Optimization Modeling of Phosphorus Removal in Reservoir and Stormwater Treatment Areas
Total Page:16
File Type:pdf, Size:1020Kb
Optimization Modeling of Phosphorus Removal in Reservoir and Stormwater Treatment Areas Daniel Smith and Lewis Hornung he Lake Okeechobee Watershed A combination of above-ground reser- Project (LOWP) is a major compo- voirs and Stormwater Treatment Areas Daniel P. Smith, Ph.D., P.E., is senior engi- nent of the Comprehensive (STAs) will be used to capture and treat T neer at the engineering and consulting Everglades Restoration Program (CERP). A runoff (Table 2). Configuration options are: firm Berryman & Henigar Inc. in Tampa. primary goal of the LOWP is to reduce phos- S Stand-alone (off-line) reservoirs that with- He is a member of the Water Environment phorus loadings to Lake Okeechobee from draw from a stream or canal and discharge Federation and the American Water Works contributing watershed planning areas north to the same or a different stream or canal. Association. Lewis Hornung is a senior of the lake (Figure 1). S Stand-alone (off-line) STAs that are simi- water resources engineering project man- Phosphorus must be reduced to restore larly configured, and reservoirs that with- ager with HDR Engineering, Inc. in West lake water quality and to meet the Total draw from a stream or canal and discharge Palm Beach. Maximum Daily Load (TMDL) of 105 metric to an STA, which then discharges to a tons per year (MTY). Best Management stream or canal. to simulate system performance: a reservoir Practices that are already planned will reduce A directly coupled reservoir/STA system model (RESOPT 3.0) and an STA simulation phosphorus loading from 433 to 235 MTY. is called a Reservoir Assisted Stormwater model. A stream matrix model was also The LOWP goal is to provide the additional Treatment Area (RASTA). For each of four employed to connect reservoirs and STAs 130 MTY needed to meet the TMDL (Table 1). basins (or Planning Areas), five Planning Area within the stream and canal network. The second LOWP goal is to provide sur- Alternatives (PAAs) were developed, consist- Simulation modeling was applied over a face water storage to help manage ecologically ing of combinations of off-line reservoirs, off- 36-year period of record to assess the effect of desirable lake levels and reduce the need for line STAs, and RASTAs (HDR Engineering, superimposing reservoirs and STAs (i.e, the damaging flood discharges from Lake Inc., 2005). Within each Planning Area, exist- LOWP) onto project Planning Areas. The Okeechobee to estuarine areas on the east and ing land uses and ecological values were used without-project conditions were the project- west coasts. The reservoirs will also provide a to pre-select land areas that could be used for ed future flows and loadings after implemen- continuous supply of water to the constructed construction of reservoirs and STAs. tation of planned BMPs, which were derived wetlands (Stormwater Treatment Areas). The number or reservoir and STA combi- from previous extensive modeling of the nations in a Planning Area was watersheds (Soil and Water Engineering potentially quite large. The Lake Technology, Inc., 2004). Okeechobee Combinatorial RESOPT 3.0 is a water budget and phos- Total P Loading Watershed Analysis Program phorus mass balance model that was devel- (Annual Average Mton/year) Planning Area (LOWCAP) was developed and uti- oped specifically to simulate LOWP reser- Future Target Future lized to evaluate the large number of (with LOWP with voirs. Phosphorus flux components in BMPs) Reduction LOWP possible combinations. LOWCAP RESOPT 3.0 are illustrated in Figure 2, and Lake Istokpoga/ computed storage volumes and water budget components are similar. Salient Indian Prairie 70.1 60.0 10.1 phosphorus load reductions, esti- features of RESOPT 3.0 are: Fisheating Creek 60.5 50.0 10.5 mated costs, and sorted alternatives S Reservoir water column control volume Kissimmee River 55.4 0 55.4 based on average annual values. S Completely mixed water column (0-D) Results were used to iden- Taylor Creek / S One-day timestep 49.0 20.0 29.0 tify a set of 20 cost-effective PAAs Nubbin Slough S Flow balance for reservoir that met the project goals. The next TOTAL 235 130 105 S Total phosphorus mass balance step was to perform more detailed S Total phosphorus (no speciation) evaluations of the PAAs to verify Table 1: Average Annual Phosphorus Loading from S Stream diversion rules storage capacities of reservoirs and Planning Areas to Lake Okeechobee. S Reservoir release rules phosphorus load reduc- For a given reservoir location, the point tion capacities and to of withdrawal from stream to reservoir and Management develop appropriate capac- Goal Approach the point of reservoir discharge were speci- Measure ities for pump stations, fied, as were capacity of withdrawal pumps. structures, and canals. Stream flow and phosphorus concentration Reduce Phosphorus Capture runoff and treat Stormwater Treatment Loading to Lake prior to discharge to the time series at the point of withdrawal were Okeechobee lake Areas Modeling Approach used to calculate the daily volume and phos- Provide improved Modeling tools were phorus mass routed into the reservoir for the management of lake Capture and detain peak Above ground reservoirs levels flows to lake needed to design reservoir specified withdrawal pump capacity. and STA systems that Precipitation time series were assembled Reduce freshwater Capture and detain peak could achieve project release to estuaries flows to lake; improved Above ground reservoirs from monitoring stations within each basin, lake management objectives within available and potential evaporation calculations pro- Table 2: LOWP Project Components. land area constraints. Two duced sinusoidally varying evaporation rates. coupled models were used Reservoir modeling included an over- 68 • JUNE 2005 • FLORIDA WATER RESOURCES JOURNAL Figure 2: Mass balance flux components for Total Phosphorus in RESOPT 3.0. Figure 1: Lake Okeechobee Watershed Planning Areas. flow feature for high-precipitation events imposed on full or near-full reservoirs; excess water and phosphorus greater than maximum working depth (8.0 ft. for all LOWP reservoirs) were routed to a receiving stream or canal via a crest overflow. Phosphorus mass balance modeling (Figure 2) included sedi- mentation and resuspension. Phosphorus sedimentation was modeled using a constant settling velocity, with the parameterization resulting in a first- order, depth-dependent removal rate constant (Smith, et al, 2004). Phosphorus resuspension was estimated using the Shallow Water Wave Model (U.S. Army Coastal Engineering Laboratory, 2002), a quadratic relationship for bottom shear stress (Lijklema et al., 1994), and resuspension rate as a linear function of bottom shear stress (Sheng and Lick, 1979). Average daily wind velocity was the driving force in the resus- pension calculation. Generally, depths of less than 1.5 feet and wind velocities of greater than 15 miles per hour were needed to discern a noticeable resuspension effect. Daily direct phosphorus deposi- tion onto the reservoir surface was predicted using a rainwater phosphorus concentration that was estimated from the average areal total deposition rates over the Lake Okeechobee watershed. DMSTA, the Dynamic Model for Stormwater Treatment Areas, is a non-steady-state model of Stormwater Treatment Figure 3: Fisheating Creek RASTA: 3240 acre reservoir and 6-cell STA Areas that simulates the hydrologic water balance and phospho- (17,000 acre). rus removal processes in treatment wetlands (Walker and Kadlec, 2004). The basic function of DMSTA is to predict phosphorus removal efficiency of an STA. For a given STA area, STA cell con- Withdrawal Point figuration (number of parallel trains, number of cells in series), Fish Eating Creek LOWP-07 To Lake cell aspect ratios, and given influent flowrate and total phospho- Calcreach 4 Okeechobee l a rus concentration time series, DMSTA predicts the flowrate and w o w l f a r r e d v h t i total phosphorus concentration in STA effluent. O The DMSTA-predicted treatment efficiency of the STA is cal- W culated by the differences between influent and effluent phospho- rus mass on a cumulative basis over a selected averaging period. Reservoir DMSTA was applied to LOWP STAs using the biokinetic param- eters for emergent macrophyte wetlands, which was the STA veg- STA etation assemblage specified by the Project Delivery Team. Stream Calcreach Node Operating Rules Controlled Flow To Lake Operating rules were needed to simulate the performance of Okeechobee Overflow reservoirs and STAs. These included specifying rules for the rates of withdrawal from canals to reservoirs, release rates from reser- Figure 4: Fisheating Creek matrix. Continued on page 70 FLORIDA WATER RESOURCES JOURNAL • JUNE 2005 • 69 to maintain a minimum in- specified; STA influent flowrates were main- Annual Average Total P Withdrawal stream flow. tained at the maximum HLR whenever possi- vs. Maximum Withdrawal Flowrate An analysis of streamflow ble. 70 rates was performed to iden- Since the maximum HLR could be met l Stream Total P Transport a 60 w w tify an efficient inflow pump only a small fraction of the time, the resulting a a r d h capacity. Typically, this average HLR was much lower, and always less t 50 i W W ) ) capacity was approximately than 5 cm/day. Furthermore, water levels in r P FEC PAA Removal Goal a l 40 e e a t y / o the 95th percentile stream STAs were never at excessively high levels that n T o t e e 30 g flow. For a stand-alone reser- could damage vegetation for continuous M ( ( a a r e v voir, release rate was based on periods. 20 A l a a analysis of mass phosphorus Based on this analysis, reservoir release u n 10 n n A removal within the reservoir, rates to directly coupled STAs, and maximum 0 and also mass phosphorus pump sizes STAs directly receiving stream or 0 1000 2000 3000 4000 5000 6000 7000 8000 removal in a downstream canal withdrawals, were based on the STA Maximum Withdrawal Flowrate (cfs) STA, if it existed.