Note: Descriptions are shown in the official language in which they were submitted.
CA 02436158 2011-06-14
1
Title: Heavy Oil Extraction Test Chamber with Configurable
Temperature Profile and Feedback Control
FIELD OF THE INVENTION
This invention relates generally to the field of heavy oil or tar
sand extraction, and more particularly to experimental techniques
and methods that may be used to model temperature sensitive insitu
extraction processes. Most particularly this invention relates to
equipment and methods to model and test, on a small scale, insitu
extraction processes for heavy oil and tar sands.
BACKGROUND OF THE INVENTION
This invention relates generally to the extraction of heavy oil
and bitumen. Heavy oils are crude oils, which have high specific
gravity and viscosity and are therefore difficult to extract
commercially because they do not readily flow. Tar sands are
geological formations in which heavy oil is trapped within a sand
formation. Achieving insitu separation of the heavy oil from the sand
is a well-known and difficult challenge.
Currently steam is the dominant thermal fluid used for insitu
recovery of bitumen and heavy oil. Injected steam raises the
temperature of the bitumen thereby reducing its viscosity and
allowing it to flow more easily. Steam extraction is subject to a
number of problems including high heat losses, clay swelling
problems, thief zones, water-oil emulsions, capillary surface tension
effects, lack of confinement for shallower zones and disposal of large
quantities of environmentally damaging salt and organic acids as a
consequence of boiler feed water purity requirements. By some
estimates, with the best currently available technologies, only 10% of
the original bitumen resource in the Athabasca tar sands are
economic to extract.
Thermal recovery processes, using steam, also require large
amounts of fuel to be burned to produce the steam and can emit
enormous amounts of greenhouse gases such as carbon dioxide.
Estimates published by Natural Resources Canadal show CO2
emissions of about 70kg/bbl for bitumen production and a total of
Can3da's Emissicn,; Outlook: an Update, December 1999,
Annex [3, pg B 6, Avail ab 1 e at www.nrcan gc calesiccoluptiote
CA 02436158 2011-06-14
2
about 120 kg/bbl for synthetic crude (i.e. upgraded bitumen usually
derived from surface mined bitumen).
Recent estimates released by the Alberta Energy Utilities
Board2 and the Canadian Association of Petroleum Producers3,
predict that bitumen (and synthetic crude) production rates will be 2
to 2.6 million bbl/day of bitumen by 2010. This level of bitumen
production will produce at least 140 million kilograms (=70x2million)
of CO2 emissions per day (i.e. 300,000,000 to 700,000,000Ibs CO2
per day depending on fuel source and the proportion of insitu vs
synthetic crude production).
Solvent extraction processes have been proposed as an
alternative to steam extraction processes. One such process is the
N-Solv process (Canadian Patent Applications, 2299790, 2351148,
2374115). However, the physical chemistry of the bitumen
extraction in solvent gravity drainage processes is not very well
understood or characterized. For example, Dunn4 et al first reported
in 1989, that for a cold solvent extraction process the measured CO2
diffusion rates in the tar sands were a factor of 460 times higher than
the theoretically predicted values. This unexpected result has been
observed and reported by many subsequent researchers using a
variety of different solvents and crude oil samples and yet the
underlying physical mechanism is still not understood.
Several computer models have been developed to predict the
extraction rates for gravity drainage bitumen extraction using solvent.
However, these computer models do not appear to be capable of
accurately describing the insitu processes. One potential problem of
such models is a lack of spatial resolution because the models are
typically far too coarse to accurately model the solvent concentration
gradients. For example, lab studies by Fisher5 have revealed that
the solvent-bitumen interface is only a couple of millimeters thick. An
appropriate gridblock size to accurately model insitu concentration
gradients should be perhaps 10 times smaller (i.e. 100-200 microns).
With typical grid block sizes of ¨0.5 m used in computer modeled
2 Alberta's Reserves 2000 and Supply/Demand outlook 2001-
010, Alberta Energy Utilities Board
- Canada's Oil Sands Development delivered by Eric Newell,
Chairman & CEO, Syncrude Canada. Available at
http://www.capp.ca/
4 Dunn, Nenniger and Rajan, A Study of Bitumen Recovery by
Gravity Drainage Using Low Temperature Soluble Gas
7njection, Canadian Journal Of Chemical Engineering Vol 67,
Dec 1989, pg 985
Fisher et al, 'Use of Magnet::: Resonance imaging and
Advanced Image Analysis as a Tool to Extract Information
from a 2D Physical Model of the Vapex Process", Society of
Petroleum Engineers Paper 59330, April, 2000
CA 02436158 2011-06-14
3
reservoir simulations (see Nghiem6), the calculated concentration
gradients in the reservoir simulators are about 500 times smaller
than values actually measured in the laboratory tests. For a 3D
computer model with an appropriate spatial resolution the number of
calculations increases by a factor of 50003 (=125,000,000,000),
which increases the model run time for a given scenario to an
unworkable duration. Since the solvent concentration gradient
provides the primary driving force for solvent penetration and
extraction, existing computer models have a significant problem in
accurately representing the process.
As noted earlier, researchers consistently measure bitumen
extraction rates with solvents that are much higher than expected.
Thus, we believe, it is necessary to use physical models (i.e.
experiments) to obtain meaningful data on bitumen yield, extraction
rate and bitumen quality. Furthermore, until the details of the solvent
extraction mechanism are better understood, we believe that it is
unrealistic to expect credible predictions from the existing computer
models.
Due to the complexity of physical processes (combined heat,
mass and momentum transfer with simultaneous asphaltene
precipitation) it may not be possible to ever develop a fully rigorous
theoretical computer model. However, empirical models can be
developed that are both accurate and useful. Such empirical models
typically require data from a large number of representative physical
experiments to be able to develop parametric sensitivities to process
variables. This type of experimentation is expensive and time
consuming, but has provided the basis for many (if not most) useful
chemical engineering processes. However, it is necessary to conduct
physical experiments which accurately represent the specific
physical processes of interest and it is necessary that the same be
accurately measured before a meaningful empirical model can be
developed.
The prior art experimental apparatuses and techniques in the
tar sand extraction field are generally intended to simulate a small
two dimensional slice of a reservoir. These experiments are typically
conducted in a thin walled rectangular can that is packed with tar
sand (see Dunn4 and Frauenfeld7) with physical properties relevant
" Nghlc-m et. al "Modeil..ng AsphatLene Precipitation and
Dispersive Mixing in the Vapex Process, SPE paper 66361,
,.gure 2
Frauenfeld t a ,
Processes for Alberta Heavy Oil Reservoirs, Journal ot
Cahadian Petroleum Technology Vol 37 co 4, 7998
CA 02436158 2011-06-14
4
to the reservoir of interest. A simulated injector well is usually located
above a simulated producer well at one end of the can. The can is
placed within a pressure vessel and external pressure is applied to
the can to mimic the overburden stresses appropriate to that
reservoir.
Tar sand extraction processes are typically based on some
type of thermal effect and therefore appropriate consideration of
thermal effects in the experiments is important. Two aspects of
thermal behavior have been identified that can greatly affect the
experimental modeling. First, there is a need to mimic to temperature
profiles and temperature gradients within the tar sand which arise
due to the thermal characteristics of the reservoir extraction process.
Second, heat may be lost through the conductive nature of the can
or sample holder of a typical experimental apparatus with the
consequent distortion of the temperature profiles within the
sandpack. Such heat loss is referred to as parasitic heat losses,
Parasitic heat loss is an ongoing problem with all thermal
gravity drainage experiments. Typically, SAGD researchers have
used the dimensional scaling criteria of Butler, to work around this
problem. Butler's scaling criteria predicts that by increasing the tar
sand permeability, the time scale can be compressed (i.e. 1 hour of
experimental time corresponds to 1 year of field time). Thus, scaled
experiments minimize the impact of parasitic heat losses by greatly
reducing the experimental time. Butler has used the analogy
between heat transfer and mass transfer to develop similar scaling
criteria for solvent processes8. However, as noted above, the
solvent extraction mechanism is not well understood so the scaling
assumptions of Butler's solvent model are in doubt. Thus a different
approach from the prior art scaling assumptions of Butler is needed.
Fig. 1 is based on prior art and illustrates the problem that the
present invention seeks to address. Figure 1 shows transient (one
dimensional) temperature profiles at different times for a section of
tar sand initially at 8C when one edge is suddenly heated to 50C. in
Figure 1 zero on the x-axis represents a bitumen interface. Figure 1
shows the temperature profiles along a 60 cm section of tar sand
initially at 8C after time intervals of 1 minute, 2 hours, one day, three
days and seven days from when one edge is suddenly heated to
50C. The physical properties of the tar sand and the temperature
profiles were calculated using the data and formulas presented by
3u,lor et_a__ A Nw Pocess (Vapex) for Recovering Heavy
Oil3 by Using 'Hot Hater and Hydrocarbon Vapour, Journal of
C-inadian Pelroleum Engineer mg Jan-Feb 19)1, vol 30, No .1
CA 02436158 2011-06-14
5
Birre119. Figure 1 also shows a temperature profile expected for a
continuous bitumen extraction process taking place in 8C tar sand
with the interface heated to 50C and an assumed extraction rate of
5cm/day again using the formulas presented by Birre119. This latter
temperature profile is referred to as a quasi-steady state profile, as it
is not expected to change further over time (since the x-axis origin is
the bitumen interface). Figure 1 shows that it takes a period of seven
days for a sample sandpack to acquire a smooth temperature profile
(assuming no parasitic heat losses out the sides of the apparatus),
but even then it does not have the same temperature profile
predicted for "quasi-steady state" operation. This figure shows that
accurate process measurements cannot be made until the
temperature profile no longer is changing over time, which even in a
small sample can take a very long time to be established.
At quasi-steady state conditions, the bitumen interface moves
with a constant velocity and the solvent condenses at a constant rate
determined by the temperature gradient (i.e. conduction heat loss) at
the bitumen interface. If we consider the temperature gradient at the
bitumen interface, (i.e. slope of the temperature profile at x=0), then
Figure 1 shows that temperature gradient is far too high, and
consequently the solvent condensation rate will be substantially in
error for the first seven days of an experiment.
In addition to the problem of achieving quasi-steady state
temperature profiles, the parasitic heat losses can be 10-100 times
larger than the expected heat delivery rate. If solvent condensation is
the only source of heat in the experiment, then these parasitic heat
losses result in solvent condensation rates 10-100 times too high.
High solvent condensation rates are undesirable and can lead to a
host of complications including flooding of the vapour chamber with
liquid and destabilization of asphaltenes. Thus, it is important to
minimize the parasitic heat losses and correctly approximate the
quasi steady-state temperature profiles for an accurate simulation to
occur.
Thus, in the absence of appropriate scaling assumptions,
which shorten the time of the physical experiments, real time
experiments are required. In real time experiments, temperature
effects become of much greater concern and represent significant
limitations on experimental accuracy. What is needed is an
experimental technique and apparatus in which the parasitic heat
Pitrel', Heat Transfer Ahead of a SACD Steam Chamber: A
Study o Thermocouple Data From Phase B of the Underground
Test :Facil3Ly ;Dover Project), Journal of Canadian Pe-roleum
Technology, March 2003
CA 02436158 2011-06-14
6
losses and temperature profiles are controlled in a way that permits
measurements to be made which accurately reflect insitu
circumstances, without an undue amount of time being required.
The data generated by such techniques can then be used to develop
accurate empirical models.
BRIEF SUMMARY OF THE INVENTION
The present invention teaches method and apparatus for
testing solvent extraction processes for the tar sands. Such testing
is desirable to determine the impact of changes in process
conditions on the bitumen yield, extraction rate and the degree of
upgrading for among other things, heated solvent processes. An
object of the present invention is to permit real time testing to
occur, without needing to run each experiment for many days
before experimental sampling can even begin.
What is desired is a means to configure temperature profiles
within testing samples to simulate predetermined insitu temperature
profiles. In this respect predetermined temperature profiles can be
selected to, for example, minimize parasitic heat losses and to also
minimize the amount of time required to establish the preferred
temperature profiles. Thus an object of the present invention is to
permit relatively quick and accurate measurements to be made of
solvent extraction processes.
An aspect of the present invention is to provide a method and
apparatus that establishes temperature boundary conditions for an
experimental sample undergoing a proposed extraction process,
which properly mimic an insitu section of the reservoir. Thus, rather
than using a scaling assumption to overcome such temperature
sensitivities as in the prior art, the present invention provides a
method of configuring a temperature profile of a sample being tested
to simulate its insitu properties.
In one preferred embodiment the present invention provides a
heat configurable sample holder, for holding the samples during the
course of the experiment. The sample holder may, for example, be
provided with an outer shell that incorporates individually controllable
and localized heaters, each of which has thermal contact through the
sample holder with the sample. Each of the heaters can be in the
form of an electrical resistor, which is periodically energized to supply
heat and a temperature sensor to measure the heater temperature.
Each heater is preferably mounted on a thermally conductive base,
which may take the form of a small tile. Each tile has sufficient
thermal mass that the temperature fluctuation through a duty cycle is
CA 02436158 2011-06-14
7
appropriate for the accuracy and precision of the temperature
sensor. The tiles have high thermal conductivity to individually
achieve a uniform temperature, but most preferably are thermally
insulated from any adjacent tiles so that temperature gradients along
the sidewall of the sample holder can be maintained without
excessive heat loss. Thus the present invention provides a heat
configurable sample holder, in which the temperature profile can be
set according to any predetermined pattern, where the resolution of
the temperature pattern is determined by the size of the individual
tiles or conductive heater bases.
The present invention also provides for an array of
temperature sensors placed within the tar sand sample being tested
which produce an output enabling the temperature profile of the
sample to be accurately measured. The sensed temperature profile
is then processed through a mapping algorithm to determine a
smoothed temperature profile within the sand. This temperature
profile is used to establish the desired temperature for each
individual tile. Thus an object of the present invention is to provide a
control system which supplies an appropriate amount of power to the
individual heater elements to allow the sample holder temperature
profile to match the local internal sample sand temperature profile
and thereby minimize any parasitic heat losses.
Therefore, according to an aspect of the present invention
there is provided a method of testing, in a lab, an insitu extraction
process comprising the steps of:
placing a sample to be tested in a sample holder
having a configurable temperature profile;
placing the sample holder in a pressure vessel;
increasing the pressure in the pressure vessel to
simulate an overburden pressure;
configuring the temperature profile of said sample
holder to match a desired temperature profile,
passing a solvent into said sample, and
measuring one or more parameters of said oil
extraction process.
In another aspect the present invention provides for:
configuring a temperature profile of said sample holder
to match a measured internal temperature profile of a sample
arising during the application of an extraction process to said
sample.
According to yet another aspect there is provided a testing
device to conduct oil extraction process experiments on a sample
CA 02436158 2011-06-14
8
to be tested, said testing device comprising a temperature
configurable sample holder to provide a desired temperature
profile to said sample.
The concepts taught in this patent may also have application
in enhancing recovery of both heavy oil and less viscous oils.
BRIEF DESCRIPTION OF THE DRAWINGS
Reference will now be made, by way of example only, to
preferred embodiments of the invention as illustrated in the
accompanying drawings and in which:
Figure 1 shows the (one dimensional) temperature profiles
along a 60 cm section of tar sand initially at 8C after time intervals
of 1 minute, 2 hours, one day, three days and ten days from when
one edge is suddenly heated to 50C;
Fig. 2a shows a schematic of a section of the heater tile array
for configuring temperatures according to the present invention;
Figure 2h shows a side view of the heater tile array of Figure
2a;
Fig. 3 shows a more detailed view of sample holder having a
sample sandpack and the placement of insitu temperatures sensors
according to the present invention;
Fig. 4 shows a schematic of a power circuit to energize
individual tiles according to the present invention;
Figure 5 shows a schematic of a data acquisition system and
the means whereby a particular heater element interacts with the
data acquisition system;
Figure 6 outlines the temperature control algorithm during
startup for the present invention;
Figure 7 outlines a tile temperature control algorithm during
operation after the solvent vapour is injected into the sample
sandpack; and
Figure 8 compares the target "quasi-steady state"
temperature profiles within a sample sandpack to predicted
temperature profiles along a side wall of a sample holder of the
present invention. Temperature profiles are shown at startup and
again after 4 days, assuming an interface velocity of 5cm/day.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Fig. 1 is calculated from the prior art as described in the
background and illustrates the problems that the present invention
seeks to address. Figure 1 shows transient (one dimensional)
temperature profiles at different times for a section of tar sand initially
CA 02436158 2011-06-14
9
at 8C when one edge is suddenly heated to 50C, for example by a
condensing solvent process. Figure 1 also shows a "quasi-steady
state" temperature profile for a bitumen extraction process taking
place in 8C tar sand with the interface heated to 50C and an
extraction rate of 5cm/day. A physical model experiment must
reproduce the appropriate "quasi-steady state" temperature profile,
to accurately model insitu processes and thus to generate
meaningful process test data. It will be understood by those skilled
in the art that Figure 1 only shows one-dimensional temperature
profiles rather than the full 2D temperature map for the sake of
clarity.
At "quasi-steady state" the sample has attained a temperature
profile that mimics an insitu profile. In figure 1 the temperature profile
of the quasi steady state curve is of constant shape, and even if the
process continues, the shape will not change. The origin represents
a bitumen interface, where the solvent vapour is condensing, and in
absolute terms this interface will be moving at an estimated speed
corresponding to the rate of extraction. However, even as the
interface moves through the sample (mimicking movement through
the reservoir in an actual extraction) the shape of the temperature
profile extending out from the interface remains the same. In this
specification the term "quasi-steady state" identifies a sample
condition that will permit process testing to begin with meaningful
measurements of process parameters being possible. Process
parameters means any measurable aspect of a process that helps in
defining the value properties or efficacy of a process, such as
extraction rate, bitumen quality or the like. The term sample shall
mean an actual sample of an insitu oil bearing formation, or a made
up sample which is provided with certain characteristics to mimic the
insitu reservoir conditions, or any other configuration of materials
which may be used to provide tests on the process of interest.
Typically, a sample will comprise a graded sand which is water
wetted and oil saturated in a known manner.
Fig. 2 shows a schematic of a section 10 of a temperature
configurable sample holder according to the present invention. The
section 10 shows a plurality of heater elements 14 embedded into
heat conductive tiles 16. As shown the heating array may take the
form of 2x2cm tiles on a 2.5cm pitch. Each tile 16 is shown with an
embedded resistor heater element 14 that is energized as needed to
control the temperature of the tile 16. Although one heater per tile is
shown, it will be understood that the present invention comprehends
that more heaters may be added if needed to provide the desired
CA 02436158 2011-06-14
10
amount of heat. However, one per tile is preferred. Each tile 16 is
also equipped with a temperature sensor 15, which is preferred to be
a type T calibrated thermocouple or the like.
The most preferred form of heater element 14 is a resistor
heater element. A 100ohm wirewound resistor provides adequate
results. The type and properties of the heater can be varied, but
what is needed is a heater power matched to a tile size to ensure
that the local temperature within the apparatus can be matched by
the tile temperature. For the purposes of the invention the resistor
may be operated at power levels exceeding its' nominal rating.
Destructive tests have shown that this is feasible, if the duration of
the heating time is limited and the resistor is adequately thermally
potted to limit temperature rise within the resistor. Consequently in
the preferred form of the invention illustrated the resistor is placed
within the tile and potted with a thermal conductive epoxy or the like
such as Supertherm 2005 manufactured by Tra-Con inc of Bedford
Massachusetts. Other configurations of heater element and tile
combinations, such as surface mounted heaters that provide the
advantages of a temperature configurable sample holder are also
comprehended.
The tiles 16 are shown mounted in an insulating matrix 20,
made from, for example, silicone rubber or the like. The tiles are
separated by a gap of, for example, 5nnm which is large enough to
reduce the heat transfer from one tile to the next but small enough to
avoid excessive temperature drop (sag) between adjacent tiles. In
some cases it is desirable to reduce the contact area of the tile 16
with the steel wall 32. Reduced contact area reduces the amount of
conduction heat flux along the steel wall 32 without increasing the
gap between the tiles. We can obtain the reduced contact area with
a "knob" profile 17 at the contacting surface of the tile. The knob can
be any convenient shape. Conventional wiring (not shown) connects
the heater elements to an appropriate power source. Although any
number of tiles may be used, it is estimated that 1024 tiles on a 2.5
cm pitch is sufficient to completely enclose a sample sandpack that
is 32 cm high 62 cm long and 12 cm wide (i.e. 13 x 25 x 5 tiles).
While it is preferred to make the array as regular as possible, various
openings may be required in the array to accommodate
thermocouple penetrations and whatever feed and drain ports that
may be needed.
Fig. 3 shows a more detailed isometric sectional view of the
both the sandpack 28 and tile assembly. In Figure 3, the sand pack
28 has temperature sensing thermocouples 30 which are
CA 02436158 2011-06-14
11
preferentially placed on a uniform spacing (i.e. 5 cm spacing).
Between the sand pack 28 and the wall of the sample holder 32 is a
rubber liner 34 or the like to smooth the adjacent tile temperatures
out so that the step profile of the tiles more nearly approximates the
desired temperature profile of the sandpack 28. The liner 34 also
helps to dampen out fluctuations arising from the power duty cycle.
The layer of heater tiles is immediately outside the wall 32 and is
physically clamped to the wall 32 to provide good thermal contact
between the tile assembly and the wall. This thermal contact is
important to ensure that each tile can determine the local
temperature of the side wall 32 of the sample holder. Beyond the
tiles are several external layers which include the packaging 35
(wiring and mounting for the tiles) as well as insulation 36 to reduce
the heat delivery requirement of the tiles.
The tiles are preferably made of a high conductivity material
such as aluminum or copper or the like, so they will individually
maintain a fairly uniform temperature across an individual tile. The
smooth temperature profile in the tar sand undergoing stimulation
is thus approximated by a sample holder temperature profile having
a number of steps, each step representing a different tile. It will be
understood by those skilled in the art that the quality of the
approximation depends on the resolution (i.e. tile size). Therefore
the more numerous and smaller the individual tiles, the more
exactly the temperature profile to the sand pack 28 can be
matched. A reasonable match is required to ensure reasonably
accurate experimental results. Furthermore, the individual tiles may
or may not be continuously energized so that their temperature
may fluctuate slightly through a duty cycle. Providing liner 34, which
has some thermal resistance, mitigates these sources of error. By
placing it on the inside of the sample holder it will act to dampen
out the duty cycle temperature fluctuations and also smooth out the
tile temperatures across the assembly.
It can now be appreciated that during any experimental run
the temperature of each tile can be set to any desired temperature.
This has two important advantages. Firstly, the desired insitu
temperature gradients can be externally imposed on the tar sand
sample before the solvent is introduced or the process experiment
otherwise begun. Then, when the actual process testing is started,
for example by injecting the solvent into the sample tar sand pack,
the individual tile setpoints can be determined by mapping the
internal temperatures within the tar sand as described below. This
strategy allows the experiment to start close to the desired "quasi-
= CA 02436158 2011-06-14
12
steady state" temperature profile and thereby minimize the duration
of the thermal transient period. This strategy provides a substantial
time/cost savings and greatly improves the usefulness of the data
since realistic solvent/oil ratios can be obtained right from the
beginning of the experiment.
In addition, once the process has begun, and heat is being
supplied to the sample by means of the process being tested,
appropriate thermal boundary conditions can be provided by the
temperature configurable sample holder of the present invention. It is
preferred to use the temperature measured at the closest adjacent
thermocouples 30 within the sand pack 28 as the basis for
calculating the tile setpoint temperature. Temperature profiles in a
heated gravity drainage process have been previously measured
(Birre119) and according to the present invention insitu temperature
profiles can be readily interpolated or extrapolated from an array of
temperature sensors in the sample. Interpolation and/or
extrapolation of these curves together with the location of the
individual tile determine the desired temperature for a particular tile.
This mapping is most convenient if the thermocouple spacing within
the tar sand pack is a simple multiple of the tile spacing. For
example, with 5cm spacing on the thermocouples it is most
convenient to have tiles on a 2.5 cm pitch.
Fig. 4 shows a power circuit 40 to energize the individual tiles.
In a preferred embodiment the power is delivered by a 24 volt 8
Amp DC Power Supply 42. The power is dissipated in a 100 ohm
resistor 44. Thus, each tile is energized with approximately 6 watts of
power. As each tile includes a temperature sensor, the control
system will apply power according to a difference between the
desired tile temperature and the actual tile temperature. The size of
the difference will determine the length of time that the power is
applied to any individual tiles. An adequate configuration is for a
ground connection multiplexer or "drain multiplexer" 46 to cycle
through each of the 32 columns in the preferred tile array with a 1
second dwell time for each tile column in the array.
An example arrangement for the drain multiplexer 46 is to use
two digital switch banks 48. Adequate results can be obtained from
National Instrument Field Point Digital Switch Banks (each having 16
software programmable switches) part number FP-DO-401 to drive
32 individual 8 amp rated relays 38. For such a mechanical relay
arrangement, it is preferential that all 32-power controller switches 39
are open circuit (La there is no current flow) during the short time
interval when the drain multiplexer switches from one column to the
= õ
= CA 02436158 2011-06-14
13
next.
Thus, in this configuration a particular tile will cool
continuously during 31 seconds of the cycle and will be energized
during the final second of the 32-second cycle in order to achieve its
target temperature. The maximum power output of this particular
configuration is about 180 watts. Other configurations and cycle
times are comprehended by the present invention, but the foregoing
yields satisfactory results.
The output power controller 41 regulates power delivery to
each of 32 individual tiles by controlling a length of time that each tile
is energized by the DC supply. Again two Field Point Digital Switch
Banks 43, can be used to actuate a total of 32 power control relays
39. For example, if 3 joules of energy are required for a particular tile
then the tile would only be energized for 0.5 seconds. The
combination of power controller, drain multiplexer and the individual
diodes 45 associated with each power resistor avoids the problem of
parallel current paths thereby ensuring that the correct tiles is
addressed and individually energized at the appropriate level. This
arrangement is particularly convenient, because it allows a common
wire to energize all the tiles in a particular row and a common drain
wire to return current to ground for a particular column. The
arrangement described can uniquely address and deliver power to
1024 individual tiles (i.e. =32columns x32rows).
A large cost is associated with providing individual power
supplies and controllers for each heater tile. The preferred
configuration uses one power supply and an appropriate
arrangement involving digital switches and diodes to individually and
uniquely address a large number of tiles. The present invention
therefore comprehends using the same to uniquely address and
energize individual tiles in a matrix consisting of 1024 tiles via 32
individual power switches 39, a resistor 44 and diode 45 pair on each
tile, and one 32 channel drain multiplexer 48 which connects to
ground.
With experiments conducted in pressure vessels, there is
always a safety/reliability concern with electrical feedthroughs
through the wall of a pressure vessel. Although feedthroughs are
well known and commonly used, they represent a possible leak path
and present a risk of catastrophic failure. Thus, it is desirable to
minimize the use of pressure wall electrical feedthroughs. For
example, with 1024 tiles there would typically be 2048 thermocouple
wires that would have to pass through the wall of the pressure
vessel. However, the preferred configuration of the present invention
CA 02436158 2011-06-14
14
is to place most of the electronics (multiplexers, digital switches,
relays, data acquision cards, power supplies, plc's etc) within the
pressure vessel so they operate at elevated pressure and then to
communicate to an external computer via a single high-speed
connection like an Ethernet link. This is outside the nominal
operating specifications of some components but pressure tests
have demonstrated that reliable operation can be achieved. This
simplifies the design and greatly increases mechanical reliability.
The present invention also comprehends using wireless, optical or
other signal communication that does not require a penetration
through the wall of pressure vessel.
Figure 5 shows a schematic of a data acquisition system 50 to
address and measure a temperature of each individual tile 16. As
previously noted, each tile has an individual temperature sensor 15.
Preferentially the sensors 15 are type T thermocouples or the like
that have sufficient sensitivity and accuracy. The sensors will be
calibrated insitu with the appropriate wiring in a known manner.
Each thermocouple is connected to an input multiplexer 52. Each
row of tiles is connected to a common multiplexer 52. Input from the
control system controls the switching of the multiplexers 52 to
selective read temperatures in a particular column of tiles. Further
input from the control system controls the switching of the
multiplexers 54 to select the appropriate row of tiles for
measurement by the data acquisition system 58. In this way, the
data acquisition system can address each tile individually.
Unfortunately, there is no easy arrangement to share common wires,
so 2048 individual wires 56 are required to measure the
temperatures of 1024 tiles. For clarity, in Figure 5, each single line 56
actually represents two thermocouple leads. The multiplexers 52, 54
are ADG732 from Analog Devices and are mounted on small printed
circuit boards (PCB) in order to provide a good mechanical for
support for the wire terminations.
Similar to the energizing of the heater tiles, a large cost is
potentially associated with providing individual data acquisition
channels for each thermocouple. The present invention provides an
efficient configuration since the thermocouples for each tile (i.e. 1024
thermocouples) are connected through a series of multiplexers to a
32-channel data acquisition system. These multiplexers provide a
means to uniquely address each individual thermocouple as
explained above.
Cold junction error could be introduced into the thermocouple
readings by the wire connections. To minimize this type of error it is
= CA 02436158 2011-06-14
15
desirable to make the PCBs small and place them in a well-
ventilated location so the PCB's are effectively isothermal, thereby
canceling out any cold junction errors. Furthermore, it is important to
use the appropriate thermocouple lead wire 60 for all subsequent
connections back to the data acquisition system Again for clarity,
each thermocouple lead wire 60 actually represents two
thermocpouple leads.
Figure 6 depicts a startup temperature control algorithm.
Good results have been obtained from software coded in National
instruments Labview programming language downloaded into a
National Instruments Field Point controller At start up, the present
invention is able to configure a desired temperature profile in the
sample holder at a given expected (either calculated or estimated)
Quasi-Steady State (QSS) temperature profile. Being able to directly
apply this temperature profile to configure the profile of the sample
sandpack greatly reduces the length of time taken to get to this
condition.
According to the present invention an estimate is made of the
bitumen interface velocity in order to calculate the QSS. However,
this may well be one of the parameters being determined by the
experiment. Where the extraction rate is not known beforehand (an
arbitrarily assumption of 5 cm/day was used in figure 1), an estimate
for the QSS can be made and as the experiment proceeds the actual
bitumen interface velocity is determined. The length of any transient
period between start up and achieving QSS will depend on the
accuracy of the first estimate.
Thus, figure 6 shows the following steps. At 100, determine
the expected QSS profile. At 102, map the individual tiles onto the
QSS grid to determine the desired temperatures for the tiles. At
104, measure the actual tile temperature and then calculate the
appropriate amount of power that should be delivered to the tile to
reach the desired temperature. At 106, energize the heater on the
tile to deliver the appropriate amount of power. At 108 the sandpack
temperature is measured. At 110 the sandpack temperature is
compared to the desired temperature. At 112 the cycle is continued
until the tile temperature and the sandpack temperatures match the
QSS target. This may take between 4 and 24 hours depending on
the size of the sandpack and the exterior insulation. When the QSS
profile is achieved then the process experimentation can proceed
and, for example, solvent can be introduced into the sandpack 114.
According to the present invention the 2D slice of the
reservoir that the experiment simulates should have no heat loss
CA 02436158 2011-06-14
16
through the sidewalls, so the two sidewalls should be perfectly
adiabatic. Thus, in addition to reducing the time to achieve QSS, the
present invention eliminates parasitic heat losses which otherwise
can distort the experimental results. Eliminating parasitic heat losses
does not mean perfect insulation. To properly mimic the insitu
conditions also requires a controlled amount of heat loss or flux,
through the top, bottom and end walls of the sample holder to
represent the temperature gradients and heat fluxes that would
extend beyond the can into the surrounding "virtual" tar sand matrix.
For example if we refer to Figure 1 it shows, for a quasi steady-state
profile, a temperature gradient occurs at 0.6 m (i.e. the end of the
can). Controlled heat loss at the top bottom and end of the can is
provided by accounting for the insulating quality of the liner and then
calculating the appropriate temperatures for individual tiles to match
the heat fluxes necessary to simulate the external temperature
gradients. For example, since the thermal conductivity of silicone
rubber is about 10% of that of typical tar sand, a rubber liner
thickness inside the can of 2.5mm has the same heat loss
characteristics as a tar sand layer having a thickness of 2.5cm. This
provides a simple and convenient way to calculate the appropriate
desired tile temperatures at the top, bottom and end of the can
needed to match the heat flux boundary conditions. Using the
equivalent "tar sand" distance of the rubber liner extrapolating from a
smoothed sandpack temperature grid the appropriate desired tile
temperatures can be found for the top, bottom and end walls.
Figure 7 shows the temperature control algorithm while in the
midst of conducting an actual process trial. Heated solvent vapour is
being injected into the sandpack causing solvent diluted bitumen to
drain from the sandpack. During this process the tile setpoint
temperatures are determined by the measured internal temperatures
within the sandpack instead of the estimated QSS temperature grid.
Figure 7 shows that the control system algorithm cycles through the
following steps. At 200, measure sandpack temperature, at 202
calculate smoothed 2D temperature grid for the sandpack. At 204
map the tiles onto the sandpack temperature grid to determine the
setpoint temperature for individual tiles. At 206 measure the actual
tile temperature and calculate the appropriate control action. And
finally, at 208 energize the tile to supply the required amount of heat.
For the 32 x 32 heater tile matrix described above, the
individual heater tiles are energized for one second out of 32. The
temperature drop during the 31seconds of cooling is determined by
the heat loss (i.e. conduction heat loss along the side of the can,
. . .
CA 02436158 2011-06-14
17
insulation + exterior temperature and by the thermal inertia of the
tile). The magnitude of any temperature "bounce" can be reduced by
increasing the heat capacity of the tile (i.e. the tile thickness),
increasing the external temperature, or increasing the insulation.
Referring to Figure 1, we can see that for an experiment at 50C the
QSS indicates that the lowest temperature is about 33C at 60cm
from the interface. Thus a minimum exterior temperature of about
25C is likely required (to allow 33C with heat conduction taking place
along the side walls). With a reasonable amount of insulation the
cooling rate of a tile at 50C is .01C/second, so the tile temperature
will drop about 0.3C in a 30 second cycle. To obtain the most
accurate tile temperature measurement for control, it is preferred to
measure its temperature just prior to applying power (i.e. at 31
seconds into the cycle in the example). With this configuration the
tile temperature typically is within 0.2C of the target setpoint
temperature.
Figure 8 compares the expected temperature profiles within
the tar sand (i.e. QSS profiles) to the temperature profiles achieved
along the side wall by the present invention. Profiles are shown at
startup and again at 4 days into an experiment assuming an
interface velocity of 5cm/day. The tiles match the temperature
profile but also introduce a temperature error due to the step
changes in temperature between adjacent tiles. Figure 8 shows that
the maximum temperature en-or (i.e. the difference between the tar
sand and the wall temperature) is about 0.3C. For a rubber liner of
1/4", this is equivalent to a parasitic heat loss near the bitumen
interface less than 1% of the target heat delivery rate. Including a
+.2 C "bounce" from the duty cycle of the control system, the total
heat loss error should be within a few percent of the ideal (i.e. insitu
reservoir) case. By comparison a highly insulated configuration
which is not configurable as is the present invention will still have
parasitic heat losses 10 to 100 times larger than the total target heat
delivery rate, making experimental results all but meaningless.
It can now be understood how the present invention may be
used to determine a true bitumen extraction rate. There are several
criteria that can be used. First, the location of any vapour chamber
that forms can be measured by means of the sandpack
thermocouples. The temperature profile will indicate a position of the
bitumen interface. By tracking its position over time an extraction
rate can be determined. A solvent delivery rate (i.e. condensation
rate) can also be measured and used to determine if the heat
delivery rate is consistent with the displacement of the bitumen
CA 02436158 2011-06-14
18
interface. (i.e. does the heat balance close?). Closing the heat
balance verifies that a true measurement was made that is not
distorted by temperature transients. We can estimate bitumen yield
(i.e. the rate of bitumen recovery) by comparing actual bitumen
production (i.e. after removing the solvent) to the interface position.
All of these may be referred to as parameters of the extraction
process. The foregoing is not intended to be restrictive and it will be
realized that other parameters can also be measured to help analyze
the efficacy of various production techniques.
It will be appreciated by those skilled In the art that while the
foregoing description relates to preferred embodiments of the
invention various alterations and modifications are possible without
departing from the broad scope of the present invention. For
example, while the foregoing discussions are centered on a
condensing solvent extraction process, the present method and
apparatus could be used on other types of thermally based insitu
recovery processes, such as being used to study bitumen extraction
in SAGD type processes. The components (i.e. power resistors,
wiring etc) would have to be appropriately sized to operate higher
temperatures which are typical of such SAGD processes. In addition
there also are processes that propose to use solvents mixed into the
steam that would be good candidates for using the present invention,
since the prior art scaling criteria for any solvent processes is in
doubt.
