The abundance in digital calibration information can be leveraged through statistical mining to improve calibration methodologies in tool manufacturing. Unable to store massive data, historical tool metrology relied on easily processed, summary performance metrics. However, with modern computing power, more computationally expensive methods of large data sets can be used to refine tool and process metrology, accuracy, and reliability. Empirical data from large sets of quartz memory gauges suggest that tools are consistently outperforming the accuracy specification derived from historical static calibration. This paper presents an alternative metrological methodology that provides additional insight into the accuracy of calibrated gauges. An analytic framework based on the Satterthwaite confidence interval (CI) is developed using an unpaired sample t-test and simultaneous field measurements from five stable pressure segments are used to validate the predicted variances between gauges. The methodology significantly improves measurement accuracy and reliability and is applicable to any instrumentation that undergoes a consistent, repeatable, and linear calibration.

In addition to reevaluating existing tool specifications, the methodology can be integrated into metrological practices in the oil industry where large digital data sets and painstakingly accurate tools and calibration devices are commonly found. Presented are results from memory gauges, but by extension the method is applicable in other formation evaluation endeavors. The analytic framework predicts a factor of proportionality between the accuracy predicted by the CI and the traditional metrics of individually calibrated gauges. Specifically, a relationship is found between the 2-norm of the mean quadratic deviation (MQD) error of two individually calibrated pressure gauges. A case study of HPHT gauges in five well tests in the North Sea and India confirm analytic predictions. The paper focuses on quartz memory gauges and improves absolute and relative accuracies of HPHT gauges by over 33% and 56%, accordingly. Comparable analysis can be applied to other tools to better evaluate metrology. The converse method is useful to determine a process' variance and bias — vital information to evaluate and compensate a facility, process, or device.

You can access this article if you purchase or spend a download.