No abstract

In this work, a reduced order model is constructed for a hydraulic mixing nozzle using proper orthogonal decomposition. Data from several CFD computations are collected to provide the data for input to the reduced order model and flow is projected onto the most dominant principal axes to study the flow structures. Flow structures are compared for the original and optimized shape nozzles to study the effect of changing the nozzle's shape on the flow characteristics. This study also addresses issues regarding the time required for collecting CFD data and how it has been reduced by using a modified convergence criterion. In this work, a reduced order model is constructed for a hydraulic mixing nozzle using proper orthogonal decomposition. Data from several CFD computations are collected to provide the data for input to the reduced order model and flow is projected onto the most dominant principal axes to study the flow structures. Flow structures are compared for the original and optimized shape nozzles to study the effect of changing the nozzle's shape on the flow characteristics. This study also addresses issues regarding the time required for collecting CFD data and how it has been reduced by using a modified convergence criterion. Disciplines Computer-Aided Engineering and Design Comments

No abstract

Engineering design is a complex and iterative process that involves multiple engineering teams sharing and communicating information during the design process. One aspect of engineering design involves the development of physics-based models and their analysis via numerical simulations that are computationally expensive. To overcome the time constraints due to the complexity of numerical simulations, reduced-order models (ROM) such as proper orthogonal decomposition are being increasingly used. Decreasing the simulation time, however, does not address the inefficiencies in communicating engineering models and analysis during the design process. This paper proposes developing and incorporating a ROM server into the design workflow. The ROM server stores all data associated with a given engineering model and automatically constructs a ROM every time a model is created or updated, thus maintaining a consistent version of information across multiple engineering teams. A common engineering workflow is compared with one using a ROM server. A cost of synchronization metric has been defined based on the parameters of data size, size of the engineering team and design iterations. This metric has been evaluated and compared for the cases with and without a ROM server and it was found that the cost of synchronization is lower when a ROM server is used in the design workflow. It is shown that as the team size increases, the ROM server helps with more efficient information storage and transfer. Finally, an example problem of a heat-exchanger fin shape design is used to demonstrate the ROM server framework.

While solving inverse problems stability of the solutions is an issue. Stability ensures that solutions obtained are physically possible and not just mathematically feasible. Techniques like Truncated Singular Value Decomposition (TSVD), Tikhonov’s regularization have been used to stabilize solutions. These are called regularization techniques, which involve selection of regularizing parameters. The choice of these regularizing parameters dictates the accuracy of solutions obtained. In this paper an Evolutionary Algorithm (EA) based optimization procedure is used to solve an inverse radiation problem in a rectangular enclosure. The fitness function and the mutation operator have been designed to eliminate the need for regularization. The evolutionary technique developed in this paper ensures a better search on the fitness landscape, without the additional effort of choosing regularization parameters.