Sunteți pe pagina 1din 9

CONFERENCE PAPER

No. 61 (1999)

Earthquake Load Sharing Between


Timber Framed and Masonry Walls
Graeme J. Beattie

Presented at the Pacific Timber Engineering Conference,


Rotorua, March 14-18, 1999

Funding for this project was provided by the Cement and Concrete Association
and the Building Research Levy.
ISSN: 0111-7505
EARTHQUAKE LOAD SHARING BETWEEN TIMBER FRAMED AND MASONRY
WALLS

G. J. Beattie
Senior Engineer, Building Research Association of New Zealand

SUMMARY

The lateral load resistance of lightly reinforced masonry walls has always been thought to be quite different from that of
timber framed walls. Timber walls were expected to respond to earthquake loading in a ductile fashion while masonry
walls were expected to behave in a brittle manner. These differences made it difficult to design structures that
incorporate both of these common forms of construction. Since the introduction of the non-specific design standards, it
is now recognised that lightly reinforced masonry walls have a limited amount of ductility, which reduces the bracing
demand. Additional ductile capacity is available in the connections between masonry walls and the ceiling. This was
quantified experimentally. The distribution of lateral force between masonry and timber framed walls was examined in
some typical structures by computer modelling. It was found that the stiffness of a masonry wall system incorporating
the bolted and nailed joint, is not significantly different from that of a timber framed wall system when there is a bolted
and nailed joint at the ceiling to wall junctions. The distribution of lateral load between walls may be obtained using
tributary areas when these joints are used. Diaphragm action must also be considered when there are continuous ceiling
“chords” perpendicular to the bracing walls. Without the presence of diaphragm action, the loads can be distributed on
the basis of tributary area.

INTRODUCTION was accepted as an “Industry Standard” for housing.


The non-specific masonry standard attempted to emulate
There has been a problem for some time relating to how the non-specific design timber code, using the same
to distribute lateral forces between lightly reinforced “bracing unit” system to describe the seismic and wind
masonry walls and timber framed walls in houses that resistance to be provided by the structure. The “bracing
are designed to the non-specific design standards [6,7]. unit system” is based upon the equivalent static force
design method.
This paper outlines the historic problem of stiffness
incompatibility between walls constructed of the two Early versions of the two standards were based on
materials and its effect on the lateral load resisting different design philosophies and the authors of the
behaviour of structures built using a combination of standards never intended them to be used jointly. The
timber framed and masonry walls. It describes changes non specific design timber standard was based on a
to the loadings standard [8] design philosophy and the working stress design method while the non specific
consequent changes to the non-specific design design masonry standard was based on a strength design
standards. method. Houses of similar weight required a greater
number of bracing units if built of masonry than if built
Experimental testing of bolted ribbon plates is of timber. Furthermore, timber walls were considered to
described. The paper then gives details of a computer respond to an earthquake loading in a ductile manner
model incorporating the results of this experimental while masonry walls responded in a brittle manner.
work and work previously undertaken to investigate the
distribution of forces between masonry walls and timber In 1992, when the loadings standard [8] was converted
framed walls. Conclusions are then drawn from the to limit states format, the non-specific design timber
results of the computer modelling. standard was also converted from a working stress to a
limit state format. The effect of this conversion was an
increase in the bracing unit demand in the non-specific
BACKGROUND design timber standard. However, the procedure for
establishing the available bracing capacity of a system
The vast majority of New Zealand houses are built to the was also amended to reflect the change in philosophy.
non-specific design code for timber [7] while a smaller
number are built with reinforced masonry walls to the There was also an amendment made at this time to the
non-specific design code for masonry [6]. non-specific design timber standard to reflect an
acceptance that reinforced masonry walls designed to
The dominance of timber framed houses in New the non-specific standard did exhibit limited ductility -
Zealand meant that the non-specific timber design code
the specified bracing ratings for masonry walls were NZS 4203:1992, showed that for houses in earthquake
increased. zone A, NZS 4203:1992 would require an increase in
the number of bracing units of up to 24% over the NZS
Houses with a reinforced masonry lower storey and a 3604 requirements.
timber framed upper storey were then able to be
designed using the two standards but there was no Clearly, there is a need to rationalise the requirements in
design procedure for houses with a mix of reinforced both non-specific design standards so that the two
masonry and timber framed walls within the same materials can be effectively used side by side in the
storey. same structure.

RELATIONSHIP BETWEEN PREVIOUSLY LOAD AND DISPLACEMENT BEHAVIOUR OF


REQUIRED BRACING UNITS AND THOSE BOTH WALL TYPES
DERIVED FROM THE 1992 LOADINGS CODE
Full load reversal pseudo-static in-plane racking tests
A comparison was recently made between the number of have been undertaken at BRANZ on both timber framed
bracing units required by NZS 4229 [6] and those and 150 series reinforced masonry walls [1,3,9]. Figure
derived using NZS 4203 [8] as part of a BRANZ 1 shows a typical load-displacement hysteresis loop for
investigation [2]. This used a house with a floor plan of a light timber framed wall. The wall panel was 2.4 m
14  7 m and a roof area of 125 m2 for the comparison. long, had nominal 100 mm by 50 mm studs at 600 mm
Tables 1 gives the bracing demands per square metre centres and was clad with a plasterboard lining nail
derived using NZS 4230:1992 for different fixed on one face. The difficulty in relating the
combinations of roof and wall materials. The number of performance of such a wall to the real world is in
bracing units required by NZS 4229:1986 are shown in accurately representing the continuity of the wall beyond
brackets. the panel under test.

It can be seen that the reduction in bracing units using To better replicate the continuity present at the outer
NZS 4203:1992 is very significant and drops to as low edges of the panel and also model the effects of
as 40% of the NZS 4229 requirements. This reduction discontinuities such as doors and windows within the
is caused partially by the recognition that lightly panel, a number of 6.4 m long walls have been subjected
reinforced non-specifically designed masonry walls are to racking tests [9]. Figure 2 gives a typical hysteresis
more ductile than was previously assumed. loop for one of these walls in which plasterboard was
fixed on one side.
In contrast to the above, a comparison [2] of the bracing
requirements of NZS 3604 [7] and those derived from
Timber Partitions Masonry Partitions
Zone A Zone B Zone C Zone A Zone B Zone C
150 series 13 (25) 10 (20) 6.5 (15) 18 (30) 14 (25) 9 (20)
+ veneer 18.5 (30) 14 (25) 9 (20) 24 (40) 18 (35) 12 (30)

a) SINGLE STOREY WITH LIGHT ROOF

Timber Partitions Masonry Partitions


Zone A Zone B Zone C Zone A Zone B Zone C
150 series 18 (30) 13 (25) 9 (20) 23.5 (40) 17.5 (35) 12 (25)
+ veneer 23 (35) 17.5 (30) 11.5 (25) 29 (45) 21.5 (40) 14.5 (30)

b) SINGLE STOREY WITH HEAVY ROOF

Zone A Zone B Zone C


200 series Upper floor 29.5 (60) 22 (50) 15 (40)
Lower floor 53 (100) 40 (85) 26.5 (70)
+ veneer Upper floor 38.5 (70) 29 (60) 20(50)
Lower floor 71 (100) 54 (85) 35.5 (70)

c) TWO STOREY WITH HEAVY ROOF - LIGHT TIMBER PARTITIONS TO THE


UPPER FLOOR AND MASONRY PARTITIONS ON THE LOWER FLOOR.

TABLE 1 COMPARISON OF BRACING REQUIREMENTS OF NZS 4203 AND NZS 4229

8 time the wall had reached a displacement of 24 mm, a


6
number of the nails connecting the plasterboard to the
bottom plate had broken out of the edge of the sheet.
4
The peak load was clearly achieved at a displacement of
2 about 12 mm. A view of the wall in the test rig is
Load kN

0
shown in Figure 3.
-25 -20 -15 -10 -5 0 5 10 15 20 25
-2

-4

-6

-8
Displacement mm

Figure 1: Load-displacement hysteresis loops for 2.4


m timber frame lined with plasterboard on one face
15

10

5
Load (kN)

0
-30 -20 -10 0 10 20 30
Figure 3: 6.4 m long wall in the racking test rig
-5

-10 The masonry wall was 4.6 m long and consisted of two
linked shear panels 1 m and 1.4 m long with window
-15 and door openings between, to represent a typical
Displacement (mm)
section of exterior wall. Only the cells containing
Figure 2: Load-displacement hysteresis loops for 6.4 vertical reinforcement and the top two courses forming a
m timber frame lined with plasterboard on one face bond beam were grouted. A view of the masonry wall
during testing is given in Figure 4 and the corresponding
During the early cycles to 6 mm, the joints between the load-displacement plot is shown in Figure 5.
plasterboard sheets had opened and this gap width
increased as the wall displacement increased. By the
The difficulty arises in distributing the lateral forces in a
house which has both masonry and timber framed walls.
For a house with a rigid diaphragm ceiling or roof and
no torsional eccentricity, then the assumption that the
lateral capacity of the whole is the sum of that of the
parts is clearly incorrect. This is because when the
stiffer masonry wall is at its ultimate strength the
displacement of the adjacent more flexible timber walls
will be less than that required to reach their ultimate
strength.

If the ceiling or roof is not rigid, as will be the case with


most sheet material linings on timber framing, then the
Figure 4: 4.6 m long masonry wall under test displacements of the two wall types will not be the same
and both types may be able to mobilise their ultimate
120 strengths at the same time. This essentially suggests that
the walls could be designed on a tributary area
80
approach.
40
Load (kN)

CONNECTION BETWEEN CEILINGS AND


0
WALLS
-20 -15 -10 -5 0 5 10 15
-40
Typical Details
-80

-120 NZS 4229 provides a number of details for connecting


Displacement (mm) the ceiling and/or roof to the masonry walls. Generally,
Figure 5: Load-displacement hysteresis loops for a either the ceiling lining is nailed directly to an
4.6 m reinforced masonry wall overhanging top plate which is in turn bolted to the top
of the masonry wall (Figure 6(a)) or it is nailed to a
Faint hairline cracks appeared in the mortar joints ribbon plate (stringer) bolted to the side face of the wall
immediately beneath the corners of the window opening (Figure 6(b)). Finishing of the junction is likely to be by
and also vertically and horizontally at the top corners of a timber fillet or similar to hide the nail heads. Both the
the door opening at a shear load of 40 kN and a bolted joint and the nailed joint provide flexibility to the
longitudinal displacement of the top of the wall of about connection between the ceiling and the wall.
1 mm. The wall was able to resist a maximum load of
approximately 70 kN before yielding occurred, but For timber framed walls, the connection between the
continued to gain strength up to about 100 kN, at a ceiling and the wall has some significant differences to
displacement of about 8 mm. Using the procedure the masonry wall connection. While the ceiling lining is
proposed by Park [5] for calculating the displacement nailed to the top plate or blocking adjacent to the top
ductility of a test specimen, a ductility of 3 was plate, the shear transfer from the ceiling to the wall is
obtained. Compare the behaviour of this wall to the through the taped and stopped joint (Figure 7(a)) or
plasterboard lined wall where a maximum load of 13 kN through a plaster cornice which is glued in place (Figure
was achieved at a displacement of 13 mm. 7(b)), both of which are stiff connections.

Clearly, the strength of the masonry wall is considerably While NZS 4229 requires M12 cast in bolts at 1200 mm
greater than the timber framed wall. Notice that the centres for fixing the plate or the stringer to the wall,
deflections at the peak load are similar in one direction. often builders will substitute proprietary expanding
In the other direction, to achieve peak capacity the masonry anchors of the same strength for ease of
timber framed wall is required to displace about one and construction. The bolted connection is expected to have
a half times as far as the masonry wall. some flexibility in either case.
Gable end in
masonry
construction
M12 bolts at 1200 crs to stringer
Building strap and solid
Wall plate blocking to ceiling joists

300mm
Ceiling
joist
Batten Ceiling diaphragm
fixed to stringer
Ceiling diaphragm
fixed to plate
M12 bolts at 1200 centres
to wall plate

(a) (b)
Figure 6: Alternative fixings for ceilings to masonry walls

Ceiling Top
joist plate

Joint Taped Plaster cornice


and stopped glued in place

Wall framing

(a) Taped and Stopped (b) Glued cornice

Figure 7: Typical ceiling/timber framed wall connections

Laboratory Testing Restraint

To quantify the load/slip characteristics of the bolted D12 bar


connection, a series of cyclic load reversal tests was Internal webs
removed
undertaken at BRANZ. Both cast-in and proprietary 50 x 50 x 3
expanding anchors were investigated. A single bolt was washer
3-2014 M12 cast in bolt (Case 1)
tested in each case, fixing a 1200 mm long nominal 150 masonry M12 Expansion anchor (Case 2)
blocks
mm x 50 mm stringer against the side face of a masonry
wall (Figure 8). 17.5 MPa grout
150 x 50 ribbon
plate
Each specimen was subjected to three cycles to each Grout
displacement from 1 mm to 5 mm in 1 mm increments at M12 cast
D12 in bolt (Case 1)
a slow loading rate. Three further cycles were Restraint 50mm dia.
completed to 10 mm, 15 mm and 20 mm, if the fixing hole in wall
Section of block
hadn’t already failed. The results showed that the peak at bolt filled with
grout
load for the M12 cast in anchors was about 13.5 kN
(equivalent to 11.3 kN per metre of wall) and for the Load

proprietary fixing 11.5 kN (9.6 kN/m). The initial


Plan on Test Setup
stiffness of both anchor types was between 8 and 10
kN/mm but the stiffness reduced quickly as the Figure 8: Bolted ribbon plate connection tests
displacement increased. The displacement of both the
cast-in anchors and the proprietary anchors was about
15 mm at the peak load. Figure 9 shows the curves of
the mean of the third cycle peak loads obtained from the
experimental hysteresis plots for the two anchor types.
20
COMPUTER MODELLING OF HOUSES
15
A selection of house formats were modelled to
10
determine the distribution of seismic forces between
5
masonry and timber framed walls. The ceiling and walls
0 were identified as the most significant components.
-20 -10 -5 0 10 20
Load (kN)
-10 M12 mean A finite element model of the ceiling was programmed
-15 Anchor mean into an Excel [4] spreadsheet. The ceiling was modelled
-20 as a series of beam elements which accommodated both
Joint Slip (mm) bending and shear deformations. The boundary chord
area and elastic modulus were used to calculate bending
Figure 9: Mean third cycle peaks of the bolted deformations. The shear modulus of the ceiling lining
stringer connections and the non-linear joint deformation characteristics were
used to calculate shear deformations. The shearwalls
Curve Fitting were modelled as non-linear springs at the ends of the
diaphragm and at up to 5 locations within its length.
An approximation to the Figure 9 response was required Forces in the model were balanced using the spreadsheet
to represent it in a model of a full house. The following solver for each arrangement of external loads.
equation has previously been used to model the load-slip
performance of nails securing plasterboard and fibre- Outputs from the program included reaction forces at
cement board to timber framing [10]: wall lines, deflections of walls, and ceiling deflections,
bending moments and shear forces.
A * Slip(mm)
Load(kN)  (1) Computer Model Spring Properties
B  Slip(mm)
where The nailed joint between the ceiling and the ribbon plate
was modelled as a non-linear spring, as was the bolted
A = the (asymptotic) ultimate strength of the joint between the ribbon plate and the masonry wall. In
bolted joint both cases, the spring characteristics were modelled
B = the characteristic deformation which is using equation (1), but with constant coefficients
that at half ultimate load (giving an initial appropriate to the stiffness of the connection.
stiffness of A/B)
The stiffness characteristics of the masonry walls and
This equation was fitted to the load-slip plot for each of the timber framed walls were modelled using the same
the cast-in anchors using a least squares method. equation form. The stiffness characteristics of the bolted
ribbon plate joint and the masonry wall were combined
Figure 10 shows the match between the experimental into a single equation to simplify the calculations.
curve and the curve generated from the above
expression. The A and B terms are displayed on the Comparison of Stiffnesses
graph.
To illustrate the stiffness relationships between the
masonry wall systems and the timber framed wall
systems, load-displacement plots have been produced
15 for the individual components and for the combination
A = 14.5 of these, all normalised to loads per metre of wall.
10
B = 0.44 These are presented in Figure 11.
5
Load (kN)

0 It is clear from the plot that the masonry wall and the
-20 -10 M12 cast-in bolts individually have high initial
-5 0 10 20
stiffnesses. However, when they are combined with the
-10 M12 bolt stiffness of the nailed joint to the ceiling, the stiffness of
-15 Match the system is markedly reduced. The system stiffness is
comparable to the stiffness of the wall lined with
-20
Joint Slip (mm)
plasterboard on both sides and the wall lined with
plasterboard on one side and fibre-cement board on the
Figure 10: Third cycle peaks and the best fit curve other.
TABLE 4
DETAILS OF HOUSE MODELS
20 PB = plasterboard
Model # Dimensions (m) Details
FC = fibre cement sheet
PB both sides
18 Timber framed end walls with plasterboard lining and one central masonry wall
1 16 x 8
PB & FC
Timber framed end walls with plasterboard lining and fibre-cement board
2 16 16 PB
x 8 nails
cladding
14 Masonry wall at one end, timber framed central wall with plasterboard lining both
3 16 PB&Bolt&Masonry
x8
Load (kN/m)

faces and timber framed end wall with plasterboard lining and fibre-cement board
12 Masonry cladding
Masonry end walls and two timber framed walls lined with plasterboard on both
4 10 16 M12
x 8 bolts
faces, each at 6 m from the end
8 Timber framed end walls with plasterboard lining and five masonry walls spaced
5 14 x 7.4
randomly between
6
Masonry end walls and a mix of two plasterboard lined timber framed and three
6 16 x 8.5
4 masonry interior walls

2
0
0 0.5 1 1.5 2 2.5 3
Displacement (mm)
Figure 11: Comparison of Stiffnesses Normalised to Loads per Metre

envelope), the distribution of force to the walls was


found to be more related to the stiffnesses of the
Computer Model Results elements, particularly the flexural stiffness of the
ceiling.
Six house arrangements were modelled. Four of these
had simple 16 m by 8 m rectangular plans and basic wall If the masonry walls were at the ends of the structure,
layouts while the remaining two contained a mix of wall they tended to attract more than their tributary area share
types, lengths and positions. Loading was considered in of load by up to 50%. When they were randomly placed
the transverse direction only. internally, the attraction of load was closer to that based
on tributary areas. The reason for this is unclear but it
Details of the plans are summarised in Table 2. may be due to the relatively short length of building load
attributable to each wall element in the latter case.
In all models the ceiling was either considered to be
lined with ordinary plasterboard or bracing plasterboard. CONCLUSIONS
The major difference between these two cases is in the
strength of the nailed connection between the ceiling This paper has presented the background to the
and the ribbon plate. The other major variation development of the two non-specific design standards
investigated in the models was the effect of the ability of and the changes to the loadings standard philosophy
the ceiling to behave as a continuous diaphragm over the since their introduction. Experimental investigations
plan of the house. have shown that the fundamental stiffness
incompatibility between the two wall types when used in
From the model outputs it was clear that when the the same structure is less of a concern when account is
ceiling was not considered to act as a diaphragm, the taken of the flexibility of the connection between the
distribution of forces to the walls was proportioned on masonry shear walls and the ceiling.
the basis of tributary areas. This would be the case
when masonry walls are utilised internally and they A computer model was developed which allowed the
penetrate the exterior envelope, causing a break in the strength and stiffness properties of both wall types, their
exterior wall top plate, which is acting as the chord of connections to the ceiling and the ceiling itself to be
the diaphragm. This could be expected if the internal incorporated. Realistic force distributions between the
wall was being provided for fire and/or sound resistance walls were obtained.
between flats or sleeping and living areas of a house.
When the ceiling was considered to act as a diaphragm The model results indicate that it is reasonable to
with a continuous top plate (a case which will exist if the attribute lateral load to the walls based on their tributary
interior masonry walls are enclosed within the building areas when the ceiling chord member is discontinuous.
However, when it is possible to have a continuous 4. Microsoft Corporation. (1992) Microsoft
chord, masonry walls placed on exterior wall lines 
Windows Excel 5.0
attract more than their tributary area share.
The influence of the ceiling flexural stiffness was found 5. Park, R. (1989) Evaluation of Ductility of Structures
to be significant but not easily quantifiable. Further and Structural Assemblages from Laboratory
work is required to establish the degree of influence Testing, Bulletin of the NZNSEE, Vol 22, No. 3.
over a range of chord types and arrangements. 6. Standards Association of New Zealand (1986) NZS
4229:1986; Code of Practice for Concrete Masonry
FUNDING ACKNOWLEDGEMENT Buildings not Requiring Specific Design, Standards
Association of New Zealand, Wellington.
The project was funded equally by the Cement and
Concrete Association and the Building Research Levy. 7. Standards New Zealand (1990) NZS 3604:1990;
Code of Practice for Light Timber Framed Buildings
References not Requiring Specific Design, Standards New
Zealand, Wellington.
1. Herbert, P.D. (1997) Report on the Indicative 8. Standards New Zealand (1992) NZS 4203:1992;
Testing of Concrete Masonry Walls for Seismic Code of Practice for General Structural Design and
Assessment. Building Research Association of New Design Loadings for Buildings (known as the
Zealand, Test Report STR 336, Judgeford, New Loadings Standard), Standards New Zealand,
Zealand. Wellington.
2. Herbert, P.D. and Beattie, G.J. (1997) Earthquake 9. Thurston, S.J. (1993) Report on Racking Resistance
Behaviour of Combined Masonry and Timber of Long Sheathed Timber Framed Walls With
Framed Walls., Building Research Association of Openings. Study Report SR 54, Building Research
New Zealand, Report SC1043, Judgeford, New Association of New Zealand, Judgeford, New
Zealand. Zealand.
3. Herbert, P.D. and King, A.B. (1998) Racking 10. Thurston, S.J. (1994) The Specific Design of
Resistance of Bracing Walls in Low-Rise Buildings Shearwalls and Ceiling Diaphragms Constructed
Subject to Earthquake Attack - Volume 2 - • •
Using Gib Braceline and Gib Fyreline. Building
Experimental Program Derivation & Assessment. Research Association of New Zealand, Report SC
Study Report SR 78, Building Research Association 5014, Judgeford, New Zealand.
of New Zealand, Judgeford, New Zealand.

S-ar putea să vă placă și