Desigining methodology for optimising the performance of underground GRP tanks

Current techniques available for GRP tank design are time consuming and use either highly analytical formulations that are based on assumptions, or rely on expensive destructive experimental tests. Because of the variety of loading conditions that these tanks are subjected to when underground, a high factor of safety is usually appied to compensate for discrepancy between theory and practice. This results in sometimes unnecessarily increasing the material cost, wright and manufacturing time. The objective of this work is to find a more deterministic design approach based on experimental and analytical methods in order to predict the properties, integrity and performance of underground GRP tanks. The design methodology is the result of correlation between computational (linear and non-linear FE analysis), analytical and experimental procedures. The methodology should lead to an empirical description for mechanical characteristics of GRP tanks. Once tested it can be proposed as a standard method that can lead to a complete solution for GRP tank design, applied to a wide range of tank sizes and laminate lay-ups and properties. This replaces the needs for the expensive field tests that are specified in the current standards. It is also expected to prevent the excessive usage of composite material (due to applying high safety factors), and at the same time, prevents the underestimation of the effect of severe loading conditions. In this research GRP material properties used for underground tanks were estimated by using experimental (ie mechanical testing of samples ytaken from GRP tanks), analytical and numerical methods. An empirical description for the determination of elastic properties of GRP material is proposed using FEA method. A series of experiments on full size tanks were designed and tested to generate repeatable data to be used for correlation with simple linear elastic analysis using FEA to establish the validity of FEA as an alternative tool. Correlation between the two types of data showed the correct modelling technique and its limitations. This investigation helped to establish the confidence level required when using a FEA method. Buckling analysis of GRP tanks both experimentally (by vacuum testing of tanks to destruction) and numerically (using geometrically non-linear FEA) was concluded and failure loads and failure modes were compared for all tests with experimental results, British (BS4995) and European (CEN) standards. Comparison has shown a good agreement with the numerical method, however, both standards results have been shown to be unreliable when used with anisotropic material. An individual investigation of the parameters that contribute to buckling failure has led to the development of an empirical design description for predicting buckling failure of GRP vesels based on a novel and numerical analysis. Finally based on the above work, a complete design methodology is proposed for the design of underground GRP tanks, and a computer program was developed to help the manufacturer to design an optimum vessel for the required application