top of page
  • Annie Vanrenterghem

Annie Vanrenterghem, PhD, infraPLAN CEO will present 3 papers at ACE2023, Toronto

Contact us at info@infraplan-llc.com


Tuesday June 13, 2023 - 9:00 AM

TUE05 - Advancement in Desktop Condition and Risk Assessment for Water Mains

“Unraveling the Mystery of a Skyrocketing Break Rate with Data-Driven Watermain Break Analytics”

The City of Huber Heights operates a water distribution system of 200 miles of predominantly cast iron and ductile iron pipes. While the system was already experiencing a relatively high break rate before 2019, the break rate from 2019 through 2021 doubled compared to what it was on average over the past 10 years. The City had to first solve the mystery of what caused the spike in breaks, not only to remedy the spike, but also to ensure that a machine learning-powered break prediction model not be skewed by that spike. This project shows 1. the danger of applying machine learning blindly without fully understanding the operational context, 2. the value of a robust statistical analysis (to gain that understanding), and 3. the capacity for a prediction model to generate valid results on a rather small system.

Speakers: Kevin Campanella, PE, Utility Planning Lead, Burgess & Niple; Russ Bergman, Engineer, Huber Heights; Annie Vanrenterghem Raven


Tuesday June 13, 2023 - 10:00 AM

TUE05 - Advancement in Desktop Condition and Risk Assessment for Water Mains

“Lessons Learned from Using Machine Learning to Predict Breaks”

By now, most water industry professionals responsible for the R&R planning of their aging distribution systems have been exposed to the use of machine learning to predict breaks with various levels of success or trust. In this presentation we will share the lessons learned from our experience developing and applying machine learning-powered break prediction models. We will explain how that experience has been enriched by our deep knowledge of advanced multi-covariates regression models has helped us understand the data at stake as well as the specific features that type of modeling requires. Finally, we will describe the features that will increase your level of comfort when embarking on the machine learning journey.

Speakers: Annie Vanrenterghem Raven; Kevin Campanella, PE, Utility Planning Lead, Burgess & Niple


Wednesday June 14, 2023 - 2:30 PM

WED23 - Pipeline Asset Management

“Using Machine Learning to design an Optimal Plan of R&R for Baltimore County Water Mains”

With a current break rate of 0.234 breaks/mi/yr. and a yearly R&R rate of approximately 0.3%, Baltimore County, recognizes that it must increase its R&R investment for its close to 2,300 miles of water mains. A machine-learning-powered model applied to design an R&R plan, predicted breaks with exceptional validation (10% of the length of mains with the worse LOF as of 2020 experienced 86.7% of the 2021-2022 breaks; a performance four (4) times better than with the scoring approach currently used). This plan will meet objectives and constraints while increasing the level of confidence; ultimately achieving more with less. We will also show how our model can be integrated within the management tools the County uses to make its R&R decisions.

Speakers: Annie Vanrenterghem Raven; Erin McKenna-Streyle, PE, Baltimore County. Co-author: Bola Fashokun, PE, CFM, Assistant Vice President, WSP


Questions, comments, contact Annie at avanraven@infraplan-llc.com












Comments


bottom of page