Digitalization of the drilling process has the potential to improve drilling data quality and consistency, providing support for drilling optimization, safety and efficiency. A significant barrier to realizing this potential is the data streams from the multitude of service companies, which changes almost daily, with variable definition of each of the real-time signals. This paper provides a solution to this problem: a method describing the semantics of real-time drilling signals in a computer readable format. For illustration, consider the calculation of mechanical specific energy (MSE) in drilling. It is possible to calculate a simple MSE signal in many ways, by using surface or downhole measurements, by applying corrections to the raw data, or by interpreting the equation in alternate ways. There is typically only a delivered value – the underlying details are lost. Semantic graphs bring transparency to the calculation by describing facts about drilling signals that are interpretable by computer systems. This semantic information encompasses details about signal measurement, and about signal calculation, correction, or conversion, yet all without exposing proprietary mathematical methods of calculation. It is possible, using semantic graphs, to assess the meaning and potential application of a signal, and whether or not the quality of the signal is suitable for its intended purpose. A semantic network relies on a vocabulary that defines a specific language dedicated to a particular topic, here drilling signals. The semantic network language is versatile: an existing language can describe new information and newly created signals. This provides a method meeting future needs without having to modify a standard constantly. In practice, each data provider exposes the meaning of its signals in the form of individual semantic networks. Merging these distinct semantic graphs provides a larger set of facts. This opens the possibility for synergies between independent data providers. For instance, applying logical rules infers new information. Since it is possible to query the semantic graph for signals that have certain properties, discovery of the most relevant signals at any time is feasible. By keeping track of modifications made to the semantic network during the drilling operation, it is also possible to post-analyze facts known about the available drilling signals, in an historic perspective. This is essential information for interpreting real-time data during offline data mining. This work is part of the D-WIS initiative (Drilling and Wells Interoperability Standards), a cross-industry workgroup providing solutions to facilitate interoperability of computer systems at the rig site and beyond. The D-WIS workgroup continues to develop the semantic vocabulary. The benefit of a computer interpretable description of the meaning of real-time signal is not limited to signals in real-time. Indeed, the method allows automatic data mining of historical data sets, facilitating the application of machine learning methods.
