Sei sulla pagina 1di 13

ISSN 0015-4628, Fluid Dynamics, 2013, Vol. 48, No. 3, pp. 291–302. © Pleiades Publishing, Ltd., 2013.

Original Russian Text © O.A. Kovyrkina, V.V. Ostapenko, 2013, published in Izvestiya Rossiiskoi Akademii Nauk, Mekhanika Zhidkosti i Gaza,
2013, Vol. 48, No. 3, pp. 12–23.

Comparison between the Theory and the Numerical


Experiment in the Problem of Dam Break on a Jump
of the Cross-Sectional Area of a Rectangular Channel
O. A. Kovyrkina and V. V. Ostapenko
Received April 26, 2012

Abstract—Within the framework of the first approximation of the spatially one-dimensional shallow
water theory the problem of flow generated by the dam break on a jump of the cross-sectional area of
a rectangular channel is solved in the case in which the upper pool is wider than the lower one. The
self-similar solutions constructed contain an euristic parameter related with the amount of the total flow
energy lost on the cross-sectional area jump. The parameter is determined by means of comparing the
one-dimensional solutions with the results of the numerical modeling of the problem on the basis of
spatially two-dimensional equations of the shallow water theory.
Keywords: shallow water theory, cross-sectional area jump, dam break problem, self-similar solutions,
numerical experiment.
DOI: 10.1134/S0015462813030022

We will consider the problem of flows generated by the dam break on a jump of the cross-sectional area
of a rectangular channel. This problem is modeled on the basis of the first approximation of shallow water
theory [1–3], whose basic conservation laws are those of mass and total momentum. However, in the case of
a nonprismatic channel the total momentum conservation law (as distinct from the mass conservation law) is
not accurate and, thus, cannot be used for deriving the Hugoniot relation across the hydraulic jump formed
on the discontinuity in the channel cross-section area. As a result, there arises a problem of selecting an
additional relation on the fixed hydraulic discontinuity.
An analogous situation takes place in one-dimensional gas flows in a tube with a discontinuity in the
cross-sectional area. In [4, 5] in solving the problem of the discontinuity breakdown on a jump of the cross-
sectional area in a tube the momentum equation allowing for the reaction p of the wall connecting the ducts
of different diameters was used as the additional relation. In [6] the deficient relation on the cross-sectional
area jump was obtained—under the assumption of flow adiabaticity—from a differential consequence of
the basic gasdynamic equations, namely, the entropy conservation law which retains its conservation-law
form when the cross-sectional area changes. An analogous approach [7] was used in deriving the deficient
relation for the equations of isentropic gas dynamics; it is based on the application of the conservation-law
equation for the total energy, along with the continuity equation, on the cross-sectional area jump.
The same approach was employed in [8, 9] in deriving the additional relation on the hydraulic discon-
tinuity arising in a rectangular channel over a bottom step (depression), when wave flows in a fluid were
described on the basis of the shallow water equations. In this case, the local momentum conservation law
was used for obtaining the additional relation on the discontinuity above the step on the bottom; on this
discontinuity this law is equivalent to the law of conservation of the total energy of the oncoming flow. The
self-similar solutions of the problem of the dam break over a step [8] or a depression [9] in the bottom thus
constructed were in fairly good agreement with the results of benchmark experiments [10, 11], as regards
the types of the waves, the speed of their propagation, and the asymptotic depths behind their fronts.
In this study, in constructing self-similar solutions of the problem of the dam break on a jump of the
cross-sectional area of a rectangular channel, we will use, as in [8], the local momentum conservation law
in deriving the relations on the hydraulic jump.

291
292 KOVYRKINA, OSTAPENKO

Fig. 1. Top view of the channel; (1) dam.

1. FORMULATION OF THE PROBLEM


The differential equations of shallow water theory in a nonprismatic rectangular channel, when disre-
garding the friction and bottom slope effects, are as follows [1]:

wt + qx = 0, (1.1)
( )
1 1
qt + qv + ghw = gh2 bx , (1.2)
2 x 2
where h = h(t, x) is the water depth, q = q(t, x) is the water flow rate in the channel cross-section. w =
w(t, x) = bh is the flow cross-section area, b = b(x) is the channel width, v = v(t, x) = q/w is the cross-
section- average flow velocity, and g is the gravity acceleration.
Equations (1.1) and (1.2) represent the differential form of the mass and total momentum conservation
laws.
For system (1.1), (1.2) we will consider the dam break problem

⎨ hl , x ≤ 0,
h(0, x) = , hl > hr , v(0, x) = 0 (1.3)

hr , x > 0
on the cross-sectional area jump
⎧ ⎧
⎨ bl , x ≤ 0, ⎨ wl , x ≤ 0,
b(x) = , w(0, x) = , (1.4)
⎩ ⎩
br , x>0 wr , x>0
where wl = bl hl , wr = br hr , and wl > wr (Fig. 1).
We will seek the solution of this problem in the form of a combination of simple waves (that is, discon-
tinuous waves propagating at a constant velocity and centered depression waves), a fixed hydraulic jump
located in the origin on the cross-section area jump, and constant flow zones connecting the former regions.
Since the discontinuous wave arising on the breakdown of discontinuity (1.3), (1.4) propagates at x > 0
along the channel section with a constant width b = br (bx = 0), the corresponding Hugoniot conditions
obtained from the mass (1.1) and total momentum (1.2) conservation laws take the form:

D[w] = [q], (1.5)

D[q] = [qv + ghw/2], (1.6)

where D = xt is the discontinuous wave velocity and [ f ] = f1 − f0 is the jump of a function f (t, x) across
its front x = x(t), where f0 = f (t, x(t) + 0) and f1 = f (t, x(t) − 0).
Since for a nonprismatic channel, that is, for bx ∕= 0, the equation for the mass (1.1) has the conservation
law form, the corresponding Hugoniot relation (1.5) remains valid on the hydraulic jump formed on the

FLUID DYNAMICS Vol. 48 No. 3 2013


COMPARISON BETWEEN THE THEORY AND THE NUMERICAL EXPERIMENT 293

cross-sectional area discontinuity (1.4). Since this jump is fixed (D = 0), from Eq. (1.5) we obtain the
relation
[q] = 0, q1 = q0 = q(t, 0), (1.7)
which means the flow rate continuity across the cross-section area jump (1.4).
In the case of a nonprismatic channel the equation for the total momentum (1.2) is not an exact conserva-
tion law; because of this, on discontinuity (1.4) its right side becomes indefinite. This means that within the
framework of the formal shallow water theory [1–3] the Hugoniot relation (1.6) corresponding to Eq. (1.2)
cannot be used for deriving the second relation on discontinuity (1.4). Hence follows that, when two char-
acteristics of system (1.1), (1.2) proceed from this discontinuity, an additional relation is required for the
closure of the conditions on this discontinuity.
Following [8] we will derive an additional relation on the hydraulic discontinuity formed on jump (1.4)
from the conservation laws of system (1.1), (1.2) which admits the conservation-law form in the case of a
nonprismatic channel as well. Along with the mass conservation law (1.1), these conservation laws include
those for the local momentum and the total energy

vt + Jx = 0, (1.8)

et + (qJ)x = 0, (1.9)

where J = v2 /2 + gh is the Bernoulli function and e = h(v2 + gbh)/2 is the total energy of the flow.
Equations (1.8) and (1.9) can be derived as differential consequences of system (1.1), (1.2).
On the fixed discontinuity occurring on jump (1.4), from the Hugoniot relations for the conservation laws
(1.8) and (1.9) we obtain relations equivalent at q = q(t, 0) ∕= 0

[J] = 0, q[J] = 0. (1.10)

In going over from the wider channel section into the narrower part a certain fraction of the total energy
of the flow is lost as a result of the water impingement on the channel wall perpendicular to the main flow
direction. This loss in the total energy e means that its certain fraction is converted to the vortex motion
energy which cannot be taken into account within the framework of the one-dimensional model of shallow
water theory. For this reason, if two characteristics of system (1.1), (1.2) proceed from discontinuity (1.4),
we will use the modified relation (1.10)
( )
1 2 1
σ J1 = J0 , σ v1 + gh1 = v20 + gh0 , (1.11)
2 2

as a relation additional with respect to Eq. (1.7). The left side of this equation includes an euristic parameter
σ ∈ (0, 1] which specifies the fraction of the total energy e conserved across discontinuity (1.4).
We will choose the value of this parameter by means of comparing with the results of the numerical
modeling of this problem on the basis of the spatially two-dimensional equations of shallow water theory.
At σ = 1 there is no loss in the total energy of the flow across the cross-sectional area jump (1.4).

2. PERMISSIBLE FLOWS ACROSS A CROSS-SECTION AREA JUMP


Rewriting Eq. (1.11) in terms of the flow rate continuity equation (1.7) yields
( )
q2 q2 √ gw20 w21 (σ h1 − h0 )
σ + 2gh1 = + 2gh0 , q2 (w1 − σ w0 ) = , (2.1)
w21 w20 w̄

where q = q(t, 0) ∕= 0 and w̄ = (w1 + σ w0 )/2.

FLUID DYNAMICS Vol. 48 No. 3 2013


294 KOVYRKINA, OSTAPENKO


We will show that w1 − σ w0 ∕= 0. Otherwise, from the second equation (2.1) we obtain
√ √
w1 = σ w0 , σ h1 = h0 , bl h1 σ br h0 = σ 3/2 br h1 , σ 3/2 = bl /br > 1,

which contradicts the condition σ ∈ (0, 1]. In view of this, the second equation (2.1) can be rewritten in the
form:
gw2 w2 (σ h1 − h0 )
q2 = 0 1 √ . (2.2)
w̄(w1 − σ w0 )
From Eq. (2.2) it follows that on discontinuity (1.4), on which the energy relation (1.11) holds, only such
flows are possible for which

(σ h1 − h0 )(w1 − σ w0 ) > 0. (2.3)
In view of the fact that σ ∈ (0, 1] and bl > br , inequality (2.3) establishes two permissible flow configu-
rations on the cross-section area jump (1.4)

σ h1 > h0 , w1 > w0 , (2.4)



w1 < σ w0 , h1 < h0 . (2.5)

According to inequality (2.3), an intermediate configuration, for which σ h1 < h0 and w1 > σ w0 , is
inadmissible. This means that the corresponding discontinuous solutions are possible only provided that
three characteristics of system (1.1), (1.2) arrive at discontinuity (1.4) and only one characteristic proceeds
from it. In this case, the flow rate continuity (1.7) is sufficient for the closure of relations (1.4) on the
discontinuity. In this study, these solutions are not considered.
Theorem 1. The flow to the left of discontinuity (1.4) at x = 0−0√is subcritical, ∣v1 ∣ < c1 , under conditions
(2.4) and supercritical, ∣v1 ∣ > c1 , under conditions (2.5), where c = gh is the small disturbance propagation
velocity.
Proof. From Eq. (2.2) we obtain

q2 gw20 (σ h1 − h0 )
v21 = = √ . (2.6)
w21 w̄(w1 − σ w0 )

We will demonstrate that under conditions (2.4) the following inequality holds

v21 < c31 = gh1 , (2.7)

while under conditions (2.5) it is the inequality

v21 > c31 = gh1 (2.8)

that is valid.
√ √
Since in the former case w1 − σ w0 > 0 and in the latter case w1 − σ w0 < 0, multiplying inequalities

(2.7) and (2.8) by bl (w1 − σ w0 ) and rewriting them in terms of Eq. (2.6) yields
√ √
bl (h1 − h0 /σ )( σ w0 )2 < w̄w1 (w1 − σ w0 ). (2.9)
√ √ √
Since under conditions (2.4) w1 > σ w0 and w̄ = (w1 + σ w0 )/2 > σ w0 , while under conditions
√ √ √
(2.5) w1 < σ w0 and w̄ = (w1 + σ w0 )/2 < σ w0 , inequality (2.9), in view of the fact that σ ∈ (0, 1]
and bl > br , follows from the chain of inequalities bl (h1 − h0 /σ 2 ) ≤ bl (h1 − h0 ) < w1 − w0 ≤ w1 − σ w0 .
The theorem is proved.
We will write Eq. (2.1) in the form:

FLUID DYNAMICS Vol. 48 No. 3 2013


COMPARISON BETWEEN THE THEORY AND THE NUMERICAL EXPERIMENT 295

σ J(h1 , bl ) = J(h0 , br ), (2.10)


where
J(h, b) = q2 /(2b2 h2 ) + gh (2.11)
is the Bernoulli function in which the flow rate q = w0 v0 = br h0 v0 = w1 v1 = bl h1 v1 is fixed. Since
q2 q2 v2
Jh (h, b) = g − = g − = g − ,
b2 h3 w2 h h
the function J(h, b) reaches a minimum with respect to h
( )
v2c 3 3 gq 2/3
min J(h, b) = J(hc , b) = + ghc = ghc =
h 2 2 2 b

in the critical flow at the point hc = v2c /g = 3 q2 /(gb2 ). For this reason, the function J(h, b) determines
supercritical flows (∣v∣ > c) at h < hc , when Jh < 0, and subcritical flows (∣v∣ < c) at h > hc , when Jh > 0.
By virtue of Theorem 1, hence follows that for stable flows Eq. (2.10) is uniquely resolvable with respect
to h1 for all√ h0 > 0 and with respect to h0 for all h1 satisfying the inequality σ J(h1 , bl ) > J(h0c , br ),
where hc = q2 /(gb2r ). If σ J(h1 , bl ) = J(h0c , br ), then within
0 3
√ the framework of stable flows Eq. (2.10)
is resolvable with respect to h0 only if h1 > h1c , where h1c = 3
q2 /(gb2l ), that is, when the flow (h1 , v1 ) is
subcritical.

3. CONSTRUCTION OF SELF-SIMILAR SOLUTIONS


We will show that for all values of σ ∈ (0, 1] the dam break problem (1.1)–(1.4) is uniquely resolvable
within the framework of conditions (1.7) and (1.11) on discontinuity (1.4). Since at hl > hr on the breakdown
of the discontinuity (1.3), (1.4) the fluid flows in the positive direction (v > 0), at x > 0 a discontinuous
wave S propagates against the background hr , while at x < 0 a depression wave R propagates against the
background hl . To the left of the hydraulic jump (1.4) (we will denote this steady discontinuity by L) there is
formed a constant subcritical flow zone connecting the discontinuity L with the depression wave R (Fig. 2a).
This means that the solution of problem (1.1)–(1.4) always involves configuration (2.4) on discontinuity L.
To the right of discontinuity L the flow pattern depends on the flow type behind the front of the discontinuous
wave S.
From Theorem 1 it follows that, provided conditions (1.7) and (1.11) on discontinuity L are fulfilled, to
the left of this discontinuity the depth h1 and the flow velocity v1 are uniquely determined by their values
h0 and v0 to the right of it. In view of this, we will consider a one-parameter family of stable discontinuous
solutions with the depths and velocities h0 = h and v0 = v(h) to the right of discontinuity L and h1 = H(h)
and v1 = V (h) to the left of it. √
Theorem 2. If a function v(h) satisfies the inequalities c ≥ v > 0 and vh > 0, where c = gh, then the
√ and V (h) satisfy the inequalities C > V > 0, Hh > 0, Vh > 0 ⇒ C > U > 0, and
associated functions H(h)
UH > 0 in which C = gh and U (H) = V (h(H)), where h(H) is the inverse function of H(h).
Proof. We will rewrite Eq. (2.10) in the form:
( 2 )
q q2
σ + 2gH = + 2gh, (3.1)
B2 H 2 b2 h2
where b = br and B = bl . Taking the total differential of both sides of Eq. (3.1) we obtain
α1 σ dH = α0 dh + (v/w − σ V /W ) dq, (3.2)
where w = bh, W = BH, α0 = (c2 − v2 )/h, and α1 = (C2 − V 2 )/H. By virtue of the conditions of the
theorem, it is the first permissible configuration (2.4), for which W > w, α0 ≥ 0, and α1 > 0, that is realized
on discontinuity L. For this reason, from Eq. (3.2) we obtain

FLUID DYNAMICS Vol. 48 No. 3 2013


296 KOVYRKINA, OSTAPENKO

Fig. 2. Characteristic wave profiles in the self-similar solutions for the subcritical (a) and supercritical (b) flows behind the
wave S front; (1) σ = 1 and (2) σ < 1.
( )
1 q(W 2 − σ w2 )
Hh = α0 + qh > 0, (3.3)
α1 σ w2W 2

where q = wv > 0 and qh = (wv)h = b(v + hvh ) > 0.


To determine the sign of the derivative Vh we will write it in the form:

Vh = (q/W )h = (W qh − qWh )/W 2 ,

where Wh = BHh . Thence, in view of inequality (3.3), we obtain

( ( ))
1 Bq q(W 2 − σ w2 )
Vh = 2 W qh − α0 + qh
W α1 σ w2W 2

B((α1 σ H + σ V 2 − v2 )qh − α0 q B((σ C2 − c2 )qh + α0 whvh


= = > 0,
α1 σ W 2 α1 σ W 2

where, in view of Eq. (2.4), σ C2 = σ gH > c2 = gh. The theorem is proved.


Since the shock adiabat

g(h + hr )
v = vs (h, hr ) = (h − hr ), h > hr , (3.4)
2hhr
on which the flow parameters behind the discontinuous wave S front lie, is a positive, rigorously monoton-
ically increasing function of h, by virtue of Theorem 2, its part presented in Fig. 3 as the segment [A1 , A2 ]
lying in the subcritical flow domain, is mapped by Eq. (3.1) across discontinuity L in the rigorously mono-
tonically increasing function v = us (h, hr ), whose plot is shown in Fig. 3 as the segment [A1 , B2 ] lying in
the subcritical flow domain. The critical flow line

v = vc (h) = gh (3.5)

is also mapped by Eq. (3.1) in the rigorously monotonically increasing function v = uc (h), whose plot for
the subcritical flow domain is presented in Fig. 3 as the segment [0, B2 ] going over into the line√B2C. Since
the shock adiabat (3.4) and the critical flow line (3.5) intersect at the single point A2 = (h∗r , gh∗r ), their
images under mapping (3.1) intersect in the subcritical flow domain at the single point B2 = (h3 , v3 ).
Since behind the depression wave R propagating against the hl background the constant flow parameters
(h, v) lie on the shock adiabat
√ √
v = vr (h, hl ) = 2( ghl − gh) (h < hl ) (3.6)

proceeding from a point hl > hr on the h axis, the unique solvability of the discontinuity breakdown prob-
lem (1.1)–(1.4) follows from a rigorously monotonic decrease of adiabat (3.6) and a rigorously monotonic

FLUID DYNAMICS Vol. 48 No. 3 2013


COMPARISON BETWEEN THE THEORY AND THE NUMERICAL EXPERIMENT 297

Fig. 3. Adiabat diagram for constructing self-similar solutions.

increase of the piecewise-smooth function



⎨ us (h, hr ), hr < h ≤ h3
v = u(h, hr ) = (3.7)

uc (h), h ≥ h3

whose plot is shown in Fig. 3 as the line A1 B2C. The region A1 B2 is the image of the subcritical part A1 A2
of the shock adiabat vs and the region B2C is the image of an interval of the critical flow line vc located to
the right of point A2 , that is, at h > h∗r .
We will denote by h∗l a point on the h axis from which there proceeds the shock adiabat vr passing through
point B2 of the bend in the line A1 B2C. Then at hl ∈ (hr , h∗l ] the adiabat vr intersects this line on the interval
A1 B2 , which leads to the formation of the flow pattern shown as line 1 in Fig. 2a. Here, the depression wave
R is associated with the interval D4 B4 of the shock adiabat vr , where B4 = (h1 , v1 ), the discontinuity L with
the shock transition B4 A4 in accordance with Eq. (2.1), where A4 = (h0 , w0 ), and the discontinuous wave S
with the interval A4 A1 of the shock adiabat vs .
At hl > h∗l the adiabat vr intersects the line A1 B2C on the interval B2C, which leads to the formation of
the flow pattern shown as the continuous line in Fig. 2b. Here, the depression wave R is associated with
the interval D3C3 of the shock adiabat vr , where C3 = (h1 , v1 ), the discontinuity L with the shock transition
C3 E in accordance with Eq. (2.1), where E = (h0 , v0 ), the depression wave R1 adjoining the discontinuity
L from the right with the interval EA3 of the wave r-adiabat
√ √
v = ur (h, h0 , v0 ) = 2( gh0 − gh) + v0 , h < h0 ,

where A3 = (h2 , v2 ), and the discontinuous wave S with the interval A3 A1 of the shock adiabat vs . The flow
at x > 0 is obtained as a result of the solution of the classical discontinuity breakdown problem with the
initial data located at points A1 and E in Fig. 3.
Since the Bernoulli function (2.11) rigorously monotonically increases in the subcritical flow region at
h > hc , from Eq. (3.1) it follows that, at a given flow (h, v) to the right of the discontinuity L, to the
left of this discontinuity the depth H rigorously monotonically increases with decrease in the parameter
σ (with increase in the energy loss on the discontinuity L) and rigorously monotonically decreases with
increase in this parameter. Contrariwise, the velocity V = q/(BH) to the left of the discontinuity L rigorously
monotonically decreases with decrease in the parameter σ and rigorously monotonically increases with its
increase. As a result, at σ ′ < σ the plot of the piecewise- continuous function (3.7), shown as the line
A1 B′2C′ in Fig. 3, is located to the right of and beneath the line A1 B2C in the same figure. Hence follows
that in the solutions presented in Fig. 2 the depth h1 increases with decrease in the parameter σ , while all
the other flow parameters, namely, the depths h0 and h2 , the flow velocities v0 , v1 , and v2 , and the velocity
of the discontinuous wave D, decrease. In Fig. 2 the corresponding solutions for a value σ ′ < σ are shown
as curves 2.

FLUID DYNAMICS Vol. 48 No. 3 2013


298 KOVYRKINA, OSTAPENKO

4. NUMERICAL MODELING ON THE BASIS OF THE TWO-DIMENSIONAL


EQUATIONS OF SHALLOW WATER THEORY
We will compare the self-similar solutions of problem (1.1)–(1.4) constructed above with the results of
its numerical simulation on the basis of the two-dimensional equations of shallow water theory. Without
regard for the friction and bottom slope effects these equations take the form [1]:

ht + (q1 )x + (q2 )y = 0, (4.1)


( )
1
(q1 )t + q1 v1 + gh2 + (q1 v2 )y = 0, (4.2)
2 x
( )
1
(q2 )t + (q2 v1 )x + q2 v2 + gh2 = 0, (4.3)
2 y

where h(t, x, y) is the water depth, q(t, x, y) = (q1 (t, x, y), q2 (t, x, y)) is the water flow rate vector,
v = (v1 , v2 ) = q/h is the depth-average flow velocity vector, and g is the gravity acceleration. Equations
(4.1)–(4.3) represent the differential form of the mass and two-dimensional total momentum conservation
laws.
Assuming that channel (1.4) possesses the longitudinal axial symmetry we will take this axis for the x
axis and direct the y axis perpendicular to the former (Fig. 1). With account for this, we will solve system
(4.1)–(4.3) in the domain
{ }
bl br
G = (x, y) : ∣x∣ ≤ X ; ∣y∣ ≤ , x ≤ 0; ∣y∣ ≤ , x > 0 (4.4)
2 2
with the initial conditions

⎨ hl , x ≤ 0,
h(0, x, y) = v1 (0, x, y) = v2 (0, x, y) = 0 (4.5)

hr , x > 0,

corresponding to the one-dimensional conditions (1.3) and the impermeability conditions on the domain G
boundary.
Since the flow under consideration contains a discontinuous wave, that is, a rapid wave process, its
numerical modeling is realized on the basis of a conditionally stable conservative difference scheme with
artificial viscosities, which does not involve sweeping techniques and iteration procedures with respect
to nonlinearity. An analogous scheme was used in [12] in modeling wave flows due to the descent of a
shore landslide, in [13] in calculating the discontinuous wave propagation along a dry bed, and in [14] in
modeling the evolution of homogeneous vorticity in a rectangular reservoir with an inclined bottom upon a
sharp deceleration of its rotation.
In Fig. 4 lines 1 and 3 show the flow depth h(t, x) obtained by means of constructing the self-similar solu-
tion of the one-dimensional problem (1.1)–(1.4). In this figure lines 2 and 4 correspond to the channel-width-
average depth values h(t, x, y) obtained in numerically solving the two-dimensional problem (4.1)–(4.5) for
the same values of the input parameters bl , br , hl , and hr . The averaging was carried out using the expression


b(x)/2
1
h̄(t, x) = h(t, x, y) dy, (4.6)
b(x)
−b(x)/2

where the integral was approximately calculated using the trapezoid formula.
Plots 1 and 2 in Fig. 4 correspond to the first-type solutions (Fig. 2a) for which hl ∈ (hr , h∗l ], which leads
to the formation of a subcritical flow behind the discontinuous wave S front. Plots 3 and 4 are associated with

FLUID DYNAMICS Vol. 48 No. 3 2013


COMPARISON BETWEEN THE THEORY AND THE NUMERICAL EXPERIMENT 299

Fig. 4. Comparison of the theoretical (1 and 3) and channel-width-average numerical (2 and 4) solutions for the subcritical
(1 and 2) t = 5 s and supercritical (3, 4) t = 4.5 s flow regimes ; (a) σ = 1, bl /br = 2; (b) σ = 0.89, bl /br = 3/2; and (c)
σ = 0.86, bl /br = 2.

the second-type solutions (Fig. 2b), for which hl > h∗l , which leads to the formation of a supercritical flow
behind the discontinuous wave S front. In both cases the initial lower-pool depth hl = 1 m, while the upper
pool depth hr = 4.5 m for the first-type solutions and hr = 9 m for the second-type solutions. The first-type
solutions are presented for t = 5 s and those of the second type for t = 4.5 s.
In Fig. 4a the exact solutions h(t, x) are obtained at σ = 1, that is, under the assumption that the total en-
ergy of the one-dimensional flow is conserved across the cross-sectional area jump. As a result, in the numer-
ical solution of the analogous two-dimensional problem, in which the total energy partially goes over into
the vortex motion energy across jump (1.4), the velocity of the discontinuous wave front propagation is con-
siderably smaller and the mean depth to the left of discontinuity L is greater than in the self-similar solutions.
In Fig. 4b and 4c, the solutions h(t, x) of problem (1.1)–(1.4) are obtained at the parameter values
σ < 1 which were chosen from the condition of the agreement between the exact and numerical solutions
concerning the location of the discontinuous wave S front. As a result, for the solutions presented in Fig. 4b
at bl /br = 3/2 the optimal value σ ≈ 0.89 and for the solutions presented in Fig. 4c at bl /br = 2 σ ≈ 0.86.
The calculations also indicate that for bl /br = 3 the optimal value σ ≈ 0.84. The comparison of the exact and
numerical solutions shows that the parameter σ determining the total energy loss across the discontinuity L
considerably depends on the value of jump (1.4), that is, on the ratio bl /br and, this ratio given, only slightly
depends on the initial discontinuity in the depth, that is, on the values hl > hr .
As follows from the calculations (Fig. 4b and 4d), a considerable difference in the averaged difference
solutions h̄(t, x) and the exact self-similar solutions is observable to the right of discontinuity L and mani-

FLUID DYNAMICS Vol. 48 No. 3 2013


300 KOVYRKINA, OSTAPENKO

Fig. 5. Two-dimensional wave surface at the moment t = 4.5 s for bl /br = 2, σ = 0.86, hl = 9 m, and hr = 1 m.

Fig. 6. Velocity field of the two-dimensional wave in the vicinity of the channel cross-section area jump.

Fig. 7. Plots of the flow depth variation across the flow in the lower pool; (1 and 2) relate to x = 2.625 and 7.575.

FLUID DYNAMICS Vol. 48 No. 3 2013


COMPARISON BETWEEN THE THEORY AND THE NUMERICAL EXPERIMENT 301

fests itself as a certain drop in the depths obtained in the difference solutions which is most clearly expressed
in the case of the first-type solutions (Fig. 2a), when the depression wave R1 is absent from the region to the
right of discontinuity L. The presence of this drop is due to the essentially two-dimensional nature of the
flow in the vicinity of jump (1.4). To illustrate this fact, in Fig. 5 for the moment t1 = 4.5 s we have plotted
the numerical function η (x, y) = h(t1 , x, y) determining the surface of the two-dimensional wave obtained
in solving problem (1.1)–(1.4) with the parameters bl /br = 2, hl = 9 m, and hr = 1 m.
Figure 6 presents the velocity field of this wave in the vicinity of discontinuity L. In Fig. 7 the plots of
the numerical functions ηi (y) = h(t1 , x, y), i = 1, 2 are presented; they show the flow depth variation across
the flow in the lower pool at x1 = 2.625 and x2 = 7.575.
From the comparison of the calculated results for different ratios bl /br (Fig. 4b and 4c) there follows
a fairly natural conclusion that the agreement between the exact self-similar and averaged difference solu-
tions worsens with increase in the ratio bl /br , that is, with increase in the cross-sectional area jump (1.4).
The numerical calculations also show that these solutions come gradually closer with time and distance from
discontinuity L, which is due to the asymptotic transition, as t, ∣x∣ → ∞, of the two-dimensional solution of
problem (4.1)–(4.5) to the one-dimensional solution governing the plane-parallel fluid flow.
Summary. The unique solvability of the problem of the dam break on a jump of the cross-sectional area
of a rectangular channel is shown under the condition that the channel width in the upper pool is greater than
in the lower pool. On the discontinuity L, which arises on the cross-sectional area jump the energy, equation
(1.11) is fulfilled; it contains an euristic parameter σ ∈ (0, 1] specifying the portion of the total flow energy
which is conserved across the discontinuity L.
A particular value of σ was chosen by means of comparing the one-dimensional solutions with the results
of the numerical modeling of the same problem on the basis of the two-dimensional equations of shallow
water theory. The exact self-similar solutions thus constructed are in fairly good agreement with the results
of the two-dimensional numerical calculations, as concerns the possible wave types, the velocities of their
propagation, and the depths behind their fronts.
In the future it is projected to compare the exact and numerical solutions obtained with the data of labo-
ratory experiments.

The study was carried out with the support of the Council on Grants of the Presidium of the Russian
Academy of Sciences for the Support of Young Scientists (grant MK-3477.2013.1), the Joint Project of
Basic Research of the National Academy of Sciences of Ukraine and the Siberian Division of the Russian
Academy of Sciences No. 8, the Project of the Basic Research of the Presidium of the Russian Academy of
Sciences No. 4.8, and the Integration Project of the Siberian Branch of the Russian Academy of Sciences
No. 132.
REFERENCES
1. J.J. Stoker, Water Waves. The Mathematical Theory with Applications, Wiley, New York (1957).
2. A.G. Kulikovskii, N.V. Pogorelov, and A.Yu. Semenov, Mathematical problems of the Numerical Solution of
Hyperbolic Systems of Equations [in Russian], Fizmatlit, Moscow (2001).
3. V.V. Ostapenko, Hyperbolic Systems of Conservation Laws and their Application to Shallow Water Theory [in Rus-
sian], Novosibirsk Univ. Press, Novosibirsk (2004).
4. V.G. Dulov, “Breakdown of an Arbitrary Discontinuity of Gas Parameters on a Cross-Section Area Jump,” Vestn.
Leningr. Un-ta. Ser. Mat. Mekh. Astron. No. 19, Issue 4, 76 (1958).
5. I.K. Yakushev, “Breakdown of an Arbitrary Discontinuity in a Channel with a Cross-Sectional Area Jump,” Izv.
Sib. Otd. Akad Nauk SSSR. Ser. Tekhn. Nauk No. 8, Issue 2, 109 (1967).
6. B.L. Rozhdestvenskii and N.N. Yanenko, Systems of Quasilinear Equations and their Applications to Gas Dy-
namics [in Russian], Nauka, Moscow (1978).
7. A. Kahane, W.R. Warren, W.C. Griffith, and A.A. Marino, “Theoretical and Experimental Study of Finite Ampli-
tude Wave Interaction with Channels of Varying Area,” J. Aeronaut. Sci. 21, 505 (1954).
8. V.V. Ostapenko, “Flows Generated by the Dam Break over a Step on the Bottom,” Zh. Prikl. Mekh. Tekhn. Fiz.
44 (4), 51 (2003).

FLUID DYNAMICS Vol. 48 No. 3 2013


302 KOVYRKINA, OSTAPENKO

9. V.V. Ostapenko, “Flows Arising upon Dam Breakup over a Depression in the Bottom,” Zh. Prikl. Mekh. Tekhn.
Fiz. 44(6), 107 (2003).
10. V.I. Bukreev, A.V. Gusev, and V.V. Ostapenko, “Breakdown of a Discontinuity of the Free Fluid Surface over a
Bottom Step in a Channel,” Fluid Dynamics 38 (6), 889 (2003).
11. V.I. Bukreev, A.V. Gusev, and V.V. Ostapenko, “Waves in an Open Channel Formed at Removal of a Shield ahead
of an Uneven Bottom of Shelf Type,” Vodn. Resursy 31, 540 (2004).
12. V.V. Ostapenko, “Numerical Modeling of Wave Flows Due to a Landslide,” Zh. Prikl. Mekh. Tekhn. Fiz. 40(4),
109 (1999).
13. M.N. Borisova and V.V. Ostapenko, “Numerical Modeling of Propagation of Discontinuous Waves along a Dry
Bed,” Zh. Vychisl. Mat. Mat. Fiz. 46, 887 (2006).
14. D.G. Akhmetov, V.V. Nikulin, and V.V. Ostapenko, “Vorticity Simulation in a Rectangular Tank with a Sloping
Bottom after a Sharp Deceleration in Rotation,” Fluid Dynamics 41 (6), 938 (2006).

FLUID DYNAMICS Vol. 48 No. 3 2013


Copyright of Fluid Dynamics is the property of Springer Science & Business Media B.V. and
its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

Potrebbero piacerti anche