Integrated equations of motion for direct integration methods

Shuenn-Yih Chang 국제구조공학회 2002 Structural Engineering and Mechanics, An Int'l Jou Vol.14 No.6

A family of dissipative structure-dependent integration methods

Shuenn-Yih Chang,Tsui-Huang Wu,Ngoc-Cuong Tran 국제구조공학회 2015 Structural Engineering and Mechanics, An Int'l Jou Vol.55 No.4

A new family of structure-dependent integration methods is developed to enhance with desired numerical damping. This family method preserves the most important advantage of the structure-dependent integration method, which can integrate unconditional stability and explicit formulation together, and thus it is very computationally efficient. In addition, its numerical damping can be continuously controlled with a parameter. Consequently, it is best suited to solving an inertia-type problem, where the unimportant high frequency responses can be suppressed or even eliminated by the favorable numerical damping while the low frequency modes can be very accurately integrated.

Performances of non-dissipative structure-dependent integration methods

Shuenn-Yih Chang 국제구조공학회 2018 Structural Engineering and Mechanics, An Int'l Jou Vol.65 No.1

Three structure-dependent integration methods with no numerical dissipation have been successfully developed for time integration. Although these three integration methods generally have the same numerical properties, such as unconditional stability, second-order accuracy, explicit formulation, no overshoot and no numerical damping, there still exist some different numerical properties. It is found that TLM can only have unconditional stability for linear elastic and stiffness softening systems for zero viscous damping while for nonzero viscous damping it only has unconditional stability for linear elastic systems. Whereas, both CEM and CRM can have unconditional stability for linear elastic and stiffness softening systems for both zero and nonzero viscous damping. However, the most significantly different property among the three integration methods is a weak instability. In fact, both CRM and TLM have a weak instability, which will lead to an adverse overshoot or even a numerical instability in the high frequency responses to nonzero initial conditions. Whereas, CEM possesses no such an adverse weak instability. As a result, the performance of CEM is much better than for CRM and TLM. Notice that a weak instability property of CRM and TLM might severely limit its practical applications.

A dissipative family of eigen-based integration methods for nonlinear dynamic analysis

Shuenn-Yih Chang 국제구조공학회 2020 Structural Engineering and Mechanics, An Int'l Jou Vol.75 No.5

A novel family of controllable, dissipative structure-dependent integration methods is derived from an eigen-based theory, where the concept of the eigenmode can give a solid theoretical basis for the feasibility of this type of integration methods. In fact, the concepts of eigen-decomposition and modal superposition are involved in solving a multiple degree of freedom system. The total solution of a coupled equation of motion consists of each modal solution of the uncoupled equation of motion. Hence, an eigen-dependent integration method is proposed to solve each modal equation of motion and an approximate solution can be yielded via modal superposition with only the first few modes of interest for inertial problems. All the eigen-dependent integration methods combine to form a structure-dependent integration method. Some key assumptions and new techniques are combined to successfully develop this family of integration methods. In addition, this family of integration methods can be either explicitly or implicitly implemented. Except for stability property, both explicit and implicit implementations have almost the same numerical properties. An explicit implementation is more computationally efficient than for an implicit implementation since it can combine unconditional stability and explicit formulation simultaneously. As a result, an explicit implementation is preferred over an implicit implementation. This family of integration methods can have the same numerical properties as those of the WBZ-α method for linear elastic systems. Besides, its stability and accuracy performance for solving nonlinear systems is also almost the same as those of the WBZ-α method. It is evident from numerical experiments that an explicit implementation of this family of integration methods can save many computational efforts when compared to conventional implicit methods, such as the WBZ-α method.

Improved formulation for a structure-dependent integration method

Shuenn-Yih Chang,Tsui-Huang Wu,Ngoc-Cuong Tran 국제구조공학회 2016 Structural Engineering and Mechanics, An Int'l Jou Vol.60 No.1

Structure-dependent integration methods seem promising for structural dynamics applications since they can integrate unconditional stability and explicit formulation together, which can enable the integration methods to save many computational efforts when compared to an implicit method. A newly developed structure-dependent integration method can inherit such numerical properties. However, an unusual overshooting behavior might be experienced as it is used to compute a forced vibration response. The root cause of this inaccuracy is thoroughly explored herein. In addition, a scheme is proposed to modify this family method to overcome this unusual overshooting behavior. In fact, two improved formulations are proposed by adjusting the difference equations. As a result, it is verified that the two improved formulations of the integration methods can effectively overcome the difficulty arising from the inaccurate integration of the steady-state response of a high frequency mode.

Assessments of dissipative structure-dependent integration methods

Shuenn-Yih Chang 국제구조공학회 2017 Structural Engineering and Mechanics, An Int'l Jou Vol.62 No.2

Two Chang-α dissipative family methods and two KR-α family methods were developed for time integration recently. Although the four family methods are in the category of the dissipative structure-dependent integration methods, their performances may be drastically different due to the detrimental property of weak instability or overshoot for the two KR-α family methods. This weak instability or overshoot will result in an adverse overshooting behavior or even numerical instability. In general, the four family methods can possess very similar numerical properties, such as unconditional stability, second-order accuracy, explicit formulation and controllable numerical damping. However, the two KR-α family methods are found to possess a weak instability property or overshoot in the high frequency responses to any nonzero initial conditions and thus this property will hinder them from practical applications. Whereas, the two Chang-α dissipative family methods have no such an adverse property. As a result, the performances of the two Chang-α dissipative family methods are much better than for the two KR-α family methods. Analytical assessments of all the four family methods are conducted in this work and numerical examples are used to confirm the analytical predictions.

A stability factor for structure-dependent time integration methods

Shuenn-Yih Chang,Chiu-Li Huang 국제구조공학회 2023 Structural Engineering and Mechanics, An Int'l Jou Vol.87 No.4

Since the first family of structure-dependent methods can simultaneously integrate unconditional stability and explicit formulation in addition to second order accuracy, it is very computationally efficient for solving inertial problems except for adopting auto time-stepping techniques due to no nonlinear iterations. However, an unusual stability property is first found herein since its unconditional stability interval is drastically different for zero and nonzero damping. In fact, instability might occur for solving a damped stiffness hardening system while an accurate result can be obtained for the corresponding undamped stiffness hardening system. A technique of using a stability factor is applied to overcome this difficulty. It can be applied to magnify an unconditional stability interval. After introducing this stability factor, the formulation of this family of structuredependent methods is changed accordingly and thus its numerical properties must be re-evaluated. In summary, a large stability factor can result in a large unconditional stability interval but also lead to a large relative period error. As a consequence, a stability factor must be appropriately chosen to have a desired unconditional stability interval in addition to an acceptable period distortion.

Integrated equations of motion for direct integration methods

Chang, Shuenn-Yih Techno-Press 2002 Structural Engineering and Mechanics, An Int'l Jou Vol.13 No.5

In performing the dynamic analysis, the step size used in a step-by-step integration method might be much smaller than that required by the accuracy consideration in order to capture the rapid chances of dynamic loading or to eliminate the linearization errors. It was first found by Chen and Robinson that these difficulties might be overcome by integrating the equations of motion with respect to time once. A further study of this technique is conducted herein. This include the theoretical evaluation and comparison of the capability to capture the rapid changes of dynamic loading if using the constant average acceleration method and its integral form and the exploration of the superiority of the time integration to reduce the linearization error. In addition, its advantage in the solution of the impact problems or the wave propagation problems is also numerically demonstrated. It seems that this time integration technique can be applicable to all the currently available direct integration methods.

Error propagation effects for explicit pseudodynamic algorithms

Chang, Shuenn-Yih Techno-Press 2000 Structural Engineering and Mechanics, An Int'l Jou Vol.10 No.2

This paper discusses the error propagation characteristics of the Newmark explicit method, modified Newmark explicit method and ${\alpha}$-function dissipative explicit method in pseudodynamic tests. The Newmark explicit method is non-dissipative while the ${\alpha}$-function dissipative explicit method and the modified Newmark explicit method are dissipative and can eliminate the spurious participation of high frequency responses. In addition, error propagation analysis shows that the modified Newmark explicit method and the ${\alpha}$-function dissipative explicit method possess much better error propagation properties when compared to the Newmark explicit method. The major disadvantages of the modified Newmark explicit method are the positive lower stability limit and undesired numerical dissipation. Thus, the ${\alpha}$-function dissipative explicit method might be the most appropriate explicit pseudodynamic algorithm.

Performances of non-dissipative structure-dependent integration methods

Chang, Shuenn-Yih Techno-Press 2018 Structural Engineering and Mechanics, An Int'l Jou Vol.65 No.1

Three structure-dependent integration methods with no numerical dissipation have been successfully developed for time integration. Although these three integration methods generally have the same numerical properties, such as unconditional stability, second-order accuracy, explicit formulation, no overshoot and no numerical damping, there still exist some different numerical properties. It is found that TLM can only have unconditional stability for linear elastic and stiffness softening systems for zero viscous damping while for nonzero viscous damping it only has unconditional stability for linear elastic systems. Whereas, both CEM and CRM can have unconditional stability for linear elastic and stiffness softening systems for both zero and nonzero viscous damping. However, the most significantly different property among the three integration methods is a weak instability. In fact, both CRM and TLM have a weak instability, which will lead to an adverse overshoot or even a numerical instability in the high frequency responses to nonzero initial conditions. Whereas, CEM possesses no such an adverse weak instability. As a result, the performance of CEM is much better than for CRM and TLM. Notice that a weak instability property of CRM and TLM might severely limit its practical applications.