Search

This paper presents complete steps for the construction and solution algorithm for a projection-based reduced order model coupled with an artificial neural net model. The reduced model is constructed based on the Galerkin projection of the equations governing the physical processes on reduced subspaces obtained by the Proper Orthogonal Decomposition method. The projection is done using numerical discretisation schemes implemented in the commonly used OpenFOAM® platform. The neural net model is trained using the Nonlinear AutoRegressive eXogenous model network (NARX) architecture. The reduced order models are demonstrated for true predictions of incompressible flows at low Reynolds numbers driven by various boundary conditions. Compared with the full CFD simulations, this model shows excellent agreement while only requiring a fraction of the computational resources.

Wake-induced vibrations are considered distinctively different from well-known vortex induced vibrations where bluff body structures vibrate under vortex shedding in the wake of flows passing through them. Understanding characteristics of wake-induced vibrations plays an important role in design of marine-offshore structures such as platform columns, risers operating at high Reynolds number currents. This work presents an extensive study of wake-induced vibrations using a computational fluid dynamics approach. The numerical model is based on an incompressible Navier-Stokes flow solver with a hybrid detached eddy simulation (DES) approach for turbulence modelling of high Reynold number flows. The hybrid DES approach is first validated for wake-induced vibrations at subcritical Reynolds number ranging from 10 3 − 10 5 . Comparison between numerical results and experiment shows very good agreement in prediction of downstream cylinder responses including amplitudes, frequency and phase angle. Data from numerical simulations is then used to characterize and study mechanisms of wake-induced vibrations by looking at detailed force components and their frequencies in relation to cylinder responses. The effect of different Reynolds number flow conditions in tandem cylinder responses are also analysed in the present work. It is found that Reynolds number has strong influence in changing response amplitude and frequency of cylinders in wake-induced vibrations.

This work presents a development of computational fluid dynamic model for numerical simulations of flow induced vibrations, important physical phenomena characterizing operations of various engineering structures, especially offshore installations including risers and platforms. We propose an application of detached eddy simulation (DES) approach for simulations of incompressible Navier–Stokes flows over cylindrical structures coupled with its rigid body motions. In this approach, a hybrid turbulence model based on Reynolds average Navier–Stokes (RANS) and large eddy simulations (LES) is implemented for modelling of high Reynolds number flows over free oscillating cylinders where near wall flow features are modelled with RANS and LES is employed for resolving wake dynamics. The proposed approach is validated against several experimental set ups of fixed cylinder as well as cylinders under vortex induced vibrations. A straightforward implementation of the proposed DES approach provides more reliable and accurate numerical results than traditional URANS for high Reynolds number flows in predictions of forces acting on cylinders and their responses. Numerical experiment using the presented approach for cylinder with single degree of freedom (cross flow oscillations) and two degree of freedoms (in-plane oscillations) shows that it is able to capture VIV fundamental characteristics and predict accurately responses of cylinders at various reduced velocities and mass ratios. Numerical prediction is in good agreement with earlier experimental studies for presented benchmark test cases. It shows a great potential of the proposed approach for design and analysis of structures under flow induced vibrations.

We propose an unstructured mesh adaptation approach for unsteady high speed compressible Navier–Stokes applications involving blasts and explosions with the presence of strong shock waves propagating in three dimensional complex domains. The idea is to identify the locations of critical physics locally and then re-mesh these regions based on solution derived metrics. The approach ensures both geometry fidelity and mesh validity, especially for areas near complex geometries, a task that is always a challenge in mesh adaptation. The proposed adaptivity is applied for simulations of blast wave propagations and compared with available data in literature. The results show that the proposed method is fully robust and efficient for computational fluid dynamics (CFD) problems in complex three-dimensional domains.

This paper contributes towards a more complete approach to capture very strong shocks in various applications of high speed compressible Navier–Stokes flows including blasts and explosions using second order finite volume method on unstructured grids. The HLLC Riemann solver is employed to solve for fluxes at cell interfaces with second order approximation of local Riemann states, thus obtaining second order accuracy. In order to stabilize solutions due to high order approximation of solutions in the presence of discontinuities, several strategies are presented in this work. Slope limiters are first explored on unstructured grid to maintain monotonicity of the solution reconstruction following local extremum diminishing (LED) or total variation diminishing (TVD) criteria. The hybrid HLLC/HLLE scheme is appended to eliminate shock instabilities in very strong shock cases. To improve resolution of shocks, a local mesh adaptation scheme is used to increase mesh resolution in areas of high gradients. The scheme only regenerates mesh locally and is proven to be robust and efficient for capturing of unsteady shock propagation applications. Comparisons on the accuracy and performance of different methods on various applications are drawn to suggest a more robust and efficient method for capturing shocks on unstructured grids.

We develop a new methodology that controls for both the timing of annual earnings news ( Asquith et al., 1989 ) and the performance prior to split announcements ( Barber and Lyon, 1996 ) to evaluate the information content of stock splits. In contrast to existing evidence, we find that stock splits in aggregate are followed by positive abnormal future earnings growth, suggesting that stock splits contain information about future, rather than past, operating performance. When we use changes in breadth of institutional ownership as a new metric of information content to corroborate our findings, we find that splits with the greatest increase in breadth experience positive post-split abnormal returns and positive abnormal earnings growth. Together, our results suggest that some splits contain positive information about future performance, and that sophisticated market participants such as institutional investors are able to select these splits.