APC 2.0 optimization fills many gaps when considering the process optimization conundrum.
APC 2.0 optimization insights
- Understand how site-wide optimization unfolds in real-time
- Understand the role of APC and multivariable control in real-time optimization
- Learn to implement real-time process control & optimization more effectively
In conventional model-based multivariable control (MPC), optimization is a complicated endeavor. A steady-state optimizer is deployed in the control layer, which is problematic in itself, because control networks are not really meant for applications like steady-state optimizers (control networks, in principle, are for concise deterministic automation programs). But moreover, it then becomes necessary to reconcile this unit level optimization solution with the site-wide optimization solution that takes place on the business side. This can lead to frequent ad hoc meetings at the control console with operators, process engineers, and control engineers, to figure out how to adjust the MPC unit optimizer to get the best result – the one that best aligns with the site-wide solution. Thus, the operating team is adjusting targets, limits, gains, etc., to coerce the desired unit optimizer solution. This stands the whole idea on its head. Site-wide optimization solutions typically change daily and weekly, so these ad hoc meetings can become a regular nuisance. Industry could have seen this coming because, at the unit level, there is vastly insufficient data to derive a unit optimization solution that will match the site-wide solution. All this begs the question, why are there two optimizers in the first place?
In APC 2.0, the site-wide optimization solution is promoted and the control layer optimization is sidelined or eliminated. In this paradigm, the APC multivariable controller implements the optimization solution, even if it doesn’t solve it in the first place. Updated targets and limits from the business side may be communicated to the control side via the network, but common industry practice is to pass most updates via the chain of command (think morning meeting) so that everyone is informed and prepared for changes that can then be implemented together in a timely coordinated manner. Stealth changes, that is, optimization results that get implemented automatically and asynchronously, have come to be frowned upon. That’s basically another reason to skip the control layer optimizer and follow the business layer optimizer.
Implementing versus solving optimization
The APC 2.0 multivariable controller implements, but does not itself solve, the optimization solution. What does this mean? Obviously, based on the discussion so far, optimization is solved in the business layer, usually in a production planning and optimization unit (PP&O) of some sort. A new updated optimization solution means certain manipulated variable (MV) and controlled variable (CV) limits and targets may change, and certain MVs or CVs may be switched in or out of service. That’s the essence of the handshaking between business side optimization and control side multivariable control – some targets, limits, and maybe some gains.
Next, implementing the optimization solution means the APC multivariable controller takes appropriate action based on all limits and targets, including when they change. Moreover, it must be able to utilize all remaining manipulated variable (MV) availability to pursue the optimization targets (one per MV, experienced users will recall) after first satisfying all constraint limits. (In multivariable control, keeping the process within the constraint limit window is the first priority; optimization is the second priority.) That’s APC multivariable control in a nutshell.
APC 2.0 Optimization paradigm
As shown in in the Figure, optimization can be drawn as two layers or circles (not really a pyramid): PP&O in the business layer, where optimization takes place (or is “solved”), and the operations layer, where each unit implements its part.
In the business layer, site-wide optimization is a complex task involving a variety of inputs, tools, timeframes, and communications. Inputs include market opportunities, schedule constraints, blending capabilities, tankage inventories, pricing, planned and current outages, etc. By and large, each input has unique needs, customized tools, and its own appropriate time frame, so that experienced personnel piece together the final overall solution semi-manually. This heterogeneous optimization process produces an asynchronous stream of outputs, such as production schedules, buy and sell orders, blending plans, and updated APC limits and targets, among other things. A single unified integrated optimization tool remains a stretch goal for industry – maybe artificial intelligence will solve that!
In the operations layer, each operating unit is responsible – counted upon – to implement their part of the PP&O master plan. In the case of the process control layer, APC controllers need to effectively manage the multivariable nature (multiple-inputs-multiple-outputs) of the control problem, in the face of limits and targets that may change anytime, whether due to a new optimization solution from PP&O, or due to an operator input for a variety of reasons, such as to address an ordinary alarm.
Only one optimizer is needed
In APC 2.0, there is only one optimizer – the site-wide optimization that takes place in the business layer. There is no optimizer in the control layer, although the APC multivariable controller remains responsible for implementing appropriate parts of the business-side solution. It still needs to manage the multivariable control problem, which includes managing all constraint limits and utilizing all MV availability to maximize optimization targets.
Allan Kern, P.E., is principal APC Consultant with Lin & Associates Inc., Phoenix, Arizona, USA. Edited by Mark T. Hoske, editor-in-chief, Control Engineering, WTWH Media, [email protected].
Keywords
Advanced process control, APC 2.0
Consider this
Are you using next-generation APC tools?
You also might like
Read parts 1 through 3 below.
APC 2.0 spotlight, Part 1: What is APC 2.0?
