CYAN MAGENTA YELLOW BLACK

CYAN MAGENTA YELLOW BLACK g n ti Injec tion o o b s e t r o f r e st Injecting oil back into a reservoir might seem an odd thing to do but it h...
Author: Brianna Floyd
5 downloads 2 Views 503KB Size
CYAN MAGENTA YELLOW BLACK

g n ti

Injec

tion

o o b

s e t r o f r e st

Injecting oil back into a reservoir might seem an odd thing to do but it has given a boost to reservoir characterization in Turkey. Instead of creating a flow-rate perturbation by producing the well, the Turkish Petroleum Corporation (TPAO) and Schlumberger put their heads together and came up with the unusual idea of carrying out an equivalent test by injecting oil into the formation a technique normally reserved for water injectors. Jorge Torre, Schlumberger Chief Reservoir Engineer for the Middle East, with Erol Memioglu and Can S. Bakiler of TPAO’s Reservoir Engineering Department, outline the tests and explain the results. Jorge and Mahmoud Latif of Egypt’s Gulf of Suez Petroleum Company (GUPCO) also show how tests using water injection have helped in measuring abnormally low formation parting pressures in a multi-layered reservoir. In addition, they explain how pulse tests in water injectors, normally used to check a formation’s hydraulic connectivity, have been harnessed to aid reservoir characterization in the Gulf of Suez.

Contributions by Eric Standen, Schlumberger, Egypt.

Injection revives Turkey's low-energy wells

Gerger K.Karakus Cendre Ak-Pinar Field Field Field B.Firat Narinca Karakus Field Guney Karakus Field Besikli Field Field

Istanbul

Kahta

hta Ka

Ankara

T U R K E Y

at

Fir

Firat

Malatya Adana

In naturally flowing wells, conventional shut-in testing is relatively easy. But what happens when a reservoir has such low energy that a well will not flow naturally to the surface? This was the problem facing TPAO's reservoir engineers on the Karakus Field when they tried to use pressure transient testing for reservoir characterization.

42

CYAN MAGENTA YELLOW BLACK

O

pen-hole log information had shown that the tight carbonate reservoir in the Karakus Field was highly fractured and that a dualporosity system existed. Determining the matrix and fracture permeabilities and storativities was therefore essential before a complete model of the entire field could be set up. But how could these figures be obtained? The conventional method of testing non-flowing wells, using a Drill Stem Test (DST), involves lowering a DST string into the borehole and setting a packer near the perforations. A valve in the DST is opened to allow reservoir fluid to enter the wellbore. Just before the borehole fluid pressure equals that of the formation (at which point flow would stop), the well is shut in to create a build-up period. Unfortunately, the maximum flow period that can be attained in Turkey’s non-producing wells is not long enough to promote full radial flow in the reservoir. This means that an analysis of DST results from this kind of well would normally give erroneous results (ie. values of reservoir permeability and skin that are too high). The problems are even worse in fractured reservoirs because the drawdown period has to be kept much longer to give sufficient time for radial flow to develop before shut-in (see box on page 46). TPAO and Schlumberger decided to test the non-flowing wells by introducing a perturbation through injection of oil into the formation prior to shut-in. This gave them control of the flow period and size of perturbation (injection rate). Injecting oil in this way is costly but the fluid returned to the formation is usually recovered when the well is put back on stream.

Diyarbakir

Deciding the length of time necessary to attain full radial flow was one of the main problems during test design. The engineers also wanted to guard against fluctuations in injection rate. Both difficulties were overcome by using surface read-out and combined downhole flow and pressure measurements (Middle East Well Evaluation Review Special Supplement on Reservoir Testing, January 1991). These simultaneous readings with the Production Logging Tool (PLT*) allowed the engineers to spot the develoment of full radial flow prior to well shut-in. In addition, the continuous flow measurements enabled the analysts to remove any noise in the pressure response which might have arisen due to flow rate variations. Accurate information about the reservoir characteristics could then be reliably extracted from the pure reservoir signal (see box overleaf). The engineers faced another problem. How could they be sure that radial flow did not start during the early time when the pressure response is dominated by wellbore storage effects? The best way to minimize these effects was to use downhole shut in. A Schlumberger Cyber Service Unit (CSU*) logging truck was used to monitor real-time injection and pressure measurements in the well during the oil injection phase. A Dowell Schlumberger pumping unit supplied fluid at a relatively constant rate to the perforated area of the wellbore (figure 4.3). A Pressure-Controlled Tester (PCT*) valve allowed full-bore access into the well. This enabled the Production Logging tool (PLT*) to be passed through the DST string. In addition, downhole memory gauges were positioned beneath the packer in readiness for

Middle East Well Evaluation Review

monitoring the pressure build up. (The wireline tools have to be removed before shut in). Since it was difficult to predict when the distinctive dip in the derivative curve might appear, two downhole memory pressure gauges had to be used. One gauge was triggered at the start of the test and recorded at a low sampling rate (one measurement every 15 minutes for the duration of the test). The other gauge was programmed to start sampling at high rate just before the shut in and had enough memory to store two hours of pressure fall-off data. The test engineers faced several practical problems. Injecting oil evenly into the well prior to shut in was difficult as the oil had to be conveyed to the site in a relay of tanker trucks. It was also very difficult to predict how long it would take before full radial flow was achieved. Eventually, for practical purposes, it was decided to inject oil for just 10 hours at a rate of 2400B/D.

Fig. 4.1: FACING UP TO FRACTURES (Above): Author Erol Memioglu examines fractures in statues in the ancient ruined city of Nemrut which lies close to the Karakus Field in Eastern Turkey. The fractured section of core (right) was taken from the Karakus Field reservoir rocks.

50m

PLT flow profile

Number 12, 1992.

Fig. 4.2: Comparison of a flow profile obtained from an open-hole PLT survey with the fracture porosity and fracture apertures derived from Formation MicroScanner* images using the Fracview* program on a computer workstation. The flow increases considerably in the fractured section of the reservoir.

KARAKUS IN A NUTSHELL The Karakus Field is located in Turkey’s most prolific oil-producing area. It lies close to the city of Adiyaman, near the famous ruined city of Nemrut (figure 4.1). The field is bounded to the northwest by the left lateral Adiyaman wrench fault. To the south, a parallel left lateral strike-slip fault separates the field from the neighbouring Güney Karakus Field. The anticlinal structure of the Karakus Field is highly deformed by the wrench fault system that has developed under compressive stress. Oil is found in Lower Cretaceous Mardin Group carbonates which overlie the Cambrian-age clastic Sosink Formation. The Mardin Group is subdivided into five main units - the Areban, Sabunsuyu, Derdere, Karababa (A, B & C members) and Karabogaz formations. The most important oilbearing unit is the 100m-thick Derdere Formation which comprises low-tomedium porosity dolomite (lower portion) and tight limestone (upper portion). The upper formations in the field are ordered from bottom to top as follows: The Sayindere, Kastel, Germav formations, the Midyat Group and the Selmo Formation. None of these formations has any hydrocarbon potential.

43

DUAL POROSITY DETECTION the pressure and pressure-derivative response of a drawdown test. The first horizontal portion of the derivative curve represents radial flow in the fracture network. A dip in the derivative curve occurs when the matrix begins to contribute to the fluid flow, after most of the fractures have been drained. The time at which the dip appears on the derivative curve depends mainly on the contrast between the matrix and fracture permeabilities (λ). The size of the dip is controlled by the ratio of the matrix and fracture storativity values (ω). The smaller the fracture storativity, the bigger the dip and vice versa. The second horizontal portion in the derivative curve indicates homogeneous behaviour and characterizes the total system response.

DS

Most of the fluid that first enters a well in a fractured reservoir comes from fracture storage. Any oil contained in the lower-permeability matrix tends not to move until the fractures become depleted. At this point a pressure difference is set up between the fractures and matrix that 'sucks' oil out of the tighter rock. Eventually the two systems (matrix and fractures) start to reach equilibrium and the pressure distribution becomes analogous to that of a homogeneous system. These three distinct phases can be clearly seen in the measured pressure response at the well. Figure 4.4 shows

Fig. 4.3: Tool string used during the downhole testing operation in the Karakus Field.

SSARV

Dimensionless pressure groups

MIRV D Bourdet et al; World Oil Oct. 1983.

10

tpD/CD=3x105 1

10-1

PCT HRT

10-2 10-1

Build-up Drawdown, transitional Drawdown, homogeneous 1

10

102

103

104

105

106

tD/CD

Perforated tail pipe Downhole memory gauge

Dimensionless pressure groups

Safety joint Positrieve packer

Fig. 4.4: In this double-porosity reservoir the derivative of the pressure curves can be used to match build-up data to drawdown type curves even though the pressure curves do not match. D Bourdet et al; World Oil Oct. 1983.

10 tpD/CD=25

tpD/CD=100

1

10-1

Wireline entry guide

10-2 10-1

Crystal

Build-up (tp/C)D=25 Build-up (tp/C)D=100 1

10

Drawdown, transitional Drawdown, homogeneous

102

103

104

105

106

tD/CD Fig. 4.5: This illustrates the effect of insufficient drawdown time prior to build-up. The well was shut in during fissure flow.

PLT

44

CYAN MAGENTA YELLOW BLACK

Middle East Well Evaluation Review

Figure 4.5 shows two build-up pressure and pressure-derivative responses for two tests performed in the same dual-porosity reservoir but with different drawdown periods prior to shutin. Note that the shape and time at which the dips occur are now affected by the length of the flowing period prior to shut-in. It is essential that the flow period is long enough to encourage full radial flow to develop in the reservoir before shut-in. This ensures that the obtained values of λ and ω remain independent of the length of the drawdown. One of the difficulties in conventional surface testing of dual porosity systems arises when the contrast between the fracture and matrix permeabilities is such that the dip in the derivative curve occurs at early time. If this happens, wellbore storage effects will mask the appearance of the dip. Therefore, the analyst will lose all the valuable information about fracture characteristics which is reflected in the shape and time of occurence of the dip. Downhole testing techniques significantly reduce wellbore storage effects and guard against this danger. However, moving downhole does not eliminate all the problems. The shape of the derivative curve in a build-up test is not only a function of the wellbore-reservoir system but also of the previous flow history (figure 4.5). The most accurate well test data is obtained by having a long period of stable well flow before the well is shut in. This allows full radial flow to develop throughout the fracture network and matrix during drawdown and minimizes the effects of production prior to shut in. Schlumberger and TPAO decided that the best method of achieving a stable flow rate perturbation in the non-flowing wells was to have a constant negative well flow - in other words inject oil into the well. In this way the flow rate and its duration could be controlled and the reservoir could be subjected to a bigger perturbation, giving accurate results.

Number 12, 1992.

∆p and derivative groups, psi

104

103 Pressure response

102

101 Pressure derivative Matrix contributes to fluid flow 100 -4 10

10-3

10-2

10-1

100

101

102

∆t, hr Fig. 4.6: A comparison of the real data and the simulated results. There is an excellent agreement between both curves, suggesting that the selected model and its parameters reliably describe the reservoir’s dynamic behaviour.

Reliable results The results shed a great deal of light on the reservoir’s behaviour. Figure 4.6 shows a comparison of the real data and the simulated results. There is an excellent agreement between both curves, suggesting that the selected model and its parameters reliably describe the reservoir's dynamic behaviour. By analysing the pressure data, the engineers obtained the characteristic parameters of the fracture system and also detected two parallel impermeable boundaries: one 245ft and the other 710ft from the well (figure 4.7).

Test results: Distance to nearest boundary Distance to furthest boundary λ ω Wellbore storage coefficient kh Skin

= = = = = = =

245 ft 710 ft -7 1.4x10 0.20 0.0116 bbl/psi 36880 md-ft -2.57

Fig. 4.7: The tests helped the Karakus Field engineers to detect two impermeable boundaries near to the well.

50

6 -1

00 -17 50 -17 0 0 80 -1 -185 0

90

-1

45

Suez

N

5800

00

56

00

52

54

G

00

u l f

56

00

o f

Badri Field

S u e z Producer El Tur Injector

0

50 km El Morgan Field 5600 0

550

5600 580

0

6000

6200

6400

Injection pressure (psia)

3100

Fig. 4.9: TAKE A BREAK: To determine the fracture pressure gradient, the injection pressure is plotted against the injection rate. When the fractures open in the formation, they produce a distinct break in slope on the plot indicating an increase in injectivity. This point also marks the fracture pressure gradient. In this plot, the fracture pressure gradient is 0.468psi/ft.

3000

2900 FG=0.468psi/ft 2800

2700

2600 1

2

3

4

5 6 7 8 Injection rate (rps)

48

CYAN MAGENTA YELLOW BLACK

Fig. 4.8: DOUBLE TROUBLE: Injectionbased well testing had to be carried out in Egypt's Badri and ElMorgan Fields to help understand problems of formation breakdown. The primary oil trapping mechanism for the Badri Field is thought to be a combination of block faulting and folding with local stratigraphic variations. The Belayim Formation consists predominantly of sandstone interbedded with finergrained dolomitic silts and shales. The sandstone beds are separated by mudstoneshale intervals. The El-Morgan Field is an elongated northwest-trending, faulted anticlinorium containing two major structural faults. The main producing horizon is the Kareem Formation which consists of stacked coarsening-upwards sequences that represent the progradation of deltaic sand lobes and openmarine mudstones.

9

10

11

Middle East Well Evaluation Review

Finding out about formation breakdown High injection rates are normally used to improve the performance of waterflood operations. But sometimes the resulting high pressures can induce fractures that seriously affect the vertical and horizontal sweep efficiencies. In Egypt’s Badri Field this problem is particularly acute because formation breakdown begins at very low injection pressures. A layered reservoir test was the most cost-effective way of

A

conventional way of measuring the fracture parting pressure is to carry out a Step Rate Test (SRT). During an SRT, the injection rate into the entire formation is increased in steps while the injection pressure is measured downhole. When fracturing starts there is a sudden change in the flow rate and this can be used to estimate the fracture gradient. The injection pressure is plotted against the injection rate (figure 4.9). Two clear slopes can be seen. The first line represents the injectivity of the formation before parting occurs. The second slope indicates the increase in injectivity due to the presence of induced fractures. However, the Badri Field consists of multiple sandstone layers with different degrees of depletion. Therefore, the simultaneous flooding of all these layers during an SRT would provide inaccurate parting pressures. To overcome this problem, the engineers decided to opt for a Layered Reservoir Test (LRT). This involved lowering the PLT to various points above and between the producing layers to measure changes in injection rate and pressure. The primary oil-trapping mechanism for the Badri Field is thought to be a combination of

block faulting and folding with local stratigraphic variations. The Belayim Formation consists predominantly of sandstone interbedded with finergrained dolomitic silts and shales. The sandstone beds are separated by mudstone-shale intervals. This study investigated two continuous beds - Hammam Faraun 1 and 3 (HF1 and HF3). The Badri Field has 21 producing and 10 injection wells (figure 4.8). The Badri D3 injection well was selected as the active well for the test which consisted of four transients: changing the well-head injection rate from 7,000BWPD to 3,000BWPD, then to 8,000BWPD, to 13,000BWPD and finally to zero by shutting-in the well. The two selected reference positions for the PLT were in the maximum flow zone (above the top perforation) and between layers HF1 and HF3. Figure 4.10 shows the injection rate sequence and the tool’s position throughout the test. In each case the PLT was moved into its new position 30 minutes before the injection rate was changed and was left there during the transient. Injection rates and stationary readings between the perforations were taken at the end of each transient.

calculating the optimum injection rates that could be safely used

13000

without fracturing the formation.

possible to determine the

Water injection (BWPD)

Using this technique, it was

8000

1000 3000

formation parting pressure for each 6440 Depth (ft)

producing interval.

HF1

Tool position

6545

HF3

6700 Fig. 4.10: Sequence of events during the layered reservoir test. The shaded areas represent the four transients and the arrows show the positions and times of the flow profile surveys. Number 12, 1992.

49

500

2800 Pressure (psia and spin) Dim

Dt = 1.5 hours Pressure variations (psi)

400

300

200 m = 92 psi/rps 100

0 0

2

4 6 8 Spinner variations (rps)

10

12

Pressure (psia and spin) Dim

2850

2770

p vs dt Scaled spin vs dt 2710

2620 FG = 0.44 psi/ft 2530

2440

2350 -3 10

10-2

101

102

Fig. 4.11: (Top left): This shows an injectivity of 92psi/rps for the bottom layer. At 1.5 hours into the transient, the fracture closure is marked by a sudden drop in injectivity.

p vs dt Scaled spin vs dt FG = 0.48 psi/ft FG = 0.46 psi/ft

Fig. 4.12: (Top right): During the first transient the pressure suddenly increases by 100psi when the fractures close. 2690

2610

2530

2450 -3 10

10-2

10-1 100 Elapsed time (hr)

101

102

Rate normalized pressure variations

500

Fig. 4.13: (Left): During the second transient there is a sudden increase in flow rate but no corresponding pressure increase. There is also a sudden pressure drop at 0.6 hours into the test. Fig. 4.14: (Left): This generalized Horner plot of the second transient indicates that radial flow only occurs prior to fracturing.

Transient 2 SUPPOS.FUN: 7.dd 400

300 Dt = 0.06 hours 200

100 Dt = 0.03 hours 0 0.4

2880 Pressure (psia and spin) Dim

10-1 100 Elapsed time (hr)

2850

0.6

0.8 1.0 1.2 1.4 1.6 1.8 Superposition time function (*104)

2.0

2.2

Fig. 4.15: During the third transient, fractures open at 0.39 hours into the test and this causes a pressure drop but the flow rate remains constant.

p vs dt NSPINDT FG = 0.49 psi/ft

2820 2790 2760 2730 2700 -3 10

10-2

10-1 100 Elapsed time (hr)

50

CYAN MAGENTA YELLOW BLACK

101

Watching the formation fracture The first transient test was created when the injection rate was allowed to drop from 7,000BWPD to 3,000BWPD. The pressure decline was measured with the tool between HF1 and HF3, at 6545ft. As the pressure drops, the fractures close. This event can be seen in figures 4.11 and 4.12. Flow rate and pressure decrease for 1.5 hours, after which the flow rate stabilizes and the pressure suddenly increases by 100psi. This value of pressure gives a fracture gradient of 0.413psi/ft. After 4.4 hours, fractures open in the top layer, causing the injection rate and the pressure to drop suddenly. This fracturing happened when the pressure reached 2,497psi, giving a fracture gradient of 0.44psi/ft. The second transient was created by increasing the injection rate from 3,000BWPD to 8,000BWPD. Prior to changing the rate, the tool was moved to a position above both layers at 6,440ft. Two main events occurred (figure 4.13). At 0.075 hours into the transient, there is a sudden increase in injectivity without a noticeable increase in the formation pressure. This suggests that there is less resistance to water flow into the formation. The pressure continues to increase up to 0.6 hours into the test, when it suddenly starts to drop. This is probably caused by either two fractures opening in succession, or vertical extension of the first fracture. This gives pressure gradients of 0.465psi/ft and 0.479psi/ft respectively.

102

Middle East Well Evaluation Review

PREDICTING FRACTURE GRADIENTS

Young's modulus 0 (psi) 20 Bulk compressibility 0.0 (psi) 2.5 Poisson's ratio 0.5 0

Hydrocarbon Water Sand Shale Dolomite

Overburden press grd 1 (ps/f) 0 Porosity Parting press grd Delta T shear (pu) 0 1 (ps/f) 0 240 (us/f) 40 100 Volume of clay Pore press gradient Delta T compressional (pu) 100 1 (ps/f) 0 240 (us/f) 40 0

Fig. 4.16: IN THE RED: Mechanical properties log (MECHPRO*) obtained from a producing well in Badri Field. The fracture parting pressures are shown in red in the second track. These are computed using log-derived rock mechanical properties (track 1) and a horizontally constrained elastic model.

In an isotropic and homogeneous reservoir the pressure needed to initiate a fracture can be estimated using a relationship which links minimum and maximum horizontal stresses, the rocks tensile strength, Biot’s elastic constant and pore pressure. However, the minimum horizontal stresses cannot be measured with current technology. Assuming the reservoir is a horizontally constrained isotropic elastic model, and neglecting deformations due to thermal variations, minimum horizontal stress can be written as a function of the overburden pressure, Poisson s ratio, Biot's constant and pore pressure†. Poisson's ratio and Biot’s elastic constant can be derived from the seismic shear and compressional wave travel times (∆ts and ∆tc). ∆ts was not recorded at the time the injector wells were drilled, but new producer wells higher up in the structure have ∆ts data. These mathematical relationships together with data from neighbouring wells have been used to investigate the effect of pressure depletion on fracture gradient reduction. The results show fracture gradients considerably higher than the recorded values obtained from the LRT test (figure 4.16). Therefore, the assumptions that the reservoir stresses are horizontally uniform and that thermal variations cause negligible deformations cannot be justified. This leads us to the conclusion that both the tectonic nature of the area and thermally induced stress reduction, due to cold water injection (see box on page 53), must be considered in any theoretical calculations.

†GR Coates and SA Denoo, 1981; Mechanical Properties Program Using Borehole Analysis and Mohr's Circle, SPWLA 22nd Ann. Log. Symp. Trans.

The rapid drop in pressure and flow rate at about 4 hours is due to a surface problem, and not a reservoir response. Figure 4.14 is the generalized Horner plot of the test. The slope of the straight line is greater than that plotted from the later fall-off test after the effect of wellbore storage and fractures have disappeared. This suggests that radial flow is happening in only a part of the tested interval. This portion of the tested interval might be related to the height of the induced fractures.

Number 12, 1992.

For the third transient, the injection rate was increased to 13,000BWPD. The tool was left in its position above the two layers for this part of the test. The response of the flow rate and pressure to the increasing injection rate can be seen in figure 4.15.

51

1000 m = 18 psi/rps 800 600 Dt = 0.03 hours

400

m = 18 psi/rps

200 0 0 3100 Pressure (psia and spin) Dim

Fig. 4.18: By 0.06 hours the injection rate has fallen to zero and the well has gone on vacuum.

Parting shots Table 4.1 summarizes the measured injection rates for each of the two reservoirs, HF1 and HF3, and the corresponding wellbore pressure potentials. This is shown graphically in figure 4.19. By extrapolating the lines into the region where they intersect, engineers calculated that the fracture gradients lay somewhere between 0.43psi/ft and 0.48psi/ft. Patterns established by the transient data showed that the changes in injectivity were caused by fractures.

1200

Pressure variation (rps)

Fig. 4.17: As the injection rate decreases the fractures begin to close at 0.03 hours.

20

25

Change in injectivity

2540

2260

1980

Table 4.1: The measured injection rates for each of the two reservoirs.

10-2

10-1 100 Elapsed time (hr)

101

Injection period

HF1 rate (B/D)

HF2 rate (B/D)

Wellbore pressure potential (psi)

1 2 3 4 5

1100 2380 2483 4050 392

2023 4857 5496 9254 -392

2471.1 2736.1 2713.1 2836.0 1706.4

102

3000

2600

2200 Top reservoir - HF1 Bottom reservoir - HF3 1800

1400

1000 -1

CYAN MAGENTA YELLOW BLACK

10 15 Spinner variations (rps)

Well goes on vacuum

1700 -3 10

Fig. 4.19: The fracture gradients can be calculated from this plot of pressure potentials.

52

5

Falloff p vs dt Scaled spin vs dt

2820

Injection pressure potential (psia)

At 0.39 hours into the test, a fracture opens or grows in one of the two layers and the pressure begins to drop. The flow rate remains constant as the tool is above the layers. This occurs at a pressure of 2,844psi and gives a fracture gradient of 0.48psi/ft. For the final transient, the injection rate was reduced to zero by shutting in the well with the tool stationed between the layers at 6,545ft. As the injection rate decreases, the plot of flow rate against pressure (figure 4.17) shows that at 0.03 hours into the fall-off, fractures are beginning to close. This is signalled by a sudden change in injectivity. This event can also be seen in figure 4.18, at a pressure of 2,667psi. The pressure pattern at 0.06 hours shows the injection rate has fallen to zero and the well begins to go on vacuum (when the pressure in the wellbore equals that of the formation). From then onwards, until the spinner stops, the injectivity is almost identical to that observed in the first transient before the fractures close.

0

2

4 6 3 Injection rate (BWPD *10 )

8

Middle East Well Evaluation Review

10

SEAWATER ON THE ROCKS Seawater is the first choice of injection fluid in the Gulf of Suez area. It reaches the reservoir at temperatures between 80°F and 90°F. This is much colder (by up to 100°F) than the reservoir itself. As a result, the reservoir is rapidly cooled from its original temperature and this introduces compressive stress and reduces the fracture gradients. A recent study of these effects on the Prudhoe Bay Field, Alaska, USA†, reported a maximum reduction in horizontal stress of 0.08psi/ft after one year’s cool-water injection. This theoretical prediction was validated with field data. Waterflooding operations in this field result in temperature reductions of 130°F and fracture gradient reductions from 0.63psi/ft before waterflooding to 0.55psi/ft after water injection. As temperature reductions in the Gulf of Suez area are known to be less severe, the minimum horizontal stress value of 0.08psi/ft can be used as an upper limit for the formations. If the Belayim Formation is considered to be horizontally uniform but is assumed to have a constant horizontal stress

reduction of 0.08psi/ft due to water cooling, the theoretical fracture gradients (TFG) and measured fracture gradients (MFG) for the two layers HF1 and HF3 would be:

Calipers 6"

16"

Fracture gradients (psi/ft) HF1

HF3

TFG 0.63

0.70

MFG 0.44

0.45

100ft

Stressful times ahead So, including thermally induced stress in the mathematical model still does not account for the fracture gradients obtained through well testing. The difference between the above values for each layer must be attributed to tectonic imbalances. In other words, uneven stress distribution. These can be observed in the caliper log in figure 4.20 which shows borehole elongation over the entire Belayim Formation.

Minimum stress † AM Garon, CY Lin and VA Dunayevsky, 1988: Simulation of Thermally-Induced Waterflooding Fracturing in Prudhoe Bay; SPE paper 17417.

Fig. 4.20: OVAL AND OUT: Caliper logs showing that the borehole cross section has an oval shape (see diagram left).

Number 12, 1992.

53

Pulse tests solve injection problems

Pressure maintenance No-flow boundary

Mode

lled re

C-4

servoir

area

C-5 C-1

C-8

C-3 C-7

Fig. 4.21: This schematic of Egypt's Badri Field shows the configuration of producing and injection wells. The yellow rectangle delineates the area modelled in the reservoir study.

The operator of Egypt's Badri Field suspected a leak towards the neighbouring El-Morgan Field and a lack of communication between injectors and producers. Could pulse testing shed some light on these problems?

54

CYAN MAGENTA YELLOW BLACK

A

pulse test was carried out on the Belayim Formation in the Badri Field to find the degree of hydraulic communication between the producing and test wells. It was also used to establish a mathematical model and its parameters for part of the reservoir. The test involved six wells, Badri C1, C3, C4, C5, C7 and C8. Well C3 was chosen as the injection well, C5 as the producing well, and the others were used for observation. Figure 4.21 shows the layout of the production/injection wells. Water was injected into C3, which lies south of the producing wells. C3 was subjected to an alternating sequence of injection and shut-in periods of 36 hours. The consequent pressure response in the observation wells was monitored for 12 days. The regions around two of the wells were mathematically modelled according to their pressure responses. The data from C7 indicated that the reservoir could be modelled as being rectan-

gular, of constant thickness, whereas C1 was best modelled as a circular reservoir with its centre at the observation well. The response in the other wells was tested against the best of these models. The pulse test sequence in well C3 consisted of four injection and three shut-in periods. After shutting in all the wells in the test, crystal gauges with extended memories were run in each observation well (C1, C4, C5, C7 and C8) and set 20ft above the top perforations. Water injection began at C3 at a flow rate of 10,400BWPD, and lasted for 35 hours. This well was then shut in for 36 hours before being opened for another injection period of 10,900BWPD for 36 hours. The well was re-opened again, and injection was resumed at a flow rate of 10,800BWPD for 47 hours. The final shut-in was for 25 hours before reopening to a 36-hour injection at 10,700BWPD. The resulting pressure changes in the observation wells were recorded.

Middle East Well Evaluation Review

50

15.0 Observed pressure variations - psi Simulated pressure variations - psi Test rate sequence - psi

40

12.0

35 30 25 20 15 10 5 0

Observed pressure variations - psi Simulated pressure variations - psi Test rate sequence - BWPD 10,000

13.5 Pressure variations - psi

Pressure variations - psi

45

10.5 9.0 7.5 6.0 4.5 3.0 1.5

0

26

52

78

0.0

104 130 156 182 208 234 260 Elapsed time (hr)

Fig. 4.22: The sequence of tests and corresponding well C7 response.

Figure 4.22 shows both the test injection and shut-in periods (blocks on the x axis), and the corresponding response of well C7 (dots). As the reservoir pressure trends affect the recorded pressure signals at each well, they are subtracted before analyzing the data. All the figures in this article are adjusted in this way. The pressure response at C7 was analyzed using history matching, ie by finding the mathematical model that shows the identical response when subjected to the same disturbances as the real system. This turned out to be equivalent to a rectangular reservoir of constant thickness, with two sides having pressure maintenance and the other two sides with no-flow boundaries (figure 4.21). The C7 reservoir model is defined by the parameters in table 4.2. The flow capacity (kh) was derived from the following equation: kh = 162.6Bw µw /m' Storativity(φhc t ) was estimated by means of the time match equation: φhct = 0.0002637kh/µwAtm where A is the reservoir area in square feet.

0

30

60

210

240

Fig. 4.23: The sequence of tests and the corresponding well C1 response.

Once the model had been defined, the simulated response was plotted together with the real pulse test data. The close fit, and consequently the accuracy of the model, can be seen in figure 4.22. The response of well C1 to the pulse test is shown in figure 4.23.

There is a good fit between actual data and the response simulated by a proposed circular reservoir model with a constant pressure boundary at a radial distance of about 2,765ft. The model’s parameters are given in table 4.2.

Table 4.2: Parameters and equations defining the C7 and C1 model reservoirs. Badri C7 model Bx By xa/Bx ya/By xob/Bx yob/By tm m' φhct kh

Badri C1 model 9,800 ft 8,375 ft 0.47 0.48 0.55 0.83 5.56E-04 l/hr 9.00E-03 psi/B/D 1.06E-04 ft/psi 8,044 md-ft

ReD RD re r m’ tm kh φhct

7,400 5,150 2,590 ft 1,804 ft 6.5E-03 psi/B/D 3.23E-03 1/hr 11,140 md-ft 1.00E-04 ft/psi

Table 4.3: Reservoir and fluid parameters: Average porosity, φ = 0.25 Average net pay thickness, hc3-c7 = 76 ft Average net pay thickness, hc3-c1 = 110 ft Water viscosity, µw = 0.44 cp Water formation volume factor, Bw = 1.012 Total compressibility, ct = 6.2E-06 1/psi Wellbore radius, rw = 0.35 ft Initial flowing pressure, pwfc7 = 1750.51 psia @ 7052 ft Initial flowing pressure, pwfc1 = 1485.49 psia @ 5594 ft

Nomenclature. Bw = water formation volume factor Bx = reservoir length By = reservoir width kh = flow capacity m’ = slope of superposition plot r = distance between active and observation wells. rD = r/rw, dimensionless wellbore radius re = radial distance to external boundary reD = re/rw, dimensionless distance to external boundary

Number 12, 1992.

90 120 150 180 Elapsed time (hr)

re xa xob ya yob φ ψ φhct µw

= = = = = = = = =

wellbore radius x coordinate of active well x coordinate of observation well y coordinate of active well y coordinate of observation well porosity pressure potential storativity water viscosity

55

If there is hydraulic communication across a formation, pressure changes in one area of the field will have repercussions in more distant places. The presence of gas between wells can act as a shock absorber and can dampen the pressure pulse as it is transmitted through the reservoir. Pulse testing gives the reservoir a sudden, sharp shock. The test is a special form of multiple well testing and uses a series of short-rate perturbations at the active wells. The test may last a few hours or several days. Pulses are created by alternating periods of injection/production and shutin. The pressure responses to the pulses are measured in one or more observation wells and since the pulses are of short duration, the pressure response is usually small. This means that special equipment is needed to measure the small pressure variations. The main advantages of the pulse test compared with interference tests are: • The short duration of the pulse • Reservoir pressure trends and noise can be automatically removed using appropriate analysis techniques. A simple way of understanding pulse testing is to imagine what happens when a stone is dropped into the middle of a duck pond. The ripples spread radially away from the stone and reach ducks floating on the pond at slightly different times and magnitudes, depending on the location of the bird. As with wells connected by reservoirs of oil or water, the arrival of the ripples tells the ducks that a stone has been thrown. A duck sitting in a reed bed will experience events differently. Its position is analogous to an observation well being separated from the injection well by free gas. The duck feels a disturbance which is much distorted and reduced due to absorption of the wave energy by the reeds. This duck is aware that something has disturbed the pond, but not much more. Another exception would be a duck floating behind a nearby concrete jetty. This duck, like the well separated by a fault or break in the fluid connection, will not feel any of the ripple caused by the splash of the stone.

56

CYAN MAGENTA YELLOW BLACK

100ft

PONDERING PRESSURE

Top of formation

Top of formation

Fig. 4.24: Gamma-Ray and Density-Neutron logs of wells C1 and C7. Note the clear difference of 70ft in the formation tops between wells.

Middle East Well Evaluation Review

25.0

20.0

Model derived from C3-C7 and Sg=0.05 Observed pressure response - psi Simulated pressure response - psi

1.8 Pressure variations - psi

Pressure variations - psi

2.0

Model derived from C3-C7 Observed pressure response - psi Simulated pressure variations - psi

22.5

17.5 15.0 12.5 10.0 7.5 5.0

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2

2.5

0.0

0.0 0

16

32

48

64 80 96 Elapsed time (hr)

112 128

Fig. 4.25: Recorded and simulated pressure response of well C8.

However, the responses of both wells cannot be matched with one model. This is partly due to different reservoir trends of C1 and C7 (see table 4.3). Also, the initial flowing pressures (pwf) of these wells are not the same and, even with gravitational terms removed, there is still a difference in pressure potential (ψ) between them: ∆ψ =pwfc7 - pwfc1 - 0.433∆h where ∆h = ∆ψ =

70ft 235psi

Logs from both boreholes show a difference in elevation of 70ft between the wells (figure 4.24). This difference in pressure potential indicates that there is fluid movement from well C7 to well C1. The reservoir model used in C7 seems to give a better represention of the overall area. The next step was to use the model derived from C7 well to predict the responses of wells C5 and C8. These could then be compared to the actual pulse test results. Plots of the simulated response for each well indicate that the characteristics of the area between C3 and C7 are quite distinct from that around wells C5 and C8. Figure 4.25 shows the comparison between recorded pressure variations at well C8 and values simulated by the C7 model. The lack of fit between both data sets are due to changes in the flow rate and/or storativity (φhct) over the zones of influence of wells C8 and C5, compared with those between wells C7 and C3. No response to injection was seen in well C5, indicating the likelihood of a sealing barrier between C5 and C3.

Number 12, 1992.

144

160

0

16

32

48

64 80 96 Elapsed time (hr)

112

128 144 160

Fig. 4.26: Modelled response assuming a 5% gas saturation.

Shock-absorbing gas The absolute pressures measured in wells C5 and C8 are about 400psi below the bubble-point pressure. This suggests that the small amplitude of the signal observed at well C8, and the lack of response from well C5, could be due to the presence of free gas towards the northwest of the reservoir area. To test this theory, a simulation was made to determine the pressure response at well C8 with a gas saturation (Sg) of 5%. Figure 4.26 shows the results, together with the actual response at well C8. Although the curves do not fit particularly well, their magnitudes are comparable. So the free gas assumption seems to hold true. The analysis techniques used throughout this article assume that the reservoir is isotropic and homogeneous in the region influenced by the test wells. This assumption may well be valid when we are dealing with reservoirs that are single-phase (eg totally oil- or gas-filled) or multi-phase when the fluid properties are similar. But in the situation where there are two distinctly different fluids - ie gas and oil the analytical solution does not hold true.

Therefore, the presence of free gas precludes us from obtaining a reliable answer using analytical solutions. In such cases, a numerical model must be used but this approach is more costly and time-consuming. Pulse testing in the Badri Field has shown there is hydraulic communication between wells C3, C1, C7 and C8, but not between these wells and C5. Therefore the producing well may lie behind a fault or other sealing boundary, preventing it from responding to injection. The responses showed that the wells are in hydraulic communication and this enabled the reservoir engineers to create a model for the area. The presence of gas towards the northwest of the Badri Field was also detected. These two results also proved that there were no leaks into the El-Morgan Field as originally suspected.

57

Suggest Documents