Abstract
A case study has been carried out to investigate how process data from Frognerparken Pumping Station can be used more efficiently to get an early detection of degradation or failures of the pump systems. The technical condition is described by use of a single number from 0 to 100, denoted as the Technical Condition Index (TCI), where maximum value describes the design condition as new and the minimum value describes the state of total degradation. As presented in this paper, the pump station is broken down to subsystems where each subsystem consists of several nodes with its own TCI value. The overall pump station's TCI number is then computed by aggregating a weighted sum of all the subsystems’ TCIs.
A method for early failure detection is also developed and is based on combining energy balance law together with the nonlinear extended Kalman filter to estimate model parameters for both the heating and the cooling phase of the pump. The model parameters are directly related to the temperature change and will give an indication of degradation during the transient phase even if the pump is working properly. During a visual inspection one of the pump had significantly higher temperature and vibration compared to the other pumps, and based on received timeseries for that specific day, we were able to validate the proposed method with realistic data.
A method for early failure detection is also developed and is based on combining energy balance law together with the nonlinear extended Kalman filter to estimate model parameters for both the heating and the cooling phase of the pump. The model parameters are directly related to the temperature change and will give an indication of degradation during the transient phase even if the pump is working properly. During a visual inspection one of the pump had significantly higher temperature and vibration compared to the other pumps, and based on received timeseries for that specific day, we were able to validate the proposed method with realistic data.
