A trace framework for analyzing utility networks: a summary of results (industrial paper)

Given a utility network and one or more starting points that define where analysis should begin, the problem of analyzing utility networks entails assembling a subset of network elements that meet some specified criteria. Analyzing utility network data has several applications and provides tremendous business value to utilities. For example, analysis may answer questions about the current state of the network (e.g., what valves need to be closed to shut off water flow to a location of a pipe leak), help to design future facilities (e.g., how many houses are fed by a transformer and can the transformer supply another house without overloading its capacity?), and help to organize business practices (e.g., create circuit maps for work crews to facilitate damage assessment after an ice storm). Analyzing utility networks is a challenging problem due to 1) the size of the data, which could have many tens of millions of network elements per utility, and billions of elements at the nationwide or continental scale, 2) modeling and analyzing utility assets at high fidelity (level of detail), and 3) the different analysis requirements across utility domains (e.g., water, wastewater, sewer, district heating, gas, electric, fiber, and telecom). This paper describes the trace framework for utility network analysis that has been implemented in ArcGIS Pro 2.1/ArcGIS Enterprise 10.6. The trace framework features algorithms in a services-based architecture for addressing analysis tasks across a wide array of utility domains. Previous approaches have focused on solving specific problems in specific domains whereas the trace framework provides a more general, scalable solution. We present experiments that demonstrate the scalability of the trace framework and a case study that highlights its value in performing a wide variety of analytics on utility networks.