Misconceptions and Misuses of Data in Infrastructure Asset Management

S. Madeh Piryonesi

Abstract:

Infrastructure asset management has been around for a few decades; however, there are still many misconceptions in the domain. Some of these problematic concepts are taught in classrooms and reprinted in textbooks. This paper investigates two key bodies of misconceptions: problematic concepts borrowed from other industries such as industrial engineering or mechanical engineering and issues related to the application of data-driven solutions to infrastructure problems. First, a few well-known concepts such as bathtub curve, replacement time and asset remaining life, which are mostly borrowed from reliability engineering, are critiqued. Two examples from road infrastructure are proposed to show the limitations of the bathtub curve. The examples are based on the roads from the Long-Term Pavement Performance database and clearly indicate that their failure rate may not perfectly follow the bathtub curve, and application of this concept to different assets should be done with more caution. Second, a few common issues related to the application of data-driven solutions to infrastructure deterioration are discussed. The focused was placed on model training and particularly data representation. An example of using population stability index as a measure for assessing data representativeness was presented.