Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

A Simplified Wood Sheathing Connection Model using

ANSYS
Andrew Blasetti
Villanova University
Rebecca Hoffman
Villanova University
David W. Dinehart
Villanova University

Abstract
This paper presents a methodology to model the complex hysteretic behavior of wood sheathing
connections using basic elements found in ANSYS. By this methodology, the single, specially
programmed elements commonly used to model these connections are replaced with a combination of basic
elements that, when assembled, produce a similar dynamic response. This both simplifies the modeling
process and adds versatility to the model, allowing it to be modified for different connections as needed.
The proposed modeling technique may be applied to nearly all structural components with hysteretic
behaviors. A typical nailed sheathing connection has been successfully modeled.

Introduction
In 1994, the Northridge, California earthquake created over $20 billion in damage to wood frame
construction. (Pardoen et. al., 2003) Shear walls play a critical role in preventing the total destruction of
residential buildings in earthquake regions as the lateral load resisting systems in wood frame structures
consist almost entirely of shear walls sheathed with plywood or OSB around the perimeter of the building.
Figure 1 shows a typical plywood shear wall framed with 50.8 x 101.6 mm [2 x 4 in.] studs, spaced at
406.4 mm [16 in.] on center.

Figure 1. Typical 8’ x 8’ shear wall

The seismic response of shear walls is governed by their ability to resist cyclic or repeated lateral
load. The wall resists the lateral load by racking; the stud frame shears while the sheathing rotates. The
fasteners attaching the frame and the sheathing bend and elongate. The lateral load capacity of the wall is
governed by the load-displacement behavior of the fasteners. Under cyclic loads, the fasteners are
repeatedly deformed in opposite directions thereby enlarging the fastener hole in the wood. This process
leaves the wall in a more compliant state for the next deformation. For this reason, sheathing-to-frame
connections have been identified as the primary source of degrading performance and are now reinforced
via reduced nail spacing, stronger nails, and thicker panels (American Wood Council, 2001). The resulting
wall, however, will not be free of damage following a seismic event. A detailed analysis shows that despite
the overall increase in strength, there is still a great deal of deterioration of the wall during dynamic cyclic
testing.
In recent years, finite element analyses have gained popularity amongst wood researchers. (Filiatrault
1990; Dolan and Foschi 1991; White and Dolan 1995; Tarabia and Itani 1997) These analyses include the
hysteretic behavior of the fasteners, bearing of sheathing panels upon one another, out-of-plane bending,
the stiffness of transverse walls, and the nonlinear behavior of the frame joints. All of these programs use a
complex hysteretic model representing the sheathing-to-stud connection, capturing the degradation in
stiffness and pinching that occurs with cyclic loading. In 2001, Folz and Filiatrault verified the capabilities
of CASHEW, a finite element model used to perform a cyclic analysis of wood shear walls. This program
utilizes nonlinear hysteretic springs to model the sheathing connections. The hysteretic behavior is defined
via ten input parameters. Though accurate, modifying the connection model becomes a challenge. One
could extract the new parameters from a set of data and input them into the model, if the data followed the
same trend as the CASHEW connection model. But if this were not the case, one would have to reprogram
the model with more appropriate hysteretic behaviors.
Because there are numerous connections in one shear wall and because of the multiple cycles of
the CUREE testing protocol, finite element models must incorporate accurate but less time-consuming
connection models than those currently available. Since the computation time-to-accuracy ratio seems to
increase nonlinearly, it is necessary to find a model that balances accuracy and practicality. This paper
presents a methodology to model complex hysteretic behaviors using a combination of basic elements
found in virtually all FEA software packages, rather than a single element requiring specific software and
advanced programming expertise. The end result is an accurate and efficient multi-linear hysteretic
sheathing-to-frame connection model that can be incorporated into a larger finite element model (i.e. a
shear wall). The hysteretic behavior is easy to modify without having to rewrite the software, making it
extremely versatile. It is also easy to replicate using other software because the required elements are
common among FEA programs and are easily defined. The proposed methodology not only applies to
typical nailed connections, but also can be applied to any hysteretic behavior including those with viscous
properties.

Procedure
Typical Sheathing-to-Stud Connection
Wood-framed shear walls dissipate energy though inelastic deformation of the sheathing-to-stud
connections, resulting in nail fatigue, nail withdrawal, and deterioration of the sheathing. (Dinehart &
Shenton, 1998) As sheathing connections are damaged, the wall degrades causing a reduction in stiffness
and energy dissipation with each passing cycle of displacement. Virtually all nailed wood connections
exhibit pinching in their hysteretic responses, representing a loss of energy dissipation. A nailed sheathing-
to-stud connection has three primary modes of dissipating energy: crushing of the wood, inelastic
deformation of the nail, and friction between the components of the connection (nail, stud and sheathing).
The last two are present throughout almost the entire range of motion. Crushing of the wood, however,
only occurs during primary cycles of loading, when the cycle displacement exceeds the maximum
displacement of all previous cycles. Hence, the secondary cycles that follow (those with less than or equal
displacements than the preceding primary cycle) cannot dissipate energy via crushing of the wood,
resulting in a pinched region on the hysteresis plot.
In order to quantify this behavior, Foley (2005) conducted a series of experiments on typical
nailed sheathing connections. The test setup is shown in Figure 2. The connection specimens were made
from 50.8x101.6 mm [2 x4 in.] Spruce-Pine-Fir (SPF) lumber (177.8 mm [7 in.] long) for the stud, a 76.2 x
203.2 mm [3 x 8 in.] section of 11.1 mm [7/16 in.] thick oriented strand board (OSB) for the sheathing,

Connects to Upper Grip

Load Cell

Upper Connection

Vice
APA approved sheathing Vertical Restraint Plate

Figure 3.7: Test Setup No. 2 SPF Stud


Roller Bearings
B
Threaded Rods
Backing Plate
Lower Connection
Connects to Lower Grip

Figure 2. A Sheathing Connection Test Setup

and nailed together with an 8d common nail. The connections were subjected to the recommended CUREE
displacement protocol (Fonseca et. al., 2002) with a reference displacement of 4.32 mm [0.17 in.] and a
frequency of 0.5 Hz. The dynamic response for a typical connection and its displacement protocol are
presented in Figure 3. Connection specimens were tested under cyclic and static loading. Results within
each category were very similar, so the connection model was derived from the average of the results rather
than just one specimen.
1.5
20

Displacement (mm) 0
1.0
0 120

-20
Time (sec)
0.5
Force (kN)

0.0
-20 -15 -10 -5 0 5 10 15 20

-0.5

-1.0

-1.5
Displacement (mm)

Figure 3. Dynamic Response of a Typical Nailed Sheathing Connection. The


Recommended CUREE Displacement Protocol for Connections is given in the
inset.

Analysis
Sheathing Connection Model
Traditionally, modeling the behavior shown in Figure 3 required a single element with
programmed response curves for loading, unloading and reloading. This methodology divides the response
into several portions, each modeled by its own element as seen in Figure 4. The result is a collection of
finite elements that work together to produce the total response of a sheathing connection. Using ANSYS,
this connection was modeled as two elements in both the X and Y directions for a total of four elements.
One of the two elements models secondary cycle response; the other models the additional response seen in
the primary cycles (cycles with displacements exceeding the maximum previously observed displacement).
All elements were modeled using COMBIN40, a connection element featuring a combination of springs,
damper, friction slider and gap. This element combines the sub-elements into one, thus reducing the total
number of elements. A schematic of the finite element connection model is presented in Figure 5. Each
element can translate in one direction (X or Y), and consists of two elements and two nodes.
The first element models the majority of the secondary cycles – the pinched region. It features a
spring (0.00875 kN/mm [50 lb/in]) and two-way friction slider (0.133 kN [30 lb]) in parallel. The 0.00875
kN/mm [50 lb/in] spring represents the slope of the pinched region and the 0.133 kN [30 lb] friction slider
represents half the thickness of the pinched region. This element models the minimum stiffness and energy
dissipation of a secondary cycle. The numerical values were taken from data obtained through experiments
of connections subjected to dynamic loading as seen in Figure 4.
P

Contact
Region
0.667 kN
0.133 kN
0.00875 kN/mm

Pinched Region

0.674 kN/mm

0.149 kN/mm

Figure 4. Input for Nailed Sheathing Connection Model

Node Fixed to Node Fixed to


Framing Sheathing

0.133 kN

Pinched Region

0.00875 kN/mm

Frame Sheathing

0.525 kN/mm 0.667 kN

Contact Region

0.140 kN/mm

Figure 5. Unidirectional Sheathing Connection Model


The second element models the behavior of the connection as the nail comes in contact with the
wood. It features a spring (0.525 kN/mm [3000 lb/in]) in series with a one-way friction slider (0.667 kN
[150 lb]), all of which is in parallel with another spring (0.140 kN/mm [800 lb/in]). It also features a gap
such that the connection offers no resistance when the gap is open, but when closed it engages the wood in
bearing. Initially, the gap is closed (no crushing of the wood). The friction slider creates a transition from
elastic stiffness (0.674 kN/mm [3850 lb/in]) to inelastic stiffness (0.149 kN/mm [850 lb/in]) of the
connection caused by crushing of the wood. The slider also allows the gap to open after the element
compresses. As the gap increases, the transition occurs at larger displacements for subsequent cycles.
With these values, the transition will occur once the element is compressed (with the gap closed) for 1.27
mm [0.05 in.] of displacement: [0.667 kN] / [0.525 kN/mm] = [1.27 mm]. This varying stiffness models
the contact between the nail and the wood, and was derived from static tests of nailed sheathing
connections. Note these values are specific to the connection tests conducted by Foley. A similar
breakdown of the data from other connection types or tests can be input into this element. Figure 6
illustrates the response of the model as the framing node moves relative to the sheathing node. The total
processing time for this connection model was less than 30 seconds for the entire range of motion with 100
sub-steps per cycle and a 2.40 GHz processor.

1.5
20
Displacement (mm)

0
0 120 1.0

-20
Time (sec)
0.5
Force (kN)

0.0
-20 -15 -10 -5 0 5 10 15 20

-0.5

-1.0

-1.5
Displacement (mm)

Figure 6. Dynamic Response of the Connection Model

The response of the connection model for any given cycle is a function of current cycle
displacement and previous maximum displacement. Therefore, it is assumed that the response is not a
function of damping or velocity. It is also assumed that the connection never fails, but rather continues to
provide resistance even at very large displacements. However, the connection model, ultimately, is part of
a larger wall model, and while the wall testing protocol requires the wall to be displaced up to 72.6 mm
[2.86 in.], and the connection model displaces considerably less. Thus, this connection model is applicable
to the entire range of motion seen in standard dynamic wall testing.
Analysis Results & Discussion
Verification of the Connection Model
Qualitatively, the response of the connection model compares quite well to that of an actual
sheathing connection. Pinching of the hysteresis plot is clearly represented, as is the reduction in area from
primary cycles to secondary cycles. There was assumed to be no degradation between consecutive
secondary cycles in order to simplify the model. Hence, unlike the test data of Figure 3, the model (Figure
6) gives only one visible hysteresis loop for each set of secondary cycles. This assumption was based on
the very small degradation between consecutive secondary cycles (Foley, 2005) and was necessary because
this model was to be implemented into a much larger finite element model consisting of 172 sheathing
connections. Another qualitative characteristic missing from the connection model is the smooth
transitions between pinched, loading, and unloading regions of the hysteresis plot. Again, this could have
been achieved, but would have required additional elements and/or elements with nonlinearities which
translate to additional computation time. This change would have had minimal effect on the shear wall
model response and was deemed unnecessary.
Three nailed sheathing connection specimens were tested and their average energy dissipation and
stiffness for each cycle of displacement was compared to those of the connection model. Values for energy
dissipation are compared in Figure 7 and those for stiffness are compared in Figure 8. All results are
plotted against cycle number. In accordance with the recommended CUREE protocol with a reference
displacement of 4.32 mm [0.17 in.], cycle 14 corresponds to a cycle displacement of 0.432 mm [0.017 in.],
cycle 35 corresponds to 4.32 mm [0.17 in.], and cycle 50 corresponds to 15.1 mm [0.595 in.].

Exp Avg Model


18

0.4
16

14 0.3
Energy Dissipation (kN-mm)

12
0.2

10

0.1
8

6 0
0 5 10 15 20

0
0 5 10 15 20 25 30 35 40 45 50

Cycle

Figure 7. Energy dissipation comparison between experimental data and connection


model
Exp Avg Model
2.5

2
Stiffness (kN/mm)

1.5

0.5

0
0 5 10 15 20 25 30 35 40 45 50

Cycle

Figure 8. Stiffness comparison between experimental data and connection model

Though the trends are the same, the values of energy dissipation and stiffness for the model do not
cohere with all cycles of the data. Energy dissipation of the connection model was lower than the
experimental data for all cycles, while its stiffness was only lower at cycles of very small displacement.
This does not imply that the connection model should not be used in the shear wall model, because not all
sheathing connections behave like those that were tested experimentally. Therefore, when using only one
type of connection model in a shear wall, it is important that the connection model represent the behavior of
all sheathing connections as much as possible.

Conclusion
This modeling methodology offers great potential as a simple alternative to the programming-
intensive techniques currently employed by the engineering community. The comparison shows how
accurate predictions can be made using a combination of basic linear elements. Not only can this
methodology be applied to wood sheathing connections; any connection, frame, wall, or other structural
element with or without hysteretic properties can be modeled using these basic element. Based on the
results of this study, the following conclusions regarding the proposed finite element modeling
methodology can be made:
1. Hysteretic behaviors can be accurately modeled using a combination of basic elements common to
virtually all commercial finite element analysis software, rather than using single complex
nonlinear elements.
2. A method has been presented for estimating the connection properties for various types of
sheathing connections based on experimental connection data. This methodology can be applied
to all types of wood sheathing-to-frame connections and, therefore, wood-frame shear walls. Its
versatility permits changes to the connection behavior as well as material properties. It also allows
the engineer to explore modifications to wood shear wall design in an effort to improve seismic
performance of wood-frame structures.
3. In addition to wood-frame shear walls, this methodology can be applied to any structural
component with hysteretic behavior.

Acknowledgments
This work was funded by a grant from the National Science Foundation (Grant No. CMS-
0229724). The authors would like to thank David Foley, a graduate of Villanova University, as well as
Adrienne Johnston, Dr. Harry Shenton, III and Gary Wenczel of the University of Delaware for their efforts
and expertise in conducting the experiments.

References
American Wood Council. (2001). Manual for Engineered Wood Construction. American Forrest and Paper
Association, Washington, DC.
Dinehart, D. W., and Shenton, H. W. III (1998). “Comparison of Static and Dynamic Response of Timber
Shear Walls.” Journal of Structural Engineering, ASCE, v124, n6, p686-695.
Dolan, J. D., and Foschi, R. O. (1991). “Structural analysis model for static loads on timber shear walls.”
Journal of Structural Engineering, ASCE, v117, n3, p851–861.
Filiatrault, A. (1990). “Static and dynamic analysis of timber shear walls.” Canadian Journal of Civil
Engineering, Ottawa, v17, p643–651.
Foley, D. (2005). An Experimental Investigation of Viscoelastic Material in Wood Connections and
Shearwalls. Master of Civil Engineering Thesis. Villanova University, Pennsylvania.
Folz, B. and Filiatrault, A. (2001). “Cyclic Analysis of Wood Shear Walls.” Journal of Structural
Engineering, ASCE, v127, n4, p433-441.
Fonseca, F.S., et. al. (2002). “Nail, Wood Screw, and Staple Fastener Connections.” CUREE Publication
No. W-16.
Pardoen, G.C. et. al. (2003). “Testing and Analysis of One-Story and Two-Story Shear Walls Under Cyclic
Loading.” CUREE Publication No. W-25.
Tarabia, A. M., and Itani, R. Y. (1997). “Static and dynamic modeling of light-frame wood buildings.”
Computer and Structures, v63, n2, p319–334.
White, M. W., and Dolan, J. D. (1995). “Nonlinear shear-wall analysis.” Journal of Structural Engineering,
ASCE, v121, n11, p1629–1635.

You might also like