Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Engineering Structures 56 (2013) 1249–1261

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Numerical evaluation of tsunami debris impact loading on wooden


structural walls
Anisa Como, Hussam Mahmoud ⇑
Department of Civil and Environmental Engineering, Colorado State University, Fort Collins, CO, USA

a r t i c l e i n f o a b s t r a c t

Article history: Water-based disasters such as tsunami, hurricane, and flood can cause significant structural damage to
Received 27 March 2013 coastal areas and ultimately result in human and socioeconomic losses. Recent studies have highlighted
Revised 12 June 2013 the difficulties associated with estimating tsunami impact loads on residential structures with greater
Accepted 14 June 2013
challenges pertaining to estimating the debris impact loads. The main challenges are associated with
Available online 31 July 2013
large variations in debris mass, shape, velocity, and impact angle. According to FEMA P-646 [1], although
impact of floating debris is required to be considered by the codes, the added mass of water behind the
Keywords:
debris and/or the potential for damming when debris is blocked by structural elements are ignored. Addi-
Debris impact
Tsunami load
tionally, there is a lack of 3D numerical models that can accurately predict debris impact loads. Due to the
Wood panel three-dimensional dynamic nature of the problem, coupled with highly variable factors affecting debris
ABAQUS loads, a 3D finite element (FE) analysis is necessary to determine the debris impact on structures as accu-
3D FEM rately as possible. To date, one of the major obstacles in modeling tsunami impact on structures is the
Dynamic load intense computational demand linked with such models. This paper presents the formulation of an effi-
cient 3D FE model with computational fluid dynamic (CFD) capabilities to study debris impact on interior
and exterior wood structural panels. Several important parameters such as initial water volume, water
velocity, debris shape, and debris density are varied throughout the analysis in order to determine the
individual effect of each parameter as well as the overall effect on wood structural walls. The results
for an interior wall panels showed that contribution from water height and velocity is more significant
the debris mass in the impact force as water height and velocity increase. Additionally, as the water
height increases, the impact force due to debris decreases for an interior panel when debris mass is con-
stant. An exterior panel is significantly closer to a rigid structure, therefore, an increase in impact force
due to debris is observed with the increase in water height and debris mass. The results can be used
by engineers to estimate debris impact load on wood panels with similar dimension and material
properties.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction other related causes such as fire following earthquake or injuries


and chronic illnesses [3].
Damage to structures due to tsunami events as well as other Other water-based natural disasters, such as hurricanes, have
similar water-based natural disasters such as hurricanes can be also have been the cause of major losses. One of the deadliest
very significant and can result in major economic, social, and hu- and most destructive in the history of the United States is the
man losses. The events are typically triggered by a continental 2005 hurricane Katrina, which exceeded 1800 fatalities and re-
earthquake in which an oceanic plate slides beneath a continental sulted in direct economic losses of over $125 billion [4]. A more re-
plate and are often more devastating than the earthquakes them- cent water-based event is the 2012 hurricane Sandy, which
selves. In the recent 2011 Tohoku earthquake/tsunami in Japan, a resulted in over $50 billion in direct economic losses and over
significant part of the damage resulted from the tsunami in addi- 140 fatalities [5]. Furthermore, the displaced population for Katrina
tion to the main earthquake. Specifically, the direct economic and Sandy exceeded 1,000,000 and at least 100,000, respectively
losses due to the Tohoku tsunami event were approximately [6]. It is clear that the socioeconomic impacts from these types of
$210 billion [2,3]. Moreover, roughly 93% of total fatalities resulted events can be disastrous; therefore, it is imperative to improve fun-
from the tsunami event rather than the main seismic event or damental understanding of the loading demands from such events
so that the residential and other structures can be improved and
optimized to better resist the expected demand.
⇑ Corresponding author. Tel.: +1 970 491 6605. Recent events have emphasized the need for more accurate
E-mail address: hussam.mahmoud@colostate.edu (H. Mahmoud). estimation of tsunami loads on structures. Recent progress has

0141-0296/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2013.06.023
1250 A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261

been made in the direction of predicting tsunami structural loads, 2. Background


which can generally be categorized into static and dynamic loads.
Static loading includes hydrostatic and buoyancy forces, while dy- 2.1. Tsunami loads
namic loading comprises hydrodynamic (drag), surge, and debris
impact loading [7]. While the static loads acting on structures Tsunami loads can be generally categorized into static and dy-
can be determined with relatively high accuracy, there are signifi- namic loads. Static loading includes hydrostatic and buoyancy
cant challenges regarding the estimation of loads due to dynamic forces, while dynamic loading involves hydrodynamic (drag),
effects. Due to the dynamic nature and multi-dimensional and surge, and debris impact loading [7]. While the static loads acting
highly variable factors associated with debris loads, a 3D FE model on structures can be determined with high accuracy, there are still
is necessary to determine the debris impact on structures as accu- challenges in estimating loads due to dynamic effects. Magnitude
rately as possible. Physical impact caused by debris can be an and load applications depend mainly on three parameters: inunda-
important contributor to the overall tsunami damage, which can tion depth, flow velocity, and flow direction [7]. These three
result in significant damage or even complete collapse of struc- parameters are dependent on tsunami wave height and period,
tures, as shown in Fig. 1. coastal topography, and roughness of the coastal inland [7].
This research study focuses on wood panels since they are FEMA 55 Coastal Construction Manual [9] addresses the design
extensively used for the construction in residential houses in for tsunami loads; however, most of the design guidelines are
coastal regions and the majority of damage caused during the based on extrapolations from flood event design procedures. How-
2011 Tohoku tsunami event was to wood structures. The aim of ever, there is a significant difference between tsunami and non-
this study is to evaluate the response of interior and edge wall pan- tsunami events in terms of water velocity, as illustrated in Fig. 2.
els in a typical wood residential construction under several varying According to FEMA P-646 [1], though hurricane storm surges
parameters. The parameters included water height, water velocity, and tsunami inundation both result in flooding, the flooding char-
debris length, and debris density. acteristic from the two events can differ significantly. Hurricane
storm surges or typical wind-generated water waves typically
inundate areas for a longer duration of several hours with repeated
pounding action, while tsunami inundation happens in a matter of
tens of minutes with rapid changes in water levels [1]. Further-
more, historic observations from tsunami events show that tsu-
nami behavior and characteristics can vary from event to event
and it is extremely difficult to derive any expected behavior from
intuition or common knowledge, and the time characteristics be-
tween tsunami and other coastal events is the main reason for
these differences [1]. As a result, extrapolation of storm surge
equations such as in hurricane events to tsunami events can cause
significant errors and it is not always applicable. Though tsunami
loads differ from other coastal events, occasionally the tsunami
runup can be characterized as a gradual rise and fall of water
(i.e., surge flooding), especially when a long-wavelength, leading-
elevation, and far-source-generated tsunami attacks land on a
Fig. 1. Structural damage due to debris impact during the Tohoku tsunami [7].
steep slope [1].

Fig. 2. Tsunami design versus stillwater depth design [10].


A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261 1251

2.2. Previous studies impermeable vertical wall, which signifies that all the water re-
turns back [1]. The ratio of small-scale experiments was approxi-
The impact force from waterborne debris such as floating drift- mately 1:100, which could introduce considerable scaling errors.
wood, lumber, boats, shipping containers, automobiles, and even From these studies, it was concluded that forces due to tsunami
buildings can be a dominant cause of structural damage, but it is run-up were higher than wave forces due to non-tsunami run-up
difficult to estimate this force accurately [1]. After some recent tsu- [1].
nami events such as the Indian Ocean tsunami in 2004 and Tohoku Haehnel and Daly [1,18] studied the maximum impact force of
in 2011, research efforts have intensified in order to determine im- logs on floodplain structures by conducting small-scale and large-
pact loads due to debris and improve design guidelines. Experi- scale experiments. The experiments consisted of colliding wood
mental efforts have been primarily confined to small scale, while logs with structures using flume and test basing laboratory facili-
numerical models have been limited due to the high computa- ties [1]. Because a towing tank was used in the basin laboratory
tional demand. Some experimental studies regarding impulsive experiments, water was stationary during the experiments; there-
loads on structures as well as previous experimental and numerical fore, debris was not propagated by the water [1,18]. This can intro-
studies regarding estimating debris impact loads are summarized duce inaccuracies for the same reason that in-air tests might
in the following paragraphs. introduce errors; the dynamic water effect is ignored. A single-de-
Significant work, especially experimental studies has been con- gree-of-freedom model of collision was also developed between
ducted in the recent years to investigate the effect of tsunami loads the logs and a rigid structure and the impact loads were estimated
on vertical walls and structures. Ramsden [11] extensively studied using three different formulations: Contact-Stiffness, Impulse-
impulse forces on vertical walls due to several different wave Momentum, and Work-Energy [18]. The resulting impact force
types. The study showed no significant initial impact forces in was a function of debris mass, impact velocity and effective stiff-
dry-bed surges, but an ‘‘overshoot’’ in force was observed in bores ness of the collision between the object and structure. The formu-
that occur once the site is initially flooded [1,11]. lation does not depend on structure properties, as long as the
More recent work was conducted by Arikawa [12] on surge structure is considered to be rigid. For cases where the rigidity
front tsunami force using physical model experiments. The study assumption is valid, the three developed formulations provided
focused on concrete and wooden walls, and different failure modes very consistent results.
were observed for different wall strengths. Wooden wall failure In terms of numerical models, though some attempts have been
was complete, while partial failure of concrete walls was observed. made at modeling impact loads due to debris, to the authors’
Full-scale tests on idealized typical coastal construction woo- knowledge, there are not any available peer-reviewed numerical
den walls were conducted by Linton et al. [13]. Transient forces models for debris impact loading due to tsunami events.
were generated by the impact of the bore on a wall shortly after
the initial impact. This was followed by a quasi-static force after 2.3. Current provisions of tsunami design criteria
the bore reflected from the structure. No impulsive forces were
observed for these tests. The ratio of the peak transient force to Current structural design codes provide very little guidance on
mean quasi-static force was 2.2 overall. It was observed that the estimating loads induced by tsunami events. The general approach
peak of transient load can be affected by standard construction in established codes has been to treat tsunami generated loads in a
up to 20%. similar fashion to storm flooding and storm surges [1]. Due to the
In addition to full-scale wall experimental tests on impulsive dynamic and varying nature of tsunami parameters affecting the
forces, tests on model residential structures have also been con- induced loads, the treatment of these loads similar to regular
ducted by van de Lindt et al. [14]. The study tested typical residen- floods may not be appropriate.
tial buildings at 1/6th scale under wave loading. The structural One of the first design provisions regarding tsunami loads is in-
models were tested transversely and longitudinally, in order to ob- cluded in the City and County of Honolulu Building Code [19]. The
serve behavior of common design components. Pushover tests loading requirements are addressed in the Coastal Flood Water De-
were also conducted in order to relate uplift forces from wave sign section and based on a 1980 Dames and Moore report [20]. The
loading to dry laboratory conditions. approach is based on impulse momentum and it is ‘‘equivalent to
Matsutomi [1,15] investigated the impulse forces of driftwood the impact force produced by a 1000-lb weight of debris traveling
experimentally. Two sets of tests were conducted: small-scale at the velocity of the flood water and acting on one square-foot
including generated bore and surge, and a full-scale in-air. For surface of the structural material where impact occurs’’ [20]. It is
the in-air experiment, a log was tied to a pendulum and swung assumed that the velocity of the body goes to zero over some small
against a load cell [1]. The in-air full-scale experiment was de- finite time interval (Dt), and the recommended Dt for wood con-
signed to account for any scaling effects, and the small-scale tests struction is 1 s [20]. According to FEMA P-646 [1], these values
were designed to account for the added water mass effect [1]. Since are unsubstantiated.
added water mass effects were ignored during the in-air tests, the International Building Code (IBC) addresses flood and tsunami
estimated impact load from the full-scale test can differ from the loads in their Appendices M and G, respectively [1,21]. For tsu-
loads introduced by a real tsunami event [1]. Though the study in- nami loads, IBC permits construction in identified tsunami zones
cluded a large amount of laboratory data, the resulting equation for if the constructed structure is designed as a Vertical Evacuation
the estimation of impact loads may be only applicable to driftwood Refuge complying with FEMA P-646 Guidelines or if designed
or logs [1]. Although the resulting equation only applies to drift to resist without collapse the hydrostatic, hydrodynamic, debris
logs, it can evaluate impact load for different wall conditions accumulation and impact, and scour effects of the Maximum
(impermeable where all the water returns back to where the wall Considered Tsunami [1]. The guidelines outlined in these two
is partially damaged and the water is allowed to flow beyond the appendices are not mandatory unless adopted by local authorities
wall) [1]. [1].
Ikeno et al. [16,17,1] conducted laboratory experiments similar ASCE/SEI Standard 7-10 Minimum Design Loads for Buildings and
to Matsutomi [1,14] in order to estimate the impact forces of the Other Structures [22] includes expressions for forces associated
objects other than driftwood or logs. The used debris shapes in- with flood and wave loads on specific types of structural compo-
cluded cylindrical, square column, and spherically shaped drift nents. Chapter 5 of this standard, Flood Loads, covers important
bodies [16,17,1]. The study only examined the impact onto an definitions that relate to flooding and coastal high-hazard areas
1252 A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261

related to tides, storm surges, and breaking waves [1]. Currently it stiff or really massive so that the debris rebounds off of the surface
does not address specifically tsunami loads, however, it is expected after impact before the structure can be displaced. Haehnel and
that in the 2016 edition, a new chapter titled Tsunami Loads and Daly [18] conducted flume experiments in order to determine a
Effects will be added, according to updates from the ASCE 7 Sub- structural stiffness required for the rigidity assumption to be valid.
committee on Tsunami Loads and Effects [23]. According to the authors, a structural stiffness of 22 MN/m was
The only tsunami design guideline currently available in the US sufficient to provide a relatively rigid structure (compared to the
is FEMA P-646 ‘‘Guidelines for Design of Structures for Vertical debris stiffness) and to obtain maximum impact forces during their
Evacuation from Tsunamis’’ [1]. The document is not a mandatory experiments. Further stiffness increases up to 120 MN/m did not
document and it cannot be incorporated into building codes as-is; result in increased maximum impact loads [18].As previously men-
on the other hand, this document is the most comprehensive work tioned, Haehnel and Daly [18] used three different approaches to
available and it is used as a reference in the International Building estimate the impact force due to debris: Contact-Stiffness, Im-
Code and serves as the basis for the development of the new tsu- pulse-Momentum, and Work-Energy. The Contact-Stiffness ap-
nami code provisions, such as in the case of the new chapter in proach was covered previously, while the Impulse-Momentum
the ASCE/SEI Standard 7-10 [23]. approach links the impulse acting on the debris in contact with
FEMA P-646 [1] presents several formulations for estimating the structure with the change in momentum of the debris [18].
impact loads. Among several formulations are three main ap- Momentum of debris is mu and it is assumed to tend to zero as
proaches: Contact-Stiffness, Impulse-Momentum, and Work-En- the debris impacts the structure. Additionally, the impact force de-
ergy, which were developed in an earlier study by Haehnel and pends on a sinusoidal relationship of time [18]. The resulting max-
Daly in 2002 [18] as discussed above. Out of the three methodolo- imum impact force is estimated as
gies, FEMA P-646 [1] suggests a variation of Contact-Stiffness ap- p um
proach to estimate the maximum force as shown below: F max ¼ ð3Þ
2 Dt
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
F max ¼ u kmð1 þ CÞ ð1Þ where Dt is the time of impact duration. For equivalency purposes
between the methods, Dt can be related to the Contact-Stiffness ap-
where Fmax = maximum impact force, u = debris velocity, m = mass proach as follows [18]:
of debris, k = contact stiffness, and c = hydrodynamic mass coeffi- rffiffiffiffiffi
p m
cient (0 for a wood log in longitudinal direction). Dt ¼ ð4Þ
2 k
Values for k and c are provided for common debris types such as
logs and shipping containers. The main challenge associated with The Work-Energy approach computes the impact force by
this methodology is in estimating the effective contact stiffness be- equating the work done on the structure with the kinetic energy
tween debris and structure during the impact. The effective stiff- of the debris element, and assuming that the velocity of the debris
ness is derived by the equation below: approaches zero as a result of the collision [18]. The stopping dis-
tance of the debris, Dx, is defined as the distance the debris travels
1 1 1
¼ þ ð2Þ from the point of contact with the target until the debris is fully
k kd ks stopped [18]. The resulting impact force is a function of the stop-
where kd and ks are the debris and structure stiffness, respectively ping distance, Dx, over which the force acts (Fmax = kDx). In order
[18]. A major assumption in the estimation of the effective stiffness to solve for stopping distance, the work and kinetic energy are
2 2
is that the structure is rigid relative to the debris; therefore, the equalized,
pffiffiffi
m
k Dx = mu . This equation canpbe ffiffiffiffiffiffiffiffi solved for Dx,
effective stiffness can be equaled to the debris stiffness. Until re- Dx ¼ u k , resulting in Dx as a function of 1=k.
cently, a k value for a wooden log debris impact has been made The Contact-Stiffness approach is preferred and recommended
available by Haehnel and Daly [18]. Recent research has provided in comparison to the other two methods mainly because k is less
a few additional debris stiffness values for some of the most com- sensitive than estimated Dt (Impulse-Momentum) and Dx
mon debris types, as shown in Table 1 [1]. (Work-Energy), even though k is only available for a limited num-
Undoubtedly, the use of the Contact-Stiffness approach is lim- ber of debris types [1]. Additionally, it is clear that Dt and Dx are
ited by the number of k values available in the literature. The accu- dependent on k. For example, regarding the evaluation of Dt in Im-
rate application of this methodology is dependent on the pulse-Momentum formulation, there is significant variability be-
estimation of stiffness for a specific debris type. Additionally, the tween assumed impact duration times. For instance, City and
rigidity assumption needs to be satisfied. According to Haehnel County of Honolulu Building code [20] gives Dt values for wood
and Daly [18], this assumption is satisfied if the structure is very construction as 1.0 s, steel construction as 0.5 s, and reinforced
concrete as 0.1 s (though these values have not been substantiated)
Table 1 [1]. FEMA 55 Coastal Construction Manual assumes the values as
Mass and stiffness of some common waterborne floating debris [1].
follows: 0.7–1.1 s for wood wall, 0.2–0.4 for reinforced concrete,
Debris type Mass of debris Debris stiffness, kd and 0.3–0.6 for concrete masonry [1]. The major variability of the
(kg) (N/m) suggested impact duration times can cause significant fluctuations
Lumber or wood oriented 450 2.4  106 in maximum impact loads due to tsunami generated debris.
longitudinally The availability of many formulations and their inherited uncer-
20-ft standard shipping container 2200 85  106
tainties emphasizes the lack of a consensus in the community and
(longitudinally)
20-ft standard shipping container 2200 80  106 the need for national systematic provisions regarding tsunami
(transverse) structural loads.
20-ft heavy shipping container 2400 93  106
(longitudinally)
20-ft heavy shipping container 2400 87  106
3. Numerical modeling – coupled Eulerian–Lagrangian
(longitudinally) approach
40-ft standard shipping container 3800 60  106
(longitudinally) The objective of this study is to obtain an accurate and effi-
40-ft standard shipping container 3800 40  106
cient numerical model which can be used to accurately determine
(transverse)
debris impact loading on structures, specifically wood buildings.
A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261 1253

The attention of this study is concentrated on obtaining an impact for computational fluid dynamic (CFD) analysis without the neces-
load due to tsunami-generated debris, and not on any structural sity of coupling any other computational techniques. The proposed
performance or failure modes of the wall panels. Even though model was constructed using a coupled Eulerian–Lagrangian (CEL)
structural failure modes are not the focus of this study, it is analysis in ABAQUS software, which is a general purpose finite ele-
important to notice that in the case of determining the behavior ment package [25]. The Eulerian capability included in ABAQUS
of the wall, a multi-dimensional approach is needed to address can be coupled with traditional Lagrangian formulations to model
different failure modes, including localized damage in the studs, interactions between highly deformable materials and relatively
the wall panel, and nail connectors. A multi-dimensional ap- stiff bodies, such as in fluid–structure interaction [25]. The avail-
proach has been presented by Cimerallo and Reinhorn for deriva- ability of this formulation significantly reduces the analysis time
tion of seismic fragility curves and the approach has been applied for fluid–structure interaction problems. Successful applications
to a case study, a hospital building [24]. Cimerallo and Reinhorn of CEL include numerical studies by Smojver and Ivancevic [26]
proposed an analytical method to derive the fragility curves that regarding bird striking damage on aircraft structures and Qiu
are a function of the return period and are based on a multidi- et al. [27] regarding geomechanical problems involving large
mensional threshold limit states (MTLS) function [24]. The pro- deformations such as soil–structure interactions. In both these
posed approach can be applied locally including parameters that studies, CEL was found to solve issues related to large mesh distor-
represent nonstructural components, and globally to describe tions, and in the second study several benchmarks were used to
the limit state of a part of a substructural system or to describe validate the CEL methodology.
the entire building structure [24].
One of the major obstacles in the way of rigorous numerical
analysis of debris impact in particular and dynamic tsunami loads 3.1. Description of wall system
in general is the cost of both computational time and power. There-
fore, time and computational requirements are important consid- The numerical model is assembled to conform to typical con-
erations in addition to accurate model formulations. The struction practices, using typical material and dimension proper-
advancements in both hardware and software have significantly ties for wood structures constructed in the US. A typical 0.5 in.
increased the potential for conducting such demanding simula- thick plywood panel of 4 ft  8 ft is selected along with typical
tions. Specifically, it is possible to construct a model which allows framing studs which consist of 2 in.  4 in. Douglas Fir-Larch at

Fig. 3. Front view of wall panel.

Fig. 4. (a) FE model of a wooden wall impacted by water and debris. (b) Spring load-deformation curve.
1254 A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261

Fig. 5. Experimental setup [18] (left) and finite element modeling in the current study (right).

Table 2
Comparison of force results between experimental and numerical modeling.
pffiffiffiffiffi
Velocity (m/s) Contact-Stiffness force 1550u m (N) Numerical modeling force (N) Difference (%)

0.381 2621 3218 54.3


0.533 3667 3218 12.2
0.800 5504 4037 26.6

16 in. on-center spacing. A front view of the wall system and its
dimensions are presented in Fig. 3.
Orthotropic material properties are used for the framing and
panel. The framing system is fixed at the bottom and at the top,
to simulate connections to top and bottom plates. The wall panel
is connected to the framing system with 8 d common nails every
6 in. on the outside studs and every 12 in. on the two inside studs.
Two sets of models are constructed in order to emulate two differ-
ent field conditions: interior panels and edge panels for a residen-
tial construction. In the case of edge panel, one of the studs is fixed
top-to-bottom.

3.2. Modeling technique

In the proposed model, the effect of several factors on the debris


impact on structures is investigated. An analysis matrix is set up in
order to conduct a parametric study and determine the sensitivity
of the impact force to the varying parameters. The considered
parameters include water height and velocity, and debris density Fig. 6. Validation of numerical results [18].
and shape. Specifically, three different water heights and velocities,
Table 3
two debris lengths, and three debris densities are varied during the
Analysis matrix.
analysis.
Both Lagrangian and Eulerian formulations are used in the mod- Component Variable Range
el to allow for small and large deformations, as needed. The water Water Run-up height Relative to wall height (Hw): 0.167Hw,
is modeled with the Eulerian part, assuming a Newtonian, nearly 0.33Hw, and 0.5Hw
incompressible, and practically frictionless fluid. The general form Maximum 1.99 m/s, 2.82 m/s, and 3.46 m/s
velocity
of Navier–Stokes equation for incompressible fluids [28] is shown Velocity profile Uniform
below:
Wall Material Panel: Plywood material properties
  properties
@v Framing: Douglas Fir-Larch material
q þ ðv  rÞv ¼ rP þ lr2 v þ f ð5Þ
@t properties
Panel 4 ft  8 ft  0.5 in.
dimensions
where q = fluid density, v = fluid velocity, P = pressure, l = dynamic Framing 2 in.  4 in. at 16 in. on center
viscosity, and f = other body forces. dimensions
It should be noted that for incompressible materials: r  v = 0 Debris Angle of impact 0°
and q = constant. Shape Rectangular
Specifically, water is defined through its density and a linear Mass 3.40, 4.25, 6.37, 13.6, 17.0, and 25.5 kg
equation of state (EOS) in energy. The particular EOS is the linear
form Us  Up of Hugoniot form [29], where Us and Up are the linear
shock velocity and the particle velocity, respectively. Three param- The Eulerian–Lagrangian general contact formulation is based
eters are required to determine water as a material: den- on an enhanced immersed boundary method. In this method the
sity = 1200 kg/m3 (seawater mixed with sand and other debris), Lagrangian structure occupies void regions inside the Eulerian
speed of sound in water = 1483 m/s, and dynamic viscosity of mesh [30]. The general contact algorithm automatically computes
water at 20 °C = 0.001 kg/(m s). and tracks the interface between the Lagrangian structure and the
A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261 1255

Eulerian materials, while penalty methods are used to couple the hourglass control are used to model the structural elements. Den-
Eulerian and Lagrangian parts [29,30]. sities for the studs, plywood, and debris are 500 kg/m3, 600 kg/m3,
Debris, wall panel, and framing are modeled using solid ele- and 600 kg/m3, respectively. The nail connections between wall
ments. Specifically, 3D elements with reduced integration with panel and framing are modeled using spring elements with

Fig. 7. Typical water–debris–panel interaction in the simulations.

Fig. 8. Normal stress history for interior panel with water velocity of 2.82 m/s.
1256 A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261

Fig. 9. Normal stress history for exterior panel with water velocity of 3.46 m/s.

Fig. 10. Impact force versus augmented velocity at impact zone.

load-deformation data for 8 d common nails from Gruber [31]. An Table 4


overview of the model and the load-deformation relation of the Resulting slopes from numerical impact force analyses.
spring elements are illustrated in Fig. 4a and b, respectively. pffiffiffiffiffiffiffiffiffiffiffi
Slope ð N=mÞ
pffiffiffiffiffi
Contact-Stiffness formulation ð1550u mÞ 1550
3.3. Modeling validation Exterior panel force at time of impact 855
Maximum exterior panel force 1218
Interior panel force at time of impact 218
In order to confirm the modeling approach, a validation of the
Maximum interior panel force 633
utilized methodology is conducted based on previous experimental
A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261 1257

tests. Available experimental data on isolated rigid systems are on were fastened to a rigid frame mounted on the flume floor [18].
small-scale tests. One set of comprehensive tests was conducted by The rounded target kept the point of impact concentrated between
Haehnel and Daly [18] in which small scale flume tests were con- the three load cells, assuring that all of the load cells were in com-
ducted to evaluate the force results from wood logs impacting a pression on impact.
stationary target. Out of a total of 8 test series conducted by Haeh- The corresponding numerical model used for verification is con-
nel and Daly [18], selective tests are chosen to verify the modeling structed using the exact experimental measurements and relative
approach used in this study, as they mimic the same hydrodynamic placements of debris, water, and load cell used by Haehnel and
conditions as the field conditions. In the actual experiments, logs of Daly [18]. Plan views of experimental set-up and corresponding
varying weight and cross-sectional dimensions were used to mea- ABAQUS model are shown in Fig. 5.
sure the impact forces on a stationary target. In addition to log Out of 48 different tests, three small-scale experiments are
weight, flow velocity, impact orientation, and target material were numerically modeled for verification purpose, which consist of a
varied. The load frame had a 7.5 cm semi-circular target mounted reduced-scale log with cross-section of 8  8 in. and length of
on a front plate, which in turn was mounted on three load cells that 36 in. Three different water velocities are selected from the actual

Fig. 11. Impact force due to debris on interior panel grouped by mass.

Fig. 12. Impact force due to debris on exterior panel grouped by mass.
1258 A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261

experiments, mainly 0.381, 0.533, and 0.8 m/s. The authors utilized 4.1. Time history of panel forces/stresses
three different approaches to estimate the impact force and to for-
mulate an equation for the impact force; Contact-Stiffness, Im- Normal stress time history is obtained for the highest stress
pulse-Momentum, and Work-Energy approach. Out of the three, point at the debris impact zone on the wall panel. Illustrative
only Contact-Stiffness Approach formulation is applicable to the examples for interior and exterior panel stresses are shown in
small-scale flume tests. The obtained force results from the ABA- Figs. 8 and 9, respectively. For the interior panel, water is not per-
QUS models are compared fairly well against the Contact-Stiffness mitted to go through or around the panel, therefore, after initial
formulation and the individual recorded experimental results. impact, there are several peaks (smaller than initial impact), which
In terms of numerical force results, two of the obtained forces could signify secondary water impacts after the initial contact. For
are within 13% and 26% of the expected values based on the Con- the exterior panel, the general pattern signifies that the stress re-
tact-Stiffness formulation, as presented in Table 2. Haehnel and duces significantly after the initial impact and continues to de-
Daly [18] observed that the flume tests exhibited some scatter crease as water is allowed to flow around one of the panel edges.
and the Contact-Stiffness formulation overestimated force for val-
ues less than 10 kN (which is the case in the small scale experi-
ments). This observation matches the numerical results, as
4.2. Impact load versus augmented velocity
illustrated in Fig. 6. The numerical results are within the range of
experimental results, thus the validation of the approach is con-
Haehnel and Daly [18] and FEMA P646 [1] both recommend an
firmed prior to conducting the parametric studies on the full-scale
impact force formulation which varies linearly with the aug-
wall panels. pffiffiffiffiffi
mented velocity expressed as u m. The impact forces obtained
numerically are plotted against the augmented velocity for both
4. Results and discussion interior and exterior panels. In addition to the force at the time
of impact, the overall maximum registered force at the impact zone
As previously mentioned, several important parameters are var- is plotted for the two panels, as shown in Fig. 10. Resulting slopes
ied during the analyses of the interior and exterior panels. The con- for each numerical model at time of impact and overall maximum
structed analysis matrix shown in Table 3 summarizes the varied impact force for both types of panels are presented in Table 4.
parameters in addition to other parameters that are kept constant While the impact force for the exterior panel exhibits an overall
during the analyses. Two numerical models are constructed in or- increasing trend, the impact force at time of impact decreases for
der to simulate interior and exterior wall panels. A general over- the interior panel. In order to analyse this behavior, the results
view of the physical behavior during an analysis of an exterior are grouped initially based on water height and velocity (and deb-
panel is illustrated in Fig. 7. ris velocity), and then according to the debris masses.

Fig. 13. Impact force on interior panel grouped by velocity.


A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261 1259

4.2.1. Effect of debris velocity and water height in Fig. 14. For the interior panel with an initial water height of
Numerical results for impact force at time of impact for interior 0.4 m, the increase in debris mass results in increase in impact
and exterior panels are presented in Figs. 11 and 12, respectively. force. However, in the other two water heights, 0.8 m and 1.2 m,
For the interior panel, as water height and velocity increase while the debris mass appears to have little effect on the impact force.
keeping mass constant, a reduction in impact force is observed. It can be observed that the increase in water height is contributing
This effect can be explained due to the flexibility of the interior pa- significantly more to the overall impact force in these two cases
nel. Specifically, the third water height corresponds to the middle than the debris mass. In general, the overall force exhibits a posi-
of the wall height where the panel is most flexible. As a result, tive trend.
since water and debris hit at a higher point compared to the other For the exterior panels, the same observations are confirmed; in-
two water heights, the panel is able to deform more and attract less crease in debris mass resulted in overall impact force increase.
force. Therefore, as panel stiffness at point of contact decreases, a While for the first two water heights (0.4 m and 0.8 m) the impact
reduction in impact force is exhibited. force due to debris and overall force at impact zone are almost iden-
For the debris impact force on the exterior panel in Fig. 12, one tical, for the third water height it can be noticed that water height
of the panel studs is fixed to simulate an exterior edge. In this case, and velocity have a more significant effect on the overall force at the
the wall panel is more rigid in comparison to the interior panel. impact zone. It is important to note, however, that impact force due
These boundary conditions bring the exterior panel closer to the to debris keeps increasing. Additionally, the exterior panel exhib-
ideal conditions of a rigid structure. As a result, an increase in im- ited much closer results to the recommended force equation, since
pact force is noticed overall with the increase of water height and its rigidity is much higher than the rigidity of the interior panel.
velocity, as it is expected for a rigid structure. While the force in-
creases in each case with increase in water height, the rate of force
increase is not always positive; this can be attributed to the com- 4.3. Impact force results summary
plex interaction between the wall and water as the water flows
past the wall. A summary of the impact force due to debris at the time of im-
pact is shown in Table 5. Impact force is normalized in order to
rank the cases with respect to the level of force impact. As ex-
4.2.2. Effect of debris velocity pected, the highest force is encountered at the exterior panel with
The effect of debris (and water) velocity is also studied from the the highest debris mass and highest water height and velocity. In
limited number of numerical analyses. Impact force is plotted general, the exterior panel attracts the higher forces, while the
against augmented velocity while velocity is kept constant and interior panel exhibits smaller impact loads, mainly due to its flex-
debris mass is varied. Force results for the interior panel are pre- ibility. The highest loads for interior panels are generally observed
sented in Fig. 13, while results for the exterior panel are shown for the lower water heights, as previously mentioned.

Fig. 14. Impact force on exterior panel grouped by velocity.


1260 A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261

Table 5
Normalized impact force results.

Panel Water height (m) Debris mass (kg) Panel location Normalized impact force
1 1.2 25.49 Exterior 1.00
2 0.8 25.49 Exterior 0.91
3 0.8 16.99 Exterior 0.79
4 1.2 13.59 Exterior 0.66
5 1.2 16.99 Exterior 0.61
6 0.8 13.59 Exterior 0.61
7 1.2 4.25 Exterior 0.61
8 1.2 6.37 Exterior 0.59
9 0.8 3.40 Exterior 0.57
10 0.4 25.49 Exterior 0.53
11 0.4 25.49 Interior 0.47
12 0.4 16.99 Interior 0.47
13 0.4 16.99 Exterior 0.47
14 0.4 13.59 Interior 0.46
15 0.4 6.37 Exterior 0.43
16 0.8 6.37 Exterior 0.42
17 0.4 13.59 Exterior 0.41
18 0.8 4.25 Exterior 0.37
19 0.4 3.40 Exterior 0.36
20 0.8 3.40 Exterior 0.35
21 0.4 4.25 Interior 0.35
22 0.4 4.25 Exterior 0.34
23 0.4 6.37 Interior 0.34
24 0.4 3.40 Interior 0.28
25 0.8 3.40 Interior 0.26
26 0.8 25.49 Interior 0.24
27 0.8 4.25 Interior 0.24
28 0.8 16.99 Interior 0.21
29 0.8 6.37 Interior 0.20
30 1.2 25.49 Interior 0.14
31 0.8 13.59 Interior 0.14
32 1.2 16.99 Interior 0.12
33 1.2 6.37 Interior 0.10
34 1.2 4.25 Interior 0.10
35 1.2 13.59 Interior 0.10
36 0.8 3.40 Interior 0.09

5. Conclusions
5.2. Large-scale parametric study
Contribution of tsunami generated debris impact loads on resi-
dential structures can be very significant. Debris impact can cause  While the results for an exterior panel exhibited similar trends
substantial damage or weaken structural components, which in to the recommended equation, significant changes were
turn can result in system collapse. Determination of structural observed for the impact loads on an interior panel.
loads due to debris impact from tsunami events is still a work-  For the exterior panel, increasing impact loads were observed
in-progress and systematic and unified methodologies are needed with the increase of both water height (velocity) and debris
in order to determine these impact loads accurately. mass.
A numerical model was constructed and verified against exist-  The overall slope compared to the force equation for wooden
ing small-scale experimental results. Following model verifica- logs was lower, but that is attributed to the not-fully rigid
tions, a parametric study was conducted to estimate the impact boundary conditions of the exterior panel.
loads on wooden residential structures by modeling interior and  For the interior panel, with the increase of water height and
exterior wall panels for typical residential structures. The paramet- debris velocity, the impact force decreased due to the panel
ric study was conducted by varying debris mass and water height flexibility.
(velocity) resulting in a total of 36 analyses. The results were com-  For the interior panel, when water height and debris velocity
pared against current recommendations for debris impact loads were kept constant and debris mass was varied, it was noticed
due to tsunami events. In general, the following conclusions are that for the lower water height debris impact force increased,
made based on analyses results: while the contribution of debris mass was not as significant
for the two greater water heights.
5.1. Small-scale verification

 Numerical results and the Contact-Stiffness force equation var-


ied from 13% to 54%. Acknowledgements
 Haehnel and Daly [18] observed that the proposed Contact-
Stiffness force equation did overestimate the load for lower The authors wish to thank Dr. R. Haehnel for providing the
force values. This observation was validated when numerical experimental flume data and for his comments and suggestions.
results were plotted against experimental results from Haehnel Thanks are also due to Mr. J.E. Daniell for allowing the authors to
and Daly [18] and the numerical results are shown to fit reason- use some of his images and Dr. M. Criswell for his advice on typical
ably well within the scatter of the experimental results. residential wood construction on the West Coast.
A. Como, H. Mahmoud / Engineering Structures 56 (2013) 1249–1261 1261

References of coastal engineering. Japan Society of Civil Engineering; 2001. p. 48 [in


Japanese].
[17] Ikeno M, Tanaka Y. Experimental study on impulse force of drift body and
[1] Applied Technology Council. Guidelines for design of structures for vertical
tsunami running up to land. In: Proceedings of coastal engineering, Japan;
evacuation from tsunamis. FEMA P-646, 2nd ed. Redwood City, California;
2003.
2012. <https://www.fema.gov/library/viewRecord.do?id=3463> [accessed
[18] Haehnel RB, Daly SF. Maximum impact force of woody debris on floodplain
07.06.13].
structures’’, technical report TR-02-2. Engineer Research and Development
[2] Ishiguro R, Kitamura S. Japan quake’s economic impact worse than first feared.
Center, US Army Corps of Engineers. February 2002, Springfield, Virginia.
Reuters; 2011. <http://www.reuters.com/article/2011/04/12/us-japan-
[19] Chock GYK, Robertson I, Riggs RH. Tsunami structural design provisions for a
economy-idUSTRE73B0O320110412> [accessed 25.03.13].
new update of building codes and performance-based engineering. Solut Coast
[3] Daniell JE, Vervaeck A, Wenzel F. A timeline of the socio-economic effects of
Disasters ASCE 2011. http://dx.doi.org/10.1061/41185(417)38.
the 2011 Tohoku earthquake with emphasis on the development of a new
[20] City & County of Honolulu. Regulations within flood hazard districts and
worldwide rapid earthquake loss estimation procedure. In: AEES 2011
developments adjacent to drainage facilities. Revised Ordinances of Honolulu,
conference, Barossa Valley, SA, Australia, November 18–20.
Chapter 16: Building Code, Article 11; 1990.
[4] Graumann A, Houston T, Lawrimore J, Levinson D, Lott N, McCown S, et al.
[21] ICC. International building code. International Code Council, Inc., Country Club
Hurricane Katrina, a climatological perspective. National Oceanic &
Hills, Illinois; 2012.
Atmospheric Administration. National Climatic Data Center, technical report
[22] American Society of Civil Engineers. Minimum design loads for buildings and
2005-01; October 2005. <http://www.ncdc.noaa.gov/oa/reports/tech-report-
other structures. ASCE/SEI Standard 7-10. American Society of Civil Engineers,
200501z.pdf> [accessed 25.03.13].
Reston, Virginia; 2010.
[5] Blake ES, Kimberlain TB, Berg RJ, Cangialosi JP, Beven II JL. Tropical cyclone
[23] Chock G. ASCE 7 and the development of a tsunami building code for the U.S.
report: hurricane sandy. National Hurricane Center; 12 February 2013. <http://
Applied Technology Council. ASCE Tsunami Loads and Effects Subcommittee;
www.nhc.noaa.gov/data/tcr/AL182012_Sandy.pdf> [accessed 25.03.13].
2012. <https://www.atcouncil.org/files/ATC-15-13/Papers/
[6] Wallace T, Kaleem J. Hurricane Sandy vs. Katrina: infographic examines
06_CHOCKpaper.pdf> [accessed 25.03.13].
destruction from both storms. Huffington Post; 11 November 2012. <http://
[24] Cimellaro GP, Reinhorn AM. Multidimensional performance limit state for
www.huffingtonpost.com/2012/11/04/hurricane-sandy-vs-katrina-
hazard fragility functions. J Eng Mech ASCE 2011;137:47–60. http://dx.doi.org/
infographic_n_2072432.html> [accessed 25.03.13].
10.1061/(ASCE)EM.1943-7889.0000201.
[7] Palermo D, Nistor I. Tsunami-induced loading on structures. Structure
[25] SIMULIA. Coupled Eulerian–Lagrangian analysis with Abaqus/explicit; 2012.
Magazine; 10 March 2008.
<http://www.3ds.com/products/simulia/services/training-courses/course-
[9] Federal Emergency Management Agency. Principles and practices of planning,
descriptions/coupled-eulerian-lagrangian-analysis-with-abaqusexplicit/>
siting, designing, constructing, and maintaining residential buildings in coastal
[accessed 07.06.13].
areas. Coastal construction manual, 4th ed. August 2011.
[26] Smojver I, Ivancevic D. Bird strike damage analysis in aircraft structures using
[10] Kong L, Robertson I, Yeh H. Structural response to tsunami loading [Lecture];
Abaqus/explicit and coupled Eulerian–Lagrangian approach. Compos Sci
2006. <http://escweb.wr.usgs.gov/share/mooney/SriL.V6.ppt> [accessed
Technol 2011;71:489–98. http://dx.doi.org/10.1016/
11.06.13].
j.compscitech.2010.12.024.
[11] Ramsden J. Forces on a vertical wall due to long waves, bores, and dry-bed
[27] Qiu G, Henke S, Grabe J. Application of a coupled Eulerian–Lagrangian
surges. J Waterway, Port, Coastal, Ocean Eng 1996;122(3):134–41. http://
approach on geomechanical problems involving large deformations. Comput
dx.doi.org/10.1061/(ASCE)0733-950X(1996) 122:3(134).
Geotech 2011;38:30–9. http://dx.doi.org/10.1016/j.compgeo.2010.09.002.
[12] Arikawa T. Structural behavior under impulsive tsunami loading. J Disaster Res
[28] Cengel Y, Cimbala J. Approximate solutions of the Navier–Stokes equation. In:
2009;4(6):377–81.
Fluid mechanics: fundamentals and applications. McGraw Hill Series in
[13] Linton D, Gupta R, Cox D, van de Lindt J, Oshnack ME, Clauson M. Evaluation of
Mechanical Engineering; 2006. p. 473–6.
tsunami loads on wood frame walls at full scale. J Struct Eng 2011. http://
[29] Abaqus Analysis Users Manual. Version 6.11. Linear Us  Up Hugoniot form
dx.doi.org/10.1061/(ASCE)ST.1943-541X.0000644.
Section 24.2.1. Dassault Systèmes; 2011.
[14] van de Lindt J, Gupta R, Cox DT, Wilson J. Wave impact study on a residential
[30] Abaqus Analysis Users Manual. Version 6.11. Formulation of Eulerian–
building. J Disaster Res 2009;4(6):419–26.
Lagrangian contact Section 14.1.1. Dassault Systèmes; 2011.
[15] Matsutomi H. A practical formula for estimating impulsive force due to
[31] Gruber JJ. Reliability and effect of partially restrained wood shear walls.
driftwoods and variation features of the impulsive force. In: Proceedings of the
Wayne State University dissertations; 2012. Paper 442. <http://
Japan Society of Civil Engineers, vol. 621; 1999. p. 111–27 [in Japanese].
digitalcommons.wayne.edu/oa_dissertations/442/> [accessed 25.03.13].
[16] Ikeno M, Mori N, Tanaka Y. Experimental study on tsunami force and
impulsive force by a drifter under breaking bore like tsunamis. Proceedings

You might also like