Search

Sustainable building construction requirements demand an efficient utilization of industrial waste as alternatives to construction materials. In this work, the nano-silica (NS) has been used as a precursor to the compressive strength of mortar and multiple mixes of 107 were produced to study the effect of the nano-silica precursor (NSP). Advanced Machine Learning (AML) techniques have been used in this research work to predict the compressive strength of the NSP mortar using 75% to 25% ratio to train and validate the models. The NS precursor with a 17% degree of importance played a substantial influence with the cement due to its contribution to the pozzolanic reaction to produce C-S-H gel in mortar. The accuracies of the developed models were compared using Taylor charts, and also, the variance distribution for the developed models was conducted. The models’ performance indices, mean average error (MAE), mean squared error (MSE), root mean square error (RMSE), sum of squared error (SSE), and the coefficient of determination (R2), were used to decide the superior model. At the end of the exercise, it has been shown that the GP model showed a poorly performed model with outliers from the NS precursor mortar UCS data entries outside the ± 25% envelop. The parametric line fit of the GP is y= 0.973x, which produced MAE of 5.62 MPa, MSE of 46.71 MPa, RMSE of 6.83 MPa, and R2of 0.680. Also, the EPR model showed a parametric line fit of y= 0.983x, which produced MAE of 4.10 MPa, MSE of 29.67 MPa, RMSE of 5.45 MPa, and R2of 0.823, while the most superior model was produced by the ANN with a parametric line fit of 0.997x, which produced MAE of 1.47 MPa, MSE of 3.84 MPa, RMSE of 1.96 MPa, and R2of 0.980. The outperformance of the ANN over the other AI techniques is supported by previous research work even though the ANN did not produce a closed-form parametric expression that allows a manual application of the model in the design and construction of buildings with mortar under NS precursor effect. Generally, the NSP has shown a reliable potential to improve the hardened strength of mortar, which confirms its application in the built environment as a sustainable pozzolanic construction material.

Soil swelling is one of the major causes of structural failure, due to excessive moisture saturation and desaturation moisture. In situ measurement of swelling stress is somewhat impossible and requires tedious routine site observation. The use of artificial intelligence to predict the swelling stress of in situ soil is highly recommended, because of the complex behavior of soil upon moisture absorption. Because of this challenge, this study is channeled towards the prediction of swelling stress using basic geotechnical properties to save the challenges of repeating geotechnical experimental tests. In this study, the swelling stress of soils collected from 15 locations in 5 sites across South Africa has been predicted by using the artificial neural network (ANN), genetic programming (GP), and evolutionary polynomial regression (EPR)-based intelligent techniques. Multiple data were collected through laboratory experiments on the predictors: gravimetric moisture content (GMC), plasticity index (Ip), dry density (γd), free swell index (FSI), degree of saturation (S), matric suction (Ψm) and the target, and swelling stress (Psm). This predictive model was aimed at proposing models, which will help earthwork designers and constructors in South Africa overcome the rampant visit to the laboratory in search of soil data needed for geotechnical engineering designs. The soils showed their potential for swelling, which was eventually confirmed by the sensitivity analysis of the intelligent models. The performance indices of the models showed that ANN outclassed the other techniques with a performance accuracy of 93.6% at an error of 1.9%. Also, the sensitivity analysis showed that the plasticity index and matric suction were the most influential to the models. With the predicted models, future earthworks in South Africa can quickly forecast swelling stress prior to designs and construction more so in hydraulically bound environments.

The hydro-mechanical behavior of a material is an imminent criterion for assessing the performance of a liner and sub-base material. Several materials have been explored in the past including waste-derived products such as gypsum-treated fly ash, bentonite sand mixtures, and biopolymer-modified cohesive soils. The nonavailability of cohesive soils has shifted the focus on the amelioration of sandy soils to suit liner and sub-base requirements. This has prompted researchers to undertake the effect of fine contents on sands. The current study investigates the prospective of utilizing three Artificial Intelligent (AI) techniques to forecast the compressive strength (U) and coefficient of permeability (k) of the sand using the collected database. The techniques employed were Genetic Programming (GP), three models of Artificial Neural Networks (ANN), and Evolutionary Polynomial Regression (EPR). Among the three techniques, ANN exhibited a better performance in the prediction of compressive strength and permeability compared to GP and EPR with more than 77% and 97% of accuracy. Among the influencing parameters, gravel content did not pose a considerable effect on the prediction of permeability values. The research findings will aid practicing engineers to come up with better design mixes of finer fraction mixed sands for liner and sub-base material applications and monitoring and assessment of the performance of constructed liners and sub-bases for a healthy environment.

The development of high-performance concrete (HPC) has shown a revolution in the built environment in terms sustainability and performance. In this research paper, the compressive strength of HPC mixed with fly ash (FA) and slag was predicted with machine learning (ML) techniques, which are the genetic programming (GP), artificial neural network (ANN), and the evolutionary polynomial regression (EPR). 1030 data entries of the HPC with cement, water, FA, Slag, superplasticizer (PL), curing age (A), fine aggregate (FAg) and coarse aggregate (CAg) as variables were divided into 80% and 20% for training and validation, respectively. The results show that the GP with line fit of y = 0.986 × has less outliers away from the ± 25% of the best fit than the EPR, which has the highest outliers while performing with a fit line of y = 0.961x. This behavior also shows in their performance indices: R2, MAE and RMSE. However, the ANN as the best model has a fit line of y= 0.988 × with the best performance in terms of R2of 0.934, MAE of 3.14 MPa, MSE of 17.44 MPa and RMSE of 4.18 MPa. The compared accuracies agree with the indices of the model performance. In addition, the ANN has the best consistent variance between model and measured values. This also shows its superiority over the GP and the EPR. These performances shown in these models agree with previous research works, which had studied the behavior of the HPC. The authors can propose that the best model can be a decisive model in the design of the behavior of the HPC especially its compressive strength for sustainable structures.

In this research paper, the gradually reducing gradient optimized response surface methodology (GRG–RSM) has been deployed on a global collection of concrete mixes of 1133 entries to propose a smart concrete mix design chart for concrete design and construction. The structural built environment is in need of technologies that save cost and time, and the development of a smart concrete mix design chart under blast furnace slag (BFS) and fly ash (FA) precursor effect is one that meets the sustainability criterion of the net-zero and efficient construction of the twenty-first century. At the end of the exercise, it was found that the mix design chart performs optimally to predicting the compressive strength (Fc) of the industrial waste precursor sustainable concrete. Meanwhile, the application of the proposed concrete mix design chart for the prediction of the BFS and FA precursor concrete can save cost and time as sustainable mechanism through which the strength of concrete (Fc) is optimized for the purposes of structural concrete net-zero production, design and concrete infrastructural performance monitoring.

The construction of flexible pavement foundations (subgrade and subbase) and landfill liners has been the cause of serious research in recent decades. In this research work, machine learning predictions have been proposed for the evaluation of the unconfined compressive strength (UCS) of geopolymer cement (GPC) reconstituted granular sand with 149 entries revised from 283 entries due to redundancy. The relations between the measured and predicted values, impact of each input on the UCS values and the best fitting of each developed model, and the comparison of the accuracies of the developed models using Taylor charts were conducted. The results show that the outliers beyond the envelops of the ± 25% are more with the GP model with a parametric line fit of y= 0.965x, R2of 0.925, MAE of 1.15 MPa, RMSE of 1.51 MPa and MSE of 2.27 MPa, while the EPR has a parametric line fit of y= 0.986x, R2of 0.944, MAE of 0.98 MPa, RMSE of 1.33 MPa and MSE of 1.78 MPa and lastly the decisive model, the ANN shows a parametric line fit of y= 0.998x, R2of 0.992, MAE of 0.39 MPa, RMSE of 0.52 MPa and MSE of 0.27 MPa. The impact of the Na/Al and the binder on the UCS model showed remarkable agreement with the models especially the ANN and GP models depicted on the impact curves. The Taylor chart agrees with the error indices and performance of the ANN as the decisive model even though the ANN can only be applied in an intelligent interface since it did not produce a closed-form parametric equation. the performance accuracies of the three models are compared with other previous works that have used the same global database to predict the UCS, which deployed the random forest (RF) with R2of 0.985, MGGP with R2of 0.924, MLSR with R2of 0.788, MVR with R2of 0.817 and the ANN with R2of 0.989. These previous works reported models based on the 283-entry database without recourse to the redundant entries in the database. In the present research work, the redundant entries were eliminated to improve the speed and performance of the models. Also, revising the original database deployed previously showed that the maximum UCS value is about 30 MPa, but assume that the measuring precision is about 1% (0.3 MPa), that means a measuring error more than 30% for the samples with UCS less than 1.0 MPa, and that is why these samples were excluded from the study due to redundancy effects to have total of 149 active entries. This procedure has given advantage to the present models, which perform better than the previous models. This agrees with previous research results as binder and Na/Al ratio theoretically increases the unconfined compressive strength of the granular sand beyond 200 and 150 MPa required for subgrade and sub-base in pavement foundations and landfill liner, respectively, and also improves the permeability coefficient to below 1E−9 m/s required for an efficient landfill construction and performance. Understandably, the liquid limit of granular sands plays a role in its performance in soaked conditions.

In the construction industry, concrete has shown to be among the most essential and commonly using construction materials. In the present era of global warming resulting from the over dependence on the use of conventional cements, there has been a necessity to employ green technologies especially in the production of concrete to meet the millennium developmental goals (MDGs) for a more sustainable infrastructure and safer environment. The reason has propelled an investigation into the utilization of recycled aggregate (RA) materials of different consistencies to explore the potentials of their application in concrete production that meets both the structural design standard and MDGs requirements. In this research, an extensive search was undertaken to gather multiple experimental data from research works concentrating efforts around the utilization of recycled aggregates among other elements in eco-friendly concrete production. Those from where 302 datasets were collected had proven the effectiveness of the RAs on improving the flexural strength (fck) of concrete. These data were deployed to propose an intelligent predictive model using the artificial neural network (ANN) of the 9-6-1 network. The database was trained and validated using 70–30% of the data, respectively. The proposed model performance has shown an accuracy of 98.5% with a unique close-form equation, which is rare with the use of ANN. This shows that the developed model can be applied in solving recycled aggregate concrete (RAC) problems both with software manipulation and manually (using the closed-form equation). Generally, overall shows the importance of predicting the production and performance of a greener concrete from recycled aggregates (RAs) of both fine and coarse consistency for sustainable structures.

To foster the sustainability of green construction materials utilized in transport infrastructure and generally in soil stabilization for the same purpose, there have been continued efforts towards innovative results for consistent improvement of the mechanical properties of soils. Metakaolin (MK) has been in use as a supplementary material due to its pozzolanic properties. However, it has always produced a limit beyond which there is recorded decline in its ability to cement and strengthen soils in a stabilization protocol. In this research work, a new innovative cementitious material made from 1:1 NaCl + NaOH blend activator mixed with sawdust ash called Ashcrete (A) has been introduced. It is blended with MK in the lateritic soil stabilization procedure. Preliminary results showed that the lateritic soil (LS) has weak consistency with plasticity index above 17%, maximum dry density (MDD) of 1.77 g/cm3and classified as A-7 soil on American Association State Highway and Transportation Officials (AASHTO) method. The MK and the Ashcrete (A) showed high compositions of aluminosilicates qualifying them as supplementary cements. The MK was used at the rate of 3, 6, and 9%, while the Ashcrete (A) was incorporated at the rate of 2, 4, 6, 8, and 10%. The results of the stabilization exercise showed that the California bearing ratio (CBR) and unconfined compressive strength (UCS) consistently increased with the addition of MK + A blend. This outcome was a shift from the previous work, which had used only MK and recorded 6% addition at which the MK-treated lateritic soil recorded its highest strength, and beyond this mark, there was a decline. The highest strength in this research work was recorded with the stabilization pattern of LS + 9%MK + 10A, which translates to that for a 200 g LS to be treated, 18 g of MK, and 20 g of A are needed to achieve the highest CBR and UCS recorded in this research paper. Finally, the recorded CBR (7-day soaked and unsoaked) and the UCS (7, 21, and 28 days) of the MK + A-treated LS fulfilled the requirements for the construction of a subgrade and subbase.

Toward carbon neutrality concrete technology (TCNCT) is the main topic in the net-zero construction industry, and designing a net-zero concrete optimized to meet the required near-zero carbon footprint specifications is one of the major goals of ecofriendly concrete technology. Several concrete mix-design methods were developed based on optimizing certain factor such as the unit weight of the concrete or the specific surface area of the aggregates. However, these available mix-design methods are limited to certain types of cement and aggregates while the construction industry today is looking for more advanced, sustainable, and ecofriendly (net-zero) materials such as blast furnace slag (BFS), fly ash (FA), and superplasticizer (SP). The aim of this study is to develop a predictive model for the compressive strength of certain net-zero concrete mixes at different ages considering the previously mentioned “towards carbon neutrality” industrial waste materials using the novel superspeed metaheuristic predictive techniques. A database of 1133 records of net-zero concrete mixes was collected from literature and studied to be used in training and testing the predictive models. The performance metrics used in this exercise are mean absolute error (MAE), mean squared error (MSE), and R2score. The Automl tools were very handy as they presented the fastest models with the best performances also. The AutoSklearn models were presented for both 5 min and 10 min execution time. At the 5-min limit execution time, a performance of R2equals 91.8%, with MAE of 3.395% and MSE of 23.224% was achieved in 262 s (4.37 min). At the 10-min set execution time, a performance of R2equals 92.2%, with MAE of 3.197% and MSE of 22.224% was achieved in 597 s (9.95 min). The AutoGluon (TabularPredictor) was also deployed to model the CCS from the global database of nine features with no set time limits. This was executed in 785 s (13.08 min) and produced a model performance with R2of 92.6%, MAE of 2.820%, and MSE of 20.918%. Though the execution time was higher than those of AutoSklearn models set for 5- and 10-min run time limits, it produced a better performance. Finally, the SVR and the RFR were deployed to produce models for the compressive strength of the hybrid concrete, and the following performances were recorded; run time of 6081 s (101.35 min), R2of 86.6%, MAE of 3.901%, and MSE of 38.158% and runtime of 6442 s (107.37 min), R2of 90.5%, MAE of 3.404%, and MSE of 27.105%, respectively. It can be learned that excessive waiting time was used by the SVR and the RFR compared to the AutoSklearn set at 5 and 10 min and the TabularPredictor, which is known as the AutoGluon. In addition to the less time advantage exhibited by the AutoML techniques, they also produced the best performance, which also outclassed previous studies that used the ANN to produce less performance models even at higher execution times. Overall, the results showed that BFS has a major impact on the overall performance of the ecofriendly concrete strength and it is a good sustainable and ecofriendly replacement for cement than FA.

Swell-shrink phenomenon experienced by foundation soils like the subgrade layer is a major cause of failures in pavement facilities. This phenomenon is fundamentally caused by rise in water table and moisture ingress from runoff, and through pavement surface cracks and lateral movement of water from location to location. Repeated laboratory trails prior to design and construction can actually be avoided by using soft computing–based predictive models to propose model expressions, which can be used during the design stage and subsequently to monitor the behavior and performance of the structure. This research utilized the intelligent abilities of genetic programming (GP), artificial neural network (ANN) and genetic algorithm (GA), and optimized polynomial linear regression (PLR) known as the evolutionary polynomial regression (EPR) to forecast the shrinkage limit of expansive soil treated with rice husk ash (RHA) and different quicklime dosage–activated rice husk ash. At the end of prediction, performance indices, i.e., R2and SSE, were used to test the accuracy of the models. It was observed that EPR outclassed ANN and GP with indices of 0.974 and 1.4%, respectively. Meanwhile, the composites of RHA showed significant improvement on the shrinkage limit of the treated soil for use as a compacted subgrade material.

In this research work, the evolutionary polynomial regression (EPR), artificial neural network (ANN), and genetic programming have been used to predict the maximum lateral displacement at pile head and maximum bending moment in flexible free head pile of a laterally loaded pile in layered soil. There have been computational solutions proposed through the application of the finite element method (FEM) which possesses non-flexible application of lateral stiffness and load functions which the application of the intelligent learning techniques tried of overcome by proposing closed-form equations for the output parameters. At the end of the model prediction exercise, performance and accuracy evaluation were conducted with the relevant indices. ANN showed the best performance in lateral displacement prediction with average error of 16% while EPR showed the best performance and accuracy among the three soft computing techniques in maximum moment prediction. Generally, the AI techniques showed superior performance when compared with the FEM and Hetenyi computational results.

The era of sustainability in the built environment is upon us and the production and usage of cement is being suppressed in line with the requirements of the United Nations Sustainable Development Goals (UNSDGs) and recently those of the conference of the parties of 27th united nations climate change (COP27). This is in agreement with the carbon-neutrality pathway for the year 2050, where greenhouses gas (GHG) emissions and carbon footprint originating from cement production and its usage in the concrete-built environment is reduced to zero. Efforts are in top gear by the concrete production expert and users to replace the use of cement with more ecofriendly cementitious materials. In this research paper, a multi-objective optimization exercise has been reported, which used machine learning (ML) techniques to predict the net-zero concrete (NZC) slump flow (SF) and slump height (SH) of fly ash (FA)- and blast furnace slag (BFS)-based concrete from a data point of 103 concrete mixes. It was reported in the literature that the increase in the addition of the binary combination of FA and BFS reduced the cement content in concrete and improved its workability and rheological behaviour. This in turn reduced the emissions of CO2. Also, the BFS was used as a replacement for the fine aggregate, which reduced the energy needed for its production and CO2 footprint associate with it. The results of the model showed that the artificial neural network (ANN) outclassed the other two techniques in forecasting SF with R2of 0.939, sum of squared error (SSE) of 441, mean squared error (MSE) of 3.92 mm, mean average error (MAE) of 1.72 mm and root mean squared error (RMSE) of 1.98 mm and SH with R2of 0.895, SSE of 3028, MSE of 26.85 mm, MAE of 4.01 mm and RMSE of 5.18 mm. With the ANN as the decisive model with good performance, it becomes imperative to use ANN to predict the design and production of the concrete mixed with FA and BFS as ecofriendly supplementary cementitious additives in terms of concrete workability.

The coupling effects of nano-geopolymer binder and preloading pressures to restrict cracks and improve the resilient modulus of an expansive subgrade were investigated in this study. A moderate alkali solution of potassium hydroxide (KOH)  with a concentration of 10 M and nano-sized fly ash (α-FA) was used to develop a nano-geopolymer binder (NGB) by synthesising the aqueous solution with 25% of α-FA. After 12 h of specimen fabrications, the geocomposites were subjected to pre-loading pressure equivalent to 10 and 20 kPa to increase the stiffness of the subgrade against cyclic cracks. The specimens were cured for 28 days under ambient conditions of 40 °C to evaluate the increased dosages of nano-geopolymer binder. Series of zero swelling tests, unconfined compressive strength (UCS) tests, split tensile strength (STS) tests, and resilient modulus tests were conducted to evaluate the synergic effects of the nano-geopolymer binder on the expansive subgrade. The cohesive zones theorem was used to analyse the subgrade through discrete fractures mechanics. The test results showed that the geocomposite specimens with 5, 15, and 20% nano-geopolymer binder recorded compressive strength of 1.4, 3.8, and 4.3 MPa, respectively. Compared to identical geocomposite specimens pre-loaded with confining pressures of 10 and 20 kPa, which rendered an increase in strength equivalent to 4.6, 7.3 and 10.12 MPa accordingly. The trend of strength development was mobilised by the pre-loading pressures. Thus, an average of 59% strength increased was obtained from the pre-loaded composites compared to the un-preloading specimens with equal percentages of the nano-geopolymer binder. The results further revealed that the un-preloading geocomposite specimens cracked at low cyclic stress, whereas the pre-loaded geocomposite specimens were fabricated with 15% and 20% nano-geopolymer, restricting the cyclic cracks at low cyclic stress. It was also observed that the cracks manifested at high cyclic stress on the specimens treated with 15% and 20% nano-geopolymer binder, which gradually led to the collapse of the geocomposite specimens irrespective of the magnitude of the pre-loading stress. The cohesive zones analysis confirmed that the coupling effects of pre-loading pressures and nano-geopolymer binder increase the traction behaviour of the treated subgrade, causing fatigue and fractures to occur at high cyclic stress. This study indicated that crack evolution was sustained at low cyclic stress at which an increase in resilient modulus was achieved due to high stiffness of the treated expansive subgrade.

There is an increasing dependence on laboratory testings in solving most of the environmental and highway subgrade failure problems. Thus, it has become necessary to forecast such problems beforehand making use of available machines and data in order to design and monitor future problems relying on predicted parameters. The present work has predicted the shear parameters of an unsaturated subgrade soil treated with hybrid cement and modified with nanotextured quarry fines using the learning algorithms of artificial neural network and group method of data handling, using multi-linear regression as a baseline regression. An A-7-6 group soil with high plasticity and undesirable for use as a subgrade foundation material was treated and multiple data were generated for several parameters, which served as independent variables in the modeling exercise. Predicted models were proposed and the models’ performance was conducted using common efficiency evaluation parameters. The performance evaluation showed that artificial neural network outclassed the other methods and with minimal error scale. However, artificial neural network and group method of data handling have proven their robustness and fitness in predicting subgrade engineering problems to be used in the design and performance monitoring of pavement infrastructures.

Nearly 51.8 million hectares of land area in India is covered with expansive soils (mainly black cotton soil (BCS)). The property of the expansive soils, in general, is that they are very hard when in dry state, but they lose all of their strength when in wet state. In light of this property, these soils pose problems worldwide and a serious challenge to overcome for the Geotechnical Engineers. One of the most important aspects for construction purposes is soil stabilization, because stabilization regime improves engineering properties of the soil like volume stability, strength and durability. Red mud is a byproduct in the process of extraction of alumina from bauxite, which has over 82% of its mass in the nano scale. The process is called Bayer’s process. It is an insoluble product and is generated after digestion with sodium hydroxide at elevated temperature and pressure. In this research work, characteristic properties of nanostructured red mud (NRM) were determined and its possible use as a supplementary cement in the expansive soil stabilization was studied. The red mud average particle size was reduced to nano-red mud (NRM) by ball-milling and passing the material through 150 nm sieve to increase its reactive surface. Basic properties like specific gravity, particle size distribution, Atterberg limits, OMC and MDD are determined in conformity with the Indian Standard Codes, British Standard and American Standards and test results are discussed in geotechnical point of view. Furthermore, intelligent models were predicted using the genetic programming, artificial neural network and evolutionary polynomial regression techniques. Generally, the results showed that NRM has the potential to improve the hydro-mechanical properties of the ameliorated BCS especially CBR and UCS to be used a pavement compacted subgrade material and in landfill liner construction with strength improvements exceeding those given by design standards. The intelligent models performed optimally with ANN with R2of 88% outperforming GP and EPR in the prediction of UCS and GP with R2of 98% outperforming ANN and EPR in the CBR prediction. But GP generated a closed form equation that allows its flexibility to be applied manually as well as intelligently and with a minimal speed of 3 h instead 12 h average speed of machine learning techniques.

Laboratory determination of subgrade soil stiffness (T) and modulus of resilient deformation (D) of subgrade lateritic material is a cumbersome exercise that requires sophisticated and expensive triaxial experimental set up. This, in most cases is lacking in most laboratories. The same condition is experienced during the experimentation of suction and capillary rise for the design of hydraulically bound pavement layers; the subgrade or unsaturated soils. To overcome the above costs, time and challenges, intelligent techniques; genetic programming (GP), artificial neural network (ANN) and genetic algorithm (GA) optimized polynomial linear regression (PLR) known as the evolutionary polynomial regression (EPR), have been adopted in this work to propose models for Tand Dunder cemented and uncemented conditions of the lateritic subgrade soil treated with geopolymer cement (G). The parameters considered in this predictive model research work were capillary rise (C), suction (S), unconfined compressive strength (U) and California bearing ratio (B) as well as the varied proportions of geopolymer cement (G). At the end of the model on G-treated uncemented and cemented soil classified as A-2-6 and highly plastic soil, performance indices (SSE and R2) were used to measure the accuracy of the models. For the uncemented case, GP and EPR showed equal accuracy with R2of 99.9% but with SSE of 2.9% and 2.6%, while ANN ended with 99.8% and 3.6% for the stiffness (T) prediction. For the resilient deformation (D) prediction, GP and ANN performed with R2of 99.3% and SSE of 6.9% and 6%, respectively, while EPR has an R2value of 99.2% and SSE of 6.7%. For the cemented case, EPR outclassed GP and ANN with R2of 99.9%; 99% and SSE 2.1%; 4.6% for the stiffness (T^) and deformation (D^) predictions, respectively. Generally, the predictive models showed consistent performance accuracy in predicting uncemented and cemented subgrade soil treated with geopolymer cement.

Artificial neural network (ANN) method has been applied in the present work to predict the California bearing ratio (CBR), unconfined compressive strength (UCS), and resistance value (R) of expansive soil treated with recycled and activated composites of rice husk ash. Pavement foundations suffer from poor design and construction, poor material handling and utilization and management lapses. The evolutions of soft computing techniques have produced various algorithms developed to overcome certain lapses in performance. Three of such algorithms from ANN are Levenberg–Muarquardt Backpropagation (LMBP), Bayesian Programming (BP), and Conjugate Gradient (CG) algorithms. In this work, the expansive soil classified as A-7-6 group soil was treated with hydrated-lime activated rice husk ash (HARHA) in varying proportions between 0.1 and 12% by weight of soil at the rate of 0.1% to produce 121 datasets. These were used to predict the behavior of the soil’s strength parameters (CBR, UCS and R) utilizing the evolutionary hybrid algorithms of ANN. The predictor parameters were HARHA, liquid limit (wL), (plastic limit (wP), plasticity index (IP), optimum moisture content (wOMC), clay activity (AC), and (maximum dry density (δmax). A multiple linear regression (MLR) was also conducted on the datasets in addition to ANN to serve as a check and linear validation mechanism. MLR and ANN methods agreed in terms of performance and fit at the end of computing and iteration. However, the response validation on the predicted models showed a good correlation above 0.9 and a great performance index. Comparatively, the LMBP algorithm yielded an accurate estimation of the results in lesser iterations than the Bayesian and the CG algorithms, while the Bayesian technique produced the best result with the required number of iterations to minimize the error. And finally, the LMBP algorithm outclassed the other two algorithms in terms of the predicted models’ accuracy.

Sugeno or Takagi–Sugeno–Kang (TSK) type fuzzy inference system ANFIS proposed by Jang and ensemble random forest (ERF) regression, an extension of bootstrap aggregation of decision trees, has been employed to forecast the triple targets of strength properties of a hydrated-lime activated rice husk ash stabilized soft clay soil. This was necessitated to deal with the incessant failure being recorded on flexible pavements around the world and the efforts being made to tackle the situation in a more smart and sustainable approach. The independent variables of this model protocol were the HARHA—hydrated-lime-activated rice husk ash, wL—liquid limit, wP—plastic limit, IP—plasticity index, wOMC—optimum moisture content, AC—clay activity, δmax—maximum dry density, while CBR—California bearing ratio, UCS28—unconfined compressive strength at 28 days curing and R—resistance value were estimated and employed as the targets (dependent variables). The natural clayey expansive soil used for this research work was investigated through preliminary experiments and classified as A-7–6 group according to AASHTO. It exhibits a very high plasticity index with high clay content, hence needed modification to be rendered as a foundation material. The soil was treated with varying percentages of HARHA, and the effect on the consistency limits, compaction, CBR, UCS, and R-value was studied. These observed values gave rise to 61 datasets. The observed datasets were deployed on the learning capacity of ANFIS and ERF regression to proposed models for the targets. The outcome of the results showed that both the models presented a close correlation between the parameters used in the model execution. Evaluation of the models was performed using a variety of statistical errors, Kendall and Spearman’s rank correlations. The results of ERF regression outclasses ANFIS model yielding a 100% coefficient of determination (R) for the triple targets. The performance evaluation and validation tests show that the coefficient of determination was more than 0.94 with minimized errors. It was concluded that ERF regression and ANFIS learning techniques are viable smart approaches to forecasting engineering problems for a more sustainable design and performance evaluation.

Soil compaction is an important exercise both in the laboratory and the field through which the mechanical properties of soils are improved through mechanical procedure. This is done using a mold and hammer of mass (m) falling from a height (H) to deliver energy (E) to the soil. But laboratory or even field time is huge that efforts are made to reduce the time spent during this procedure. One of those efforts is either to minimize time/maximize speed technique or minimize time by maximizing speed. To fully compact soil to achieve maximum density desired for safe earthworks, a number of blows are required by design standards for each kind of laboratory exercise. This takes time. Meanwhile, in this work, Euler–Lagrange-solved Brachistochrone problem has been formulated for the time (t) taken for a compaction hammer of mass (m) to fall through height (H) under gravity (g) to a mold to deliver energy of compaction to a compacted soil mass. First, the Brachistochrone path equation for time (t) was formulated and was solved to formulate the parametric equations upon which the time (t) is minimized and the parametric equations substituted into the minimization equation to evaluate (t) after applying appropriate boundary conditions for the soil compaction cases.

Support vector machine techniques (polynomial and radial basis function) and random forest regression techniques have been used to predict erodibility of an unsaturated soil collected from an erosion watershed located at Amuzukwu, Nigeria. This location has been on the spotlight due to the level destruction resulting from erosion gullies. The soil, which was found to be of “Benin formation”, has been treated with composites of waste-based binders in a cleaner and greener technology to avoid the conventional emissions emanating from the use of ordinary cement. These materials were applied in a nanotextured scale, which improved the reactive surface and nucleating ability of the supplementary cementitious additives. This work is focused on proposing more reliable intelligent models, which can be used in future designs, construction and monitoring of the performance of the facilities at the erosion site without relying on repeated laboratory experiments. The performance indices, which included r, R2, MSE, RMSE, MAE and MAPE were used to test the validity and accuracy of the models. The three models showed agreement with the performance accuracy though SVM-radial basis function (SVM-RBF) outclassed both the previously used ANN and the SVM-p and RFR with R2of 0.999 and minimal errors. The additives also proved to be good greener and cleaner cementing additives in the stabilization of soils with substantial improvements.

Gene expression programming (GEP) and multi-expression programming (MEP) have been employed to predict models for the unconfined compressive strength of soil under unsaturated (partially saturated to 60%) conditions. Due to the complexity and time consumed during the laboratory estimation of UCS for flexible pavement subgrade design, there is a great need to employ prediction models to overcome this setback. Soils are erratic and change from time to time and from season to season especially unsaturated soils. Therefore, this behavior poses a big problem to the design and performance monitoring of foundations subjected to hydraulically bound effects. The soil belonged to A-7–6 group in the AASHTO method of classification and poorly graded. It was also observed to be highly plastic with high clay content (CH). The soil was treated with hybrid cement (HC); a composite binder made from the blend of rice husk ash and hydrated lime as activator and nanostructured quarry fines (NQF); another binder sourced from pulverized quarry dust sieved with 200-nm sieves to achieve fineness. Multiple experiments were conducted to generate datasets of the selected parameters, which were the predictors of the model exercise. A baseline regression model was conducted using the MLR to ascertain the level of agreement between the target variable and its multiple predictors (HC, NQF, C, AC, CC, Cu, δmax, wmax,wL,wP,IP,Nc,∅∘, γunsat). The results of this regression showed that only 6 predictor variables (HC,AC,Cu,wP,∅,γunsat) out of the 14 variables showed close-correlation between 0.987 and 0.995 to influence the behavior of the dependent variable (target). Therefore, these were further deployed in the GEP and MEP models. The outcome of the models showed that the GEP outclassed the MEP and MLR with a coefficient of determination (R2) of 0.99 and 0.988, relative root-mean-square error (RRMSE) of 0.0728 and 0.0741 and proximal performance indicators of 0.036 and 0.037 for training and validation, respectively. Moreover, the model with a minimum O’Brien and Fleming (OBF) of 0.31 was selected, which was GEP to show that there was no over-fitting. Finally, the sensitivity analysis showed that the parameters influenced the model in this order; wP˃ ∅∘˃ HC, ˃γunsat˃ Cu˃AC. Generally, GEP proved to be the better soft computing technique in this exercise, although MEP equally performed creditable.

Unsaturated lateritic soils are complex soils to work with due to moisture effects. So, the determination of its properties requires lots of time, labor and equipment. For this reason, the application of evolutionary learning techniques has been adopted to overcome these complexities. Lateritic soil under unsaturated condition classified as poorly graded and A-7–6 group was subjected to treatment by using hybrid cement and nanostructured quarry fines in a stabilization method. The clay activity, clay content and frictional angle were determined through multiple experiments at different proportions of the additives. 121 datasets were collected through the multiple testing of treated specimens and 70% and 30% of the datasets were used in the model training and testing, respectively to predict the coefficients of curvature and uniformity (Cc and Cu) of the unsaturated lateritic soil. Fist, the multi-linear regression (MLR) model showed that the selected input parameters correlated well with the output parameters. The model performance evaluation and validation selected indicators; R2, RMSE and MAE showed that ANFIS with 0.9999, 0.0021 and 0.0015 respectively, for the training and 0.9994, 0.0077 and 0.0059 respectively outclassed all its hybrid techniques and MLR in both training and testing. However, ANFIS-PSO with performance indicators 0.9996, 0.0062 and 0.0050 respectively (training) and 0.9989, 0.0095 and 0.0073 respectively (testing); followed by ANFIS-GA; 0.9991, 0.0094, and 0.0065 respectively (training) and 0.0089, 0.0099, and 0.0079 (testing) outclassed the other learning techniques for the Cc prediction model while ANFIS-GA; 0.9949, 0.1000, and 0.0798 respectively (training) and 0.9954, 0.0983, and 0.0807 respectively, followed by ANFIS-PSO; 0.9893, 0.1347, and 0.1011 respectively (training) and 0.9951, 0.1127, and 0.0924 respectively outclassed the other techniques for the Cu prediction model. Finally, ANFIS and its evolutionary hybrid techniques have shown their usefulness and flexibility in predicting stabilized unsaturated soil properties for sustainable earthwork design, construction and foundation performance monitoring.

Volumetric stability and erodibility are important soil properties influenced by moisture through raindrops and eventual runoff and the rise in water tables during wet seasons. Compacted subgrade materials made of clay respond to water ingress through swelling and shrinking in turn during drying and this poses a problem for foundation structures. Supplementary cementitious materials have been used to treat soils, in a cleaner procedure to improve the mechanical properties and to overcome undesirable behavior during changes in seasons. However, design and construction of foundation structures exposed to these problems become necessary and common, which requires constant visits to the laboratory and equipment needs. In order to overcome this, machine learning-based predictive models have been proposed in this work for the estimation of durability (Sv) via loss of strength on immersion technique and erodibility (Er) of agro-based ashes. Genetic programming (GP) (six levels of complexity), artificial neural network (ANN) (sigmoid activation function), evolutionary polynomial regression (EPR) (GA optimized PLR method) techniques have been used to conduct this intelligent prediction exercise. The performance of the models was conducted using the sum of squared errors (SSE) and coefficient of determination (R2) indices. The results show that EPR’s Er and Sv prediction with SSE of 5.1% and 2.7% respectively and R2of 97.2% and 92.9% respectively outclassed GP and ANN. However, both GP and ANN showed minimal error and acceptable R2above 0.85, which showed their ability to predict with good performance accuracy.

The morphology and mineralogy of the soil treated with rice husk ash (RHA) under different molding moisture conditions. Leachate condition in landfills built with compacted clay soil is damaging the underground water flow with the hazards released from disposed and decomposing waste materials. This makes landfills dangerous infrastructure. The leakage can be dealt with through the deployment of green materials developed from agricultural waste. One of such wastes is rice husk combusted to derive ash. The test soil used in this exercise has been classified as highly plastic and poorly graded. The treated soil was examined by scanning electron microscopy and x-ray diffractometer methods. From the test results, the presence of goethite alongside quartz and kaolinite were observed in the XRD (X-ray Diffraction) spectra of 6% and 10 % RHA treated soil. The Goethite possessing an inner needle-like structure with a closed packed striated structure makes the composite a promising material for constructing landfill liners. This is because the closed packed striated structure of the goethite present in the composite will slow down the vertical seepage of leachate to allow its collection and removal by the leachate collection system. The composite will form a barrier between groundwater, soil, and substrata, and waste. From the SEM (Scanning Electron Microscopy), the uniformly distributed grain boundaries and smaller grain size of the composite (lateritic soil and RHA) will serve as a barrier to the movement of contaminants and other leachates to the groundwater and thus, making the composite a viable material for landfill liner system.

The ability of the compacted soils and treated/compacted soils to withstand loads as foundation materials depends on the stability and durability of the soils. The design of such phenomena in treated soils whether as subgrade of pavements or embankments, backfills, etc., is a crucial phase of foundation constructions. Often, it is observed that soil mechanical and structural properties fall below the minimum design and construction requirements and this necessitates the stabilization in order to improve the needed properties. It can be observed that for this reason, there is a steady use of the laboratory and equipment prior to any design and construction as the case may be. In this work, genetic programming (GP) has been employed to predict the unconfined compressive strength of unsaturated lateritic soil treated with a hybridized binder material called hybrid cement (HC), which was formulated by blending nanotextured quarry fines (NQF) and hydrated lime activated nanotextured rice husk ash. Tests were conducted to generate multiple values for output and inputs parameters, and the values were deployed into soft computing technique to forecast UCS adopting three (3) different performance complexities (2, 3 and 4 levels of complexity). The results of the prediction models show that the four (4) levels of complexity GP model outclassed the others in performance and accuracy with a total error (SSE) of 2.4% and coefficient of determination (R2) of 0.991. Generally, GP has shown its robustness and flexibility in predicting engineering problems for use in design and performance evaluation.

The use and its advantage in overcoming time and equipment needs of an evolutionary prediction technique known as the genetic programming have been studied using unsaturated sample of soft soil treated with multiple binders. The soil classified as weak and highly plastic was stabilized and multiple experiments were conducted to measure the effect of the dosages of the treatment on the selected properties. The geotechnics of the exercise showed that the studied parameters substantially improved with increased proportion of hybrid cement (HC) and nanostructured quarry fines (NQF). These measured selected properties were further deployed to predict the compression index of the soil. The prediction operation proposed four-model equation by the degree of importance, sensitivity and influence of the independent parameters. This shows eventually that plasticity index has the greatest sensitivity on the compression behaviour of clay soils. The performance analysis shows that the models have very low error with model trial 4 presented in Eq. 7: CCGP=IP-Hc·NQFσpart/σmaxNQFIp/wmaxLnwmax+3.0, showing the least error with more consideration for the influence of more of the selected variables. It also exhibited the highest degree of determination. Generally, GP has proven to be flexible, fast and able to predict models for engineering problems for use in design and performance study.

Sensitivity and error analyses and machine-based prediction have been conducted on the erodibility response of erodible unsaturated soil (degree of saturation 60%) treated with local cement and modified with nanostructured quarry fines. The machine-based exercise has become necessary because of the incessant washing away of soil on erosion watersheds causing devastating gullies around the developing world and the need to propose model equations to study, design and proffer future solutions to this environmental problem. Also, in order to overcome complex experimental setup needed to repeatedly study erosion problems, there is also need to forecast model equations by employing variables that can easily be determined as predictors of the model. This work was aimed at the prediction of erodibility and generating a model equation using the ANN learning technique. The erodible soil was collected and classified as poorly graded, highly plastic and as an A–7–6 group. 121 datasets were generated from multiple experiments for the input parameters and deployed in model training and testing in the ratio of 70 to 30%, respectively. The model performance was validated and error analysis was conducted using R2, MAE, MSE, RMSE and MAPE indices. The performance showed that the model has R2of more than 0.95 in both training and testing between the predicted and measured values. Also, the error indices showed significantly small values, which showed good performance. Finally, the sensitivity analysis outcome showed that the liquid limit was the most influential on the erodibility model results. Generally, ANN technique has shown to be very flexible in forecasting civil engineering problems and fundamentally in proposing model equations.