The Flint Water Treatment staff and their consultants struggled to meet the Safe Drinking Water Act levels at the water treatment plant. Then residents noticed unpleasant changes in the smell, color, and taste of the water coming out of their taps. Tests showed high levels of bacteria that forced the city to issue boil advisories. In response, the city upped its chlorine levels to kill the pathogens. This created too many disinfectant byproducts, which are carcinogens and corrosive. Then the corrosive water began leaching lead, other metals and whatever else was in the biofilm on the old pipes into the water in the homes.
Flint’s water department might have been able to avert this disaster by having a corrosion management plan and using additives to diminish the corrosiveness of the water at a negligible cost, but there was an underlying problem that effects not only Flint-lead service lines- the lines that connect homes to the city water system.
By 2016 replacing the service lines became a top priority for the City of Flint. Though, technically, these lines are owned by the homeowner, the Michigan state legislature appropriated $27 million towards the expense of replacing these lines and later the U.S. Congress allocated almost another $100 million for Flint.
With money in hand the problems became that the City of Flint did not know how many lead service lines existed and where they were located. Service line information are theoretically documented during original construction or renovation in the building department files, but those records were found to be incomplete or lost.
Because digging up an entire water service line pipe under a resident’s year cost thousands of dollars, unnecessarily digging up a line that turns out not to be a lead service line is wasteful and the city did not have money to waste. Flint believed there might be as many as 50,000 homes potentially needing service line replacements, the city was facing costs up to $250 million.
Google believed data science could expedite the process of identifying specific homes in need of replacement service lines and donated to the University of Michigan and Flint’s Community Foundation. Dr. Jacob Abernethy, then an assistant professor at University of Michigan and now with Georgia Tech, and his Data Science Team of students were called in. With the help of Captricity.com who digitized a set of over 100,000 index cards, water sampling data, and hand-annotated maps digitized by the University of Michigan-Flint GIS Center, Dr. Abernethy and his team were able to analyze the available data to provide statistical and algorithmic support to guide decision making and data gathering.
By 2016 replacing the service lines became a top priority for the City of Flint. Though, technically, these lines are owned by the homeowner, the Michigan state legislature appropriated $27 million towards the expense of replacing these lines and later the U.S. Congress allocated almost another $100 million for Flint.
With money in hand the problems became that the City of Flint did not know how many lead service lines existed and where they were located. Service line information are theoretically documented during original construction or renovation in the building department files, but those records were found to be incomplete or lost.
Because digging up an entire water service line pipe under a resident’s year cost thousands of dollars, unnecessarily digging up a line that turns out not to be a lead service line is wasteful and the city did not have money to waste. Flint believed there might be as many as 50,000 homes potentially needing service line replacements, the city was facing costs up to $250 million.
Google believed data science could expedite the process of identifying specific homes in need of replacement service lines and donated to the University of Michigan and Flint’s Community Foundation. Dr. Jacob Abernethy, then an assistant professor at University of Michigan and now with Georgia Tech, and his Data Science Team of students were called in. With the help of Captricity.com who digitized a set of over 100,000 index cards, water sampling data, and hand-annotated maps digitized by the University of Michigan-Flint GIS Center, Dr. Abernethy and his team were able to analyze the available data to provide statistical and algorithmic support to guide decision making and data gathering.
Dr. Abernethy and his team developed a machine learning program and statistics to accurately estimate the locations of home needing lead service line replacement as well as those with safe pipes. Their predictive model was found to be empirically accurate for estimating whether a Flint home’s pipes were safe/unsafe with an accuracy rate of 97%. The team estimated that this increase in accuracy would save the City of Flint about $10 million which could be used to replace the pipes at an additional 2,000 homes. This model can be used in other cities to help them identify and replace their lead service lines most cost effectively.
To read the full article:
Jacob Abernethy, Cyrus Anderson, Chengyu Dai, Arya Farahi, Linh Nguyen, AdamRauh, EricSchwartz, WenboShen, GuangshaShi, JonathanStroud,etal. 2018. ActiveRemediation: The Search for Lead Pipes in Flint, Michigan. arXiv:1806.10692v2 [cs.LG] August 2018.
To read the full article:
Jacob Abernethy, Cyrus Anderson, Chengyu Dai, Arya Farahi, Linh Nguyen, AdamRauh, EricSchwartz, WenboShen, GuangshaShi, JonathanStroud,etal. 2018. ActiveRemediation: The Search for Lead Pipes in Flint, Michigan. arXiv:1806.10692v2 [cs.LG] August 2018.
No comments:
Post a Comment