The new set of health advisory levels (HALs) on multiple per- and polyfluoroalkyl substances (PFASs) released by the US EPA warrants a need for utilities to quickly design and implement PFAS separation technologies such as granular activated carbon (GAC). To this end, a tool was developed based on a set of predictive machine learning (ML) models coupled with an user-interface (UI) developed in Jupyter Notebook for improved access by users without coding knowledge. The tool can be used to estimate GAC bed life in a matter of seconds and select effective GACs for the removal of PFASs in a wide range of water types.  GAC adsorption is frequently considered to control recalcitrant organic micropollutants (MPs) such as in both drinking water and wastewater. To predict full-scale GAC adsorber performance, bench- and/or pilot- scale experiments are widely conducted, which in turn has generated a wealth of breakthrough data. Using such data, ML models were built from adsorbent, adsorbate, and background water matrix properties to predict MP breakthroughs. These models provide a simple and fast tool to predict GAC performance ; specifically the bed volumes of water that could be treated until MP breakthrough reached ten, twenty, thirty, forty and fifty percent of the influent MP concentration (BV10, BV20, BV30, BV40 and BV50). Although mechanistic models, as well as simpler data driven linear regression models were developed by others, these models either still rely on physical experiments or cannot provide accurate predictions. The data set on which the ML models were built is the most comprehensive database to date ;it encompassed 60 MPs [including pharmaceuticals, volatile organic contaminants, PFASs], 3 GAC types by base material (20 unique GAC products), and 49 water matrices (groundwater, surface water, and treated wastewater). Approximately 620 data points were split into training, validation, and test sets. Features such as MP properties, background water matrix characteristics, and GAC properties, were explored in ML models to predict log-10-transformed BV10 (logBV10) through BV50 (logBV50). Mean absolute errors on test set predictions of BV10 through BV50 averaged between 3000 to 5000 bed volumes. To put into perspective, this translates to 42 to 70 days of error assuming an empty bed contact time of 20 minutes. Focused application of such ML models on the prediction of PFAS removal by GAC was demonstrated via predictions on multiple sets of pilot studies, including a set of recently completed GAC pilot at a North Carolina utility.

• Cost is $30 per attendee.

Webinar Information and Requirements:

Webinar Technical Requirements

• NC One Water will provide each attendee with the link to access the webinar and call-in audio instructions. NC One Water is not responsible for technical support before, during or after the webinar.
• Attendees must ensure that they have the needed equipment, that it is functioning, that it is compatible with the webinar system and that they know how to use it.
• Refunds will be handled according to NC One Water’s current seminar refund policy and will not be issued for attendees that experience technical or equipment failures during the webinar.
• NC One Water webinars are conducted using Zoom. It is highly recommended that you test your computers compatibility before planning to attend a webinar. How-to videos and additional Zoom resources can be found here: https://learn-zoom.us/show-me.
• To participate in the webinar, you must have access to a computer with an internet connection AND an audio connection (phone line or VOIP connection). It is recommended that you participate from a semi-private, quite location where there will be few distractions.

Webinar Continuing Education Units Requirements

• Attendees must be on camera at all times during the presentation. Visibility and participation will be monitored. Failure to participate and be visible will result in no CEUs being given.
• Attendees must pre-register and provide a valid certification numbers issued by the NCWPCSOCC, NCWTFOB or NC Professional Engineers.
• Attendees must be logged in at the beginning and must remain logged in through the end of the session. Attendees must promptly respond to brief quizzes that appear randomly throughout the webinar.
• Each attendee must be logged into a separate computer so that their participation may be accurately tracked.
• No partial credit will be given.
• After the webinar, NC One Water staff will email CEU documentation/certificates to attendees meeting the CEU requirements. Reports will also be sent to the appropriate certification boards.
• NC One Water will report all cases of fraudulent attendance to the appropriate licensing boards.
• Inability to complete CEU requirements due to technical difficulties will result in no CEU’s being issued.