An Automatic Differentiation Based Nonlinear Reduced Order Modeling Technique for Unsteady Separated Flows

Author(s):  
Jeffrey Thomas ◽  
Earl Dowell ◽  
Kenneth Hall
1997 ◽  
Vol 119 (3) ◽  
pp. 439-447 ◽  
Author(s):  
M. P. Castanier ◽  
G. O´ttarsson ◽  
C. Pierre

The analysis of the response statistics of mistuned turbomachinery rotors requires an expensive Monte Carlo simulation approach. Simple lumped parameter models capture basic localization effects but do not represent well actual engineering structures without a difficult parameter identification. Current component mode analysis techniques generally require a minimum number of degrees of freedom which is too large for running Monte Carlo simulations at a reasonable cost. In the present work, an order reduction method is introduced which is capable of generating reasonably accurate, very low order models of tuned or mistuned bladed disks. This technique is based on component modes of vibration found from a finite element analysis of a single disk-blade sector. It is shown that the phenomenon of mode localization is well captured by the reduced order modeling technique.


Author(s):  
Bartolome´ Segui´ ◽  
Euro Casanova

This paper presents a reduced-order modeling technique, based on a component mode synthesis method specifically tailored for bladed disks, that allows the resulting low-order model to be attached to a shaft. Mistuning is included in the bladed disk model and the shaft is modeled using uniaxial finite elements according to the rotordynamic approach. The proposed formulation is applied to an example finite element model of a bladed disk, for both tuned and mistuned blades. Comparisons are made between the reduced model and the full finite element solution for free and forced responses in order to assess the methodology. The forced response amplitudes of the blades are found to vary significantly with the inclusion of a flexible shaft. This work suggest that stage independent analyses might not be adequate for predicting the global dynamic response of rotating assemblies of turbomachines.


1999 ◽  
Vol 121 (3) ◽  
pp. 515-522 ◽  
Author(s):  
R. Bladh ◽  
M. P. Castanier ◽  
C. Pierre

This paper presents important improvements and extensions to a computationally efficient reduced order modeling technique for the vibration analysis of mistuned bladed disks. In particular, this work shows how the existing modeling technique is readily extended to turbomachinery rotors with shrouded blades. The modeling technique employs a component mode synthesis approach to systematically generate a reduced order model (ROM) using component modes calculated from a finite element model (FEM) of the rotor. Based on the total number of degrees of freedom, the ROM is typically two or three orders of magnitude smaller than the FEM. This makes it feasible to predict the forced response statistics of mistuned bladed disks using Monte Carlo simulations. In this work, particular attention is devoted to the introduction of mistuning into the ROM of a shrouded assembly. Mistuning is modeled by projecting the mistuned natural frequencies of a single, cantilever blade with free shrouds onto the harmonic modes of the shrouded blade assembly. Thus, the necessary mistuning information may be measured by testing individual blades.


SPE Journal ◽  
2019 ◽  
Vol 25 (01) ◽  
pp. 326-346 ◽  
Author(s):  
Zhaoyang Larry Jin ◽  
Timur Garipov ◽  
Oleg Volkov ◽  
Louis J. Durlofsky

Summary A reduced-order-modeling (ROM) framework is developed and applied to simulate coupled flow/quasistatic-geomechanics problems. The reduced-order model is constructed using POD-TPWL, in which proper orthogonal decomposition (POD), which enables representation of the solution unknowns in a low-dimensional subspace, is combined with trajectory-piecewise linearization (TPWL), where solutions with new sets of well controls are represented by means of linearization around previously simulated (training) solutions. The overdetermined system of equations is projected into the low-dimensional subspace using a least-squares Petrov-Galerkin (LSPG) procedure, which has been shown to maintain numerical stability in POD-TPWL models. The states and derivative matrices required by POD-TPWL, generated by an extended version of the Stanford University Automatic Differentiation General Purpose Research Simulator (AD-GPRS), are provided in an offline (preprocessing or training) step. Offline computational requirements correspond to the equivalent of five to eight full-order simulations, depending on the number of training runs used. Run-time (online) speedups of O(100) or more are typically achieved for new POD-TPWL test-case simulations. The POD-TPWL model is tested extensively for a 2D coupled problem involving oil/water flow and geomechanics. It is shown that POD-TPWL provides predictions of reasonable accuracy, relative to full-order simulations, for well-rate quantities, global pressure and saturation fields, global maximum- and minimum-principal-stress fields, and the Mohr-Coulomb rock-failure criterion, for the cases considered. A systematic study of POD-TPWL error is conducted using various training procedures for different levels of perturbation between test and training cases. The use of randomness in the well-bottomhole-pressure (BHP) profiles used in training is shown to be beneficial in terms of POD-TPWL solution accuracy. The procedure is also successfully applied to a prototype 3D example case.


1994 ◽  
Author(s):  
Gisli Ottarsson ◽  
Matthew Castanier ◽  
Christophe Pierre

Sign in / Sign up

Export Citation Format

Share Document