Abstract
The work described in this thesis was undertaken to support the development and application of Shell Global Solutions Light Touch oil and gas exploration technique, which uses inverse dispersion modelling to predict a plume’s source location from atmospheric concentration measurements. The overall aim was to develop a better understanding of dispersion issues relevant to the interpretation of measurements from fixed and mobile sensors. This translated into understanding how dispersion models assisted this activity and, where appropriate, developing modelling capability. A LightTouch held trial, conducted in the Algerian Sahara is analysed. Concentration measurements, collected using a survey aircraft, revealed a complex, non-Gaussian plume structure. It is shown by the application of the Atmospheric Dispersion Modelling System (ADMS) that the data demonstrate an episodic release in the presence of an existing methane background. An accurate reconstruction of the observed behaviour is demonstrated. The analysis of the held trial highlighted a need for better understanding of plume behaviour in the presence of a non-uniform background. A series of wind tunnel experiments was then conducted to address this topic. These adopted a common methodology, simulating neutral atmospheric boundary layer conditions in the University of Surrey’s EnFlo wind tunnel. The approach how conditions are characterised and the techniques for analysing experimental data outlined. The development of a pair of plumes with longitudinally separated sources is analysed. The data show that for small source separation relative to the downwind fetch, the plumes are indistinguishable in terms of time averaged concentration. However, the concentration fluctuations profile within the combined plume differs from that of a single plume. On the centre-line, the fluctuation levels are increased by the combined, correlated effect of the two plumes. However, at the fringes of the combined plume the fluctuations are smaller, as the effect of the two plumes results in reduced intermittency. It is also shown that at relatively large source separations the downwind plume is ‘visible’ against the ‘background’ plume for several boundary layer heights downwind. The space-time correlation of concentration fluctuations within a plume was shown to be a fundamental issue and this was investigated in some detail. The data demonstrate that for pure spatial separation, the longitudinal correlation decays on a scale of order a tenth of a boundary layer height. However, when the plume advection time is included in the analysis, significant longitudinal correlations in the concentration fluctuations are observed for separations up to half a boundary layer height downwind and lateral correlations are significant for separations of order one standard deviation of the plume width. A synthetic plume model was then developed, applying the observations made during the experimental phase of the project. The aim of the model is to simulate multiple time series in a plume, with realistic inter-relations between the concentration fluctuations at each point. The output needed to be suitable for investigating questions of plume sampling but was not intended as a faithful simulation of all details of concentration fluctuations. The model is shown to provide a reasonable representation of the observed concentration fluctuation correlations. The model is further developed to simulate integral path measurements, demonstrate their variability and, hence, assess their usefulness for identifying plumes in existing backgrounds.