Abstract This paper discusses the implementation of interfacial transport in the systematic model... more Abstract This paper discusses the implementation of interfacial transport in the systematic modelling methodology by Preisig (2012) and how to lecture the approach as well as its implications to engineering students. We first elaborate on the theoretical concept of interfaces before abstracting them into topologies that allow for comprehensible as well as precise mathematical assessments of the transport processes involved. Thereby, any interfacial transfer is rooted in its fundamental parts of transport towards, across and away from the interface. These steps can be assessed by only a handful of algebraic equations from an ontology, which covers the entirety of physical concepts, state derivatives, and conservation principles. Due to this, interfacial transport can rapidly be implemented in process models, significantly aiding subsequent model simplification as well as model fitting.
Abstract This paper presents a new modelling framework called the Process Modeller Suite, which c... more Abstract This paper presents a new modelling framework called the Process Modeller Suite, which combines ontologies and graph-based modelling to do automatic code generation for models. The process models are represented by a process graph where the graph components capture system knowledge in the ontology in a systematic manner. The result is a software tool that promotes correct model formulation, interoperability between different modelling environment and reusability of the models and also generates a proper model documentation.
Abstract Visual modelling serves the purpose of designing process models, discussing them on the ... more Abstract Visual modelling serves the purpose of designing process models, discussing them on the back of an envelope, a serviette, the meeting-room board and define & edit them in the graphical user interface ModelComposer, which is a component of our ontology-based simulation environment. Models that spread over large range of time-scales do not connect straightforwardly as the fast time-scale usually is computationally intensive to the extent where on the large time-scale one uses surrogate models derived from the detailed models. The network-of-network approach extends readily to multi-scale system – from quantum to mechanical properties.
Abstract We present a new modelling framework called the ModelComposer. This framework combines w... more Abstract We present a new modelling framework called the ModelComposer. This framework combines well-known concepts such as equation-based modelling, graph-based modelling and code generation to produce simulation models automatically. Moreover, the framework uses ontologies as the bases, which means that interoperability of models produced by the modelling framework is high. The model information is stored in JSON-files, and code generation for different programming languages is used to transform the defined models into executable simulation. A result is a modelling tool that promotes the reuse of models and interoperability between networks of models of different domains. This paper will present the overall layout of the ModelComposer and how the framework handles internal conversions within a network and how different networks communicates. This is exemplified by a case study presenting the model construction of a dynamic mixing tank and the simulation of the resulting model.
Abstract We proposed a dynamic model for butanol production from glucose and xylose as representa... more Abstract We proposed a dynamic model for butanol production from glucose and xylose as representative lignocellulosic sugars. The model consists of the extended Monod model for growth on mixed sugars with the effect of noncompetitive inhibition between them. The Luedeking-Piret model was used for butanol production. Sugar uptake equations were derived through a semi-empirical approach by employing yield coefficients. The proposed model has 10 parameters. We estimated the parameters by using data from a feed interval of a fed-batch fermentation experiment, which we performed with glucose and xylose by Clostridium beijerinckii NCIMB 8052. Parameter estimation imposed a constrained optimization problem where the objective was to minimize the sum of squared error (SSE) between the predicted and observed values of components. The constrained nonlinear optimization problem was solved using fmincon in MATLAB 2017b Optimization Toolbox based on the interior point algorithm. The problem was solved after 90 iterations and 1141 function evaluations resulting an SSE value of 8.9417e-14. We validated the model with data sets of four other feed intervals, and the index of model accuracy, which was sum of squared relative error (SSRE) was 2.2245. We conducted the sensitivity analysis with end point deviations to identify the critical parameters. Substrate affinity constant for xylose, KsX had the highest impact on cell mass, xylose and butanol concentrations, and had the second greatest effect on glucose concentration. This is the first study which addresses the dynamic modeling of simultaneous utilization of lignocellulosic sugars for butanol production. The proposed model is the first attempt to incorporate the interaction between sugars. The methodology we proposed provides an understanding in the dynamics of the butanol production from glucose and xylose, and it can further be used for guiding optimal process design and operation.
Abstract This paper describes the concepts and philosophy behind the multi-scale software framewo... more Abstract This paper describes the concepts and philosophy behind the multi-scale software framework developed in conjunction with the project ”Modelling of morphology development of micro- and nanostructures” (MoDeNa). MoDeNa provides a framework for systematically generate and update simple models acting as surrogates for the complex counter part, which they approximate in a given domain. The framework also provides an automatic procedure for expanding the domain if so found necessary during the course of the simulation. To our knowledge this is the framework that is specifically designed to incorporate all these elements and simultaneously facilitate run-time parameter-estimation in addition to providing a consistent way of defining surrogate models and their incorporation into applications.
Abstract This paper discusses the implementation of interfacial transport in the systematic model... more Abstract This paper discusses the implementation of interfacial transport in the systematic modelling methodology by Preisig (2012) and how to lecture the approach as well as its implications to engineering students. We first elaborate on the theoretical concept of interfaces before abstracting them into topologies that allow for comprehensible as well as precise mathematical assessments of the transport processes involved. Thereby, any interfacial transfer is rooted in its fundamental parts of transport towards, across and away from the interface. These steps can be assessed by only a handful of algebraic equations from an ontology, which covers the entirety of physical concepts, state derivatives, and conservation principles. Due to this, interfacial transport can rapidly be implemented in process models, significantly aiding subsequent model simplification as well as model fitting.
Abstract This paper presents a new modelling framework called the Process Modeller Suite, which c... more Abstract This paper presents a new modelling framework called the Process Modeller Suite, which combines ontologies and graph-based modelling to do automatic code generation for models. The process models are represented by a process graph where the graph components capture system knowledge in the ontology in a systematic manner. The result is a software tool that promotes correct model formulation, interoperability between different modelling environment and reusability of the models and also generates a proper model documentation.
Abstract Visual modelling serves the purpose of designing process models, discussing them on the ... more Abstract Visual modelling serves the purpose of designing process models, discussing them on the back of an envelope, a serviette, the meeting-room board and define & edit them in the graphical user interface ModelComposer, which is a component of our ontology-based simulation environment. Models that spread over large range of time-scales do not connect straightforwardly as the fast time-scale usually is computationally intensive to the extent where on the large time-scale one uses surrogate models derived from the detailed models. The network-of-network approach extends readily to multi-scale system – from quantum to mechanical properties.
Abstract We present a new modelling framework called the ModelComposer. This framework combines w... more Abstract We present a new modelling framework called the ModelComposer. This framework combines well-known concepts such as equation-based modelling, graph-based modelling and code generation to produce simulation models automatically. Moreover, the framework uses ontologies as the bases, which means that interoperability of models produced by the modelling framework is high. The model information is stored in JSON-files, and code generation for different programming languages is used to transform the defined models into executable simulation. A result is a modelling tool that promotes the reuse of models and interoperability between networks of models of different domains. This paper will present the overall layout of the ModelComposer and how the framework handles internal conversions within a network and how different networks communicates. This is exemplified by a case study presenting the model construction of a dynamic mixing tank and the simulation of the resulting model.
Abstract We proposed a dynamic model for butanol production from glucose and xylose as representa... more Abstract We proposed a dynamic model for butanol production from glucose and xylose as representative lignocellulosic sugars. The model consists of the extended Monod model for growth on mixed sugars with the effect of noncompetitive inhibition between them. The Luedeking-Piret model was used for butanol production. Sugar uptake equations were derived through a semi-empirical approach by employing yield coefficients. The proposed model has 10 parameters. We estimated the parameters by using data from a feed interval of a fed-batch fermentation experiment, which we performed with glucose and xylose by Clostridium beijerinckii NCIMB 8052. Parameter estimation imposed a constrained optimization problem where the objective was to minimize the sum of squared error (SSE) between the predicted and observed values of components. The constrained nonlinear optimization problem was solved using fmincon in MATLAB 2017b Optimization Toolbox based on the interior point algorithm. The problem was solved after 90 iterations and 1141 function evaluations resulting an SSE value of 8.9417e-14. We validated the model with data sets of four other feed intervals, and the index of model accuracy, which was sum of squared relative error (SSRE) was 2.2245. We conducted the sensitivity analysis with end point deviations to identify the critical parameters. Substrate affinity constant for xylose, KsX had the highest impact on cell mass, xylose and butanol concentrations, and had the second greatest effect on glucose concentration. This is the first study which addresses the dynamic modeling of simultaneous utilization of lignocellulosic sugars for butanol production. The proposed model is the first attempt to incorporate the interaction between sugars. The methodology we proposed provides an understanding in the dynamics of the butanol production from glucose and xylose, and it can further be used for guiding optimal process design and operation.
Abstract This paper describes the concepts and philosophy behind the multi-scale software framewo... more Abstract This paper describes the concepts and philosophy behind the multi-scale software framework developed in conjunction with the project ”Modelling of morphology development of micro- and nanostructures” (MoDeNa). MoDeNa provides a framework for systematically generate and update simple models acting as surrogates for the complex counter part, which they approximate in a given domain. The framework also provides an automatic procedure for expanding the domain if so found necessary during the course of the simulation. To our knowledge this is the framework that is specifically designed to incorporate all these elements and simultaneously facilitate run-time parameter-estimation in addition to providing a consistent way of defining surrogate models and their incorporation into applications.
Uploads
Papers by Heinz A Preisig