My analysis showed that water mains broke predominantly during winter months (December, January, February), leading to higher total repair costs during these months as well. Water main breaks across mostly involved cast iron pipes. Older water mains (~60 years or older) used mostly cast iron, while younger water mains (~60 years or younger) used mostly ductile iron. Based on these findings and my knowledge about the topic from the semester, I provided 2 actionable recommendations: Switch to Ductile Iron-based Water Mains & Establish Yearly Maintenance Plans. 

Even though my analysis and recommendations were specific to Ann Arbor, the background knowledge I obtained from developing my project is fundamental to tackling issues outside of our city -- This includes current legislation gaps, corrosive materials, and the fragmented nature of national infrastructure. Despite this theme not being my central interest, I was grateful for the opportunity to step outside my comfort zone and become more informed about such a prevalent issue, all while advancing my technical skills. As students, our impact begins with these projects that allow us to start small and grow into national and/or global initiatives.

View the final project report here.

This project was my first experience working on a long-form, data-driven project, so I truly learned so much. Not only was I exposed to the water issues around Ann Arbor that I was blind to before, but working on this project strengthened my grasp on coding and taught me how to use analytical insights to help mitigate real issues. My project focused on stormwater in Ann Arbor and whether it increased Nitrate or E. Coli levels in local waterways or drinking water to a dangerous degree. 

Through my research, I found statistically significant weak but positive correlations between the amount of stormwater and the levels of Nitrate/E.Coli in Barton Pond. Adding onto this initial insight, it appeared that as the average monthly amount of E. coli found in Barton Pond increased, the monthly average amount of E. coli found in the drinking water intake increased as well. These outcomes could support further investment into the Ann Arbor Water treatment systems, ensuring clean and healthy drinking water for all residents. Hopefully, these results will also encourage Ann Arbor residents to reflect on whether their living and buying practices could harm nearby water if caught in stormwater. Whether it's cleaning up pet waste or swapping out household items like fertilizer, there are ways for citizens to live more eco-friendly lives.

View the final project report here.

The outcome of the project led to two conclusions for the City of Ann Arbor. The first one was that when building future water infrastructure, the city should build shorter watermains around 250 meters in length. The second recommendation was to build future watermains with diameters close to 7 inches. Based on an extensive analysis based on each attribute and in the context of optimizing different variables, I was able to see which attributes had strongest correlations.

These findings can be helpful to city staff as they incorporate the number of breakages and severity levels with the cost of repair. To optimize the repair costs of individual watermain breaks, the City of Ann Arbor should consider longer watermains around 700 meters over shorter watermains around 100-200 meters even though longer watermain breaks around 700 meters are more frequent. This finding aligns with expectations because shorter watermains are often found in proximity to infrastructure limiting access to the site. From a cost perspective, having longer watermains may be more ideal. However, this may change based on labor costs as sending crews more frequently to fix watermains may be more expensive. Additionally, the city should further investigate the materials they use in watermains. Pipes made out of ductile iron cost more on average to repair than cast iron (Figure 5). This plot conflicts with expectations of the city staff, so this should be further investigated. Considering the severity and cost, lower severity repairs are associated with lower costs and vice-versa (Figure 6). This aligns with expectations as less severe breakages may be less urgent and require fewer materials and labor. When optimizing the cost of a watermain break, changing the length will likely be most effective as there is some conflicting data with diameter size. 

View the final project report here.

The quantitative measures obtained from the research showed the material and subtype that had the highest ratio of breaks. The findings aligned with my initial expectations to some extent. The identification of cast iron and ductile iron as materials with the highets number of broken pipes was expected, given their prevalence in the dataset. However, after adjustments made by calculating the ratio of broken pipes to total pipes for each material, the data revealed that steel had the highest ratio. Unfortunately, the limited number of installed steel pipes did not provide sufficient data reliability and variability, which caused me to shift my focus towards the second highest ratio, cast iron. The next analysis showed that Subtype 1 had the highest ratio of broken pipes for cast iron. These findings can help stakeholders and the city in reevaluating the material selection policies. Despite having the most amount of installed pipes, ductile iron had the second to last highest ratio. Policymakers might consider focusing on ductile iron as the main material for these watermains going forward, as well as allocate resources to focus on materials and subtypes with higher vulnerability. This could include prioritizing inspections, repairs, and replacement for watermains of these attributes.

View the final project report here.

I learned a great deal about the interactions between watermain materials and watermain features such as age, dimensions, and pressure zones and their correlations to watermain breaks in the Ann Arbor water infrastructure system. My findings revealed that cast iron mains are particularly susceptible to breaks, while steel main breaks are more influenced by age, dimensions, and pressure zones than other materials. Additionally, PVC mains have the lowest break rate across all materials, especially compared to steel and cast iron mains, which have the highest break rates. Lastly, the variables age, dimension, and pressure zone do not seem to affect PVC and contribute to breakages in the same way they do for other materials. These insights offer valuable guidance for prioritizing repairs and optimizing conditions for different watermain materials in future infrastructure enhancements.

Aside from my findings, this project offered me valuable experience in applying data science to a real-world scenario. I specifically learned that cleaning data requires a creative approach and a lot of patience. Overall, this course and project taught me that data science is so much more than coding graphs and charts. It is a role that encompasses client communication, thorough planning, collaborative teamwork, constructive critique, and so much more.

View the final project report here.

Through this project, I found that older water mains, longer water mains, and water mains made out of cast iron were more likely to break, which could help the City of Ann Arbor in making decisions about replacing or installing water mains. These findings would benefit the residents of Ann Arbor by keeping water service running and ensuring good water quality, while also helping the city in managing its budget. Personally, this project provided an excellent opportunity to develop my data analysis skills while using these skills to attack an issue that has real-world implications.

View the final project report here.