A Quantitative Approach to Forecast Well Productivity While Drilling

Li, Maowen (CNOOC) | Lei, Guowen (Baker Hughes, a GE Company) | Zhang, Mingjie (CNOOC) | Coskun, Sefer B. (Baker Hughes, a GE Company) | Sy, Resksmey C. (Baker Hughes, a GE Company) | Hardikar, Nikhil P. (Baker Hughes, a GE Company)

OnePetro 

Abstract

The WEIZHOU-XYZ project is a newly developed offshore oilfield located in South China Sea. A quick and accurate productivity estimate would add valuable information to the decision-making on further field development, which is divided into three phases as per priority. A luxury dataset comprising seismic, drilling, well logs, well testing and production was acquired from most wells drilled in Phase-1. The objective of this paper is to establish a fast productivity forecasting method that can be used for the newly drilled wells of Phase-2 and Phase-3 after acquiring logging-while-drilling (LWD) data only.

The well-productivity forecasting model was based on uncertainty analyses using the Monte-Carlo method. Starting with equations of the production rate and the productivity index, each parameter of the equations has been investigated based on LWD data, and with a reference to the Phase-1 wells. Two key reservoir data used for the forecasting model are LWD formation testing (formation pressure while drilling - FPWD) and LWD nuclear magnetic resonance (NMR). These two dataset are the main information collected while drilling along with LWD resistivity and gamma ray in Phase 2. The FPWD data provides mobility, thus indicating the ability of fluid flow through the permeable reservoirs. The LWD NMR data provide continuous porosity measurement. Additionally, the T2 relaxation sensitives to both the pore size distribution and fluid properties providing estimation of formation permeability and fluid viscosity. Differing from the conventional way of a constant value input to the production rate equation, the proposed method sets all productivity related parameters (permeability, thickness, formation volume factor, viscosity, drainage radius etc.) under an uncertainty distribution range. The productivity prediction model is processed and evaluated using Monte-Carlo simulations. A scenario of 10,000 runs were conducted to account for the possible production distribution. As a result, an expected value of production rate or productivity index is used for the delivery of the possible forecasting outcome.

In this study, a successful application of this forecasting model has been proven by a good match with actual results from a well test. The observed difference is less than 5% between the real production rate and the expected value from the forecasting model.

This paper shows that formation testing while drilling and NMR while drilling together provide valuable inputs for productivity forecasting. The integrated method would be very helpful and meaningful for the production evaluation and decision-making for the Phase-2 and Phase-3 development wells, which will have limited data acquisition. A Monte-Carlo simulation workflow has been proposed for quantitative prediction.