Evaluation of the Structure of Levee Transitions on Wave

Run-up and Overtopping by Physical Modeling

Drake Oaks1; Billy Edge2; and Patrick Lynett3
1

Abstract: Coastal regions are continually plagued by high water levels induced by river flooding or hurricane-induced storm surge. As with
any protective structure, it is essential to understand potential problematic locations that could result in structural failure and devastating loss.
Common coastal protective systems are composed of floodwalls and levees, for each of which practiced methodologies have been used to
estimate their performance under design conditions. Methodologies concerning spatial variability are limited, however, and transitions where
earthen levees merge with floodwalls are considered areas vulnerable to erosion and possible breaching. Physical modeling of a levee tran-
sition is undertaken in a three-dimensional wave basin to evaluate this hypothesis, and the detailed results of this assessment are presented in
this paper. From the physical model testing, analysis of the data reveals that overtopping rates tend to be larger immediately near the transition
than away from it. The run-up values and floodwall wave heights tend to show potential problematic areas and mimic the variation of
overtopping along the levee transition. Under the design conditions tested, extreme overtopping conditions and associated water level values
indicate that for the structure to sustain the hydraulic conditions, it must be well armored. It is shown that the variation of the still water level
plays the largest role in the magnitude of the measured values, and increasing the peak wave period and wave heights also yields greater
overtopping and water levels at the structure. This study highlights the need to understand specific spatial variability along coastal protective
systems, and provides a better understanding of the mechanisms affecting overtopping for the specific structure tested. DOI: 10.1061/
(ASCE)WW.1943-5460.0000103. © 2012 American Society of Civil Engineers.
CE Database subject headings: Levees; Dikes; Floods; Wave overtopping; Coastal environment.
Author keywords: Levee; Floodwall; Overtopping; Physical model.

Introduction the Netherlands, physical modeling of dikes has been conducted to

better understand the mechanisms of run-up and overtopping, and it
Throughout many coasts and low-lying areas globally, periodical has been refined into a sophisticated method producing accurate
floods and storm surges cause significant and catastrophic damage. estimations beneficial to final designs (Schüttrumpf and van Gent
For regions plagued by these incidents of drastic water level in- 2003). These models further serve to validate or disprove design
creases, a common first and ultimate line of defense is levees guidelines and calibrate numerical models to ensure efficiency
and floodwalls. Levees and floodwalls offer sufficient protection and integrity in the final design. More recent events, such as
against high rises in water level and can protect large areas from Hurricane Katrina in 2005, have begun to spark even more interest
inundation; however, the levee and floodwall system is only as in the evaluation of these coastal protective structures. Hurricane
durable as its most tenuous area. One such area, highlighted by this Katrina caused 200 miles of damage to the floodwalls and levees,
research, occurs where an earthen levee transitions into a floodwall which is more than 60% of the total Hurricane Protection System
with an incorporated levee, referred to in this paper as a levee (HPS), encompassing the City of New Orleans (Link 2009). Of the
transition. 50 breaches in the HPS caused by Hurricane Katrina, all but 4 were
Ultimately, the most significant consequence of hydraulic load- induced by overtopping and run-up (Hughes 2008; Nicholson
ing that jeopardizes the structural integrity of levees and floodwalls 2007; Sills et al. 2008). The majority of design guidelines used
is erosion. The two most common mechanisms that provoke ero- to predict wave overtopping and run-up are purely empirical
sion are wave run-up and overtopping. Since the 1953 inundation in (Pullen et al. 2007; Technical Advisory Committe for Flood
Defence in the Netherlands (TAW) 2002; van der Meer and Janssen
1
Graduate Research Assistant, Ocean Engineering Program, Texas 1995) and take into account various parameters that detail the levee
A&M Univ., College Station, TX 77843, USA. E-mail: drakeoaks@ geometry and associated hydraulic conditions (Schüttrumpf and
[Link] van Gent 2003). However, there are many assumptions inherent
2
Professor, Sustainable Coastal Processes and Engineering Program, within the empirical design formulas. Overall, it is important to
North Carolina State Univ., Raleigh, NC, 27695, USA. E-mail: bledge@ understand and evaluate the degree of accuracy of the estimations
[Link] provided by the empirical design guidelines.
3
Associate Professor, Ocean Engineering Program, Texas A&M Univ.,
College Station, TX 77843, (corresponding author). USA. E-mail: plynett
@[Link]
Note. This manuscript was submitted on November 18, 2010; approved Scope of Work
on May 2, 2011; published online on December 15, 2011. Discussion per-
iod open until June 1, 2012; separate discussions must be submitted for The geometry of the structure under evaluation is closely associated
individual papers. This paper is part of the Journal of Waterway, Port, with typical levee and floodwall geometry found in the HPS. The
Coastal, and Ocean Engineering, Vol. 138, No. 1, January 1, 2012. research and the experiments discussed in this paper focus on
©ASCE, ISSN 0733-950X/2012/1-0–0/$25.00. wave-only overtopping and run-up.

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 1
 The basic objectives of this research are as follows: The incorporated levee acts as a 1V:10H berm, which dissipates
• Conduct laboratory investigation of levee transition in three- some of the wave energy before encountering the 9.14-m-high
dimensional shallow water wave basin, floodwall. The backside, or protected side, of the levee section typ-
• Analyze the resulting data and identify potential problem areas ically integrates a berm in addition to the 1V:3H slope, but is modi-
within structure, fied in this case to optimize placement within the laboratory, as
• Provide comparison of experimental overtopping measurements subsequently discussed. To form the transition, the levee section
with typical empirical design formulas, and is rounded at the end to form a levee head. The floodwall section
• Develop relationships between hydraulic conditions and ob- is then placed against the levee section forming the contours at the
served overtopping for the specific structure tested. transition, as illustrated in the three-dimensional rendering
in Fig. 3.
Levee Transition
Hydraulic Conditions
As mentioned, the levee transition referred to in this paper is the
transition between an earthen levee and a vertical floodwall. Fig. 1 The testing parameters used for evaluation of the levee transition
provides a generalized picture. The established dimensions of the are given in Table 1, and are approximately representative of
two sections are addressed in Fig. 2. In the figures, the vertical da- design-level conditions along the HPS of New Orleans. Eight
2 tum of both sections is at the toe of the levee, which is þ0:0 m. The tests are conducted with variations in the still water level
levee section is a composite sloped levee with a 1V:10H berm and a (SWL), significant wave height (H s ), and peak period (T p ). Note
levee crest elevation of 8.23 m. The floodwall section consists of an that the significant wave heights provided are the target heights,
internal floodwall embedded within an incorporated levee. and will be matched with a spectral H mo . The characteristic values

Fig. 1. Example levee transition (courtesy of Google Maps, ©2011)

Fig. 2. Cross-sectional prototype dimensions of (top) levee section and (bottom) floodwall section

2 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
 Table 1. Prototype Hydraulic Conditions These traditional empirical equations [Eqs. (1)–(4)] are applied
Test No. SWL (m) H s (m) T p (S) to specific cross sections of levees or floodwalls under specific
wave conditions, indirectly assuming that the cross sections are
1 5.79 2.29 6 uniform for a given levee or floodwall span; in other words, the
2 5.79 2.29 7 equations do not provide any temporal or spatial variability.
3 5.79 2.29 8 Clearly, the structure under scrutiny does not satisfy the assumption
4 5.79 2.29 9 of cross-sectional uniformity.
5 6.89 2.74 6
6 6.89 2.74 7 Testing Environment
7 6.89 2.74 8 All testing was conducted at the Reta and Bill Haynes ’46 Coastal
8 6.89 2.74 9 Engineering Laboratory at Texas A&M University, College Station,
Texas. A shallow water wave basin is contained within the labo-
ratory and has the following dimensions: length of 36.6 m, width
listed in the table serve to formulate a target unidirectional Texel/ of 22.9 m, and water depth up to 1.22 m. The basin houses a direc-
tional wave basin that operates to a depth of 1 m, creating waves up
3 Marsen/Arsloe (TMA) wave spectrum, which is to be produced by
the wave generator in the testing facility. To provide an averaged to 61 cm. The piston-type wave generator is composed of 48 in-
result and a measure of the experimental scatter in the resulting dependent paddles, enabling wave directionality and the generation
measured data, three 10-min runs per test are completed, yielding of multiple spectral models. The waves are generated from the wave
24 tests in total. All waves are propagated normal to the levee struc- generator and propagate over a flat bottom until interaction with the
ture. All SWLs are referenced to the originally established datum of tested structure. Waves that transmit beyond the structure encounter
the levee toe. The significant wave heights are referenced to a rock beach, which dissipates approximately 85% of the wave
the SWL. energy.
Of course, laboratory effects potentially influence measure-
Design Guidelines ments during testing. The largest model effect was generated by
the sudden drop-off at the model sides. This sudden drop-off caused
As mentioned, one objective is to compare the experimental results the wave crests to refract toward the levee, resulting in abnor-
with typical design formulas. The typical design formulas pre- mally large overtopping rates near the end of the levee section
sented as a comparison are overtopping calculations provided in and floodwall section. Consequently, overtopping measurements
the European overtopping manual (Pullen et al. 2007) and the and floodwall wave gauge measurements recorded near the levee
TAW manual (2002), as originally defined by van der Meer and end and floodwall end are not represented. Additionally, seiching
Janssen (1995). The Iribarren number, or surf similarity parameter, within the basin was limited by additional wave absorbers at the
is calculated as sidewalls.
tan α
ξ op ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1Þ
H s =Lop
Experimental Setup and Physical Modeling Process
where α is the flood-side slope of the structure, H s is the offshore
significant wave height (defined as the average of the highest one- Levee Transition Model
third waves), and Lop is the offshore wavelength dependent on the As noted earlier, the levee transition consists mostly of two parts,
peak wave period through the floodwall section and the levee section. Fig. 3 illustrates the
gT 2p transition between these two sections. The levee section, high-
Lop ¼ ð2Þ lighted in green, is rounded at the transition as it intersects with
2π the floodwall section, which is presented in white. The model is
where g is gravitational acceleration, and T p is the peak wave composed of a rock core with additional wooden supports within.
period. Pullen et al. (2007) provide guidance for calculation of Model scaling is 1∶20 using Froude similitude. On top of the rock
the deep water wavelength using the mean wave period, T m , in core, a 10-cm concrete shell is placed to provide a smooth surface
which T p ¼ 1:1T m . The overtopping calculations in this paper and enable accurate representation of the detailed contours present
are based on the use of T m , requiring use of the ratio on the flood side of the model, especially at the transition. The
  floodwall is composed of a wooden frame with sheet metal on
q 0:067 R 1 either side. The floodwall is built within the incorporated levee,
pffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffi γb ξ op exp
4:3 c ð3Þ
gH 3mo tan α H mo ξ op γb γf γβ γv as shown in Fig. 2, to allay any movement of the wall during test-
ing. The 1∶20 scale yields a model that is approximately 10.7 m
with a maximum of long and 4.4 m wide with a floodwall crest elevation of
 
q R 1 45.72 cm and a levee crest elevation of 41.15 cm.
pffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2 exp
2:3 c ð4Þ
gH 3mo H mo γf γβ
Instrumentation and Analysis
where q is the average wave overtopping discharge (e.g., m3 =m), As illustrated in Fig. 3, the instruments referenced in this paper and
H mo is the spectral significant wave height at the toe of the levee, γf used in the data collection are of four major types: four floodwall
is the reduction factor for slope roughness, γβ is the reduction factor wave gauges, four levee run-up gauges, nine overtopping contain-
for oblique wave attack, γb is the reduction factor for a berm, γv is ers, and five wireless wave gauges. The spacing of the gridlines in
the reduction factor for a vertical wall on the slope, and Rc is the Fig. 3 is 1.52 m (5 ft) in model dimensions or 30.48 m (100 ft) in
free crest height above the SWL. The free crest, or freeboard, is the prototype dimensions. The floodwall gauges are located approxi-
vertical distance between the SWL and the crest of the levee mately 5–8 mm in front of the floodwall. Only three of the four
or dike. gauges shown are used in the data analysis and results. The gauge

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 3
 Table 2. Prototype Hydraulic Conditions Achieved
Test No. SWL (m) H mo (m) T p (S)
1 5.79 1.77 5.92
2 5.79 1.82 7.35
3 5.79 1.87 8.52
4 5.79 1.86 9.10
5 6.89 1.91 5.92
6 6.89 2.03 6.73
7 6.89 2.09 7.99
Fig. 3. Flood side of modeled levee transition and associated labora- 8 6.89 2.10 8.87
tory equipment: (1) floodwall wave gauges; (2) levee run-up gauges;
(3) overtopping containers; (4) wireless capacitance three-gauge array
sheet metal and fit either directly on the floodwall or on the back-
side slope of the levee. Each is positioned and created so that the
closest to the tank side wall was subject to model edge effects, no- containers are immovable during testing and no water is capable of
tably refraction along the edge of the model, which resulted in unre- going under the container. The volume of water in the container, the
alistic and inaccurate representations of the prototype structure; time required to fill the container to the measured level, and the
therefore, this gauge’s data were omitted. From the three repeatabil- width of the opening of the are needed to determine the overtopping
ity runs per each 10-min test, three statistical pieces of information rate (l=s=m). This array of overtopping containers enables calcula-
are obtained from the three floodwall gauges’ time series, each of tion of an overtopping distribution along the levee transition and
which is determined from a zero upcrossing method of the time allows correspondence of overtopping values to the measured wave
series: 2% floodwall crest elevation (H 2% ), floodwall significant heights and run-up values. As mentioned, the measured overtop-
wave height (H s ), and floodwall mean water level (MWL). From ping rates are directly compared with the empirical equations
the zero upcrossing analysis, all wave heights recorded at the flood- [Eqs. (1)–(4)]. In addition to the calculated values from the deter-
wall were ranked and the significant heights determined. The 2% ministic approach, values from a probabilistic approach (van
floodwall crest elevation is defined as the floodwall crest elevation Ledden et al. 2007) are provided. These values, based on a
that is exceeded by only 2% of the incident waves, where the indi- Monte-Carlo approach, are listed as q50 and q90 and are defined
vidual waves are likewise identified through an upcrossing analy- as the overtopping values exceeded by 50 and 10% of the storm
sis. The significant floodwall wave height is the average floodwall events, respectively. In this analysis by van Ledden et al., storm
wave height of the highest one-third waves, whereas the floodwall wave and water level conditions are chosen randomly from a speci-
fied distribution, and thus these statistical exceedence values are
MWL is defined as the average water level during the time series
representative of uncertainty in the hydrodynamic forcing. These
and is essentially the value of the SWL plus the wave-induced setup
empirical values are based on hydraulic conditions provided in
at the floodwall. Validation of the floodwall wave gauge measure-
Table 1.
ments are based on visual observation and data analysis. Video and
Three wave gauges are positioned in front of the levee section to
photography during the testing compared well with the measure-
create a three-gauge array for measurement of the incident wave
ments. From this video and data comparison, it can be stated that,
heights created by the wave generator. This three-gauge array al-
in general, water splashing off the wall and hitting the wave gages
lows decomposition of the full wave spectrum into incident and
was not registered as a high elevation by the instrument. However,
reflected wave spectra by a least-squares method (Mansard and
any anomalous spikes in the time series were filtered, and repeat-
Funke 1980). The three-gauge array is located in the constant-depth
ability runs provided consistent results.
portion of the tank away from the model, and can be interpreted as
Four run-up gauges are positioned along the upper levee section
the wave height at the toe of the structure. Table 2 lists the hydraulic
surface at a slope of 1V:4H. The gauges are positioned 1 mm away
achieved conditions inside the wave basin. The hydraulic condi-
from the levee face to ensure the gauge does not remain wet as the
tions achieved differ from the requested conditions, influenced
water recedes down the levee’s surface. Similar to the floodwall
mostly by wave-breaking energy loss very close to the wave maker,
gauges, a zero upcrossing analysis of the gauges’ time series en-
suggesting that the created waves are close to the steepest that can
ables calculation of the following statistics: 2% free surface eleva-
exist over a flat bottom.
tion (2% η) and mean shoreline position (MSP). The 2% free
Fig. 4 is a photograph of the levee transition model under testing
surface elevation is defined as the surface elevation as measured at the Haynes Coastal Engineering Laboratory’s shallow water
by the gauge that is exceeded by only 2% of the incoming waves. wave basin. The wave parameters of Test No. 7 provide extreme
This value is somewhat analogous to a run-up value, but because floodwall wave heights and levee run-up values, which conse-
there is overtopping of the structure, any run-up value is ill-defined. quently yield significant overtopping rates along the entire levee
The MSP is synonymous with the floodwall MWL and is the value transition. As a reminder, the floodwall crest elevation is 9.14 m
of the SWL plus the wave-induced setup on the flood-side slope of in prototype dimensions.
the levee. Having multiple gauges along the structure allows the
lateral variation of the measured values to be plotted, and use of
MWL and MSP allows determination of the spatial variation of Experimental Results
the wave-induced setup along the levee transition.
A total of nine overtopping containers are used in the data analy-
sis and results. Just as the far-side floodwall gauge was subject to Levee Run-up and Floodwall Wave Heights
the refraction at the edge of the model, so is the overtopping con- Fig. 5 provides dimensional plots of two representative tests: No. 3
tainer located at this location; therefore, the last overtopping con- and No. 7. The top plots illustrate the floodwall wave heights and
tainer shown in Fig. 3 is omitted. The containers are fabricated from levee water elevations in prototype dimensions. The vertical datum

4 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
 Overtopping Rates
The bottom subplots in Figs. 5(a) and 5(b) are comparisons of the
measured overtopping rates and the empirically calculated overtop-
ping rates based on Eqs. (1)–(4). The blue lines represent the aver-
aged measured overtopping rates for the given test. The total range
of values is again represented by the max/min bars at each meas-
urement location. The floodwall side is indicated by the negative
distance along the levee transition, and the levee section is indicated
by the positive distance. As noted, the predictions from the deter-
ministic equations [Eqs. (1)–(4)], as well as the probabilistic over-
topping estimates, are provided to compare the measured results
with expected, empirical results. The dashed black line represents
Fig. 4. Levee transition model during Test No. 7 the deterministic overtopping rate predicted from Eqs. (1)–(4). The
green line is representative of the statistical q90 value, and the red
dashed line is representative of the q50 value.
Immediately visible are the overtopping rates associated with
for these statistical values is the toe of the levee, or the basin floor.
the levee side of the protective structure. Clearly, the values of
The horizontal datum is the midpoint of the model, or at the tran-
the levee side are much greater than those of the floodwall side.
sition between the two sections, where positive values are indica- This is expected, and is consistent among all tests. Though the val-
tive of the levee section, and negative values represent the floodwall ues may be much smaller on the floodwall side, according to the
section. Each piece of statistical information from the floodwall European overtopping manual (Pullen et al. 2007), discharges
wave gauges and levee run-up gauges is portrayed on the graphs. greater than 1:0 L=s=m can still induce erosion. For reference, dis-
Each plotted value is the average of the repeatability runs for each charges greater than 10 L=s=m will erode poorly protected levees
test. In addition to this value, the total variation in results among the and floodwalls; discharges greater than 50 L=s=m will result in
three runs is indicated by the error bar plots; however, in some re- damage to the crest and rear slopes, unless these areas are well
sults the precision among the three runs was very high, and it is protected.
difficult to see the error bars. There are many differences between There are considerable differences among the tests with differ-
the lower and higher SWL tests. As shown for Test No. 3, a ent SWLs, especially in the overall magnitude of the results. The
recognizable feature of the floodwall wave heights is the gradual empirical equations predicted no overtopping along the floodwall
increase in the heights toward the transition during the lower SWL for Test Nos. 1–3, which evidently is not the case as noted by the
tests. The variations in floodwall wave heights and levee run-up measured values. During the lower SWL tests, the floodwall sec-
values are directly influenced by the variations in MWL and tion experienced overtopping rates ranging from 1 to 5 L=s=m.
MSP. As the wave-induced setup increases, so will the overall There is a gradual increase in the overtopping rates toward the tran-
height of the depth-limited waves at the floodwall and the run- sition along the floodwall, which is mimicked by the gradual
up values along the levee. increase in water level values along the floodwall. Also clearly
Essentially, the MWL and MSP are equivalent values and form a shown are the underestimated overtopping rates on the levee side
mean water level distribution across the entire model. As expected, of the structure; this too is consistent for all tests. Only the q90
the wave-induced setup is greater on a sloped structure, such as the values from Test Nos. 1–3 provide a reasonable estimation of
levee section, than at a vertical wall. The variation in wave-induced the measured overtopping rates on the levee side. For the lower
setup ultimately influences the overall height of the floodwall wave SWL tests, the range of overtopping rates for the levee section
heights and levee run-up. Because the MSP at the levee is greater range from approximately 10 L=s=m to more than 50 L=s=m,
than the MWL at the floodwall on the lower SWL tests, the MWL which could result in significant damage and erosion.
As predicted, the overtopping rates increased considerably for
tends to increase toward the transition and levee section. This in-
the higher SWL tests. The floodwall side experienced overtopping
crease in MWL directly increases the floodwall wave heights (with
values ranging from approximately to nearly 175 L=s=m. The
respect to the toe of the levee) toward the transition as well. For the
range is significant, and there is distinct variability along the flood-
larger SWL tests, as shown for Test No. 7, the opposite is true for
wall, which is discussed further. For the levee side of the structure,
the variation in floodwall wave heights; instead, the values decrease
tests beyond Test No. 3 yield measured overtopping rates well be-
toward the transition. Also noticeable is that the MWL is actually low the provided q90 value, concluding that the q90 value offers a
higher than the levee crest elevation and, therefore, larger than the conservative design approach for Test No. 4 and for the higher
MSP. This occurs because the floodwall crest elevation is greater SWL tests, as shown in Fig. 5(b). For the floodwall side, tests
than the levee crest elevation, which generates a flow around the beyond Test No. 6 also yield measured overtopping rates that
floodwall crest at the transition, inducing a divergence at the tran- are below the q90 value; therefore, the q90 value can also be con-
sition. As a result, the MWL decreases toward the transition and so sidered conservative along the floodwall side for the extreme cases
do the floodwall wave height values. of Test Nos. 7 and 8. Overtopping rates for the high SWL tests
Overall, all wave height and run-up values are substantial, even would most likely exceed the design conditions for a well-protected
for the lower SWL tests. For the majority of the tests, the floodwall structure. Again, the empirical overtopping rates calculated using
wave crest elevations are well above the floodwall crest elevation, Eqs. (1)–(4) tend to continually underestimate the measured values,
and 2% η is nearly equivalent to the levee crest elevation, indicating and the estimate begins to converge closer to the q50 value on the
that a large fraction of waves are overtopping the structure. Of levee side, indicating that the empirical overtopping equations are
course, with an increase in the SWL, the run-up values and flood- not well suited for calculation of the overtopping rates for this par-
wall wave heights increase drastically, to 30 and 10%, respectively. ticular, spatially variable, protective structure.

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 5
 Levee Transition Crest Elevation Profile
Test No. 03
12
2% Crest Elevation Floodwall Hs Floodwall MWL 2% η Mean Shoreline Position
11

10

Water Level (m)

Floodwall Crest
9
Levee Crest
8

5
-80
80 -60
60 -40
40 -20
20 0 20 40 60
Distance Along Levee Transition (m)
(a)

Prototype OT Profile Along L evee Transition

Test No. 03
50
Measured OT Rate
ototype OT Rate (L/s/m)

Deterministic OT Rate
40
q50
q90
30

20

10
Pr o

0
-80 -60 -40 -20 0 20 40 60
Distance Along Levee Transition (m)
(b)

Levee Transition Crest Elevation Profile

Test No. 07
16 2% Crest Elevation Floodwall Hs Floodwall MWL 2% η Mean Shoreline Position

14
er Level (m)

12

10 FloodwallCrest
Wate

Levee Crest
8

-80
80 -60
60 -40
40 -20
20 0 20 40 60
Distance Along Levee Transition (m)
(c)

Prototype OT Profile Along L evee Transition

Test No. 07
600
Measured OT Rate
ototype OT Rate (L/s/m)

500 Deterministic OT Rate

q50
400 q90

300

200
Pro

100

0
-80 -60 -40 -20 0 20 40 60
Distance Along Levee Transition (m)
(d)

Fig. 5. Dimensional floodwall wave heights, levee water elevations, and overtopping (OT) rate distribution results from (a) Test No. 3 and
(b) Test No. 7

6 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
 value of ζ is chosen. In the following dimensionless parameters, the
superscript ‘t’ indicates a nondimensional parameter, and H i is the
measured incident characteristic wave height:
 
H Rc
H t2% ¼ 2% ð5Þ
Hi h
 
H s Rc
H ts ¼ ð6Þ
Hi h
 
MWL Rc
MWLt ¼ ð7Þ
Hi h
 
2%η Rc
2%ηt ¼ ð8Þ
Hi h
Fig. 6. Visual effects of undertow (highlighted by arrows)  
MSP Rc ζ
MSPt ¼ ð9Þ
Hi h
In addition to the increase in SWL, the increase in peak period
also directly increases overtopping rates; these results are mimicked A dimensionless long-shore distance, xl , is defined as
by the general increase in floodwall wave heights and levee run-up 2π
values. One consistent trend in the spatial distribution of overtop- xl ¼ kxΔk ¼ ð10Þ
Li
ping rates for all tests is the local overtopping peaks that occur at
approximately
25 m on the floodwall side and at þ15 to þ30 m where x is the distance along the levee transition; the transition, or
on the levee side. These results cannot be explained using empirical midpoint of the model, serves as the datum for the long-shore dis-
methods, which assume a uniform cross section along the length of tance. The long-shore distance is normalized by the characteristic
a structure. In fact, these variations are generated from the three- incident wavenumber, k, calculated at the three-gauge array, based
dimensional contours at the transition, which in turn generate an on the peak spectral period at the structure toe water depth as mea-
undertow, as portrayed in Fig. 6. sured by the incident wave gauges. The wavenumber is defined as
At the transition, the contours cause the receding water to flow the ratio in Eq. (10), where Li is the characteristic incident wave
not only down the flood side of the structure, but also toward the length.
floodwall side of the structure. As water flows down the contour in As noted, the floodwall wave heights differed as SWL
the longshore direction and down the structure in the cross-shore increased. Not only did the magnitude of the wave heights in-
direction, a more prevalent undertow is generated near the
25-m crease, but the overall trend of the data differed. The wave
region of the floodwall side. This undertow then causes the incident heights tended to increase toward the transition for the lower
waves to refract. Also noticeable is the levee-side wave refraction, SWL tests and decreased toward the transition for the higher
which occurs at approximately þ20 m toward the levee. Because SWL tests; thus, the exponent ζ is unnecessary and is omitted in
water is flowing down the contours away from the levee side, a Eqs. (5)–(7).
divergence zone is created at the transition. As a result, water from As representative plots of the gauge data at the floodwall and
the levee side attempts to fill this divergence; this flow, coupled levee, the dimensionless floodwall MWL and levee MSP are plot-
with the receding water, causes this undertow and incident wave ted in Fig. 7. On each graph, all values from each test are plotted.
refraction evident at þ20 m. In essence, there are two circulations, As already noted, MSP and MWL are essentially equivalent and are
one rotating clockwise on the levee and another rotating counter- a combination of SWL and the vertical rise in water level resulting
clockwise over the transition region. Because of the wave refraction from the wave-induced setup. Unlike the dimensional plots, inner-
and current interaction, wave energy is focused at these locations, most run-up and floodwall gauge are not coincident because of the
yielding higher overtopping values. difference in the dimensionless parameter. Immediately noticeable
in Fig. 7(a) is the separation between the different SWL tests. The
Dimensionless Floodwall Wave Heights and Levee resulting dimensionless values tend to mimic the trends observed in
Run-up the dimensional plots. In Fig. 7(b), however, there is not as much
differentiation between the two SWLs, and the levee-side values
To analyze further the variable trends of the measured water level
tend to follow the same trend for both SWLs. In this case, the opti-
values and overtopping rates along the levee transition, as well as to
mization variable ζ is used to obtain the most accurate collapse to
compare these results with the empirical equations, a dimensionless
obtain a R2 value of approximately 0.60.
analysis is conducted. To collapse the wave heights and water
elevations experienced at the floodwall and the levee sections, these Dimensionless Overtopping Rates
parameters are normalized by a combination of the incident char-
acteristic wave height, H i , the freeboard, Rc , and the SWL, h. A free Overtopping rates are normalized by methods outlined in the liter-
parameter, ζ, is introduced as an exponent in the normalized values. ature, and as before, the original ratio must be further scaled to
To optimize the value of ζ, a MatLab® script is created to test a alleviate the separation induced by the variation in water depth
multitude of possibilities for the exponent’s value. Ultimately, among the tests. The resulting dimensionless overtopping rate
the script enables the user to define an array of values for ζ; it then parameter is
applies this value to the dimensionless parameter and fits the result-  ζ
Q R
ing values with a least-squares fit. The ensuing coefficient of de- q ¼ pffiffiffiffiffiffiffiffi3ffi c
t ð11Þ
termination, or R2 value, is then computed, and the most optimum gH i h

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 7
 Dimensionless Floodwall Mean Water Level
0.3
Test No. 01
Test No. 02
0.28 Test No. 03
Test No. 04
Test No. 05
Test No. 06
0.26
Test No.
No 07
Test No. 08
Curve Fit Low SWL, R2=0.46008
0.24
Curve Fit High SWL, R2=0.56641
Dimensionless MWL
((MWL/Hi)*(Rc/h)

0.22

0.2

0.18

0.16

0.14

0.12
0 12
-8 -7 -6 -5 -4 -3 -2 -1
kx (Dimensionless long-shore distance, Transition is datum)
(a)
Dimensionless Mean Shoreline Position
1.5

1.4

1.3
(MSP/Hi)*(Rc/h)ζ , ζ = (-0.46)
Dimensionless MSP

1.2

1.1

Test No. 01
1 Test No. 02
Test No. 03
Test No.
No 04
Test No. 05
0.9 Test No. 06
Test No. 07
Test No. 08
Curve Fit, R2=0.61119
0.8
0 1 2 3 4 5 6 7 8
kx (Dimensionless long-shore distance, Transition is datum)
(b)

Fig. 7. Dimensionless water level value plots of (a) floodwall wave heights and (b) levee run-up

where Q represents the measured overtopping rate, the superscript overtopping rates experienced at the floodwall section compared
‘t’ denotes the nondimensionality, and the exponent ζ provides the with the levee section, the two sections are independently analyzed.
means of optimizing the ratio, generating the most efficient col- Fig. 8(a) depicts the floodwall overtopping rates. The value of ζ is
lapse of the data, as already detailed. iteratively calculated to be 4.415, yielding an R2 value of nearly
The resulting plots of the dimensionless overtopping rates 0.40. The quadratic least-squares fit provides the general fit to
are displayed in Fig. 8. Because of the significantly different the data, expressing an overall increase in the dimensionless

8 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
parameter toward the transition, conveying that the overtopping equaling 2.81 and an R2 value of only 0.18, the plot suggests a
generally increases toward the transition. Fig. 8(b) provides the re- parabolic tendency along the levee section, denoting that the maxi-
sults of the dimensionless overtopping rates on the levee side. mum overtopping generally occurs around kx ¼ 2.
These values did not collapse as well as the floodwall overtopping It is reasonable to state that the dimensionless analysis does
rates, but they did maintain the overall parabolic trend. With a ζ not provide enough conclusive evidence to generate a confident

Floodwall Dimensionless Overtopping Rates

0.015
Test No. 01
Test No. 02
Test No. 03
Test No. 04
Test No. 05
Test No. 06
Test No.
No 07
(Q/(g*H3i )1/2)*(Rc/h)ζ , ζ = (4.415)

Test No. 08
Curve Fit, R2=0.38815
0.01
Dimensionless q

0.005

0
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0
kx (Dimensionless long-shore distance, Transition is datum)
(a)

Levee Dimensionless Overtopping Rates

0.7
Testt
T N
No. 01
Test No. 02
Test No. 03
0.6 Test No. 04
Test No. 05
Test No. 06
Test No. 07
(Q/(g*H3i )11/2)*(Rc/h)ζ , ζ = (2.81)

0.5 Test No. 08

Curve Fit, R2=0.17636
Dimensionlesss q

0.4

0.3

0.2

0.1

0
0 1 2 3 4 5 6 7 8
kx (Dimensionless long-shore distance, Transition is datum)
(b)

Fig. 8. Dimensionless overtopping rates at the (a) floodwall and (b) levee sections

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 9
expression relating the hydraulic and geometric parameters of the References
test to the overtopping rates experienced along the levee transition
structure. There is extreme variability along the levee transition, Hughes, S. A. (2008). “Levee overtopping design guidance: What we know
and what we need.” Proc. Solutions Coastal Disasters Congr., Vol. 78,
and the generic nondimensional methodologies are not applicable
L. Wallendorf, L. Ewing, C. Jones and B. Jaffe, eds., American Society
to such a nonlinear and three-dimensional evaluation. However, for of Civil Engineers, New York, 867–877.
engineering problems these results can at least provide approximate Hughes, S. A., and Nadal, N. C. (2008). “Laboratory study of combined
guidance for a problem where there is little. wave overtopping and storm surge overflow of a levee.” Coast. Eng.,
56(3), 244–259.
Link, L. E. (2010). “The anatomy of a disaster, an overview of Hurricane
Summary and Conclusions Katrina and New Orleans.” Ocean Eng., 37(1), 4–12.
Mansard, E. P. D., and Funke, E. R. (1980). “The measurement of incident
The research presented here outlines the evaluation and assessment and reflected spectra using a least squares method.” Proc. 17th Int.
of a specific levee transition structure tested under extreme design Coast. Eng. Conf., Vol. 2, American Society of Civil Engineers,
conditions. Tests varied hydraulic conditions such as incident wave New York, 154–172.
Nicholson, P. G. (2007). “Preliminary reconnaissance and observations
period, incident wave height, and still water level. Response of the
of the New Orleans levee and floodwall failures by independent
structure was investigated within a three-dimensional shallow water assessment teams.” Proc. Geo-Denver 2007: New Peaks Geotechnics
wave basin stationed within the Reta and Bill Haynes’46 Coastal Conf., Geotechnical Specialty Publications: Embankments, Dams,
Engineering Laboratory at Texas A&M University, College Station, and Slopes, Vol. 13, American Society of Civil Engineers, New York,
Texas. The results of the testing were analyzed both dimensionally 233–224.
and nondimensionally to help present the mechanisms contributing Pullen, T., Allsop, N. W. H., Burce, T., Kortenhaus, A., Schüttrumpf, H.,
to the measured floodwall wave heights, levee run-up, and overtop- and van der Meer, J. W. (2007). “EurOtop: Wave overtopping of sea
ping rate distribution along the levee transition structure. From the defences and related structures: Assessment manual.” 〈www
.[Link]〉.
results, there is evidence that the methodologies employed to es-
Schüttrumpf, H., and van Gent, M. R. A. (2003). “Wave overtopping at
timate the overtopping rates on the levee transition underestimate seadikes.” Proc. Coast. Struct. Conf., American Society of Civil Engi-
the measured values. The observed trends indicate spatial variabil- neers, New York, 431–443.
ity among the measured values along the levee transition caused by Sorensen, R. M. (2006). Basic Coastal Engineering, 3rd Ed., Springer Sci-
inherent three-dimensional effects, such as observed undertow gen- ence, NY, NY. 4
erated by the levee transition contours, which would otherwise be Sills, G. L., Vroman, N. D., Wahl, R. E., and Schwanz, N. T. (2008). “Over-
unnoticed without three-dimensional physical modeling. For the view of New Orleans levee failures: Lessons learned and their impact on 5
tests undertaken at the laboratory, the most influential parameter national levee design and assessment.” J. Geotech. Geoenviron. Eng.,
134(5), 556–565.
for the overtopping rates and water levels measured at the structure
Technical Advisory Committee for Flood Defence in the Netherlands
was the variation in the SWL. Longer peak periods also resulted in (TAW). (2002). “Wave run-up and wave overtopping at dikes.” Tech-
higher overtopping rates and water level values at the structure. nical Report compiled at the request of the Road & Hydraulic Engi-
Measured overtopping rates exceeded 300 L=s=m on the levee sec- neering Institute under the auspices of TAW, TAW, Delft.
tion of the structure and 100 L=s=m on the floodwall section for the Van der Meer, J. W., and Janssen, W. (1995). “Wave run-up and wave over-
extreme hydraulic conditions, all of which were underestimated by topping at dikes.” Wave Forces on Inclined and Vertical Wall Struc-
existing empirical methods. Floodwall wave crest elevations were tures, N. Kobayashi and Z. Demirbilek, eds., American Society of
measured up to 4–5 m above the floodwall crest elevation, and Civil Engineers, New York, 1–27.
Van Ledden, M., Lynett, P., Resio, D. T., and Powell, N. J. (2007). “Prob-
wave-induced setups allowed mean water levels to essentially equal
abilistic design method of levee and floodwall heights for the hurricane
the height of the structure. Overall, the measured values were sub- protection system in the New Orleans area.” Proc. 10th Int. Workshop
stantial, and would ultimately require the structure to be well pro- Wave Hindcasting Forecasting Coastal Hazard Symposium. JCOMM,
tected against the erosion effects induced by the hydraulic available online at: 〈[Link]
conditions. VanLeddenetal_Hawaii_2007.PDF〉.

10 / JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012
 Queries
1. Please check that ASCE Membership Grades (Member ASCE, Fellow ASCE, etc.) are provided for all authors that are members.
Also, please include the full location (eg., city, state, and/or country, as appropriate) for each author's affiliation.
2. Please confirm that 'a' was the top and 'b' the bottom in Fig. 2
3. Do you mean Texel/Marsen/ Arsloe by TMA?
4. Please indicate a cross reference for the Sorensen 2006 journal citation or indicate that it should be deleted from the References.
5. A check of online databases revealed a possible error in this reference. (Sills, G. L., Vroman, N. D., Wahl, R. E., and Schwanz, N.
T. [2008]) The issue has been changed from 'none' to '5'. Please confirm this is correct.

JOURNAL OF WATERWAY, PORT, COASTAL, AND OCEAN ENGINEERING © ASCE / JANUARY/FEBRUARY 2012 / 11