3/22/23

Hydraulic Engineering

GRADUALLY VARIED FLOW

Review: what do we know about nonuniform flow?

• By now, we understand that UF does not occur frequently in nature
• If So or A in a uniform OCF are abruptly changed, UF may
transition to gradually varied flow (GVF) or rapidly varied flow
(RVF)

Q: What might cause an abrupt change in So or A?

1
 3/22/23

Review: what do we know about nonuniform flow?

• By now, we understand that UF

does not occur frequently in
nature
• If So or A in a uniform OCF are
abruptly changed, UF may
abruptly transition to gradually
varied flow (GVF) or rapidly
varied flow (RVF)

Q: What might cause an abrupt

change in So or A?
A: An obstruction, channel
contraction, erosion of bed
material, presence of a hydraulic
structure, etc.

Review: what do we know about nonuniform flow?

• Recall that nonuniform flow is classified as
• GVF when y changes gradually over a long distance along the channel (dy/dx << 1)
• RVF when y changes markedly over a relatively short distance in the flow direction
(such as the flow of water past a partially open gate or over a falls) (dy/dx ≳ 1)

• A GVF region typically occurs between RVF and UF regions

2
 3/22/23

Nonuniform Flow
• GVF and RVF can be steady or unsteady
• Both types of varied flow can accelerate (dV/dx > 0) or decelerate (dV/dx < 0), depending
on the variation of velocity in the direction of flow due to changes in So and A
• In GVF, the change in the flow depth in the x direction is very gradual and the flow is
treated as being quasi-uniform with a constant V

V
So

So

Gradually Varied Flow (GVF)

• Characterized by gradual variations in y and V
• So and Sw are small
• No abrupt changes in the flow depth
• The free surface remains largely undisturbed (i.e. smooth)

Site C Dam (Fort St. John, BC) (@ Abbotsford News) The Saskatoon weir (Saskatoon, SK) (@ CBC News)

3
 3/22/23

Assumptions of GVF Analysis

1. The flow is nonuniform.
2. The flow is steady.
3. As with UF analysis, there is no lateral
(e.g. cross-stream) inflow or outflow.
4. Bed slope So is small (e.g. So < 0.05); ycosθ y
sinθ ≈ tanθ ≈ θ, the angle between the
bed and the horizontal datum.
• Accordingly, cosθ ≈ 1 and y
measured vertically or normal to the
bed are very similar (e.g. ycosθ ≈ y)

Assumptions of GVF Analysis

5. The channel is prismatic, or the flow
conditions are such that prismatic
conditions are approximated.
6. The pressure distribution in GVF can
be taken as hydrostatic. The
streamline curvature is very small
and the streamlines are relatively
straight and parallel.
7. Head losses in GVF may be
determined using the same
procedure as for UF, as has been
proven effective by water surface
profile measurements in model and
prototype investigations.
8

4
 3/22/23

Governing Equations of GVF

• Recall that the energy grade line (EGL) is the
line along which
𝑉!
𝐻 =𝑧+𝑦+a
2𝑔

• Differentiating both sides with respect to x and

taking V = Q/A gives (a is included in equation [1] to generalize)

"# "% "& '(" " +

"$
= "$ + "$ + )* "$ ,"
[1]

• Understanding that the change in bed elevation z

is equal to the bed slope So which is in the
streamwise (x) direction

"%
"$
= −𝑆- [2]

Governing Equations of GVF

#$ #& #' ()! # +
• Similarly, the change in the elevation of the EGL, #%
=
#%
+ +
#% !* #% ,!
[1]
H, is equal to the slope of the EGL, which is also
negative in the x direction

"#
"$
= −𝑆. [3]

• In the third term on the RHS of [1]

𝑑 1 𝑑 1 𝑑𝐴
=
𝑑𝑥 𝐴! 𝑑𝐴 𝐴! 𝑑𝑥
𝑑 1 𝑑𝐴 𝑑𝑦
=
𝑑𝐴 𝐴! 𝑑𝑦 𝑑𝑥

10

5
 3/22/23

Governing Equations of GVF

#$ #& #' ()! # +
• By the very definition of GVF, the change in flow depth Δy = + + [1]
#% #% #% !* #% ,!
is small

• The change in cross-sectional area of flow, ΔA, can

therefore be taken as BΔy, where B is the lateral channel
width.
𝑑 1 𝑑𝐴 𝑑𝑦 −2𝐵 𝑑𝑦
= -
𝑑𝐴 𝐴! 𝑑𝑦 𝑑𝑥 𝐴 𝑑𝑥

• The third term on the RHS of [1] then becomes

!
𝑄 !
!
𝛼𝑄 −2𝐵 𝑑𝑦 −𝛼𝐵𝑄 𝑑𝑦 𝑑𝑦 𝑑𝑦 𝑑𝑦
= =− 𝐴 = −𝛼𝐹.! ≅ −𝐹.!
2𝑔 𝐴- 𝑑𝑥 𝑔𝐴- 𝑑𝑥 𝑔𝐴 𝑑𝑥 𝑑𝑥 𝑑𝑥
𝛼𝐵

11

Governing Equations of GVF

• Substituting [2] and [3] into [1] and
rearranging then gives

𝑑𝑦 𝑑𝑦
−𝑆. = −𝑆- + − 𝐹1)
𝑑𝑥 𝑑𝑥
𝒅𝒚 𝑺𝒐-𝑺𝒇
𝒅𝒙
= 𝟏-𝑭𝟐𝒓
[4]

• This is the governing equation used

for classifying and sketching water
surface profiles in GVF

12

6
 3/22/23

Bed Slope Classification

• At this point, we understand that it is important to be able to predict y under varying Q and
channel geometry conditions for channel design

• In order to achieve this, it is necessary to plot the water surface profile, which is a plot of
the flow depth y as a function of streamwise location x

• There are many methods to computationally estimate water surface profiles in GVF
(which we will be discussing in a forthcoming lecture)

• However, it is important to be able to first qualitatively classify the water surface profiles
using So and y, in order to then select an appropriate computational method

13

Bed Slope Classification

• When classifying water surface profiles, the
bed slope is classified as mild (M), steep (S),
critical (C), horizontal (H) and adverse (A) NDL
• Bed slope classification is based on the free
surface location relative to the critical-depth
line (CDL), along which y = yc

• The CDL is the location of the free surface in CDL

critical flow

• Classification is also dependent on the flow

depth relative to the normal-depth line (NDL),
along which y = yn

• The NDL is the location of the free surface

under uniform flow conditions

14

7
 3/22/23

Bed Slope Classification

• If the NDL is located above the CDL (yn >

yc), the bottom slope is classified as mild NDL
(M)

• If the CDL is located above the NDL (yn <

yc), the bottom slope is classified as steep
(S)
CDL
• The bottom slope is critical (C) if yn = yc

• If So = 0, the bottom slope is horizontal

(H)

• If So < 0 (e.g. the slope is negative), the

bottom slope is adverse (A)

15

Water Surface Profile Naming Convention

• Water surface profile nomenclature is
comprised of two parts
NDL
• The first is the bed slope classification
shorthand (M, S, A, H, C, etc.)

• The second is based on the zone of flow in

which they lie
CDL
• This is dependent on the locations of:
• the flow depth y,
• the CDL, and

• the NDL

in relation to one another

16

8
 3/22/23

Water Surface Profile Naming Convention

• 3 zones corresponding to these profile
classifications are demarcated
NDL
• Zone 1 is the region located above both the NDL
and the CDL

• Zone 2 is the region between the NDL and the

CDL

• Zone 3 is the region between the lower of the CDL

NDL and the CDL and the channel bed
• The second part of the water surface profile
classification refers to the zone in which the
profile is located
• For example, the water surface profile located in
Zone 2 for a horizontal bed slope is called an H2
profile

17

Slope Types
MILD
Zone 1

yn Zone 2
yc
Zone 3
STEEP
yc Zone 1
yn
Zone 2

Zone 3

18

9
 3/22/23

Slope Types
CRITICAL
Zone 1
yn yc Zone 2
ADVERSE
Zone 3
Zone 3

yc
HORIZONTAL
Zone 2

yc Zone 3

19

Sketching Water Surface Profiles

• Recall that

𝒅𝒚 𝑺𝒐 − 𝑺𝒇 NDL
=
𝒅𝒙 𝟏 − 𝑭𝟐𝒓

• So= bed slope

• Sf = slope of the EGL
• Fr = Froude number CDL

• If y < yn, Sf > So ; So – Sf is negative

• If y > yn , Sf < So ; So – Sf is positive

21

10
 3/22/23

Sketching Water Surface Profiles

• Recall that

𝒅𝒚 𝑺𝒐 − 𝑺𝒇 NDL
=
𝒅𝒙 𝟏 − 𝑭𝟐𝒓

• So= bed slope

• Sf = slope of the EGL
• Fr = Froude number CDL

• If y < yn, Sf > So ; So – Sf is negative

• If y > yn , Sf < So ; So – Sf is positive
• If y < yc, Fr2 > 1 ; 1 – Fr2 is negative
• If y > yc, Fr2 < 1 ; 1 – Fr2 is positive

22

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.

• Recall that

𝒅𝒚 𝑺𝒐6𝑺𝒇 Zone 1
𝒅𝒙
= 𝟏6𝑭𝟐
𝒓

yn Zone 2
yc
Zone 3

23

11
 3/22/23

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.

• Recall that
𝒅𝒚 𝑺𝒐 )𝑺𝒇
=
𝒅𝒙 𝟏)𝑭 𝟐
𝒓
Zone 1
• In Zone 1
yn Zone 2
• y > yn and so Sf < So
• y > yc and so Fr < 1 yc
Zone 3
• From which
"& 9,69- ;
• "$
= +6:/
=;= +
.

• Therefore, the depth of flow is increasing in the streamwise direction in Zone 1

24

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.

• How do we know what form the water

surface profile takes?

Zone 1
• There are many surface profiles that
would satisfy the condition that
yn Zone 2
dy/dx > 0
yc
Zone 3

25

12
 3/22/23

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.

• We do this by establishing the limits of the

dy/dx function
• We know that y is increasing in the
downstream direction and so it is
conversely decreasing in the upstream Zone 1
direction
• In effect, as we move in the upstream
direction (which we do because the flow is yn Zone 2
subcritical and the control is downstream),
the water surface profile (i.e. y) is
approaching the NDL (i.e. yn) yc
Zone 3
𝒅𝒚 𝑺𝒐 − 𝑺𝒇
=
𝒅𝒙 𝟏 − 𝑭𝟐𝒓

• As y → yn, Sf → So and dy/dx → 0, i.e.

approaches the short dash line

26

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.

• We do this by establishing the limits of the

dy/dx function
• We know that y is increasing in the
downstream direction and so it is
conversely decreasing in the upstream Zone 1
direction
• In effect, as we move in the upstream
direction (which we do because the flow is yn Zone 2
subcritical and the control is downstream),
the water surface profile (i.e. y) is
approaching the NDL (i.e. yn) yc
Zone 3
𝒅𝒚 𝑺𝒐 − 𝑺𝒇
=
𝒅𝒙 𝟏 − 𝑭𝟐𝒓

• As y → yn, Sf → So and dy/dx → 0, i.e. a

approaches the short dash line

27

13
 3/22/23

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.
• In Zone 2

• y < yn and so So < Sf

• y > yc and so Fr < 1 Zone 1

𝑑𝑦 𝑆- − 𝑆. − yn Zone 2
= = = −
𝑑𝑥 1 − F<) +
yc
Zone 3
• So we know that y is decreasing in the
downstream direction

28

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.
• In Zone 2

• y < yn and so So < Sf

• y > yc and so Fr < 1 Zone 1

𝑑𝑦 𝑆- − 𝑆. − yn Zone 2
= = = −
𝑑𝑥 1 − F<) +
yc
Zone 3
• So we know that y is decreasing in the
downstream direction

• Again, there are several forms that the

profile can take

29

14
 3/22/23

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.
• We know that y in this region is
approaching yn in the upstream direction
(as with Zone 1)

• So again, as y → yn, Sf → So and dy/dx → Zone 1

0; so the NDL becomes an asymptote of
the water surface profile in the US
direction
yn Zone 2
• In Zone 2, the flow is subcritical because y
> yc yc
Zone 3
• As y decreases in the DS direction, y will
approach and eventually reach yc (where
Fr = 1)
𝑑𝑦 𝑆- − 𝑆. −
• Fr approaches unity and the denominator = = =∞
approaches 0, so dy/dx becomes infinitely 𝑑𝑥 1 − 𝐹1) 0
large in the downstream direction

30

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.
• We know that y in this region is
approaching yn in the upstream direction
(as with Zone 1)

• So again, as y → yn, Sf → So and dy/dx → Zone 1

0; so the NDL becomes an asymptote of
the water surface profile in the US
direction
yn Zone 2
• In Zone 2, the flow is subcritical because y
> yc yc
Zone 3
• As y decreases in the DS direction, y will
approach and eventually reach yc (where
Fr = 1)
𝑑𝑦 𝑆- − 𝑆. −
• Fr approaches unity and the denominator = = =∞
approaches 0, so dy/dx becomes infinitely 𝑑𝑥 1 − 𝐹1) 0
large in the downstream direction

31

15
 3/22/23

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.
• Repeat for Zone 3

• y < yn
• y < yc Zone 1

𝒅𝒚 𝑺𝒐 − 𝑺𝒇 − yn Zone 2
= =+
𝒅𝒙 𝟏 − 𝑭𝟐𝒓 −
yc
Zone 3
• Flow in this zone is supercritical, so we
look from upstream to downstream for
the shape of the profile (because the
control is upstream)

32

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.
𝑑𝑦 𝑆- − 𝑆. −
= = =+
𝑑𝑥 1 − 𝐹1) −

Zone 1
The flow depth is increasing in the DS
direction
yn Zone 2
• As y → yc, Fr2
→ 1 and so the
yc
denominator approaches 0 and dy/dx
→∞ Zone 3
• Conversely, in the US direction, as y
becomes small, the denominator
becomes very large, and dy/dx
becomes practically small (i.e. tends
towards zero)

33

16
 3/22/23

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.
• As y → yc, Fr2 → 1 and so the
denominator approaches 0 and dy/dx
→∞
• Conversely, in the US direction, as y Zone 1
becomes small, the denominator
becomes very large, and dy/dx
becomes practically small (i.e. tends yn Zone 2
towards zero)
yc
Zone 3

34

Example 1. Sketching Water Surface Profiles for a Mild Slope

Sketch the water surface profiles in zones 1, 2 and 3 when the bed slope is mild.
• The profiles for each zone are then
named as previously described
M1

• Water surface profiles for critical, Zone 1

steep, adverse and horizontal bed M2
slopes can be derived using the same yn
procedure Zone 2
yc M3
• Recall that for horizontal and adverse Zone 3
slopes, uniform flow is not possible
and so the NDL does not exist and we
can take its location as being infinitely
large while sketching profiles

35

17
 3/22/23

36

37

18
 3/22/23

Examining Slope and Surface Profile Types

38

NDL - different
in both channels
CDL - the same

39

19
 3/22/23

Flow profile

40

Example

41

20
 3/22/23

Please verify the table below

42

Critical Slope

43

21
 3/22/23

44

Computation of Water Surface Profiles

• The sketches in the preceding sections will provide us with an idea of the general shape
of the free surface profile

• Furthermore, they provide a good qualitative basis for accurate computation of the water
surface profiles, which is required for engineering applications in practice

• Detailed water surface information is required at specific discharges to carry out planning,
design and operation of open channels with respect to control structures, hydraulic
structures and channel modification

• In computational modelling of steady flows, the initial conditions drive the simulation

• If the initial conditions are incorrect because they have been improperly estimated, errors
may be incurred in the simulation and the results

45

22
 3/22/23

Governing Equation of GVF: A Closer Look

• In hydraulic engineering practice, in order to estimate the water surface profile
computationally, we are generally given a design Q and select So, n and A based on
additional design considerations and site information
• However, in order to determine A and other channel properties (e.g. free surface width B),
some channel control (e.g. yn, yc or some other control y) must be determined
• Recall Equation 4 - GVF

𝒅𝒚 𝑺𝒐 − 𝑺𝒇
=
𝒅𝒙 𝟏 − 𝑭𝟐𝒓

• From the expansion of Equation 1, we derived the above expression by taking

𝜶𝑩𝑸𝟐
≈ 𝑭𝟐𝒓
𝒈𝑨𝟑

46

Governing Equation of GVF: A Closer Look

• If we return to the original form of Equation 4, this gives

"& 9,69-
"$
= 012/
[5]
+6
345

• Equation [5] is a first-order ordinary differential equation (ODE); we will return to the
significance of this momentarily

• In Eq. [5], x is the independent variable and y is the dependent variable

• Essentially
𝑑𝑦
= 𝑓 𝑦, 𝑄, 𝐵, 𝐴
𝑑𝑥

47

23
 3/22/23

Governing Equation of GVF: A Closer Look

𝑑𝑦
= 𝑓 𝑦, 𝑄, 𝐵, 𝐴
𝑑𝑥

• However, B and A are also both f(y)

• Furthermore, So, n and Q are known at the outset of the problem

• This allows us to reduce the RHS of Eq. [5] to a function of y, which is a function of x in
GVF, such that
"&
"$
= 𝑓 𝑥, 𝑦 [6]
and
𝑆4 − 𝑆5
𝑓(𝑥, 𝑦) =
𝛼𝐵𝑄!
1−
𝑔𝐴-

48

Computing dy/dx
• You may recall that ODEs are solved by actual integration or finite difference
approximations
• However, a closed-form solution of Equation [5] is only available for very simplified cases
because f(x,y) is a nonlinear function

"& 96 697
"$
= 89:"
[5]
+6
;<=

• Therefore, numerical methods are used for integration to determine y at specific x

locations
How is this done?

49

24
 3/22/23

Computing dy/dx
• You may recall that ODEs are solved by actual integration or finite difference
approximations
• However, a closed-form solution of Equation [5] is only available for very simplified cases
because f(x,y) is a nonlinear function
• Therefore, numerical methods are used for integration to determine y at specific x
locations
How is this done?

There are two methods by which this can be y

carried out:

Given y and x at location i, yi and xi yi xi, yi xi+1, yi+1

1. The value of yi+1 can be determined at yi+1
a prescribed xi+1
2. The value of xi+1 can be determined for
a prescribed yi+1 x
xi xi+1

50

Scenario 1: Computing yi+1 from xi+1

"&
"$
= 𝑓 𝑥, 𝑦 [6]
• We will start by examining scenario 1, in which the value of yi+1 is determined at xi+1
• Multiplying both sides of [6] by dx and integrating gives

'!"# %!"#
H 𝑑𝑦 = H 𝑓 𝑥, 𝑦 𝑑𝑥
'! %!

• In applying the limits of integration,

$
𝑦A;+ = 𝑦A + ∫$ 678 𝑓 𝑥, 𝑦 𝑑𝑥 [7]
6

• From [7], we can see that yi+1 can be determined by evaluating the integral term on the
RHS and so computing the flow depth at location i+1

51

25
 3/22/23

Scenario 2: Computing xi+1 from yi+1

"& 9,69-
"$
= 012/
[5]
+6
345
• From above, Eq. 5 may be written as
"&
"$
= 1/𝐹 𝑥, 𝑦 [8]
where
012/
+6
345
𝐹 𝑥, 𝑦 = 9,69-
[9]
• By multiplying both sides of Eq. [8] by dy, integrating, and applying the limits of
integration, we obtain
&
𝑥A;+ = 𝑥A + ∫& 678 𝐹 𝑥, 𝑦 𝑑𝑦 [10]
6

• From [10], we can see that xi+1 can be determined by evaluating the integral term on the
RHS and so computing the water surface profile

52

Computing xi+1 from yi+1 Using the Direct Step Method

• Problem statement for scenario 2: The flow depth yi at streamwise distance xi is known
and we are required to determine the streamwise distance xi+1 at which specified flow
depth yi+1 will occur
• This is called the Direct Step Method, which solves the equation of GVF explicitly
• dx is determined from dy, in a process in which we first specify dy
• In such a case, the values of So, Q and n are known
• Since the specific energy E is a function of y and y varies with streamwise distance x, the
slope of the EGL (Sf) is also a function x
• Since we will not know what dx is prior to the computation, the value of Sf must be
estimated using some expression
• Several methods of determining Sf have been investigated, and it has been determined
that as long as dy and dx are sufficiently small, using the following expression for
estimating Sf yields satisfactory results:

+
𝑆0. = ) (𝑆.,A + 𝑆.,A;+ ) [11]

53

26
 3/22/23

The Direct Step Method

+
𝑆0. = ) (𝑆.,A + 𝑆.,A;+ ) [11]

• The above expression can be used to estimate the head loss hL

+
ℎC = ) (𝑆.,A + 𝑆.,A;+ )(𝑥A;+ − 𝑥A ) [12] hL/L = Sf

• Substituting [11] and [12] into Bernoulli’s equation then gives

+
𝑧A;+ + 𝐸A = 𝑧A + 𝐸A;+ + ) (𝑆.,A + 𝑆.,A;+ )(𝑥A;+ − 𝑥A ) [13]

• Recall that
𝑧A;+ = 𝑧A − 𝑆- (𝑥A;+ − 𝑥A ) [14]

54

The Direct Step Method

• Now, substituting [14] into [13] and rearranging, we obtain

D6786D6
𝑥A;+ = 𝑥A + 8 [15]
9,6 (9-,6;9-,678)
/

• From [15], the streamwise location (xi+1) of yi+1 can be determined; similarly, xi+2…xi+n can
be determined for yi+2…yi+n in order to obtain the water surface profile over the desired
channel reach

• As previously noted, the step size in both the x (e.g. xi+1 – xi) and y (e.g. yi+1 – yi)
directions should be very small in order to avoid large errors

• Furthermore, it is very important to use proper significant digits in the computations and
exercise care in rounding off values

55

27
 3/22/23

Drawbacks of the Direct Step Method

• There are two main disadvantages of the Direct Step Method

• The flow depth has not been computed at all x locations; therefore, if y is required at a
specific location, it will need to be interpolated and the corresponding cross-sectional
details will be estimated, reducing accuracy

• Furthermore, the Direct Step Method is cumbersome to apply to non-prismatic channels,

since the computational method relies on straightforward calculations which utilize So

56

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Consider a trapezoidal prismatic channel with Manning’s n = 0.025 and So = 0.001.
Compute the water surface profile for a design Q of 30 m3/s if the channel ends in a free
overfall B

1 y
2
• Recall
>$ !/- 8m
𝑄= 𝐴 𝑅@ 𝑆 +/!
?

𝐸BC+ − 𝐸B
𝑥BC+ = 𝑥B +
1
𝑆4 − (𝑆5,B + 𝑆5,BC+ )
2

57

28
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Determining yi

B
• In order to select yi, both yn and yc need to be determined
1 y
𝑘? !/- 2
𝑄= 𝐴 𝑅@ 𝑆 +/!
𝑛
8m
1 !/-
30 m- /s = 𝐴 𝑅@ (0.001)+/!
0.025

• Recall that
• 𝐴 = 𝑦? 𝑏 + 2𝑦?
• 𝑝 = 𝑏 + 2 5𝑦? yn = 1.754 m
,
• 𝑅@ =
E

58

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Determining yc

B
F
• Fr =
*'%
1 y
2
• Setting Fr = 1 and V = Q/A
) ,%
&/( 8m
= ; from which yc = ?
* G #/(

M2
yn

yc

59

29
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Determining yc

B
F
• Fr =
*'%
1 y
2
• Setting Fr = 1 and V = Q/A
) ,%
&/( 8m
= ; from which yc = 1.0298 m
* G #/(

Classifying the water surface profile

M2
yn
• We know that the slope is mild (yn > yc)
• We are told that the profile ends in overfall yc
• This is an M2 profile

60

Example 2. Computing the Water Surface Profile Using the Direct Step Method

• With an M2 profile, we know that the surface

profile meets the CDL in the downstream B
direction, and the NDL acts as an asymptote of
the profile in the upstream direction
1 y
2
• This knowledge can inform the point at which
8m
we will begin our computation of the water
surface profile

• For this example, we will begin with y = yc and

work our way in the upstream direction at
M2
increments of x yn

yc
• However, theoretically we could begin at any y
< yn and still arrive at the same solution

61

30
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• We can begin the computations at any location
where the flow depth for the specified
discharge is known B

• It is common practice to start the computations 1 y

at a downstream control section if the flow is 2
subcritical and proceed in the upstream
direction 8m

• For supercritical flows, however, we start at an

upstream control section and compute the
profile in the downstream direction
M2
yn
• Unfortunately, this fact has been incorrectly
attributed in many well-known publications to yc
indicate that the computations become
unstable or yield incorrect results if this
convention is not followed

62

Example 2. Computing the Water Surface Profile Using the Direct Step Method

• Other than the fact that the flow depth is

known at a control section, there appears to B
be no reason why we should proceed in either
the upstream or downstream direction
1 y
2
• This is because all we are doing is either 8m
numerically solving a differential equation for
the specified initial condition or solving a
nonlinear algebraic equation

M2
• Whether we proceed in the positive or yn
negative x direction should make little
difference provided the computational step is
properly selected in the computations yc

63

31
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Direct Step Computation Procedure

1. Start with yi = yc as previously calculated (yi = 1.0298 m)

2. Calculate Ai for the trapezoidal section

3. Calculate Rh,i

(1) (2) (3)

Location y (m) A (m2) Rh (m)
i 1.0298

64

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Direct Step Computation Procedure

1. Start with yi = yc as previously calculated (yi = 1.0298 m)

2. Calculate Ai for the trapezoidal section

𝐴B = 𝑦B 𝑏 + 2𝑦B = 1.0298 m 8 m + 2 ∗ 1.0298 m = 10.35938 m!

3. Calculate Rh,i

𝐴B 𝑦B 𝑏 + 2𝑦B 10.35938 m!
𝑅@,B = = = = 0.82182 m
𝑝B 𝑏 + 2 5𝑦B 8 m + 2 5(1.0298 m)

(1) (2) (3)

Location y (m) A (m2) Rh (m)
i 1.0298 10.35938 0.82182

65

32
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
4. Calculate Vi

5. Calculate Ei

6. Calculate Sf,i from Manning’s equation

(1) (2) (3) (4) (5) (6)

Location y (m) A (m )2 Rh (m) V (m/s) E (m) Sf
i 1.0298 10.35938 0.82182

66

Example 2. Computing the Water Surface Profile Using the Direct Step Method
4. Calculate Vi
𝑄 𝑄 30 m- /s
𝑉B = = = = 2.89593 m/s
𝐴B 𝑦B 𝑏 + 2𝑦B 10.35938 m!

5. Calculate Ei
m
𝑉B! (2.89593 )!
𝐸B = 𝑦B + = 1.0298 m + s = 1.45724 m
2𝑔 2𝑔

6. Calculate Sf,i from Manning’s equation

𝑘? m- 1 ! +
!/- +/!
𝑄= 𝐴B 𝑅@,B 𝑆5 = 30 = 10.35938 m! 0.82182 m - 𝑆5! ; 𝑆5 = 0.00681
𝑛 s 0.025
(1) (2) (3) (4) (5) (6)
Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681

67

33
 3/22/23

Computing dy/dx
y
Given y and x at location i, yi and xi
1. The value of yi+1 can be determined at yi xi, yi xi+1, yi+1
a prescribed xi+1 yi+1
2. The value of xi+1 can be determined
for a prescribed yi+1
x
xi xi+1
The Direct Step Method
• Since the specific energy E is a function of y and y varies with streamwise distance x, the slope
of the EGL (Sf) is also a function x
• Since we will not know what dx is prior to the computation, the value of Sf must be estimated
using some expression
• Several methods of determining Sf have been investigated, and it has been determined that as
long as dy and dx are sufficiently small, using the following expression for estimating Sf yields
satisfactory results:
+
𝑆0. = ) (𝑆.,A + 𝑆.,A;+ )

68

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Direct Step Computation Procedure
1. Start with yi = yc as previously calculated (yi = 1.0298 m)
2. Calculate Ai for the trapezoidal section
3. Calculate Rh,I
4. Calculate Vi
5. Calculate Ei
6. Calculate Sf,i from Manning’s equation

(1) (2) (3) (4) (5)

Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681
i+1

69

34
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• Recall that the procedure for determining the slope of the EGL requires Sf at locations i
as well as i+1:
+
𝑆5,B = (𝑆5,B + 𝑆5,BC+ ) [11]
!

• So at this point, we cannot proceed any further at location i M2

• We move on to location i+1 for the time being yn
• We first need to select yi+1
yc
• We have previously established that Δy should be small
• We will select Δy = 0.01 m, such that yi+1 = yi + Δy = 1.0398 m
• Recall that the flow depth is increasing, since we have an M2 profile and we are moving in
the upstream direction
(1) (2) (3) (4) (5)
Location y (m) A (m )2 Rh (m) V (m/s) E (m) Sf 𝑆_5
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681
i+1 1.0398

70

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• Now, steps 1 - 6 are now repeated using yi+1 to determine Ai+1, Rh,i+1, Vi+1, Ei+1 and Sf,i+1

7. Now, we can return to location yi and calculate 𝑆5,B using Eq. [11]

1
𝑆5,B = 𝑆 + 𝑆5,BC+
2 5,B

(1) (2) (3) (4) (5) (6) (7)

Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658

71

35
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• Now, steps 1 - 6 are now repeated using yi+1 to determine Ai+1, Rh,i+1, Vi+1, Ei+1 and Sf,i+1

7. Now, we can return to location yi and calculate 𝑆5,B using Eq. [11]

1 1
𝑆5,B = 𝑆 + 𝑆5,BC+ = 0.00681 + 0.00658 = 0.00669
2 5,B 2

• ΔE at any i+n cannot be calculated without E at the preceding location (i+n-1), which does
not exist when n = 0 (as it is for location i)

𝐸BC+ − 𝐸B
𝑥BC+ = 𝑥B +
1
𝑆4 − (𝑆5,B + 𝑆5,BC+ )
2
(1) (2) (3) (4) (5) (6) (7)
Location y (m) A (m )2 Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658

72

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• In order to proceed further at yi, we first need to proceed to yi+2

• So we complete steps 1 through 6, using yi+2 (calculated using the same Δy, 0.01 m) to
determine Ai+2, Rh,i+2, Vi+2, Ei+2 and Sf,i+2

7. Now, we can return to i+1 and calculate 𝑆5,BC+ using Eq. [11]:

1
𝑆5,BC+ = 𝑆 + 𝑆5,BC!
2 5,BC+

8. We are now able to calculate ΔEi+1

(1) (2) (3) (4) (5) (6) (7) (8)
Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658
i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636

73

36
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• In order to proceed further at yi, we first need to proceed to yi+2

• So we complete steps 1 through 6, using yi+2 (calculated using the same Δy, 0.01 m) to
determine Ai+2, Rh,i+2, Vi+2, Ei+2 and Sf,i+2

7. Now, we can return to i+1 and calculate 𝑆5,BC+ using Eq. [11]:

1 1
𝑆5,BC+ = 𝑆5,BC+ + 𝑆5,BC! = 0.00658 + 0.00636 = 0.00647
2 2

8. We are now able to calculate ΔEi+1 from 𝐸BC+ − 𝐸B = 1.45740 − 1.45724 = 0.00016
(1) (2) (3) (4) (5) (6) (7) (8)
Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016
i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636

74

Example 2. Computing the Water Surface Profile Using the Direct Step Method
9. Now, we can determine Δxi+1 by rearranging Eq. [15]:

∆𝐸BC+
∆𝑥BC+ = 𝑥BC+ − 𝑥B =
𝑆4 − 𝑆5,BC+

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx (m) ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016
i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636

75

37
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
9. Now, we can determine Δxi+1 by rearranging Eq. [15]:

∆𝐸BC+ 0.00016
∆𝑥BC+ = 𝑥BC+ − 𝑥B = = = −0.02847 m
𝑆4 − 𝑆5,BC+ 0.001 − 0.00647
• Note: Δx is negative because we are moving in the upstream direction (recall that we
have started at the location of the CDL for an M2 profile, and are moving upstream in
positive y increments, since dy/dx is increasing in the upstream direction)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx (m) ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016 -0.02847
i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636

76

Example 2. Computing the Water Surface Profile Using the Direct Step Method
9. Now, we can determine Δxi+1 by rearranging Eq. [15]:

∆𝐸BC+ 0.00016
∆𝑥BC+ = 𝑥BC+ − 𝑥B = = = −0.02847 m
𝑆4 − 𝑆5,BC+ 0.001 − 0.00647
• Note: Δx is negative because we are moving in the upstream direction (recall that we
have started at the location of the CDL for an M2 profile, and are moving upstream in
positive y increments, since dy/dx is increasing in the upstream direction)

10. Now we can calculate the cumulative value of Δx, which is the quantity which will be
used to plot the water surface profile
• Since we have no preceding value for Δx, ∑Δxi+1 = Δxi+1

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx (m) ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016 -0.02847 -0.02847
i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636

77

38
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• Steps 1-10 are now repeated at subsequent i+n locations

• How do we know when to truncate the computation?

• Again, we know that the NDL for an M2 profile is an asymptote of the water surface
profile

• For each y location, we can calculate y/yn and stop calculations when y = 0.99yn

• Finally, we plot ∑Δx vs. y, which gives the water surface profile.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
y/yn Location y (m) A (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx (m) ∑Δx
0.587 i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
0.593 i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016 -0.02847 -0.02847
0.599 i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636

78

Final solution

Proper significant figures used after computations were completed.

79

39
 3/22/23

80

Computing yi+1 from xi+1

• Recall that there are two methods by which the water surface profile is computed from the
governing ODE of GVF
• In the Direct Step Method, we were able to determine the streamwise distance x at which
some prescribed flow depth y occurs
• Conversely, there are several methods by which the flow depth y at some prescribed
streamwise distance x is determined
• These methods are distinguished as either single-step or predictor-corrector

• As in the direct step method, in the single step method the unknown depth at a
section is expressed in terms of f(x,y) at a neighboring point where y is known or has
been computed during the previous step

• We will be focusing on predictor-corrector methods, in which the value of yi+1 is first

predicted based on the quantities calculated for the previous location, and then
refined by an iterative procedure during the corrector step until a solution of desired
accuracy is attained

81

40
 3/22/23

Computing yi+1 from xi+1

• We can compute the rate of change of the flow depth with respect to distance, dy/dx, at a
specified streamwise location xi
"&
𝑦AG = "$ ⃒A = 𝑓(𝑥A , 𝑦A ) [16]

• In Eq. [16], the subscript i refers to quantities at distance xi, and a prime (') on y indicates
a derivative of y with respect to x

9,69-,6
𝑓(𝑥A , 𝑦A ) = 162/
[17]
+6
345
:,6

• In the above expression, all of the variables on the RHS are known and so 𝑓 𝑥B , 𝑦B (which
is equal to 𝑦BH ) can be computed

82

The Euler Method

• By assuming that the rate of variation of y (𝑦BH ) is constant over the interval xi to xi+1, yi+1 can be
determined from
𝑦;<= = 𝑦; + 𝑦;> ∆𝑥 [18]
• In Eq. [18], Δx = xi+1 – xi
• From Eq. [16], Eq. [18] then becomes

𝑦;<= = 𝑦; + 𝑓 𝑥; , 𝑦; ∆𝑥 [19]

• Eq. [19] is known as the Euler method

• If we were to expand yi+1 in a Taylor series such that

𝑦;<= = 𝑦; + 𝑦;> ∆𝑥 + 𝑂(∆𝑥)?

• We can see that O (Δx)2 is the local truncation error; from Eq. [19], the global error is O (Δx)
and so this method is first-order accurate

83

41
 3/22/23

The Improved Euler Method

• The accuracy of the method could be improved by using the slope of y at both xi and xi+1 to
solve for yi+1 at xi+1

• However, since we don’t know the slope of y at xi+1 since we do not yet know yi+1, this is not
possible using Eq. [19] in its current form

• If we initially predict a value of yi+1 using the Euler method, we may then evaluate the slope of
y, f(yi+1), at xi+1

∗
• We will call the predicted flow depth at xi+1 obtained from Eq. [19] as 𝑦BC+ such that
∗
𝑦;<= = 𝑦; + 𝑦;> ∆𝑥

∗
• Using 𝑦BC+ , we can compute the slope of the solution curve y at x = xi+1

G ∗
𝑦A;+ = 𝑓 𝑥A;+ , 𝑦A;+ [20]

84

The Improved Euler Method

• In order to further improve accuracy, we may use the average value of the slopes of the
solution curve at xi and xi+1, from which Eq. [19] becomes

=
𝑦;<= = 𝑦; + (𝑦;> + 𝑦;<=
>
)∆𝑥 [21]
?

which can be written as either Eq. [22a] or Eq. [22b]

= ∗
𝑦;<= = 𝑦; + 𝑓 𝑦; + 𝑓 𝑦;<= ∆𝑥 [22a]
?
= ∗
𝑦;<= = 𝑦; + 𝑓 𝑥; , 𝑦; + 𝑓 𝑥;<= , 𝑦;<= ∆𝑥 [22b]
?

• Eq. [22] is called the improved Euler method; by expanding Eq. [22] in Taylor series, the
method is shown to be second-order accurate

• This method is a predictor-corrector method whose solution can be refined by multiple inner
iterations; however, if this is not properly executed, the solution may diverge

85

42
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
Consider a trapezoidal prismatic channel with Manning’s n = 0.025 and So = 0.001.
Compute the water surface profile for a design Q of 30 m3/s if the channel ends in a drop.
B

• From Example 2
1 y
• yn = 1.754 m 2
• yc = 1.0298 m
8m
for this problem

1. Recall that we are still dealing with an M2 profile, so we will still start with yi ≈ yc;
however, we will choose a slightly higher value for yi in order to avoid an error which
may occur in calculating dy/dx

(1)

xi (m) yi (m)

0 1.0300

86

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
2. Calculate Ai for the trapezoidal channel using yi
3. Calculate Rh,i
B
4. Calculate Bi
1 y
2

8m

(1) (2) (3) (4) (5) (6) (7) (8)

xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi)

0 1.0300

87

43
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
2. Calculate Ac,i for the trapezoidal channel using yi
3. Calculate Rh,i
B
4. Calculate Bi
5. Calculate Fri 1 y
!
𝐵B 𝑄! 2
Fr =
𝑔𝐴-J,B
8m
2. Calculate Vi = Q/Ai

(1) (2) (3) (4) (5) (6) (7) (8)

xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi)

0 1.0300 10.3618 0.8220 12.1200

88

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
7. Calculate Sf,i using Manning’s equation

B
8. Calculate f(yi) from the GVF equation
1 y
𝑑𝑦 𝑆4 − 𝑆5,B 2
= f(yi) =
𝑑𝑥 1 − FrB!
8m

(1) (2) (3) (4) (5) (6) (7) (8)

xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi)

0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068

89

44
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
7. Calculate Sf,i using Manning’s equation

8. Calculate f(yi) from the GVF equation

𝑑𝑦 𝑆4 − 𝑆5,B
= f(yi) =
𝑑𝑥 1 − FrB!

(1) (2) (3) (4) (5) (6) (7) (8)

xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi)

0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871

90

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗
9. Calculate 𝑦BC+ from Eq. [ 19] by prescribing Δx

∗
𝑦BC+ = 𝑦B + 𝑓(𝑦B )∆𝑥

• The value of Δx does not need to be held constant throughout the computation process
• It is common practice to begin with smaller intervals and then gradually increase the step
size as the computation proceeds
• We will begin with a small step size of Δx = 0.005 m, such that x at location i+1 = - 0.005
m (recall that we are moving upstream in our computation)

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871

91

45
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗
9. Calculate 𝑦BC+ from Eq. [ 19] by prescribing Δx

∗
𝑦BC+ = 𝑦B + 𝑓(𝑦B )∆𝑥

• The value of Δx does not need to be held constant throughout the computation process
• It is common practice to begin with smaller intervals and then gradually increase the step
size as the computation proceeds
• We will begin with a small step size of Δx = 0.005 m, such that x at location i+1 = - 0.005
m (recall that we are moving upstream in our computation)

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871

-0.005

92

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗
9. Calculate 𝑦BC+ from Eq. [ 19] by prescribing Δx

∗
𝑦BC+ = 𝑦B + 𝑓(𝑦B )∆𝑥

• The value of Δx does not need to be held constant throughout the computation process
• It is common practice to begin with smaller intervals and then gradually increase the step
size as the computation proceeds
• We will begin with a small step size of Δx = 0.005 m, such that x at location i+1 = - 0.005
m (recall that we are moving upstream in our computation)

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871 1.0849

-0.005

93

46
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗
10. Calculate A*i+1, Rh*,i+1, B*i+1, Fr*i+1, Vi+1, and Sf*,i+1 from 𝑦BC+

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871

-0.005

94

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗
10. Calculate A*i+1, Rh*,i+1, B*i+1, Fr*i+1, Vi+1, and Sf*,i+1 from 𝑦BC+
11. Calculate f(y*i+1) from GVF equation

𝑑𝑦 𝑆4 − 𝑆 ∗5,BC+
= f(y∗i+1) = ∗!
𝑑𝑥 1 − FrBC+

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871 1.0849 11.0337 0.8585 12.3397 0.9180 2.7190 0.0057

-0.005

95

47
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗
10. Calculate A*i+1, Rh*,i+1, B*i+1, Fr*i+1, Vi+1, and Sf*,i+1 from 𝑦BC+
11. Calculate f(y*i+1) from GVF equation

𝑑𝑦 𝑆4 − 𝑆 ∗5,BC+
= f(y∗i+1) = ∗!
𝑑𝑥 1 − FrBC+
12. Calculate yi+1 from Eq. 22a
+ ∗
𝑦BC+ = 𝑦B + 𝑓 𝑦B + 𝑓 𝑦BC+ ∆𝑥
!

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871 1.0849 11.0337 0.8585 12.3397 0.9180 2.7190 0.0057 -0.0297

-0.005

96

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗
10. Calculate A*i+1, Rh*,i+1, B*i+1, Fr*i+1, Vi+1, and Sf*,i+1 from 𝑦BC+
11. Calculate f(y*i+1) from GVF equation

𝑑𝑦 𝑆4 − 𝑆 ∗5,BC+
= f(y∗i+1) = ∗!
𝑑𝑥 1 − FrBC+
12. Calculate yi+1 from Eq. 22a
+ ∗
𝑦BC+ = 𝑦B + 𝑓 𝑦B + 𝑓 𝑦BC+ ∆𝑥
!
• Repeat for subsequent steps
• Plot xi vs yi

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ai (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871 1.0849 11.0337 0.8585 12.3397 0.9180 2.7190 0.0057 -0.0297

-0.005 1.0575

97

48
 3/22/23

Tutorial Assignment
• For this week’s practice problem, complete Examples 2 and 3 (e.g. computation of the
water surface profile using both the Direct Step Method and the Improved Euler Method)
• Show all calculations for one iteration of each method, including preliminary
calculations and analysis which precede the computation procedure (yc, yn, classification
of water surface profile, etc.)
• Develop spreadsheets as shown in class for both methods and plot the water surface
profiles
• Plots should be appropriately formatted with axis titles, etc.

98

Notes on the Improved Euler Method

On selection of Δx:
As previously noted, the value of Δx should gradually increase as the computation
proceeds. For the assignment, please follow the following convention for uniformity:

• Begin with xi = 0, then take Δx = -0.005 m

• From -0.01 m to -0.1 m, use 0.01 m increments
• From -0.1 m to -1 m, use 0.1 m increments
• From -2 m onwards, use 2 m increments

On truncation of the computation:

f(yi) will be negative at the beginning of the computation and then will become positive
stop the computation when f(yi) becomes negative again (indicating that y has approached
the NDL)

99

49
 3/22/23

Solutions

100

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Determining yi

B
• In order to select yi, both yn and yc need to be determined
1 y
𝑘? !/- 2
𝑄= 𝐴 𝑅 𝑆 +/!
𝑛 J @
8m
1 !/-
30 m- /s = 𝐴 𝑅 (0.001)+/!
0.025 J @

• Recall that
• 𝐴J = 𝑦? 𝑏 + 2𝑦?
• 𝑝 = 𝑏 + 2 5𝑦? yn = 1.754 m
,
• 𝑅@ = %
E

101

50
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Determining yc

F
B
• Fr = *'%
1 y
2
• Setting Fr = 1 and V = Q/A (taking yc= Ac/B for a trapezoidal channel)
&/(
8m
) ,%
= #/( ; from which yc = 1.0298 m
* G

Classifying the water surface profile

M2
yn
• We know that the slope is mild (yn > yc)
• We are told that the profile ends in overfall yc
• This is an M2 profile

102

Example 2. Computing the Water Surface Profile Using the Direct Step Method

• With an M2 profile, we know that the surface

profile meets the CDL in the downstream B
direction, and the NDL acts as an asymptote of
the profile in the upstream direction
1 y
2
• This knowledge can inform the point at which
8m
we will begin our computation of the water
surface profile

• For this example, we will begin with y = yc and

work our way in the upstream direction at
M2
increments of x yn

yc
• However, theoretically we could begin at any y
< yn and still arrive at the same solution

103

51
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• We can begin the computations at any location
where the flow depth for the specified
discharge is known B

• It is common practice to start the computations 1 y

at a downstream control section if the flow is 2
subcritical and proceed in the upstream
direction 8m

• For supercritical flows, however, we start at an

upstream control section and compute the
profile in the downstream direction
M2
yn
• Unfortunately, this fact has been incorrectly
attributed in many well-known publications to yc
indicate that the computations become
unstable or yield incorrect results if this
convention is not followed

104

Example 2. Computing the Water Surface Profile Using the Direct Step Method

• Other than the fact that the flow depth is

known at a control section, there appears to B
be no reason why we should proceed in either
the upstream or downstream direction
1 y
2
• This is because all we are doing is either 8m
numerically solving a differential equation for
the specified initial condition or solving a
nonlinear algebraic equation

M2
• Whether we proceed in the positive or yn
negative x direction should make little
difference provided the computational step is
properly selected in the computations yc

105

52
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
Direct Step Computation Procedure

1. Start with yi = yc as previously calculated (yi = 1.0298 m)

2. Calculate Ac,i for the trapezoidal section

𝐴J,B = 𝑦B 𝑏 + 2𝑦B = 1.0298 m 8 m + 2 ∗ 1.0298 m = 10.35938 m!

3. Calculate Rh,I

𝐴J,B 𝑦B 𝑏 + 2𝑦B 10.35938 m!

𝑅@,B = = = = 0.82182 m
𝑝B 𝑏 + 2 5𝑦B 8 m + 2 5(1.0298 m)

(1) (2) (3)

Location y (m) 2
Ac (m ) Rh (m)
i 1.0298 10.35938 0.82182

106

Example 2. Computing the Water Surface Profile Using the Direct Step Method
4. Calculate Vi
𝑄 𝑄 30 m- /s
𝑉B = = = = 2.89593 m/s
𝐴J,B 𝑦B 𝑏 + 2𝑦B 10.35938 m!

5. Calculate Ei
m
𝑉B! (2.89593 )!
𝐸B = 𝑦B + = 1.0298 m + s = 1.45724 m
2𝑔 2𝑔

6. Calculate Sf,i from Manning’s equation

𝑘? m- 1 ! +
!/- +/!
𝑄= 𝐴J,B 𝑅@,B 𝑆5 = 30 = 10.35938 m! 0.82182 m - 𝑆5! ; 𝑆5 = 0.00681
𝑛 s 0.025
(1) (2) (3) (4) (5) (6)
Location y (m) Ac (m2) Rh (m) V (m/s) E (m) Sf
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681

107

53
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• Recall that the procedure for determining the slope of the EGL requires Sf at locations i
as well as i+1:
+
𝑆5,B = (𝑆5,B + 𝑆5,BC+ ) [11]
!

• So at this point, we cannot proceed any further at location i M2

• We move on to location i+1 for the time being yn
• We first need to select yi+1
yc
• We have previously established that Δy should be small
• We will select Δy = 0.01 m, such that yi+1 = yi + Δy = 1.0398 m
• Recall that the flow depth is increasing, since we have an M2 profile and we are moving in
the upstream direction
(1) (2) (3) (4) (5)
Location y (m) 2
Ac (m ) Rh (m) V (m/s) E (m) Sf 𝑆_5
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681
i+1 1.0398

108

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• Now, steps 1 - 6 are now repeated using yi+1 to determine Ac,i+1, Rh,i+1, Vi+1, Ei+1 and Sf,i+1

7. Now, we can return to location yi and calculate 𝑆5,B using Eq. [11]

1 1
𝑆5,B = 𝑆5,B + 𝑆5,BC+ = 0.00681 + 0.00658 = 0.00669
2 2

• ΔE at any i+n cannot be calculated without E at the preceding location (i+n-1), which does
not exist when n = 0 (as it is for location i)

𝐸BC+ − 𝐸B
𝑥BC+ = 𝑥B +
1
𝑆4 − (𝑆5,B + 𝑆5,BC+ )
2
(1) (2) (3) (4) (5) (6) (7)
Location y (m) Ac (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658

109

54
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• In order to proceed further at yi, we first need to proceed to yi+2

• So we complete steps 1 through 6, using yi+2 (calculated using the same Δy, 0.01 m) to
determine Ac,i+2, Rh,i+2, Vi+2, Ei+2 and Sf,i+2

7. Now, we can return to i+1 and calculate 𝑆5,BC+ using Eq. [11]:

1 1
𝑆5,BC+ = 𝑆5,BC+ + 𝑆5,BC! = 0.00658 + 0.00636 = 0.00647
2 2

8. We are now able to calculate ΔEi+1 from 𝐸BC+ − 𝐸B = 1.45740 − 1.45724 = 0.00016
(1) (2) (3) (4) (5) (6) (7) (8)
Location y (m) Ac (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016
i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636

110

Example 2. Computing the Water Surface Profile Using the Direct Step Method
9. Now, we can determine Δxi+1 by rearranging Eq. [15]:

∆𝐸BC+ ∆𝐸BC+ 0.00016

∆𝑥BC+ = 𝑥BC+ − 𝑥B = = = = −0.02847 m
𝑆4 − 𝑆5,BC+ 𝑆4 − 𝑆5,BC+ 0.001 − 0.00647
• Note: Δx is negative because we are moving in the upstream direction (recall that we
have started at the location of the CDL for an M2 profile, and are moving upstream in
positive y increments, since dy/dx is increasing in the upstream direction)

10. Now we can calculate the cumulative value of Δx, which is the quantity which will be
used to plot the water surface profile
• Since we have no preceding value for Δx, ∑Δxi+1 = Δxi+1

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Location y (m) Ac (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx (m) ∑Δx
i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016 -0.02847 -0.02847
i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636

111

55
 3/22/23

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• Steps 1-10 are now repeated at
subsequent i+n locations
• How do we know when to
truncate the computation?
• Again, we know that the NDL for an
M2 profile is an asymptote of the
water surface profile
• For each y location, we can
calculate y/yn and stop calculations
when y = 0.99yn
• Finally, we plot ∑Δx vs. y, which
gives the water surface profile:

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
y/yn Location y (m) Ac (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx (m) ∑Δx
0.587 i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
0.593 i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016 -0.02847 -0.02847
0.599 i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636 0.00626 0.00046 -0.08773 -0.11620

112

Example 2. Computing the Water Surface Profile Using the Direct Step Method
• Steps 1-10 are now repeated at 2.0
subsequent i+n locations
• How do we know when to 1.5
truncate the computation?
• Again, we know that the NDL for an
y (m)

M2 profile is an asymptote of the 1.0

water surface profile
• For each y location, we can 0.5
calculate y/yn and stop calculations
when y = 0.99yn
0.0
• Finally, we plot ∑Δx vs. y, which -2000 -1500 -1000 -500 0
gives the water surface profile:
x (m)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
y/yn Location y (m) Ac (m2) Rh (m) V (m/s) E (m) Sf 𝑆_5 ΔE Δx (m) ∑Δx
0.587 i 1.0298 10.35938 0.82182 2.89593 1.45724 0.00681 0.00669
0.593 i+1 1.0398 10.48077 0.82851 2.86239 1.45740 0.00658 0.00647 0.00016 -0.02847 -0.02847
0.599 i+2 1.0498 10.60256 0.83519 2.82951 1.45786 0.00636 0.00626 0.00046 -0.08773 -0.11620

113

56
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
Consider a trapezoidal prismatic channel with Manning’s n = 0.025 and So = 0.001.
Compute the water surface profile for a design Q of 30 m3/s.
B

• From Example 2
1 y
• yn = 1.754 m 2
• yc = 1.0298 m
8m
for this problem

1. Recall that we are still dealing with an M2 profile, so we will still start with yi ≈ yc;
however, we will choose a slightly higher value for yi in order to avoid an error which
may occur in calculating dy/dx

(1)

xi (m) yi (m)

0 1.0300

114

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
2. Calculate Ac,i for the trapezoidal channel using yi
3. Calculate Rh,i
B
4. Calculate Bi
5. Calculate Fri 1 y
6. Calculate Vi = Q/Ai 2

7. Calculate Sf,I using Manning’s equation 8m

8. Calculate f(yi) from the GVF equation

𝑑𝑦 𝑆4 − 𝑆5
= f(yi) =
𝑑𝑥 1 − 𝐹.!

(1) (2) (3) (4) (5) (6) (7) (8)

xi (m) yi (m) Ac,i (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi)

0 1.0300

115

57
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
2. Calculate Ac,i for the trapezoidal channel using yi
3. Calculate Rh,i
B
4. Calculate Bi
5. Calculate Fri 1 y
6. Calculate Vi = Q/Ai 2

7. Calculate Sf,I using Manning’s equation 8m

8. Calculate f(yi) from the GVF equation

𝑑𝑦 𝑆4 − 𝑆5
= f(yi) =
𝑑𝑥 1 − 𝐹.!

(1) (2) (3) (4) (5) (6) (7) (8)

xi (m) yi (m) Ac,i (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi)

0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871

116

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗ ∗
9. Calculate 𝑦BC+ from 𝑦BC+ = 𝑦B + 𝑦BH ∆𝑥 = 𝑦B + 𝑓(𝑦B )∆𝑥 by prescribing Δx
∗
10. Calculate A*i+1, Rh*,i+1, B*i+1, Fr*i+1, Vi+1, and Sf*,i+1 from 𝑦BC+
B
11. Calculate f(y*i+1) from GVF equation
12. Calculate yi+1 from Eq. 22a 1 y
+ ∗ 2
𝑦BC+ = 𝑦B + 𝑓 𝑦B + 𝑓 𝑦BC+ ∆𝑥
!
8m

• Repeat until yi begins to decrease (indicating that y has approached the NDL)
• Plot xi vs yi

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ac,i (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871

1.0575

117

58
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method
∗ ∗
9. Calculate 𝑦BC+ from 𝑦BC+ = 𝑦B + 𝑦BH ∆𝑥 = 𝑦B + 𝑓(𝑦B )∆𝑥 by prescribing Δx
∗
10. Calculate A*i+1, Rh*,i+1, B*i+1, Fr*i+1, Vi+1, and Sf*,i+1 from 𝑦BC+
B
11. Calculate f(y*i+1) from GVF equation
12. Calculate yi+1 from Eq. 22a 1 y
2
+ ∗
𝑦BC+ = 𝑦B + 𝑓 𝑦B + 𝑓 𝑦BC+ ∆𝑥 8m
!

• Repeat until yi begins to decrease (indicating that y has approached the NDL)
• Plot xi vs yi

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ac,i (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) Fr*i+1 Sf*,i+1 f(y*i+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871 1.0849 11.0337 0.8585 12.3397 0.9180 2.7190 0.0057 -0.0297

1.0575

118

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method

1 y
2

8m

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ac,i (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) F*i+1 Sf*,i+1 f(yi+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871 1.0849 11.0337 0.8585 12.3397 0.9180 2.7190 0.0057 -0.0297

-0.005 1.0575 10.6971 0.8403 12.2302 0.9574 2.8045 0.0062 -0.0624 1.0582 10.7048 0.8408 12.2327 0.9565 2.8025 0.0062 -0.0609

-0.01 1.0579 10.7009 0.8405 12.2314 0.9570 2.8035 0.0062 -0.0617 1.0591 10.7160 0.8414 12.2363 0.9551 2.7996 0.0062 -0.0589

119

59
 3/22/23

Example 3. Computing the Water Surface Profile Using the Improved Euler
Method

1 y
2

8m

(1), (12) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
A*c,i+1 Rh*,i+1 V*i+1
xi (m) yi (m) Ac,i (m2) Rh,i (m) Bi (m) Fri Vi (m/s) Sf,i (m) f(yi) y*i+1 (m) B*i+1 (m) F*i+1 Sf*,i+1 f(yi+1)
(m2) (m) (m/s)
0 1.0300 10.3618 0.8220 12.1200 0.9997 2.8952 0.0068 -10.9871 1.0849 11.0337 0.8585 12.3397 0.9180 2.7190 0.0057 -0.0297

-0.005 1.0575 10.6971 0.8403 12.2302 0.9574 2.8045 0.0062 -0.0624 1.0582 10.7048 0.8408 12.2327 0.9565 2.8025 0.0062 -0.0609

-0.01 1.0579 10.7009 0.8405 12.2314 0.9570 2.8035 0.0062 -0.0617 1.0591 10.7160 0.8414 12.2363 0.9551 2.7996 0.0062 -0.0589

120

A review

121

60
 3/22/23

Flow profiles
with various controls

122

61