Apr-2015

Advanced process control in FCC and hydrocracking units

Implementation of advanced process control of two major units produced significant savings with short payback times

GRZEGORZ OLESZCZUK and MARTA DYLEWSKA
Honeywell Advanced Solutions

Article Summary

Given the competitive and cost pressures facing refineries, solutions to boost company profits and reduce expenses are highly sought after. Advanced process control (APC) in this respect is among the most attractive solutions. Through process optimisation, APC not only delivers and often exceeds expectations for cost savings and productivity gains, but does so with relatively little outlay. Few technological changes in the units addressed are required, and there is no need to invest in new equipment.

Nevertheless, this should not diminish the importance of the implementation and the expertise required. The fact that APC does not require the investment or disruption associated with other options does not mean it is simpler. For every quick win there are complex problems that must be overcome. Two aspects are of particular note. First, the APC should reflect current market conditions for product margins, feed costs and other variables to ensure the process is effectively optimised. Second, empirical models should underpin the solution by employing step tests and real-world observations of plant responses. Results from solutions relying on analytical models of the process will invariably deliver disappointing results.

Two APC applications implemented at PKN Orlen in Plock, Poland, in 2013 illustrate these points. The refinery, Central Europe’s biggest, chose the hydrocracker and fluid catalytic cracking (FCC) units to act as pilot schemes with a view to implementation of APC throughout the group.

“We need to continuously improve our already high production standards, so we decided to undertake a production optimization program for the whole group. Advanced process control is one of the main components, because it impacts the production results very quickly and positively,” said  Krystian Pater, Member of the Management Board for Production at PKN Orlen.

Technology used
Both projects used Honeywell’s Profit Suite of APC and optimisation applications. The key software, Profit Controller, reads process data from the existing control system and, based on the process model and its range control algorithm (RCA), sends optimum set point values back to the control system.

Three types of inputs and outputs are used:
• Controlled variables (CV) which are usually process values (such as qualities, flows, temperatures and pressures) that must be kept in safe and optimal ranges
• Disturbance variables (DV) which are read-only variables such as feed quality and ambient temperature, or parameters outside the local control system’s control such as disturbances from other process units
• Manipulated variables (MV), the values the APC sends back to the control systems, usually represented by the list of set point values for PID controllers.

The RCA algorithm calculates and predicts CV values based on the process model and MV movement. The objective of APC is to push the steady state MV values closer to the predetermined ideal operation point.

“Honeywell executed two very big APC projects on the HCK and FCC units in parallel. Both covered more than 10 multivariable controllers and several inferential calculations. Cooperation with Honeywell had a positive impact on our APC Team through sharing Honeywell’s experience and knowledge in the field of refinery process optimization,” said Marek Bożek, PKN Orlen APC Team Manager.

FCC unit
The FCC unit (see Figure 1) is among the most important process units in oil refineries, but it handles a physically complex process that is difficult to operate and control. The reactor, regenerator and main fractionator sections are highly sensitive to process dynamics and face multiple constraints. Units are also required to run under different operating modes, reflecting changing economics, market demand, mechanical constraints and refinery feedstock. The significant economic incentive to push FCC constraints close to limits is matched by operators’ reluctance to do so for fear of process upsets. In short, FCC units are ideally suited to advanced control solutions.

APC’s benefits derive from several sources:
• Feed maximisation: the FCC unit operates on various types of feed from different units in the refinery. Consequently, throughput can be a downstream constraint
• Conversion maximisation results in higher propylene yields by optimising the reaction severity in the reactor and regenerator section
• Distillation control and yield shift are achieved by producing more valuable products from the main fractionator and gas plant section. The most valuable products for the latter are propylene and light gasoline, which is routed downstream to the Prime-G desulphurisation unit.

In view of the specific processes in each section of the FCC unit, the APC’s targets were divided by operating area. The top three, for the unit feed system and reactor and regenerator section, for example, were to maximise throughput (taking into consideration unit constraints and planning requirements), improve conversion control and maximise propylene yield. Those for the fractionation section included: maximising the yield of the most valuable main fractionator distillates, subject to quality specifications; minimising CLO draw; maintaining the heat and liquid/vapour loading balance using pumparound streams; and reducing energy consumption. Objectives were similarly developed for all of the FCC unit’s sections.

Based on these objectives, six multivariable controllers were designed for the FCC unit. One controller divided into two sub-controllers covered the reactor/regenerator section and fractionation section. Another six were designed for the reactor/regenerator and main fractionator; the debutaniser; depropaniser; butylene splitter; the new propylene splitter; and the old propylene splitter.

All controllers were designed to execute their calculations at one-minute frequency. Under-pinning the controllers were multivariable process models for each area.

The role of inferentials
Inferentials, or ‘soft sensors’, are crucial elements in many APC projects. Calculations using historical data from lab samples and key process parameters can predict quality parameter values in place of lab analyses. A number of inferentials were implemented for the FCC unit to meet product quality specifications. Since the implemented inferentials predict lab values in just one minute, against the eight hours required for lab analysis, they enable more responsive control. Testing indicated that these inferentials were highly accurate. Figure 2 shows the trend for the inferential and real lab samples for 90% distillation of gasoline. Green dots represent lab samples; the brown line is the soft sensor prediction. Even lab samples far away from the main trend were accurately predicted (as in the sample with the green arrow).

Results
All requirements for the APC project were fully realised. A key objective – a 1% increase in throughput – was surpassed (see Figure 3). The multivariable controller increased feed smoothly despite oxygen constraints in the regenerator, compressor capacity, and product quality constraints measured after the main fractionating column.

The most valuable benefit, however, came from improving propylene purity, so boosting the yield. During test run, the propylene purity was approximately 99.63% (v/v, based on laboratory data), against a specification value maximum of 99.6%. Figure 3 shows the trend of propylene purity during APC start-up. The process stabilises immediately.