The ability to precisely control a plant and optimise its operation has always been an important priority for refiners, but software can help address these and other challenges, explains Paul Turner of AspenTech.
Process plants often don’t perform as well as their operators would hope. Plant functions that depend on human operation can lead to unreliability, which in turn leads to variations in product quality, high energy costs and inconsistent volumes of materials processed through the plant; all of these can cause operations to fail to run to their actual limit. To improve quality, operators need to consider changing their plant-control strategies.
Instead of operators manually adjusting control units for specific variables in the plant, advanced systems provide models that automate regulatory and constraint control, as well as delivering process optimisation.
Advanced process control (APC) is a general term used to describe different types of process control tools and methodologies frequently used for solving multivariable control or discrete control problems. It helps to improve the operation of production processes, resulting in the continuous management and optimisation of complex process interactions. A dynamic, multivariable interaction model is designed with empirical control software to predict the future path of the process, compare the information with the operating constraints and implement an efficient plan to the target set by the company. The resulting benefits of APC are maximum profits, increased capacity and improved operational performance.
Studies have revealed that APC can save significant annual operating costs and is able to generate additional revenue, with payback in less than six months.
Steps to identify the opportunity
APC is about real-time data processing to positively affect the overall optimisation of the production process in order to operate in the most profitable way.
The project starts by identifying the business opportunity. Once an agreement has been made by the key stakeholders that there is scope to further maximise the capabilities within a plant, then a plan is developed to deploy the multivariable predictive control within the operation.
This plan will also determine the factors to achieve higher margins with good return on investment. Once the business feasibility study has been carried out to determine likely benefits, a pre-test is conducted with technical staff from the operator and the software vendor. This stage is to ensure that the process is understood by all stakeholders. A functional design specification will then be created by a software vendor’s lead engineer. This document will outline what the controller will do, what process parameters APC will manipulate and where the process will generate the highest profit from the plant.
The next step is to collect data from the process in order to build the process model. The software vendor will plan step testing typically over two visits — the first being the pre-test — and time will be spent in the control room to ensure that all the relevant regulatory controllers (PID controllers) are tuned as required.
Once laboratory data has been collected appropriately and verified, the process engineers will run a series of single variable step tests one at a time, observe how the process responds and electronically record the process changes that have been made to help optimise the specific area of the operation.
This data will allow the engineers to build a preliminary model of the process. Once this has been analysed at the head office, simulations will be carried out and the software vendor will know what further tests need to be conducted to complete the final model and finish the proper step-testing phase. At the same time, the functional design will be edited once more towards its final state.
Step testing can be completed today within a short period of time without causing production loss or process interruptions.
Recent innovations in step-testing technology also allow data suitable for model building to be generated in a non-invasive manner where an intelligent controller is implemented using the models generated from the pre-test data (realising immediate benefits), which then balances robust closed-loop control against making setpoint moves that will generate data with suitable characteristics for model identification.
This unique capability saves enormous time for the customer and does not cause costly disturbances to the operation.
The Aspen SmartStep automated tester and the new ‘Calibrate Mode’ technology of Aspen Adaptive Modelling knows all the interactions in the process and uses the preliminary process model to be able to process simultaneous calculations, elicit information and refine the model accordingly with the minimum interruptions to the plant, while achieving constraint control of the process.
This entire process can uniquely reduce the implementation timeline from months to weeks, helping to manage the efficiency and time constraints set by the customer.
The process industries encompass an incredible range of products, processes and plant configurations. A variety of APC solutions are needed to address such a diverse set of production processes.
Many companies have utilised solutions from AspenTech. For example, aspenONE APC provides several approaches for advanced control applications: Aspen DMCplus Controller, Aspen Non-linear Controller and Linear State-Space Control — all encapsulated in a single, common environment called the Aspen Control Platform.
Prediction and inferential measurements are essential for operators today. These application tools are powerful for the modelling of inferred product qualities. Inferentials are a supplement for infrequently measured qualities or critical sensors and are also used to support environmental compliance. APC provides a rich choice of model types, making it possible to implement linear or non-linear inferential sensors online. Flexible analyser and laboratory modules automatically adjust the inferred qualities to ensure accuracy.
APC enables the use of online simulation based on both empirical and rigorous models. Typical applications include improving controller models, generating shadow targets for operators, providing what-if simulations to aid operators and engineers in identifying and resolving process problems and generating key performance indicators (KPIs) for real-time performance management.
In a recent case study, a Czech refinery and petrochemical group reduced energy costs with the APC solution while realising estimated annual energy savings of approximately €2m (£1.6m). The company also reported a return on investment from the technology within just six months.
APC is a powerful technology that enables companies to increase throughput, improve product quality, reduce energy and raw-material usage and increase overall operational efficiency while keeping the process between the safe limits of reliable operation. Simply, there are many benefits, which help process manufacturers maintain quality and satisfy customer demand.
Today, the use of APC has been widely adopted across the process industries. Not only can it efficiently scale to any control problem size, but it has also been successfully applied to virtually every control problem in refining, chemicals and petrochemicals processing.
In the highly regulated environment in which process industries operate today, APC software tools can significantly help manufacturers to reach the limits of the operation and optimise performance while also squeezing out the last drop of profit that can be achieved from the business.
Paul Turner, senior design manager, AspenTech