Regression Analysis of the Variation in Rainfall Derived Inflow and Infiltration

Abstract
Rainfall derived inflow and infiltration (RDII) into sanitary sewers is often a major factor contributing to sanitary sewer overflow (SSO) and water in basement (WIB) complaints. SSO poses serious problems by contaminating the environment, causing property damage and threatening public health. Control of SSO is therefore a priority of many sewer system agencies throughout the United States and Canada. Sewer collection systems are commonly modeled to facilitate the control of SSO. Understanding the variation in RDII as a function of rainfall conditions is critical to improving the simulation results of sanitary sewer system models.
This chapter presents a statistical analysis of variation in RDII and its relationship to rainfall characteristics and other weather conditions. As part of the City of Columbus’s long term collection system modeling, sewer flow and rainfall data have been collected for a period of nearly ten years, providing a large dataset for analysis. RDII was analyzed using the USEPA’s recently released sanitary sewer overflow analysis and planning (SSOAP) toolbox program to generate the total RDII capture fraction R, which is commonly used to model RDII in sanitary sewer systems, as well as the rainfall characteristics. Multivariate regression analysis was performed on these data using Minitab 15 in order to examine the relationship between the total R and rainfall characteristics such as rainfall volume, rainfall duration, peak intensity and antecedent dry days, and other weather factors such as temperature. A generalized linear regression model was used to reveal the possibility of nonlinear relationships.
The results showed significant relationships between total R and rainfall volume, antecedent dry days, peak intensity and temperature. Seasonal or monthly unit hydrograph (RTK) parameters for RDII are typically used in SWMM 5 modeling for long term continuous simulation of RDII in sanitary sewers. Integration of an RDII regression model into a SWMM 5 model could enable storm-specific RTK values to be used, which might improve the continuous simulation of sanitary sewer systems.
This paper is only available in PDF Format: View full text PDF