as that of Type C, respectively. The differences on stiffness between
HOWTEC method and visually method were small.
As the shear stresses at panel zone are concentrated in moment transmitting joints, the
deformations of panel zone must occur as long as we use elastic material, not rigid body. The
deformations of panel zone make the rotation of joint unavoidable. Therefore, for making
rigid joints, the improvements of panel zone are effective solution. As the elasticity of timber
is low, to avoid the concentrated shear stress at panel zone with elongating the panel zone
using improved column, such as Type S, can be thought as the effective solution in the timber
moment transmitting joints.
Type EType CType S
Off-set value (rad)
Fig 8 Definition and values of off-set
Next, we will discuss about the effects of the clearance between bolt and predrilled hole. In
practice, the clearances are always required to erect the structures on the sites. However, the
clearances tend to bring undesired initial slag in bolted cross lapped joints. Fig. 8 (a) shows
the initial slag. Fig. 8 (a) shows definition of the off-set value due to the initial slag. As shown
in this Fig., the ratio of the average off-set value of type E and type S to type C were half and
quarter, respectively. More information can be obtained from Fig. 8 (b), which shows the
dispersions of the off-set values in both Type E and S types were much smaller than that of
type C. The reason was discuss as follows separately. In Type E, it was thought that the
rotational angle was not same as the story drift angle. While, in the type S, the distance
between each bolt and shaft was much lager than that of Type C, Type C of 73.5 mm, Type S
of 237 mm. As the initial slip was roughly in inverse proportion to the distance at the same
rotational angle of joint, the slip of type S was around one-forth times as large as that of Type
Generally, the secondary stresses due to shrinking of the member caused by thermal and
moisture changes were, in statically undetermined structures, occurred, i.e.: the portal frame
structures using knee joints with adhesive and no clearance moment joints like expanded tube
joints, While, in case of the portal frame structure with bolted knee joint having some
clearances, i.e.: the structure proposed in this article, it was thought that the secondary stresses
can be avoided to release using the clearances. Because there are poor information associated
with this problem in timber engineering. Therefore, it is thought that both proposed portal
frames have also advantage for secondary stress.
Strength and Ductility
Average yield, ultimate and maximum strengths and standard deviations for the three different
types of portal frames are shown in Table 3. As can be seen in Table 4, the strength of Type E
and S are around 1.25 and 1.45 times as large as that of Type C, respectively. The difference
on strength ratio among the types of specimen was small. As both Type E and C made
collapse mechanisms with bolted cross lapped joints in the same manner, we tried to explain
the difference between the strength of Type C and Type E using classic yield collapse model,
as shown in Fig. 9.
(a) Type C (b) Type E
Fig 9 Collapse mechanism
In this article, two types of wooden portal frame structures were proposed. Both structures
have improved columns. First type was the structures changed the location of moment
transmitting ductile connection with improved columns. The second type of structure whose
panel zone was extended using improved column. From the test results, the stiffness were
improved around 1.7 and 3.5 times as large as that of control, the strength were improved
around 1.25 and 1.45 times, respectively. Therefore, the portal frame structures with improved
columns have structural advantages, especially stiffness.
(1)Leijten A.J.M.(1998) Reinforced joints with expanded tube fasteners In; Densified veneer
wood reinforced timber joints with expanded tube fasteners, Delft University press, Delft, pp
(2)Leijten A.J.M (1988): Steel reinforced joints with dowels and bolts. In; Proceedings of
International Confidence on Timber Engineering, Washington, vol. 2, pp. 475-488
(3)Haller P.,Chen C. J, Natterer J (1996): Experimental study on glass fibre reinforced and
densified timber. In; Proceedings of International Wood Engineering Conference, New
Orleans, USA, vol. 1, pp. 308-314
(4)COST C1 edited by P. Haller (1999) Semi-rigid timber joints – Structural behaviour,
modelling and new technologies Final report of Working Group ”Timber joints ”
(5)Leichti,R.J, Tjahyadi, A, Bienhaus, A, Gupta, R, Miller, T, Duff, S (2002) Design and
behaviour of frictio dampers for two-dimensional braced and moment-resisting timber frames.
In; Proceedings of World Conference on Timber engineering, Shah Alam, Malaysia, vol. 2,
(6) Rodd P. D.(1996) Resin injected dowels in moment transmitting joints. In; Proceedings of
International Wood Engineering Conference, New Orleans, USA, pp.169-176
(7) Ohashi Y, Sakamoto I. (1989) Study on laminated timber moment resisting joint. In;
Proceedings of the Secund Pacific Timber Engineering Confidence, Auckland, Australia, vol.
2, pp. 37-42
(8)Komatsu, K, Karube M., Harada M., Fukuda I., Hara Y., Kaihara H.(1996) Strength and
Ductility of Glulam Portal Frame Designed by Considering Yield of Fasteners in Part. In;
Proceedings of International Wood Engineering Conference, New Orleans, USA
(9) The building center of Japan (2002) the building letter, 443, 27-32
(10)Noguchi M, Komatsu, K (2003) A new proposal for estimating method of stiffness and
strength in the bolted timber-to-timber joints and its verification by experiments (II): bolted
cross lapped beam to column joints. Journal of wood science, 50, 391-399
(11)Architectural Institute of Japan (AIJ) edited (1995) Moment resisting joints In; Structural
Design Note for Timber Structures, Maruzen, Tokyo, (in Japanese) pp. 184-221