Content area
Taube et al discuss a digital twin-based approach to distillation control education. The traditional methods of teaching process control often fail to provide practical lessons on distillation control. Distillation columns in modern chemical and petrochemical production facilities operate at tighter specifications and higher recovery rates, making control more challenging. The article highlights the need for continuing education and competence building in distillation process control. The current methods of teaching process control are highly abstract, resulting in a limited understanding of real-world applications. They propose the use of digital twin-based training simulations to bridge this gap. These simulations provide a platform for engineers, operators, and managers to learn about distillation operations in a hands-on and interactive manner.They also emphasize the importance of process understanding, control systems, process dynamics, and instrumentation in achieving desired plant performance. Digital twin models developed on process simulators offer a realistic representation of distillation processes and can be used for operator training.
Process control education often fails to teach students the practical lessons of distillation control. Digital-twin-based training simulations may hold the key to more robust process control education.
Aside from the control of complex reactions, the robust and on-specification control of distillation columns is one of the most challenging aspects of operating modern-day chemical and petrochemical production facilities. Many distillation columns are operated at tighter product specifications and at higher product recovery rates than in previous decades, while stricter environmental regulations mean bottom and side stream specifications must also be tightly managed (1).
Distillation columns can often have challenging operational specifications (2) and/or complex arrangements (3), which makes control, not to mention optimization, of these processes a great challenge; and, with the drive for improved energy efficiencies, they are getting increasingly more complicated (4). As a result, even experienced process operators and engineers need to think about the complex interactions between thermodynamics, internal hydraulics, mass and energy balance limitations, and controller configuration when making operational changes. Considering these shifts and the inevitable loss of experienced operators and engineers, continuing education and competence building are needed in the domain of distillation process control.
However, as many students, as well as practicing engineers, will attest, the way in which process control is taught in many chemical engineering curricula, as well as many industrial continuing education courses, is highly abstract. As a result, students and practitioners have a limited grasp of the influence of process dynamics, as well as steady-state behavior, on the real-time performance of an operating plant. That is, they are unable to apply the theory as it is taught in the traditional way to the real world as they find it (1). Furthermore, those that seem to have mastered the craft of process control can also exacerbate the issue by being unable to effectively communicate with both engineers and operators how the controls affect the process, in both steady state and dynamic aspects. Thus, process control is often perceived and treated as a "black art," mastery of which is reserved to those who speak a language and describe concepts indiscernible by others!
This article discusses the development of a digital-twinbased approach to continuing education of real-time control of distillation processes. This approach is accessible to engineers in industry, operators, and even managers.
Challenges of process control education
From a chemical engineering education perspective, a key hurdle that hinders continuing education in control theory is the use of abstract mathematical concepts, such as Laplace transformations, to teach distillation control. While these concepts might be suitable in a purely academic or mathematical environment, this approach does not work well for continuing education in a plant operations setting. Many individuals find these abstract mathematical concepts hard to reconcile with industrial distillation column operations, which frequently exhibit nonlinear and other complex behavior. Even if these concepts can be applied to accurately model complex distillation processes, engineers, who may have been familiar with these abstract mathematical concepts in their academic education, will likely have forgotten these concepts (due to lack of practical use). Furthermore, the mathematical complexity alienates many operators and other personnel who have never been taught these concepts.
As a result, many continuing education programs attempt to teach distillation operations through "thought exercises" or experiments that discuss complex phenomena and identify how these phenomena can be tackled. Often, these thought exercises focus on identifying potential control strategies rather than holistic control structure design, implementation, and tuning in operations.
While distillation courses have played a great role in transferring tacit knowledge and improving the general understanding of distillation among engineers and operators, one disadvantage they suffer is a lack of a platform where the individuals can practice and explore distillation operations and dynamics in real-time (or faster). Another disadvantage is the need for highly experienced "authority" level instructors who can, through simple thought exercises, shed some light on distillation control.
Due to reduced investment in in-house workforce training over the last few decades, the number of experienced authorities available in the marketplace has rapidly declined to a mere fraction of that from previous decades. Consequently, there are too few potential instructors and/or mentors available to educate the less-experienced staff, who now comprise an increasing portion of the workforce.
The push from the industry toward digitalization opens the possibility of using operator training simulators, a type of digital twin, as a potential tool for continuing education. The main benefit is the ability of engineers and plant operators to have "hands-on" experience in handling complex situations, carrying out root cause analysis, and understanding the consequences of operational decisions.
In industries such as commercial aircraft and pipeline operations, this type of operator training simulator is used extensively to train individuals on how to appropriately respond to adverse events, which cannot be recreated in real life without putting lives and equipment at risk. The objective of these exercises is to improve an individual's overall knowledge in handling these situations and to enable them to rapidly build up the tacit understanding needed to tackle future complex real-life problems. To this end, using digital-twin-based operator training simulators to provide both engineers and operators with a platform to learn about distillation operations holds the key to the future of continuing distillation and control education.
The purpose of process control
In both academic and industrial treatises on process control, the stated purpose for the design of control schemes is to reduce process variation (Figure 1). While it is generally understood that having well-designed and tuned controls is vital to achieving the desired real-time plant performance, this definition for the purpose of process control falls short of giving a full and complete purpose that is worthy of study and mastery (1). To this end, we suggest the following definition: To stably, robustly, and predictably maintain product qualities in the face of measured and unmeasured disturbances with the least total and incremental energy input (i.e., minimal movement) to the process, while satisfying all physical constraints and limits.
Another element that is often underappreciated about process control is that achieving the desired performance is predicated on the combination of multiple bodies of knowledge. As shown in Figure 2, process understanding is foun- that resembles their distributed control system (DCS) operating graphics. As a result, the proposed approach enables individuals and organizations to focus on gathering practical and applicable process dynamic and control insights as opposed to learning complex mathematical manipulations that they will find difficult to apply to a real-world problem.
Adopting a hands-on learning approach
Traditional continuing education courses often use a classic classroom approach in which experienced instructors encourage participants to bring practical distillation operations and control questions to be discussed as part of the course. The remainder of the course is traditionally lecture-based. For operators and engineers who are eager to learn, this type of course offers an opportunity to cover multiple aspects of distillation control. Some examples include understanding the influence of a given controller pairing, or the need for extra attention when using one type of process equipment (e.g, reboiler design) as opposed to another. However, many of these aspects are limited to thought experiments.
On the other hand, by using a hands-on active learning approach enabled by digital models, individuals are able to try out different control structures in a safe environment. They also learn to analyze real-time data and explore potential scenarios that they may encounter. As a result, they take back specific knowledge to their organization that is readily applicable to revamping control structures and improving operational performance. This type of course can also be the catalyst for encouraging younger process engineers to delve further into the details of process dynamics and control structure development, which could ultimately lead them to become process control engineers or, at a minimum, more capable process engineers.
A fundamental paradigm shift in the understanding of distillation concepts, particularly for participants from refining backgrounds, is commonly noted as a key takeaway from the hands-on digital-twin-based learning experience. One example of this conceptual shift relates to participants' understanding of the reflux ratio. Traditionally, students are taught that reflux is an independent variable; digital-twin simulations help exemplify why this isn't true.
If one views distillation from the perspective of a "black box," that is, not knowing what is inside the box, and knowing only the inputs and outputs, then one observes a distillation column as shown in Figure 4a: the feed stream and energy input to the column as inputs and two product streams as outputs. However, knowing that the mass balance must be maintained, one realizes that the two product streams are not independent of each other and their movements must be coordinated, as illustrated in Figure 4b: as one outlet stream flowrate increases, the other must decrease and vice versa. Lastly, to give context to this concept, a typical two-product distillation tower is placed inside the box, as shown in Figure 4c. Here one observes that reflux is not, in fact, an independent variable as it is inside the box and not available for independent manipulation (1).
Example one-day workshop
The immersive workshop schedule shown in Table 1 is designed to equip process engineers with practical insights and skills needed for designing and troubleshooting distillation control applications. Based on the book, A Real-time Approach to Distillation Process Control (1), this course strikes a balance between process and control fundamentals with real-world scenarios. The workshop leverages prebuilt process simulations to train attendees on developing control applications using real-world process information. This approach provides hands-on training on the intricacies of distillation processes in a safe, controlled, and realistic environment. The workshop is also ideal for graduate students seeking practical knowledge of distillation process control applications in industrial settings. By participating in this workshop, learners gain practical skills that can be immediately applied to their work in distillation operations.
Half of the course is dedicated to carrying out hands-on exercises. The first hands-on exercise involves setting up dynamic industrial process simulations of distillation processes. The second focuses on developing the competencies to implement robust base-layer control structures.
Initial trials of the one-day workshop. The course has been tested at the Univ, of Auckland with a group of senior undergraduate and masters-level students, where it was well received. In particular, the students found the hands-on sessions to be informative and the skills learned in setting up dynamic distillation simulations to be readily transferable to other unit operation models. The students also found the time-domain-based process control concepts to be easily grasped with no need for "mathematical gymnastics" while also finding it to be a more natural extension of process design and steady-state operational analysis. Overall, this initial trial shows the proposed approach of hands-on digital model-based learning taken to distillation process control as a promising alternative to traditional approaches. In comparison to the traditional approaches, the proposed digitaltwin-based approach is intuitive and, based on the nature of the approach, provides skills that are readily transferable and applicable to real-world problems.
To efficiently carry out this workshop, instructors (and trainers) who have a sufficient level of knowledge and experience in developing dynamic process simulations in a given simulation platform are required. For the Univ, of Auckland workshop, the simulation platform chosen was Aspen HYSYS; however, the course content and fundamentals are simulator-agnostic. While a full course would require students to build simulations from scratch (i.e., start with a blank flowsheet), a short course has basic models already developed for use. This allows the attendees to focus on the design of control structures, rather than having to learn all the intricacies of the specific simulator software. Figure 5 illustrates an example of a dynamic distillation column simulation with a base layer control (in Aspen HYSYS).
It is also beneficial for the simulation exercise to be carried out in groups consisting of two participants as this was shown to spur-on constructive debates between the participants. Having multiple instructors actively engaged with the course participants was also beneficial. It should be noted that this course can be carried out by all trainers who have a general competency in developing dynamic simulations on similar platforms and have practical experience in setting up time-domain-based industrial process control systems. Ref. 1 contains sufficient information and worked examples to facilitate such a development.
Conclusions and recommendations
The digital model-based approach to distillation process control has been executed successfully in an academic context as a senior undergraduate elective at the Univ, of Auckland over the past three years. The course is now being targeted to practicing engineers and operators alike and, hence, takes a more practical approach to tackling the problem of process controller analysis, design, and tuning. In comparison to established methods of highly mathematical Laplace-based process control design and tuning, this approach is much more intuitive, as participants do not require an advanced understanding of complex mathematics. As a result, the lessons learned from this approach are directly transferable to real-world engineering problems that are found at every operating plant.
Literature Cited
1. Young, B. R., et al., "A Real-time Approach to Distillation Process Control," Wiley, Hoboken, NJ (2023).
2. Udugama, I. A., et aL, "Side Draw Control Design for a High Purity Multi-Component Distillation Column," ISA Transactions, 76, pp. 167-177 (May 2018).
3. Halager, N. S., et aL, "Modelling and Control of an Integrated High Purity Methanol Distillation Configuration," Chemical Engineering and Processing - Process Intensification, 169, #108640 (Dec. 2021).
4. Udugama, 1. A., et aL, "Separation of Middle Boiling Trace Compounds by Distillation: An Investigation of Practical Implications of Complex Column Arrangements on an Industrial Methanol Distillation Case Study," Asia-Pacific Journal of Chemical Engineering, doi: 10.1002/apj.2588 (Oct. 2020).
Copyright American Institute of Chemical Engineers Jan 2024