Date of Award

2015-01-01

Degree Name

Master of Science

Department

Mechanical Engineering

Advisor(s)

Calvin M. Stewart

Abstract

Inspection and maintenance of industrial gas turbines (IGTs) cost millions of dollars. Growing demand of obtaining higher IGT efficiency leads to higher temperature and pressure operating conditions. Long exposure of turbine components at elevated temperature and pressure

makes creep damage critically important to consider during planning, designing and operating conditions. Effective and economic maintenance requires accurate creep deformation, damage

evolution and rupture life prediction information. Creep prediction models are used to determine the state of the turbine components and to schedule the inspection, maintenance and replacement time periods. The more accurate the prediction model, the less is the overall cost by the reduction of over maintenance, lay off time and premature replacement of components. There are many popular models available for each of these creep phenomena. Different models are developed on

different assumptions. Use of different models to predict these phenomena may lead to inconsistent creep prediction. It is very important to have one set of constitutive equations that can predict all these creep phenomena. At the same time, the set of equations has to be accurate and easy for application. In the present work an improved Sin-hyperbolic (Sinh) model is introduced. Analytical derivation, numerical analysis and FEM simulations are performed to

characterize the capacity to predict creep deformation, damage evolution and creep rupture life. Comparison with other popular models, scope of applicability, advantages over other models and

limitations are discussed in detail. Experimental data collected from literature for 304 Stainless Steel (SS) of creep deformation and creep rupture are used to compare the Sinh model with other

models at different level of temperature and stress. A process to determine the material constants of both models is clearly elucidated. The results and experimental data fitting curves produced

by these model shows that the sinh model produces a continuous damage (from zero to unity) by normalizing the experimental data. It is found that overall the new Sinh model offers more

flexibility and prediction accuracy. FEM simulations on notched specimen are conducted. Notches enable the ability to study the effect of stress concentration evolution of creep damage and rupture in components during life prediction modeling. Numerical analysis on notched specimen can provide prediction of creep damage or crack propagation of materials containing defects or initial cracks. In this work, the Sin-hyperbolic (Sinh) model is demonstrated to

significantly mitigate mesh dependency and exhibits a nonlocal damage distribution around the crack when compared to the other classical model. Prediction model equations are implemented

into FORTRAN using ANSYS Mechanical APLD (ANSYS Parametric Design Language) code to perform the analysis. USERMAT (user material routine) of ANSYS UPF (user-programmable

feature) is used to define the mechanical constitutive behavior and stress-strain relations. Finite element analysis of circular notched specimen of 304 Stainless-Steel (SS) show that the Sinh

model overcomes the mesh dependency problem, converges to a unique result upon mesh element refinement. Constant load is applied and the modulus of elasticity is degraded as crack propagates with each time steps. Elastic modulus decreases with damage increment and accelerates deformation. A series of simulation at different mesh size are used to check the mesh dependency. Results of single element FEM analysis are matched with experimental data

collected from literature to validate the model constants and iteration accuracy. Contour plots of damage evolution, and mesh dependency of crack growth rate are discussed in detail. It is demonstrated that the Sinh model offers better creep damage, and crack growth analysis and significantly improves the stress sensitivity, damage localization, and mesh-dependency of numerical results.

Language

en

Provenance

Received from ProQuest

File Size

84 pages

File Format

application/pdf

Rights Holder

Mohammad Shafinul Haque

Download

Share

COinS