Flow Measurement and Instrumentation 23 (2012) 56–65 Contents lists available at SciVerse ScienceDirect Flow Measurement and Instrumentation journal homepage: www.elsevier.com/locate/flowmeasinst Estimation of volume fractions and flow regime identification in multiphase flow based on gamma measurements and multivariate calibration Benjamin Kaku Arvoh a,∗ , Rainer Hoffmann b , Maths Halstensen a a Telemark University College, P.O Box 203, N-3901 Porsgrunn, Norway b Statoil ASA, Research Centre Porsgrunn, 3908 Porsgrunn, Norway article info Article history: Received 12 April 2011 Received in revised form 19 October 2011 Accepted 8 November 2011 Keywords: Gamma-ray Volume fraction Multiphase Chemometrics Flow regime identification abstract Gamma measurements combined with multivariate calibration were applied to estimate volume fractions and identify flow regimes in multiphase flow. Multiphase flow experiments were carried out with formation water, crude oil and gas from different North Sea gas fields in an industrial scale multiphase flow test facility in Porsgrunn, Norway. The experiments were carried out with a temperature of 80 ◦ C and 100 bar pressure which is comparable to field conditions. Different multiphase flow regimes (stratified- wavy, slug, dispersed and annular) and different volume fractions of oil, water and gas were investigated. A traversable dual energy gamma densitometer instrument consisting of a 30 mCi Ba 133 source and a CnZnTd detector with a sampling frequency of 7 Hz was used. 111 partial least square prediction models were calibrated based on single-phase experimental data. These models were used to predict all the volume fractions and also to identify the different flow regimes involved. The results from the flow regime identification were promising but the first results for the predictions of volume fractions were not acceptable. Principal component analysis was then applied to the calibration data and some of the calibration and test data in combination. The results from the PCA showed that there were differences between the calibration and test data. An average linear scaling technique was developed to improve the models volume fraction prediction performance. This technique was developed from half of the three-phase data sets and tested on the other half. The root mean square error of prediction (RMSEP) for the test data for gas, oil and water was 37.4%, 39.2% and 6.3% respectively before this technique was applied and 6.5%, 8.9% and 4.4% respectively after this technique was applied. Average linear scaling also improved the flow regime identification plots. Average scaling was then applied to predict the volume fractions and to identify the flow regimes of both the Gas/Oil and Gas/Water two-phase data sets. The RMSEP for gas, oil and water for Gas/Oil test data was 4.8%, 6.0% and 6.8% respectively. In the case of Gas/Water, the RMSEP for gas, oil and water were 6.2%, 9.2% and 5.8% respectively. Likewise their respective flow regimes were also easier to identify after this technique was applied. © 2011 Elsevier Ltd. All rights reserved. 1. Introduction Multiphase flow metering has been and continues to be one of the major areas of interest in oil and gas industries operating both onshore and offshore. Liquids and gases are the main components of oil and gas reservoirs and these components are transported through pipelines. Quantitative estimates of the individual components are necessary in determining whether or not it is beneficial to continue drilling. With adequate information on the volume fractions of the individual components, the separating process can be optimized. There is the need to identify ∗ Corresponding author. Tel.: +47 35 57 51 34; fax: +47 35 57 50 01. E-mail address: benjamin.k.arvoh@hit.no (B.K. Arvoh). the type of flow regime in the transportation process and also the volume fractions of the individual components. This is due to the fact that the flow regime directly affects the efficiency of the separating process whilst the volume fractions of the individual components provide indication as to whether the drilling process should be continued or stopped (i.e. directly related to the economics of the process). The cost of production in the oil industry is relatively high and thus an efficient drilling and separating process greatly determines the profit margins. Recently, there has been a higher interest in development and research in the area of non-invasive measurement principles due to the fact that it is possible to apply these techniques without any need to modify the existing process. In most production pipelines some of the properties of the fluid that are of great interest include temperature, pressure, flow pattern, flow rate, volume fractions of oil, water and gas, 0955-5986/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.flowmeasinst.2011.11.002