Computation of critical and normal depths for prismatic open channels
Free surface flows are gravity-driven flows that occur in open water systems such as canals,
rivers, lakes and oceans. The study of free surface flows can be simplified by assuming that
the curvature of streamlines is small, leading to negligible vertical velocities and
accelerations, and consequently to hydrostatic pressure distribution (Figure 1). The
assumption of this type of ‘quasi-horizontal’ flows facilitates the mathematical solution since
the pressure can be estimated directly from the water depth.
Figure 1 (a) Water pressure over linear slope bed. (b) Water pressure over convex shaped
bed. (c) Water pressure over concave-shaped bed.
One-dimensional open channel flow
The main assumptions made for the mathematical treatment of one-dimensional open channel
flows are the following:
1. The longitudinal dimension of the flow domain is much larger than the
transverse dimensions (width and depth of the channel).
2. The unknown variables are the average cross-sectional velocity u(x,t)
and the water depth y(x,t) measured from the free surface to the lowest
point of the channel bed. This depth can be expressed either as the sum
of the initial depth, ho(x) and the surface elevation ζ(x,t) above the initial
depth, or the difference between the elevation of the free surface and the
elevation of the channel bed, measured from a certain reference datum.
3. Due to small bed slopes, the hydrostatic pressure is approximated by
considering the vertical depth (y), instead of the depth perpendicular
to the bed (hp) (Figure 1).
4. The approximation of nearly horizontal flow and that of hydrostatic pressure
distribution is valid.
The behaviour of these flows depends heavily on the Froude number:
𝑢
𝐹𝑟 = (1)
√𝑔𝑦
For Fr = 1 flow is critical, for Fr < 1 flow is subcritical and for Fr > 1 flow is supercritical.
The Froude number is indicative of the balance between inertia and gravity forces.
Another important classification for open channel flows is their variability in time, t, or
𝜕𝜑 𝜕𝜑 𝜕𝜑
distance, x. Thus flows can be steady ( 𝜕𝑡 = 0) or unsteady ( 𝜕𝑡 ≠ 0), uniform ( 𝜕𝑥 = 0) or
𝜕𝜑
non-uniform ( 𝜕𝑥 ≠ 0), where 𝜑 is either
the depth y(x,t) or the velocity u(x,t). Furthermore, non-uniform flows can be gradually
varied (surface profiles) or rapidly varied.
Steady-state gradually varied non-uniform flow
Two important features of any one-dimensional open channel flow are the
normal depth, yn, and the critical depth, yc. The normal depth corresponds
to the water depth under uniform flow conditions and can be estimated using
Manning’s equation (Equation 2) or Chezy’s equation (Equation 3):
𝑄 1 2⁄ 1
𝑢 = 𝐴 = 𝑛 𝑅ℎ 3 𝑆 ⁄2 (2)
𝑜
𝑢 = 𝐶𝑧 √𝑅ℎ 𝑆𝑜 (3)
where Rh is the hydraulic radius; So is the bed slope; and n and Cz are
the Manning’s and the Chezy’s coefficients of friction, respectively. The hydraulic radius is
expressed as
𝐴
𝑅ℎ = 𝑃 (4)
𝑤
where A is the flow cross-sectional area and Pw is the wetted perimeter defined as the water–
solid boundary interface (Figure 2).
Figure 2 Schematic illustration of open channel characteristics. (a) Longitudinal view
and (b) cross-sectional area.
It should be noted that in Equations 2 and 3 the bed slope (So) is being used for the estimation
of the normal depth, that is, uniform flow. For non-uniform flow, the energy gradient (Se)
should be used instead.
For an arbitrary-shaped cross-section area, the critical depth (yc) is estimated from the
equation
𝑄2 𝑇
=1 (5)
𝑔𝐴3
where Q is the discharge and T is the surface flow width. For rectangular channels, Equation
5 can be simplified and solved for the critical depth (yc) as
3 𝑞2
𝑦𝑐 = √ 𝑔 (6)
where q is the flow discharge per unit width.
Since both yn and yc are implicitly involved in Equations 2, 3 and 5,
solutions can be obtained very effectively using a numerical iterative algorithm such as the
Newton-Raphson method.
By considering a trapezoidal shaped channel (side slope 1 Vertical: m Horizontal), the
equations that describe its geometric characteristics are as follows:
𝐴 = (𝑏 + 𝑚𝑦)𝑦 (7)
𝑃𝑤 = 𝑏 + 2𝑦√1 + 𝑚2 (8)
𝑇 = (𝑏 + 2𝑚𝑦) (9)
These equations can be also used for rectangular channels (m = 0) or for triangular ones (b =
0).
Combining Equations 2, 7 and 8 leads to the following implicit nonlinear function, fn(yn), for
the normal depth:
5
√𝑆 [(𝑏+𝑚𝑦𝑛 )𝑦𝑛 ] ⁄3
𝑓𝑛 (𝑦𝑛 ) = ⁄2 −𝑄 =0 (10)
𝑛
[(𝑏+2𝑦𝑛 √1+𝑚2 )] 3
Similarly, combining Equations 5, 7 and 9 results into the following implicit non-linear
function, fc(yc), for the critical depth and its first derivative:
3
[(𝑏+𝑚𝑦𝑐 )𝑦𝑐 ] ⁄2 𝑄
𝑓𝑐 (𝑦𝑐 ) = − =0 (11)
√𝑏+2𝑚𝑦𝑐 √𝑔
For the computations of normal and critical depth, first derivatives of Equation 10 (wrt yn)
and Equation 11 (with respect to yc) is obtained.
Once the functions and the derivatives are established, the Newton Raphson iterative solution
algorithm can be applied as
𝑓(𝑦 (𝑘) )
𝑦 (𝑘+1) = 𝑦 (𝑘) − 𝑓′ (𝑦 (𝑘)) (12)
Starting with a very small initial value for y(0), the algorithm converges very rapidly to the
solution for either the yn or yc.