DC WRRC Report No. 178 AN EXPERIMENTAL STUDY OF THE OPTIMAL THICKNESS OF A SAND LAYER IN A SAND FILTER WATER QUALITY STRUCTURE July 1994 D.C. Water Resources Research Center University of the District of Columbia 4200 Connecticut Ave, NW Building 50, MB 5004 Washington, DC 20008
AN EXPERIMENTAL STUDY OF THE OPTIMAL THICKNESS OF A SAND LAYER IN A SAND FILTER WATER QUALITY STRUCTURE Submitted by: Farshad Amini Fred Chang and Jutta Schneider
DC WRRC REPORT NO. 153 AN EXPERIMENTAL STUDY OF THE OPTIMAL THICKNESS OF A SAND LAYER IN A SAND FILTER WATER QUALITY STRUCTURE July 1994 D.C. Water Resources Research Center University of the District of Columbia 4200 Connecticut Ave, NW Building 50, MB 5004 Washington, DC 20008
TABLE OF CONTENTS 1 INTRODUCTION.... 1 2 OBJECTIVES...... 2 3 SCALE MODEL AND TESTING PROCEDURE... 2 4 RESULTS.... 4 5 CONCLUSIONS...... 11 6 REFERENCES...... 11 APPENDIX I. GRAIN SIZE DISTRIBUTION FOR THE THREE SANDS TESTED APPENDIX II. DETAIL OF LABORATORY TESTS
AN EXPERIMENTAL STUDY OF THE OPTIMAL THICKNESS OF A SAND LAYER IN A SAND FILTER WATER QUALITY STRUCTURE 1.- INTRODUCTION Conventional infiltration devices are often used for water quality control of runoff in an urban environment. These types of best management practices (BMPs) adversely impact groundwater quality (e.g. EPA, 1983, Nightingale, 1987). In addition, these conventional BMPs may not be feasible in ultra.-urban environments because of the large land areas required for their installation. In an effort to address these limitations, in the District of Columbia, an alternative solution, i.e. underground confined Sand Filter Water Quality (SFWQ) Structure has been considered (Truong et al., 1993). A sand filter can be used to remove fine particles in water. While the water moves through the filter media, the fine particles in the water are trapped in the voids of the filter media. This process continues until the size of the voids has been reduced considerably. The water ceases to flow when all the porous spaces have been completely filled. When the flow rate reduces to a certain limit, the filter media needs to be replaced or the deposited particles needs to be removed from the filter media. Sediment particles trapped in sand filters may be cleaned by back-washing with clean water. For a sand filter in the field, back-washing is impractical because clean water is hardly accessible. As a result, the filter media of a sand filter needs to be replaced when the flow rate reaches its lower limit. This device may be used in the following areas: a. Surface parking lots; b. Parking apron, taxiway and runway shoulders at airports; c. Underground parking lots or multi-level garages; d. Emergency stopping and parking lanes, and sidewalks; e. Vehicle maintenances area; f. On street parking aprons in residential area; g. Recreational vehicle camping area parking pads; h. Private roads, easement service roads, and fire lanes; i. Industrial storage yards and loading zones; j. Driveways for residential and light commercial use; k. Office complexes. This research was funded by a grant from the U.S. Geological Survey through the DC WRRC of the University of the District of Columbia. The research was performed in the Soil Mechanics Laboratory of the Civil Engineering Department at the University of the District of Columbia.
2. OBJECTIVES The grain size of the filter needs to be chosen such that the sand filter is as effective as possible. If the grain size is too small, incoming particles can be trapped easily, but the effective zone will be limited to the uppermost layer of the filter medium. If the grain size is too large, no incoming particles will be trapped, but will instead flow through the filter medium. As a result, the grain size of the filter medium needs to be selected such that the whole thickness of the filter medium is used effectively. The first objective of this study was to investigate the effect of particle size on the effectiveness and efficiency of sand filter, with the ultimate goal of establishing the optimal grain size for the sand filters to be installed in the District of Columbia. In addition, though a thicker sand layer has more space to trap particles, the particles are not trapped evenly throughout the layer. In fact, the distribution of particles is highly skewed. The concentration is very high at the upstream, or inflow, side of the sand layer, and drops rapidly. The second objective was to determine the relationship between the thickness of a sand filter and its effectiveness in removing fine particles suspended in water. In this case, the ultimate goal is to establish the optimal sand bed thickness of a sand filter to be installed in the District of Columbia. To accomplish these objective, laboratory tests were conducted using a scale model of a sand filter. Three types of sands, and sand thicknesses ranging from 6 inches to 12 inches were utilized. It may be noted that the research plan does not consider the effect of the size of the incoming particle. A study by Sherard et al. (1984) indicated that uniform sand filters will catch particles with diameter of about 0.11D 15 when the particles are carried in suspension in seeping water. 3. SCALE MODEL AND TESTING PROCEDURE The sand filter model, as shown in Figure 1, was built in the Soil Mechanics Laboratory at the University of the District of Columbia. The model is built of plexiglass, and it consists of a sand filter, a 2 inch thick gravel, and a filter fabric (geotextile) to separate sand from gravel. A 1 1/2 inch perforated pipe in a gravel bed is used to discharge and direct the flow to the outflow pipe. For this experiment, three types of sands, namely fine, medium, and coarse sand were used. The grain size distribution for these three sand types are shown in Appendix I. The coefficient of permeability for the fine, medium and coarse sand was found to be approximately 0.02, 0.10, and 0.2 ft/min, respectively. To study the effect of sand thickness on the efficiency, three sand thicknesses, 6", 9", and 12" sand, were used for each sand type. Water-sediment mixtures were obtained from various storm drain inlet locations in the District of Columbia. After the sediment was added to the inflow tank, it was thoroughly mixed
INFLOW FROM FLOW SEPARATOR OUTFLOW TO STORM
with water. Every thirty minutes, two (2) samples, one 0.5 ft above the filter media to represent the inflow concentration, and the other from the outflow to indicate the concentration of the outflow were taken. Each sample was about 100 ml. Two series of tests for each soil type and each sand thickness were conducted to obtain an average value. Each series were done for about 4 hours. To obtain weight of sediment for both inflow and outflow conditions, filters with a normal pore size of about 1.0 um were used. After the sample was filtered, it was placed in an oven at 110 C. The weight of the sediment was obtained by subtracting the total weight of the filter with sediment after it had been dried, from the weight of the filter. The sediment concentration by weight (ppm) for both inflow and outflow conditions can be computed using the following equation. Sediment Concentration (ppm) = weight of sediment x 10 6 weight of the sample Sediment concentration (ppm) were obtained for each sand type and each sand thickness at 30- minute time interval. After two series of tests were completed, average values of ppm could be determined. A detail of testing procedure is presented in Appendix II. The efficiency of the sand filter was then determined for each test. The efficiency is one minus the ratio outflow/inflow sediment concentration. 4. RESULTS Plots of sediment concentration (ppm) Vs. time for both inflow and outflow conditions for the 12 inch sand filter and for each sand type are shown in Figures 2 through 4. The sediment concentration (ppm) for both inflow and outflow conditions decreased with time for all sand types. The sediment concentration (ppm) values corresponding to the outflow condition was much less than the ones for inflow conditions for all sand types, indicating the effectiveness of the scale model sand filter. A comparison of the effect of sand type on the efficiency of the sand filter is shown in Figure 5. The efficiency of the sand filter system with the medium sand was the highest, indicating a relatively better performance for the medium sand than the other sand types. The efficiency for the sand filter with the fine sand was slightly higher than the one with the coarse sand. The effect of sand thickness on the efficiency for each sand type is shown in Figures 6 through 8. Within the range of variables studied, as filter thickness increased from 6 inches to 12 inches, the efficiency also increased.
inflow & outflow (ppm X 1000) FINE SAND time (hours) Figure 2. Inflow and Outflow Sediment Concentration As a Function of Time for Fine Sand-
Figure 4. Inflow and Outflow Sediment Concentration As a Function of Time for Coarse Sand.
100 Fine Sand Efficiency 90 8 70 60 50 Time (hours) Figure 6. Effect of sand thickness on the efficiency of sand filter for fine sand
5- CONCLUSIONS The results of this preliminary study revealed that the use of the medium sand provided a scale model sand filter with the highest efficiency. The use of the coarse sand was found to be the least effective. Within the range of variables studied, a thicker sand bed resulted in a more effective SFWQ. A more comprehensive study may need to be conducted to include the effect of a variety of factors including uniformity of sand particle size, sand particle shape, incoming sediment particle size, and longer durations. In addition, a comparison of the results between scale model laboratory testing and full scale testing in the field may be beneficial. 6- REFERENCES 1. Results of the Nationwide Urban Runoff Program, Final report, U.S. Environmental Protection agency, Water Planning Division, 1983, 2. Nightingale, H. T., "Water Quality Beneath Urban Runoff Water Management Basins," Water Resources Research, 23(2), 1987, 197-208. 3. Truong, H. V., Burrel, C. R., and Phua, M. S., "Application of Washington, DC Sand Filter for Urban Runoff Control," D.C. Dept. of Consumer and Regulatory affairs, Environmental Regulation Administration, 1993. 4. Sherard, J. L., Lorn, P. D., and Talbot, J. R., "Basic Properties of Sand and Gravel Filters," Journal of Geotechnical engineering, ASCE, 110(6), June 1984, 684-700.
APPENDIX I GRAIN SIZE DISTRIBUTION FOR THE THREE SAND TYPES
Fine Sand Percent finer 0 0.01 0.1 1 10 Grain Size, D (mm) Medium Sand 100 Percent finer 80 60 40 20 0 L ~ L J - J 0.01 0.1 1 10 Grain Size, D (mm)
Coarse Sand 100 Percent finer 80 60 40 20 0 0.01 0.1 1 10 Grain Size, D (mm)
APPENDIX II DETAIL OF LABORATORY TESTS
APPENDIX II Detail of Laboratory Specific Objective The specific objective of this laboratory test is to study the effect of filter sand size and sand thickness on the removal efficiency of suspended solids. Scope This experiment is intended to be a pilot study to be followed at a later date by a more detailed investigation which may be used for developing a maintenance manual. The test covers the following. 1. Steady flow and constant head condition. 2. Three sizes of the sand: fine, median and coarse 3. Suspended sediment collected at one field site. 4. Sediment concentration in and out of the sand filter will be measured. No BOD or chemical properties of the concentration will be included. Test Procedure: 1. Discharge and Head Control The purpose of this part of the test is to establish the discharge and head control. The experiment shall be run with a constant head with the water surface elevation slightly (0.1') below the crest of the overflow weir. The discharge in this condition will be measured. For each filter, this control flow condition needs to be determined. The procedure to determine the discharge is as follows: 1. Introduce water from water supply pipe gradually increase the discharge. 2. When the water surface (ws) is getting near the crest of the overflow weir, slow down the discharge until the ws is 0.1 ft blow the crest. 3. Observe if the water surface is steady. 4. When the water surface is steady, measure the flow rate by obtaining the total weight of the water within a time t. Q = (Wt - Wb) / (Y{) Where Q = discharge in cfs W t = total weight of the bucket with water caught within time t (lb) W b = weight of the bucket, lb V = specific weight of water, 62.41b1fe
t = catching time, sec. 5. Repeat the flow measurements two more times. 6. Mark the position of the faucet valve and record it. 7. Let the fresh water run until the outlet water is clear. 8. Also obtain coefficient of permeability of the sand. II Test for Filter Efficiency This part of the test requires continuous data collections at 30 minute time interval. It involves the following: 1. Clean the bottom of the outflow well. 2. Turn on the water, fill the tank to set up the control flow condition. 3. After the flow becomes steady, the experiment starts. 4. The sediment taken from the field shall be added at a rate such that its concentration is at 1000 ppm by weight. The weight of the sediment input every 10 minutes shall be W S =0.6VQ =37.4Q Where W S = weight of inflow per 10 minutes, lb Q = discharge, cfs 5. Mix the water in inflow well after sediment is added. 6. Every 30 minutes, two (2) samples each from the water 0.5 ft above the filter media and the outflow well are to be collected. The size of each sample is about 100 ml. 7. The following data need to be collected at ever 30 minutes time interval: Time from starting: Taken By: Selby Sample Number Size(ml) Weight of Sediment (Filter) Concent. (ppm) Average Concent. Remarks Inflow 1 Inflow 2 Outflow 3 Outflow 4
8. To obtain the weight of sediment, use filter with a normal pore size of about 1.0 um to filter the sample and then evaporate water at 100 C in an oven. The weight of the sediment is the total weight of the filter with sediment after dried minus the weight of the filter. 9. The sediment concentration by weight can be computed using the following equation ppm = (weight of sediment x 1000)/(total weight of sample with container - weight of container) III Presentation of results Plot average concentration of inflow and outflow Vs. time for each sand type on a regular graph paper.