應用遺傳基因演算法求解擾動狀態概念最佳材料參數之研究

The Optimal Material Parameters Estimation for Disturbed State Concept Model by Genetic Algorithms

林佩君(Pei - Chun)

指導教授 ： 蔡佩勳

朝陽科技大學/理工學院/營建工程系/碩士(2009年)

摘要

因Desai所提出的擾動狀態概念模式（Disturbed State Concept, DSC）可模擬土壤受剪時應變硬化與軟化的受剪行為，其材料參數共有七個，這些參數的決定一般皆由土壤三軸試驗結果來決定，然而Desai建議的材料參數決定方法對體積應變-軸向應變的模擬卻不甚理想。
本研究以實數型基因演算法來決定「擾動狀態概念模式」的材料參數，並進行擾動狀態概念材料參數的敏感度分析。因FLAC分析軟體並未內建DSC模式供使用者使用，因此本研究將以Fish語言撰寫成FLAC可執行之副程式。本研究也將進行多組壓密排水三軸試驗，探討渥太華砂(ASTM C-109)在三種不同的圍壓(69kPa、207kPa、345kPa)以及不同的緊密程度(Dr=20%、90%)之受剪行為，比較Desai建議的方法所得到的DSC材料參數與本研究所採用的基因演算法得到的DSC材料參數的結果，由分析結果可發現Desai建議的材料參數在軸差應力-軸向應變曲線可得較佳的模擬，然而體積應變-軸向應變的曲線模擬結果卻不佳；而本研究所建議的基因演算法所得的最佳材料參數可較佳地模擬渥太華砂三軸試驗的試驗結果。
為了瞭解DSC材料參數之敏感度分析，分別對七個DSC材料參數進行敏感度分析。由分析結果發現，n越小軸差應力則越大；體積應變則越小。γ越小軸差應力越小；體積應變越小。β越小軸差應力則越大；體積應變則越大。a1越小軸差應力則越大；體積應變則越小。η1越小軸差應力則越大；體積應變則越小。A越小軸差應力則越大；體積應變則越小。B越小軸差應力則越小；體積應變則越小。

關鍵字

基因演算 ； 數值分析 ； 擾動狀態概念

並列摘要

The Disturbed State Concept (DSC) by Desai is a powerful fool for the simulation of the shear behaviors of strain hardening or softening of soil. There are seven material parameters in this model. Those material parameters can be decided by the results of the triaxial test of soil. However, there are some unsatisified simulation results on volumetric strain by using the past procedure by Desai for the determination of DSC material parameters.
In this study, the material parameters of Disturbed State Concept (DSC) were determined by Real-parameter Genetic Algorithms and the sensitivity analysis of material parameters of Disturbed State Concept were performed. Because the DSC model has not been built in FLAC, it is necessary to use the Fish language to compile this model into a subprogram which can be implemented by FLAC. The study implemented some Consolidated Drained Triaxial Test to understand the stress-strain behaviors of Ottawa sand (ASTM C-109) in three confining pressures (69kPa, 207kPa, 345kPa) with various relative densities (Dr＝20%、90%). The difference of predicting stress-strain behavior was discussed between the past procedure for the determination of DSC material parameter and Genetic Algorithms. From the results of this study, we can find that it would be better for the simulation of deviator stress-axial strain relationship by using the past procedure by Desai for the determination of DSC material parameter. However, the preduicting result of volumetric strain-axial strain is not satisified enough. The determination of DSC material parameters by using Genetic Algorithms in this study can obtain the less difference between the predicting result of the triaxial test and the experiment results.
In order to study the influence of DSC material parameters, the sensitivity analysis were performed. Based on the results of the sensitivity analysis, we can find that the deviator stresses increase with the decrease of the n, beta, a1, eta1, and A value. And the deviator stresses decrease with the decrease of the r, B value. In the volumetric strain part, they decrease with the decreasing n, r, a1, eta1, A and B values. But they decrease with the increasing Beta value.

並列關鍵字

Numerical Simulation ； Genetic Algorithms ； Disturbed State Concept

參考文獻

【1】 C. S. Desai., Mechanics of materials and interface the disturbed state concept, CRC Press, Boca Raton, Florida (2000).

【2】 C. S. Desai and J. Toth., “Disturbed state constitutive modeling based on stress-strain and nondestructive behavior,” International Journal Solids Structures Vol. 33, No.11, pp. 1619-1650 (1996).

【4】 A. Varadarajan, K. G. Sharma, K. Venkatachalam and A. K. Gupta., “Testing and modeling two rockfill materials,” International Journal of Geomechanics, Vol. 129, No. 3 (2003).

【5】 A. Varadarajan, K. G. Sharma, S. M. Abbas and A. K. Dhawan., “Constitutive model for rockfill materials and determination of material constants,” International Journal of Geomechanics, Vol. 6, No. 4 (2006).

【6】 C. S. Desai and J. Y. Chen., “Parameter optimization and sensitivity analysis for disturbed state constitutive model,” International Journal of Geomechanics, Vol. 6, No. 2 (2006).