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.
232 CHAPTER 6. Conclusions and Recommendations 6.1 Overview The objective of this research project is to determine the relationships among individual scour components observed in the same flow event at a bridge, and determine how to combine them to produce realistic estimates of total scour depth for safe and economical design of bridge foundations. The scour components of interest are lateral contraction scour, abutment scour, vertical contraction or pressure scour, and local pier scour. Contraction scour results from acceleration of the flow caused by the constriction of the bridge opening, either due to the bridge embankment narrowing the channel width or to the water level exceeding the height of the bridge forcing submergence and orifice flow. Local abutment and pier scour are the result of several different types of vortex action caused by flow obstruction by the bridge and the turbulence that is generated. Although these scour processes occur simultaneously and interact under some conditions, current scour depth prediction methods for the design of bridge foundations treat them as independent without interaction by adding them together as a conservative assumption. The degree to which current scour design practices are overly conservative is unknown. For this reason, investigation of the degree of interaction of simultaneous scour processes was undertaken in this study. The research approach was a combination of experimental and numerical techniques at a more advanced and comprehensive level than has been attempted in most previous bridge scour research projects that focused on only one component of scour under idealized flow conditions. The experimental study was designed to be as realistic and comprehensive as possible. Three different flumes were utilized to study the effects of model scale, clear-water vs. live-bed scour, compound channel geometry, abutment length, flow types from free to submerged orifice to overtopping cases, pier shape and configuration, and abutment type. The bridge embankment slopes were protected by rock riprap, and riprap aprons were installed around the abutment toe. For each scour experiment, detailed distributions of velocity and turbulence were measured in the approach flow and all the way through the bridge opening using a fixed bed at the beginning of scour. Then the movable bed was installed in both the floodplain and main channel, and the experiments were run for three to five days as required until scour equilibrium was reached, after which the bathymetry of the scour hole was carefully measured. A suite of computational fluid dynamics (CFD) models was applied to the experimental setup at Georgia Tech and to a field example. The CFD models included large-eddy simulation (LES), three-dimensional, Reynolds-Averaged Navier-Stokes equations (3D RANS), and 2D RANS. The purpose of the CFD models was to gain insight into the coupling between turbulence structures and the scour interaction processes, to augment the experimental observations, and to explore options for estimating the predictor parameters to be used in the proposed scour prediction formulas.
233 6.2 Conclusions The aims of this investigation were satisfied successfully by dividing scour interactions into four categories and developing predictive equations for each type of interaction as shown previously in Table 4-7. The categories are related to the abutment length (long setback abutment (LSA), short setback abutment (SSA), and bankline abutment (BLA)) subject to free (F), submerged orifice (SO) and overtopping (OT) flows. The categories are given by: I. Abutment/Lateral Contraction Scour with or without Vertical Contraction Scour for LSA: combination of local scour around the abutment induced by the turbulent structure of the flow and acceleration of the flow resulting from the width constriction offered by the bridge opening for free flow around LSA, with or without the addition of vertical contraction scour in SO and OT flows; II. Abutment/Lateral Contraction Scour with or without Vertical Contraction Scour for SSA/BLA: same combinations as Category I, but for SSA and BLA; III. Abutment/Lateral Contraction Scour for LSA with pier scour in F, SO, and OT flow; IV. Vertical Contraction Scour and Pier Scour in the floodplain outside the zone of influence of the abutment. Based on information extracted from the LES on flow structure and a dimensional analysis, it was found that combinations of lateral contraction scour (L), abutment scour (A), and vertical contraction scour (V) for both Category I and Category II could be formulated in terms of the flow intensity, defined as the ratio of approach flow velocity to its critical value for incipient sediment motion (V1/Vc), the backwater ratio given by the ratio of the approach flow depth to the unconstricted value in the bridge section (Y1/Yo), and the unit discharge ratio defined as the mean discharge per unit width in the bridge section to that in the approach flow section (q2/q1). It was shown that these three dimensionless variables also appear in the theoretical equation for idealized long contraction scour. Previously, it has been suggested that the total scour depth can be calculated as an amplification factor multiplied times the theoretical contraction scour depth, but in this study, it was found that a more effective approach was to use a similar mathematical form of the idealized equation but with the parameters determined by regression analysis of the very large data set developed in this study. The result was a separate best-fit regression equation for Category I and Category II scour depths, but the prediction equations are both dependent on the same variables and have the same mathematical form as shown previously in Table 4-7. The validity of the scour prediction equations was confirmed through comparisons with experimental results of other investigators and with a few extreme field events. The clear-water scour (CWS) results from the Georgia Tech and University of Auckland flumes collapsed into the same prediction equation for Category I even though they were obtained from experiments conducted with a different compound channel geometry and at a different scale. This finding confirmed the value of conducting clear-water scour experiments in two very different flumes as
234 intended in the research plan. A separate equation for Category II CWS scour prediction was developed, but it was of the same form as the Category I equation with the backwater ratio neglected because of the larger depths in the main channel. The live-bed scour (LBS) experiments at Auckland were all Category II and were dominated by contraction effects and dynamic bed forms which made them more difficult to conduct. Nevertheless, the same Category II equation developed for SSA and BLA in CWS was found to be suitable for the LBS results as well with the flow intensity, Vm1/Vmc1, set equal to 1.0. In many of the Category II CWS experiments, sediment conditions in the main channel approached incipient live-bed scour which might explain similar results for Category II LBS and CWS experiments. All of the pier scour interaction experiments were conducted at Georgia Tech according to plan so there were no LBS results for Category III and Category IV. Category III scour is identified by interactions of Category I scour processes with pier scour. This interaction is two-way in terms of the influence of pier scour on abutment/contraction scour and vice versa. Pier location was varied from a position very close to the toe of the abutment to a position in the floodplain well out of the zone of influence of the abutment and contraction scour processes. Both dual rectangular column piers and wall piers were studied with no appreciable difference in the results for maximum scour depth. Partly due to the influence of some portions of the abutment riprap blanket rolling into the scour hole in the early phases of scour, the deepest point of abutment/contraction scour tended to move near the downstream face or just outside the bridge. As a result, the pier scour had little effect on the abutment/contraction scour depth. On the other hand, defining pier scour as the maximum scour depth just upstream of the pier, it was found that pier scour was affected by abutment/contraction scour, depending on pier position. The Category III interaction of abutment on pier was formulated as an excess pier scour caused by the abutment added to the calculated isolated pier scour under the same flow conditions. The excess pier scour was found to depend on the same variables as Category I abutment/embankment scour and on the distance of the pier from the abutment. The ratio of excess pier scour to abutment/contraction scour is applied to the Category I scour equation estimate with the ratio determined by the distance of the pier from the abutment. The excess pier scour depth is calculated and then added to the scour depth from the isolated pier scour formula to obtain the combined scour at the pier. The maximum abutment/contraction scour depth is not influenced by the pier regardless of its lateral position, so no pier scour is added to it. Because the main channel in the experiments was not wide enough for a systematic series of pier positions, the case of pier interaction in the main channel in Category II scour utilizes the same excess pier scour ratios as for the long setback abutment but applied to the Category II abutment/contraction scour equation. Additional experimental data for a wider main channel in the compound channel geometry would be helpful to verify this procedure in the main channel. Category IV scour occurred as an interaction between pier scour and vertical contraction scour, in either overtopping or submerged orifice flows. In this particular category, the two scour processes were somewhat independent, at least in a first order analysis. The vertical contraction scour was calculated in terms of the ratio of velocity under the bridge to its critical value in
235 contrast to using the approach flow velocity ratio and the relative vertical geometric contraction ratio. Some previous investigators have also formulated vertical contraction scour alone in terms of flow intensity under the bridge. This method agreed with isolated vertical contraction scour data collected in this study. The supposition is that velocity under the bridge is a more direct vertical contraction scour predictor. On the other hand, pier scour in the presence of vertical contraction scour depends primarily on pier width and approach flow velocity immediately upstream of the bridge. Under these circumstances, it was found that the current procedure of adding scour components did not result in excessive overprediction. In this approach, the Lyn (2008) equation was used for prediction of pressure scour, and either the pier scour prediction equation by Sheppard and Melville or the CSU equation can be used. In Category I and Category II, the LES results further confirmed the interaction among turbulent flow structures associated with flow separation around the abutment and formation of an intense shear zone, a large recirculation area downstream of the abutment coupled to the shear zone, regions of increased velocity under the bridge in submerged orifice flow, and vertical circulations downstream of the abutment in overtopping flow. Each of these flow features could be identified by values of elevated turbulent kinetic energy (TKE) and increased bed shear stress extracted from the LES. Furthermore, areas of increased TKE and bed shear stress coincided with scour contours from the experimental results for local scour around the abutments while increased bed shear stress was more prominent for overtopping and submerged orifice flows which caused vertical contraction scour across the floodplain. Although this raises the possibility of a direct physical relationship between scour depth and TKE, it was found that at first order, the discharge contraction ratio could serve a similar purpose in simpler scour prediction formulas. The 3D RANS results were less satisfying than LES because of significant overprediction of the size of the recirculation zone in the floodplains just downstream of the bridge, which appears to be very important in the scour process based on the LES results. Overprediction of recirculation areas is a known problem of RANS-based simulations. The 3D RANS also significantly overestimated the turbulent kinetic energy at one abutment and underestimated it for the other. While it may be possible to improve this result with a different, calibrated turbulence submodel, matching the good performance of the LES model seems unlikely. On the other hand, the LES captured the highly three-dimensional zone downstream of the left abutment for submerged orifice and overtopping flows very well. It is suggested that further LES work is warranted to continue exploration of the turbulence structures responsible for bridge scour. Application of the total scour prediction equations suggested for the four scour categories requires estimates of the unit discharge ratio, flow intensity, and bridge backwater ratio. Although the 3D RANS model predicted these parameters well in the laboratory results, it did not perform as well in predicting the turbulence properties. As a practical matter, the 2D model results were just as effective in predicting integral scour parameters associated with the mean flow velocity field such as q2/q1 needed in the proposed scour prediction equations. Under these circumstances, a 2D RANS model can be used to predict the required scour parameters so that the more complicated 3D RANS model is unnecessary with the caveat that it cannot reproduce
236 secondary currents that might be generated in a bend upstream of a bridge. Furthermore, the 2D RANS model was demonstrated to perform well when applied to a field example at prototype scale. The 1D HEC-RAS model did not perform as well for this field example because of the inherent shortcomings of determining velocity distribution based on calculated conveyance ratios in 2D and 3D flow regions, especially near the abutments and the piers. A detailed application procedure with a flow chart and a comprehensive example are given in Chapter 5. Estimated scour depths using current methods are excessively large at the abutments in comparison with the proposed equations. A more useful comparison is given in Table 4-7 which shows statistically the performance of existing methods and proposed equations using experimental and field data. In direct answer to the question posed by the study objective, it has been shown that scour interactions produce less scour than predicted by current methods of adding individual scour components in Categories I, II, and III. Overestimation of maximum scour depth ranged from 20% to 45% at the 95% level of confidence for these three categories. On the other hand, approximately 80% of maximum scour depth estimates using equations developed in this study fell within ±10% of the line of perfect agreement. Using current methods, zero to 15% of the data points fell within ±10% of the line of perfect agreement. For Category IV scour, 100% of the data points fell within ±20% of the line of perfect agreement using the equations suggested herein. These results suggest that current scour prediction methods do overestimate total scour by a significant amount, and that the relatively straightforward prediction formulas reported herein can rectify this situation. Even though this study has been conducted to apply to a wide variety of flow types and channel and bridge geometries, wingwall and spill-through abutments, twin rectangular column piers, Clear-water scour, and live-bed scour, it is limited to erodible embankments protected by riprap with riprap aprons... The riprap protection is not so much a limitation as it is a condition of similar, consistent end points for all of the experiments without catastrophic failure of the abutment by exposure of the abutment stub with the bridge deck dropping into the stream. The implied design philosophy is one of protection against total failure rather than designing for a failure condition. 6.3 Recommendations The following recommendations are made based on the findings of this study: ï· Implement the scour interaction equations proposed herein for estimating maximum depth of scour for four separate categories of interactions; ï· Investigate additional properties of turbulent flow features using LES, especially for overtopping flows, and their quantitative connection with scour depths; ï· Add a sediment transport model to the LES model to obtain direct estimates of scour depth; ï· Adopt the Bureau of Reclamation model SRH-2D or a similar 2D model using k-epsilon
237 turbulence closure and including submerged orifice and overtopping flows as well as free flows for estimating scour prediction parameters; ï· Extend current riprap protection design procedures for embankments and riprap aprons to include more energetic submerged orifice and overtopping flows, especially in live-bed scour; ï· Commit to a long-term scour monitoring program at bridges to obtain a more robust set of field data for model validation at a wide variety of bridges.
