Semantic Network

As mentioned earlier, the description is in the form of a subject (node), a relation (verb) and an object (another node). This nomenclature is called a subject-verb-object, or SVO, triplet. A node can be both the subject of several SVO triplets and the object of other triplets. [Cayeux 2019]

A semantic rule is a Boolean-valued function: if the evaluation of the predicate is true then the semantic of the rule is respected, otherwise the semantic of the network is flawed for the node in question. A semantic rule is expressed as a logical expression, i.e. any combination of logical statements combined with logical operators like "and", "or", "not" and predicate functions such as "Is Of Type" and cardinality tests, here noted "#Relations", on the number of relations that respect a given pattern. A relation pattern specifies the type of the relation, the type of the subject, noted here "From" and the type of the object, referred here as "To". For pattern matching purposes, the symbol "[this]" is used to refer to the node being evaluated. The various figures in this document provide examples of this.

Reference system

Signals can be seen as coordinates within a specific vectorial space. This space needs to be specified. This may imply specifying the dimensions of the space, its nature (cartesian, spherical, polar), the meaning of the origin and of base vectors.

Examples:

· an azimuth signal lies in a 1-dimensional polar system, whose origin may be the magnetic north, the true north or the cartographic north.

· A pressure lies in a 1-dimensional cartesian system, whose origin may be 0 pressure, atmospheric pressure or a reference pressure.

· A drilling measured depth lies in a 1-dimensional curvilinear system, whose origin is a specified datum and base curve is given by trajectory calculations.

· An elevation lies in a 1-dimensional cartesian system, whose origin is a specified datum (drill-floor, sea-bed) with a unit vector pointing upward.

Time reference system

Every signal is associated with a time stamp. This time stamp can be seen as another signal, whose underlying reference system (the time axis) also needs to be specified. This implies specifications of the corresponding clock (synchronization...). Note that there may be more than one time stamp: the sensor may time stamp the reading, or the data acquisition system (DAQ) may stamp when the reading is recorded, or the next element in the chain may stamp the time the reading is received at that element. This is more significant with slow transmission systems such as mud pulse telemetry. As an example, the reading’s time stamp in the internal memory storage of a MWD tool will differ from the new time stamp generated when the reading is received by the surface storage device.

Quantity

To each signal is associated a quantity. This quantity is the combination of two elements:

· The “derived” quantity: a combination of integer exponents that relate to base quantities (mass, length, time)

· A meaningful precision, that indicates how to test equality between two values. For example, mud shear stress, downhole pressure and formation strength are all pressures (M=1, L = -1, T = -2) but correspond to different physical notions. Using the same accuracy to compare shear stresses and formation strengths is obviously wrong.

A list of derived quantities can be found there: https://en.wikipedia.org/wiki/List_of_physical_quantities

Measurement

The measurement is the value observed for a specific quantity at a given time. The quantity is a static element to define the data system. The measurement is a dynamic reading or count of that specific quantity.

Contextual dependencies

Some signals (not all) can only be correctly interpreted when supplied with additional contextual information. Typical examples are the pressure-temperature dependence of fluid mass density, elevation dependence of some pressure types...

Logical position

Most signals (not all?) do quantify some characteristics of the drilling physical system. For example, the SPP is the pressure at a particular point in the hydraulic system. The location and nature of the signal with respect to the relevant system should be described. Those relevant systems are, for example:

· Hydraulic system

· Mechanical system

· Thermal system

Physical location

Most signals (not all?) are associated to a specific geographical location. This location should be specified. The location is unique, but it can have several representations depending on the coordinate system.

A location is described by (possibly multiple) pairs of:

· Coordinates

· Reference system.

For example, a downhole sensor’s location can be described by:

· A fixed coordinate (10m)

· The curvilinear coordinate system with the bit location as origin, the well’s trajectory as base curve and pointing upward

The bit location, which is the origin of the above reference system, can be described by:

· A dynamic coordinate (the bit depth)

· The curvilinear coordinate system with the rig-floor as origin, the well’s trajectory as base curve and pointing downward.

By combining the two systems, one always has access to the sensor’s location.

Signal type

Signals on a rig have different natures. One encounters measurements, parameters, set-points, etc...

One can distinguish between:

· the signal’s functions: a signal can have several functions. It can be a set-point, a limit value, an input to some processing unit, an expected value...

· The signal’s source: any signal is the result of a single process. This process can be a measurement, a correction, a conversion, a duplication, a computation...

For example: a DWOB can be the result of a downhole measurement (source), and an input to an auto-driller controller (function).

Uncertainty

Most signals have an associated uncertainty. Classical uncertainty relates to sensor precision and accuracy, but any estimated value is also uncertain.

This uncertainty can take several forms, from sensor uncertainty (typically precision and accuracy) to propagated uncertainty (any probability distribution).

Note that the uncertainty can be dynamic (and therefore associated to other signals):

· Pump pressure real-time estimations have an uncertainty that depend on the used flow-rate, rpm and block velocity, the model’s calibration status and general accuracy.

· Trajectory uncertainty evolves as the well gets deeper.

Signal processing

Exposed signals on a rig are always processed values:

· Measurements are derived from electrical signals via some transfer function

· Measurements are often filtered and/or averaged

· Estimated values are estimated...

· Commands are generated by a controller

The nature of the processing should be exposed. All details cannot be provided, but the main characteristics and parameters can be specified:

· Moving window averaging over 2.56 s

· Low-pass filter with a given frequency

Note that the processing characteristics can perhaps be merged with the source aspects from the signal type topic.