Ensuring Excellence: Dimensions of Data Quality In Modern Operations ??

In today’s data-driven landscape, the ability to leverage high-quality data is critical for organizations to make informed decisions. With AI transforming operations, ensuring the quality of data—including completeness, accuracy, congruence, precision, timeliness, and cohesion—is paramount. These dimensions shape the effectiveness of business insights, driving operational efficiency and minimizing costly errors.?

1. Completeness?

Completeness refers to the availability of all necessary data. Missing information can compromise the value of the dataset.?

Description: A dataset lacking key data points can render it ineffective. For instance, if data from a frac operation is incomplete—missing channels like Chemical or Environmental data—it may distort insights critical for operational decisions.?

Industry-Specific Example: When receiving data at 1 hertz for an hour, the total set should comprise of 3,600 data points.?

In frac operations, ensuring complete data capture from all relevant channels—such as Frac, Chemical, Consumable, and Environmental Monitoring—is critical for accurate, real-time decision-making. Since data is rarely delivered at a strict 1Hz rate, completeness must account for variations in frequency and duration. By calculating the expected number of data points based on these factors, Cold Bore ensures your team receives comprehensive, actionable insights. This approach minimizes risks, maximizes efficiency, and gives you a competitive edge by providing the full scope of data needed to make informed decisions.?

?

2. Accuracy

Accuracy measures how closely data reflects the real-world entities or events it intends to represent.?

Description: Accuracy is a straightforward dimension. Essentially, it represents that the data reported directly represents the underpinning reality. Methods to detect issues regarding accuracy include (but are not limited to) using one or many external authoritative sources to triangulate with the evaluated dataset.?

Example: Validating a manual time log of well operations against real-time pressure data helps pinpoint errors in the log, confirming the data’s correctness.?

If you plot the miles driven by a truck based on the driver's mileage log and cross-reference it with an external authoritative source, such as the truck's GPS module, the resulting graph should match the provided table exactly—1-to-1—for each data point.?

Industry-Specific Example: An example of ensuring data accuracy is using an external data source to confirm channel validity. For instance, validating a manual time log of “well open and closed” events against an external Wellbore Pressure Sensor involves correlating these events with sudden pressure changes that signal valve actions.

If the manual log indicates a closed valve while the sensor records pressure changes, this discrepancy clearly suggests an inaccurate manual entry. See example below.

3. Congruence

Congruence refers to?statistical consistency across different observations of the same data element. It can also refer to the inherent business logic of a data element that prevents the data from having a particular shape.?

Description: When comparing data over time, congruence ensures consistency. Variations outside expected ranges signal potential errors. One can use historical observations of specific datasets to derive their statistical properties. With a statistically significant amount of those, new observations can be compared to those statistics and divergences of many standard deviations can be used to detect potentially erroneous data.?

Example: Tracking the miles driven by a truck should show a consistent, positive slope. If the data suddenly dips, it signals an issue.?

Industry-Specific Example:

Using data from the same source over time to confirm channel validity exemplifies data congruence.

For instance, the Proppant Rate channel in a frac data source should display a similar profile across stages, assuming a consistent stage design and steady formation factors.

Similarly, the Wireline Depth channel from a wireline data source should show a gradually changing profile across stages. The difference in the Max Wireline Depth summary metric between stages should remain consistent or close to the same for each stage.

4. Precision?

Precision is the number of digits in a number and its scale, representing the number of digits to the right of the decimal point.

Description: Different applications require different levels of granularity in the data it consumes. For instance, expressing water volume in gallons with one decimal point is sufficient for measuring the volume of tanks at a water treatment plant. Still, this level of precision isn’t appropriate when calculating the required water volume in a lab experiment.?

Example: An electricity provider bills at $0.10 per kWh. If the meter reading rounds to the nearest whole kWh (without decimals), then a usage of 1.49 kWh would be rounded down to 1 kWh, costing $0.10. However, a usage of 1.50 kWh would round up to 2 kWh, doubling the cost to $0.20.?

Industry Example: A digitally generated time log from a frac monitoring system provides precise start and end times for each activity. Manually editing this log and rounding times to the nearest minute reduces its precision score.?

Similarly, manually editing the log while using start and end times from an unsynchronized time source may preserve a high precision score but could cause it to fail other metrics. Notably, a precise format alone does not guarantee data accuracy.?

5. Timeliness?

Timeliness is the time delta between the moment a datum is created and the moment that same datum is consumed for its end application.?

Description: Typically refers to how “fresh” the data is. Some applications can be efficient only if they can impact a process in close to real-time.?

Example: A self-guided missile’s thruster system relies on high timeliness from its radar system’s data stream, as any data point received with a delta of more than a millisecond may put the missile off its path to the target.?

Industry Specific Example:?

Timeliness as above commonly refers to the time between data collection and available consumption.? This metric gets more complex when there are multiple points of access vs. several datums.?

The channel Treating Pressure from the frac service is streamed live on site and received in near real time by the data capturing device. This data is transmitted to an edge device on pad to be available for the on site rep (OSR) (Timeliness 1).?This data is streamed to cloud to be visible for an ROC (Remote Operations Center) (Timeliness 2) and is made available in the client’s data warehouse for post stage analytics (Timeliness 3).?The Timeliness metric entirely depends on use cases.?

?6. Cohesion?

Cohesion refers to the relationships between different datums. Observing those relationships over time, and recognizing when they diverge in a statistically significant way, can be very useful in supporting data quality.?

Description: Identifying the statistical properties of the relationship between datums from related data sets allows you to verify those same properties when looking at new, similar data sets. The existence (or non-existence) of those statistical properties can give great insights into data quality.?

Example: When plotting the amount of air and the amount of liquid in a water tank, the relationship should be inversely proportional.?

Cohesion - Industry Specific Example:?

The channel Wireline Speed from a wireline data source during the activity “Wireline Running in Hole” is greater than 0. On some jobs, the channel Pump Down Rate from a pump down data source is greater than 0. If this relationship isn’t true, Pump Down Rate and Wireline Speed are non-cohesive for these specific jobs.?