Oct-2013

Using APC in refinery energy systems — develop your own solutions

Advanced Process Control (APC) has been used extensively to assist in operating refinery and petrochemical processes smoothly, safely and profitably.

Zoran Milosevic, KBC Process Technology
Andreas Polster, Andreas Mai, Ute Seifert and Manfred Wagner, Total Leuna Refinery

Article Summary

However, with the constantly increasing fuel and environmental costs, and the related pressures to squeeze every percent of efficiency out of existing facilities, APC applications have gradually found their way into energy systems as well. The control of distillation columns, fired heaters, large turbogenerators, and the complex heat exchange schemes can all be subject to a tighter control, that maintains stable operation, achieves and sustains high energy efficiency of a part of a process or a piece of equipment, or strikes an optimum between yield and energy performance of a process unit. The use of APC, however is not without its controversies, as a number of not-so-successful applications have recently been reported. The paper discusses the potential use of APC in oil refinery energy systems, and describes cost-effective, real-life applications that will be of particular interest to refiners who favour the development of their own APC solutions.

Introduction
Advanced Process Control is generally seen as one of the highest level of technology and process control sophistication, the efficacy of which is well established and documented. However, APC is currently being re-evaluated by the industry, as an increasing number of poor-performing solutions have been reported (Jubien and McIlwain, 2009). “While pleasant and pleasing on an academic or simulation level, APC runs into many limitations in a real-world process control” (Kern, 2010).

In search for the reasons of lower-than-expected APC performance, one school of thought maintains that it is the unnecessarily high complexity employed in APC designs, and the ever changing plant economics. A solution is proposed that combines the use/preference of inferential control models, and enhanced “ownership” of APC application (Friedman, 2010).

Another line of thought, represented by Kern, argues that APC faces many limitations in the real-world process control. The “realism factors” reduce the initial large matrix to a number of “clamped” manipulative variables, and a small handful of control overrides, explaining under-performance. The proponents of this hypothesis argue that actual optimisation takes place in the refinery knowledge “cloud” that includes engineers, planners and operators, and results in achieving a near-optimum operation by this collaboration, which the control system has only to maintain. This should be process control’s primary aim, they claim, and not to try to squeeze another small percentage point from the process. “Much of what the APC can achieve can be done through the DCS, more easily and at lower cost” (Kern, 2010).

Yet another group of authors, whilst asserting that “there are likely more poor performing APC applications in existence than well-performing ones”, discusses how, given the dynamic nature of the plant environment, APC applications should be best designed to maximize robustness and reduce the maintenance effort (Jubien and McIlwain, 2009).

Another problem, that all agree upon, may be staffing. Some refiners have reduced their process control staff to the minimum necessary just to maintain the systems, but some are still able to justify larger teams and extend their work to APC.

Total’s Leuna refinery has extensive experience in APC development and has its own view to share. Of particular interest is the often mentioned use of APC in energy systems, as the industry seeks novel ways to reduce energy consumption, precisely trying to “squeeze another small percentage” in view of tightening fuel quality and environmental legislation. The authors maintain that energy systems lend themselves well to APC because of high confidence that benefits will be achieved, calculation robustness, relative simplicity of control variables used to maintaining the economic optimum, and relatively simple and predictable cost calculation (avoiding the uncertainty of intermediate product prices, so critical in yield-related optimisations).

APC in Leuna Refinery: Applications, Successes, Failures and Reasons for Success/Failure
The cost of energy has increased rapidly during the last years and with it the incentive to use APC in energy optimisation. Basically, APC produces benefits by pushing constraints and performing more consistently and accurately than an operator can by acting alone. An important mechanism of APC effectiveness is that it tends to reduce the operator’s conservatism, bringing the operation closer to the optimum than would the operator do. Examples include furnace stack oxygen control, and minimisation of reflux and reboiling in distillation columns while maintaining tight control of product specification.

Leuna refinery developed their approach to APC over a period of time, climbing the learning curve, basically by carefully evaluating and selecting from the following list of options that many APC decision makers face:
• APC’s role: optimiser, or a controller, or operator’s tool (an “open loop consultant”)
• APC versus enhanced use of DCS
• Own APC solution versus third party solutions (including pre-engineered solutions)
• Small applications versus complex systems
• Use of empirical and / or physical models, or a combination of the two
• And for those who, like Leuna refinery, decide to give preference to building their own applications, the selection of priorities and the methodology.

Leuna decided to base their APC systems on own solutions, developed by a dedicated team of process control engineers; recognising that this may not necessarily be a possible option for everyone.

APC Role – Optimisation, Control, or Operator’s Advisor?
The strategy of process optimisation at Leuna is to rely on combined planning, operations and process engineering expertise to propose the optimum operating scenario. Total’s Leuna refinery therefore strives to achieve an overall plantwide optimisation that is more wide-ranging than what Real-Time Optimization of any complexity could handle. Therefore, APC is the highest level of process control in Leuna and it finds its role in maintaining a stable unit operation around an optimum identified in the refinery’s “knowledge cloud”.

The above approach changes the role of APC. It becomes increasingly more important to provide operators with the right information upon which they can base their process decisions. The overall optimisation is carried out by planning, process and maintenance departments, who determine throughputs, product specifications and key operating parameters for availability, reliability, process stability and optimisation. The focus of APC is therefore changed to providing soft sensors , which operators use to adjust the required specifications. Depending on process stability, soft sensors can be applied either as open-loop indicators, or incorporated into a closed loop control system of varying complexity.

Optimisation by APC is usually quite effective in cases were feed forward control is needed. Leuna refinery has successful applications in sulphur recovery units (minimizing pure oxygen consumption), the gasification reactors (optimising supply of gasification agents), and in distillation columns (controlling product specifications). The distillation application does not necessarily need to be a closed loop system. Actually, when a reboiler is heat integrated (e.g. heated by a side draw of the main fractionator), closed loop applications become cumbersome and contain a number of constraints. It then becomes more convenient to give the operator an optimum set point through the open loop system, which he can then maintain as tight as practical, depending on the overall unit performance.

In summary, Leuna refinery maintains that APC has to be integrated into refinery’s overall optimisation process, interfacing between operators, the planning department and process engineers, and resolving the three main dilemmas:

Dilemma 1: APC versus Enhanced DCS
The authors maintain that much of what APC can achieve, the operations can accomplish by a smart use of DCS. Enhanced DCS applications are suitable for small feed forward control loops that can cover a considerable portion of APC’s potential use. Unless the mathematics and calculation efforts prove too complex, the choice between enhanced DCS and APC application is usual made according to process stability and safety considerations. It is to be noted that while the classical APC applications allow operators to define various constraints for the manipulated, disturbance and controlled variables, the enhanced DCS applications cannot easily provide such flexibility. Larger APC application, especially those that optimise complete process units, cannot be realised directly in DCS as the calculation efforts and constraint management become too complex.

Dilemma 2: Small versus Complex
The view of Leuna refinery is that in terms of controlled variable selection, APC applications must be more selective and also smaller, and avoid the use of large matrices in the hope that such a matrix will “find optimum”. Automatic optimum-finding rarely happens because complex applications are usually controlling against many constraints. These are better controlled by operator’s action, not automatically by empirical algorithms. Secondly, in a well operated refinery, processes are already running against at least one of the main constraints, in which case the operators may be apprehensive and careful in how the important variables are moved by the APC.

The APC organisation in Leuna allows a step-by-step development of APC applications. Unit optimization usually starts by removing the most critical process bottleneck. After relaxing that constraint the operator will try to move the process performance to a new optimum, until another constraint is hit. This step-by-step optimisation makes the decision “small vs. complex” quite straightforward and makes the APC solutions as “small” as possible. Furthermore, this strategy leads to optimization of APC engineering itself - the development of complex applications is automatically divided into smaller pieces and starts with the most valuable loop. The optimization project stops when the economic benefit of the next step decreases below a predefined level. Hence, APC-projects are directed just towards their economic optimum, which is an important factor that affects the success of APC engineering.