Advanced process control improves refinery, chemical plant operations

Next-generation advanced process control (APC) shatters processing barriers for process facilities.

By Hiroshi Wakasugi and Sanjay Venugopal August 7, 2019

To survive challenging conditions, process industry firms, including in the downstream oil & gas industry are actively seeking opportunities to optimize their entire supply chain. The economics of national operating companies are vastly different from international downstream organizations. Yet all are seeking ways to increase production margins, improve safety performance, and maximize equipment and plant reliability.

Feedstocks and energy are the major costs items, so downstream companies must view the entire supply chain to optimize feedstock usage while shopping for the best value without compromising quality. Likewise, downstream companies strive to lower energy consumption and minimize waste, such as octane giveaway in transportation fuels. One option downstream companies are pursuing to improve their competitive position is implementing automation/optimization projects, with advanced process control (APC) being a critical component of these efforts.

Tools to innovate

Automation and optimization strategies focus on how to schedule operations to produce high-demand transportation fuels, and petrochemical- and chemical-based products, while improving yields and conserving energy. Especially in petrochemical and chemical manufacturing, being a cost leader is an advantageous position, and cutting costs requires limiting process disturbances and downtime.

Since the 1980s, companies have applied various automation/optimization programs, including early APC systems, to attain better control of manufacturing costs and improve daily operations. First-generation APC programs suffered from numerous problems, including:

  • They were tedious to develop with available computer/software and field instrumentation technologies.
  • The APC software programs were difficult to maintain due to process/feed/grade changes.
  • Early APC models did not function well with nonlinear processing conditions.
  • The systems required frequent attention and maintenance from control engineers.

Newer APC technologies are addressing these issues, and delivering other improvements as well.

Over the past 20 years, APC applications have benefited greatly from the new computational power of PCs, more sophisticated “smart” field instrumentation and wireless technologies. Leading-edge computer capabilities and software can support complex simulation models, enabling APC programs to push operations closer to processing/equipment constraints. At the heart of this development is the ability to more closely control processing operations. The continuing development of multivariable control (MVC) accurately defines processing conditions and is thus a key component of APC.

Improved MVC supports predictive control via dynamic process models. This enables:

  • Maximization of process capacity ─ pushing process/reaction constraints under safe conditions
  • Minimization of downtime and increased unit/equipment availability
  • Freeing of operators’ time so they can complete other vital tasks
  • Reduction of product giveaway (minimizing waste/off-spec product generation)
  • Optimization of energy use across process units and entire plants.

MVC is one of the two main components of APC; the other one is enhanced regulatory control (ERC). Figure 1 is a graphical representation of benefits delivered by these components per expenditure of capital cost. Optimization also delivers significant benefits, but it not considered part of APC.

As Figure 1 shows, the APC components with the greatest benefit/cost ratio are ERC and MVC. Next-generation APC technologies are driving down costs in these two areas. An example is the platform for advanced control and estimation (PACE), co-developed by Shell Global Solutions (Shell) and Yokogawa, with similar products available from other vendors, with all these products here referred to as next-generation APC.

Next-Generation APC

As mentioned earlier, MVC and ERC are the key components of a successful APC strategy. Next-generation APC combines these two components into one software application (Figure 2).

The most successful APC installations take a holistic view of the plant to develop complex plant-wide applications within the same workspace and sequenced by multi-processors.

MVC models can be linear, linear dynamic or nonlinear. Process reactions and unit operations require using different models to create MVC models. Better modeling of the process vastly improves the overall control performance of the APC. Especially in the design stage, inclusion of additional measurement or process data is very useful for generating better models.

The controller uses process models to control constraints of the application. It can be configured by specifying controlled variables (CVs), disturbance variables (DVs), manipulated variables (MVs) and intermediate process out variables (POVs). The controller model structure can be confirmed using a graphic model viewer to provide clear visual information on CVs, DVs, MVs, POVs and model relationships. The graphic model viewer shows the control designer what POVs impact which CVs, and if particular CV’s are highly correlated with each other.

The estimator relates MVs and DVs to POVs. POVs can also feed other POVs, which provides more much flexibility in defining the relationship amongst variables, and in particular, how prediction errors are forecasted to other POVs. Only later in the design process does the control designer select POVs to become controlled variables (CVs). Model identification using POVs is often simpler and more robust as the control designer is fitting a series of models among POVs rather than an MV-to-CV relationship which may have many process operations between them.

A comprehensive simulation of the APC strategy is achieved via several steps:

  • Defining the simulation sequence: Set the processor execution and controller configuration details
  • Creating scenarios to simulate: Identify model disturbances or operator actions
  • Developing cases: Define one or more scenarios to review
  • Conducting the simulation: Collect and store simulation results for review
  • Comparing simulation results: Run simulation for multiple scenarios, and review by comparing simulation results.

Holistic APC development

Early APC technologies needed to work with other applications such as quality estimators and user calculations. Next-generation APC improves deployment of the control application via sequenced multiple processors (controller, estimator, and more). As illustrated in Figure 3, the process MVs and DVs are jointly used by the control and estimation layers.

Full estimation functionality is integrated into the controller, with the estimator using non-linear relationships. With this functionality, projections from the estimation layer calibrate the control layer, and gains in the control layer can be updated automatically. Next-generation APC enables the use of different model blocks in the estimation and control layers. Control engineers can configure a nonlinear model in the estimation layer to handle the nonlinearities of actual operations.

Nonlinear functions built by POVs mirror the true nonlinear dynamics of processing conditions. Such cross-utilization of data reduces application development time and greatly improves control capabilities. The graphic model viewer enables visualization of cause-effect relationships among MVs, DVs, POVs, and CVs ─ allowing the engineer to review selected variables in a static view.

As with any software installation, understanding the lifecycle is essential for success. The lifecycle phases for a next-generation APC implementation are:

  • Design
  • Staged
  • Live
  • Archive

The design (time) phase uses a single workspace for data management, process dynamics modeling, and scenario-based simulations, while the staged and live phases rely on the Run Time components of Human-Machine Interface, Data Repository and OPC Interface (Figure 4).

Preliminary control matrixes and model results from the design time are shared by the run time environment. As part of the Next-generation APC, two modes of operation are possible: Staged (read-only) and Live (read and write). Running two models in parallel allows the control engineer to validate any modifications in the Staged model before activating new control changes in the Live version.

APC in Action

A chemical company had several initiatives in progress at their plant in Spain. In 2016, the company decided to implement an APC project on a selected unit, with execution a joint effort with Yokogawa.

The unit selected for this APC project was the Cyclohexane (CX) distillation unit because it displays the typical process characteristics in which an APC value proposition can be applied:

  • High energy consumption
  • Operators faced with controllability challenges
  • Operation versus process constraints needs to be managed.

Potential positive impact on process performance was, in this case, the opportunity to improve the selectivity of the reactor fed by the Cyclohexane product from the distillation.

The “Platform for Advanced Process Control and Estimation” is Yokogawa’s multivariable control technology, jointly developed with Shell. This APC platform was released in Q4 2015 and is implemented worldwide, including on this project.

Process background

A simplified flow diagram of the process is shown below in Figure 5.

The scope of APC in this project includes stage 1 & 2 of the CX distillation. The CX is oxidized with air in the presence of a catalyst, yielding the main reaction products ON (Cyclohexanone) and OL (Cyclohexanol). The reactor conversion is very low, so most of the output stream is non-reacted CX, which is recycled to the reactor from the distillation units. These units consist of several distillation columns, with the main product the non-reacted CX. ON + OL is further processed in other units to produce caprolactam. Caprolactam is the precursor to Nylon 6, which is a widely used synthetic polymer.

Objectives and challenges

The APC project’s main objectives were:

  • Control the long recycle impurities (% ON + OL) as measured by an on-line analyzer. The impurities should be minimized to ensure good selectivity in the reactor. Before APC, this variable was not under automatic control because the analyzer is a chromatograph with a sample period of 14 minutes, resulting in a long lag time between process variable updates, and consequently poor PID control performance.
  • Maximize the short recycle and long recycle streams to improve reaction selectivity and reduce undesired by-products. This requires additional energy in the form of more steam required by the distillation columns, so process constraints must be kept under control when the recycle streams are high.

The second distillation stage has two columns in a double-effect configuration. Controlling the level in the bottoms of the second column at a desired target by adjusting the first column reboiler steam flow maintains unit mass balance because the steam rate influences the addition of fresh CX feed.

Plant operation is not straightforward for the operator and requires a lot of attention. One reason is the difficulty of controlling the above-mentioned level because of a long response time and a strong interaction between the first column steam rate and the recycle rates. Before APC, the level was controlled with a regular PID loop, but the performance was not adequate. Another difficulty for the operator is managing the unit constraints since it is desired to maximize the CX recycle rates, which requires frequent manual adjustments.

Maximizing the recycle will lead to increased column refluxes to maintain the purity of the recycle stream. Increasing refluxes will drive the unit against some constraints. Setting a high purity setpoint and increasing the recycle are both desirable in terms of impact on the selectivity, but a trade-off must be found between the two. If one sets a lower purity setpoint, one will be able to reach a higher recycle stream, and vice versa. Proper targets and limits for the APC were set based on observation of the actual unit performance, and further optimization is an ongoing process.

Project steps

The project was executed over a period of five months, with the following sequence of activities:

Phase 1: Feasibility study and economic evaluation.

Preliminary plant tests were carried out to verify the unit responses. A first pass APC design was developed. Expected economic benefits were estimated in order to verify the viability of the project.

Phase 2: APC implementation.

Base layer control checking to ensure proper performance of regulatory controls for the purpose of the APC, and fine tuning of some PID loops. This ensured the foundation was sound for the APC application

Step-testing: Step the relevant manipulated variables to collect data on the actual dynamic response. The data was then used to develop the model required for the controller.

Controller design and simulation: Configure the controller in a development system, test in simulation, and prepare the controller for implementation in the run-time control system

Commissioning: Closing of the APC loops, and monitoring of performance and fine tuning

Post-Implementation study: Verification of actually achieved benefits

All the above activities involved very close co-operation between the chemical company’s staff from their technology and operations department and Yokogawa’s APC engineers.

Economic benefits

The achieved APC benefits were found to be in line with the feasibility study estimates. The benefits come from improved selectivity in the reactor as a result of improved CX recycle purity control and CX recycle flow maximization, both made possible by the APC application.

General industry trends have resulted in downsizing of technical staff at many downstream sites, requiring outsourcing of some complex and specialized projects, such as APC implementations. Success when outsourcing an APC project depends on the experience and expertise of the selected APC consultant, so this decision must be made with care.

In general, APC projects take about six to twelve months for design and installation. A typical APC project involves:

Feasibility study: During the study, process/base-layer control and process flow diagram reviews and recommendations are conducted across the plant. Taking a plantwide view generates greater opportunities for improved operations. The study should also include a benefit analysis to show exactly how the project will improve plant operations.

Enhanced regulatory control: To be successful, an APC project requires a solid foundation. Basic loop tuning and enhancements are therefore often necessary before APC installation.

APC implementation: As each downstream facility is unique, a detailed APC design and review is necessary. Generic APC solutions fail to meet the fine details in controlling individual process units and downstream facilities. Applying a quality estimator to plant data allows fine tuning of the APC strategy. Step-testing generates a robust model to validate any plant responses to constraints. Simulation and testing are very important in finalizing the APC model before live commissioning. As with any rollout requiring process changes, operator training and documentation are essential.

Here are some of the key steps in APC implementation:

  • Pre-test
  • Step-test
  • DCS interface engineering
  • Commissioning
  • Handover

Throughout each step, the consultant should provide constant monitoring to advise operators with respect to setting the right limits for the controller variables, updating loop tuning parameters as necessary, and updating models to accommodate drastic changes in process dynamics due to turnaround or equipment changes.

Once the control system is activated, post implementation reviews of the new APC performance are highly recommended. Process conditions constantly change; consequently, audits and performance reviews confirm that the APC strategy is adhering to the design goals.

Innovating for tomorrow

APC technologies have been available for more than 30 years. As new hardware and software innovations appeared, the power of APC was quickly adopted by downstream companies. There are many APC solutions available. However, the application of APC has had problems. Looking forward, APC users want:

  • Quickly deployable APC solutions requiring less engineering time to construct and maintain
  • Visualization tools that allow operators and engineers to track and monitor process unit operations while checking control variables
  • Integrated platforms for improved engineering efficiency.

The selected next-generation APC technology needs to meet these user requirements. The graphic model viewer interface uses an organized hierarchy for easy visualization of process conditions and control-system status. Operators and engineers can quickly access relevant information and implement changes to the APC when necessary.

By combining all these tools and functionality, APC can be implemented and maintained to deliver significant and sustainable improvements to plant performance.


Author Bio: Hiroshi Wakasugi is the general manager of the advanced process solutions department at Yokogawa; Sanjay Venugopal is an advanced process control specialist at Yokogawa.

Related Resources