Detroit will utilize predictive modeling used in Flint to find its 80,000 lead service lines
Detroit — The same technology used to find lead pipes in Flint will be used in Detroit to reduce the amount of excavation in the search for service lines by more than 95%, city water and sewerage officials announced Thursday.
The Detroit Water and Sewerage Department will save an estimated $165 million by using Ann Arbor-based startup BlueConduit's predictive modeling to create a lead service line inventory in the city. The city faces a January 2025 deadline to give the inventory to state environmental officials.
The department said it will only have to tap 384 stop boxes — the valve attached to the service line to bring treated drinking water from Detroit's water system to homes and businesses — instead of more than 300,000.
The data collected from those 384 digs, along with property and permit data, will be included in BlueConduit software to provide a report of the probable locations and amount of lead service lines in the city.
Officials said the expense of excavating all 300,000 water services lines to verify whether the pipes contain lead would have been passed on to water customers through a rate increase if not for the software and $254,000 in grants from the Rockefeller Foundation, the Kresge Foundation and the Michigan Department of Environment, Great Lakes & Energy.
The city's $450 million lead service line replacement program is the most robust in America, said Palencia Mobley, DWSD deputy director and chief engineer. Based on previous estimates, the department projected Detroit has 80,000 lead services lines.
Since 2018, DWSD has been replacing service lines as water mains have been replaced. By the end of 2020, DWSD replaced 1,155 lead service lines and projects an additional 1,200 pipes will be replaced in the next two years.
During a Thursday virtual press conference, Mobley said every decision they make keeps affordability in mind.
"This partnership with BlueConduit and the funding...allows us to have mapping of probable lead service line locations for planning and regulatory reporting, without digging up every single service line in the city of Detroit which would have likely increased water rates to pay for the work," Mobley said.
The technology startup is co-founded by University of Michigan professor Eric Schwartz and Georgia Tech professor Jacob Abernethy. They said their BlueConduit platform was successfully used in Flint and has since been used in more than 50 cities across the country. The accuracy rate is about 85% to 90%, he said.
"We're going to do our absolute best...in an aim to reduce time Detroit residents are living with lead, that's the goal," Schwartz said during the press conference. "We will help find lead in the city by finding the lead pipes and directing the crews and contractors to go to the areas where they would be most likely lead."
Not every pipe could get flagged. However, DWSD spokesman Bryan Peckinpaugh said the city will have to eventually expose each service line to ensure that when water mains are replaced, service lines are only connected to copper piping.
The program will help tell replacement crews where to go next, Schwartz said.
"We know that it is the case that older homes are more likely to have lead pipes and where in the city they may be. We can know the size, the data of the home and once we have enough information to the point we know the material underground, we can connect and find patterns to...provide the best guess. That's all these models are doing," Schwartz said.
Michigan's Lead & Copper Rule, which was enacted in June 2019, requires water utilities to replace all lead service lines within the next 20 years.
Mobley said Michigan's rule is the most stringent in the country. It also requires utilities and municipalities to provide a Complete Distribution Systems Material Inventory by January 2025. EGLE is encouraging water providers to use predictive modeling to meet the goal.
"This is an unfunded mandate. We hope this system will allow us to do more," Mobley said.