Conference Agenda

Temperature regulating in liquid tanks is critical in the chemical industry and typically relies on thermometer feedback. However, due to the complexity of flow and thermal fields, unsensed local temperatures can deviate from desired limits, highlighting the need for improved tank temperature modeling. The absence of internal thermal or flow data, however, presents a challenge for developing and validating predictive or control models.
In this study, a virtual model for regulating liquid tank temperature was developed using computational fluid dynamics (CFD) based on the Navier-Stokes and energy equations. The inlet temperature was set to a constant value (10 °C for cooling and 50 °C for heating) with a steady flow rate, 770 ml/min. The CFD model was validated against experimental data within a water tank for temperature and flow fields in typical heating and cooling modes with adiabatic walls. Using this virtual CFD model, several new cases were simulated, involving two mechanisms (1) a fuzzy set to trigger the feeding when the temperature of a virtual sensor falls outside [24 °C, 26 °C] and (2) the imposition of unfavorable temperatures on the walls representing ambient influences. The simulations revealed temperature response discrepancies between the sensor and other interior points, which can be over 2°C. Thus, the constructed model can be used to generate valuable datasets for temperature regulation for liquid tanks.
This virtual CFD model offers an economical and reliable approach for advancing temperature prediction and control models in the chemical industry, supporting improved material quality control and energy efficiency.

Dynamic or Programmable Catalysis is an innovative strategy to improve heterogeneous catalysis processes.¹ The technique modulates the binding energies (BE) of adsorbates to the catalytic surface, enabling the periodic favoring of different reaction steps. Such forced energetic oscillations can overcome limitations imposed by the Sabatier Principle, allowing for higher overall reaction rates, unattainable through conventional steady-state methods. Researchers confirmed the effects of dynamic catalysis computationally through the sequential simulation using forward integration of the differential-algebraic equation systems that govern catalytic processes.² In this work, we implemented a simultaneous simulation approach by formulating the problem as a boundary value problem with limit cycle or periodic boundary conditions, directly solving for the cyclic steady state (CSS). The approach was implemented using the optimization algebraic modeling language Pyomo.DAE³, to support automatic transcription of the differential equations, and the solver IPOPT.⁴ The methodology improved run times by orders of magnitude. The computational efficiency of the simultaneous approach allowed the implementation of derivative-free optimization methods to determine optimal parameters (within bounds) for the shape of the forcing signal that describes BE oscillations. For the continuous forcing functions, it was possible to implement gradient-based methods and modify the Pyomo/IPOPT framework to determine the waveform parameters that directly optimize the overall rate of reaction (avTOF) using IPOPT. For the square wave, we verified an increase of four orders of magnitude of the avTOF when compared to the peak of the Sabatier Volcano of the static system. These results demonstrate both the potential of dynamic catalysis and the value of using optimization techniques for waveform design. We now aim to extend the methodologies to more complex systems of industrial interest. Wittreich et al. conducted the most comprehensive study on dynamic catalysis in complex systems, working on ammonia synthesis⁵. Applying the simultaneous simulation approach to such a system should show that the methodology is extensible to more intricate scenarios. Furthermore, applying waveform optimization to this system could lead to even higher reaction rates than those reported. We also aim to further develop mathematical approaches for more direct analyses of dynamic catalysis system behavior.

References

(1) Ardagh, M. A.; Abdelrahman, O. A.; Dauenhauer, P. J. ACS Catal. 2019, 9 (8), 6929–6937. https://doi.org/10.1021/acscatal.9b01606.

(2) Ardagh, M. A.; Birol, T.; Zhang, Q.; Abdelrahman, O. A.; Dauenhauer, P. J. Catal. Sci. Technol. 2019, 9 (18), 5058–5076. https://doi.org/10.1039/C9CY01543D.

(3) Nicholson, B.; Siirola, J. D.; Watson, J.-P.; Zavala, V. M.; Biegler, L. T. Math. Program. Comput. 2018, 10 (2), 187–223. https://doi.org/10.1007/s12532-017-0127-0.

(4) Wächter, A.; Biegler, L. T. Math. Program. 2006, 106 (1), 25–57. https://doi.org/10.1007/s10107-004-0559-y.

(5) Wittreich, G. R.; Liu, S.; Dauenhauer, P. J.; Vlachos, D. G. Sci. Adv. 2022, 8 (4), eabl6576. https://doi.org/10.1126/sciadv.abl6576.

Mathematical modeling and simulation are pivotal components in process systems engineering (Vassiliadis, 2024). Focusing on polymerization processes, identifying microscopic properties of polymers is necessary for advancing kinetic comprehension and facilitating industrial applications. Mathematical modeling methods for polymerization processes can be categorized into two types: the deterministic method solving mass balance equations and the stochastic simulation using the Monte Carlo (MC) method. For complex polymerizations, like the controlled radical polymerization (CRP) with reversible reaction pairs, it is challenging to derive a rigorous model and solve it deterministically. Direct solution of the resulting ordinary differential equations is computationally infeasible due to memory resource limitations and model stiffness. These hurdles can be surpassed via coarse graining and order reduction for computational efficiency, albeit with a trade-off in applicability. In contrast, the kinetic Monte Carlo (kMC) method, based on Gillespie’s (1977) stochastic simulation algorithm, is conceptually straightforward, focusing on individual molecules. The kMC method enables characterizing individual polymer chains and the tracking of system state evolution in a mathematically straightforward way. However, in kMC simulations for polymerization processes with complicated mechanisms, computational costs can be prohibitive due to the large number of tracked events and simulated polymeric chains.

The superbasin strategy was originally proposed to speed up molecular simulations (Chatterjee, 2010). To address the computational bottlenecks in kMC simulations, this strategy has great potential for enhancing polymerization process modeling. This work presents a developed superbasin-aided kMC model to accurately and efficiently simulate polymerization processes featuring complex kinetic mechanisms with microscopic resolution. The core concept of the developed model is to algorithmically regularize the discrepancy in reaction rates between fast reactions and slow reactions, thereby reducing the computational resources demanded by fast reactions. The implementation of the superbasin-aided kMC simulation integrates three critical steps into the standard kMC framework: (i) on-line classification of fast and slow reactions, (ii) feasibility assessment of scaling, and (iii) scaling operation for adjusting fast reaction rates. Compared to conventional stochastic simulations, the developed model significantly improves computational efficiency by bypassing the meticulous tracking of fast reaction events that have negligible impact on the overall polymer structures.

Photoiniferter-RAFT (PI-RAFT), a representative light-mediated CRP technique, is used as a case study to demonstrate the superbasin-aided kMC model. The accuracy and efficiency of the developed model are validated against the deterministic method of moments method and the conventional kMC simulation. A thousand-fold speedup is achieved for predicting the evolution of microscopic properties of polymers without compromising simulation accuracy. Furthermore, with light irradiation as an external stimulus, light-mediated CRP presents high temporal control over polymer chain growth by modulating light irradiation. The performance of temporal control under intermittent light irradiation is investigated through the proposed model. The kinetic insights gained from this modeling and acceleration work offer potential for further online control and optimization of light-mediated CRP processes.

Corresponding author: a.a.kiss@tudelft.nl

The current trends for sustainability require the chemical industry to migrate from non-renewable feeds to green raw materials. Formic Acid (FA) production from green hydrogen and captured CO₂ can be a good candidate to mitigate greenhouse gas emissions, as it is a widely applicable chemical. The greatest challenge on the usage of renewable sources is their intermittent nature. Because of this reason, control is of outmost importance to ensure the specifications of the product and process parameters. With processes becoming more complex, the role of plantwide control (PWC) is becoming increasingly prominent.

The original contribution of this work is to design a new PWC system for a highly intensified FA production plant that maintains the throughput of 50 kta and product purity of FA to 85% (%wt.). An already established conceptual design was used, which comprises of two sections [1]. The process starts with the production of CO from CO₂ and green H₂, which is followed by the production of FA using Methyl Formate as an intermediate. In this context, COPure and divided wall column (DWC) are identified as the most challenging sub-processes. The project was kick-started by firstly addressing the robustness of the Aspen Plus V12 steady state flowsheet towards the identification of the most sensitive equipment pieces. Additionally, by implementing relative specifications in Aspen Plus, it was possible to achieve a flexible steady state simulation for different feed flow rates.

When converting into dynamic mode, several modifications in the steady state flowsheet were needed for successful initialization. These include, adding the necessary equipment and pressure gradients to achieve a proper pressure-driven dynamic simulation. From this point on, the designed PWC system was implemented using two levels. The first level focuses on controlling individual equipment (but it was not adequate to handle different feed flow rates), whereas the second level revolves around controlling the whole process from a plant-wide perspective, which allows for a smooth transition between different production capacities.

The evaluation of the PWC scheme showed that the designed control system maintained FA purity and production rate for all throughput disturbances tested (up to ±20% change), despite several intricacies of the FA process, such as multiple recycles and the DWC. The deviation of the FA flow rate in dynamic mode from the steady state simulation for the +20% and -20% change in throughput was +0.6% and -0.2%, respectively.

Overall, the evaluation of the PWC design of Formic Acid under unsteady hydrogen supply conditions proved the feasibility of the suggested PWC design on a conceptual level.

[1] N. Kalmoukidis, A. Barus, S. Staikos, M. Taube, F. Mousazadeh, A. A. Kiss, Novel process design for eco-efficient production of green formic acid from CO2, Chemical Engineering Research and Design, 210, 425-436, 2024.

Demographic changes in operator populations as well as substantially higher levels of automation in chemical plants are leading to a decline in experience and skill levels required to operate these plants in a safe and efficient manner.

Operator training simulators (OTS) have become essential tools within BASF to enhance and develop the experience levels of operators in the plant.

The OTS consists of a dynamic model describing the real process. Besides the model, the OTS environment includes a mimic of the control system and safety logics which are connected to the model. The operator interacts with the OTS via similar control system graphics as in the real plant or control room. In this way the OTS environment becomes a very realistic environment which creates an immersive training experience.

OTS allows operators to practice and improve their skills in this safe and controlled environment. These simulators offer a range of benefits, including reducing training costs, minimizing operational risks, and increasing overall efficiency and experience levels.

OTSs are extensively used to train operators on various aspects of plant operation, such as process understanding and optimization, procedural training and disturbance handling. By providing a realistic simulation environment of the process, OTS enables operators to gain hands-on experience in handling critical situations, troubleshooting problems, and making informed decisions.

Different levels of training can be handled by different types of OTSs. Where the training of a starting operator is mainly focused on understanding (parts of) the process, an experienced operator will focus on handling critical situations to avoid damage or production loss. Therefore, it is important to define upfront the objective of the training to be performed with the OTS, since this will have a significant impact on the scope, level of detail and setup of the OTS.

Due to the high accuracy level of OTS models, OTS are used as well to support activities for HAZOP (Hazard and Operability) studies, debottlenecking and optimization studies or (advanced) control design.

The integration of OTS into BASF's training programs has led to improvements in operator competency and operational efficiency. These simulators have also facilitated knowledge transfer between experienced and new operators, ensuring a smooth transition and continuity in operations.

In conclusion, operator training simulators have become an indispensable tool within BASF for training operators. They offer a safe and realistic environment for operators to practice and enhance their skills, leading to improved performance, reduced incidents, and increased operational efficiency.

The transition to a sustainable industrial sector is a complex, capital-intensive challenge that requires the integration of a diverse array of technologies. As a result, it is crucial to understand when, where, and how to deploy both emerging and mature energy technologies synergistically. This integration must be accomplished while minimising any adverse effects on production and managing the inherent increasingly volatile material and energy supply and demand chains. To achieve this, Process Systems Engineering (PSE) provides the advanced conceptual frameworks and software tools to formulate and optimise well-considered integrated solutions that could accelerate the sustainability transition. However, many of the traditional PSE platforms have struggled to fully embrace modern computing technologies and delve into the broader challenges of designing sustainable multi-scale systems.

The landscape of advanced PSE, or PSE+, is poised to undertake a considerable transformation with the rise in popularity of open-source and script-based software platforms with predictive modelling capabilities based on modern mathematical optimization techniques. This paper provides a review of three leading equation-based platforms—Pyomo/IDAES, Modelica and Gekko—that are increasingly utilised for the modelling, simulation, and optimisation of complex systems within the PSE+ domain. Pyomo/IDAES and Modelica have seen considerable attention, forming the basis of an ecosystem of standard models and extensions. Gekko, in contrast, is a light-weight and fast library with minimal overhead and deployable in many industrial control systems. Each platform is critically examined for its capabilities, strengths, and current limitations, highlighting their roles in addressing both conventional and evolving challenges in process systems analysis and integration.

Beyond the current state, this paper explores potential future directions for the development of the PSE+ ecosystem. In particular, the paper discusses the ongoing development of an online platform featuring a graphical user interface (GUI) that is accessible, user-friendly, and reasonably modelling platform agnostic. This cloud-based platform, called the Ahuora Digital Twin platform, is built on modern software engineering principles, including structured data architectures, database integration, modularity, and containerisation, thereby enhancing scalability, flexibility, and accessibility for both researchers and practitioners. The development of such a platform could significantly reduce the barriers to entry for utilising advanced PSE+ methods, making them more accessible to industries and sectors that have not traditionally employed these techniques.