Skip to contentUnited States Department of Transportation - Federal Highway Administration FHWA Home
Research Home
Public Roads
Featuring developments in Federal highway policies, programs, and research and technology.
This magazine is an archived publication and may contain dated technical, contact, and link information.
Federal Highway Administration > Publications > Public Roads > Vol. 70 · No. 6 > Testing Bottomless Culverts

May/Jun 2007
Vol. 70 · No. 6

Publication Number: FHWA-HRT-07-004

Testing Bottomless Culverts

by Kornel Kerenyi and Jorge Pagán-Ortiz

The Maryland State Highway Administration and FHWA partnered to study the effects of scour on these environmentally friendly and economic roadway structures.

This bottomless culvert on Whitehall Road over Euclid Creek in Cuyahoga County, OH, is both attractive and environmentally friendly. The culvert spans 12.8 meters (42 feet), rises 3.4 meters (11 feet), and is 12.8 meters (42 feet) long. Photo: CONTECH.
(Above) This bottomless culvert on Whitehall Road over Euclid Creek in Cuyahoga County, OH, is both attractive and environmentally friendly. The culvert spans 12.8 meters (42 feet), rises 3.4 meters (11 feet), and is 12.8 meters (42 feet) long. Photo: CONTECH.

Although hydraulic considerations form the basis of culvert design, the behavior of fish, the characteristics of local soils, and the cost of construction can influence the type of culverts used on highway applications at stream crossings. Bottomless, or three-sided, culverts use the stream's natural channel bed and often are economic and environmentally attractive alternatives to traditional closed culverts.

"We make the selection of culvert type based on all factors involved, taking into account both environmental and engineering issues," says Andrzej Kosicki, assistant chief of the Bridge Design Division, Maryland State Highway Administration (SHA). "Bottomless culverts may be right in some situations, while paved invert culverts may be right in other cases. If we have good soils, especially when the structure foundations can be placed directly on the underlying bedrock without unreasonably high expense, we are more likely to use a bottomless culvert. In order to consider them totally environmentally friendly, we also would look into channel stability questions."

Two counties in Maryland — Frederick and Charles Counties — currently are using bottomless culverts on selected stream crossings. These structures typically are supported on spread footings, which have large lateral dimensions that distribute loads to the surrounding soils. Because of the use of spread footings, designers need to address as part of their designs the depth of the footing that will be required to resist to the effects of scour caused by the flowing water. The scour problem is similar to abutment and contraction scour in a bridge opening and can be treated in much the same way.

"In situations where we have to deal with highly erodible, easily scourable soils," says Kosicki, "we are more likely to consider paved bottom culverts unless our studies indicate that foundations can be designed to protect the structure from scour at a reasonable cost."

He adds, "In the mid-1990s, Stan Davis and Dr. Fred Chang, both contractors to Maryland SHA at that time, developed the computer program Abutment Scour (ABSCOUR) to conduct all of our scour studies. But because the equations were derived theoretically from limited experimental data, we wanted to validate them in a lab setting. So about 5 years ago, Stan Davis contacted the Federal Highway Administration's [FHWA] J. Sterling Jones, then the manager of FHWA's Hydraulic Research Laboratory." FHWA and Maryland SHA arranged funding for a cooperative study. Kosicki adds, "FHWA used some excellent equipment to make the scour measurements and to analyze the data obtained."

Design of the Study

Two suppliers, CONTECH® Construction Products, Inc., and CON/SPAN®, agreed to provide models of the configurations that State departments of transportation typically use for highway applications. The primary objective was to compare results from a simple rectangular shape to the results from shapes that are available commercially. For this effort, the FHWA researchers focused on flows assumed to be perpendicular to the roadway.

A major consideration in estimating scour in culverts with open bottoms is the flow distribution at the entrance of the culvert, especially when there is side flow that is being funneled as it passes through the culvert opening.

FHWA researchers conducted the study in two phases. The first focused on measuring maximum scour depths at the culvert entrance and developing an analysis procedure using methods described in the Bottomless Culvert Scour Study: Phase I (FHWA-RD-02-078) to approximate prescour hydraulic parameters. The researchers conducted submerged entrance or fixed-bed experiments during this first phase to measure prescour hydraulic parameters.

The second phase expanded the investigation to include scour measurements at the entrance and outlets for submerged flow conditions. At the locations where maximum scour occurred, the second phase also involved detailed measurements of velocity with a prescour fixed bed. The researchers conducted additional tests to evaluate the use of various measures to reduce scour, including wing walls (short retaining walls that guide a stream into a culvert), pile dissipators (vertical arrays of circular piles buried just below the channel bed), riprap (rough stones placed to prevent scour), and cross vanes (sets of upstream-angled lines of boulders, connected by a section of smaller rocks upstream).

This two-dimensional diagram shows the flow into a culvert. The left side of the diagram is a rectangular approach area to the culvert, which is depicted as a smaller rectangle abutting the center of the approach area. The primary flow, indicated by arrows, enters the approach area, curves into the culvert entrance, narrows in the initial portion of the culvert, and then expands to fill the culvert. In the initial portion of the culvert where the primary flow narrows, secondary flows, or eddies, occur. The secondary flows are indicated by pinwheels at the sides of the culvert, just inside the opening.
In this diagram, water flows from the left into a culvert. Embankments constrict the primary flow near the upstream entrance and force the water through the opening. This flow contraction creates a secondary flow of eddies and strong turbulence at the sides of the culvert just downstream of the opening. Source: FHWA.
The two-dimensional diagram shows water levels in a culvert after scour. The view is into the culvert entrance, which is in the center of the diagram. Each side of the culvert entrance is a straight vertical line. The top of the culvert entrance is formed by an extension of the side lines to join and form a semiellipse facing downward. The width of the culvert entrance is labeled Width. The approach section to the culvert entrance extends to the left and right on each side of the entrance. The width of the approach section is labeled Width, Approach. The water surface in the approach section is indicated by a horizontal line that intersects the culvert entrance near the entrance's top. Three beds are indicated in the diagram. A horizontal line to the left and right of the culvert entrance, and a short distance above the bottom of the culvert entrance, is the original bed. The other two beds are inside the culvert entrance. The theoretical bed for contraction scour is indicated by a horizontal line that is a short distance above the bottom of the culvert entrance but below the level of the original bed line. The actual bed with local scour is the other bed inside the culvert entrance. At each end, the actual bed with local scour is above the bottom of the culvert entrance but below the horizontal line indicating the theoretical bed for contraction scour. In the approximate center two-thirds of the culvert entrance, the actual bed with local scour is above the theoretical bed for contraction scour. The water surface in the culvert prior to scour is lower than the water surface in the approach section. The distance from the water surface in the culvert to the original bed is labeled y subscript o. The distance from the water surface in the culvert to the theoretical bed for contraction scour is labeled y subscript max. The distance from the water surface in the culvert to the highest point of the actual bed with local scour is labeled y subscript 2. The distance between the original bed and the theoretical bed for contraction scour is labeled y subscript s.

Where:

y1 = flow depth in the approach channel at a distance of 3 times the culvert width upstream of the culvert
y0 = water depth at the culvert entrance before scour occurs
ymax = maximum water depth in the culvert after scour hole develops
y2 = equilibrium water depth after scour hole develops
yS = maximum depth of scour in the culvert

This definition sketch shows a cross section of a flume after a scour experiment for unsubmerged flow conditions. Inside the culvert, measurements show the actual bed with local scour and a theoretical bed for contraction scour. Source: FHWA.

Presentations of status reports on this study to drainage engineers at meetings of the American Association of State Highway and Transportation Officials and at hydraulics conferences revealed a widespread interest in the research, because this type of culvert is considered less intrusive to the environment.

Scour Is a Critical Concern

Experiments show that scour generally is deepest near the corners of the upstream entrance to a culvert because of the contraction or narrowing of the water flow. The vortices and strong turbulence just downstream of the inlet, generated by the contraction of flow typically called secondary flow, occurs in the so-called separation zone that is between the side walls of the culvert and the primary flow. This flow pattern is similar to the one observed at bridge abutments that sometimes can lead to large amounts of scour at or near an abutment.

Two reports, Testing Abutment Scour Model and the Bottomless Culvert Scour Study: Phase I, have suggested that bridge abutment scour can be analyzed as a form of equilibrium scour related to the primary flow by incorporating an empirical adjustment factor to account for vorticity (the swirling motion of a vortex) and turbulence. The earlier research mentioned above established that this empirical adjustment factor could be derived from laboratory results. The FHWA researchers built on this previous work to formulate the theoretical background for analyzing data on culvert scour.

Researchers used this Plexiglas® model of a rectangular bottomless culvert for measuring velocity and depth in a controlled environment. The model's upstream entrance is in the foreground, and the bottom of the flume is sand.
Researchers used this Plexiglas® model of a rectangular bottomless culvert for measuring velocity and depth in a controlled environment. The model's upstream entrance is in the foreground, and the bottom of the flume is sand.

The first variable used in the analysis for unsubmerged flow conditions is the unit discharge, which remains constant as the scour hole develops. If no sediment is being transported into the scour hole, as was the case with all of the experiments in the study, then no sediment could be transported out of the scour hole after equilibrium scour develops. In this case, the local velocity would be reduced to the critical velocity (threshold velocity at which sediment will erode) for the sediment size after the equilibrium scour depth develops. Other variables include the representative (local) velocity at the entrance of the culvert and the critical velocity at which incipient sediment motion occurs.

Also important to note is the assumed representative unit discharge across the scour hole at the beginning of scour. Once the researchers determined the representative velocity and the critical incipient motion velocity, they could derive the equilibrium flow depth that reflects the scour attributed to the incoming flow distribution. The researchers considered three alternative equations for the representative velocity: the average velocity in the culvert inlet, the potential flow velocity (flow theory that is characterized by no curl throughout the velocity field), and the measured flow velocity.

The ABSCOUR program uses the average velocity in the culvert for the representative velocity. This average velocity is the volumetric flow rate divided by the cross-sectional area of flow in the culvert. Chang, in his presentation "Maryland SHA Procedure for Estimating Scour at Bridge Abutments Part 2 — Clear Water Scour," published in the 1998 ASCE Proceedings of the International Water Resources Engineering Conference, used potential flow principles to derive a velocity adjustment expression to approximate the representative velocity that should be used for computing bridge abutment scour. This adjustment compensates for the flow contraction at the culvert inlet. The FHWA researchers adapted Chang's expression for use in analyzing bottomless culverts. Because the present study produced accurate measurements for the local (representative) velocities in the approach section of the culvert, the researchers adjusted the potential flow theory to match the measured flow velocities at the corners of the culvert inlet.

The researchers looked at two alternatives for calculating the threshold velocity at which sediment will erode. The first was a method described by E.M. Laursen in his article "An Analysis of Relief Bridge Scour," published by the American Society of Civil Engineers in the Journal of the Hydraulics Division in 1963. The other was based on research published by C.R. Neill in the 1973 publication Guide to Bridge Hydraulics. Laursen's equation for the critical velocity is summarized in appendix C of the FHWA Hydraulic Engineering Circular No. 18.

Scour, usually deepest near the corners at the upstream entrance, is evident after testing of this rectangular culvert model with 45-degree wing walls. The rectangle is surrounded by sediment that is lowest at the edges of the rectangle and peaks at its center.
Scour, usually deepest near the corners at the upstream entrance, is evident after testing of this rectangular culvert model with 45-degree wing walls. The rectangle is surrounded by sediment that is lowest at the edges of the rectangle and peaks at its center.

Neill developed a family of curves for estimating critical velocities for sediments for varying flow depths with grain sizes ranging from 0.3 to 300 millimeters (0.01 to 11.7 inches). Neill defined the critical velocity as the flow velocity just fast enough to move the bed material and used a combination of field data and laboratory data to develop his family of curves. He used an equation similar to Laursen's to estimate the critical velocity for grain sizes greater than about 30 millimeters (1.2 inches). For a grain size of 0.3 millimeter, Neill used design equations developed from field data collected in stable, fine sediment canals in Pakistan for estimating the critical velocity. Having defined critical velocities for a grain size of 0.3 millimeter and for grain sizes greater than 30 millimeters, he hand-drew transition curves for grain sizes between 0.3 and 30 millimeters.

This research on scour in bottomless culverts revealed that the maximum scour depth, measured at the corners of the culvert, was always greater than the computed equilibrium depth, regardless of which equations for representative velocity and critical velocity were used. Thus, an empirical coefficient, similar to Chang's in ABSCOUR, was needed to explain the additional scour depth.

Measured Representative Velocities

The y-axis is measured representative velocity, 0 to 1.2 meters open parenthesis 0 to 4 feet close parenthesis per second. The x-axis is potential representative velocity and ranges from 0 to 1.2 meters open parenthesis 0 to 4 feet close parenthesis per second. The plotted data points fall between x-axis coordinates of 1 and 3 and y-axis coordinates of 1 and 3.7. A regression line through the data points rises from left to right. The regression equation of the line is measured representative velocity equals 1.2781 times potential representative velocity.
This chart compares measured representative velocities near the upstream corners of culverts to representative velocities computed based on potential flow theory. The plot shows that the measured velocities are approximately 30 percent higher than the computed velocities. Source: FHWA.

Model Setup

The experiments were conducted in a rectangular flume 21.35 meters (70 feet) long and 1.83 meters (6 feet) wide with a recessed section 2.4 meters (8 feet) long and 1.83 meters (6 feet) wide to allow for scour hole formation. A 9.15-meter (30-foot) approach section from the head box to the test section consisted of foam boards constructed 0.1 meter (4 inches) above the stainless steel flume bottom. The walls of the flume were constructed from smooth glass. The flume was set at a constant slope of 0.04 percent, and an adjustable tailgate located at the downstream end of the flume controlled the depth of flow. Flow was supplied by a pumping system with a maximum output rate of 0.3 cubic meter per second (10 cubic feet per second). An electromagnetic flow meter measured the discharge. A 13-millimeter (0.5-inch) spherical electromagnetic velocity sensor measured equivalent two-directional mean velocities in a plane parallel to the flume bed. Scour maps were generated using a laser distance sensor mounted on an automated flume carriage fitted to the main flume.

The upper part of the figure shows a scour map, or plan, view, using laser distance sensor. The lower part of the figure is a longitudinal scour profile of the culvert and scour. For each figure, the approach bed and culvert entrance are to the left, the culvert barrel is in the center, and the culvert exit and exit bed are to the right. In the upper part of the figure, culvert entrance is 600 millimeters open parenthesis 2 feet close parenthesis wide, and the culvert barrel is 1,600 millimeters open parenthesis 5.3 feet close parenthesis long. Areas of darker shading indicating scour are found at the corners of the culvert entrance; in the exit bed, a pronounced area of scour is near the center of the figure, and a slightly smaller area of scour is just below the pronounced area.
This scour map and profile for a bottomless culvert with wing walls show that the deepest scour (darkest shading) was observed at the upstream corners and downstream of the culvert outlet. Source: FHWA.

Three bottomless culvert shapes — rectangular, CON/SPAN, and CONTECH models — were constructed of Plexiglas® and tested for the first phase study. All three were evaluated with and without 45-degree wing walls. Marine-grade plywood was used for the headwalls and wing walls of the models. The models were mounted in the centerline of the flume.

For the second phase experiments, the laboratory model consisted of a rectangular bottomless culvert 0.60 meters (2 feet) wide and 0.15 meters (0.49 feet) high that was mounted in the centerline of the flume. The culvert and headwall of the model were constructed of Plexiglas, and the wing walls were made from polystyrene foam.

Researchers conducted steady flow outlet scour experiments for approach flow depths ranging from 0.10 to 0.23 meter (0.33 to 0.75 foot) and approach velocities ranging from 0.07 to 0.16 cubic meter per second (0.25 to 0.57 cubic foot per second). The discharges to obtain the approach flow conditions varied from approximately 0.026 to 0.080 cubic meter per second (0.9 to 2.8 cubic feet per second). Particle size was set at 2.0 millimeters (0.08 foot) for the outlet scour experiments. The researchers tested several scour countermeasure configurations, including various wing wall angles, use of pile dissipators (vertical array of circular piles buried just below the channel bed), and the Maryland Standard Plan as a scour countermeasure design. The Maryland Standard Plan employs wing walls at the inlet and outlet of the culvert and lines the wing walls and the inside walls of the culvert with riprap.

The y-axis is the dimensionless empirical adjustment factor k subscript s using measured representative velocity and Neills critical velocity curves as a function of an independent variable using blocked discharge normalized by the acceleration of gravity and the computed equilibrium scour, and ranges from 0 to 2. The x-axis is the dimensionless independent variable using blocked discharge normalized by the acceleration of gravity and the computed equilibrium scour, and ranges from 0 to 1.6. Two sets of data points are plotted, one for bottomless culverts with wing walls and one for bottomless culverts without wing walls. Most of the data points are located in the upper left quadrant of the graph. The regression equation for a line through the no wing walls data points is y equals the product of 1.44 times the 0.23 power of x. The regression equation for a line through the with wing walls data points is y equals the product of 1.52 times the 0.06 power of x.
The two sets of data plots on the chart compare the scour amplification factor against an independent variable for a bottomless culvert model with wing walls and a model without wing walls. Source: FHWA.

Speed and Turbulence

Representative velocities near the upstream corners of the culverts were measured using prescour fixed-bed experiments. Researchers then compared the measured values to the values from the potential flow theory that allows derivation of theoretical velocity values. to derive a multiplier. A linear regression of the results shows that measured velocities for bottomless culvert applications are 1.28 times the velocities derived from potential flow theory.

The researchers performed extensive scour analysis using various combinations of equations for resultant velocity and critical velocity to determine the empirical adjustment factor to account for turbulence and vorticity at the upstream corner of the culvert. They used a laser distance sensor to generate scour maps through the entire bottomless culvert model and observed deepest scour at the upstream corners and downstream of the culvert outlet.

The researchers conducted a regression analysis for vertical face and wing wall entrance datasets and derived separate equations for the two datasets. They compared the scour amplification factor against an independent variable using the discharge blocked by the embankment normalized by the acceleration of gravity and the computed equilibrium scour depth.

Researchers found that the amplification factor is higher for a culvert without wing walls, indicating deeper scour compared to a culvert with 45-degree wing walls. Wing walls at the culvert entrance guide the flow more smoothly into the culvert, reducing turbulence and associated scour at the upstream corners.

Signs of significant scour appear at the downstream outlet of this model with 45-degree wing walls.
Signs of significant scour appear at the downstream outlet of this model with 45-degree wing walls.

Traditionally, wing walls have been constructed with highway culverts to increase flow capacity. The inlet configuration is implicit in the inlet control equations and reduces the severity of erosion and scour of the channel and adjacent banks at both the inlet and outlet. The researchers investigated various inlet and outlet wing wall configurations under unsubmerged flow conditions to determine the overall effects of wall shape, length, and orientation on scour hole formation.

Experimental results indicate that turbulence is reduced, and "vortex shedding" caused by abrupt changes in pressure is almost eliminated with the use of a streamlined shape for the wing walls. In other words, the streamlined wall eliminates flow separation and decreases turbulence.

With the streamlined bevel, vortices do not propagate downstream, and the resulting turbulence is more evenly distributed and not concentrated in a single location. With sharp corners at the exit, the resulting abrupt change in pressure induces vortex shedding and increases scour depths. Researchers also tested 8-degree outlet wing walls, because streamlined wing walls may be impractical in the field. The 8-degree outlet wing wall results also revealed reduced turbulence and scour depth at the outlet.

Significant Findings

Using the flow distribution at the culvert entrance to compute the primary scour depth component and adjusting with an empirical factor based on laboratory data appears to be valid for bottomless culverts. Culvert shapes tested in these experiments did not significantly influence the scour, but the entrance conditions did. The use of 45-degree inlet wing walls decreases the scour at the upstream corners considerably.

The experimental setup had some limitations. These experimental results are based on laboratory flume experiments with a flat approach cross section with uniform flow conveyance, which is not typical of field conditions. The experiments also were conducted under clear-water approach flow conditions with no sediment being transported into the culvert, which models a "worst case" condition.

Scour map and profile, for free surface flow through the culvert with streamlined wing wall at exit. The upper part of the figure shows a scour map, or plan, view, using laser distance sensor. The lower part of the figure is a longitudinal scour profile of the culvert and scour. The approach bed and culvert entrance are to the left, the culvert barrel is in the center, and the culvert exit and exit bed are to the right. In the upper part of the figure, the culvert entrance is 600 millimeters open parenthesis 2 feet close parenthesis wide, and the culvert barrel is 1,600 millimeters open parenthesis 5.3 feet close parenthesis long. Areas of darker shading indicating significant scour are found at the corners of the culvert entrance and reduced scour at the exit.
For a bottomless culvert with a streamlined wing wall, this scour map and profile show a diminished flow separation and decreased turbulence, resulting in reduced scour at the outlet. Source: FHWA.

These results have not been tested for field conditions; they are offered, however, as initial guidance for field applications. An anticipated next step is that the Maryland SHA will adopt the results as preliminary design guidelines and test them at field sites to decide whether the applications are reasonable.

The outlet scour experimental results showed the effects of using different outlet wing wall configurations at the outlet. Reducing the angle of the wing walls reduces the turbulent shear stress and thus reduces the scour depth created. The outlet experiments clearly demonstrated that outlet scour can be reduced substantially by using outlet wing walls with a streamlined shape. Test results from the 8-degree outlet wing walls revealed reduced turbulence and scour depth at the outlet. This is an encouraging finding, because wing walls with an 8-degree flare are easy to construct or order prefabricated, which may make this design more cost effective than the streamlined design.

"The basic advantage of bottomless culverts," says Kosicki, "at least from the perspective of environmental agencies, is the natural substrate in the bottom of the culvert that creates favorable conditions for fish passage. But natural substrate also can be simulated in circular culverts by burying them below the existing channel invert. When done properly, that offers the same advantage as bottomless culverts in terms of fish passage and additional advantages in terms of eliminating safety concerns due to scour that might exist in bottomless culverts."

Some advantages others have cited for using bottomless culverts, says Stan Davis, a consultant with Maryland SHA, "include ease and speed of construction, reduced road closure time, use of prefabricated elements instead of in-place casting, and wide selection of products to meet specific site criteria."


Kornel Kerenyi is a hydraulics research engineer in the FHWA Office of Infrastructure R&D. He coordinates hydraulic and hydrological research activities with State and local agencies, academia, and various partners and customers. He also manages the FHWA Hydraulics Laboratory. Kerenyi was previously a research engineer at a private company and supervised support staff in the Hydraulics Lab. He holds a Ph.D. in fluid mechanics and hydraulic steel structures from the Vienna University of Technology in Austria.

Jorge Pagán-Ortiz is a principal bridge engineer in the FHWA Office of Bridge Technology in Washington, DC. He earned a bachelor's degree in civil engineering from the University of Puerto Rico at Mayagüez and a master's degree in water resources engineering from The George Washington University. Pagán-Ortiz serves as leader of the Geotechnical and Hydraulics Team and as leader of the FHWA national hydraulics team. He is the lead engineer responsible for FHWA's primary publications on scour, stream stability, and countermeasures: Hydraulic Engineering Circular numbers 18, 20, and 23.

For more information, contact Kornel Kerenyi at 202-493-3142 and kornel.kerenyi@fhwa.dot.gov or Jorge Pagán-Ortiz at 202-366-4604 and jorge.pagan@fhwa.dot.gov. References for this article are available online at www.fhwa.dot.gov/publications/publicroads/index.cfm.

ResearchFHWA
FHWA
United States Department of Transportation - Federal Highway Administration