By
·
2 minute read
In the oil and gas industry, where capital-intensive processes meet continuous output, maintaining quality and efficiency is paramount. Variability in raw materials poses a significant challenge, as it can profoundly influence the process. Traditional methods, based on scientific principles and experience, can be hit or miss. Enter Statistical Process Control (SPC) charts, a valuable tool to signal process instability.
Recent developments in Minitab have made modeling techniques more powerful and accessible. Process models, created using supplier measures and controlled inputs, establish relationships with key outputs. These models, deployed in Minitab Model Ops, make predictions based on new data, enabling the creation of SPC charts. If unfavorable trends emerge, the model can be reviewed to identify variables that can be adjusted to mitigate the trend. All of this is done before the process is executed, significantly reducing quality risks.
In an example involving 16 continuous variables, including supplier certifications, discrete variables, and key settings, a robust linear multiple regression model was developed, providing a good fit for historical data.
Figure 1. Response optimizer plot.
Figure 1, the response optimizer plot, provides a visual of how crucial predictor variables influence the process. Notably, the supplier certification measure and initial pressure exhibit strong, linear associations, where even slight value adjustments lead to significant shifts in the primary outcome. In contrast, the cooling temperature shows a milder effect, while changes in the unit settings appear to have a collective influence on the primary outcome.
Figure 2. Regression model.
The regression model effectively forecasts historical data's key response outcomes. Engineers seamlessly deploy the model into Model Ops using Minitab Statistical Software, simplifying the process with just a single click as shown in Figure 2.
Figure 3. Publish Model to Minitab Model Ops from Minitab Statistical Software.
Gathering fresh data for key output predictions involves supplier measures, fixed input settings, and generated process data with known variation. Take processing temperatures, for instance: real-world variations around set controls are factored in, using parameters derived from process measurements or equipment specs, creating a realistic process scenario.
Figure 4. Example 1 of a Predictive Statistical Process Control (SPC) chart.
Minitab Connect fetches data hourly, sending it to Model Ops for predictions. Monitoring with individual and moving range control charts is essential. Accurate historical parameters are crucial for control limits; using simulated data is inappropriate. In Figure 4, the predictive SPC chart indicates a stable process until the last three observations dipped below the lower control limit, posing potential quality concerns if instability persists.
Figure 5. Example 2 of a Predictive Statistical Process Control (SPC) chart.
The engineering team identifies a straightforward solution: raising initial pressure from 90 to 120. The model optimizer predicts this change will counterbalance the supplier measure shift. After adjusting the pressure in the data table, Connect data retrieval and charting are restarted. The resulting Figure 5 confirms the pressure adjustment's positive impact on supplier measure change.
In Conclusion
SPC's essence lies in prompt responses to evolving trends, preventing process instability. This avoids quality risks, safeguarding product integrity. In oil and gas, these benefits can mean millions in saved revenue. Minitab simplifies predictive SPC for integral quality management.
Ready to revolutionize your company with Minitab? Contact Laura Avery, our Director of Sales & Business Development, at laura@bowriversolutions.com. Bring your data to life.