Larger APC applications
A wider scope allows for a more robust control and optimization solution as it brings unit-level interactions within scope. Traditionally, this approach has involved building a larger application that might cover, for example, a crude and vacuum unit. However, this introduces additional problems in terms of control horizons and settling times for sub areas.
A more holistic application considers the settling time of the entire scope. Consider an overhead product stream from one distillation column, which is dependent on reflux flow and reboiler duty. In normal practice, with separate control applications, this flow would be counted as a disturbance variable in any downstream application, or as an independent variable. However, combining these controllers into larger applications means the flow is no longer independent in the downstream application. This means the modelling associated with its use in other applications needs to be reconsidered, providing true interaction mapping.
In fact, application scope could be increased to cover the entire scope of a refinery. For optimal performance, this would require detailed information on shadow prices and operational boundaries.
First Principles Simulation
This type of application is the result of numerous successive stages. In the process simulation stage, for example, considerable engineering support is needed to ensure the simulation continually replicates the process units. Minor day-to-day changes, debottlenecks, or even control-loop tuning can mean the simulation and the process drift apart.
This approach also requires the simulation be supported with the latest product prices, shadow and energy prices, and a steady process operation. For instance, increases in processing speed mean latency is reduced and the solution has a higher applicability.
Meanwhile, reaction kinetics and thermodynamics help to determine the optimally feasible solution. An increase in severity on a fluidized cat cracker may increase the yield of naphtha, for example, but the result in higher catalyst circulation, cyclone velocity and catalyst attrition would cost more.
Historically, the key issue has been the ability to link simulations together to achieve convergence on multi-unit optimization. The degrees of freedom involved on multiple units coupled with time to run sequential models has resulted in latency being the biggest concern, with recurrence of the steady-state issue.
Whichever approach is taken, refiners need to be sure to review six key factors:
- Planning and scheduling performed at site(s)
- Strategy on APC schemes (if any) already implemented
- Layered technologies for intermediary control focused on wide process optimization
- Knowledge and the construct of operational boundaries within the refinery
- Technology layer (if any) used in conjunction with relevant applications
- Desire to implement closed-loop control and optimization
Achieving refinery-wide optimization calls for solutions that cover scheduling requirements, constraints on tankage, process dynamics, unit interactions and, more importantly, factors in dynamic constraints.
Leading refiners of the future will be those who pursue the approach that has the best fit for their plant. This may involve a hybrid approach to maximize optimization potential. Attention to the refinery workforce and their receptivity to changes in the technology will also be critical for successful, wide-scope optimization.