JUST ADD WATER Renovation of the Filters at the Soldier Canyon Filter Plant in Fort Collins, Colorado Thomas M. Getting, P.E., DEE, Filtration Product Manager John Geibel, P.E., Chief Engineer Michael Wild, Regional Manager The F. B. Leopold Co., Inc. 227 South Division Street Zelienople, Pennsylvania 16063 Thomas F. Ullmann, P.E., Principal The Engineering Company 2310 East Prospect Road, Suite B Ft. Collins, Colorado 80525 ABSTRACT The Soldier Canyon Filter Plant in Fort Collins, Colorado has completed the renovation of the existing filtration system and construction of new filters. A total of eight filters were renovated in two stages and four new filters were constructed. Each stage has a different media configuration and backwash scheme. The new filters were constructed with a mixed media filter bed and use an air-water backwash. The existing filters were renovated with a modified dual media filter bed and use a water-only backwash scheme. This paper examines the operation and performance differences after the renovation and is a unique opportunity to compare the two types of media beds on a full-scale, side-by-side operating basis. Indications are that the dual media filters using polymer as the coagulant had substantially longer filter run times than the mixed media filters, with no difference in effluent water quality. Further investigations revealed a difference in filter performance depending upon which feed water was used from either plate or tube settlers. The plate settlers produced water that provided longer filter run times. INTRODUCTION The Soldier Canyon Filter Plant was constructed in 1963 on the west side of Fort Collins, Colorado. The plant is jointly owned by the East Larimer County, Fort Collins-Loveland, and the North Weld County Water Districts and supplies water to the rural areas of eastern Larimer County and northern Weld County. The plant has a current capacity of 30 million gallons per day (MGD) and serves a population of approximately 55,000. In 1995, the plant joined the Partnership for Safe Drinking Water program and was only the 16 th facility in the nation to receive The Director s Certificate of Recognition from the U.S. Environmental Protection Agency (EPA). However, optimization of the treatment plant has been an on-going process since 1989, when the need to improve the plant s performance was recognized by the staff and the administrators. The optimization has involved the construction of improvements to the plant, upgrading of the plant staff, changes in the plant management structure, and changes in the attitude and philosophy of the administrators and staff. This paper documents changes made to the filter system and attempts to show the improvements that have been achieved.
BACKGROUND Raw water for the plant is from Horsetooth Reservoir, which is a component of the Colorado Big Thompson Project. The reservoir is filled with water that originates on the Western Slope of the Colorado Rockies and is delivered via pipelines and open channels by the Northern Colorado Water Conservancy District (NCWCD). Horsetooth Reservoir was constructed in 1949 and has a 151,000-acre foot capacity. The intake to the plant from the reservoir is approximately 200 feet below the surface. The intake structure does not currently have provisions for multiple level withdrawal. Currently, Horsetooth Reservoir is the only source of raw water supplied to the Soldier Canyon Filter Plant. However, a secondary surface water source is being constructed for joint use with two local facilities. The raw water in Horsetooth Reservoir is of excellent quality and is stable. However, the plant does experience problems with fluctuating manganese levels of 0.00 to 400 ug /l during the spring and fall. The treatment plant was originally designed as a conventional treatment plant with a capacity of 4 MGD. In 1970, the plant was expanded to a capacity of 6.0 MGD, and the treatment process was changed from conventional to in-line direct filtration. Growth in the area forced another expansion in 1973 to bring the plant capacity to 10.0 MGD. In 1978, growth again forced the plant to expand from 10.0 MGD to 20.0 MGD. In 1991, improvements were made to convert the treatment process from direct in-line filtration to conventional treatment. TREATMENT PROCESS The following is a description of the current water treatment process being employed: Rapid Mix The raw water is metered through a 42-in. venturi meter and the flow rate is controlled with a 30-in. sleeve valve. The primary oxidant, chlorine dioxide (ClO 2 ) dosed at 0.3 to 0.5 mg/l, is injected immediately downstream of the sleeve valve. The ClO 2 is primarily added for manganese removal but also has been shown to assist with filtration and particle removal. 1 Either PACL or a PACL/alum combination is added to the raw water ahead of an in-line jet mixer. Streaming current monitors control the coagulant feed rate. Flocculation The flow enters four, serpentine flow, four-stage flocculation basins. Each stage contains two 12-ft-diameter, horizontal mixing paddles that have variable speed drives. Floc-aid is added as necessary, usually in the mid-point of the first stage. Depending upon flow demands and the time of year, a combination of one, two, or three basins are utilized. Sedimentation There are currently four sedimentation basins available for use. Depending on the flow demands, various combinations of basins can be utilized to supply high-quality settled water to the filters consistently below 1.0 NTU. Sedimentation Basins 1 and 2 have tube settlers and usually provide flow to Filters 1-4. The rated capacity for these basins is 10 MGD and the average flow to the basins is at the design rate of 10 MGD. Sedimentation Basins 3 and 4 have plate settlers and usually provide flow to Filters 5-12. The rated 2
capacity of these basins is 40 MGD, though the normal flow rate is 20 MGD. All of the sedimentation basins utilize submerged sludge collectors for withdrawing accumulated solids. However, the characteristics of the particles left in the forward flow appears to be different and the effect on the filters will be shown later in this paper. Filtration The plant currently has 12 filters that have a combined surface area of 4,320 ft 2. Each filter is approved to operate at a rate of 7 gal/min/ft 2, though microscopic particulate analysis testing is required during any month in which filter rates exceed 5 gal/min/ft 2. Filters 1-8 were rehabilitated in 2000. The existing clay tile underdrain was replaced with Leopold low profile Type SL dual parallel lateral underdrain and Leopold IMS cap porous plate support in place of gravel. The media was replaced with 12 in. of 0.45 to 0.55 mm effective size (ES), <1.4 uniformity coefficient (UC) sand under 33 in. of 0.95 to 1.05 mm ES <1.4 UC anthracite. The media depth to ES ratio (l/d) calculates as 1,448 ([12 in. x 25.4 in./mm 0.5 mm ES] + [33 in. x 25.4 in./mm 1.0 mm ES]). These filters will be referred to as the dual media filters. It should be noted that Filters 1-4 receive flow only from Sedimentation Basins 1 and 2 while Filters 5-8 usually receive flow from Sedimentation Basins 3 and 4. However, Filters 5-8 are valved so that they, too, can receive flow from Sedimentation Basins 1 and 2. Filters 9-12 were constructed new in 1997. The Leopold Type S dual parallel lateral underdrain with Leopold IMS cap was used. The media consists of 4.5 in. of 0.2 to 0.3 mm ES <1.8 UC garnet sand under 9 in. of 0.45 to 0.55 mm ES <1.4 UC sand with a cap of 22.5 in. of 0.95 to 1.10 mm ES <1.4 UC anthracite. The l/d ratio calculates as 1,486 ([4.5 in. x 25.4 in./mm 0.25 mm ES] + [9 in. x 25.4 in./mm 0.5 mm ES] + [22.5 in. x 25.4 in. 1.0 mm ES]). These filters will be referred to as the mixed media filters. These filters can receive flow from any of the sedimentation basins but usually receive flow from Sedimentation Basins 3 and 4. All filters are equipped with on-line turbidimeters, monitoring filter effluent continuously. Particle counting has been incorporated into the control scheme. Corrosion Control/Disinfection/Clearwell Corrosion control is accomplished by the addition of soda and lime to the clearwell. The lime is fed as needed throughout the year while the soda ash is fed continuously to the influent of the clearwell dependent upon incoming manganese levels from the reservoir. The finished water Langlier Index goal of 1.0 has proven effective in meeting the Lead and Copper Rule. Disinfection is applied as break-point chlorination to the combined effluent. The operating range of free residual chlorine is 1.0 to 1.4 mg/l leaving the plant. Dosage is dependent upon the flow rates to the three water systems supplied and is adjusted as necessary. FILTER OPERATION Under normal operating conditions, criteria for filter operation require filters to initiate backwashing between 6 to 8 feet of headloss. Also taken into consideration before backwashing are turbidity values no higher than 0.10 NTU and particle counts, >2 microns, no higher than 15 particles/ml. Even though the filters do not have filter-to-waste 3
capabilities, turbidity spikes to the clearwell after backwashing are consistently below 0.10 NTU. Part of this performance is due to filling the filter headspace with backwash water at the end of the backwash cycle and spiking the incoming raw water with 500 ml of a cationic polymer immediately after a backwash. Although automatic backwashing equipment has been installed, the plant operators believe that manual washing has proven more effective with changing raw water conditions. The dual media filters use a standard water-only hydraulic backwash. Rotary agitators are used to break up surface particle build-up. The mixed media filters have both rotary agitators and air scour capabilities for backwash. These filters are usually hydraulically backwashed, with air scour used every third backwash. The plant has an extensive SCADA system that monitors, acquires, and stores the operation parameters of the various treatment plant unit operations. The parameters monitored in the operation of filters are headloss, filter run time, and effluent turbidity. Figure 1 presents typical data from Filter 10 (mixed media) and Filter 6 (dual media) on June 10, 2001. A direct comparison is made between the initial filter runs. The settled water for both filters was fed from the plate settlers. The effluent turbidity was the same as shown on the bottom of the graph. Note that the dual media filter had an additional 20 hours of run time. For this time period, the major coagulant was 7 mg/l of polymer. Flow to the filters was between 3.7 and 4.3 gal/min/ft 2. 12.0 6/10/01 (3.7-4.3 gpm/sq ft) 0.12 11.0 10.0 20 Hrs. Additional 0.1 HEADLOSS (Ft.) 9.0 8.0 7.0 6.0 5.0 4.0 3.0 Mixed Media Dual Media 0.08 0.06 0.04 EFFLUENT TURBIDITY (NTU) 2.0 0.02 1.0 0.0 0 0 5 10 15 20 25 30 35 40 45 50 55 60 TIME (Hrs.) 7.28 mg/l Summit 820B Filt 6 Filt 10 Filt 6 Turb Filt 10 Turb Figure 1 4
Figure 2 presents typical data from Filter 10 (mixed media) and Filter 6 (dual media) on July 5, 2001. A direct comparison is made between the initial filter runs. As shown on the bottom of the graph the effluent turbidity was the same. Again, the settled water for both filters was fed from the plate settlers. The dual media filter had an additional 11 hours of run time. For this time period, the major coagulant was 8 mg/l of polymer and the flow to the filters was between 4 and 4.5 gal/min/ft 2. 7/5/01 (4.06-4.55 gpm/sq ft) 12.0 0.06 11.0 11 Hrs. Additional 10.0 0.05 HEADLOSS (Ft.) 9.0 8.0 7.0 6.0 5.0 4.0 Mixed Media Dual Media 0.04 0.03 0.02 EFFLUENT TURBIDITY (NTU) 3.0 2.0 0.01 1.0 0.0 0 0 5 10 15 20 25 30 35 40 45 50 55 60 TIME (Hrs.) 8.0 mg/l Summit 820B Filt 6 LOH Filt 10 LOH Filt 6 Turb Filt 10 Turb Figure 2 Figure 3 presents the data comparing Filter 6 (dual media) to Filter 10 (mixed media) on July 29, 2001. During this time period alum was added in addition to polymer for coagulation. Alum is added during peak periods when the flow rate to the filters is increased and influent water quality changes are detected. During this time period the flow rate to the filters was increased to about 5 gal/min/ft 2. With effluent turbidities the same, the dual media filter run time was an additional 16 hours using the plate settled influent water. 5
12.0 11.0 7/29/01 (4.84-5.14 gpm/sq ft) 16 Hrs. Additional 0.06 10.0 0.05 HEADLOSS (Ft.) 9.0 8.0 7.0 6.0 5.0 4.0 Mixed Media Dual Media 0.04 0.03 0.02 EFFLUENT TURBIDITY (NTU) 3.0 2.0 0.01 1.0 0.0 0 0 5 10 15 20 25 30 35 40 TIME (Hrs.) 7.2 mg/l Summit 820B, 5.2 mg/l Alum Filt 6 LOH Filt 10 LOH Filt. 6 Turb Filt 10 Turb Figure 3 EFFECT OF DIFFERENT SETTLERS As mentioned above, the plant has two sedimentation trains that employ tube settlers in one train and plate settlers in the other. The tube settlers usually provide flow to the dual media filters, Filters 1-4, while the plate settlers usually provide flow to the remaining dual media and mixed media filters. The above data reflects that usual condition. However, it is possible to divert flow so that the tube settler basins can feed the mixed media filters. Because of this dual train condition and to quantify any effect that it has on the filters, data acquired from the sedimentation process units was investigated. Also, it was noticed that there was a run time difference between dual media Filters 1-4 and dual media Filters 5-9. Figure 4 presents the 2001 settling performance for the first seven months comparing the settled turbidity of the tube settlers to the plate settlers. On average, the settled turbidity of the plate settlers is relatively the same as that of the tube settlers. Since it was determined that there was no difference in the settled turbidity performance between the sedimentation trains, it was decided to try to compare the performance of the filters on alternate settled water. Figure 5 presents filter performance using alternate feeds. While Filter 2 (dual media/tube feed) and Filter 10 (mixed media/plate feed) were maintained with their normal settled water feed, Filters 5 and 6 (dual media) were switched 6
1.60 SCFP 2001 SETTLING PERFORMANCE 1.40 Plates Tubes 1.20 SETTLED WATER TURBIDITY (NTU) 1.00 0.80 0.60 0.40 0.20-1/1 1/31 3/2 4/1 5/1 5/31 6/30 7/30 8/29 9/28 10/28 11/27 12/27 Figure 4 12.0 11.0 7/29/01 (4.84-5.14 gpm/sq ft) Mixed Media/Plates 10.0 9.0 Dual Media/Tubes Dual Media/Plates 8.0 HEADLOSS (Ft.) 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 0 5 10 15 20 25 30 35 40 TIME (Hrs.) 7.2 mg/l Summit 820B, 5.2 mg/l Alum Filt 2 Filt 10 Filt 6 Filt 5 Figure 5 7
from tube settled water to plate settled water. Note that the dual media/tube fed filter tracks the same as the mixed media/plate fed filter. The dual media filters switched to plate settled water had an additional 18 hours of run time. This is double the run time of the dual media filter fed by tube settled water and the mixed media filter fed by plate settled water. To confirm the results in the above test using the mixed media filters, another comparison of the effect that the settled water difference has on the filters was performed. Filter 10 (mixed media) was fed with tube settler effluent while Filter 9 (mixed media) was fed with plate settler effluent. As shown in Figure 6, the difference in run times was an additional 4 hours for Filter 10 being fed the plate settler effluent or a decrease in run time for a mixed media filter that is normally fed plate settled water. After another filter run with tube settled water with a similar run time to the initial tube settled effluent (approximately 11 hours to reach 8 feet of headloss), Filter 9 was returned to plate settler effluent. The run time of Filter 9 returned immediately to match the run time of Filter 10, as shown in Figure 6. The only explanation of the filter performance difference offered to date comes from the plant operators. They feel that the particle size in the tube settled water is slightly larger than the particle size of the plate settled water. The larger particle size may contribute to higher solids loading in the top of the media and thereby cause shorter filter runs. Part of the reason for the particle size difference could be that the tube settlers are operated at capacity, while the plate settlers are operated at approximately one-half capacity. 12.00 4 Hrs Additional FILTERABILITY (Plate vs. Tube Effluent) Tube Settler Effluent 10.00 Plate Settler Effluent 8.00 LOSS OF HEAD (FT) 6.00 4.00 Plate Settler Effluent 2.00 0.00 0 10 20 30 40 50 60 7.28 mg/l Summit 820B Filt 10 Filt 9 Mixed Media Figure 6 8
CONCLUSION Over the years, the staff at the Soldier Canyon Filter Plant has endeavored to optimize their water treatment system through several upgrades and expansions. During the most recent upgrade, a difference in run times was noticed in the operation of the filters producing the same effluent water quality. The dual media filters had longer run times, up to double the mixed media filters, producing the same effluent turbidity and utilizing the same settled water. During the study an investigation was made to determine if the different sedimentation equipment affected the operation of the filters. It was found that even though the tube settled effluent had the same turbidity as the plate settled effluent, the filters operating on the plate settled effluent had longer run times. This could be due to the plate settlers being operated at one-half of their capacity. REFERENCES ¹ Ullmann, Turner, and Reed. Optimization of a Medium-Sized Water Treatment Plant: The Evolution of the Soldier Canyon Filter Plant. Presented at the American Water Works Association 1998 Water Quality Technology Conference, San Diego, California 9