Choose another country to see content specific to your location

//Select Country

Steps towards condition-based calibration through data models and statistical analysis

06 February 2020

Steps towards condition-based calibration through data models and statistical analysis

CalibrationAs flow metering technologies become more digital, the volume of data generated has increased dramatically. As the devices get smarter, so must we. Condition-based monitoring (CBM), where data output by digital devices is used to assess performance, has been around for a few years now, however
TÜV SÜD National Engineering Laboratory is working on a project focused on the diagnostic and predictive capability of these systems and the ultimate end-user experience.

The measurement of flow is a key requirement across industry, with the range of available meter types, technologies and manufacturers growing every year. Flow meters such as ultrasonic flow meters (USMs) and Coriolis flow meters can produce large volumes of digital data which, in theory, can be used for diagnostic purposes, to indicate device performance and installation conditions. Potential flow measurement errors such as incorrect installation, deposition or the presence of a second phase can be indicated by a drift in the meters baseline measurements. However, there has been a challenge in distinguishing between the specific causes of errors.

As part of a larger project on behalf of the UK Government’s Department for Business, Energy and Industrial Strategy, our Digital Services Team have published a new report; Data-Driven Models and Statistical Analysis of Ultrasonic Meter Data, which describes some of their latest work using statistical analysis techniques to better understand the dynamic system behaviour within USM data. The aim of this project is to support the introduction of
condition-based calibration as a feasible proposition to time-based calibration. Dr Yanfeng Liang, Mathematician and report author commented: “This report provides detailed analysis of USM data using data-driven predictive models which can predict the cause of drifts in diagnostic variables with a high level of accuracy.”

The output from this project to date has increased our capabilities to a level where we are now able to offer consultancy services in the field of data analytics and predictive modelling. This is an exciting new direction for the team, and we are looking forward to applying our newly developed toolsets to both live and historical big data applications” commented Technical Lead for Digital Services, Dr Gordon Lindsay.

The models constructed in this report can be extended and applied to other flow meter types and, despite being created with data from the oil and gas sector, are equally applicable to the emerging energy transition and other flow measurement industry sectors such as healthcare and water.

For further details contact Communications Manager

Next Steps

Select your location





Middle East and Africa