[V7N2] – Success and Failure of First Principle Models for the Control of Refining Processes – Part II

First Principles Models aim to describe the behavior of a system from fundamental physical laws or principles, often using mathematical equations and computational methods. These models are advantageous when empirical data is limited, unreliable, or unavailable or when a deep understanding of the underlying physics of a system is necessary.

In our context, let us briefly discuss the First Principle Model for Fuel Blending, or FPBM. This model governs how the fuel manufacturing process is modeled and manufactured. Simple as it sounds, it is not because it involves many processes and procedures, all linked and integrated, to get that gasoline in your car. A refinery’s fuel blending system involves eleven automation modules and sub-control systems.

1709425683939

The first principle model for blending is a set of fundamental equations to describe the mixing rules. It is highly non-linear due to interaction, non-deterministic, that is, qualities are more than the equations controlling them and many other factors. A typical FPBM looks like this.

1709425714036

The First Principle Model for Blending is discussed excellently in topic OEA14P on our OMS eLearning Academy Portal.

First principle models (FPMs) have shown success and failure in controlling refining processes.

Here’s a breakdown of their strengths and weaknesses:

Successes:

  • Transparency and Interpretability: FPMs are built from fundamental physical and chemical laws, making them transparent and interpretable. The FPM allows engineers to understand how process changes will impact outputs, aiding in troubleshooting and optimization.
  • Robustness and Generalizability: FPMs can be more robust to unexpected disturbances and generalize better to different operating conditions than data-driven models. This observation is so because they capture underlying principles rather than relying solely on historical data.
  • High-Fidelity Simulation: For complex processes, FPMs can provide high-fidelity simulations that can be used for training operators, testing new control strategies, and predicting process behavior.

Failures:

  • Complexity and Development Costs: Developing and maintaining FPMs can be expensive and time-consuming, requiring deep domain expertise and significant computational resources.
  • Accuracy and Calibration: Accurately capturing all relevant physical and chemical phenomena in an FPM can be challenging. Additionally, calibrating the model to real-world data is crucial for reliable predictions.
  • Limited Applicability to Entire Process: FPMs might not be suitable for modeling the entire refining process due to its sheer complexity. They are often applied to specific sub-processes where the fundamental physics is well understood.

Takeaways of FPMs:

FPMs offer valuable insights and control capabilities for refining processes but come with challenges. Their success depends on the specific application, available resources, and expertise. Refineries have failed to implement FPBM to achieve their goal for product qualities to minimize giveaways and be on specs as tightly as possible. They fail or lose 25-30 million dollars annually because they do not do due diligence to model the process.

Refineries want to implement AI, as we discussed in the first part of the series, to ride on the bandwagon unthinkingly but fail to implement basic building blocks like analyzers, model corrections, feedback, etc., etc. AI/Machine is not the refinery’s savior; it is just a helping hand if they can hold it tight and with faith. Otherwise, it is just a buzzword in the vocabulary.


Disclaimer: OMS eLearning Academy and Google Gemini collaborated as Humans and AI to generate this article for you.


In the next  topic of the series,  we will discuss alternatives and supplements to FPM, focusing on blend control systems such as AI/Machine learning.

Part-III: Can AI-Based Machine Learning Stand On Its Own For Fuel-Blending?

In the subsequent series, we will deep dive into the concept of AI/machine learning for fuel blending and discuss steps to improve FPBM to reach the desired quality target of fuel products.


Stay tuned for more groundbreaking publications and enrich your expertise with OMS Academy. 🌐✨

Sign up today to start your learning journey.


➡️ Sign Up Now 🚀


Thank you for being a part of the OMS eLearning Academy community.

If you have any questions or feedback, please contact us.

“At OMS eLearning Academy, excellence in education isn’t just our goal; it’s our promise – a promise to fuel minds and forge futures in Downstream Refining.”

– Success and Failure of First Principle Models for the Control of Refining Processes – Part II" sku=""]
Explore Our Courses