This paper describes a model for numerically simulating themal recovery processes, The primary focus is on the simulation of in-situ combustion, but the formulation also represents /ire-and-water flooding, steam/looding, hot water flooding, steam stimulation, and spontaneous ignition as well. The simu Iator describes the flow of water, oil, and gas, and includes gravity and capillary effects. Heat tran.r/er by conduction, convection, and vaporization-conden~a~ion of both water and hydrocarbons are included. The rigorous but general nature of the simulator is obtained by employing conservation balance equations for . oxygen, inert gases, a Iigbt hydrocarbon pseudocomponent, a heavy hydrocarbon. pseudo component, water, coke, and energy. Vaporization-condensation is governed by vaporliquid equilibrium using temperature and pressure-dependent equilibrium coefficients. Four chemical reactions are -accounted for: formation of coke from the heavy hydroctirbon component and the oxidation of coke and botb heavy and light hydrocarbon components. Formulation details, numerical solution procedures, and computational results are pr~sented. The computational results include both one-and two-dimensional cross-sectional studies. The simulator represents a major improvement in the ability to simulate thermal recovery processes urder complex conditions. P G9 P Cw %? %? Sw Sw Pg, T, Y5

This paper presents the results of numerical simulation of dry, forward combustion tube experiments. The kinetic aspects of in-situ combustion processes also are discussed. The goals of the study are to investigate processes also are discussed. The goals of the study are to investigate the fuel deposition mechanism and to identify the key parameters affecting the performance of in-situ combustion processes. The thermal simulator developed at Gulf R and D Co. was used in the study. It was modified to include the capillary outlet effects for a more realistic description of the oil and water productions. The following experimental data were matched: cumulative water and oil productions, position of the combustion front as a function of time, fuel consumption, position of the combustion front as a function of time, fuel consumption, temperature as a function of time and position, and the pressure drop across the tube. History matches were performed for two crude oils with distinctly different physical properties (gravities of 26.5 and 13 API [0. 896 and 0. 979 g/cm3]). The agreements between experimental data and simulation results were excellent. Results indicate that the component equilibrium K-values and the kinetics of cracking reactions are the most important parameters affecting the fuel deposition, and that the fuel deposition mechanism, the fuel composition, and the locations and sizes of the transient zones depend on the crude oil and reservoir rock properties. Simulation results are always sensitive to the K-values of the light oil component but insensitive to the K-values of the heavy oil component. Results are sensitive to the kinetics of cracking reaction only if the cracking reaction is catalytic or the peak temperature and the fuel consumption are sufficiently high. Furthermore, the fuel available may or may not be solely in the form of coke. Our study suggests that further investigations of the catalytic effect of reservoir rocks and reaction kinetics of the cracking reaction are needed. Also, more than two crude oil components may be required to simulate the evaporation effect of crude oil accurately. Introduction In in-situ combustion processes, many physical changes as well as chemical reactions take place simultaneously or sequentially in the vicinity of the combustion front. It is generally believed that the combustion zone is preceded by a cracking or superheated steam zone, where coke is formed and preceded by a cracking or superheated steam zone, where coke is formed and deposited on the sand grains, and some lighter crude oil components evaporate and move forward with the flowing gas phase. The kinetics of combustion and cracking reactions in the combustion zone and the cracking zone has been discussed widely in the literature. The mechanisms of the physical changes and chemical reactions occurring around the combustion zone can be studied effectively through numerical simulation by using a thermal simulator. Although a number of numerical simulations of combustion tube experiments have been performed with different thermal simulators, no conclusions regarding the mechanism of fuel deposition can be drawn from these studies. The mentioned simulations either neglect the formation of coke from cracking reaction or use a high cracking rate so that no residual oil will be present in the combustion zone. The mechanism of fuel deposition is controlled by two important processes: the evaporation of crude oil components and the kinetics of the processes: the evaporation of crude oil components and the kinetics of the cracking reaction. These two processes determine how much fuel eventually will be burned and how much fuel will be in the form of coke. It has been reported, that low-temperature oxidation can have a significant effect on the fuel deposition and fuel characteristics. However, this reaction is important only when oxygen is available downstream of the combustion front. If oxygen is used completely in a combustion tube experiment, low-temperature oxidation will not play an important role in the fuel deposition mechanism. For a system with a high cracking reaction rate, it is likely that all of the crude oil in the cracking zone will be either evaporated or coked so that coke is the sole source of fuel. However, if the cracking rate is so low that only a portion of crude oil in the cracking zone is evaporated or coked, then some residual crude oil also will be burned in the combustion zone. This is supported strongly by the experimental data of Hildebrand who conducted a number of combustion tube experiments using clean, crushed Berea sandpacks with a variety of crude oils. SPEJ p. 657

Experimental phase equilibrium data are presented for three reservoir oils at conditions approximating those encountered in in-situ thermal recovery processes. The fluid systems involved consist of three major groups of components: flue gas, water, and crude oil. Data were measured at temperatures from 204.4 to 371.1°C (400 to 700°F) and pressures from 6996.0 to 20785.6 kPa (1,000 to 3,000 psia). Experimental phase equilibrium data were used to develop a correlation of binary interaction coefficients of crude-oil fractions required for the Peng-Robinson equation of state. Phase equilibrium data predicted using the Peng-Robinson equation of state, using our interaction coefficients, are compared with experimental data. Generally, the Peng-Robinson equation of state predictions were in close agreement with the experimental data. Effect of feed gas/oil ratio and water/oil ratio on the equilibrium coefficients was examined through the Peng-Robinson equation of state. A study on the feasibility of representing the crude oil by only two fractions was made also. This study includes a procedure for lumping the crude-oil fractions and examples showing the importance of mixing rules in determining the pseudo critical properties of lumped fractions. Introduction The steady growth of commercial thermal recovery processes1 has created a need for basic data on phase equilibria that involve water and hydrocarbons ranging from methane to high boiling-point fractions. The in-situ thermal recovery processes often are operated at pressures above 6800 kPa (1,000 psia) and temperatures above 200°C (400°F). Experimental data and theoretical correlations on phase equilibria approximating these systems are virtually nonexistent. Early work by White and Brown2 dealt with high boiling-point hydrocarbon phase equilibria. However, the highest pressure studied was 6894.8 kPa (1,000 psia) and the lightest component was pentane. Poettmann and Mayland,3 on the basis of an empirical correlation,4 constructed charts of equilibrium coefficients, or K values, as functions of pressure and temperature for various boiling-point fractions. But the maximum pressure studied was 6894.8 kPa (1,000 psia). Later, Hoffmann et al.5 studied phase behavior of a gas-condensate system with the highest pressure reaching 20 684.3 kPa (3,000 psia) but the highest temperature investigated was only 94.2°C (201°F). In 1963, Grayson and Streed6 reported experimental vapor/liquid equilibrium data for high-temperature and high-pressure hydrocarbon systems. They also extended the Chao-Seader correlation to cover the higher temperature ranges. However, the. major light component in Grayson and Streed's system was hydrogen. Recently, because of the increasing activity in carbon dioxide flooding processes, the phase equilibria of systems involving carbon dioxide and crude oil has received attention. Simon et al.7 studied phase behavior and other properties of carbon-dioxide/reservoir-oil systems. Shelton and Yarborough8 examined phase behavior in porous media during carbon dioxide or rich-gas flooding. No extensive data on equilibrium coefficients were reported in those papers, and the temperature ranges (out of physical reality) were below 93.5°C (200°F). None of these papers surveyed included water as a component.

Summary. This paper presents two kinetic models for representing the thermal cracking of crude oils, which incorporate the cracking rate parameters and stoichiometric coefficients to correlate experimental data. parameters and stoichiometric coefficients to correlate experimental data. The models presented show that the first-order kinetics generally accepted for pure components are unsatisfactory for multicomponent systems characterized by pseudocomponents. We conclude that three corrections to the existing first-order model are needed for modeling thermal cracking of mixtures. First, the apparent reaction order is always greater than one. Second, the reaction order is a decreasing function of temperature. Third, coke may also be formed from intermediate products. These corrections are incorporated into the models. In the first model, crude oil is split into two pseudocomponents, while in the second model, crude oil is represented by three pseudocomponents. The models can be easily extended to any number of pseudocomponents. The models can be easily extended to any number of pseudocomponents. pseudocomponents. The models successfully correlated experimental data of four systems available in the literature. Furthermore, it was confirmed that coke is not always the same source of the fuel burned in an in-situ combustion process. process. Introduction It is generally believed that in in-situ combustion processes, the combustion zone is preceded by a cracking or processes, the combustion zone is preceded by a cracking or superheated steam zone where coke is formed from the thermal cracking (pyrolysis) of crude oil. The kinetics of the cracking reaction may be a crucial process mechanism affecting the performance of combustion processes because it not only produces solid-like coke for combustion but also upgrades the remaining oil, which affects the vaporization behavior. As a result, the cracking reaction will strongly influence the total amount of fuel available in the combustion zone. The effects of cracking reactions on the fuel deposition mechanism and the fuel composition have been discussed elsewhere. The reaction mechanisms of hydrocarbon cracking are very complex. Even for a pure component, it is almost impossible to describe the mechanism precisely. Nevertheless, it is possible to use simple global rate expressions to represent the reaction rate of the reactant. For pure hydrocarbons, it is well established that the cracking reaction can be properly modeled by a first-order rate expression, although self-inhibition (decreasing first-order rate constant with increasing conversion) was generally observed. This is caused by the formation of olefins, which are known to be good inhibitors of free-radical reactions. As a general rule, the reaction rate constant for normal paraffins increases with increasing carbon number. while the activation energy decreases with increasing carbon number. Global rate expressions were also applied to the pyrolysis of gas oil and crude oils by use of first-order pyrolysis of gas oil and crude oils by use of first-order kinetics. Most of these studies lumped the multicomponent mixture into one oil component, while McNab et al. assumed that only the heavy-oil fraction contributes to the cracking reaction and arbitrarily chose 80% of the residue of the distillation of the original crude as the heavy-oil component. This two-pseudocomponent approach was adopted by Henderson and Weber. In all the above studies, only the reaction rates of the crude oil components were considered, and no attempt was made to correlate the stoichiometry of the reaction. In an attempt to match the product distribution of Athabasca bitumen pyrolysis, Hayashitani et al. constructed a number of complex kinetic models. The oil was divided into three to five components, and six to eight first-order reactions were included in each model. They found that these complex models cannot satisfactorily correlate all aspects of experimental behavior of cracking reactions. They concluded that during the course of cracking reactions, each pseudocomponent might have changed its characteristics, A comprehensive reaction scheme for catalytic cracking of gas oils was recently developed by Jacob et al., who simulated the gas-oil cracking using four pseudocomponents (heavy oil, light oil, gasoline, and gas plus coke). pseudocomponents (heavy oil, light oil, gasoline, and gas plus coke). For each of the light- and the heavy-oil components, the pseudocomponent was further split into paraffinic, pseudocomponent was further split into paraffinic, naphthenic, aromatic, and aromatic substitute groups. As a result, the reaction scheme involves 10 species (components) and 20 first-order reactions. The Arrhenius constants and the activation energies for these reactions were assumed to be universal constants (i.e., independent of gas-oil composition), This model was claimed to work well for a wide spectrum of gas oils. However, the applicability of this model to heavy oils with a wider range of composition remains to be proved. P. 54

This paper presents and examines in detail and computer core storage required. These tests indicate that the higher-order Galerkin methods require the [east amount of computer time for a given range of accuracy.