You are on page 1of 8

A Shock-Capturing TVD-WAF Finite Volume Model for Solitary Wave Run-Up

1 2
A. Mahdavi and N. Talebbeydokhti
1
Graduate student, Department of Civil Engineering, Shiraz University, Shiraz, Iran.
Email: amahdavi.ir@gmail.com
2
Professor, Department of Civil Engineering, Shiraz University, Shiraz, Iran.
Email: taleb@shirazu.ac.ir

Abstract
In this paper, a finite volume model based on the non-linear shallow water (NLSW)
equations is developed to investigate the propagation and run-up of non-breaking and
breaking solitary waves. The total variation diminishing version of weighted average
flux (TVD-WAF) explicit method in conjugation with the Harten-Lax-van Leer
(HLL) approximate Riemann solver estimates the numerical flux in the governing
equations. The bed topography and friction source terms are treated in a fully
implicit way. The accuracy of the model is verified by recourse to available
experimental data as well as existing run-up laws.
Introduction
The numerical study of the solitary wave propagation and run-up on a sloping beach
has received extensive attention in the coastal hydrodynamics. It may be due to the
fact that many characteristics of tsunamis can be simulated using the solitary waves
or combinations of negative and positive solitary-like waves (Li and Raichlen, 2001).
The non-linear shallow water (NLSW) equations have been widely employed to
describe the evolution and run-up process of solitary waves. Unfortunately, the
mathematical difficulties associated with wave breaking often cause drawback to
attain a fully theoretical approach. Therefore, efforts have been made in seeking for
numerical solutions of the NLSW equations. There have been a relatively large
number of numerical models that treat the moving shoreline by adding new grid
points during run-up and subtracting the points that were not covered by water during
run-down (e.g., Hibbert and Peregrine, 1979; Titov and Synolakis, 1995; Kowalik,
2001). Some numerical run-up models invoke the minimum depth criterion to locate
the shoreline position (e.g., Hu et al. 2000; Que and Xu 2005; Mahdavi and
Talebbeydokhti, 2009). A different shoreline-tracking approach is based on a
computing domain mapping which transforms the time-varying physical domain onto
a time invariant fixed-length computational domain (e.g., Zhang, 1996).
In the present work, a general numerical scheme has been developed for one-
dimensional free-surface flow problems including dry bed conditions and/or flow
discontinuities. The model was especially adopted for the calculation of solitary
wave run-up. This involves solving an initial value problem for NLSW equations.
The moving shoreline is automatically handled in the numerical scheme and no
ad-hoc term is required to eliminate the spurious oscillations near the breaking wave
front. The capability of the model has been verified by proper numerical tests.

33rd IAHR Congress: Water Engineering for a Sustainable Environment


c 2009 by International Association of Hydraulic Engineering & Research (IAHR)
Copyright °
ISBN: 978-94-90365-01-1
33rd IAHR Congress: Water Engineering for a Sustainable Environment

Governing equations
The propagation and run-up of two dimensional solitary waves can be appropriately
described by the non-linear shallow water equations as follows:
∂U ∂F
+ =S (1)
∂t ∂x
where the vector of conserved variables U, the flux vector F, and the source term S
are defined as:

 hu 
h     0 
U =   , F(U) = , S (U ) =   .
 hu   hu 2 + 1 gh 2   gh (S 0 − S f ) (2)
 2 
In the above equations, t denotes time, x is the horizontal coordinate taken to be
positive seaward with x=0 at the initial position of the shoreline, h(x,t) is the water
depth, u(x,t) is the depth-averaged velocity and g is the gravitational acceleration.
The bottom slope S 0 and the friction slope S f are respectively given by:
2
dz n Manning uu
S0 = − b Sf = (3)
dx h 4/3
where zb(x) is the bottom elevation and nManning denotes the Manning’s friction
coefficient. It is also appropriate to define the free surface elevation
asη(x,t)=h(x,t)+zb(x).

Evaluation of numerical flux


There are vast shock-capturing methods for solving the general initial value problem
for the shallow water equations with wet bed situations. However, the necessity of
reconstruction of the fluid motion over the dry coast is an important issue for the
numerical description of the run-up process. To alleviate this problem, the Harten-
Lax-van Leer (HLL) method is used herein. Based on the HLL approach, the
Riemann problem with data U L and U R is simplified as three constant states
separated by two waves. The numerical flux in the intermediate region of the wave
structure is determined by:

S R F ( U L ) − S L F ( U R ) + S R S L (U R − U L )
F ( U* ) = (4)
SR −SL
This flux together with F(U L ) ≡ F(U i ) and F(U R ) ≡ F(U i +1 ) will be further used to
estimate the inter-cell flux Fi +1/ 2 in which index i+1/2 denotes the interface between
two neighboring cells i and i+1. In (4), S L and S R represent the wave speed
estimates on the left and right sides of the cell interface i+1/2, respectively. These
wave speeds can be calculated by (Toro, 1992):

2477
33rd IAHR Congress: Water Engineering for a Sustainable Environment

S L = min(u L − ghL ,u * − gh * ) , S R = max(u R + ghR , u * + gh * ) (5)

In above expressions h * and u * denote, respectively, the flow depth and velocity in
the intermediate region of the wave structure. Using the two-rarefaction approach,
these flow variables can be evaluated by these closed-form solutions (Toro, 2001):
2
1 1 1  1
h* =  ( ghL + ghR ) + (u L − u R )  , u * = (u L + u R ) + ghL − ghR (6)
g 2 4  2

The aforementioned estimates of wave speeds are valid only when the entire
computational domain is covered by a finite water depth. However, in presence of a
dry bed region, these wave speeds should be replaced by the following analytical
wave front speeds which are essentially emerged from the exact solution of the dry
bed Riemann problem.

S L = u L − ghL S L = u R − 2 ghR


 (right dry bed)  (left dry bed) (7)
 R
S = u L + 2 gh L  R
S = u R + gh R

Indeed, this approach provides a relatively simple treatment of dry bed states as it
does not require significant modification to the basic numerical scheme or additional
damping term to deal with the moving wet-dry front. To compute the inter-cell
numerical fluxes, the weighted average flux (WAF) method is employed in which a
limiter function enforces the total variation diminishing (TVD) constraint on the
scheme. Thereby, sufficient dissipation is added to the scheme to guarantee the
monotonicity near large gradients of the solution. The TVD-WAF scheme preserves
second order accuracy in space and time and may be written as:

1 1 N
Fi +1/ 2 = (Fi + Fi +1 ) − ∑ sign(c k )φi(+k1/) 2 ∆Fi(+k1/) 2 (8)
2 2 k =1

where N denotes the number of waves in the solution of Riemann problem,


∆Fi(+k1/) 2 = Fi(+k1/+1)2 − Fi(+k1/) 2 is the flux jump across wave k in which Fi(1)
+1/ 2 = F ( U L ) ,

+1/ 2 = F ( U ) and Fi +1/ 2 = F ( U R ) , c k is the Courant number for wave k and φi +1 / 2


, Fi(2) * (3) (k )

is the limiter function. Some suitable choices for limiter function are reported by
Toro (2001). The well-known van Albada’s limiter has been used for the present
applications. It is given by:
1 for r ( k ) ≤ 0,

φi(+k1)/ 2 =  (1 − c k )r ( k ) (1 + r ( k ) ) (9)
1 − (k ) 2
otherwise.
 1 + (r )
(k)
where r is the ratio of the upwind change to the local change in flow depth.

2478
33rd IAHR Congress: Water Engineering for a Sustainable Environment

 ∆hi(−k1/) 2
 (k ) for c k > 0,
(k )  ∆hi +1/ 2
r =  (k ) (10)
 ∆hi +3 / 2 otherwise.
 ∆h ( k )
 i +1/ 2

where ∆hi(+kα) ( α = − 12 , 12 , 32 ) defines the jump in h across wave k in the solution


U i +α (x , t ) of the Riemann problem with data ( U 1 ,U 1 ).
i +α − i +α +
2 2

Treatment of the source term


In previous section, a numerical procedure for solving the homogeneous NLSW
equations has been explained. When a source term exists, the TVD-WAF scheme can
be implemented unaltered, provided that the source term is incorporated with either
one or more separate integration steps. This technique may offer a second-order
accurate splitting scheme for the governing equations that relies on solving the
following set of three initial value sub-problems in a successive manner.
dU 
ODEs: = S ( U )  ∆t ′
dt   → U (1)
ICs: Un 

∂U ∂F 
PDEs: + = 0  ∆t
∂t ∂x  → U
(2)
(11)
U (1) 
ICs: 
dU 
ODEs: = S ( U )  ∆t ′
dt   → U n +1
ICs: U (2) 

where superscript n denotes the current time level, ∆t is the time step size and
∆t′=∆t/2. To maintain the numerical stability, the magnitude of time step has to be
governed by the Courant–Friedrichs-Lewy (CFL) criterion. In the present work, the
trapezoidal time integration method is utilized to solve the source term parts. This
implicit scheme is second-order accurate and can be written for the first sub-problem
in Eq. (11) as:
 ∆t ′  ∂S(U ) n 
I −  ∆U i = ∆t ′ S(U i )
n
 
2  ∂U i 
(12)


In which I is the identity matrix and ∆U i = U i(1) − U in is the temporal jump in


conserved variables. The first intermediate value U (1) can be calculated by using
(12). By taking U (1) as the initial condition, the TVD-WAF scheme is then utilized in
an explicit conservative scheme to obtain the second intermediate value U (2) as:
∆t
U (2) = U (1) − (Fi +1 / 2 − Fi −1/ 2 ) (13)
∆x

2479
33rd IAHR Congress: Water Engineering for a Sustainable Environment

where ∆x is the spatial step size. Finally, applying the source term operator to
U (2) gives the solution at the new time level.

Initial conditions for modeling of solitary waves


The behavior of solitary waves propagating in a region of constant depth h0 and
running up a simple plane beach of angle β is investigated herein. The initial surface
profile is the first-order solitary wave of height H with the wave crest centered at the
position x=X.

 3H 
η (x , 0) = H sech 2  3 (
x − X ) (14)
 4h0 
The initial flow velocity is given by:

u (x , 0) = −η (x , 0) g h0 (15)
The minus sign in Eq. (15) implies that the wave moves toward the initial position of
the shoreline.

Numerical tests
a) Breaking solitary wave
The motion of non-linear breaking solitary wave with initial wave height H/h0=0.3 on
1:19.85 sloping beach was simulated by the scheme. The initial wave was located at
X=14h0 and the computational domain extending from x=−20h0 to x=25h0 was
discretized by employing 500 cells. These model inputs are analogous to those used
by Que and Xu (2005) except that 1000 cells were utilized in their gas-kinetic
Bhatnagar-Gross-Krook finite volume model. Furthermore, in our computation, a
Manning coefficient nManning=0.01m−1/3s defines the surface roughness of the material
used in the experiments of Synolakis (1986). In Fig. 1, the numerical results are
shown and checked against the experimental data. As can be seen in snapshots (b, c),
the wave forms a vertical front face and changes into a bore-type structure when it
propagates over the sloping surface. The laboratory wave commences to break at
some point in the above-mentioned snapshots. The breaking process terminates as
the bore collapses near the initial position of the shoreline (Fig. 1–d). Thereafter, the
attenuated wave generates a tongue propagating up the sloping beach (Fig. 1–e, f, g).
After the fluid reaches its maximum level, the run-down process is initiated in which
a thin layer of fluid accelerates down the beach. This supercritical flow creates a
hydraulic jump when impacting the water of wave tail indicating the breaking of this
solitary wave during run-down motion (Fig. 1–j, k, l). Toro (2001) used the Rankine-
Hougonit jump condition for NLSW equations and proved that a shock wave cannot
exist nearby a region of dry bed. In spite of this fact, some of the previous numerical
models failed to form a realistic wave profile at the first instances of motion of a
breaking wave over a dry bed (e.g., Titov and Synolakis, 1995; Zelt, 1991).
Therefore, a comparison is made in Fig. 2 to illustrate that the proposed model
provides a more accurate representation for this phenomenon.
2480
33rd IAHR Congress: Water Engineering for a Sustainable Environment

0.4 a) t*=0 0.4 g) t*=30


* *
0.2 0.2
0 0
−10 −5 0 5 10 15 20 −10 −5 0 5 10 15 20
0.4 b) t*=5 0.4 h) t*=35
* *
0.2 0.2
0 0
−10 −5 0 5 10 15 20 −10 −5 0 5 10 15 20
0.4 c) t*=10 0.4 i) t*=40
* *
0.2 0.2
0 0
−10 −5 0 5 10 15 20 −10 −5 0 5 10 15 20
0.4 d) t*=15 0.4 j) t*=45
* *
0.2 0.2
0 0
−10 −5 0 5 10 15 20 −10 −5 0 5 10 15 20
0.4 e) t*=20 0.4 k) t*=50
* *
0.2 0.2
0 0
−10 −5 0 5 10 15 20 −10 −5 0 5 10 15 20
0.4 f) t*=25 0.4 l) t*=55
* *
0.2 0.2
0 0
−10 −5 0 5 10 15 20 −10 −5 0 5 10 15 20
x* x*

Fig. 1. Run-up of a solitary wave with H/h0=0.3 on 1:19.85 sloping beach at


different non-dimensional times t*=t(g/h0)1/2. Solid lines represent numerical
simulation and symbols are experiments by Synolakis (1986). (η*=η/h0; x*=x/h0)

0.3
Titov and Synolakis(1995)
Zelt (1991)
0.2 Synolakis (1986)
Present Model
*
0.1

−0.1
−5 0 5 10 15 20
x*
Fig. 2. Bore collapsing for a solitary wave with H/h0=0.3 on 1:19.85 sloping
beach: Comparison with experimental data and other numerical models.

b) Prediction of maximum run-up height


Figure 3 depicts the normalized maximum run-up versus the normalized incident
wave height for different beach slopes of 1:11.43 and 1:15. The run-up heights

2481
33rd IAHR Congress: Water Engineering for a Sustainable Environment

corresponding to breaking criterion of Synolakis (1986), i.e. H/h0=0.818(cot β)−10/9 ,


are also computed for these slopes. In each slope structure, the numerical results
show a bilinear form with the two branches representing the non-breaking and
breaking regimes. The following run-up laws are used for comparison purposes in
these regions.
5/ 4 9/4
R H  H 
Non-breaking: = 2.831(cot β )1/ 2   + 0.293(cot β )3 / 2   (16)
h0  h0   h0 
1/ 2
R  MF 
Breaking: = (1.39 − 0.027 cot β )  2 
(17)
h0  ρ gh0 
In above equations, R is the maximum run-up height measured from the still water
level, ρ is the mass density of water and MF is the maximum depth-integrated wave
momentum flux across the unit width of the channel. The interested reader is referred
to Hughes (2004) for the definition of this parameter together with its physical
interpretation. A semi-theoretical expression for MF as a function of H/h0 is also
presented therein. The non-braking formula (Eq. 16) was derived by Li and Raichlen
(2001) based on a higher order theoretical solution to NLSW equations in which the
bottom friction effects were neglected. In order to compare the results with the
aforementioned run-up laws, a frictionless bed was assumed in the non-breaking
region while in the breaking region the bottom roughness was represented by a
friction coefficient of nManning=0.01m−1/3s. The results of this computational test
provide confidence for the capability of this shallow water model as a predictive tool
in estimation of the maximum run-up of both non-breaking and breaking solitary
waves.

1 0
10 10

0
10
R/h0 R/h0 -1
10
-1
10

(a) (b)
-2 -2
10 -3 -2 -1 0
10 -3 -2 -1 0
10 10 10 10 10 10 10 10
fgfgfg H/h
0
H/h
0

Fig. 3. Normalized maximum run-up (R/h0) versus normalized incident wave


height (H/h0) for solitary waves climbing on slopes: (a) 1:11.43 and (b) 1:15.
Comparison between numerical results (Squares: Non-breaking waves;
Triangles: Breaking criterion of Synolakis (1986); Circles: Breaking waves) and
two different run-up laws (Solid lines: Eq. 16; Dash lines: Eq. 17).

2482
33rd IAHR Congress: Water Engineering for a Sustainable Environment

Concluding remarks
A shock-capturing numerical scheme was presented for one-dimensional non-linear
shallow water equations. The developed model was applied to investigate the
propagation and run-up of solitary waves. The wave breaking was approximated in
the frameworks of bore propagation and hydraulic jump during the run-up and run-
down processes, respectively. Comparison with other numerical models showed that
the scheme provides a more accurate prediction of wave motion at the instant of bore
collapsing. Despite the depth-averaged limitation of the governing equations, the
predicted results agreed well with two existing run-up laws. This suggests the
potential capabilities of the model in flood prediction, hazards mitigation and
inundation modeling.

References
Hibbert, S., Peregrine, D. H. (1979). “Surf and run-up on a beach: a uniform bore.” J.
Fluid Mechanics, 95, 323–345.
Hu, K., Mingham, C. G., Causon, D. M. (2000). “Numerical simulation of wave
overtopping of coastal structures using the non-linear shallow water
equations” Coastal Engineering, 41, 433–465.
Hughes, S. A. (2004). “Estimation of wave run-up on smooth, impermeable slopes
using the wave momentum flux parameter.” Coastal Engineering, 51, 1085–
1104.
Kowalik, Z. (2001). “Basic relations between tsunami calculation and their physics.”
J. Science of Tsunami Hazards, 19(2), 99–115.
Li, Y., Raichlen, F. (2001). “Solitary wave run-up on plane slopes.” Journal of
Waterway, Port, Coastal, and Ocean Engineering, ASCE, 127, 33–44.
Mahdavi, A., Talebbeydokhti, N. (2009). “Modeling of non-breaking and breaking
solitary wave run-up using FORCE-MUSCL scheme.” Accepted by Journal
of Hydraulic Research.
Que, Y.-T., Xu, K. (2005). “The numerical study of roll-waves in inclined open
channels and solitary wave run-up.” International Journal for Numerical
Methods in Fluids, 50, 1003–1027.
Synolakis, C. E. (1986). The run-up of long waves, PhD Thesis, California Institute
of Technology, Pasadena, CA.
Titov, V. V., Synolakis, C. E. (1995). “Modeling of breaking and non-breaking long-
wave evolution and run-up using VTCS-2.” Journal of Waterway, Port,
Coastal, and Ocean Engineering, ASCE, 121, 308–461.
Toro, E. F. (1992). “Riemann problems and the WAF method for solving the two-
dimensional shallow water equations.” Philos. Trans. R. Soc. London, Ser. A,
338, 43–68.
Toro, E. F. (2001). Shock-capturing methods for free-surface shallow flows. Wiley,
West Sussex, England.
Zelt, J. A. (1991). “The run-up of non-breaking and breaking solitary waves.”
Coastal Engineering, 15(3), 205–246.
Zhang, J. E. (1996). Run-up of ocean waves on beaches, PhD Thesis, California
Institute of Technology, Pasadena, CA.

2483

You might also like