Designing Channels

Run Manning's equation backward, and it designs the channel.

Before this Lesson

A new subdivision will send a 150 ft³/sec design peak down a rectangular concrete channel 8 ft wide, with n=0.013n = 0.013. Two questions about it sound almost the same. What slope would carry that flow 2 ft deep? Manning's equation answers in one line. How deep will the water run on the 0.002 slope the ground allows? Now every quantity but the depth is known, and no rearranging of the equation will ever hand you a formula for it.

Both are the same equation with one blank in it. What changes is where the blank sits.

In Manning's Equation, the channel was given and the discharge was the answer. In design, the watershed sets the discharge, and the question asks for one other quantity in the Handbook's Manning's equation:

Q=Kn ARH2/3S1/2Q = \frac{K}{n}\,A R_H^{2/3} S^{1/2}

If the blank is the slope SS or the roughness nn, it sits outside the section's geometry, and algebra pulls it out. If it's the depth yy or the bottom width bb, it hides inside both AA and RHR_H, and you have to search. Once the depth is known, one last check asks whether the water moves too fast for its lining.

Slope and roughness: one line of algebra

The slope is what pushes the water along. Divide both sides of Manning's equation by everything but S1/2S^{1/2}, then square:

S=(QnKARH2/3)2S = \left(\frac{Qn}{K A R_H^{2/3}}\right)^2

The square is where people slip. QQ goes as the square root of SS, so SS goes as Q2Q^2: pushing twice the flow through the same section at the same depth takes four times the slope. Leave the square off and you get the square root of the right answer, and for slopes as small as channels have, that's many times too steep.

That settles the Hook's first question. At 2 ft deep, the 8 ft channel has A=16 ft2A = 16\ \text{ft}^2, P=8+2(2)=12P = 8 + 2(2) = 12 ft, and RH=1.333R_H = 1.333 ft, so

S=(150(0.013)1.486(16)(1.333)2/3)2=0.0046S = \left(\frac{150(0.013)}{1.486(16)(1.333)^{2/3}}\right)^2 = 0.0046

Roughness comes out the same way, with no square, since nn appears to the first power:

n=KQ ARH2/3S1/2n = \frac{K}{Q}\,A R_H^{2/3} S^{1/2}

This is the roughest lining that still carries the design flow at that depth. Any lining with a lower nn carries it lower in the channel. The same line works backward from the field: measure the flow and the uniform depth in an existing channel, and it tells you how rough the channel really is. Flow in Sewers makes the same move with the velocity form, finding the slope that keeps sewage moving fast enough to carry its solids.

Worked Example The slope for an earth channel
1 of 3

A trapezoidal earth channel with n=0.025n = 0.025 has a 6 ft bottom width and 3:1 side slopes. What bed slope will carry a design flow of 120 ft³/sec at a depth of 2.5 ft?

Normal depth: no formula, so search

Now the Hook's second question. In a long channel, a steady flow settles to the one depth at which gravity's pull balances the boundary's drag. That's the Handbook's uniform flow, where depth and velocity don't change along the channel, and its depth is the normal depth, written here yny_n. The Handbook doesn't name it, but every question about how deep a design flow runs asks for it.

Put everything you know on one side of Manning's equation and everything that depends on the depth on the other:

ARH2/3=QnKS1/2A R_H^{2/3} = \frac{Qn}{K S^{1/2}}

The right side is one number, the target, fixed by the flow, the lining, and the slope. The left side, called the section factor, depends only on the shape and the depth. For the Hook's channel, the target is

150(0.013)1.486(0.002)1/2=29.3\frac{150(0.013)}{1.486(0.002)^{1/2}} = 29.3

Now try to solve for yy. In this rectangle, the section factor is (8y)5/3/(8+2y)2/3(8y)^{5/3}/(8 + 2y)^{2/3}. Clear the fractional powers and you're left with a fifth-degree polynomial in yy, and those have no general formula. The depth isn't hard to find. It just can't be written down.

What saves you is that in a rectangle or trapezoid, the section factor only grows as the water rises, so exactly one depth hits the target, and every guess tells you which way to go. Try 2 ft: 16(1.333)2/3=19.416(1.333)^{2/3} = 19.4, short of 29.3, so go deeper. Try 3 ft: 24(1.714)2/3=34.424(1.714)^{2/3} = 34.4, past it. Interpolating between them,

y≈2+29.3−19.434.4−19.4=2.66 fty \approx 2 + \frac{29.3 - 19.4}{34.4 - 19.4} = 2.66\ \text{ft}

A check at 2.7 ft gives 29.7, just over, so the depth sits a hair below 2.7 ft: yn=2.68y_n = 2.68 ft.

0 1 2 3 4 0 10 20 30 40 50 60 flow depth, y (ft) A RH2/3 (ft8/3) target = 29.3 y = 2 ft: 19.4 y = 3 ft: 34.4 yn = 2.68 ft
The 8 ft channel's section factor rises with depth and meets the target once
The trial depths of 2 ft and 3 ft fall on either side of the target, which the curve crosses at the normal depth.

On the exam, the answer options are your guesses. Compute the target once, then test an option near the middle: if it falls short, the answer is deeper, and if it overshoots, shallower. With no options, guess and interpolate as above, or hand Manning's equation, written in terms of yy, to the calculator's equation solver. A bottom width works the same way: fix the depth and test widths.

The wide-channel shortcut

When a channel is many times wider than it is deep, the banks add almost nothing to the wetted perimeter. Then P≈bP \approx b, so RH=by/b≈yR_H = by/b \approx y, and the depth comes out of hiding. Per unit width, with the unit discharge q=Q/bq = Q/b:

q=Kn y5/3S1/2⇒yn≈(qnKS1/2)3/5q = \frac{K}{n}\,y^{5/3} S^{1/2} \quad\Rightarrow\quad y_n \approx \left(\frac{qn}{K S^{1/2}}\right)^{3/5}

In a rectangle, pretending RH=yR_H = y overstates the hydraulic radius, so the shortcut always comes out shallow: about 7 percent at a width of 10 times the depth, and 4 percent at 20 times. That's fine for a broad river. The Hook's channel is only 3 times as wide as it is deep, and the shortcut gives 2.18 ft, 19 percent short of the true 2.68 ft. (In a trapezoid, the water over the sloped sides pulls the other way, but the rule is the same.) The Handbook doesn't print the shortcut, so check the width against the depth you get before you trust it.

Worked Example Normal depth from the options
1 of 3

A trapezoidal earth channel with n=0.025n = 0.025 has a 3.0 m bottom width and 2:1 side slopes, on a slope of 0.0004. For a design flow of 10 m³/s, the normal depth is most nearly: (A) 1.4 m, (B) 1.6 m, (C) 1.8 m, or (D) 2.0 m.

Too fast for the lining?

With the normal depth known, continuity gives the velocity: v=Q/Av = Q/A, with AA taken at yny_n. The Hook's channel runs at 150/(8×2.68)=7.0150/(8 \times 2.68) = 7.0 ft/sec.

Speed matters because the water drags on the boundary as hard as the boundary drags on the water, and that drag grows roughly as the square of the speed. Past some speed, bare soil scours, grass tears out, and the stones of a riprap lining roll. So each lining has a maximum permissible velocity, a ceiling the design velocity must not pass. The Handbook prints no such limits, so problems state them. Concrete takes the Hook's 7.0 ft/sec in stride, but grass or bare earth is usually rated for less.

Take the area at the normal depth, not at the top of the banks. The dry freeboard above the water carries nothing, so the full bank-height area makes the water look slower than it is. When the velocity is too high, flatten the slope, perhaps with drop structures that step the channel down, or switch to a tougher lining such as riprap or concrete.

Checkpoint
4 ft 2.5 ft 1.5 ft 3 1
A grass-lined channel at its design flow
Vertical scale exaggerated 2×.

A grass-lined trapezoidal channel has a 4 ft bottom width, 3:1 side slopes, and banks 2.5 ft high. At its design flow of 60 ft³/sec, it runs at a normal depth of 1.5 ft. The grass is rated for a maximum permissible velocity of 4.0 ft/sec. Which conclusion is correct?

A.
The velocity is 2.1 ft/sec, so the grass holds.
B.
The velocity is 4.7 ft/sec, safely above the 4.0 ft/sec limit, so the grass holds.
C.
The velocity is 10.0 ft/sec, so the grass will erode.
D.
The velocity is 4.7 ft/sec, so the grass will erode.

Recap

  • A design question is Manning's equation with one blank. Slope and roughness come out by algebra: S=[Qn/(KARH2/3)]2S = [Qn/(K A R_H^{2/3})]^2, squared because QQ goes as S1/2S^{1/2}, and n=(K/Q)ARH2/3S1/2n = (K/Q) A R_H^{2/3} S^{1/2}.
  • The normal depth yny_n hides in both AA and RHR_H, so compute the target Qn/(KS1/2)Qn/(K S^{1/2}) once and test depths, starting with the options, until ARH2/3A R_H^{2/3} matches it. A bottom width is found the same way.
  • When the width is many times the depth, yn≈[qn/(KS1/2)]3/5y_n \approx [qn/(K S^{1/2})]^{3/5} with q=Q/bq = Q/b. In a rectangle, it always runs a little shallow.
  • Check v=Q/Av = Q/A at the normal depth, not the bank top, against the lining's maximum permissible velocity, which is a ceiling.