GOING WITH THE FLOW: DESIGN TECHNIQUES TO IMPROVE RAW WATER QUALITY

Transcription

1 GOING WITH THE FLOW: DESIGN TECHNIQUES TO IMPROVE RAW WATER QUALITY ABSTRACT Porter Rivers, III, PE URS Corporation Keith Cannon, PE URS Corporation Geoff Smith, PE URS Corporation Ken Green, Eastern Band Cherokee Indians The Eastern Band of Cherokee Indians (EBCI) owns and operates a 6.0 million gallon per day (MGD) surface water treatment plant (WTP) which uses the Oconaluftee River as a raw water source. As a shallow mountain stream, the Oconaluftee is subject to periodic episodes of high turbidity (due to runoff) as well as seasonal impacts resulting from foliage during the fall. The original intake at the raw water pump station was relocated (situated approximately 400 feet upstream of the raw water pump station) due to underground petroleum contamination near the existing pump station. However, relocation of the intake has resulted in significant operational and maintenance issues that hinder efficient operation of the WTP. These issues include: 1. Excess sediment enters the intake which damages the intake pumps and collects in the prefiltration units of the WTP, requiring frequent cleanout and WTP downtime. 2. Leaf collection on the intake reduces water withdrawal, requires frequent maintenance, and lowers the efficiency of WTP operations. 3. Icing of the intake screens reduces peak intake volume and requires maintenance by WTP staff under potentially hazardous conditions. To address these issues, a range of options were proposed including in-stream (ie, j-hook and cross vane structures) and off-stream (ie, sedimentation, mechanical bar screens, etc) methods. Evaluation of these options was accomplished by performing a detailed assessment of the intake area as well as the natural characteristics of the Oconaluftee River. Analysis of stream flow, sediment transport, and storm flow velocities was used to determine channel flows and particle size distribution through the water column. Following completion of sediment transport modeling, the proposed design options were compared to determine the most feasible approach. This presentation will cover the results of the field assessment and sediment transport analysis as well as giving an overview of the final design options and recommendations provided to EBCI for retrofitting of their raw water intake. KEYWORDS Sediment transport, silt deposition, stream velocity INTRODUCTION The original 1996 raw water intake consisted of a static bar screen structure located on the river bank. The structure used two bar screens constructed of 3/8-inch by 2-inch bars, angled at 45 degrees in the direction of river flow and spaced at 1-1/4 inches. Raw water entered through the screens into a two-chamber flow division box. The division box is connected to the RWPS wet well chambers using two 24-inch diameter pipes, each equipped with sluice gates for isolation of the wet well chambers.

2 During the autumn months, the screens would routinely require cleaning due to leaf packs that would collect at the screen inlet and clog the screens. Ice accumulation during the winter months was also a maintenance issue. An underground gasoline tank leak at the Golden Eagle Exxon gas station (located across Highway 441 from the raw water pump station) caused EBCI to cover the existing bar screens and relocate the raw water intake upstream of the spill site. The raw water intake was upgraded in 2008 by replacing the static bar screen with four in-stream, passive intake screens, each with a capacity of 2.0 MGD. The location of the new screens is indicated in Figure 1. Figure 1. Relocated Intake Screens Operational and maintenance issues related to the existing intake screens include: Ø Screen blinding from debris in the river and sand/silt deposits around the screens, Ø Sand/silt inflow to the RWPS, and Ø Constant stream maintenance due to the debris and sand/silt deposition. The existing screens are cylinders with openings around the entire circumference. To maintain screen performance, 14 inches clearance from both the river bottom and the water surface is required. Performance of the screens is hindered by varying water levels in the river. When the required depth over the screens is not provided, the inlet velocity increases which allows additional suspended sand/silt particles to enter the screens. However, as the water depth increases, the velocity of the river decreases. Sand/silt deposits as well as leaf packs collect around the screen due to the decreased stream velocity. The deposits around the

3 screens result in sand/silt inflow to the RWPS during normal water withdrawal. The air scour system is insufficiently sized to remove these deposits, which causes reduced screen performance resulting in EBCI staff having to regularly maintain the screens by dredging the deposits. Typically, dredging operations occur semi-annually with heavy equipment and more frequently by manual means. Although the screens are located upstream of the spill site (thus reducing the risk of raw water contamination), there is still risk of the screens being damaged by debris traveling down the river, especially during heavy rain events, when the stream velocity increases significantly. In addition, the stream dredging poses safety risks for EBCI staff, particularly during cold months. Raw Water Pump Station The RWPS was also constructed in 1996 and consists of a below-grade, two-chamber, concrete wet well and an at-grade pump building constructed over the wet well. Raw water pumping is accomplished using two vertical turbine pumps w rated for 2,100 gallons per minute (GPM), giving the pump station a reliable capacity of 3.0 MGD. The pumps discharge into a 16-inch transmission main to the WTP. Operational and maintenance issues related to the RWPS include: Ø Inadequate reliable capacity relative to the WTP capacity, Ø Decreased flow variation capability due to constant speed pumps, and Ø Excessive sand/silt pumped to the WTP due to the inflow from screens and lack of sand pumps in the wet well. In addition to the operational concerns, there is the on-going concern of contamination of the raw water intake and wet wells at the RWPS associated with the benzene and MTBE plumes that currently exist under and around the RWPS site. Figure 2 shows the extent of the contamination plume as of early 2013 (Martin and Slagle, 2013). There is evidence of undermining of the sidewalk around the raw water pump building, which could be the result of surface water erosion or leaking from the wet well. If there are leaks in the wet well, there is a risk of contaminated groundwater infiltration as well as possible structural instability. There is also evidence of leaking around the 24-inch raw water intake line connection to the flow division box, which also results in the risk of contaminated groundwater to enter the wet well. The existing raw water transmission main to the WTP and the finished water line each cross the contamination plume, which poses an additional risk to public health. METHODOLOGY To evaluate the existing conditions, URS developed a field assessment plan to determine hydrologic and hydraulic conditions occurring at the intake screens. The primary components of the field assessment included collecting data for development of a stream model, collecting sediment samples during varying flow conditions, and analyzing leaf loading conditions. The field work and resulting analyses were coordinated through EBCI staff. Both URS and EBCI personnel participated in completing the assessment. Stream Modeling The first step in the field assessment process was to collect sufficient data for development of an accurate hydrologic and hydraulic model to determine flow characteristics at the raw water intake. EBCI staff collected six cross sections along the river, beginning just upstream of the Acquoni Road Bridge and ending at the existing RWPS. These cross sections are indicated in Figure 3.

4 Figure 2. Contaminant Plume around Existing Raw Water Pump Station Figure 3. Cross Sections used from Stream Modeling

5 The cross sections are defined as follows (see Figure 3): 1. Riffle: Shallow, fast water 2. Riffle Run: Shallow fast water, then deepening and slowing 3. Run Pool: Continuing to deepen to slowest velocity 4. Pool Glide: Channel begins to get shallower, stream velocity still low 5. Glide Riffle: Continuing to get shallow, increased water velocity 6. Riffle Run: Shallow fast water, then deepening and slowing Concurrent with the cross sections, URS staff developed a channel profile along the centerline of flow in the river between the upstream and downstream cross sections. At each cross section, URS collected stream velocity, bedload sediment samples, and in-stream suspended sediment samples. Velocity and sediment samples were collected through the vertical profile (depth) of the cross section to determine how differing water depths were influencing channel shear, velocities, and ultimately sediment transport throughout the system. This data was utilized to evaluate the overall sediment load to the raw water intake and how specific particle sizes were reacting during normal flow conditions. Sediment Transport The primary area of concern related to the raw water intake was the passing of sediment through the screens. Conversations with EBCI staff indicated that a high level of sedimentation occurred around the screens and through the intake during high rainfall events. URS worked with EBCI to develop a suspended sediment sampling plan to evaluate sediment loads during these rainfall events. A sediment sampling device was installed along the downstream face of the Acquoni Road Bridge. EBCI staff collected several sediment samples during rainfall events through the winter of URS evaluated this data and determine that four sample events would be used to develop the projected sediment load to the intake screens. Leaf Loading Since leaf litter loading on the intake screens was a particular concern of EBCI, URS completed an evaluation of leaf litter loading to determine transport characteristics throughout the river. During the fall of 2012, URS performed three leaf pack analyses of the river during normal flow conditions. Based on empirical data regarding the study of leaf litter transport, it was anticipated that the highest volume of leaf litter would occur during storm events in the fall season. However, the likelihood of settling and ultimately clogging the intake screens would occur prior or subsequent to these storm events, when water levels and flow velocities were lower. Therefore, URS completed a leaf loading assessment during normal flow conditions within the river. Samples and corresponding stream channel velocities were collected across the six surveyed cross sections as well as in areas of noticeably high leaf pack collection. DISCUSSION Based on the cross section and profile data collected, in conjunction with existing LIDAR topographic data provided by EBCI, a detailed model of the stream channel was developed between the upstream limits of the study area and the RWPS. This model was used to evaluate operational impacts on the existing raw water intake. Findings from this analysis were used to provide EBCI with potential in-stream recommendations for meeting the project goals.

6 Stream Modeling The physical characteristics of the river above and below the raw water intake consist of a stable natural stream channel. The channel banks are fairly well stabilized and provide marginal input with regards to sediment loading. This evaluation indicated that natural sediment transport associated with a high gradient mountain stream is present, and except for minor spot erosion associated with tree fall or development, there is little in the way of potential for excessive sediment loading from the study area itself. Physical assessment of the plan view of the river (i.e. the flow pattern, radii and curvature), indicate that the raw water intake is located along the outside of a meander bend within the river. The profile of the stream channel indicates that the existing raw water intake is located within a glide area of the river. While the river has a stable profile structure, i.e. riffle (shallow fast water) run (deepening and slowing of water) pool (slowest and deepest areas) glide (shallowing and increasing velocity), the actual location of the raw water intake is located in the pool-glide portion of the river. The combination of the intake being located in both the outside of a meander and in the pool-glide portion of the profile provides a worst-case scenario for sediment deposition. At this point, water is extremely deep relative to other areas of the river, the overall river width is wider and flow velocities are very low. This provides an ideal location for sediment deposition within healthy stable stream systems such as the Oconaluftee River. Additionally, this pool-glide area is noticeably longer than other pool-glide areas along the river through Cherokee. The flow and velocity evaluation of the river correlate directly with the physical characteristic evaluation. As the channel enters the meander bend at Cross Section 3, the velocity lowers as the flow enters the pool structure. As the flow reaches the end of the pool and beginning of the glide section (Cross Section 4) the velocity drops at the centerline of flow. This would result in the deposition of sediment at this location. The raw water intake is located directly in the centerline of flow for Cross Section 4. Additionally, the vertical velocity profile was evaluated, and again, at Cross Section 4, the lowest bed channel velocity was observed. This would indicate a higher likelihood that small particles could settle out at this location. Sediment Transport Based on in-stream data collection, it was determined that the location of the raw water intake had the highest potential for particle settling during normal flow conditions within the river. Further examination of sediment transport within the river was conducted by looking at a series of sediment sample collection points. Sampling of in-stream sediment was inconclusive with regards to total suspended sediment (TSS) and settling through the river. Baseline sampling at each cross section resulted in an extremely low baseline TSS (less than 5.0 mg/l). Thus, to determine where sediment settled and how the particle sizes were distributed, further investigation was necessary. The first step was to evaluate the bedload sediment samples. URS completed an evaluation of particle size distribution across the length of each channel cross section. For Cross Sections 1, 2, 5, and 6, there are no particles within the bedload smaller than 1.0 mm. This indicates that velocity and channel shear is high enough to preclude the deposition of these smaller particles at those cross sections. However, Cross Sections 3 and 4 do indicate the presence of a minimum of 20% of materials smaller than 1.0 mm. The distribution of this material across the length of the cross section is dramatically different for these two sections. Field data indicated that a majority of the smaller material within Cross Section 3 was located along the left bank as part of a sediment bar, while the majority of the smaller

7 particles within Cross Section 4 were located at the center of flow, directly corresponding with the location of the raw water intake screens. Further analysis was performed between the sediment samples collected at the raw water intake and within the water treatment plant (WTP). A substantial deposition of material within the WTP was routinely occurring that required frequent cleaning and maintenance. The evaluation of the sediment load was utilized to determine the material particle size reaching the treatment plant versus the material that was being deposited at the intake screens. What can be concluded from the assessment of the two samples is that approximately 90% of the material that reaches the WTP has a particle size less than 1.0 mm, while 80% of the material that remains in the stream around the intake screens is greater than 1.0 mm. By evaluating the flow data and the sediment load data together, conclusions were drawn with respect to reducing the overall sediment load that enters the intake screens (ultimately reaching the RWPS and water treatment plant). Thus, it was concluded that even small storm events can increase the sediment loading through the system by well over 2000%. Table 1 compares the sediment loading at various stream flow volumes. Date Table 1. Sediment Load Sampling Results Sediment Load Comparison Sediment Concentration Stream Flow Sediment Load** % Change from** (mg/l) (MG/day) (tons/day) Baseline Average Baseline* Storm 12/17/ % Storm 1/14/ % Storm 1/15/ % * Highest measured baseline concentrations used for analysis - at Cross Section 2 ** Loading and % change determined and evaluated utilizing available sampling data. The water depths and in-stream velocities precluded the measurement of a velocity profile during these storm events at the intake structure. However, empirical sediment transport modeling indicated this location maintained low velocities across the bed, approximately 1.2 ft/sec, which still permitted the settling of small particulates suspended in the storm flows. This was verified through grab sample visual evaluation following the 100-year storm event, where the substrate at the intake remained high in fine sand and silt particulates. Leaf Loading Leaf pack sampling was completed on three separate occasions during the leaf fall season in These events were during the initial leaf fall, during high leaf loading and once following the final leaf fall in November. Sampling was completed at the same points along each of the six cross sections where velocity was sampled in addition to visual leaf packs along the banks and profile of the river. Leaf samples were collected through a dip net attached to the flow meter to correlate suspended leaf loading to channel velocity. Overall, suspended leaf loads were higher at the surface of the river and the bottom third of the velocity profile. Very little leaf transport was observed during the center of the water column, which reflects average channel velocities. While actual leaf loads varied by cross section during each sampling event, general conclusions for leaf transport based on velocity and channel characteristics have been developed:

8 Ø Leaf transport through the system was highest at the surface, where velocities were also high. Ø Little vertical transport, top-to-bottom or bottom-to-top, was visually observed. Ø Leaf transport was observed from low velocity areas along the channel banks while slowly settled to the bottom of the channel through gravity transport. Leaf packs were highest in length, depth and concentration along the channel banks, where corresponding channel velocities were less that 1.0 ft/sec and average channel depth less than 0.5 feet. Leaf snags were observed at higher velocity areas downstream of Cross Section 1 and at Cross Section 6 at the RWPS. These were contributed to large boulders which snagged leaves. These packs disappeared during high flow events. Leaf packs were observed along the channel bottom at two primary locations, Cross Section 4 and Cross Section 5. The deposition at Cross Section 4 began just upstream of the intake screens resulting from a debris jam and transverse sediment bar, which significantly decreased velocities at this point in the channel, (velocity at leaf packs was between 0.3 and 0.1 ft/sec). This leaf pack was evident during all three sampling events. The deposition at Cross Section 5 was solely attributed to a transverse bar which created an obstruction blockage. However, at this point, velocities in the channel at the leaf pack were significantly higher, 1.1 ft/sec. This leaf pack was observed for the first and second sampling event, but was absent at the final sampling event, indicating that the channel velocities transported these leaves over time. CONCLUSIONS Based on the physical, hydrologic, hydraulic and leaf loading analyses, the following conclusions have been drawn. Ø To reduce the potential sediment intake through the existing screens and ultimate loading to the water treatment plant, the velocity along the intake screens needs to be increased to greater than 1.5 ft/sec. This should increase the size of the particles that settle in this location and in turn, would increase the transport of smaller particles through the system. Ø An increase in velocity would also transport leafs through the system and reduce the potential for leaf pack formation at the intake. As observed through the leaf analysis, the intake would also need to be reconfigured to reduce the amount of surface area facing upstream to eliminate the potential for the development of leaf snags that were observed in other areas of the river. Proposed In-stream Modifications To accomplish this change in velocity, URS evaluated the placement of an in-stream structure similar to a j-hook or partial cross vane (using the stream model). These structures can be made of wood or large boulders and are in the shape of a v with the point facing upstream. The point of the v facilitates an increase in velocity and channel shear directly downstream. Typical placement of these structures would result in natural excavation downstream to create a deep pool in order to naturally reduce velocities. In the EBCI application, this pool would not form due to the bedrock located below the intake screens. Thus, a higher velocity would be maintained throughout the velocity profile at the intake structure. This would result in a clean system and only larger particulates could settle out at the intake. These larger particles would be less likely to be drawn into the RWPS due to the screen size and intake velocity. Initially, this alternative was considered viable to rectify the sediment and leaf loading issues related to the intake. Further analysis (including a risk assessment) indicated potential drawbacks to this alternative. The size of the rock necessary to resist sliding due to channel shear would be extensive. Observations made during the 100-year event in January 2013 indicated the transport of large woody debris and channel velocities greater than 10.0 ft/sec. Potential for moving of these in-stream structures

9 by impact from debris exists, and should the structure become dislodged, it would likely damage or destroy the intake structure. Since no secondary intake exists for the RWPS, damage to the intake would be catastrophic to the ability of the water plant to provide drinking water to the community. In discussions between EBCI and URS it was determined that this risk, however remote, resulted in the placement of an in-stream structure at the existing location becoming a less desirable option. However, placement of an in-stream structure could be used in conjunction with an intake located along the channel bank to reduce potential sediment loading to the water plant with reduced risk of damage to the overall intake. These structures could be much smaller than the large in-stream structure due to the lower velocities along the banks of the channel and could be more easily anchored into the channel bank to prevent movement. Public Health and Safety Assessment The overall assessment of public health and safety associated with the raw water intake and pump station focused directly on the ramifications of the fuel spill at the former Golden Eagle fueling facility. While the in-stream raw water intake structure does provide some level of risk to public safety due to its proximity to recreational activities, this risk is negligible in relation to the potential contamination associated with the existing benzene and MTBE plumes. Since the spill, the existing Golden Eagle facility has been demolished and removed. Martin & Slagle, (geotechnical contractor under the direction of the EPA) has developed a corrective action plan to remediate the site. It is anticipated that it will take a minimum of three years (2016) to complete the remediation of the contaminated soils to a level that is considered safe. During the interim, the potential remains for contamination of the existing wet wells at the RWPS as well as the transmission line that carries raw water from the pump station to the WTP and the distribution line that carries treated water from the WTP to the system. While there currently is no indication that contamination exists in the treated water, the potential for contamination is a concern to the community with respect to ensuring a safe water supply to the community. In addition to the risk to public health, if the existing plume were to penetrate the existing wet well, transmission line, or distribution line, EBCI would be forced to take emergency measures to bypass these parts of the system, which would result in another major disruption to service. During periods of high tourism, this could have a serious financial impact. Recommended Solution Based on the evaluation of the data collected and the project goals established by EBCI, a recommended solution was proposed. Based upon input from EBCI staff, the actual intake will be located upstream of the pump station (above Miller Creek) on the bank. This provides increased protection from the risk of future contamination as well as minimizes potential damage to the intake from high velocity debris during storm events. The new intake will feed into a new raw water pump station that will be located upstream of the contaminant plume. In addition, portions of the raw water transmission line to the WTP and the distribution line from the WTP that are currently located in the contaminant plume will be relocated outside of the contaminant plume. Figure 4 illustrates the proposed recommended improvements.

10 Figure 4. Recommended Solution for EBCI Raw Water Intake and Pump Station ACKNOWLEDGEMENTS We would like to acknowledge the input of the WTP Operations staff as well as the EBCI Engineering Department for their participation and contribution to this project. REFERENCES Martin and Slagle (2013), Corrective Action Plan, Golden Eagle Exxon, Facility ID , UST Incident No 28436, MSGS Project No PE 11-01, Black Mountain, NC. Going With the Flow