Using WinSLAMM v10 to Predict Stormwater Pollutant Loads and Evaluate LID Management Approaches Overview

Transcription

1 Evaluating CSO Management Goals through Monitoring and Modeling of Green Infrastructure: Kansas City EPA National Demonstration Project (and some comments from on going Cincinnati Projects) Using WinSLAMM v10 to Predict Stormwater Pollutant Loads and Evaluate LID Management Approaches Overview Low Impact Development Symposium, St. Paul Minnesota August 18, Kansas City s CSO Challenge Combined sewer area: 58 mi 2 Fully developed Rainfall: 37 in./yr 36 sewer overflows/yr by rain > 0.6 in; reduce frequency by 65%. 6.4 billion gal overflow/yr, reduce to 1.4 billion gal/yr Aging wastewater infrastructure Sewer backups 2 Poor receiving water quality 1

2 KC s Modeling Connections SUSTAIN SWMM Individual LID Drainage (Transport) Multi scale Subarea Optimization Weight of Evidence KCMO XP SWMM Drainage (Transport) Design Objectives WinSLAMM Land Surface Charact Drainage (Transport) Design Options Stormwater Beneficial Uses Multi scale 3 Adjacent Test and Control Watersheds 4 2

3 Summary of Constructed Stormwater Controls in Test Area Design plan component Number of this type of stormwater control units in 100 acre test (pilot) area Device as a % of the drainage area Average drainage area for each unit (ac) Bioretention 24 (no curb extensions) (with curb extension) (shallow) Bioswale 1 (vegetated swale) Cascade 5 (terraced bioretention cells in series) Porous sidewalk 18 (with underdrains) or pavement 5 (with underground storage cubes) Rain garden 64 (no curb extensions) Total area treated by these devices (ac) 8 (with curb extension) Example micro flow and drainage area analysis for a set of stormwater controls in the test area, examining both direct runoff area to biofilters and overflows from upgradient biofilters. 6 3

4 7 Characteristics of Areas Draining to Stormwater Controls vs. Areas without Controls Land Component Area in subwatersheds with no devices Area (acres) Percent of subarea Area in subwatersheds with stormwater control devices Area (acres) Impervious, directly connected Impervious, draining to pervious areas Pervious areas Percent of subarea 8 Total area:

5 Preliminary Test and Control Watershed Flow Comparisons Mann Whitney Rank Sum Test Group N Median 25% 75% Pre construction pilot area to control area Rv ratio Post construction pilot area to control area Rv ratio Mann Whitney U Statistic= 206 T = 242; n (small) = 8; n (big) = 78; (p = 0.12) Therefore the difference is not significant at the 0.05 level for the eight sets of sample pairs available to data (p = 0.12). The apparent change between the initial monitoring period and the after construction monitoring period is about 30%. Individual biofilter monitoring and modeling suggest differences of 50+%. Therefore, need to detect a difference of at least 30% between the two groups. With a COV of about 0.4, this would require about 25 sample sets (for 95% confidence and 80% power). 9 Observed and Modeled Flows in the Test Watershed after Construction of Stormwater Controls 100,000 Modeled Total Area Runoff Quantity (ft3), with 1 in/hr native soil infiltration rates below biofilters 10,000 1,000 1,000 10, ,000 Observed Total Area Runoff Quantity (ft3) 10 5

6 One of the Kansas City rain gardens being monitored (zero surface discharges during the three years of monitoring; this rain garden is 20% of roof drainage area) WinSLAMM Production Function: Percentage Reductions of Annual Runoff Flows with Rain Gardens Percent reduction in annual roof runoff Percent of roof area as rain garden 12 6

7 WinSLAMM Production Functions: Effects of Underdrains in Biofilters on Annual Runoff Reductions (0.5 in/hr subsurface soil infiltration rates) Annual runoff reduction (%) no underdrain SmartDrain 3" drain Biofilter as a percenage of residential drainage area 13 WinSLAMM Production Functions: Clogging Potential for Biofilters in the Kansas City Test Area 14 7

8 th St. monitoring location, biofilter and adjacent porous concrete sidewalk (one of monitored, plus system) Three Study Areas for Green Infrastructure Effectiveness Monitoring in Cincinnati, OH (more than 20 GI demos in the city) 8

9 Available Flow Data at GI Demonstration Projects About 3 years of high resolution (5 minute) flow measurements from in system flow monitors located in combined and separate sewers on or adjacent to several green infrastructure installations Determine the base flow and the dry and wet weather flow components from the flow time series in sewer lines Flow (MGD) Cincinnati State Technical College August, 2012 Tuesday Manhole Number: /7/2012 8/14/2012 8/21/2012 8/28/2012 Tuesday :00 AM 3:00 AM 6:00 AM 9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM Trend Analysis 9

10 Cincinnati State Technical College 8/10/2012 Manhole Number: Flow Base Flow Rain 5 Rain: 0.32 in 0 Cincinnati State Technical College Manhole Number: /10/2012 Rain Flow Rain Rain: 0.32 in Flow (MGD) Rain (in) Flow (MGD) Rain (in) :00 AM 3:00 AM6:00 AM9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM :00 AM 3:00 AM 6:00 AM 9:00 AM 12:00 PM 3:00 PM 6:00 PM 9:00 PM Prepare individual storm event data summaries that are coordinated with the rain data for each monitoring point, including: start/end time of rain, rain duration, antecedent dry days, total rain, peak and average rain intensity, pipe flow start/end time, total pipe flow discharge volume, total runoff, peak and average flow discharge rates, R v (the ratio of runoff to rainfall depth). 0.5 Cincinnati State Technical College Level spreaders and biofilters 10

11 Cincinnati State Technical College Watershed Analysis Cincinnati State College (southern drainage with extensive bioinfiltration and rain gardens) R v values for different study periods for Cincinnati State College separate sewer system Kruskal Wallis One Way Analysis of Variance on Ranks test post hoc comparison test (Tukey s test) 11

12 Cincinnati Zoo Cincinnati Zoo Main Entrance with Extensive Paver Block Use 176 events monitored after construction 1 Main entrance of the Cincinnati zoo - Manhole Runoff Depth (in) Rain Depth (in) 12

13 African Savannah Zoo Kruskal Wallis One Way Analysis of Variance on Ranks of Rv Values Group N Missing Median 25% 75% Before During After There is a statistically significant difference (P = <0.001) All Pairwise Multiple Comparison Procedures (Dunn's Method): Comparison P<0.05 During vs After Yes During vs Before No Before vs After Yes Clark Montessori High School Porous Conc. Pavement Rain garden MH:

14 Clark Montessori High School Grouping Information Using Tukey Method for Rv Values N Mean Grouping Before & During A After B Location Preliminary Performance from Cincinnati Monitoring at Green Infrastructure Sites Cincinnati State College Southern Area (bioinfiltration and rain gardens) Cincinnati Zoo Main Entrance (extensive paver blocks) Cincinnati Zoo African Savannah (rainwater harvesting system and pavement removal) Clark Montessori High School (green roofs and parking lot biofilters on small portion of watershed) Runoff Volume Reduction (%) Compared to Pre Construction Data 80 Average Rv values after construction: 0.1 (compared to about 0.8 for conventional pavement in area)

15 Conclusions and Recommendations for Flow Monitoring for Green Infrastructure Performance Monitor both test and control areas both before and after construction of stormwater controls, if possible, for the greatest reliability (to account for typical year to year rainfall variations and to detect sensor problems early). Test areas should have most of their flows treated by the control practices to maximize measurable reductions. Any untreated upgradient areas should be very small in comparison to the test areas. Difficult to subtract two large numbers (each having measurement errors and other sources of variability), such as above and down gradient monitoring stations, and have confidence on the targeted flows. Conclusions and Recommendations for Flow Monitoring for Green Infrastructure Performance (Cont.) Most monitored flows from common rains may only result in shallow depths in the sewerage, a flow condition that is difficult to accurately monitor. Flow sensors may fail more often than expected. Costs of flow monitoring is small compared to green infrastructure investment. Use redundant sensors, such as an area velocity sensor (or bubbler) in addition to an acoustic depth sensor mounted on the crown. Calibrate the flow sensors at the beginning and periodically throughout the project period. Review flow data frequently and completely to identify sensor failures or other issues. Supplement the flow sensors with adequate numbers and placement of rain gages in the watersheds. 15

16 Conclusions and Recommendations for Flow Monitoring for Green Infrastructure Performance (Cont.) Monitor sufficient numbers of events to have statistically valid results for the performance expectations. As an example, with a COV of 1 (a typical value for stormwater), 50 pairs of samples would enable differences of about 50% or greater to be detected with 95% confidence and 80% power. It is very difficult to detect small differences with suitable confidence and power (the reason why most of the runoff needs to be treated). 16