Production Optimization: Improving Upon Traditional Regulatory Control

/

Most process units in refineries and chemical plants are governed by regulatory control. This involves a large collection of individual pressure, flow, temperature, level and other control loops with some degree of coordination. Most loops are governed by a PID mechanism.

This is common, basic control theory. So, what’s the problem? Here are two things worth thinking about.

First, multiple loops are interconnected, so when a process is operating erratically, keeping the extremes clear of limits requires moving the overall setpoint away from that limit, hurting overall average production and reducing plant profitability.

Second, who or what sets the setpoint? Why is the unit running at the level it is? Because it’s stable? Some other considerations?

Improved software and control techniques can be used to resolve these issues, and to optimize production based on any one of several desired outcomes.

Production Optimization Through Digital Transformation

How does this new approach work, and why is it better? First, plant managers have to determine their most critical objectives. The obvious answer is maximum output, but somewhere underneath, there is usually another variable that can’t be measured in real time by any kind of instrument. Maybe it’s product purity or some other key attribute.

We can show you how that key attribute can be measured using a “soft sensor,” taking readings from multiple instruments and performing calculations to create a synthesized variable. We call this Robust Quality Estimation, and it delivers the desired characteristics when a group of process operating conditions are realized. It is now possible to see the key attribute in real time, but performance is not always optimal when using conventional regulatory control, for the reasons discussed earlier.

Controlling a loop to a synthesized variable is not an easy task, but digital transformation makes it possible, along with Yokogawa’s Platform for Advanced Control and Estimation (PACE) software, which can replace or supplement individual PID loops by using new tools, including:

  • Multivariable Model Predictive Control

  • Digital Twin

  • AI

  • Big Data Analysis

  • Statistical Process Control

  • Linear Programming

  • Rigorous Simulation

The correct combination of tools must be tailored to a given process unit, and it governs operation from a central point to optimize production based on a desired set of goals. As control is improved, the amount of deviation from setpoint for each variable is reduced with far more consistency than the old regulatory approach.

Once this new control regime is established, you now have the ability to place the attribute wherever you like. It is a simple matter to shift production using PACE to whatever point is best. It is possible to maximize or optimize the key attribute, while maintaining more basic considerations such as overall output.

Optimize versus Maximize

There are situations where maximizing output or some other variable is not as important as optimization. Running at maximum output may reduce efficiency, causing the additional output to shift the cost picture unfavorably. In these types of situations, reducing output may be more profitable overall.

But let’s apply this concept to a larger picture. If your company has multiple sites and multiple production units, there’s a good chance some of the products are intermediates that are processed further in another unit. In other words, the output of Unit 1 becomes the feedstock for Unit 2. These types of relationships and interdependencies might extend even farther and be more intertwined from unit to unit.

How is it possible to determine output for all those units, so they all produce exactly the amount of product to match the needs of the next manufacturing step? The same techniques to optimize a single unit can be applied company-wide using a variety of tool, which in many cases will include Yokogawa’s Dynamic Real Time Optimizer (RT-OP). This tool finds optimum operating points for multiple interdependent units using current process data.

This approach works just as well on a small or large scale, and it can sustainably maximize plant and fleet performance to increase return on investment.

Yokogawa’s Production Optimization

In addition to PACE and RT-OP, Yokogawa provides software tools to create digital twin models based on simulation technology. As shown in the diagram below, this contributes to profit improvement by increasing added value using the digital twin’s prediction functionality (left loop), and by increasing efficiency through the use of the digital twin’s advanced control and optimization technologies (right loop).

Inside.png

These and other technologies embedded in Yokogawa’s digital twin platform enable optimization by maximizing throughput and production of the highest value products, whilst minimizing variable costs such as energy.

Click here to download our eBook and discover more about Yokogawa’s DX approach

Author.PNG