The current study presents an intelligent method for calculating natural gas compressibility factor. The method requires three easily measurable properties including pressure, temperature, and speed of sound as inputs. As sound speed could be measured with ultrasonic flow meters, temperature, and pressure with appropriate sensors, the real-time natural gas compressibility factor could be calculated easily. The presented method eliminates the high cost of determining compressibility based on measuring natural gas composition. Artificial Neural Network is employed to develop the method. The artificial neural network is trained in a way to accept pressure, temperature, and speed of sound as inputs. To train an artificial neural network, the 30,000 random datasets of natural gas compositions were utilized. To check the validity and accuracy of the developed artificial neural network, four different natural gas compositions are selected and the compressibility factor are compared with the GERG-2008 equation of state (as standard and accurate method for calculating natural gas properties) results. The results reported the average absolute percent deviation is less than 0.7% for the compressibility factor calculation by utilizing the proposed method.

Keywords