Research On Optimal Control Of Polymer Flooding In High-temperature And High-salinity Reservoir

With the further development, the reservoir conditions become worse. As a result,high-temperature and high-salinity reservoirs increase. How to enhance the oil recovery ofthese reservoirs becomes the base of output stabilization of oil fields. Polymer flooding is oneof the most important enhanced oil recovery technologies for high-temperature andhigh-salinity reservoirs. However, polymer is very expensive and the injection process is longand complex. In order to make the development schemes scientifically and economically, anoptimal control method was investigated for injection schemes optimization in thisdissertation.An optimal control model of polymer flooding in high-temperature and high-salinityreservoir is established, which involves the performance index as maximum of the profitgained from oil recovery, the governing equations as the fluid flow equations of polymerflooding, and some constraints such as the maximum polymer usage, the injectionconcentration boundary and the terminal water cut limitation. The control variables are chosenas the time stage length, the injection concentration of each slug, and the terminal floodingtime. The continuous and discrete necessary conditions of this optimal control problem (OCP)are deduced by using the maximum principle. An improved method of discrete adjointequations construction is proposed which can enhance the computation efficiency.A non-uniform control vector parameterization (NCVP) approach is presented forsolving the OCP of polymer flooding in high-temperature and high-salinity reservoirs. Byintroducing a normalized time variable, the original OCP is transformed into a uniformparameterization problem, in which the time stage lengths of every slug are considered as newcontrol variables. The discrete adjoint equations are solved to obtain the gradients of theobjective and constraint functions. Then the numerical solution is obtained by sequentialquadratic programming method.To deal with the sensitive initial and local optimal value problem of NCVP, a hybridgenetic algorithm (HGA) is proposed, which makes use of the position displacement strategy of the bare bones particle swarm optimizer as a mutation operation. It can use the historyinformation of the chromosomes and share information among the population. The HGA notonly keeps the good property in global search, but also makes the local search more efficient.The computing cost is typically high. Moreover, the time stage length of every slug is aninteger variable in days for practical polymer flooding projects. To overcome the twodrawbacks, a mix-integer iterative dynamic programming algorithm is proposed, whichincorporates a time normalization strategy and a special truncation procedure. The novelalgorithm has faster convergence speed and less dependence on initial control values.Since the crude oil prices are important factors for polymer flooding implementation, thefuture trend of these prices are forecasted based on the wavelet neural network. Consideringthe two performances including profit maximization and cost minimization, a muti-objectiveoptimal control model of polymer flooding in high-temperature and high-salinity reservoir isestablished. And its solving algorithms are presented.In order to show the effectiveness of the optimal control method, the polymer floodingscheme optimization problem of the Li21block of the Shengli Oil Field is studied. For theOCP of single objective under fixed oil price, the profit, oil recovery and production arecomputed and compared. It is illustrated that the development schemes obtained by theproposed method can obviously increase the profit and oil production. For the OCP ofmuti-objective under variable oil price, the optimization results with different index weightsare given. It indicates that the optimal control method can make reasonable developmentschemes according to the variable oil price and differernt weights.