Identification of oil-sludge interface using a numerical vibro acoustic approach

点击翻译按钮获取中文摘要

This research project in collaboration with INENCO is aiming at developing a tool to determine the sludge profile that is accumulated in large oil storage tanks. Currently, INENCO is interested in determining the sludge levels that are accumulated in large oil storage tanks. In practice, 5 to 15 feet of sludge is accumulated inside the tank where the sludge level is low near the outlet nozzle and much higher at the far end of the tank opposite the outlet nozzle. The tank height is typically around 40 to 45 feet and consists of a floating roof, which when lowered will have contact with the non-uniform distribution of the sludge inside the tank and can increase the risk of damage to the roofs which is very costly. Consequently, operation is usually kept at high crude oil levels to avoid floating roof failure. However, this will result in loss of storage capacity and thus, a big loss in production. Knowing the sludge profile levels inside the tanks will not only maximize the storage capacity inside the tanks, but also reduce the cost of production while at the same time, reducing the risk of damage to the floating roofs. The method proposed is based on vibro-acoustics, where by assuming that the wave propagation solution in a tank is unique to a specific interface location, we want to develop a tool that provides this surface location. Therefore, a shape parameterization will be utilized in order to define a unique shape with a limit number of parameters. Randomly generated shapes can be produced by selecting random values of the parameters. For each oil and sludge interface level, the geometry will be meshed and a numerical computation of the wave propagation problem modeled by the Helmholtz equation in the frequency domain for the heterogeneous media is then performed. The numerical results will yield the approximate pressure fluctuations in the domain which includes both the sludge and oil. The solution to each randomly generated shape will be simplified into a relation between the wave properties at a fixed point and the values of the shape parameters. A large set of simulation will be performed to generate a neural network that given the wave properties at a few points and outputs the parameters that define the shape.