Skip to contentUnited States Department of Transportation - Federal Highway AdministrationFHWA HomeFeedback

Hydraulics Engineering


Bridge Scour and Stream Instability Countermeasures: Experience, Selection, and Design Guidance-Third Edition

Design Guideline 4 Riprap Revetment


Revetments are used to provide protection for embankments, streambanks, and streambeds. They may be flexible or rigid, and can be used to counter virtually all erosion mechanisms. They do not significantly constrict channels or alter flow patterns. However, revetments do not provide resistance against geotechnical instability, such as slumping failure in saturated streambanks and embankments. In addition, they are relatively unsuccessful in stabilizing streambanks and streambeds in degrading streams. Special precautions must be observed in the design of revetments for degrading channels. This design guideline provides recommendations for the design and installation of rock riprap as an armoring-type bank protection against erosion.


Flexible revetments include rock riprap, partially grouted rock riprap, rock-and-wire mattresses, gabions, pre-cast articulating concrete blocks, rock-filled trenches, windrow revetments, used tire revetments, and vegetation. Because rock riprap is almost always installed in a layer that is multiple particles thick, it adjusts to distortions and local displacement of materials without complete failure of the revetment installation. This aspect of rock riprap behavior is often referred to as a "self-healing" characteristic. In contrast, flexible rock-and-wire mattresses, gabions, articulating concrete blocks, used-tire systems, and grout-filled mats may sometimes span over voids in the underlying soil, but usually can adjust to gradual distortions. Discussion of design guidelines for flexible revetments other than rock riprap can be found separately in the following section.

4.2.1 Flexible Revetments Other Than Rock Riprap

Design guidelines, installation recommendations, and suggested specifications for flexible revetments other than rock riprap are provided in this document, as follows:

Wire Enclosed Riprap Mattresses: Design Guideline 6
Articulating Concrete Blocks: Design Guideline 8
Grout-Filled Mattresses: Design Guideline 9
Gabion Mattresses: Design Guideline 10
Partially Grouted Riprap: Design Guideline 12
4.2.2 Design Guidelines for Revetment Riprap

NCHRP Report 568, "Riprap Design Criteria, Recommended Specifications, and Quality Control" (Lagasse et al. 2006) provides design guidance for sizing the rock for dumped riprap used for bank protection. That NCHRP study evaluated numerous procedures for sizing revetment riprap, and recommends using the method developed by Maynord (1989, 1990) and published by the U.S. Army Corps of Engineers (USACE) as Engineering Manual No. 1110-2-1601 (EM-1601) (USACE 1991). The procedure uses both velocity and depth as its primary design parameters.

The EM-1601 equation can be used with uniform or gradually varying flow. Coefficients are included to account for the desired safety factor for design, specific gravity of the riprap stone, bank slope, and bendway character. The EM-1601 equation is:

Equation 4.1: d sub 30 equals, y times (S sub f times C sub s times C sub v times C sub t) times [the quotient of (V sub des) and square root of (K sub 1 times (S sub g minus 1) times g times y) ] raised to the power 2.5. Descriptions of algebraic symbol used are listed in the text. (4.1)


d30 = Particle size for which 30% is finer by weight, ft (m)
y = Local depth of flow, ft (m)
Sf = Safety factor (must be > 1.0)
CS = Stability coefficient (for blanket thickness = d100 or 1.5d50, whichever is greater, and uniformity ratio d85/d15 = 1.7 to 5.2)

= 0.30 for angular rock

= 0.375 for rounded rock
CV = Velocity distribution coefficient

= 1.0 for straight channels or the inside of bends
= 1.283 - 0.2log(Rc/W) for the outside of bends (1.0 for Rc/W > 26)
= 1.25 downstream from concrete channels
= 1.25 at the end of dikes
CT = Blanket thickness coefficient given as a function of the uniformity ratio d85/d15. CT = 1.0 is recommended because it is based on very limited data.
Vdes = Characteristic velocity for design, defined as the depth-averaged velocity at a point 20% upslope from the toe of the revetment, ft/s (m/s)

For natural channels,
Vdes=Vavg(1.74 - 0.52 log (Rc/W))
Vdes=Vavg for Rc/W > 26

For trapezoidal channels,
Vdes=Vavg (1.71 - 0.78 log (Rc/W))
Vdes=Vavg for Rc/W > 8

Vavg = Channel cross-sectional average velocity, ft/s (m/s)
K1 = Side slope correction factor

The side slope correction factor K sub 1 equals the square root of (1 minus (sine (bank angle, theta, minus 14 degrees) divided by sine 32 degrees) raised to the power 1.6)

where: θ is the bank angle in degrees
Rc = Centerline radius of curvature of channel bend, ft (m)
W = Width of water surface at upstream end of channel bend, ft (m)
Sg = Specific gravity of riprap (usually taken as 2.65)
g = Acceleration due to gravity, 32.2 ft/s2 (9.81 m/s2)

The values of the coefficients used in the EM-1601 equation are provided in the variable definitions as given above. They can also be determined graphically from charts provided in Appendix B of EM 1601 (USACE 1991). The EM-1601 document can be downloaded from USACE websites if additional guidance is desired.

Using the recommended riprap gradations from NCHRP Report 568, the d30 size of the riprap is related to the recommended median (d50) size by:

Equation 4.2: d sub 50 equals 1.2 times d sub 30 (4.2)

The flow depth "y" used in Equation 4.1 is defined as the local flow depth. The flow depth at the toe of slope is typically used for bank revetment applications; alternatively, the average channel depth can be used. The smaller of these values will result in a slightly larger computed d30 size, since riprap size is inversely proportional to (y0.25).

The blanket thickness coefficient (CT) is 1.0 for standard riprap applications where the thickness is equal to 1.5d50 or d100, whichever is greater. Because only limited data is available for selecting lower values of CT when greater thicknesses of riprap are used, a value of 1.0 is reasonable for all applications.

The recommended Safety Factor Sf is 1.1 for bank revetment. Greater values should be considered where there is significant potential for ice or impact from large debris, freeze-thaw degradation that would significantly decrease particle size, or large uncertainty in the design variables, especially velocity.

A limitation to Equation 4.1 is that the longitudinal slope of the channel should not be steeper than 2.0% (0.02 ft/ft). For steeper channels, the riprap sizing approach for overtopping flows should be considered and the results compared with Equation 4.1 (see Design Guide 5).

Once a design size is established, a standard size class can be selected, if design criteria and economic considerations permit. Using standard sizes the appropriate gradation can be achieved by selecting the next larger size class, thereby creating a slightly over-designed structure, but economically a less expensive one. Recommended size classes and gradation characteristics are derived from NCHRP Report 568, and are provided in Section 4.2.4 of this design guide.

4.2.3 Thickness of Riprap

All stones should be contained reasonably well within the riprap layer thickness, with little or no oversize stones protruding above the surface of the riprap matrix. The following criteria are recommended in NCHRP Report 568 for revetment riprap:

  1. Layer thickness should not be less than the spherical diameter of the D100 stone nor less than 1.5 times the spherical diameter of the D50 stone, whichever results in the greater thickness.
  2. Layer thickness should not be less than 1 ft (0.30 m) for practical placement.
  3. Layer thickness determined either by criterion 1 or 2 should be increased by 50% when the riprap is placed underwater to compensate for uncertainties associated with this placement condition.
4.2.4 Riprap Shape and Gradation

Riprap design methods typically yield a required size of stone that will result in stable performance under the design loadings. Because stone is produced and delivered in a range of sizes and shapes, the required size of stone is often stated in terms of a minimum allowable representative size. For example, the designer may specify a minimum d50 or d30 for the rock comprising the riprap, thus indicating the size for which 50 or 30% (by weight) of the particles are smaller. Stone sizes can also be specified in terms of weight (e.g., W50 or W30) using an accepted relationship between size and volume, and the known (or assumed) density of the particle.

Shape: The shape of a stone can be generally described by designating three axes of measurement: Major, intermediate, and minor, also known as the "A, B, and C" axes, as shown in Figure 4.1.

Sketch of a rough oblong cubic shape with major long axis labeled as length A, intermediate width axis labeled as B and minor thickness axis labeled as C
Figure 4.1. Riprap shape described by three axes.

Riprap stones should not be thin and platy, nor should they be long and needle-like. Therefore, specifying a maximum allowable value for the ratio A/C, also known as the shape factor, provides a suitable measure of particle shape, since the B axis is intermediate between the two extremes of length A and thickness C. A maximum allowable value of 3.0 is recommended:

A over C is less than or equal to 3. 0. (4.3)

For riprap applications, stones tending toward subangular to angular are preferred, due to the higher degree of interlocking, hence greater stability, compared to rounded particles of the same weight.

Density: A measure of density of natural rock is the specific gravity Sg, which is the ratio of the density of a single (solid) rock particle γsto the density of water γw:

Equation 4.4: Specific gravity S sub g equals gamma sub s divided by gamma sub w. Terms are explained in the text. (4.4)

Typically, a minimum allowable specific gravity of 2.5 is required for riprap applications. Where quarry sources uniformly produce rock with a specific gravity significantly greater than 2.5 (such as dolomite, Sg = 2.7 to 2.8), the equivalent stone size can be substantially reduced and still achieve the same particle weight gradation.

Size and weight: Based on field studies, the recommended relationship between size and weight is given by:

Equation 4.5: Weight W equals 0.085 times (density of stone, gamma sub s, times size of intermediate axis d to the power 3) (4.5)


W = Weight of stone, lb (kg)
γs = Density of stone, lb/ft3 (kg/m3)
d = Size of intermediate ("B") axis, ft (m)

Table 4.1 provides recommended gradations for ten standard classes of riprap based on the median particle diameter d50 as determined by the dimension of the intermediate ("B") axis. These gradations conform to those recommended in NCHRP Report 568 (Lagasse et al. 2006). The proposed gradation criteria are based on a nominal or "target" d50 and a uniformity ratio d85/d15 that results in riprap that is well graded. The target uniformity ratio d85/d15 is 2.0 and the allowable range is from 1.5 to 2.5.

Table 4.1. Minimum and Maximum Allowable Particle Size in Inches.
Nominal Riprap Class by Median Particle Diameter d15 d50 d85 d100
Class Size Min Max Min Max Min Max Max
I 6 in 3.7 5.2 5.7 6.9 7.8 9.2 12.0
II 9 in 5.5 7.8 8.5 10.5 11.5 14.0 18.0
III 12 in 7.3 10.5 11.5 14.0 15.5 18.5 24.0
IV 15 in 9.2 13.0 14.5 17.5 19.5 23.0 30.0
V 18 in 11.0 15.5 17.0 20.5 23.5 27.5 36.0
VI 21 in 13.0 18.5 20.0 24.0 27.5 32.5 42.0
VII 24 in 14.5 21.0 23.0 27.5 31.0 37.0 48.0
VIII 30 in 18.5 26.0 28.5 34.5 39.0 46.0 60.0
IX 36 in 22.0 31.5 34.0 41.5 47.0 55.5 72.0
X 42 in 25.5 36.5 40.0 48.5 54.5 64.5 84.0

Note: Particle sized corresponds to the intermediate ("B") axis of the particle.

Based on Equation 4.5, which assumes the volume of the stone is 85% of a cube, Table 4.2 provides the equivalent particle weights for the same ten classes, using a specific gravity of 2.65 for the particle density.

Table 4.2. Minimum and Maximum Allowable Particle Weight in Pounds.
Nominal Riprap Class by Median Particle Weight W15 W50 W85 W100
Class Weight Min Max Min Max Min Max Max
I 20 lb 4 12 15 27 39 64 140
II 60 lb 13 39 51 90 130 220 470
III 150 lb 32 93 120 210 310 510 1100
IV 300 lb 62 180 240 420 600 1,000 2,200
V 1/4 ton 110 310 410 720 1,050 1,750 3,800
VI 3/8 ton 170 500 650 1,150 1,650 2,800 6,000
VII 1/2 ton 260 740 950 1,700 2,500 4,100 9,000
VIII 1 ton 500 1,450 1,900 3,300 4,800 8,000 1,7600
IX 2 ton 860 2,500 3,300 5,800 8,300 13,900 30,400
X 3 ton 1,350 4,000 5,200 9,200 13,200 22,000 48,200

Note: Weight limits for each class are estimated from particle size by: W = 0.85(γsd3) where d corresponds to the intermediate ("B") axis of the particle, and particle specific gravity is taken as 2.65.

4.2.5 Recommended Tests for Rock Quality

Standard test methods relating to material type, characteristics, and testing of rock and aggregates typically associated with riprap installations (e.g., filter stone and bedding layers) are provided in this section and are recommended for specifying the quality of the riprap stone. In general, the test methods recommended in this section are intended to ensure that the stone is dense and durable, and will not degrade significantly over time.

Rocks used for riprap should only break with difficulty, have no earthy odor, no closely spaced discontinuities (joints or bedding planes), and should not absorb water easily. Rocks comprised of appreciable amounts of clay, such as shales, mudstones, and claystones, are never acceptable for use as fill for gabion mattresses. Table 4.3 summarizes the recommended tests and allowable values for rock and aggregate.

4.2.6 Filter Requirements

The importance of the filter component of revetment riprap installation should not be underestimated. Geotextile filters and granular filters may be used in conjunction with riprap bank protection. When using a granular stone filter, the layer should have a minimum thickness of 4 times the d50 of the filter stone or 6 inches, whichever is greater. When placing a granular filter under water, its thickness should be increased by 50%.

The filter must retain the coarser particles of the subgrade while remaining permeable enough to allow infiltration and exfiltration to occur freely. It is not necessary to retain all the particle sizes in the subgrade; in fact, it is beneficial to allow the smaller particles to pass through the filter, leaving a coarser substrate behind. Detailed aspects of filter design are presented in Design Guideline 16 of this document.

Some situations call for a composite filter consisting of both a granular layer and a geotextile. The specific characteristics of the base soil determine the need for, and design considerations of the filter layer. In cases where dune-type bedforms may be present at the toe of a bank slope protected with riprap, and where adequate toe down extent cannot be ensured, it is strongly recommended that only a geotextile filter be considered.

Table 4.3. Recommended Tests for Riprap Quality.
Test Designation Property Allowable value Frequency(1) Comments
AASHTO TP 61 Percentage of Fracture < 5% 1 per 20,000 tons Percentage of pieces that have fewer than 50% fractured surfaces
AASHTO T 85 Specific Gravity and Water Absorption Average of 10 pieces:

Sg > 2.5

Absorption < 1.0%
1 per year If any individual piece exhibits an Sg less than 2.3 or water absorption greater than 3.0%, an additional 10 pieces shall be tested. If the second series of tests also exhibits pieces that do not pass, the riprap shall be rejected.
AASHTO T 103 Soundness by Freezing and Thawing Maximum of 10 pieces after 25 cycles:

< 0.5%
1 per 2 years Recommended only if water absorption is greater than 0.5% and the freeze-thaw severity index is greater than 15 per

ASTM D 5312.
AASHTO T 104 Soundness by Use of Sodium Sulfate or Magnesium Sulfate Average of 10 pieces:

< 17.5%
1 per year If any individual piece exhibits a value greater than 25%, an additional 10 pieces shall be tested. If the second series of tests also exhibits pieces that do not pass, the riprap shall be rejected.
AASHTO TP 58 Durability Index Using the Micro-Deval Apparatus
> 90Severe
> 80Moderate
> 70Mild
1 per year Severity of application per Section 5.4, CEN (2002). Most riverine applications are considered mild or moderate.

Splitting Tensile Strength of Intact Rock Core Specimens Average of 10 pieces:

> 6 MPa
1 per year If any individual piece exhibits a value less than 4MPa, an additional 10 pieces shall be tested. If the second series of tests also exhibits pieces that do not pass, the riprap shall be rejected.

D 5873
Rock Hardness by Rebound Hammer See Note (2) 1 per 20,000 tons See Note (2)
Shape Length to Thickness Ratio A/C
< 10%,d50 < 24 inch
< 5%,d50 > 24 inch
1 per 20,000 tons Percentage of pieces that exhibit A/C ratio greater than 3.0 using the Wolman Count method (Lagasse et al. 2006)

D 5519
Particle Size Analysis of Natural and Man-Made Riprap Materials 1 per year See Note (3)
Gradation Particle Size Distribution Curve 1 per 20,000 tons Determined by the Wolman Count method (Lagasse et al. 2006), where particle size "d" is based on the intermediate ("B") axis

(1) Testing frequency for acceptance of riprap from certified quarries, unless otherwise noted. Project-specific tests exceeding quarry certification requirements, either in performance value or frequency of testing, must be specified by the Engineer.

(2) Test results from D 5873 should be calibrated to D 3967 results before specifying quarry-specific minimum allowable values.

(3) Test results from D 5519 should be calibrated to Wolman Count (Lagasse et al. 2006) results before developing quarry-specific relationships between size and weight; otherwise, assume W = 85% that of a cube of dimension "d" having a specific gravity of Sg

4.2.7 Edge Treatment and Termination Details

Riprap revetment should be toed down below the toe of the bank slope to a depth at least as great as the depth of anticipated long-term bed degradation plus toe scour (see Volume 1, Section 4.3.5). Installations in the vicinity of bridges must also consider the potential for contraction scour.

Recommended freeboard allowance calls for the riprap to be placed on the bank to an elevation at least 2.0 feet greater than the design high water level. Upstream and downstream terminations should utilize a key trench that is dimensioned in relation to the d50 size of the riprap. Where the design water level is near or above the top of bank, the riprap should be carried to the top of the bank. Figures 4.2, 4.3, and 4.4 are schematic diagrams that summarize these recommendations. If toe down cannot be placed below the anticipated contraction scour and degradation depth (Figure 4.2), a mounded toe approach (Figure 4.3) is suggested.

Sketch showing riprap placed on sloped bank. Maximum slope indicate as 1 vertical to 1.5 horizontal. Riprap placed from Minimum freeboard, 2 foot, 0.6 meters, above design high water, to below ambient bed elevation to toe down at maximum scour depth. Maximum scour depth equals contraction scour plus long-term degradation scour plus toe scour. Riprap thickness indicated as larger of 1.5 times the D50 or the D100 of the riprap. A geotextile or granular filter is under the riprap.
Figure 4.2. Riprap revetment with buried toe.

Sketch showing riprap placed on sloped bank with riprap at double layer thickness at slope base. Double layer mound height equals desired toe down depth. Maximum slope indicate as 1 vertical to 1.5 horizontal. Riprap placed from Minimum freeboard, 2 foot, 0.6 meters, above design high water, to ambient bed elevation. A geotextile or granular filter is under the riprap.
Figure 4.3. Riprap revetment with mounded toe.

Sketches indicating details of: riprap placement on curves; on slopes with a toe down section or launchable riprap; and key in for upstream edge of riprap.
Figure 4.4. Riprap revetment details.


Riprap is to be designed for a 100 ft (30.5 m) wide natural channel on a bend that has a centerline radius (Rc) of 500 ft (152.4 m). The radius of curvature divided by width (Rc/W) is 5.0. The revetment will have a 1V:2H sideslope (26.6°) and the rounded riprap has a specific gravity of 2.54. A factor of safety (Sf) of 1.2 is desired. Toe scour on the outside of the bend has been determined to be 2.5 ft (0.76 m) during the design event.

The following data were obtained from hydraulic modeling of the design event.

Variable English Units SI Units
Value Units Value Units
Average Channel Velocity ft/s 7.2 m/s 2.19
Flow Depth at Bank Toe ft 11.4 m 3.47

Step 1: Compute the side slope correction factor (or select from graph on Plate B-39 of EM 1601):

The side slope correction factor K sub 1 equals the square root of (1 minus (sine (bank angle, theta minus 14 degrees) divided by sine 32 degrees) raised to the power 1.6)equals the square root of (1 minus (sine (26.6 degrees minus 14 degrees) divided by sine 32 degrees) raised to the power 1.6) equals 0.87

Step 2: Select the appropriate stability coefficient for rounded riprap: Cs = 0.375

Step 3: Compute the vertical velocity factor (Cv) for Rc/W = 5.0:

Cv = 1.283 - 0.2log(RC --/W) = 1.283 - 0.2log(5.0) = 1.14

Step 4: Compute local velocity on the side slope (Vdes) for a natural channel with Rc/W = 5.0:

Vdes = V avg [1.74 - 0.52log(RC --/W) = 7.2[1.74 - 0.52log(5.0)]

= 9.9 ft/s (3.01 m/s)

Step 5: Compute the d30 size using Equation 4.1:

d sub 30 equals, y, times (S sub f times C sub s times C sub v times C sub t) times [(V sub des) divided by the square root of ((S sub g minus 1) times K sub 1 times g times y) ] raised to the power 2.5. Descriptions of algebraic symbol used are listed in the text.

Equals, 1.2, times (0.375 times 1.14 times 11.4) times [9.9 divided by the square root of ((2.54 minus 1) times (0.87) times 32.2 times 11.4)] raised to the power 2.5 equals 0.78 feet (0.24 meters)

Step 6: Compute the d50 size for a target gradation of d85/d15 = 2.0:

d50 = 1.2d30 = 1.2(0.78) = 0.94 ft = 11.2 inches (0.29 m)

Note: Use next larger size class (see Volume 1, Chapter 5)

Step 7: Select Class III riprap from Table 4.1 of this design guide: d50 = 12 inches (0.3 m)

Step 8: Determine the depth of riprap embedment below the streambed at the toe of the bank slope:

Since toe scour is expected to be 2.5 ft (0.76 m), the 1V:2H slope should be extended below the ambient bed level 5 ft (1.52 m) horizontally out from the toe to accommodate this scour. Alternatively, a mounded riprap toe 2.5 ft (0.76 m) high could be established at the base of the slope and allowed to self-launch when toe scour occurs.

4.4.1 At the Quarry

The Wolman Count method and Galay transect approach are designed to determine a size distribution based on a random sampling of individual stones within a matrix. Both methods are widely accepted in practice, and rely on samples taken from the surface of the matrix to make the method practical for use in the field. Details of the methods can be found in: Bunte and Abt 2001; Galay et al. 1987; and Wolman 1954. In general, these three references provide detailed descriptions of sampling methods, as well as analysis and reporting procedures for determining the size distribution of rock samples. The Wolman count method is illustrated in this section. The Galay transect approach is discussed in Section 4.4.2.

Material gradations for sand size and small gravel materials are typically determined through a sieve analysis of a bulk sample. The weight of each size class (frequency by weight) retained on each sieve is measured and the total percent of material passing that sieve is plotted versus size (sieve opening). The Wolman (1954) procedure measures frequency by size of a surface material rather than a bulk sample. The intermediate dimension (B axis) is measured for randomly selected particles on the surface.

One field approach for cobble size and larger alluvial materials is to select the particle under one's toe after taking a step with eyes averted to avoid bias in particle selection. Another field approach is to stretch a survey tape over the material and measure each particle located at equal intervals along the tape. The equal interval method is recommended for riprap. The interval should be at least 1 ft for small riprap and increased for larger riprap. The B axis is then measured for one hundred particles. The longer and shorter axes (A and C) can also be measured to determine particle shape. Kellerhals and Bray (1971) provide an analysis that supports the conclusion that a surface sample following the Wolman method is equivalent to a bulk sample sieve analysis. One rule that must be followed is that if a single particle is large enough to fall under two interval points along the tape, then it should be included in the count twice. It is probably better to select an interval large enough that this occurs infrequently.

Once 100 particles have been measured, the frequency curve is developed by counting the number of particles less than or equal to specific sizes. To obtain a reasonably detailed frequency curve, the sizes should increase by (2)1/2. For uniform riprap the sizes may need to increase by (2)1/4 to obtain a detailed frequency curve. The starting size should be small enough to capture the low range of sizes, with 64 mm being adequate for most riprap. This process should be repeated to obtain several samples at the riprap installation.

Figure 4.5 shows one of two riprap stockpiles that were sampled using a Wolman Count to determine whether the sizes met the design criteria of d50 equaling 6 and 12 inches (0.15 to 0.3 m). Three samples of 100 stones were measured at each pile and gradations curves were developed for each of the six samples. Table 4.4 includes the data and results for sample number 1 on the 12-inch (0.3 m) stockpile. The B axis was measured to the nearest 10 mm and the percent less than or equal to each size was computed. The starting size of 64 mm was used and size classes increased by (2)1/2 (64 mm, 91 mm, 128 mm...). For 100 stones, the percent passing is equal to the number of stones less than or equal to a given size.

Photograph of a large stockpile of angular riprap stone
Figure 4.5. Riprap stockpile.

Table 4.4. Example Gradation Measurement Using Size by Number Technique.
Count mm Count mm Count mm Count mm Size mm Percent Passing
1 540 26 560 51 500 76 400 - -
2 510 27 670 52 480 77 340 - -
3 180 28 550 53 180 78 470 - -
4 250 29 220 54 450 79 450 - -
5 250 30 290 55 300 80 280 - -
6 530 31 400 56 420 81 340 64 0
7 450 32 320 57 200 82 940 91 0
8 170 33 270 58 360 83 600 128 0
9 200 34 520 59 290 84 530 181 9
10 180 35 650 60 650 85 230 256 24
11 520 36 550 61 600 86 400 362 52
12 520 37 380 62 400 87 220 512 77
13 360 38 180 63 520 88 180 724 98
14 300 39 200 64 300 89 300 1024 100
15 400 40 190 65 320 90 540 1448 100
16 390 41 340 66 300 91 530 2048 100
17 170 42 420 67 220 92 270 - -
18 330 43 440 68 260 93 280 - -
19 600 44 300 69 320 94 210 - -
20 380 45 420 70 160 95 200 - -
21 340 46 510 71 470 96 230 - -
22 300 47 540 72 730 97 300 - -
23 280 48 600 73 470 98 390 - -
24 330 49 180 74 200 99 710 - -
25 450 50 290 75 200 100 500 - -

Figure 4.6 shows the results of the gradation measurements of the two stockpiles. The average gradation was developed by averaging the three samples. The target d50 was achieved for the average sample for each stockpile. Also shown is the target or allowable range of sizes based on the recommended gradation discussed earlier. The recommended gradation is based on a target d50 and uniformity ratio (St = d85/d15) ranging from 1.5 to 2.5, which are the limits identified by CUR (1995) as "well-graded" riprap (Figure 4.6 and Section 4.2.4). The average curve for the 6-inch (0.15 m) material meets this gradation target but the 12-inch (0.3 m) material exceeds the target maximum d84 by 10%. This indicates that the 12-inch (0.3 m) material is approaching "quarry run" with the uniformity ratio for the 12-inch (0.3 m) material of d85/d15 = 510/187 = 2.7. One solution to correcting this slight deficiency is to exclude the largest particles during placement. However, that would also reduce d50 so the smallest particles should also be excluded from the stockpile.

Graph showing percent passing from Wolman Counts on 12-inch and 6-inch Riprap Stockpiles. On the Cartesian graph the y-axis is total passing 0 to 100 percent, the x-axis is Sieve size in millimeters reverse logarithmic from 1000 to 10. For each size or gradation three samples are shown. Averages indicate 6-inch is within the graphed target range but the average 12-inch D84 is outside the target size range.
Figure 4.6. Example gradations from 6- and 12-inch (0.15 and 0.31 m) d50 stockpiles.

4.4.2 On Site

In reporting on Canadian practice, Galay et al. (1987) notes that typically, stone material used in the construction of riprapped banks and aprons is specified for design as a gradation on a by-weight basis. If it were required to monitor the stones being placed during construction, hypothetically it would be necessary to obtain a volumetric sample of the stone and pass it through a set of sieves. The accumulated weight retained on each sieve would then be plotted as a percentage of the total sample weight in relation to the grid sizes of each sieve. A volumetric or bulk sample in this instance would involve removal of all placed stones to total riprap layer depth within a specified surface area, or all stones within one or more truckloads being transported to the project site.

As this procedure is not practicable, a variety of methods have evolved to check the size gradation of stones being placed as riprap. Generally, the approach has been to assess stone sizes visually while having some impression of what the maximum, minimum, and average sizes of stone look like. This impression is sometimes obtained by actually weighing stones to find typical examples of these three sizes. For projects where extremely large amounts of stone are involved, inspectors sometimes go to the extent of dumping randomly selected truckloads of stone and sorting the stones into several piles of different size ranges. Each of these piles is weighed and related to the total sample weight and a typical size of stone for each pile (Galay et al. 1987).

There has been an effort to develop a simple but effective means of monitoring gradations of stone riprap material (Galay et al. 1987). Basically, what has evolved is a surface sampling technique, whereby stones exposed on the surface of a completed riprap layer are measured with respect to their sizes. Sampling is done in such a way that the measured stones give a representative picture of the proportional area occupied by various sizes. Rather than analyzing the distribution of the sample sizes on a by-weight basis, a by-number analysis is used instead. A gradation curve is then drawn relating stone sizes and frequency distribution. Since riprap specifications are typically provided in terms of stone weight, a link has to be established between stone size and weight. Several methods have been used to describe stone size, including: (1) a single measurement of a stone's intermediate dimension; or (2) relating a stone's volume to an equivalent spherical diameter. In any case, a sample set of stones is weighed and size dimensions determined so that the stone size versus weight relationship can be determined (see Section 4.2.4).

One approach has been to take line samples (that is, stretch a measuring tape across the riprap surface and select stones at even intervals) or an areal sample (select every surface stone within a randomly established boundary). The intermediate dimension of each sample stone is measured and the distribution plotted on a by-number basis in relation to stone size. A predetermined relationship between a stone's size and weight is then used to establish the gradation in terms of weight.

Another approach is to use field-testing procedures related to a visual interpretation of the stone weights that are being placed. Some stones are weighed so that the inspector can gain some appreciation of what minimum, mean, and maximum stone sizes look like. Frequently, this set of stones is marked and set aside at the quarry or the project site for reference by the loader operator and inspector. Rarely would large volumetric or bulk samples be collected so that individual stones could be weighed and the total sample analyzed on a by-weight basis. Occasionally, bulk samples might be collected and sizes segregated into several piles. Each pile would then be weighed and a representative size established for each pile; the distribution would then be plotted on a by-weight basis.

Basic to the argument that an analysis of surface samples can be considered reasonably equivalent to analysis of bulk sample is a paper by Kellerhals and Bray (1971). Although the subject of interest in the paper is sampling of river bed gravels, the conclusions presented are assumed to apply to all coarse materials, including riprap stone, i.e., 'grid sampling with frequency analysis by number is the only sampling procedure capable of describing a surface layer one grain thick, in equivalence with customary bulk sieve analysis' (Galay et al. 1987).

Figure 4.7 presents a plot of sampled stone sizes and their respective measured stone weight, which were selected from a quarry site in Alaska. During placement of stones from this quarry, line samples were collected and their distributions were plotted on a by-number basis. Figure 4.8 shows the results plotted for five samples in relation to the specified gradation envelope curves.

Graph of Stone weight versus stone size for riprap material. Stone size is taken at its middle dimension. On the Cartesian graph the y-axis is stone weight 0 to 7000 pounds, the x-axis is Stone size in feet 0 to 4.5. The Galay data and Galay trendline indicate the volume of the stone falls between the plotted value for a cube and a sphere and close to the equation Volume equals 0.85 times diameter cubed. (Lagasse et al. 2006).
Figure 4.7. Stone weight versus stone size for riprap (Lagasse et al. 2006).

Graph of Stone size versus percent of stones larger than. On the Cartesian graph the y-axis is stone size in inches from 0 to 36, the x-axis is percent of Stones larger than from 0 to 100. Comparing the five samples to the specified gradation curves the five samples have more larger stones than specifications and less stones in the middle to small size than specifications. (Galay et al. 1987)
Figure 4.8. Stone riprap gradations: specified and sampled (Galay et al. 1987).

In this instance, stone placement was unsatisfactory and production procedures were subsequently revised in an attempt to achieve a more widely graded distribution. This required an inspector to be present at the quarry, continually working with the equipment operators to ensure that more stones in the middle and lower range were being loaded and hauled to the site (Galay et al. 1987). A similar field test, the Wolman Count, suitable for both quality control and post-construction/post-flood inspection of riprap is discussed in Section 4.4.1.

4.5.1 Overview

Riprap is placed in a riverine or coastal environment to prevent scour or erosion of the bed, banks, shoreline, or near structures such as bridge piers and abutments. Riprap construction involves placement of rock and stone in layers on top of a bedding or filter layer composed of sand, gravel and/or geotechnical fabric. The basis of the protection afforded by the riprap is the mass and interlocking of the individual rocks.

Factors to consider when designing riprap installations begin with the source for the rock, the method to obtain or manufacture the rock, competence of the rock, and the methods and equipment to collect, transport, and place the riprap. Rock for riprap may be obtained from quarries, by screening oversized rock from earth borrow pits, by collecting rock from fields, or from talus deposits. Screening borrow pit material and collecting field rocks present different problems such as rocks too large or with unsatisfactory length to width ratios for riprap.

Quarry stones are generally the best source for obtaining large rock specified for riprap. However, not all quarries can produce large stone because of rock formation characteristics or limited volume of the formation. Since quarrying generally uses blasting to fracture the formation into rock suitable for riprap, cracking of the large stones may only become evident after loading, transporting, and dumping at the quarry, after moving material from quarry to stockpile at the job site, or from the stockpile to the final placement location.

In most cases, the production of the rock material will occur at a source that is relatively remote from the construction area. Therefore, this discussion assumes that the rock is hauled to the site of the installation, where it is either dumped directly, stockpiled, or loaded onto waterborne equipment.

Quarry operations typically produce rock for riprap that falls into one of three broad categories based on gradation limits: (1) quarry run, (2) graded (blasted or plant run), and (3) uniform riprap.

Quarry run riprap sizing is established by controlling the borehole spacing and blasting technique. Some sorting may be required at the shot pile or a rock breaker may be used to reduce oversized rock to within the maximum size allowed.

Graded riprap sizing is established by controlling the borehole spacing and blasting technique, along with removal of small sizes by running the material over a grizzly, or by sizing it through a crusher. This material is more expensive

Uniform riprap is produced by removing the over- and undersized material by a series of grizzlies. This produces a one-sized gradation within a narrow size limit as dictated by the size of the grizzlies. Of the three types of riprap discussed here, this material is the most expensive to produce.

The objectives of construction of a good riprap installation are (1) to obtain a rock mixture from the source that meets the design specifications and (2) to place that mixture in a well-knit, compact and uniform layer without segregation of the mixture. The best time to control the gradation of the riprap mixture is during the quarrying operation. Sorting and mixing later in stockpiles or at the construction site is not satisfactory. In the past, control of the riprap gradation at the job site has almost always been carried out by visual inspection. Therefore, it is helpful to have a pile of rocks with the required gradation at a convenient location where inspectors can see and develop a reference to judge by eye the suitability of the rock being placed (see Sections 4.4.1 and 4.4.2).

The guidance in this section has been developed to facilitate the proper installation of riprap systems to achieve suitable hydraulic performance and maintain stability against hydraulic loading. The proper installation of riprap systems is essential to the adequate functioning and performance of the system during the design hydrologic event. Guidelines are provided herein for maximizing the correspondence between the design intent and the actual field-finished conditions of the project. This section addresses the preparation of the subgrade, placement of the filter, riprap placement, and measurement and payment.

4.5.2 General Guidelines

The contractor is responsible for constructing the project according to the plans and specifications; however, ensuring conformance with the project plans and specifications is the responsibility of the owner. This is typically performed through the owner's engineer and inspectors. Inspectors observe and document the construction progress and performance of the contractor. Prior to construction, the contractor should provide a quality control plan to the owner (for example, see USACE ER 1180-1-6, 1995, "Construction Quality Management") and provide labor and equipment to perform tests as required by the project specifications.

Designers should include construction requirements for riprap placement in the project plans and specifications. Standard riprap specifications and layout guidance are found in Section 4.2 of this document. Recommended requirements for the stone, including the tests necessary to ensure that the physical and mechanical properties meet the requirements of the project specifications are provided. Field tests can be performed at the quarry and/or on the job site, or representative samples can be obtained for laboratory testing. Additional riprap specifications can be found in manuals of most governmental agencies involved in construction (Federal Highway Administration 1981), (USACE 1991), (Racin et al. 2000).

Typically, one or more standard riprap gradations are specified and plan sheets show locations, grades, and dimensions of rock layers for the countermeasure. Additional drawings clarify features at the toe, at the end of the revetment, at transitions, or at other unusual changes in the structures. The stone shape is important and riprap should be blocky rather than elongated, platy or round. In addition, the stone should have sharp, angular, clean edges at the intersections of relatively flat surfaces.

Stone size and riprap layer thickness are related. Layer thickness is generally defined as not less than the spherical diameter of the upper limit W100 stone or not less than 1.5 times the spherical diameter of the upper limit of the W50 stone, whichever results in the greater thickness. Typically, project specifications call for a 50% increase in layer thickness if the riprap is to be placed underwater. Riprap should be placed on bedding stone and/or geotextile filter material.

On-site inspection of riprap is necessary both at the quarry and at the job site to ensure proper gradation and material that does not contain excessive amounts of fines. Breakage during handling and transportation should be taken into account. Segregation of material during transportation, dumping, or off-loading is not acceptable. Inspection of riprap placement consists of visual inspection of the operation and the finished surface. Inspection must ensure that a dense, rough surface of well-keyed graded rock of the specified quality and sizes is obtained, that the layers are placed such that voids are minimized, and that the layers are the specified thickness.

Inspection and quality assurance must be carefully organized and conducted in case potential problems or questions arise over acceptance of stone material. The engineer and inspectors reserve the right to reject stone at the quarry, at the job site or stockpile, and in place in the structures throughout the duration of the contract. Stone rejected at the job site should be removed from the project site. Stone rejected at the quarry should be disposed or otherwise prevented from mixing with satisfactory stone.

Construction techniques can vary tremendously due to the following factors:

  • Size and scope of the overall project
  • Size and weight of the riprap particles
  • Whether placement is under water or in the dry
  • Physical constraints to access and/or staging areas
  • Noise limitations
  • Traffic management and road weight restrictions
  • Environmental restrictions
  • Type of construction equipment available

Competency in construction techniques and management in all their aspects cannot be acquired from a book. Training on a variety of job sites and project types under the guidance of experienced senior personnel is required. The following sections provide some general information regarding construction of riprap installations and provide some basic information and description of techniques and processes involved.

4.5.3 European Installation Techniques

In Europe, riprap is considered an effective and permanent countermeasure against channel instability and scour, including local scour at bridge piers. Considerable effort has been devoted to techniques for determining size, gradation, layer thickness and horizontal extent, filters, and placement techniques and equipment for riverine and coastal applications (TRB 1999). Engineers in Europe emphasize the need for designing the riprap for a specific site, and in many cases a hydraulic model study will be performed to verify riprap stability. The intensity of turbulence in relation to the structure to be protected is analyzed to assist in developing the most economical riprap design, with larger rock being specified for areas of high turbulence (CUR 1995).

Great care is taken in placing the riprap at critical locations, and in many cases stones are placed individually in the riprap matrix. Highly specialized equipment has been developed by construction contractors in Europe for placing riprap, particularly for coastal installations. The use of bottom dump or side dump pontoons (barges) is common in both Germany and the Netherlands. By loading pontoon "bins" selectively with different sizes of rock, a design gradation in the riprap can be achieved. For large installations, vessels for placing riprap are equipped with dynamic positioning systems using Differential Global Positioning System technology and thrusters to maintain position, and echo sounders (or divers) to verify the coverage of the riprap layer. Some of the smaller pontoon systems, particularly the bottom dump pontoons developed in Germany could be used to place riprap in water at larger bridges (Figure 4.9).

Bottom dump pontoon barge showing riprap supported between the pontoons ready to dump. (TRB 1999).
Figure 4.9. Bottom dump pontoon barge used in Germany for placing riprap (TRB 1999).

4.5.4 Materials

As noted, the best time to control the gradation of the riprap mixture is during the quarrying operation. Generally, sorting and mixing later in stockpiles or at the construction site is not recommended. Inspection of the riprap gradation at the job site is usually carried out visually. Therefore, it is helpful to have a pile of rocks with the required gradation at a convenient location where inspectors can see and develop a reference to judge by eye the suitability of the rock being placed. On-site inspection of riprap is necessary both at the quarry and at the job site to ensure proper gradation and material that does not contain excessive amounts of fines. Breakage during handling and transportation should be taken into account.

The Wolman Count method (Wolman 1954) as described in Section 4.4 may be used as a field test to determine a size distribution based on a random sampling of individual stones within a matrix. This method relies on samples taken from the surface of the matrix to make the method practical for use in the field. The procedure determines frequency by size of a surface material rather than using a bulk sample.

Filter Layer

Geotextile: Either woven or non-woven needle punched fabrics may be used. If a non-woven fabric is used, it must have a mass density greater than 12 ounces per square yard (400 grams per square meter). Under no circumstances may spun-bond or slit-film fabrics be allowed. Each roll of geotextile shall be labeled with the manufacturer's name, product identification, roll dimensions, lot number, and date of manufacture. Geotextiles shall not be exposed to sunlight prior to placement.

Granular filters: Samples of granular filter material shall be tested for grain size distribution to ensure compliance with the gradation specification used in design. Sampling and testing frequency shall be in accordance with the owner or owner's authorized representative.

Subgrade Soils

When placing in the dry, the riprap and filter shall be placed on undisturbed native soil, on an excavated and prepared subgrade, or on acceptably placed and compacted fill. Unsatisfactory soils shall be considered those soils having excessive in-place moisture content, soils containing roots, sod, brush, or other organic materials, soils containing turf clods or rocks, or frozen soil. These soils shall be removed, backfilled with approved material and compacted prior to placement of the riprap. Unsatisfactory soils may also be defined as soils such as very fine noncohesive soils with uniform particle size, gap-graded soils, laminated soils, and dispersive clays, per the geotechnical engineer's recommendations.

4.5.5 Installation
Subgrade Preparation

As noted, the subgrade soil conditions shall meet or exceed the required material properties described in Section 4.5.4 prior to placement of the riprap. Soils not meeting the requirements shall be removed and replaced with acceptable material.

When placing in the dry, the areas to receive the riprap shall be graded to establish a smooth surface and ensure that intimate contact is achieved between the subgrade surface and the filter, and between the filter and the riprap. Stable and compacted subgrade soil shall be prepared to the lines, grades and cross sections shown on the contract drawings. Termination trenches and transitions between slopes, embankment crests, benches, berms and toes shall be compacted, shaped, and uniformly graded. The subgrade should be uniformly compacted to the geotechnical engineer's site-specific requirements.

When placing under water, divers shall be used to ensure that the bed is free of logs, large rocks, construction materials, or other blocky materials that would create voids beneath the system. Immediately prior to placing the filter and riprap system, the prepared subgrade must be inspected.

Placing the Filter

Whether the filter is comprised of one or more layers of granularmaterial or made of geotextile, its placement should result in a continuous installation that maintains intimate contact with the soil beneath. Voids, gaps, tears, or other holes in the filter must be avoided to the extent practicable, and replaced or repaired when they occur.

Placement of Geotextile: The geotextile shall be placed directly on the prepared area, in intimate contact with the subgrade. When placing a geotextile, it should be rolled or spread out directly on the prepared area and be free of folds or wrinkles. The rolls shall not be dragged, lifted by one end, or dropped. The geotextile should be placed in such a manner that placement of the overlying materials (riprap and/or bedding stone) will not excessively stretch or tear the geotextile.

After geotextile placement, the work area shall not be trafficked or disturbed in a manner that might result in a loss of intimate contact between the riprap stone, the geotextile, and the subgrade. The geotextile shall not be left exposed longer than the manufacturer's recommendation to minimize potential damage due to ultraviolet radiation; therefore, placement of the overlying materials should be conducted as soon as practicable.

The geotextile shall be placed so that upstream strips overlap downstream strips. Overlaps shall be in the direction of flow wherever possible. The longitudinal and transverse joints shall be overlapped at least 1.5 feet (46 cm) for dry installations and at least 3 feet (91 cm) for below-water installations. If a sewn seam is to be used for the seaming of the geotextile, the thread to be used shall consist of high strength polypropylene or polyester and shall be resistant to ultraviolet radiation. If necessary to expedite construction and to maintain the recommended overlaps anchoring pins, "U"-staples or weights such as sandbags shall be used. Figure 4.10 illustrates the placing of a geotextile for a coastal shoreline application.

Photograph of workers pulling geotextile into place down a sloped embankment. The longitudinal seam in the geotextile can be seen as well as riprap placing equipment and riprap ready to be placed.
Figure 4.10. Hand placing geotextile prior to placing riprap. Note sewn seam.

Placing Geotextiles Under Water: Placing geotextiles under water can be problematic for a number of reasons. Most geotextiles that are used as filters beneath riprap are made of polyethylene or polypropylene. These materials have specific gravities ranging from 0.90 to 0.96, meaning that they will float unless weighted down or otherwise anchored to the subgrade prior to placement of the riprap (Koerner 1998).

Flow velocities greater than about 1.0 ft/s (0.3 m/s) create large forces on the geotextile. These forces cause the geotextile to act like a sail, often resulting in wavelike undulations of the fabric (a condition that contractors refer to as "galloping") that are extremely difficult to control. The preferred method of controlling geotextile placement is to isolate the work area from river currents by a temporary cofferdam. In mild currents, geotextiles precut to length can be placed by divers, with sandbags to hold the filter temporarily.

Placement of Granular Filter: When placing a granular filter, front-end loaders are the preferred method for dumping and spreading the material on slopes milder than approximately 1V:4H. A typical minimum thickness for granular filters is 0.5 to 1.0 feet (0.15 to 0.3 m), depending on the size of the overlying riprap and whether a layer of bedding stone is to be used between the filter and the riprap. When placing a granular filter under water, the thickness should be increased by 50%. Placing granular media under water around a bridge pier is best accomplished using a large diameter tremie pipe to control the placement location and thickness, while minimizing the potential for segregation. NOTE: For riverine applications where dune-type bed forms may be present, it is strongly recommended that only a geotextile filter be considered.

Placing the Riprap

Riprap may be placed from either land-based or water-based operations and can be placed under water or in the dry. Special-purpose equipment such as clamshells, orange-peel grapples, or hydraulic excavators (often equipped with a "thumb") is preferred for placing riprap. Unless the riprap can be placed to the required thickness in one lift using dump trucks or front-end loaders, tracked or wheeled vehicles are discouraged from use because they can destroy the interlocking integrity of the rocks when driven over previously placed riprap. Water-based operations may require specialized equipment for deep-water placement, or can use land-based equipment loaded onto barges for near-shore placement. In all cases, riprap should be placed from the bottom working toward the top of the slope so that rolling and/or segregation does not occur as shown in Figure 4.11.

Photograph of riprap being placed with tracked hydraulic excavators on a sloped embankment
Figure 4.11. Placing riprap with hydraulic excavators.

Riprap Placement on Geotextiles: Riprap should be placed over the geotextile by methods that do not stretch, tear, puncture, or reposition the fabric. Equipment should be operated to minimize the drop height of the stone without the equipment contacting and damaging the geotextile. Generally, this will be about 1 foot of drop from the bucket to the placement surface (ASTM Standard D 6825). Further guidance on recommended strength properties of geotextiles as related to the severity of stresses during installation are provided in Part 1 of this document. When the preferred equipment cannot be utilized, a bedding layer of coarse granular material on top of the geotextile can serve as a cushion to protect the geotextile. Material comprising the bedding layer must be more permeable than the geotextile to prevent uplift pressures from developing.

Riprap Placement Under Water: Riprap placed in water requires close observation and increased quality control to ensure a continuous well-graded uniform rock layer of the required thickness (ASTM Standard D6825). A systematic process for placing and continuous monitoring to verify the quantity and layer thickness is important. Typically, riprap thickness is increased by 50% when placement must occur under water.

Excavation, grading, and placement of riprap and filter under water require additional measures. For installations of a relatively small scale, diversion of the stream around the work area can be accomplished during the low flow season. For installations on larger rivers or in deeper water, the area can be temporarily enclosed by a cofferdam, which allows for construction dewatering if necessary. Alternatively, a silt curtain made of plastic sheeting may be suspended by buoys around the work area to minimize environmental degradation during construction.

Depending on the depth and velocity of the water, sounding surveys using a sounding pole or sounding basket on a lead line, divers, sonar bottom profiles, and remote operated vehicles (ROV) can provide some information about the riprap placement under water.


Detailed guidance for inspecting riprap installations is provided in NCHRP Report 593 (Lagasse et al. 2007). The guidance includes inspection during construction, periodic inspection, and inspection after flood events (see Volume 1, Appendix D).

4.5.6 Measurement and Payment

Riprap satisfactorily placed can be paid for based on either volume or weight. When using a weight basis, commercial truck scales capable of printing a weight ticket including time, date, truck number, and weight should be used. When using a volumetric basis, the in-place volume should be determined by multiplying the area, as measured in the field, of the surface on which the riprap was placed, by the thickness of the riprap measured perpendicular as dimensioned on the contract drawings.

In either case, the finished surface of the riprap should be surveyed to ensure that the as-built lines and grades meet the design plans within the specified tolerance. Survey cross-sections perpendicular to the axis of the structure are usually taken at specified intervals. All stone outside the limits and tolerances of the cross sections of the structure, except variations so minor as not to be measurable, is deducted from the quantity of new stone for which payment is to be made. In certain cases, excess stone may be hazardous or otherwise detrimental; in this circumstance, the contractor must remove the excess stone at his own expense. Payment will be full compensation for all material, labor, and equipment to complete the work.


Rock-filled trenches are structures used to protect banks from caving caused by erosion at the toe. A trench is excavated along the toe of the bank and filled with rocks as shown in Figure 4.12. The size of trench to hold the rock fill depends on expected depths of scour.

As the streambed adjacent to the toe is eroded, the toe trench is undermined and the rock fill slides downward to pave the bank. It is advantageous to grade the banks before placing riprap on the slope and in the toe trench. The slope should be at such an angle that the saturated bank is stable while the stream stage is falling.

Sketch in cross section of riprapped embankment slope and toe of bank trench. Shown is both before toe erosion and after toe of slope erosion with riprap movement into eroded bed section. (after Richardson et al. 2001).
Figure 4.12. Rock-filled trench (after Richardson et al. 2001).

An alternative to a rock-fill trench at the toe of the bank is to excavate a trench above the water line along the top of the bank and fill the trench with rocks. As the bank erodes, stone material in the trench is added on an as-needed basis until equilibrium is established. This method is applicable in areas of rapidly eroding banks of medium to large size streams. Note that if a geotextile filter is used beneath the entire width of the trench, it will remain in place as adjustment occurs, whereas a granular filter is likely to be removed by particle displacement.

Windrow revetment (Figure 4.13) consists of a supply of rock deposited along an existing bank line at a location beyond which additional erosion is to be prevented. When bank erosion reaches and undercuts the supply of rock, it falls onto the eroding area, thus giving protection against further undercutting. The resulting bank line remains in a near natural state with an irregular appearance due to intermittent lateral erosion in the windrow location. The treatment particularly lends itself to the protection of adjacent wooded areas, or placement along stretches of presently eroding, irregular bank line.

The effect of windrow revetment on the interchange of flow between the channel and overbank areas and flood flow distribution in the flood plain should be carefully evaluated. Windrow installations will perform as guide banks or levees and may adversely affect flow distribution at bridges or cause local scour. Tying the windrow to the highway embankment at an abutment would be contrary to the purpose of the windrow since the rock is intended to fall into the channel as the bank erodes. This would potentially expose the abutment.

Note that the final configuration and thickness of the layer of launched stone is completely uncontrolled. In addition, there is no possibility of establishing any kind of filter (neither geotextile nor granular) with this type of placement.

The following observations and conclusions from model investigations of windrow revetments and rock-fill trenches may be used as design guidance. More definitive guidance is not presently available (USACE 1981).

  • Application rate of stone is a function of channel depth, bank height, material size, and estimated bed scour.
  • A triangular windrow is the least desirable shape, a trapezoidal shape provides a uniform blanket of rock on an eroding bank, and a rectangular shape provides the best coverage. A rectangular shape is most easily placed in an excavated trench.

Sketch showing three stages of riprap windrow revetment: cross section of initial installation of rock riprap into longitudinal trench along bank - riprap is covered with mulch and fill; riprap windrow as the bank is eroded and riprap is launched down the bank; cross section showing the initial streambank and the erosion of streambank with riprap facing the eroded slope
Figure 4.13. Windrow revetment, definition sketch (after USACE 1981).

  • Bank height does not significantly affect the final revetment; however, high banks tend to produce a nonuniform revetment alignment. Large segments of bank tend to break loose and rotate slightly on high banks, whereas low banks simply "melt" or slough into the stream.
  • Stone size influences the thickness of the final revetment, and a smaller gradation of stone forms a more dense, closely chinked protective layer. Stones must be large enough to resist being transported by the stream, and a well-graded stone should be used to ensure that the revetment does not fail from leaching of the underlying bank material. Large stone sizes require more material than smaller stone sizes to produce the same relative thickness of revetment. In general, the greater the stream velocity, the steeper the side slope of the final revetment. The final revetment slope will be about 15% flatter than the initial bank slope.
  • A windrow segment should be extended landward from the upstream end to reduce the possibility of outflanking of the windrow.

Rigid revetments are generally smoother than flexible revetments and thus improve hydraulic efficiency and are generally highly resistant to erosion and impact damage. They are susceptible to damage from the removal of foundation support by subsidence, undermining, hydrostatic pressures, slides, and erosion at the perimeter. They are also among the most expensive streambank protection countermeasures. For the above reasons, rigid erosion protection measures such as cast-in-place concrete, fully grouted riprap, and rigid grout-filled mats are generally not recommended for bankline revetment applications.

Note that partially-grouted riprap is considered flexible in that its construction is designed to allow breaking of the partial grout, under stress, to result in conglomerate particles which are much larger than the individual stones of the matrix (see Design Guideline 12). Additional guidance on rigid revetments in this document include:

  • Soil Cement - Design Guideline 7
  • Grout-Filled Mattresses - Design Guideline 9

Concrete paving should be used only where the toe can be adequately protected from undermining and where hydrostatic pressures behind the paving will not cause failure. This might include impermeable bank materials and portions of banks which are continuously under water. Sections intermittently above water should be provided with weep holes.


Burlap sacks filled with soil or sand-cement mixtures have long been used for emergency work along levees and streambanks during floods (Figure 4.14). Commercially manufactured sacks (burlap, paper, plastics, etc.) have been used to protect streambanks in areas where riprap of suitable size and quality is not available at a reasonable cost. Sacks filled with sand-cement mixtures can provide long-term protection if the mixture has set up properly, even though most types of sacks are easily damaged and will eventually deteriorate. Sand-cement sack revetment construction is not economically competitive in areas where good stone is available. However, where quality riprap must be transported over long distances, sack revetment can often be placed at a lesser cost than riprap.

Sketch in cross section of sand-cement bag revetment of a sloped embankment. Sketch shows: the extents from design high water to below scour elevation (below bed); extension back into the bank and overlapping sack placement.
Figure 4.14. Typical sand-cement bag revetment.

If a sack revetment is to be constructed, the sacks should be filled with a mixture of 15% cement (minimum) and 85% dry sand (by weight). The filled sacks should be placed in horizontal rows like common house brick beginning at an elevation below any toe scour (alternatively, riprap can be placed at the toe to prevent undermining of the bank slope). The successive rows should be stepped back approximately one-half-bag width to a height on the bank above which no protection is needed. The slope of the completed revetment should not be steeper than 1:1. After the sacks have been placed on the bank, they can be wetted down for a quick set or the sand-cement mixture can be allowed to set up naturally through rainfall, seepage or condensation. If cement leaches through the sack material, a bond will form between the sacks and prevent free drainage. For this reason, weepholes should be included in the revetment design. The installation of weepholes will allow drainage of groundwater from behind the revetment thus helping to prevent a pressure buildup that could cause revetment failure. This revetment requires the same types of toe protection as other types of rigid revetment.


Bunte, K. and Abt, S.R., 2001, "Sampling Surface and Subsurface Particle-Size Distributions in Wadable Gravel- and Cobble-Bed Streams for Analyses in Sediment Transport, Hydraulics, and Streambed Monitoring," Gen. Tech. Rep. RMRS-GTR-74, U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station, Fort Collins, CO, 428 p.

Centre for Civil Engineering Research and Codes (CUR), 1995, "Manual on the Use of Rock in Hydraulic Engineering," CUR/RWS Report 169, Road and Hydraulics Division, A.A. Balkema Publishers, Rotterdam, Netherlands.

Comité Européen de Normalisation (CEN), 2002, "European Standard for Armourstone," Report prEN 13383-1, Technical Committee 154, Brussels, Belgium.

Federal Highway Administration, 1981, "Standard Specifications for Construction of Roads and Bridges on Federal Highway Projects," U.S. Department of Transportation, FP-79 (Revised June 1981, Washington D.C.

Galay, V.J., Yaremko, E.K, and Quazi, M.E., 1987, "River Bed Scour and Construction of Stone Riprap Protection," in Hey, R.D., Bathurst, J.C., and Thorne, C.R. (Eds.), Sediment Transport in Gravel-Bed Rivers, John Wiley & Sons Ltd., pp. 353-383.

Kellerhals, R. and Bray, D., 1971, "Sampling Procedures for Coarse Fluvial Sediments," Proc. Am. Soc. Civ. Engrs., J. Hyd. Div., 97(HY7).

Koerner, R.M., 1998, "Designing with Geosynthetics," 4th Edition, Prentice-Hall, Inc., Englewood Cliffs, NJ, 761 p.

Lagasse, P.F., Clopper, P.E., Zevenbergen, L.W., and Ruff, J.F., 2006, "Riprap Design Criteria, Recommended Specifications, and Quality Control," NCHRP Report 568, Transportation Research Board, National Academies of Science, Washington, D.C.

Lagasse, P.F., Clopper, P.E., Zevenbergen, L.W., and Girard, L.G., 2007, "Countermeasures to Protect Bridge Piers from Scour," NCHRP Report 593, Transportation Research Board, National Academies of Science, Washington, D.C.

Maynord, S.T., Ruff, J.F., and Abt, S.R., 1989, "Riprap Design," ASCE Journal of Hydraulic Engineering, Vol. 115, No. 7, pp 937-939.

Maynord, S.T., 1990, "Riprap Stability Results from Large Test Channel," Hydraulic Engineering, Proceedings of the 1990 ASCE National Conference, Volume 1, Chang, H.H. and Hill, J.C. (eds), San Diego, CA.

Racin, J.A., Hoover, T.P., and Crossett-Avila, C.M., 2000, "California Bank and Shore Rock Slope Protection Design," Final Report No. FHWA-CA-TL-95-10, Caltrans Study No. F90TL03 (Third Edition - Internet), California Department of Transportation, Sacramento, CA.

Richardson, E.V., Simons, D.B., and Lagasse, P.F., 2001, "River Engineering for Highway Encroachments - Highways in the River Environment," Report No. FHWA NHI 01-004, Hydraulic Design Series No. 6, Federal Highway Administration, Washington, D.C.

Transportation Research Board (TRB), 1999, "1998 Scanning Review of European Practice for Bridge Scour and Stream Instability Countermeasures," National Cooperative Highway Research Program, Research Results Digest, Number 241, Washington, D.C.

U.S. Army Corps of Engineers, 1981, "The Streambank Erosion Control Evaluation and Demonstration Act of 1974," Final Report to Congress, Executive Summary and Conclusions.

U.S. Army Corps of Engineers, 1991, "Hydraulic Design of Flood Control Channels," EM 1110-2-1601, Department of the Army, Washington, D.C.

U.S. Army Corps of Engineers, 1995, "Construction Quality Management," Engineering Regulation No. 1180-1-6, Washington, D.C.

Wolman, M.G., 1954, "A Method of Sampling Coarse Bed Material," American Geophysical Union, Transactions, 35: pp. 951-956.

Updated: 09/21/2011

United States Department of Transportation - Federal Highway Administration