Application of general regression neural network (GRNN) for indirect measuring pressure loss of Herschel–Bulkley drilling fluids in oil drilling

• GRNN has been used to predict pressure losses of Herschel–Bulkley fluids in oil drilling industry.
• Proposed technique is both time and cost effective respect to experimental measurements.
• Proposed model has some advantages such as faster training speed and simple network architecture respect to BPNN model.
• Predictions are good and comparable to experimental measurements.

Experimental measurements of the pressure losses in a well annulus are costly and time consuming. Pressure loss calculations in annulus is generally conducted based on an extension of empirical correlations developed for Newtonian fluids and extending pipe flow correlations. However, correct estimation of pressure loss of non-Newtonian fluids in oil well drilling operations is very important for optimum design of piping system and minimizing the power consumption. In this paper, a general regression neural network (GRNN) was applied to predict the pressure loss of Herschel–Bulkley drilling fluids in concentric and eccentric annulus. Experimental data from literature were used to train the GRNN for estimating pressure losses in annulus. The predicted values using GRNN closely followed the experimental ones with an average relative absolute error less than 6.24%, and correlation coefficient (R) of 0.99 for pressure loss estimation.