Season's Greetings
Season's Greetings
Dear co-workers, project partners, colleagues, friends, and former members of the dyn group,
In 2019, the Department of Biochemical and Chemical Engineering at TU Dortmund celebrated its 50th anniversary with many activities and events. I was particularly happy that in this context TU Dortmund awarded an Honorary Dr.-Ing. Degree to Prof. Costas Pantelides from Imperial College and PSE Ltd., acknowledging both his scientific contributions and his leadership role in developing and rolling out advanced modelling and simulation tools to the process industry worldwide. Prof. Ignacio Grossmann who received the same honor in 2012 gave one of the Jubilee Lectures of the department and several seminars when he visited as a Gambrinus Fellow of TU Dortmund in February. And Nobel Prize winner Frances Arnold gave an enthusiastic Jubilee lecture to a huge audience in the Audimax of TU Dortmund.
During the celebrations, I realized that I have been with the department for more than half of its lifetime, serving in many functions on the department and university level and helping to master also some quite critical times. It was always enjoyable and rewarding to be a professor here. And I think that all past and present members of the dyn group can be proud to have contributed to the standing of the department throughout these years in many ways, e.g. by attracting external funding, the main indicator of success at TU Dortmund, and by reaching out internationally in our projects, study programs, and exchange activities.
I am very happy to report that in November Daniel Hasskerl received the award for the best dissertation in process automation from NAMUR, the International User Association of Automation Technology in the Process Industries. Thanks to his and several others’ hard work, model-based optimizing control was demonstrated successfully at a real pilot-scale plant of the FVT group, made possible by the ERC Advanced Investigator Grant MOBOCON.
Since this summer, we are part of the consortium that has prepared a large proposal (KEEN) on the application of artificial intelligence in the process industries. As it was positively evaluated, we now look forward to future collaborations with several industrial partners on this topic. In the 1990’s we had already done some nice work on the use of neural nets and fuzzy logic in process control that however at that time did not gain much interest from industry. Now, 20 years later, the AI wave rolls again, this time maybe with a bigger impact.
2019 was another successful year for us, and I would like to thank all group members, project partners and colleagues for the pleasant and rewarding collaboration! I wish you enjoyable holidays and a successful and happy year 2020!
Sebastian Engell
dyn @ PAAT 2019
From November 11th to 12th, 2019, the former group member and PhD candidate Corina Nentwich and the PhD students Lukas Samuel Maxeiner and Simon Wenzel represented the DYN at the PAAT2019
(Jahrestreffen der ProcessNet-Fachgemeinschaften “Prozess-, Apparate- und Anlagentechnik” unterstützt durch “Sustainable Production, Energy and Resources”) in Dortmund, Germany.
Process technology, apparatuses, and plant engineering for sustainable and competitive production were in the focus of this year's annual meeting. Over 200 chemical engineers, plant constructors,
process engineers, and technical chemists from science and industry had the opportunity to present research results, discuss requirements from industrial practice, and jointly develop solutions for
new processes in the chemical process industries and other sectors.
They DYN contributed the following talks to the program:
- Gemeinsame Optimierung von Anlagenverbünden ohne Austausch sensitiver Informationen – geht das? S. Wenzel; L. Maxeiner; S. Engell
- Preisbasierte Optimierung des Einkaufs technischer Gase. R. Lemoine; C. Maul; L. Maxeiner; S. Engell
- Overcoming the modeling bottleneck – Effiziente MILP Modellierung von Verbundstandorten und deren Logistik. L. Maxeiner; S. Wenzel; Y. Misz; S. Engell
- Optimierung chemischer Prozesse unter Verwendung von Surrogatmodellen. C. Nentwich; S. Engell
dyn @ IJCNN 2019
From July 14th to 19th, 2019, Corina Nentwich represented the DYN at the IEEE International Joint Conference on Neural Networks (IJCNN 2019) in Budapest, Hungary.
She presented the following paper in her talk:
- C. Nentwich, C. Varela and S. Engell: Optimization of chemical processes applying surrogate models for phase equilibrium calculations
dyn @ the 5th IEEE International Symposium on Systems Engineering
Between 1-3 October 2019, Marina Rantanen-Modeer represented the DYN at the symposium in Edinburgh, Scotland.
Her presented paper proposes generating abstractions of detailed sub-models of Cyber-Physical Systems, and then embedding those abstractions into system-wide models formulated in other formalisms.
- M. Rantanen-Modeer and S. Engell: Design and validation of Cyber-Physical Systems through model abstraction
dyn @ ECCE12 2019
From September 15th to 19th, 2019, several DYN PhD students presented the following papers:
- K. Klanke, L. Maxeiner and S.Engell: Price-based coordination of shared resources with external suppliers
- M. Cegla, T. Janus, S. Tlatlik, P. Krause, T. Bäck, A. Gottschalk and S. Engell: Flexible and efficient process synthesis and optimization based on Aspen Plus simulations - MTBE production case study
- A. Gottu Mukkula and S.Engell: Application of Iterative Real-time Optimization in an Intensified Continuous Plant at Pilot Plant Scale
- S. Wenzel, Y.Misz, K. Rahimi-Adli, B. Beisheim and S.Engell: Optimal Site-Wide Planning of A NH3 Network – A Study on Uncertain Logistic Constraints
- A. Elekidis, V. Yfantis, F. Corominas, M. Georgiadis, S. Engell: Optimal Production Scheduling in the Packaged Consumer Goods Industry
- J. Pitarch, C. Jasch, M. Kalliski, Y. Misz, M. Marcos, C.Prada, G. Seyfriedsberger, S. Engell: Energy-efficient Operation of a Multi-unit Recovery Cycle in EU’s largest Viscose Fiber Plant
- E.Leo, K.Rahimi-Adli, B.Beisheim, R.Gesthuisen and S.Engell: Applying Stochastic Optimization to Demand-Side Management of a Combined Heat and Power Plant
DYN @ ECC 2019
The PhD students, Sakthi Thangavel and Sankaranarayanan Subramanian, represented the DYN at the European Control Conference in Napoli, Italy from the July 25th to 28th. They presented the following papers:
- S. Subramanian, M., Aboelnour and S. Engell, Robust Tube-Enhanced Multi-Stage Output Feedback MPC for Linear Systems with Additive and Parametric Uncertainties
- S., Thangavel, S. Subramaianan, R. Paulen and S. Engell, Robust Multi-Stage NMPC under Structural Plant-Model Mismatch without Full-State Measurements
dyn @ the 10th IFAC Symposium on Intelligent Autonomous Vehicles
The IFAC Symposium on IAV took place in Gdansk, Poland, between 3-5 July 2019. Marina Rantanen-Modeer represented the group with a paper which proposed
a filter for the preprocessing of raw position data before the trilateration-based position calculation of an RFID-based positioning system for an experimental pipe-less plant.
- M. Rantanen-Modeer, S. Vette, S. Engell: Compensating Signal Loss in RFID-Based Localization Systems
dyn at the 22nd International Conference on Process Control
From June 11-14, 2019, Prof. Sebastian Engell and the research associates Sakthi Thangavel and Afaq Ahmad attended the 22nd International Conference on Process Control in Štrbské Pleso, Slovak Republic.
Prof. Dr.-Ing. Sebastian Engell was invited to give a plenary talk with title ‘Robust NMPC by Multistage Optimization - Basic Idea and Further Developments’.
Further, the following contributions were presented:
- Handling Plant-model Mismatch Using Multi-stage NMPC with Model-error Model Thangavel, S. and Engell, S.
- Model Adaptation with Quadratic Approximation in Iterative Real-Time Optimization Ahmad, A.; Gottu Mukkula, A.R. and Engell, S
dyn @ DYCOPS, IN FLORIANOPOLIS, BRAZIL
From April 23th to 26th, 2019, part of the DYN group attended the 12th IFAC Symposium on Dynamics and Control of Process Systems, including Biosystems in Florianopolis, Brazil.
DYCOPS symposium takes place every three years and has a long history initiated in 1986 in Bournemouth (UK). The scope of the conference is the analysis and control of process systems. In this year, the
following contributions were presented:
- Economics Optimizing Control with Model Mismatch Based on Modifier Adaptation, Reinaldo Hernández and Sebastian Engell
- Extension of the Do-MPC Development Framework to Real-Time Simulation Studies, Alexandru Tatulea-Codrean, Clemens Lindscheid, Rafael Farrera Saldana and Sebastian Engell
- Robust NMPC Using a Model-Error Model with Additive Bounds to Handle Structural Plant-Model Mismatch, Sakthi Thangavel, Sankaranarayanan Subramanian and Sebastian Engell
- Demand Side Management Scheduling Formulation for a Steel Plant Considering Electrode Degradation, Giancarlo Dalle Ave, Jesus Hernández, Luca Onofri, Iiro Harjunkoski and Sebastian Engell
dyn Members 2019
We all would like to thank our partners, colleagues, students and alumni for their support and the fruitful collaboration all throughout 2019. We wish you a bright, happy and successful year 2020!
Farewell to four of our colleagues
Corina Nentwich
Corina started her research career with the dyn group in December 2014. She has been focusing on surrogate modelling of phase equilibrium calculations and on the application in process simulation and optimization. She is very happy that this interesting topic will also be part of her new position in the industry. Since September 2019, Corina is employed at the Evonik Technology & Infrastructure GmbH, in the Digital Process Technologies group in Marl. She will always think back to the days spent at the dyn, especially for the open discussions and the work atmosphere. And she reckons that she will miss teaching.
Simon Wenzel
Simon did his BSc in a dual study program in the textile industry and in Krefeld at the Hochschule Niederrhein, where he focused on chemical technology. After his bachelor thesis at P&G he studied Process Systems Engineering in Dortmund and was absorbed quite quickly by the DYN, where he started work as a Research Associate in December 2014. Simon was very active in both teaching and research projects. His research interests are real-time optimization, site-wide optimization with MILP models, and distributed optimal shared resource allocation with market-like coordination algorithms, which he got to exercise in EU projects DYMASOS and CoPro. Simon will be employed at the Evonik Technology & Infrastructure GmbH as of January 2020. He thinks of the DYN as a sort of a large family that lives spread across the industrial spectrum and meets often at their family meetings like the PAAT, the ProcessNet, the Namur HS, or at the birthdays of the DYN group.
Simone Herchenröder
Simone started her work at the group in 2010, as Prof. Engell's secretary. Her favorite day of the week is Monday and she will never skip the dyn breakfast to see all of her favourite researchers, even if she now works for the Dean’s office (since the 1st November). She remembers fondly the many students and colleagues she has had the pleasure of working with during the last 9 years. When her daughters grow up she hopes they, too, will study at TU Dortmund.
Vassilios Yfantis
Vassilios Yfantis studied Chemical Engineering at TU Berlin and TU Dortmund. After his Master Thesis at Bayer AG in Leverkusen on production scheduling he joined the group of Process Dynamics & Operations at TU Dortmund to work on planning and scheduling in the H2020 project CoPro. Since April 2019 he is working at TU Kaiserslautern on the integration of Demand-side Management and Model Predictive Control.
New dyn Members
Sandra Isabel Gröning
Isabel is the department's newest addition, as she takes over secretary tasks from Simone and Lisa. She started work at TU Dortmund in October 2019, where she is quickly adapting to the dyn life. She has had training as assistant tax consultant before joining the group. Isabel is also a proud mother of two children, ages 5 and 8. Welcome to the dyn , Isa!
Robin Semrau
Robin Semrau received his B.Sc. in Chemical Engineering from the Paderborn University in 2017. In 2019, he finished his Master in Chemical Engineering at TU Dortmund University. The topic of his Master’s thesis was „Validation of an Output-feedback NMPC Scheme for a CFI Polymerization Reactor“. Since May 2019, he is working as a research associate on the process control of co-polymerization and crystallization processes.
Yannik-Noel Misz
Yannik-Noel Misz joined the group in November. He got his bachelor degree in biochemical engineering at TU Dortmund and has been involved in the group’s research since his bachelor’s thesis. For his master’s degree he took the step into the Automation and Robotics program focusing his profile on process automation. He now works on optimization of large scale systems with limited resources.
Christian Klanke
Christian joined the dyn group in February 2019 as a research associate. During his biochemical engineering master studies, he was a working student with the engineering service provider UTEK at the Bayer site in Bergkamen. Afterwards, he conducted his Master thesis entitled “Price-based coordination of shared resources with external suppliers” at the group and received his Master's degree in December 2018. Currently, Christian is researching topics of industrial scheduling applications within the CoPro project and beyond. Apart from that, he declares himself a true Dortmunder, who is passionate about football and has played for a local club ever since he was a child.
Jens Ehlhardt
Jens Ehlhardt studied Chemical Engineering at TU Dortmund. He completed his bachelor's thesis "Experimental studies on innovative visualization concepts for resource efficiency indicators" and his master's thesis "Realisation of NMPC in a lab-scale polymerisation reactor" at the Group of Process Dynamics and Control. During his studies, he spent a semester at the Hong Kong University of Science and Technology. He joined the group in December 2019.
Yehia Abdelsalam
Yehia obtained a Bachelor's in 2009 in Information Engineering and Technology (IET), with a major in communications from the German University in Cairo (GUC). He then worked in the oil and gas industry for 7 years as an instrumentation and control engineer, after which he decided to return to university again in 2016. Yehia joined the program Automation and Robotics in October 2016, and finished in December 2018. He started his research career at the DYN in March 2019, where he is working on stabilizing formulations for model predictive control for plants with parametric uncertainties and additive disturbances. Currently, the focus of his research is on adaptive MPC schemes as well as stabilizing economic MPC for plants with structural uncertainties. Although he was once sure to have left the chemical industry behind him, Yehia now finds the challenges of MPC, and APC in general, within this field to be amongst some of the most interesting.
Journal Articles 2019
Flexible automation with compact NMR spectroscopy for continuous production of pharmaceuticals
Analytical and Bioanalytical Chemistry, 411, 3037-3046, 2019
Modular plants using intensified continuous processes represent an appealing concept for the production ofpharmaceuticals. It can improve quality, safety, sustainability, and profitability compared to batch processes; besides, it enables plug-and-produce reconfiguration for fast product changes. To facilitate this flexibility by real-time quality control, we developed a solution that can be adapted quickly to new processes and is based on a compact nuclear magnetic resonance (NMR) spectrometer. The NMR sensor is a benchtop device enhanced to the requirements ofautomated chemical production including robust evaluation ofsensor data. Beyond monitoring the product quality, online NMR data was used in a new iterative optimization approach to maximize the plant profit and served as a reliable reference for the calibration ofa near-infrared (NIR) spectrometer. The overall approach was demonstrated on a commercial-scale pilot plant using a metal-organic reaction with pharmaceutical relevance.
Optimal experiment design in nonlinear parameter estimation with exact confidence regions
Journal of Process Control, 83, 187-195, 2019
A model-based optimal experiment design (OED) of nonlinear systems is studied. OED represents a methodology for optimizing the geometry of the parametric joint-confidence regions (CRs), which are obtained in an a posteriori analysis of the least-squares parameter estimates. The optimal design is achieved by using the available (experimental) degrees of freedom such that more informative mea- surements are obtained. Unlike the commonly used approaches, which base the OED procedure upon the linearized CRs, we explore a path where we explicitly consider the exact CRs in the OED framework. We use a methodology for a finite parametrization of the exact CRs within the OED problem and we introduce a novel approximation technique of the exact CRs using inner- and outer-approximating ellip- soids as a computationally less demanding alternative. The employed techniques give the OED problem as a finite-dimensional mathematical program of bilevel nature. We use two small-scale illustrative case studies to study various OED criteria and compare the resulting optimal designs with the commonly used linearization-based approach. We also assess the performance of two simple heuristic numerical schemes for bilevel optimization within the studied problems.
A study of model adaptation in iterative real-time optimization of processes with uncertainties
Computers & Chemical Engineering, 122, 218-227, 2019
In real-time optimization, plant-model mismatch can be handled by adding bias and gradient correction terms to the model-based optimization problem in order to meet the first-order necessary conditions of optimality. However, since these correction terms do not ensure the satisfaction of the second-order condition of optimality upon convergence, the model that is used in the optimization can be inadequate. In the framework of iterative modifier-adaptation, this paper proposes to only use effective model parameter updates to ensure and to speed up the convergence to the process optimum. Additionally, this paper shows that model adequacy can and should be enforced explicitly in model parameter adaptation. By means of a simulation study of maximizing the product yield in a fed-batch reactor, we demonstrate that the proposed model adaptation procedure computes model parameters which make the iterative real-time optimization with modifier-adaptation converge faster and more reliably to the plant optimum.
Education in Process Systems Engineering: Why it matters more than ever and how it can be structured
Computers & Chemical Engineering, 126, 102-112, 2019
This position paper is an outcome of discussions that took place at the third FIPSE Symposium in Rhodes, Greece, between June 20–22, 2016 (http://fi-in-pse.org). The FIPSE objective is to discuss open research challenges in topics of Process Systems Engineering (PSE). Here, we discuss the societal and industrial context in which systems thinking and Process Systems Engineering provide indispensable skills and tools for generating innovative solutions to complex problems. We further highlight the present and future challenges that require systems approaches and tools to address not only ‘grand’ challenges but any complex socio-technical challenge. The current state of Process Systems Engineering (PSE) education in the area of chemical and biochemical engineering is considered. We discuss approaches and content at both the unit learning level and at the curriculum level that will enhance the graduates’ capabilities to meet the future challenges they will be facing. PSE principles are important in their own right, but importantly they provide significant opportunities to aid the integration of learning in the basic and engineering sciences across the whole curriculum. This fact is crucial in curriculum design and implementation, such that our graduates benefit to the maximum extent from their learning.
Surrogate modeling of phase equilibrium calculations using adaptive sampling
Computers & Chemical Engineering, 126, 204-217, 2019
Equation of state models as the Perturbed-Chain Statistical Associating Fluid Theory (PC-SAFT) model are accurate and reliable prediction models for phase equilibria. But due to their iterative nature, they are difficult to apply in chemical process optimization, because of long computation times. To overcome this issue, surrogate modeling – replacing a complex model by a black-box model – can be used. A novel surrogate modeling strategy for phase equilibria is presented, combining the training of a classifier model with regression models for the phase composition using a mixed adaptive sampling method. We discuss the selection of the parameters of the sampling algorithm and a suitable stop criterion for the example ternary liquid-liquid equilibrium system of n-decane, dimethylformamide and 1-dodecene in detail. The sequential mixed adaptive sampling method is compared to the one-shot Latin hypercube sampling design.
Integration of production scheduling and energy-cost optimization using Mean Value Cross Decomposition
Computers & Chemical Engineering, 129, 106436, 2019
Integrated optimization of the procurement cost of electric energy with production planning is increasingly considered in various industries. The traditional approach in industry is production driven, i.e. the production is scheduled first, followed by the energy supply optimization to find the best available energy portfolio, which is usually sub-optimal. The combined scheduling and energy procurement optimization can be formulated as an integrated monolithic optimization model, resulting in intractable problems, even if solutions to the two isolated problems are available. We propose to use Mean Value Cross Decomposition for solving the combined problem by iterating between energy-aware production scheduling and energy-cost optimization, possibly building on existing solutions. We apply the approach to a pulping process and a steel production process. MILP-based models are employed for the two scheduling problems and for the energy cost optimization a Minimum-Cost Flow Network model is used, resulting in good quality solutions within reasonable computation times.
A non-uniform grid approach for scheduling considering electricity load tracking and future load prediction
Computers & Chemical Engineering, 129, 106506, 2019
Large consumers’ electricity bills depend on many factors including: different electricity purchasing contracts and markets, deviation penalties, and grid fees. Two key markets are the intraday and the day-ahead markets. On the day-ahead market, the consumer commits to an amount of electricity that stacks upon their longer-term contract commitments. The next day, the consumer must follow this total demand profile to avoid paying deviation penalties. This total committed demand curve can be modified within the current day on the intraday market. In this work, a novel demand side management formulation is proposed that looks at a two-day scheduling horizon considering both the intraday and the day-ahead markets. As a result, instead of modeling a single-day load tracking problem a two-day problem is considered looking at both load following and future load prediction. Furthermore, the computational complexity of the problem is reduced with a non-uniform grid-based formulation. Results show that the novel formulation is able to effectively combine the intraday and day-ahead market concerns resulting in more realistic demand profiles than current formulations. Additionally, the non-uniform grid-based approach captures the key aspects of the formulation with a much lower computation time than the full-space approach.
Surrogate Modeling of Fugacity Coefficients Using Adaptive Sampling
Industrial & Engineering Chemistry Research, 58, 18703-18716, 2019
Complex thermodynamic models such as the perturbed chain statistical associating fluid theory (PC-SAFT) model describe the phase equilibria in a chemical process in a very precise way; however, because of their implicit and complex nature, the application of such models in process simulation and optimization can lead to a high computational effort, which may prevent the direct application of such models in process simulation and optimization. In this contribution, we replace the iterative calculation of the fugacity coefficient using PC-SAFT with explicit surrogate models that are trained using a novel adaptive sampling method.
An optimization model for site-wide scheduling of coupled production plants with an application to the ammonia network of a petrochemical site
Optimization and Engineering, 20, 969-999, 2019
This contribution presents the modeling and optimization of the operation of production plants that are coupled via distribution networks and applies it to a part of the petrochemical production site of INEOS in Köln in Germany. The problem is formulated as a mixed-integer linear problem and solved to generate an optimal monthly plan for a set of plants, tanks, and loading/unloading facilities, while respecting various constraints arising from technical limitations, physical couplings between the plants, production targets, and the schedule for import and export across the company borders via ships and trains. The optimization problem takes into account varying energy prices, the influence of the ambient temperature on the processes, and the inventory management for different types of storages. We solve the optimization problem for the particular case and compare the results for a 1 month scenario to recorded data and show that a significant energy saving potential exists. We discuss the current limitations and outline potential improvements in the context of the application of the optimization model to optimal site planning that leads to an improved coordination of the production in the process industries.
Energy performance analysis of continuous processes using surrogate models
Energy, 183, 776-787, 2019
Energy intensity is a commonly used key performance indicator (KPI) for the energy performance of production processes and often serves as an Energy Performance Indicator (EnPI). The energy performance of a process depends on a variety of factors like capacity utilization, ambient temperature and operational performance. Understanding the influence of these factors on the relevant KPI or EnPI helps to distinguish between influenceable and non-influenceable contributions and to identify the improvement potential. By describing the best historically observed performance as a function of the non-influenceable factors, valuable information on the efficiency of the current operation of a plant and the improvement potential is provided to plant managers and operators. In this contribution, a method is proposed to identify a surrogate performance model for the attainable energy performance considering the relevant factors. The modeling method is based solely on the evaluation of historical process data and employs a novel combination of known surrogate modeling techniques using clustering, model fitting and model simplification by backward elimination. The method is applied to real process data of a large industrial production plant and the use of the model for process performance monitoring and reporting in accordance with energy management system requirements is illustrated and discussed.
Conference Articles 2019
Model Adaptation with Quadratic Approximation in Iterative Real-Time Optimization
22nd International Conference on Process Control, Strbske Pleso, Slovakia, 2019
This paper deals with the iterative real-time optimization (RTO) of chemical processes under plant-model mismatch. Modifier-adaptation can cope with the plant-model mismatch by adding bias- and gradient-correction terms to the underlying model based optimization problem. These correction terms ensure the convergence to the true plant optimum via enforcing the first-order necessary-conditions of optimality of the plant despite plant-model mismatch. However, the estimation of the empirical gradients from noisy measurement data is a limiting factor and also the added correction terms do not guarantee that the second-order conditions of the optimality are satisfied upon convergence. Modifier adaptation can be combined with the quadratic approximation approach used in derivative-free optimization to ensure the convergence to the process optimum. In the framework of modifier-adaptation with quadratic approximation, this contribution proposes to add a model adaptation step such that the second-order optimality conditions are met at the plant optimum. Also to improve the estimates of the model parameters may speed up convergence. The performance of the proposed scheme is demonstrated by using a fed-batch reactor case-study.
Handling Plant-model Mismatch Using Multi-stage NMPC with Model-error Model
22nd International Conference on Process Control, Strbske Pleso, Slovakia, 2019
We propose a robust nonlinear model predictive controller (NMPC) to address the presence of parametric and structural plant-model mismatch in a multi-stage NMPC framework using a model-error model (MEM). The scenario tree of the multi-stage NMPC is generated for different realizations of the uncertain parameter and the model-error model. The modelerror model consists of a linear model followed by an unknown nonlinear operator with bounded gain and captures the dynamics of the plant that are not described by the nominal model. The advantages of the proposed multi-stage NMPC with model-error model scheme are demonstrated for a benchmark case study.
Start-up and Shut-down Conditions for Iterative Real-Time Optimization Methods
6th Indian Control Conference, Hyderabad, India, 2019
Iterative real-time optimization methods are able to identify the real process optimum in the presence of structural and parametric plant-model mismatch. However, upon converging to the process optimum they may suffer from generating ongoing process perturbations in response to measurement noise which are inefficient. In this paper, we propose a strategy for the shut-down of the iterative optimization schemes upon convergence to the plant optimum and a strategy for the start-up of the iterative optimization when a change in the process behavior occurs, in order to avoid a loss of performance. We employ techniques from statistical process monitoring to formulate appropriate conditions to detect a change in the process. The performance of the proposed start-up and shut-down strategies in combination with a powerful real-time optimization method namely, modifier adaptation with quadratic approximation (MAWQA), is analyzed using a chemical engineering case study.
Coordination of coupled systems of systems with quadratic approximation
15th IFAC Symposium on Large Scale Complex Systems, Delft, The Netherlands, 2019
The process industries becomes increasingly digital, integrated, and connected. Chemical production sites consist of many subsystems which are tightly interconnected and physically coupled by flows of material and energy. Coordination of these systems of systems is essential but challenging. Various barriers prohibit a joint optimization of the overall system. One of the barriers results from the complex management structure of these systems which does not allow for the exchange of all necessary information for a holistic formulation and solution of the optimization problem. Market-like coordination is one way to address this challenge by establishing a micro-market for shared resources between the competing subsystems. However, market-like coordination is known for its slow rate of convergence. Using quadratic approximation for an improved rate of convergence was proposed in Wenzel et al. (2016). In this contribution, we present an extension of this algorithm and apply it to a number of randomly generated medium- to large-scale test problems. The performance of the algorithm is compared to the performance of subgradient-based coordination and to the performance of the Alternating Direction Method of Multipliers (ADMM).
Robust Multi-stage NMPC under Structural Plant-model Mismatch Without Full-State Measurements
18th European Control Conference, Naples, Italy, 2019
We address the problem of robust nonlinear model predictive control (NMPC) under plant-model mismatch in the absence of full-state measurements. We propose an approach that is based on the use of a model-error model (MEM) to handle the estimation errors and the structural plant-model mismatch in a Multi-stage NMPC framework. The MEM which consists of a linear model followed by a nonlinear operator with bounded gain captures the estimation error along with the unmodeled dynamics of the plant. Multi-stage NMPC explicitly considers the presence of feedback in the problem formulation, hence it is less conservative than other robust NMPC schemes. The advantages of the proposed scheme are demonstrated on a benchmark reactor problem.
Robust tube-enhanced multi-stage output feedback MPC for linear systems with additive and parametric uncertainties
18th European Control Conference, Naples, Italy, 2019
We propose a new robust output feedback Model Predictive Control (MPC) scheme that can handle both additive and parametric (multiplicative) uncertainties using the tube-enhanced multi-stage (TEMS) MPC framework. In TEMS MPC, the significant parametric uncertainties are handled using the multi-stage MPC formulation and the additive uncertainties are handled using the tube-based MPC framework resulting in an improved trade-off between optimality and complexity. The proposed output feedback approach can handle estimation errors in addition to multiplicative and additive uncertainties. A key advantage of the proposed formulation is that it is independent of the state estimation scheme employed and hence it can be easily combined with any estimation scheme. The recursive feasibility and stability of the proposed approach in the linear case are proven. The advantages of the proposed approach are demonstrated for two examples.
Optimization of chemical processes applying surrogate models for phase equilibrium calculations
2019 International Joint Conference on Neural Networks, Budapest, Hungary, 2019
The calculation of the thermo-physical phase equilibrium of a multicomponent mixture is employed in chemical process development to model the number and composition of phases in order to predict and to optimize the reaction and separation performance of a chemical process plant. Complex thermodynamic models as equations of state provide reliable predictions of phase equilibria over a broad operating range. But due to the need for iterative calculations, they are hardly applicable in optimization. This work shows how a combination of classification and regression can be used to replace these calculations in the optimization of the process of hydroformylation of 1-dodecene.
Robust NMPC using a model-error model with additive bounds to handle structural plant-model mismatch
12th IFAC Symposium on Dynamics and Control of Process Systems, Florianopolis, Brazil, 2019
We address the problem of robust nonlinear model predictive control (NMPC) under structural plant model mismatch in a multi-stage framework using a model-error model (MEM). Multi-stage NMPC models the presence of future feedback information in the predictions, hence it is less conservative than the other existing robust NMPC approaches. MEM consists of a stable linear model, an unknown nonlinear operator with bounded gain and a bounded additive mismatch. The computational burden of the proposed scheme is reduced by using two different approximations; 1. Constraint tightened multi-stage NMPC 2. Tube-enhanced multistage NMPC. Both the schemes are real-time implementable with a small loss in performance when compared to the standard multi-stage NMPC. The advantages of the proposed schemes are demonstrated for a benchmark continuous stirred tank reactor (CSTR) example.
Economics optimizing control with model mismatch based on modifier adaptation
12th IFAC Symposium on Dynamics and Control of Process Systems, including Biosystems, Florianopolis, Brazil, 2019
It is well-know that the use of an inaccurate model in any model-based control scheme might lead to a suboptimal operation and to violation of the constraints. In this contribution, an economics optimizing control scheme is proposed with the aim of achieving closed-loop optimality despite the presence of structural model-plant mismatch. Model uncertainties are handled by augmenting the nominal model with bias and gradients correction terms which are iteratively updated using plant information. The basic idea is to enforce the matching of the optimality conditions between the plant and the augmented model by linear correction terms similar to modifier adaptation. The proposed scheme is discussed and illustrated by simulation studies of a benchmark problem.
Extension of the do-mpc development framework to real-time simulation studies
12th IFAC Symposium on Dynamics and Control of Process Systems, including Biosystems, Florianopolis, Brazil, 2019
Nonlinear Model Predictive Control (NMPC) has become a serious option for industrial applications, with advances in software and algorithms making economics optimizing control suitable even for large scale systems. However, the industrial applications are posing a number of engineering challenges, related to the performance of NMPC in real plant environments. In this paper we address the issue of validating NMPC solutions in a real-time simulation environment, which mimics precisely the behavior of the controlled plant. We propose a validation strategy and describe the simulation framework that was developed in order to support the validation process. The software platform used for this purpose is based on the do-mpc framework, benefiting from its modularity and efficient implementation of NMPC algorithms. The focus of this work is the extension of the software environment to a real-application oriented platform where control scenarios can be simulated in real time. The necessity of the proposed validation strategy is demonstrated for a semi-batch polymerization example, where it is shown that feedback delays have a significant impact on the real-time performance. The necessary steps for a transition from simulation studies to the real application are discussed and a method for improving the NMPC performance is proposed based on the real-time simulation studies.
Demand side management scheduling formulation for a steel plant considering electrode degradation
12th IFAC Symposium on Dynamics and Control of Process Systems, including Biosystems, Florianopolis, Brazil, 2019
Stainless steel production consumes a large amount of electricity. Demand side management (DSM) has been identified as a key way to reduce costs in steel plant operations by responding to time-varying electricity prices. The electric arc furnace (EAF) is responsible for the majority of a steel plant's electricity consumption and has been identified as having sufficient flexibility to participate in DSM schemes as it is run in batch mode with variable operating settings. Often neglected in steel plant scheduling is the degradation of the electrodes of the EAF units and the associated cost of replacing them. However, there exists a complex tradeoff between the intensity at which the EAF is operated, and the lifetime of the electrode. The intensity in this case also strongly affects both electricity consumption and the length of a batch. This work proposes a novel scheduling formulation that explicitly considers DSM and its relation to electrode degradation. Results show on the one hand, that the novel formulation is able to more realistically model the EAF operations by explicitly considering electrode consumption. On the other hand, the model is also able to effectively balance the trade-offs between DSM-related costs and electrode replacement costs by extending the lifetime of the electrodes while still having the flexibility to be able to participate in DSM schemes.
Optimization of a hydroformulation process in a thermomorphic solvent system using a commercial steady-state process simulator and a memetic algorithm
29th European Symposium on Computer Aided Process Engineering, Eindhoven, The Netherlands, 2019
The economic evaluation of alternative process configurations is an important step in process development. It should be based on optimization to correctly investigate the potential of different process routes and process variants. In many companies, such design studies are performed using block-oriented flowsheet simulators such as Aspen Plus to utilize the extensive model libraries and the ease of model building. We developed an optimization framework that integrates a process simulator (specifically Aspen Plus) with a memetic algorithm (MA). This MA combines an evolution strategy (ES) with derivative-free (DFO) local search methods. The ES addresses the global optimization of all design variables, whereas the DFO method locally optimizes the continuous sub-problems that arise by fixing the discrete variables. In this work, the performance of the memetic algorithm is evaluated for different local methods, involving different DFO methods or the internal equation-oriented optimization engine of Aspen Plus. We discuss the results and the efficiency of the different local methods.
An Explicit Online Resource-Task Network Scheduling Formulation to Avoid Scheduling Nervousness
29th European Symposium on Computer Aided Process Engineering, Eindhoven, The Netherlands, 2019
Scheduling is a decision-making process that is often based on the assumption of nominal operating conditions. In reality, scheduling is a dynamic process and uncertainties often arise during the execution of a schedule. One way to handle uncertainty in schedule is via reactive scheduling, or rescheduling. Rescheduling accounts for uncertainty by revising the existing agenda in response to real-time events. The downside to frequent revisions of the schedule is that it often leads to scheduling nervousness. In this work, a novel set of constraints based on the Resource-Task Network (RTN), coupled with a region-based penalty approach is proposed to combat scheduling nervousness. Results show that the approach is able to produce more stable schedules, while still giving rescheduled agendas the flexibility needed to pursue the original scheduling objective.
A model identification approach for the evaluation of plant efficiency
29th European Symposium on Computer Aided Process Engineering, Eindhoven, The Netherlands, 2019
Regulations and the public expectations on improving efficiency, reducing the carbon footprint and lowering the environmental impact drive the process industry towards improved operation and the development of new technologies. The efficiency of an existing production plant depends on a variety of factors like capacity utilisation, raw material quality, ambient temperature or operational performance. Identifying the influence of these factors on the performance of the plant helps to take suitable measures to drive it towards a more efficient operation. One approach to assess the resource efficiency potential of a plant is the comparison of the actual performance with the best possible operation under the given circumstances. This work presents a surrogate modelling approach for the identification of the best possible operation based on historical data. The surrogate model is compared to a more detailed rigorous model and advantages and possible shortcomings of the surrogate approach are discussed based on real production data at INEOS in Köln.
Iterative Medium-Term Production Scheduling of an Industrial Formulation Plant
29th European Symposium on Computer Aided Process Engineering, Eindhoven, The Netherlands, 2019
In this contribution, a discrete-time medium-term production scheduling model of an industrial formulation plant for crop protection chemicals is presented. As the optimization model has to provide feasible solutions in a real-world environment, all relevant processing characteristics, as e.g. sequence- and unit-dependent changeovers, timing relations between different units and processing tasks or shift patterns must be represented in order to yield feasible and implementable schedules. In order to cope with the computational complexity of the problem, an iterative scheme is employed in which the production orders are scheduled in a sequential fashion. After each iteration a set reduction step is performed, removing all infeasible unit allocations from the feasible set of the next iteration. The proposed framework is tested on a realistic industrial case study.
A framework for the optimization of the operation of an industrial power plant under demand uncertainty
Jahrestreffen der ProcessNet-Fachgemeinschaft "Prozess-, Apparate- und Anlagentechnik", Dortmund, Germany, 2019
The cogeneration of heat and power helps to increase the energy-efficiency, to reduce CO2 emissions and to increase the robustness of the energy infrastructure through decentralized generation. For a typical chemical production site, the power plant is of a high importance, as it incinerates the off-gases produced by the production plants and provides them with steam and electricity. The significance of the Combined Heat and Power (CHP) plants is also recognized by the European Union through the CHP directive (Directive 2004/8/EC, 2004), which promotes the construction and operation of cogeneration plants.
The typically heat-driven operation of a power plant in the chemical industry is determined by the steam demand of the production plants. This demand is often uncertain due to inaccuracies in the production plan and fluctuations induced by the operation of the plants. The demand uncertainty can result in an inefficient operation due to the surplus/deficiencies of the steam needed to balance the steam network.
In this contribution an approach for the optimization of the operation of the power plant at INEOS in Köln that was developed in the framework of the European project CoPro is presented. The approach is based on a two-stage stochastic mixed-integer linear programming formulation, which is solved consecutively on a rolling horizon. At each iteration of the rolling horizon scheme, the model parameters are updated according to the new information acquired from the plants and the optimization is re-executed.
Hedging against steam demand uncertainty results in the reduction of the vented steam and a better tracking of the time-sensitive electricity price, which enables significant savings in the power plant e.g. in the natural gas consumption.
Price-based coordination of shared resources with external suppliers
12th European Congress of Chemical Engineering, Florence, Italy, 2019
Worldwide, chemical production sites are significant consumers of energy and resources and improving their efficiency has been pursued continuously over the past decades. Optimizing plant operations is an economically attractive way of improving their efficiency because the investments are relatively low, and the duration of the projects is not overly long. In large chemical sites, the plants are interconnected by networks of raw and intermediate materials and of carriers of energy, especially steam at different pressure levels. Coordinating the production and consumption of these shared resources provides an impactful lever towards a more efficient production.
While the site-wide optimization of the operation of the interconnected plants is theoretically feasible, practically it is desirable to perform a distributed optimization with some central coordination. One strong argument in favour of such schemes is that information on the internals of the individual plants, e.g. their demands, revenues and profit functions does not have to be shared so that such schemes can also be applied if the plants are operated by different business units. This can be achieved by so-called price-based coordination schemes in which a central coordinator balances the networks by adapting the transfer prices, which enter in the individual cost functions.
A limitation that the previously proposed coordination schemes suffer from is that many of the shared resources produced and/or consumed on the site are also available via external suppliers on short notice or can be sold on the spot market. Such exchanges of materials and carriers of energy are subject of medium- or long-term contractual agreements. What we propose here is a scheme that determines dynamic transfer prices in the context of such agreements without providing the parties involved access to sensitive data, thus improving flexibility and enhancing fairness. This is an essential agreement of the practical implementation of industrial symbiosis.
Flexible and efficient process synthesis and optimization based on Aspen Plus simulations - MTBE production case study
12th European Congress of Chemical Engineering, Florence, Italy, 2019
The decisions taken during process development are critical for the economic success of process engineering projects. The design engineer needs to screen, develop and evaluate a large number of process alternatives. Even when sophisticated software tools are applied for modelling and analysis, iteratively carrying out the design usually requires month to years. Due to the strong involvement of human work force, high personnel costs are incurred and the results depend on the experience of the developers. Obviously, also the risk of human errors e.g. by prematurely ruling out alternative solutions is always present.
Marking a breakthrough in boosting the efficiency of process development a novel process optimization tool has been established in a joint project between divis, SUPREN and TU Dortmund. Starting from an existing simulation flowsheet that was created by a design engineer, a memetic algorithm is applied to investigate, globally optimize and assess alternative process configurations and conditions in an automated fashion. The tool supplements the process understanding and creativity of an engineer with the reliability and efficiency of optimization. Replacing human labor by computations, the repetitive work in the course of process development is accelerated while the related costs decrease.
Application of Iterative Real-time Optimization in an Intensified Continuous Plant at Pilot Plant Scale
12th European Congress of Chemical Engineering, Florence, Italy, 2019
It is always of great interest to identify the optimal point of a plant and to operate it at this point. Usually, the optimal operating point of a plant is identified by solving a model-based optimization problem. Developing an accurate mathematical model and estimating its parameters require a large amount of time and effort. Sometimes it may not be possible at all to develop an accurate model due to the complex phenomena that take place in the plant. Modifier adaptation with quadratic approximation (MAWQA) is an iterative optimization scheme which uses measurements from the plant to drive the plant to its true optimum in spite of the presence of significant plant-model mismatch.
Optimal Site-Wide Planning of A NH3 Network – A Study on Uncertain Logistic Constraints
12th European Congress of Chemical Engineering, Florence, Italy, 2019
Many companies in the process industries operate in a highly competitive market environment and are thus obliged to improve continuously their production in terms of energy and material efficiency and to become greener. In general, the value chain of chemical production consists of several, sometimes many production steps and complex interactions between these production steps exist, which lead to highly integrated and physically coupled systems of systems. In large (petro-) chemical production sites, complex networks of flows of material and carriers of energy link the production plants. For an optimal operation of the overall system, mathematical modeling of the constituent systems and a joint optimization has to be employed to find site-wide optimal and feasible operating conditions according to different objectives such as material efficiency, reduction of carbon footprint, or economic performance indicators. The scope of such an optimization can, however, not be limited to the units and buffers of the production site, but it also has to consider exogenous influences such as fluctuating market prices for energy and raw materials and the logistics of their provision and distribution. Recently, there has been an increased interest in expanding the domain of optimization and thus enlarging the scope of the formulated optimization problems to include these aspects.
Optimal Production Scheduling in the Packaged Consumer Goods Industry
12th European Congress of Chemical Engineering, Florence, Italy, 2019
The ever-increasing competitiveness of large-scale industries necessitates the increase of productivity and minimization of production costs. Production scheduling has a direct impact on the overall efficiency of industrial facilities by minimizing, among others, changeover times which cause production downtimes, waste and energy consumption/use of resources. Potential plant reconfigurations also provide significant room for improvements of the overall productivity. Although production scheduling problems have received significant attention, the majority of the work is focused on relatively small to medium size problem instances. A few recent contributions have studied the application of scheduling techniques in medium- and large-scale industries. In this work two MILP-based approaches are proposed for the optimization-based scheduling of a real-life large-scale consumer goods production plant. The first approach focuses on the optimal scheduling of the current plant layout while the second one is applied to a reconfigured flexible layout allowing the decoupling of the formulation and packing stages.
Energy-efficient Operation of a Multi-unit Recovery Cycle in EU’s largest Viscose Fiber Plant
12th European Congress of Chemical Engineering, Florence, Italy, 2019
Lenzing AG in Austria is the world’s largest viscose fiber production plant. Lenzing’s fibers are made from wood: They are botanic products derived from renewable sources and processed with resource conserving technologies. The production of these high-quality viscose fibers is a multi-step chemical technological process. The key role in terms of resource efficiency within the viscose-fiber production belongs to the spinbath recovery cycle. The recovery cycle itself is a sequence of basic operations carried out in several multi-unit networks. Due to its high-energy de-mand, the LENZING use case within the SPIRE project CoPro is particularly focused on the evaporation stage and the heat-recovery network.
Applying Stochastic Optimization to Demand-Side Management of a Combined Heat and Power Plant
12th European Congress of Chemical Engineering, Florence, Italy, 2019
Integrated production sites in the chemical industry typically consist of many large-scale plants which are highly interconnected through material (e.g. base chemicals or intermediates) and utility networks. A power plant is of major significance, as it provides utilities to the plants, e.g. electricity and process steam. Furthermore, power plants are able to collect off-gases from other plants and use them as fuels. The importance of power plants is also acknowledged by the European Union with the CHP Directive (Directive 2004/8/EC, 2004) that sets a framework to promote growth of cogeneration plants.
In this work we address the integrated electricity procurement and operation of a power plant that supplies steam and power to a chemical process and interacts with the power grid. Adopting a stochastic mixed-integer programming formulation, the aim of this work is to optimally determine for a planning horizon of one week: a) the amount of electricity purchased from power contracts; b) a daily day-ahead commitment with hourly discretization to purchase electricity from the day-ahead market under time-sensitive prices; c) the hourly production levels of steam and electricity; d) the amount of fuels stored and the off-gases incinerated.
Master's Theses 2019
Optimale Strategie der modellbasierten Versuchsplanung für hybride Bioprozessmodelle
supervised by: Hille, Gottu Mukkula
Ein schnell wachsender Markt für Biopharmazeutika, strenge regulatorische Vorgaben und die Durchsetzung der Quality by Design Initiative stellen neue Herausforderungen an industrielle Bioprozesse. Neben genetischer Optimierung der Zellkulturen und Optimierung von Kulturmedien, konnten modellbasierte Methoden erfolgreich zur Reduzierung des Versuchsaufwandes, Erhöhung der Prozessrobustheit und höherer Prozesstransparenz eingesetzt werden. Speziell mechanistische Modelle auf Basis der Metabolischen Netzwerkanalyse stellen einen vielversprechenden Ansatz dar. Mithilfe von gemessenen Daten und bekannten intrazellulären Reaktionen können damit die Reaktionswege der untersuchten Zellkultur ermittelt werden. Die Metabolische Netzwerkanalyse weist jedoch gewisse Limitierungen auf – so können damit solche Einflussfaktoren wie Temperatur, pH-Wert, Rührerdrehzahl oder DO-Wert nicht beschrieben werden, daher sind zur Modellierung dieser Größen datengetriebene Ansätze erforderlich.
Im Rahmen dieser Arbeit werden die Methoden der statistischen Versuchsplanung im Kontext der Entwicklung eines hybriden Bioprozessmodells untersucht. Nach der Definition der Anforderungen an Modellkandidaten, Auswahl und Untersuchung geeigneter Modelle wird ein Versuchslauf durchgeführt, um die modellbasierten Methoden der Versuchsplanung an reellen Daten zu untersuchen. Es wird auf Gemeinsamkeiten, Unterschiede, Vorteile und Limitierungen von unterschiedlichen Methoden eingegangen.
Als Ergebnis wird eine DoE-Strategie zur Entwicklung datengetriebener Untermodelle vorgestellt.
Hybrid scheduling of an industrial formulation plant by metaheuristic optimization and constraint programming
supervised by: Klanke
Due to increasing global competition, not only the process industry requires tools to efficiently plan and schedule production processes. Production scheduling plays a major role to cope with the challenges arising from interdisciplinary functional entities in a company. Research in this area concentrates mainly on the development of optimization models and methods. This thesis focuses on developing an algorithm to solve a multi-product batch process. The developed hybrid evolutionary algorithm is tested by using a case study which represents a formulation plant in the chemical industry. The results show that feasible solutions can be found in reasonable amount of computation time to solve a two-stage production scheduling problem with unlimited intermediate storage.
The algorithm can be further improved by including constraints that restrict the intermediate storage tanks to their real capacity limits. This is potential to improve the quality of the algorithm. However, this is less of a conceptual but rather a quantitative implementation aspect. Also, one further termination criterion may be included to allow the scheduler to terminate when a feasible solution is obtained and has not improved after certain generations. The computational time results reveal that the proposed algorithm can find feasible solutions in reasonable time.
Optimal startup procedure for an industrial evaporation process
supervised by: Ebrahim
Two modes of operation dominate the life cycle of chemical production processes, the first one is the productive steady-state operation which accounts for most of the operating time, and the second mode refers to the large transitions such as start-up, shut-down as well as switching of the operating point in normal operating conditions which only occur infrequently. The control goal during the steady-state operation is mostly achieved using purely continuous process models. Although large transitions only occur infrequently, they are usually operated on heuristic basis and thus associated with material, energy waste and high safety risks. The optimization of large transitions is a complicated task because it is usually associated with interaction between discrete and continuous dynamics. The discrete dynamics arise due to the switching of dynamics or due to the usage of discrete actuating devices. Therefore, new methods which are neither pure continuous nor discrete are needed to deal with the hybrid optimal problems associated with large transients. In this thesis a HOCP without state jumps is solved using the technique of outer convexification and relaxation. Multiphase MIOCP can be used to solve the HOCP with known switching sequence and unknown switching times. Two case studies are considered. Egerstedt Standard Problem contains the discrete controls is solved using outer convexificaiton and relaxation. The second one is the optimal start-up problem of evaporation process. Evaporation processes are widely used in industry to concentrate liquids. Start-up process exhibits hybrid dynamics due to the phase changes and the discrete control devices. The problem of optimal start-up is solved by multiphase MIOCP method.
Robust nonlinear model predictive control for switched systems
supervised by: Ebrahim
Switched hybrid systems arise as a result of the interaction between the continuous time driven and the discrete event driven dynamics. The importance of such class of systems is a consequence of the high potential of applications and the associated technical challenges. In switched systems, the dynamical system evolves according to a set of vector fields describing the continuous evolution in different sections of the state space, or so called modes of operation. In addition to that, a switched mechanism determines the active mode at each time instant along the time horizon. Model predictive control (MPC/NMPC) proves to be an appealing scheme for multivariable and constrained systems. The basic idea lies around solving an optimal control problem (OCP) for the control inputs iteratively in real time with the availability of each new set of measurements from the process. In addition to the difficulties encountered in the continuous case, optimal control for switched systems imposes more challenges due to the discontinuities and the associated combinatorial increase in the problem size. Embedding transformation is a relatively new scheme for solving OCP for switched systems. It relies on embedding the switched system into a larger family of continuous systems and seeking the optimal solution, which is either valid for the switched system or an arbitrary close switched solution can be found, e.g., by rounding. A novel insight is adopted in this thesis, where a class of switched system is regarded as continuous system subject to additive disturbance due to rounding. Thus a new approach to achieve stability of switched systems by purely switching control is proposed, where stability of nominal switched systems can be achieved by employing robust MPC controllers to handle the rounding error. In this context, three robust controllers are proposed for the nominal case, where knowledge about the source of additive disturbance is exploited by means of switching time optimization (SWTOP), in order to minimize the rounding error. Since presence of plant-model mismatch is almost inevitable, a tube-like enhanced multistage with SWTOP is proposed to handle switched systems with parametric uncertainty. The proposed controller is composed of two levels, each level comprising a multi-stage NMPC. The higher level computes a relaxed sequence robust to possible discrete realizations of the uncertainty, meanwhile the lower level optimizes the switching instants of the rounded sequence in order to minimize the average cost over all the considered scenarios. Moreover, a moving horizon estimator (MHE) for uncertain switched systems is proposed assuming full state information is not available. The performance of the proposed controllers for the state feedback as well as output feedback case is demonstrated on a switching continuous stirred reactor.
Metaheuristic-based Scheduling of an Agile Consumer Goods Production Plant
supervised by: Yfantis
Manufacturing scheduling plays a very important role in successful operation of the production planning and control department of an organization. It also offers a great theoretical challenge to the researchers because of its combinatorial nature. Earlier, researchers emphasized classical optimization methods such as linear programming and branch-and-bound method to solve scheduling problems. However, these methods have the limitation of tackling onlysmall-sized scheduling problems because of the consumption of high computational (CPU) time. As a result, various efficient metaheuristics optimization methods such as genetic algorithms and simulated annealing have been applied to scheduling problems for obtaining near optimal solutions in short computation times. These computational tools are currently being utilized successfully in various sectors. In this thesis, a two-stage simulated annealing algorithm is used to optimize the scheduling of an consumer goods production plant. The two-stage simulated annealing algorithm allows the acceptance of worse product sequences to escape from a local minimum and reach a near global optimal solution. The algorithm is applied to two layouts of the plant, one in which the production stages are coupled and one in which a buffer is employed to decouple production and packing. Quite good schedules are achieved in reasonable computational time and feasible good solutions are even obtained in less than one minute.
Application of State and Parameter Estimation in a Lab-Scale Polymerization Reactor
supervised by: Lindscheid
Most polymerization processes are carried out in batch or semi-batch reactors. One of the main advantages of batch and semi-batch operation is the flexibility for multi-product processes. However, these processes possess an inherent variability of the product quality, requiring more sophisticated process instrumentation and control to reduce product variability. In industrial polymerization processes, a common quality indicator is the product viscosity. Polymer viscosity is intrinsically tied to its molecular weight distribution. Therefore, a better control of a polymer's molecular weight distribution can lead to less variability in product quality. State and parameter estimation can be a useful tool in gaining knowledge about system states that are not measured online or about parameters that vary during the process. Having knowledge about these unmeasured states is a requirement when applying more advanced control techniques such as nonlinear model predictive controllers. Several examples of successful application of NMPCs in industry have already been reported. Therefore, the successful application of a state estimator to the plant is of great interest. In this work, the homopolymerization of acrylic acid by solution radical polymerization in a semi-batch reactor is considered. Previous work performed by Azimzadegan is discussed and improvements to the plant are investigated and proposed. A semi-batch recipe is designed based on the reaction model and finally the application of state and parameter estimation to the plant is presented.
Optimizing Control of a Continuous Oscillatory Baffle Crystallizer
supervised by: Hernandez
Crystallization is one of the oldest and widely used unit operations in the process industry; with significant relevance in the production of pharmaceuticals, food and fine chemicals. Recently, the industry has been under an increasing pressure to switch the traditionally operated batch crystallizers to continuous crystallizers in an attempt to reduce cost, improve quality consistency and increase productivity. Unfortunately, there are still some barriers that need to be overcome for a successful transition from batch to continuous. One of them is the development of reliable advanced control schemes that are able to ensure constant product quality despite the presence of the inherent uncertainties associated to the process.
In this work, a model-based approach is proposed for the optimal operation of a continuous oscillatory baffled crystallizer (COBC). A full detailed model was developed, which includes a population balance to describe the crystal size distribution and semi-empirical correlation, which characterize the non-ideal flow behaviour of the equipment. Furthermore, the individual modules of the complete model were validated by using available experimental data from the literature. In order to overcome the numerical difficulties that arise during the simulation of the resulting large-scale problem, different discretization methods were evaluated including finite difference and high resolution schemes (e.g. WENO).
Finally, it is propose to use the developed model in a closed loop RTO scheme for precise control of the CSD. Robustness to model errors is provided by means of iterative adaptation of the nominal optimization problem according to the well-known Modifier Adaptation scheme. In silico experiments showed the potentiality of the proposed approach as a method to tailor CSD in crystallization process despite the presence of uncertainties.
Realisation of NMPC in a lab-scale polymerisation plant
supervised by: Lindscheid
Tighter economic and ecological demands in the chemical industry foster the need for novel methods in process operations. Especially the production of polymers, which are widely used in many different applications, require a tight control of the product specification, while at the same time operating economically efficiently and with little energy consumption. Polymers are often produced in semi-batch reactors, whose operation shows strong nonlinearities due to the exothermic reaction.
Model based approaches such as nonlinear model predictive control (NMPC) -an optimizing controller based on a dynamic nonlinear process model- are suitable for meeting these demands. Tracking the product quality as well as keeping important operational constraints within limits can lead to an improved economic performance, as has been pointed out in many simulation studies. However, while many simulation studies show very promising results, there are only very few practical realizations of NMPC.
In this master’s thesis, a NMPC is realized on a lab-scale reactor for the homopolymerization of acrylic acid. The major adaptions to be performed for a transition from a simulation study to a real application are investigated in detail: Process and basic automation are adjusted so that they form a suitable baseline for the application, a basic recipe is found, the model is updated according to the compromise between accuracy and complexity, the NMPC algorithm is refined towards real plant applications, appropriate tuning parameters are found and the interfaces between process and NMPC are designed for an easy connection of the external NMPC hardware to the control system.
First comparisons between a PID-cascade structure and the NMPC for temperature tracking at the lab-scale plant revealed that further adjustments are necessary to show the full potential of NMPC applications. However, with this master’s thesis a solid basis for further improvements in the operation of semi-batch polymerization reactors is provided.
Derivative-free Refinement Methods for Flowsheet Optimization of Chemical Processes
supervised by: Janus
During process design, engineers apply optimization on chemical flowsheet to find the most suitable process configuration. Depending on the available model information different sets of optimization methods are applicable. A derivative free optimization method, like a memetic algorithm can be applied when model information is hidden in a process simulator. In this work the performance of the memetic algorithm with different local derivative-free optimization solvers has been compared on two case studies that model chemical processes of different complexity.
Performance Improvement of a Blackbox Process Optimization with Surrogate Models
supervised by: Janus
Designing optimal chemical production processes regarding investment and operation costs is a challenging and crucial step in the lifecycle of a process. In this phase changes can be incorporated with low costs but the effect on process performance is high. The design task can be formulated as an optimization problem based on a process model. In this context, the use of commercial process simulators involves the advantages of straightforward model building with large standard libraries and widespread application cases. However, the optimization becomes time-consuming due to a lack of detailed model information.
This work aims to resolve this disadvantage by applying surrogate models within the optimization procedure. Strategies to improve the performance of an algorithm for process optimization are elaborated and their potential is estimated based on historic data. Finally, the developed models are implemented into the existing optimization framework. They are tested on the hydroformylation of dodecene in a thermomorphic solvent system and the industrial MTBE production process as case studies. Faster convergence and an advantageous distribution of solutions compared to the original algorithm are obtained by applying the proposed surrogate strategies.
Validation of an Output-feedback NMPC Scheme for a CFI Polymerization Reactor
supervised by: Tatulea-Codrean
The production of polymers is an integral part in modern process industry. In contrast to the commonly used batch or semi batch reactors, the continuous
polymerization in a tubular reactor has several advantages, such as the absence of idle time, or the better heat removal, but deals with challenges, such as the
avoidance of blockage. The complex non-linear behaviour of the polymerization and the tubular reactor leads to a difficult control task. A suitable approach
dealing with these problems is the non-linear model predictive control. In several previous work the copolymerisation of Acrylamide with 2-Acrylamido-2-methylpropane sulfonic acid
was investigated and a reactor concept was developed. In this work, the application of the NMPC to the existing continuous plant is investigated. First, a new set-up of the plant is developed. On the
basis of the previous work, an accurate model is devised and verified against plant data. In order to observe the process, the model is used to analyse the
estimation problem and tune an Extended Kalman Filter. The performance of the Extended Kalman Filter is tested in simulations studies. The application
of the nonlinear model predictive control is tested and analysed concerning the performance and the runtime behaviour. Finally the estimator and the
nonlinear model predictive controller are combined to an offset feedback NMPC and their performance tested in asynchronous simulation studies.
Optimized autonomous driving of a miniature vehicle
supervised by: Tatulea-Codrean
Autonomous driving is one of the biggest challenges in the automotive research. In this thesis the driving optimization problem is approached making use of the F1/10 race car platform. The F1/10 community, proposing the use of 1:10 scale racing cars, enables research on a small but realistic scale and therewith allows research to explore paths that would be too costly or hazardous in reality. The approach presented in this thesis uses nonlinear model predictive control (NMPC) to control a 1:10 scale autonomous vehicle (AV), by putting it in the center of the optimization scheme and standardizing the environment frame. The problem has therefore a local interpretation of the track, disassociating it from the global shape of the circuit. The optimal driving problem is made compact by joining in a single layer approach the computation of the best trajectory and the controlling of the vehicle. From the need of having a fast controller, making use of training data gathered with the NMPC method, an ANN model is trained. It is shown how it is possible to teach an optimal driving pattern making use of a rather small training dataset. The ANN, thanks to the standardization process adopted for the NMPC, is capable of driving on differently shaped unknown tracks. In order to make the machine-learning controller even more flexible to unexpected situations, obstacle avoidance is applied.
Tuning and performance evaluation of direct optimizing control schemes in a plant-wide NMPC application
supervised by: Tatulea-Codrean
This thesis deals with the centralized, direct optimizing control (DOC) of the Tennessee Eastman Challenge (TEC), which is a benchmark for plant-wide control (PWC). For the direct optimizing control a NMPC solution is applied to the model of the process, which is carefully adapted to the original data. The tuning of various objectives functions, in particular economic objective functions, is done by multi-objective optimization. With the weighted sum and Normalized Normal constraint (NNC) approaches a pareto frontier can be constructed as basis for decision. This way a sufficient tuning for the NMPC of the TEC can be found. The implementation delivers a feasible solution with good profit, computational time and no constraint violation. Several uncertainties can affect the TEC process, from which two with the largest effect are selected in a sensitivity analysis. To robustify the NMPC application, different multi-stage NMPC configurations are tested and evaluated, such that an optimal scenario tree design for the TEC application is found. With this configuration the NMPC can cope with changes in uncertain parameters as well as with changing control objectives.
State estimation for a continuous polymerization process considering uncertain measurements
supervised by: Schweers
Due to the wide usage of poly-acrylic acid and the different applications, different properties are required, which is given by the molecular weight distribution (MWD). Therefore, the implementation of a state estimator combined with a communication system is used in this work, to gain knowledge about non-measurable states of a polymerization process within the used ContiPlant. Through the knowledge of all states, the inputs can be controlled, to achieve a desired MWD. In the first part of the thesis the theoretical background is explained, containing the derivation of the mathematical model and the used state estimator, Extended Kalman Filter (EKF). Afterwards, a communication system is implemented to allow the transfer of the measurements from LabVIEW to MATLAB and the estimates from MATLAB to LabVIEW. Following that, first online estimations are performed, to test the communication system as well as the implemented state estimator. After performing a tuning of the EKF another online state estimation is performed. Due to the success of this online state estimation the used temperature model is extended by three concentration equations, and then simulated. Before the estimation of the full model is performed, an eigenvalue determination over time for the non-linear system is presented, to investigate, for which sampling times the EKF has convergence properties.
Experimental investigation, rigorous model parametrization, and chemical reaction model validation in a tubular reactor for acrylic acid homo-polymerization
supervised by: Schweers
Poly-acrylic acid (PAA) is an important product in the chemical industry. It is mandatory in homo-polymerized form for pharmaceutical formulation and in cosmetic products as a thickener. The product quality of polymers and its purpose changes with the number of crosslinks as well as the molecular weight distribution (MWD). The MWD can be influenced by many input conditions like the inlet concentration of reactants and system characteristics like temperature, pressure, and the used reactor type. It is crucial to get control over the desired product specifications in the production process. This can be performed by a model predictive controller (MPC), which is based on a rigorous model that describes the reaction system. For control the validity of the model is highly important. In this thesis, a theoretical investigation of the reaction system is made. The influence of the reaction conditions on the product specifications and its interactions are carried together and the need to model the influences is discussed. Three different analytical methods for the substance system are investigated and evaluated. In laboratory experiments, acrylic acid polymerizations with the aim of narrow and similar MWD´s are performed, in which reproducible thermal behavior of the plant is reached. An experimental plan for full validation of the reactor model of the tubular reactor is set up. As conclusion of this thesis, a reactor model validation can be done with analytical methods of (multi detection) Size Exclusion Chromatography and Raman Spectroscopy. A validation can be done by investigating the conversion and temperature behavior as well as the influence of the input parameter, especially the ratio of initiator to monomer, on the MWD.
Experimental investigation of the residence time distributions and the heat transfer of viscous flows in a lab-scale continuous polymerization plant for acrylic acid polymerization in aqueous solution
supervised by: Schweers
In der chemischen Industrie erfährt jeder Produktionsprozess zeitliche Änderungen. Dies gilt insbesondere für Batch Prozesse, für welche die Zeit einen wesentlichen Einflussparameter darstellt. Aber auch kontinuierliche Verfahren, die für einen oder mehrere stationäre Zustandspunkte ausgelegt sind, weisen in der Regel dynamische Bereiche auf. Diese ergeben sich mitunter als Resultat externer Störungen. Darunter fallen beispielsweise Schwankungen der Eduktqualität, des Zulaufstroms oder auch Wetterbedingungen. Während die stationäre Prozessmodellierung durch die Implementierung entsprechender Regelungssysteme für den Fall geringer Abweichungen noch eine hinreichende Genauigkeit bietet, ist sie für andere Fälle unzureichend. Unter diese Szenarien fallen beispielsweise auch Prozesse, welche für den Betrieb chemischer Anlagen unerlässlich sind. Hier sind insbesondere die An und Abfahrprozesse der Produktionsanlagen zu nennen, welche trotz ihrer verhältnismäßig geringen Dauer einen sehr bedeutsamen Teil darstellen. Aufgrund der Störanfälligkeit dieser Prozesse können bei fehlender Erfahrung schnell Schwierigkeiten auftreten, die hohe Kosten verursachen. Die dynamische Modellierung bietet somit die Möglichkeit, die Wahrscheinlichkeit kostspieliger Schäden zu senken. Ein weiteres Einsatzgebiet stellen Produktwechsel dar. Durch die realistische mathematische Abbildung der Übergangsprozesse wird die Implementierung entsprechender Regelungssysteme ermöglicht. Mit deren Hilfe lassen sich Prozesse über optimierte Trajektorien von den einen in den anderen Zustand bringen. Dadurch werden sogenannte Waste-Ströme reduziert. Dabei handelt es sich um Nebenprodukte oder Produkte minderer Qualität, welche sich nicht für den Verkauf eignen. Auf diese Weisen lassen sich durch den Einsatz dynamischer Modellierungen somit Rohstoffe und Kosten einsparen.
Application of online state estimation algorithms on a lab-scale continuous polymerization plant for acrylic acid polymerization in aqueous solution
supervised by: Schweers
The present work deals with the application of online state estimation algorithms on a lab-scale continuous polymerization plant for acrylic acid homo-polymerization in an aqueous solution. For this purpose a state-of-the-art server client system communicating via the OPC UA protocol is set up for data transfer and storage. As state estimation algorithms an Extended Kalman Filter and a Moving Horizon Estimator are implemented and tested against a variety of ODE systems in an online and offline estimation environment as well as a model of the polymerization plant. Single- and multi-rate measurement capabilities are implemented for the EKF whereas the MHE is equipped with a variety of functionalities, ranging from a Riccati update of the covariance matrix to a variable length arrival cost.
Calibration of Stationary Distillation Models based on Process Data for Multi-Objective Resource Efficiency Optimization
supervised by: Janus
In the scope of man-made climate change, increasing the resource efficiency of our existing production processes is an important intermediate step towards a sustainable future. Chemical plants have, for the mostly been operated over decades and offer improvement potential without physically changing the equipment.
Within this thesis, the concept of indicator-based resource efficiency optimization is transferred to distillation columns. A framework is developed, whose application enables exploiting the given improvement potential by combining information from process data with rigorous process models. On the prerequisite of a set of resource efficiency indicators (REI) the framework consists of: the automated detection of steady-states from historic process data, an efficient workflow for multi-scenario model calibration and multi-objective optimization for decision support. Herein, the model calibration step is of particular importance to ensure reliable model prediction over a wide range of process inputs. A novel method was developed, which allows automated model parameter fitting on an arbitrary number of steady-states overcoming previous limits of Aspen Plus. The framework is successfully applied to a real-world distillation column for solvent recovery of ethanol. For this application case only minor trade-offs between the most economic and most resource efficient operation are detected. By stabilizing the process closer to the constraints, significant improvements in the energy efficiency, the material efficiency and profitability can be achieved simultaneously. The thesis subsequently shows that REI-based decision support can be generated with two approaches: data-based analysis of the historic process data and model-based optimization.
Optimizing a Chemical Process by using Evolutionary Algorithms and the Process Simulator CHEMCAD
supervised by: Janus
Evolutionary Algorithms (EAs) is used in this work for the hydroformylation of long chained hydrocarbon. Genetic Algorithm (GA) and Memetic algorithm (MA) are used for the process wide optimization hyydroformylation of 1-dodecene process. Optimization problem is to minimize the overall cost which consists of capital, raw material and utility costs. The effect of catalyst leaching is also the part of the optimization problem. The design variables consisting of mixture of continuous and discrete variables. A commercial process simulator, CHEMCAD is used for the flowsheeting of the hydroformylation process. CHEMCAD offers COM interface i.e. the information exchange between flowsheet and some external algorithm can be done. This opportunity is taken to optimize the process using evolutionary algorithms. Different solver and solver setting has been employed and tried to find a suitable evolutionary algorithm and its settings for the hydroformylation process.
Optimizing control of SMB processes
supervised by: Gerlich
Preparative chromatography is preferably used in difficult separation tasks involving sensitive products. In preparative chromatography the standard operation mode is batch. However, their large-scale use is expensive due to the high desorbent equirements. The simulated moving bed (SMB) process, a continuous counter-current chromatographic separation process, has proven the ability to improve the separation in terms of productivity, desorbent consumption and purity. The counter-current movement is simulated by periodically switching of ports in the direction of liquid flow. The advantages of SMB processes, however, are achieved with higher complexity in terms of operation and design. In order to enhance the economic potential, advanced optimizing control schemes based on rigorous models can be employed. In this work, economic NMPC is applied to the SMB process using the framework do-mpc, in which the solution of the underlying dynamic optimization problem is based on a direct collocation approach. The fundament of the rigorous model is the transport dispersive model (TDM) for describing the dynamic behavior of the columns which is combined with the finite volume WENO scheme for spatial discretization. Following results conclusions are drawn from this thesis. The framework do-mpc is suitable to use for applying economic NMPC to a complex, periodic process such as the SMB. Using CasADi, the symbolic building block of do-mpc, improved the required computation times significantly. Also, a simultaneous approach with direct collocation solved the underlying dynamic optimization problem efficiently. Finally, a comparison of objective functions with different complexity has shown that a straightforward objective is sufficient to optimize the case study process.
Online process monitoring in SMB processes
supervised by: Gerlich
In simulated moving bed (SMB) processes, only scarce measurement information is available. This is a barrier for the application of online optimal control methods for improved plant operation. In this thesis a state and parameter estimation scheme is adapted for the separation of phenylalanine and tryptophane in a SMB pilot plant. The column dynamics are modeled with the transport dispersive model discretized with finite volumes method and the weighted essentially non-oscillatory (WENO) method. The parameter estimation is based on a dynamic optimization problem which is cast into a large-scale nonlinear programming problem by direct collocation. A tuning for the optimization problem, which balances computational speed and parameter convergence, is developed. In simulation studies, the convergence of estimated states and parameters is shown for multiple cases including step changes in a single parameter and an offset in all estimated parameters of up to 30 % at different operating points. It is shown that the estimation scheme is not dependent on cyclic steady state (CSS) condition and can be started shortly after the start of operation. In another experiment, it is shown that the scheme is capable of compensating plant-model mismatch in a parameter which is not estimated. The adaption of the state and parallel estimation scheme yields a real time feasible online monitoring tool for future integration into the pilot plant operation.
Robust MPC schemes to handle structural plant-model mismatch
supervised by: Thangavel, Subramaianan
The presence of structural plant-model mismatch poses a challenge to model based control. Even-though extensive research has been done to handle additive and parametric uncertainties present in the plant model, structural plant-model mismatch is rarely considered. In this thesis, different databased approaches are proposed to handle structural plant-model mismatch by capturing the unmodelled dynamics using different methods. The first method captures the unmodeled dynamics using a linear model with parametric uncertainty. An offline optimization problem is formulated to get minimal intervals on the bounds of the parameters that would explain all the observed measurements. The second method is augmenting a nominal model with a model error model (MEM) which captures the error between the nominal model and the true plant measurements. The uncertainty in the model is handled using three different robust controllers (Multi-stage MPC, tube MPC and tube enhanced multi-stage MPC).
The performance of the different controllers are evaluated on a continuous-stirred tank reactor benchmark case study.
Bachelor's Theses 2019
Analysis of Raman Spectroscopy in Continuous Copolymerization
supervised by: Semrau
Although polymers are one of the most produced products currently, their process of obtaining is limited to the use of batch reactors, which have a series of disadvantages such as low thermal efficiency and idle times. Therefore, the transition to a continuous reactor seems to be a good alternative to overcome these disadvantages. In order to use the benefits of continuous reactors, it is also necessary to make additional efforts in certain areas, such as in the control of the reactor. A basic element of the control system is the measurement of the controlled variables. Unlike variables such as temperature and pressure, the concentration is a bigger challenge, since it is not obtainable by trivial methods and is a variable that cannot be ignored, since one of the main advantages of continuous reactors is ensure a constant quality product.
Previously in other works a continuous polymerization reactor has been developed to copolymerize Acrylamide with 2-Acrylamido-2-methylpropane sulfonic and an approach has been made to measure the concentrations through Raman spectroscopy. In this work, the analysis of the spectra generated by Raman spectroscopy was deepened to achieve a better interpretation and modeling of their peaks. It began with the implementation of a secondary set up of the plant to measure the desired sample spectra more easily. Then the influences that certain variables have on the measurement of the spectra were determined, such as the flow rate, and some alternatives were proposed to minimize their effects. Likewise, the feasibility of distinguishing the spectra of the monomers from the ones of the polymers was analyzed, and once this was verified, the main part of this work was dealt with, the development of a methodology and a program capable of determining the hard models of the pure components, for its subsequent use in the modeling of multicomponent mixture spectra.
Optimal Design of Experiment for Kinetic Models of Reductive Amination
supervised by: Kaiser
Almost all chemical processes require a rigorous kinetic model. It is used to predict reaction rates, yield, chemical selectivity and changes in enthalpy in the reaction medium. These can be used for simulation and optimization of reactions for safe and profitable plant operation. In order to find satisfying kinetic models experiments must be performed, because the reaction constants are dependent on complex, computationally expensive, molecular interactions that currently are not feasible to be calculated without experimental data. Designing these experiments is itself very important since a lot of resources and time need to be invested to do these experiments. Therefore their relevance, statistical significance and ability to discern between different models should be maximised. Both, the design and the conduction of the experiments will be presented in this thesis.
Prof. Engell's 65th Birthday
On the 8th of March, we had the pleasure to celebrate the 65th birthday of prof. Engell. His family, friends, co-workers and PhD students gathered together for discussions and presentations on the topic "Diversity as a Resource", a topic that certainly characterized prof. Engell's scientific carrier. The following presentations were given:
- "Less is more or less is a bore? Looking back on my research design" by Sebastian Engell (dyn@TUDO)
- "Lephroig or Johnny Walker ― Diversity in research explained with whiskey" by Stefan Krämer (Bayer AG, Leverkusen)
- "When is “more” more? Benefits and limitations of diversity in multi-partner projects" by Svetlana Klessova (inno TSD, Nice, France)
- "Diversity as a resource for knowledge-generation in the reality of a scientific research group― a model-based analysis" by Corina Nentwich and Lukas Maxeiner (dyn@TUDO)
- "Connecting systems, connecting people, connecting disciplines" by Kai Dadhe (Evonik Infrastructure and Services, Hanau)
- "Language is everywhere ― Notes on interdisciplinarity from a linguist‘s point of view" by Uta Quasthoff (Emerita, TU Dortmund)
- "Obstacles and opportunities related to diversity in academic education ― personal experiences" by Kirsten Lindner-Schwentick (BCI@TUDO)
Fresh dyn doctors
From left to right: Prof. G. Schembecker, Prof. J. B. Jorgensen, D. Haßkerl, Prof. S. Engell, Prof. A. Gorak, Prof. M. Thommes
2019 was a special year for Daniel Haßkerl, who had the chance to successfully defend his PhD thesis.
We congratulate the new doctor on his hard work and dedication and wish him all the best in his future endeavours, scientific or otherwise!
- Daniel Haßkerl: "Economic Performance Optimization by Direct Optimizing Control Applied to Reactive Distillation Processes", 25.02.2019
NAMUR Award for Daniel Haßkerl
From the left: Prof. Dr.-Ing. Sebastian Engell, 2019 NAMUR-Award Winner Dr.-Ing. Daniel Haßkerl (now with BASF), and the Chairman of the Board of NAMUR Dr.-Ing. Felix Hanisch, Bayer AG, also a DYN PhD graduate. Source: NAMUR
On the NAMUR Annual General Meeting 2019 in Bad Neuenahr, DYN alumni Dr.-Ing. Daniel Haßkerl has been awarded the NAMUR-Award 2019 for his dissertation “Economic Performance Optimization by Direct Optimizing Control Applied to Reactive Distillation Processes”.
NAMUR is the international user association of automation technology in the process industries. The main goal of the association is to further develop the field of process automation by gathering
expertise and proposing new ideas across companies. The Annual General Meeting of NAMUR is a congress for members and invited guests of NAMUR. It is attended by about 650 participants, predominantly
from the industry. NAMUR awards one PhD student per year for his or her outstanding contribution that addresses intelligent process control and operational management or other elements of process automation. We congratulate our former colleague Daniel Haßkerl for winning this outstanding award!
dyn winter trip
1 / 6
Visit at TUDo power plant
2 / 6
Visit at TUDo power plant
3 / 6
The bench
4 / 6
Stadium tour
5 / 6
Borussia!
6 / 6
Südtribune
This year the winter excursion focused on two of Dortmund's highlights: the TU and the Signal Iduna Park. The day started off with a very interesting visit of the TU Dortmund power plant, where we learned about the challenges of supplying the entire campus with electricity, heat and cooling capacity. The visit included tours through the engine rooms and the control rooms, as well as a walk through the infamous subterranean tunnel network of the university. After a well-deserved break for Glühwein and cookies, the day continued with a tour of the BVB stadium, where we established that we would do well on the Südtribune. Later on, the evening continued with a pleasant dinner at the Hövels restaurant in the city.
dyn summer trip
This year's summer trip taught the DYN group about the production process of carbonate ("The stuff diamonds are moade of") at the KG Deutsche Gasrußwerke GmbH & Co in Dortmund.
Afterwards, we visited one of the oldest parts of the City of Düsseldorf, Kaiserwerth. We came back to the city center of Düsseldorf, where we spent the afternoon with chats
and drinks, with a panorama boat tour on the river Rhein.
International Summer Program 2019
This year, 27 students from our oversea partner universities in Asia and America were welcomed to Dortmund in the 17th edition of the International Summer Program.
The ISP, organized by the DYN chair in close collaboration with the faculty of American Studies and the International Office, has provided the students with bothF an interesting learning environment
for various academic subjects and a balanced cultural program. The participants have given the program best grades for the effort that has been invested which underlines the acceptance and the value
of the program for the international activities of the faculty and of the university.
Excursion to COVESTRO
On June 25, 2019, as part of the course "Sicheres und optimiertes Betreiben von Anlagen in der Chemie- und Pharmaindustrie", the students visited the production site of Covestro AG in Leverkusen. The students visited the
pilot plants at Covestro. The day was concluded with some presentations on advanced process control at Covestro.
dyn at the Science Slam 2019
The TU-Dortmund Science Slam is an annually event organized by the Graduate Center to offer junior researchers a platform to present their research to a non-specialist audience and make it visible.
This year the PhD candidate Corina Nentwich gave a very interesting talk entitled "Can an AI write my dissertation?" ("Kann eine KI meine Dissertation schreiben?").
Besides research: the BCI Cup
Our DYN football team, the DYNosaurs, participated at the BCI-Cup 2019, the faculty tournament where mixed teams of students and researchers face eachother in fierce competition.
The event took place on the 21th September on the grounds of the TU Dortmund Sportzentrum. This year the DYNosaurs gave their best reaching the semi-finals of the tournament and finishing at the 4th place.
The tournament was a great success for the DYN and we are eager to do better in next year's BCi-Cup!
Further "athletic" endeavours : sport and creativity
On May 22nd, 2019, the Process Dynamics and Operations Group (DYN), had a sensational participation in the annual “Campus-Run” contest at the Technical University of Dortmund.
The DYN Run Club, led by the DYNosaur, participated the annual “Campus-Run” contest of TU Dortmund. Not only the team performed well in the run marathon
of 2.5 km through the TU Dortmund campus, but also, thanks to the creativity of the team, attracted the attention of the audience and the organizing staff in the “Custom competition” contest.
The outstanding custom of the team became the photo attraction to not only many photographers on site, but also the official press “Ruhr Nachrichten”.
Overall, the team successfully took part in the event and gathered more experience for prominent participations in the following years.
PSE project group presentation at MERCK.
The group from the international study program "Process Systems Engineering" took the chance to present and discuss their final results with the audience at Merck.
The project on "Continuous Production of High Quality Perhydropolysilazane in a Modular and Automated Plant" was proposed by Merck and supervised by Reinaldo Hernández and Pourya Azadi from the DYN group.
The Automation & Robotics students presented their projects
On Friday March 1st two groups of Automation and Robotics MSc students successfully presented the results of their project work carried out the past three months at the DYN chair.
The following presentations were given:
- Group 1: Designing and implementing a distributed control structure for a multi-agent system using ROS and RaspberryPi
- Group 2: Cost Efficient Wireless Process Automation
The General Consul of India visited the dyn laboratories
Pratibha Parkar, the Consul General of India in Frankfurt, visited the TU Dortmund University on 9th of April, 2019. As part of her visit to the TU Dortmund, the DYN group had the honor of showing her their student and research laboratories. She also met students from the Master's program Process Systems Engineering. During the subsequent discussion in the International Department, Ms. Parkar also met Indian students and program coordinators from other faculties of the TU Dortmund. The dicussions were focused in particular on the support options for Indian students in Germany.
Visit from Prof. Ignacio Grossmann
From February 4th to 12th 2019, Professor Ignacio Grossmann from the Carnegie Mellon University, USA, visited the BCI department.
Prof. Grossmann gave a 1-day course and 3 lectures at the department, including the Gambrinus and BCI Jubilee Lecture celebrating the 50th anniversary of the BCI department.
- 1-day course: "Recent Theoretical and Computational Advances in the Optimization of Process Systems under Uncertainty"
- Lecture 1: "Recent Advances in Computational Models for the Discrete and Continuous Optimization Models of Industrial Process Systems"
- Lecture 2: "Global Optimization of Nonconvex Nonlinear Generalized Disjunctive Programs"
- Gambrinus and BCI Jubilee Lecture: "Process Systems Engineering: Evolution, Accomplishments and Future Research Trends"