Using advanced control for PROFITABILITY

For companies operating today's process plants there is an ever-increasing drive to maximise throughput and yields and to minimise energy consumption and environmental emissions. The often-conflicting aims of maximising profit whilst minimising the cost to the environment can be achieved with plantwide advanced process control (APC).

In the last ten years techniques such as model-based predictive control have become a practical industrial reality, particularly in the oil refining industry.

Connecting computers together into a plantwide network overcomes some of the barriers to implementing plantwide APC. With the advent of IT, it is now possible to integrate these advanced control techniques with business planning and scheduling systems, resulting in remarkable economic benefits.

Today's improved control systems, combined with modern IT, provide a rich diversity of control techniques encompassing basic regulatory control, advanced multivariable control and process optimisation. When expertly applied in refineries this extensive control suite can yield significant benefits by reducing energy and material costs and increasing quality and production.

For many years regulatory controls have been used to achieve maximum throughput and yields, with minimum energy use and environmental emissions. Standard regulatory controls, which include simple control and cascade feedback loops, do a good job of controlling the basic variables on the plant - flows, levels, pressure, temperature and so on. This form of control has always played an important part in process control and will continue to do so. However, in business-related control situations, regulatory controls perform less satisfactorily because they do not focus on the variables and parameters on which the plant's economic performances depend.

This is where the strength of APC lies. APC improves profits by increasing stability and allowing the process to operate more closely to its economic and process constraints. APC is designed to sit above regulatory controls and to adjust the set points of the basic loops, achieving better control of the `higher level' variables. It therefore depends on basic regulatory controls to implement the constrained setpoints. If these controls are poorly designed or tuned, APC cannot produce the desired returns. Accurately tuned regulatory control is therefore a prerequisite for proper operation of APC. The purpose of basic regulatory control is to hold the process variables near steady state to allow APC to meet its objectives.

Advanced regulatory control, on the other hand, uses standard calculation blocks available in today's DCSs (distributed control systems) to create multi-layer control applications which drive the process towards the operating objectives and minimise the control effort required to adjust quality. Advanced regulatory control strategies include feed-forward, ratio, adaptive and self-tuning control.

For certain types of processes, traditional and advanced regulatory constraint controllers still give unsatisfactory performance - processes with difficult dynamics, for example, or multivariable processes with strong coupling, processes with dead time, processes with measurable disturbance and processes with constraints on the manipulated and controlled variables. These limitations can restrict the potential gains available through cost-beneficial operation of industrial processes.

Model-based Predictive Control (MPC) addresses control problems where regulatory control fails. MPC achieves higher plant throughput and brings added economic benefit where there is an opportunity to push the throughput against constraints.

The reliability of model-based multivariable controllers depends on having reasonably accurate real-time plant data and, normally, on the assumption that the process is linear. But these conditions cannot be assumed in all plant situations. Processes that are poorly understood or whose behaviour is variable are very difficult to model but newer techniques based on expert systems, fuzzy logic, neural networks, genetic algorithms and multivariate analysis are available for these types of control problems.

The highly interactive nature of most processes, especially in the oil industry, means that they are subject to large changes in feed composition and product constraints. To run these plants at their optimum economic constraints is very difficult, if not impossible, under manual supervision. Optimum operation and profitability of this type of plant is realised by dynamically adjusting feed rates and operating conditions as unit constraints and market demands change.

Traditional advanced process control and off-line, open-loop optimisation may not deliver substantial economic benefits because of the constantly changing conditions of today's plants. For plants to be competitive, economic benefits must be realised through improved process control techniques, which is why a number of companies have now decided to pursue the use of multivariable model predictive control and closed loop Real-Time Optimisation (RTO).

RTO uses detailed chemical engineering plant models that are regularly fitted to real-time plant data as the basis for a non-linear optimisation problem. Generally speaking, the optimiser maximises an economic objective function whilst respecting equipment and scheduling constraints.

As outlined earlier, MPC ensures that plants operate at an optimum economic and safe point. RTO determines these points in the optimal way by gradually pushing the process in a controlled manner to the most profitable constraints. Once the optimum setpoint values are calculated, they are passed directly to lower level advanced controllers for implementation. Identifying the profitable set of process constraints and then driving the process to operate at these constraints produces economic benefits such that the RTO system payback can be less than a year. Some of the many benefits of implementing RTO include decreased product variability, decreased energy consumption and increased yields.

Moving on from RTO, the next level of plantwide APC is global site optimisation using real-time data and plant models. This in effect represents a large-scale version of RTO technology that recognises inter-unit constraints and prevents localised optimisation. Several companies are developing open-loop versions of global site optimisation software. There is a potential for financial benefits in this technology, especially in energy management.

Thanks to advancement in IT technology, planning and business systems can now also easily be interfaced to RTO and overall plant control systems. Typically market requirements, product inventories and recent production histories are presented to the planning computer which then generates daily production targets for specific process units based on feedstock requirements and current product values. The RTO system will use the output from the scheduling and planning system as its setpoint (target), applying specific constraints (for example, qualities and short-term targets) to drive its objective function. In this way the plant avoids overproduction when sales are down and maximises production when demand is high. The cost of storage is reduced to a minimum and energy is rationally utilised.

On many plants the opportunity exists to increase profitability by increasing plant throughput. This is dependent on several factors including the availability of additional feedstock and extra market demand, and the plant being constrained by the quality of its present control system. Plantwide APC technologies also improve plant stability by reducing the effects of disturbances on plant. The plant therefore runs more smoothly than with regulatory controls. This increase in stability creates the headroom required to move the plant to a more profitable operating point, and gives operating staff confidence to work towards a more profitable regime. The financial benefits are typically in the range 2-6% turnover - for a plant with a turnover of £50m/yr, the annual benefits are in the order of £2m. Table 2 shows the benefits from APC and Optimisation for a typical refinery.

It is clear that the emerging new automation systems have begun to take on many of the characteristics of high-level information systems. Today's challenge is how to take advantage of these powerful new tool sets and apply them directly to the problems faced by manufacturing. PE

This article is based on an original paper by Tony L Moro of Foxboro GB.

{{Benefits from APC and optimisation in oil refining

Refining unit Benefits Typical unit Typical benefits c/bbl size `000bbl/d $million/y

Crude/vacuum 5-10 135 2.5-5.0FCCU 20-40 38 2.8-5.6Reformer 5-25 25 0.5-2.3Hydrocracker 10-30 26 1.0-2.9Alkylation 5-35 10 0.2-1.3Delayed coker 10-40 31 1.1-4.5Light ends 10-20 40 1.5-2.9Isomerisation 5-15 30 0.5-1.6Lubes vacuum 10-40 10 0.4-1.5Aromatics 15-20 15 0.8-1.1MTBE refinery 28-40 4 0.4-0.6}}