چكيده به لاتين
Nowadays, Underground gas storage is an essential approach to control consumer market fluctuations, especially in the cold months and large population cities.
Ensuring the reservoir production in the cold seasons of the year is important and strategic, so the purpose of this study is to manage and optimize the production and injection process in the gas storage project in a real gas field. Achieving optimal reservoir variables such as injection and production rate of each well, minimum of production tubing pressure, optimal amount of cushion gas, selecting new wells for production and injection And suggesting a location for drilling new well by performing sensitivity analysis and evolutionary optimizing techniques like Genetic algorithm.
There are many uncertainties in the static properties of the reservoir model such as porosity, rock permeability and net to Gross ratio, which has always made the decision on choosing the optimal scenario by senior managers very challenging. In order to solve this problem, in this study, the "optimization under uncertainty" approach has been used.
For this purpose, after building a reservoir dynamics model like the reservoir parameters of one of the real gas storage fields in Iran, first using the sequential conditional Gaussian Simulation method, 100 Realization of each of the properties like porosity, Rock permeability and Net to gross were constructed.
Afterthat, Fuzzy C-means clustering algorithm was used, from which out of 1000 realizations, four models were selected as the final representatives of the reservoir properties.
Before performing the optimization operation, in order to find the range of changes in the input variables of the problem, sensitivity analysis was performed on these parameters. Objective function called NPV, including the cumulative gas injection in each injection cycle, the cumulative gas production and the gas condensate in each production and considering the price of gas and condensate produced in the cold season, as well as injected gas in the hot season made. Then, the mean and variance NPV functions of the selected clusters, which are inherently opposite, were selected as the two objectives of the problem optimization.
Finally, a two-objective problem with three input variables using genetic algorithm led to Pareto answers. In the final stage, after making two-dimensional maps of a combination of effective petrophysical properties in the storage process, 2 new locations for infill drilling were selected and optimal parameters were obtained with new wells according to the mentioned objective functions.
كليدواژه هاي لاتين
optimization,Genetic algorithm, Realization, Fuzzy C-means, Net present value, objective functions, Pareto front, Multi objective-optimization, Principal component analysis, Sensitivity analysis, Parallel -Computing, Objective function
