The hydraulic fracturing (HF) technique has been widely used for measuring in situ stresses of intact rocks in the mining and petroleum industry, etc., due to its simple operation and no theoretical limit of testing depth. However, when there are pre-existing fractures in rock masses, the ‘hydraulic tests on pre-existing fractures’ (HTPF) is usually preferred (Cornet and Valette 1984). The HTPF is a generalization of classical hydraulic fracturing (Cornet et al. 1997), and aims to measure the normal stress supported by the pre-existing fracture plane. Different from HF, the HTPF is independent of pore pressure effects and does not require any material property determination. In order to determine a complete stress tensor, however, the HTPF requires testing more than eight fractures with at least six different orientations, due to the need of correction for uncertainties (Cornet et al. 1997). Furthermore, when test intervals are distant from one another by more than 50 m, stress gradients must be considered and additional tests are required. With the HTPF technique, the stress tensor is evaluated so as to best fit the normal stress measurements obtained for all the tested fractures. This requires a parameterization of in situ stress field and the definition of a misfit function (Haimson and Cornet 2003). Cornet and Valette (1984), Cornet (1993) and Cornet et al. (1997, 2003) made a detailed discussion of the inversion method for the regional stress field, and obtained complete stress tensors using the least-squares algorithm, the Monte-Carlo method and the genetic algorithm, respectively. Nevertheless, it should be noted that, during the search for the best solution for the stress field, the assumption of linear variations of the stresses with depth from the ground to the bottom along the borehole axis is not always appropriate, especially for cases where geological conditions are complex. As an alternative, an assumption of linear variation of the stresses with depth limited in the zone of interest may be more reasonable. In the past several decades, underground liquefied petroleum gas (LPG) storage facilities have been extensively constructed in many countries in response to the rapidly increasing demand for energy consumption and additional stockpiles. However, the commercialized underground storage of liquefied natural gas (LNG) in unlined rock caverns was progressing slowly. A major difficulty in LNG underground storage is the prevention of leakage of liquid and gas from the containment system to the rock mass caused by tensile failures due to shrinkage of the rock mass around the caverns. Park et al. (2007, 2010) developed a new concept for storing LNG in a lined rock cavern, to provide a safe and economical solution. The concept consisted of protecting the host rock against the extremely low temperature and providing a liquid-tight and gas-tight liner. They demonstrated its technical feasibility by examining the overall performance of a pilot underground liquefied nitrogen storage cavern in Daejeon, Korea. In addition to the tightness of underground caverns, the stability of caverns and the hydrogeological Y. Liu (&) H. Li J. Li B. Liu X. Xia State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, Wuhan 430071, Hubei, People’s Republic of China e-mail: yqliu@whrsm.ac.cn