Achieving an Enhanced Understanding of Stormwater Pipe Condition Through Data Mining of CCTV Inspection Records

Stormwater pipe condition is commonly assessed using closed-circuit television (CCTV) inspection. These inspections tend to be expensive and most municipalities have been limited to inspecting small portions of their aging stormwater systems. Data mining is proposed as a means of extracting valuable information related to stormwater pipe deterioration from existing CCTV inspection records. Classification tree algorithms are used to identify the influence of pipe-specific attributes (i.e. year of construction, material, diameter, length and slope) on stormwater pipe condition in Guelph, Ontario. The municipality has inspected 25% of their system and the developed trees can be used to predict the condition of individual pipes that have not yet been inspected. The data mining approach presented in this research allows greater value to be extracted from inspection efforts and enhances stormwater pipe operation and management practice.