This paper presents experimental results of laboratory model tests of geosynthetic reinforced highway base layers subjected to repeated loads. The potential benefit of geocell and geogrid, when used in the base layer as a reinforcement material, was examined and assessed by performance indicators including total, permanent (plastic), and elastic (resilient) deformation, traffic benefit ratio (TBR), rut depth reduction (RDR), percentage of elastic deformation, and resilient modulus (M_R). In the light of the results of this study, geocell and geogrid-reinforced bases outperform the unreinforced base in terms of elastic and permanent deformations. Further experimental results reveal that although the thickness of the base layer was decreased to a rate of 33%, geocell and geogrid enable a considerable reduction in the rut depth at an approximate rate of 21% and 13%, respectively. Moreover, when the thickness of the base layer remains constant, about 48% to 43% reduction in the rut depth was provided by geocell and geogrid reinforcement.

Accurate determination of scour depth (d_s) around bridge piers is a major concern and an essential criterion in the safe and economical design of bridge pier foundation. The estimation of d_s by the conventional empirical methods is difficult due to the very complex mechanism of the 3D flow around the bridge piers. This paper proposes the Sequential Minimal Optimization Regression (SMOREG) approach for local pier scour depth estimation. Additionally, Gradient Boosted (GBM), K-Nearest Neighbors (K-NN), and Random Forest (RF) methods were developed to compare the statistical performance of SMOREG. The numerous reliable databases from the literature includes six input parameters such as pier width (b), pier length (l), skew of the pier to approach flow (θ), mean velocity (v), flow depth (y), the particle size for which 50 percent of the bed material (D₅₀), and an output parameter d_s. It revealed that the SMOREG can build a relationship between d_s and flow characteristics and provides an estimation with an R-value of 0.85 and a mean absolute error (MAE) of 0.35. The comparison between models developed in this study showed that SMOREG and RF gave higher prediction performance than GBM and K-NN with respect to synchronic evaluation between RMSE, R, and Standard Deviation. The sensitivity analysis were also performed to determine the efficiency of each input parameter in the estimation of d_s. It is found that pier width and mean velocity of the flow are the most effective parameters than the other parameters to estimate d_s. The SMOREG models for sensitivity yielded MAE values in the range of 0.34–0.39.

Accurate determination of the resilient modulus (M_R) of subbase materials and subgrade soil is a major concern and an essential criterion in the design process of the flexible pavement. The experimental determination of M_R involves a challenging process that requires ordinarily very difficult test procedures and extreme cautions and labour. For this reason, soft computing approaches and numerical simulation techniques are becoming more popular and have increasing importance. Most of the current studies cannot provide flexible usage and consistent prediction of the M_R for practical engineering. In the present study, it is intended to investigate the bagged and unbagged with Random Forest (RF) and M5P tree regression models for forecasting the M_R. On the other hand, the numerical simulation established to examine the effect of soil properties on deformation characteristics of subbase materials and subgrade soil subjected to repeated loading. A database employed for developing the models consists of a large amount of data collected from various published research. It includes routine properties of soil such as the dry unit weight (γ_d), uniformity coefficient (C_u), percent passing a No. 200 sieve (#200), unconfined compressive strength (q_u), plasticity index (PI), confining stress (σ_o), deviator stress (σ_d), degree of saturation (S_r) water content (w) and optimum water content (w_opt). The performance of models was evaluated comprehensively by some statistical criteria. The results revealed that the models are a fairly promising approach for the prediction of M_R and capable of representing the complex relationship between M_R and fundamental material properties. The statistical performance evaluations showed that the RF model significantly outperforms the M5P models in the sense of training performances and prediction accuracies. The numerical analysis showed that the mechanical parameter like elastic modulus is the dominant parameter on the behaviour of the materials subjected to repeated loading.

This paper focuses on developing the predictive models for International Roughness Index (IRI) of flexible pavement using several soft computing systems. About 564 sections with 4229 observations of the flexible pavement from the database of the long-term pavement performance (LTPP) program were used for developing and validating the prediction models. Equivalent single axle load (ESAL), structural number (SN), thickness, resilient modulus (M_R) of the base layer, age, annual truck volume trend, total annual precipitation, freezing index, and maintenance treatments were the predictive attributes used in models. Various statistical criteria were used for evaluating the performance of the developed models, comprehensively. The results showed that the models developed in the current study are fairly promising tools for the prediction of IRI and capable of representing the complex relationship between the IRI and the predictive attributes selected. The models created through the Gradient Boosting Method (GBM) and Random Forest (RF), which were markedly outperformed the results of other methods, produced a correlation coefficient (R) greater than 0.90 and mean absolute error lower than 0.2. Furthermore, the SN was found to be the most important parameter for the IRI according to the variable importance score of the best performing prediction model.