A Study of Optimal Simulation Model on the workability of High-performance Concrete

• Main Authors: HSU, HSUN-HSIN, 徐訓新
• Other Authors: CHEN, LI
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/23691967653758139949

博士 === 中華大學 === 土木工程學系 === 104 === HPC is better than traditional concrete in terms of its strength,elastic modulus,workability,slump,volume stability,durability,abrasion resistance, water tightness,completeness,economy,and other special functions,such as low hydration heat,etc.However,most of the development of HPC nowadays focuses on mix design in strength.There are few researches on the prediction of workability such as slump and slump flow.Yet slump and slump flow would determine the workability of concrete. The less the slump or slump flow, the less easier the concrete would flow and the worse its workability is. In addition, to keep the strength, the mix design often ignores the economic benefits of the costs of materials. Therefore, this study not only enhances the expandability of GAOT with AGAOT to more accurately construct the operating modes of predicting slump and slump flow with more layers, but also adopts GANN, the combination of GA’s preferential choice scheme and NN’s learning ability. In comparison with NN, it can make more reasonable interpretation in weight calculation (without excessive learning) and accurately predict slump and slump flow. Besides, the study also predicts slump and slump flow through Multiple Regression of LR and compares the Correlation Coefficient (R) with RMSE. Four sets of data generated after the sequence of the data is arranged randomly are calculated in various methods. Then the ones with better results in each method are compared and discussed subsequently. The results of the study show that in slump and slump flow, the AGAOT and GANN, and other nonlinear methods developed in the study can predict slump and slump flow more accurately. In the capability of predicting, AGAOT is better than GANN as well as previous studies. Therefore, AGAOT can construct the mode of workability prediction for slump and slump flow.The variable combos have better physical significance than those of GANN.AGAOT not only helps the study of subsequent verification and application, but the structural variables’ ability of optimization can also explore the importance of the variable combos in the mode to slump and slump flow in subsequent studies.It can locate important indicators for the assessment of HPC’s workability (slump and slump flow, etc.).Moreover,the study assesses and compares the superiority and inferiority of the efficiency and accuracy of the results of different methods through SS, a method often applied to weather forecast. The foundation of the assessment is in accordance with RMSE.The results show that AGAOT is better than GANN and MLR because it can increase improvement rate of MLR slump by 41.87% and that of MLR slump flow by 42.12% respectively.Therefore, the formula of slump and slump flow constructed by AGAOT is indeed a workability prediction mode with high reliability and low deviation.Finally, the study assesses the extent of superiority of HPC’s variables’(including mix design such as W/C, W/B, W/S, and A/B) contribution to mode prediction by exploring the slump and slump flow constructed by AGAOT in terms of Sensitivity Analysis.According to the data and graphics,in AGAOT,the mode of workability prediction for HPC’s slump and slump flow is mainly composed of water, SP, gravel, W/B,W/S, etc. fly ash, water, and A/B. Through the amount of the variables (materials) in this mode, we can know that the performance of slump and slump flow, and the material variables including water, W/B, and A/B are the major impact factors, playing an important role in HPC’s Mix Design Proportion.Thus,this study successfully improves the weaknesses of previous GAOT and NN through AGAOT by increasing insufficient layers of algorithm for the former and providing reasonable calculation mechanism for the latter.Both assess HPC’s slump and slump flow more effectively. In the modes of workability prediction constructed through sensibility analysis, the extent of influence of each concrete material on HPC can serve as a basis for the assessment of risks in the industry.For the follow-up studies, the paper will not only research on the construction of new variable combos to establish mode of slump prediction more rapidly and accurately, but will also rule out singular values.The prediction and comparison will be made through these methods.The constructed modes will be applicable for the slump values changed in each section and enhance the prediction ability and accuracy of the mode. Moreover, we can effectively optimize the material costs of HPC by conducting experiments for verification while making HPC and by assessing the risks of mix design through the combination of theoretical techniques and practical data so as to minimalize factors of disturbance.The best mix design can be located through the control of the proportion of related materials and ingredients. In the future, the costs can be analyzed through the variables (materials) of workability modes so that the amount of materials used for HPC and the costs can meet the limitations of strength and workability.This information will serve as practical reference for the industry, ensuring HPC is an engineering material with stable constructability with economic profits. Keywords:High Performance Concrete, Slump, Slump Flow, Advanced Genetic Algorithm of Operation Tree, Genetic Algorithm Neural Network