DEPARTMENT OF THE ARMY EM 1110-2-1601

U.S. Army Corps of Engineers Change 1

CECW-EH-D Washington, D.C. 20314-1000

Engineer Manual
No. 1110-2-1601 30 June 1994

Engineering and Design

HYDRAULIC DESIGN OF FLOOD CONTROL CHANNELS

1. This Change 1 to EM 1110-2-1601, 1 Jul 91:

a. Updates Chapter 2.

b. Updates Chapter 3.

c. Adds Chapter 5, which describes methods for predicting n values for the Manning equation.

d. Updates the Table of Contents to reflect the changes in Chapters 2 and 3 and the addition of
Chapter 5.

e. Updates the preceding chapters to reflect the addition of Chapter 5.

f. Adds references in Chapters 3 and 5 to Appendix A.

g. Adds updated plates in Chapter 3 to Appendix B.

h. Inserts page F-18, which was inadvertently omitted.

i. Updates Appendix H.

j. Adds symbols in Chapter 5 to Appendix I.

2. Substitute the attached pages as shown below:

Chapter Remove page Insert page

Table of Contents i and ii i and ii

2 2-1 and 2-2 2-1 and 2-2
3 3-1 thru 3-10 3-1 thru 3-12
5 — 5-1 thru 5-16

Appendix A A-1 thru A-7 A-1 thru A-8

Appendix B B-35 thru B-60 B-35 thru B-61
Appendix F F-17 F-17 and F-18
Appendix H H-1 and H-2 H-1 and H-2
Appendix I I-1 thru I-4 I-1 thru I-4
 EM 1110-2-1601
Change 1
30 Jun 94

3. File this change sheet in front of the publication for reference purposes.

FOR THE COMMANDER:

2
 EM 1110-2-1601
1 July 1991

US Army Corps
of Engineers

ENGINEERING AND DESIGN

Hydraulic Design of Flood

Control Channels

ENGINEER MANUAL
 DEPARTMENT OF THE ARMY EM 1110-2-1601
U.S. Army Corps of Engineers
CECW-EH-D Washington, D.C. 20314-1000

Engineer Manual
No. 1110-2-1601 1 July 1991

Engineering and Design

HYDRAULIC DESIGN OF FLOOD CONTROL CHANNELS

1. Purpose. This manual presents procedures for the design analysis and criteria of design for improved
channels that carry rapid and/or tranquil flows.

2. Applicability. This manual applies to major subordinate commands, districts, and laboratories having
responsibility for the design of civil works projects.

3. General. Procedures recommended herein are considered appropriate for design of features which are
usable under most field conditions encountered in Corps of Engineers projects. Basic theory is presented
as required to clarify presentation and where the state of the art, as found in standard textbooks, is limited.
In the design guidance, where possible, both laboratory and prototype experimental test results have been
correlated with current theory.

FOR THE COMMANDER:

ROBERT L. HERNDON
Colonel, Corps of Engineers
Chief of Staff

________________________________________
This manual supersedes EM 1110-2-1601, 1 July 1970
 DEPARTMENT OF THE ARMY EM 1110-2-1601
U.S. Army Corps of Engineers Change 1
CECW-EH-D Washington, DC 20314-1000

Engineer Manual
No. 1110-2-1601 30 June 1994

Engineering and Design

HYDRAULIC DESIGN OF FLOOD CONTROL CHANNELS

Table of Contents

Subject Paragraph Page Subject Paragraph Page

Chapter 1 Section III

Introduction Design Guidance for Stone Size
Purpose . . . . . . . . . . . . . . . . . . . . . . 1-1 1-1 General . . . . . . . . . . . . . . . . . . . . . . . . . 3-5 3-4
Scope . . . . . . . . . . . . . . . . . . . . . . . . 1-2 1-1 * Design Conditions . . . . . . . . . . . . . . . . . 3-6 3-5*
References . . . . . . . . . . . . . . . . . . . . 1-3 1-1 Stone Size . . . . . . . . . . . . . . . . . . . . . . . 3-7 3-5
Explanation of Terms . . . . . . . . . . . . . 1-4 1-1 * Revetment Top and End
Channel Classification . . . . . . . . . . . . 1-5 1-1 Protection . . . . . . . . . . . . . . . . . . . . . . 3-8 3-8*
Preliminary Investigations for Selection
of Type of Improvement . . . . . . . . . . 1-6 1-1 Section IV
Revetment Toe Scour Estimation and Protection
Chapter 2 * General . . . . . . . . . . . . . . . . . . . . . . . . 3-9 3-9*
Open Channel Hydraulic Theory * Revetment Toe Protection
Physical Hydraulic Methods . . . . . . . . . . . . . . . . . . . . . . 3-10 3-10*
Elements . . . . . . . . . . . . . . . . . . . . . 2-1 2-1 * Revetment Toe Protection
Hydraulic Design Design . . . . . . . . . . . . . . . . . . . . . . . . 3-11 3-10*
Aspects . . . . . . . . . . . . . . . . . . . . . . 2-2 2-1 * Delivery and Placement . . . . . . . . . . . . 3-12 3-11*
Flow Through Bridges . . . . . . . . . . . . 2-3 2-5
Transitions. . . . . . . . . . . . . . . . . . . . . 2-4 2-8 Section V
Flow in Curved Channels . . . . . . . . . . 2-5 2-12 Ice, Debris, and Vegetation
Special Considerations . . . . . . . . . . . . 2-6 2-14 * Ice and Debris . . . . . . . . . . . . . . . . . . . 3-13 3-11*
Stable Channels . . . . . . . . . . . . . . . . . 2-7 2-15 * Vegetation . . . . . . . . . . . . . . . . . . . . . 3-14 3-12*

Chapter 3 Section VI
Riprap Protection Quality Control
* Quality Control . . . . . . . . . . . . . . . . . . 3-15 3-12*
Section I
Introduction Chapter 4
General. . . . . . . . . . . . . . . . . . . . . . . 3-1 3-1 Special Features and Considerations
Riprap Characteristics . . . . . . . . . . . . . 3-2 3-1 Sediment Control Structures . . . . . . . . . . 4-1 4-1
Air Entrainment . . . . . . . . . . . . . . . . . . . 4-2 4-3
Section II Hydraulic Jump in Open
Channel Characteristics Channels . . . . . . . . . . . . . . . . . . . . . . . 4-3 4-3
Side Slope Inclination . . . . . . . . . . . . 3-3 3-4 Open Channel Junctions . . . . . . . . . . . . . 4-4 4-5
Channel Roughness, Shape, Hydraulic Model Studies . . . . . . . . . . . . 4-5 4-8
Alignment, and Gradient. . . . . . . . . . . 3-4 3-4

i
 EM 1110-2-1601
Change 1
30 Jun 94

Subject Paragraph Page Subject Paragraph Page

* Chapter 5 Appendix D
Methods for Predicting n Values Computer Program for Designing Banked Curves
for the Manning Equation for Supercritical Flow in Rectangular Channels
Introduction . . . . . . . . . . . . . . . . . . . 5-1 5-1
Approach . . . . . . . . . . . . . . . . . . . . . 5-2 5-1 Appendix E
Hydraulic Roughness by Theory of Combining Flow at Open Channel
Handbook Methods . . . . . . . . . . . . . 5-3 5-1 Junctions (Confluences)
Base n Values (nb) for
Channels . . . . . . . . . . . . . . . . . . . . 5-4 5-1 Appendix F
Hydraulic Roughness by Report on Standardization of Riprap Gradations
Analytical Methods . . . . . . . . . . . . . 5-5 5-2
Composite n Values and Appendix G
Hydraulic Radius . . . . . . . . . . . . . . . 5-6 5-12 Velocity Estimation Based on Field Observations
Expansion and Contraction
in a 1-D Model . . . . . . . . . . . . . . . . 5-7 5-14 Appendix H
Unforeseen Factors . . . . . . . . . . . . . . 5-8 5-14 * Examples of Stone Size Calculations

Appendix A Appendix I
References Notation

Appendix B
Plates

Appendix C
Notes on Derivation and Use of Hydraulic
Properties by the Alpha Method

ii
 EM 1110-2-1601
1 Jul 91

Chapter 1 velocities with Froude numbers greater than 1 (F > 1),

Introduction and tranquil flows have subcritical velocities with Froude
numbers less than 1 (F < 1).

b. Slopes. Invert slopes in general are greater than

1-1. Purpose critical slopes (So > Sc) for rapid flow and less than criti-
cal slopes (So < Sc) for tranquil flow.
This manual presents procedures for the design analysis
and criteria of design for improved channels that carry c. Channel storage. Channel storage is usually
rapid and/or tranquil flows. negligible in rapid flow, whereas it may be appreciable in
natural rivers with tranquil flow.
1-2. Scope
d. Discharge. All discharges are normally confined
Procedures are presented without details of the theory of within the channel for rapid flow (no overbank flow).
the hydraulics involved since these details can be found in
any of various hydraulic textbooks and publications avail- Other characteristics such as standing waves, surges, and
able to the design engineer. Theories and procedures in bed configuration that differ under the influence of rapid-
design, such as flow in curved channels, flow at bridge or tranquil-flow conditions should be recognized and
piers, flow at confluences, and side drainage inlet struc- considered as the occasion demands. Rapid and tranquil
tures, that are not covered fully in textbooks are discussed flows can occur within a longitudinal reach of a channel
in detail with the aid of Hydraulic Design Criteria (HDC) with changes in discharge, roughness, cross section, or
charts published by the US Army Engineer Waterways slope. Channel improvements may bring about changes in
Experiment Station (USAEWES). The charts and other flow characteristics.
illustrations are included in Appendix B to aid the desig-
ner. References to HDC are by HDC chart number. The 1-6. Preliminary Investigations for Selection of
use of models to develop and verify design details is Type of Improvement
discussed briefly. Typical calculations are presented to
illustrate the principles of design for channels under vari- The investigation required in selecting the type of channel
ous conditions of flow. Electronic computer program- improvement to be adopted involves three considerations:
ming techniques are not treated in this manual. However, physical features of the area, hydraulic and hydrologic
most of the basic hydraulics presented herein can be aspects, and economy.
adapted for computer use as illustrated in Appendix D.
a. Physical features. The topography of the area
1-3. References controls in a general way the channel alignment and
invert grades. Of prime importance, also, are width of
References are listed in Appendix A. available right-of-way; location of existing channel; and
adjacent existing structures, such as bridges, buildings,
1-4. Explanation of Terms transportation facilities, utility structures, and outlets for
local drainage and tributaries. Invert slopes may be con-
Abbreviations used in this manual are explained in the trolled by elevations of existing structures as well as by
Notation (Appendix I). The symbols employed herein general topography, elevations at ends of improvements,
conform to the American Standard Letter Symbols for and hydraulic features.
Hydraulics (American Society of Mechanical Engineers
1958) with only minor exceptions. b. Historical and observed elements. The flow char-
acteristics noted in historical records and indicated from
1-5. Channel Classification detailed observation of existing conditions will usually be
basic to the selection of type of improvement or design.
In this manual, flood control channels are considered With the flood discharges determined, the interdependent
under two broad classifications: rapid- and tranquil-flow factors that determine improvement methods and general
channels. The most important characteristics that apply to channel alignment are slope of invert, width and depth of
rapid and tranquil flows are listed below: flow, roughness coefficient, the presence or nature of
aggradation and degradation processes, debris transporta-
a. Velocities. Rapid flows have supercritical tion, bank erosion, cutoffs, and bar formations.

1-1
EM 1110-2-1601
1 Jul 91

c. Preliminary layout. A preliminary map or aerial of the proposed channels are then made with a view
mosaic of the area showing the topography and other toward establishing the most efficient channel improve-
control factors to a scale satisfactory for plotting the cen- ment from the standpoint of hydraulic efficiency and
ter line of the channel should be obtained. A scale of economic feasibility.
1 inch (in.) to 100 feet (ft) with 2-ft-contour interval is
suggested, although judgment based on local conditions e. Economy. Approximate cost estimates are pre-
should be used. A preliminary profile should be prepared pared, including costs of channel construction, appurtenant
that will show all pertinent elevations of the ground and works and bridges, and rights-of-way. It may be neces-
existing structures along the banks and along the center sary to consider several channel alignments, cross sec-
line of the proposed channel. tions, and construction materials before the least-cost
design consistent with sound engineering principles is
d. Preliminary alternative designs. From a study of determined. Assured performance, consistent with project
the preliminary plan, profiles, and available widths, tenta- formulation based on sound engineering judgment, is a
tive channel cross sections are adopted. These are gener- necessary part of economic consideration. With an
ally rectangular or trapezoidal sections. Low velocity optimum general design thus tentatively established, and
flows can usually be carried in natural-bottom trapezoidal provided the cost is economically feasible for the project
channels with or without stone-revetted side slopes. as a whole, the detailed hydraulic design is presented in
High- velocity flows normally would be carried in Chapter 2.
concrete-lined channels. Preliminary hydraulic analyses

1-2
 EM 1110-2-1601
Change 1
30 Jun 94

Chapter 2 when the width of the right-of-way is limited. A small,

Open Channel Hydraulic Theory low-flow channel in the center of the cross section pro-
vides a flow way into which subdrainage can be emptied.
In cold climates, the low-flow channel reduces the invert
area subjected to the deleterious effects of freezing and
2-1. Physical Hydraulic Elements thawing. In some cases the low-flow channel may serve
as a fishway.
a. General. The physical hydraulic elements con-
cerned in hydraulic design of channels consist of invert c. Roughness. The concept of surface roughness as
slope (So), cross-sectional area (A), wetted perimeter (P), the basic parameter in flow resistance (friction) is almost
and equivalent boundary surface roughness (k). The universally accepted. Absolute roughness is determined
hydraulic radius (R) used in resistance formulae is the from the physical dimensions of the wetted surface irregu-
ratio A/P . The invert slope of proposed channel im- larities and is normally of theoretical interest only.
provement is controlled primarily by elevations of the Equivalent roughness is a linear dimension (effective
ground along the alignment as determined by preliminary roughness height) directly related to the boundary resis-
layout discussed in paragraph 1-6d. A center-line profile tance of the channel (Plate 3). The relations between
between controlling elevations along the proposed align- roughness and the various coefficients for friction formu-
ment will indicate a preliminary channel slope. lae are adequately covered by Chow (1959, chap 8).
* Friction formulae and their uses are discussed in para-
b. Channel cross section. graph 2-2, and methods for predicting Manning’s rough-
ness coefficient n are discussed in Chapter 5. *
(1) The proper channel cross section for a given
reach is the one that has adequate hydraulic capacity for a d. Composite roughness. Where there is material
minimum cost of construction and maintenance. The variation in roughness between various portions of the
economics must include the costs of right-of-way and wetted perimeter such as might be found in natural chan-
structures such as bridges. In rural areas a trapezoidal nels or channels with protected banks and natural inverts,
cross section may be least costly, whereas in urban areas an equivalent or effective roughness or friction coefficient
a rectangular cross section is often the least costly. for each stage considered should be determined. Appen-
dix C illustrates a method for determining a composite
(2) Plate 11 shows a sample cost computation and value of k for each stage. Plates 4 and 5 give the rela-
related cost curve for a reach of curved rectangular con- tion between k and Manning’s n for flows in the rough
crete channel. Similar procedures may be applied to flow zone shown in Plate 3. HDC sheets 631-4 and
compute the cost for any type of cross section considered 631-4/1 also give a procedure for determining an effective
for design. Special types of concrete channel cross sec- Manning’s n .
tions are shown in Plate 2: the V-bottom channel and the
modified trapezoidal channel. The latter has a small e. Hydraulic efficiency. The problem of the most
low-flow channel in the center. efficient cross section is treated by Brater and King
(1976, see pp 7-5 to 7-7) and Chow (1959, see
(a) In the V-bottom channel, low flows are concen- paragraph 7-6).
trated along the channel center line. This prevents moder-
ate flow from meandering over the entire channel width, 2-2. Hydraulic Design Aspects
which would result in random deposition of material
across the invert as in the case of a horizontal bottom. a. General. This presentation assumes that the
Deposition in the center of the V-bottom is removed by design engineer is fully acquainted with the hydraulic
larger flows. Because the wear caused by bed load is theories involved in uniform and gradually varied flows,
also concentrated near the center line, maintenance cost is steady and unsteady flows, energy and momentum princi-
reduced. ples, and other aspects such as friction related to
hydraulic design normally covered in hydraulic texts and
(b) In the modified trapezoidal cross section, vertical handbooks such as those by Brater and King (1976) and
sidewalls reduce the top width. This design is desirable Chow (1959). The following is presented as guidance in
______________________ the method of application of textbook material and to give
1
Plates mentioned in this and succeeding chapters are additional information not readily available in reference
included in Appendix B as Plates B-1, B-2, etc.

2-1
EM 1110-2-1601 *
1 Jul 91

material. The use of k is emphasized herein because The relation between the coefficients in these equations
computational results are relatively insensitive to errors in can be expressed as
assigned values of k . However, use of Manning’s n
has been retained in several procedures because of its C R 1/6 10.8 (2-4)
wide acceptance and simplicity of use. This applies 1.486 n f 1/2
particularly to varied flow profiles, pulsating flow, and the
design of free-surface hydraulic models.
(2) When determining friction coefficients, it should
b. Friction losses. be recognized that the energy grade line and therefore the
friction coefficient include uniformly occurring turbulence
(1) The importance that friction plays in the determi- and eddy losses as well as the friction loss. Equivalent
nation of flow characteristics in channels cannot be over- roughness for the same reason. Special, locally occurring
stressed. Three equations (Chezy’s, Manning’s, and turbulence and eddy losses are to be determined sepa-
Darcy’s) are in general use for the determination of losses rately as covered in hydraulic textbooks and elsewhere in
due to friction. These equations expressed as friction this manual.
slope Sf , i.e., slope of the energy grade line, are
c. Friction coefficients.
(a) Chezy:
(1) The equations for using equivalent roughness to
determine friction coefficients (Plate 3) are
V2 (2-1)
Sf
C 2R (a) For hydraulically smooth channels

(b) Manning:  
 5.2Rn  (2-5)
C 32.6 log10  
 C 
V 2n 2 (2-2)
Sf
2.21R 4/3
(b) For hydraulically rough channels

(c) Darcy:
 12.2R  (2-6)
C 32.6 log10  
 k 
fV 2 (2-3)
Sf
8Rg
where Rn is the Reynolds number.

where (2) For the channel surface to be hydraulically

smooth, the equivalent roughness must be less than the
V = velocity critical value given by paragraph 8-12 of Chow (1959).

C = Chezy coefficient
 5C  ν
kc     (2-7)
f = Darcy-Weisbach resistance   V
coefficient  g 

g = acceleration of gravity
where ν is the kinematic viscosity of water.

2-2 *
 EM 1110-2-1601
1 Jul 91

(3) Most channels (including concrete-lined chan- d. Flow classification. There are several different
nels) with appreciable velocity are hydraulically rough. types of flow classification. Those treated in this para-
Plates 4 and 5 are furnished as an aid for determining graph assume that the channel has a uniform cross-
friction coefficients from equivalent roughness. Irrigation sectional rigid boundary. The concepts of tranquil and
and power canals generally fall in the transition zone rapid flows are discussed in (1) below. The applicability
shown in Plate 3. of the newer concepts of steady rapid flow and pulsating
rapid flow to design problems are treated in (2) below.
(4) Table 2-1, extracted from HDC sheets 631 to All of these concepts are considered from the viewpoint
631-2, provides acceptable equivalent roughness values of uniform flow where the water-surface slope and energy
for straight, concrete-lined channels. grade line are parallel to the bottom slope. Flow classifi-
cation of nonuniform flow in channels of uniform solid
(5) See Chapter 3 for friction coefficients for riprap. boundaries or prismatic channels is discussed in (3)
below. The design approaches to flow in nonprismatic
(6) Values of k for natural river channels usually channels are treated in other portions of this manual.
fall between 0.1 and 3.0 ft (see Table 8-1 of Chow
(1) Tranquil and rapid flows.

(a) The distinction between tranquil flow and rapid

Table 2-1 flow involves critical depth. The concept of specific
Acceptable Equivalent energy He can be used to define critical depth. Specific
Roughness Values
energy is defined by
Design Problem k , ft

V2
Discharge Capacity 0.007 He d α (2-8)
2g
Maximum Velocity 0.002

Proximity to Critical
Depth1 where

Tranquil Flow 0.002 d = depth

Rapid Flow 0.007
α = energy correction factor
Note:
1. To prevent undesirable undulating waves, ratios of flow depth V2/2g = velocity head
to critical depth between 0.9 and 1.1 should be avoided where
economically feasible. Plate 6 shows a specific energy graph for a discharge q
of 100 cubic feet per second (cfs) (two-dimensional
1959). These values will normally be much larger than flows). Each unit discharge has its own critical depth:
the spherical diameters of the bed materials to account for
boundary irregularities and sand waves. When friction
coefficients can be determined from experienced flow  2 1/3
information, k values should then be computed using the q  (2-9)
dc  
relations described in Equation 2-6. The k values so g
determined apply to the surfaces wetted by the experi-
enced flows. Additional wetted surfaces at higher stages
should be assigned assumed k values and an effective The development of this equation is given by pp 8-8 and
roughness coefficient computed by the method outlined in 8-9 of Brater and King (1976). It may be noted that the
Appendix C if the increased wetted surfaces are estimated critical depth occurs when the specific energy is at a
to be appreciably smoother or rougher. Values of k for minimum. Flow at a depth less than critical (d < dc) will
natural channels may also be estimated from Figures 8 have velocities greater than critical (V > Vc), and the flow
and 9 of Chow (1959) if experimental data are not is described as rapid. Conversely, when d > dc and V
available. < Vc , the flow is tranquil.

2-3
EM 1110-2-1601
1 Jul 91

(b) It may be noted in Plate 6 that in the proximity

of critical depth, a relatively large change of depth may  2/3
0.0463R 1/6 F (2-10)
occur with a very small variation of specific energy. 4.04 log10 F 
n  s
Flow in this region is unstable and excessive wave action
or undulations of the water surface may occur.
Experiments by the US Army Engineer District (USAED), The limiting Froude number Fs for use in this equation
Los Angeles (1949), on a rectangular channel established was derived by Escoffier and Boyd (1962) and is given
criteria to avoid such instability, as follows: by

Tranquil flow: d > 1.1dc or F < 0.86

ξ
Fs (2-11)
Rapid flow: d < 0.9dc or F > 1.13 g ζ3/2 (1 Zζ)

where F is the flow Froude number. The Los Angeles

District model indicated prototype waves of appreciable where ξ , the flow function, is given by
height occur in the unstable range. However, there may
be special cases where it would be more economical to
Q
provide sufficient wall height to confine the waves rather ξ
than modify the bottom slope. b 5/2

(c) Flow conditions resulting with Froude numbers

near 1.0 have been studied by Boussinesq and Fawer. where Q is the total discharge and ζ , the depth-width
The results of their studies pertaining to wave height with ratio, is given by
unstable flow have been summarized by Jaeger (1957,
pp 127-131), including an expression for approximating
d
the wave height. The subject is treated in more detail in ζ
paragraph 4-3d below. Determination of the critical depth b
instability region involves the proper selection of high and
low resistance coefficients. This is demonstrated by the
example shown in Plate 6 in which the depths are taken where b is the bottom width.
as normal depths and the hydraulic radii are equal to
depths. Using the suggested equivalent roughness design Plate 7 shows the curves for a rectangular channel and
values of k = 0.007 ft and k = 0.002 ft , bottom slope trapezoidal channels with side slopes Z of 1, 2, and 3.
values of So = 0.00179 and So = 0.00143 , respectively,
are required at critical depth. For the criteria to avoid the (3) Varied flow profiles. The flow profiles discussed
region of instability (0.9dc < d < 1.1dc), use of the smaller herein relate to prismatic channels or uniform cross sec-
k value for tranquil flow with the bottom slope adjusted tion of boundary. A complete classification includes
so that d > 1.1dc will obviate increased wall heights for bottom slopes that are horizontal, less than critical, equal
wave action. For rapid flow, use of the larger k value to critical, greater than critical, and adverse. However,
with the bottom slope adjusted so that d < 0.9dc will the problems commonly encountered in design are mild
obviate increased wall heights should the actual surface be slopes that are less than critical slope and steep slopes
smoother. Thus, the importance of equivalent roughness that are greater than critical slope. The three types of
and slope relative to stable flow is emphasized. These profiles in each of these two classes are illustrated in
stability criteria should be observed in both uniform and HDC 010-1. Chow (1959) gives a well-documented
nonuniform flow design. discussion of all classes of varied flow profiles. It should
be noted that tranquil-flow profiles are computed proceed-
(2) Pulsating rapid flow. Another type of flow ing upstream and rapid-flow profiles downstream. Flow
instability occurs at Froude numbers substantially greater profiles computed in the wrong direction result in diver-
than 1. This type of flow is characterized by the gences from the correct profile. Varied-flow computa-
formation of slugs particularly noticeable on steep slopes tions used for general design should not pass through
with shallow flow depth. A Manning’s n for pulsating critical depth. Design procedures fall into two basic cate-
rapid flow can be computed from gories: uniform and nonuniform or varied flow. Many

2-4
 EM 1110-2-1601
1 Jul 91

graphs and tables have been published to facilitate compu- channel where both bottom width and side slope vary.
tation of uniform flow. Brater and King (1976) have Successive approximations of water-surface elevations are
specially prepared tables for trapezoidal channels based on made until a balance of energy is obtained. Friction
the Manning equation. HDC 610-1 through 610-4/1-1 losses hf are based on the Manning equation.
give graphs that afford rapid solution for the normal depth
in trapezoid channels. Nonuniform or varied flow in
prismatic channels can be solved rapidly by use of the n 2V 2 V2 (2-1 and 2-2 bis)
Sf
varied flow function. (It should be noted that different 2.21R 4/3 C 2R
authors have used the terms "nonuniform" flow and "var-
ied" flow to mean the same thing; "varied flow" is used in
this manual.) Varied flow in nonprismatic channels, such
as those with a gradually contracting or a gradually ex- For the sample computation a mild slope upstream and
panding cross section, is usually handled by "step meth- steep slope downstream of sta 682+40 have been
ods." It should be noted that short, rapidly contracting or assumed. Critical depth would occur in the vicinity of
expanding cross sections are treated in this manual as sta 682+40 and has been assumed as the starting condi-
transitions. tion. Initially, column 21 has the same value as column
10. The computations proceed downstream as the flow is
(a) Prismatic channels. A prismatic channel is char- rapid. The length of reach is chosen such that the change
acterized by unvarying cross section, constant bottom in velocity between the ends of the reach is less than
slope, and relatively straight alignment. There are three 10 percent. The energy equation is balanced when
general methods of determining flow profiles in this type column 21 checks column 10 for the trial water surface of
of channel: direct integration, direct step, and standard column 5. Plate 9 repeats the computation, substituting
step, as discussed in Chow (1959, pp 252-268). The k = 0.002 ft for n = 0.014 . For rough channel
direct integration and direct step methods apply exclu- conditions
sively to prismatic channels, whereas the standard step
method applies not only to prismatic channels but is the
only method to be applied to nonprismatic channels. The  12.2R  (2-6 bis)
C 32.6 log10  
direct integration method (with certain restrictions as to  k 
the constancy of hydraulic exponents) solves the varied
flow equation to determine the length of reach between
successive depths. Use is made of varied-flow-function 2-3. Flow Through Bridges
tables to reduce the amount of computations. This
method is not normally employed unless sufficient Bridge piers located in channels result in energy losses in
profiles and length of channel are involved to warrant the the flow and create disturbances at the bridge section and
amount of precomputational preparation. The direct step in the channel sections immediately upstream and down-
method determines the length of reach between successive stream. As bridge pier losses materially affect water-
depths by solution of the energy and friction equations surface elevations in the vicinity of the bridge, their
written for end sections of the reach. The standard step careful determination is important. Submergence of
method is discussed in (b) below. bridge members is not desirable.

(b) Nonprismatic channels. When the cross section, a. Abutment losses. Bridge abutments should not
alignment, and/or bottom slope changes along the channel, extend into the flow area in rapid-flow channels. In
the standard step method (Chow 1959, p 265) is applied. tranquil-flow channels they should be so designed that the
This method determines the water-surface elevation flow depth between abutments or between the abutment
(depth) at the reach extremity by successive approxima- and an intermediate pier is greater than critical depth.
tions. Trial water-surface elevations are assumed until an The Bureau of Public Roads (BPR) (Bradley 1978) has
elevation is found that satisfies the energy and friction published design charts for computing backwater for
equations written for the end sections of the reach. Cross various abutment geometries and degrees of contraction.
sections for this method should, in general, be selected so The design procedure and charts developed by BPR are
that velocities are increasing or decreasing continuously recommended for use in channel designs involving bridge
throughout the reach. EM 1110-2-1409 contains further abutments. For preliminary designs, a step backwater
information on this method. Plate 8 shows a sample computation using abrupt expansion and contraction head
computation for a gradually contracting trapezoidal losses of 1.0 and 0.5, respectively, times the change in

2-5
EM 1110-2-1601
1 Jul 91

velocity head may be used. This method under the same V = average channel velocity, feet per
circumstances may be applied to bridge openings contain- second (fps)
ing piers.
g = acceleration of gravity, ft/sec2
b. Pier losses. Rapid, tranquil, or a combination of
rapid- and tranquil-flow conditions may occur where only In Equation 2-12 β is generally assumed to be equal to
bridge piers are located in the flow area. Flow through 1.0. Since
bridge piers for this condition is classified as class A, B,
or C, according to the depth of flow in relation to critical
Q AV (2-13)
depth occurring upstream, between piers, and downstream.
Plate 10 is a graphic description of these classes, which
are discussed below. Plate 11 is useful in determining the
class of flow in rectangular channels. Equation 12 can be written

(1) Class A flow (energy method). Chow (1959,

paragraph 17-10) presents a discussion and several energy γQ 2 (2-14)
M
loss formulae with appropriate coefficients that may be gA
used for computing bridge pier losses for tranquil flow
(class A). While the momentum method presented below
may also be used for class A flow, the energy method (c) The total hydrostatic force m (in pounds) in the
usually gives better results. channel section can be expressed as

(2) Classes B and C flows (momentum method).

m γyA (2-15)
(a) A graph (example shown in Plate 12) constructed
from the equation proposed by Koch and Carstanjen
(Chow 1959) and based on the momentum relation can be _
where y is the distance from the water surface to the
used for determining graphically the flow classification at
center of gravity (centroid) of the flow section.
constrictions due to bridge piers. In addition, the graph
can be used for estimating unknown flow depths. A
(d) Combining Equations 14 and 15 results in
summary of the equation derivation follows.

(b) In a given channel section the momentum per γQ 2

unit time of the flow can be expressed by m M γyA (2-16)
gA

 γQV 
M β   (2-12)
By the momentum principle in an unconstricted channel
 g 

γQ 2 γQ 2 (2-17)
where ma mb
gAa gAb
M = momentum per unit time, pounds (lb)
(from pounds-second per second
where ma and mb are the total hydrostatic forces of
(lb-sec/sec))
water in the upstream and downstream sections,
respectively, lb.
β = momentum correction coefficient
(e) Based on experiments under all conditions of
γ = specific weight of water, pounds per
open-channel flow where the channel was constricted by
cubic foot (pcf)
short, flat surfaces perpendicular to the flow, such as with
Q = total discharge, cfs

2-6
 EM 1110-2-1601
1 Jul 91

bridge piers, Koch and Carstanjen (Koch 1926) found that Q = total discharge, cfs
the upstream momentum force had to be reduced by
(Ap/A1)(γQ2/gA1) to balance the total force in the m1 = total hydrostatic force of water in
constriction. upstream section, lb

(f) Equating the summation of the external forces m2 = total hydrostatic force of water in
above and below the structures with those within the con- pier section, lb
tracted section yields
m3 = total hydrostatic force of water in
downstream section, lb

 A   2 
γQ 2   p  γQ  mp = total hydrostatic force of water on
m1  A   gA  pier ends, lb
gA1   1  1 
(2-18)
A1 = cross-sectional area of upstream
γQ 2 channel, square feet, ft2
m2 mp
gA2
A2 = cross-sectional area of channel
within pier section, ft2
and
A3 = cross-sectional area of downstream
channel, ft2
γQ 2 γQ 2 (2-19)
m2 mp m3
gA2 gA3 Ap = cross-sectional area of pier
obstruction, ft2

Combining these equations results in (g) Curves based on the Koch-Carstanjen equation
(Equation 2-21) are illustrated in Plate 12a. The resulting
flow profiles are shown in Plate 12b. The necessary
computations for developing the curves are shown in
 A   2 
γQ 2   p  γQ   Plate 13. The downstream depth is usually known for
m1  A   gA   m2 tranquil-flow channels and is greater than critical depth.
gA1   1  1  (2-20) It therefore plots on the upper branch of curve III in
γQ 2
γQ 2
Plate 12a. If this depth A is to the right of (greater force
mp m3
gA2 gA3 than) the minimum force value B of curve II, the flow is
class A and the upstream design depth C is read on curve
I immediately above point A. In this case, the upstream
depth is controlled by the downstream depth A plus the
This reduces to the Koch-Carstanjen equation pier contraction and expansion losses. However, if the
downstream depth D plots on the upper branch of curve
III to the left of (less force than) point B, the upstream
γQ 2 γQ 2 design depth E is that of curve I immediately above point
m1 mp A1 Ap m2
2 gA2
gA1 (2-21) B, and critical depth within the pier section B is the con-
γQ 2 trol. The downstream design depth F now is that given
m3 mp
gA3 by curve III immediately below point E. A varied flow
computation in a downstream direction is required to
determine the location where downstream channel condi-
tions effect the depth D.

where (h) In rapid-flow channels, the flow depth upstream

of any pier effect is usually known. This depth is less
γ = specific weight of water, pounds than critical depth and therefore plots on the lower branch
per cubic foot (pcf) of curve I. If this depth G is located on curve I to the

2-7
EM 1110-2-1601
1 Jul 91

right of point B, the flow is class C. The downstream extensions to accomplish this is illustrated in USAED,
design depth H and the design flow depth within the pier Los Angeles (1943), and USAEWES (1957). The general
section I are read on curves III and II, respectively, im- statements relative to bridge pier extensions for class B
mediately above depth G. A varied flow computation is flow also apply to class C flow. However, in the latter
required to determine the location where downstream case, the degree and extent of the disturbances are much
channel conditions again control the depth. However, if more severe than with class B flow. Excellent illustra-
the unaffected upstream rapid-flow depth J plots on the tions of the use of bridge pier extensions in high-velocity
lower branch of curve I to the left of point B, the design channels are given in USAED, Los Angeles (1943), and
upstream depth K is read on curve I immediately above USAED, Walla Walla (1960). The bridge pier extension
point B. The design downstream depth L is read on curve geometry shown in Plate 18 was developed by USAED,
II immediately below point B. In this case, class B flow Los Angeles, and pier extensions of this design have been
results with a hydraulic jump between depths J and K found to perform satisfactorily.
(Plate 12b) upstream of the pier as controlled by critical
depth within the pier section B. A varied flow computa- d. Model studies. Where flow conditions at bridge
tion is again required to determine the location where piers are affected by severe changes in channel geometry
downstream channel conditions control the flow depth. and alignment, bridge abutments, or multiple bridge
crossings, consideration should be given to obtaining the
(3) Design charts, rectangular sections. A graphic design flow profile from a hydraulic model study.
solution for class A flow in rectangular channels,
developed by USAED, Los Angeles (1939), and published 2-4. Transitions
as HDC 010-6/2, is reproduced in Plate 14. The drop in
water surface H3 in terms of critical depth is presented a. General. Transitions should be designed to ac-
as a function of the downstream depth d3 and critical complish the necessary change in cross section with as
depth in the unobstructed channel. Separate curves are little flow disturbance as is consistent with economy. In
given for channel contraction ratios of between 0.02 and tranquil flow, the head loss produced by the transition is
0.30. In rectangular channels, α is the horizontal con- most important as it is reflected as increased upstream
traction ratio. The basic graph is for round nose piers. stages. In rapid flow, standing waves produced by
The insert graph provides correction factors (γ) for other changes of direction are of great concern in and down-
pier shapes. Use of the chart is illustrated in Plate 15. stream from the transition. Streamlined transitions reduce
Plate 16 (HDC 010- 6/3) presents the USAED, Los head losses and standing waves. As transition
Angeles, (1939), solution for class B flow using the mo- construction costs exceed those of uniform channel cross
mentum method. Plate 17 (HDC 010-6/4) presents the section and tend to increase with the degree of streamlin-
USAED, Chicago, solution for class B flow by the energy ing, alternative transition designs, their costs, and the
method. The use of these charts for rectangular channel incremental channel costs due to head losses and/or stand-
sections is shown in Plate 15. ing waves should be assessed.

c. Bridge pier extension. Upstream pier extensions b. Types. The three most common types of transi-
are frequently used to reduce flow disturbance caused by tions connecting trapezoidal and rectangular channels are
bridge piers and to minimize collection of debris on pier cylindrical quadrant, warped, and wedge, as shown in
noses. In addition, it is often necessary and economical Plate 19. For comparable design, the wedge-type transi-
to make use of existing bridge structures in designing tion, although easier to construct, should be longer than
flood channels. In some instances there is insufficient the warped because of the miter bends between channel
vertical clearance under these structures to accommodate and transition faces. Warped and wedge types can be
the design flow. With class B flow, the maximum flow used generally for expansions or contractions.
depth occurs at the upstream end of the pier and the criti-
cal depth occurs within the constriction. Field observa- (1) Tranquil flow. Each of these three transition
tions and model studies by USAED, Los Angeles (1939), types may be used for tranquil flow in either direction.
indicate that the minimum depth within the constricted The cylindrical quadrant is used for expansions from
area usually occurs 15 to 25 ft downstream from the rectangular to trapezoidal section and for contractions
upstream end of the pier. Pier extensions are used to from trapezoidal to rectangular section. An abrupt or
effect an upstream movement of the control section, straight-line transition as well as the quadrant transition
which results in a depth reduction in the flow as it enters can be used in rectangular channels.
the constricted pier section. The use of bridge pier

2-8
 EM 1110-2-1601
1 Jul 91

(2) Rapid flow. The cylindrical quadrant is used for

transitions from tranquil flow in a trapezoidal section to tan β1 1 8F1 sin2 β1
2
3
rapid flow in a rectangular section. The straight-line tan θ (2-22)
2 tan β1 8F sin β1
2 2 2
transition is used for rectangular sections with rapid flow. 1 1 1
Specially designed curved expansions (c(2)(b) below) are
required for rapid flow in rectangular channels.

c. Design. y2 1
8F1 sin2 β1
2
1 1 (2-23)
y1 2
(1) Tranquil flow. Plate 20 gives dimensions of
plane surface (wedge type) transitions from rectangular to
trapezoidal cross section having side slopes of 1 on 2; 1 and
on 2.5, and 1 on 3. In accordance with the recommenda-
tions of Winkel (1951) the maximum change in flow line
has been limited to 6.0 degrees (deg). Water-surface  
y1  2 
1 y1  y2
 y  2 (2-24)
profiles should be determined by step computations with 1  2 
1 
2
F F1
y2  2 y2  y1 y
2
less than 20 percent (less than 10 percent in important   1 
instances) change in velocity between steps. Adjustments
in the transition should be made, if necessary, to obtain a
water-surface profile that is as nearly straight as
practicable. where

(2) Rapid flow. In rapid flow, stationary waves θ = wall deflection angle
result with changes in channel alignment. These distur-
bances may necessitate increased wall height, thereby F = Froude number
appreciably increasing construction costs. USAED, Los
Angeles, uses the criterion in Table 2-2 for the design of β = wave front angle
straight-line or wedge-type transitions to confine flow
disturbances within the normal channel freeboard y = flow depth
allowance:
The subscripts 1, 2, and 3 refer to the flow areas indicated
on the sketches in Plate 21. For straight-line convergence
Table 2-2 (Plate 21b), the maximum flow disturbance results when
Recommended Convergence and Divergence Transition Rates the initial wave front intersection, point B, occurs at the
downstream transition CC’. When the reflected waves BD
Wall flare for each
Mean channel wall (horizontal to and BD’ intersect the channel walls below or above sec-
velocity, fps longitudinal) tion CC’, diamond-shaped cross waves develop in the
channel. However, the change in wall alignment at sec-
10-15 1:10 tion CC’ results in negative wave disturbances that should
tend to decrease the downstream effects of positive wave
15-30 1:15
fronts. This should result in somewhat lower depths
30-40 1:20 where the waves meet the downstream walls. The mini-
mum disturbance occurs when the reflected waves BD and
BD’ meet the channel walls at section CC’. This, theoreti-
(a) Rectangular contractions. Ippen (1950), Ippen cally, results in the flow filaments again becoming parallel
and Dawson (1951), and Ippen and Harleman (1956) ap- to the channel center line. If the reflected waves meet the
plied the wave theory to the design of rectangular channel walls upstream from section CC’, the waves would be
transitions for rapid flow and developed the following deflected again with a resulting increase in depth.
equations for computing flow depths in and downstream Graphic plots of Equations 2-22 through 2-24 have been
from the convergence: published (Ippen 1950, Ippen and Dawson 1951, and

2-9
EM 1110-2-1601
1 Jul 91

Ippen and Harleman 1956). Plate 22 presents design

curves based on these equations. The extent of the curves  3/2
Z 1  X  1 (2-28)
has been limited to flow conditions normally occurring in b1 2  b1F1  2
rapid-flow flood control channels. The required length of
the transition is a function of the wall deflection angle θ
and the channel contraction b1 - b3 , or
where
b1 b3 Z = transverse distance from channel
L (2-25)
2 tan θ center line

b1 = approach channel width

where
X = longitudinal distance from
b1 = upstream channel width, ft beginning of expansion

b3 = downstream channel width, ft F1 = approach flow Froude number

The theory indicates that the surface disturbances are Equation 2-28 is for an infinitely wide expansion. Opti-
minimized when L = L1 + L2 (Plate 21). The equations mum design of expansions for rapid flow necessitates
for L1 and L2 are control of wall curvature so that the negative waves gen-
erated by the upstream convex wall are compensated for
by positive waves formed by the downstream concave
b1 wall. In this manner, the flow is restored to uniformity
L1 (2-26)
2 tan β1 where it enters the downstream channel. A typical design
of a channel expansion is shown in Plate 24b. Plate 24a
reproduces generalized design curves presented in Rouse,
Bhoota, and Hsu (1951). It is to be noted that the convex
and wall curve equation is appreciably less severe than that
indicated by Equation 2-28. Equations for laying out the
transition and a definition sketch are given in Plate 24b.
b3 The data given in Plate 24 should be adequate for prelimi-
L2 (2-27)
2 tan (β2 θ) nary design. In cases where the wave effects are critical,
the design should be model tested. Laboratory experi-
ments based on the generalized curves have indicated that
the downstream channel depths may be appreciably in
The correct transition design for a given change in chan- excess of those indicated by the simple wave theory. The
nel width and Froude number involves selection of a simple wave theory can be applied to the design of
value of θ so that L = L1 + L2 . A computation illus- straight-line transitions. An illustration of the computa-
trating the design procedure is given in Plate 23. tion procedure is given on pages 9-10 through 9-12 of
Brater and King (1976). It is to be noted that this compu-
(b) Rectangular expansions. In channel expansions tation does not include any wave effects reflected from
the changes in flow direction take place gradually in one sidewall to the other. Also, an abrupt positive wave
contrast to the steep wave front associated with contrac- exists where the expanding wall intersects the downstream
tions. In 1951, Rouse, Bhoota, and Hsu (1951) published channel wall. Application of this method of characteris-
the results of a study of expanding jets on a horizontal tics is illustrated on pages 9-12 through 9-16 of Brater
floor. A graphical method of characteristics, described in and King (1976).
Ippen (1951), was used for the theoretical development
of flow depth contours. These results were verified (c) Nonrectangular transitions. The necessary tech-
experimentally. The following equation based on theoreti- niques for applying the wave theory to channel transitions
cal and experimental studies was found to give the most involving both rectangular and trapezoidal sections have
satisfactory boundary shapes for the expansion of a not been developed, and generalized design curves are not
high-velocity jet on a horizontal floor. available. Limited tests on straight-line and warped-wall

2-10
 EM 1110-2-1601
1 Jul 91

channel transitions for trapezoidal to rectangular sections

and for rectangular to trapezoidal sections have been h1 Cc∆hv (2-29)
made at Pennsylvania State University (Blue and Shulits
1964). Tests were limited to three different transition
shapes for Froude numbers of 1.2 to 3.2. Each shape was for contractions and as
tested for five different transition lengths. The trapezoidal
channel invert was 0.75 ft wide. The rectangular channel
was 1.071 ft wide. Generalized design curves were not h1 Ce∆hv (2-30)
developed. However, the study results should be useful
as design guides.
where
(3) Rapid to tranquil flow.
Cc = contraction coefficient
(a) The design of rapid-flow channels may require
the use of transitions effecting flow transformation from Ce = expansion coefficient
rapid to tranquil flow. Such transitions normally involve
channel expansions in which the channel shape changes for expansions. Equations 2-29 and 2-30 have been
from rectangular to trapezoidal. obtained and published (Chow 1959, Brater and King
1976, US Bureau of Reclamation (USBR) 1967). The
(b) Channel expansions in which the flow changes values in Table 2-3 are generally accepted for design
from rapid to tranquil are normally of the wedge type. purposes.
The flow transformation can be accomplished by means
of the abrupt hydraulic jump or by a gradual flow change
involving an undular-type jump. In either case, it is Table 2-3
necessary that the flow transformation be contained in the Transition Loss Coefficients
transition section. The use of a stilling-basin type of
transition to stabilize the hydraulic jump is illustrated in Transi-
USAED, Los Angeles (1961) and USAEWES (1962). A tion
C C
typical example of this type of transition is given in Type c e Source
Plate 25.
Warped 0.10 0.20 Chow
1959,
(c) USAED, Los Angeles (1958, 1961, 1962) has
Brater
designed and model tested a number of transitions trans- and King
forming rapid flow in rectangular channels to tranquil 1976
flow in trapezoidal channels without the occurrence of an
abrupt hydraulic jump. The high-velocity jet from the Cylin- 0.15 0.20 Chow
drical 1959
rectangular channel is expanded in the transition by means
Quadrant
of lateral and boundary roughness control in such a man-
ner that an undular-type jump occurs in the downstream Wedge 0.30 0.50 USBR
reach of the transition. Plate 26 illustrates a typical 1967
design developed through model tests.
Straight 0.30 0.50 Chow
Line 1959
d. Transition losses.
Square 0.30 0.75 Chow
(1) Tranquil flow. Transitions for tranquil flow are End 1959
designed to effect minimum energy losses consistent with
economy of construction. Transition losses are normally
computed using the energy equation and are expressed in (2) Rapid flow. Transition losses may be estimated
terms of the change in velocity head ∆hv from upstream for rapid-flow conditions from the information supplied in
to downstream of the transition. The head loss h1 (1) above. However, the effects of standing waves and
between cross sections in the step computation may be other factors discussed in c(2) above make exact
expressed as

2-11
EM 1110-2-1601
1 Jul 91

determinations of losses difficult. Model tests should be where

considered for important rapid-flow transitions.
∆y = rise in water surface between a
2-5. Flow in Curved Channels theoretical level water surface at
the center line and outside water-
a. General. surface elevation (superelevation)

(1) The so-called centrifugal force caused by flow C = coefficient (see Table 2-4)
around a curve results in a rise in the water surface on the
outside wall and a depression of the surface along the V = mean channel velocity
inside wall. This phenomenon is called superelevation.
In addition, curved channels tend to create secondary W = channel width at elevation of
flows (helicoidal motion) that may persist for many chan- center-line water surface
nel widths downstream. The shifting of the maximum
velocity from the channel center line may cause a disturb- g = acceleration of gravity
ing influence downstream. The latter two phenomena
could lead to serious local scour and deposition or poor r = radius of channel center-line
performance of a downstream structure. There may also curvature
be a tendency toward separation near the inner wall, espe-
cially for very sharp bends. Because of the complicated Use of the coefficient C in Equation 2-31 allows compu-
nature of curvilinear flow, the amount of channel align- tation of the total rise in water surface due to
ment curvature should be kept to a minimum consistent superelevation and standing waves for the conditions
with other design requirements. listed in Table 2-4. If the total rise in water surface
(superelevation plus surface disturbances) is less than
(2) The required amount of superelevation is usually 0.5 ft, the normally determined channel freeboard (para-
small for the channel size and curvature commonly used graph 2-6 below) should be adequate. No special
in the design of tranquil-flow channels. The main treatment such as increased wall heights or invert banking
problem in channels designed for rapid flow is standing and spiral transitions is required.
waves generated in simple curves. These waves not only
affect the curved flow region but exist over long distances
downstream. The total rise in water surface for rapid Table 2-4
flow has been found experimentally to be about twice that Superelevation Formula Coefficients
for tranquil flow.
Channel
Flow Type Cross Section Type of Curve Value of C
(3) Generally, the most economical design for rapid
flow in a curved channel results when wave effects are re- Tranquil Rectangular Simple Circular 0.5
duced as much as practical and wall heights are kept to a Tranquil Trapezoidal Simple Circular 0.5
minimum. Channel design for rapid flow usually involves Rapid Rectangular Simple Circular 1.0
Rapid Trapezoidal Simple Circular 1.0
low rates of channel curvature, the use of spiral transi- Rapid Rectangular Spiral Transitions 0.5
tions with circular curves, and consideration of invert Rapid Tapezoidal Spiral Transitions 1.0
banking. Rapid Rectangular Spiral Banked 0.5

b. Superelevation. The equation for the transverse

water-surface slope around a curve can be obtained by (1) Tranquil flow. The amount of superelevation in
balancing outward centrifugal and gravitational forces tranquil flow around curves is small for the normal chan-
(Woodward and Posey 1941). If concentric flow is nel size and curvature used in design. No special treat-
assumed where the mean velocity occurs around the ment of curves such as spirals or banking is usually
curve, the following equation is obtained necessary. Increasing the wall height on the outside of the
curve to contain the superelevation is usually the most
V 2W economical remedial measure. Wall heights should be
∆y C (2-31)
gr increased by ∆y over the full length of curvature. Wall
heights on the inside of the channel curve should be held

2-12
 EM 1110-2-1601
1 Jul 91

to the straight channel height because of wave action on to be made for superelevation around the curve. The
the inside of curves. minimum length of spiral should be 30 times the amount
of superelevation (∆y) (USAED, Los Angeles, 1950).
(2) Rapid flow. The disturbances caused by rapid
flow in simple curves not only affect the flow in the
Ls 30∆y (2-33)
curve, but persist for many channel widths downstream.
The cross waves generated at the beginning of a simple
curve may be reinforced by other cross waves generated
farther downstream. This could happen at the end of the The detailed design of spiral curves is given in
curve or within another curve, provided the upstream and Appendix D. A computer program for superelevation and
downstream waves are in phase. Wall heights should be curve layout is included. Banked inverts are not used in
increased by the amount of superelevation, not only in the trapezoidal channels because of design complexities and
simple curve, but for a considerable distance downstream. because it is more economical to provide additional free-
A detailed analysis of standing waves in simple curves is board for the moderate amount of superelevation that
given in Ippen (1950). Rapid-flow conditions are usually occurs in this type of channel.
improved in curves by the provision of spiral transition
curves with or without a banked invert, by dividing walls c. Limiting curvature. Laboratory experiments and
to reduce the channel width, or by invert sills located in field experience have demonstrated that the helicoidal
the curve. Both the dividing wall and sill treatments flow, velocity distribution distortion, and separation
require structures in the flow; these structures create around curves can be minimized by properly proportion-
debris problems and, therefore, are not generally used. ing channel curvature. Woodward (1920) recommends
that the curve radius be greater than 2.5 times the channel
(a) Spiral transition curves. For channels in which width. From experiments by Shukry (1950) the radius of
surface disturbances need to be minimized, spiral transi- curvature should be equal to or greater than 3.0 times the
tion curves should be used. The gradual increase in wall channel width to minimize helicoidal flow.
deflection angles of these curves results in minimum wave
heights. Two spiral curves are provided, one upstream (1) Tranquil flow. For design purposes a ratio of
and one downstream of the central circular curve. The radius to width of 3 or greater is suggested for tranquil
minimum length of spirals for unbanked curves should be flow.
determined by (see Douma, p 392, in Ippen and Dawson
1951) (2) Rapid flow. Large waves are generated by rapid
flow in simple curves. Therefore a much smaller rate of
change of curvature is required than for tranquil flow. A
VW 1969 study by USAED, Los Angeles (1972), of as-built
Ls 1.82 (2-32)
gy structures shows that curves with spiral transitions, with
or without banked inverts, have been constructed with
radii not less than

where y is the straight channel flow depth.

4V 2W (2-34)
rmin
(b) Spiral-banked curves. For rectangular channels, gy
the invert should be banked by rotating the bottom in
transverse sections about the channel center line. Spirals
are used upstream and downstream of the central curve where
with the banking being accomplished gradually over the
length of the spiral. The maximum amount of banking or rmin = minimum radius of channel curve
difference between inside and outside invert elevations in center line
the circular curve is equal to twice the superelevation
given by Equation 2-31. The invert along the inside wall V = average channel velocity
is depressed by ∆y below the center-line elevation and
the invert along the outside wall is raised by a like W= channel width at water surface
amount. Wall heights are usually designed to be equal on
both sides of the banked curves and no allowance needs y = flow depth

2-13
EM 1110-2-1601
1 Jul 91

The amount of superelevation required for spiral-banked in the channels; and variation of resistance or other coeffi-
curves (b above) is given by cients from those assumed in design.

(2) Local regions where water- surface elevations are

V 2W
∆y C (2-35) difficult to determine may require special consideration.
gr Some examples are locations in or near channel curves,
hydraulic jumps, bridge piers, transitions and drop
structures, major junctions, and local storm inflow struc-
However, this study indicates that the maximum allowable tures. As these regions are subject to wave-action
superelevation compatible with Equation 2-34 is uncertainties in water-surface computations and possible
overtopping of walls, especially for rapid flow, conserva-
tive freeboard allowances should be used. The backwater
2∆y W tan 10 0.18W (2-36)
effect at bridge piers may be especially critical if debris
accumulation is a problem.

or (3) The amount of freeboard cannot be fixed by a

single, widely applicable formula. It depends in large part
on the size and shape of channel, type of channel lining,
∆y 0.09W consequences of damage resulting from overtopping, and
velocity and depth of flow. The following approximate
freeboard allowances are generally considered to be satis-
d. Bend loss. There has been no complete, sys- factory: 2 ft in rectangular cross sections and 2.5 ft in
tematic study of head losses in channel bends. Data by trapezoidal sections for concrete-lined channels; 2.5 ft for
Shukry (1950), Raju (1937), and Bagnold (1960) suggest riprap channels; and 3 ft for earth levees. The freeboard
that the increased resistance loss over and above that for riprap and earth channels may be reduced somewhat
attributable to an equivalent straight channel is very small because of the reduced hazard when the top of the riprap
for values of r/W > 3.0 . For very sinuous channels, it or earth channels is below natural ground levels. It is
may be necessary to increase friction losses used in de- usually economical to vary concrete wall heights by 0.5-ft
sign. Based on tests in the Tiger Creek Flume, Scobey increments to facilitate reuse of forms on rectangular
(1933) recommended that Manning’s n be increased by channels and trapezoidal sections constructed by channel
0.001 for each 20 deg of curvature per 100 ft of channel, pavers.
up to a maximum increase of about 0.003. The small in-
crease in resistance due to curvature found by Scobey was (4) Freeboard allowances should be checked by
substantiated by the USBR field tests (Tilp and Scrivner computations or model tests to determine the additional
1964) for r/W > 4 . Recent experiments have indicated discharge that could be confined within the freeboard
that the channel bend loss is also a function of Froude allowance. If necessary, adjustments in freeboard should
number (Rouse 1965). According to experiments by be made along either or both banks to ensure that the
Hayat (Rouse 1965), the free surface waves produced by freeboard allowance provides the same degree of protec-
flow in a bend can cause an increase in resistance. tion against overtopping along the channel.

2-6. Special Considerations b. Sediment transport. Flood control channels with

tranquil flow usually have protected banks but unprotected
a. Freeboard. inverts. In addition to reasons of economy, it is some-
times desirable to use the channel streambed to percolate
(1) The freeboard of a channel is the vertical dis- water into underground aquifers (USAED, Los Angeles,
tance measured from the design water surface to the top 1963). The design of a channel with unprotected inverts
of the channel wall or levee. Freeboard is provided to and protected banks requires the determination of the
ensure that the desired degree of protection will not be depth of the bank protection below the invert in regions
reduced by unaccounted factors. These might include where bed scour may occur. Levee heights may depend
erratic hydrologic phenomena; future development of on the amount of sediment that may deposit in the chan-
urban areas; unforeseen embankment settlement; the accu- nel. The design of such channels requires estimates of
mulation of silt, trash, and debris; aquatic or other growth sediment transport to predict channel conditions under
given flow and sediment characteristics. The subject of

2-14
 EM 1110-2-1601
1 Jul 91

sediment transport in alluvial channels and design of 2-7. Stable Channels

canals has been ably presented by Leliavsky (1955).
Fundamental information on bed-load equations and their a. General.
background with examples of use in channel design is
given in Rouse (1950) (see pp 769-857). An excellent (1) The design of stable channels requires that the
review with an extensive bibliography is available (Chien channel be in material or lined with material capable of
1956). This review includes the generally accepted resisting the scouring forces of the flow. Channel armor-
Einstein approach to sediment transport. A comparative ing is required if these forces are greater than those that
treatment of the many bed-load equations (Vanoni, the bed and bank material can resist. The basic principles
Brooks, and Kennedy 1961) with field data indicates that of stable channel design have been presented by Lane
no one formula is conclusively better than any other and (1955) and expanded and modified by Terrell and Borland
that the accuracy of prediction is about +100 percent. A (1958) and Carlson and Miller (1956). An outline of the
recent paper by Colby (1964b) proposes a simple, direct method of channel design to resist scouring forces has
method of empirically correlating bed- load discharge been given in Simons (1957). The most common type of
with mean channel velocity at various flow depths and channel instability encountered in flood control design is
median grain size diameters. This procedure is adopted scouring of the bed and banks. This results from rela-
herein for rough estimates of bed-load movement in flood tively large discharges, steep channel slopes, and normally
control channels. limited channel right-of-way widths. These factors fre-
quently require the use of protective revetment to prevent
c. Design curves. Plate 27 gives curves of bed-load scouring.
discharge versus channel velocity for three depths of flow
and four sediment sizes. The basic ranges of depths and (2) While clay and silt are fairly resistant to scour,
velocities have been extrapolated and interpolated from especially if covered with vegetation, it is necessary to
the curves presented in Colby (1964a) for use in flood provide channel revetment when tractive forces are
control channel design. Corrections for water temperature sufficiently high to cause erosion of channels in fine
and concentration of fine sediment (Colby 1964a) are not material. Little is known about the resistance of clay and
included because of their small influence. The curves in silt to erosion as particles in this size range are influenced
Plate 27 should be applicable for estimating bed-load dis- to a large extent by cohesive forces. A summary of some
charge in channels having geologic and hydraulic charac- of the effects is given by the Task Committee on Prepara-
teristics similar to those in the channels from which the tion of Sedimentation Manual (1966). Suggested maxi-
basic data were obtained. The curves in this plate can mum limiting average channel velocities for noncohesive
also be used to estimate the relative effects of a change in materials are listed in c below and plotted in Plate 28.
channel characteristics on bed-load movement. For exam-
ple, the effect of a series of check dams or drop structures b. Prevention of scour. Scour and deposition occur
that are provided to decrease channel slope would be most commonly when particle sizes range from fine sand
reflected in the hydraulic characteristics by decreasing the to gravel, i.e., from about 0.1 mm through 50 mm
channel velocity. The curves could then be used to esti- (Plate 28). Erosion of sands in the lower range of sizes is
mate the decrease in sediment load. The curves can also especially critical as the sand particle weight is small,
be used to approximate the equilibrium sediment dis- there is no cohesion between grains, and there is usually
charge. If the supply of sediment from upstream sources little vegetation along the channel. This particle size
is less than the sediment discharge computed by the rating range comprises the majority of the bed and suspended
curves, the approximate amount of streambed scour can load in many streams. Paragraph 2-6 above discusses
be estimated from the curves. Similarly, deposition will sediment movement and presents a sediment rating curve
occur if the sediment supply is greater than the sediment as a guide to predicting channel stability.
discharge indicated by the rating curves. An example of
this is a large sediment load from a small side channel c. Permissible velocity and shear. The permissible
that causes deposition in a major flood channel. If the velocity and shear for a nonerodible channel should be
location of sediment deposition is to be controlled, the somewhat less than the critical velocity or shear that will
estimated size of a sediment detention facility can be erode the channel. The adoption of maximum permissible
approximated using the curves. An example of the use of velocities that are used in the design of channels has been
a sediment discharge equation in channel design is given widely accepted since publication of a table of values by
in USAED, Los Angeles (1963). Fortier and Scobey (1926). The latest information on

2-15
EM 1110-2-1601
1 Jul 91

critical scour velocities is given by the Task Committee permissible velocity or shear values should be based on
on Preparation of Sedimentation Manual (1966). reliable field experience or laboratory tests. Channels
Table 2-5 gives a set of permissible velocities that can be whose velocities and/or shear exceed permissible values
used as a guide to design nonscouring flood control cha- will require paving or bank revetment. The permissible
nnels. Lane (1955) presents curves showing permissible values of velocity and/or shear should be determined so
channel shear stress to be used for design, and the Soil that damage exceeding normal maintenance will not result
Conservation Service (1954) presents information on from any flood that could be reasonably expected to occur
grass-lined channels. Departures from suggested during the service life of the channel.

Table 2-5
Suggested Maximum Permissible Mean Channel Velocities

Mean Channel
Channel Material Velocity, fps

Fine Sand 2.0

Coarse Sand 4.0

Fine Gravel1 6.0

Earth
Sandy Silt 2.0
Silt Clay 3.5
Clay 6.0

Grass-lined Earth
(slopes less
than 5%)2
Bermuda Grass
Sandy Silt 6.0
Silt Clay 8.0

Kentucky Blue
Grass
Sandy Silt 5.0
Silt Clay 7.0

Poor Rock (usually

sedimentary) 10.0
Soft Sandstone 8.0
Soft Shale 3.5

Good Rock (usually

igneous or hard
metamorphic) 20.0

Notes:
1. For particles larger than fine gravel (about 20 millimetres (mm)
= 3/4 in.), see Plates 29 and 30.
2. Keep velocities less than 5.0 fps unless good cover and proper
maintenance can be obtained.

2-16
 EM 1110-2-1601
Change 1
30 Jun 94

Chapter 3 more likely to be eroded by channel flow. If used, the

Riprap Protection rounded stone should be placed on flatter side slopes than
angular stone and should be about 25 percent larger in
diameter. The following shape limitations should be spec-
ified for riprap obtained from quarry operations:
Section I
Introduction (1) The stone shall be predominantly angular in
shape.
3-1. General
(2) Not more than 30 percent of the stones distri-
* The guidance presented herein applies to riprap design for buted throughout the gradation should have a ratio of a/c
open channels not immediately downstream of stilling greater than 2.5.
basins or other highly turbulent areas (for stilling basin
riprap, use HDC 712-1, Plates 29 and 30). The ability of *
riprap slope protection to resist the erosive forces of chan- (3) Not more than 15 percent of the stones distri-
nel flow depends on the interrelation of the following buted throughout the gradation should have a ratio of a/c
factors: stone shape, size, weight, and durability; riprap greater than 3.0.
gradation and layer thickness; and channel alignment,
cross-section, gradient, and velocity distribution. The bed (4) No stone should have a ratio of a/c greater than
material and local scour characteristics determine the 3.5.
design of toe protection which is essential for riprap reve-
tment stability. The bank material and groundwater con- To determine stone dimensions a and c , consider that
ditions affect the need for filters between the riprap and the stone has a long axis, an intermediate axis, and a short
underlying material. Construction quality control of both axis, each being perpendicular to the other. Dimension a
stone production and riprap placement is essential for is the maximum length of the stone, which defines the
successful bank protection. Riprap protection for flood long axis of the stone. The intermediate axis is defined
control channels and appurtenant structures should be by the maximum width of the stone. The remaining axis
designed so that any flood that could reasonably be is the short axis. Dimension c is the maximum dimen-
expected to occur during the service life of the channel or sion parallel to the short axis. These limitations apply
structure would not cause damage exceeding nominal only to the stone within the required riprap gradation and
maintenance or replacement (see ER 1110-2-1150). not to quarry spalls and waste that may be allowed.
While the procedures presented herein yield definite stone
sizes, results should be used for guidance purposes and b. Relation between stone size and weight. The
revised as deemed necessary to provide a practical protec- ability of riprap revetment to resist erosion is related to
tion design for the specific project conditions. the size and weight of stones. Design guidance is often
expressed in terms of the stone size D% , where %
3-2. Riprap Characteristics denotes the percentage of the total weight of the graded
material (total weight including quarry wastes and spalls)
The following provides guidance on stone shape, that contains stones of less weight. The relation between
size/weight relationship, unit weight, gradation, and layer size and weight of stone is described herein using a spher-
thickness. Reference EM 1110-2-2302 for additional ical shape by the equation
guidance on riprap material characteristics and
construction.
 6W 1/3
a. Stone shape. Riprap should be blocky in shape D%  %  (3-1)
rather than elongated, as more nearly cubical stones  πγ s 
“nest” together best and are more resistant to movement.
where
The stone should have sharp, angular, clean edges at the
intersections of relatively flat faces. Stream rounded
D% = equivalent-volume spherical stone diameter, ft
stone is less resistant to movement, although the drag
force on a rounded stone is less than on angular, cubical
W% = weight of individual stone having diameter of D%
stones. As rounded stone interlock is less than that of
equal-sized angular stones, the rounded stone mass is

3-1
EM 1110-2-1601
1 Jul 91

γs = saturated surface dry specific or unit weight of stone, Division (LMVD) Standardized Gradations presented in
pcf Appendix F. The LMVD gradations are similar to the
gradations listed in Table 3-1 except the LMVD
Plate 31 presents relations between spherical diameter and W50(max) and W15(max) weights are larger, which can
weight for several values of specific or unit weight. make the LMVD gradations easier to produce. Most
Design procedures for determining the stone size required graded ripraps have ratios of D85/D15 less than 3. Uniform
to resist the erosive forces of channel flow are presented riprap (D85/D15 < 1.4) has been used at sites in the US
in paragraph 3-5 below. Army Engineer Division, Missouri River, for reasons of
economy and quality control of sizes and placement.
c. Unit weight. Unit weight of stone γs generally
varies from 150 to 175 pcf. Riprap sizing relations are (3) Rather than a relatively expensive graded riprap,
relatively sensitive to unit weight of stone, and γs should a greater thickness of a quarry-run stone may be consid-
be determined as accurately as possible. In many cases, ered. Some designers consider the quarry-run stone to
the unit weight of stone is not known because the quarry have another advantage: its gravel- and sand-size compo-
is selected from a list of approved riprap sources after the nents serve as a filter. The gravel and sand sizes should
construction contract is awarded. Riprap coming from the be less by volume than the voids among the larger stone.
various quarries will not be of the same unit weight. This concept has resulted in considerable cost savings on
Under these circumstances, a unit weight of stone close to large projects such as the Arkansas and Red River Navi-
the minimum of the available riprap sources can be used gation Projects. Not all quarry-run stone can be used as
in design. Contract options covering specific weight riprap; stone that is gap graded or has a large range in
ranges of 5 or 10 pcf should be offered when sufficient maximum to minimum size is probably unsuitable.
savings warrant. Quarry-run stone for riprap should be limited to D85/D15 ≤
7.
d. Gradation.
(4) Determining optimum gradations is also an
(1) The gradation of stones in riprap revetment economics problem that includes the following factors:
affects the riprap’s resistance to erosion. Stone should be
reasonably well graded throughout the in-place layer (a) Rock quality (durability under service conditions)
thickness. Specifications should provide for two limiting
gradation curves, and any stone gradation as determined (b) Cost per ton at the quarry (including capability
from quarry process, stockpile, and in-place field test of quarry to produce a particular size)
samples that lies within these limits should be acceptable.
Riprap sizes and weights are frequently used such as (c) Number of tons required
D30(min), D100(max), W50(min), etc. The D or W
refers to size or weight, respectively. The number is the (d) Miles transported
percent finer by weight as discussed in b above. The
(max) or (min) refers to the upper or lower limit (e) Cost of transportation per ton-mile
gradation curves, respectively. Engineer Form 4794-R is
a standard form for plotting riprap gradation curves (f) Cost per ton for placement
(Plate 32). The gradation limits should not be so restric-
tive that production costs would be excessive. The choice (g) Need for and cost of filter
of limits also depends on the underlying bank soils and
filter requirements if a graded stone filter is used. Filters (h) Quality control during construction (it is easier to
may be required under riprap revetments. Guidance for ensure even coverage with a narrow gradation than with a
filter requirements is given in EM 1110-2-1901. Filter wide gradation)
design is the responsibility of the Geotechnical Branch in
each District. (i) Number of different gradations required. Some-
times cost savings can be realized by using fewer
(2) Standardized gradations having a relatively narrow gradations.
range in sizes (D85/D15 of 1.4-2.2) are shown in Table 3-1.
Other gradations can be used and often have a wider See EM 1110-2-2302 for further discussion of these
range of allowable sizes than those given in Table 3-1. factors.
One example is the Lower Mississippi Valley

3-2
 EM 1110-2-1601
Change 1
30 Jun 94

Table 3-1
Gradations for Riprap Placement in the Dry, Low-Turbulence Zones

Limits of Stone Weight, lb1, for Percent Lighter by Weight

D100(max) 100 50 15 D30(min) D90(min)
in. Max Min Max2 Min Max2 Min ft ft

Specific Weight = 155 pcf

* 9 34 14 10 7 5 2 0.37 0.53 *
12 81 32 24 16 12 5 0.48 0.70
15 159 63 47 32 23 10 0.61 0.88
18 274 110 81 55 41 17 0.73 1.06
21 435 174 129 87 64 27 0.85 1.23
24 649 260 192 130 96 41 0.97 1.40
27 924 370 274 185 137 58 1.10 1.59
30 1,268 507 376 254 188 79 1.22 1.77
33 1,688 675 500 338 250 105 1.34 1.94
36 2,191 877 649 438 325 137 1.46 2.11
42 3,480 1,392 1,031 696 516 217 1.70 2.47
48 5,194 2,078 1,539 1,039 769 325 1.95 2.82
54 7,396 2,958 2,191 1,479 1,096 462 2.19 3.17

Specific Weight = 165 pcf

* 9 36 15 11 7 5 2 0.37 0.53 *
12 86 35 26 17 13 5 0.48 0.70
15 169 67 50 34 25 11 0.61 0.88
18 292 117 86 58 43 18 0.73 1.06
21 463 185 137 93 69 29 0.85 1.23
24 691 276 205 138 102 43 0.97 1.40
27 984 394 292 197 146 62 1.10 1.59
30 1,350 540 400 270 200 84 1.22 1.77
33 1,797 719 532 359 266 112 1.34 1.96
36 2,331 933 691 467 346 146 1.46 2.11
42 3,704 1,482 1,098 741 549 232 1.70 2.47
48 5,529 2,212 1,638 1,106 819 346 1.95 2.82
54 7,873 3,149 2,335 1,575 1,168 492 2.19 3.17

Specific Weight = 175 pcf

* 9 39 15 11 8 6 2 0.37 0.53 *
12 92 37 27 18 14 5 0.48 0.70
15 179 72 53 36 27 11 0.61 0.88
18 309 124 92 62 46 19 0.73 1.06
21 491 196 146 98 73 31 0.85 1.23
24 733 293 217 147 109 46 0.97 1.40
27 1,044 417 309 209 155 65 1.10 1.59
30 1,432 573 424 286 212 89 1.22 1.77
33 1,906 762 565 381 282 119 1.34 1.94
36 2,474 990 733 495 367 155 1.46 2.11
42 3,929 1,571 1,164 786 582 246 1.70 2.47
48 5,864 2,346 1,738 1,173 869 367 1.95 2.82
54 8,350 3,340 2,474 1,670 1,237 522 2.19 3.17

Notes:
1. Stone weight limit data from ETL 1110-2-120 (HQUSACE, 1971 (14 May), “Additional Guidance for Riprap Channel Protection, Ch 1,”
US Government Printing Office, Washington, DC). Relationship between diameter and weight is based on the shape of a sphere.
2. The maximum limits at the W50 and W15 sizes can be increased as in the Lower Mississippi Valley Division Standardized Gradations
shown in Appendix F.

3-3
EM 1110-2-1601
Change 1
30 Jun 94

e. Layer thickness. All stones should be contained plans to determine the most economical and practical
within the riprap layer thickness to provide maximum combination of channel factors and stone size. Resistance
resistance against erosive forces. Oversize stones, even in coefficients (Manning’s n) for riprap placed in the dry
isolated spots, may result in riprap failure by precluding should be estimated using the following form of
mutual support and interlock between individual stones, Strickler’s equation:
causing large voids that expose filter and bedding
materials, and creating excessive local turbulence that
n K D90(min) 1/6 (3-2)
removes smaller size stone. Small amounts of oversize
stone should be removed individually and replaced with
proper size stones. The following criteria apply to the where
riprap layer thickness:
K = 0.036, average of all flume data
(1) It should not be less than the spherical diameter
of the upper limit W100 stone or less than 1.5 times the = 0.034 for velocity and stone size calculation
spherical diameter of the upper limit W50 stone, whichever
results in the greater thickness. = 0.038 for capacity and freeboard calculation
(2) The thickness determined by (1) above should be D90(min) = size of which 90 percent of sample is finer,
increased by 50 percent when the riprap is placed under- from minimum or lower limit curve of
water to provide for uncertainties associated with this type gradation specification, ft
of placement. At one location in the US Army Engineer
Division, Missouri River, divers and sonic sounders were The K values represent the upper and lower bounds of
used to reduce the underwater thickness to 1.25 times the laboratory data determined for bottom riprap. Resistance
dry placement thickness. data from a laboratory channel which had an irregular
surface similar to riprap placed underwater show a
Section II Manning’s n about 15 percent greater than for riprap
Channel Characteristics * placed in the dry. Equation 3-2 provides resistance losses
due to the surface roughness of the riprap and does not
3-3. Side Slope Inclination include form losses such as those caused by bends. Equa-
tion 3-2 should be limited to slopes less than 2 percent. *
The stability of riprap slope protection is affected by the
steepness of channel side slopes. Side slopes should ordi-
Section III
narily not be steeper than 1V on 1.5H, except in special
Design Guidance for Stone Size
cases where it may be economical to use larger
hand-placed stone keyed well into the bank. Embankment
3-5. General
stability analysis should properly address soils characteris-
tics, groundwater and river conditions, and probable Riprap protection for open channels is subjected to hydro-
failure mechanisms. The size of stone required to resist dynamic drag and lift forces that tend to erode the revet-
the erosive forces of channel flow increases when the side ment and reduce its stability. Undermining by scour
slope angle approaches the angle of repose of a riprap beyond the limits of protection is also a common cause of
slope protection. Rapid water-level recession and piping- failure. The drag and lift forces are created by flow
initiated failures are other factors capable of affecting velocities adjacent to the stone. Forces resisting motion
channel side slope inclination and needing consideration are the submerged weight of the stone and any downward
in design. and lateral force components caused by contact with other
stones in the revetment. Stone availability and experience
3-4. Channel Roughness, Shape, play a large part in determining size of riprap. This is
Alignment, and Gradient particularly true on small projects where hydraulic param-
eters are ill-defined and the total amount of riprap
As boundary shear forces and velocities depend on chan- required is small.
nel roughness, shape, alignment, and invert gradient, these
factors must be considered in determining the size of
stone required for riprap revetment. Comparative cost
estimates should be made for several alternative channel

3-4
 EM 1110-2-1601
Change 1
30 Jun 94

3-6. Design Conditions upstream bends, large values of R/W should be used,
resulting in constant values of VSS/VAVG . Very few
Stone size computations should be conducted for flow channels are straight enough to justify using VSS/VAVG
conditions that produce the maximum velocities at the < 1 . A minimum ratio of VSS/VAVG = 1 is recom-
riprapped boundary. In many cases, velocities continue to mended for side slopes in straight channels. Rock stabil-
increase beyond bank-full discharge; but sometimes back- ity should be checked for both side slopes and the channel
water effects or loss of flow into the overbanks results in bottom. In bendways, the outer bank side slope will
velocities that are less than those at bank-full. Riprap at generally require the largest rock size. In straight reaches,
channel bends is designed conservatively for the point the channel bottom will often require the largest stone
having the maximum force or velocity. For braided chan- size. Velocities in the center of a straight channel having
nels, bank-full discharges may not be the most severe equal bottom and side slope roughness range from 10 to
condition. At lesser flows, flow is often divided into 20 percent greater than VAVG . Plate 34 describes VSS
multiple channels. Flow in these channels often impinges and Plate 35 shows the location in a trapezoidal channel
* abruptly on banks or levees at sharp angles. bend of the maximum VSS . Velocity downstream of
bends decays at approximately the following rate: No
3-7. Stone Size decay in first channel width downstream of bend exit;
decay of VSS/ VAVG = 0.1 per channel width until
This method for determining stone size uses depth- VSS/VAVG = 1.0. Plate 36 shows the variation in
averaged local velocity. The method is based on the idea velocity over the side slope in a channel. The straight
that a designer will be able to estimate local velocity channel curve in Plate 36 was found applicable to both
* better than local boundary shear. Local velocity and local 1V:2H and 1V:3H side slopes. The bend curve for
flow depth are used in this procedure to quantify the R/W = 2.6 was taken from a channel having strong sec-
imposed forces. Riprap size and unit weight quantify the ondary currents and represents a severe concentration of
resisting force of the riprap. This method is based on a high velocity upon the channel side slope. These two
large body of laboratory data and has been compared to curves represent the extremes in velocity distribution to be
available prototype data (Maynord 1988). It defines the expected along the outer bank of a channel bend
stability of a wide range of gradations if placed to a thick- having a riprap side slope from toe of bank to top of
* ness of 1D100(max). Guidance is also provided for thick- bank. Knowing VSS from Plate 33, the side slope
ness greater than 1D100(max). This method is applicable * velocity distribution can be determined at the location of
to side slopes of 1V on 1.5H or flatter. VSS. An alternate means of velocity estimation based on
field observation is discussed in Appendix G. The alpha
a. Velocity estimation. The characteristic velocity for method (Appendix C), or velocities resulting from sub-
side slopes VSS is the depth-averaged local velocity over sections of a water-surface profile computation, should be
the slope at a point 20 percent of the slope length from used only in straight reaches. When the alpha method is
* the toe of slope. Plate 33 presents the ratio VSS/VAVG , used, velocity from the subsection adjacent to the bank
where VAVG is the average channel velocity at the subsection should be used as VSS in design of bank
upstream end of the bend, as a function of the channel riprap.
geometry, which is described by R/W , where R is the
center-line radius of bend and W is the water-surface b. Stone size relations. The basic equation for the
width. VAVG , R , and W should be based on flow in the representative stone size in straight or curved channels is
main channel only and should not include overbank areas.
The trapezoidal curve for VSS/VAVG shown in Plate 33 is
based on the STREMR numerical model described in
 1/2  2.5
Bernard (1993). The primary factors affecting velocity  γ V 
distribution in riprap-lined, trapezoidal channel bendways D30 SfCsCVCTd  w 
  (3-3)
 γ
 s γ w 
are R/W , bend angle, and aspect ratio (bottom width/  K1gd 
depth). Data in Maynord (1992) show a trapezoidal chan-
nel having the same bottom width but side slopes ranging
from 1V:1.5H to 1V:3H to have the same maximum where
VSS/VAVG at the downstream end of the bend. Plate 33
should be used for side slopes from 1V:3H to 1V:1.5H. D30 = riprap size of which 30 percent is finer by weight,
For straight channels sufficiently far (>5W) from length

3-5
 EM 1110-2-1601
Change 1
30 Jun 94

Sf = safety factor (see c below) floating debris. Impact will produce more damage to
alighter weight riprap section than to a heavier section.
* Cs = stability coefficient for incipient failure, For moderate debris impact, it is unlikely that an added
D85/D15 = 1.7 to 5.2 * safety factor should be used when the blanket thickness
exceeds 15 in.
= 0.30 for angular rock
(2) The basic stone sizing parameters of velocity,
* = 0.375 for rounded rock * unit weight of rock, and depth need to be determined as
accurately as possible. A safety factor should be included
CV = vertical velocity distribution coefficient to compensate for small inaccuracies in these parameters.
If conservative estimates of these parameters are used in
= 1.0 for straight channels, inside of bends the analysis, the added safety factor should not be used.
The safety factor should be based on the anticipated error
= 1.283 - 0.2 log (R/W), outside of bends (1 for in the values used. The following discussion shows the
(R/W) > 26) importance of obtaining nearly correct values rather than
relying on a safety factor to correct inaccurate or assumed
= 1.25, downstream of concrete channels stone sizing parameters. The average velocity over the
toe of the riprap is an estimate at best and is the para-
= 1.25, ends of dikes meter to which the rock size is the most sensitive. A
check of the sensitivity will show that a 10 percent
CT = thickness coefficient (see d(1) below) change in velocity will result in a nearly 100 percent
change in the weight limits of the riprap gradation (based
* = 1.0 for thickness = 1D100(max) or 1.5 D50(max), on a sphere) and about a 30 percent change in the riprap
whichever is greater * thickness. The riprap size is also quite sensitive to the unit
weight of the rock to be used: a 10 percent change in the
* d = local depth of flow, length (same location as V) * unit weight will result in a 70 percent change in the
weight limits of the riprap gradation (based on a sphere)
γw = unit weight of water, weight/volume and about a 20 percent change in the riprap thickness.
The natural variability of unit weight of stone from a
* V = local depth-averaged velocity, VSS for side slope stone source adds to the uncertainty (EM 1110-2-2302).
riprap, length/time * The rock size is not nearly as sensitive to the depth para-
meter.
K1 = side slope correction factor (see d(1) below)
(3) Vandalism and/or theft of the stones is a serious
2
g = gravitational constant, length/time problem in urban areas where small riprap has been
placed. A W50(min) of 80 lb should help prevent theft
* Some designers prefer to use the traditional D50 in riprap and vandalism. Sometimes grouted stone is used around
design. The approximate relationship between D50 vandalism-prone areas.
and D30 is D50 = D30 (D85/D15)1/3. Equation 3-3 can be
used with either SI (metric) or non-SI units and should be (4) The completed revetment will contain some
limited to slopes less than 2 percent. * pockets of undersized rocks, no matter how much effort is
devoted to obtaining a well-mixed gradation throughout
c. Safety factor. Equation 3-3 gives a rock size that the revetment. This placement problem can be assumed
should be increased to resist hydrodynamic and a variety to occur on any riprap job to some degree but probably
of nonhydrodynamic-imposed forces and/or uncontrollable more frequently on jobs that require stockpiling or addi-
physical conditions. The size increase can best be accom- tional handling. A larger safety factor should be consid-
plished by including the safety factor, which will be a ered with stockpiling or additional hauling and where
* value greater than unity. The minimum safety factor is placement will be difficult if quality control cannot be
* Sf = 1.1 . The minimum safety factor may have to be expected to address these problems.
increased in consideration for the following conditions:
(5) The safety factor should be increased where
(1) Imposed impact forces resulting from logs, severe freeze-thaw is anticipated.
uprooted trees, loose vessels, ice, and other types of large

3-6
 EM 1110-2-1601
Change 1
30 Jun 94

The safety factor based on each of these considerations (a) Determine average channel velocity (HEC-2 or
should be considered separately and then the largest of other uniform flow computational methods, or
these values should be used in Equation 3-3. measurement).

d. Applications. (b) Find VSS using Plate 33.

(1) The outer bank of straight channels downstream (c) Find D30 using Plate 37.
of bends should be designed using velocities computed for
the bend. In projects where the cost of riprap is high, a (d) Correct for other unit weights, side slopes, verti-
channel model to indicate locations of high velocity might cal velocity distribution, or thicknesses using Plates 38
* be justified. Equation 3-3 has been developed into through 40.
Plate 37, which is applicable to thicknesses equal to
* 1D100(max), γs of 165 pcf, and the Sf of 1.1. Plate 38 is (e) Find gradation having D30(min) ≥ computed
used to correct for values of other than γs of 165 pcf * D30. Alternately Equation 3-3 is used with Plates 39
(when D30 is determined from Plate 37). The K1 side and 40 to replace steps (c) and (d). *
slope factor is normally defined by the relationship of
Carter, Carlson, and Lane (1953) (3) This procedure can be used in both natural chan-
nels with bank protection only and prismatic channels
having riprap on bed and banks. Most bank protection
sin2 θ (3-4) sections can be designed by direct solution. In these
K1 1 cases, the extent of the bank compared to the total perim-
sin2 φ eter of the channel means that the average channel veloc-
ity is not significantly affected by the riprap. The first
where example in Appendix H demonstrates this type of bank
protection.
θ = angle of side slope with horizontal
(4) In some cases, a large part of the channel perim-
φ = angle of repose of riprap material (normally eter is covered with riprap; the average channel velocity,
40 deg) depth, and riprap size are dependent upon one another;
and the solution becomes iterative. A trial riprap gra-
Results given in Maynord (1988) show Equation 3-4 to be dation is first assumed and resistance coefficients are
conservative and that the repose angle is not a constant computed using Equation 3-2. Then the five steps
40 deg but varies with several factors. The recommended described in (2) above are conducted. If the gradation
relationship for K1 as a function of θ is given in found in paragraph (e) above is equal to the assumed trial
Plate 39 along with Equation 3-4 using φ = 40 deg. gradation, the solution is complete. If not, a new trial
* Using the recommended curve for side slope effects, the gradation is assumed and the procedure is repeated. The
least volume of rock per unit length of bank line occurs second example in Appendix H demonstrates this type of
on a 1V:1.5H to 1V:2H side slope. Also shown on * channel riprap.
Plate 39 is the correction for side slope when D30 is
(5) In braided streams and some meandering
determined from Plate 37. Correction for the vertical
streams, flow is often directed into the bank line at sharp
* velocity distribution in bends is shown in Plate 40. Test-
* angles (angled flow impingement). For braided streams
ing has been conducted to determine the effects of blanket
having impinged flow, the above stone sizing procedures
thickness greater than 1D100(max) on the stability of rip-
require modification in two areas: the method of velocity
* rap. Results are shown in Plate 40. The thickness coeffi-
estimation and the velocity distribution coefficient CV .
cient CT accounts for the increase in stability that occurs
All other factors and coefficients presented are applicable.
when riprap is placed thicker than the minimum thickness
of 1D100(max) or 1.5 D50(max), whichever is greater. *
(a) The major challenge in riprap design for braided
* (2) The basic procedure to determine riprap size using streams is estimating the imposed force at the impinge-
the graphical solution of this method is as follows: ment point. Although unproven, the most severe bank *
*

3-7
 EM 1110-2-1601
Change 1
30 Jun 94

* attack in braided streams is thought to occur when the * e. Steep slope riprap design.
water surface is at or slightly above the tops of the mid-
channel bars. At this stage, flow is confined to the multi- In cases where unit discharge is low, riprap can be used
ple channels that often flow into or “impinge” against on steep slopes ranging from 2 to 20 percent. A typical
bank lines or levees. At lesser flows, the depths and application is a rock-lined chute. The stone size equation
velocities in the multiple channels are decreased. At is
higher flows, the channel area increases drastically and 1.95 S 0.555 q 2/3
D30 (3-5)
streamlines are in a more downstream direction rather
g 1/3
than into bank lines or levees.

(b) The discharge that produces a stage near the tops where
of the midchannel bars is Qtmcb . Qtmcb is probably
highly correlated with the channel-forming discharge S = slope of bed
concept. In the case of the Snake River near Jackson,
Wyoming, Qtmcb is 15,000-18,000 cfs, which has an q = unit discharge
average recurrence interval of about 2-5 years. Using
cross-section data to determine the channel area below the Equation 3-5 is applicable to thickness = 1.5 D100, angular
tops of the midchannel bars and Qtmcb allows determina- rock, unit weight of 167 pcf, D85/D15 from 1.7 to 2.7,
tion of the average channel velocity at the top of the slopes from 2 to 20 percent, and uniform flow on a down-
midchannel bars, Vtmcb . slope with no tailwater. The following steps should be
used in application of Equation 3-5:
(c) Field measurements at impingement sites were
taken in 1991 on the Snake River near Jackson, (1) Estimate q = Q/b where b = bottom width of
Wyoming, and reported in Maynord (1993). The maxi- chute.
mum observed ratio VSS/Vtmcb = 1.6 , which is almost
identical to the ratio shown in Plate 33 for sharp bend- (2) Multiply q by flow concentration factor of
ways having R/W = 2 in natural channels, and this ratio 1.25. Use greater factor if approach flow is skewed.
is recommended for determining VSS for impinged flow.
The second area of the design procedure requiring modifi- (3) Compute D30 using Equation 3-5.
cation for impinged flow is the velocity distribution coef-
ficient CV , which varies with R/W in bendways as (4) Use uniform gradation having D85/D15 ≤ 2 such
shown in Plate 40. Impinged flow areas are poorly as Table 3-1.
aligned bends having low R/W , and CV = 1.25 is
recommended for design. * (5) Restrict application to straight channels with side
slope of 1V:2.5H or flatter.
(6) Transitions in size or shape may also require
riprap protection. The procedures in this paragraph are (6) Use filter fabric beneath rock.
applicable to gradual transitions where flow remains tran-
quil. In areas where flow changes from tranquil to rapid The guidance for steep slope riprap generally results in
and then back to tranquil, riprap sizing methods applicable large riprap sizes. Grouted riprap is often used instead of
to hydraulic structures (HDC 712-1) should be used. In loose riprap in steep slope applications. *
converging transitions, the procedures based on Equa-
tion 3-3 can be used unaltered. In expanding transitions, 3-8. Revetment Top and End Protection
flow can concentrate on one side of the expansion and
design velocities should be increased. For installations Revetment top and end protection requirements, as with
immediately downstream of concrete channels, a vertical all channel protective measures, are to assure the project
velocity distribution coefficient of 1.25 should be used benefits, to perform satisfactorily throughout the project
due to the difference in velocity profile over the two economic life, and not to exceed reasonable maintenance
surfaces.

3-8
 EM 1110-2-1601
Change 1
30 Jun 94

costs. Reference is made to ER 1110-2-1405, with bend. In a trapezoidal channel, the maximum velocities
emphasis on paragraph 6c. along the outer bank are often located in the straight reach
immediately downstream of the bend for relatively large
a. Revetment top. When the full height of a levee is distances downstream. In a natural channel, the limit of
to be protected, the revetment will cover the freeboard, protection on the downstream end should depend on
i.e., extend to the top of the levee. This provides protec- where the flow crosses to the opposite bank, and should
tion against waves, floating debris, and water-surface consider future bar building on the opposite bank, result-
irregularities. Similar provisions apply to incised channel ing in channel constriction and increased velocities. Gui-
banks. A horizontal collar, at the top of the bank, is dance is generally lacking in this area, but review of
provided to protect against escaping and returning flows aerial photographs of the subject location can provide
as necessary. The end protection methods illustrated in some insight on where the crossover flow occurs. Model
Plate 41 can be adapted for horizontal collars. Plate 36 tests in a sand bed and bank flume (USACE 1981) were
provides general guidance for velocity variation over conducted to determine the limits of protection required to
channel side slopes that can assist in evaluating the prevent scour that would lead to destruction of the revet-
economics of reducing or omitting revetment for upper ment. These tests were conducted in a 110-deg bend
bank areas. Revetment size changes should not be made having a constant discharge. The downstream end of the
unless a sufficient quantity is involved to be cost effec- revetment had to be 1.5 channel widths downstream of the
tive. Many successful revetments have been constructed end of the bend. Geomorphic studies to determine revet-
where the top of the revetment was terminated below the ment ends should be considered.
design flow line. See USACE (1981) for examples.
Section IV
b. Revetment end protection. The upstream and Revetment Toe Scour Estimation and Protection
downstream ends of riprap revetment should be protected
against erosion by increasing the revetment thickness T 3-9. General
or extending the revetment to areas of noneroding veloci-
* ties and relatively stable banks. A smooth transition Toe scour is probably the most frequent cause of failure
should be provided from where the end protection begins of riprap revetments. This is true not only for riprap, but
to the design riprap section. The keyed-in section should also for a wide variety of protection techniques. Toe
satisfy filter requirements. The following guidance * scour is the result of several factors, including these three:
applies to the alternative methods of end protection illus-
trated in Plate 41. a. Meandering channels, change in cross section
that occurs after a bank is protected. In meandering
(1) Method A. For riprap revetments 12 in. thick or channels the thalweg often moves toward the outer bank
less, the normal riprap layer should be extended to areas after the bank is protected. The amount of change in
where velocities will not erode the natural channel banks. cross section that occurs after protection is added is
related to the erodibility of the natural channel bed and
(2) Method B. For riprap revetments exceeding original bank material. Channels with highly erodible bed
12 in. in thickness, one or more reductions in riprap thick- and banks can experience significant scour along the toe
* ness and stone size may be required (Plate 41) until of the new revetment.
velocities decrease to a noneroding natural channel
velocity. b. Meandering channels, scour at high flows. Bed
profile measurements have shown that the bed observed at
(3) Method C. For all riprap revetments that do not low flows is not the same bed that exists at high flows.
terminate in noneroding natural channel velocities, the At high flows the bed scours in channel bends and builds
ends of the revetment should be enlarged, as shown in up in the crossings between bends. On the recession side
* Plate 41. The decision to terminate the revetment in of the flood, the process is reversed. Sediment is eroded
erosive velocities should be made with caution since from the crossings and deposited in the bends, thus
severe erosion can cause the revetment to fail by progres- obscuring the maximum scour that had occurred.
sive flanking.
c. Braided channels. Scour in braided channels can
c. Length. Riprap revetment is frequently carried too reach a maximum at intermediate discharges where flow
far upstream and not far enough downstream of a channel in the channel braids attacks banks at sharp angles.

3-9
 EM 1110-2-1601
Change 1
30 Jun 94

Note that local scour is the mechanism being addressed 3-11. Revetment Toe Protection Design
herein. When general bed degradation or headcutting is
expected, it must be added to the local scour. When The following guidance applies to several alternative
scour mechanisms are not considered in the design of methods of toe protection illustrated in Plate 43.
protection works, undermining and failure may result.
* Plate 42 may be used for depth of scour estimates. The a. Method A. When toe excavation can be made in
design curve in Plate 42 represents an upper limit for the dry, the riprap layer may be extended below the exist-
scour in channels having irregular alignments. For bend- ing groundline a distance exceeding the anticipated depth
ways having a relatively smooth alignment, a 10 percent of scour. If excavation quantities are prohibitive, the
reduction from the design curve is recommended. Neill * concept of Method D can be adapted to reduce
(1973) provides additional information on scour depth excavation.
estimation.
b. Method B. When the bottom of the channel is
3-10. Revetment Toe Protection Methods nonerodible material, the normal riprap should be keyed
in at streambed level.
Toe protection may be provided by two methods:
c. Method C. When the riprap is to be placed
a. Extend to maximum scour depth. Place the lower underwater and little toe scour is expected (such as in
extremity below the expected scour depth or found it on straight reaches that are not downstream of bends, unless
nonerodible material. These are the preferred methods, stream is braided), the toe may be placed on the existing
but they can be difficult and expensive when underwater bottom with height a and width c equal to 1.5T and
excavation is required. 5T , respectively. This compensates for uncertainties of
underwater placement.
b. Place launchable stone. Place sufficient launch-
able stone to stabilize erosion. Launchable stone is d. Method D. An extremely useful technique where
defined as stone that is placed along expected erosion water levels prohibit excavation for a toe section is to
areas at an elevation above the zone of attack. As the place a launchable section at the toe of the bank. Even if
attack and resulting erosion occur below the stone, the excavation is practicable, this method may be preferred
stone is undermined and rolls/slides down the slope, stop- for cost savings if the cost of extra stone required to
ping the erosion. This method has been widely used on produce a launched thickness equal to or greater than
sand bed streams. Successful applications include: * T plus the increase shown in Table 3-2 is exceeded by the
cost of excavation required to carry the design thickness
(1) Windrow revetments: riprap placed at top of T down the slope. This concept simply uses toe scour as
bank. a substitute for mechanical excavation. This method also
has the advantage of providing a “built-in” scour gage,
(2) Trench-fill revetments: riprap placed at low water allowing easy monitoring of high-flow scour and the need
level. for additional stone reinforcement by visual inspection of
the remaining toe stone after the high flow subsides or by
(3) Weighted riprap toes: riprap placed at intersec- surveyed cross sections if the toe stone is underwater. It
tion of channel bottom and side slope. is readily adaptable to emergency protection, where high
flow and the requirement for quick action make excava-
Trench-fill revetments on the Mississippi River have * tion impractical. Shape of the stone section before
successfully launched to protect for a vertical scour depth launching is not critical, but thickness of the section is
of up to 50 ft. On gravel bed streams, the use of launcha- important because thickness controls the rate at which
ble stone is not as widely accepted as in sand bed rock is released in the launching process. For gradual
streams. Problems with using launchable stone in some scour in regular bendways, the height of the stone section
gravel bed rivers may be the result of underestimating before launching should be from 2.5 to 4.0 times the bank
stone size, scour depth, or launchable stone volume protection thickness (T). For rapid scour in impinged
because the concept of launchable stone has been success- flow environments or in gravel bed streams, the stone
ful on several gravel bed rivers. section height before launching should be 2.5 to 3.0 T. In

3-10
 EM 1110-2-1601
Change 1
30 Jun 94

* prevent leaching of lower bank material through the

Table 3-2
Increase in Stone Volume for Riprap Launching Sections launched riprap. Launchable stone should have
D85/D15 ≥ 2. *
Volume Increase, Percent
Vertical Launch Dry Underwater
Distance, ft1 Placement Placement 3-12. Delivery and Placement

≤ 15 25 50 Delivery and placement can affect riprap design. See

EM 1110-2-2302 for detailed guidance. The common
> 15 50 75
methods of riprap placement are hand placing; machine
Note: placing, such as from a skip, dragline, or some form of
1
From bottom of launch section to maximum scour. bucket; and dumping from trucks and spreading by bull-
dozer. Hand placement produces the most stable riprap
revetment because the long axes of the riprap particles are
any case, the thinner and wider rock sections represented
oriented perpendicular to the bank. It is the most expen-
by the lower values of thickness have an apparent advan-
sive method except when stone is unusually costly and/or
tage in that the rock in the stream end of the before-
labor unusually cheap. Steeper side slopes can be used
launch section has a lesser distance to travel in the
with hand-placed riprap than with other placing methods.
launching process. Providing an adequate volume of
This reduces the required volume of rock. However, the
stone is critical. Stone is lost downstream in the launch-
greater cost of hand placement usually makes machine or
ing process; and the larger the scour depth, the greater the
dumped placement methods and flatter slopes more eco-
percentage of stone lost in the launching process. To
nomical. Hand placement on steep slopes should be con-
compute the required launchable stone volume for
* sidered when channel widths are constricted by existing
Method D, the following assumptions should be used:
bridge openings or other structures when rights-of-way are
costly. In the machine placement method, sufficiently
(1) Launch slope = 1V on 2H. This is the slope
small increments of stone should be released as close to
resulting from rock launched on noncohesive material in
their final positions as practical. Rehandling or dragging
both model and prototype surveys. Launch slope is less
operations to smooth the revetment surface tend to result
predictable if cohesive material is present, since cohesive
in segregation and breakage of stone. Stone should not be
material may fail in large blocks.
dropped from an excessive height or dumped and spread
as this may result in the same undesirable conditions.
(2) Scour depth = existing elevation - maximum
However, in some cases, it may be economical to increase
scour elevation.
the layer thickness and stone size somewhat to offset the
shortcomings of this placement method. Smooth, compact
* (3) Thickness after launching = thickness of the bank
riprap sections have resulted from compacting the placed
revetment T .
stone sections with a broad-tracked bulldozer. This stone
must be quite resistant to abrasion. Thickness for under-
* To account for the stone lost during launching and for
water placement should be increased by 50 percent to
placement underwater, the increases in stone volume listed
provide for the uncertainties associated with this type of
in Table 3-2 are recommended. Using these assumptions,
placement. Underwater placement is usually specified in
the required stone volume for underwater placement for
terms of weight of stone per unit area, to be distributed
vertical launch distance less than 15 ft = 1.5T times
uniformly and controlled by a “grid” established by
launch slope length
* shoreline survey points.
= 1.5T times scour depth times 5
Section V
Ice, Debris, and Vegetation
= 3.35T (scour depth)
3-13. Ice and Debris
Add a safety factor if data to compute scour depth are
unreliable, if cohesive bank material is present, or if
Ice and debris create greater stresses on riprap revetment
monitoring and maintenance after construction cannot be
by impact and flow concentration effects. Ice attachment
guaranteed. Guidance for a safety factor is lacking, so to
to the riprap also causes a decrease in stability. The Cold
some extent it must be determined by considering conse-
Regions Research Engineering Laboratory, Hanover, NH,
* quences of failure. Widely graded ripraps are recom-
should be contacted for detailed guidance relative to ice
mended because of reduced rock voids that tend to

3-11
EM 1110-2-1601 *
1 Jul 91

effects on riprap. One rule of thumb is that thickness Recognizing that vegetation is, in most instances, inimical
should be increased by 6-12 in., accompanied by appro- to riprap installations, planned use of vegetation with
priate increase in stone size, for riprap subject to attack riprap should serve some justifiable purpose, be accounted
by large floating debris. Riprap deterioration from debris for in capacity computations, be controllable throughout
impacts is usually more extensive on bank lines with the project life, have a strengthened riprap design that will
steep slopes. Therefore, riprapped slopes on streams with withstand the additional exigencies, and account for
heavy debris loads should be no steeper than 1V on 2.5H. increased difficulty of inspection.

3-14. Vegetation Section VI

Quality Control
The guidance in this chapter is based on maintaining the
riprap free of vegetation. When sediment deposits form 3-15. Quality Control
lowflow berms on riprap installations, vegetation may be
allowed on these berms under the following conditions: Provisions should be made in the specifications for samp-
roots do not penetrate the riprap; failure of the riprap ling and testing in-place riprap as representative sections
would not jeopardize project purposes prior to repairs; and of revetment are completed. Additional sample testing of
the presence of the berm and vegetation does not signifi- in-place and in-transit riprap material at the option of the
cantly reduce the discharge capacity of the project. For Contracting Officer should be specified. The primary
riprap areas above the 4 or 5 percent exceedence flow concern of riprap users is that the in-place riprap meets
line, consideration may be given to overlaying the riprap specifications. Loading, transporting, stockpiling, and
with soil and sod to facilitate maintenance by mowing placing can result in deterioration of the riprap. Coordi-
rather than by hand or defoliants. This may be par- nation of inspection efforts by experienced staff is neces-
ticularly appropriate for riprap protecting against eddy sary. Reference EM 1110-2-2302 for detailed sampling
action around structures such as gate wells and outlet guidance and required sample volumes for in-place riprap.
works in levees that are otherwise maintained by mowing.

3-12 *
 EM 1110-2-1601
1 Jul 91

Chapter 4 the sheet piling is set at an elevation required by the

Special Features and Considerations above-mentioned criteria. Plate 47 is used with the
known discharge to compute the energy head at 5dc
upstream of the structure. The head H on the structure
4-1. Sediment Control Structures
is determined from the energy equation and used with
Plate 46 to estimate the required derrick stone size. The
a. General. Two basic types of control structures
curves in Plates 29 and 30 should be used as guides in the
are used:
selection of riprap sizes for the less critical flow area.
(1) stabilizers designed to limit channel degradation
c. Drop structures.
and
(1) Description and purpose. Drop structures are
(2) drop structures designed to reduce channel slopes
designed to check channel erosion by controlling the
to effect nonscouring velocities.
effective gradient, and to provide for abrupt changes in
channel gradient by means of a vertical drop. They also
These structures also correct undesirable, low-water-
provide a satisfactory means for discharging accumulated
channel meandering. Gildea (1963) has discussed channel
surface runoff over fills with heights not exceeding about
stabilization practice in USAED, Los Angeles. Debris
5 ft and over embankments higher than 5 ft provided the
basins and check dams are special types of control struc-
end sill of the drop structure extends beyond the toe of
tures that are used to trap and store bed-load sediments.
the embankment. The hydraulic design of these structures
may be divided into two general phases, design of the
b. Stabilizers.
notch or weir and design of the overpour basin. Drop
structures must be so placed as to cause the channel to
(1) A stabilizer is generally placed normal to the
become stable. The structure must be designed to pre-
channel center line and traverses the channel invert.
clude flanking.
When the stabilizer crest is placed approximately at the
elevation of the existing channel invert, it may consist of
(2) Design rules. Pertinent features of a typical drop
grouted or ungrouted rock, sheet piling, or a concrete sill.
structure are shown in Plate 48. Discharge over the weir
The stabilizer should extend into or up the channel bank
should be computed from the equation Q = CLH3/2 ,
and have adequate upstream and downstream bed and
using a C value of 3.0. The length of the weir should
bank protection. Plate 44 illustrates the grouted stone
be such as to obtain maximum use of the available chan-
type of stabilizer used in USAED, Los Angeles. Stabiliz-
nel cross section upstream from the structure. A
ers may result in local flow acceleration accompanied by
trial-and-error procedure should be used to balance the
the development of scour holes upstream and downstream.
weir height and width with the channel cross section.
As indicated in Plate 44, dumped stone should be placed
Stilling basin length and end sill height should be deter-
to anticipated scour depths. Maximum scour depths usu-
mined from the design curves in Plate 48. Riprap
ally occur during peak discharges.
probably will be required on the side slopes and on the
channel bottom immediately downstream from the
(2) Laboratory tests on sheet piling stabilizers for the
structure.
Floyd River Control Project were made by the University
of Iowa for USAED, Omaha (Linder 1963). These stud-
d. Debris basins and check dams.
ies involved the development of upstream and down-
stream bed and bank riprap protection for sheet piling
(1) General. Debris basins and check dams are built
stabilizers in a channel subject to average velocities of
in the headwaters of flood control channels having severe
14 fps. The final design resulting from these tests is
upstream erosion problems in order to trap large bed-load
shown in Plate 45. Plate 46 is a general design chart giv-
debris before it enters main channels. This is done to
ing derrick stone size required in critical flow areas as a
prevent aggradation of downstream channels and deposi-
function of the degree of submergence of the structure.
tion of large quantities of sediment at stream mouths.
Plate 47 presents design discharge coefficients in terms of
Also, the passage of large debris loads through reinforced
the sill submergence T and critical depth dc for the
concrete channels can result in costly erosion damage to
channel section. Use of Plates 46 and 47 is predicated on
the channel. Such damage also increases hydraulic
the condition that the ratio T/dc is greater than 0.8. For
roughness and reduces channel capacity. A general sum-
smaller values the high-velocity jet plunges beneath the
mary of data on the equilibrium gradient of the deposition
water surface, resulting in excessive erosion. The top of
profile above control structures has been presented by

4-1
EM 1110-2-1601
1 Jul 91

Woolhiser and Lenz (1965). The principles of design and (e) An outlet tower and conduit through the embank-
operation of large debris basins as practiced by USAED, ment at the spillway for basin draining.
Los Angeles, have been presented by Dodge (1948).
Ferrell and Barr (1963) discuss the design, operation, and Plate 49 shows general design plans for a debris basin.
effects of concrete crib check dams used in the The basin shape, the inlets, and the outlet should be
Los Angeles County Flood Control District on small located so that the debris completely fills the basin before
streams. debris discharge occurs over the spillway.

(2) Debris storage. Debris basins, usually located (4) Design criteria. The slope of the upper surface
near canyon mouths at the upper end of alluvial fans, are of the debris deposit must be estimated to determine the
designed to settle out and provide storage space for debris proper basin shape and to estimate the total debris ca-
produced from a single major storm. In the Los Angeles pacity of the basin. A value of 0.5 times the slope of the
area, the debris basin design capacity has been based on natural debris cone at the basin site has been used for
100,000 cubic yards (cu yd) per square mile of drainage design. The basin side embankments should be of suffi-
area, or 62 acre-feet per square mile. This quantity was cient height and extend far enough upstream to confine
obtained as an envelope curve of observed debris pro- the maximum debris line slope projected upstream from
duction during the storm of 1938 (Dodge 1948). Later the spillway crest. The spillway should be designed to
estimates by Tatum (1963), taking into account factors pass the design flood discharge with the basin filled with
affecting debris production such as fire history of the debris. The tops of the basin embankments should pro-
area, indicated a value of about twice this amount. Debris vide 5 ft of freeboard with the foregoing conditions. The
storage in the basin is usually maintained by reexcavation design criteria for debris basins in the Los Angeles area
after a major storm period. The debris stored in the basin should be used only for general guidance because of large
after any one flood should not be allowed to exceed differences in geology, precipitation patterns, land use,
25 percent of the basin capacity. When permanent debris and economic justification in different parts of the coun-
storage is more economical than periodic excavation, the try. The following conditions are peculiar to the
average annual rate of debris accumulation multiplied by Los Angeles area:
the project life should be used for storage capacity. Data
from the Los Angeles County Flood Control District (a) Phenomenal urban growth in the desirable land
(Moore, Wood, and Renfro 1960) on 49 debris dams and area of the lower alluvial fans.
basins give a mean annual debris production of 5,500 cu
yd per square mile of drainage basin. This figure applies (b) Large fire potential.
in the Los Angeles and similar areas, and can be used to
determine the economic feasibility of long-term storage (c) Hot, dry climate over a large portion of the year
versus periodic debris removal. which inhibits vegetative growth.

(3) Debris basin elements. A debris basin consists (d) Sudden torrential rainfall on precipitous mountain
of five essential basic parts: slopes during a short rainy season.

(a) A bowl-shaped pit excavated in the surface of the (e) Unstable soil conditions subject to voluminous
debris cone. slides when saturated.

(b) An embankment, usually U-shaped in plan, con- Debris and sediment production rates vary throughout the
structed from pit material, located along the two sides and country depending on many factors, some of which are
the downstream end of the pit, and joining the hillside at controllable by man. Extensive construction, strip mining
each end where possible. operations, intensive agricultural use, and timber cutting
operations are only a few examples of land uses that can
(c) One or more inlet chutes at the upstream end of have a profound local effect on sediment production and
the pit, when necessary to prevent excessive streambed thus determine the type of sediment control necessary.
degradation upstream of the debris basin. Formulation of a sediment control plan and the design of
associated engineering works depend to a large extent on
(d) A broad-crested spillway at the downstream end local conditions.
of the basin leading to a flood control channel.

4-2
 EM 1110-2-1601
1 Jul 91

4-2. Air Entrainment S = sine of angle of chute inclination

a. General. Air entrainment should be considered in q = discharge per unit width of channel
the design of rapid- flow channels. The entrainment of
air may result in bulking of the flow and necessitate in- d. Design criteria. Use of Equation 4-2 or HDC
creased wall heights. Presently available data indicate 050-3 requires the assumption _that the experimental water
that appreciable air entrainment should not occur with flow depth dw in the term C = da/(da + dw) where da
Froude numbers less than about 1.6. is depth of air-water mixture, ft, is the same as the
theoretically computed flow depth. The Minnesota data
b. Early design criteria. The USAED, Sacramento, indicate that this assumption is valid only for small
developed the following equation based on data reported Froude numbers. For large Froude numbers, the theoreti-
by Hall (1943): cally computed depths for nonaerated flow were found to
be 50 to 75 percent greater than the observed experimen-
tal flow depth. For this reason and for convenience of
V2 (4-1) design, the Minnesota and Kittitas data have been com-
m
200gd puted and plotted in terms of the observed total flow
depth (air plus water) and the theoretical flow depth and
Froude number for nonaerated flow (Plate 50a). The
where resulting design curve has been extrapolated for low
Froude numbers and replotted as Plate 50b. This plate
m = air-water ratio should be used for air-entrained flows in flood control
channels. A comparison of HDC 050-3 and Plate 50b
V = theoretical average flow velocity indicates that this plate results in more conservative
without air design for low Froude numbers.

d = flow depth including air 4-3. Hydraulic Jump in Open Channels

The term V2/gd is the Froude number squared. Equa- a. General. Flow changes from the rapid to tranquil
tion 4-1 with minor differences in the definition of terms state will usually occur in the form of a hydraulic jump.
has been published by Gumensky (1949). The basic The hydraulic jump consists of an abrupt rise of the water
equation has been used extensively for design purposes in surface in the region of impact between rapid and tranquil
the past. flows. Flow depths before and after the jump are less
than and greater than critical depth, respectively. The
c. Modern investigations. The mechanics of zone of impact of the jump is accompanied by large-scale
self-aerated flow in open channels with sand grain sur- turbulence, surface waves, and energy dissipation. The
faces has been studied at the University of Minnesota by hydraulic jump in a channel may occur at locations such
Straub and Anderson (1960). The results of the as:
Minnesota tests have been combined with selected Kittitas
chute prototype data (Hall 1943) and published as HDC (1) The vicinity of a break in grade where the chan-
050-3. The chart includes the following suggested design nel slope decreases from steep to mild.
equation:
(2) A short distance upstream from channel constric-
tions such as those caused by bridge piers.
 
0.701 log10 
S  (4-2)
C 0.971
1/5  (3) A relatively abrupt converging transition.
q 
(4) A channel junction where rapid flow occurs in a
tributary channel and tranquil flow in the main channel.
where
(5) Long channels where high velocities can no
_ longer be sustained on a mild slope.
C = ratio of experimentally determined
air volume to air plus water volume

4-3
EM 1110-2-1601
1 Jul 91

b. Jump characteristics. (Equation 4-4 for rectangular channels). The location of

the jump is estimated by the sequent depths and jump
(1) The momentum equation for the hydraulic jump length. The mean location is found by making backwater
is derived by setting the hydrodynamic force plus momen- computations from upstream and downstream control
tum flux at the sections before and after the jump equal, points until Equation 4-3 or 4-4 is satisfied. With this
as follows: mean jump location, a jump length can be obtained from
Plate 51 and used for approximating the location of the
jump limits. Because of the uncertainties of channel
roughness, the jump should be located using practical
Q2 Q2 (4-3) limits of channel roughness (see paragraph 2-2c). A
A1y1 A2y2
gA1 gA2 trial-and-error procedure is illustrated on page 401 of
Chow (1959).

_ (2) The wall height required to confine the jump and

where y is the depth to the center of gravity of the the backwater downstream should extend upstream and
stream cross section from the water surface. For a rectan- downstream as determined by the assumed limits of chan-
gular channel the following jump height equation can be nel roughness. Studies also should be made on the height
obtained from Equation 4-3: and location of the jump for discharges less than the
design discharge to ensure that adequate wall heights
extend over the full ranges of jump height and location.
y2 1 2
1 8F1 1 (4-4)
y1 2 (3) In channels with relatively steep invert slopes,
sequent depths are somewhat larger than for horizontal or
mildly sloping channels and jump lengths are somewhat
smaller than those given in Plate 51. Peterka (1957) sum-
where the subscripts 1 and 2 denote sections upstream and marizes the available knowledge of this subject. This
downstream of the jump, respectively. Equation 4-3 also reference and HDC 124-1 should be used for guidance
gives good agreement for trapezoidal channels as shown when a jump will occur on channel slopes of 5 percent or
by tests reported by Posey and Hsing (1938). However, more.
flood channels should not be designed with jumps in
trapezoidal sections because of complex flow patterns and d. Undular jump. Hydraulic jumps with Froude
increased jump lengths. numbers less than 1.7 are characterized as undular jumps
(Bakhmeteff and Matzke 1936) (see Plate 52). In addi-
(2) The energy loss in the hydraulic jump can be tion, undulations will occur near critical depth if small
obtained by use of the energy equation and the derived disturbances are present in the channel. Jones (1964)
jump height relation (Chow 1959). This results in an shows that the first wave of the undular jump is consider-
equation that is a function only of the upstream Froude ably higher than given by Equation 4-4. The height of
number. The relations between the Froude number, the this solitary wave is given by
jump height (Equation 4-4), and the energy loss
(Equation 15-1, Brater and King 1976) are presented in
Plate 51. The relation between the Froude number and a 2
F1 1 (4-5)
the jump length, based on the data by Bradley and Peterka y1
(1957) for rectangular channels, is also presented in this
plate.
where a is the undular wave height above initial depth
c. Jump location. y1 . Additional measurements were also made by
Sandover and Zienkiewicz (1957) verifying Equation 4-5
(1) The location of the hydraulic jump is important and giving the length of the first undular wave. Other
in determining channel wall heights and in the design of measurements with a theoretical analysis have been re-
bridge piers, junctions, or other channel structures, as its ported by Komura (1960). Fawer (Jaeger 1957) has also
location determines whether the flow is tranquil or rapid. given a formula for the wavelength based on experimental
The jump will occur in a channel with rapid flow if the data; Lemoine (Jaeger 1957) used small-amplitude wave
initial and sequent depths satisfy Equation 4-3 theory to give the wavelength of the undular jump. The

4-4
 EM 1110-2-1601
1 Jul 91

results of these investigations are summarized in Plate 52, by Bowers (1950) indicate that a hydraulic jump may
which gives the undular jump surge height, breaking surge form in one or both of the inlet channels, depending on
height (Equation 4-4), and the wavelength of the first the flow conditions.
undular wave. Also shown in this plate is a relation given
by Keulegan and Patterson (1940) for the height of the (2) Wave conditions that may be produced by rapid
first undulation flow in and downstream of a typical junction are shown
in Plate 53. One area of maximum wave height can
occur on the side channel wall opposite the junction point
  and another on the main channel right wall downstream
a 3  y2 y1  (4-6)
2  y1 

y1 from the junction. Behlke and Pritchett (1966) have
conducted a series of laboratory tests indicating that wave
pileup against the channel walls can be up to 7 times the
initial depth with a flow Froude number of 4. The design
Experiment and theory indicate that the undular wave will of walls to contain these wave heights over long channel
begin to spill at the first crest when the Froude number distances is usually not economical. The practical remedy
exceeds about 1.28. Undulations persist, however, until is to reduce or minimize standing waves.
the Froude number exceeds about 3 (≈ 1.7). This is
the limit for breaking waves when Equation 4-4 gives a (3) Peak flows from the side channel may not occur
value of y2/y1 = 2 . Further configuration information on simultaneously with peak flows in the main channel.
undular jumps may be obtained from Figures 44, 45, and Laboratory tests by Behlke and Pritchett (1966) indicate
46 of USBR (1948). that occurrence of the design flow in one of the channels
with zero flow in the other can result in very high wave
e. Stilling basins. Stilling basin design for high pileup on the junction walls. Plates 54a and b show
Froude numbers is covered in EM 1110-2-1603. The maximum wave height as a function of upstream Froude
design of stilling basins in the range of Froude numbers number for conditions of zero flow in the side channel
from 1.0 to about 1.3 is complicated by undular waves and main channel, respectively. This plate demonstrates
that are dissipated only by boundary friction with increas- the need for keeping the angle of the junction intersection
ing distance downstream. This range of Froude numbers relatively small. The data are also useful in designing
should be avoided whenever possible because of flow wall heights; for example, the maximum wave pileup on
instability. The hydraulic jump with Froude numbers of the main channel wall would be greater than twice the
1.3 to 1.7 is characterized by breaking undulations with side channel flow depth for F2 = 3.0 , a junction angle of
very little energy dissipation (see Plate 51). Wall heights 15 deg, and no flow in the main channel.
in this range of Froude numbers should be designed to
contain waves up to the value given by the Keulegan and c. Wave height criteria. Behlke and Pritchett’s
Patterson (1940) limit. (1966) recommended criteria for the design of channel
junctions in rapid flow to minimize wave effects are listed
4-4. Open Channel Junctions below:

a. General. The design of channel junctions is com- (1) Enlarge the main channel below the junction
plicated by many variables such as the angle of intersec- apex to maintain approximately constant flow depths
tion, shape and width of the channels, flow rates, and type throughout the junction.
of flow. Appendix E presents a theoretical analysis,
based on the momentum principle, that can be used for (2) Provide equal water-surface elevations in the side
several types of open channel junctions. The design of and main channels in the vicinity of the junction.
large complex junctions should be verified by model tests.
(3) Ensure that the side channel wave originating at
b. Wave effects. the junction apex impinges on the opposite side channel
wall at its intersection with the enlarged main channel
(1) Standing waves (Ippen 1951) in rapid flow at wall.
open channel junctions complicate flow conditions. These
waves are similar to those created in channel curves de- (4) Provide tapered training walls between the main
scribed in paragraph 2-4, and may necessitate increased channel and the side channel flows.
wall heights in the vicinity of the junction. The studies

4-5
EM 1110-2-1601
1 Jul 91

(5) Ensure that maximum wave heights occur with width is required downstream from the confluence, the
maximum flows. Plate 55 illustrates typical design transition should be made gradually.
examples for rectangular and trapezoidal channels using
these criteria. Important junctions in rapid flow designed e. Design procedure. The design procedure for the
to reduce wave effects should be model tested at all prob- typical open channel confluence shown in Plate 56 in-
able flow combinations as well as at design flow. volves the following steps:

d. Confluence design criteria. (1) Determine side-channel requirements relative to

discharge, alignment, and channel size.
(1) The results of several model studies in USAED,
Los Angeles, indicate that some general guides can be (2) Select junction point to obtain an entrance angle
adopted for the design of confluence junctions. Gildea less than 12 deg. This angle requirement may necessitate
and Wong (1967) have summarized some of these criteria: a long, spiral curve for the side channel upstream from
the junction.
(a) The design water-surface elevations in the two
joining channels should be approximately equal at the (3) Determine the increase of channel width ∆b3
upstream end of the confluence. from the Q2/Q3 ratio curve in Plate 57. Compute the
required downstream channel width b3 = b1 + ∆b3 and
(b) The angle of junction intersection should be the confluence width bc = b1 + 2∆b3.
preferably zero but not greater than 12 deg.
(4) Make the confluence layout on a straight-line
(c) Favorable flow conditions can be achieved with basis by setting the main channel walls parallel to and at
proper expansion in width of the main channel below the distances of (1/2)b3 and bc - (1/2)b3 from the center line
junction. as shown in Plate 56a.

(d) Rapid flow depths should not exceed 90 percent (5) Connect the left walls of the side and the main
of the critical depth (Froude number should be greater channels by a curve determined by the apex angle θ and
than 1.13) to maintain stable rapid flow through the junc- a radius rL given by
tion (paragraph 2-2d(1)).

(2) Model tests of many confluence structures indi- 4V 2b2 (4-7)

rL 400
cate very little crosswave formation and turbulence at the gy
junction if these criteria are followed. Moreover, experi-
ence has shown that the momentum equation approach
given in Appendix E can be used for junctions involving
small angles and equal upstream water-surface elevations. Equation 4-7 results from a study of a number of con-
fluences built by USAED, Los Angeles. The term
(3) Typical confluence layouts model tested by (4V2b2)/gy is the same as that used in Equation 2-34.
USAED, Los Angeles, and proven to have good flow
characteristics are shown in Plate 56. The design with the (6) Make the right wall of the side channel concen-
offset in the main channel center line is normally used tric with the left wall and locate the junction intersection
(Plate 56a). When the main channel center-line alignment point. The right wall radius rR is given by
cannot be offset, a layout with a transition on the wall
opposite the inlet side should be used (Plate 56b). The
rR rL b2 (4-8)
proper amount of expansion in the main channel down-
stream of the confluence is very important in maintaining
good flow conditions. Plate 57 gives the USAED, Los
Angeles, empirical curve for the required increase in (7) Determine the average depth of flow at midpoint
channel width, ∆b3 , as a function of the discharge ratio. of the confluence by the momentum method (Appendix E)
If the junction angle is zero, the width of the channel at assuming bm = (1/2)(b1 + b2 + bc) .
the confluence will be equal to the sum of the widths of
the main and side channels plus the thickness of the (8) Set the side-channel invert elevation so that the
dividing wall between the channels. If a reduction in design water-surface levels in both channels approximate

4-6
 EM 1110-2-1601
1 Jul 91

each other. A stepped invert in either of the channels (c) Determine the side-channel flow depth d at the
may be required. upstream end of the spillway.

(9) Determine the length of transition and invert (d) Set the side-channel invert so that the spillway
slope required to reduce the channel width from bc to b3 approach depth is equal to d - H .
without exceeding the criterion y/yc ≤ 0.90 in the transi-
tion. Convergence rates should be in agreement with (e) Determine the side-channel convergence required
those recommended in paragraph 2-4. to maintain a constant flow depth in the side channel
behind the spillway. This should result in a reasonably
f. Side drainage inlets. Flow disturbances occur constant unit discharge over the spillway equal to that
where storm drains or industrial waste lines discharge into computed by the equation in (b) above.
flood control channels, commonly referred to as "inlets."
Small side-drainage flows are commonly conveyed in a (f) Plot the computed side-channel alignment points
pipe storm drain system. Criteria for box and pipe culvert obtained from step (e) on the channel plan and connect
inlet design are given in h below. Economical design for them by a smooth curve or straight line to intersect the
intermediate tributary flows normally requires free surface main channel wall so that the side channel has a mini-
structures. A side-channel spillway type of inlet for this mum width of 2 ft behind the spillway.
range of discharge has been developed by USAED, Los
Angeles, which reduces disturbances to a minimum in the (g) Adjust the side-channel convergence and repeat
main channel. This type of junction is described in g step (e) if the spillway length in step (f) does not approxi-
below. The conventional confluence structure described mate that determined in step (b).
in d above should be used for large tributary discharges.
h. Box and pipe culvert inlets. Gildea and Wong
g. Side-channel drainage inlet. (1967) have determined design criteria for pipe inlets.
The variables to be considered in the design are width of
(1) The side-channel spillway type of drainage inlet the main channel, angle of entrance of the storm drain,
was developed and model tested by USAED, Los Angeles size of the storm drain, volume and velocity of flow, and
(1960b). The recommended structure consists of a com- elevation of the storm drain with respect to the channel
mon wall between the side channel and the main channel. bottom. Model tests (USAED, Los Angeles, 1960b,
A weir notched in this wall allows the tributary flow to 1964) have shown that flow disturbances in the main
enter the main channel with minimum disturbance. A channel are minimized when side-drain openings are small
typical design of this type of structure is illustrated in and side- drainage flows are introduced reasonably par-
Plate 58. A small drain should be placed at the lowest allel to the main flow. The following criteria should be
point of the side channel. The objective of this design is used for design:
to discharge the side flow with reduced velocity into the
main channel gradually over a relatively long spillway (1) The maximum angle of entrance for side culverts
inlet. Model tests (USAED, Los Angeles, 1960b) indicate should be:
that this effectively reduces wave action and disturbances
in the main channel for all flow combinations. Satisfac- (a) 90 deg for diameters of 24 in. or less.
tory operation may require periodic sediment removal
from behind the weir. (b) 45 deg for diameters from 24 to 60 in.

(2) The procedure for designing the side-channel (c) 30 deg for diameter 60 in. or greater.
spillway inlet structure follows:
(2) The culvert invert should be placed no more than
(a) Set the spillway crest 0.5 ft above the parallel to 18 in. above the main channel invert to give the maxi-
the design watersurface level in the main channel. mum submergence practicable.

(b) Determine the required length L of the crest by (3) Automatic floodgates or flap gates should be
the equation, L = Q/(CH3/2) , so that the maximum H is installed when damage from backflooding from the main
not greater than 1.5 ft with critical depth over the crest channel would exceed that resulting from local pondage
C equal to 3.097. caused by gate operation. These gates should be recessed

4-7
EM 1110-2-1601
1 Jul 91

to prevent projecting into the main channel flow when in structure. Scale ratios of 1:15 to 1:30 are usually
a full-open position. Head loss coefficients for flap gates employed where supercritical flow wave problems are
are given in HDC 340-1. involved. They are also used for sectional models of drop
structures, spillways, etc. The smaller scale ratios (1:30
4-5. Hydraulic Model Studies to 1:70) are used for general model studies where long
channel lengths are reproduced. The accuracy of possible
a. General. The use of hydraulic models has model construction and flow measurements may control
become a standard procedure in the design of complex the permissible scale ratios. Most models of channels are
open channels not subject to analytical analyses or for generally built to give depths of flow about 0.5 ft or more
which existing design criteria based on available model and channel widths of about 1 to 2 ft. The most common
and field tests are inadequate. Hydraulic models afford a scale ratios used by the USAED, Los Angeles, Hydraulic
means of checking performance and devising modifica- Laboratory for channel model studies are from 1:25 to
tions to obtain the best possible design at minimum cost. 1:40.
Model tests should be used to supplement but not replace
theoretical knowledge, good judgment, and experience of c. Model roughness. Turbulent flow will prevail
the design engineer. They often indicate design changes with model channel velocities and depths commonly used
that save substantial amounts in construction costs as well in testing. In most cases, the channel flow is
as effect improvements in operation. Model tests of large rough-turbulent or nearly so; therefore, hydraulic resis-
flood control channels are generally desirable where sup- tance is determined primarily by the relative size of the
ercritical flow results in standing waves and other major roughness elements. However, the model Reynolds num-
disturbances in channels containing junctions, transition ber will always be smaller than the prototype, and this
structures, alignment curvature, multiple bridge piers, or will to some extent cause scale distortion of certain phe-
stilling basins. nomena such as zones of separation, wave dissipation,
flow instability, and turbulence in the model. Particular
b. Model design. care should be taken in interpreting those effects that are
known to be strongly dependent on viscous forces.
(1) The theory of model design is treated in
EM 1110-2-1602 and other publications (Rouse 1950, d. Slope distortion. An empirical equation of the
Davis and Sorenson (1969), American Society of Civil Manning type may be used to give the required model
Engineers (ASCE) 1942). For open channel models, the roughness (Rouse 1950) for large-scale models where
gravity force will dominate the flow and similitude will fully rough-turbulent flow prevails. This condition is
require equality of Froude number in the model and pro- expressed by the equation
totype. The Froudian scale relations (model-to-prototype)
in Table 4-1 apply to undistorted models. The length ratio
Lr is the model-to- prototype ratio Lm/Lp . These nr Lr
1/6 (4-9)
transfer relations are based on equal force of gravity and
density of fluid in model and prototype. The procedure
for initiation of model studies is discussed in
EM 1110-2-1602. If this roughness criterion cannot be fulfilled, slope ad-
justment or distortion must be applied to the model so
(2) Model scale ratios for flood control channels that prototype flow conditions can be simulated in the
have ranged from 1:15 to 1:70, depending on the type of model. The amount of additional slope required is given
problem being studied, the relative roughness of the by the equation (Rouse 1950)
model and prototype, and the size of the prototype

Table 4-1
Scale Relations
Manning’s
Length Area Volume Time Velocity Discharge n

2 3 1/2 1/2 5/2 1/6

Lr Lr Lr Lr Lr Lr Lr

4-8
 EM 1110-2-1601
1 Jul 91

2 2/3
nr Rr
Sr (4-10) nr (4-12)
1/3 1/2
Lr Lr

Equation 4-10 applies only when the model and prototype For a wide channel Equation 4-12 reduces to
channels are geometrically similar in cross section. With-
out slope distortion (Sr = 1), this equation would reduce to 2/3
Equation 4-9. yr
nr (4-13)
1/2
Lr
e. Scale distortion.

(1) Distorted scales are generally used in models of

river channels, floodways, harbors, and estuaries.
Movable-bed models are distorted in order to ensure the The required roughness in the model can be computed by
movement of particle-size bed material under model flow Equation 4-12 and used as a guide in designing the
conditions. Flood control projects for the improvement of model. Distorted models should be verified using mea-
river channels through urbanized areas often require the sured field data or computed prototype data prior to test-
reproduction of long channel lengths and wide floodway ing of improvement plans. Flood control channel models
widths. Most such channels have mild slopes and the should be built to as small a distortion as is economically
flows are tranquil at very low Froude numbers. In order feasible. A distortion of 3 or less is desirable, but de-
to fit this type of model in a reasonably economical pends to some extent on the type of information needed
space, the horizontal scale ratio has to be limited and from the model study. It may sometimes be economically
vertical scale distortion selected to give measurable depths feasible to divide a long channel study into several prob-
and slopes as well as to ensure turbulent flow in the lem areas and model each one independently. In this
model. The use of distorted models should be generally manner different scales could be used as required by the
limited to problems involving tranquil flows. A number problem to be studied in each reach.
of reports (USAEWES 1949a, 1949b, 1953) have been
published that illustrate the application of distorted models f. Movable-bed models. Open channel studies in-
for the solution of complex local flood protection prob- volving problems of sediment erosion, transportation, or
lems and channel improvements. deposition require a bed of sand or other material that will
move when subjected to flow. Rouse (1950), Davis and
(2) The scale relations for distorted models are given Sorenson (1969), and ASCE (1942) give considerable
in ASCE (1942). If the bed slope ratio is made equal to detail on design, construction, verification, and use of
the energy slope ratio, the slope ratio will also be equal to movable-bed models. Qualitative indication of bed move-
the amount of model distortion. ment has been used in flood control channel models for
design purposes. For example, the effectiveness of a
hydraulic jump to dissipate energy is often obtained
yr
Sr (4-11) through the relative extent of downstream scour. The sta-
Lr bility of riprap protection can also be obtained from
model studies. A typical example of a study to determine
the relative scour and design of riprap protection at inlet
and outlet channels is given in USAED, Los
where yr is the vertical scale ratio and Lr is the hori- Angeles (1960a).
zontal scale ratio, model to prototype. The Manning
equation can then be used to obtain a roughness criteria
for model design (Rouse 1950).

4-9
 EM 1110-2-1601
Change 1
30 Jun 94

* Chapter 5 channels to stream power. These data indicate that a sig-

Methods for Predicting n Values for the nificant change can occur in n values as the stream bed
changes from ripples to dunes to plane bed to antidune.
Manning Equation
Subsequently, work by Limerinos (1970) and Brownlie
(1983) provided regression equations for calculating bed
roughness in mobile boundary channels. Note that chan-
5-1. Introduction nel bed roughness is just one component of the total
n value for a reach.
This chapter describes the prediction of the total
Manning’s roughness coefficient (n value) for a reach by e. Compositing. The procedure for combining dif-
establishing physically based component parts and deter- ferent roughnesses across a section into a single value for
mining the contribution from each. The following compo- hydraulic computations is called compositing. The com-
nent parts were selected: bed roughness, bank roughness, posited value may change if a different method for com-
surface irregularities, obstructions, vegetation roughness, positing is chosen. Therefore, the handbook methods are
and expansion/contraction losses. probably more dependable as sources of n values than the
analytical methods because the compositing is included in
5-2. Approach the field observation.

Hydraulic roughness is a major source of uncertainty in 5-3. Hydraulic Roughness by Handbook Methods
water surface profile calculations. Field data at each
project are required to confirm selected values. When Arcement and Schneider (1989) summarize the state of
field data are not available, the traditional approach is to the art in selecting n values for natural channels and flood
use handbook methods or analytical methods to predict plains. This work was performed for the U.S. Department
the hydraulic roughness values. of Transportation and subsequently will be called the
USDT method in this chapter. The basic approach fol-
a. Handbook method. In this approach the engineer lows that proposed by Cowan (Chow 1959):
uses “calibrated photographs” and other subjective meth-
ods to associate hydraulic roughness values with condi- n (nb n1 n2 n3 n4) m (5-1)
tions observed and anticipated in the project reach. Chow
(1959) and Barnes (1967) are the dominant sources of
calibrated photographs. More recently, Arcement and where
Schneider (1989) extended the work to include flood-
plains. Other sources, like hydraulics and agricultural nb = base n value
handbooks, add variation but not much additional insight.
n1 = addition for surface irregularities
b. Analytical methods. A second approach for
predicting roughness coefficients is to relate hydraulic n2 = addition for variation in channel cross section
roughness to the effective surface roughness and irreg-
ularity of the flow boundaries. This approach is called n3 = addition for obstructions
analytical methods in this chapter. The classic example is
the Moody-type diagram for hydraulic roughness in open n4 = addition for vegetation
channel flow (Plate 3). The procedure shown in para-
graph 2-2c is still the state of the art in n values for m = ratio for meandering
concrete-lined channels. It is based on the Keulegan
equations for velocity distribution (Chow 1959). The 5-4. Base n Values (nb) for Channels
Iwagaki relationship has been included in the determina-
tion of the coefficients for the roughness equations. On page 4 of their report, Arcement and Schneider state,
“The values in [their] Table 1 for sand channels are for
c. Grass-lined channels. Manning’s n values for upper regime flows and are based on extensive laboratory
grass-lined channels were reported by the Soil Conserva- and field data obtained by the U.S. Geological Survey.
tion Service (Chow 1959). When using these values, a check must be made to ensure
that the stream power is large enough to produce upper
d. Mobile boundary channels. Simons and
Richardson (1966) related bed forms in mobile boundary *

5-1
 EM 1110-2-1601
Change 1
30 Jun 94

* regime flow.” Although the base n values given in values to use. Scobey (Chow 1959) proposed increasing
Table 5-1 for stable channels are from verification studies, the n value by 0.001 for each 20 degrees of curvature.
the values have a wide range because the effects of bed Chow suggested that should not exceed a total of 0.002
roughness are extremely difficult to separate from the even in flumes having pronounced curvature.
effects of other roughness factors. The choice of n values
from Table 5-1 will be influenced by personal judgment a. Effective surface roughness height ks. For the
and experience. The n values for lower and transitional design of concrete channels, Corps of Engineers values
regime flows are much larger generally than the values for ks are shown in Chapter 2 (Table 2-1). Chow (1959)
given in Table 5-1 for upper regime flow. Also, the gives a table of ks values (Table 8-1) for other boundary
vegetation density method of Petryk and Bosmajian materials such as ks for natural rivers. Please note that, at
(1975) is presented for the vegetation component n4. this point in time, the profession has not adopted tables of
Although the work was published in the mid-1970’s, it ks values as they have Manning’s n values. Moreover,
has not received widespread attention in the profession. It there is no generally accepted technique for measuring
has considerable appeal as a design procedure, however, this property geometrically. Therefore, the use of
and deserves additional evaluation. Table 8-1 is discouraged. Instead, use the Strickler or the
Keulegan equations and calculate ks from available
a. Example. Figure 5-1 is the proposed design for a sources of Manning’s n value. (Note: These equations
levee project in which the sponsor proposes vegetation do not necessarily give the same results.)
along the project. The hydraulic roughness values for this
section are estimated from several different handbook b. Relative roughness. Relative roughness refers to
sources in Tables 5-1 and 5-2. Note that handbooks the ratio of the effective surface roughness height, ks to
divide n values into two categories: channel bed and the hydraulic radius R. The relative roughness parameter
bank and flood plains. is R/ks.

b. Sensitivity of calculations to n values. The cal-

c. Strickler equation, rigid bed. The Strickler func-
culated water depth is shown in Table 5-3 using the mean
tion (Chow 1959) is shown in Figure 5-2. Notice that the
values of both channel and overbank roughness. The
effective surface roughness height ks is correlated with the
mean values are considered to be the best estimate,
D50 of the bed sediment in this figure. However, ks can
statistically.
be correlated with other measures of the surface
roughness depending on what is representative of the
Both n values were increased by adding their standard
surface roughness height of the boundary materials. For
deviation. The resulting water surface elevation increased
example, riprap research at WES has shown that the
about 0.7 ft, from 9.4 ft to 10.1 ft. This standard devia-
Strickler equation (Equation 5-2) will give satisfactory
tion in n values is really quite small. However, it demon-
n values when ks is taken to be the D90 of the stone.
strates how sensitive water depth is to n value.

n
1/6
C ks (5-2)
5-5. Hydraulic Roughness by Analytical Methods

Investigators continue to explore physically based

where
hydraulic roughness equations. These are the methods in
which hydraulic roughness is calculated from the effective
surface roughness ks. The new Hydraulic Design Package C = 0.034 for riprap size calculations where ks = D90
(SAM), under development at the U.S. Army Engineer
Waterways Experiment Station (WES) (Thomas et al., in = 0.038 for discharge capacity of riprapped
preparation), offers nine analytical methods for n values channels where ks = D90
(Table 5-4). None of the n value equations account for
momentum or bend losses. Presently, the only technique = 0.034 for natural sediment where ks = D50
for bend losses is to increase the n values by a factor. (Chow 1959)
Cowan (Chow 1959) proposed a multiplier in Equa-
tion 5-1, and both Chow and the USDT report suggest *

5-2
 EM 1110-2-1601
Change 1
30 Jun 94

*
Table 5-1
Hydraulic Roughness, Channel Bed and Banks
Reference m nb n1 n2 n3 n4 n

USDT (Arcement and Schneider 1989), 1.0 0.024 0.002 0.002 0.001 0.005 0.034
pp 4 & 7
Barnes (1967), p 78 - 0.037 - - - - 0.034
Chow (1959), p 109, Table 5-5, 1.0 0.024 0.005 0.0 0.0 0.00 0.034
Fine Gravel
Chow (1959), p 112, Table 5-6, D-1a3 - 0.040 - - - - 0.040
Chow (1959), p 120, Figure 5-5(14) - 0.030 - - - - 0.030
Brater and King (1976), p 7-17, - 0.035 - - - - 0.035
Natural

Mean 0.035
Standard deviation 0.003

Note:
n = (nb + n1 + n2 + n3 + n4)m

where

nb = base n-value
n1 = addition for surface irregularities
n2 = addition for variation in channel cross section
n3 = addition for obstructions
n4 = addition for vegetation
m = ratio for meandering

Figure 5-1. Design cross section

5-3
 EM 1110-2-1601
Change 1
30 Jun 94

*
Table 5-2
Hydraulic Roughness, Floodplain
Reference nb n1 n2 n3 n4 n

USDT (Arcement and Schneider 1989), 0.028 0.010 - 0.012 0.050 0.100
pp 4 & 9
Barnes (1967), None Given - - - - - -
Chow (1959), p 113, Table 5-6, D-2c5 0.100 - - - - 0.100
Chow (1959), p 123, Figure 5-5(23) 0.125 - - - - 0.125
Brater and King (1976), None Given - - - - - -

Mean 0.108
Standard deviation 0.012
Note: Same n value equation as channel bed and banks.

Table 5-3 Table 5-4

Sensitivity of Depth to n Value n Value Equations and Compositing Methods in SAM
n Value n Value Methods for
Flood- Water Equations Compositing
Case Channel plain Surface
Manning’s n Alpha Method
Keulegan Equal Velocity Method
Mean 0.035 0.108 9.4
Strickler Total Force Method
+1 Standard Deviation 0.038 0.120 10.1
Limerinos Total Discharge Method
Brownlie
Grass E1
Grass D1
Grass C1
Grass B1
Grass A1
1
Note: Grass type described in Table 5-7.

Figure 5-2. The Strickler function (Chow 1959) (courtesy of McGraw-Hill Book Company, Inc.)

5-4
 EM 1110-2-1601
Change 1
30 Jun 94

* d. Keulegan equations, rigid bed. The procedure in e. The Iwagaki relationship.

Chapter 2 is still the state of the art in n values for rigid
boundary channel design. It is a relative roughness (1) Chow presents Keulegan’s equation for the
approach based on the Keulegan equations for velocity average flow velocity V in the following form
distribution (Chow 1959). Keulegan classified flow types
as hydraulically smooth flow, hydraulically rough flow,   
U 6.25 5.75 log10  
and a transition zone. His equations, presented in Chap- R (5-4)
V
ter 2 and repeated as follows, are written in terms of the  k
 s 
Chezy coefficient because of the simpler powers involved.
The conversion to Manning’s n value follows. where

(1) The equation for fully rough flow is

U* = boundary shear velocity gRS
 12.2 R  (2-6 bis)
C 32.6 log10  
 k  g = acceleration of gravity

S = slope
(2) For smooth flow the equation is
6.25 = coefficient for fully rough flow
 
 5.2 Rn  (2-5 bis)
C 32.6 log10   (2) Substituting a variable, Ar, for the constant, 6.25,
 C  substituting the Chezy equation for velocity, and substi-
tuting gRS for U* gives
(3) The equation showing the relationship of n value
and Chezy C is (see Equation 2-4)  
5.75 log10  
V C R (5-5)
Ar
U g  ks 
1.486 1/6 (5-3)
n R
C
  
g Ar 5.75 log10  
R (5-6)
C
where   ks 

Rn = Reynolds number The form shown in Chapter 2 can be written as follows:

= 4RV/ν  
 A gr   (5-7)
C 32.6 log10 10 32.6  
R
where   k 
  s 
V = average flow velocity
where Ar is the Iwagaki coefficient for rough flow.
ν = kinematic viscosity of water
From Keulegan’s study of Bazin’s data, the value of
and 32.6, 12.2 and 5.2 are empirical coefficients deter- Ar was found to have a wide range, varying from
mined from laboratory experiments. These equations, 3.23 to 16.92. Thus, a mean value of 6.25 for Ar
when graphed, produce a Moody-type diagram for open may be used.
channel flow (Plate 3). *

5-5
 EM 1110-2-1601
Change 1
30 Jun 94

* “A further study was made by Iwagaki on experimen- where F is the Froude number. Data ranged from 0.2 <
tal data obtained from many sources. The results of F < 8.0.
the study have disclosed that resistance to turbulent
flow in open channels becomes obviously larger than (2) Using an equation of the same form, the rela-
that in pipes with increase in the Froude number. tionship for As is
Iwagaki reasoned that this is due to the increased
instability of the free surface at high Froude numbers” As 24.739 log10 (F 10) 29.349 (5-10)
(Chow 1959, p 204).

(3) The Iwagaki relationship is shown in Figure 5-3. (3) When the values of Ar and As are 6.2411 and
3.25, the coefficients in the roughness equations are 12.2
and 5.2, respectively. These are the values shown in
Equations 2-5 and 2-6. Using Equations 5-9 and 5-10,
those values correspond to Froude numbers of 1.88 and
1.35, respectively.

g. Transition zone. The limit of the fully rough

zone is

Rn / C
> 50 (5-11)
R / ks

The roughness equation in the transition zone is a com-

bination of the equations for smooth and fully rough flow
as follows:

 
 4C ks 
C 32.6 log10   (5-12)
Figure 5-3. The Iwagaki relationship (Chow 1959)
 A g
s
Ar g 
(courtesy of McGraw-Hill Book Company, Inc.)  g R 10 32.6 R 10 32.6 
 n 
(4) The comparable form of the equation for smooth
flow is h. Comparison of n-values, from Strickler and
Keulegan equations. Table 5-5 is a comparison of
 As g   n values calculated by the Strickler and Keulegan equa-
 32.6  g Rn  (5-8) tions. Flow is fully rough. Notice the Strickler equation
C 32.6 log10 10  
  4C  uses the effective surface roughness height ks, and not
relative roughness. Therefore, the n value does not vary
with hydraulic radius R. On the other hand, the Keulegan
where As is the Iwagaki coefficient for smooth flow. equation uses relative roughness, and that requires both ks
and R. The constant in the Strickler equation, 0.034, is
f. Ar and As coefficients. that recommended by Chow (1959). The resulting
n values match the Keulegan results adequately. For
(1) The Ar and As coefficients are shown graphically example, the ks for concrete is 0.007. That converts to an
in Figure 5-3, but the equations for the curves were not n value of 0.015 using Strickler and to 0.014-0.018 using
provided. It can be shown that the equation for Ar is of Keulegan.
the form

Ar 27.058 log10 (F 9) 34.289 (5-9)

*

5-6
 EM 1110-2-1601
Change 1
30 Jun 94

*
Table 5-5
n Values Calculated by Strickler and Keulegan Equations

Effective Keulegan Equation

Roughness R, ft

Strickler
ks, mm ks, ft n = 0.034 . ks1/6 F 1 5 10 20 50

8 0.012 0.013 0.014 0.015 0.016

0.10 0.0003281 0.009 1.88 0.010 0.011 0.012 0.013 0.014
0.2 0.009 0.011 0.011 0.012 0.013

8 0.017 0.017 0.018 0.019 0.020

1.00 0.003281 0.013 1.88 0.013 0.014 0.015 0.015 0.017
0.2 0.012 0.013 0.014 0.015 0.016

8 0.019 0.019 0.020 0.020 0.021

2.13 0.007 0.015 1.88 0.014 0.015 0.016 0.017 0.018
0.2 0.013 0.015 0.015 0.015 0.018

8 0.026 0.025 0.025 0.025 0.026

10 0.03281 0.019 1.88 0.018 0.018 0.019 0.019 0.020
0.2 0.016 0.017 0.017 0.018 0.019

8 0.049 0.037 0.035 0.034 0.033

64 0.20997 0.026 1.88 0.026 0.024 0.024 0.025 0.025
0.2 0.022 0.022 0.022 0.022 0.023

8 0.060 0.042 0.039 0.037 0.036

100 0.3281 0.028 1.88 0.029 0.026 0.026 0.026 0.027
0.2 0.024 0.023 0.023 0.024 0.024

8 0.084 0.048 0.043 0.041 0.039

152.4 0.500 0.030 1.88 0.033 0.029 0.028 0.028 0.028
0.2 0.027 0.025 0.025 0.025 0.026

8 ____ ____ 0.092 0.073 0.061

1,000 3.2808 0.041 1.88 ____ ____ 0.043 0.040 0.039
0.2 ____ ____ 0.036 0.034 0.034

Note:
C 32.6 log10 Coef2 R/ks

g Ar/32.6
Coef2 10

Ar 27.058 log10 F 9 34.289

5-7
 EM 1110-2-1601
Change 1
30 Jun 94

* i. Bed roughness in mobile boundary streams. j. Limerinos n-value predictor, mobile bed.

(1) In mobile boundary channels the bed roughness is (1) Limerinos developed an empirical relative rough-
composed of grain roughness and form roughness. The ness equation for coarse, mobile bed streams using field
grain roughness refers to the effective surface roughness data (Limerinos 1970). He correlated n values with
height of the mixture of sediment particles on the stream- hydraulic radius and bed sediment size. The following
bed. Form roughness refers to bed features described as equation resulted:
ripples, dunes, transition, plain bed, standing waves, and
antidunes. These bed features, called bed forms, are 0.0926 R 1/6
grouped into the general categories of lower regime, tran- n
  (5-13)
2.0 log10  
sitional, and upper regime. R
1.16
d
 84 
(2) Regime, in this usage of the term, does not refer
to whether the flow is sub- or supercritical. The Froude where
number may remain less than 1, and the bed regime may
still shift from lower to upper and back. Neither does it n = Manning’s n value. Data ranged from 0.02 to
refer to channel dimensions, flow velocity, nor slope. It 0.10.
is simply the category of bed forms that are contributing
to the hydraulic roughness. However, the amount of R = hydraulic radius, ft. Data ranged from 1 to 6 ft.
hydraulic loss produced by bed form roughness may
exceed that produced by grain roughness. Therefore, it d84 = the particle size, ft, for which 84 percent of the
cannot be ignored. sediment mixture is finer. Data ranged from 1.5 to
250 mm.
(3) The significant difference between mobile bound-
ary streams and rigid boundary streams is in the require- (2) Data were from relatively wide, straight streams
ment to predict when the bed forms change from one having a simple trapezoidal shape and no overbank flow.
regime to another. It seems to be related to flow velocity, There was very little increase in width with depth, and the
flow depth, water temperature, and effective sediment banks were stable. Irregularity was minimal. The amount
particle size. of vegetation on the bed and banks was negligible.

(4) Two functions are presented in this chapter for (3) Grain sizes in Limerinos’s data ranged from very
calculating n values in mobile boundary channels: coarse sand to large cobbles. The objective was to select
Limerinos (1970) and Brownlie (1983). However, only field sites at which the bed forms would not change with
the Brownlie method includes predicting the change from flow hydraulics during the measurement. Consequently, it
one bed regime to the other. These relationships are follows that this equation is applicable to gravel/cobble
described in more detail in the following paragraphs. bed streams and to bed regimes similar to those found in
such streams.
(5) It is important to establish which portion of the
channel cross section is bed and which is bank because (4) N values predicted with the Limerinos equation
the bed roughness predictors apply only to the channel are sufficiently larger than those predicted by the Strickler
bed. That is, typically the vegetation roughness and bank equation to indicate that some loss other than grain rough-
angle do not permit the bed load to move along the face ness must have been present. However, the Limerinos
of the banks. Therefore, the Limerinos and Brownlie equation is not applicable to lower regime flow nor does
n value equations should not be used to forecast bank it forecast the transition between upper and lower
roughness. regimes.

(6) On the other hand, the point bar is a natural (5) Burkham and Dawdy (1976) showed the
source-sink zone for sediment transport. Consequently, it Limerinos equation could be used in sand bed streams
is a location at which the Limerinos and Brownlie equa- provided the regime was plain bed. In that analysis they
tions apply. *

5-8
 EM 1110-2-1601
Change 1
30 Jun 94

* extended the range of the relative roughness parameter as l. The Brownlie bed-roughness predictor, mobile
follows: bed.

R (1) In sediment transport calculations it is important

600 < < 10,000 to link n values to the bed regime. This is particularly
d84
true when hydraulic conditions shift between upper
regime and lower regime flow. There are several
k. Comparison of Strickler and Limerinos n values. methods in Vanoni (1975) that express n value in terms of
sediment parameters, but Brownlie (1983) is the only
(1) Table 5-6 shows n values calculated by the method that calculates the transition. This method post-
Strickler and the Limerinos equations. For a hydraulic dates Vanoni (1975).
radius of 1 ft, the Limerinos values are higher than
Strickler’s by 15 to 57 percent. (2) Brownlie sought to reconstitute the most funda-
mental process--the discontinuity in the graph of hydraulic
(2) Furthermore, for ks up to about 10 mm the radius versus velocity (Figure 5-4). In the process of this
Limerinos n values increase with depth, which is the same research, he collected the known sediment data sets--77 in
trend as seen in the Keulegan n values in Table 5-5. all, containing 7,027 data points. Of the total, 75 percent
However, the Limerinos n values are larger than were from flume studies and 25 percent from field tests.
Keulegan’s by 7 to 52 percent. These consistent differ- He used 22 of these data sets and demonstrated a signifi-
ences lead one to suspect some bed irregularities in cant agreement with both field and laboratory data.
Limerinos’ field data in addition to grain roughness.
(3) Brownlie’s basic equations were modified for
(3) Arcement and Schneider (1989, p 6) state, “If a SAM to display bed roughness as a coefficient times the
measured d84 is available or can be estimated, grain roughness.
[Limerinos] may be used to obtain a base n for sand
channels in lieu of using Table 1.” However, n values n [BED FORM ROUGHNESS] (5-14)
calculated by Limerinos, shown in Table 5-6 herein, are × [STRICKLER GRAIN ROUGHNESS]
considerably smaller than the values shown in Table 1 of
Arcement and Schneider even though they state their
Table 1 is for upper regime flow.

Table 5-6
n Values Calculated by Strickler and Limerinos Equations
Effective Limerinos Equation
Roughness R, ft
Strickler
ks, mm ks, ft n = 0.034 . ks1/6 1 5 10 20 50

0.10 0.0003281 0.009 0.011 0.013 0.013 0.014 0.015

1.00 0.003281 0.013 0.015 0.016 0.017 0.017 0.019
2.13 0.007 0.015 0.017 0.018 0.018 0.019 0.020
10 0.03281 0.019 0.022 0.022 0.022 0.023 0.024
64 0.20997 0.026 0.037 0.031 0.030 0.030 0.030
100 0.3281 0.028 0.044 0.034 0.033 0.032 0.032
152.4 0.5 0.030 0.053 0.038 0.036 0.035 0.034

Note:
0.0926 R 1/6
Limerinos Equation: n
1.16 2 log (R/ks)

5-9
 EM 1110-2-1601
Change 1
30 Jun 94

* d50 = the particle size, ft, for which 50 percent of the

sediment mixture is finer

S = bed slope. Probably the energy slope will be more

representative if flow is nonuniform.

σ = the geometric standard deviation of the sediment

mixture (is shown as σg in Figure 5-4)

d d50 
σ 0.5  84 (5-17)
 d50 d16 

(c) Transition function: If the slope is greater than

0.006, flow is always upper regime. Otherwise, the tran-
sition is correlated with the grain Froude number as
follows:

V
Fg (5-18)
(ss 1) g d50
Figure 5-4. Velocity versus hydraulic radius in a
mobile bed stream (courtesy of W. M. Keck Labora- 1.74
Fg′ (5-19)
tories of Hydraulics and Water Resources (Brownlie S 1/3
1981))

If Fg ≤ Fg′ , then lower regime flow

This makes it easy to compare the results with the skin
friction for fixed bed systems as presented in Plate 3. The If Fg > Fg′ , then upper regime flow
resulting forms of the equations for lower and upper
regimes are as follows:
where
(a) Lower regime flow:
Fg = grain Froude number
 
  0.1374  V = velocity of flow
n 1.6940   S 0.1112 σ0.1605 0.034 d50
R 0.167

  d50   ss = specific gravity of sediment particles

(5-15)
The transition occurs over a range of hydraulic radii and
(b) Upper regime flow: not at a point. Over this range, then, it is a double-valued
function, and the transition test will give different regimes
  depending on which equation is being solved for rough-
  0.0662  ness at that iteration. That is realistic since one expects
n 1.0213   S 0.0395 σ0.1282 0.034 d50 0.167
R
the rising side of a hydrograph to trigger the transition at
  d50   a different discharge than does the falling side.
(5-16)
m. Soil Conservation Service (SCS) n values, grass
where cover. Hydraulic roughness curves for five types of grass
cover were published by SCS (US Department of Agricul-
R = hydraulic radius, ft, of the bed portion of the cross ture 1947) (Figure 5-5). Each curve type, A
section
*

5-10
 EM 1110-2-1601
Change 1
30 Jun 94

* through E, refers to grass conditions described in

Table 5-7.

n. Example. To use analytical methods, the engineer

is faced with assigning physically based parameters, like
surface roughness or material type, to each subdivided
area in a cross section. The subdivided areas are bounded
by vertical lines between successive coordinate points on
the boundary and the water surface. Table 5-8 illustrates
the development of n values for the cross section in
Figure 5-1 by the application of analytical equations. The
analytical methods are in the Hydraulic Design Package
SAM. The cross section is coded as station and elevation
starting at the levee on the left, Area 1.
Figure 5-5. n value relationships for grass cover

Table 5-7
Characteristics of Grass Cover

Type Cover Condition

A Weeping love grass Excellent stand, tall (average 30 in.)

Yellow bluestem (Andropogon ischaemum) Excellent stand, tall (average 36 in.)
B Kudzu Very dense growth, uncut
Bermuda grass Good stand, tall (average 12 in.)
Native grass mixture (little Good stand, unmowed
bluestem, blue grama, other long
and short midwest grasses)
Weeping love grass Good stand, tall (average 24 in.)
Lespedeza sericea Good stand, not woody, tall (average
19 in.)
Alfalfa Good stand, uncut (average 11 in.)
Weeping love grass Good stand, mowed (average 13 in.)
Kudzu Dense growth, uncut
Blue grama Good stand, uncut (average 13 in.)

C Crabgrass Fair stand, uncut (10 to 48 in.)

Bermuda grass Good stand, mowed
Common lespedeza Good stand, uncut (average 11 in.)
Grass-legume mixture--summer Good stand, uncut (6 to 8 in.)
(orchard grass, redtop, Italian
ryegrass, and common lespedeza)
Centipede grass Very dense cover (average 6 in.)
Kentucky bluegrass Good stand, headed (6 to 12 in.)

D Bermuda grass Good stand, cut to 2.5-in. height

Common lespedeza Excellent stand, uncut (average 4.5 in.)
Buffalo grass Good stand, uncut (3 to 6 in.)
Grass-legume mixture--fall, Good stand, uncut (4 to 5 in.)
spring (orchard grass, redtop,
Italian ryegrass, and common lespedeza)
Lespedeza sericea After cutting to 2-in. height; very
good stand before cutting

E Bermuda grass Good stand, cut to 1.5-in. height

Bermuda grass Burned stubble

5-11
 EM 1110-2-1601
Change 1
30 Jun 94

*
Table 5-8
Hydraulic Roughness from Surface Properties
Area
No. Station Elevation n Value ks, ft Comment
0.0 18.00
1 Grass D: Bermuda grass cut to 2.5 in.
From Soil Conservation Service
(Chow 1959, pp 179-184)
50.0 5.50
2 0.100 Left Floodplain, (USDT (Arcement and
Schneider 1989), Table 3)
n = ( nb+ n1 + n2 + n3 + n4)
= (0.028+0.010+0.012+0.050)
125.0 2.00
3 1 Strickler ks-ft; Assumed (Chow, p 206)
129.0 0.00
4 Brownlie bed roughness equations
(Brownlie 1983)
D84 = 6.5 mm, D50 = 1.7 mm, D16 = 0.4 mm
154.0 0.00
5 1 Same as left bank (Area 3)
158.0 2.00
6 0.125 Right Floodplain, (USDT (Arcement and
Schneider 1989), Table 3)
n = (0.028+0.010+0.012+0.075)
168.0 5.50
7 Same as left levee (Area 1)
218.0 18.00

(1) Area 1 is designed to be a mowed grass surface. (7) Area 7, the right levee, will be the same as the
The n value will depend on the flow depth and velocity left levee.
over the panel.
5-6. Composite n Values and Hydraulic Radius
(2) Area 2 is the left floodplain. The best source for
n values in large, woody vegetation is the USDT proce- The calculations that transform the complex geometry and
dure, referenced in Table 5-2. Therefore, that n value roughness into representative one-dimensional hydraulic
will be coded directly. parameters for flow depth calculations are called composi-
ting hydraulic parameters. That is, in a complex cross
(3) Area 3 is the left bank of the channel. Rough- section the composite hydraulic radius includes, in addi-
ness will be calculated by estimating a surface irregularity tion to the usual geometric element property, the variation
ks for the bank line to be 1 ft. of both depth and n values. There are several methods in
the literature for compositing. The Alpha method,
(4) For Area 4, the channel bed roughness will be described in Appendix C, was selected as the default for
calculated from the bed sediment gradation using the SAM. Two other methods are provided as options: equal
Brownlie bed roughness equations. That method predicts velocity and sum of forces.
whether the roughness is lower or upper regime. It uses
the d84, d50, and d16 grain sizes of the bed surface. a. Equal velocity method. Cox (1973) tested three
methods for determining the equivalent roughness in a
(5) Area 5 is the right bank. It will be the same as rectangular channel: the equal velocity method, which is
the left bank. sometimes called the Horton or the Einstein method after
the developers; the Los Angeles District method; and the
(6) Area 6 is expected to have a more dense stand of Colbatch method.
vegetation than on the left side.

5-12
 EM 1110-2-1601
Change 1
30 Jun 94

* (1) Perhaps a more rational method for vertical walls

is the equal velocity method. It was proposed indepen-
dently by Horton and by Einstein (Chow 1959), and is
one which prevents dividing by zero.

1.5 1.5 1.5 2/3

p1 n1 p2 n2 ... pN nN (5-20)
n
P 2/3

where

n = the composite n value for the section

pN = wetted perimeter in subdivided area n

nN = n value in subdivided area n

N = the last subdivided area in the cross section

P = total wetted perimeter in the cross section

Since only wetted perimeter, and not hydraulic radius, Figure 5-6. Definition sketch for Los Angeles District
appears in this equation, it is always well behaved. and Colbatch methods

(2) The equations for the Los Angeles District (Equ- (3) Horton’s method is retained here because of its
ation 5-21) and Colbatch (Equation 5-22) methods simplicity. It is adequate for the simple cross-section
(Figure 5-6) are as follows: shapes, and it is programmable for the complex cross-
section shapes. The other methods that Cox tested would
a1 n1 a2 n2 aN nN be very difficult to program for automatic computations in
n (5-21) complex cross sections.
A
b. Alpha method.
2/3
1.5 1.5 1.5
a1 n1 a2 n2 aN nN (5-22) (1) The Chezy equation forms the basis for this
n
A 2/3 method. The cross section is subdivided into areas
between coordinate points.

where (2) Calculations always begin at the first area in the

cross section, and the geometric properties are calculated
aN = end area associated with subdivided area n and saved for each wet area across the section. The
hydraulic radius and Chezy C are then calculated and the
A = total area in cross section compositing parameters summed. Computations move
area by area to the end of the cross section.
As a result of these experiments, Cox concluded that
Horton’s method was not as accurate as the Los Angeles (3) The alpha method fails when there is a vertical
District method or the Colbatch method. Based on one of wall.
Cox’s figures, the Horton method gave a composite
n value as much as 8 percent higher than measured for (4) James and Brown reported that the “Manning or
the combination of rough walls and a smooth bed. One Chezy equations do not accurately predict the stage-
test, a combination of smooth walls and a rough bed, gave discharge relation in a channel-floodplain configuration
an effective n value about 4 percent lower than measured. for shallow depths on the floodplain (1.0 < Y/D < 1.4;
*

5-13
 EM 1110-2-1601
Change 1
30 Jun 94

* where Y = main channel depth and D = main channel ‘

bank full depth) without adjustments to either the resis- = A/P
tance coefficient or the hydraulic radius.... the effects of
geometry seem to disappear at the higher stages, i.e., for K = total conveyance of cross section = K1 + K2
Y/D > 1.4, it no longer became necessary to make any + .........+ Kn
correction to the basic equations” (James and Brown
1977, p 24). e. Example. Flow depth calculations using n values
calculated by the analytical methods are shown in
c. Sum of forces method. This method was proposed Tables 5-9 through 5-11. Note the column headed
by Pavlovskii, by Muhlhofer, and by Einstein and Banks “ni value” in Table 5-10. The value for each area is
(Chow 1959). It is based on the hypothesis that the total shown, and at the bottom of that column the composited
force resisting the flow is equal to the sum of the forces value for the entire cross section is 0.062. Table 5-11
resisting the flow in each area. The resulting composite shows the equivalent n value for the conveyance method
n value is to be 0.051. It is important not to mix n values deter-
mined by different compositing methods.
2 2 2
p1 n1 p2 n2 pN nN (5-23)
n 5-7. Expansion and Contraction in a 1-D Model
P 1/2
If the handbook approach is used, the expansion and
d. Conveyance method. The traditional approach to contraction losses are included in the n2 term. That is the
compositing by the conveyance method requires the cross contribution from variation in cross sections. Therefore,
section to be subdivided into subsections between channel if contraction and expansion coefficients are being used,
and overbanks. Conveyance is calculated for each sub- leave that term out.
section as follows:
If the analytical methods are used, no terms for expansion
1.486 Ai R
2/3 or contraction will be included. They would have to be
i (5-24)
Ki added separately--perhaps by increasing the ks value.
ni Values from the n2 component in the handbook method
would be appropriate. They would have to be included in
where ks .

Ki = conveyance in subsection i 5-8. Unforeseen Factors

Ai = end area of subsection i a. Seasonality. This affects water temperature and

vegetation. Both can cause significant changes in n value.
Ri = hydraulic radius in subsection i
b. Tubeworms and barnacles. The Corps built a
ni = n value in subsection i concrete channel in Corte Madera Creek only to find that
marine creatures called tubeworms were attracted to it.
The composite n value is calculated from the total con- They create a substantial increase in the surface roughness
veyance and the hydraulic radius as follows: in the zone below sea level. Rather than the usual ks of
0.007 ft, WES estimated the zone with the tubeworms had
1.486 AR 2/3 a ks of 0.08 ft (Copeland and Thomas 1989).
n (5-25)
K
c. Roughness from gravel moving in a concrete
channel. In recent experiments at WES, gravel movement
was modeled along a hard bottom flume to determine how
where much the n value would increase (Stonestreet, Copeland,
and McVan 1991). As long as it moved, the increase was
A = total end area of cross section only about 10 percent. That was the case for concentra-
tions up to about 3,000 ppm. When the concentration
R = hydraulic radius for the entire cross section exceeded that, bed deposits began to form. That effect on
*

5-14
 EM 1110-2-1601
Change 1
30 Jun 94

*
Table 5-9
Water Surface Elevations Using the Alpha Method
Normal Depth Using Composite Properties by Alpha Method

Water Boundary
Surface Top Composite Shear
Discharge Elevation Width R Slope Composite Velocity Froude Stress
**** N cfs ft ft ft ft/ft n Value fps Number psf

**** 1 2,300.00 9.58 150.6 7.77 0.000800 0.0621 2.64 0.17 0.39

Table 5-10
Water Surface Elevations Using the Alpha Method
Flow Distribution by Alpha Method, Discharge = 2,300.00 cfs

Percentage Wetted
Increase Area Ai Perimeter pi Ri = ks ni Velocity
Station Discharge sq ft ft Ai/pi ft Value fps

0.0
3.06 33.2 16.8 1.98 1.179 0.0312 2.11
50.0
25.74 437.0 75.1 5.82 624.9 0.1000 1.35
125.0
7.10 34.3 4.5 7.67 1.000 0.0342 4.76
129.0
51.31 239.4 25.0 9.58 4.563 0.0383 4.93
154.0
7.10 34.3 4.5 7.67 1.000 0.0342 4.76
158.0
2.64 58.3 10.6 5.50 2,384.0 0.1250 1.04
168.0
3.06 33.2 16.8 1.98 1.179 0.0312 2.11
218.0
100.00 869.9 153.2 7.77 18.59 0.0621 2.64

Table 5-11
Water Surface Elevations Using the Alpha Method
Equivalent Hydraulic Properties using Conveyance Method

Hydraulic
Radius Subsection
Velocity Manning’s Discharge Area Velocity
ft n Value cfs sq ft fps

5.68 0.0506 2300.00 869.86 2.64

5-15
 EM 1110-2-1601
Change 1
30 Jun 94

* n value is very significant and requires a sedimentation f. Wetlands. Measurements by the South Florida
investigation. Water Management District in connection with the res-
toration of the Kissimmee River produced n values of
d. Bed form roughness in concrete channels. After 1.011. That coincided with flow depths below the top of
the Corte Madera Creek channel went into operation, sedi- the marsh vegetation. They chose to use an n value of
ment deposited over the smooth concrete bed in the 0.3 for the levee design calculations because the flow
downstream portion. A sedimentation study was con- depth was considerably above the top of the dense marsh
ducted, after the fact, using HEC-6 (Copeland and vegetation. However, that was judgment rather than
Thomas 1989). They determined the channel n value to experiment. (Once flow depth exceeds the top of vegeta-
be 0.028 using high-water marks and the known water tion, it seems reasonable to reduce n values.)
discharge. The calculated depth and gradation of bed
deposits matched prototype values very nicely. This g. Marsh. Studies for a flood at Kawanui Marsh,
n value is not suggested as a design value. It is presented Hawaii, resulted in an n value of 0.95. That is attributed
to illustrate surprises that can come from a fixed-bed to a dense vine that was growing on the water surface. It
hydraulic approach. was attached to the bed from place to place, but when the
flood occurred, it piled the vine into accordion-like folds.
e. Large woody debris. Large woody debris refers to Subsequent measurements, on smaller floods, were used
downed trees and log jams. This is a condition that to develop the n value.
exists, but its effect on the hydraulic roughness during *
large floods is not well documented.

5-16