Professional Documents
Culture Documents
Electrical Heating
K. VINSOME
Dyad Engineering Inc.
B.C.W. McGEE
McMillan-McGee Canada Company
FE.VERMEULEN,FS.CHUTE
University of Alberta
Power
Conditioning
Unit
Current Flow
----,~
Surface
---~~----~
Overburden
Reservoir
Underburden
5 meters
Electrical Resistance Heating Figure I describes how the process works. The specific system
described here is similar to Gill, 1969; Spencer, 1988; and Rice,
Electrical heating is a thermal process which can be applied to 1992.
a well to increase its productivity. The productivity increase is
The essential components of an electrical heating system are:
substantial and comes about because of the removal of thermal
adaptable skin effects (visco-skin for example) and the reduction • power supply,
of the oil viscosity in the vicinity of the wellbore. Salient features • power delivery system,
of the process are: • electrode assembly, and
• It is a continuous, not a cyclic process. Electrical heating • ground return.
occurs simultaneously with production of fluids. The variable frequency (2 to 60 Hz), power supply (Isted,
• Low frequency power (not microwave frequency) is used. 1992), is capable of delivering up to 300 kW of power. The power
• All the downhole equipment can be contained within a delivery system may consist of tubing, cables or a combination of
single wellbore. both. The electrode assembly consists of bare casing pipe with
The visco-skin is a zone of high oil viscosity that develops in where the variable \jf is the root mean square (rms) value.
the low-pressure region near the wellbore. It occurs in most natu- The electrical conductivity is given by an Archie type equation:
rally producing oil wells, but is especially prevalent in saturated
heavy oils of 10 to 24° API gravity (McGee, 1989). Visco-skin S2
can best be described by reference to Figure 2. 0,=~nO'w---..l£.+(1-~)·O'r
a (3)
Radially approaching the wellbore from the reservoir, the pres-
sure decreases rapidly to the producing pressure. As the pressure where the electrical conductivity of the water and rock, O'w and O'r
drops, more and more gas evolves from the oil into the gaseous are temperature dependent functions.
phase. A result of gas evolving from the oil is a viscosity profile
like that shown in Figure 2. The oil viscosity reaches a maximum
at the wellbore and decreases rapidly to the original oil viscosity Model Formulation
in the reservoir. The region of high oil viscosity usually extends Equation (1) is of a form familiar in reservoir simulation, and
only 1 to 2 meters into the reservoir. can be treated as such using standard finite differencing tech-
As a result, the flow is impaired. The magnitude of the produc- niques. However Equation (2) is more problematical because of
tivity decrease (visco-skin) depends on the ratio of the oil viscosi- the quadratic nature of the way the potential enters into it.
ty at the wellbore to live oil viscosity, (viscosity parameter PIl). In Standard forms of differencing on this equation can lead to signif-
heavy oil reservoirs, PIl is typically greater than 10 and the pro- icant errors in the calculation of the energy generation within a
ductivity decrease caused by visco-skin is typically two to three grid block. The best way we have found to eliminate this problem
times (McGee, 1991). .is to avoid finite difference forms of (2) altogether, and use
30 The Journal of Canadian Petroleum Technology
Which using the preceding relationships can be shown to
integrate to:
-2h (\IIm-\IIi)2
Qe - n cr i 1 rm
n-
ri (10)
1= -2nrhcr d\ll
Iw ~ 21thcr ('InI'r '1';)
Gt'1
rw (11)
dr (4)
where G is the geometrical factor, in this case equal to
Let i and j denote two adjacent grid blocks, and let 'lfm be the 11:
cri\lli + cr j\ll j
Qe = I w '(\IIw -\IIi) (12)
In(~) In( ~)
\11m = cr. cr· Boundary conditions at wells can be specified as either current
__ I_+ __J _ or potentials, in much the same way as rates or bottom hole pres-
In( r~) In( ;~ ) .......................................................................(6) sures can be imposed for regular wells. Automatic switching
between the constraints is also available, so that for example a
(\IIj-\IIi) well can have a current constraint Iw subject to a maximum bot-
I =- 2n rmh cr eft -'---'----'-- tom hole potential 'lfw' The techniques that have been developed
L1 (7) for multicompletion wells are also directly applicable, and have
been implemented.
where ~ is the spacing between the grid nodes, and
Solution Procedure
L1 rm
_ _ _ =_r_i + ~
In In.!l The solution of the finite difference equations for the electric
rmcreft. cr i crj (8)
potential is decoupled from the reservoir fluid flow and energy
conservation equations. The procedure is straightforward and very
efficient. It involves the following steps:
Equation (7) is the finite difference form for the current
1. Calculate the conductivity explicitly at the beginning of each
between two grid blocks. Because of the radial geometry the
time step.
effective conductivity (8) for calculating the interblock transmissi-
2. Solve the finite difference equations for the electric potential
bility is a harmonic radial average. Other geometric shapes give
other averages, but they are all some variation of a harmonic aver-
'If. This is done once per time step.
3. Evaluate the energy generation rate Qe in each grid block.
age. In the absence of sources and sinks, the finite difference form
of (1) is a sum of terms like (7) over all grid block faces. This is also only required once per time step.
The rate of energy generation in grid block i between points i 4. Use the Qe as source terms in the energy equation and solve
the reservoir flow and energy conservation equations in the
and m is then given by:
normal fashion (usually fully implicitly via a Newton-
Raphson iteration).
r 2
No stability problems have ever been encountered due to the
J
Qe = 2nrhcri ( ~~) dr explicit calculation of the conductivity. It appears to be a perfectly
ri (9) adequate approximation for all reasonable ranges of conditions
where ETotal is the total internal energy in the grid block, ~t is the A.exp[_D)
time step, cxunits is the unit conversion factor, and L is a length 1 Tabs - E
K - - - - -----'-,.------'-
characteristic of the grid block. It has been found by trial and error - x(r) - p(r)
..........................................................(17)
that if:
where x is the mole fraction of gas component in the oil phase.
L = min( Llx, Lly, Llz) (15) Thus at constant temperature, combining Equation (16) with
Equation (17), results in the following mole fraction distribution
then exactly the same convergence tolerance can be chosen as for away from the wellbore:
the normal simulator conservation equations.
Semi-analytical Model
TETRAD is compared to a semi-analytical model developed
for scouting electrical heating for heavy oil type calculations
(McGee, 1990). The model takes into account the visco-skin. The
viscosity, pressure and temperature dependencies are coupled with The above equations can be rewritten in the traditional form of
Darcy's steady state radial-flow and transient-energy equations. Darcy's radial flow equation and a visco-skin:
The energy equation is solved analytically. Initially a steady state
pressure distribution is assumed. A pressure and temperature per-
turbation is introduced into the equations and the flow rate 2nkh(Pe - Pwj )
Qo = ----;::-'-------"----:0-
obtained from the perturbed solution. The procedure is iterative
/l wBo [In( ;: )+ svs ]
until the flowrate converges to a solution. . (24)
c2 = _ e J [ (\If wb - \If re ) 1 over the length of the system with Gaussian quadratures. This
process continues until convergence is achieved.
(pC)t 1n( ;: )
.............. ; (30)
Then the solution to Equation (28) is,~ Analytical Comparison and Field Case
Numerical Simulation Study
C2 (r)
T(r,t)=T * (r,t)+T re +~ln-f Two sets of data are summarized in Table 1. Data associated
1 (31) with the "Model Case" are used to compare the numerical simula-
tor to the semi-analytical model. The "Field Case" data are used to
T*(r,t)=-C2 1n [ re ] compare the numerical calculations to a field test of the electrical
cl ~r2 +2c1t heating process. The Field Case data are obtained from the Frog
..........................................................(32) Lake Well llD-15-56-3W4M.
if
Semi-analytical Comparison
The purpose of this comparison is to demonstrate the utility of
the visco-skin equation [Equation (25)] when used in combination
with TETRAD. This example is for simulation of electrical heat-
Otherwise, ing in a reservoir with visco-sIan.
The utility of Equation (25) comes from being able to model
the visco-skin problem using a single component oil in TETRAD.
T*(r,t)=O.
A constant oil viscosity as a function of pressure and thus radius
during primary production results. Therefore, to account for the
The flow rate term in Equation (29) is determined iteratively visco-skin resultant of a varying oil viscosity in the reservoir, the
because of the dependency of viscosity on temperature and pres- wellbore radius and the drainage radius are used in Equation (25).
sure. However, the temperature distribution is not known, so the If the oil phase is defined to vary as a function of pressure in
flow integral must be solved numerically. The following equation TETRAD, then the wellbore radius and the radius of the first grid
is used to calculate the flow rate with the flow integral in the block are used to define the visco-skin. The oil viscosity Ilw, used
denominator: in Equation (25), and TETRAD is the estimated oil viscosity at
1"\
.... 100
~
~
l!! 80
60
"'.; -x- TETRAD
---+-
_x_
TETRAD
Field
Analytical
Q)
c.
40
i---- ---+-
E
Q)
I- I
20
I
o +--- 50 100 150 200 250
o 4 10
Production Days
Radius [meters]
FIGURE 4: TETRAD and analytical model temperature FIGURE 5: Production history of the field case well and TETRAD
distribution. match.
TABLE 2: TETRAD-analytical model comparison. The extent of the visco-skin during primary production had to
be calculated. The simulator was set to compositional mode and
Parameter TETRAD Analytical Model the oil viscosity distribution after 153 days was calculated and is
Primary Production 3.60 m3/day 3.60 m3/day shown in Figure 6. As shown in the figure, the region of high oil
viscosity is within one meter of the wellbore. The viscosity para-
Stimulat$d Production 17.32 m3/day 17.06 m3/da.y
meter is assumed to be about ten, which is based on the experi-
Productivity Factor 4.81
mental work of Beal (Beal, 1946). History matching the primary
production is based on the cumulative oil production, and is
initial reservoir conditions (the live oil viscosity with dissolved shown in Figure 5.
gas at initial reservoir pressure) The procedure is straightforward: At the onset of electrical heating, more than a threefold
1. Construct the grid, fluid, rock, and thermal properties to be increase in production was observed in the field. This was
consistent with the analytical model, (note that heat loss is achieved at initial power rates of less than 5 kW. When the elec-
not accounted for in the analytical model, so is disabled in trical heating option was turned on in the simulator, a production
TETRAD). response match was attained. The simulator verified the removal
2. Calculate the visco-skin factor using Equation (25). of a visco-skin as a mechanism of production stimulation.
3. Set the skin value in the wellbore model of the simulator to The actual productivity in the field was more than 4 times the
the calculation made in Step 1, primary production. The calculated was close to 3112 times prima-
4. At the onset of electrical heating set the skin value to zero ry. The discrepancies in the calculations with the field data are
(the high temperature around the wellbore caused by electri- probably from the assumptions regarding the viscosity parameter,
cal heating has dramatically reduced the oil viscosity there other reservoir and fluid properties and the presence of skin other
and therefore removed the visco-skin), than visco-skin. What is deemed important is the response of the
The results of the calculations demonstrate the accuracy of the well to electrical heating and verification by numerical simulation.
simulator when compared to the simple analytical model. Both It is important to estimate the operating temperature of the elec-
results are essentially the same (11/2 % difference) as shown in trode during electrical heating since the downhole equipment may
Table 2. fail at temperatures above 100°C. Figure 7 shows the calculated
Very little difference is found in the temperature profiles temperature distribution in the reservoir around the electrode.
between the numerical simulation calculations and analytical as These calculations are based on oil flow of 10 m3/day and input
shown in Figure 4. power of 30 kW. Since the flowrate changes during the life of the
It is noted that heat conduction in the analytical model is well, a curve showing the input power necessary for an electrode
assumed to be negligible in comparison to the other heat transfer temperature of 100°C for various flowrates is required. This curve
mechanisms; electrical heating and convective heat transfer. is shown in Figure 8, and is referred to as the P-Q Curve (Power
TETRAD does account for conductive heat transfer from the well- Flowrate). Operating the system above the line will result in peak
bore which may be significant in electrical heating of a horizontal temperatures greater than 100°e.
well, where the fluid flow velocities are much smaller and heat The analytical model predicts a productivity increase of 4.1 for
transfer by conduction may be comparatively significant. our Field Case. The difference between the analytical and simula-
tion calculation is the assumption of a uniform temperature distri-
Field Case Comparison bution along the entire length of the electrode for the case of the
analytical model. In actuality, the power intensity is greater at the
The well was drilled into the Sparky formation in the Frog ends of the electrode because of its finite length and a temperature
Lake area and completed for electrical heating in June 1988. The distribution such as shown in Figure 7 is more realistic. Thus the
oil there is heavy and oil can be produced under primary condi- average temperature of the electrode is lower and so the produc-
tions. Figure 5 shows the production history of the well. Peak pro- tion increase calculated will also be lower These differences are
duction was 7.1 m 3/day and declined to 3.0 m3/day before electri- consistent with calculations made by Baylor (Baylor, 1990). The
cal heating. The well produced for 153 operating days and then analytical model calculates an optimistic productivity increase.
was electrically stimulated Immediately the production rate
increased to over 12 m3/day. The input power during stimulation
averaged 15 kW.
The development and subsequent removal of the visco-skin in Conclusions
the near-wellbore region is one explanation to account for the oil A reservoir simulator, TETRAD, has been modified to incorpo-
production during primary production and the rapid increase in rate the electrical heating equations. The simulator includes treat-
production after a short period of thermal stimulation. TETRAD, ment of the electrical conductivity as a function of temperature,
in 2D radial coordinates, was used to match the production salinity, and saturation. The simulator was validated against ana-
response of the well during primary production and electrical lytical calculations and field data. It has been used to design sev-
heating. As well, the simulator was used to determine the power eral electrode completions and assist in developing operating
requirements to maintain an electrode operating temperature of strategies for field implementation of the electrical heating
100°C for varying flow conditions. process.
34 The Journal of Canadian Petroleum Technology
Top of Reservoir
.2 .4 .6 .8 1.0 Meters
8
13,500 cP"
50 -
~
~
.-/
6 1,500 cp 40 11 Peak Electrode Temperature> 100 Ci ~
Electrode &
i ./"
~
Perforated ~ 30 V
Interval 4 =.. V
Lines of Constant
~ 20
V
Oil Viscosity a.
10
/ IPeak Electrode Temperature < 100 g
2 /V J
o
o 2.5 7.5 10 12.5 15 17.5 20
h thickness metres
I current amps
Top of Reservoir
Iw well current amps
.2 .4 .6 .8 1.0 Meters J current density amp/m3
8 L characteristic grid block length metres
\