Water Treatment Modeling Tools for Removing PFAS and other ...
Transcript of Water Treatment Modeling Tools for Removing PFAS and other ...
Water Treatment Modeling Tools for Removing PFAS and other Contaminants
Thomas Spetha, Jonathan Burkhardta, David Handb, Rajiv Kheraa, Pat Ransomc, Craig Pattersona, and David Wahmana
a U.S. Environmental Protection Agencyb Michigan Technological University
C Abt Associates
Water Research Webinar SeriesApril 29, 2020epa.gov/water-research/water-research-webinars
Disclaimer
The views expressed in this presentation are those of the individual authors and do not necessarily reflect the views and policies of the US EPA. Mention of trade names or commercial products does not constitute endorsement or recommendation for use
Overview
Topics to Cover1) Background 2) Modeling programs available on EPA’s website3) Demonstration on how EPA is utilizing the GAC models
1) Relevant background 2) Types of outputs useful to communities
4) Future of the models1) Availability2) Development of existing models3) Development of new models
3
EPA’s PFAS Drinking Water Research
Problem: Utilities lack treatment technology cost data for PFAS removalAction:
• Gather performance and cost data from available sources (DOD, utilities, industry, etc.)• Conduct EPA research on performance of treatment technologies including home
treatment systems• Update EPA’s Treatability Database and Unit Cost Models • Connect EPA’s Treatability Database to EPA’s Unit Cost Models for ease of operation• Model performance and cost, and then extrapolate to other scenarios
• Variable source waters• Variable PFAS concentrations in source water• Different reactivation/disposal options• Document secondary benefits• Address treatment impact on corrosion
• Evaluate reactivation of granular activated carbon Impact: Enable utilities to make informed decisions about cost-effective
treatment strategies for removing PFAS from drinking water4
Modeling Agreement
ETDOT Modeling ToolA number of years ago, a series of treatment models, data sets, and parameter estimation tools were put together under the moniker of the Environmental Technologies Design Option Tool (ETDOT) at Michigan Technological University (MTU). ETDOT was developed by National Center for Clean Industrial and Treatment Technologies.
Members of the ETDOT team were David Hand, John Crittenden, David Hokanson, Eric Oman, Tony Rogers, Alex Mayer, Jim Mihelcic, and numerous graduate students and collaborators through the years.
The models were sold as a package for numerous years. In 2019, EPA signed an agreement with MTU to make this suite of water and air treatment models available to the public at no cost.
5
Expected interest
ETDOT Modeling ToolIt is expected that these established and new models will be useful to the following entities
• Consulting engineers• University academicians• Water utilities with experience in running models
The models can be currently found through USEPA’s website…https://www.epa.gov/water-research/environmental-technologies-design-option-tool-etdot
That site will lead to a GitHub site where the models can be found6
Available Modeling Tools
7
Available Modeling Tools2
Models available at the GitHub site…Adsorption Design Software for Windows (AdDesignS) Version 1.0Advanced Oxidation Process Software (AdOx) Version 1.0.2Aeration System Analysis Program (ASAP) Version 1.0Biofilter Design Software Version 1.0.27Continuous Flow Pore Surface Diffusion Model for Modeling Powdered Activated Carbon Adsorption Version 1.0Dye Study Program (DyeStudy) Version 1.0.0Predictive Software for the Fate of Volatile Organics in Municipal Wastewater Treatment Plants (FaVOr) Version 1.0.11Ion Exchange Design Software (IonExDesign) Version 1.0.0Software to Estimate Physical Properties (StEPP) Version 1.0
The engines are written in FORTRAN with a Visual Basic front end.
https://www.epa.gov/water-research/environmental-technologies-design-option-tool-etdot 8
GAC Column ModelingAdDesigns Package
• Equilibrium column model (ECM)• Used to predict simple breakthrough in multi-component mixtures (Ideal adsorbed solution theory - IAST)• Ignores mass transfer resistance (vertical breakthrough profiles)
• Constant Pattern Diffusion Model (CPDM)• Used for single solute breakthrough when computational time is at a premium • For favorably adsorbed compounds, the mass transfer zone will increase until it attains a constant shape• Assumes that the bed length is long enough and loading rate is low enough for the mass transfer wave to become
constant
• Pore Surface Diffusion Model (PSDM)• Used for the most complex GAC modeling efforts• Multicomponent mixtures (IAST)• Accounts for external diffusion resistance, pore diffusion, surface diffusion• Can incorporate NOM pre-loading or GAC fouling mechanisms (select fouling parameters are provided)
Because processing speeds have increased tremendously in recent decades, there is less need for the ECM or CPDM; however, there are times when they come in handy.
9
Adsorption Mechanisms
Film transfer resistance through stagnant boundary layerFlow
Surface diffusion
Pore diffusion
10
Model InputsPore Surface Surface Diffusion Model
• Bed Length • Adsorber diameter• Mass of GAC in the bed• Flowrate or Empty Bed Contact Time (EBCT)• Total simulation time• Time at which first data point is displayed• Time step between two consecutive points• Apparent density of the adsorbent• Average particle radius• Porosity of the adsorbent• Particle shape factor (typically set to 1.0)• Number axial collocation points• Number of radial collocation points• Number of axial elements
AdDesigns package contains a database of GAC properties
Default values are provided. Greater numbers will increase run time, but typically add definition and decrease instabilities
Film transfer
11
Model Inputs (continued)Pore Surface Diffusion Model (continued)
• Contaminant molecular weight• Contaminant concentration (can be variable)• Contaminant Freundlich K and 1/n• Film mass-transfer coefficient
Gnielinski correlation provided: molar volume of contaminant (MW/density), water viscosity, liquid density, bed void fraction, interstitial velocity,
• Surface diffusion coefficientSontheimer correlation provided: influent concentration Co, molar volume of contaminant (MW/density), water viscosity, solid-phase concentration, void fraction, apparent density, tortuosity
• Pore diffusion coefficient From liquid diffusivity coefficient (Hayduk and Laudie) plus tortuosity: molar volume of contaminant (MW/density), water viscosity, tortuosity
• NOM preloading fouling adjustment • Must choose a fouling correlation (factors for six waters are provided),
must choose a fouling correlations (factors for eight contaminants provided, factors for 16 compounds can be found in Magnuson and Speth but are not available in AdDesignS yet), andmust click the “Use Pore Diff. Corr. For Tortuosity” button in Kinetics page
#
AdDesignS package contains a database of contaminant properties including K and 1/n and correlation packages for estimating K and 1/n. See also EPA’s Treatability Database.
12
Treatment Information
Publicly Available Drinking-Water Treatability Database• Interactive literature review database that contains over
88 regulated and unregulated contaminants and covers 34 treatment processes commonly employed or known to be effective (thousands of sources assembled on one site)
Currently available:
• PFOA, PFOS, PFTriA, PFDoA, PFUnA, PFDA, PFNA, PFHpA, PFHxA, PFPeA, PFBA, PFDS, PFHpS, PFHxS, PFBA, PFBS, PFOSA, FtS 8:2, FtS 6:2, N-EtFOSAA, N-MeFOSAA, and GenX
Contains GAC information including Freundlich K and 1/n values
https://www.epa.gov/water-research/drinking-water-treatability-database-tdbSearch: EPA TDB
13
Treatability Database
Agency Landing Page Database Homepage
14
https://www.epa.gov/water-research/drinking-water-treatability-database-tdbSearch: EPA TDB
Data Needs for Fitting Field Data• Influent concentrations:
• PFASs (both removal targets and other PFASs)• Other competing contaminants (sulfate, nitrate)• Influent water characteristics (TOC)
• Study scale (RSSCT, pilot, full-scale)• Type of GAC/resin tested• Definition of breakthrough associated with a given bed life value. For example, does
“breakthrough” mean:• Initial detection (and at what detection limit)• A specified target PFAS concentration (e.g., 70 ppt)• Percent breakthrough (e.g., 10% breakthrough/90% removal)
• Raw data (concentration vs time/bed volumes) are useful
15
Applying the PSDM Mode
• Fitting Pilot-/Full-scale Data • Predicting Results for Consistent Design
• Allows for comparison across technologies by cost
• Allows for Predicting other Scenarios• Other designs: number of contactors,
contactor EBCTs, different treatment goals, etc.
• Other influent conditions: Changing concentrations of PFAS or background constituents, changing demand, etc.
16
Fitting other Waters (PFOA)
0
5
10
15
20
0 500 1000 1500
PFO
A Co
ncen
trat
ion
(ng/
L)
Time (days)
Calgon ACT Test - Plainfield Twp., MI GroundwaterCarbon Type: F400 (12X40) Mesh GAC
Perfluorooctanoic Acid = 21 ng/L
Scaled ACT Column Data
PSDM Model Fit
Fitting other waters help with increasing confidence in model predictions
• Allows comparison of the capacity and kinetic parameter assumptions across contaminants and waters needed to obtain the fits
17
Fitting other Waters (PFAS)
0
0.5
1
1.5
2
2.5
3
3.5
0 200 400 600 800 1000
PFO
S Co
ncen
trat
ion
(ng/
L)
Time (days)
Calgon ACT Test - Plainfield Twp., MI GroundwaterCarbon Type: F400 (12X40) Mesh GACPerfluorooctanesulfonic Acid = 18 ng/L
Scaled ACT Column Data
PFOS Model Fit
Fitting other waters help with increasing confidence in model predictions
• Allows comparison of the capacity and kinetic parameter assumptions across contaminants and waters needed to obtain the fits
18
Predictions Allow for Design Evaluation
0
50
100
150
200
250
0 5 10 15 20
Spec
ific
Thro
ughp
ut (m
^3/k
g)
EBCT (minutes)
Relationship Between Specific Throughput and EBCT
PFOA
PFOS
PFBS
PFHpA
PFHxA
PFPeA
PFBA
Co = 10 ng/L, Cto = 2 ng/LExample: EBCT Choice
• Can evaluate EBCT for various PFAS for different PFAS
19
Predictions Allow for Design Evaluation1
0
20
40
60
80
100
120
0 5 10 15 20
Spec
ific
Thro
ughp
ut (m
^3/k
g)
EBCT (minutes)
Relationship Between Specific Throughput and EBCT
PFOA
PFOS
PFBS
PFHpA
PFHxA
PFPeA
PFBA
Co = 100 ng/L, Cto = 2 ng/L Example: EBCT Choice• Can evaluate EBCT for
various PFAS for different PFAS at different influent concentrations, effluent goals, etc.
20
Predictions Allow for Design Evaluation2
Example: Bed Replacement • Can evaluate how many and how often beds need to be brought online or replaced.
21
Predictions Allow for Design Evaluation3
Example: Bed Replacement • Can evaluate how many and how often beds need to be brought online or replaced.
22
Impact of Reducing Feed Concentration by Half
0
20
40
60
80
100
120
140
160
180
0 50 100 150 200 250 300
Conc
entr
atio
n (p
pt)
Time (days)1/2 Initial Conc. (C1-F400) Sum - EBCT20 (C1-F400) 1/2 Initial Conc. (C2-F300)
Sum - EBCT20 (C2-F300) 1/2 Initial Conc. (C3-PF-F400) Sum - EBCT20 (C3-PF-F400)
1/2 Initial Conc. (C4-PF-AC1230) Sum - EBCT20 (C4-PF-AC1230)
Cost Modeling: Additional Data Needs
• As-built costs:• Installed equipment cost• System engineering and other indirect cost• Annual operating cost
• The more detail, the better:• Detailed breakdown of cost by line item• Total cost with list of categories included, for example:
o “Equipment includes vessels, piping, valves, instrumentation, concrete pad, buildings”o “Indirect includes engineering, permitting, pilot testing, site work, mobilization”o “Operating cost includes media replacement, labor, electricity”
• Total only
• Associated flow rates, vessel sizes, materials of construction for major components (e.g., stainless steel, fiberglass)
24
EPA‘s Drinking Water Cost Models
• Adsorptive media• Anion exchange*• Biological treatment*• Cation exchange• GAC*• Greensand filtration• Microfiltration /
ultrafiltration• Multi-stage bubble
aeration*
• Non-treatment• Packed tower
aeration • POU/POE#
• Reverse Osmosis / Nanofiltration
• UV disinfection• UV Advanced
Oxidation
* Search: EPA WBS http://www2.epa.gov/dwregdev/drinking-water-treatment-technology-unit-cost-models-and-overview-technologies
# For POU/POE: Temporarily taken off web: Please contact Rajiv Khera at [email protected]
25
Costs for PFAS Treatment: One GAC Example
Primary Assumptions:• Two vessels in series• 20 min EBCT Total• Bed Volumes Fed
1,1-DCA = 5,560 (7.5 min EBCT)Shorter Chain PFCA = 4,700Gen-X = 7,100Shorter Chain PFS = 11,400PFOA = 31,000PFOS = 45,000
• 7 % Discount rate• Mid Level Cost
26
Future Plans
For the models housed at the GitHub site…
• Update GUIs to work with Windows 10• Provide Python code for PSDM GAC model to automate the optimization routines shown here
• Specific throughput and carbon use rate calculations for multiple scenarios• Automated fitting of parameters• Automated optimal bed configuration• Automated optimal EBCT selection• Automated evaluation of bed replacement frequency• Evaluation of multiple feed conditions• Evaluation of multiple flow conditions
• Provide code for ion exchange models for• Include competition (e.g., inorganic ions, and PFAS)• Continuous flow (columns) and batch (isotherm and kinetics)• Gel (HSDM) and macroporous (PSDM) resins• with automation features
https://www.epa.gov/water-research/environmental-technologies-design-option-tool-etdot 27
Ultimate Goal
For the treatment/cost models housed at EPA…
• Provide tools and approaches to accurately predict the performance and cost of treating PFAS in waters
ETDOT Models: https://www.epa.gov/water-research/environmental-technologies-design-option-tool-etdotTreatability Database: Search EPA TDB or go to https://www.epa.gov/water-research/drinking-water-treatability-database-tdbCost models: Search EPA WBS or go to http://www2.epa.gov/dwregdev/drinking-water-treatment-technology-unit-cost-models-and-overview-technologies
28