WIND-RESISTANCE OF COMPOSITE STRUCTURAL INSULATED PANELS (CSIPS) by LI DONG NASIM UDDIN, COMMITTEE CHAIR JASON T. KIRBY TALAT SALAMA A THESIS Submitted to the graduate faculty of The University of Alabama at Birmingham, In partial fulfillment of the requirements for the degree of Master of Science in Civil Engineering BIRMINGHAM, ALABAMA 2012
139
Embed
Wind Resistance of Composite Structural Insulated Panels ... · wind-resistance of composite structural insulated panels (csips) by li dong nasim uddin, committee chair jason t. kirby
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
WIND-RESISTANCE OF COMPOSITE STRUCTURAL INSULATED PANELS
(CSIPS)
by
LI DONG
NASIM UDDIN, COMMITTEE CHAIR JASON T. KIRBY TALAT SALAMA
A THESIS
Submitted to the graduate faculty of The University of Alabama at Birmingham, In partial fulfillment of the requirements for the degree of
Master of Science in Civil Engineering
BIRMINGHAM, ALABAMA
2012
Copyright by Li Dong
2012
iii
WIND-RESISTANCE OF COMPOSITE STRUCTURAL INSULATED PANELS (CSIPS)
LI DONG
DEGREE: MS PROGRAM: CIVIL ENGINEERING
ABSTRACT
Extreme windstorms are frequent and destructive natural hazards in the United
States. Each wind disaster proves that traditional wood houses are too fragile to withstand
high air pressures. The proposed Composite Structural Insulated Panels (CSIPs) are made
of a low-cost, thermoplastic orthotropic glass/poly-propylene (glass-PP) laminate as the
face sheet and expanded polystyrene (EPS) foam as the core, with a high face sheet/core
moduli ratio.
In order to evaluate the wind-resistance of CSIPs, panels with cores of 16 kg/m3
(1 PCF) and 48 kg/m3 (3 PCF) density were tested at the University of Florida. Stepwise
and dynamic simulated wind pressure was generated by the High Airflow Pressure
Loading Actuator. The connections of the CSIPs were 2”×6” lumbers as used in
traditional constructions. Top and bottom lumbers were fixed to steel plates with bolts. In
stepwise windstorm tests, step-by-step pressure was applied to CSIPs; the highest
pressure was equivalent to a wind speed of 124.44 m/s (280 mph). The CSIPs wall
system demonstrated superior capacity except local debonding between the face sheet
and core; there were connection failures at lumbers in all but one sample. In dynamic
windstorm tests, cyclic pressure was used to simulate actual exposure. Only one sample
failed at the lumber connection, and there were minor cracks in the lumber connections
and local debondings in three of the eight samples. The intact condition of all face sheets
iv
and cores after tests indicate the excellent wind-resistance of CSIPs for structural wall
applications.
Corresponding finite element modeling was developed and initially validated, in a
dynamic windstorm test, within the first second for the 1 PCF core density CSIP. The
modeling was further investigated with longer durations of exposure. The results were
compared, and modifications were projected for future modeling to improve the
6.2 Further FE modeling and Experimental Results ............................................. 107 6.2.1 Stepwise Windstorm Test ........................................................................... 108 6.2.2 Dynamic Windstorm Test ........................................................................... 110
atmosphere), and service (connecting to the test chamber). The amount of air traveling
from the test chamber to the exhaust (intake) port was modulated by the air valve. The air
leakage through the specimen was equivalent to the difference between the exhaust rate
and intake rate. The air valve was connected to a rotary actuation system, which consists
11
of a Metronix ARS2340 servo-positioning controller, a Sumimoto Drive Technologies
CHF 6135 Y-11 11:1 drive reduction, and a Metronix SBL-T6-2900 brushless servo
motor. A two-pole resolver in the motor provided positioning feedback. The pressure in
the chamber was controlled by a customized Labview 8.5 on a National Instruments PXI
system through a 100 Hz PID controller that received feedback from a pressure
transducer attached to the “airbox.” The test chamber was a stiff 2.4 m × 2.4 m × 0.3 m (8
ft x× 8 ft × 1 ft) hollow steel frame box lined with 14-gauge galvanized sheet steel on
five sides. The test specimen was placed on the sixth side, as shown in Figure 3 and
Figure 4. The CSIP was vertically installed, and the air inside the chamber was increased
slowly, creating horizontally distributed air pressure on the panel. A transparent plastic
membrane covering the sixth side of the airbox and a red rubber pipe along the edge of
the box were used to seal the chamber to prevent leaking. As indicated in Figure 5, the air
pressure in the chamber was measured through the small hole by use of a thin pipe
connecting to the data acquisition and control system [35].
12
Figure 2 High airflow pressure loading actuator [35]
Note: From “Water penetration resistance of residential window and wall systems subjected to steady and unsteady wind loading” by Lopez C., Masters, F. J., Bolton S., 2011, Building and Environment,46, p. 1333. Copyright 2011 by Elsevier Ltd. Reprinted with permission.
13
Figure 3 Front view of HAPLA
Figure 4 HAPLA with sample ready
14
Figure 5 Pressure Transducer
2.3 Boundary Conditions
Due to the limits of the HAPLA testing system, both the top and bottom
connections used the same assembly. Figure 6 and Figure 7 show the joint configuration
in detail. A hot knife was used to remove foam at the ends.
When cutting the foam with hot knife, the CSIP could be placed only horizontally.
The foam melted down due to gravity, which could produce an unsymmetrical bonding
between the two face sheets and the core. As shown in Figure 8, one side of the face sheet
disbonded slightly before the tests. It is possible that, during tests, this disbonding caused
earlier debonding failures. More details are presented in Chapter 5. Disbonding is due to
defects in manufacture and debonding is developed during loading.
15
Figure 6 Panel installation
Figure 7 Connection detail sketch
Figure 8 Disbonding
16
As stated previously, the foam core of the CSIP was 139.7 mm (5.5’’). In order to
fit it into the gap, 2” × 6” southern yellow pine lumber was used. As shown in Figure 9
and Figure 10, the wood connection was similar to the traditional wood wall assembly,
which can speed up construction because no new techniques are needed.
Figure 9 Connection details
Figure 10 Connection details side view
17
At both ends, the face sheet and the lumber were nailed together with a gas gun,
as shown in Figure 11.
Figure 11 Fastener details
2.4 Instrumentation Setup
Six strain gauges were attached at quarter points of both sides, as shown in Figure
13. The laser deflection measurement system on the black tripod monitored the mid-span
and provided a precision of 0.0001 inch. The lumber on the tripod made a perfect
horizontal angle for the laser. Figure 13 is a front view of all the instruments immediately
before a test. The yellow straps were used to keep the panel from crashing out when there
was a sudden connection failure at a high pressure; it did not touch the panel during tests.
18
To achieve minimal air leakage, several C-clamps were used to fix the sample onto the
chamber. All the stepwise and dynamic windstorm tests described in the thesis were led
and performed by the Co-PI of this project Dr. Forrest J. Masters, and his graduate
student, George Fernandez, from the University of Florida.
Figure 12 Strain gauges deployment
19
Figure 13 Instruments setup
20
CHAPTER 3
STEPWISE WINDSTORM TEST
3.1 Target Load
According to ASCE 7-10 [36], the wind load was determined with an assumption
of a building occupancy category II, an exposure classification of C, a building width of
14.33m (47’), a length of 42.67m (140’), a CSIP exterior wall, and a flat roof with a mean
height of 2.44m (8’). The topography was assumed to be homogeneous. The proposed
pressure is shown in Figure 14. The pressure started at 0.24 kPa (5 PSF) with increments
of 0.48 kPa (10PSF) every 90 seconds.
Figure 14 Pressure history of stepwise windstorm test
Equation 1 from Section 27.3.2 in ASCE 7-10 [36] was used to find the
As stated above, the load data were collected from the aerial dynamic database in
NIST. In order to quantitatively prove the capability of the dynamic windstorm test, the
Rayleigh distribution was employed to provide the theoretical exposure time of the
dynamic load, which means fatigue load as well, in a 500-year return period [38]. The
Rayleigh probability distribution function is displayed here. ( ) = 1 − exp − ( ) (Eq.2)
Where:
P (V) = the probability of a wind gust of a given speed or less occurring at a given
moment
V = the wind speed under consideration V = the average mean wind speed for the location under consideration
Thus:
[1-P (V)]*Lifetime = Probable exposure time at or above a given wind velocity
The Rayleigh distribution is a special form of the Weibull probability distribution,
which is typically used to model the probability of occurrence for certain wind velocities,
with a shape factor of 2.0 and constant ratio of standard deviation to mean value. In this
case, a 500-year return period, a 170 mph design wind speed, and a mean annual wind
speed of 13.7 mph were used to determine the probable exposure time. The 170 mph
wind speed is the largest value in Figure 49. The predicted exposure times are displayed
in Table 12.
53
Figure 49 Model estimated 500-year return period 3 sec gust wind speeds at a height of 10 m in open terrain including wind modeling uncertainty [39]
Note: From “U.S. hurricane wind speed risk and uncertainty” by Vickery, P. J., Wadhera, D., Twisdale, L. A., Lavelle, F. M., 2009, Journal of Structural Engineering, 135, 3, p. 318. Copyright 2009 by ASCE Reprinted with permission.
54
Table 12 Rayleigh probability of exposure at or above given wind speed
• Because no evident permanent deformation was found after removing the CSIPs
from the frame, the materials work within elastic range.
• Except for a control problem with panel D-1102-3, which caused it experience
only thirteen stages of the dynamic windstorm tests, all of the panels completed
the pre-programmed pressure stages.
• Only one panel, D-1003-1, experienced a connection failure. The bottom lumber
cracked through longitudinally. Before the final increase, indicating plate failure,
the mid-span deflection was about 12 cm.
• Debonding was observed in the tests for both panels with 1 PCF core density and
one 3 PCF core density panel, D-1103-3. Although the bonding of CSIP with the
rigid core showed no defects on its sides and corners, and both connection
lumbers of some panels remained intact during the tests, the slope changes in the
deflection plot indicate there are debondings in the middle of the panel, which
were not visible.
• Panels D-0912-3, D-0921-3 and D-1102-3 were completely intact. No debonding
was evident on the sides and corners, and there was no crevice on the wood
plates.
• D-1025-3 and D-1101-3 had partial cracks on the bottom and top lumbers,
respectively, but neither had debondings.
• The mid-span deflection was similar for the two 1 PCF core density CSIPs, but it
varied considerably for the 3 PCF core density CSIPs, ranging from 4.2 cm to
14.8 cm. Some samples showing similar conditions after the tests had different
94
deflections. D-0921-3 and D-1102-3 had no obvious damage, but the former panel
had a mid-span deflection of 14.8 cm; the deflection for the latter was only 7 cm.
A reasonable explanation for the inconsistency is that more debondings in the
middle of the panel, which were not visible, developed in D-0921-3 than in D-
1102-3.
• Typically, the dynamic windstorm tests lasted for about 100 min for 1 PCF core
density panels and about 175 min for 3 PCF core density panels. The repeated
load cycles, especially stage 7 to stage 16, make this experiment an effective
check for fatigue load.
4.13 Summary
• Although the HAPLA system did not match the high-peak pressures in later
stages from 8 to 16 and the trough pressure in each stage, the control system
worked well in most load traces, as shown in figures of 100-second excerpts.
• In the dynamic windstorm tests, all materials of the CSIPs were strong and ductile
enough to function in the elastic range.
• The CSIPs with 1 PCF core density deflected more and were more prone to face
sheet debonding and connection failures; only slight damage occurred in a few of
the 3 PCF core density CSIPs.
• All CSIPs were able to bear an equivalent wind velocity as high as 92.84 m/s (208
mph). They have the potential to bear even higher wind loads and remain
structurally intact.
• Under repeated pressures as high as 3.83 kPa (79.97 PSF), the CSIPs
demonstrated effective fatigue resistance.
95
• In the dynamic windstorm tests, face sheet debonding was not a substantial
problem for 3 PCF core density CSIPs. Nevertheless, in future research,
nondestructive techniques may be used to determine debondings in the middle of
the panels.
• Debonding in the middle of the panels may account for the discrete mid-span
deflections.
• The connections performed well in the dynamic windstorm tests. Only one
connection failure occurred in CSIPs, D-1003-1.
• The deflection plots (Figure 96) proved the repeatability of the test.
96
CHAPTER 5
AS-BUILT CONDITIONS
During the tests, debonding between the face sheet and the core is caused
primarily by wrinkling of the face sheet in compression. But “disbonding,” a preexisting
condition due to defects in manufacture, will be covered in this chapter.
For an instance, Figure 94 shows one of four disbonding corners of CSIP S-0513-
3 before the stepwise test. Such disbonding is large enough to be considered. The
disbonding probably contributed considerably to the deflection, and may account for the
fact that curve S-0513-3 (Figure 50) is different from the other 3 PCF core density CISPs.
Figure 94 Corner disbonding
The connection is expected to model the realistic fixed condition in order to find
the wind resistance of the CSIPs. The ideal fixed boundary condition with no rotations
97
can hardly be achieved. Figure 95 shows the fixtures, including ½’’ Simpson strong ties
and bolts after a test. The warpage is insignificant, which means the fixed boundary
condition was essentially achieved.
Figure 95 Simpson strong tie
98
CHAPTER 6
FINITE ELEMENT MODELING
The commercial software package LS-DYNA was used to perform the finite
element analysis for CSIPs.
6.1 Initial Modeling Validation
Because the relatively long test duration and the time-sensitive calculation
characteristics of LS-DYNA, only the first one second of the dynamic windstorm test on
CSIP with 1 PCF core density was selected to validate the finite element modeling in
order to perform modeling efficiently.
6.1.1 Simulation Details
The dimensions of the CSIP are in Figure 96 and Figure 97. Through the finite
element modeling, material models, explicit and implicit methods, and element quantity
were investigated in order to find the model that generated the best matched deflection-
time curve. Because the debonding stress between the face sheet and the core was
unknown, it was assumed that the two components were bonded well through the test,
thus CONTACT_TIED_SURFACE_TO_SURFACE was used. The load was modeled as
a uniformly distributed surface load, and LOAD_SEGMENT_SET and SET_SEGMENT
were used to produce the time-changing pressure, which varied every 0.01 second. The
load history is seen in Figure 98. For simplicity, the wood connection was not considered
in the modeling, and the edge nodes were constrained for all six degrees of freedom
99
(Figure 99). The material properties used in the model are in Table 30 and Table 31. Due
to the long time of the experiment and the computation characteristics of LS-DYNA, only
data for the first second were used for the validation.
Table 30 Material properties of face sheet RHO EX EY EZ PRXY PRYZ PRXZ GXY GYZ GXZ
16 kg/m3
11035 MPa
11035 MPa
1030 MPa
0.11 0.22 0.22 1800 MPa
750 MPa
750 MPa
Table 31 Material properties of core
RHO EX PRXY
1783 kg/m3 1.2 - 1.5 MPa 0.25
Figure 96 Connection dimensions
100
Figure 97 Panel dimensions
101
Figure 98 Load history: Pressure - Time
Figure 99 Geometry
102
6.1.1.1 Material Models
MAT161 and MAT54 were evaluated for the 3.04-mm glass fiber / PP face sheet,
and MAT003, MAT012, and MAT063 were evaluated for the EPS foam core. The stress-
strain curve for MAT012 was obtained from Croop et al. [40]. The baseline material
models for the core and face sheet are MAT003 and MAT054, which, after several trials,
were determined to be the optimum types (see Figure 100).
Figure 100 Deflection - Time curves for different materials
Shell element was used instead of solid element because the element with an
aspect ratio closer to unity yields better results [41] and considering the cost of central
processing units (CPU). Although MAT161 was considered as a great fit for fiber
polymer materials, it did not show appropriate behavior in this modeling. MAT161 may
be better for modeling fiber failure, but the relatively low pressure did not cause any
103
damage to this model. Further, MAT161 is not applicable for 4-node shell element. Due
to the thin (3.04-mm) face sheet and the large dimensions of the model, the CPU cost
would be substantial if the element aspect ratio meets the unity requirement.
6.1.1.2 Explicit and Implicit Solver
The terms implicit and explicit refer to time integration algorithms. For explicit
analysis, the maximum time-step size is limited by the Courant condition. It is well suited
for dynamic simulations but can become prohibitively expensive in conducting static
analyses or those of long duration. For implicit analysis, the time-step size may be
selected by the user, but extensive efforts are required to form, store, and factorize the
stiffness matrix [42]. CONTROL_IMPLICIT_GENERAL was used to activate the
implicit solver. The comparisons among experiments, explicit and implicit, can be seen in
Figure 101. For the two methods, the results do not vary appreciably. However, the CPU
cost for implicit calculations was larger than that for the explicit method. Thus, the
explicit method was chosen for modeling.
104
Figure 101 Deflection – Time curve of explicit and implicit methods
For the implicit solver, the default iterative nonlinear solver used is the BFGS
method that employs a ‘Quasi-Newton’ method, in which the global stiffness matrix is
reformed only at each ILIMIT step. This default method often fails to converge when the
non-linearity of the problem grows or when there is substantial contact involved [43].
This specific modeling has a substantial amount of contact between the face sheet and the
core. In this case, as Suri Bala [43] recommends, a more expensive Full-Newton method
is used by setting ILIMIT = 1 to force LS-DYNA to reform the stiffness matrix at every
iterative step and to yield a more accurate estimation of stiffness. Further, the MAXREF
was set to 50 to allow a sufficient number of stiffness reformations before terminating or
reducing the solution time step.
105
6.1.1.3 Element Quantity
To balance the relationship between CPU cost and consistency of results, different
element numbers were investigated for this model. With different mesh sizes, 32
elements, 96 elements, 1128 elements, and 64843 elements were evaluated. Results for
the four modeling did not differ appreciably, and were consistent with the experiment
curve (Figure 102). Also the computation time for the modeling with 32 elements, 96
elements and 1128 elements were similar. Thus, it was appropriate to use the model with
1128 elements.
Figure 102 Deflection – Time curve of different element numbers
Although the model had dimensions of 2.4 m × 1.2 m × 0.14 m (8 ft × 4 ft × 5.5
in), the results were consistent with mesh size of 100 mm and element number of 1128.
106
The reason why this model is not sensitive to the element size may be due to its simple
geometry and load pattern.
Results
As all the validation modeling shows, a finite model with MAT054 for face sheet,
MAT063 for core, CONTACT_TIED_SURFACE_TO_SURFACE between face sheet
and core, and a mesh size of 100 mm with element number of 1128 were consistent with
the experiment using explicit solver. The variance is within an acceptable range (Figure
103). The deflection contour can be seen in Figure 104.
Figure 103 Comparison between experiment and FE modeling
107
Figure 104 Deflection contour
6.2 Further FE modeling and Experimental Results
After validation of the first one second of the CSIP with 1 PCF core density, the
modeling was further developed in stepwise and dynamic windstorm tests of longer
duration. In the following, fixed and simply supported conditions and material models
MAT003 and MAT063 were evaluated in a parametric study of 1 PCF core density
CSIPs in stepwise tests; only boundary conditions were investigated for 3 PCF core
density CSIPs in stepwise tests. In dynamic windstorm tests, all models were in fixed
condition, and different material models were investigated only for 1 PCF core density
108
CSIPs. Due to the lack of debonding-related data, debonding was not considered in
modeling.
6.2.1 Stepwise Windstorm Test
The pressure history in Figure 105 was the input to the load-modeling file. The
deflections from experiment and finite element modeling for all fifteen stages of CSIPs
with 1 PCF core density can be seen in Figure 10. There was little difference between
different boundary conditions, and MAT003 gave a slightly better result than MAT063.
Nevertheless, after the first three stages, there were large divergences between deflections
from the experiment and the finite element analysis.
Figure 105 Deflection comparisons of 1 PCF core density CSIPs in stepwise windstorm
test
109
With the load input described in Figure 14, the deflections of experiment and
modeling for 3 PCF core density CSIPs were obtained (Figure 106). Again, fixed and
simply supported conditions were investigated. From this plot, it can be seen that the
modeling matches well with the experiment for only the first two stages; the difference
between them is greater at later stages.
Figure 106 Deflection comparisons of 3 PCF core density CSIPs in stepwise windstorm
test
The total deflection of the mid-span is the sum of bending deflection and shear
deflection, which are described by the first and second terms, respectively, of the
following formula.
U
PLk
D
PLk sb +=Δ3
(Eq. 3)
110
In the formula, P is total load acting perpendicular to neural axis, L is span, D and
U are bending and shear stiffness, and bk and sk are bending and shear constants
depending on loading, boundary conditions, and points of interest. For example, if mid-
span is the point of interest, the boundary condition changes from fixed to simply-
supported, and other conditions stay the same. In this case, bk changes from 1/384 to
5/384. In Figure 105 and Figure 106, the difference between fixed and simply supported
boundary conditions was not substantial, indicating that the deflection due to bending
was small enough to be ignored.
6.2.2 Dynamic Windstorm Test
Because the load file of LS-DYNA had an input limit and also considering the
expensive CPU cost and large volume of result files, only the first 300 seconds were used
as load input for the modeling (Figure 48).
The deflections of experiment and modeling of dynamic windstorm test of 1 PCF
core density CSIPs are illustrated in Figure 107. The deflections of experiment and
modeling followed a similar up-and-down pattern, and the FE63 modeling curve on top,
which used MAT063, gave a less noisy curve. The peak values of experiment and FE
modeling, which used MAT003 as core material (as evaluated previously in initial
validation), were within an acceptable difference; the mean values of the two varied
considerably, especially after 100 seconds.
Figure 108 shows the deflections of experiment and modeling for CSIPs with 3
PCF core densities. The deflection of modeling is at the bottom of all curves. The
111
modeling matched well with D-1101-3 before 150 seconds; thereafter it may not be a
validated model due to the relatively large variance.
112
Figure 107 Deflection comparisons of 1 PCF core density CSIPs in dynamic windstorm test
0 50 100 150 200 250 3000
0.5
1
1.5
2
2.5
3Deflection vs Time
Time/s
De
fle
cti
on
/cm
FEEX0926EX1003FE63
113
Figure 108 Deflection comparisons of 3 PCF core density CSIPs in dynamic windstorm test
0 50 100 150 200 250 3000
0.5
1
1.5
2
2.5Deflection vs Time
Time/s
De
fle
cti
on
/cm
FEEX0912EX0921EX1005EX1101EX1102EX1103
114
6.3 Summary
Although the modeling did not match well with the experimental work, there are still
points that can be made:
• MAT003 gave smaller deflections than experiments. Consequently, MAT063 was
tried because of its realistic behavior in other engineering modeling [40], and
more importantly it gave a larger deflection than MAT003.
• In stepwise windstorm tests, the small differences between different boundary
conditions may indicate that the deflection due to bending is small enough to be
ignored and that the deflection due to shear dominates. This is reasonable because
the foam is weak in shear resistance, and debonding between the face sheet and
the core also contributes.
• In dynamic windstorm tests, the modeling with MAT003 as core matched well
with the experiment for 1 PCF CSIPs, although the modeling with MAT063 was
less noisy. More closely matched results with MAT063 may be achieved with
stress-strain data from the mechanical test of the foam.
• The modeling will be more realistic when debonding-related data are obtained
from future tests and included in the analyses.
115
CHAPTER 7
SUMMARY & CONCLUSION
7.1 Summary
Composite Structural Insulated Panels (CSIPs) are proposed as a substitute for
traditional wood-made structural insulated panels (SIPs), which are typically used for
building envelopes. Made of composite materials, the CSIPs possess various structural
merits, most of which have been established by laboratory tests that have been described
in previous statements. By addressing the increasing loss due to wind hazards in recent
years, this thesis concentrated on the wind resistance of CSIPs. Simulated wind pressure
was applied, in stepwise and dynamic ways, to full-scale CSIPs using the High Air
Pressure Load Actuator (HAPLA) system developed at the University of Florida. The
commercial finite element software, LS-DYNA, was used to evaluate the modeling
validation of the experiments.
In stepwise windstorm tests, the HAPLA system produced simulated air pressure
starting from 5 PSF with increments of 10 PSF up to 145 PSF with 90 seconds for each
pressure period. During tests, CSIPs ended with equivalent wind velocities ranging from
227 mph to 280 mph. This stepwise test was considered conservative because it is an
extremely rare case that a structure is subject to a large wind load for 90 seconds.
Although four of five specimens failed at their connections, one CSIP experiencing the
largest velocity of 280 mph showed no cracks at the connections, which demonstrates the
116
wind resistance potential of CSIPs. For all specimens, debonding was observed at most
corners. The mid-span deflections were significant in tests, and CSIPs with thicker or
denser cores should solve that problem. An assumption of wrinkling of the face sheet on
the compression side was made due to the decreasing compression strain values. Future
tests can evaluate this possibility.
In dynamic windstorm tests, the HAPLA system generated fluctuating wind
pressure. The input data were obtained from the Aerodynamic Database of National
Institute of Standards and Technology (NIST). The peak equivalent wind velocity ranged
from 130 mph to 208 mph, which is 22% greater than the largest wind speed in the 500-
year hurricane map based on uncertainty calculations. The Rayleigh distribution was used
to predict the probable exposure time during the 500-year period. This dynamic
windstorm test was considered conservative because the actual exposure time during
dynamic windstorm tests was larger than the predicted time from the Rayleigh
distribution. In dynamic windstorm tests, four of eight CSIPs experienced connection
failures. Three of the four were 3 PCF core density samples that had minor cracks; the
other was a 1 PCF core density sample, which had a crack through the lumber. Three of
eight samples experienced corner debondings, two of the three were in 1 PCF core
density panels. Because the deflection plots for the 3 PCF core density panels were not
consistent, an assumption was made that debondings appeared in the middle of the face
sheet during tests, which would not be noticed but could be evaluated by nondestructive
techniques.
The stepwise and dynamic windstorm tests validate the wind resistance of the
CSIPs as a single full-scale wall panel. The connection details, construction problems,
117
and the performance of CSIPs after the cut for the windows, doors, wire line, plumbing
etc. will be the focus of this project in the next phase.
The commercial finite element software LS-DYNA was used to perform finite
element analysis. Due to the expensive CPU cost and computing time, initial modeling
validation was performed for the first second of the dynamic windstorm tests of 1 PCF
core density CSIP. After several trials, a 4-node shell element and MAT054 for face
sheet, 8-node solid element and MAT003 for core,
CONTACT_TIED_SURFACE_TO_SURFACE, element size of 100 mm, and explicit
solver were considered to be a good fit. With the load input from stepwise windstorm
tests and the first 300 seconds of dynamic windstorm tests, further modeling validation
was performed. Due to the uncertain stress-strain curve of the foam core and the
unknown debonding stress between the face sheet and the core, the modeling merely
developed a similar plot pattern. The results from stepwise windstorm test modeling led
to an assumption that the deflection due to shear dominates the total deflection and that
the weak shear-resistance foam core and the weak bonding between the face sheet and the
core justified the assumption.
7.2 Conclusion
• The results of stepwise and dynamic windstorm tests prove that the full-scale
CSIPs are capable of resisting an equivalent wind velocity of more than 200 mph
without showing substantial structural damage.
• In the tests, debonding was a typical damage mode for CSIPs, but the wind-
resistance capacity of CSIPs did not decrease appreciably after debonding.
118
• CSIPs with 1 PCF core density deflected more at the mid-height of panels than
the ones with 3 PCF core density. Denser or thicker foam cores are expected to
decrease the mid-height deflection to more comfortable criteria.
• For future tests, nondestructive techniques are recommended to determine if any
debondings fall away from the edges of the panels. At present, this cannot be
visually observed.
• The finite element modeling indicates that the deflection due to shear is a primary
factor in the structural behavior of CSIPs.
• The finite element modeling gave similar patterns in deflection plots. More
mechanical properties of foam cores, which are expected to be found in future
tests, will likely make the modeling more consistent with experiments.
119
LIST OF REFERENCES
1. Insurance Information Institute. Available at, http://www.iii.org/facts_statistics/catastrophes-us.html/
2. Storm Prediction Center, National Weather Services, National Oceanic and Atmospheric Administration. Available at, http://www.spc.noaa.gov/climo/torn/fataltorn.html
3. Blake, E.S., Landsea, C.W., and Gibney, E.J. (2011): The Deadliest, costliest, and most intense United States Tropical Cyclones from 1851 to 2010 (and Other Frequently Requested Hurricane Facts). NOAA, Technical Memorandum NWS NHC-6.
4. Gurley, K, Masters, F. (2011) Post 2004 hurricane field survey of residential building performance. Natural Hazards Review, 12(4), 177-183.
5. Mehta K. C., Minor J. E., Reinhold T. A. (1983) Wind speed-damage correlation in Hurricane Frederic. Journal of Structural Engineering 1, 37-49.
6. Kareem, A. (1985) Structural performance and wind speed-damage correlation in Hurricane Alicia. Journal of Structural Engineering, 111(12), 2596-2610.
7. Guillermo F., Green R., Khazai B., Smyth A., Deodatis G. (2010) Field damage survey of New Orleans homes in the aftermath of Hurricane Katrina. Natural Hazards Review, 11, 7-18.
8. Alwin, H. Z. (2002) Development of a method to analyze structural insulated panels under transverse loading. Master Thesis, Washington State University.
9. Vaidya, A. S. (2009) Lightweight composites for modular panelized construction. Ph.D. Dissertation, University of Alabama at Birmingham.
10. Mousa, M. (2007), Optimization of Structural Panels for Cost-Effective Panelized Construction. MS Thesis, CCEE Department, University of Alabama at Birmingham.
120
11. Khotpal, A. (2004) Structural Characterization of Hybrid Fiber Reinforced Polymer (FRP)-Autoclaved Aerated Concrete (AAC) Panels. MS Thesis, CCEE, Department, University of Alabama at Birmingham.
12. Shelar, K. (2006) Manufacturing and Design Methodology of Hybrid Fiber Reinforced Polymer (FRP)-Autoclaved Aerated Concrete (AAC) Panels and its Response under Low Velocity Impact. MS Thesis, CCEE Department, University of Alabama at Birmingham.
13. Uddin, N., and Fouad, H. (2007).Structural Behavior of FRP Reinforced Polymer-Autoclaved Aerated Concrete Panels. ACI Structural Journal, 104(6), 722-730.
14. Mousa, M., Uddin, N. (2009) Experimental and Analytical Study of Carbon Fiber-Reinforced Polymer (FRP) /Autoclaved Aerated Concrete (AAC) Sandwich Panels. Journal of Engineering Structures, 31, 2337-2344.
15. Nguyen, T. T. (2009) Analysis and design of innovative composite storm shelter against wind hazards. M.S. Thesis, University of Alabama at Birmingham.
16. Mousa, M. (2011) Composite structural insulated panels (CSIPs) for hazards resistant structures. Ph.D. Dissertation, University of Alabama at Birmingham.
17. Hartness, T., Husman, G., Koenig, J., Dyksterhouse, J. (2001) The
characterization of low cost fiber reinforced thermoplastic composites produced by the DRIFT process. Composites - Part A: Applied Science and Manufacturing, 2(8), 1155-1160.
18. Company literature: U.S. Liner Company – A Division of American Made, LLC, 19 Leonberg Road Cranberry Township, PA 16066 U.S.A.
19. Company literature: Universal Foam Products 16 Stenersen Lane, Suite 4B Hunt
Valley, MD 21030 U.S.A.
20. Annual book of ASTM standards C393-00 (2000), standard test method for flexural properties of sandwich construction, 100 Barr Harbor Drive, West Conshohocken, PA 19428-2959, United States
21. Vaidya, A. (2007) Thermoplastic composite structural insulated panels (CSIPS)
for building construction. Proceedings of Asia-Pacific Conference on FRP in Structures (APFIS 2007), 2007 International Institute for FRP in Construction
22. Vaidya, A., Uddin, N., Vaidya, U. (2010) Structural characterization of composite structural insulated panels for exterior wall applications. Journal of Composites for Construction, 14(4), 464-469.
23. AnnuaconduDrive
24. Mousinsula
25. MouspanelsNovem
26. MinimAmer
27. Mousinsula
28. Comp43782
29. Comp01749
30. CompBoule
31. Cente2003.
32. AnnuaSurfacHarbo
33. AnnuaSurfacWest
34. AnnuaSpecifHarbo
35. Lopezresideloadin
al book of ucting streng, West Cons
a, M. A., ated wall pan
a, M., Uddis. Journal mber 2010
mum Desigican Society
a, M., Uddinated floor pan
pany literatur2 U.S.A.
pany literatu9 U.S.A.
pany literatuevard, P.O. B
r for Comp1.
al book of ce Flammabor Drive, We
al book of ce Burning Conshohock
al book of fic Optical or Drive, We
z, C., Masteential windong. Building
ASTM stangth tests of shohocken, P
Uddin, N. nels. 32(2), 7
in, N. Deboof Reinforc
gn Loads fy of Civil En
n, N. (2011)nels. Advanc
re: Hollinee
ure: Avtec I
ure: Great Box 2200, W
posites Manu
ASTM stanility of Mate
est Conshoho
ASTM stanCharacterist
ken, PA 1942
ASTM stanDensity of
est Conshoho
ers, F. J., Bow and wal
and Environ
ndards E-72panels for b
PA 19428-29
(2011) Gl766-772.
onding of coced Plastics
for Buildingngineers.
) Flexural beced Compos
Glass Fiber
Industries, 9
Lakes CheWest Lafayett
ufacturing F
ndards E-162erials Using ocken, PA 1
ndards E-84tics of Build28-2959, Un
ndards E-662f Smoke Geocken, PA 1
Bolton, S. (l systems snment, 46(7)
2-05 (2005)building con959, United
obal buckli
omposites ss and Comp
g and Othe
ehavior of fusite Material
rs, 9702 Iron
9 Kane Ind
emical Corpte, Indiana 4
Final Report
2-95 (1995)a Radiant H9428-2959,
4-10 (2010),ding Materinited States.
2-95 (1995)enerated by9428-2959,
(2011) Watsubjected to), 1329-1342
). “Standardnstruction” States
ing of com
tructural insposites, 29(
er Structure
ull-scale comls, 20(6), 547
n Point Rd S
dustrial Driv
poration, O47906-0200
t. (2003) FT
), Standard Heat Energy
United State
, Standard Tals, 100 Ba
), Standard y Solid Mat
United State
ter penetratio steady and2.
d test metho100 Barr H
mposite struc
sulated sand(22), 3380-3
es. ASCE
mposite struc7-567.
SE Shawnee
ve, Hudson,
ne Great L
TA-AL-26-7
Test MethoSource, 100es.
Test Methoarr Harbor D
Test Methoterials, 100 es.
ion resistancd unsteady
121
od of Harbor
ctural
dwich 3391,
7-05,
ctural
e, OH
MA
Lakes
7001-
od for 0 Barr
d for Drive,
od for Barr
ce of wind
122
36. Minimum Design Loads for Building and Other Structures. ASCE 7-10, American Society of Civil Engineers.
37. Sharaf, T., Shawkat, W., Fam, A. (2010) Structural performance of sandwich wall panels with different foam core densities in one-way bending. Journal of Composite Materials, 44(19), 2249-2263.
38. Frankl, B. A. Structural behavior of insulated precast prestressed concrete sandwich panels reinforced with CFRP grid. Ph.D. Dissertation, North Carolina State University, 2008.
39. Vickery, P. J., Wadhera, D., Twisdale, L. A., Lavelle, F. M., (2009) U.S.
Hurricane Wind Speed Risk and Uncertainty, Journal of Structural Engineering, 135(3), 301-320.
40. Croop, B., Lobo, H. (2009) Selecting Material Models for the Simulation of Foams in LS-DYNA. 7th European LS-DYNA Conference Proceedings.
41. Bhavikati, S.S. (2005), Finite Element Analysis. New Age International (P) Ltd., Publishers, New Delhi, India.
42. LS-DYNA Version 971 Keyword User’s Manual, 2006.
43. http://blog2.d3view.com/ Website visited on Dec 15, 2011.