Regression to Experimental PVT Data

Abstract

A procedure is presented for regression of equation of state parameters to experimental PVT-data. The starting point is a predictive C7+ characterization based on the available analytical data. If the agreement between the experimental and calculated PVT data is unsatisfactory, the first step is to critically evaluate the analytical data. If this still does not lead to satisfactory results, an adjustment of the equation of state volume translation parameter is performed. This parameter is chosen because it influences the liquid phase densities without having any influence on the phase equilibrium results. Any additional parameter regression needed is performed by adjustment of the two most sensitive coefficients in the expressions used to determine the equation of state parameters. It is shown that the applied procedure may be used to match experimental PVT data without having a major influence on properties which may be derived from an equation of state, but for which no experimental data exist for the actual composition. Also it is shown that reasonable results are obtained for the measured PVT properties at conditions not used in the regression. Introduction The requirements to a PVT simulation program are not limited to prediction of volumetric properties, phase fractions and saturation points at reservoir conditions. PVT simulation software is also expected to be able to predict the phase behaviour at process plant and transport conditions. Not only saturation points and volumetric properties need to be calculated but also derived properties as for example enthalpies, entropies, heat capacities, Joule- Thomson coefficients and sound velocities. This has to do with the frequent use of PVT simulation packages to generate the PVT property tables needed as input to reservoir and flow simulation programs. Petroleum reservoir fluids consist of thousands of different hydrocarbon constituents. The diversity in chemical structure of the individual components increases with carbon number. It is, therefore, unpractical to analyse for all C7+ components. A standard composition analysis most often stops at either C7+, C10+ or C20+. In PVT simulations the C7+ fraction is usually represented through a number of pseudo-components. Previously the detailed composition of the plus-fraction was not given much attention when selecting the pseudo-components. Though different in their detailed approach, the formerly used characterization procedures had in common that experimental PVT data were needed to be able to assign equation of state parameters to the pseudocomponents. The applied experimental data most often originated from PVT experiments (constant mass expansion, constant volume depletion and differential liberation) carried out at reservoir temperature. One of the most extensive works on how to perform a component pseudorization without estimating the composition of the plus-fraction has been presented by Coats(1). One reason for previously making no attempt to estimate the detailed composition of the plus-fraction was lack of high quality analytical data. Composition analyses to above C7+ were rare and often the information available about the C7+ fraction was limited to its mole fraction. New analytical techniques have made it possible to develop C7+ characterization procedures based on fairly accurate estimations of the molar composition of the plus fraction.

Reservoir Oil Characterization for Compositional Simulation of Solvent Injection Processes

“…Although there is no well-defined regression scheme for step 4 due to its high flexibility in the conventional method, Figure S1 in the Supporting Information depicts the conventional regression scheme used in this research, which is based on Pedersen and Christensen 15 and Christensen. 42 Adjustments are made for T C , P C , and ω of pseudocomponents to match the P SAT at the reservoir temperature. Adjustment parameters are selected based on their sensitivities to P SAT calculation (Voulgaris et al 43 ).…”

Analysis of Caprock Tightness for CO2 Enhanced Oil Recovery and Sequestration: Case Study of a Depleted Oil and Gas Reservoir in Dolomite, Poland

“…The SRK equation of state with parameters described above is typically sufficient to predict fluid properties. However, its results can be improved by the method of EoS regression [45] to experimental data.…”

EoS Modeling for Two Major Kuwaiti Oil Reservoirs

“…All routine PVT data could be matched after marginal adjustments of the coefficients in the property correlations used to assign critical temperatures and critical pressures to the heavy end pseudo-components (Christensen, 1999). The regression algorithm applied was the Levenberg-Marquardt algorithm as proposed by Marquardt (1963).…”

“…Extensive routine PVT and EOR data has been measured for both oils and will be used to develop EoS models. Christensen (1999) has shown that routine PVT data can be matched by adjusting only the critical temperatures and critical pressures of the heavy end pseudo-components. However, miscibility in an oil reservoir undergoing miscible gas injection develops through a critical point and it is therefore a requirement for any EoS model to be used for simulating miscible gas injection processes that the critical point on the Swelling Test saturation point curve is matched accurately (Negahban et al, 2010).…”