Failure of Spill-Through Bridge Abutments during Scour: Flume and Field Observations

Technical Note Failure of Spill-Through Bridge Abutments during Scour: Flume and Field Observations Robert Ettema, M.ASCE 1 ; Kam Ng, M.ASCE 2 ; Ram Chakradhar 3 ; Joshua Fuller 4 ; and Edward W. Kempema 5 Abstract: This paper presents early findings from laboratory tests and field observations on the failure of spill-through abutments subject to abutment scour. These findings show that geotechnical and hydraulic processes interact to erode embankment soil during abutment scour, producing lesser scour depths than predicted using leading abutment scour equations. A major failure location is the flow waterline beginning at an abutment s upstream corner where soil is exposed to the highest values of flow velocity and turbulence. Undercutting and toppling of soil blocks occurs sequentially along the face of the spill slope, eroding it back and eventually exposing the abutment column. Further erosion then may breach the embankment. The laboratory findings, based on uniform sand compacted to varying densities and thereby shear strengths, show that soil strength influences scour depth. DOI: 10.1061/(ASCE)HY.143-700.00006. 2015 American Society of Civil Engineers. Introduction Spill-through abutments, a widely used form of bridge abutment, are prone to erode and fail during scour of bridged waterways [e.g., Ettema et al. (2010)]. They structurally comprise a compacted earth-fill embankment and an abutment column, which is enclosed by an earth-fill spill slope, and are usually located on floodplains or gently sloped channel banks as Fig. 1 illustrates. Although abutment scour has been extensively studied, the question of how spill-through abutments actually fail during scour has remained largely unaddressed. Prior studies focus largely on the depth and location of a scour hole formed near the toe of an abutment, but do not relate scour to abutment failure [e.g., Sturm et al. (2011) and Melville and Coleman (2000)]. This paper presents flume observations and data regarding abutment failure and shows that a prevalent failure mode has been largely overlooked: spill slope erosion along the abutments waterline, beginning at the upstream corner. It is also shown that abutment failure involves a combination of hydraulic and geotechnical processes that cause spill slope erosion and abutment scour to be influenced by the shear strength of the compacted earth fill forming the spill slope. Flume Tests Flume tests were conducted at the University of Wyoming using a 1:30 geometric-scale model of a spill-through abutment dimensioned for a standard two-lane road. Observations from the tests 1 Professor, Dept. of Civil and Architectural Engineering, Univ. of Wyoming, Laramie, WY 2071 (corresponding author). E-mail: rettema@ uwyo.edu 2 Assistant Professor, Dept. of Civil and Architectural Engineering, Univ. 3 Graduate Student, Dept. of Civil and Architectural Engineering, Univ. 4 Graduate Student, Dept. of Civil and Architectural Engineering, Univ. 5 Research Scientist, Dept. of Civil and Architectural Engineering, Univ. Note. This manuscript was submitted on June 16, 2014; approved on December, 2014; published online on January 21, 2015. Discussion period open until June 21, 2015; separate discussions must be submitted for individual papers. This technical note is part of the Journal of Hydraulic Engineering,, ISSN 0733-42/06015001(5)/$25.00. were compared with field observations of actual failed spill-through abutments. The scale-model abutment was located on a simulated erodible floodplain formed of compacted sand whose median particle diameter (d 50 ) and coefficient of uniformity (d 60 =d 10 ) were 0.70 mm and 2.7, respectively. All tests involved clear water scour conditions run in an 1.30-m-long by 2.44-m-wide flume, which included a 0.30-m-deep sediment recess upon which the model abutment was formed (Fig. 2). For each test the approach flow depth was set at 0.14 m and the flow adjusted so that the cross-section averaged bed shear stress relative to critical value for bed sediment entrainment was τ=τ c ¼ 0.0, with critical shear stress, τ c, determined using the Shields diagram [e.g., ASCE (2006), Chapter 2]. Two main series of laboratory tests were conducted: observations of spill-through abutment failure during scour and influence of abutment soil strength on scour depth. The tests used three model soils: a compacted uniform sand, a mix of 0% sand with 20% clay, and a weak, naturally occurring clayey sand. The same uniform sand was used for the model abutment soil as for the model floodplain. For the clayey sand, these variables were 0.40 mm and 175, respectively, while the 0% sand 20% clay mix had a distinctly bimodal distribution of particle sizes related to the uniform sand and clay particles used for the mix. Each model abutment was constructed with spill slope and side slope slopes of 1.5 horizontal to 1 vertical, as commonly used for actual abutments. Model soil was placed in layers that were compacted until the full height of the model was attained. The tests on soil strength influence concentrated particularly on uniform sand as the model soil whose as-built strength could be controlled via compaction, at least over a fairly narrow range. Additional exploratory tests on the influence of soil strength involved the two other model soils, although further research is needed to determine ways to better control and measure the shear strength of model soils used in scaled hydraulic models. For example, further investigation is needed on how clay type, particle grading, and moisture content can be manipulated to control model soil strength. Chakradhar (2014) and Fuller (2012) document the tests. Table 1 summarizes data from the tests. The duration of each test corresponded to the period needed for the spill slope to erode. As the erodible abutments eroded, the flow area increased at the abutment axis and scour ceased deepening. Because values of model soil strength vary with the model s length scale (ASCE 2000), the strengths of the model soil needed to be approximately 1/30th of the strengths of soils in geometrically full-scale abutments. Accordingly, the tests required a technique to 06015001-1

prepare, quantify, and control a range of reduced soil strengths for the model abutments. The technique entailed a direct-shear box to relate shear-strength and density values of compacted sand, and then used a needle penetrometer to relate penetration resistance to sand density and shear strength. The representative overall values of density and shear strength of sand compacted to form the model abutments in the flume were estimated using correlation relationships with penetration resistance measurements. For the model abutments formed of permeable compacted sand, the actual shear strength of soil had to be adjusted to take into account porewater pressure associated with the 0.14 m of water depth in the flume. The vertical effective soil stress at the abutment base is σv0 ¼ γh γ w Y Fig. 1. Spill-through bridge abutment on a floodplain (image by authors) ð1þ where H = height of abutment earth fill; Y = average flow depth; γ = unit weight of compacted soil; and γ w = unit weight of water. A small vane shear device was used to measure the shear strength of model abutments formed of clayey sand soil. A scale effect that aided the present tests using sand as model soil was the capillary action of water in sand. Surface tension within water and adhesion of water to sand elevated the water surface within the dry sand approximately 55 mm above the elevation of stationary water in the flume, coinciding with the estimated meniscus rise for water in a medium having 0.70-mm-diameter openings (Batchelor 167). Capillary rise contributed apparent cohesion to the behavior of sand in the zone of capillary rise (practically the full height of model abutment above the waterline), and thereby enabled wet sand exposed to air to stand with a vertical face and fail in blocks. The shear strength of sand in the capillary zone is estimated as τ ¼ σv0 tanðϕþ ¼ ½γz þ γ w ðd zþ tanðϕþ ¼ ðγ γ w Þz tanðϕþ þ Dγ w tanðϕþ Fig. 2. Layout and dimensions of the model abutment for the flume tests; the abutment, with an erodible nose formed of model soil, was built on a sediment recess located in the flume floor ð2þ where D = depth of the capillary zone (i.e., 55 mm); z = depth below the abutment s top; and ϕ = friction angle of the sand. Apparent cohesion altered the as-built strength of the model abutments formed of sand and caused the model abutment to behave as if it had a layer of weakly cohesive soil immediately above the waterline. The model abutments formed of compacted sand and sand-clay mix could therefore visually replicate the main failure processes associated with riverbank erosion. The clayey sand was much less permeable and did not exhibit the same extent of capillary rise during the relatively short duration of a test. Table 1. Summary of Laboratory Data Model spill-slope soil Sheet metal Compacted uniform sand 0% sand 20% clay mix Clayey sand Unit weight of compacted soil (kn=m3 ) Effective shear strength (kpa) Scour depth (cm) Typical toppled block width (cm) Run time (min) 15.31 15.50 15.66 15.3 16.06 16.10 16.17 16.3 16.0 15.72 1.76 1.4 5.30 6.0 6.73 7.76.2.41.67.45 23.4 17.5 2.06 31.25 10.14 2.67 2.1 2.44 2.67 2. 3.26 3.4 3.63 3.04 2.22 6.5 6.37 3 10 3 10 50 7 10 11 14 11 06015001-2

For several of the tests the model abutment was armored with two layers of model riprap stone; the angular stone had an average diameter of 16 mm. To determine a benchmark scour depth associated with a nonerodible model abutment, a test was run with the model abutment clad with a sheet metal and fitted with a protective stone apron around its toe. This test ran until equilibrium scour depth occurred. Flume Observations Scour development and spill slope failure were observed in the flume for two situations: Failure of an unprotected spill slope, a situation that is analogous to erosion of a convex section of exposed river bank; and Failure of a riprap-protected spill slope, reflecting the usual situation for spill-through abutments. The progressive failures of the unprotected and riprap-protected spill slopes are depicted in Figs. 3 and 4, respectively. For both spill slope situations, scour began on the channel bed at the toe of the abutment s upstream corner. As scour developed it began to undercut spill slope soil immediately adjacent to the forming scour depression. Scour also began along the sandy floodplain at a location slightly downstream of the abutment s centerline. Flow field contraction and turbulence cause scour to begin at these locations [e.g., Sturm et al. (2011)]. However, immediately noticeable along the waterline at the upstream corner of each model abutment were swift flow velocities and energetic water-surface oscillation and turbulence associated with flow contraction and separation at the abutment s upstream corner. The unprotected spill slope began visibly eroding at its upstream corner [Fig. 3(a)] as flow hydraulically eroded the model soil and undercut the spill slope at the waterline. What subsequently unfolded was the progressive toppling of blocks of spill slope soil, such that the corner eroded back to form a vertical face along the front of the spill slope [Fig. 3(b)]. A recurring feature of spill slope failure was the appearance of tension cracking above each undercut block of abutment soil, and the action of gravity eventually caused soil blocks to rotate about a point at the water surface where undermining was greatest. Blocks toppling into the channel were quickly eroded by flowing water and exposed additional spill slope soil to undercutting. The observed erosion cycle was essentially the same as often described for eroding riverbanks [e.g., ASCE (2006), Chapter 7]. No circular failure surface was observed for the spill slopes. The cycle of undercutting, toppling, and spill slope erosion continued until the exposed face of the spill slope retreated to form a circular arc at the waterline [Fig. 3(b)]. Arc radius, r, approximately fitted the geometry recommended for rounded orifices and nozzles; e.g., for bell mouth nozzles, Brater and King (176) give r ¼ 1.63d, where d = nozzle diameter (or, in the present case, the waterway surface width across the downstream transect of the bridge). As the spill slope eroded, d increased and the arc flattened in curvature. The model spill slopes formed of sand-clay mixtures failed in much the same way as the spill slopes of uniform sand, except for two additional factors that complicated the tests. For the 0% sand 20% clay mix, flow around the abutment bled clay particles from the failure zone at the spill slope face, reducing model soil strength to essentially that of sand alone. The greater shear strength of the clayey sand model soil resulted in commensurately wider blocks of failed spill slope soil (Table 1 includes approximate measurements of the block widths observed) and toppled blocks did not erode as rapidly as did the sand blocks; consequently, the flow area at the abutment did not increase as rapidly as for the failed weaker abutments formed of sand. The model riprap-protected spill slopes formed of compacted sand began failing when flow velocities and water-surface oscillation cyclically shook riprap stone at the spill slope s upstream corner, causing gaps to open between riprap stones and spill slope soil to be winnowed from between the stones. Fig. 4(a) depicts the observed failure sequence, starting with the failure of riprap at the upstream corner, and resulting in erosion of the spill slope back along the bridge axis. Spill slope failure occurred largely at and above the waterline. As Fig. 4(b) shows, the spill slope remained practically intact below the waterline. Riprap stone eroded from spill slope deposited in sparse concentration over the bed closely downstream of the spill slope. Influence of Soil Strength The effective soil strengths mentioned herein are the values associated with the model abutments surrounded by water immediately prior to beginning a test. Abutment erodibility and soil strength influenced scour depth at the model abutments, although the strength of the model soil was difficult to control and measure immediately at the start of a test. Capillarity effects, along with the bleeding of clay particles from sand-clay mixtures and the different time rates of scour development on the channel bed Fig. 3. Erosion and failure of the compacted sand abutment without riprap protection: (a) erosion beginning along the waterline at the abutment s upstream corner; (b) the eroding spill slope assumed a bell mouth curvature 06015001-3

Fig. 4. Erosion and failure of the compacted sand abutment with protective riprap cover: (a) time series of overhead photos showing erosion progressing along the waterline of the abutment, starting at the abutment s upstream corner; (b) oblique view of the abutment in the drained flume showing erosion along the abutment s waterline (compare this figure to the eroded prototype abutment in Fig. 6) and geotechnical erosion of the model spill slope soil, complicated interpretation of the influence of soil strength. Fig. 5 presents the maximum scour depths versus the effective shear strengths associated with the abutments formed of the model soil. The main results are as follows: The scour depths obtained with the compacted, uniform sand embankments were approximately 25 to 40% of the depth obtained with the nonerodible abutment. For the lower values of shear strength the measured maximum scour depths were practically constant for the uniform sand owing to the difficulty of controlling the compaction of sand forming the spill slope. Only for the most densely compacted and stronger model embankments was the soil strength in the spill slope appreciably greater than that for the weaker model embankments. For the compacted, uniform sand abutments, the tests show a 40% increase in scour depth over an 0% increase in shear strength of the model soil. The data in Fig. 5 infer that additional increase in the strength of embankment soil would result in scour depth increases that trend upwards toward the scour depth obtained with the nonerodible abutment. For the three higher values of shear strength, spill slope erosion by means of undercutting and soil block toppling failure occurred more slowly, thereby causing the flow to scour the channel bed more deeply. Also, it was observed that greater undercutting of the spill slope could occur, resulting in somewhat larger distance from the eroding spill slope face to the tension cracks and thus toppling of larger sand blocks. Flow bled clay particles from the spill slopes formed of the 0% sand and 20% clay mix, and essentially reduced soil strength to that for the compacted sand. The resulting scour depths therefore were in the same range as those for the model abutments formed of compacted uniform sand. The scour depths obtained with abutments formed of the strongest model soil, the clayey sand, approached 65 to approximately 0% of the depth obtained with the erosion-resistant abutment, and they were larger than obtained with the abutments formed of weaker model soil. Field Observations Many field observations of failed abutments reflect the failure mode observed in these flume tests. Additionally, the reported scour 06015001-4

12 10 6 4 2 0 0 5 10 15 20 25 30 35 Fig. 5. Flume data on maximum scour depth versus effective shear strength of model soil at the beginning of an experiment; the dashed line indicates the scour depth for the nonerodible abutment formed of sheet metal; rapid erosion of clay from the 0% sand 20% clay mix caused this model soil to behave like the uniform sand Fig. 6. Field example of spill slope failure at a spill-through abutment at Cottonwood Creek, Wyoming; the riprap failed, causing the abutment s compacted earth fill to erode by undercutting and toppling and expose the abutment column; the failure appears similar to that shown in Fig. 4(b) (image courtesy of William Bailey, Wyoming Department of Transportation) depths often are relatively modest, as are those observed in the flume tests. A brief qualitative comparison of these flume and field observations is offered here. Fig. 6 depicts a representative example of a failed spill-through abutment, where the riprap stone failed at the waterline, causing the spill slope to begin eroding along the waterline and continue eroding back toward the abutment column; the riprap was notably large and resisted entrainment by water flow. The abutment was left with a vertical back face at the abutment column. There was little evidence of substantial scour in the channel adjoining the abutment. Conclusions Flume tests and field observations regarding the failure of spillthrough bridge abutments during scour are reported. The observations show how geotechnical and hydraulic processes interact to scour the channel or floodplain bed on which the abutment is situated and to erode the abutment s spill slope. Hydraulic erosion of the spill slope begins along the waterline at the spill slope s upstream corner and progressively undercuts spill slope soil, causing blocks of spill slope soil to topple. Soil strength affects the rate of undercutting and subsequent erosion of toppled blocks of soil. This failure process is similar to that often observed for riverbank erosion. More research is needed concerning several geotechnical and hydraulic aspects of abutment failure during scour, including research on countermeasures for protecting the upstream corner of spill slopes and on laboratory and hydraulic tests involving combined geotechnical and hydraulic processes such as bridge abutment failure. References ASCE. (2000). Hydraulic modeling: Concepts and practice, Manual of Practice 7, Reston, VA. ASCE. (2006). Sedimentation engineering: Processes, measurements, modeling and practice, Manual of Practice 110, Reston, VA. Batchelor, G. K. (167). An introduction to fluid dynamics, Cambridge University Press, Cambridge, U.K. Brater, E. F., and King, H. W. (176). Handbook of hydraulics, 6th Ed., McGraw-Hill, New York, 4 20. Chakradhar, R. (2014). Laboratory investigation of geotechnical and hydraulic influences during abutment scour. M.S. thesis, Univ. of Wyoming, Laramie, WY. Ettema, R., Nakato, T., and Muste, M. (2010). Estimation of scour depth at bridge abutments. National Cooperative Highway Research Program, Final Rep. for NCHRP Project 24-20, Transportation Research Board, Washington, DC. Fuller, J. (2012). Observations of scour at spill-through abutments with erodible spillslopes. M.S. thesis, Univ. of Wyoming, Laramie, WY. Melville, B. W., and Coleman, S. E. (2000). Bridge scour, Water Resources Publications, Littleton, CO. Sturm, T. W., Ettema, R., and Melville, B. W. (2011). Evaluation of bridge scour research: Abutment and contraction scour processes and predictions. National Cooperative Highway Research Program, Final Rep. for NCHRP Project 24-27(02), Transportation Research Board, Washington, DC. 06015001-5