Oil and gas flow meters need periodic recalibration to remain accurate and comply to contractual specifications. Recalibrations are costly and preferably delayed. However, even small meter errors in the field can lead to loss of revenues. So how do you know meters perform well in the field? DNV GL will set up a worldwide comprehensive metering database and use machine learning to determine the critical indicators of flow meter field performance.
We will use big data analytics, advanced visualization of meter performance and machine learning algorithms to estimate live meter errors of in-field flow meters, based on a database of field and calibration data from dozens of operators worldwide. We aim to bridge the gap between metering diagnostics and the actual in-field meter error.
By knowing the live flow meter errors, operators get insight in meter population performance. Meters with large errors can be replaced early to minimize the loss of revenues and meters without errors can be operated longer to save calibration cost.
An ultrasonic meter measuring 500 Million m3/yr which has a -0.5% error will be identified. The unmeasured gas amounts to 2.5 Million m3/yr.
The operator can recalibrate the meter earlier than planned, saving lost revenues of roughly 250k USD/yr for each year up to the year of planned recalibration.