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.
8Details of the data acquisition and the quality control procedures used to identify the final data sets are presented in this chapter. As stated earlier, assessing the accuracy of the scour evolution data required introducing assumptions about the maturity of the scour holes at the end of the tests. Therefore, this data was used as published. Only field data for piers of simple shape (round, square, rectangular, etc.) were included in the field data set. Equilibrium Local Scour Data Relatively large quantities of equilibrium local scour data have been published (943 field and 569 laboratory data points). The sources and quantities of these data are listed in Table 2; however, it includes only data where all the relevant parameters are known and the sediment is cohesionless. Only 15 field data points were excluded due to missing information. Laboratory data are derived from experiments that were carefully performed with all pertinent parameters (flow speed and direction relative to the pier, sediment size and size distribution, scour depths, etc.) given. There are, how- ever, potential scale effects when the laboratory results are used for prototype piers. Flow regimes are usually different between model and prototype resulting in differences in the relative magnitudes of the forces involved. This is particularly true for such complex mechanisms as sediment transport and scour. Field studies have the advantage of little or no scale effects provided the structure size is approximately the size of interest. However, the measurement accuracy for both independent (flow velocity and duration, sediment properties, etc.) and dependent (contraction and local scour depths) quantities is, in general, less than that for laboratory conditions. This is especially true for high-velocity flows where there is signifi- cant suspended sediment in the water column. In some of the reported cases, the substructure shape and dimensions were not known (Zhuravlyov 1978). Another important parameter that is usually missing in field data is the level of maturity of the scour hole at the time of measurement. Therefore, the potential errors associated with field data must be considered when verifying predictive equations. The available laboratory and field equilibrium scour data are presented in matrix plots in Figures 1 through 4. Each of the independent parameters are plotted versus the other independent parameters, for example, a or a* versus V1; y1 versus D50; V1 versus y1, a, D50; etc. The matrix plots are also given in dimensionless form in terms of the ratios a/y1, V1/Vc and a/D50. The histograms in the matrices provide an easy way to illustrate the distribution of existing data and thus where data gaps exist. Figures 1 through 4 provide a qualitative view of the range and distribution of the variables covered in the laboratory and field data sets compiled in this study. The information in the matrices can be interpreted as follows. In Figure 1 the top row in the matrix has four elements. Starting from the left, the first element is a histogram showing the distribution of laboratory data with water depth. The vast majority of the data is for water depths less than 1 ft with only a small number of tests conducted at depths beyond 2 ft. The second element shows the values of water depth and pier size covered by the data. The 3 ft diameter pier tests were performed at water depths ranging from about 1 ft up to 6 ft. There is a data gap for pier diameters between 1 and 3 ft. The third element in the top row shows the values of water depth and flow velocity for the laboratory data. All of the high-velocity (greater than 2 ft/s) tests were performed with water depths less than or equal to 2 ft. The fourth element shows the values of water depth and sediment size. The horizontal axis is logarithmic so the value of â0.5 corresponds to a sediment diameter of 0.32 mm, 0.5 to a diameter of 3.2 mm, etc. The sediment diameters range from 0.1 to 7.0 mm, but only a few tests were conducted at water depths greater than 2 ft. The other elements in Figures 1 through 4 can be interpreted in this manner. C H A P T E R 3 Data Acquisition and Analyses
A significant data gap exists in the available laboratory data for a/D50 > 1,800. At present the largest laboratory value for a/D50 is 4,159. The histogram in the lower right corner of Figure 4 shows the distribution of a/D50 in the field data. This illustrates the range of this parameter for prototype piers. Because the horizontal axis in this plot is logarithmic, the â4â corresponds to a/D50 = 104. A gap in available laboratory data also exists for a/y1 > â¼10. Figure 5 shows the distribution of normalized scour depth, ys/a, and V1/Vc for the field data set. The largest value of ys/a is approximately 1.8. Assessment of Data Quality The quality of the data cannot be assessed by using the descriptions of the experimental conditions mainly because the information in the publications is insufficiently detailed. All of the Yanmaz and Altinbilek (1991) laboratory data were discarded due to the short durations of the tests (from 4 to 6 h), which were significantly less than that required for the develop- ment of equilibrium scour depths. The method developed and used to identify and eliminate outliers in the data sets included the following steps: 1. Place the data in a three-dimensional Euclidean space with coordinates log(a/D50), V1/Vc, and y1/a. 2. Compute the variance of the data in each direction. 9 Data Source Number of Data Points Clear Water Live Bed Field Data Zhuravlyov (1978) 40 147 Froehlich (1988) 17 60 Gao et al. (1993) 119 133 Mueller and Wagner (2005) 171 241 Total field 347 581 Laboratory Data Chabert and Engeldinger (1956) 87 6 Ettema (1976) 19 0 Shen et al. (1969) 2 21 Jain and Fischer (1979) 2 32 Ettema (1980) 90 7 Chiew (1984) 11 90 Chee (1982) 1 36 Yanmaz and Altinbilek (1991) 14 19 Graf (1995) 3 0 Dey et al. (1995) 18 0 Melville (1997) 17 0 Melville and Chiew (1999) 27 0 Ettema et al. (2006) 6 0 Coleman (unpublished, personal communication) 6 0 Jones (unpublished, personal communication) 15 2 Sheppard et al. (2004) 12 2 Sheppard and Miller (2006) 4 20 Total laboratory 334 235 Table 2. Equilibrium local scour data sources and quantity. Note: Many of the photographs, figures, and tables in this report have been converted from color to grayscale for printing. The electronic version of the report (posted on the TRB website at http://www.trb.org/Main/Blurbs/164161.aspx) retains the color versions. Figure 1. Dimensional plots of laboratory local equilibrium scour data.
10 Note: The dashed lines indicate the wide-pier boundary. Figure 2. Plots of normalized laboratory local scour data. Note: The dashed lines indicate the wide-pier boundary. Figure 3. Dimensional plots of field local scour data.
11 Note: The dashed lines indicate the wide-pier boundary. Figure 4. Plots of normalized field local scour data. 0 1 2 3 4 5 6 7 8 0 0.5 1 1.5 2 V1/Vc y s /a * Figure 5. Normalized scour versus normalized velocity for field data. 3. Normalize the data values by the variances [i.e., the value of log(a/D50) for a data point was divided by the variance of the data in the log(a/D50) direction, etc.]. 4. Compute the three-dimensional distance between data points. 5. For each data point, determine its four closest neighbors that are not from the same experimental data set. 6. Determine the variance in measured scour depths for the five data points (scour variance). 7. Determine the variance in the distances from the point in question to its neighboring four points (distance variance). 8. Compute the ratio of scour variance to distance variance. This ratio is referred to here as the âproximity parameter.â 9. Plot the proximity parameter versus experiment number. 10. Establish a cutoff criterion. 11. Identify the data that exceed the cutoff criteria. These data are considered outliers and eliminated from the data sets.
12 The laboratory data were analyzed following the above procedure and the results presented in Figure 6. Many of the data points with a proximity parameter value greater than 2.0 were from one source: Chabert and Engeldinger (1956). These data were removed from the data set, the process repeated, and the proximity parameter for the other points recalculated. The results shown in Figure 7 indicate that removing these data changed the values of the proximity parameter to below 2.0 for most of the remaining data. A cutoff value of 2.0 was selected, which resulted in the removal of two additional data points as shown in Figure 7. Most of the outliers had V1/Vc values less than 0.8 and very large equilibrium scour depths. This methodology could not be used to determine outliers in the field data because it was not known if the measured scour depths were equilibrium values for the specified flow, sediment, and structure conditions. In some cases, where scour depths were measured at the same pier for different flow events, it appeared that the scour hole had not recovered from a pre- vious, more severe, event at the time of the measurement. For this reason, field scour depth measurements can be less or greater than the equilibrium depth for the conditions at the time of the measurement. The Zhuravlyov (1978) report included both laboratory and field data from several sources. Data from one field site (denoted as the Amu Darya River data) appeared to have unusually large scour values for the reported structure, sedi- ment, and flow conditions. A comparison of these data with data from Froehlich (1988) and Mueller and Wagner (2005) with similar values of a/D50 clearly showed an inconsistency Figure 6. Proximity parameter for all laboratory data. 0 100 200 300 400 500 600 0 0.5 1 1.5 2 2.5 3 3.5 4 Experiment Number Pr o x im ity Pa ra m e te r Figure 7. Proximity parameter for laboratory experiments [excluding data reported by Chabert and Engeldinger (1956)]. 0 100 200 300 400 500 600 0 0.5 1 1.5 2 2.5 Experiment Number Pr o x im ity Pa ra m e te r All Discarded
in the measured scour depth values. Data points from the three sources are shown in Figures 8 and 9. Figure 8 shows the similarity in their a/D50 values. Figure 8 shows that the Mueller and Wagner data and Froehlich data have larger values of y1/a than that from the Amu Darya River site and therefore should have greater scour depths. The values of V1/Vc were very large for all of these data so flow duration should not have been a problem. That is, all of the scour depths should have been near equilibrium values. However, as can be seen in Figure 10, the scour depth measurements from the Amu Darya River site are much larger. It is very difficult to make accurate measurements of flow speed and direction and scour depths during a high- velocity flow event, even with todayâs sophisticated instrumen- tation. The Mueller and Wagner data were obtained during 1965 and 1997 (with 69% being obtained after 1988) using state-of-the-art instrumentation and methodology. They also made attempts to distinguish between local and other types of scour (i.e., degradation and contraction) by making meas- urements away from as well as at the piers. There is only one Froehlich data point in the range of variables of the Amu Darya River data and it is in agreement with the Mueller and Wagner data. The dates and methods used to obtain the data in the Zhuravlyov report (including the Amu Darya River data) are not given. However, all of the data in the report were obtained prior to 1978 (most likely in the late 1960s and early 1970s). There is also no information available about the review process for the Zhuravlyov report. Of the conflicting data sets, the Mueller and Wagner (2005) and Froehlich (1988) data were considered to be the more accurate and credible due to the information provided in their reports regarding instrumentation and methods used. For this reason the Amu Darya River data were eliminated from the field data set. 13 Figure 8. Normalized scour depth versus a/D50. 10 4.14 10 4.15 10 4.16 10 4.17 10 4.18 10 4.190 0.5 1 1.5 2 a/D50 y s /a Mueller Zhuravlyov Froehlich Figure 9. Normalized scour depth versus y1/a. 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 y1/a y s /a Mueller Zhuravlyov Froehlich
Local Scour Evolution Data Laboratory Data A relatively large number of laboratory-scale local scour evolution rate experiments, with simple shaped structures, has been conducted and reported in the literature. A total of 195 scour time series data sets was compiled for circular or square piles. A list of contributing researchers with their number of reported clear-water and live-bed experiments is presented in Table 3. Figures 11 and 12 show the distribution of local scour evolution data collected in laboratory experiments. The matrices in these figures show the ranges and distribution of variables covered by the scour evolution data. For example, the third row from the top in Figure 11 shows the values of velocity and three of the other variables for the data. The first plot from the left shows that the data covers a wide range of velocities from 1 ft/s up to 7 ft/s but that with few exceptions the water depths are less than 1 ft. Also, there are no data for high-velocity flows and large depths. The second plot shows that most of the tests were performed with piers less than 1.5 ft in width. The third plot is a histogram showing the distribution of velocities for the tests, the majority being around 1 ft/s. The last plot in this row shows velocities and sediment size for the tests. The sediment sizes are grouped into three ranges: 1 mm and smaller, 3.0 to 3.5 mm, and the remainder in the 5.0 to 5.5 mm range. The higher- velocity tests were conducted with the smaller sediment. A limited number of laboratory experiments have been performed with complex shaped piers (i.e., piers composed of a pile group, pile cap, and a column). Methods have been developed to estimate scour depths at complex piers using the equations developed for single cylindrical piers (Richardson and Davis 2001 and Sheppard and Renna 2005). The effective diameter of a complex pier is obtained using known pier, sediment, and flow values. The effective diameter is the diam- eter of a single circular pile that will experience the same local scour as the complex pier under the same flow and sediment conditions. This approach seems to work well for equilibrium scour depths for the limited data that exist. However, this approach has not been tested for scour evolution rates. Also, it is not known how scour evolution rates for complex piers compare with those for their equivalent circular piers. Field Data The only time-dependent local scour field data obtained in the information and data search portion of this study were that reported by Walker (1995). In that study, the scour hole at a 2.0 ft wide square pile on a bridge over a tidal inlet on the Northwest Florida Coast (East Pass near Destin, Florida) was filled with the surrounding sand and the redevelopment of the scour hole monitored for a period of approximately 10 days. The unsteady (tidal) flow at this site, which exceeded sediment critical velocities at peak flow, reversed direction approxi- mately every 14 h. After 10 days the scour depth was near the original value. The laboratory and field equilibrium scour data compiled during this project are given in Appendix D (available on the NCHRP Report 682 summary web page: www.trb.org/Main/ Blurbs/164161.aspx). 14 Figure 10. Normalized scour depth versus V1 /Vc. 1 2 3 4 5 6 7 8 0 0.5 1 1.5 2 V1/Vc y s /a Mueller Zhuravlyov Froehlich Data Source Number of Data Sets Oliveto and Hager (2002) 80 Rajasegaran (1997) 14 Grimaldi (2005) 3 Melville and Chiew (1999) 21 Sheppard et al. (2004) 14 Sheppard and Miller (2006) 24 Total 156 Table 3. Sources and number of scour evolution data sets.
15 0 2 4 D50 (mm) 2 4 6 V1 (ft/s) 0 2 4 6 0 2 4 y1 (ft) D 50 (m m ) 2 4 6 V 1 (ft/ s ) 0 2 4 6 y 1 (ft) 0 1 2 3 a (ft) 0 1 2 3 a (ft) Figure 11. Plots of the data for local scour evolution rates. 0 5 10 y 1 /a 2 4 6 V 1 /V c 2 3 4 log(a/D50) 2 4 6 2 3 4 V1/Vc lo g(a /D 50 ) 0 5 10 y1/a Note: The dashed lines indicate the wide-pier boundary. Figure 12. Plots of normalized data for local scour evolution rates.
