InfoMaster now provides users with the ability employ multiple statistical models for calculating the reliability and survivability of their pipes. InfoMaster features the Weibull, Herz, and Cox proportional hazards models for modeling the aging processes of water and wastewater network pipes and accurately estimating their lifetimes and structural/functional failure potential and distribution over time. Here are some highlights and insight in to the different techniques available:
The Weibull and Herz Cohort methods are incredibly easy to configure and tolerant of ‘noisy’ data. Herz was designed specifically for the water industry, is very flexible and is unlikely to yield unrealistic results, conversely, Weibull is almost always capable converging but use caution in interpreting the results. Users may want to screen out the anomalies from their analysis, for example, if we were using failure history data to define ‘life expectancies’ (inputs to the analysis) for our pipes, and a small number of pipes failed right off the bat – we would wish to exclude these pipes from our analysis as the may skew the rest of results for their cohort (groups of pipes based on failure behavior). The Herz and Weibull probability density functions will have a single peak.
The Cox Regression technique provides detailed results by evaluating separate variables and the relationship between these covariates (such as material, lining, soils, diameter, etc). The Cox technique will require a high volume of detailed data and is driven by initial input failure data, failure data can take the form of CCTV surveys (Peak scores, for example), failure History data (bursts, breaks, etc) or probability of failure results. The Cox proportional hazard model’s probability density functions may have 2 peaks (as opposed to just 1), and users can use these curves to determine when to repair pipes (at the firs peak) versus when to replace (the second peak) depending on the stage of life that a pipe may currently be in.