3.
GRADUALLY-VARIED FLOW (GVF) AUTUMN 2010
3.1 The Gradually-Varied-Flow Equation
At any horizontal location x the total head is
V2 h V2
H = zs + = zb + + (1)
2g cos 2g
where zs is the level of the free surface and zb is the level of the bed.
zs
g
h
zb cos θ h
θ
Although not crucial, we make the small-slope assumption, that cos ≈ 1 , and make no
distinction between the depth h perpendicular to the bed (which is used to get the flow rate)
and the vertical depth h / cos (which is required in the hydrostatic-pressure calculation).
In terms of specific energy,
H = zb + E (2)
where E is the head relative to the bed:
V2
E = h+ (3)
2g
In frictionless flow, H = constant; i.e. the energy grade line is horizontal. In practice, H
decreases with distance due to bed friction (cf pipe flow).
Differentiate (2):
dH dz b dE
= + (4)
dx dx dx
Write:
dH
= −S f (5)
dx
dz b
= −S 0 (6)
dx
Sf is the downward slope of the energy grade line, or friction slope; (more about how this is
calculated later). S0 is the actual geometric slope. Then
dE
= S0 − S f (7)
dx
Thus, the specific energy only changes if there is a difference between the geometric slope
and the friction slope; i.e., a difference between the rates at which gravitational forces drive
the flow and friction forces retard it.
Hydraulics 3 Gradually-Varied Flow - 1 Dr David Apsley
Equations (5) and (7) are two forms of the gradually-varied-flow equation. However, the
third, and most common, form rewrites dE/dx in terms of the rate of change of depth, dh/dx.
V2 dE dh V dV
E = h+ ⇒ = + (8)
2g dx dx g dx
But, with bs the width of the channel at the surface:
Q dV Q dA
V = ⇒ =− 2 bs
A dx A dx
dV Qb dh h
dA = b s dh ⇒ = − 2s A
dx A dx
Vb dh
=− s
A dx
V dh
=−
h dx
Substituting in (8) gives the rate of change of E in terms of that of h:
dE dh V2
= (1 − )
dx dx gh
dh
= (1 − Fr 2 )
dx
Combining this with (7) gives, finally,
Gradually-Varied-Flow Equation
dh S 0 − S f
= (9)
dx 1 − Fr 2
3.2 Finding the Friction Slope
Since the flow is only gradually-varying the local friction slope Sf is evaluated on the “quasi-
uniform-flow” assumption that it is the same rate of energy loss as that in normal flow of the
same depth; e.g. using Manning’s equation:
1
V = Rh2 / 3 S 1f / 2
n
Inverting for the friction slope:
n 2V 2
S f = 4/3 (10)
Rh
Both V and Rh should be written in terms of the depth h (depending on the channel shape).
In general, the deeper the flow then the smaller the velocity and friction losses. For example,
for a wide channel, V ∝ 1/h and Rh = h, so Sf ∝ h–10/3. Qualitatively:
greater depth … lower velocity … smaller Sf ;
smaller depth … higher velocity … greater Sf .
Hydraulics 3 Gradually-Varied Flow - 2 Dr David Apsley
3.3 Profile Classification
For any given discharge (but independent of slope) there is a particular critical depth hc, at
which Fr = 1. For example, in a wide or rectangular channel, hc = (q 2 /g )1 / 3 .
For any given discharge and slope there is a particular normal depth hn, associated with
uniform flow. For example, in a wide channel, hn = (nq / S ) 3 / 5 .
The normal depth is that to which the flow would tend given a long enough undisturbed
fetch. A downward slope is classed as steep if the normal depth is less than the critical depth
(i.e. the normal flow is supercritical) and mild if the normal depth is greater than the critical
depth (i.e. the normal flow is subcritical).1
In general, given the actual depth h, critical depth hc and normal depth hn, simple inspection
of signs of numerator and denominator on the RHS of the GVF equation
dh S 0 − S f
=
dx 1 − Fr 2
will tell us whether depth is increasing or decreasing with distance. In the special case Sf = S0
we have dh/dx = 0; i.e. normal flow. Otherwise, we note that:
• S0 – Sf > 0 if and only if h is greater than the normal depth;
• 1 – Fr2 > 0 if and only if subcritical.
Considering the whole range of possibilities allows a two-character classification of slopes
(S1, M3 etc) where:
• the first character is S, C, M, H, A for Steep, Critical, Mild, Horizontal, Adverse;
• the second character is 1, 2, 3 depending on where h lies with respect to hc and hn.
Typical profiles and examples of where they occur are given in the table overleaf.
A backwater curve is a streamwise increase of depth because of a downstream obstruction or
because the slope is insufficient to maintain that rate of flow. A reduction in depth is called a
drawdown curve. Profiles with second character 1 or 3 are backwater curves (numerator and
denominator of the GVF equation have the same sign); profiles with second character 2 are
drawdown curves (numerator and denominator of the GVF equation have opposite signs).
Note that:
• the normal depth is approached asymptotically.
• the critical-depth line is approached at right angles (1 – Fr2 in the denominator).
• above normal depth a backwater asymptotes a horizontal surface (M1, C1, S1)
1
An alternative statement is that a downward slope is steep if it exceeds the critical slope (that slope at which
the normal flow is critical).
Hydraulics 3 Gradually-Varied Flow - 3 Dr David Apsley
Type Symbol Definition Sketches Examples
h > hc > hn
S1 Hydraulic jump upstream with obstruction or
S1 hc reservoir controlling water level downstream.
STEEP S2
hn
(normal flow is S2 hc > h > hn Change to steeper slope.
supercritical) S3
S3 hc > hn > h Change to less steep slope.
CRITICAL C1 h > hc = hn h c= h n C1
(undesirable;
C3
undular unsteady
flow) C3 hc = hn > h
h > hn > hc Obstruction or reservoir controlling water level
M1 hn M1 downstream.
MILD M2
hc
(normal flow is M2 hn > h > hc Approach to free overfall.
subcritical) M3
hn > hc > h Hydraulic jump downstream; change from steep to
M3
mild slope or downstream of sluice gate.
H2
HORIZONTAL H2 h > hc Approach to free overfall.
(limiting mild hc
slope; hn ∞) hc > h H3 Hydraulic jump downstream; change from steep to
H3
horizontal or downstream of sluice gate.
A2
A2 h > hc
ADVERSE
(upslope) hc A3
A3 hc > h
Hydraulics 3 Gradually-Varied Flow - 4 Dr David Apsley
3.4 Qualitative Examples of Open-Channel Flow Behaviour
A control point is a point where there is a known relationship between water depth (or head)
and discharge (aka “stage-discharge relation”). Examples include critical-flow points (weirs,
flumes, sudden changes in slope, free overfall), sluice gates, entry or discharge to a reservoir.
A hydraulic jump can also be classed as a control point. Control points usually provide a
location where one can start a GVF calculation.
Some general rules:
(i) Supercritical ⇒ controlled by upstream conditions.
Subcritical ⇒ controlled by downstream conditions.
(ii) Given a long-enough fetch the flow will try to revert to normal flow.
(iii) A hydraulic jump occurs between two regions of gradually-varied flow at the point
where the jump condition for the sequent depths is correct.
(iv) Where the slope is mild (i.e. the normal flow is subcritical) and there is no influential
downstream control, a hydraulic jump can usually be assumed to jump directly to the
normal depth.
Flow over a weir (mild slope)
normal M1
normal
hc
hn h1
WEIR h2 M3 hydraulic hn
jump
CP CP
Flow under a sluice gate – (a) mild slope
normal M1
normal
hn h1
h2 M3 hydraulic hn
jump
CP CP
Flow under a sluice gate – (b) steep slope
S1
normal
hn h1
h2 S3 normal
hn
CP
Hydraulics 3 Gradually-Varied Flow - 5 Dr David Apsley
Flow from a reservoir – (a) mild slope
CP
normal
hn
RESERVOIR
Flow from a reservoir – (b) steep slope
CP
hc S2
normal
RESERVOIR
Outflow to reservoir (mild slope)
normal M1 CP
hn
RESERVOIR
Free overfall (mild slope)
normal
M2 CP
hn
hc
critical
Hydraulics 3 Gradually-Varied Flow - 6 Dr David Apsley
3.5 Numerical Solution of the GVF Equation
Analytical solutions of the GVF equation are very rare and it is usual to solve it numerically.
The process yields a series of discrete pairs of distance xi and depth hi. Intermediate points
can be determined, if required, by interpolation.
All methods employ a discrete approximation to one of the following forms of GVF equation:
dH V2
= − S f (where total head H = z b + h + ) (11)
dx 2g
dE V2
= S 0 − S f (where specific energy E = h + ) (12)
dx 2g
dh S 0 − S f
= (13)
dx 1 − Fr 2
In any of these the friction slope can be obtained by inverting Manning’s equation:
n 2V 2
S f = 4/3 (14)
Rh
and the Froude number is
V
Fr = (15)
gh
where h = A / b is the mean depth (= actual depth for a rectangular or wide channel).
VERY IMPORTANT!
Integration starts at a control point and proceeds:
forward in x if the flow is supercritical (upstream control); or
backward in x if the flow is subcritical (downstream control).
3.5.1 The “Standard-Step” Method h4
h3
This solves for depth h at specified distances x, h1 h2
intervals x. h0
It uses the total-head form of the GVF equation: CP
dH
= −S f
dx ∆x ∆x ∆x ∆x
which is discretised as
H i +1 − H i S f ,i + S f ,i +1
= −( ) (16)
x 2
solving sequentially for h1, h2, h3, … starting at the depth at the control point h0.
The method operates by adjusting hi+1 (iteratively) at each step so that LHS and RHS of (16)
are equal.
This is a good method, but since it requires iterative solution at each step it is better suited to
a computer program than hand or spreadsheet calculation.
Hydraulics 3 Gradually-Varied Flow - 7 Dr David Apsley
3.5.2 The “Direct-Step” Method
This solves for distance x at specified depth intervals h. The advantage is that it can be used
to calculate the profile starting from a critical point (where 1 – Fr2 = 0 and the standard-step
method would fail).
supercritical: subcritical:
direction of integration direction of integration
∆h ∆h
∆h ∆h
∆h ∆h
CP CP
x0 x1 x2 x3 x3 x2 x1 x0
It uses either the specific-energy or depth forms of the GVF equation.
Specific-Energy Form
dE V2 q2
= S0 − S f (where E = h + = h+ for a rectangular channel)
dx 2g 2 gh 2
This is inverted to make x the dependent variable:
dx 1
=
dE S 0 − S f
and then discretised and rearranged for the distance increment at each step as:
E
x=
( S 0 − S f ) av
where
E = Ei +1 − Ei
In the commonest form, ‘av’ is the average of values at start (i) and end (i+1) of an interval.
Depth Form
dh S 0 − S f
=
dx 1 − Fr 2
This is inverted to make x the dependent variable:
dx 1 − Fr 2
=
dh S 0 − S f
and then discretised and rearranged for the distance increments as:
1 − Fr 2
x= h
S −S
0 f
Hydraulics 3 Gradually-Varied Flow - 8 Dr David Apsley
As before, the bracketed term on the RHS can be taken as the average of values at the start
and end of an interval or (my own preference) by evaluation at the ‘mid-point’ depth
hmid = 12 (hi + hi +1 ) .
Comment. Different authors adopt different ways of solving the GVF equation numerically:
particularly in choosing whether to use the specific-energy or depth form, and how to form
the average derivative (e.g., average of values at the ends of the interval or simply the single
value at the midpoint). All should give the same answer when the step size h becomes very
small, but may differ for the larger step sizes typical of hand calculations. The specific-
energy form seems to be slightly more common in the literature, but my own experience is
that the depth form, with derivative evaluated just once at the mid-point of the interval, seems
to give slightly better results for large step sizes. Note that surface profiles become highly
curved near critical points and more steps, with a smaller h, should be used there.
Example (Examination, January 2007 – modified).
A long rectangular channel of width 4 m has a slope of 1:5000 and a Manning’s n of
0.015 m–1/3 s. The total discharge is 8 m3 s–1. The channel narrows to a width of 1 m as a
venturi flume over a short length.
(a) Determine the normal depth for the 4 m wide channel.
(b) Show that critical conditions occur at the narrow 1 m wide section.
(c) Determine the depth just upstream of the venturi where the width is 4 m.
(d) Determine the distance upstream to where the depth is 5% greater than the normal
depth using two steps in the gradually-varied flow equation given below; (you may
use either form).
Data
In standard notation,
dh S 0 − S f dE
= or = S0 − S f
dx 1 − Fr 2 dx
Solution.
(a) For the normal depth,
1 2 / 3 1/ 2 h
Q = VA where V = Rh S , Rh = , A = bh
n 1 + 2h / b
b S h5/ 3
⇒ Q=
n (1 + 2h / b) 2 / 3
Rearrange as an iterative formula for h:
3/ 5
nQ
h = (1 + 2h / b) 2 / 5
b S
or, in this case, (with h in m):
Hydraulics 3 Gradually-Varied Flow - 9 Dr David Apsley
h = 1.570(1 + 0.5h) 2 / 5
Iterate to get a normal depth:
hn = 2.090 m
(b) First assume that critical conditions do occur at the throat and calculate the corresponding
total head. If this exceeds the total head in the approach flow then the assumption of critical
flow (evidenced by the flow backing up) will be correct.
At the throat the discharge per unit width is
Q
q max = = 8 m 2 s −1
bmin
The critical depth for this rectangular section is
1/ 3
q2
hc = max = 1.869 m
g
The corresponding critical specific energy is
3
E c = hc = 2.804 m
2
and, since the bed of the flume is flat, this is also the total head.
The total head at the normal depth is
V2 Q2
H = hn + n = hn + = 2.137 m
2g 2 gb 2 hn2
Since this is less than the head corresponding to critical flow in the venturi then the flow must
back up and go critical through the throat.
(c) Just upstream of the venturi the total head is H = 2.804 m and we are looking for a
subcritical solution of
V2 Q2
H =h+ = h+
2g 2 gb 2 h 2
Rearrange for the deeper solution as
Q2
h=H−
2 gb 2 h 2
or, in this case (with h in m):
0.2039
h = 2.804 −
h2
Iterate (from, say, 2.804) to get a depth just upstream of the venturi as
h = 2.778 m
(d) Need to solve upstream from a depth h = 2.778 to a depth h = 2.194 (i.e. 1.05×hn) using 2
steps. Hence, h = –0.292 m.
(i) Using the depth form of the GVF equation:
dh S 0 − S f
=
dx 1 − Fr 2
Hydraulics 3 Gradually-Varied Flow - 10 Dr David Apsley
dx 1 − Fr 2
⇒ =
dh S 0 − S f
x 1 − Fr 2
⇒ ≈
h S 0 − S f
mid
1 − Fr
2
⇒ x≈ h
S −S
0 f mid
( ‘mid’ mean the mid-point of the interval; i.e. half way between hi and hi+1. It is often written
as ‘i+1/2’.)
For convenience, work out numerical expressions for Fr2 and Sf in terms of h:
V2 Q2 / b2 0.4077
Fr =
2
= 3
=
gh gh h3
nQ (1 + 2h / b)
2 4/3
(1 + 0.5h) 4 / 3
Sf = = 9 × 10 − 4
b h10 / 3 h10 / 3
Hence,
0.4077
1−
dx 1 − Fr 2
h3
= =
dh S 0 − S f (1 + 0.5h) 4 / 3
2 × 10 − 4 − 9 × 10 − 4
h10 / 3
With
dx
x= h and h = –0.292 m
dh mid
working may then be set out in tabular form.
i hi xi hmid (dx/dh)mid x
0 2.778 0
2.632 10810 –3157
1 2.486 –3157
2.340 18850 –5504
2 2.194 –8661
This gives a distance of about 8660 m upstream.
(ii) Using the specific-energy form of the GVF equation:
dE
= S0 − S f
dx
dx 1
⇒ =
dE S 0 − S f
x 1
⇒ ≈
E ( S 0 − S f ) av
E
⇒ x≈
( S 0 − S f ) av
Hydraulics 3 Gradually-Varied Flow - 11 Dr David Apsley
(‘av’ means the average of values calculated at start and end of each interval in h.
Here:
V2 Q2 / b2 0.2039
E = h+ = h+ 2
=h+
2g 2 gh h2
and the same expression as before may be used for Sf, so that:
(1 + 0.5h) 4 / 3
S 0 − S f = 2 × 10 − 4 − 9 × 10 − 4
h10 / 3
With
E
x=
( S 0 − S f ) av
working may then be set out in tabular form.
i hi xi Ei S0 –Sf E (S0 –Sf)av x
0 2.778 0 2.804 1.046×10–4
–0.285 8.883×10–5 –3208
1 2.486 3208 2.519 7.304×10–5
–0.283 4.851×10–5 –5834
2 2.194 9042
This gives a distance of about 9040 m upstream.
Smaller steps h will give more ‘accurate’ results (and closer agreement between the two
methods). However, the accuracy with which one generally knows n does not justify this.
Example (Examination, January 2008 – modified)
A long, wide channel has a slope of 1:2747 with a Manning’s n of 0.015 m–1/3 s. It carries a
discharge of 2.5 m3 s–1 per metre width, and there is a free overfall at the downstream end.
An undershot sluice is placed a certain distance upstream of the free overfall which
determines the nature of the flow between sluice and overfall. The depth just downstream of
the sluice is 0.5 m.
(a) Determine the critical depth and normal depth.
(b) Sketch, with explanation, the two possible gradually-varied flows between sluice and
overfall.
(c) Calculate the particular distance between sluice and overfall which determines the
boundary between these two flows. Use one step in the gradually-varied-flow
equation.
Hydraulics 3 Gradually-Varied Flow - 12 Dr David Apsley