Increasing pipe failure rate observed in a water distribution system prompted an investigation of pipe condition. Exponent worked with the Alameda County (CA) Water District to assess remaining life of some of their existing asbestos-cement water pipes. Gradual degradation of the pipe due to calcium leaching was identified and a service-life prediction model was developed to assist in creating an optimum pipe replacement schedule.
Research was undertaken to investigate the stability of the asbestos-cement (AC) pipe infrastructure of the Alameda County Water District (ACWD) located in California’s San Francisco Bay area. The study included the evaluation of current condition and prediction of remaining service life of the AC distribution pipe. Laboratory assessment of 47 pipe samples removed from the distribution system revealed that, while pipe strengths were in general satisfactory, progressing deterioration of AC pipe due to leaching of calcium from the cement matrix was identified. Extrapolation of the observed trends suggested that the age at which pipe strength would fall below the design strength is on average 70 to 80 years.
Analysis of leak records between 1987 and 2006 revealed that pipe age and nominal size are the most critical factors affecting leak rate, with an increasing rate of leaks observed for older pipes and a decreasing rate of leaks observed for larger diameter pipes. An observed non-uniform spatial distribution of leaks in the system suggested contribution from additional factors, such as climate, soil type, groundwater conditions, conveyed water chemistry and operating pressure. While these factors were also included in the analysis, insufficient data precluded incorporating them in a prediction model. A general, semi-empirical model for predicting future leak rates and cumulative number of leaks over time was developed based on pipe age and nominal pipe diameter.
To overcome the lack of well-defined correlations with factors that ultimately lead to non-uniform spatial distribution of leaks, the entire distribution system was divided into equal-sized areas (grid cells). A method was developed to correct the cumulative number of leaks within each grid cell to account for 40 years of unrecorded leak data. The general prediction model was individually calibrated for each cell to reflect the unique leak rate history observed for the specific cell, thus increasing the reliability of prediction of future number of leaks. The estimated remaining service life of the AC pipe in each cell was determined by comparison of predicted cumulative leaks to a threshold number of leaks. A sensitivity analysis of the remaining service life to the selected threshold number of leaks suggested that an acceptable threshold value is within the range from 6 to 10 leaks per mile over the life of the pipe. The developed methodology will be incorporated in strategic planning of future AC pipe replacement.