Assessing the Severity of Rainfall-Derived ... - ACS Publications

Sep 8, 2011 - for abrupt failure also allows for better asset management.8. However, most .... download.htm) and processed using ArcGIS 9.3 software (...
0 downloads 0 Views 2MB Size
ARTICLE pubs.acs.org/est

Assessing the Severity of Rainfall-Derived Infiltration and Inflow and Sewer Deterioration Based on the Flux Stability of Sewage Markers Jessica M. Shelton,† Lavane Kim,† Jiasong Fang,§ Chittaranjan Ray,†,‡ and Tao Yan*,† †

Department of Civil and Environmental Engineering, ‡Water Resources Research Center, University of Hawaii at Manoa, Honolulu, Hawaii, 96822 United States § Department of Natural Sciences, Hawaii Pacific University, Honolulu, Hawaii, 96813 United States ABSTRACT: This study investigated the flux stability of select chemical and biological sewage markers, including caffeine, total nitrogen (TN), total suspended solids (TSS), E. coli, and enterococci, and their suitability in assessing the severity of rainfall-derived infiltration and inflow (RDII) in a residential sewershed. To quantify and compare marker flux stability, concentrations of the candidate markers in two dry-weather periods were determined and the one-day lag autocorrelation coefficients (r) of their mass fluxes were calculated. TN (r = 0.820.88) exhibited higher and more consistent flux stability than TSS (r = 0.490.82), caffeine (r = 0.560.58), E. coli (r = 0.360.87), and enterococci (by culture; r = 0.400.52), all of which except enterococci by qPCR (r = 0.100.21) showed significant autocorrelation. Sewage flows and marker concentrations were also monitored in two wet-weather periods, and the severity of RDII (RRDII) were calculated using either flow measurements or marker concentrations independently. Corresponding to its outstanding flux stability, RRDII values estimated by TN predicted all severe RDII instances and gave the highest and most consistent correlation (r = 0.740.78) among the different sewage markers. Overall, the study illustrated the feasibility of using the flux stability of sewage markers in assessing the severity of RDII and thereby deterioration levels in sewer systems.

’ INTRODUCTION Well-maintained wastewater collection systems are important for proper conveyance of sanitary sewage to treatment facilities prior to environmental discharge. However, the ca. 600 000 miles of sewer pipes in the United States are eroding with age, use, and neglect.13 It is estimated that approximately 75% of sewer systems are functioning at less than half of their design capacities,4,5 and about 23 00075 000 sanitary sewer overflows (SSOs) occur every year in the United States as a result of deteriorated sewer systems.6,7 To adequately protect the environment and public health, significant sewer rehabilitation and replacement works are urgently needed, which would cost approximately $270 billion through the year 2019 according to a 2002 U.S. EPA estimate.3 Efficient implementation of such massive rehabilitation and replacement efforts would require a thorough assessment of sewer conditions to rank the different sewer sections based on the level of deterioration.4 Prioritization of high-risk areas would allow utilities to focus their efforts and proactively treat problems, avoiding the higher costs associated with emergency repairs. Ranking the sewers in terms of deterioration severity and potential for abrupt failure also allows for better asset management.8 However, most current inspection methods are not suitable for large-scale sewer condition assessment at the sewershed or sewer system level. Sewer inspection is predominantly conducted by physical methods, with the most commonly used being CCTV examination.9 These methods are often too detail-focused and r 2011 American Chemical Society

labor-intensive to achieve high-throughput system-wide condition assessment.8,10 One potential approach for system-wide sewer deterioration severity assessment is to use rainfall-derived infiltration and inflow (RDII) as an indicator of sewer structure deterioration.8 RDII is a symptom of sewer deterioration and is also a contributing cause of overloading the system, SSOs, and reduced efficiency of treatment facilities.4,11 Because RDII increases with increased severity of sewer deterioration, given the same precipitation conditions, the levels of RDII are directly related to sewer deterioration levels in different sewer sections.7 However, determining RDII by direct flow measurement is limited by the technical difficulties of obtaining accurate flow measurements12 and the high costs associated with simultaneous measurement of sewage flow at numerous locations in a sewer system, which is required for system-wide assessment efforts. Alternatively, RDII may be quantified through chemical and bacteriological analyses, which are usually less costly to perform and can be easily automated for higher throughputs than by direct flow measurements. Among the countless chemicals and organisms present in sanitary sewage, some may exhibit stable diurnal fluxes (i.e., having relatively constant total mass over a Received: June 7, 2011 Accepted: September 8, 2011 Revised: September 8, 2011 Published: September 08, 2011 8683

dx.doi.org/10.1021/es2019115 | Environ. Sci. Technol. 2011, 45, 8683–8690

Environmental Science & Technology given time period in any day) due to their specific sources and inputs from households in a residential community. Such markers may include fecal bacteria such as Escherichia coli and enterococci,13,14 caffeine,15,16 total nitrogen,17 and total suspended solids (TSS).17 For the sewage markers with stable fluxes, sewage flow dilution caused by RDII would proportionally decrease their concentrations; conversely, the concentration changes can be quantified to determine the RDII level. Because of the high-throughput analysis techniques available for these sewage markers, numerous sewer sections in a sewer system can be sampled and analyzed simultaneously under the same weather conditions, and the resulting RDII levels can be used to infer sewer deterioration severity. The overall goal of this study was to develop a marker-based approach to determine RDII levels for assessing sewer deterioration severity at the system-wide level. It was hypothesized that certain sewage markers exhibit stable diurnal fluxes and their concentration changes under wet-weather conditions are a function of the severity of RDII and sewer deterioration. Three specific tasks were carried out in this study to achieve the goal. First, concentration dynamics of select candidate sewage markers, including TSS, total nitrogen (TN), caffeine, E. coli, and enterococci, were investigated under dry-weather conditions in the Manoa sewershed (a residential sewershed on the Island of Oahu, Hawaii). Second, the fluxes of the candidate markers were determined for the eight 3-h time periods of a day under dry-weather conditions, and the markers with stable diurnal fluxes were identified. Finally, sewage flow and sewage marker concentrations under wet-weather conditions were monitored. The severity levels of RDII were estimated by sewage marker concentrations, which were then compared with those determined by direct flow measurement.

’ EXPERIMENTAL SECTION Sewershed and Sampling Site. Manoa sewershed serves the predominantly residential Manoa valley (Figure 1) which includes the University of Hawaii. Sanitary sewage from Manoa valley flows through two parallel sewer pipes through the University of Hawaii-Manoa campus. The sampling location (sewer manhole 368092) is located on the campus grounds but is just upstream of wastewater discharges from the university facilities (Figure 1), which avoids fluctuations and atypical inputs caused by university generated activities. The sewer pipe to be sampled is 15 in. in diameter and accessible through a 6.6 ft deep manhole. Data for the sewershed geography and the sewer pipe network were obtained from Hawaii Statewide Geographic Information System (http://hawaii.gov/dbedt/gis/ download.htm) and processed using ArcGIS 9.3 software (ESRI; Redlands, CA). Flow Monitoring and Sewage Sample Collection. Flow measurements at the sampling site were conducted using an ISCO 2150 area velocity flow meter (Teledyne, Lincoln, NE), and data acquisition and analysis were accomplished using the FlowLink 5 software (Teledyne). Sewage samples were collected using a Sigma 900 automatic field sampler (Hach Company, Loveland, CO). The area velocity sensor of the flow meter and inlet tube of the autosampler were submerged into sewage flow and mounted to the bottom of the sewer pipe using an ISCO street level manhole installation apparatus (Teledyne). Flow monitoring and sewage sample collection were conducted in two dry-weather (DW) periods and two wet-weather

ARTICLE

Figure 1. Sewer pipe network in the Manoa sewershed and the location of Manoa watershed on the island of Oahu, Hawaii. The sampling manhole location is indicated by a dark star.

(WW) periods. The two DW periods were Aug. 25Sept. 1, 2010 and Feb. 1017, 2011, which flank the two WW periods which were Sept. 30Oct. 7 and Nov. 310, 2010. Flow measurements were taken every 15 min, and the flows were averaged over 3-h periods to coincide with the 3-h sewage sampling interval. Sewage samples (150 mL) were taken every 15 min and every 12 samples were combined into a 3-h composite sample. For each sampling period, the flow measurements and sewage sampling continued for seven days, resulting in eight samples per day for a total of 56. The sewage samples within the autosampler were kept at approximately 4 °C using ice packs, and transported back to the laboratory every day for immediate processing. Sample Analysis. The sewage samples were analyzed for various parameters, including pH, salinity, TSS, TN, caffeine, E. coli, and enterococci. pH was measured using a Denver Instruments pH meter (Denver Instruments, Bohemia, NY), and salinity was measured using an Orion 150A Plus conductivity meter (Thermo Scientific, Beverley, MA). TSS was measured by filtering sewage samples (50 mL) through 1.1-μm glass fiber filters (VWR Grade 693; West Chester, PA) and determining dry weight difference prior to and after filtration according to Standard Methods 2540D.18 TN, Caffeine measurement, and bacterial enumeration (either by cultivation or by real-time PCR (qPCR)) are described in detail below. TN Quantification. To measure TN in sewage samples, 30 mL of each sample was first filtered through a 0.45-μm cellulose-ester membrane filter, and the filtrates were acidified to pH 2 or less using 1 N HCl before being analyzed based on an oxidative combustion-chemiluminescence method using a Shimadzu TOC/TN analyzer (Shimadzu, Columbia, MD). TN in the sample was converted into nitrogen monoxide by combustion 8684

dx.doi.org/10.1021/es2019115 |Environ. Sci. Technol. 2011, 45, 8683–8690

Environmental Science & Technology

ARTICLE

and then to nitrogen dioxide by reacting with the oxidizer ozone, and changes in chemiluminescence were detected by the instrument, which were compared against external calibration standards to determine TN concentrations. Caffeine Quantification. A SPE-GCMS procedure developed by the U.S. Geological Survey 19 was modified to quantify caffeine in sewage samples. One liter of each sewage sample was first prefiltered through 0.7-μm glass-fiber filters to remove suspended solids. The filtrates were then amended with 60 g of NaCl to increase the ionic strength and stabilize the solution, and 100 μL of 13C3-caffeine at 20 ng/μL in methanol was added to each 1-L sample as the recovery surrogate. Samples were acidified to pH 4.3 using sodium acetate buffer (30 g/L of acetic acid and 15 g/L of sodium acetate). Solid phase extraction (SPE) of the acidified samples was performed using HLB Oasis cartridges (Waters, Milford, MA) with a sample flow-through rate of ca. 15 mL per min. The sample bottles were rinsed with potassium phosphate buffer (30 g/L of K2HPO4 and 20 g/L of KH2PO4, pH 7), which was also passed through the SPE cartridge to neutralize in preparation for elution. The cartridges were airdried, and then eluted with six passes of 3 mL of the solvent mixture (hexanes:dichloromethane; 50:50). The eluents were evaporated using a N-Evap nitrogen evaporator (Thompson, Clear Brook, VA) to approximately 0.4 mL. After amendment with 20 μL of 100 ng/μL of Phenanthrene-d10 in dichloromethane as an internal quantification standard, the samples were evaporated to the final volume of 0.4 mL for subsequent GC-MS quantification. All samples were analyzed on an Agilent 7890A gas chromatograph (GC) interfaced with an Agilent 5975C Mass Selective Detector and an Agilent 7683B autoinjector. Analytical separation of the analytes was accomplished using a 30 m  0.25 mm (i.d.) DB-5 fused-silica capillary column (J&W Scientific, Folsom, CA). Column temperature was programmed from 50 (held for 1 min) to 190 at 40 °C min1, then to 220 at 3 °C min1, and finally to 300 at 30 °C min1. Selective ion monitoring (SIM) for target analytes (Table 1) was conducted for quantification, and a full scan run was performed to confirm the SIM analysis results. Compounds were identified based on mass spectra. Recovery efficiencies for the analytes were calculated based on the quantity of the surrogate standard recovered. Concentration of each compound was calculated based on the GC-MS response relative to that of the internal standard after correction for recovery efficiency, which averaged 79%. Any samples lower than these amounts were discarded and not used during the analysis. In all cases, samples were assumed to be free from contamination if samples that contained only solvents and reagents (i.e., blanks) produced chromatograms free of peaks. Bacterial Enumeration. E. coli and enterococci in sewage samples were enumerated following the standard cultivationbased modified mTEC agar method20 and mEI agar method,21 respectively. Briefly, 10-fold serial dilutions of each sewage sample were made, and 20 mL of appropriate dilutions (usually the 106 and 105 dilutions for E. coli and the 104 and 103 dilutions for Table 1. GC/MS-SIM Acquisition Parameters for Caffeine Analysis target compound

quantification ion

confirmatory ion

caffeine

194

109, 82

13

197

110

phenanthrene-d10

188

C3-caffeine

enterococci) were filtered through sterile 0.45-μm cellulose-ester membrane filters using vacuum filtration and placed on the modified mTEC agar plates (for E. coli) and the mEI agar plates (for enterococci). The modified mTEC agar plates were incubated at 37 °C for 2 h before being transferred to 44 °C for overnight incubation, and the mEI plates were incubated at 42 °C for 24 h before colony forming units (CFU) were counted. qPCR Quantification. Enterococci in sewage samples were also quantified using a qPCR method.22 Briefly, sewage samples were filtered through 0.45-μm cellulose ester membrane filters and subjected to DNA extraction using the UltraClean soil DNA isolation kit (Mo-Bio; Carlsbad, CA). The qPCR reactions include 1X TaqMan universal PCR master mix (ABI; Carlsbad, CA), 8 μM primers ECST784F and ENC854R, 4 μM TaqMan probe GPL813TQ, 0.2% BSA (by weight), and 1 μL of DNA template. The thermal cycler conditions included an initial denaturing at 95 °C for 10 min and subsequent 50 cycles of 95 °C for 15 s and 60 °C for 1 min. Absolute quantification of qPCR was achieved with five-point external calibration standard curves constructed using stationary-phase cells of Enterococcus faecalis ATCC 29212. Calibration curves were established for each batch of qPCR reactions, with r2 ranging from 0.962 to 0.998, and the concentrations were represented as calibrator cell equivalent per mL (CCE/mL). Data Analysis. The mass flux of sewage markers (mass/time) was calculated by multiplying concentrations of the sewage markers in the composite sewage sample by the average sewage flow within the sampling period. Bacterial measurements (in either CFU or CCE) were log transformed in the flux calculation based on the assumption of log-normal distribution. Linear correlations between time-series data were tested using Pearson’s coefficients (r) using Microsoft Excel with a statistical add-in (Statistixl, Australia). Autocorrelation coefficients of timeseries data were calculated based on a 1-day lag to indicate the daily reproducibility of the signal patterns of marker fluxes. Average marker fluxes within each of the 3-h time intervals and their standard deviations across different sampling days were calculated to indicate flux stability. Average marker flux differences and their 95% confidence intervals between the two dry-weather sampling periods, which flanks the wet-weather sampling periods, were calculated and plotted to indicate secular variations of marker fluxes. The default significance level in statistical tests was P e 0.05 unless otherwise stated. Precipitation data from the weather station at Lyon Arboretum operated by the National Ocean and Atmospheric Administration (NOAA) and streamflow data from the U.S. Geological Survey Kanewai station (USGS 16242500) were used in verifying the weather conditions during sampling periods. Stable Flux Model. A sewage marker with stable flux should have a flux pattern reproducible each day and a limited secular variation within the study period. For such markers, fluxes under dry weather conditions should be equal to fluxes under wetweather conditions (eq 1), where Cdry is the average marker concentration under dry-weather conditions, Qdry is the average sewage flow under dry-weather conditions, Cwet is the wetweather marker concentration under the influence of RDII, and Qwet is the wet-weather sewage flow with RDII contribution. The wet-weather sewage flow (Qwet) can be expressed as the summation of a dry-weather average flow component (Qdry) and the RDII contribution (QRDII) (eq 2), and the ratio between QRDII and Qdry is defined as RRDII (eq 3), which is an indicator of the severity of RDII and sewer deterioration. Independently, 8685

dx.doi.org/10.1021/es2019115 |Environ. Sci. Technol. 2011, 45, 8683–8690

Environmental Science & Technology

ARTICLE

RRDII can also be expressed as a function of sewage marker concentrations (eq 3). Cdry Qdry ¼ Cwet Qwet

ð1Þ

Qwet ¼ Qdry þ QRDII

ð2Þ

RRDII ¼

Cdry  Cwet QRDII ¼ Qdry Cwet

ð3Þ

’ RESULTS AND DISCUSSION Concentration Dynamics of Sewage Markers. Sewage flow and the five candidate sewage markers, including caffeine, TN, TSS, E. coli, and enterococci, were monitored in two DW periods that were devoid of significant precipitation in the Manoa sewershed (Figure 2). As expected, sewage flow in the DW periods exhibited a typical diurnal fluctuation with two peaks daily (Figure 2A), and a high level of similarity in sewage flow patterns between the two DW periods was detected (Pearson’s r = 0.80). The autocorrelation coefficients for sewage flow in the two DW periods were 0.71 and 0.83, indicating that the sewage flow under dry-weather conditions exhibited a strong daily pattern. The DW periods 1 and 2 had flow ranges of 16.552.3 L/s and 14.738.4 L/s, respectively. Concentration dynamics of the candidate sewage markers exhibited daily fluctuation patterns with different levels of reproducibility between the two DW periods (Figure 2BG). Among the physical and chemical candidate markers, TN concentration in sewage samples appeared to be the most stable, with a single major peak present in the beginning of each day (Figure 2C). The concentrations of caffeine (Figure 2B) and TSS (Figure 2D) also showed reproducible patterns, but with less regularity than that of TN. The observed concentration ranges for caffeine, TN, and TSS within the two DW periods were 5.0103.4 μg/L, 9.234.2 mg/L, and 18456 mg/L, respectively. Concentrations of the bacterial marker E. coli and enterococci in sewage samples were determined by cultivation-based membrane filtration methods (Figure 2EF). A DNA-based qPCR method, which affords higher throughput, was also evaluated for quantifying the enterococci marker in sewage samples (Figure 2G). Among the bacterial markers, enterococci by cultivation (Figure 2F) exhibited better diurnal repeatability than enterococci by qPCR (Figure 2G) and E. coli by culture (Figure 2E). The observed concentration ranges for enterococci (by cultivation), enterococci (by qPCR), and E. coli (by culture) were 4.96.3 log CFU/100 mL, 3.47.3 log CCE/100 mL, and 6.38.9 log CFU/100 mL, respectively. Overall, the bacterial markers exhibited larger fluctuation in concentration than the physicochemical sewage markers. Flux Stability of Sewage Markers. Mass fluxes of the candidate sewage markers in the two DW sampling periods were calculated to identify those markers with a stable flux. Autocorrelation coefficients of marker fluxes with a one-day lag indicate the reproducibility of flux patterns across different sampling days (Table 2). Among the candidate markers, TN flux exhibited the highest and most consistent regularity in both DW periods (autocorrelation r = 0.820.88). Significant autocorrelations were also observed for the fluxes of TSS, caffeine,

Figure 2. Dry-weather sewage flow (A) and concentration dynamics of caffeine (B), TN (C), TSS (D), E. coli by culture (E), enterococci by culture (F), and enterococci by qPCR (G) in Manoa sewershed. The two sampling weeks were Aug. 26Sept. 2, 2010 (b) and Feb. 1017, 2011 (O). The bright and shaded areas represent daytimes (8 a.m.8 p.m.) and nighttimes (8 p.m.8 a.m.).

Table 2. Autocorrelation Coefficients of Sewage Flow and Marker Fluxes in the Dry-Weather Sampling Periods correlation coefficients r (P value) parameter (units)

8686

DW period 1

DW period 2

flow (L/s)

0.71 (