On the use of prompt gamma-ray neutron activation analysis for determining phase amounts in multiphase flow

Prompt gamma-ray neutron activation analysis (PGNAA) is considered for the measurement of the in situ multiphase flow amounts of oil, gas, water and salt in a deep sea oil well. PGNAA has the advantages for this application that: (1) useful characteristic prompt gamma rays are produced by neutron interactions with almost all elements, (2) it is a rapid non-destructive measurement method, (3) a large sample volume is measured and (4) it can be used under the relatively extreme conditions present for undersea oil recovery. Feasibility calculations have been made with the previously developed Monte Carlo-library least-squares (MCLLS) measurement approach used with the specific purpose Monte Carlo code named CEARCPG that was previously developed at CEAR for PGNAA bulk material analysis. A slight modification of the MCLLS measurement approach previously developed for the nonlinear PGNAA and energy dispersive x-ray analysis (EDXRF) measurement applications is used for the present application. This modification allows the use of the very accurate forward Monte Carlo calculation of the PGNAA response and consists of using first the three components oil plus gas, water and salt as library spectra rather than the normal use of individual elemental libraries. Then the gamma-ray transmission density gauge response from the Cs-137 source is used to obtain the amount of gas. This approach allows one to determine the four parameters of primary interest directly. The arrangement considered is the use of a Cf-252 neutron source and a Cs-137 gamma-ray source with a large NaI detector placed on the opposite side of a right circular cylindrical sample holder for an assumed homogeneous mixture of oil, gas and seawater. A background that was previously obtained experimentally in bulk analysis applications was added in various amounts to the response here to make the calculations more reasonable. More experimental results for benchmarking will be taken in the future. Results indicate that this approach would be accurate and is feasible.