In the bustling city of Ningbo, China, a groundbreaking study led by YE Ru of the Construction Branch of Ningbo Rail Transit Group Co., Ltd. is revolutionizing how we approach metro foundation pit engineering, particularly when it comes to managing confined aquifers. The research, published in *Chengshi guidao jiaotong yanjiu* (Urban Rail Transit Research), offers a novel method for accurately determining hydrogeological parameters, a critical factor in ensuring the safety and efficiency of metro construction.
The study focuses on the dewatering process at the Sports Center Station on Ningbo Rail Transit Line 7. Dewatering, the process of removing groundwater from construction sites, is a crucial step in metro foundation pit engineering. However, accurately predicting groundwater behavior is a complex challenge, especially in confined aquifers where water is trapped between impermeable layers. “Accurate determination of hydrogeological parameters is essential for formulating effective dewatering schemes,” YE Ru explains. “This is a critical prerequisite for enhancing the safety of metro station foundation pit construction.”
To tackle this challenge, YE Ru and his team developed a three-dimensional transient groundwater seepage model using Modflow6 software, called via the Flopy module in Python. But what sets this study apart is the introduction of an LSTM (long- and short-term memory) deep learning model to build a surrogate model of confined aquifer water level variations. This surrogate model, combined with a particle swarm optimization algorithm, enabled the team to conduct an inverse analysis of the confined aquifer’s permeability and storage coefficients based on field-measured data.
The results were impressive. The inverted vertical permeability coefficient was found to be 0.76×10-5 m/s, the horizontal permeability coefficient 1.38×10-5 m/s, and the storage coefficient 6.42×10-5 m-1. When these parameters were input into the numerical seepage model, the calculated data closely matched the measured data throughout the entire process, from the initial rapid water level drop to the gradual recovery after pumping stops. “The use of deep learning-based surrogate modeling combined with optimization algorithms enables efficient and accurate inversion analysis of groundwater parameters,” YE Ru asserts.
The implications of this research are significant for the construction industry, particularly for metro foundation pit engineering. Accurate determination of hydrogeological parameters can lead to more effective dewatering schemes, enhancing safety and reducing construction time and costs. Moreover, the method proposed by YE Ru and his team could be applied to other areas of construction where groundwater management is a concern, such as in the energy sector for oil and gas extraction or geothermal energy projects.
As cities around the world continue to expand their metro systems, the need for accurate and efficient groundwater management will only grow. This research provides a promising solution, paving the way for safer, more efficient metro construction. And with the increasing availability of field-measured data and advancements in deep learning and optimization algorithms, the future of hydrogeological parameter inversion looks bright.