Welcome back to #TechTalk. In our last entry, we focused on Norfolk Southern’s use of data analytics to more intelligently forecast the wear and tear on our tracks. This week, we’re focusing on another cutting-edge application of big data, exploring how NS uses predictive analytics to detect locomotive-health issues before they occur.
One of the most prevalent issues NS encounters in maintaining the health of its locomotive fleet is the loss of cooling water used to moderate engine-oil temperature. If a locomotive develops an undetected leak in its cooling-water tank, water levels can drop to a point where the locomotive is unable to keep its engine oil sufficiently cooled. When oil overheats, the locomotive will shut down and may incur costly damage to key engine components.
Cooling-water leaks can be difficult to detect in advance, but our crack team of data scientists had a solution. NS locomotives already wirelessly stream real-time data from hundreds of sensors to NS' operations headquarters. Using this information, our data-science team was able to design a machine-learning-based model to detect cooling-water problems before they begin to cause issues.
Currently, NS’ models are capable of predicting a cooling-water event roughly a week in advance of when it would affect operations. With the help of NS' locomotive reliability team, NS now is able to proactively generate work orders and shop at-risk locomotives before they break down in the field or incur expensive damage.
This glimpse into the future health of our fleet has significantly improved operations, both by preventing line-of-road failures and by increasing the reliability and efficiency of daily train movements. One more way NS is unlocking the secrets of big-data to reimagine possible.