Underground Space Development In Singapore

Singapore has been exploring underground space development since1990 as an alternative solution to their limited land resource problem [1]. Recently, Singapore has completed the Jurong Rock Caverns (JRC) as the first commercial underground rock cavern storage facility in Southeast Asia. JRC is a subsea cavern located at a depth of 150 m below the ground [2], with storage capacity of 1.47 million m3 or 580 Olympic-sized pools worth of space [3]. The construction of deep subsea cavern is very risky and greatly affected by geological uncertainties [4].

One of the major challenge is to seal the water inflow into the cavern [5]. Water inflow can cause construction delays, enormous financial losses or even worse, unwanted fatal accidents [6]. Rock pre-grouting is the preferred method employed by engineers to reduce the water inflow problem [7]. Grouting is the method of filling gaps/ cracks using a dense fluid which generally is a mixture of water and cement to strengthen the structure [8]

.Sealing tunnels and caverns from excessive water inflow has been considered to be notoriously unpredictable and only possible to manage by experienced engineers and their knowledge intuition [9]. It is important that pre-grouting volume needed to seal the water inflow is determined correctly to prevent unwanted delays, cost overrun and accidents. For example, the cost of stopping the water inflow by post-grouting is 30-60 times higher than pre-grouting [10].

Past researches have been conducted to accurately predict the pre-grouting volume needed to seal the water inflow. However, such researches that employed empirical approach [11], numerical approach [12], and probabilistic approach [13] have failed to give a satisfactory prediction of pre-grouting volume.This project aims to give an accurate prediction of pre-grouting volume using a new approach called the Artificial Neural Network (ANN). The ANN refers to the computing systems whose main idea is borrowed from the analogy of biological neural networks.

The processing ability of the ANN comes from the interconnected assembly of nodes or simple processing elements [14]. The ANN takes arbitrary amount of inputs and is trained to come up with a prediction or an output. However, a 100% accuracy is impossible to achieve due to lack of data available. In this project we aim to achieve 80% accuracy using the input data given by the JRC project owner. This method is suited to predict the pre-grouting volume as it needs to process different parameters such as rock quality, hydraulic conductor, aperture size (input) to predict the pre-grouting volume (output). With an ANN model that can accurately predict the pre-grouting volume needed, underground construction would become more efficient, cost-effective and safer for all stakeholders.

References

1. Y. Zhou and J. Cai, “ROCK CAVERN SPACE DEVELOPMENT IN SINGAPORE,” Society for Rock Mechanics and Engineering Geology, Singapore, Nov. 2015.
2. “Jurong Rock Caverns | JTC - Creating Tomorrow's Industry Spaces,” JTC Corporation. [Online]. Available: https://www.jtc.gov.sg/industrial-land-and-space/Pages/jurong-rock-caverns.aspx. [Accessed: 22-Sep-2018]
3. “Five things to know about the Jurong Rock Caverns,” The Straits Times, 05-Jun-2018. [Online]. Available: https://www.straitstimes.com/singapore/five-things-to-know-about-the-jurong-rock-caverns. [Accessed: 22-Sep-2018].
4. Palmström, “The challenge of subsea tunnelling,” Tunnelling and Underground Space Technology, vol. 9, no. 2, pp. 145–150, 1994.
5. D. Zhang, Q. Fang, and H. Lou, “Grouting techniques for the unfavorable geological conditions of Xiangan subsea tunnel in China,” Journal of Rock Mechanics and Geotechnical Engineering, vol. 6, no. 5, pp. 438–446, 2014
6. G. Anderson and B. Roskrow, The channel tunnel story. New York: E & FN Spon, 1994 1. D. Mao, M. Lu, Z. Zhao, and M. Ng, “Effects of Water Related Factors on Pre-grouting in Hard Rock Tunnelling,” Procedia Engineering, vol. 165, pp. 300–307, 2016
7. D. A. Bruce, G. S. L. John, and A. M. C. Naudts, “Grouting materials for ground treatment: A practitioner’s guide .”
8. G. Gustafson and H. Stille, “Prediction of groutability from grout properties and hydrogeological data,” Tunnelling and Underground Space Technology, vol. 11, no. 3, pp. 325–332, 1996.
9. G. F. Knut, Pre-Excavation Grouting in Tunneling. BASF Construction Chemicals Europe Ltd.1. L. Giovanni, “international conference on grouting and grout treatment,” in grouting of rock masses, vol. 3rd.
10. O. Saeidi, H. Stille, and S. R. Torabi, “Numerical and analytical analyses of the effects of different joint and grout properties on the rock mass groutability,” Tunnelling and Underground Space Technology, vol. 38, pp. 11–25, 2013.
11. W. Zhang, A. Goh, and J. Wong, “Probabilistic assessment of stability of underground rock caverns and cavern shape optimization,” Harmonising Rock Engineering and the Environment, pp. 1843–1847, 2011.
12. S. Suwansawat and H. H. Einstein, “Artificial neural networks for predicting the maximum surface settlement caused by EPB shield tunneling,” Tunnelling and Underground Space Technology, vol. 21, no. 2, pp. 133–150, 2006.
13. Y. Lecun, L. Bottou, G. B. Orr, and K.-R. Müller, “Efficient BackProp,” Lecture Notes in Computer Science Neural Networks: Tricks of the Trade, pp. 9–50, 1998.
14. H. Lisa, B. Christian, F. Åsa, G. Gunnar, and F. Johan, “A hard rock tunnel case study: Characterization of the water-bearing fracture system for tunnel grouting,” Tunnelling and Underground Space Technology, vol. 30, pp. 132–144, 2012.