Countermeasures to Protect Bridge Abutments from Scour (2007)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

118 8.1 Experimental Work 8.1.1 Introduction The aim of the experiments reported in this chapter was to investigate the use of riprap and cable-tied blocks as bridge abutment scour countermeasures. Both riprap and cable-tied block aprons were placed around spill-through bridge abut- ments to protect them from scour, which could otherwise potentially undermine the abutments. A series of experiments was conducted in the Fluid Mechan- ics Laboratory of the School of Engineering, University of Auckland, in New Zealand, with clear-water conditions at flow velocities just below the threshold velocity of the sediment. A spill-through abutment model, molded from the bed material, was sited on the floodplain of the compound channel. For these experiments, abutment length, floodplain width, and apron extent were systematically varied for both riprap and cable-tied blocks to determine the minimum required apron extent to sufficiently protect the abutment. The results from these exper- iments are presented in Section 8.3. Flow fields around the bridge abutments were measured for all abutment and compound channel configurations to provide insight into the development of scour and the inter- action between the developing scour formations and the apron countermeasures protecting the abutment. The flow field measurements are presented in Section 8.2. The measured flow fields described above were compared with flow distributions obtained from a two-dimensional shallow-water numerical model.The two-dimensional shallow- water model developed for analysis of bridges for Federal Highway Administration, FESWMS (Finite Element Surface Water Modeling System), is used by highway agencies throughout the United States; therefore, it was used in this study. The results are given in Section 8.4. A further set of experiments was undertaken in a relatively large-scale flume at the University of Iowa to validate the main recommendations from the extensive parametric flume experiments described above. These experiments are pre- sented in Section 8.5. 8.1.2 University of Auckland Experimental Equipment and Set-Up A 2.4-m wide, 0.3-m deep, and 16.5-m long nonrecirculat- ing flume was used to conduct the clear-water, spill-through abutment scour countermeasure study. The flume is sup- ported by two universal beams that pivot about a central support. Screw jacks support the beams at either end so that the flume slope is adjustable. The flume consists of an inlet tank, a flow straightener, a 13-m long channel, a sediment col- lection tank, and an outlet tank. A 2.8-m long, 0.45-m deep sediment recess is located 7 m downstream of the inlet tank. Figure 8-1 shows the flume in the upstream direction, and Figure 8-2 shows a longitudinal cross section of the flume. Water is supplied to the flume inlet from the laboratory constant-head tank via two 150-mm and one 200-mm diame- C H A P T E R 8 Lab Results III: Aprons at Spill-Through Abutments Figure 8-1. The 2.4-m wide flume in the upstream direction.

119 ter pipes. The discharge from each pipe is regulated by butterfly valves, and the flow rate is measured by measuring the pressure difference across an orifice plate in each pipeline. The inlet tank consists of a 1.8-m high header tank with a baffle system to regulate and distribute the flow evenly across the flume. The flow passes through a baffle system at the bot- tom of the header tank, through a wire mesh screen to smooth the flow, and then through a flow straightener into the channel section of the flume. The flume channel consists of fiberglass sides and a concrete floor. The recess was filled with bed sediment such that the surface of the sand was level with the flume floor prior to any scouring. From the flume channel, the flow passes into the sediment collection tank. The tailgate controls the water depth in the flume. A sediment feed system is located 1.2 m downstream of the flow straightener, as seen in Figure 8-3. Dry sediment stored in the hopper falls onto a conveyor belt, which carries the sedi- ment away. The sediment discharge is controlled by a variable- sized triangular orifice. A rotating brush sweeps the sediment from the conveyor belt into the flume. Sediment can be prevented from being fed into the flood- plain section by removing sections of the rotating brush, so that some of the sand on the conveyor belt is carried across the flume, where it falls off the belt and is deposited over the side of the flume. A moveable floodplain was constructed along the length of the rectangular channel section of the flume, as shown in Figure 8-4. The main channel bank was constructed in sec- tions of sheet metal folded into wedge profiles 150 mm high with a side slope of 2:1 (H:V). The wedge sections were placed lengthwise in the flume, with the top edge of the main chan- nel bank running parallel to the side of the flume at the desired floodplain width, Bf. The sheet metal lining the main channel bank and the top of the floodplain was painted and sprinkled with sand to simulate the roughness of a sediment bed. The four different floodplain widths, Bf, used in this study are 0.8, 1.2, 1.6, and 2.0 m. Figure 8-4 shows a flood- plain in the flume. In the figure, Bf equals 1.6 m. Four different spill-through abutments were used in this study.All the abutments had the same frontal shape and dimen- sions, but varied in length, L. The lengths of the abutments used in this study are 0.4, 0.6, 0.8, and 1.0 m. The dimensions of the frontal section of the abutment are given in Figure 8-5. The abutments were placed on the floodplain in the sediment recess section of the flume and were molded from the same sediment as that used for the floodplain and in the recess of the flume. Sheet metal molds were used to construct the abutments. The Figure 8-2. Longitudinal cross section of the 2.4-m wide flume. Figure 8-3. Sediment feed system in the upstream section of the 2.4-m wide flume.

120 mold was split into a 0.4-m frontal section and several extension sections with varying lengths, as shown in Figure 8-6. 8.1.3 Bridge Abutment Flow Field Measurements A particle-tracking velocimetry (PTV) technique was devel- oped to measure the two-dimensional surface flow fields around the spill-through abutment models. PTV is based on the principle of capturing sequences of images of specially illu- minated, particle-seeded fluid flow, from which quantitative information about the flow field can be extracted. A particle- seeding density of approximately 15-blocks/m2 was used. Uniform flow was established along the length of the flume, with a flow depth of 100 mm on the floodplain and 250 mm in the main channel. For each compound channel configura- tion (Bf  0.8, 1.2, 1.6, and 2.0 m), the flow field was measured in the test section of the compound channel using the PTV technique. The flow distribution across the flume was adjusted at the flume inlet by blocking off sections in the flow straight- ener until there was no net transfer of flow over the main channel bank boundary. The average flow velocity in the main channel was set to the threshold velocity for sediment move- ment, while the average velocities on the floodplain were typ- ically 80 to 90 percent of the threshold velocity. Figure 8-7 shows the surface velocity distributions across the flume for the four different compound channel configurations. A small Figure 8-4. Experimental set-up for the spill-through abutment study. Figure 8-5. Spill-through abutment dimensions. Figure 8-6. Spill-through abutment molds.

diagram of the cross section of the corresponding compound channel is shown under each velocity distribution. The criti- cal velocity for sediment entrainment Vc, determined using Shields criterion adjusted for lateral slope, is shown on each velocity distribution. The adjustment for lateral slope was made using the method given in Vanoni (1977). For the spill-through abutments, the flow measurements were undertaken using the experimental set-up in Figure 8-8 at the initial (i.e., unscoured) stage of the experiments. The flow fields were measured for the abutment lengths 0.4, 0.6, 0.8, 1.0, 1.2, 1.4, and 1.6 m, situated on the various floodplain widths, Bf, of 0.8, 1.2, 1.6, and 2.0 m. 8.1.4 University of Auckland Experimental Procedure The purpose of the study was to determine the amount of scour countermeasure protection required to protect spill- through bridge abutments from failure. The test models used in this experimental study were designed to be representative of spill-through abutments situated on the floodplain of wider compound river channels. At such river crossings, the natural vegetation typically protects the flood channels from general erosion, and approach-flow conditions can be taken as clear-water during floods. An idealized compound channel geometry was used, as shown in Figure 8-8. Similitude between laboratory experiments and field scale was satisfied by the use of the aforementioned u*/u*c ratio, of which a value of just below 1.0 represents a condition called “clear-water scour.” This condition is extreme for scouring because the velocity is as high as possible without the move- ment of the channel bed, which causes infilling of the sedi- ment hole. The primary objective was to determine (a) the scour hole geometry under clear-water conditions due to variations in the compound channel and abutment geometries and (b) the extent and type of scour protection provided. The extent of apron pro- tection W, the length of the abutment and bridge approach embankment L, and the width of the floodplain Bf were system- atically varied for both riprap and cable-tied block protection Figure 8-7. Velocity distributions across the 2.4-m wide flume for the four different compound channel geometries. 121

122 (Figure 8-8). The aspect ratio of the floodplain width Bf/yf ranged from 8 to 20, and the aspect ratio of the abutment length L/yf ranged from 4 to 10,where yf is flow depth on the floodplain. The compound channel consisted of an erodible sand bed for the floodplain and erodible sand boundaries for the main channel bed and bank in the test section. Uniform coarse sand was used as the bed material for all the experiments. The sed- iment properties are identical to those of the sediment used in the wing-wall experiments described in Chapter 7 and are summarized in Table 8-1. Riprap protection was placed on the erodible bank and bed of the main channel. This was necessary to prevent erosion of the main channel bank, which would have occurred in the absence of any abutment structure due to the influence of the bank slope. Filter fabric was placed over the abutment and covered with riprap or cable-tied block protection. The filter layer was also placed beneath the horizontal apron for cable- tied block protection (i.e., to prevent winnowing of the bed sediment), but was not used for riprap protection because this could induce edge failure of the riprap (Eve, 1999). Experiments were run under uniform flow conditions (with flow depths of 100 mm on the floodplain and 250 mm in the main channel). Initially, experiments were run with no apron protection (W  0,as per the experimental set-up in Figure 8-8). Instead, the spill-slope protection was extended below the sur- face of the floodplain to a depth greater than the expected scour hole depth to protect the toe of the abutment, based on existing recommendations. Next, experiments were run with a 0.5-m wide apron, and thereafter the aprons were successively reduced in size by 0.1 m until the edge of the equilibrium scour hole occurred adjacent to the abutment slope. Figure 8-9 shows such experimental set-ups in the flume. All data apply to the case of “no failure” of the abutment and approach embankment. For experiments where the scour hole would encroach on the main channel bank, the riprap stones lining the main channel bank were removed just before they were about to fall into the scour hole. In this way, the scour hole formation was not affected by the riprap protection covering the main chan- nel bank. Using the Richardson and Davis (1995) method for calcu- lating the required riprap size to prevent dislodgement of the stones by the flow, the required riprap size, d50, for the most critical flow conditions was calculated to be 15 mm. The riprap size (Type R1 in Table 7-2 and Figure 7-7) used for the experimental study was larger than the required riprap size, ensuring the stability of the riprap for all the abutment- compound channel configurations tested. In the same way, the cable-tied blocks used in the experiments (Type R1 in Table 7-2 and Figure 7-7) were sized so that they would remain stable for all test conditions. Figure 8-8. Experimental configuration of the countermeasure placement for the spill-through abutment study. Table 8-1. Sediment properties. Description d16(mm) d84 (mm) d50 (mm) σg Ss u*c (ms-1) Filter sand 0.62 1.04 0.82 1.30 2.65 0.020

The experiments were carried out with a geotextile placed underneath the countermeasure apron to eliminate the win- nowing of sand from between the riprap stones or cable-tied blocks. The geotextile used for testing (shown in Table 7-3) was flexible enough to ensure that the riprap or cable-tied blocks would be in contact with the bed at all times. All experiments were run for 72 hours to ensure that the local scour hole had reached the equilibrium depth. At the conclu- sion of each experiment, the resulting scour hole formation was contoured in 50-mm increments for photographic purposes. Figures 8-10 and 8-11 show examples of the contour lines. The position of the deepest point of the scour hole defined by x and y, the depth of the scour hole relative to the flood- plain dsf, the horizontal distance from the floodplain wall to the opposite edge of the scour hole e, and the minimum distance between the edge of the scour hole and the abutment Wmin were measured. Figure 8-12 shows these measurements. The accu- racy of the measurements of dsf was 5 mm, whereas the accu- racy of the measurements of x, y, e, and Wmin was 10 mm. 8.2 Bridge Abutment Flow Fields 8.2.1 Introduction The flow fields around the model bridge abutments were measured for all abutment and compound channel configu- rations with a flat fixed bed to provide insight into the devel- opment of scour and the interaction between the developing scour formations and the apron countermeasures protecting the abutment. 8.2.2 Data Analysis The measured flow field data sets were used to plot time- averaged, two-dimensional velocity-vector fields. An example 123 (b) Leveled floodplain with riprap protection extended below the bed (a) Riprap extended below the surface of the floodplain (c) Riprap apron protection (d) Cable-tied block apron protection Figure 8-9. Initial set-up of experiments.

124 of the velocity vector field around a 0.8-m long spill-through abutment situated on a 1.6-m wide floodplain is shown in Figure 8-13. The associated vorticity, , for each flow field was calcu- lated by using the time-averaged velocity measurements and adopting a central-difference approximation for the follow- ing vorticity expression: (8-1) Where Vx and Vy are velocity components in the x and y direc- tions, respectively, as defined in Figure 8-4 and x and y  − Vx y y x y x y yV V y +[ ] [ ] −[ ] − +   2 2  = ∂ ∂ − ∂ ∂ ≈ − ++[ ] [ ] −[ ]V x V y V V V x y x y x x y x y x x   2 2 0.1 m. An example of the calculated vorticity fields at the spill- through abutment is shown in Figure 8-14. The bed shear stress, , was calculated by using the time- averaged surface velocity measurements. Assuming a loga- rithmic velocity profile, the shear velocity u* was estimated by an iterative process: (8-2) Where V is the velocity, measured using an acoustic Doppler velocimeter at an elevation z above the bed level and ks  2d50. The parameter ar is given by the following: (8-3) (8-4) (8-5) Where  is the kinematic viscosity of the water in the flume. The shear stress exerted on the bed by the flow was calcu- lated as follows: (8-6) where is the density of the water in the flume. The bed shear stress values were normalized using idealized values of c, the critical stress for sediment entrainment, estimated from (8-7) Where C is the nondimensional critical shear stress param- eter obtained from the Shields diagram, the specific gravity of the bed sediment Ss is 2.65, g is 9.81 ms-1, and d is the sediment diameter. For the spill-through abutment flow fields, the c val- ues were adjusted for localized bed slope wc using the follow- ing equation given by Vanoni (1977): (8-8) Where the main channel bank slope angle s is 26.6 degrees, or 2:1 (H:V), the repose angle of the bed material  is 30 degrees and  is the direction of the flow defined in Figure 8-14. An example of the normalized bed shear − ′ sin sin tan            wc c = ′ ⎛ ⎝⎜ ⎞ ⎠⎟ + − ′ sin sin tan cos tan tan 2 2 1 ⎛ ⎝⎜ ⎞ ⎠⎟ ⎡ ⎣ ⎢⎢ ⎤ ⎦ ⎥⎥ − 2   c C sS gd= −( )1  = ∗ u2 if u ks∗ >  70ar = 0 85. if 3 5 70. < <∗u ks  a u k r s = ⎛⎝⎜ ⎞⎠⎟ +∗5 75 8 5. log . if u ks∗ <  3 5.a u kr s= ⎛⎝⎜ ⎞⎠⎟ +∗5 75 5 5. log . V u z k a s r ∗ = +5 75. log Figure 8-10. Contour lines placed in the equilibrium scour hole for a spill-through abutment protected by a riprap apron (Bf  1.6 m, L  0.8 m, and W  0.3 m). Figure 8-11. Contour lines placed in the equilibrium scour hole for a spill-through abutment protected by a cable-tied block apron (Bf  1.6 m, L  0.8 m, and W  0.4 m).

Figure 8-12. Scour hole parameters measured for each experiment in the 2.4-m wide flume. Figure 8-13. Velocity vector field around a spill-through abut- ment (Bf  1.6 m; L  0.8 m). 125

126 stress fields at the spill-through abutment is shown in Figure 8-15. 8.2.3 Results Figures 8-16 and 8-17 give lateral distributions of flow velocity. The series of velocity, vorticity, and normalized bed shear stress plots are shown in Figures 8-18, 8-19, and 8-20, respec- tively, for the spill-through abutment flow fields. In these three figures, the abutment length L increases down the page from 4yf to 16yf, the floodplain width Bf increases across the page from 8yf to 20yf, and the flow velocity V/Vc increases down the page from 1.1 to 2.2. The increase in velocity at the abutment, shown in Figure 8-16, causes a local increase in bed shear stress on the floodplain, shown in Figure 8-18. This observation is most obvious for the case where L/yf is 8 and Bf /yf is16. Upstream of the abutment, the backwater on the floodplain caused by the abutment diverts flow from the floodplain into the main channel. The velocity in the main channel increases as the flow accelerates through the contracted sec- tion, causing an increase in bed shear stress in the main channel, as seen in Figures 8-16 and 8-18. These two obser- vations were also observed by Biglari and Sturm (1998) and Lim and Nugroho (2004) in both their experimental and numerical work on flow around abutments in compound channels. Figure 8-14. Vorticity field around a spill-through abutment (Bf  1.6 m; L  0.8 m). Figure 8-15. Normalized bed shear stress field at a spill-through abutment (Bf  1.6 m; L  0.8 m).

There is a small counterclockwise rotation in the flow field at the upstream corner of the abutment (also observed by Kwan, 1984) and a larger counterclockwise rotation in the flow field downstream of the abutment, which extends out past the end of the abutment (see Figure 8-16), the latter increasing with abutment length. There is also a smaller clockwise rotation in the flow field at the downstream corner of the abutment next to the larger region of counterclockwise rotation. This clockwise rotation is more obvious with increasing abutment length and is most obvious for the spill-through abutment case where L/yf is 10 and Bf /yf is 16. Atayee (1993) studied the stability of riprap in aprons around spill-through abutments situated on the floodplain of a compound channel. Atayee observed that failure occurred consistently at the toe of the embankment just downstream of the end of the abutment, regardless of abutment length and proximity to the main channel. From Figure 8-17, it can be seen that the vorticity strength of the flow field around the abut- ment is strongest at the toe of the embankment just down- Figure 8-16. Velocity distributions across the bridge section (in terms of Vx and Vy, defined in Figure 8-7) for all the experimental configurations given in Figure 8-32. 127

128 stream of the abutment end, irrespective of abutment length and proximity to the main channel. The zone of strongest vorticity at the abutment corresponds to the zone where riprap shear failure occurred in the study by Atayee, suggesting that the vorticity strength may be the dominant parameter initiat- ing riprap shear failure in an apron around an abutment. The bed shear stress fields were calculated based on the assumption that the vertical velocity distribution could be rep- resented by a logarithmic velocity profile. The validity of this assumption has not been verified by measuring the bed shear stress at different points in the flume, so the results should be interpreted with caution. Lim and Nugroho (2004) measured vertical velocity profiles around a vertical-wall abutment situ- ated on the floodplain of a compound channel both before and after scour occurred. Their work showed that in the unscoured state (when the flow is mainly two-dimensional), the velocity distributions are represented reasonably well by the log-law relationship. This suggests that it is reasonable to Figure 8-17. Velocity distributions across the bridge section (in terms of the velocity magnitude) for all the experimental configurations given in Figure 8-32.

129 assume a logarithmic velocity distribution to calculate the bed shear stress. Although the normalized shear stress plots must be inter- preted with caution, they can effectively indicate zones where the shear stress increases relative to other areas in the flow field. Figures 8-16 through 8-20 all show that the velocity, vorticity strength, and shear stress at the end of the abutment increase with increasing abutment length as a result of more flow being diverted around longer abutments. It can also be seen that the velocity in the main channel increases with increasing abutment Figure 8-18. Velocity contour plots for various flow scenarios.

130 length because of a greater contraction at the bridge section. Figure 8-16 shows that the velocity component across the flume in the y-direction increases with increasing abutment length as the flow is diverted around the abutment. Equation 8-8 allows for flow direction in the calculation of the critical shear stress on the main channel bank. Consequently it can be seen from the normalized shear stress plots that the normalized shear stress on the main channel bank increases considerably with increasing abutment length because of the increasing component of the flow in the y-direction. Figure 8-19. Vorticity contour plots for various flow scenarios.

131 Figures 8-16 and 8-17 show that, for most cases, the veloc- ity, vorticity strength, and shear stress at the abutment end increase slightly with increasing floodplain width. These trends are explained as follows. As the floodplain width increases, the flow area at the bridge section decreases, increasing the flow velocity at the bridge section. As a result of the increasing flow velocity at the bridge section, stronger vortices are shed from the end of the abutment. Sections 8.3.3 and 8.3.4 describe the relationship between the measured flow fields around the spill-through abutments Figure 8-20. Bed shear stress contour plots for various flow scenarios.

and the development of scour at the abutments. These sec- tions also describe how the flow fields can be used to deter- mine the zones around the abutment that need to be protected from scour. 8.3 Spill-Through Abutment Clear- Water Study 8.3.1 Introduction The aim of the study was to investigate the use of riprap and cable-tied blocks as spill-through abutment scour coun- termeasures. Both riprap and cable-tied block aprons were placed around abutments to protect them from scour, which could potentially undermine them if no protection were pro- vided. A series of experiments were conducted with clear- water conditions just below the threshold velocity of the sediment. Abutment length, floodplain width, and apron extent were systematically varied for both riprap and cable- tied blocks to determine the minimum required apron extent to sufficiently protect the abutment. Similitude between laboratory experiments and field scale was satisfied by the use of the aforementioned u*/u*cratio, of which a value of just below 1.0 represents a condition called “clear-water scour.” This condition is extreme for scouring because the velocity is as high as possible without the move- ment of the channel bed, which causes infilling of the sedi- ment hole. 8.3.2 Experimental Results The measurements from the experiments are summarized in Table 8-2. For the experiments marked with an asterisk in Table 8-2, dsf and Wmin were also measured during the exper- iments. These measurements are given in Table 8-3. The dis- tance to the deepest point of the scour hole from the abutment end Rdmax (as shown in Figure 8-12), is also given in Table 8-2 and can be determined as follows: (8-9) The F values given in Table 8-2 were developed as part of the data analysis and are discussed in Section 8.3.4. The position of the deepest point of the scour hole for all of the experiments is shown in Figure 8-21. As the abutment length increases, the distance to the deepest point of the scour hole from the end of the abutment increases. The scour hole position is independent of the floodplain width. For the case where the scour hole forms entirely on the floodplain (e  Bf), the size of the scour hole tends to decrease slightly with increasing apron width and increase slightly with increasing floodplain width. Conversely, for the case where part of the scour hole forms beyond the floodplain (e  Bf), the size of Rd x ymax 2 2 2 = +  the scour hole increases with increasing apron width and decreases with increasing floodplain width. As the apron width increases, the scour hole is deflected farther away from the end of the abutment. The systematic trends of the scour hole geometry inherent in each of the experimental series is discussed in Section 8.3.4, including the equations derived from the data, which allow pre- diction of the position, extent, and depth of the scour hole and provide a method for estimating the minimum extent of apron protection to ensure adequate toe protection for the abutment. 8.3.3 Experimental Observations The apron protection around the abutment inhibited the development of scour at the abutment toe. Scour was initiated at the edge of the apron and increased in depth with the pas- sage of time. As the scour hole deepened, bed material on the sides of the scour hole fell into the scour hole, progressively undermining the protection apron. The response of the apron to the undermining process depended on the protection type. As the riprap aprons were undermined, the stones at the outer edge would roll into the scour hole, protecting the bed of the hole from further scour. This would deflect the erosion zone farther away from the abutment. As the cable-tied block aprons were undermined, the outer edge of the cable-tied block apron folded down onto the side of the scour hole because the cables prevented the blocks from sliding into the scour hole. As the apron folded down onto the side slopes of the scour hole, the horizontal distance between the toe of the abutment and the edge of the apron decreased, allowing the erosion zone to move closer toward the abut- ment. The scouring process would continue until the equilib- rium scour depth was reached. The velocity flow fields measured at the abutments showed that the velocity at the contracted bridge section increased with increasing abutment length and floodplain width. Both parameters have the effect of reducing the flow area at the contracted bridge section, thereby increasing the velocity and flow strength. Consequently, the vorticity and bed shear stress also increase with increasing abutment length and floodplain width. Similar effects were observed regarding the influence of the abutment length and floodplain width on the equilib- rium scour hole depths, showing that the scour hole size is related to the flow field around the abutment. The flow fields were overlaid on the associated equilibrium abutment scour formation photographs to compare some of the flow field features with the resultant scour hole forma- tions. This comparison was undertaken only for the experi- ments where the spill-slope protection was extended below the surface of the floodplain (W  0)—that is, for experi- ments with no apron. For the experiments with the protec- tion aprons (W  0), comparison of the flow field features 132

133 Bf (m) L (m) W (m) dsf (m) αx (m) αy (m) αe (m) Wmin (m) Rdmax (m) F Riprap Protection 0.800 0.400 0.000 0.200 0.620 0.250 0.980 - 0.669 0.27 0.800 0.400 0.100 0.190 0.660 0.290 0.990 0.000 0.721 0.28 0.800 0.400 0.200 0.185 0.790 0.350 1.000 0.070 0.864 0.29 0.800 0.400 0.300 0.210 0.880 0.430 1.080 0.170 0.979 0.38 0.800 0.400 0.400 0.285 0.960 0.530 1.230 0.250 1.097 0.53 0.800 0.400 0.500 0.280 1.050 0.560 1.350 0.310 1.190 0.61 0.800 0.600 0.000 0.355 0.780 0.320 1.430 - 0.843 0.89 0.800 0.600 0.300 0.330 1.000 0.500 1.500 0.005 1.118 0.91 0.800 0.600 0.400 0.380 1.040 0.550 1.730 0.060 1.176 0.96 0.800 0.600 0.500 0.370 1.080 0.625 1.810 0.135 1.248 0.97 0.800 0.800 0.000 0.435 0.850 0.435 2.020 - 0.955 1.00 0.800 0.800 0.500 0.400 0.900 0.700 2.200 0.000 1.140 1.00 1.200 0.400 0.000 0.180 0.380 0.155 0.860 - 0.410 0.00 1.200 0.400 0.100 0.160 0.450 0.260 1.010 0.040 0.520 0.00 1.200 0.400 0.200 0.155 0.650 0.290 1.065 0.120 0.712 0.00 1.200 0.400 0.300 0.100 0.780 0.360 1.120 0.250 0.859 0.00 1.200 0.400 0.400 0.080 0.860 0.410 1.180 0.380 0.953 0.00 1.200 0.400 0.500 0.070 1.050 0.410 1.300 0.465 1.127 0.07 1.200 0.600 0.000 0.250 0.500 0.220 1.260 - 0.546 0.07 1.200 0.600 0.200 0.280 0.650 0.410 1.360 0.010 0.769 0.17 1.200 0.600 0.300 0.300 0.810 0.560 1.480 0.180 0.985 0.28 1.200 0.600 0.400 0.315 0.920 0.600 1.540 0.280 1.098 0.32 1.200 0.600 0.500 0.325 1.030 0.650 1.610 0.350 1.218 0.38 1.200 0.800 0.000 0.315 0.795 0.440 1.630 - 0.909 0.54 1.200 0.800 0.300 0.365 0.990 0.550 1.840 0.010 1.133 0.69 1.200 0.800 0.400 0.385 1.100 0.690 2.000 0.125 1.298 0.77 1.200 0.800 0.500 0.400 1.120 0.725 2.100 0.210 1.334 0.81 1.600 0.400 0.000 0.205 0.450 0.230 1.060 - 0.505 0.00 1.600 0.400 0.100 0.205 0.600 0.290 1.090 0.010 0.666 0.00 1.600 0.400 0.200 0.210 0.750 0.440 1.260 0.065 0.870 0.00 1.600 0.400 0.300 0.215 0.790 0.440 1.280 0.165 0.904 0.00 1.600 0.400 0.400 0.200 0.870 0.475 1.350 0.275 0.991 0.00 1.600 0.400 0.500 0.210 1.010 0.530 1.390 0.385 1.141 0.00 1.600 0.600 0.000 0.290 0.500 0.270 1.470 - 0.568 0.00 1.600* 0.600* 0.200* 0.305* 0.790 0.490 1.600 0.020* 0.930 0.00 1.600 0.600 0.300 0.300 0.830 0.530 1.680 0.150 0.985 0.05 1.600 0.600 0.400 0.270 0.950 0.580 1.710 0.230 1.113 0.06 1.600 0.600 0.500 0.260 1.010 0.720 1.770 0.350 1.240 0.10 1.600 0.800 0.000 0.315 0.610 0.330 1.690 - 0.694 0.08 1.600 0.800 0.200 0.330 0.820 0.510 1.780 0.000 0.966 0.14 1.600 0.800 0.300 0.375 0.880 0.650 1.950 0.090 1.094 0.26 1.600 0.800 0.400 0.400 1.060 0.750 2.060 0.210 1.298 0.33 1.600 0.800 0.500 0.390 1.120 0.780 2.120 0.320 1.365 0.36 2.000 0.400 0.000 0.200 0.470 0.190 0.960 - 0.507 0.00 2.000 0.400 0.100 0.190 0.530 0.260 1.000 0.000 0.590 0.00 2.000 0.400 0.200 0.180 0.640 0.310 1.080 0.110 0.711 0.00 2.000 0.400 0.300 0.160 0.770 0.370 1.130 0.250 0.854 0.00 2.000 0.400 0.400 0.120 0.955 0.380 1.190 0.380 1.028 0.00 2.000 0.400 0.500 0.105 1.040 0.400 1.210 0.480 1.114 0.00 2.000 0.600 0.000 0.260 0.540 0.240 1.420 - 0.591 0.00 2.000 0.600 0.200 0.260 0.630 0.425 1.505 0.025 0.760 0.00 2.000 0.600 0.300 0.270 0.720 0.500 1.610 0.180 0.877 0.00 2.000 0.600 0.400 0.255 0.810 0.530 1.700 0.290 0.968 0.00 2.000 0.600 0.500 0.250 0.880 0.615 1.750 0.410 1.074 0.00 2.000 0.800 0.000 0.290 0.580 0.180 1.715 - 0.607 0.00 2.000 0.800 0.200 0.295 0.650 0.510 1.820 0.000 0.826 0.00 2.000 0.800 0.300 0.310 0.750 0.610 1.950 0.120 0.967 0.00 2.000 0.800 0.400 0.300 0.800 0.600 1.990 0.220 1.000 0.00 2.000 0.800 0.500 0.305 0.990 0.715 2.090 0.320 1.221 0.04 2.000 1.000 0.000 0.325 0.660 0.370 2.080 - 0.757 0.05 Table 8-2. Equilibrium scour hole measurements. (continued on next page)

with the scour formation was difficult, because the scour hole development occurred away from the abutment because of the presence of the apron. Figure 8-22 shows an example of the equilibrium scour hole at a 0.8-m long abutment situated on a 1.6-m wide floodplain, with spill slope protection extended below the surface of the floodplain. Figures 8-23, 8-24, and 8-25 show the corresponding velocity, vorticity, and relative bed shear stress, respectively, overlaid on top of the equilibrium scour formation at the abutment. It is apparent that the region of high vorticity in the wake of the abutment corresponds to areas where scour is initi- ated, as shown in Figure 8-24. During the development of the scour hole, erosion develops down the face of the riprap protection surface and downstream along the line of high vorticity. The effect of placing a riprap apron or similar protection layer around an abutment is to prevent the initiation of scour at the point where the vorticity is strongest. The scour hole is deflected downstream and generally reduces in size.Figure 8-24 shows that the vorticity strength decreases farther away from the abutment, which is consistent with the reduced scour depth observed when wider aprons are placed around the abutments. The deepest point of scour occurs on a line extending from the end of the abutment at an approximate angle of 30 de- grees to the downstream direction, irrespective of the apron size (discussed in Section 8.3.4). Figure 8-24 shows that the dividing line of positive and negative vorticity at the end of the abutment also occurs at an approximate angle of 30 degrees to the downstream direction. This shows a strong correlation between the scour hole position and zone of strongest vor- ticity at the abutment. Therefore, using a two-dimensional numerical model to determine strong vorticity regions at a bridge abutment could enable better prediction of the ex- pected scour hole location. Coupled with regions of strong vorticity at the abutment end are regions of increased flow velocity and associated bed shear stress, where the bed shear stress exceeds the critical bed shear stress. Such regions typically occur upstream of, and in front of, the abutment. The excess shear stresses are responsi- ble for the erosion that occurs farther toward the main chan- nel where the vorticity is weaker. Because the bed of the main channel was fixed for these experiments, the increase in the bed shear stress in the main channel as a result of the con- tracted flow did not result in contraction scour. 134 0.800 0.400 0.200 0.210 0.480 0.190 0.950 0.000 0.516 0.23 0.800 0.400 0.300 0.235 0.580 0.260 0.990 0.050 0.636 0.28 0.800 0.400 0.400 0.275 0.620 0.380 1.100 0.100 0.727 0.41 0.800 0.400 0.500 0.260 0.690 0.410 1.150 0.160 0.803 0.45 0.800 0.600 0.400 0.370 0.640 0.360 1.530 0.000 0.734 0.92 0.800 0.600 0.500 0.385 0.735 0.405 1.620 0.100 0.839 0.94 0.800 0.800 0.500 0.415 0.755 0.260 2.040 0.000 0.799 1.00 1.200 0.400 0.200 0.185 0.300 0.185 0.950 0.015 0.352 0.00 1.200 0.400 0.300 0.180 0.420 0.280 1.010 0.060 0.505 0.00 1.200 0.400 0.400 0.170 0.530 0.320 1.100 0.200 0.619 0.00 1.200 0.400 0.500 0.130 0.700 0.360 1.230 0.375 0.787 0.02 1.200 0.600 0.300 0.300 0.380 0.250 1.310 0.000 0.455 0.12 1.200 0.600 0.400 0.315 0.480 0.370 1.400 0.070 0.606 0.21 1.200 0.600 0.500 0.295 0.540 0.480 1.480 0.130 0.722 0.28 1.200 0.800 0.400 0.380 0.615 0.440 1.715 0.010 0.756 0.61 1.200 0.800 0.500 0.405 0.760 0.510 1.900 0.085 0.915 0.72 1.600 0.400 0.300 0.215 0.425 0.260 1.110 0.025 0.498 0.00 1.600 0.400 0.400 0.200 0.520 0.350 1.240 0.150 0.627 0.00 1.600 0.400 0.500 0.215 0.635 0.440 1.310 0.250 0.773 0.00 1.600* 0.600* 0.300* 0.305* 0.360 0.280 1.390 0.000* 0.456 0.00 1.600 0.600 0.400 0.300 0.530 0.330 1.500 0.050 0.624 0.00 1.600 0.600 0.500 0.280 0.615 0.380 1.630 0.145 0.723 0.02 1.600 0.800 0.400 0.355 0.400 0.360 1.800 0.000 0.538 0.16 1.600 0.800 0.500 0.350 0.630 0.400 1.850 0.070 0.746 0.19 2.000 0.400 0.200 0.190 0.310 0.180 0.940 0.000 0.358 0.00 2.000 0.400 0.300 0.190 0.430 0.250 1.035 0.100 0.497 0.00 2.000 0.400 0.400 0.190 0.470 0.365 1.160 0.130 0.595 0.00 2.000 0.400 0.500 0.200 0.630 0.390 1.250 0.330 0.741 0.00 2.000 0.600 0.300 0.250 0.340 0.250 1.320 0.000 0.422 0.00 2.000 0.600 0.400 0.250 0.460 0.360 1.420 0.080 0.584 0.00 2.000 0.600 0.500 0.250 0.550 0.440 1.540 0.180 0.704 0.00 2.000 0.800 0.400 0.305 0.410 0.350 1.790 0.005 0.539 0.00 2.000 0.800 0.500 0.300 0.460 0.460 1.860 0.100 0.651 0.00 2.000 1.000 0.500 0.330 0.530 0.420 2.150 0.005 0.676 0.10 Cable-Tied Block Protection Bf (m) L (m) W (m) dsf (m) αx (m) αy (m) αe (m) Wmin (m) Rdmax (m) F Table 8-2. (Continued).

If the flow fields at the abutment had been remeasured after the development of a scour hole at the abutment, the velocity around the abutment would have decreased because the flow depth would have been deeper as a result of the scour at the abutment. Consequently, the bed shear stresses around the abutment would also have decreased. It is postulated that the bed shear stresses around the abutment would progressively decrease with the development of scour at the abutment. This process would continue until the stage where the bed shear 135 Riprap Protection W = 0.2 m Cable-Tied Block Protection W = 0.2 m Cable-Tied Block Protection W = 0.3 mTime (hrs) dsf (m) Wmin (m) dsf (m) Wmin (m) dsf (m) Wmin (m) 0.0 0.000 0.200 0.000 0.200 0.000 0.300 0.5 0.050 0.190 0.040 0.275 1.0 0.080 0.190 0.080 0.140 0.060 0.265 1.5 0.090 0.190 0.105 0.100 0.090 0.250 2.0 0.100 0.160 0.125 0.100 0.115 0.225 2.5 0.105 0.130 0.135 0.075 0.125 0.220 3.0 0.110 0.130 0.145 0.030 0.135 0.200 3.5 0.115 0.130 0.150 0.010 0.140 0.170 4.0 0.125 0.130 0.155 0.000 5.0 0.135 0.110 0.160 0.000 0.160 0.150 6.0 0.155 0.100 0.170 0.125 7.0 0.170 0.100 0.175 0.115 8.0 0.170 0.100 9.0 0.180 0.090 0.190 0.080 10.0 0.195 0.080 11.0 0.200 0.070 12.0 13.0 0.220 0.070 0.210 0.050 14.0 0.225 0.070 15.0 0.225 0.030 30.0 0.255 0.025 72.0 0.305 0.020 0.305 0.000 Table 8-3. Non-equilibrium scour measurements for the experiments marked with an asterisk in Table 8-2 (Bf  1.6 m and L  0.6 m). Figure 8-21. Position of the deepest point of scour relative to the spill-through abutment for the experimental data recorded in Table 8-3. Figure 8-22. Equilibrium scour hole at a spill-through abutment with the spill-slope riprap protection extended below the floodplain (Bf  1.6 m; L  0.8 m).

stresses would decrease to the critical bed shear stress level, at which point equilibrium scour conditions would be attained. The problem with remeasuring the flow fields around the abutment after the development of a scour hole at the abut- ment is that the measured flow fields would not be represen- tative of the actual flow fields at the abutment. The reason is that the measuring technique used is a surface particle- tracking technique. At the initial stages of the experiment, the flow can be represented reasonably well by a two-dimensional flow field, but at equilibrium scour conditions the flow field becomes significantly three-dimensional. Consequently, the surface velocity field is not representative of the flow structure around the abutment, as shown by the experimental work of Lim and Nugroho (2004). 8.3.4 Discussion A comparison between the experimental scour depth data with the data presented in Sturm and Chrisochoides (1998a) for the scour depths at abutments situated in compound channels is shown in Figure 8-26. The data from Sturm and Chrisochoides apply to unprotected solid abutment struc- tures. Therefore, only the experiments where the spill-slope protection was extended below the surface of the floodplain (W  0) are included for the purpose of comparison. The flow directly upstream of the bridge opening Qo, as a fraction of the total flow in the compound channel upstream of the bridge crossing QT, was determined from the velocity flow fields. The data from the present study agree reasonably well 136 Figure 8-24. Vorticity field overlaid on the corresponding equilibrium scour formation. Figure 8-23. Velocity vector field overlaid on the corresponding equilibrium scour formation.

forms beyond the floodplain (αe > Bf), the data points lie above the envelope suggested by Melville and Coleman (2000). For the latter experiments, scour depths are affected by the compound channel geometry, which increases the scour depths compared with those for abutments situated in equivalent rectangular channels. This effect is discussed further in the following section. Scour Hole Position The position of the center of the scour hole was recorded for each experiment in terms of αx and αy. The position of the center of the scour hole can also be defined by R and  (see Figure 8-12) using trigonometric expressions. The values of R were calculated for each experiment using 137 Reprinted with permission from ASCE. Figure 8-26. Normalized scour depth as a function of the upstream flow parameters (reproduced from Sturm and Chrisochoides, 1998a). Also included are data from the current experiments where the riprap protection was extended below the surface of the floodplain. Figure 8-27. Normalized scour depth as a function of the normalized flow depth (reproduced from Melville and Coleman, 2000). Also included are scour data from the current spill-through abutment experiments. Figure 8-25. Normalized bed shear stress field overlaid on the corresponding equilibrium scour formation. with the trend for spill-through abutments given by Sturm and Chrisochoides (1998a). Figure 8-27 compares the scour depths from the present study with the results of experimental studies by Gill (1972), Wong (1982), Tey (1984), Kwan (1984, 1988), Kandasamy (1989), and Dongal (1994), as presented by Melville and Coleman (2000). The envelope suggested by Melville and Coleman (2000) gives a reasonable upper estimate of the scour depth for the experiments where the scour develop- ment at the abutment occurred on the floodplain only (e < Bf). However, for experiments where part of the scour hole

(8-11) (8-12) Where C1, C2, , and  depend on the type of protection. For riprap, C1  C2  4,   0.4, and   0.1, while for cable-tied blocks, C1  1.0, C2  2.2,   0.9, and   0.4. Longer abutments (larger L) induced deeper scour holes and correspondingly increased Rdmax and e. Increasing W deflects the scour hole farther away from the end of the abut- ment, thereby increasing Rdmax and e, consistent with the results from the experimental studies of Croad (1989), Eve (1999), Hoe (2001), Cheung (2002), and Martinez (2003). Scour holes at abutments protected by cable-tied blocks form closer to the abutment than scour holes at abutments pro- tected by equivalent riprap aprons—that is, R and e are larger for riprap protection than for cable-tied block protec- tion, as discussed in Section 8.3. Scour Depth Three different cases of scour development were identified in this study, depending on the position of the scour hole in relation to the boundary between the main channel and the floodplain (shown schematically in Figure 8-31): • The scour hole formed entirely on the floodplain (e  Bf). • The scour hole extended beyond the floodplain onto the main channel bank and, in some cases, into the main channel (e  Bf).   e f f fy C L y W y = ⎛ ⎝⎜ ⎞ ⎠⎟ + ⎛ ⎝⎜ ⎞ ⎠⎟2 0 7 1 . R y C L y W yf f f = ⎛ ⎝⎜ ⎞ ⎠⎟ + ⎛ ⎝⎜ ⎞ ⎠⎟1 0 2 1 .  138 Figure 8-28. Normalized longitudinal position of the deepest point of the scour hole as a function of the normalized lateral position of the deepest point of the scour hole. Figure 8-29. Comparison of the predicted distance from the end of the abutment to the deepest point of the scour hole using Equation 8-9 with the measured data. Equation 8-9, while the angle  was calculated from the following: (8-10) Figure 8-28 is a plot of y against x for all of the experi- ments. The gradient of the line of best fit in Figure 8-28 is tan, where a value of   30 degrees was determined. Expressions for R and e were derived from the data of Table 8-2 using regression analysis. Figures 8-29 and 8-30 show the results of the regression analysis for R and e, respectively, which were found to be dependent on L and W and independent of Bf. The expressions for R and e are as follows: tan   = y x Figure 8-30. Comparison of the predicted distance from the side of the flume to the outer edge of the scour hole using Equation 8-10 with the measured data.

• The scour hole formed in the main channel (e  Bf). For all the cases, the scour depth is measured relative to the bed level of the floodplain dsf and can be related to the scour depth relative to the local bed level ds by the follow- ing equation: (8-13) Where F is a function that accounts for the effects of e/Bf and L/Bf on the scour development, and e/Bf and L/Bf rep- resent the relative position of the scour hole and abutment in the compound channel, as illustrated in Figure 8-12. F, which takes values between zero and unity, is given as follows: when (8-14) when (8-15) The exponent in Equation 8-14 was determined from a regression analysis of the data, as discussed below. Equation 8-14 is plotted in Figure 8-32 for different values of e/Bf and L/Bf. Equations 8-13 and 8-14 apply only to the channel geometry cases C and D, which means that L ≤ Bf. For this rea- e fB < 1F = 0 e fB >1F L Bf B e f = − − ⎛ ⎝⎜ ⎞ ⎠⎟ − ⎛ ⎝⎜ ⎞ ⎠⎟ 1 1 2 1 d d F y ysf s m f= + −( ) son, L/Bf takes values between zero and unity only, as shown on the horizontal axis of Figure 8-32. When the scour hole forms entirely on the floodplain— that is, when e  Bf and F  0—then Equation 8-13 reduces to dsf  ds. When L  Bf, then the scour depth forms mostly in the main channel, F  1, and dsf  ds  (ym-yf). Values of F for each of the experiments are given in Table 8-4. For the riprap experiments, the normalized (i.e., equilib- rium) scour depth data of Table 8-2 are plotted in Figure 8-33 in terms of L/yf, W/yf, and Bf/yf. The solid and hollow symbols signify scour data where e  Bf and e  Bf, respectively. Figure 8-33 shows that the influences of W/yf and Bf/yf on the scour depth depend on e/Bf, as discussed in the following paragraphs. 139 Figure 8-31. Three different cases of scour development at the spill-through abutment. Bf (m) L (m) Vtip (ms-1) ωmax (s-1) dsf (m) F ds (m) 0.800 0.400 0.36 1.9 0.200 0.27 0.146 0.800 0.600 0.40 2.3 0.355 0.89 0.224 0.800 0.800 0.39 2.9 0.435 1.00 0.285 1.200 0.400 0.34 0.8 0.155 0.00 0.155 1.200 0.600 0.37 2.0 0.250 0.07 0.219 1.200 0.800 0.45 2.6 0.315 0.54 0.223 1.200 1.000 0.49 3.2 - - - 1.200 1.200 0.50 3.5 - - - 1.600 0.400 0.37 2.4 0.205 0.00 0.205 1.600 0.600 0.40 2.6 0.290 0.00 0.290 1.600 0.800 0.44 3.2 0.315 0.08 0.301 1.600 1.000 0.51 3.7 - - - 1.600 1.200 0.61 3.9 - - - 1.600 1.400 0.70 4.7 - - - 1.600 1.600 0.63 5.5 - - - 2.000 0.400 0.36 2.4 0.200 0.00 0.200 2.000 0.600 0.43 2.6 0.260 0.00 0.260 2.000 0.800 0.48 3.5 0.290 0.00 0.290 2.000 1.000 0.55 3.5 0.325 0.00 0.325 2.000 1.200 0.62 3.8 - - - 2.000 1.400 0.66 4.6 - - - 2.000 1.600 0.73 4.9 - - - Table 8-4. Flow field characteristics and scour depths (W  0). Figure 8-32. Plot of Equation 8-14 for the F function.

The lines in Figure 8-34 represent Equation 8-13 for the four different Bf/yf values used in the experimental study. As in Figure 8-33, the solid and hollow symbols in Figure 8-34 signify scour data where e  Bf and e  Bf, respectively. The lower line applies to the case where e  Bf (i.e., F  0), while the upper lines incorporate the adjustment corresponding to F. When e  Bf, the data follow the power-law relation of Equation 8-13. However, when e  Bf, the data diverge from this line. This divergence can also be seen in Figure 8-33, par- ticularly for the L/yf  4 data with increasing W/yf. When e  Bf, the scour depth tends to decrease with larger W/yf (Figure 8-33), and increase slightly with increasing Bf/yf (Figure 8-34), although this trend is less marked. As Bf increases, the flow area at the bridge section decreases and the mean flow velocity is correspondingly higher, resulting in deeper scour. This trend is consistent with that noted by Sturm and Janjua (1993) and was observed also in flow field measurements undertaken in the present study (discussed in Section 8.1). For smaller apron widths (W/L  0.75), scour depth is approximately independent of W, while for wider aprons, scour depth decreases with increasing W/yf. The reduced scour for wider aprons is a consequence of the lateral flow distribution in the flood channel. The strongest flow occurs at the end of the abutment, and the flow strength decreases away from the abutment. When e  Bf, scour depth increases with increasing W/yf (Figure 8-33) and decreases as Bf/yf increases (Figure 8-34). The latter trend is a consequence of the abutment set-back distance being larger with larger Bf/yf, for which more of the scour hole is situated on the floodplain. With wider protec- tive aprons, the scour hole develops farther away from the abutment. When e  Bf, the scour hole development occurs farther into the main channel, where deeper scour occurs rel- ative to the floodplain bed level. Figure 8-35 compares the measured scour depth data with the predicted scour depths using Equation 8-13. Most of the scour data from Table 8-3 are included in Figure 8-35, although a few data points for very wide aprons at short abut- ments (W  0.75L) are excluded because such aprons are impractically wide. Figure 8-35 demonstrates a good agreement between the measured and predicted scour depths, supporting the incorpo- ration of the new F function (to account for compound chan- nel effects for Cases C and D, as shown later in Figure 8-46) in the scour depth prediction equation of Melville and Coleman (2000). 140 Figure 8-33. Variation of the scour depth with abutment length, floodplain width, and apron extent. Figure 8-34. Variation of the scour depth with abut- ment length. Figure 8-35. Comparison of the predicted scour depth using Equation 8-13 with the measured scour depth data.

Minimum Apron Extent The apron extent Wo, for which Wmin  0, was measured for each experimental configuration—that is, for every combination of L/yf and Bf/yf. The data are plotted in Figure 8-36, where the encircled data represent values for which Wo was measured directly. In the plot, Wo is approximated by W-Wmin for data in the range W/L  0.75. This assumption is discussed below. The plot also includes data by Hoe (2001) and Eve (1999). The following expression for Wo is an envelope to the equi- librium scour data (shown in Figure 8-36) and additional nonequilibrium scour depth data for a few experiments: (8-16) Where C3  0.5 and 1.4 and   1.35 and 1.0 for riprap and cable-tied blocks, respectively. In applying Equation 8-16 to nonequilibrium scour depths, it is assumed that the equation is applicable during scour development as well as at the equi- librium condition. The assumption is reasonable, given the similarity of the scour hole shape throughout its development and the inclusion of some nonequilibrium scour depth data from Table 8-3 in Figure 8-36, as noted above. As a consequence of the dependencies of scour depth on L and Bf, Equation 8-16 implies that Wo/yf increases with increasing L/yf and tends to decrease as Bf/yf increases when e  Bf. Larger scour holes are developed at longer abutments, requiring wider aprons for protection. Narrower aprons are required for abutments situated on wider floodplains because a greater portion of the scour hole develops on the floodplain as the floodplain width increases, resulting in smaller scour depths relative to the floodplain. W y C d y o f sf f = ⎛ ⎝⎜ ⎞ ⎠⎟3  The dependency of the minimum required apron width with scour depth is consistent with the experimental studies of Hoe (2001) and Cheung (2002). Both of the experimental studies showed that with increasing approach-flow velocity (resulting in deeper scour holes), wider aprons were required to prevent abutment failure. It is apparent from Figure 8-36 that Wo is larger for cable- tied block aprons, compared with equivalent riprap aprons, because scour holes at abutments protected by cable-tied blocks form closer to the abutment than for equivalent riprap aprons (as discussed in Section 8.3.3). Therefore, a narrower riprap apron will afford a greater level of protec- tion to the base of the abutment spill slope compared with an equivalent cable-tied block apron. This implies that cable-tied block aprons need to be larger than riprap aprons to afford the same level of protection at abutments, consistent with the experimental studies of Croad (1989), Eve (1999), Hoe (2001), and Cheung (2002). The cable-tied block experimental studies of Hoe (2001) and Cheung (2002) showed that cable-tied block apron widths equal to twice the flow depth did not provide sufficient protection for the abutment, whereas the riprap experimental studies of Croad (1989) and Eve (1999) showed that riprap apron widths equal to twice the flow depth provided adequate protection for the abutment. This implies that the use of riprap aprons is preferable to the use of cable-tied block aprons to protect spill-though abutments from clear-water scour. By definition, W  Wo when Wmin  0. Thus, Wo defines the minimum apron width to prevent erosion of the toe of the abutment. If the toe is not sufficiently protected (i.e., W  Wo), the scour hole undermines the spill-slope fill material, causing the fill material to slump into the scour hole. For the case when W  Wo (i.e., Wmin  0), the edge of the scour hole is deflected away from the toe of the abutment, and Wmin is given as follows: (8-17) If the apron width increases to a width greater than W/L  0.75, the scour hole reduces in size until eventually the apron is wide enough to eliminate all local scour at the abutment. 8.3.5 Comparison of the Scour Depth Prediction Method with Other Experimental Data Other experimental data for scour depth at abutments sit- uated on floodplains of compound channels are reported in Sturm and Janjua (1993) and Cardosso and Bettess (1999). Figure 8-37 compares these experimental data with the W W Wmin = − 0 141 Figure 8-36. Minimum apron extent Wo as a function of scour depth.

predicted scour depths of Equation 8-13. Figure 8-37 also includes the data given in Figure 8-35. Figure 8-37 shows that the data of Sturm and Janjua (1993) and Cardosso and Bettess (1999) are overpredicted by Equation 8-13. The measured data from Sturm and Janjua are smaller than predicted because the scour experiments were run for only 12 hours—that is, the scour depths recorded were less than the equilibrium scour depths. The measured data from Cardosso and Bettess are smaller than predicted because the velocities in the main channel were above the threshold conditions for sediment motion. For the cases when the scour hole formed partly in the main channel, the live-bed condi- tions in the main channel reduced the maximum clear-water scour depth. Flow Field Correlations with Scour Hole Depths The velocity just outside the separation zone at the end of the abutment Vtip and the vorticity strength max for each abut- ment and compound channel configuration were determined from the flow field measurements. These are summarized in Table 8-4. The corresponding equilibrium scour depths dsf and adjusted scour depth ds are also given in Table 8-4 for the W  0 experiments. Figure 8-38 shows a positive correlation between ds and Vtip, consistent with the results of Sturm and Janjua (1993). Figure 8-38 also shows a positive correlation between ds and max, consistent with the experimental work of Kirkil et al. (2004), which showed that the normalized scour depth increases with increasing vorticity strength at bridge piers. The scour hole depth increases with increasing velocity at the abutment end and vorticity strength because the increas- ing flow strength is capable of eroding more bed material from the scour hole. The Vtip and max parameters were used in a regression analysis to develop an empirical expression to predict the scour depth using the values obtained from the measured abutment flow field data. The results from the regression analysis of the normalized scour depth data (ds/yf) with Vtip and max are shown in Figure 8-39. Figure 8-40 shows the effect of a dimensionless expression containing max on normalized ds. It can be seen that the dimensionless expression 2.4(maxVtip/g)0.35 collapses the data well and could be used as an alternative scour depth predic- tion method when the flow fields around bridge abutments are modeled numerically. Thus, the alternative scour depth prediction method can be expressed as follows: (8-18) d y V g s f tip = ⎛ ⎝⎜ ⎞ ⎠⎟2 4 0 35 . . max 142 1993 Figure 8-37. Comparison of the predicted scour depth using Equation 8-13 with the measured scour depth data shown in Figure 8-35 and the measured data from Sturm and Janjua (1993) and Cardosso and Bettess (1999). Figure 8-38. Normalized scour depth plotted against the measured velocity at the end of the abutment just outside the separation zone. Figure 8-39. Normalized scour depth plotted against the maximum vorticity in the flow field at the abutment.

Figure 8-41 compares scour depth predictions using Equa- tion 8-13 with those using Equation 8-18 in terms of the flow field parameters listed in Table 8-4. Figure 8-41 shows that there is a good agreement between Equations 8-13 and 8-18. By substituting Equation 8-18 into Equation 8-13, the geo- metric parameters of the abutment can be related to the flow field parameters at the abutment as follows: (8-19) Equation 8-19 shows that scour depth increases with increasing L, Vtip, and max, consistent with the observations noted above. The direct implication is that Vtip and max d Ly y V gs f f tip = = ⎛ ⎝⎜ ⎞ ⎠⎟2 4 0 35 . . max increase with L, consistent with the observations noted in Section 8.3. It is important to note that Equations 8-18 and 8-19 are limited to when the abutment is aligned perpendic- ular to the flow—that is, when V/Vc  1, yf /ym  0.4, and 4  L/yf  10. Further investigation into the effects of Vtip and max on the scour depth at the abutment is needed before Equations 8-18 and 8-19 can be applied beyond the data range tested. 8.4 Two-Dimensional Modelling of Flow Around a Small-Scale Model Abutment 8.4.1 Introduction Selection of countermeasures to protect bridges from scour requires estimates of velocity distributions in the bridge opening. Estimates of the peak velocity in what is typically a highly nonuniform flow distribution near the tip of the abut- ment is necessary to determine whether countermeasures are necessary and, if so, to determine the type, size, and extent of countermeasures to protect bridge abutments from scour. Laboratory physical models have been developed to deter- mine the size, type, and location of protection for a relatively small range of flow conditions at bridges; however, the labo- ratory models represent very simplistic geometric conditions. Effective transfer of laboratory model results to the complex hydrodynamic conditions of real bridge sites requires that flow velocity be predicted in the vicinity of bridge abutments using numerical models; however, the degree to which numerical models, typically used by highway engineers, can represent the highly nonuniform flow around abutments has not been examined. The main purpose of this component of the study was to compare flow distributions obtained from a two-dimensional shallow water numerical model to flow distributions meas- ured in small-scale laboratory model studies of flow around abutments. The two-dimensional shallow water model FESWMS (Finite Element Surface Water Modeling System), developed for analysis of bridges for the Federal Highway Administration, is used by highway agencies throughout the United States; therefore, it was used in this study. Also, the mesh-generating and postprocessing program SMS 8.1 was used in developing the computational mesh and in postpro- cessing the numerical results. This study illustrates the com- parison of numerical and physical model results and an evaluation of the effectiveness of the two-dimensional model to simulate flow around abutments under small-scale labora- tory conditions. As noted previously, a particle-tracking velocimetry (PTV) was used to estimate the surface velocity of flow for two model abutments and four different channel and floodplain 143 Figure 8-40. Comparison of the predicted scour depth using the flow field parameters from Table 8-4 in Equation 8-16 with the measured scour depth data for W  0. Figure 8-41. Comparison of the predicted scour depths of Equation 8-13 with the predicted scour depth of Equation 8-18 for W  0.

configurations. The physical model’s configuration of abut- ment, channel, and floodplain, diagrammed in Figure 8-42, was chosen to evaluate the numerical model. The physical model configuration included a spill-through abutment 600mm long with 1:1 side slopes, a floodplain 1.6m wide, and a main channel 0.8m wide. The flow rate was con- stant at 127.4 l/s; the flow depth was 100mm in the floodplain and 250mm in the main channel. The flume slope was set at 0.002m/m. The floodplain surface was sand with a d50 of 0.7mm, and the channel surface was roughened concrete except for the test section, which was composed of sand of the same gradation as the floodplain. The abutment face was cov- ered with uniformly graded riprap with intermediate axis between 18 and 22mm. 8.4.2 Finite Element Mesh and Hydraulic Coefficients of the Two-Dimensional Model A finite element mesh was created using SMS Version 8.1. Figure 8-43 shows the developed mesh at two different scales. The mesh was composed of a combination of 11,966 nine-noded quadrilateral elements and 679 six-noded tri- angular elements, requiring 49,786 nodes. Element dimen- sions varied from 5mm in the region of suspected flow separation around the abutment to 352mm at the model upstream and downstream boundaries. Variation of mesh size was controlled by restricting the change in element area by adjacent elements to less than 40 percent. Nine-noded quadrilateral elements were used wherever possible. Trian- gular elements were used to change element density or where ambiguous bed slope developed in quadrilateral ele- ments, thereby requiring that one quadrilateral be divided into two triangular elements. Slopes along the main channel, on the face of the abut- ment, at the downstream end of the flume false floor, and at the tailgate all exceed 10 percent. At the locations where sub- stantial components of the flow are in the direction of the slope, significant error in flow acceleration is possible (Froehlich, 2002). The two-dimensional model roughness coefficient for the channel and floodplain was calibrated such that the numeri- cal model depth matched the measured depth (100mm) in the physical model floodplain and, to the extent possible, the velocity distribution observed for the floodplain at the upstream and downstream extents of the flume and for the cross-flume velocity distribution prior to placement of the abutment in the model. This two-dimensional model calibration resulted in a Manning n of 0.019 for the channel and 0.016 for the floodplain. The eddy viscosity for the main channel was also calibrated as 0.001m2/s. The two- dimensional model eddy viscosities for the floodplain varied from 0.00002 to 0.002m2/s. The largest values of eddy viscos- ity were applied in the entrance and exit of the two-dimen- sional model and in straight reaches with large elements, whereas the smallest values of eddy viscosity were applied in 144 Figure 8-42. Plan view of the physical model as represented in the numerical model for the 600-mm spill-through abutment.

the low-velocity regions of the wake. No further calibration was performed. The Peclet number is defined as: Where: L  characteristic length of an element (m), V  representative velocity magnitude (m/s), and Ev  kinematic eddy viscosity (m2/s). The Peclet number varied from 4.5 in the low-velocity region of the wake downstream of the abutment to 150 near the high-velocity regions of the flow boundary. In the flood- plain and in the complex flow region near the abutment, the element size and the eddy viscosity were reduced until flow separation was simulated upstream of the abutment center- line. A combination of an element length of 5mm and an eddy viscosity of 0.0001 was required to model separation upstream of the abutment centerline. 8.4.3 Boundary Conditions and Roughness Characteristics of the Two- Dimensional Model Boundary conditions for the numerical simulation included upstream flow input (127.4 l/s), downstream water Pe L V Ev = surface elevation based on the flow depth (100 mm), and semi-slip conditions along the walls and abutment (Manning n for flume walls). A flume slope of 0.002 m/m, as determined from a survey, was used in the model. Manning n values were determined from the roughness characteristics of the chan- nel, bank, and floodplain of the model. The downstream extent was complicated by the end of the false floor used to model the floodplain and trap sediment and the tailgate. Ear- lier simulations indicated that a large recirculation zone downstream of the abutment extended beyond the end of the false floor. Consequently, flow was entering the computa- tional model from a large part of the downstream boundary. To numerically model the entire recirculation zone, the mesh was extended to the flume tailgate. 8.4.4 Two-Dimensional Model Results The two-dimensional simulation showed an increase in velocity both on the floodplain and in the main channel, as indicated in Figures 8-44 through 8-46. The location of sim- ulated maximum increase in velocity associated with flow contraction was predicted to be on the abutment’s sloping face upstream of its centerline; however, the flow in the remaining floodplain and channel continues to contract and accelerate to a point well downstream of the abutment. The point of highest simulated velocity on the embankment face is in an area of the flow field with high curvature; therefore, 145 Figure 8-43. Finite element mesh composed primarily of nine-node quadrilateral and six- node triangular elements.

the depth-averaged velocity at this location is likely to be inaccurate. Depth-averaged velocity and computed boundary shear stress were plotted for four cross-flume transects, as shown in Figures 8-47 and 8-48. The first transect was obtained at the flume station of 5 m and represents the approach flow, although backwater effects have developed at this location. The second transect, at station 8.18m, was across the abut- ment, approximately 40mm upstream of the transect repre- senting the abutment centerline. This transect at 8.18m passes through the region of highest local velocity (0.48m/s) and highest boundary stress (5.0N/m2) computed by the two- dimensional model. The high boundary stress can, in part, be attributed to a combination of high local flow velocity and high local roughness caused by the layer of gravel on the face of the abutment slope. The rapid decrease in boundary shear stress in the 8.1-m transect is caused by the rapid change in roughness from the abutment slope to the adjacent model floodplain (sand grain roughness). Flow separation, as indi- cated in Figure 8-47, was computed to occur along a transect approximately 50mm upstream of the abutment centerline transect and approximately the same distance downstream of the maximum velocity and stress transect. Although flow separation from the abutment tip was sim- ulated in the approximate location anticipated (upstream of the abutment centerline), the increase in flow velocity in the vicinity of the abutment tip was not as high as expected. 8.4.5 Comparison of Flume PTV-Measured and Two-Dimensional Modelled Velocity Magnitude Figures 8-49 and 8-50 show plots of both PTV and two- dimensional model velocity magnitude data. Again, keep in 146 Figure 8-44. Numerically simulated velocity field over the entire model reach of the flume for 600-mm spill- through abutment (velocity contours in m/s). Figure 8-45. Numerically simulated velocity field in the region of the significant channel and floodplain contraction and expansion for 600-mm spill-through abutment (velocity contours in m/s).

147 Figure 8-47. Depth-averaged velocity along transects across flume. The transects are located at flume stations 5.00 m, 8.18 m, 8.22 m, and 8.91 m, as indicated in Figure 8-42. 0.47 0.45 0.420.37 0.35 0.32 0.30 0.27 0.25 Less than 0.100 0.40 Figure 8-46. Numerically simulated velocity field near the abutment tip for 600-mm spill-through abutment (velocity contours in m/s).

148 Note: two-dimensional model velocities are depth-averaged, and PTV velocities are from the water surface only. Figure 8-49. Comparison of PTV-measured velocity magnitude along transects upstream of the model abutment with that computed from the two-dimensional model simulation. Figure 8-48. Bed shear stress along transects across flume. The transects are located at flume stations 5.00 m, 8.18 m, 8.22 m, and 8.19 m, as indicated in Figure 8-42.

mind that the PTV data represent flow velocity at the water surface in the physical model, while two-dimensional model data represent computed depth-averaged flow velocity. Comparisons of local peak velocity and average velocity over the floodplain and main channel are provided below. Local Peak Velocity Because of the tendency for flow to accelerate around abut- ments, the local peak in flow velocity is used in some scour pre- diction methods or in the design of scour protection. Table 8-5 shows that local peak velocity on the floodplain and in the main channel was predicted well by the two-dimensional model, although the location of the local maximum was not. In addition, the two-dimensional model computes the highest velocity along the transect over the riprap; PTV data, however, were not available at precisely the same location. Table 8-6 compares the average velocities of the two mod- els for the floodplain of the flat part of the channel. Regions 149 Note: two-dimensional model velocities are depth-averaged, and PTV velocities are from the water surface only. Figure 8-50. Comparison of PTV-measured flow velocity magnitude along a transect 40 mm upstream of the model abutment centerline, with flow velocity magnitude obtained from the two-dimensional model simulation. Source of Data Transect Station (m) Over Riprap Slope of Abutment (m/s) Over Floodplain (m/s) In Main Channel (m/s) Two-Dimensional Model 8.22 0.43 0.45 0.54 Two-Dimensional Model 8.18 0.48 0.44 0.54 Flume PTV 8.18 0.41 0.45 0.53* Two-Dimensional Model/PTV Difference 8.18 17% 2% 2% *Velocity taken from center of channel rather than along flume wall, where a slightly higher velocity was reported. Table 8-5. Direct comparison of maximum velocity magnitude at abutment transects. Source of Data Transect Station (m) Over Floodplain (m/s) In Main Channel (m/s) Two-Dimensional Model 5.00 0.31 0.46 Two-Dimensional Model 6.67 0.29 0.46 Flume PTV 6.67 0.36 0.46 Two-Dimensional Model 8.18 0.39 0.50 Flume PTV 8.18 0.44 0.52 Two-Dimensional Model 8.22 0.34 0.50 Two-Dimensional Model/PTV Difference 6.67 20% 0% Two-Dimensional Model/PTV Difference 8.18 11% 4% Table 8-6. Comparison of average velocity magnitude.

of flow over the abutment slope and over the sloping channel bank were not included in the averages for the floodplain and channel, respectively. The average velocity in the channel is well predicted by the two-dimensional model. Flow velocity observations obtained from the flume PTV measurements and those computed by FESWMS are plotted in Figures 8-51 and 8-52, respectively. As shown in Figure 8-51, one unusual velocity observation in the PTV measurements causes the PTV point data to deviate from the two-dimensional model results. The average velocity for the PTV floodplain data is significantly higher than that of the two-dimensional model for transects 8.18 and 8.67. The velocity magnitudes were plot- ted along transects perpendicular to the flume walls in the flow upstream of the modeled abutment and near the abutment centerline. The transect representing the upstream extent of PTV measurements (approximately 1,550mm upstream of the abutment centerline transect) is compared with two- dimensional, depth-averaged velocity magnitudes at the same location and at a second location 3,220mm upstream of the abutment centerline. The PTV-measured velocity in the flood- plain (from 0m to 1.6m along the transect) is higher than that computed by FESWMS. Overall, good agreement is demonstrated between the two- dimensional model results and the PTV measurements. 8.5 Large-Scale Tests of Riprap Apron Performance 8.5.1 Introduction Described here are the findings from relatively large-scale flume experiments conducted to validate the main recom- mendations from the extensive parametric flume experi- ments, which are described in Section 8.3. In particular, these tests had two objectives: • To ascertain the minimum width (W0) of riprap apron placed around a spill-through abutment that has a much larger size than, though nominally similar flow geometry to, the abutments described in Section 8.1. • To obtain additional detailed information on apron per- formance. The large-scale tests were run for clear-water scour condi- tions that replicated a spill-through abutment located on a floodplain and at some distance back from the main channel of a river. However, the large-scale tests were done at a geo- metric size approximately four times larger than the abut- ment size used in Section 8.1; here, the standard size is taken to be the top (i.e., road) and base widths of abutment, as well as abutment height. Flow depth was slightly more than five times the flow depth; a larger depth factor was dictated by constraints in flume operation. The extent of the riprap apron 150 Figure 8-51. Velocity observations obtained from the flume PTV measurements (apron length  0.80 m). (a) Contours of bed elevation after scour (b) Unit discharge distribution around the abutment (a) Contours of bed elevation (b) Unit discharge distribution around the abutment 7 Figure 8-52. Velocity observations computed from FESWMS (apron length  0.50 m).

and the stone size conform in scale to the values mentioned in Section 8.1. Also, the clear-water approach-flow condition matched those for the experiments described in Section 8.1. Flume tests on abutment scour at this scale have not been undertaken heretofore. Morales (2006) fully documents the flume tests. Similitude between laboratory experiments and field scale experiments was satisfied by the use of the aforementioned u*/u*c ratio, of which a value of just below 1.0 represents a condition called “clear-water scour.” This condition is extreme for scouring because the velocity is as high as possi- ble without the movement of the channel bed, which causes infilling of the sediment hole. 8.5.2 Tests The test layout consists of a subcase of the test program described in Section 8.1. The layout is not an exact up-scaling of an abutment and floodplain layout in Section 8.1. The lay- out was designed with the intent of having maximum length and height of abutment for the given constraints of flume width and pump capacity. Test Layout The experiments were performed using a large flume at a facility in IIHR, a unit of the University of Iowa’s College of Engineering (Figure 8-53). The flume, shown in Figure 8-54, was fitted with a simulated portion of floodplain. The flume consisted of a rectangular open channel 3.05 m (10 ft) wide, 19.81 m (65 ft) long, and 2.29 m (7.5 ft) deep. The test abut- ment was placed on a sediment recess in the simulated flood- plain. The recess was 0.9 m (3 ft) deep and 7.00 m (23 ft) long. The approach to the test section was roughened with blocks so as to trip the flow boundary layer and thereby create a fully turbulent velocity profile for the flow approach to the 151 Figure 8-53. Environmental flow facility. Figure 8-54. Layout of the large-scale experiment in the flume.

abutment. Downstream of the roughened entry was a 4.87-m (16-ft) long planar surface that was of sand-grain roughness; the grain diameter was 1.05 mm. At the downstream end of the simulated floodplain, an adjustable tailgate was used to control water surface elevations preceded by a sediment trap to prevent the sand from going into the pump. The flow was provided by two 36-in. diameter propeller pumps driven by variable-speed motors operated to produce a regular velocity distribution of the flow upon entering the test area with the desired average velocity. The discharge range attainable with these pumps is 0.60 to 3.54 m3/s (22 to 125 ft3/s). A flow depth of 0.53 m (1.76 ft) was selected because it gave the prescribed clear-water condition of approach flow, yet suitably exceeded the minimum discharge limit for the flume’s pumps; a flow depth of 0.40 m would have required too low a discharge for the pumps to deliver. Flow depth through the test section was controlled by means of a tail gate at the end of the test section. A bed of uniform quartz sand was formed in the recess. The sand had a median diameter of 1.05 mm and a geomet- ric standard deviation of 1.3. The estimated value of the crit- ical shear velocity for the sand is u*c  0.024 m/s. The abutment model was four times the top and base widths, as well as height, of the spill-through abutment used in Section 8.1. It was constructed of wood and sheet metal, painted with a layer of epoxy paint, and placed on the sand recess; it was held in position by internal weighting and ver- tical struts. The abutment’s location in the flume is shown in Figure 8-55, and its form and main dimensions are shown in Figure 8-56; the main dimensions are a top width of 0.40 m, a bottom width of 2.00 m, a height of 0.80 m, and a length of 1.20 m. The abutment’s length was chosen so that the abutment would not contract flow entirely between the abutment and the opposite wall of the flume. The base of the model abutment was surrounded by a band of sheet metal, which in effect simulated a sheet pile skirt. Figure 8-57 shows the skirt. This configuration differed from the abut- ment form used for the experiments conducted at Auckland University insofar that the latter used an abutment form whose slope continued into the bed. 152 Figure 8-55. Layout of the model abutment in the flume, including the lighting and benchmark system used for the LSPIV measurements of water-surface velocities.

An initial experiment was conducted with an abutment formed of loose sand embankment placed around a standard stub abutment structure, as shown in Figure 8-58. This exper- iment sought to illustrate scour development around an unprotected abutment formed of an erodible earthfill embankment. This abutment form was wider that the fixed abutment because of the sand’s angle of repose used for the embankment. The apron of riprap was placed as a circumferential band around the abutment. The apron was formed of two stone layers, giving an average thickness of about 1.5 times the median diameter of stone, and was placed on the simulated floodplain surface. The riprap stone consisted of uniform, crushed rock of median diameter estimated by sieve analysis to be 75 mm and a geometric standard deviation of about 1.4. 153 Figure 8-56. Dimensions of the model abutment and apron. Figure 8-57. Model abutment with skirt. (a) Before scour (b) After scour Figure 8-58. Scour at abutment without apron.

The estimated critical shear velocity for this stone u*C was cal- culated to be 0.26 m/s. Apron width was the principal variable in the experiments, and it was adjusted from 0 to 1.0 m; in terms of flow depth, Y, apron width W varied from 0 to about 2Y; this upper limit is the design width recommended by Lagasse et al. (1997). The width restriction of the flume limited the upper value of W. Figure 8-58a shows the prerun set-up of riprap around a semi-circular spill-through abut- ment fitted with a 0.50-m wide riprap apron with a leveled sediment bed (simulating a floodplain). Figure 8-58b shows the abutment after the experiment. In terms of the parameters used in Section 8.1, the present abutment’s details are V/u*C  0.90, Driprap/d50  75, Bf/yf  5.75, L/yf  2.23, Bf/L  2.5, and W/L  0.33 to 0.83. The values of Bf/yf, L/yf, and Bf/L are less than those discussed in Section 8.1. Flow Conditions For the flow depth set at 0.53 m, and with u*/u*c  0.90, the average velocity of flow V was 0.48 m/s. The values of flow velocity in the approach to the abutment were determined by means of acoustic Doppler velocimetry (ADV), and velocity profiles over the flow depth were confirmed to conform to the general form associated with flow in a fully developed turbu- lent boundary layer. Measurements Large-scale particle image velocimetry (LSPIV) was used to visualize and document the free surface of the flow field around the abutment. Figure 8-55 indicates the system of lighting and benchmarks used in obtaining the LSPIV data. The flow field was documented at the start and end of each test. The flow field measurements were analyzed to obtain estimated distributions of depth-averaged velocity, unit dis- charge, and bed-shear stress around the abutment and its apron. Depth-averaged values of velocity are estimated using the velocity profiles associated with fully turbulent flow in the channel. The LSPIV technique and the associated analyses are documented by Morales (2006). The velocity measurements obtained by way of LSPIV and ADV were also augmented by numerical simulations of the flow field around the abutment. These simulations, which were carried out using the depth-averaged, two-dimensional simulation code FESWMS-2DH, are also documented by Morales (2006). Only selected findings from the simulations are presented herein. The bed elevations at the end of the experiment were taken using a Total Station TOPCON® GTS 226 with resolution up to 0.0003 m. Readings were taken at regular intervals. Downstream of the scour hole, measurements were taken at 0.25-m intervals. Around the abutment and in the scour hole, measurements were taken at 0.10-m intervals. Because there was no erosion of the bed near the abutment, the bed eleva- tions there were measured at 0.45-m intervals. Test Program Table 8-7 provides the test program, as well as the maxi- mum depths of scour measured for each test. Note that, because the abutment was placed on top of the sand bed and did not extend into the sand bed, as scour developed the abutment was undermined. Had the abutment been built of an earthfill embankment around an abutment structure, the embankment would have failed by way of slope instability. 154 Test Number Abutment Length L (m) Channel Half-Width Bf (m) Flow Depth yf (m) Apron Width W (m) Maximum Scour Depth dsmax (m) Scour Depth at Axis dc (m) 1 1.20 3.05 0.53 0.00 0.315 0.300 2 1.20 3.05 0.53 0.15 0.355 0.330 3 1.20 3.05 0.53 0.20 0.347 0.312 4 1.20 3.05 0.53 0.25 0.200 0.170 5 1.20 3.05 0.53 0.25 0.385 0.307 6 1.20 3.05 0.53 0.30 0.397 0.305 7 1.20 3.05 0.53 0.40 0.457 0.352 8 1.20 3.05 0.53 0.50 0.443 0.306 9 1.20 3.05 0.53 0.60 0.451 0.257 10 1.20 3.05 0.53 0.70 0.337 0.254 11 1.20 3.05 0.53 0.70 0.335 0.193 12 1.20 3.05 0.53 0.80 0.228 0.155 13 1.20 3.05 0.53 1.00 0.281 0.183 W = average width of apron. Actual width varied ±d50/2, where d50 = median diameter of riprap stone. Table 8-7. Test program and maximum scour depth measured in the experiments.

8.5.3 Results The results presented herein include data on the maxi- mum depth and location of the consequent region of scour, along with information illustrating the variations in flow field corresponding to the scour region. As clear-water scour asymptotically approached an equilibrium condition over time, each experiment was run for about 72 to 80 hours until negligible change was observed in the scour hole dimensions. Scour Depth and Location The location and depth of scour at the model abutment depend on the resistance of the abutment form to erosion and on the placement of the abutment within (or on) the flood- plain base. The scour tests with the abutment surrounded by a skirt, and with an abutment of erodible embankment, pro- duced markedly different scour forms than those that appeared when the abutment was surrounded by an apron. Without the protection of an apron of riprap stone, the scour region developed immediately around the edge of the fixed abutment and exposed the simulated sheet pile skirt underneath the abutment toe. Figure 8-58 shows this effect as observed in the flume experiments. Two experiments were carried out on the abutment without an apron. In both experiments, the deepest scour was located approximately below the midpoint of the upstream round corner. The sub- sequent discussion of the flow field around the abutment shows that this particular region in the channel is subject to the highest level of turbulence and shear stress. Figure 8-58a shows the initial abutment condition, and Figure 8-58b shows the exact location of the deepest scour. The measured 155 Figure 8-59. View of the model abutment and apron before test (W  0.50 m). Figure 8-60. View of apron launched into scour region downstream of abutment after the test (W  0.50 m). maximum scour depth was 0.315 m 5 mm in both cases. This scour depth was proportionately less than that pre- dicted using FESWMS. Also observed in these experiments was the presence of small dunes inside the scour region, especially on the exit slope of the scour hole. These dunes, having a maximum height of about 15 mm, increase the resistance to flow through the scour region. In effect, the flat dunes add form resistance to flow. By so doing, the dunes in the scour area dis- sipate flow energy and act to reduce scour depth. The abutment formed of the unprotected embankment eroded completely, leaving the stub abutment structure fully exposed. Scour then developed around the stub abutment. Though the abutment was completely washed out, the actual scour depth was comparatively modest. By way of illustration of one test with an apron, Figure 8-59 shows the abutment and a 0.50-m apron before a test, and Figures 8-60 and 8-61 show the apron after the test. The apron stayed intact around the upstream perimeter of the abutment, but the apron frayed around the end of the abut- ment. As expected from the findings described earlier in this

chapter (and earlier chapters), the apron did not prevent scour development, but shifted its position away from the abutment. The apron, though, did inhibit scour immedi- ately at the abutment. Riprap stone forming the apron slid (or launched) into the scour region. Subsequently, in this section, Figure 8-61 presents the bed bathymetry of the scour region. It is important to mention the geotechnical stabilizing influence of an apron. In viewing Figures 8-58 through 8-61, along with Figure 8-62, it is evident that, by forcing the scour region to shift away from the abutment, the apron not only impedes scour at the toe of the abutment but also reduces the effective slope of the embankment face. This effect, indicated in Figure 8-62, increases the geotechnical slope stability of the embankment face. Seen in terms of a slope-stability failure surface, the apron serves to lengthen the arc of the failure surface (thereby adding stability to the embankment slope). Additionally, the counterweight effect of the apron at the toe of the embankment slope has a stabilizing effect. Figure 8-63 summarizes the overall scour trends obtained from the tests by presenting a matrix of views and bathym- etry measurements of the scour regions that occurred as the apron width was widened. The views and bathymetry data assembled in Figure 8-63 show that increasing apron width from 0.76y to about 1.00y barely alters scour depth or loca- tion. As apron width increases from about 1.00y to 2.00y, scour depth drops significantly, although the location of maximum depth changes only slightly, as evident in Table 8-7. When W  2.00y, the maximum depth of scour corre- sponds approximately to the thickness of the apron, which is about 0.15 m. In effect, for the widest apron tested, scour occurs mainly because of turbulence shed by flow passing over the edge of the apron and impinging on the bed down- stream. The eventual scour form produced by a relatively wide apron is akin to scour immediately downstream of a submerged apron in the absence of the abutment. 156 Figure 8-61. View of upstream condition of apron around abutment after the test (W  0.50 m). (a) No apron (b) Apron adds stability by lengthening slip circle or counter-ballasting slip circle in embankment earthfill Figure 8-62. Influence of apron on slope stability of abutment embankment.

Figure 8-64 plots maximum scour depth, dsmax, versus apron width, W. This figure indicates three regions of scour depth trends: • Scour attributable to abutment form and erodibility, • Scour development attributable to the combined structural form of abutment and apron, and • Scour attributable to flow over apron. The first region is likely to be highly variable in depth value, because scour depth and location depend on the abutment foundation condition and erodibility of the abutment embank- ment. The second region shows a reduction in scour depth as apron width increases,until apron width is sufficiently large that scour around the apron’s trailing edge is not substantially influ- enced by abutment presence. The third region shows that flow over the end of the apron causes some scour of the bed. For the abutment with apron, the location of dsmax was slightly downstream of the abutment, as evident in Figure 8-64. Also indicated in Figure 8-64 is the scour depth immediately in front of the abutment, dSO, along the abutment’s centerline axis. The figure also indicates the average thickness of the apron, about 0.12 m. The initial placement of an apron (of width 0.76y) around the abutment substantially reduced dsmax, but then slight increases of apron width only slightly reduced dsmax further. As W increases so that W/y approaches and exceeds about 2, however, dsmax decreased to a minimum approximately equivalent to the nominal thickness of the apron. Values of dSO decreased monotonically for the range of widths considered as apron width increased. Values of the resulting minimum extent of apron, Wmin, at the end of each test versus initial apron width, W0, are plotted in Figure 8-65. The trend here indicates convergence of W0 and Wmin as W0 increases. The tests suggest that riprap stone at the edge of the apron may usually fray from the apron, such that Wmin  W0 only when W0 becomes very large. Figure 8-66 shows views of the minimum extent of riprap after each test for values of W  0.4, 0.5, 0.6, 0.7, 0.8, and 1.0 m. 157 Test #2: W = 0.40 m Test #3: W = 0.50 m Test #4: W = 0.60 m Test #5: W = 0.70 m Test #5a: W = 0.70 m Test #6: W = 0.80 m Figure 8-63. Assembled views of the scoured region for the variable apron widths.

Flow Field Over Apron The trends for scour depth and location are explainable in terms of the flow field around the abutment and over the apron, as well as in terms of apron extent. In this regard, the flow field insights provided by the LSPIV measurements, along with the findings from the two-dimensional numerical simulation, usefully show the following trends in flow field as apron width increases: • The median mean value of depth-averaged velocity of the approach flow to the abutment, with the 0.40-m wide apron, is 0.45 m/s.As the flow passes around the abutment, the flow contracts, producing an overall depth-averaged velocity of 0.55 m/s at the plane extending through the center of the abutment. For all the tests, the maximum value of depth- averaged velocity around the abutment was 0.60 m/s before scour had developed. This velocity occurred a short distance downstream of the abutment and was slightly beyond the downstream edge of the apron, thereby more or less coin- ciding with the area of maximum scour depth. • Slight reductions in maximum velocity and unit discharge over the scour region were observed for increasing apron 158 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0.00 W/yf d sm ax /y f, d c/ y f S C O U R A T A B U T M E N T dc/yf dsmax/yf COMBINED SCOUR SCOUR AROUND APRON No Apron 2.502.001.501.000.50 Figure 8-64. Maximum scour depth, dsmax, versus apron width, W. Figure 8-65. Minimum final width of apron, Wmin , versus initial width, W0, for test conditions.

159 (a) W = 0.4 m (c) W = 0.6 m (d) W = 0.7 m (e) W = 0.8 m (f) W = 1.0 m (b) W = 0.5 m Figure 8-66. Views of the minimum extent of riprap after each test.

lengths. The scour deepening of the bed drew more flow to the scour region, thereby locally increasing unit discharge of water in the region of scour. The lesser scour depths for the wider aprons drew less flow to the scour region. Figures 8-51 and 8-52 show the bathymetry contours and distributions of unit discharge for apron widths of 0.50 to 0.80 m. • Because bed shear stress and flow velocity are related (  u2), it is possible to estimate the maximum bed shear stress near the location of the maximum scour depth. For the approach flow, the bed shear stress is on average 0.34 N/m2; the shear rises at the section through the center of the abutment to attain a value of 0.43 N/m2. The maximum bed shear downstream of the abutment is 0.62 N/m2. • The numerical simulations show that increased apron width mildly decreases the bed shear stress near the abut- ment, though increasing it away from the abutment. This influence of an apron is attributable to the influence of apron roughness in reducing flow velocities near the abut- ment. 8.5.4. Comparison with University of Auckland Results The overall scour forms observed in the large-scale tests concurred with those found in the flume tests described in Section 8.1, especially those described in Figures 8-7, 8-10, and 8-11. Relative to abutment position, the locations of deepest scour coincide reasonably well. Scour depths, though, were proportionately less for the large-scale tests. The difference in location of maximum scour for the pres- ent test with the large abutment without apron protection is due to the different form of the abutment below the bed level; the model abutment at Auckland continued its side slope below the bed level. The depth of scour was smaller, relative to flow depth, at the unprotected abutment in the present tests than at the test abutments at Auckland. The lesser scour depth was due to the larger scale of the model. Ettema et al. (2006) explain how scale effects that are incurred with simu- lating the vorticity of eddies generated by flow around a cylin- der may produce an amplified scour depth in a smaller model. The vorticity of eddies is smaller in the larger model. Addi- tionally, as mentioned above, the presence of dunes inside the scour region formed on the exit slope of the present large- scale abutment increased the resistance to flow through the scour region and thereby reduced scour depth. The scour depth trend obtained for the large-scale abut- ment, which replicates a short abutment on a comparatively wide floodplain, is similar to the trend obtained in the small- scale tests when L/Bf  0.2. Scour depth decreases in two stages as apron width is increased. For small values of L/Bf, the presence of the main channel does not affect the abutment flow field and scour development. The lesser depths for the large abutment can be attributed to two factors: • The value of L/yf for the large abutment is 2.23, which is less than the values used for the small-scale abutments (L/yf  4 to 8). The values of L/Bf, however, are in the same range; L/Bf  0.39 for the large-scale abutment and L/Bf  0.20 to 0.50 for the small-scale abutment. A smaller value of L/yf, for an equivalent value of L/Bf, means that less floodplain flow must pass around the abutment. Accordingly, flow velocities at the abutment are proportionately less; therefore, less scour depths will result. • Magnitudes of flow vorticity generated by flow around the abutment and over the apron are proportionately larger in the small-scale tests than in the large-scale tests.As both series of tests were conducted with the parameter V/VC as the prin- cipal criterion for dynamic similitude, and both test series involved beds of coarse sand, tests at the smaller scale have the greater exaggeration of flow vorticity and therefore experi- ence greater entrainment and movement of bed sediment. The design recommendation can be used to estimate the minimum initial width of apron, W0, such that scour would not fully launch the apron and thereby begin to undermine the embankment of the abutment used in the present, large- scale tests. From Equation 8-16, the scour depth estimated for the present test is 0.74 m; from yf  0.53 m, L  1.20 m, and V/VC  0.90. Equation 8-16 is repeated here as (8-20) Where: C3  0.9 and   1.35. In accordance with this equation, the predicted value of W0 is 0.42 m. This value compares favorably with the test result, in which W0 is 0.4 m, as indicated in Figures 8-64 and 8-65. In this test, the minimum width of apron Wmin was about one riprap stone width. The large-scale test results also agree with the design recommendation that, for relatively short abut- ments at least, an apron width of 2yf practically eliminates substantial scour in the vicinity of an abutment. However, the tests show that some modest extent of scour will occur around the edge of even a wide apron. Given the differences in geometric scale and layout of abut- ment for the large-scale test, this agreement is a substantial validation of the design relation developed from the small- scale tests presented in Section 8.1. W y C d yf sf f 0 3= ⎛ ⎝⎜ ⎞ ⎠⎟  160