# SI-Cluster

When calculating the risk of pipe failure with SI-Rehab, we multiply the probability of failure with the effect of failure for each pipe segment in the network. The probability can be calculated as an average failure frequency of the total population (or cohort) of pipes as a function of age with SI-Regression. SI-Cluster does not require the year of installation as input parameter, but acknowledges that ‘local unknown factors’ may cause leakages.

**Local unknown factor**

Often it’s difficult, if not impossible, to give the exact reason of the failure of a pipe. Has the material worn out as a consequence of ageing? Did traffic play a role? Settling of ground? Or have the pipes been installed by a less competent contractor? SI-Cluster groups pipes in the network with similar characteristics and determines for each cluster the mean-time-between-failure (MTBF), while ignoring the reason of failure. Please note that we do exclude leakages caused by third parties from the calculations.

The example below shows two clusters of pipes of circa 3.4 km length, PVC 110 mm from 1960. When relying upon SI-Regression solely, the failure rate would be 0,057 failures/km/year. In reality they differ substantially in the studied time period. For this analysis, we ignore the reason of failure: it has failed, so it is likely that it will fail again in future if the circumstances do not change and as long as the pipe is not renewed.

**This makes SI-Cluster unique**

- Takes the ‘local unknown factor’ into account
- Assessment of groups of pipes with similar characteristics
- Does not require the age of pipes as input

SI-Cluster and SI-Regression are powerful methods to calculate the probability of failure of each pipe in the network available on SI-Platform, our dashboard for software tools.

**Next step**

We are looking forward to reveal the details and essence of our algorithm. Please contact Marc Felten to select an area to test the software and define a pilot project.

CONTACT

"Apply SI-Cluster and add reality to your calculation of the probability of failure"