InfoAsset Planner's Deterioration Models allow for statistical analysis to help predict future pipe failures. These deterioration models are available for the Gravity Mains facility type in the Sewer network and the Pressurized Mains facility type in the Water network.
Before running any of InfoAsset Planner's Deterioration Models, a word of caution. These models may only be as accurate as the data provided to them. It is advisable that users should initially use caution when making conclusions based off of their deterioration models. Then as the models accumulate reliable data points and grow with the InfoAsset Planner project, users may place increasing faith in these models. Deterioration Models are not a required component for successful InfoAsset Planner projects, so if Deterioration Modeling options can be found within the Operation Center, within the Analysis tab, below the Risk section.
As seen in the screenshot above, the Deterioration Model module consists of four parts. For more information on these parts, see the appropriate sections of the Help File or our Knowledge Base and User Forum resources.
Failure Definitions are the first building block in creating a model. They define the pipe attributes that indicate a failed pipe. Failure Definitions essentially split the pipes into three pools: failed pipes, non-failed pipes, and pipe excluded from the analysis all together. How users create Failure Definitions to define these pools is crucial to the success of any deterioration or statistical model. Because Failure Definitions may be saved in a list, they can be mixed and matched between different Sensitivity Analyses, Cohort, or Regression models.
Sensitivity Analyses are the second building block used to create full Deterioration Models. Sensitivity Analyses basically consumed the Failure Definition results and then evaluate parameters input by the user to determine whether any of those parameters have a statistical relationship to pipe failure. For example, a Sensitivity Analysis may evaluate pipe attributes like diameter, material, contractor used, length, depth, etc. A Sensitivity Analysis may be able to say that diameter and contractor used were significant factors in determining pipe failure, but there was no pattern or relationship with the other parameters and the failure/non-failure pools. This can help focus our Cohort Models and is a requirement for Regression analysis.
Cohort Models group like pipes together to make more generalized statistical models about groups or 'cohorts' of pipes. Cohort Models are useful in that users may not have enough failure data to determine life expectancies for any individual pipe, but perhaps they can determine the average life expectancy for an average pipe within a particular 'cohort'. It is advisable to first run Cohort Models either with the Weibull or Herz models before moving onto Regression Models. Cohort Models generally require fewer data points to produce more reliable results compared to Regression Models.
Regression Models are the most advanced statistical analysis within InfoAsset Planner. Regression analyses look at the results from a Sensitivity Analysis and show how user input variables (Diameter, Pipe Length, Material, etc.) affect the overall pipe survival probability and likelihood of failure. By adjusting the parameter values within a Regression Model result, users can extract detailed curves for specific pipe criteria. Four different types of Regression Models are available within InfoAsset Planner: Cox, Linear Extended Yule Progression (LEYP), Non-Homogeneous Poisson Process (NHPP), and Non-Homogeneous Markov Chain (NHMC).