Study of Sediment Deposits for Reservoir Dredging Tiao J. Chang1, Travis D. Bayes2, Scott McKeever3 Abstract Many reservoirs in the country are aging, and dredging has been the most common means to maintain continuing uses of these reservoirs. By investigating geographic distribution of sediment deposits in a reservoir, this study was to gather information for developing the guidance for future dredging programs for other similar reservoirs. Based on the recent topographical survey and the original topographical map of the studied lake, sixty sampling locations were selected to geographically represent the lake. A gravity corer, two feet long and three inches in diameter, was used to collect sediment samples. These samples were first classified by color and then the general physical characteristics of the sediment deposits were recorded in a field book. Each sample of sediment deposit was then bagged and marked with its geographical location using a global positioning system and physical landmarks in conjunction with a detailed map. Samples were dried and analyzed by mechanical sieve analysis to develop a particle-size distribution curve for all sampling locations. Grain sizes of sediment deposits at sampling locations, assumed to be geographically referenced variables, were spatially interpolated to form a regional distribution. The analysis of sediment grain sizes and texture classification resulted in grids of raster-based values that can be expressed by spatial images. The analyzed results were used to develop the guidance for future dredging programs. Introduction Particles in water such as sediment deposits settle if the densities of particles are greater than that of water. The motion of this settling process is essentially involved with three forces, namely, the particle weight, W, the buoyancy force, Fb, and the drag force, Fd. W W is a downward force while Fb and Fd are downward forces. When the particle reaches a constant velocity, U, in the falling process, the net force is equal to zero, i.e., W Fb Fd 0, (1) Let w be the density of water, D be the diameter of the falling particle, and µ be the dynamic viscosity of water, then the Reynolds number, Re, can be expressed as UD (2) Re w , Based on the Stokes law, if the Reynolds number is less than one, the drag force, Fd, can be estimated by (Streeter and Wylie, 1985) Fd ( 1 24 wU 2 D 2 )( )( ) 3UD, Re 2 4 (3) Professor, Civil Engineering Department, Ohio University, Athens, OH 45701 Research Assistant, Civil Engineering Department, Ohio University, Athens, OH 45701 3 Manager of Special Projects, Muskingum Watershed Conservancy District, New Philadelphia, OH 44663 2 where the constant velocity, U, is called the settling velocity. Let s be the density of the falling particle, then Equ. 1 can be rewritten as 4 3 D 2 4 3 D 2 s ( ) ( )3 w ( ) ( )3 3UD 0, (4) where g stands for the gravitational acceleration. Equ. 4 can be rearranged to result in an expression for the settling velocity as follows: U ( s w ) gD , 18 (5) Based on Equ. 5, the settlement of a sediment particle depends mainly upon the density and diameter of the particle in a given water body. Sediment particles may not be spherical, but Stokes’ law can still give a good representation of the settling velocity. In general, the diameter of the particle carries the most weight in determining the settling velocity. The settling velocity of a particle increases linearly with an increase of particle diameter. This has an implication on the sizes of sediment deposits settling at the bottom of a reservoir or river. The settling velocity expressed in Equ. 5 is based on the calm water body. Turbulence of the flow may also affect the particle deposit. If the average velocity of the current is constant and fluctuations in the current’s turbulence are small, the sediment will be well graded (Trash, 1950). Soil particles that are small enough to stay suspended will do so, while the larger particles fall out. On the other hand, when the current changes in velocity or turbulence, the large particles have deposited earlier, so the range of particle size is small. Therefore, in an ideal reservoir, sediment deposits can be predictable; the coarser soil particles will be deposited near the entrance stream and the finest material will settle near the dam. The flow velocity decreases and results in finer material deposits as it gets closer to the dam. The depth of the sediment would generally be the greatest near the dam as shown in a schematic diagram in Figure 1 (Annandale, 1987). Figure 1. Schematic diagram for reservoir sedimentation Background The studied reservoir, Charles Mill Lake, is on the edge of the glaciated region of Ohio. The reservoir has a shallow slope while it resides in several valleys so that it has an irregular shape as shown in Figure 2. Because of this, the amount of sediment deposits and the size of the sediment are not easily predicted. Reports have been made that the sediment deposition in the lake is gradually filling the lake. Residents in the surrounding areas and users of the lake have complained about poor navigability. In addition, the reservoir has reduced its effectiveness in controlling floods. Charles Mill Lake is located four miles east of Mansfield, Ohio in Richland and Ashland counties. It receives its inflow from the Black Fork Creek, located on the north of the lake. The water from the lake ultimately discharges to the Ohio River via the Muskingum River. The dam holding Charles Mill Lake was constructed in 1935 for the purpose of flood control. The Muskingum Watershed Conservancy District (MWCD) owns the lake and surrounding land and is responsible for the conservation management and recreational activities. The dam is owned and operated by the U.S. Army Corps of Engineers. The man-made lake contains three natural lakes, Mifflin, Bell, and Mud, which existed before the dam was built. It also includes fourteen islands. The average depth of the lake has decreased by about one foot in the last 60 years. It means a decrease of about 20% of the volume of the lake. The watershed draining into Charles Mill Lake is 217 mi2. Table 1 lists facts about Charles Mill Lake and its watershed. Table 1. Facts of Charles Mill Lake and its associated watershed Lake Length 20,700 feet 6,300 meters Lake Breadth 6,200 feet 1,900 meters Original Average Depth 5 feet 1.5 meters Current Average Depth 4 feet 1.2 meters Maximum Depth 34 feet 10.4 meters Original Volume 11,369 acre-feet 14,034,230 km3 Current Volume 8,129 acre-feet 10,034,678 km3 Water Surface Area (normal pool) 1,339.5 acres 5.42 km2 Shoreline Length 34 miles 53.5 km Lake Elevation (normal pool) 997.1 feet (MSL) 304.0 meters Lake Elevation (spill way) 1020.0 feet (MSL) 311.0 meters Watershed Area 217 mile2 562 km2 Methodology Digitized topography and hydrography of the studied area were downloaded from the Geographic Information System Support Center, Ohio Department of Administrative Services and converted for use in ArcView GIS. This data were 1:24,000 scale Digital Line Graphs. An original topographic map of the land, before the dam was built, was obtained and digitized. Once the map was digitized, the elevation point data were interpolated over the whole lake to create a continuous grid and a raster image as shown in Figure 3a. The recent topographic survey of Charles Mill Lake was obtained from the Army Corps of Engineers (ACOE). The ACOE surveyed the lake in 1998 with cooperation with the MWCD using a Global Positioning System (GPS) and sounding equipment. These data points were also interpolated to create a raster image as shown in Figure 3b. The analysis of these two topographic images indicates that most of the sediment deposits has been located in the main channel, especially in the north section of the lake. Based on this result, the sampling locations were chosen to best represent the lake and sediment deposits as shown in Figures 4. The sampling was taken from each sample location in July of 1998. The samples were retrieved with a gravity corer that could collect a core two feet long and three inches in diameter. Each sample was described in the field book, and the location was taken down on the map and with a GPS. The description included the sample number, color, general soil classification, location, length of core, and time taken. The sample was bagged and marked with identification. Moisture content testing and mechanical sieve analysis was conducted for each sample. The results were input into the associated coordinates along with the sample’s general description. The data was further analyzed by a spline interpolation method to obtain a spatial distribution (Mitus and Mitasova, 1988). The particle-size distribution curve was developed for all samples. From the curves the D10, D30, D50, D60, and D90 values were interpolated and used for data points at the respective sampling locations. The D10 is the particle diameter that corresponds to 10% finer on the particle-size distribution curve. In other words, 90% of the particles, by weight, are larger than the D10 value. The particle-size distribution curves were further used to determine the percentages of four-classified materials, namely, gravel, coarse sand, fine sand, and silt/clay. It is noted that the total of these four percentages should be equal to one. The size of the soil in each class is listed in Table 2. Table 2. Soil classification by particle sizes Soil Classification Gravel Coarse Sand Fine Sand Silt/Clay Minimum Size (>) 0.0787 in 0.0157 in 0.00295 0.00000 in Maximum Size () 0.0787 in 0.0157 in 0.00295 in The spline interpolation imposes that the surface must pass exactly through the data points. In addition, the cumulative sum of the squares of the second derivative terms of the surface, taken over each point on the surface, must be a minimum. The spline method fits a mathematical function to a specified number of points, while passing through the sample points. Since it can be assumed that the sediment deposit data are continuous and only moderate variations occur between the data points, this method worked well to smoothly interpolate sampled data. Results Data of the four-classified sediment deposits are geographically referenced variables. The interpolation using the spline method was conducted for each soil class resulting in a set of grid values that can be expressed as an image. Figure 5 is an example of gravel image that shows the geographical distribution of gravel deposit in the reservoir. It can be seen that two locations in the main channel have the high percentage of gravel deposit. One of the locations is the downstream of a highway bridge and likely has a high velocity. It is noted that the percentage addition of the four classes should is equal to one at any given location based on the defined classification. The spline method resulted in 5% error for only 2.6% of the grid analyzed; only 0.4% of the grid contained more than 10% error. The total sediment deposit in the lake contained only a small amount of gravel. This makes sense because stream has to maintain a rapid current to carry suspended gravel to the reservoir. Since the gradient of the stream is small, the flow is moderate. The total sediment deposit contains an average of 4.4% gravel. The maximum percentage of gravel is about 50% near the bridge and in the southwest side of the lake as shown in Figure 5. The bridge constricts the only connection from the north section to the south section of the lake and causes the downstream scouring. The smaller particles are eroded while the larger particles remain. The average D90 for the sediment is 0.070 inches, where the D90 for downstream of the bridge is 0.30 inches as seen in Figure 6. On the other hand, the average D10, or effective size, for the sediment deposits is 0.0046 inches. The smallest effective sized sediment deposits are located near the entrance of the Black Fork shown in Figure 7. This may be due to the velocity decrease before the flow entering the lake. Once the streamflow enters the lake, the water becomes calm because of the constriction. The effective size increases after the bridge constriction in the downstream end. The percent of silt and clay is the greatest near the entrance of the Black Fork, around 20% as shown in Figure 8. However, an unusual result also sown in Figure 8 is the percentage of silt and clay decreases nears the dam. This is against the general rule as explained in the schematic diagram of Figure 1. The uniformity coefficient (Cu) was used as a factor to further exam the composition of sampled sediment deposits (Braja, 1994). It is defined as D (6) C u 60 , D10 where D60 and D10 are the diameters of particles obtained form the particle-size distribution curve for 60% and 10%, respectively. The sediment deposit is usually gravely if the uniformity coefficient is greater than four; it is sandy if the uniformity coefficient is greater than six. The main sandy area is in the middle of the southern section of the lake as shown in Figure 9. Most of the sediment qualifies as non-uniform since its uniformity coefficient is less than four. The coefficient of gradation (Cc) is used to determine how varied particles of sediment deposits are graded (Braja, 1994). It is defined as 2 D30 Cc , (7) D60 * D10 where D30, D60, and D10 are the diameters of particles obtained form the particle-size distribution curve for 30%, 60%, and 10%, respectively. When the coefficient of gradation is between one and three, the sediment deposits are considered well graded. Figure 10 shows the area upstream of the highway bridge is not well graded, but the area downstream of the bridge is well graded. Figure 11 is the depth of the sediment deposited in the lake based on the differences of ground elevation between 1935 and 1998. The original channel in the southern section has the greatest amount of sediment. Because of the sedimentation, the bottom of the lake seems to have become smoother and flatter. The sediment deposits seem to be more concentrated along the originally mainstem of the river and in just upstream and downstream of a highway bridge. Conclusions It can be concluded that the major sediment deposits are along the original mainstem of the river. There is a minimum percentage of gravel in the composition of sediment deposits, and the settlement of gravel deposits mainly occur at two apparent locations as graphically shown in the analyzed result. Further, the images of uniformity and gradation provide also the geographical distribution of sediment deposits. Based on these results, a working program for dredging can be developed in terms of priorities. Appendix - References Annandale, G. W., 1987. Reservoir Sedimentation, Elsevier Science Publishers Company, New York. Braja, M. D., 1994. Principles of Geotechnical Engineering, Third Ed., PWS Publishing Company, Boston, MA. Mitas, L. and Mitasova, H., 1988. "General Variational Approach to the Interpolation Problem," Computers & Mathematics with Applications, Vol. 16, No. 12, pp. 983992. Streeter, V. L.. and Wylie, E. B., 1985. Fluid Mechanics, Eighth Ed., McGraw-Hill, New York. Trash, P. D., 1950. Applied Sedimentation, John Wiley & Sons, New York.