- Email: [email protected]
Multiscale Modelling Framework for Chemical Product-Process Design
19th European Symposium on Computer Aided Process Engineering – ESCAPE19 J. JeĪowski and J. Thullie (Editors) © 2009 Elsevier B.V. All rights reserved.
495
Multiscale Modelling Framework for Chemical Product-Process Design Ricardo Morales-Rodríguez, Rafiqul Gani CAPEC, Department of Chemical and Biochemical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark, [email protected]
Abstract The objective of this paper is to present a novel computer-aided model-based framework for product-process design that also includes multiscale modelling features. To develop this framework, a combination of different computational tools, such as, property prediction packages, modelling tools, simulation engines, solvent selection software, etc, are necessary together with a set of established systematic work-flow and data-flow for various types of design problems. This framework allows the user to cover a wide range of problems at different scales (of length and time) and disciplines of chemical engineering and science in an easy and efficient manner; achieving in this way the development of a product-process with the desired end-use characteristics. Development of a pesticide formulation product where its uptake into the plant is used as a product performance measure, is used to highlight the work-flow and data-flow in the multiscale modelling framework.
Keywords: Multiscale modelling, product-process design, framework, virtual lab 1. Introduction The design, development and reliability of a chemical product and the process to manufacture it, need to be consistent with the end-use characteristics of the desired product. One of the common ways to match the desired product-process characteristics is through trial and error based experiments which can be expensive and time consuming. An alternative approach is the use of a systematic model-based framework according to an established work-flow in product-process design, replacing some of the time consuming and/or repetitive experimental steps. Furthermore, for many chemical products the appropriate models for product-process design need to have multiscale features as the properties defining the chemical structure and the product end-use characteristics are dependent on parameters of different size and time scales. The advantages of the use of multiscale modelling approach in this case is that in the design, development and/or manufacturing of a product-process, the knowledge of the applied phenomena can be provided at diverse degrees of abstractions and details. Some authors [1-4] have highlighted the importance of the multiscale and multidiciplinary approach in product-process design and identified design issues related to different scales of size, time and complexity. The development of a computer-aided framework for product-process design including a multiscale modelling option is very important for analysis, design, and/or identification of feasible chemical product candidates because it allows one to consider
R. Morales-Rodríguez and R. Gani
496
processing issues during the development of the product. The multiscale modelling framework should include the product and process design components, modelling tools and templates (work-flow) for guiding the user through the appropriate design steps. The integration of computational tools is also necessary to increase the application range of the computer-aided product-process framework; where the connection between computational tools could be established through well-defined COM-objects or the CAPE-OPEN standards. A novel computer-aided model-based framework for product-process design, that also includes multiscale modelling features, is presented in this paper. To develop this framework, a combination of different computational tools, modelling tools, simulation engines, molecular and mixture design software, solvent selection software, etc, are integrated within a set of established systematic work-flow and data-flows for various types of design problems. The performance of the multiscale model-based framework, the associated models and the work-flow for a specific product-process design is illustrated through a case study involving the modelling and design of a pesticide product; here, the pesticide uptake inside the leaf is designed and evaluated through the use of the framework, where a suite of models, database of properties, chemicals, and, a modelling tool are employed.
2. Multiscale Modelling Framework for chemical product-process design: Requirements. The multiscale modelling framework allows the user to cover a wide range of problems at different scales (of length and time) and disciplines of chemical engineering and science in an easy and efficient manner; achieving in this way the development of a product-process with the desired end-use characteristics. Another requirement of the modelling framework and the software architecture is a feature that provides the means for integration and merging of methods and tools from different sources. This architecture needs to accommodate models used for the prediction of the product behaviour/performance. Here, ICAS-MoT is the main modelling tool, which provides interactions with modelling engines, external software through the use of COM-objects, and also with external simulators through with the use of CAPE-OPEN standards. More details about this synergy is described by Morales-Rodríguez et al.[5] . Figure 1 illustrates the multiscale modelling framework for chemical product-process design where four main parts can be found:
problem definition, product design, product-process modelling, product-process evaluation.
Each of them have sub-steps that guide the user through a systematic work-flow and data-flow to solve specific design problems.
Multiscale Modelling Framework for Chemical Product-Process Design
497
Fig. 1. Multiscale modelling framework for chemical product-process design.
Problem definition: The multiscale modelling framework for chemical product-process design starts with the conceptual definition of the design problem, which concerns with setting the desired characteristics of the product, its properties, special qualities, ingredients to make the product, etc., that might apply for a new product or for existing products that need to be improved. Product Design: Often, information needed to perform product behaviour analysis is lacking. To overcome these gaps of information, computer-aided methods and tools are employed. That is, the use of specialized computer-aided tools such as, databases containing properties of chemicals, property prediction packages, molecular and mixture design software, solvent selection tools, etc. (all these computational tools can be found in the ICAS software developed by CAPEC at Technical University of Denmark) are employed in order to address the issue of missing information. The first sub-step involves generation of data/knowledge related to the product needs, ingredients, assumptions in the conceptual design work-flow, historical records, etc., that are needed in the subsequent steps. Another sub-step in this part consists of the selection of the materials to be added to the product taking into consideration, its application; functional property values; primary property values; main active ingredients; solvents; coatings; etc., as well as the calculation of the necessary properties (such as, diffusion coefficients, partition coefficient, surface tension, etc.) related simulation model to be used in product performance evaluation. Product-Process Modelling: Once the necessary information for evaluating the product performance through the generated models has been retrieved, a modelling tool is used to assist in the simulation and generation of alternatives and verification of the formulated properties through the ICAS-MoT (available in ICAS). ICAS-MoT is essentially a modelling tool able to generate, analyze and simulate mathematical models without extra programming efforts. In this way, the predicted product behaviour is compared against the desired targets specified at the beginning, and decided if the targets have been matched.
498
R. Morales-Rodríguez and R. Gani
Product-Process Evaluation: If the design targets are not matched, a new product design problem is started, or, a model-based analysis is made and the subsequent steps repeated to find another solution. If the targets are matched, before proceeding to manufacturing the product; an option to evaluate product-process performance is provided through the use of ICAS-MoT. Furthermore, an option to make sustainability analysis in order to evaluate the environmental and economical impact for the production and performance of the product is provided. More details of the systematic multiscale modelling framework for chemical product-process design of microcapsules for controlled release of active ingredients can be found in Morales-Rodríguez and Gani (2008) [6] .
3. Case Study for pesticide uptake prediction through the use of the multiscale modelling framework for chemical product-process design. To highlight the use of the modelling framework, a pesticide uptake example has been chosen due to its multiscale features from the modelling point of view.
Fig. 2. Multiscale description of pesticide uptake
Figure 2 shows that the modelling of the pesticide uptake can be carried out in an entire field that basically corresponds to the macroscale where the behaviour in a specific area is analyzed, and/or the impact of the pesticide to the environment is calculated; an analysis in a droplet scale is also performed, that is, the behaviour of the active ingredient inside the droplet as well as the vaporization of the liquid phase are predicted (this plays an important role in pesticide uptake); further analysis in a smaller scale (the behaviour of the pesticide uptake inside the leaf can be predicted) is performed. Here, the importance of the pesticide mass transfer phenomenon between the droplet and the leaf as well as inside the leaf is considered. A computer-aided tool called “Virtual Product-Process Design Lab” has been developed, where the multiscale modelling framework has been included, for chemical product-process design. The models at the various scales, needed for design and analysis of pesticide uptake is available in the model library. The case study involving
Multiscale Modelling Framework for Chemical Product-Process Design
499
the design of the pesticide formulation and the prediction of the pesticide uptake is highlighted for the active ingredient (AI) cyanazine, which is to be sprayed in a wheat field together with a surfactant called C13E11. Figs. 3a-3d depict the workflow for the design of cyanazine based pesticide formulation and the models needed at various scales, from the droplet scale to the overall pesticide uptake. Fig. 3a (problem definition) shows the step where the generation of information related with the product properties, ingredients, assumptions, etc. is carried out and saved in a documentation file for future examination and/or use.
Fig. 3. Design of one Pesticide using the multiscale modelling framework
Fig. 3b (product design) shows the selection of the plant where the pesticide will be applied, also the selection of the pesticide to be used in the formulated product as well as the surfactant. All the information that is collected in this part is transferred to the modelling tool. Fig. 3c (product-process modelling) shows the different mathematical models [7] that are available in the model-based library for this problem, each model takes into account different phenomena as well as different assumptions allowing the user to have a wider range of applicability in the pesticide formulation design. Note that this is one of the reasons for having a large model-based library. Once the virtual design has been done, an evaluation of the product behaviour performance is carried out; Fig. 3d (product-process evaluation) shows the results highlighting the calculated amounts of pesticide uptake by the leaf. Here, it is allowed to ask if the performance criteria have been satisfied. If “Yes” a new product alternative has been developed and verification
R. Morales-Rodríguez and R. Gani
500
by experiment can now be performed, if necessary. Otherwise, it is possible to return to the appropriate product design step and repeat with other options until the desired performance criteria are matched. The size of the mathematical model depends on the scenario being evaluated and the number of discretization points. The model consists of a set of differential and algebraic equations representing the phenomena of evaporation rate of the droplet, mass transfer of the pesticide AI and solvent through wax and cuticle layers of the leaf. It is important to note the multidisciplinary nature of the problem (different sources of information from different parts or fields of science, for instance, the information about the leaf conformation, diffusion coefficients for biological systems, weather conditions that effect in the behaviour of the plant, etc.) and the multiscale characteristics of the model.
4. Conclusion A systematic multiscale model-based framework for product-process design has been developed and its application illustrated through the design/analysis of the uptake cyanazine as the pesticide AI. The usual trial and error experimental-based approach has been replaced with a virtual product/process lab, which allows some of the time consuming and repetitive steps to be performed virtually through a model-based framework. In this way, the resources of experimental work are reserved for the final verification of the product, when a small number of candidates matching the desired end-use characteristics of the product have been identified. Finally, for the virtual product-process design lab to succeed, reliable multiscale models must be available in a model-library and used through an appropriate model-based framework, that can also help to generate models, when they are not available. The framework also contains models for design of devices for controlled release of active ingredients and for design/analysis of direct methanol fuel cells.
5. References [1] Charpentier, J.C., Comp. Chem. Eng., 57 (2002) 4667. [2] Charpentier, J.C., ESCAPE 17: Vol. CACE 24, (2007) 11. [3] Gani, R., Comp. Chem. Eng., 28, (2004) 2441. [4] Klatt, K.U. and Marquardt, W., ESCAPE 17: Vol. CACE 24, 19. [5] Morales-Rodríguez, R., Gani, R., Déchelotte, S., Vacher, A. and Badouin, O., Chem. Eng. Res. Des., 86 (2008) 823. [6] Morales-Rodríguez, R. and Gani, R. Proceedings of Chempor 2008. Braga Portugal. [7] Rassmusen, J.K., Prediction of Pesticide Uptake in Plants, Master Thesis, Department of Chemical engineering of Technical University of Denmark.
495
Multiscale Modelling Framework for Chemical Product-Process Design Ricardo Morales-Rodríguez, Rafiqul Gani CAPEC, Department of Chemical and Biochemical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark, [email protected]
Abstract The objective of this paper is to present a novel computer-aided model-based framework for product-process design that also includes multiscale modelling features. To develop this framework, a combination of different computational tools, such as, property prediction packages, modelling tools, simulation engines, solvent selection software, etc, are necessary together with a set of established systematic work-flow and data-flow for various types of design problems. This framework allows the user to cover a wide range of problems at different scales (of length and time) and disciplines of chemical engineering and science in an easy and efficient manner; achieving in this way the development of a product-process with the desired end-use characteristics. Development of a pesticide formulation product where its uptake into the plant is used as a product performance measure, is used to highlight the work-flow and data-flow in the multiscale modelling framework.
Keywords: Multiscale modelling, product-process design, framework, virtual lab 1. Introduction The design, development and reliability of a chemical product and the process to manufacture it, need to be consistent with the end-use characteristics of the desired product. One of the common ways to match the desired product-process characteristics is through trial and error based experiments which can be expensive and time consuming. An alternative approach is the use of a systematic model-based framework according to an established work-flow in product-process design, replacing some of the time consuming and/or repetitive experimental steps. Furthermore, for many chemical products the appropriate models for product-process design need to have multiscale features as the properties defining the chemical structure and the product end-use characteristics are dependent on parameters of different size and time scales. The advantages of the use of multiscale modelling approach in this case is that in the design, development and/or manufacturing of a product-process, the knowledge of the applied phenomena can be provided at diverse degrees of abstractions and details. Some authors [1-4] have highlighted the importance of the multiscale and multidiciplinary approach in product-process design and identified design issues related to different scales of size, time and complexity. The development of a computer-aided framework for product-process design including a multiscale modelling option is very important for analysis, design, and/or identification of feasible chemical product candidates because it allows one to consider
R. Morales-Rodríguez and R. Gani
496
processing issues during the development of the product. The multiscale modelling framework should include the product and process design components, modelling tools and templates (work-flow) for guiding the user through the appropriate design steps. The integration of computational tools is also necessary to increase the application range of the computer-aided product-process framework; where the connection between computational tools could be established through well-defined COM-objects or the CAPE-OPEN standards. A novel computer-aided model-based framework for product-process design, that also includes multiscale modelling features, is presented in this paper. To develop this framework, a combination of different computational tools, modelling tools, simulation engines, molecular and mixture design software, solvent selection software, etc, are integrated within a set of established systematic work-flow and data-flows for various types of design problems. The performance of the multiscale model-based framework, the associated models and the work-flow for a specific product-process design is illustrated through a case study involving the modelling and design of a pesticide product; here, the pesticide uptake inside the leaf is designed and evaluated through the use of the framework, where a suite of models, database of properties, chemicals, and, a modelling tool are employed.
2. Multiscale Modelling Framework for chemical product-process design: Requirements. The multiscale modelling framework allows the user to cover a wide range of problems at different scales (of length and time) and disciplines of chemical engineering and science in an easy and efficient manner; achieving in this way the development of a product-process with the desired end-use characteristics. Another requirement of the modelling framework and the software architecture is a feature that provides the means for integration and merging of methods and tools from different sources. This architecture needs to accommodate models used for the prediction of the product behaviour/performance. Here, ICAS-MoT is the main modelling tool, which provides interactions with modelling engines, external software through the use of COM-objects, and also with external simulators through with the use of CAPE-OPEN standards. More details about this synergy is described by Morales-Rodríguez et al.[5] . Figure 1 illustrates the multiscale modelling framework for chemical product-process design where four main parts can be found:
problem definition, product design, product-process modelling, product-process evaluation.
Each of them have sub-steps that guide the user through a systematic work-flow and data-flow to solve specific design problems.
Multiscale Modelling Framework for Chemical Product-Process Design
497
Fig. 1. Multiscale modelling framework for chemical product-process design.
Problem definition: The multiscale modelling framework for chemical product-process design starts with the conceptual definition of the design problem, which concerns with setting the desired characteristics of the product, its properties, special qualities, ingredients to make the product, etc., that might apply for a new product or for existing products that need to be improved. Product Design: Often, information needed to perform product behaviour analysis is lacking. To overcome these gaps of information, computer-aided methods and tools are employed. That is, the use of specialized computer-aided tools such as, databases containing properties of chemicals, property prediction packages, molecular and mixture design software, solvent selection tools, etc. (all these computational tools can be found in the ICAS software developed by CAPEC at Technical University of Denmark) are employed in order to address the issue of missing information. The first sub-step involves generation of data/knowledge related to the product needs, ingredients, assumptions in the conceptual design work-flow, historical records, etc., that are needed in the subsequent steps. Another sub-step in this part consists of the selection of the materials to be added to the product taking into consideration, its application; functional property values; primary property values; main active ingredients; solvents; coatings; etc., as well as the calculation of the necessary properties (such as, diffusion coefficients, partition coefficient, surface tension, etc.) related simulation model to be used in product performance evaluation. Product-Process Modelling: Once the necessary information for evaluating the product performance through the generated models has been retrieved, a modelling tool is used to assist in the simulation and generation of alternatives and verification of the formulated properties through the ICAS-MoT (available in ICAS). ICAS-MoT is essentially a modelling tool able to generate, analyze and simulate mathematical models without extra programming efforts. In this way, the predicted product behaviour is compared against the desired targets specified at the beginning, and decided if the targets have been matched.
498
R. Morales-Rodríguez and R. Gani
Product-Process Evaluation: If the design targets are not matched, a new product design problem is started, or, a model-based analysis is made and the subsequent steps repeated to find another solution. If the targets are matched, before proceeding to manufacturing the product; an option to evaluate product-process performance is provided through the use of ICAS-MoT. Furthermore, an option to make sustainability analysis in order to evaluate the environmental and economical impact for the production and performance of the product is provided. More details of the systematic multiscale modelling framework for chemical product-process design of microcapsules for controlled release of active ingredients can be found in Morales-Rodríguez and Gani (2008) [6] .
3. Case Study for pesticide uptake prediction through the use of the multiscale modelling framework for chemical product-process design. To highlight the use of the modelling framework, a pesticide uptake example has been chosen due to its multiscale features from the modelling point of view.
Fig. 2. Multiscale description of pesticide uptake
Figure 2 shows that the modelling of the pesticide uptake can be carried out in an entire field that basically corresponds to the macroscale where the behaviour in a specific area is analyzed, and/or the impact of the pesticide to the environment is calculated; an analysis in a droplet scale is also performed, that is, the behaviour of the active ingredient inside the droplet as well as the vaporization of the liquid phase are predicted (this plays an important role in pesticide uptake); further analysis in a smaller scale (the behaviour of the pesticide uptake inside the leaf can be predicted) is performed. Here, the importance of the pesticide mass transfer phenomenon between the droplet and the leaf as well as inside the leaf is considered. A computer-aided tool called “Virtual Product-Process Design Lab” has been developed, where the multiscale modelling framework has been included, for chemical product-process design. The models at the various scales, needed for design and analysis of pesticide uptake is available in the model library. The case study involving
Multiscale Modelling Framework for Chemical Product-Process Design
499
the design of the pesticide formulation and the prediction of the pesticide uptake is highlighted for the active ingredient (AI) cyanazine, which is to be sprayed in a wheat field together with a surfactant called C13E11. Figs. 3a-3d depict the workflow for the design of cyanazine based pesticide formulation and the models needed at various scales, from the droplet scale to the overall pesticide uptake. Fig. 3a (problem definition) shows the step where the generation of information related with the product properties, ingredients, assumptions, etc. is carried out and saved in a documentation file for future examination and/or use.
Fig. 3. Design of one Pesticide using the multiscale modelling framework
Fig. 3b (product design) shows the selection of the plant where the pesticide will be applied, also the selection of the pesticide to be used in the formulated product as well as the surfactant. All the information that is collected in this part is transferred to the modelling tool. Fig. 3c (product-process modelling) shows the different mathematical models [7] that are available in the model-based library for this problem, each model takes into account different phenomena as well as different assumptions allowing the user to have a wider range of applicability in the pesticide formulation design. Note that this is one of the reasons for having a large model-based library. Once the virtual design has been done, an evaluation of the product behaviour performance is carried out; Fig. 3d (product-process evaluation) shows the results highlighting the calculated amounts of pesticide uptake by the leaf. Here, it is allowed to ask if the performance criteria have been satisfied. If “Yes” a new product alternative has been developed and verification
R. Morales-Rodríguez and R. Gani
500
by experiment can now be performed, if necessary. Otherwise, it is possible to return to the appropriate product design step and repeat with other options until the desired performance criteria are matched. The size of the mathematical model depends on the scenario being evaluated and the number of discretization points. The model consists of a set of differential and algebraic equations representing the phenomena of evaporation rate of the droplet, mass transfer of the pesticide AI and solvent through wax and cuticle layers of the leaf. It is important to note the multidisciplinary nature of the problem (different sources of information from different parts or fields of science, for instance, the information about the leaf conformation, diffusion coefficients for biological systems, weather conditions that effect in the behaviour of the plant, etc.) and the multiscale characteristics of the model.
4. Conclusion A systematic multiscale model-based framework for product-process design has been developed and its application illustrated through the design/analysis of the uptake cyanazine as the pesticide AI. The usual trial and error experimental-based approach has been replaced with a virtual product/process lab, which allows some of the time consuming and repetitive steps to be performed virtually through a model-based framework. In this way, the resources of experimental work are reserved for the final verification of the product, when a small number of candidates matching the desired end-use characteristics of the product have been identified. Finally, for the virtual product-process design lab to succeed, reliable multiscale models must be available in a model-library and used through an appropriate model-based framework, that can also help to generate models, when they are not available. The framework also contains models for design of devices for controlled release of active ingredients and for design/analysis of direct methanol fuel cells.
5. References [1] Charpentier, J.C., Comp. Chem. Eng., 57 (2002) 4667. [2] Charpentier, J.C., ESCAPE 17: Vol. CACE 24, (2007) 11. [3] Gani, R., Comp. Chem. Eng., 28, (2004) 2441. [4] Klatt, K.U. and Marquardt, W., ESCAPE 17: Vol. CACE 24, 19. [5] Morales-Rodríguez, R., Gani, R., Déchelotte, S., Vacher, A. and Badouin, O., Chem. Eng. Res. Des., 86 (2008) 823. [6] Morales-Rodríguez, R. and Gani, R. Proceedings of Chempor 2008. Braga Portugal. [7] Rassmusen, J.K., Prediction of Pesticide Uptake in Plants, Master Thesis, Department of Chemical engineering of Technical University of Denmark.