CA2114456C

CA2114456C - Thermal recovery process for recovering oil from underground formations

Info

Publication number: CA2114456C
Application number: CA002114456A
Authority: CA
Inventors: Thomas James Boone; Peter Richard Kry; Richard John Gallant; Harshad Nathubhai Patel
Original assignee: Imperial Oil Resources Ltd
Current assignee: Imperial Oil Resources Ltd
Priority date: 1994-01-28
Filing date: 1994-01-28
Publication date: 2004-08-31
Anticipated expiration: 2014-01-28
Also published as: CA2114456A1

Abstract

This invention relates to thermal recovery processes for recovering oil from underground formations, where steam, hot water or any other hot liquid or gas is used to heat the oil in the reservoir, thereby reducing the viscosity of the oil and allowing the oil to be recovered at economic rates. The invention is especially suited to the recovery of highly viscous oil or heavy oil, wherein significant viscosity reduction upon heating occurs, and provides for increased efficiency of recovery in conjunction with increased and controlled areal and vertical conformance. The process of the invention comprises:

(a) providing an injection/production well in fluid communication with the underground formation;

(b) establishing a desired maximum rate of injection Qi of fluid into the formation to horizontally fracture the formation;

(c) calculating for the injection rate Qi the duration of the injection to provide a desired pattern or areal extent of fracture extending from said well, and (d) injecting one or more pulses of said fluid through the well at the desired rate Qi, each for the period of time calculated in step (c) in order to fracture the formation.

Description

BACKGROUND OF THE INVENTION
This invention relates to thermal recovery processes for recovering oil from underground formations, where steam, hot water or any other hot liquid or gas is used to heat the oil in the resexwoir, thereby reducing the viscosity of the oil and allowing the oil to be recovered at economic rates. The invention is especially suited to (but not limited to) the recovery of highly viscous oil or heavy oil, wherein significant viscosity reduction upon heating occurs, and will hereinafter be described in that context. Heavy oils typically have a viscosity in excess of about 100 cp at initial reservoir temperatures.
There are many oil reservoirs in the world which contain heavy oil with such high viscosities that the oil cannot be produced using conventional production methods. Large deposits of this type are located in the Province of Alberta, Canada. Many methods of enhanced production from these reservoirs have been documented including steam injection, solvent injection, in-situ combustion and electrical heating.
One of the more common methods of recovering heavy oil, by employing steam injection, is referred to as cyclic steam stimulation (CSS). CSS involves continuous injection of steam into a well for a period of several days to a few months. Following the injection period, the well may be shut-in for a period known as the soak period, then the well is brought on production for an interval varying from several days to several months. Subsequently, the cycle of injection, soak and production is repeated. The injection of steam enhances heavy oil production by (i) heating the oil, thereby reducing its viscosity and allowing it to flow into the wellbore more readily, and (ii) enhancing reservoir drive mechanisms such as compaction, steam flashing, solution gas and thermal expansion.
A second common thermal recovery method is known as steam drive. In this method a pattern of wells is employed which includes some wells dedicated to continuous steam injection and others dedicated to continuous fluid production. Again, production of heavy ~~ ~. ~~ ~~ ~ ~i oil is enhanced by heating. However, the drive mechanism is now a result of a pressure drop between in3ection wells and production wells.
Often, CSS processes can only recover economically a relatively small fraction (20-30X) of the overall oil in place. This econ~~ic life of a process is limited by the efficiency with which the heavy oil is recovered. Efficiency is commonly measured in terms of the volumetric oil (produced) to steam (injected) ratio (OSx), which inevitably decreases with time. While this low economic recovery can be attributed to many factors, a primary consideration is both horizontal and vertical conformance. Or stated simply, most of the oil tends to be produced from localized regions while other regions remain unaffected. In the later stages of CSS processes, steam tends to flow more readily into zones which are already hot and from which much of the oil has previously been produced. The process would be more efficient and economical if the steam could be directed to the cooler regions in the reservoir which still have high heavy oil saturations by control of the high permeability paths created by fracturing the formation, thereby increasing areal and/or vertical conformance.
In both CSS and steam drive processes, fracturing may occur during steam injection. Indeed, it is often necessary to in3ect at pressures that cause fracturing in order to inject steam at a rate which will allow economic production of the heavy oil. While in the past fracturing has been considered necessary for economic reasons, it has typically been avoided wherever possible for technical reasons. However, a number of patents have previously been granted for processes where fracturing is exploited as a mechanism for creating high permeability paths in reservoirs. representative of these are Canadian Patent ftos. 1,235,652 (Harding et al.) and 1,122,113 (Britton et al.); and U.S. Patent No. 3,330,353 (Flohr).
However, none of these adequately provides for both increased and controlled areal and/or vertical conformance.

SUMMARY OF THE IftVENTIOft The present invention is based upon the discovery that by injecting at rates that are high enough to initiate fracturing and by controlling the rate of steam injection and the duration of the injection period (or equivalently the volume injected), the eXtent of both the fracture and the heated zone can be designed and controlled. The previously unheated reservoir can be heated and the heavy oil in the region can then be mobilized and produced.
Fractures that are created during steam injection can usually be classified as leak-off dominated fractures, which are defined as fractures where the rate of fluid flow injected into the fracture is approximately equal to the rate of fluid loss from the fracture and the rate of fluid storage in the fracture is small compared to the injection anc~ leak-off rates. Equations governing leak-off and storage dominated fractures are given by Geertsma and Deklerk, as more specifically described hereinafter. For leak-off dominated fractures, considering only single-phase flow, the areal extent Af, is given by a simplification of Carter's Model (Howard, G.C. and Fast, C.R.: "Optimum Fluid Characteristics for Fracture Extension", Drill. and Prod. Prac., API (1957) 261):
Af - Qi (~rct)1/2 (1) ~r~cAP
where Qi is the injection rate (commonly quoted in units of m3/day, 1 m3/day = 1.16~10-5 m3/s, in which case Af would be quoted in units of m2) cold mater equivalent volume; ~c is the fluid mobility of the injected fluid in the formation and is equal to k/u, where k is the permeability (units of Darcies, 1 Darcy = 1D = 10-12 m2), a is the apparent viscosity of the fluid (units of cp, 1 cp = 1 mPa~s = 1'10-3 Pa's); DP is the difference between the injection pressure or fracture pressure and the reservoir pressure (units of MPa, 1 MPa =
106Pa); c is the pore pressure diffusivity (units of m2/s or m2/day) and t is injection pulse duration (units of seconds or days).
Utilizing this relationship in the method of the present invention allows injection rates and volumes to be designed appropriately. Critical to the present invention are the following 'I ~
.~ ~ ~~ :~ J
_4_ properties of leak-off dominated fractures: they have extremely high fluid conductivities or effective permeabilities; the primary factor controlling their orientation and direction of propagation is the stress state in the formation; and because the fractures are a parting of the formation, the heat capacitance of the fracture is limited (compared to even a thin but highly permeable layer of sand or rock), which allows heat to be transported effectively along fractures.
To maximize the heat flux into the fracture (and minimize the flux into the previously heated zone).it is advantageous to periodically insect the steam in a series of pulses (i.e. at high injection rates for short periods of time) rather than at a lower continuous rate. If Qbp is defined as the rate at which steam can be injected into a well at 3ust below the pressure required to initiate a fracture, during high rate injection when steam enters the well at a rate Qi, the fraction of the steam entering the fracture (a measure of the efficiency of this process) can be estimated as (Qi-Qbp)/Qi.
Therefore, the efficiency with which fluid is directed into the fracture as opposed to the previously heated zone increases with increased injection rates.
It can be determined from Equation (1) that as Qi is increased, the areal extent of a fracture for the same volume of injection increases. 1"he duration of high-rate injection pulses should be limited or designed so that the fractures extend to the desired radius and do not adversely affect performance of surrounding wells.
Pulsed injection therefore provides an especially efficient and controlled method in accordance with a preferred embodiment of the invention for improving thermal conformance in heavy oil reservoirs.
Thus, in its broadest terms, the invention may be summarized as a thermal recovery process for oil situated in an underground formation which comprises:
(a) providing an injection/production well in fluid ' communication with said formation;
(b) establishing a desired maximum rate of injection Qi of fluid into said formation to horizontally fracture said formation;

~'~ ~ ~" 'r~ ~ J

(c) calculating for said injection rate Qi the duration t of said injection to provide a desired pattern or areal extent of fracture extending from said well, and (d) injecting one or more pulses of fluid through said well at said rate Qi, each for said period of time t as calculated i~i'step (c), in order to fracture said formation.
The invention is especially effective in the recovery of heavy oils having viscosities in excess of about 100 cp at initial reservoir temperatures.
One simple method to calculate the areal extent Af (or conversely, to calculate the time t for a desired area Af) is to apply Equation (1) as described above. If it can be assumed that the fracture is circular, then Af = ~rr2 where r is the radius of fracture. Then, for a desired fracture radius r, the injection pulse time duration t is calculated according to the formula:

r = Q ~ (~rct)1/4 (2) Expressed in terms of t, this becomes:
-2 r4 (2a) t =
~cAP ~ ~c In some situations, the formula above may be an over-simplification of the inter-relationship between Af, Qi and t, for optimization of the fracture pattern or areal extent.
Alternatively, one may employ any or a combination of (i) more complex analytical models, (ii) numerical models or (iii) empirical relationships based on field data (i.e, data from observations), to establish the inter-relationship between Af, Qi and t. Furthermore, a key consideration in optimizing the fracture extent is the distribution of heat deposition in the reservoir. Therefore, one must consider the temperature profile or heat distribution along the fracture which typically requires any or a combination of (i) more complex analytical models, (ii) numerical models or (iii) empirical relationships, as well, if it is found that Carter's model does not provide the required level of optimization.

~1144~~

Assuming one has reasonable estimates of the thermal properties of the formation, as well as those properties identified in Equation (1), the temperature distribution along a fracture and the heated zone around it can be estimated using relatively simple analytical models (e.g. Boone, T.J. and Bharatha, S.: "Temperature '°
distributions along Propagating Leak-Off Dominated Fractures," SPE
25791, First presented at the Thermal Operations Symposium, Bakersfield, CA, February 1993). However, the heat distribution may also be determined using any or a combination of (i) more complex analytical models, (ii) numerical models or (iii) empirical relationships, as well.
Preferably, the fluid is steam, either alone or in combination with cold or warm water ix~ ections. Cold or warm water serves to transfer heat from hot portions of the resexwoir to colder reservoir portions, thereby increasing thermal conformance and allowing more oil to be recovered at a potentially lower cost compared to steam in3ection only.
The injection fluid may also be hot or cold water without the presence of bteam. In the case of cold water, it is assumed that it would be heated as it flows through the formation.
As yet a further alternative, the injection fluid may be a gas, for example one produced by combustion of hydrocarbons with air or oxygen. In this case, significantly higher injection rates with shorter durations may be achieved.
For the reasons discussed above, maximum efficiency is attained by in3ecting the fluid at a high rate - preferably at least 1,000 m3/day of cold water equivalent fluid volume and more preferably at least 2,000 m3/day - for short periods of time. Indeed, the rate of in3ection should be as high as possible, consistent with the practical and economic constraints imposed by existing facilities.
For economic.and practical purposes, the preferred injection rate is up to about 3,000 m3/day of cold water equivalent fluid volume but ' much higher rates are attainable (and would be desirable) by modification of existing facilities).
The injection pulse duration may be from a few hours to several days but for most formations is preferably from about 3 to about 24 _,_ hours. Generally speaking, as cumulative injection and production from a well increases, it is found that pulse durations should also increase and'may eventually become measured in days, rather than hours. Nevertheless, the invention enables maximum efficiency to be maintained by continued application of the high rates of injection.
If desired, the formation may be pre-pressurized and heated by low rate steam injection prior to high-rate fracturing injections, in order to increase the horizontal stress and thus the likelihood that fracturing will occur horizontally.
Vertical conformance can be improved by recompleting or reperforating the well at new locations along the wellbore.
In some situations it may be desirable to initially operate the well for one or more fluid in3ection cycles at conventional in3ection rates and then in a subsequent cycle to implement the high injection rates enabled by application of the present invention. Thus a recovery schedule for a given formation might consist of (i) one or more cycles of conventional CSS, (ii) followed by implementation of pulsed injection employing high injection rates with pulses of less than 24 hours (with or without pre-pressurization); and (iii) eventually increasing the duration of the pulses to several days whilst maintaining the high injection rates consistent with maximum efficiency. The precise conditions for recovery from a given formation will dictate the chosen schedule and will be reflected by the calculations performed in order to maximize the infection rates, as herein described.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 illustrates the stages in a conventional CSS process and also the stages in a high-rate pulsed injection process according to a preferred embodiment of the invention;
Figure 2(a) is an areal plot of a well and a surrounding heated zone and fracture pattern that develops during a conventional CSS or steam drive process;
Figure 2(b) is an areal plot of a well and a surrounding heated zone and fracture pattern that develops with high rates of injection;

_8_ Figure 3 illustrates how cold or warm water injection and fracturing ct~n be employed as a means of increasing thermal conformance;
Figure 4 illustrates the layout of an injection well and a series of observation wells, and illustrates the extent of previously heated zones surrounding such wells;
Figures 5 and 6 are respectively plots of temperatures and well head pressures measured at an observation well D3-OB2 during conventional CSS processing;
Figure 7 is a plot of typical injection rates and injection pressures during a six hour high-rate pulse performed in accordance with the invention;
Figure 8 is a plot of temperatures measured at observation well D3-OB2 during a test of high-rate pulsed injection in accordance with the invention;
Figure 9 is a plot of temperature change measured at selected thermocouples on observation wells D3-OB2 and D3-OB4 during a high-rate pulsed injection HRPI test;
Figure 10 is a plot of temperatures measured at observation well D3-OB6 during the HRPI test;
Figure 11 illustrates the areal extent of the fracture resulting from the HRPI test;
Figure 12 illustrates a series of temperature profiles from successive surveys at observation well D3-OB1 during the HRPI test;
Figure 13 is a plot of bottom hole pressures measured at injection well D3-8 during the HRPI test;
Figure ''4 is a plot of pressures at well D3-8 and five observation wells during the final pulse of tha. HRPI test; and Figure 15 illustrates a further preferred embodiment of the invention directed to improvement in vertical conformance.
DESCRIPTIOft OF THE PREFERRED EMBODIMENTS ' Heavy oils found in the Clearwater formation near Cold Lake, Alberta have been produced using CSS. Typically this process has consisted of repeated cycles where between 5,000 m3 and 20,000 m3 (cold water equivalent volumes) of steam are injected into a vertical well. Subsequently, the well has been placed on production for periods ranging from a few months to more than a year, the injection rates typically being between 200 and 300 m3/day. Initially, these rates are sufficient to immediately fracture the reservoir. However, as the cycles progress, the rate at which the pressure increases during injection decreases and it generally takes significant volumes of infected steam before the pressure reaches a level where the reservoir fractures. Eventually, one can expect that at these injection rates the entire volumes of steam will enter the formation without fracturing. As one would expect, with progressive cycles, the effective permeability of the saad matrix to steam and water increases significantly as the bitumen is extracted. An associated implication is that the hot water and steam are preferentially flowing through channels and zones in the reservoir which have high water or gas saturations and have, therefore, been depleted of their bitumen. As a result, the heat carried by the water and steam is deposited in regions of the reservoir which have relatively low bitumen saturations. This typically manifests itself as a decline in the oil to steam ratio (OSx) as cycles progress.
At Cold Lake, it has been estimated that only 20X of the oil in place can be economically recovered using CSS. The formation consists of a sand with an absolute permeability of 1D and 35X
porosity which is initially 70% saturated with bitumen. The bitumen's viscosity is approximately 100,000 cP at the ambient reservoir temperature of 13°C. A major impediment to increasing this recovery level is the inability to uniformly distribute heat in the reservoir and thereby attain an optimal thermal conformance. When the economic limit of CSS is reached, large volumes within the reservoir will remain near ambient temperatures with high bitumen saturations. A simple, but relatively costly, solution to the problem is to drill more wells. Alternatively, one can devise infection strategies that fundamentally alter the distribution of heat deposition in the reservoir. When in3ecting at rates below the fracture pressure, especially in later cycles, it is the formation's effective permeability to water and steam which is controlling where ~~.~.~4~~
-lo-heat is deposited in the reservoir. Fundamentally, when injecting heat through steam-induced fractures, it is the stress state which controls the orientation and shape of the fractures and, therefore, the deposition of the injected heat. The injection strategy described herein is designed to exploit high-rate injection and~°
fracturing to preferentially deposit heat in the previously uncontacted regions of the reservoir.
Previous fracture tests and observations made in the Alberta oil sands have involved inducing fractures from as injector and monitoring observation wells, or subsequently drilling observation wells to determine the extent, shape and orientation of the fractures. The results have generally indicated that these fractures are not confined to vertical or horizontal planes. A conclusion that can be drawn from these results is that the three principal stresses are nearly equal in these formations.
Factors Controlling Steam-Induced Fractures The following is a brief overview of the fundamental principles that either control or influence fractures in the reservoir and the associated deposition of heat. Oil sands differ from most reservoir rocks due to their lack of cementation or cohesion between grains.
However, ma~,y of the fundamental factors which control fractures is rock are applicable to sands as well. For example, the extent of a leak-off dominated fracture can be estimated by balancing the injection rate with leak-off from the fracture and the fracture can be expected to orient itself perpendicular to the minimum principal stress thereby minimizing the in~ectioa pressure.
First, the extent of fracture propagation can be predicted using the special case of Carter's model for leak-off dominated fractures [Howard and Fast] described above. The area, Af, is expressed as a function of time, t, as follows:
(1) Af(t) = Q (~rct)1/2 where Qi is the injection rate, DP is the difference between the injection pressure Pi and the initial reservoir pressure Po, ~c is the fluid mobility and c is the pore pressure diffusivity of the formation. The fluid mobility ~c is equal to k/u where k is the permeability (Darcies) and a is the apparent viscosity of the fluids in centipoises. Alternatively, one can substitute V = Qit where V is the volume of the injected fluid and attain:
Af(t) = 1 ~ ~~rcQiV 1/2 (3) ~r~cA [,P
Equation (3) demonstrates that fracture area can be increased by increasing the injection rate and maintaining constant volume. i~lhen ix~jecting steam, Equations (1) and (3) can be reasonable approximations if it is assumed that the steam condenses at or near the fracture~face. Given this assumption, it is appropriate to employ the equivalent cold water injection rate and the effective permeability to water in the formation.
It has recently been shown that the temperature distribution along fractures of this type are bounded by self-similar profiles (Boone, T.J. and Bharatha, S.: "Temperature Distributions Along Propagating Leak-Off Dominated Fractures", SPE 25791, First presented at the 1993 Thermal Operations Symposium, Bakersfield, CA, February 1993). These profiles are referenced to the current fracture area and are independent of time and injection rate. This allows one to first estimate the extent of the fracture from Equation (1) and then estimate the interior region which has been heated. One can also make estimates of the arrival time of heat at a point in the reservoir and the quantity of heat that will be deposited. Most importantly, it becomes apparent that, while one can control rate and infection period in order to optimize placement of heat in the reservoir, one also requires knowledge of the effective permeability of the formation to the in3ection fluid.
Geertsma and DeRlerk (Geertsma, J. and DeKlerk, F.: "A rapid Method of Predicting Width and Extent of Hydraulically Induced Fractures", JPT (Dec. 1969) 21, No. 12, 1571-1581) provide some simple equations, based on an assumed logarithmic pressure ~~.~~4~

distribution in a radial fracture, for estimating the fracture width at the wellbore, wf:
wf ~ 2 uQ;a 1/4 l G
where Qi is the injection rate, x is the fracture radius, G is the shear modulus for the formation and a is the viscosity of the in3ected fluid. The average fracture width is 2wf/3. At the well bore the fracture width, given parameters appropriate for Cold Lake and a fracture radius of 60 m, is between 1 and 2 mm. The pressure distribution in the fracture, Pf, is given as:
Pf(r) = Sn - ~ ln(r) (5) 4tt8 x where r is the radial distance from the center of the wellbore and Sn is the stress normal to the fracture. For the fractures at Cold Lake, this equation implies that the pressure at the perforations (r = 0.1 m) ~,s only 0.1 MPa greater than Sn; (1 Ira = 106Pa).
The effective permeability at a point along a fracture is simply w2/12 where w is the width of the fracture at that point. Assuming a 1.3 mm fracture width at the wellbore, the effective permeability is 150 kD and multiplying by the width, the inventors determined a fracture conductivity of 200 Darcy-meters (D-m). This compares with a typical reservoir conductivity of 50 D-m, assuming 1 D absolute permeability and a 50 m thick reservoir. It may be more appropriate to compare the fracture conductivity to the reservoir's effective conductivity to water at its residual oil saturation which is estimated to be 2.5 D-m. Clearly, a steam-induced fracture in oil sands has orders of magnitude greater permeability than any expected communication channels in the formation.
Equation (1) is based on the assumption that there is negligible ..
fluid storage within the fracture so that the rate of fluid injection is balanced exactly by the rate of fluid leak-off. This dictates the areal extent of the fracture but not its shape. As a general principle, a fracture will extend in a manner or shape that minimizes the injection pressure and, therefore, the energy required to inject the fluid into the formation. Typically, this means a fracture will be oriented perpendicular to the minimum stress. Oil sands are typically found at relatively shallow depths where vertical and horizontal stresses are nearly equal. From a reservoir engix~eering perspective, horizontal fractures are often favoured because they are more effective at areally distributing heat and are contained with the formation. One method of increasing the likelihood of horizontal fractures is to increase the reservoir pressure before injecting at rates which induce fractures. It is well known from practice and simple elastic theory (Felsenthal, M. and Ferrell, H.H.: "Fracturing Gradients in Waterfloods of Low-Permeability, Partially Depleted Zones", JPT (June 1971) 727-730) that both pressurizing and heating of the reservoir tend to increase the horizontal stresses more than the vertical stress.
Assuminl~ the fracture is horizontal, it will expand within that plane in a manner or shape that minimizes the injection pressure. In conventional fracture treatments, this principle is exploited in cases where there are relatively high stresses, typically in shales, that confine vertical fractures within the reservoir. Finite element analyses incorporating temperature and pore pressure changes from thermal reservoir simulations of the Cold Lake reservoir have shown that areal variations in the vertical stress of 1 to 2 MPa are induced by the CSS process. Both the areal variation in the vertical stress and the pressure drop along the fracture will influence its shape. However, given the limited pressure drops as estimated from Equation 5, the areal variations in the vertical stress are likely to be the dominant influence (which is supported by the field evidence as hereinafter described and explained). This leads to the conclusion that horizontal fractures will tend to propagate into regions of lowest vertical stress. These are the regions with the least change in temperature and pressure. This is a critical and ' highly desirable aspect of fracturing since it favours transport of heat to previously unheated reservoir. Steam-induced fracturing is potentially a mechanism for diverting heat away from the high-permeability channels in the formation's matrix and into cooler regions of the reservoir where the effective permeability in the matrix is still very low.
In uncemented oil sands, relatively large porosity increases can occur especially under conditions of low effective stress.
Associated recompaction plays a significant role in bitumen recovery as a drive mechanism. Finite element analyses have shown that the region of highest porosity change is in the near vicinity of the fracture faces. This can be attributed to three factors: high temperatures, high pore pressures and low effective stresses. While high temperatures and high pore pressures may occur elsewhere due to flow through the sand matrix, the boundary condition normal to a fracture face is necessarily zero effective stress. These large porosity changes allow for increased effective permeabilities and fluid storage near the fracture face. As a result, a smaller fracture area is required to balance the injection rate and leak-off from the fracture (Settari, A., P.B. Kry and C.-T. Yee: "Coupling of Fluid Flow and Soil Behaviour to Model In3ection into Uncemented Oil Sands", J. Can. Pet. Tech. (Jan-Feb. 1989) 28, No. 1).
High-rate Pulsed Infection During steam in3ection into oil sands, one can control relatively few parameters at a specific well: rate; volume; and energy content. To date, steam strategies have focused on exploiting infection volumes along with field-scale steaming patterns (Gallant, x.J., Stark, S.D., and Taylor, M.D.: "Steaming and Operating Strategies ar. a Mid-Life CSS Operation", (SPE 25791, First presented at the 1993 Thermal Operations Symposium, Bakersfield, CA, February 8-10, 1993). In an especially preferred embodiment, the present invention uses high-rate pulsed in3ection (HBPI) as a tool for improving areal thermal conformance. In essence, the concept is that (1) high injection rates cause the formation to fracture; (2) the shape and orientation of fractures in the reservoir are controlled by the stress state (rather than existing high permeability channels);
(3) by controlling the injection rate and pulse duration, the extent of fracture propagation can be controlled; and (4) in turn, through ~~.~~~~5 knowledge of the heat transport capabilities of fractures, heat can be placed optimally in the reservoir.
Conceptually, the injection strategy is illustrated in Figure 1. Typically, a CSS cycle consists of an infection period at a constant rate followed by a soak period and then a much longer period of production. HRPI involves altering only the injection period so that rather than in3ecting continuously at typical rates (200-300 m3/day), steam is in3ected at high rates for periods which may be from a few hours to several days. For most formations the duration of steam in3ection is preferably between three and 24 hours at rates of from 2,000-3,000 m3/day. Between pulses, the well can be shut-in or itt3ection can be maintained at a nominal rate.
At typical injection rates, it was expected that, if a fracture was induced, it would likely be confined within the existing heated (depleted) zone as shown in Figure 2(a). At high injection rates, a fracture extends well beyond the existing heated zone and thereby improves areal thermal conformance as shown in Figure 2(b).
Essentially, by injecting at high rates, one overwhelms the formation's capacity for accommodating flow through the matrix.
Figure 3 shows how cold water injection and fracturing can be employed as a mechanism for increasing thermal conformance. The cold water is heated as it passes through the heated zone near the wellbore and deposits that heat outside that zone in cooler regions of the reservoir. Thus, heat is transferred from hot portions to colder portions of the reservoir thereby increasing thermal conformance and allowing more oil to be recovered at a potentially lower cost than steam injection alone.
Since steam will be flowing from the well's perforations directly into both the fracture and the formation, it is desirable to maximize the in3ection rate and thereby the fraction of the flow that is entering the fracture. The fracture pressure is effectively limited by the minimum stress in the formation. It also limits the ' injection pressure and therefore the flow rate into the matrix. The period for high-rate injection should be designed so that the extent of the fractures and heat deposition are optimized. At very high rates and for long periods, fractures could extend far beyond the ~.~~4~~~

ad3acent infection wells. The period between high-rate pulses allows both pressure and temperature to diffuse away from the near vicinity of the fracture layer.
Field Test of the Invention The primary purpose of this field test was to prove that the concept of controlled fracturing according to this invention is a feasible mechanism for improving areal thermal conformance. The location chosen for the test was an injection well that had completed seven cycles. of the cyclic steam stimulation prior to the test (see Table 1 below). The well was also surrounded by six observation wells which had provided temperature and pressure histories during the previous five years. The location was also unique since 3D
seismic imaging had been used to identify anomalous zones around the injection wells during the sixth cycle of production. Based on prior stress tests in this reservoir and previous injection pressures, it was expected that horizontal fractures would be induced. The layout of the injection well, known as D3-8, the surrounding observation wells, D3-OB1 through D3-OB6 and the zones imaged from the 3D seismic work are shown in Figure 4.
To prove the feasibility of the invention, it was required that (i) it be clearly demonstrated that horizontal fractures were induced and (ii) it be shown it is possible, given facility constraints and the depleted state of the reservoir, to propagate the fractures and, thereby, transport heat into the undepleted regions of the reservoir.
In cycles one through six of the prior cyclic steam stimulation, injection volumes averaged about 8,000 m3 of steam and, in the seventh cycle, the volume was increased to approximately 19,000 m3 as summarized in Table 1:

Table 1 IftJECTIOft VOLUMES AftD PRESSURES BY CYCLE

.<

___~______ s _--_ Typical ~ Maximum o Volume ~ I~ection ~ Injection Injected~ Pressure , Pressure i Cycle (m3) ~ (MPa) 3 (MPa) ~_~-________.~__ _._.~~_.____~.__..____-_.____________~___ .__ __ I s ;

1 7932 ' 10.5 ~ 10.7 j 2 7027 ~ 11.0 ~ 12.1 3 8034 ( 11.0 ~ 12.0 c 4 i 7904 ~ 10.0 ~ 10.7 i 5 ~ 10113 ~ 8.5 to 11 11 !

6 ~ 7741 ~ 8 to 8.9 ~ 8.9 7 g 18723 ~ 9.5 to 11 11 ~

8(HRPI) ? 17200 ~ 12.2-12.5 12.6 ~

-_'-.__._______.~__.~'~~~___~____T.______..__.___.._____,_______._____ Approximately equivalent volumes were injected into the other operating wells on the pad. Injection rates during each of these cycles were typically between 200 and 300 m3/day.
The six observation wells surrounded D3-8 as shown in Figure 4.
Each of the observation wells had thermocouples attached to the casing and cemented in place when they were installed. The Clearwater oil sands are approximately 55 m thick at this location between depths of 420 and 475 m. Six thermocouples were installed in each well at approximate depths of 0, 10, 20, 30, 40 and 50 m below the top of the Clearwater. Each of the observation wells, except D3-OB6, was perforated over an eight meter interval at a depth of 459 ' to 467 m, wh3~ch allowed measurement of the reservoir pressure through the annulus. Measurements of temperature and pressure were typically recorded four times daily.

The down hole temperature responses and wellhead pressure responses measured at observation well D3-OB2 between 1986 and 1991 are plotted in Figures 5 and 6. These plots are also representative of the other observation wells. During an injection cycle, the wellhead pressures rose to between 4 and 5 MPs which equates.to~-bottomhole pressures between 8.6 and 10.6 MPs. The observed temperatures remained below 50°C except for occasional spikes in the early cycles and D3-OB6 which saw temperatures up to 120°C during the seventh cycle. This temperature response at D3-OB6 was attributed to communication with an injection well to the northeast (D3-3 or D3-4) rather than D3-8 which is southwest of the observation well.
Table 1 lists typical and peak wellhead pressures at the injector, D3-8, for each cycle. These pressures were between 0 and 1 MPs above the bottom hole pressure. Based on plots of wellhead pressures for each cycle, it can be inferred that fracturing likely occurred during most of cycle one to three and parts of cycles five and seven. It is unlikely that fracturing occurred during the fourth and sixth cycles. The original vertical stress at the perforation depth of D3-8 was estimated at 9.7 MPs from density logs. The wellhead pressures were consistent with the expectation that horizontal fractures had been induced in the reservoir.
A 3D seismic survey, centered at D3-8, was performed approximately four months after the completion of sixth cycle injection. Figure 4 plots the approximate limit of the 60°C contour for the region surrounding D3-7, D3-8 and D3-9 based on interpretation of the seismic image and comparable numerical simulations. This interpretation was consistent with the observed temperatures at the surrounding obsexwation wells. All the observation wells were outside what is interpreted as the 60°C
contour with D3-OB2 located closest to the zone. This layout was fortuitous far the purposes of the test described below since the observation wells were ideally positioned for observing heat ' transported outside the previously heated zone by steam-induced fractures.
The HRPI test (cycle 8) was initiated at noon on December 16, 1991, after D3-8 had been steamed for about one week at a rate of approximately 200 m3/day. Steam injection at D3-8 and the surrounding wells, prior to the test, raised the reservoir pressure to approximately 5 MPa. A series of 21 injection pulses were then performed at~approximately two day intervals with most pulse durations being close to six hours. The first series was followed by two 24 hour pulses which were 11 days apart. Figure 7 plots injection rates and the wellhead pressure recorded during a typical pulse. With relatively limited modifications to existing facilities, maximum rates exceeding 2000 m3/day were attainable. The rate typically decreased from 2000 m3/day to approximately 1700 m3/day during each pulse. The wellhead pressures increased gradually during this period. This increase can be attributed to poroelastic effects or backstress which caused the stress normal to the fracture and, therefore, the fracture pressure to increase with time (Boone, T.J., Rry, P.B., Bharatha, S., and Gronseth, J.M.: "Poroelastic Effects related to Stress Determination by Micro-frac Tests in Permeable Rock", 8och Mechanics as a Multi-Disciplinary Science, Proc. of the 32nd U.S. Bock Mechanics Symposium, J.C. xoegiers (ed.), Norman OR, (1991)). The increasing injection pressure combined with facility constraints caused the injection rates to decrease.
The temperature responses from the thermocouples at D3-OB2 are plotted in Figure 8. Note that prior to the start of pulsed injection, the temperature at the 40 m thermocouple increased from 40°C to 110°C then declined to 80°C. Given the close vicinity of the heated zone to D3-OB2, as determined from the seismic image in Figure 3, this initial thermal communication was not surprising.
Cyclic temperature response was apparent from the start of pulsed injection at wells D3-0B1, D3-OB2 and D3-OB4, as can be seen for D3-0B2 from Figure 8. Each pulse was characterized by a rapid temperature increase at these three observation wells. Figure 9 plots the temperature response at selected thermocouples from D3-OB2 and D3-OB4 during the first pulse. They showed significant response ' within hours of commencing the test. Wells D3-OB3 and D3-OB5 apparently were not contacted by hot fluids at suvy time during the test and did not show any significant temperature response. The response at D3-OB6 is plotted in Figure 10. It appears to be communicating with one of the other in~ectioa wells as it had on the previous cycle. However, superposed on this response is a clear signature from pulsed infection at D3-8 as evidenced by the spikes in the 20m and 30m plots. Initially, the response is negligible but after approximately tea pulses it becomes quite distinct. Oae can speculate that either initially the fracture did not intersect D3-OB6 or, if it did, the fluids were at reservoir temperatures.
Subsequently, either the fracture grew to the point it intersected D3-0B6 or, if the fracture had intersected it initially, heat began to be transported to D3-0B6.
The observation wells showed temperature responses through much of the thickness of the reservoir. However, it would be expected that the fracture would only heat a very narrow zone. The broader temperature ~espoase at the observation wells was attributed to a fracture intersecting the wellbore at or near the perforations.
Fluid can they flow into the wellbore, compressing gas at the top of the wellbore and it can potentially flow through pathways along the exterior of the casing or cement. This broad temperature response is, therefore, an artifact of the observation wells. However, it is somewhat advantageous since it negates the need for detailed temperature logging to determine if a fracture intersected the observation well.
The rapid arrival time for heat at the observation wells was highly indicative of fracturing. Given the infection rates sad steam quality 070x), it was estimated using the model of Marx and Laagenheim (Marx, J.N. and Langeaheim, H.H.: "reservoir Heating by Hot Fluid I~ection", Trans. AIMS, Vol. 216, (1959) 312-315) that the steam must be flowing through a channel ao more than a few centimeters thick. Since there was no prior evidence of such channels, it was reasonable to assume that it was a steam-induced fracture.
Based oa the temperature response at the observation wells, estimates of the fracture area were made. Assuming as elliptical fracture, Figure 11 plots approximate extents of the fracture gives that for the first 27 days (1) it intersects D3-OB1, D3-OB2 and D3-0B4, and after 27 days (2) it also intersects D3-OB6. For the two cases, the /i t~ a ~ !a areas were roughly 11,000 m2 and 15,000 m2, where the overlap with the previously heated zone was about 5,000 m2 in each case. These estimates of the fracture size are based on temperature observations. However, the actual extent of the fracture could be greater since it is likely that the injected fluid had cooled tos reservoir temperature before it reached the fracture front and the arrival of fluid at reservoir temperature would not be observed.
It is consistent with Equation (1) that the fracture area would increase with successive pulses since the reservoir pressure was also continually increasing. A larger fracture area was then required to compensate for reduced leak-off. Similarly, as the region around the fracture plazie was progressively heated, heat was transported to larger areal extents with progressive pulses.
Finally, it is important to note that the fracture apparently propagated away from the previously heated zone as determined from seismic results and into a region of the reservoir which was relatively cool and undepleted. This is consistent with the assertion that the areal variation in the vertical stress would control the fracture shape and promote fracture growth into regions with lower pressures and temperatures.
Several temperature logs were taken at the obsexwation wells during the test. Unfortunately, the presence of cold bitumen in observation wells D3-OB2 and D3-OB4 prevented logging of these wells on several occasions. The logs at D3-OB3 and D3-0B5 confirm that there was not any significant heat transport to these wells. Figure 12 plots the temperature logs taken at various times for well D3-OB1. These surveys show significant increases both at the interim survey and at the final survey about 7 weeks later. The latter is strongly indicative of fracturing, given the peak in the temperature at a depth of 460 m, about 1 m above the top of the perforations at D3-8 which extend from a true vertical depth of 461 to 469 m.
The temperature logs were integrated to provide estimates of the change in the areal heat density at the observation wells. However, due to convection of heat along the wellbore, it is apparent that the data from observation wells overestimate the average change in heat density that was occurring away from the well. The best estimate is w that approximately 50X of the total heat injected was deposited outside the previously heated zone.
The bottomhole pressure response at the injection well, D3-08, was measured down hole through a gas filled bubble tube contained within the wellbore. The pressure response is plotted in Figure~l3 along with results from a simple numerical simulation. The field data was analyzed in detail for each injection pulse except for some pulses where the data collection system was not functioning. From plots of the derivative of pressure with respect to time versus pressure, both the fracture initiation pressures (Pi) and instantaneous shut-in pressures (Pisip) or fracture closure pressures could be extracted. As a general trend, both the Pi and Pisip tended to increase with progressive cycles. The bottom hole pressure during injection, 3ust prior to shut-in, typically exceeded the Pisip by about 0.6 MPa.
The numerical simulation results are from a single well radial model which assumed single phase flow. It also accounted for changes in the vertical stress and the fracture pressure due to temporal changes in both temperature and pressure throughout the reservoir using an integral approach (Boone, T.J., Rry, P.x., Bharatha, S., and Gronseth, J.M.: "Poroelastic Effect$ related to Stress Determination by Micro-frac Tests in Permeable rock", xoch Mechanics as a Multi-Disciplinary Science, Proc. of the 32nd U.S. Bock Mechanics Symposium, J.C. xoegiers (ed.), Norman OR, (1991)). In spite of its simplicity, the model produced a reasonable match to the pressure response and captured the rise in injection pressure with subsequent cycles. The match was attained with assumed effective permeabilities of 1.1 mD vertically and 7 mD horizontally. The vertical permeability was the primary factor controlling the areal extent of the fracture and the horizontal permeability was what largely controlled the pressure decline between cycles.
Figure 14 overlays the bottomhole pressure response at D3-8 with that obtained from the observation wells during the final 24 hour pulse. It shows a rapid but delayed response at the observation well during each r~ulse. Wells D3-OB1 and D3-OB4 had similarly rapid response whereas wells D3-OB3 and D3-OBS, which were apparently not intersected by the fracture, responded more slowly.
The fracture initiation pressure (Pi) is identified as 10.7 MPa in Figure 14 and the instantaneous shut-in pressure Pisip is 11.0 MPa. The latter was the best estimate of the fracture pressure-outside the near wellbore region where most of the pressure drop was expected to occur during injection. The pressure immediately dropped by 0.6 MPa after shut-in to the value of the Pisip~ It was deduced that the best estimate of the vertical stress at the time of fracture initiation was 10.1 MPa or 0.6 MPa less than the value Pi. The difference between the Pisip at the injector and the pressures measured at the observation wells was only a few tenths of an MPa.
However, the pressure at the injectors and the observation wells were at least 1 MPa greater than original estimate of the vertical stress (9.7 MPa). These observations were consistent with the theoretical expectations described above that there should be only a small pressure drop along the fracture and an order of magnitude larger areal variation in the vertical stress.
There were numerous observations that supported the hypothesis that what in, essence were conventional, leak-off dominated fractures were induced in the reservoir which was composed of uncemented oil sands. In fact, this test was unique in terms of the variety of tools and facilities that could be used to monitor and interpret the behavior of the fractures. The evidence for horizontally-induced fractures can be summarized as follows:
(1) Heat Transport: The rapid temperature response, as illustrated in Figure 8, which occurred within hours of the initiation of pulsed injection indicated that a thin highly permeable channel formed in the reservoir. Based on the model of Marx and Langenheim, one would expect that the channel must be no more than a few centimeters in thiclrness. There was no prior evidence of a high permeability channel at this depth.
(2) Infection Pressure: The maximum injection pressure was comparable to that reached on previous cycles at much lower injection rates. This was indicative of fracturing which essentially capped the injection pressure at a value close to the stress due to the weight of the overburden and independent of the formation's effective permeability to the injected fluid.
(3) 'Fracture' Orientation: It was expected that with the reservoir pressure at 5 MPa at the start of pulsed injection that the stress state would favour horizontal fractures. At D3-OB1 the peak temperature response on the final survey was at a depth of 460 m.
This is approximately one meter shallower than the top of the perforations at the injection well which was 80 m away. The observed response at the other three observation wells which saw significant temperature changes was also consistent with a nearly horizontal fracture.

(4) Changes in Injection Pressure: As the reservoir pressure and temperature increased the fracture pressure should also increase due to local changes in the vertical stress. The field data supports this assertion since the fracture closure pressure as determined from the Pisip increased from 9.5 to 11 MPa over the injection period.

(5) Numerical Simulations: The match between the injection pressure and that determined from a simulated fracture model, Figure 12, adds credence to the fracture concept. In particular, the effective permeabilities used to attain the match were consistent with expectations and useful design parameters.

(6) Pressure Gradients: From the pressures measured during the last injection pulse, it was determined that the Pisip was 11 MPa which was the best estimate of the fracture pressure outside the near wellbore region. At that point in time the pressures measured at D3-OB2 and D3-OB4 were both 10.8 MPa, respectively. This small pressure differential between the observation wells and D3-08 and the rapid rise times for the pressures were also indicative of fracturing. As discussed above, a pressure drop in the order of 0.1 l~a would be expected over the radius of a fracture. Fluid flow through the matrix would be expected to give rise to much higher pressure drops given the injection rate and lrnown matrix ' permeabilities.

(7) Areal Shape and Extent: Given the observed increase in the fracture pressure, it was deduced that, in the reservoir, there must be an areal variation in the vertical stress with a magnitude on the order of 1 to 2 MPa. Since this is significantly greater than the pressure drop along the fracture as determined from theory and observation,'it follows that areal variations in the vertical stress were controlling the fracture shape. This is consistent with the field evidence, as shown in Figure 11, since the fracture propagated away from the previously heated zone where the vertical stresses are predicted to be greatest.
Vertical Conformance Figure 15 illustrates how vertical conformance can be improved by recompleting or reperforating the in3ection well at new locations along the wellbore. There is shown firstly a depleted zone underlying a high bitumen saturated zone. A re-perforation is formed in the wellbore in communication with the high bitumen saturation zone and an injection pulse is applied in the manner of the invention. This provides a horizontal fracture extending from the vicinity of the re-perforation and steam and heat are thus diverted away from the depleted zone. Re-perforation can be repeated at a number of levels in the formation, so that a series of horizontal production zones results, with correspondingly improved vertical conformance.
Imvlications for Thermal Reservoir Enzineerina Some of the implications for thermal reservoir engineering arising from the present invention are as follows:
1. It has been demonstrated that high-rate steam injection and induced fracturing are an effective means of rapidly increasing areal thermal conformance (even in eighth cycle CSS at Cold Lake).
2. It has been demonstrated that HRPI is a thermally efficient mechanism since it is estimated that approximately 50~ of the injected heat was placed outside the previously heated zone in the _ vicinity of the fracture.
3. Controlling the pulse duration provides a mechanism for limiting the areal extent of fracturing and thereby the extent of areal heat transport.

4. Pulsed in3ection provides a distinctive trace to communication events so that they can be clearly tied to the source.
5. It has been demonstrated that at Cold Lake the fractures are expected to remain nearly horizontal over radial distances of at least 80 m. This confirms the potential to exploit HRPI as a ~-mechanism for altering vertical conformance by re-completion at various depths. Fractures can be expected to initiate at or near the top of the perforated interval. By re-perforating, so that the top of the perforations align with another zone of high bitumen saturation, steam and heat can be diverted away from a connected zone which has been previously depleted or initially had high water saturation.
6. It has been shown that areal variations in the vertical stress are the dominant mechanism controlling the fracture shape and that this mechanism causes fractures to grow into regions of the reservoir that were not previously heated. This leads to the conclusion that HRPI preferentially deposits heat in the previously uncontacted cooler reservoir.

Claims

1. A thermal recovery process for recovering oil from an vaderground formation which comprises:

(a) providing as injection/production well in fluid cammtmication with said formation;

(b) establishing a desired maximum rate of injection Qi of fluid into said formation to horizontally fracture said formation;

(c) calculating for said injection rate Qi a duration of time t of said injection to provide a desired pattern or areal a:teat of fracture extending from said well, and (d) injecting one or more pulses of said fluid through said yell at said'rate Qi, each for said period of time t as calculated in step (c) in order to fracture said formation.

2. Thermal recovery process of claim 1, wherein said oil is a heavy oil having a viscosity in excess of about 100 cp.

3. The thermal recovery process of claim 2, wherein said time t is calculated by application of Carter's model:
there .DELTA.f is the desired areal eactent of said fracture, K is the fluid nobility of the injected fluid in the formation, .DELTA.P is the difference between the pressure required to fracture the formation and the reservoir pressure and c is the pore pressure diffusivity constant for the formation.

4. The thermal recovery process of claim 3, wherein said desired fracture pattern is circular and has a radius r calculated by application of the formula:

5. The thermal recovery process of claim 2, 3 or 4, wherein said fluid ie steam.

6. The thermal recovery process of claim 2, 3 or 4, wherein said fluid is selected from the group comprising cold water with steam, warm water with steam, cold water without steam, and warm water without steam.

7. The thermal recovery process of claim 2, 3 or 4, wherein said fluid comprises the gaseous product of combustion of a hydrocarbon with air or oxygen.

8. The thermal recovery process of claim 2, 3 or 4, wherein the horizontal stress in said formation is increased by pressurization and heating of said formation prior to said fluid injection, is order to optimize conditions for said fluid injection to fracture said formation horizontally.

9. The thermal recovery method of claim 2, 3 or 4, wherein said fluid is injected in a single pulse.

10. The thermal recovery method of claim 2, 3 or 4, wherein said fluid is infected in a series of pulses.

11. The thermal recovery process of claim 2, 3 or 4, wherein said fluid is infected at a rate of at least about 1,000 m3/day cold water equivalent volume.

12. The thermal recovery process of claim 5, wherein said steam is injected at a rate of at least about 1,000 m3/day cold water equivalent volume.

13. The thermal recovery process of claim 11 or 12, wherein said fluid is infected at a rate of about 2,000-3,000 m3/day cold water equivalent fluid volume.

14. The thermal recovery process of claim 6, wherein said steam is injected in a series of pulses of duration from about 3 hours to about 24 hours each.

15. The thermal recovery process of claim 2, 3 or 4, wherein the horizontal stress in said formation is increased by pre-pressurizing and heating said formation by steam injection prior to said fluid injection pulses at said rate Qi for said time t, is order to optimize conditions for said pulses to fracture said formation horizontally.

16. The thermal recovery process of claim 2, 3 or 4, which further comprises recompleting the bore of said well at a sew location within said formation, said location being is alignment with a zone of high oil saturation, and injecting fluid through said well as aforesaid in order to fracture said formation in said zone of high oil saturation sad thus create a new production zone.

17. A thermal recovery process for oil from an underground formation which comprises:
(a) providing an injection/production well in fluid communication with said formation;

(b) establishing a desired maximum rate of injection Qi of fluid into said formation to horizontally fracture said formation, whrein said fluid ie selected from the group consisting of cold water, hot water and steam and said injection rate Qi is at least about 1,000 m3/day cold water equivalent volume;

(c) calculating for said injection rate Qi the duration t of said injection to provide a desired pattern or areal extent of fracture extending from said well, and (d) injecting a series of pulses of fluid through said well at said rate Qi, each for said period of time t as calculated in step (c) in order to fracture said formation, and each said period being from about 3 hours to about 24 hours duration.

18. Thermal recovery process of claim 17, wherein said oil is a heavy oil having a viscosity in excess of about 100 cp.

19. The thermal recovery process of claim 18, wherein said time t is calculated by application of Carter's model:

where A f is the desired areal extent of said fracture, k is the fluid mobility of the injected fluid in the formation, .DELTA.P is the difference between the pressure required to fracture the formation and the reservoir pressure and c is the pore pressure diffusivity constant for the formation.

20. The thermal recovery process of claim 19, wherein said desired fracture pattern is circular and has a radius r calculated by application of the formula:

21. The thermal recovery process of claim 18, 19 or 20, wherein the horizontal stress in said formation is increased by pre-pressurization sad heating of said formation prior to said fluid injection, in order to optimize conditions for said fluid injection to fracture said formation horizontally.

22. The thermal recovery process of claim 18, 19 or 20, wherein the horizontal stress in said formation is increased by injecting steam into said formation, prior to said fluid its action pulses, is order to pre-pressurize and heat said formation and optimize conditions for said pulses to fracture said formation horizontally.

23. The thermal recovery process of claim 18, 19 or 20, which further comprises reperforating or recompleting the bore of said well at a new location within said formation, said location being in alignment with a cone of high oil saturation, and injecting fluid through said well as aforesaid in order to fracture said formation in said zone of high oil saturation and thus create a new production zone.

24. A thermal recovery process for recovering oil from an underground formation which comprises:

(a) providing as injection/production well in fluid communication with said formation;

(b) in an initial recovery phase, establishing a rate of infection of fluid into said formation which will horizontally fracture said formation;

(c) in a subsequent recovery phase, increasing the rate of fluid injection to a desired maximum infection rate Qi;

(d) calculating for said injection rate Qi a duration of tine t of said infection to provide a desired pattern or steal extent of fracture extending from said well, and (e) infecting one or more pulses of said fluid through said well at said rate Qi, each for said period of time t as calculated in step (d) is order to fracture said formation.

25. The thermal recovery process of claim 24, Wherein said oil is a heavy oil having a viscosity in excess of about 100 cp.

26. The thermal recovery process of claim 25, wherein said time t is calculated by application of Carter's model:
where Af is the desired steal ezteat of said fracture, K is the fluid mobility of the injected fluid in the formation, .DELTA.P is the difference between the pressure required to fracture the formation sad the reservoir pressure and c 18 the pore pressure diffusivity constant for the formation.

27. The thermal recovery process of claim 26, wherein said desired fracture patters is circular and has a radius r calculated by application of the formula:

28. The thermal recovery process of claim 25, 26 or 27, wherein said fluid is steam.

29. The thermal recovery process of claim 25, 26 or 27, wherein acid fluid is selected from the group comprising cold eater with steam, warm water with steam, cold water without steam, sad warm water without steam.

30. The thermal recovery process of claim 25, 26 or 27, wherein said fluid comprises the gaseous product of combustion of a hydrocarbon with air or oxygen.

31. The thermal recovery process of claim 25, 26 or 27, wherein the horizontal stress in said formation is increased by pressurization sad heating of said formation prior to said fluid injection at said rate Qi, in order to optimize conditions for said fluid injection to fracture said formation horizontally.

32. The thermal recovery method of claim 25, 26 or 27, wherein said fluid is injected at said rate Qi in a single pulse.

33. The thermal recovery method of claim 25, 26 or 27, wherein said fluid is infected at said rate Qi in a series of pulses.

34. The thermal recovery process of claim 25, 26 or 27, wherein said fluid is injected at said rate Qi of at least about 1,000 m3/day cold water equivalent volume.

35. The thermal recovery process of claim 28, wherein said steam is infected at said rate Q i of at least about 1,000 m3/day cold water equivalent volume.

36. The thermal recovery process of claim 34 or 35, wherein said fluid is injected at said rate Q i of about 2,000-3,000 m3/day cold water equivalent fluid volume.

37. The thermal recovery process of claim 29, wherein said steam is injected at said rate Q i in a series of pulses of duration from about 3 hours to about 24 hours each.

38. The thermal recovery process of claim 25, 26 or 27, wherein the horizontal stress is said formation is increased by pre-pressurizing and heating said formation by steam injection prior to said fluid injection pulses at said rate Q i for said time t, in order to optimize conditions for said pulses to fracture said formation horizontally.

39. The thermal recovery process of claim 25, 26 or 27, which further comprises recompleting the bore of said well at a new location within said formation, said location being is alignment with a zone of high oil saturation, sad injecting fluid through said well as aforesaid in order to fracture said formation is said zone of high oil saturation sad thus create a new production zone.

40. A thermal recovery process for oil from an underground formation which comprises:
(a) providing an injection/production well in fluid communication with said formation;
(b) in an initial recovery phase, establishing a rate of injection of fluid into said formation which will horizontally fracture said formation;
(c) in a subsequent recovery phase, increasing the rate of fluid injection to a desired maximum injection rate Q i, said fluid being selected from the group consisting of cold water, hot water and steam and said injection rate Q i being at least about 1,000 m3/day cold water equivalent volume;
(d) calculating for said infection rate Q i the duration t of said injection to provide a desired pattern or areal extent of fracture extending from said well, and (e) injecting a series of pulses of fluid through said well at said rate Q i, each for said period of time t as calculated is step (d) in order to fracture said formation, and each said period being from about 3 hours to about 24 hours duration.

41. Thermal recovery process of claim 40, wherein said oil is a heavy oil having a viscosity in excess of about 100 cp.

42. The thermal recovery process of claim 41, wherein said time t is calculated by application of Carter's model:
where A f is the desired areal extent of said fracture, .kappa. is the fluid mobility of the injected fluid in the formation, .DELTA.P is the difference between the pressure required to fracture the formation and the reservoir pressure and c is the pore pressure diffusivity constant for the formation.

43. The thermal recovery process of claim 42, wherein said desired fracture pattern is circular and has a radius r calculated by application of the formula:

44. The thermal recovery process of claim 41, 42 or 43, wherein the horizontal stress in said formation is increased by pre-pressurization sad heating of said formation prior to said fluid infection at said rate Q i, in order to optimize conditions for said fluid injection to fracture said formation horizontally.

45. The thermal recovery process of claim 41, 42 or 43, wherein the horizontal stress is said formation is increased by injecting steam into said formation, prior to said fluid infection pulses at said rate Q i, in order to pre-pressurize and heat said formation and optimize conditions for said pulses to fracture said formation horizontally.

46. The thermal recovery process of claim 41, 42 or 43, which further comprises recompleting the bore of said well at a new location within said formation, said location being in alignment with a zone of high oil saturation, and injecting fluid through said well as aforesaid in order to fracture said formation is said zone of high oil saturation sad thus create a new production zone.

47. The thermal recovery process of any one of claims 1 to 46, wherein between the one or more pulses there is no production.