Scholarlink

<p>In this research work, the influence of Crab Shell Ash (CSA) on the mechanical properties of lateritic material for sustainable flexible pavement subgrade and subbase construction has been studied and predicted by using the analysis of variance (ANOVA) and the artificial neural network (ANN). The studied lateritic soil was found to belong to the A-2-6 group according to the AASHTO classification system and high swelling behavior. The CSA was used in varying proportions to stabilize the soil. The CSA showed remarkable influence on the behavior of the mechanical properties of the reconstituted lateritic soil. The liquid limit (LL), plastic limit (PL), free swell index (FSI), optimum moisture content (OMC) and the maximum dry density (MDD) were observed to improve with increased addition of CSA. The FSI reduced consistently with the addition of the CSA between 4% and 28%. The MDD reached optimal value with the addition of 16 wt % CSA to attain a value of 1.829&#xa0;g/cm3. The CBR increase to 97% at 16 wt % of CSA and agreed with the optimal value of MDD. Also, the UCS attained its highest value of 0.932&#xa0;MPa at 16 wt % of CSA. The attained CBR and UCS values have shown that at 16 wt % addition of CSA to reconstituted lateritic soil, the application of the treated soil as subgrade and subbase material is tenable in line with appropriate design conditions. Three different ANN topologies were applied in this model as 6-1-2, 6-2-2, and 6-3-2 and these produced different degrees of error with the modeled CBR and UCS. However, 6-3-2 produced the most reliable outcome with minimal error and therefore was used to produce the final ANN model for the CBR and the UCS of the CSA reconstituted soil. The PL showed a relatively higher influence of 0.24 based on the ANN prediction even though the CSA influence of 0.14 was also better compared to the MDD parameter. From the ANN prediction, it can be observed that the lines of fit between the experimental and modeled values produced linear parametric relationship of 1.004x and 0.998x for the CBR and UCS respectively. The observed R<sup>2</sup>, mean absolute error (MAE) and root mean squared error (RMSE) for the CBR model are 0.998, 0.9%, and 1.2% and for the UCS are 0.996, 0.01&#xa0;MPa, and 0.002&#xa0;MPa, respectively. These results show that the ANN models produced a more reliable system with which the UCS and CBR of the CSA reconstituted A-2-6 soil can be modeled in the design and construction application.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predictive Analysis of Crab Shell Ash Impact on Lateritic Soil for Eco-Friendly Pavement Subgrades and Subbases

  • 期刊论文

  • 发表日期：

    2026年3月8日

查看全文
  • Shaik Subhan Alisha,
  • Dumpa Venkateswarlu,
  • Vemu Sreenivasulu

摘要

In this research work, the influence of Crab Shell Ash (CSA) on the mechanical properties of lateritic material for sustainable flexible pavement subgrade and subbase construction has been studied and predicted by using the analysis of variance (ANOVA) and the artificial neural network (ANN). The studied lateritic soil was found to belong to the A-2-6 group according to the AASHTO classification system and high swelling behavior. The CSA was used in varying proportions to stabilize the soil. The CSA showed remarkable influence on the behavior of the mechanical properties of the reconstituted lateritic soil. The liquid limit (LL), plastic limit (PL), free swell index (FSI), optimum moisture content (OMC) and the maximum dry density (MDD) were observed to improve with increased addition of CSA. The FSI reduced consistently with the addition of the CSA between 4% and 28%. The MDD reached optimal value with the addition of 16 wt % CSA to attain a value of 1.829 g/cm3. The CBR increase to 97% at 16 wt % of CSA and agreed with the optimal value of MDD. Also, the UCS attained its highest value of 0.932 MPa at 16 wt % of CSA. The attained CBR and UCS values have shown that at 16 wt % addition of CSA to reconstituted lateritic soil, the application of the treated soil as subgrade and subbase material is tenable in line with appropriate design conditions. Three different ANN topologies were applied in this model as 6-1-2, 6-2-2, and 6-3-2 and these produced different degrees of error with the modeled CBR and UCS. However, 6-3-2 produced the most reliable outcome with minimal error and therefore was used to produce the final ANN model for the CBR and the UCS of the CSA reconstituted soil. The PL showed a relatively higher influence of 0.24 based on the ANN prediction even though the CSA influence of 0.14 was also better compared to the MDD parameter. From the ANN prediction, it can be observed that the lines of fit between the experimental and modeled values produced linear parametric relationship of 1.004x and 0.998x for the CBR and UCS respectively. The observed R2, mean absolute error (MAE) and root mean squared error (RMSE) for the CBR model are 0.998, 0.9%, and 1.2% and for the UCS are 0.996, 0.01 MPa, and 0.002 MPa, respectively. These results show that the ANN models produced a more reliable system with which the UCS and CBR of the CSA reconstituted A-2-6 soil can be modeled in the design and construction application.