Cavern wall design principles

Sammanfattning: This licentiate thesis deals with the design of lined rockcaverns used for the storage of gas under high pressures (20-25MPa). This storage technique has been developed in Swedenduring the last 20 years. The concept has been named LRC (LinedRock Cavern).The goal of the research has been to develop a methodologyfor the design of the cavern wall so that it can fulfil thesafety demands put placed upon it by the society. To achievethis, an extensive knowledge about the properties of the wallmaterials, and how they interact, is required.The proposed design methodology is based on the cavern wallbuild-up principles that have emerged during the many years ofdevelopment of the LRC concept. The cavern wall consists, inshort, of (from inside outwards): a gastight steel lining, asliding layer of bitumen, a reinforced concrete lining, ashotcrete layer and finally the rock mass.In the thesis, a step-by-step approach is proposed toanalyse the deformations and strains that the cavern wall issubjected to as the rock caverns expands as a result of thepressurisation. The analysis begins with an assessment of thedeformation properties of the rock mass and proceeds with anestimation of the deformation in the most strained part of therock cavern. The analysis continues with an examination of howthe rock deformation is transmitted through the wall structure,ending in an assessment of the strain in the most strained partof the steel lining. The final step in the methodology is tocompare the calculated steel strain with the acceptable strain,derived from acceptable risk criteria or extracted fromrelevant codes. As an LRC storage is subjected to hundreds ofload cycles (with potentially high strain levels in the steellining) during its lifetime, the dimensioning load case is lowcycle fatigue.The methodology is based on several assumptions regardingthe properties of the structural parts included in the cavernwall and how they react to the pressure load, both as singleparts and in interaction with adjacent structural parts. Theassumptions regard e.g. the deformation properties of the rockmass, the behaviour of the rock mass during repeatedhigh-pressure loading, the crack distributing effect of thereinforcement and the properties of the sliding layer.Support for these assumptions has been gathered fromtheories, previous experiences and performed practical tests.The long concept development work has included numeroustechnical studies and tests, which have been used in thecreation of the design methodology. The experiences gained fromthe Pilot tests in Grängesberg have been especiallyvaluable to this end. A series of laboratory tests on themechanical properties of the sliding layer, for the load casein question, have been performed as part of this licentiatework.The proposed design methodology for the cavern wall is basedon a probabilistic approach. This approach has been chosen forseveral reasons, among others because LRC is a new technologywithout established design practice and because a probabilisticview is a good way to manage the uncertainties, which in thepresent case originates from the stochastic nature of the rockmass. The properties of the rock mass vary within the volumeand are better described by an interval or a distributioninstead of a deterministic value.The calculation tools used are in themselves rather simple.The basis is an elastoplastic analytical solution for thedeformation of the rock cavern during pressurisation. Thecalculations are performed as Monte Carlo simulations in aspreadsheet model. The choice of calculation tool was done fortwo main reasons, partly to get a lucid method where each stepin the process can be followed, partly because of limitationsin time and budget. However, a large number of FEM calculationshave been used, in addition to the observed behaviour of thePilot Plant, to verify and calibrate the model.The proposed methodology shows one practicable way ofdesigning an LRC storage. The methodology has already beenapplied in the design of the world?s first large scale LRCstorage (the LRC Demo Plant at Skallen, near Halmstad insouth-western Sweden).Areas where it is judged possible to improve or supplementthe proposed design methodology are:?A thorough evaluation of the deformationbehaviour of the Demo Plant can be used to check the validityof the assumptions made. Depending on the outcome of such acheck, this might lead to a modification of the designmethodology.?The sliding layer is of great importance for thestress and strain in the steel lining. It is urgent to continuethe development and testing of the sliding layer material andits properties.?It should be examined if, and in which case how,the use of stochastic FEM analysis, asa calculation tool,could improve the handling of uncertainties in the designmethodology.