The synergetic effects of alkaline red mud (RM) and sulfate-based phosphogypsum (PG) on the undrained triaxial behavior of cement-admixed clay were explored in this study. A series of isotropically consolidated undrained triaxial tests were performed on stabilized clay with respect to different admixed RM/PG proportions. The triaxial behavior of stabilized clay is presented in terms of a stress-pore pressure-strain relationship, failure mode, undrained deformation modulus, stress path, and failure envelope. Scanning electron microscopy (SEM) tests were conducted to survey microscopic evolution. The results showed that the brittleness of the specimen intensified with a high RM content, which was manifested by a predominant postpeak strength reduction. As the PG content increased, the strain-softening behavior weakened and gradually evolved into strain-hardening. The failure mode changed from local shear failure to the single cone failure and bulging failure correspondingly. The RM played a role in increasing soil cohesion, whereas PG contributed to a larger frictional angle at the postyield stage. Microscopic observations indicated that the alkali source from RM significantly promoted pozzolanic reactions and strengthened cementation bonds, which increased the peak strength, deformation modulus, and cohesion. In addition, the sulfate in PG contributed to ettringite generation among clay particles and clusters, resulting in a more ductile behavior and a larger frictional angle due to large clusters formed.

Cement-admixed clay (CAC) is a widely-used soil stabilization technique for enhancing the strength and stiffness of soft clay. However, the stress-strain behavior of CAC is complex and nonlinear, and also depends on various factors such as mixing proportion, confining pressure, stress path, and shearing condition. In this study, we propose a novel approach for modeling the stress-strain behavior of CAC using recurrent neural networks (RNNs), which are a type of deep learning (DL) technique that can well capture the temporal dependencies and nonlinearities in sequential data. We compare three types of RNNs: traditional RNN, long short-term memory (LSTM) neural network, and gated recurrent unit (GRU) neural network, and evaluate their performance in simulating the strain- and stress-controlled triaxial test results of 25 CAC specimens with different mixing proportions and confining pressures. The results demonstrate that the LSTM model, incorporating a 2-time step backward, exhibits superior prediction accuracy and generalization capability compared to other evaluated models, achieving a mean absolute percentage error (MAPE) of 4%. This LSTM model is capable of capturing the stress-strain behavior of CACs across various loading conditions and mixing proportions within a unified framework. Therefore, we suggest that the LSTM model is a promising tool for modeling and analyzing the mechanical behavior of CAC in geotechnical engineering applications.