Experience, Insights and Opportunities Using ZAPS Technologies
Transcription
Experience, Insights and Opportunities Using ZAPS Technologies
Experience, Insights and Opportunities Using ZAPS Technologies LiquID Station Dan Hanthorn WWRP Operations Supervisor City of Corvallis, Oregon Key Points Today • ZAPS Technologies, Inc. • What the LiquID Station does • How the LiquID Station works • LiquID Station applications • Testing and validation • Corvallis’ experience Chapter 1 ZAPS Technologies, Inc. About ZAPS ZAPS Technologies Inc. is based in Corvallis, Oregon, which has developed and patented the LiquID Station for continuous Real Time on line monitoring of a wide range of water quality parameters. The LiquID is based on over 25 years of research and development by Dr. Gary Klinkhammer, a preeminent global geochemist with over 25 years of experience in environmental monitoring and earth sciences. Detect. Respond. 5 ZAPS Technologies, Inc. • Founded and based in Corvallis, Oregon • Technology developed at Oregon State University • Roots in oceanography • US Navy and Defense Advanced Research Projects Agency (DARPA) funded 17 years of research and development • Brought to market for water, wastewater, industrial and environmental monitoring Chapter 2 ZAPS LiquID Station Liquid Station Parameters Detect. Respond. 9 Liquid Station Web Interface Liquid Station Corvallis Data Chapter 3 How it works • High Energy flash lamp • Fiber optic light path • 144 configurable optical filter pairings • Sensitive photon receptor How the Liquid Station Works • Zero Angle Photon Spectroscopy (ZAPS) • Optical based, solid state • Broad multi-spectrum (hyperspectral) • Deep UV • Visible • Infrared • 4 to 7 optical pairings per parameter • 400 individual tests per optical pair • No chemical or reagents How the Liquid Station Works Advanced Optical Analysis • Hybrid • Absorption, fluorescence, and reflectance measurements from a single optical platform • Hyperspectral • Deep UV, Visible, Infrared Z.-M. Zhang et al. sform technique to Raman spectral et al.[16] used a numerical method transform (DWT) for the analysis of ectroscopy measurements performed uld overcome the presence of noise and aused by light diffusion or fluorescence ison between derivative preprocessing ial baseline correction [Lieber and Leger and Ryder[17] found that neither e other, but the LMJ method simplifies preprocessed spectra. With the help filtration, Clupek et al.[18] treated the tively, but high computing power was ultiresolution advantages provided by ate the varying background of spectral eliminate the background in Raman o et al.[20] have developed an improved orescence removal based on modified hich used a peak-removal procedure d a statistical method to take the signal by some parameters [signal-to-noise ratio (SNR), ridge length, differences and etc.]. The paper is organized as follows. In next section, the principles of peak detection based on wavelet, wavelet derivatives calculation method for peak-width estimation, and penalized least squares for background fitting are presented. In the following section, our results are presented and discussed. Starting with two simulated data, we use them to demonstrate the intelligence of the proposed algorithm for correction of various backgrounds (slope, curve, etc.). Then, through applications to real Raman spectra, we demonstrate the validity of our algorithm. Finally, some conclusions and outlooks are drawn from this paper. How the Liquid Station Works • Thin slices of light probe the sample • Molecular bonds selectively react • Absorbed light detected Materials and Methods • Fluorescent light detected Peak detection Therelight are various criteria to detect peaks, such as SNR, intensity • Reflected detected threshold, slopes of peaks, local maximum, shape ratio, ridge lines, model-based criterion, peak width, etc.[21] In this study, we follow the SNR and ridge lines method proposed by Du et al.[22] Let us start with a brief introduction to continuous wavelet transformation (CWT). CWT is defined as the sum over all time of the signal multiplied by scaled, shifted versions of the wavelet function ψ. Mathematically, this process of CWT is represented as follows: • Photons counted – all of them wavelet mial fitting,[13,14,17,20] [4,12,17] d derivatives are three maund-correction algorithms in Raman widely used in Raman applications, hortcomings in certain aspects. For nomial fitting is not so effective and s on the user’s experience.[13] For the ng, the performance is poor in low o-background ratio environments.[20] change original peak shapes after the use difficulty in the interpretation of 7] Wavelet is a powerful tool in signal ound-correction algorithms based on m the signals into different frequency remove the varying low-frequency • Data processed on-board LiquID Station • Algorithm for each parameter C(a, b) = !+∞ −∞ 1 t−b s(t)ψ a,b (t)dt, ψ a,b (t) = √ ψ( ), a ∈ R+ , b ∈ R a a (1) where s(t) is the signal, a is the scale, and b is the shifting. Here ψ(t) is the mother wavelet, while ψ a,b (t) is the scaled and shifted one. The result C is a two-dimensional (2D) matrix of wavelet coefficients. Unlike the nonredundant, more efficient DWT, CWT allows How the Liquid Station Works 400 “Tests” Per Filter Pairing Absorbance Signal 400 Replicates 2 Minute Intervals Fluorescence Signal Biofouling Management Self Cleaning - Self Calibrating Signal strength recovered following manual clean and automatic self-calibration Starting signal strength Signal decay due to biofouling User defined Alert point at 50% signal loss Over 1600 automatic cleans and calibrations checks during the month without operator intervention Detect. Respond. Chapter 4 Applications LiquID Station Applications • Source water (drinking water) protection • Drinking water process monitoring • Wastewater process and effluent monitoring • Recycled/Reuse water applications • Wastewater pretreatment & source control • Treatment process modeling LiquID Station Applications • Food processing • Research • Industrial processes • Environmental monitoring and detection • Public Health applications • Agricultural irrigation monitoring Applications Sea Water Aquarium Sea water intake, multiple compounds monitored to ensure water quality (LiquID in shed) Pulp and Paper Mill Pre-treated effluent, primarily monitoring BOD load Agriculture Rinse water used to wash produce, primarily monitoring chlorine Research Vessel Sea water, multiple compounds monitored to map coastal water quality (LiquID in ship) 22 Chapter 5 Testing and Validation Validation Trials • 4 years of NELAC accredited lab comparisons • Third-party “blind” proficiency testing • Algorithm tuning • Build and replicate a 24-hr composite sample • Robust and reliable service record • Documented low O&M expense • Proven low maintenance history Validation Trials LiquID Brief – TOC Sensitivity Sensitivity andZAPS Accuracy October 2011 Calibration of the LiquID system used in the clean water test discussed above was performed using ZAPS – Yabby, our in-house designed and constructed automated reactor cell system. The Yabby system allows for precise aliquots of standard materials to be added to a continuously flowing sample at constant volume, resulting in an extremely well defined linear standard regression. The graph below is an example of a 30-point, 15 hour low-level TOC calibration experiment for potassium hydrogen phthalate (KHP) standard, further demonstrating the unmatched sensitivity and resolving power of the ZAPS LiquID. concentration TOC (mg/L) LiquID ultra-low TOC concentration calibration 1.2 1.0 0.8 0.6 R² = 0.998 2- measurement errors LOD = 0.015 (mg/L) 0.4 0.2 0.0 0.00 0.05 0.10 0.15 0.20 Absorption 0.25 0.30 0.35 0.40 BOD (mg/l) – ZAPS using HMA LiquID vs 5210B 600 500 400 300 200 100 0 0 100 200 300 400 BOD (mg/l) – standard method 5210B 500 600 LiquID Data vs Standard Method 5210B Installation Locations 7 Number of LiquID Stations 11 Grab Samples (N) 100 Correlation (R2) 0.94 Detect. Respond. 26 cBOD (mg/L) Regulatory Composites 350 300 250 200 150 100 50 0 12/1/2012 5/30/2013 Technique Municipal Lab ZAPS LiquID Report Type Daily Composite Daily average # of Days 78 (43%) 181 (100%) # of Readings 78 82,165 Average Value 145.2 156.7 Detect. Respond. 27 Chapter 6a Corvallis’ Experience About Corvallis Corvallis, Oregon WWTP • Population 55,000 • Combined sewers • 6 operators • One shift -10 hrs/day • Less than 1,000 kW per MGD w/Influent pumping • Landfill leachate • Oregon State University • Selective Trickling Filter/Selective Activated Sludge Corvallis Unit Processes Trickling Filters X 2 160’ X 8 161,000 cu. ft. each Speed controlled distributors TF Circulation Pumps X 2 11.5 MGD each 40 hp each Snail Sumps X 2 Aeration Basins X 2 75’ X 40’ X 15’ each Process Blower 35 hp Secondary Clarifiers X 2 115’ X 18’ Inboard Weirs Process Flow Diagram – STF/SAS Corvallis’ Experience - Matrices • 4 years monitoring with ZAPS LiquID Stations • Raw/Screened Influent • Primary Effluent • Aeration Basin MLSS • Secondary Clarifier effluent • Final Effluent Corvallis Process Flow Diagram Chapter 6c Corvallis’ Experience O&M and Data Capabilities Corvallis’ Experience LiquID Station O&M • Simple installation • No membranes, no chemicals, no reagents • Multi-parameters - simultaneously • Auto cleaning • Self calibrating • Very low maintenance • Infrequent manual clean • Station preventative maintenance 1/yr • Robust and reliable – think appliance 35 Corvallis’ Experience – Lab & Data • Eliminates sample degradation, handling errors and chain of custody • Continuous data – new values every 2 minutes • Real-time vs. composite data • Build a flow weighted composite sample • Diurnal impacts • Seasonal variations • Event detection Chapter 6d Corvallis’ Experience Event Detection Oregon State University Football Event kickoff tailgating half-time Oak Creek Event – E.coli Dairy Waste Spill Cheese Whey Discharge - cBOD Food Processor Discharge - cBOD Chapter 6e Corvallis’ Experience Selector TF/AS Process – Before ZAPS Process Control - BEFORE § Low energy requirements § Low maintenance § Effective snail control § Enhanced process control § Controlled sloughing – consistent TSS & BOD loading § Eliminated filter fly episodes § Improved “SRT” or biofilm persistence § Enhanced but uncontrolled nitrification & high Cl2 demand Trickling Filters BNR Selector Activated Sludge BNR Selector - BEFORE Challenge & Opportunities • Optimal nitrification mode • Partial nitrification Cl2 demand • Partial de-nitrification • Filaments • Carbon source control • TF odors Challenge & Opportunities (continued) • Optimize alkalinity recovery • Real-time vs. composite data • Diurnal impacts • Seasonal variations • Automated process control opportunities • Wastewater Master Planning projects Chapter 6f Corvallis’ Experience Selector TF/AS Process Understanding Diurnal Loading • Influent monitoring has led to the identification and tracking down of two significant source control problems - both were food processors with illicit discharges • Diurnal BOD range is significantly greater than diurnal flow • Very low BOD values at night Trickling Filter cBOD Loading Diurnal Loading • BOD removal rates in the primaries of up to 90% before 6:30 a.m. • Between 6:30 a.m. and 7:00 a.m. the removal efficiency drops to 10%. • Limited testing of chemically enhanced primary settling has demonstrated that night time removal is not complete - without the chemical addition. • Material is settling just enough to keep most BOD, TSS & E.coli inside the tanks at low flow • Material is carried over the launders in the morning with the slightest increase in plant flow. Diurnal Loading • The carryover results in a sudden and very high increase in loading to the biological processes • At 7:00 a.m. the highest loading rate of the day is experienced • Peak loading occurs 3 hours ahead of peak diurnal flow rates • The sudden change in loading ripples through the plant in a cascade of consequences…including increased E.coli and an increased demand for sodium hypochlorite Diurnal Loading of Effluent Ammonia vs. Nitrite+Nitrate Effluent Ammonia vs. Nitrite+Nitrate Impacts of Flow & Temperature 11.20C rain event Nitrification Affected by Flow and Temperature rain event Low pH 11.20C Trickling Filter Loading & Ventilation • TFs operated in series optimize and stabilize BOD and NH3 removal. • Compartmentalizing the process enhances the removal of both constituents and reduces the incidence of partial nitrification • Transient dips in TF BOD & NH3 removal were linked to insufficient TF ventilation from two separate cause/ effect relationships. • The most common is a sharp reduction in natural draft when the ambient air temperature is less than +/-2 degrees F from the process water temperature Trickling Filter Ventilation • Insufficient temperature differential to generate a draft - either up or down. • Most commonly this occurs as air temperature swings past the water temperature in the morning and evening and usually lasts less than an hour • May persist more than a day during times of "just right" atmospheric temperature stability. Trickling Filter De-nitrification Trickling Filter Ventilation • The other occurrences of insufficient ventilation is during a combination very warm weather with near 100% nitrification across the TFs. • Demand outstriped O2 provided by a strong natural draft • To the extent that anoxic conditions and denitrification are evident across the filters. Activated Sludge Loading • Soluble BOD is only about 1-2 mg/L following the TFs • Methanol and crude glycerol have been as a supplemental carbon source in the past • Using primary clarifier effluent, bypassed around the TFs • 15% bypass rate during the day; at night the bypass rate is nearly doubled due to the low BOD in the PE at low flow conditions. • A LiquID Station identified the optimal dose rate of PE VFAs Activated Sludge DO Control • Nitrification can be successfully sustained in the A/Bs during the night with much lower D.O. residuals than needed during the day • Within a reasonable D.O. range, maintaining a similar "oxidation pressure" (intensity X duration) at low flows does not inhibit the conversion of NH3. • A LiquID Station monitors the secondary effluent as well, used as a sentinel for BOD & TSS anomalies. Activated Sludge BNR Selector Optimization - AFTER Process Control - AFTER • Enhanced primary clarification and control • Improved process understanding • Process diagnosis • Blower control based on CBOD or NH3 • BNR selector optimization • Nitrification/de-nitrification chemical control – VFAs, methanol, glycerol • Disinfection optimization (hypo & UV) • Process modeling Chapter 6f Corvallis’ Experience Other Process Enhancements LiquID Station - Other Uses • Monitors final effluent E.coli • Began dosing hypo based on E.coli rather than Cl2 residual • Cut hypo and NaBs consumption by more than half • Use to provide continuous BOD & TSS data for modeling the existing secondary clarifiers • The MLSS Station and SE Station were used to document clarifier stress testing alongside the customary grab samples • Event monitoring and Pretreatment Program Source Control Prior To Disinfection Optimization High Cl2 Consumption & Cost Cl2 Initial Cl2 Residual Nabs Nabs Residual Initial Cl2 Dose = 1.9 – 2.1 mg/L Initial Cl2 Residual = 0.9 – 1.1 mg/L Final Cl2 Residual = 0.4 mg/L Nabs Dose = 0.7mg/L Nabs Residual = 0.3 mg/L Final Cl2 Residual Process Optimization Lower Initial Cl2 Demand Cl2 2 - Cl2 Feed 3 - Low Range Analyzer 1 – Real time E.coli data Initial Low-‐range (LR) Cl2 Residual 4 - Partial Nitrification Cl2 Nabs Final Cl2 LR Residual + ZAPS 5 – Hypo Degradation Nabs Residual Initial Cl2 Dose = 1.0 – 1.1 mg/L Initial Cl2 Residual = 0.2 mg/L Final Cl2 Residual = 0.0 mg/L Nabs Dose = 0.0mg/L Nabs Residual = 0.0 mg/L Process stabilization minimizes 1) partial nitrification, and 2) false-‐positive E.coli counts using approved test methods. 0 3/7/13 0:00 3/7/13 1:03 3/7/13 2:05 3/7/13 3:02 3/7/13 4:06 3/7/13 5:02 3/7/13 6:04 3/7/13 7:01 3/7/13 8:08 3/7/13 9:15 3/7/13 10:17 3/7/13 11:13 3/7/13 12:19 3/7/13 13:16 3/7/13 14:16 3/7/13 15:12 3/7/13 16:17 3/7/13 17:21 3/7/13 18:22 3/7/13 19:18 3/7/13 20:22 3/7/13 21:20 3/7/13 22:21 3/7/13 23:17 Reduced Chemical Consumption 120 100 80 60 40 20 ZAPS LiquID IDEXX Colilert-‐24 Hach ColiBlue Reduced Chemical Consumption Annual chemical savings are $75,000 or 62% of the former annual chemical budget; 15% of the M&S budget Effluent BOD is Strongly Tied to TSS Effluent TSS (floc carryover) Tied to Limited De-nitrification Future Process Enhancement Projects And Lessons Learned • Combined systems have complex interactions • Consider the insults • Use primary sludge fermentation to generate VFAs • Mixed liquor recycle at 3-4X • Trickling filter forced ventilation • If you are non measuring it, you can’t fix it Corvallis’ Experience - Savings • Lab testing • COD test reagent - $2,400/year • BOD – fewer bottles used • E.coli – IDEXX Quanti-Tray - $2,340/year • Lab labor down 33% • Chemical savings down 62% for Hypo and Sodium Bisulfite - $76,000 per year • Energy – reduced A/B dissolved oxygen - $22,500/ year • BNR Capital Project $50M 74 Common Questions • Are the same algorithms used in every LiquID Station? • Yes….and no • Are the results approved for compliance monitoring? • Pending for BOD, cBOD, COD, TSS….. • Phosphorus? • No – similar to most other metals • Other parameters? • Under development • raBOD (VFAs), bio-available iron, free Cl2, soluble parameters Common Questions • How much? • Standard configurations: water & environmental; wastewater; water reuse; industrial • Custom configurations available • Communications options • Cost to operate? • ~$0.30 per day – same as a 120 watt light bulb • Outputs? • web page • 4-20ma • Modbus Questions? Dan Hanthorn Operations Supervisor City of Corvallis, Oregon PO Box 1083, Corvallis Oregon 97339 dan.hanthorn@hotmail.com