U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-17-013     Date:  February 2017
Publication Number: FHWA-HRT-17-013
Date: February 2017


Hydraulic Performance of Shallow Foundations for The Support of Vertical-Wall Bridge Abutments


The following descriptions briefly summarize relevant references.

MACKY (1986)(6)

G.H. Macky studied the effectiveness of methods of protecting bridge abutments from scour under clear-water conditions.(6) Macky undertook laboratory experiments using an idealized l:40scale model of the Waiharakeke River Bridge on State Highway 25 near Whangamata, NewZealand. Macky attempted to model typical construction practice for scour countermeasures rather than recommended construction practice. This meant that the protection measures were extended only slightly below the existing bed level. The protection methods tested were riprap, concrete Akmon units (commonly used at coastal sites), flexible concrete mattresses, gabions laid on the embankment slope, stacked gabions staggered up the embankment slope, and boulder-filled wire mattresses laid on the bed sediment with stacked gabions on the embankment slope.

The significant findings of the study were that while the downstream side of the abutment required nominal protection only, the upstream corner of the abutment was subject to strong attack by the flow, requiring protection not only above the existing bed, but also on the slope of the developing scour hole. The high initial velocities caused damage to the abutment structure, which could possibly be avoided by preexcavating a scour hole. By using typical scour countermeasure placement practice, the abutment slopes slumped to a less steep armored slope, even though the abutment structure itself remained stable. The bridge pier adjacent to any abutment needed special protection, because it was possible that it could be sited in the abutment scour hole.

CROAD (1989)(7)

R.N. Croad studied the performance of riprap protection at abutments under clear-water conditions with preexcavated scour holes.(7) Croad also conducted some experiments with riprap placed just below the initial bed level and with riprap placed down to the initial bed level with a horizontal launching apron. The work was intended to build on that of Macky.(6)

PAGÁN-ORTIZ (1991)(8)

J.E. Pagán-Ortiz studied the stability of riprap protection placed in the apron and measured point velocities at model abutments.(8) Vertical-wall and spill-through abutment models were modeled. The riprap was placed directly on the floor of the flume, rendering the study essentially a fixed-bed investigation (i.e., scour did not occur at the abutment). Therefore, the study was useful in determining the likely position in which riprap, placed in an apron around an abutment, would first fail while the riverbed remained level. This information was useful only until a scour hole began to develop, at which time the flow regime changed, and riprap in other positions of the apron might become unstable. The tests were carried out under clear-water conditions, with V/Vcsvalues of approximately 0.9, where V is the mean approach-flow velocity and Vcs is the critical velocity. The abutment model was placed against the side wall of the flume and surrounded by an observation area consisting of a gravel bed placed on the floor of the flume.

The significant findings from this study were as follows:

ATAYEE (1993) AND ATAYEE ET AL. (1993)(9, 10)

A.T. Atayee studied the stability of a riprap apron using a model spill-through abutment situated on the floodplain of a compound channel. (9)The study was intended to build on that of Pagán-Ortiz by measuring the threshold of movement of the gravel material used to protect the floodplain and channel in the vicinity of the abutment.(8) The hydraulic conditions that initiate gravel movement were measured.

Failure was defined as occurring at the instant when the unprotected surface (in this case, the bed of the flume) was clearly exposed. Degradation of the gravel layer to expose the flume bed could occur very rapidly (in seconds). In all experiments, failure occurred at the toe of the embankment just downstream of the abutment centerline.

EVE (1999)(11)

N.J. Eve studied criteria for selection of riprap protection at spill-through bridge abutments with launching apron protection under clear-water and live-bed conditions.(11) The size of the riprap and the extent of the launching apron were varied systematically in the tests. The abutment embankments were constructed using the bed sediment material. The following three failure conditions were defined:

For the clear-water tests, two abutment lengths and three riprap sizes were used. Eve measured the position at which the maximum scour depth occurred in all experiments. Generally, the point of maximum scour moved away from the toe of the abutment as the size of the launching apron and riprap stone size increased, as expected. The lateral extent of the riprap-launching apron was initially set at twice the flow depth, based on the HEC-23 recommendations.(3) This criterion was found to be conservative in all cases. For subsequent experiments, the lateral extent was reduced in increments until failure occurred. An equation was proposed for the apron extent on the basis of these tests.

The live-bed experiments were conducted at 125 and 150 percent of the threshold velocity for the bed sediment. These tests were preliminary in nature, and Eve recommended further study under live-bed conditions.

In all cases, the riprap failed rapidly, apparently because of winnowing of bed sediment through voids between the riprap stones. Some of the tests were repeated with the addition of a filter fabric, which was found to improve the stability of the protection, especially at the lower flow velocity. At the higher flow velocity, the abutments failed, in spite of the presence of the geotextile, because of undermining of the abutment toe, which led to slumping of the sediment beneath the filter fabric. The following three types of failure were observed in the live-bed tests:

HOE (2001)(12)

D.A. Hoe undertook preliminary tests to investigate the use of CTBs to protect spill-through bridge abutments from scour under clear-water conditions.(12) Hoe used the same experimental setup as that used in the clear-water experiments by Eve.(11) Ceramic tiles were used to model the CTB blocks. They were joined together by gluing them onto a flexible loose-woven fabric. The embankment slopes of the abutment were covered with the ceramic tiles, and an apron with a width equal to twice the flow depth was laid around the abutment.

The study showed that the apron protected the abutment from being undermined for approach flow conditions less than 66 percent of the threshold velocity. For higher flow velocities (resulting in deeper scour), the abutment toe was undermined causing abutment failure.

Some experiments were repeated with an increased level of protection, which was achieved by adding additional sections of the ceramic blocks to the standard apron This had the effect of deflecting the scour hole slightly away from the abutment, thereby protecting the abutment from failure at the higher approach flow velocities. Hoe observed that the additional apron was vulnerable to overturning when located in the contracted zone of the flow and recommended that further testing focus on methods to prevent overturning of CTB mats.

CHEUNG (2002)(13)

M. Cheung studied the effectiveness of CTB scour countermeasures around piers and wing wall abutments under clear-water conditions.(13) Cheung used two types of CTBs in the experimental study—ceramic tiles and truncated square pyramid blocks. The blocks were intended to be more representative of commercially available prototype CTBs. Both types of blocks were joined together to form an apron by gluing them onto a flexible loose-woven fabric.

Experiments were conducted with and without the apron attached to the abutment face and with and without geotextile filters placed underneath the aprons. At the end of each experiment, the level of protection provided by the countermeasure was assessed. Failure criteria were developed based on the ability of the foundation structure to continue to support the bridge. An abutment was considered to have failed if the CTB protection could no longer provide any protection for the abutment by allowing excessive amounts of sediment to be scoured away, which would eventually undermine the abutment foundations. The following failure conditions were defined by Cheung:

MARTINEZ (2003) AND KORKUT (2004)(14, 15)

E. Martinez and R. Korkut undertook laboratory experiments to evaluate the use of geobags as a scour countermeasure to protect bridge abutment foundations from failure.(14,15) Geobags are geotextile cloth bags filled with the local sediment or concrete. The experiments focused on the performance of geobag aprons, which were placed around a pile-supported wing-wall abutment retaining an erodible embankment. Martinez focused mainly on geobag performance under clear-water conditions while Korkut studied geobag performance under live-bed conditions. The experiments were intended to investigate the possible use of geobags as a scour countermeasure.

Martinez concluded that local scour could not be eliminated by a geobag apron, but that it could shift the scour region away from the abutment. In this regard, geobag aprons could be effectively used to protect abutments from scour, so long as the shifting of the scour hole did not imperil other hydraulic structures nearby. Martinez suggested that geobags were an effective scour countermeasure for wing-wall abutments when the width of the apron W, exceeded the length of the abutment L.(14)

Korkut used a wing-wall abutment model similar to that used by Martinez, but it was supported by piles and retained an erodible embankment. An equivalent experiment was conducted with a riprap apron. The resulting scour at the abutment was similar for the two experiments, showing that untied geobag aprons functioned in a similar manner to riprap aprons. In both cases, significant scour occurred at the abutment face, exposing the piles of the wing-wall structure.(15)

As a result, the embankment material was winnowed out from underneath the abutment causing the embankment retained by the wing-wall structure to collapse. Other experiments were run with geobags placed underneath the abutment, geobags tied together in the apron, and additional geobags placed under the apron around the perimeter of the apron.(15)

To adequately protect the abutment from scour, Korkut suggested that it was necessary to tie the geobags in the apron together and that the apron should have a toe or skirt that extended below the apron with at least two geobag thicknesses. The live-bed conditions proved to be more critical for abutment protection (compared to the clear-water experiments by Martinez), mainly due to bed-form destabilization of the geobags around the edges of the apron.(14) The passage of the troughs of the bed forms undermined the apron, dislodging the individual geobags or causing the tied geobag apron to fold down.

MELVILLE ET AL. (2006A)(17)

B.W. Melville et al. studied the performance of riprap and CTB aprons as a scour countermeasure for a spill-through abutment.(17) The abutment model was placed on the flood plain of a compound channel. The compound channel comprised a fixed-bed main channel in the approach and an erodible sand bed in the test section. The bank slope was erodible but protected with riprap in the test section and with a V:H slope of 1:2 for all tests. The scour countermeasure experiments were run in clear-water conditions. Abutment length, floodplain width, and apron extent were varied for both apron types.

The purpose of the study was to determine the minimum required apron extent to sufficiently protect the abutment from failure and to determine the scour-hole geometry under clear-water conditions due to variations in the flood plain width, abutment length, and apron extent. Uniform coarse sand was used for all experiments. The riprap protection to the abutment slopes and in the apron comprised uniform-sized gravels with a thickness of 1.5 × D50, equivalent to two riprap layers. A filter fabric was placed over the abutment and covered with riprap or CTBs. The filter fabric was also placed beneath the CTB apron but not the riprap apron because this could induce edge failure of the riprap.

The study had the following significant findings:

MELVILLE ET AL. (2006B)(18)

B.W. Melville et al. studied the performance of riprap and CTBs as scour countermeasures for wing-wall abutments under live-bed conditions for different flow depths, flow velocities, apron extents, and apron burial depths.(18) The aim of the experiments was to investigate the minimum required apron extent to sufficiently protect abutment from scour under live-bed flow conditions.

The majority of the experiments were conducted in a compound channel with a fixed-bed floodplain and a mobile-bed main channel. The wing-wall abutment was sited at the bank of the main channel. These experiments were designed such that the scour process was dominated by bed-form trough migration. In addition, four experiments were run with a different abutment model to investigate the effects of local and contraction scour only. For these experiments, the floodplain was removed, and the abutment model was simplified to a trapezoidal-shaped structure. The two models were referred as the “compound channel model” and the “rectangular channel model.” Uniform coarse sand was used as mobile-bed material for all experiments.

For both channel configurations, the settlement of CTB and riprap aprons at the end of each experiment were measured, and it was defined as the distance from the average bed level to the top of the apron after settlement. Two cases of apron settlement were observed. For case I, apron settlement occurred at the outer edge, whereas for case II the entire apron settled (i.e., scour at the abutment face also occurred). Melville et al. observed a trend of increased apron settlement of the outer edge with increasing flow depth and flow velocity. Also, scour at the upstream corner of the abutment was deeper than at the downstream corner. For case II, scour was deepest at the outer edge of the apron. The burial depth of the apron did not affect its settlement depth at the outer edge, but it did affect the stability of the apron.

Melville et al. concluded that the scour at wing-wall abutments under live-bed conditions was directly related to the level of deepest bed-form trough that propagated past the abutment, together with any localized scour that might occur. Only a vertical settlement was observed at the outer edge of CTB aprons, allowing the scour to occur closer to the abutment face than for an equivalent riprap apron, which tended to settle and move away from the abutment, deflecting the maximum scour depth further away from the abutment. The researchers developed an equation for prediction of the minimum width of apron that remained horizontal after erosion.

MELVILLE ET AL. (2007)(19)

B.W. Melville et al. conducted an experimental investigation to determine the stone size D50 of riprap aprons around a wing-wall abutment necessary to withstand shear failure induced by the approach flow.(19) The experiments were conducted in a compound channel with a fixed-bed floodplain and an erodible main channel under live-bed conditions. The experimental results were then compared with predictions of riprap stone size D50 given by existing equations.

The stability of the riprap stones against shear failure was observed after each run. Experiments were classified as stable (no entrainment) and shear failure (entrainment of rocks). It was observed that the dislodged rocks were transported some distance downstream. Often these rocks were entrained from the edges of the apron (i.e., edge failure was a failure mechanism linked).

PETERSEN ET AL. (2015)(5)

T.U. Petersen et al. conducted two experiments to investigate the mechanisms of edge scour in currents to better understand its causes.(5) The first type of experiment was conducted under live-bed conditions for scour testing, while the subsequent test was conducted in the clear-water regime to obtain clear particle image velocimetry (PIV) measurements.

The results were measured with a variety of devices, including a mini underwater video camera that videotaped the overall time development of the scour process, a laser rangefinder that measured the scour profiles, and a laser Doppler anemometer that measured the velocity profiles across the water depth. The miniature underwater video camera also used the movement of the sand grains in the video recordings as flow tracers. The data were then used to visualize the flow structures alongside the toe of the stone layer and individual stones. The second experiment used PIV measurements to investigate the three-dimensional flow in the cross-sectional plane at the meeting of the sand and stone layer. Objectives for the experiment included obtaining the secondary currents and estimating the turbulence intensity field over the two layers.

Based on the data that were collected, the secondary flow very close to the bed in the sand section was directed away from the stone layer toe, and that over the stones was directed into the stone layer. This pattern resulted in a pattern of deposition occurring inside the toe of the stone layer, scour occurring at the edge of the stone layer, and finally a slight deposition of sediment. These secondary currents originated from either transition in surface roughness between the sand and stone layer or from corner effects.

The mechanisms governing the edge scour was the turbulence in the primary flow as it neared the intersection between the stone layer and sand bed and the secondary flow. The sediment was brought into suspension by the turbulence and was subsequently carried away by the secondary flow. Two types of turbulence together caused the sediment to be brought into suspension: turbulence in the primary flow generated in the fully developed boundary layer over the sand bed and stone layer, and the turbulence generated locally around individual stones at the toe, which formed horseshoe-vortices and lee-wake flow. As the scour around the stones occurred, the process removed sand below the stones and subsequently caused the stones to slump into the scour hole. Shear failure of the sand also contributed to the movement of the stones and might have caused some stones to sink into the sediment bed. As the lower rocks slumped, the upper rocks also moved toward the edge. Horseshoe vortex and vortex shedding occurred and was revealed by sediment, which was used as flow tracers. These mechanisms increased the turbulence intensity and when paired with contraction of streamlines at the side edge of the stone caused scour holes around the toe stones. In addition, sand infiltrated into the cover stone layer, and a substantial quantity of deposition took place within the stone layer. Consequently, the combination of primary and secondary flow caused deposition to occur both in the stone layer and elsewhere. The governing parameters in controlling the three-dimensional flow field, including the secondary currents and turbulence adjacent to cover-stone layer, were the size of the roughness elements and, to some degree, the side slope.

JESSON ET AL. (2013)(20)

M. Jesson et al. conducted and reported the first attempt to use the Shiono-Knight method to model velocity and boundary stress distributions in an open channel.(20) An attempt to compare physical and numerical simulations of the open channel flow was made. It had been shown in many experiments that secondary flows not only form in homogeneously rough rectangular and trapezoidal channels but also in heterogeneous channels where the secondary flows were caused by changes in roughness. These secondary flows along with general boundary shear stress conditions affected the flow structures, conveyance capacity, and transport of sediment in the channel.

Smooth and rough sections were placed at intervals based on the idea that biotope changes could occur between three and four times in the stream-wise direction of a length equivalent to 10channel widths. In accordance with this concept, the length of the patches used in the experiment was set at three channel widths. It was observed that the flow field adjusted quickly to changes in boundary roughness.

Jesson noted that the Reynolds stresses were complex, which indicated that higher order velocities unlike stream-wise velocities took longer to adjust. In addition, cross sections upstream and downstream of a change in roughness configuration showed very little observable variation in boundary shear stress profiles. This indicated that boundary shear stress adjusted relatively quickly to local changes in surface roughness. These results were promising and indicated that robust modeling could assist environmental regulators and river managers to assess and manage river stability and sediment transport.



Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101