Production Allocation of Commingled Reservoir Fluids by On-Site Spectroscopic Analysis

Allocation of production fluids is a key aspect for reservoir management purposes. Many consolidated techniques exist, but they have the drawback of being expensive (multiphase flowmeters, production logging tool, spectral noise logging) or not directly portable at wellhead (geochemical production allocation). For these reasons, we developed a new rapid and accurate method employing Fourier Transform InfraRed (FTIR) spectroscopy coupled with regression methods, and successfully applied to a real case of reservoir commingled fluids. After testing different spectroscopic techniques, we realized that FTIR was the best method to perform allocation. FTIR spectra were acquired with a portable spectrometer operated in transmission mode on oils loaded in standard cells for liquids (0.1 mm optical path, KBr windows). The portable instrumentation yielded equally informative signals as the laboratory one for our needs. After suitable baseline subtraction, a machine learning workflow written in R language was applied to select the most informative spectral regions for the deconvolution of single component contribution in analysis of mixtures. Through a minimization algorithm, we are able to get the concentration of end members samples into the commingled samples. To validate our technology, we first took the end member oils (coming from two different layers of the same reservoir), we mixed them, performed the IR analysis with our portable instrument and then applied our regression modelling approach, getting results that are both accurate and precise (less than 2% of average error). Based on that, we applied our workflow directly on 9 real commingled samples coming from the same aforementioned reservoir, getting results that are in very good agreement with multi-phase flowmeters measurements. We then think that the technology is very promising and can be considered a real, low-cost and affordable opportunity among all the reservoir allocation best practices. Combination of spectroscopic portable IR hardware with regression software for the sake of allocation directly at wellhead is an innovative solution for the old problem of allocation.