An uncertain solution to optimise flow meter management

Calibrations and maintenance tasks are often completed on a calendar-based timescale despite the fact that this method does not take into account the quality of the measurement, which is the risk associated with the measurement in terms of the financial exposure caused by any inaccuracies. 

During calendar-based calibrations, a meter that is operating well within acceptable performance limits could still be removed and sent for calibration – an added cost that is not actually required. Conversely, the meter could have exceeded its calibration performance within the first few months of service – operating with an undetectable error for a significant length of time, potentially costing the company money. Only when the next calibration was completed would this error be spotted and rectified.

Obviously, neither of these options are ideal. It would make a lot more sense financially to have another system in place, particularly when the number of meters requiring calibration in any one installation is considered. One alternative method is a risk-based approach where uncertainty and the financial exposure of the measurements are used to optimise the calibration period.

Essentially, this method makes use of historic calibration data and uses this to predict the meter’s expected performance over the next year. Intuitively, this makes sense as many of the operating conditions and factors that affect performance will not change significantly from year-to-year, such as the stream physical properties, fluid cleanliness, concentration of erosive particles etc. Therefore, the same meter, in the same conditions, should operate in a similar manner to the way it has previously. 

The use of historic data also allows for different stream conditions to be given different criteria for calibration, rather than a blanket calendar cycle. If a stream was particularly erosive, for example, it would make sense that the meter would change, exceeding its calibration performance limits quicker than a stream which is less erosive.

Using previous history alone is a valid method, but coupling it with uncertainty in measurement, and hence financial exposure, can create a powerful tool for optimising calibrations. 

Uncertainty analyses are essential to determine whether meter measurement systems are capable of meeting performance targets, but they can also be used to determine where improvements in the measurement can be made. There are different sources of uncertainty that all contribute to the final value. Some are constant over time, such as calibration uncertainty, and there are some that accumulate over time, causing measurements to drift from calibration values such as bearing wear on mechanical devices. Assessing the contribution of each source to the overall measurement uncertainty estimate allows end-users to better reduce uncertainty by focussing efforts on the largest source or sources.

In terms of calibration optimisation, sources of uncertainty based on meter drift are important as they determine the financial exposure faced by the company through changes in measurement performance from calibration conditions. If the calibration timescale question is then answered in terms of financial exposure and cost, it provides a more useable criterion on which to base the calibration. For example, knowing the additional exposure caused by drift and the average cost of calibration/maintenance of equipment over time, it is possible to evaluate the optimal calibration period – and hence the most financially efficient timescale to calibrate the meter and minimise costs.

The above method essentially bases the calibration and maintenance timescale on financial risk as opposed to a routine calendar period. Another advantage of the risk method is that it can be updated every time a new calibration data set is obtained. By this method, the produced result becomes increasingly statistically relevant as it is based on the latest data.

In summary, by using data readily available for most end-users it is possible to set calibration timescales of equipment based on the financial risk from the measurement itself. Considering the number of measurement instruments in use today, this could save a significant amount of time, effort and money.