You are on page 1of 14

Chemical Engineering Science 94 (2013) 200213

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Science


journal homepage: www.elsevier.com/locate/ces

Numerical simulation of thermal and reaction fronts for oil shale upgrading
M.S.K. Youtsos n, E. Mastorakos, R.S. Cant
Hopkinson Laboratory, Engineering Department, University of Cambridge, UK

H I G H L I G H T S
c c c c c

We model shale oil extraction by in situ thermal upgrading using an in house code. Hot gas injection and conduction heating are found to be viable methods. Reaction wave progression can be tracked solely by monitoring the thermal wave. Dimensionless ow rate governs oil recovery by hot gas injection. Dimensionless depletion region length governs oil recovery by conduction heating.

a r t i c l e i n f o
Article history: Received 5 September 2012 Received in revised form 14 February 2013 Accepted 15 February 2013 Available online 27 February 2013 Keywords: Petroleum Shale Computational uid dynamics Multiphase ow Heat transfer Pyrolysis

a b s t r a c t
This paper analyses reaction and thermal front development in porous reservoirs with reacting ows, such as those encountered in oil shale upgrading. A set of dimensionless groups and a 1D code are developed in order to investigate the important physical and chemical variables of such reservoirs when heated by in situ methods. Theory necessary for this study is presented, namely shale decomposition chemical mechanisms, governing equations for multiphase ow in porous media and necessary closure models. Plotting the ratio of the thermal front speed to the uid speed allows one to infer that the reaction front ends where this ratio is at a minimum. The reaction front follows the thermal front closely, thus allowing assumptions to be made about the extent of decomposition solely by looking at thermal front progression. Furthermore, this sensitivity analysis showed that a certain minimum permeability is required in order to ensure the formation of a traveling thermal front. Compared to varying deposit porosities and kerogen activations energies, varying temperature, pressure and permeability are more important. & 2013 Elsevier Ltd. All rights reserved.

1. Introduction 1.1. Motivation Enhanced oil recovery methods and unconventional hydrocarbon sources represent the next steps in ensuring todays ever growing demands for oil and gas are met. Shale deposits around the world are estimated to contain ca. 3 trillion barrels worth of oil (Fan et al., 2009). There has been a renewed interest in shale in recent years, driven heavily by rising oil prices, resulting in the development of new in situ oil shale upgrading technologies, for example by Shell (Vinegar, 2006), ExxonMobil (Symington et al., 2010) and Chevron (Looney et al., 2010).

In situ oil shale upgrading (OSU) and heavy oil extraction are both achievable by thermal methods, which involve a mix of thermal decomposition and combustion reactions (Fan et al., 2009). These methods are analyzed in this paper by the construction of governing equations and their numerical solution and exploration. 1.2. In situ oil shale upgrading Oil Shale is a solid clay-rich sedimentary deposit containing organic matter (kerogen) that generates oil upon being heated (Fan et al., 2009). The kerogen weight content in oil shale can range from 0 to 40% (Lee, 2010). For instance, rich shale in the Green River deposit in the USA typically will give 0.00035 m3 of oil per kg of shale; whereas lean shale might be somewhere around 0.00013 m3/kg (Qian and Yin, 2010). Shale oil has been produced by mining and subsequent surface retorting methods for many years, in countries like Australia,

Corresponding author. Tel: 44 1223 332641; fax: 44 1223 3 32662. E-mail address: msy21@cam.ac.uk (M.S.K. Youtsos).

0009-2509/$ - see front matter & 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ces.2013.02.040

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

201

China, Estonia and Canada (Qian and Yin, 2010). The present contribution however focuses only on in situ heating methods. These methods include: conduction heating, hot gas injection and in situ combustion. The rst method, makes use of simple conduction heaters drilled into the reservoir and is the preferred method of Shell (Vinegar, 2006). The latter two methods involve uids being injected into the reservoir. In the case of the hot gas injection, an inert gas is injected with the intent of heating the reservoir solely by its sensible enthalpy. Chevron have outlined a method whereby CO2 is injected (Looney et al., 2010). In the case of in situ combustion, hot air is injected in order to combust the shale organic component (kerogen) and subsequently release its products (Looney et al., 2010). These methods result in the formation of thermal and reaction fronts. They are traveling structures with a characteristic set of properties, dependent on the properties of the injected uid(s) and the initial reservoir uids (Woods, 1999). The interaction between the different fronts is of prime interest if one wishes to truly understand how a reservoirs properties will change in each of these processes. While there exists an abundant amount of literature on the description of thermal and reaction fronts for more conventional reservoir processes (Akkutlu and Yortsos, 2003; Mailybaev et al., 2010; Phillips, 1991; Woods, 1999) (for instance, in situ combustion) there are no such contributions for shale reservoir processes. To the knowledge of the authors, there are four other studies, published in the open literature, which deal with the simulation of the in situ extraction of shale oil (Bauman and Deo, 2011; Bauman et al., 2010; Fan et al., 2009; White et al., 2010). Conduction heating and kerogen combustion were investigated. Higher heater temperatures result in faster recoveries, but potentially unfavorable oil decompositions (Fan et al., 2009; White et al., 2010). Combining the two methods was found to be benecial (Bauman et al., 2010), for example when combustion follows conduction heating in order to displace the deposited heavy oils. A timespan of 510 years is recommended for heating (Bauman and Deo, 2011; Bauman et al., 2010; Fan et al., 2009; Sun et al., 2006; White et al., 2010), while EROI ratios of as high as four (Bauman et al., 2010) have been reported. EROI is dened as the energy output (the potential heat of formation when combusting the fuel) to the energy input (in the form of pressurized and heated uid injected into the reservoir) ratio. Numerical studies employed either in-house (White et al., 2010; Fan et al., 2009) or commercial (Bauman et al., 2010) reservoir simulators, with empirically derived closure models. Shells study is the only large scale in situ oil shale upgrading study to date (Bauman et al., 2010). An important conclusion of White et al. (2010) was that only considering kerogen decomposition ignoring oil and coke decompositions yields residual liquid oil in the formation, something which is inconsistent with lab and eld observations. Studies have focused extensively on well patterns, extraction rates and efciencies for conduction heating (Bauman and Deo, 2011; Bauman et al., 2010; Fan et al., 2009; Sun et al., 2006; White et al., 2010). There seems to be no analytical work on the subject of oil shale upgrading, or any contribution which investigates oil shale upgrading by hot gas injection. Finally the evolution of the various thermal and reaction fronts developed through heating shale has not been analyzed systematically. 1.3. Objectives It is thus the objective of this investigation to implement a model for the thermal and reaction fronts developed when shale is heated in situ by conduction and hot gas injection. Coupled with non-dimensional analysis, a deeper physical understanding of these processes is obtained. In order to assess the performance

of each method, a preliminary comparison of oil shale upgrading by thermal conduction and hot inert gas injection is presented and discussed.

2. Model formulation 2.1. Summary of the model This study was performed using an in-house code to simulate reacting multiphase ow in porous media. A kinetic model developed by Burnham and Braun (1990) which is based on a multiple pathway thermal decomposition model, was used. Kerogen decomposes into components, which at industrially relevant temperatures and pressures, include gases, liquids and solids. Reservoir properties such as permeability, porosity and depth are inputs to the model and in this paper, values typical of the Green River shale formation in the US are used. The OSU process is modeled as a Cartesian problem, with heat entering the reservoir vertically through an injection well and extracted at the bottom by a collector well. These wells are represented by appropriate boundary conditions. The phase behavior of the liquids and gases was described using the PengRobinson equation of state. Oil and gas species densities vary according to reservoir pressure and temperature. The output of the model is the concentration of liquid and gaseous species, per unit volume of the reservoir, as a function of distance and time. In greater detail, the model is given in the following sections, while the Appendix includes some details on the closures used and the non-dimensionalizations. 2.2. Governing equations Several phases are considered, a solid phase (the deposit matrix) and a gaseous phase made of many species. The species/lumped species considered are given in Table 1. In greater detail, there will be three distinct phases, two of which (the gas mixture and the oil) will be present within the pore space of the shale (the void continuum), resulting in the need for multiphase ow equations. The third phase is the solid matrix. Since with reservoir ow inertial forces are orders of magnitude smaller than viscous forces, the velocity eld is derived simply by applying Darcys law (Phillips, 1991): ua kkra

ma

rPa rg rz

where k is the absolute permeability of the shale matrix, f is the porosity, kra is the relative permeability of phase a, ma is the viscosity of the phase and rPa is the spatial gradient operator and
Table 1 Species considered in this simulation, their symbols and their respective phases. Variable Symbol Phase Molecular weight 44.01 339.85 130.88 44.10 16.04 647.00 13.00 13.00 13.00 P crit (MPa) 7.38 1.40 2.78 4.25 4.6 T crit (K) 304.18 893.70 646.30 370.18 190.18

Carbon dioxide Heavy oil Light oil Hydrocarbon gas Methane Kerogen Coke-1 Coke-2 Coke-3

CO2 ho lo hc me k c1 c2 c3

Gaseous Liquid/ gaseous Liquid/ gaseous Gaseous Gaseous Solid Solid Solid Solid

202

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

z is the depth. The phase symbol a can either denote a gas (g) or oil (o). The conservation of the liquid (oil) and gas mixture phases is governed by @fra Sa r ra ua qa @t 2

where ra is the density of the phase, Sa is the saturation of the phase (ratio of phase volume to total non-solid volume) and qa is the source term in kg/m3/s. The transport of components within the phases is governed by @frai cai r rai cai ua r Di rai rcai qai @t 3

where cai is the mass fraction of a component i within the formation and Di is the mass diffusivity of component i in the formation. When this equation is summed, Eq. (2) is obtained, since Di D. The pressure eld is found by combining mass conservation, Darcys law for the gaseous phase and the gas law

PW ZRT

Z is the gas deviation factor. The following pressure equation is obtained:    Pa W a kkrg @fSa Pa W a =Z a T P W r rPa a a g rz qa 5 @t mZ a RT Za T where T is the formation temperature, R is the universal gas constant and W is the mass averaged molecular weight of the gaseous phase. The term q represents the pressure source term due to chemical reactions. Capillary pressures are accounted for through the empirical model provided by Chen et al. (2009). Finally, the governing equation for the temperature is presented. Only one energy governing equation for the solid matrix and the void continuum is required. The time needed for the ow to traverse a grain tow (10 4 s) is much greater than the time for a rock grain to absorb the uids heat theat (10 7 s) (Phillips, 1991; Woods, 1999), meaning that all uid phases will be at the same temperature as the rock grains. The formation temperature equation is described as in Chen et al. (2009) and is written as " # Na Na X @ X ra cpa T 1frs cps T r ra ua cpa T rlrT qe @t a a 6 where a is the thermal diffusivity of the formation, qe is the energy source term from chemical reactions. Following Woods (1999) we dene Ga as

injection and collector wells. The injector well boundary condition is that of a constant pressure. As such, the injection rate shall vary depending on the reservoir conditions, prompting a maximum ow rate to be set at the collector and injector. Vertical 1D direct line drive is simulated through the use of cartesian coordinates. As Liu et al. (2011) note, such a system promotes gravity stable displacement. One must note the limiting effects of a 1D system as mainly being constrained to a homogenous reservoir. Inhomogeneities are certainly an important feature of any reservoir, however as this is the rst study of its kind on thermal and reaction fronts for shale oil upgrading, homogeneity is assumed to isolate key physical mechanisms. There may be some heat loss when thermal OSU methods are employed. However, Vinsome and Westerveld (1980) have noted that for large Peclet numbers, longitudinal heat losses are few. Liu et al. (2011) have used heat loss models for 1D simulations. These models cannot account for how heat losses affect the energy transport at the boundaries and thus they are unt for use in this study. However, even without taking into account heat loss, this study can provide much insight into the front dynamics of thermal OSU. The schematic shown in Fig. 1 can represent our solution. Heat and mass transfer occur only in the lateral direction. 2.3. Porous medium and transport properties Lee (2010) has compiled much experimental data on Shale, outlining certain ranges of values that are typical of a Oil Shale sample. The shale properties shall be modeled here as if the shale were ca. 1 km underground, thus subject to high temperatures and very high pressures. The shale deposit will be lled with methane initially. The initial conditions of the shale deposit are presented in Table 2. Of the several studies that simulated oil extraction from Green River formation shale (Bauman and Deo, 2011; Bauman et al., 2010; Fan et al., 2009; White et al., 2010) initial permeability used varied from 0.003 mD to 10 mD. A higher value on this range was chosen for three reasons. First, kerogen concentrations used in this study are typical of lean deposits which have higher permeabilities typical of oil shale reservoirs that would be targeted as economically recoverable. Second, hot gas injection would only be applied to oil shale

Ga

PN a

ra cpa

a ra cpa 1frs cps

cpa is the specic heat capacity of phase a, Na is the number of phases (in this case there are two) and the subscript s denotes the solid phase. This can be physically understood as the dimensionless heat capacity of the injected uid. Finally, qe is the energy source term from the chemical reactions. This formulation necessitates the use of two-phase relative permeability which is obtained from Chen et al. (2009) and details are given in the Appendix. With regards to oil shale upgrading, the authors are unaware of any empirical studies to describe relative permeabilities specic to hydrocarbons derives from oil shale upgrading. As such, this study employs empirical data specic to typical black oil models, as other studies on oil shale upgrading have (Bauman and Deo, 2011; Bauman et al., 2010; Fan et al., 2009; White et al., 2010). Oil shale upgrading is achieved by heating through a borehole at the top of the formation, with a single collector at the bottom. Boundary conditions at either end of the domain simulate the

Fig. 1. Schematic for simulation grid. 1N refer to the grid nodes for the numerical solution.

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

203

Table 2 Summary of formation initial conditions. P, SK, SC, t and T denote pressure, kerogen volumetric saturation, coke volumetric saturation, time and temperature respectively. Property Permeability Temperature Density kerogen Density inorganics Concentration of kerogen Concentration of inorganics Kerogen Mass Pct. Thermal conductivity Value 7 mD 313 K 1200 kg/m3 2200 kg/m3 240 kg/m3 1960 kg/m3 10 1.2 W/m K Constant? Function of NO NO NO YES NO YES NO YES P, SK, SC (White et al., 2010) t, T t, T t, T T (Nottenburg et al., 1978) P, SK, SC (White et al., 2010) T (Qian and Yin, 2010) t

Table 3 Initial inert gas properties for base case at inlet. Property Temperature Density Pressure Specic heat capacity Viscosity Value 573 K 141.15 kg/m3 230 bar 1110 J/kg K 0.00007942 Pa s Constant? YES NO YES NO NO Function of P, T T P, T (Chen et al., 2009)

Porosity Specic heat capacity Pressure of formation Mass diffusivity

0.1 1335 J/kg K 100 atm 4.8 10 9 m2/s

NO NO NO YES

alternate pathway mechanism, by which oil is formed from kerogen directly or from bitumen if such bitumen has been produced by the kerogen. Reactions 15, used here, are taken from Burnham and Braun (1992) and detailed in the Appendix (Table 3). Mineral decomposition is ignored since the maximum temperature encountered during simulation is insufcient to activate these reactions. Water presence is ignored as only small quantities may be present in most shales (Fan et al., 2009; Rahm, 2011). Table 4 summarizes each reactions kinetic parameters. 2.5. Non-dimensionalization

reservoirs with relatively high initial permeabilities, thus it makes sense to assess said method on such reservoirs. Finally, Shell (Vinegar, 2006) and Exxon (Symington et al., 2010) have recently reported that initial hot water ooding could dissolve much inorganic material resulting in ca. an order of magnitude increase in initial permeability. As such, typically higher values of oil shale initial permeability are very likely to exist in practice before any upgrading takes place. Kerogen decomposition will result in increased pore space, less the deposited coke residue. Pressure increases will act to increase the pore radii which result in increased permeability. These phenomena are captured in the empirical equation for permeability variation, detailed in the appendix. It was decided to use CO2 as the inert gas for injection, as per Chevrons method. The transport properties of CO2 are mostly temperature dependent, though the large pressure variations do vary some of the properties signicantly (e.g. dynamic viscosity). The initial conditions of the inert gas, within the borehole, are shown below for 573 K and 230 atm, the chosen base case condition for inert gas injections. The conduction base case is simply a temperature boundary condition at 573 K with no uid injection. A minimum bottom hole pressure, as well as a maximum collection rate, is dened. Fluids are allowed to pass through the collector boundary at a rate specied by each phases velocity, but no more than the maximum allowable collection rate. As the uids travel through the reservoir, they are eventually cooled. The lighter oil fraction will condense to its liquid phase. Thermodynamic transport properties are detailed in the Appendix. Relative permeability equations are also detailed in the Appendix for the oil and gaseous phases. 2.4. Chemistry Since there is no oxygen in the gas injected, only pyrolysis reactions are considered. The decomposition of kerogen into Oil and other products has been very widely studied but not one single model exists that is used by a majority of these researchers. However, the prevalent model is that of Burnham and Braun (1992) and there is a prevalent tendency to take primary kerogen decomposition as being a rst order reaction (Braun and Burnham, 1986; Burnham and Braun, 1992; Fan et al., 2009; White et al., 2010). These authors pointed out that all data are consistent with an

A set of non-dimensional equations was developed for the case of oil shale upgrading, based upon the methods in Phillips (1991) and Jupp and Woods (2003). By working with the non-dimensional coordinate Z, dened as

Z Gg ug =lxGut

nine dimensionless groups were derived in the Appendix and are given below.

P1

lrk rk F k, ho
2 u2 g fC ho, max Gg

P2

lrho
2 u2 g fGg

p1

lrk r k F k, lo rho r ho F ho, lo 2 u2 g fC lo, max Gg lHk, rxn lHo, rxn , E2 2 2T 2 G2 u G g g inlet g ug T inlet

10

E1

11

lrk l , Le 2 D G2 g ug 1f
1fGg , R uo ug

12

fGg

13

Another dimensionless group was added, in accordance with the observed phenomena by Jupp and Woods (2003).

ug Dx

14

The variable ri refers to the rate of pyrolysis of the species i. F i, j refers to the mass fraction of j produced from the pyrolysis of species i. It must be noted that the thermal front will travel with the speed of approximately Gg ug (Woods, 1999), i.e. based on the Darcy velocity of the gas, since the Darcy velocity of the oil phase is signicantly (23 orders of magnitude) smaller, based on typical gas and oil viscosities (Chen et al., 2009). The rst two terms can be thought of as ratios of production/ depletion rates of heavy oil to thermal front speed. The more positive the value, the larger the rate of production given a front speed. Similarly, p1 is a measure of the rate of production of light oil to the speed squared of the thermal front. Alternatively they can be thought of as ratio of the time scale of formation of the thermal front (l=Gg ug 2 ) to the time scale over which the specic reaction occurs (1=r ).

204

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

Table 4 Kinetic data for decomposition reactions (Burnham and Braun, 1992). Coke 3 and H/C gas activation energies are far too high to be activated at this studys temperatures of interest (Burnham and Braun, 1992). Decomposition reaction Heavy oil (HO) - 0.373 LO 0.156 Gas 0.03 Methane 0.441 Coke 2 (2) Light oil (LO) - 0.595 Gas 0.115 Methane 0.290 Coke 3 (3) Kerogen - 0.279 HO 0.143 LO 0.018 Gas 0.005 Methane 0.555 Coke 1 (1) Coke 1 - 0.031 Gas 0.033 Methane 0.936 Coke 2 (4) Coke 2 - 0.003 Gas 0.033 Methane 0.964 Coke 2 (5) Frequency factor (1/s) 1.0 1013 5.0 1011 3.0 1013 1.0 1013 5.0 1011 Activation energy (kJ/mol) 213.4 225.9 225.9 225.9 225.9

DH (J/mol)
46 500 46 500 3 35 000 46 500 46 500

E1 and E2 are simply the non-dimensionalized energy release terms. The rate of energy production (power) is nondimensionalized by the rate of heat traveling past a point per unit second due to the thermal front. Y is a measure of the rate of depletion of kerogen to the speed square of the thermal front. Le is the Lewis number, which is a measure of thermal to mass diffusivity. The dimensionless number V is a measure of the speed of the uid front to the speed of the thermal front. R is the ratio of the oil phase to the gaseous phase velocity. Finally, the term b is a ratio of the volumetric ow rate per unit length of uid to the thermal diffusivity (effectively a measure of the inuence of convection to that of conduction) and plays an important role in dening the temperature prole, as will bee seen later in the paper. 2.6. Numerical method In order to solve the set of differential equations presented above the Method of Lines (MOLs) was used. By this method, all the PDEs were discretized in spatial dimensions but not in the time dimension. Central differences are used for diffusion and upwind differencing for convection terms. This leaves us with a set of ODEs for which initial value conditions have been provided. The DVODPK (Variable-coefcient Ordinary Differential equation solver, written in FORTRAN 77) (Brown et al., 1989) was used and run in fully implicit stiff mode. This solver has been used for a wider variety of reacting ow problems (de Paola et al., 2009; Wright et al., 2009) and has been shown to handle stiff sets of equations well (Byrne, 1992).

100 Species Concentration (kg/m3) Light Oil Heavy Oil Coke 1 Coke 2 Coke 3 Gas mixture

80

60

40

20

0 0.5 1 1.5 2 2.5 Distance 3 3.5 4

150 Species Concentration (kg/m3) Light Oil Heavy Oil Coke 1 Coke 2 Coke 3 Gas mixture

100

50

3. Results and discussion 3.1. In situ gas injection Fig. 2(a) shows the species concentrations at 30 days for hot gas injection. The oil phases terminate near the reaction front and there is little heavy oil decomposition (there is very little of the coke 2 species). Fig. 3(a) shows the extent of kerogen decomposition with time, as the rate at which the reaction front penetrates the reservoir is ca. 6 10 6 m/s but decreasing slowly with time. The velocity of the gaseous phases drops as the gases move outward into the reservoir, thus it is no surprise that the thermal front progression also slows down as time passes (Fig. 3(b)). Insignicant thermal decomposition of heavy oil is taking place (there are very small amounts of Coke 2 in the system which is a product of heavy oil decomposition). The thermal gradients are relatively steep when compared to what thermal conduction achieves, as will be discussed later. This means that the injection rate is large enough in order to ensure that thermal advection is the dominant heat transfer mechanism. In Fig. 4, the normalized decomposition rates of heavy oil and kerogen vs. Z are plotted. The reaction rates are normalized with

0.5

1 Distance

1.5

Fig. 2. (a) Species proles for Hot gas injection after 30 days and (b) species proles for Heat conduction after 90 days.

the rate of decomposition either species would undergo were its temperature at the injected temperature. Both reaction fronts retreat to more negative Z values as time passes. Our nondimensional coordinate Z is based on the location of the thermal front. Since the thermal fronts plateau is not horizontal and is growing in the x-direction, the thermal front (as tracked by Z, lags behind the true thermal front. Despite this, the front decomposition rate for kerogen is slowly receding due to slowly decreasing thermal front temperatures. However, heavy oil decomposition rates are growing because of the increasing presence of heavy oil. Fig. 5(a) yields some very interesting insight into the front dynamics of the system. The normalized temperature prole shows us where the thermal front is located at 15 days into the heating program. The net heavy oil production of heavy oil is maximum at the thermal fronts edge. Right ahead of the reaction

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

205

Normalised Kerogen Decomposition Rate

1 15 days 30 days 45 days

0.3 15 days 30 days 45 days

0.8 Kerogen (kg/m3)

0.25

0.2

0.6

0.15

0.4

0.1

0.2

0.05

0.5

1.5

2 2.5 Distance (m)

3.5

0 10

Normalised Heavy Oil Decomposition Rate

600 550 Temperature (K) 500 450 400 350 300 15 days 30 days 45 days

0.03 15 days 30 days 45 days

0.025

0.02

0.015

0.01

0.005

4 6 distance (m)

10

0 10

Fig. 4. Hot gas injection: (a) proles of the kerogen pyrolysis rate and (b) proles of the heavy oil pyrolysis rate.

Fig. 3. Hot gas injection: (a) proles of kerogen at 573 K and (b) the evolution of the thermal front as a function of distance for different times.

front, the ratio of the thermal to the uid front speed is minimum. To understand why these two observations occur, the following three graphs of Fig. 5 are considered. Fig. 5(b) depicts the variable b vs. the dimensionless coordinate Z. As the heating progresses, the b prole shifts upwards, implying thermal convections increasing importance. This would mean that the velocity of the gas phase is increasing as a result of permeability increasing at high pressures and the freeing up of pore space from the kerogen. The b proles are at local minimum around the thermal front, downstream of which they rapidly increase. Downstream of the thermal and reaction fronts, the gas is less hindered by the banks of oil being built up upstream of the thermal front. The local minimum occurs because the heavy oil hinders the ow of the gaseous phase. Fig. 5(c) shows the plots of the dimensionless depletion/production terms of Heavy Oil after 15 days. These plots capture several important aspects of the system. They show the decreasing reaction rate, upstream of the thermal front, with time due to kerogen depletion. Also, the abrupt decline just downstream of the thermal front implies there is very little to no kerogen decomposition happening. This leads to a very important result: the given reaction timescales and convection heating timescales have led to the point where the thermal front and reaction front are clearly aligned throughout the heating process. This was seen in Fig. 5(a). The non-dimensional production of heavy oil stays positive for the whole of the heating program. P1 P2 quanties the extent

of dominance that kerogen decomposition has over heavy oil decomposition and would be negative at a higher temperature. Since a very low temperature will yield a very high dominance, yet low kerogen decomposition rates, this marker can not be the sole point of reference for the extent of kerogen depletion. Fig. 5(d) depicts the proles of the dimensionless number 1/V. The variable 1/V quanties the ratio of the thermal to uid front speed. Perhaps the rst thing noticed is that all the radial values are small. This makes sense because the uid front will always move faster than the thermal front. However, two distinct regimes on this graph are noticed. Upstream of the thermal front, there is an increase in the importance of the thermal versus the uid front. This would be driven by the initial formation of the thermal front followed by the increasing presence of oil, which slows down the gas front. What is now observed differently from b is that downstream of the thermal front, the curves converge. Slightly further downstream of the thermal front a minimum value for 1/V is reached. This value could be considered a characteristic marker for this base case, as it does not seem to be affected by time. Further to this, downstream of the thermal front, the pressure gradients develop and thus would affect the thermal and uid front speeds in equal proportions. This would explain why the curves converge downstream of the thermal front. If indeed it is the case that the decomposition reactions are what lead to the temporal evolution of the 1/V curves upstream of the thermal front, then this must mean that the reaction front ends where the thermal front speed is at a minimum.

206

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

1.1 normalized 1/V normalized temperature normalized 1 + 2

2000 1800 1600 1400 1200

15 days 30 days 45 days

0.9

1000 800 600

0.8

2.5 2 1.5 1 0.5 0 0.5

400 200

0.7 3

0 10

5.6 0.08 15 days 30 days 45 days 5.4 5.2 5 1/V 4.8 4.6 0.02 4.4 4.2 0 3 2.5 2 1.5 1 0.5 0

x 10

7.5 days 10 days 15 days

0.06 1 + 2

0.04

4 4

Fig. 5. Hot gas injection: (a) proles of 1/V, P1 P2 and normalized temperature vs. Z after 15 days. Net heavy oil production peaks at the edge of the thermal front. The thermal to uid front speed is at minimum right ahead of the reacting front , (b) proles of b vs. Z, (c) proles of P1 P2 vs. Z and (d) proles of 1/V vs. Z.

When running this base case for 180 days, the nondimensional production of heavy oil converges ca. 0.036 (at the thermal front). As such, for a given 1/V one can infer characteristic value of P, thereby being able to comment on the extent of decomposition solely by looking at the thermal front. The nal recovery at 142 days for the gas injection base case was 0.78 and the energy returned on energy invested (EROI) was ca. 3.0. These recoveries are not achievable in reality due to heat losses and heterogeneities. However, they serve as a benchmark for real world comparison as well as comparison with other ideal cases of oil shale upgrading methods. The recovery of hydrocarbons was dened as the ratio of the total amount of heavy and light oils extracted at the well node to the maximum recoverable amount of oils (assuming no heavy and light oil pyrolysis). Residual oil, which could not be displaced, accounted for most of the unrecovered oil for the base case run. 3.2. In situ conduction heating The conduction base case (the gas injection boundary condition was removed) was run up to 270 days. For the conduction base case, the thermal front tails are exceedingly long (shown in Fig. 6(a)) and they take much longer to develop, compared to the gas injection case. The thickness of the thermal front heats shale volumes where no reactions are taking place. Due to this, the residence time is long enough now to permit signicant decomposition of heavy oil and coke-1. Fig. 2(b) shows species concentrations for heat conduction. There is much heavy oil and coke 1 decomposition. The reaction front takes much longer to progress than the gas injection case, because there is hardly any gas to carry the heat downstream. Conduction heating does not produce a moving thermal front, but rather an ever stretching thermal prole.

Fig. 6(d) displays two separate regions. The rst one is a small region between Z 0 and ca. Z 0:25. Here, the 1/V curve evolves with time. This region, as will be shown, is the region where kerogen decomposition is being completed, i.e. the depletion region. This depleted region terminates quite abruptly at ca. Z 0:25, while ahead of the depletion region there is very little change in time because no reactions have occurred there yet. Fig. 6(c) lends credence to the previous reference about the depletion region terminating at ca. Z 0:25. The P1 P2 curves collapse onto each other ahead of Z 0:25. The peak represents the reaction front, which decreases in rate with time. The volumetric ux per unit length of uids is constant throughout time upstream of the reaction front (Fig. 6(b)). Further to this, since the reaction front moves in tandem with the 1/V curve depletion region, the extent of decomposition based solely on the kerogen products inuence on the thermal fronts progression is inferred. The nal recovery at 3012 days for the conduction base case was 0.75 and the energy returned on energy invested (EROI) was ca. 3.4. Compared with hot gas injection, this method appears to be more energetically efcient due to the much lower heat input rates. Compressing gas, heating it and then injecting it is much more energy intensive than using resistance heaters. 3.3. Sensitivity analysis Having described the base case, key parameters in the system were varied so as to investigate their respective inuence on thermal and reaction front progression. Variation of the initial permeability, kerogen reaction rates, temperature and pressure of the gas injected into the reservoir were investigated. Permeability and pressure variations have more effect on gas injection, thus only this method was considered in those sensitivity analyses. For

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

207

Kerogen Decomposition Rate (kg/m3s)

3.5 3 2.5 2

x 10

2000 270 days 135 days 90 days 1500 270 days 135 days 90 days

1000

1.5 1 0.5 0 2 1 0 1 2 0 2 500

150 270 days 135 days 90 days 100 1 + 2 1/V

2.5

x 104 270 days 135 days 90 days

1.5

50 1

0 1.5

0.5

0.5

1.5

0.5

0.5

1.5

2.5

Fig. 6. Heat conduction: (a) temperature proles, (b) proles of b vs. Z, (c) proles of P1 P2 vs. Z and (d) proles of 1/V vs. Z.

the conduction heating method, only temperature and kerogen reaction rates were varied and are presented alongside the variations of temperature and kerogen reaction rates for the hot gas injection simulations.

3.3.1. Permeability variations Permeability can inhibit uid progression greatly, leading to a dominance of conduction as a heat transfer mechanism, or it can permit very high ow rates, thus creating a very steep traveling thermal front. It is thus no surprise that permeability enhancing techniques are becoming more and more advanced (Farcas and Woods, 2007). Any oil shale upgrading process which aims to rely on hot gas injection should make sure that reservoir permeability is at a favorable level. A high nal permeability permits efcient oil shale upgrading relatively quickly. However, below a certain permeability, the convection as a heat transfer mechanism fails to dominate conduction. This leads to very slow overall decomposition rates. Fig. 7(d) shows the thermal front greatly suffering, with respect to the uid front, (predominantly upstream of Z 0) from the decrease in permeability. But this is not only because the thermal front travels faster (evidence of that provided in Fig. 7(a)), but rather because as is shown in Fig. 7(b) the uid fronts ability to convect energy declines, as indicated by lower b values. Thus, conduction plays the signicant role in the overall shape of the thermal front. In fact for 10 mD, the thermal conduction is the stronger of the two heating mechanisms, and thus no formation of a traveling thermal front is observed. As such, it is possible to nd a minimum permeability for which hot gas injection would not induce a traveling thermal front for the given boundary and initial conditions. In order to give a characteristic b value to each of the simulation runs in Fig. 7(b), the minimum value ahead of the thermal front is selected. This minimum value happens to coincide with the edge of

the thermal and depletion front, as discussed earlier. This is expected because at this point the hot injected gas is the only uid owing through shale that has not seen an increase in permeability due to kerogen decomposition. It is observed that for bmin o 1, a traveling thermal front does not develop. For 10 o bmin o 1000 thermal fronts are forming, however thermal tails are still signicant in their inuence on the efciency of heating. Finally, for bmin 4 1000 very small thermal tails and a fast traveling thermal front are expected. The graph of non-dimensional net production of heavy oil for varying permeability (Fig. 7(c)), shows a decrease for increasing permeabilities, because of kerogen depletion. Recovery rates for high permeability reservoirs are noticeably larger, with the added benet that residence times are decreased thus avoiding unwanted heavy oil decomposition.

3.3.2. Pressure variations Three inlet pressures for the inert gas were run: 210, 230 and 250 bar. Fig. 8(a) and (d) show that higher pressures lead to quicker thermal and reaction fronts. There is no evidence that the thermal front shape benets from an increase in pressure, only that it progresses faster. Although there is a noticeable increase in volumetric ow rate per unit length with respect to the thermal diffusivity (Fig. 8(b)) upstream of the thermal front, this is not the case downstream of it. Finally, an intermediate pressure leads to a higher nondimensional net production of oil (Fig. 8(c)) at 15 days. Recovery rates for highly pressurized injectants are noticeably larger, with the added benet that residence times are decreased thus avoiding unwanted heavy oil decomposition. There is a limit to the increase in efciency for higher pressure injection, due to the added energy required to compress the uids.

208

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

650 600 Temperature (K) 550 500 450 400 350 300 1 2 3 Distance (m) 4

1mD 7mD 10mD

5000 4000 3000 2000 1000 0 20

1mD 7mD 10mD

15

10

0.08

0.06 1 + 2

1mD 7mD 10mD

10 9 8 1/V 7 6

x 10

1mD 7mD 10mD

0.04

0.02

5 4 6

0 2

1.5

0.5

0.5

Fig. 7. Hot gas injection: (a) temperature proles for permeability variations, (b) proles of b vs. Z for permeability variations, (c) proles of P1 P2 vs. Z for permeability variations and (d) proles of 1/V vs. Z for permeability variations.

650 600 Temperature (K) 550 500 450 2000 400 350 300 1 2 3 Distance (m) 4 5 1000 210 MPa 230 MPa 250 MPa 5000 210 MPa 230 MPa 250 MPa

4000

3000

0 10

10 0.07 0.06 0.05 1 + 2 1/V 0.04 0.03 0.02 0.01 4 0 2 1.5 1 0.5 0 0.5 1 210 MPa 230 MPa 250 MPa 9 8 7 6 5

x 10

210 MPa 230 MPa 250 MPa

Fig. 8. Hot gas injection: (a) temperature proles for pressure variations, (b) proles of b vs. Z for pressure variations, (c) proles of P1 P2 vs. Z for pressure variations and (d) proles of 1/V vs. Z for pressure variations.

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

209

650 600 Temperature (K) 550 500 450 400 500 350 300 1 2 3 4 5 Distance (m) 0 10 553K 573K 593K 2000 553K 573K 593K 1500

1000

0.08 553K 573K 593K 0.06

6.5 6 5.5

x 103 553K 573K 593K

1 + 2

0.04

1/V 1.5 1 0.5 0

5 4.5 4

0.02

0 5 4 3 2 1 0 1 2

Fig. 9. Hot gas injection: (a) temperature proles for temperature variations, (b) proles of b vs. Z for temperature variations, (c) proles of P1 P2 vs. Z for temperature variations and (d) proles of 1/V vs. Z for temperature variations.

3.3.3. Temperature variations Fig. 9(a) shows that increasing the temperature at the inlet will strengthen the thermal front. Increasing the temperature even more (to 593 K) increased the thermal front speed, however there is a drop in the 1/V prole at 593 L, upstream of the reaction front. This is shown in Fig. 9(b) shows a difference in the rate of kerogen decomposition at higher temperatures. At the highest temperature, the non-dimensional production of heavy oil has decreased because of ensuing decomposition. For a higher temperature, while the kerogen decomposes much more readily, heavy oil decomposition is more likely to ensue. Too low a temperature delays the decomposition of kerogen by too much and leads to low overall decomposition rates (Fig. 9(c,d)). This makes sense because the heavy oil has decomposed back into a solid material and thus increased the solid mass that the thermal front must heat and hinders the subsequent ow of gas. Even though the thermal tails of the proles do not diverge that much (Fig. 10(a)) the 1/V proles in Fig. 10(b) clearly show an increasingly wide depletion region front for the in situ conduction method. The wider, the more efcient because the thermal tail is smaller and does not heat as much kerogen which is not decomposing. Recovery rates for high temperature injectants are noticeably larger. The competing effect of unwanted heavy oil decomposition results in a presumably optimal temperature which is neither too low nor too high. High temperature conduction recovery rates for temperature variations display similar behavior.

decomposition rate case because pore space is freed up from kerogen much more rapidly. The reaction front is also much thinner (Fig. 10(d)) due to the increased decomposition rate. Thus, knowing the kerogen chemistry will allow us to infer depletion region lengths solely by looking at the progression of the thermal front. Hot gas injection cases are now considered. Fig. 11(a) shows that increasing the kerogen decomposition rate has little effect on the thermal front. This is to be expected since the heat release rate is very small compared to the heat injection rate. In fact, quicker decomposition leads to impeded hot gas ow (Fig. 11(b)) and thus a weaker convective front, as observed. Fig. 11(c) shows, as expected, quicker decomposition rates for increased chemistry rates. Fig. 11(d) shows the 1/V proles from which much useful information is extracted. The thermal front propagation is hindered at high reaction rates due to increased oil presence. Downstream of the front, the proles converge. This signies that no decomposition is taking place, denitive proof the thermal and reaction fronts are aligned.

4. Conclusions A set of governing equations and closure models for multiphase reacting ows in Shale reservoirs have been developed and solved numerically for a one-dimensional formation. The nal recovery at 142 days for the gas injection base case was 0.78 and the EROI was ca. 3.0. The nal recovery at 3012 days for the conduction base case was 0.75 and the EROI was ca. 3.4. It is observed that for migrating thermal fronts, the reaction front follows the thermal front closely, thus allowing one to roughly calculate the extent of decomposition solely by looking

3.3.4. Kerogen decomposition rate variations The increase in depletion region length is clearly illustrated, for the heat conduction case, in Fig. 10(c) and (d). A more rapid kerogen decomposition rate results in deeper reaction front penetration. The gas travels quicker through the high

210

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

650 600 Temperature (K) 550 500 450 400 350 300 1 2 3 Distance (m) 4

573 K 593 K 603 K

x 10

573 K 593 K 603 K

1/V

Kerogen Decomposition Rate (kg/m3s)

120 100 80 beta 60 40 20 2 1 0 1 2

0.5 x Base Base 2 x Base

1.2 1 0.8 0.6 0.4 0.2 0 1

0.5 x Base Base 2 x Base

0.5

0.5

Fig. 10. Heat conduction: (a) temperature proles for temperature variations, (b) proles of 1/V vs. Z for temperature variations, (c) proles of b vs. Z for kerogen decomposition rate variations and (d) kerogen decomposition rates vs. Z for kerogen decomposition rate variations.

650 600 Temperature (K) 550 500 450 400 500 350 300 1 2 3 Distance (m) 4 5 0 10 1000 0.5 x Base Base 2 x Base 2000 0.5 x Base Base 2 x Base

1500

0.08 0.5 x Base Base 2 x Base

6.5 6 5.5

x 10

0.06 1 + 2

0.5 x Base Base 2 x Base

0.04

1/V 1.5 1 0.5 0 0.5

5 4.5 4

0.02

0 5 4 3 2 1 0 1 2

Fig. 11. Hot gas injection: (a) temperature proles for kerogen decomposition rate variations, (b) proles of b vs. Z for kerogen decomposition rate variations, (c) proles of P1 P2 vs. Z for kerogen decomposition rate variations and (d) proles of 1/V vs. Z for kerogen decomposition rate variations.

at thermal front progression. This was true for both the conduction and hot gas injection base cases. There was very little (less than 1%) heavy oil decomposition observed for heavy oil in the

hot gas injection base case, however this was not true for the conduction base case where large residence times allowed for signicant heavy oil decomposition.

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

211

Hot gas injection was found to be more effective, in terms of speed and heavy oil recovery, for heating the formation. The more dominant thermal conduction is in the formation of the thermal front, the larger the tail of the temperature prole. Heavy Oil decomposition becomes very active at high temperatures and since it has a high viscosity, it will remain in the reservoir for some time before it is extracted; this being more prominent in the conduction base case. This will mean that it will almost certainly decompose completely if the heating temperature is too high, regardless of heating method. Plotting the ratio of the thermal front speed to the uid speed (1/V) allows one to infer that the reaction front ends where this ratio is at a minimum, for both heating methods. It was found that for given boundary conditions and initial conditions, a certain characteristic value for each 1/V can be assigned that will characterize the behavior of the induced thermal front and reaction front. For heat conduction, a depletion region can be dened to characterize each run. For hot gas injection, the dimensionless variable b must have a minimum value of ca. 1 in order to produce a thermal front which migrates through the reservoir. Compared to varying deposit porosities and kerogen activations energies, varying temperature, pressure and permeability, in either method, are more important. Varying the gas injection pressure had little effect on much else than varying the progression of the thermal and reaction fronts. The temperature had a great effect on the depletion rate, the shape of the thermal front and its progression and the upstream volumetric ow rate per unit length. Nomenclature A C E F H P S T V W Y Z c f I k kr r frequency factor (s 1) concentration (kg/m3) activation energy (kJ/mol) product mass fraction from pyrolysis reaction enthalpy of reaction (J/mol) pressure (Pa) saturation V pores =V total temperature (K) volume of uid (m3) molecular weight non-dimensional kerogen pyrolysis rate gas deviation factor species mass fraction fraction of decomposing species species subscript permeability relative permeability pyrolysis rate (kg/s) non-dimensional uid heat capacity non-dimensional rate of production of heavy oil non-dimensional heavy oil pyrolysis rate phase subscript ratio of inuence of convection to conduction heating non-dimensional coordinate that follows thermal wave thermal conductivity (W/m K) uid viscosity (Pa s) non-dimensional rate of production of light oil uid density (kg/m3) porosity

Appendix A. Non-dimensionalization The injection of a uid in a 1D pore space at a constant Darcy speed ug is considered. The kerogen and heavy/light oil decompose at a rate given by Q k r k C k Q o r o C o A:1 A:2

where C is the mass concentration of a species per unit reservoir volume. The oil (non-aqueous liquid phase) ows through the pore space as described by the following equation:

@ro r uo ro Q o, p Q o, d @t

A:3

Two lumped oil species are considered: light oil and heavy oil. They are both described by Eq. (A.3). For the purposes of this analysis, the decomposition of light oil is neglected since it is very slow compared to the decomposition of heavy (Burnham and Braun, 1992). Thus, the behavior of the heavy oil decomposition reaction will dominate in terms of affecting the uid dynamics and chemistry of the system. Also, the rate at which kerogen is depleted is: @rk Q k @t 1f A:4

The following procedure is developed based upon the methods in Phillips (1991) and Jupp and Woods (2003). It is desirable to write the differential equations in dimensionless form, thus the dimensionless coordinate system is based on that which is of most interest: the speed of the thermal front, Gu:

Z Gg ug =lxGg ug t t Gg ug 2 t=l

A:5 A:6

The dimensionless coordinate Z moves with respect to the dimensional position x, so that x 0 is given by Z t. The concentration variables (which have units kg per total formation volume) are non-dimensionalized with the maximum possible concentrations they can attain. For instance, if there is Ck initially within the reservoir, this is C k, max , and the maximum possible concentration for heavy oil (C ho, max ) is 0:279C k, max according to reaction (1).

O C ho =C ho, max o C lo =C lo, max


K C k =C k, max

A:7 A:8

Y T =T inlet

Thus the equations are re-written as @Y @Y @2 Y E1 E2 @t @Z @Z2 @K YK @t @O 1 2 1 V RrdO r O P1 K P2 O Le @t @o 1 2 1 V Rrdo r o P2 o Le @t with initial conditions and boundary conditions A:9

G P1 P2

a
b

A:10 A:11 A:12

Z
l

m p1 r
f

PZ, 0 0, pZ, 0 0, YZ, 0 1, K z, 0 1, Yt, t 1


A.1. Permeability

A:13

Acknowledgments The authors would like to acknowledge the helpful discussions with Prof. Andrew Woods of the BP Institute, Cambridge. M. Youtsos acknowledges the nancial support of National Oil Shale Holdings LLC.

Of the several studies that simulated oil extraction from Green River formation shale (Bauman and Deo, 2011; Bauman et al.,

212

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

2010; Fan et al., 2009; White et al., 2010) initial permeability used varied from 0.003 mD to 10 mD. White et al. (2010) used a model which is a function of porosity and weight composition of the solid. This makes physical sense because it is expected that the permeability increases due to the pore pressure expanding the pore radii and due to the decomposition of kerogen which creates new pore space. The model from White et al. (2010) is employed, namely   1 1Ske Sco 3 f 1 k k0  A:14 2 1 f1Ske Sco where k0 is the initial permeability at formation pressure and Si is the formation volume saturation of a component i. A.2. Porosity Chen et al. (2009) model the porosity as

have been tted to the data set. These relations are


4 3 2 kr, g 93:954S5 o 238:07So 223:42So 91:395So 13:026So 0:41153

A:22
5 4 3 2 kr, o 100154:2S6 o 341:2So 286:3So 120:2So 26:3So 2:85So 0:117

A:23 where So is the volume fraction of the oil phase within the pore space. It is interesting to note that the gaseous and oil phase velocities are likely to drop by almost an order of magnitude because of their combined presence. This means it is crucial to ensure that oil ows as easily as possible so that the injected gas can also ow uninhibited. A.6. Oil compressibility The oleic phase compressibility is calculated by the following equation from Chen et al. (2009)

f0 f0 1 C R pp0

A:15

co

1433 17:2T 1180Y G 12:61API 100 000P b

A:24

The same rock compressibility factor as White et al. (2010) was used. A compressibility factor of 4.0 10 11/Pa was used, in line with typical Shale compressibility factors (Lee, 2010). f0 was modeled as being purely dependent on f0 , the initial porosity (prior to any decomposition) at the formation pressure to be investigated, and the solid volumetric composition of the formation. Namely

where YGS is the gas gravity (unit for air), Pb is the bubble point pressure and API refers to then gravity of the oil. A.7. Oil and gas viscosity The American Petroleum Institute gravity (API gravity) measures how heavy a petroleum fraction is compared to water. If the API is larger than 10 (1000 kg/m3), it is lighter and oats on water (Chen et al., 2009). The heavy oil is taken to be ca. 15 API (965.0 kg/m3) and light oil is taken as ca. 25 API (903.7 kg/m3). The BeggsRobinson equation (Beggs and Robinson, 1980) was used to calculate oil viscosities based on each fractions API and temperature. For simplicity, no gases were allowed to dissolve into the oil. Thus the dead oil viscosity formula from the BeggsRobinson equation was used. The relation is

f 1 f0 Ss C s =rs

A:16

where f0 is the initial porosity for a given formation pressure and subscript s refers to summing all solid species within the formation. A.3. Thermal conductivity of Shale The thermal conductivity model was based on 30 gal/tonne. As such, the empirical formulae for lateral and vertical permeabilities are based on Nottenburg et al. (1978):

mo 10C 1
where C 10C T 1:163 ,
0

A:25

lv 7 10 T 0:0078T 2:9505, R 0:8951 lh 5:9172 104 T 1:2775, R2 0:9873


Units are W/mK A.4. Specic heat capacity of Shale

7 2

A:17 A:18

C 0 3:03240:02023API

A:26

Qian and Yin (2010) provide separate relations for specic heat capacity for kerogen, coke and the inorganic components. The following are the heat capacity models for each component cp, ke 1753:6 3:3367T cp, in 687:27 0:92945T cp, co 1008:2 2:8276T J=kg K J=kg K J=kg K A:19 A:20 A:21

Formulas give viscosities in cP therefore must multiply by 1000 to get Pa s (SI unit). The gas viscosity calculation is a lot more complicated to calculate. From Chen et al. (2009): The gas viscosity mg was evaluated based on an estimation of the gas density using the real gas law (with a Z-factor correction). The pseudocritical pressure and temperature are corrected for non-hydrocarbon components. Here, mg was calculated by the LeeGonzalez correction (Dempsey, 1965):

mg

eF mc T red

A:27

where F is a function of the reduced temperature (Tred), the reduced pressure (Pred). These reduced properties are simply the normalization of each individual gas temperature and pressure by its critical temperature and pressure. The corrected gas viscosity mc is dened by Carr et al. (1954) as

A.5. Relativity permeability The presence of oil in the porous media will hinder the ow of gas and vice versa. There is little theory for how to describe the relative permeabilities of each phase vs. the saturation of those phases. Data has been provided by Chen et al. (2009) and curves

mc 1:709 105 2:062 106 Y G T Y CO2 6:24 103


8:188 103 6:15 103 logY CO2 9:08 103 logY G A:28 where YG is ratio of the gas mixture density to air density and Y CO2 is the mass fraction of CO2 within the gas mixture.

M.S.K. Youtsos et al. / Chemical Engineering Science 94 (2013) 200213

213

A.8. Phase fugacities The fugacity, f, of a real gas is the effective pressure which replaces the true mechanical pressure for chemical equilibrium calculations (Chen et al., 2009). The fugacity of component i is f i, a pa xi, a Fi, a A:29

where xi, a and Fi, a are the mass fraction and fugacity coefcient of a component i in phase a. For thermodynamic equilibrium, the fugacities of every hydrocarbon component to be distributed by the relation po xi, o Fi, o pg xi, g Fi, g A:30

for i 1, 2, . . . , N c . The phase mass fractions of each oil and gas fraction are given by xi, o and xi, g . Phase pressures of gas and oil are given by pg and po. The fugacity coefcients, Fi, a are functions of the gas deviation factor, Z a , from the PengRobinson equation of state, oi is the acentric factor of component i and nally kij , which is the binary interaction parameter between components i and j. The acentric factor of a component roughly expresses the deviation of the shape of the component molecule from a sphere (Fan et al., 2009). Necessary physical properties of the components used are detailed in Fan et al. (2009). References
Akkutlu, I.Y., Yortsos, Y.C., 2003. The dynamics of in-situ combustion fronts in porous media. Combust. Flame 134 (3), 229247. Bauman, J.H., Deo, M.D., 2011. Parameter space reduction and sensitivity analysis in complex thermal subsurface production processes. Energy Fuels 25 (1), 251259. Bauman, J.H., Huang, C.K., Gani, M.R., Deo, M.D., 2010. Oil shale: a solution to the liquid fuel dilemma. Modeling of the In-Situ Production of Oil from Oil Shale. American Chemical Society. (Chapter 7). Beggs, H.D., Robinson, J.R., 1980. Estimating the viscosity of crude oil systems. J. Pet. Technol. 27 (9), 11401151. Braun, R.L., Burnham, A.K., 1986. Kinetics of Colorado oil shale pyrolysis in a uidized-bed reactor. Fuel 65, 218222. Brown, P.N., Byrne, G.D., Hindmarsh, A.C., 1989. VODE, a variable coefcient ODE solver. SIAM J. Sci. Stat. Comput. 10, 10381051. Burnham, A., Braun, R., 1990. Development of a detailed model of petroleum formation, destruction, and expulsion from lacustrine and marine source rocks. Org. Geochem. 16, 2739. Burnham, A., Braun, R., 1992. Pmod: a exible model of oil and gas generation, cracking, and expulsion. Org. Geochem. 19, 167172.

Byrne, G.D., 1992. Computational ordinary differential equations. In: Cash, J., Gladwell, I. (Eds.), Pragmatic Experiments with Krylov Methods in the Stiff ODE Setting. Oxford University Press, pp. 323356. Carr, N.L., Kobayashi, R., Burrows, D.B., 1954. Viscosity of hydrocarbon gases under pressure. J. Pet. Technol. 6 (10), 4755. Chen, Z., Huan, G., Ma, Y., 2009. Computational Methods for Multiphase Flows in Porous Media. SIAM. de Paola, G., Kim, I.S., Mastorakos, E., 2009. Second-order conditional moment closure simulations of autoignition of an n-heptane plume in a turbulent coow of heated air. Flow Turbulence Combust. 82, 455475. Dempsey, J.R., 1965. Computer routine treats gas viscosity as a variable. Oil Gas J. 63, 141143. Fan, Y., Durlofsky, J., Tchelepi, H., 2009. Numerical simulation of the in-situ upgrading of oil shale. SPE J. 15, 368381. Farcas, A., Woods, A.W., 2007. On the extraction of gas from multilayered rock. J. Fluid Mech. 581, 7995. Jupp, T.E., Woods, A.W., 2003. Thermally driven reaction fronts in porous media. J. Fluid Mech. 484, 329346. Lee, J.M., 2010. The In Situ Hydrocarbon Extraction Process from Oil Shale. Ph.D. Thesis proposal. Ph.D. thesis. Colorado School of Mines, Golden, CO. Liu, Z., Jessen, K., Tsotsis, T., 2011. Optimization of in-situ combustion processes: a parameter space study towards reducing the CO2 emissions. Chem. Eng. Sci. 66, 27232733. Looney, M., Lestz, S., Hollis, K., Taylor, C., Kinkead, S., Wigand, M., 2010. Kerogen Extraction from Subterranean Oil Shale Resources. Technical Report US 7,789,164 B2, United States Patent Ofce. Mailybaev, A., Bruining, J., Marchesin, D., 2010. Analysis of in situ combustion of oil with pyrolysis and vaporization. Combust. Flame 158 (6), 10971108. Nottenburg, R., Rajeshwar, K., Rosenvold, R., DuBow, J., 1978. Measurement of thermal conductivity of green river oil shales by a thermal comparator technique. Fuel 57 (12), 789795. Phillips, O.M., 1991. Flow and Reactions in Permeable Rocks. Cambridge University Press. Qian, J., Yin, L., 2010. Oil Shale Petroleum Alternative. China Petrochemical Press. Rahm, D., 2011. Regulating hydraulic fracturing in shale gas plays: the case of Texas. Energy Policy 39 (5), 29742981. Sun, N., Sun, A., Gallagher, B., 2006. The forward and inverse problems in oil shale modeling. In: 26th Oil Shale Symposium Golden, CO, USA. Symington, W.A., Kaminsky, R.D., Meurer, W.P., Otten, G.A., Thomas, M.M., Yeakel, J.D., 2010. ExxonMobils electrofrac process for in situ oil shale conversion. Vinegar, H., 2006. Shells in-situ conversion process. In: 26th Oil Shale Symposium, Golden, CO, USA. Vinsome, P.K.W., Westerveld, J.D., 1980. A simple method for predicting cap and base rock heat losses in thermal reservoir simulators. J. Can. Pet. Technol. 19 (3). White, M., Chick, L., McVay, G., 2010. Impact of geothermic well temperatures and residence time on the in situ production of hydrocarbon gases from green river formation oil shale. In: 30th Oil Shale Symposium, Golden, CO, USA. Woods, A.W., 1999. Liquid and vapour ow in superheated rocks. Annu. Rev. Fluid Mech. 31, 171199. Wright, Y.M., Boulouchos, K., de Paola, G.E.M., 2009. Multi-dimensional conditional moment closure modelling applied to a heavy-duty common-rail diesel engine. SAE Int. J. Engines 2, 714726.

You might also like