Understanding Errors
The difference between an observed or calculated value and a true value. Specifically: variation in measurements, calculations, or observations of a quantity due to mistakes or to uncontrollable factors.” (Merriam-Webster Online Dictionary)
We constantly observe to understand what happens around us. We may observe to optimize an outcome as well as to try to predict the future. In any attempt to understand reality, whether through measurements or calculations, it is important to be aware of factors beyond our control, uncertainties, and how they affect the accuracy. We have to relate to deviations from reality, errors, acknowledge their existence, and have a notion of their origin to assess their impact on any conclusions we draw.
To understand and conceptualize error, or accuracy, we can use archery as an analogy. Below, is the result of three different archers shooting ten arrows each.
The archer to the left has a large spread of the shots, and we can agree that the marksmanship is not accurate. In this case, it is hard to predict where the next arrow will end up. The archer in the middle groups the shots well, but they are not centered. This indicates good repeatability, but there is an offset, a bias. In comparison to the first archer, we have a good notion of where the next arrow will hit. While marksmanship is not accurate, there is a clear way to improve the result if the aim is shifted up and to the left. Finally, the archer on the right demonstrates good accuracy, that is good repeatability without a bias.
Continuing with the archery analogy, the two archers below both distribute their shots around the center, but their accuracies differ in terms of how closely they group their arrows.
There can be various explanations for the different accuracy levels above. One potential explanation is the amounts of practice the two archers have put in, with more practice the archer on the left may achieve similar results as the one on the right. Another potential explanation is that the archer on the right has better equipment for precision shooting. In both cases, it can be argued that a higher accuracy comes at a higher cost, either monetary or in terms of time. To have the best return on investment, it is important to ask which level of accuracy is really required; there is no universal answer and assessments should be made on a case-by-case basis.
Finally, related to the level of accuracy is the level of confidence, which can be thought of as the frequency at which we can expect a certain outcome. Turning to archery again, we can look at the target below and we see that the archer hits the center with 9 out of 10 arrows. If this is repeated over time, we can state that the archer has a 90% confidence level of hitting the center.
No system is perfect and able to account for all eventualities, e.g., a sudden gust of wind may force the arrow off target, and the confidence level sets an expectation on “perfection.” As with increasing the level of accuracy, better confidence levels come at a cost, and the confidence level should reflect the actual needs on a case-by-case basis. However, there is an important difference – confidence levels directly impact risk assessments and the likelihood of not detecting critical events.
At Turbulent Flux, we strive to deliver life-of-field valid solutions to the oil and gas industry. To achieve this, we have developed self-adjusting technology which utilizes existing sensor data and reference measurements to provide the best possible accuracy at any point in time. Firstly, this accounts for changing conditions to avoid bias. Secondly, it caters for improved predictions as more relevant data becomes available. Thirdly, it offers the ability to detect discrepancies between predictions and measurements to flag potential challenges at an early stage. All in all, this ensures accurate solutions with little need for human intervention which are fast to deploy.