November 2019

Special Focus: Instrumentation and Automation

Process engineering, optimization and advanced process control

A recent Hydrocarbon Processing “Industry Perspectives” survey found that process engineering and optimization is the number-one topic that readers want to read about most. Process engineering and optimization rated higher than the next-highest topic (maintenance and reliability) by a factor of 2:1, and higher than every other topic by a factor of at least 4:1. Process control and instrumentation made the top ten list, albeit below process engineering and optimization by a factor of 7:1.

Kern, A., APC Performance LLC

A recent Hydrocarbon Processing “Industry Perspectives” survey found that process engineering and optimization is the number-one topic that readers want to read about most.1 Process engineering and optimization rated higher than the next-highest topic (maintenance and reliability) by a factor of 2:1, and higher than every other topic by a factor of at least 4:1. Process control and instrumentation made the top ten list, albeit below process engineering and optimization by a factor of 7:1.

These results suggest that managers and engineers want optimization more than ever, but also that optimization makes the most sense to them in the context of process engineering, rather than process control. This raises the question: Exactly how do optimization, process engineering and advanced process control (APC) work together in an industrial process operation environment?

Optimization by the layers

Fig. 1. A typical process optimization schematic, with additions to highlight important aspects of APC 2.0, especially the distinction between optimization solution generation vs. implementation, and the role of APC multivariable control in a closed-loop implementation support role, rather than acting as a separate, independent optimization activity.
Fig. 1. A typical process optimization schematic, with additions to highlight important aspects of APC 2.0, especially the distinction between optimization solution generation vs. implementation, and the role of APC multivariable control in a closed-loop implementation support role, rather than acting as a separate, independent optimization activity.

Fig. 1 is a common high-level view of process optimization. However, what does optimization look like from a more detailed, low-level view, especially for sites that consist of multiple interconnected units, such as oil refineries or petrochemical plants?

The top tier is site-wide production planning and optimization (PPO). PPO sets production plans and optimization goals for each unit based on feedstocks, prices, commitments, blendstocks, equipment in service, etc., in addition to (at least rudimentary) models of individual unit performance capabilities and the inter-relationships between units. From this activity, each unit receives its production “marching orders,” usually on a daily basis. This tier is the best-known aspect of optimization, and indeed is the only tier that can generate a complete site-wide optimization solution, comprising updated production targets, optimization goals and constraint limits, across all units. Subsequent (lower) tiers ideally should implement or enforce the PPO optimization solution, not generate new or different solutions.

The second tier in optimization takes place at the unit operation level, where each unit operating team implements their piece of the site-wide optimization solution. This tier is perhaps best represented by process engineers (also sometimes called production or operation engineers), who are often the most active players when it comes to following up on production plans and optimization goals in actual operation throughout the day, week and month. They tend to ensure that targets, both short-term and long-term, are on track, and often pursue possibilities for exceeding targets, such as extra volume, higher yield or greater efficiency. When actual production either exceeds or falls short of PPO plans, it is often the process engineer who troubleshoots operation, identifies options and feeds this plant intelligence back to the business planning side for continuous improvement of the planning cycle. While the first tier is optimization planning, this tier is optimization boots on the ground.

A third tier of optimization activity is APC. The role of APC in operations and optimization is best understood as being ancillary to the role of process engineering—i.e., to implement the PPO optimization solution, rather than to generate new or different optimization solutions. Additionally, the important distinction between the role of APC and the role of the process engineer is that APC is closed-loop and on the job continuously, whereas process engineers (and other operating personnel) take breaks and have competing priorities. The closed-loop nature of APC brings many obvious opportunities to improve upon the task of optimization implementation, over what can be expected from the open-loop efforts of engineers and operators. APC brings the same well-understood benefits as closing single-loop controllers, except APC controllers in effect close multiple loops together, using multivariable control algorithms. This tier is optimization automation.

Optimization in the control layer?

In this low-level look, the role of the conventional embedded APC optimizer at the control layer comes into question (noting that conventional APC comprises a multivariable controller and an embedded optimizer). What is the role of the embedded APC optimizer at the control layer, relative to the PPO solution at the business layer?

The embedded APC optimizer may have been necessary in the 1980s, when few other computer optimization programs existed in either the business or control layers, but today the embedded APC optimizer can be seen as extremely limited relative to modern business layer PPO solutions, which encompass more extensive (site-wide) information and employ modern planning and optimization tools that have kept better pace with technology. Can the PPO solution in the business layer be adequately leveraged by the APC controller in the control layer, thereby potentially eliminating the APC embedded optimizer—which (after model maintenance) has been one of the largest sources of APC’s continued high maintenance and high total cost of ownership?

Experience has shown that communication requirements between the optimization solution and the APC controller are not nearly as extensive or time-sensitive as it was once thought they might become. APC control must run at high frequency, since process values change in real time, but optimization does not need to run at high frequency, because the optimization solution normally does not change in real time. Typically, the PPO plan is updated daily and affects only a handful of parameters that need to be pushed to the control layer, which are normally passed down through the operating chain of command rather than via network connectivity. This allows for vetting and awareness of planned changes before they are made, and allows operations to time the changes when necessary, based on conditions on the ground. This picture suggests that leveraging the PPO solution by APC is not only possible, but for many purposes has already become established as industry best practice.

Another rationale for the embedded APC optimizer is its role in arbitrating the use of manipulated variables (MVs) when more than one is available to address a particular constraint control or optimization objective. However, the majority of APC applications in industry resorted long ago to using APC optimizer prices as course tuning parameters, rather than real prices. This has the effect of simply prioritizing the use of MVs, rather than assigning them based on (often fragile) economics.

In industrial process operation, MV choice is primarily an operation question and not a purely economic question. This experience suggests that in many APC applications, a straightforward prioritization scheme may be a more reliable and effective approach to MV arbitration than an economic optimizer.

New optimization and APC paradigm

Fig. 2. A proprietary APC 2.0 solution implements multivariable control and optimization without reliance on detailed models or an embedded optimizer.
Fig. 2. A proprietary APC 2.0 solution implements multivariable control and optimization without reliance on detailed models or an embedded optimizer.

The reader survey results reflect sentiment for a more effective optimization para-digm, especially in view of several inefficient legacy aspects of the conventional “APC 1.0” paradigm. The APC 1.0 paradigm remains on a trajectory of high cost and maintenance, and end users are no longer confident that this trajectory can (or should) be overcome without fundamental changes.

One potential change in a new “APC 2.0” paradigm, from an optimization standpoint, is elimination of the embedded APC optimizer in favor of better leveraging of the PPO solution. This has the potential to bypass problematic aspects of the embedded APC optimizer, while leveraging increased value from the PPO activity (Fig. 2).

Another potential change is an organizational change, wherein APC fills a straightforward closed-loop optimization implementation support role, rather than acting as its own optimization activity, separate and independent of PPO, which has led to several inefficiencies. HP

LITERATURE CITED

  1. Nichols, L., “Industry Perspectives: The topics our readers want most,” Hydrocarbon Processing, July 2019.

The Author

Related Articles

From the Archive

Comments

Comments

{{ error }}
{{ comment.comment.Name }} • {{ comment.timeAgo }}
{{ comment.comment.Text }}