Along with many more subtle updates to InfoMaster 8.0, what used to be called Reliability Analysis in InfoMaster has now been upgraded and renamed as Deterioration Modeling. The Deterioration tools in InfoMaster allow users to select from a multitude of industry-adopted deterioration models. These complex models will provide estimated failure probability for pipes as a function of time. Advanced knowledge of statistics is NOT required to run these models! Rather, the user must simply have an understanding of which pipes have already failed (and the criteria that defines such), and of course, the data to back it up.
Getting started with these models is as simple as working your way down through the four steps in the Deterioration Modeling module.
First, users will define pipe failure in the Failure Definition. This is where the user defines criteria (observed or calculated) that amounts to pipe failure. In other words, this is where the user defines what criteria is indicative of failed pipes. Failed pipe data can come from many sources such as CCTV surveys, Work Manager data, or a particular likelihood of failure. The “Assume Missing Data as Non-Failure Data” allows users to specify how many of the non-failed pipes should be treated as reliable. For example, entering 100% assumes total confidence in non-failed pipes while entering 80% assumes that 80% of the pipes have indeed not failed, but the other 20% will be excluded from the analysis.
Second, users will generate a Sensitivity Analysis. This sensitivity analysis will show users which covariates are most indicative in determining pipe failure. In other words, what pipe attributes such as material, CCTV Scores, diameter, service requests, etc., are correlated with pipe failure. The output from this analysis will tell users whether a covariate was significant to pipe failure or not.
Third, users can leverage their Failure Definitions and Sensitivity Analysis data to generate Cohort and/or Regression models. The “Cohort” models are meant to be applied to groups of pipes that will have similar expected failure behavior over time (these groups are in fact the ‘Cohorts’). For this reason, it is a good idea to create cohorts based on available variables from a sensitivity analysis. Users can also create cohorts based on a field within the pipe feature class if they prefer. The end result of a cohort analysis will show how different groups of pipes (for example steel vs. cast iron vs. PVC) deteriorate over time.
Finally, regression models look at all the variables defined as indicative from the sensitivity analysis and calculates failure curves based off of the changes in those variables over time. For example, users can compare failure curves between hypothetical pipes with identical material, diameter, length values, but with different CCTV Peak Scores. In this way, users can analyze the impact of individual covariates, such as CCTV Peak Scores, in this case.
InfoMaster’s Deterioration Modeling tool gives users powerful statistical models to incorporate into the Rehabilitation Plan, Risk Analysis, or other areas within InfoMaster. These are extremely powerful tools for users wanting to predict pipe lifespans based on past data! But with great power comes greater responsibility, so make sure the data you are feeding these models is reliable before making any conclusions. For more information on this feature, please refer to the InfoMaster User Guide, check the helpfile, or call/email our support staff.