• 대한기계학회논문집A
• /
• 제20권4호
• /
• pp.1290-1300
• /
• 1996

The finite element method, based on rigid-plastic formulation, is widely used to simulate metal forming processes. In order to improve the computational efficiency of the rigid-plastic FEM, one-point integration is used to evaluate the stiffness matrix with four-node rectangular elements and eight-node brick elements. In order to control the hourglass modes, hourglass strain rate components were introduced and included in the effective strain rate definition, Numerical tests have shown that the proposed one-point integration scheme reduces the stiffness matrix evaluation time without deteriorating the convergence behavior of Newton-Raphson method. Simulations of a ring compression, a plane-strain closed-die forging and the three-dimensional spike forging processes were carried out by using the proposed integration method. The simulation results are compared to those obtained by applying the conventional integraiton method in terms of the solution accuracy and computational efficiency.

• Structural Engineering and Mechanics
• /
• 제21권5호
• /
• pp.539-551
• /
• 2005

Finite elements based on isoparametric formulation are known to suffer spurious stiffness properties and corresponding stress oscillations, even when care is taken to ensure that completeness and continuity requirements are enforced. This occurs frequently when the physics of the problem requires multiple strain components to be defined. This kind of error, commonly known as locking, can be circumvented by using reduced integration techniques to evaluate the element stiffness matrices instead of the full integration that is mathematically prescribed. However, the reduced integration technique itself can have a further drawback - rank deficiency, which physically implies that spurious energy modes (e.g., hourglass modes) are introduced because of reduced integration. Such instability in an existing stiffness matrix is generally detected by means of an eigenvalue test. In this paper we show that a knowledge of the dimension of the solution space spanned by the column vectors of the strain-displacement matrix can be used to identify the instabilities arising in an element due to reduced/selective integration techniques a priori, without having to complete the element stiffness matrix formulation and then test for zero eigenvalues.