Camille Ri 2017
Camille Ri 2017
L. Camilleri, M. El Gindy, A. Rusakov, I. Ginawi, and H. Abdelmotaal, Schlumberger; E. Sayed, T. Edris, and
M. Karam, Khalda Petroleum Company
This paper was prepared for presentation at the SPE Electric Submersible Pump Symposium held in The Woodlands, Texas, USA, 24-28 April 2017.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.
Abstract
To arrest production decline without infill drilling, one must maximize production from existing wells,
typically by identifying wells with skin and increasing drawdowns on wells with good pressure support or
lower water cut. This paper examines how high-frequency, high-resolution flow rate measurements on ESP
wells can identify such opportunities without the need for buildups which cause production deferment. The
application of this workflow was examined for wells in Egypt.
To obtain flow rate measurements at frequencies greater than once an hour, without dedicating a test
separator or multiphase flowmeter to each well, the method relied on real-time data to calculate liquid rate
and water cut. The liquid flow rate calculation was based on the principle that the power absorbed by the
pump is equal to that generated by the motor. Water cut was calculated by modelling the production tubing
as a gradiometer. Analytical equations ensured that the physics were respected at all times, which yields
greater repeatability and resolution than analogous methods based on correlations and artificial intelligence.
The well analysis in Egypt demonstrated that the evolution of depletion and skin could be identified
qualitatively using plots of rate-normalized differential pressure. These diagnostic plots are only possible
with high-frequency and high-resolution flow rate measurements and could not be generated using
traditional monthly production test data. The case studies also illustrated how frequency and resolution
enabled real-time measurement of the impact of small changes in pump speed on both the reservoir inflow
characteristic as well as production. This qualitative technique makes it possible to fine-tune production
iteratively without the need for time-consuming simulation, which was nevertheless also conducted to
quantify the changes in reservoir pressure and skin on the wells considered in this case study. Furthermore,
with a water cut resolution of less than 1%, potential water coning can be identified rapidly, which allows
the production operator to test small drawdown increases. Finally, this method also has the advantage that
it can reduce the mobilization of testing equipment to the well site to measure the change in production,
thereby minimizing and eliminating health, safety and environment (HSE) risks in remote locations while
also optimizing the use of the available test packages.
This novel use of real-time gauge data demonstrates how a cost-effective method can improve well testing
quality and thereby identify production optimization opportunities, providing the means to arrest decline.
This case study provided a proof of concept on specific wells, however fieldwide application is necessary
2 SPE-185144-MS
to identify the wells with the highest production optimization potential because, typically, most of the gain
is obtained from a minority of the wells in a given field.
Introduction
Khalda Petroleum Company (KPC) produces over 670 wells in the Western Desert of Egypt with over 230
wells (34%) lifted with ESPs and the rest of the wells utilizing other forms of artificial lift such as beam
Pumps and progressive cavity pumps (PCPs). Roughly 54% of the total production of 151,000 BOPD comes
from 34% of the wells lifted using ESPs, which is as expected because the ESP lifted wells represent the
highest producers. Production covers several fields and reservoirs with a very wide range of lithologies and
reservoir drive mechanisms, which results in each well having its own unique production characteristics and
challenges. Testing and inflow monitoring are therefore essential to production optimization and arresting
the natural decline of production, especially in a mature field environment where infield drilling is often
uneconomical in a low oil price environment, which increases the importance of producing each well to
its full potential.
In addition to the wide heterogeneity, challenges are further compounded by many wells suffering from
low well testing frequency, which means that meaningful flowrate trends are difficult to obtain. KPC
carry out well testing with mobile packages, which are mobilized to the well site for each flowrate test.
Consequently, testing is prioritised because of the limited number of testing packages per field and the
logistical challenges of moving equipment in the desert. Well testing resources are typically allocated by
KPC in the following way:
– High priority is given to testing wells following a workover.
– High priority is also given to wells that are either high producers or where additional reservoir
information is required. For such wells, testing frequency can be as high as once per week, as will
be seen further on, well SIWA 3-R6 is such an example.
– On lower profile wells, testing is only conducted when there is a change in trend, which is observed
from real time pressure readings where the ESP is fitted with a gauge and this is connected to a
SCADA system.
As a result many wells are only tested once per quarter, which hinders effective reservoir surveillance.
In an effort to generate meaningful liquid rate trends, a trial was conducted to evaluate if testing with high
frequency, high resolution and high repeatability could be performed cost effectively on all ESP installations.
The trial also examined if these liquid rate trends would enable calculation or measurement of the following
parameters:
– Skin calculation without build-ups for example pressure transient analysis (PTA) in drawdown.
– Depletion evolution
– Connected pore volume and therefore drainage area.
The field trial was conducted on four wells, which were selected from four different fields and reservoirs
to provide a representative sample for the trial across KPC’s various reservoirs. These wells are located
in three concessions (Razak, Kallabsha and Siwa) in the Western Desert which are shown in a map in
Fig. 1. A summary of the four wells is provided in Table 1 along with key information on their associated
reservoir characteristics and fluid properties. Geological maps are also shown in Fig. 2, Fig. 3, Fig. 4
and Fig. 5. These figures serve to demonstrate that the reservoirs are highly faulted, that each well must
be monitored independently and it is not possible to rely on just macro-level production profiles. This
complexity reinforces the importance of dynamic measurement of drainage areas. For the purpose of
the trial, Khalda conducted frequent physical wells tests with their mobile test separator to compare the
calculated liquid rate with the measured test separator results.
SPE-185144-MS 3
Lithology and datum Clastic Sandstone AEB 1. Sandstone Upper Sandstone AEB-3C Sandstone Lower SAFA
Unit 1 at 7658 ft TVD SS Bahareya (UBAH) Toppay Datum depth Datum depth 13900 TVD SS
at 5,980 ft. TVD SS 9,580 SS TVD SS
2. -Carbonate Abu
Rawash G (ARG)
at 5,700 ft. TVD SS
Completion Type Single Reservoir 2 Comingled Zones Single Reservoir Single Reservoir
Drive Mechanism Active Water Drive but Depletion Partial Water Drive Active Water Drive
cannot see constant pressure
boundary as neighbouring
producers create notional
no-flow boundary
Top of Perforations Just 7651 ft TVD SS 4933 ft TVD SS 9436 TVD SS 13581 TVD SS
show a single number for
top of perfs TVD depth
ESP Setting depth TVD 5561 ft TVD 4890 ft TVD 8725 ft TVD 8137 ft TVD
Differential Vertical Depth 2090 ft TVD 43 ft TVD 711 ft TVD 5713 ft TVD
between ESP gauge and top
of perfs
API 33 20 to 25 38 36
Bubble Point 300 - 500 psia 300 psia 100 psia 100 - 200 psia
Reservoir Temperature 200 deg F 170 deg F 250 deg F 260 deg F
a. As will be discussed further on, WC requires a digital wellhead pressure gauge, which is not always
available. Therefore this technique decouples liquid rate (oil + water rate) from WC calculations,
which was important during the trial because the two wells in the Razak field were not fitted with
gauges at the wellhead.
b. Most of the inflow characterization such as productivity index (PI) measurement, PTA in
drawdown and flowing material balance to calculate reserves can be performed with just liquid rate
measurements, especially in an undersaturated reservoir. The only caveat is if the ESP is set at a
substantial depth above the perforations, in which case WC is required to correct flowing pressure
from gauge to perforation depth in order to capture reservoir response as opposed to wellbore
hydraulic effects.
4. Unique solution across the entire pump curve not just the recommended operating range: The power
equation provides a unique solution across the full flow-rate range of the pump curve irrespective of
the pump type, which is not the case with the traditional pump-rate calculation method, which are
usually limited to the recommended operating range. A comparison of the two methods is explained
below and illustrated in Fig. 7.
a. The traditional method used in the industry for calculating liquid flow rate involves using the
pump characteristic head-flow curve and a measurement of the pump differential pressure. This
can be impossible to use at flow rates below the best efficiency point (BEP) because some pumps
in the industry exhibit either a very flat or saddle curve shape; see Fig. 7 for an illustration of the
three main head-flow curve types found. In such cases, the flow-rate solution for a given measured
differential head is non-unique.
b. In Fig. 7, the ratio ηp/Qp is plotted against normalized flow rate for the same three pump types to
illustrate that there is a unique solution for any measured DP/Power (ratio of pump differential
pressure over power). This demonstrates that the power equation method provides a unique solution
across the full flow rate range of any pump independently of the pump head-flow curve shape.
5. Ability to capture transients: The power equation takes advantage of the low inertia of an ESP and
the quasi instantaneous response of current to changes in required power caused by a change in flow,
differential head and/or specific gravity. It is therefore very well suited to capturing flow transients
such as start-up conditions and flowrate variations due to pump surging caused by gas.
6. PHI: The use of the power equation allows the user to take advantage of the PHI in the two ways
listed below with details of the algorithm provided in Appendix C.
a. The PHI provides an "alarm" defining when the liquid rate model requires recalibration i.e. when
PHI deviates from 1.0. This is equivalent to having an alarm of pump degradation.
b. The initial calibration of the liquid rate model can be performed without a physical meter.
7. Real-time flow-rate calculations: Because the algorithm does not require any iterative calculations, it
lends itself to being programed into an engine, which can service multiple wells with high frequency
data e.g. updates twice per minute, which is equivalent to 1 million points per year per analogue signal.
This was one of the aspects that was tested during the field trial (i.e. testing whether liquid rate and
WC could be delivered to the production and reservoir engineer’s desk in real time via the Internet).
This was proven during this field trial as was also the case during another field trial documented by
Camilleri et al. (ADIPEC 2016). The IT architecture deployed is shown in Fig 8. and has the advantage
that it can receive and send data to any historian.
8. Provides real-time pump operating point: Most pump operators keep a close eye on where the pump is
operating relative to the BEP. This is because, depending on the pump mechanical configuration and
material selection, there is less mechanical wear on the pump and power consumption is minimized
when operating within the recommended range (i.e., as close to BEP as possible). Furthermore,
8 SPE-185144-MS
head degradation caused by free gas is increased when operating at flow rates below the minimum
recommended operating flow rate, as explained by Gamboa and Prado (2012). Calculating the liquid
rate using the power equation provides the downhole average flow rate through the pump directly
without the need to correct the surface measured rate to downhole conditions using formation volume
factor (FVF) effects. Therefore, this analysis is independent of WC calculation and fluid pressure/
volume/temperature (PVT) properties and solely dependent on the downhole liquid rate calculation.
To illustrate this aspect of the workflow, the calculated rates for the four wells have been plotted in Fig.
9, Fig. 10, Fig. 11 and Fig. 12 to show both the absolute downhole rate as well as the operating point,
which is expressed as a fraction of the flowrate at BEP, which is therefore frequency independent.
– The liquid rate calculation effectively automates an often laborious and repetitive task for many
field production engineers who conduct nodal analysis on a regular basis to monitor pump
operating point.
– Since there is often uncertainty associated with bubble point and FVF data, this method provides
a means to calculate downhole rate without any prior knowledge of PVT.
– Subsequent ESP installations benefit from this analysis because it provides advice on whether
additional stages are required to operate an ESP within its operating range.
Figure 7—Nondimensional pump curves illustrating the three main types of pumps found in the industry
and the ensuing impact on the two main methods for calculating liquid flow rate based on gauge data
Figure 9—Production log for WKAL-A4 with and without calculated liquid rate and WC
the back pressure on the pump. Although real-time WHP was not available, this could be seen in the
discharge pressure readings which showed a step change increase of 35 psi (see Fig. 12), which should
be associated with a drop in flow, which was not capture by the test separator. Once again, one notes
that test separator rates can lack repeatability. In this case, the test separator did not capture the true
change in well performance associated with the choke change.
Figure 13—Calibration of liquid rate using PHI on well WKAL-A4. There are multiple points showing how the
ESP operates across a wide range of points on the pumps curve as a result of transients during each start-up
One of the key lessons to extract from the field trial is that although test separators provide the initial
calibration, test separator data must be treated with caution as it often suffers from a lack of repeatability,
which is often caused by the short duration of testing, especially on wells with a low flowrate. This is
explained by Theuveny and Mehdizadeh (2002) who demonstrated that testing accuracy improves as test
durations increases. Test data validation is done in one or all three of the following ways:
i. Traditionally this is done with nodal analysis, preferably in time-lapsed mode to observe trends.
ii. Another way is to compare test and calculated rates after initial calibration. When the test rates are
greater than a previously calibrated PCL rate, then it is highly likely that the test separator rate is
suspect because ESPs do not gain in efficiency.
iii. However the best method for differentiating between pump degradation and incorrect test separator
data is with the PHI as the user can eliminate pump performance degradation where PHI=1.0.
An important outcome of this particular trial was that the power equation did provide correct flowrates
using a single calibration when PHI=1.0 and the two main reasons are:
i. An analytical solution is used, therefore the physics are respected at all times.
ii. The solution is independent of SG, therefore calibration is maintained despite the large variations
in WC.
calculate the voltage that would minimize the current and thereby minimize both consumed power and
motor operating temperature without any change in pump frequency and production. An example of such
real-time calculations is shown in Fig 14. for well WKAL-A4. Normalized voltage is used, as expressed in
Eq. 1, to remove the effect of frequency when plotting versus time.
(1)
In this case, one can observe the increase in absorbed power as the load factor rises to a maximum of
1.10 as a result of rising WC. The motor voltage was too low at 0.9 of nameplate and should have been
increased to at least 1.0 to reduce the current and motor operating temperature.
To illustrate the effect on motor temperature, this example shows how, in January 2016, the voltage ratio
was increased by 3% from 0.9 to 0.93, which reduced the current by 8% from 64 to 59 amps, which in
turn reduced the motor winding temperature from 330 to 315o F. This change in voltage, not only reduced
the KVA load by 5%, but also reduced the motor winding temperature by 15o F, which has an important
impact on run life.
by the equations in Appendix B, which is why the uncalibrated WC trends uses the tubing differential
pressure as a good proxy, however it assumes that hold-up is constant, which is not always true and requires
care. The constant hold-up assumption is a reasonable assumption for wells with high WC and low gas/
oil ratio (GOR); however in view of the fact that it was known that some of the wells started with WCs
less than 20% and that hold-ups were expected to change as WCs increased, an alternative method was
sought. The method deployed was to calibrate a multiphase correlation which can then be used to convert
tubing differential pressure to WC. The multiphase correlation and fluid PVT models provide the method
for predicting the liquid hold-up variation in time as WC varies.
Although real time wellhead was not available on wells WRZK-82 and NRZK-16, WC was nevertheless
calculated assuming constant WHP. This proved to be a reasonable assumption because there were no choke
movements combined with the fact that the facility back pressure is stable and dominates well performance.
Note that this constant WHP assumption approach should be treated with caution as it can potentially
incorrectly predict WC, where there is a change in tubing head pressure and a digital real-time WHP gauge
should always be the preferred option.
Inflow Measurements
The objective is to measure key parameters such as depletion rate, PI and connected pore volume i.e.
reserves and ideally without performing build-ups which cause deferred production i.e. simply using the
calculated flowrates and measured flowing pressures. The challenge is how to perform such measurements
when rates are changing continuously due to changes in frequency and choke position as well as changing
inflow parameters and flow regimes. For this reason, the authors recommend using plots of (Pi-Pwf)/Q (rate
normalized pressure drop) versus time and material balance time because they are analytical solutions i.e.
proven from first principles.
Eq. (2) is the usual starting point for conventional vertical oil wells. This is derived by combining pseudo-
steady state inflow with material balance for an under saturated reservoir. This derivation is explained in
detail by Dake (1978).
(2)
This can be rearranged by dividing throughout by the total liquid rate "Q" to yield Eq. (3). This rate
normalization allows the user to compare all production periods even if there are variations in flow i.e. this
is the unit flow response.
(3)
where:
– Pi is initial reservoir pressure, usually taken from gauge intake pressure during initial ESP installation
as the well has had several days to build-up.
– Pwf equals flowing pressure, usually assumed to be equal to the gauge pressure at time t
– Q is total downhole liquid flowrate; traditionally one uses surface rates, which are then converted to
reservoir conditions. As the power equation provides the rate at the pump conditions, this provides
a quasi-reservoir liquid flowrate that does not require conversion, which is very convenient and
removes the need to know FVF and WC.
– A is the inverse of the PI including skin.
– B is the value that captures the rate of depletion, which is why in the analytical form in Eq. (2), it is
inversely proportion to the hydrocarbon pore volume (HCPV).
What this equation demonstrates is that when points lie on a straight line, then the drainage area depletion
("B" value) and well PI (including skin effect) are constant for that period of time. Conversely a change in
inflow parameters can also be detected. An illustration of the application of this equation is shown in Fig.
17 for well WKAL-A4. After three months of production, one detects both depletion and repressurisation
in alternate sequence, which is corroborated by the fact that the nearby injector was turned on and off at the
same time and one can observe directly the reaction in the producer. This is useful as one can measure the
value of water injection for the well productivity and take a decision on whether to increase water injection
to improve the production and recovery factor. It is interesting to note that because well WKAL-A4 has a
18 SPE-185144-MS
high PI (~10 BPD/psi), the (Pi-Pwf)/Q response is dominated by reservoir pressure (i.e. the "B" value) as the
inverse of the PI which is the "A" value becomes small with high PIs.
To observe hydraulic fracture flow regimes as well as BDF (Boundary Dominated Flow), one can write a
more generalized form of Eq. (3) to be as per equation Eq.(4) below, which is applicable to both vertical and
horizontal wells and both with and without hydraulic fractures. This is demonstrated from first principles
by Jones and Britt (2009). When this is plotted in log-log, one can readily identify the exponent "n" as the
slope of the curve.
(4)
where:
– a and b are constants, which have different meanings depending on whether the well is in
fracturetransient or boundary dominated flow. Analytical equations exist for these terms and are
summarized by Jones and Britt (2009) but these do not need to be known in to determine the flow
regime.
– The exponent "n" defines the flow regime, which can be summarized as follows:
∘ n = 0.25 bi-linear flow associated with a fracture and/or horizontal wells.
∘ n = 0.5 linear flow associated with a fracture and/or horizontal well
∘ n = 1.0 determines BDF
SPE-185144-MS 19
An important point is that the flow regime identification can be done equally well with an uncalibrated as
well as a calibrated calculated liquid rate trend. This was demonstrated by Camilleri et al. (2016 URTEC),
but can be understood intuitively on the basis that, when the ESP power equation generates a type curve,
it is the flowrate trend (or rate of change) that provides the slope of the (Pi-Pwf)/Q versus time curve and
not the absolute flowrate value.
To illustrate this technique, Fig. 18, Fig. 19 and Fig. 20 show the flow regime for wells NRZK-16,
WRZK-82 and SIWA-3R6 using log-log plots of (Pi-Pwf)/Q versus elapsed time. These graphs demonstrate
the value of liquid rate trend with high frequency, high repeatability and high resolution. The importance
of the metrology is explained below.
High frequency. Simple observation of the three graphs of (Pi-Pwf)/Q versus elapsed time illustrates the
importance of high frequency liquid rate calculation when compared with episodic well test measurements.
However, to quantify the minimum required frequency, one can postulate that to identify a straight line
trend on the log-log (Pi-Pwf)/Q versus elapsed time graphs, five consecutive points would need to be aligned.
Using the analysis of these three wells, the required testing frequency would therefore need to be as follows:
– NRZK-16 (Fig. 18): to identify how the flow regime switches from bi-linear to linear and then back
to bi-linear, one would need to test approximately once every 3 to 5 daysi.e. 10 times per month.
– WRZK-82 (Fig. 19): BDF occurs twice for 10 and 13 days only, so five points would require a test
frequency of once every 2 days i.e. 15 times per month.
– SIWA-3R6 (Fig. 20): BDF occurs for 37 days, therefore 5 points would require a testing frequency
of once every 7 days.
The above calculations are optimistic and testing frequency would need to be even greater if:
– Repeatability is not achieved because some of the points would not be valid.
– There was a need to identify the start and end of BDF to within a few days of production.
Because real-time data is available every 5 to 10 minutes, the frequency of the calculated flowrate by far
exceeds the required minimum and provides the necessary confidence to identify the various flow regimes.
It can be seen why it is logistically difficult to provide the necessary testing frequency with classical test
separators.
High repeatability is also essential. This is illustrated by the NRZK-16 well where the test separator
measurements under-estimated the liquid rate for a period (see Fig. 12), which resulted in missing the BDF
unit slope, see Fig. 18. Although this example is extreme, a more subtle example is trying to identify the
switch from bi-linear to linear flow on well NRZK-16, Fig. 18 confirms that the test separator measurements
do not have the repeatability to capture such detail and cannot confirm the presence of communication with
a fracture.
As can be seen, two of these wells (NRZK-16 and SIWA 3R6) exhibit linear flow, which is probably
due to the presence of a fracture, which reinforces the point that the pseudo-steady state PI i.e. without
and after the effect of the fracture should be measured during BDF. This can be used for ESP design and
optimization as long as the effect of depletion is taken into consideration with a reduced static pressure on
the inflow performance relationship (IPR) curve. Note that the fracture communication on NRZK16 was
confirmed by the open-hole log which suggested that the well inadvertently had communication with a
second carbonate layer.
(5)
Where:
22 SPE-185144-MS
Key conditions to be respected when using this equation are the following:
▪ The well should be in boundary-dominated flow regime. The production engineer can ensure that this
condition is met by identifying the periods when a unit slope is apparent on the (Pi-Pwf)/Q versus time
curves. Note that if the well has not reached BDF and there is a straight line on the flowing material
balance curve, the slope simply underestimates the HCPV, which can still be useful as it provides a
conservative estimate of the HCPV.
▪ It is an under saturated reservoir (i.e., reservoir pressure is greater than bubble point). This condition
is necessary to use the simplified form of the material balance equation. It is satisfied for the four
wells in the trial because the bubble points are very low.
▪ The original derivation by Mattar and Anderson (2003) for bpss was for a vertical well in a cylindrical
reservoir. However, we are not trying to identify the components of the pseudo steady-state equation
in this application of Eq. (5) because we only need to capture the effect of the "bundled" PI which is
the y-axis intercept during BDF. This equation can therefore be used on any type of well (horizontal
and vertical) where the objective is purely to capture the slope of the curve and measure the speed
of depletion as long as we are in BDF.
▪ Finally, the original derivation by Blasingame used FVF to convert surface rates to downhole
conditions. In this application, all the calculations have been performed using total calibrated
downhole liquid rate as calculated by the power equation and therefore there is no need convert to
downhole conditions, which is an inherent advantage of the power equation rate calculation.
With Eq. 5 in mind, (Pi-Pwf)/q versus tmb was plotted and this is displayed in Fig. 21, Fig. 22 and Fig.
23 for wells NRZK-16, WRZK-82 and SIWA-3R6 respectively. We are pleasantly surprised with the ease
and confidence with which one can fit a straight line to the data despite varying pressures and flow rates.
The separator data are shown by red dots for the plots in all three wells and it is obvious that this type of
analysis would not have been possible with low frequency measurements which were typically one to two
times per month during this trial.
SPE-185144-MS 23
For NRZK-16 (Fig. 21), the reservoir size (slope) observed during BDF can be observed again later in
the life of the well which indicates that the HCPV being drained does not change in time and therefore
no interference from neighbouring producing wells can be observed. In the case of SIWA 3R6 (Fig. 23), a
similar observation can be made, although there seems to be a slight increase in the HCPV (slope becoming
smaller).
For well WRZK-82 (Fig. 22), two periods of BDF are observed and the measured HCPV differs by a
factor of nearly 2. At the time of writing this paper, the cause was not finalized but the most probably
hypothesis was that as flowing pressure dropped, cross-flow was eliminated and the second layer contributed
to production.
With these linear relationships in hand, the production engineer can easily develop rough estimates of
production profiles for late-time production (i.e., boundary dominated) and can examine different drawdown
scenarios. This is one of many techniques for taking advantage of flowrate trends with high frequency,
resolution and repeatability to estimate reservoir properties, however one should consider other techniques
to corroborate any answers, such as simulation, although the flowing material balance technique has the
merit of being fast and simple to use. Finally, it is rarely possible to use this technique with monthly well
testing as observed in the plots on this trial.
the data is less noisy, whereas the derivative function in the log-log plot amplifies the noise. This means
that we can identify IARF and thereby quantify skin, which was found to be 6.5 on this well. Similar PTA
drawdown analysis was performed for WRZK-82 and shown in Fig. 25 and in this case there is no skin.
It is interesting to note that one can observe the constant pressure boundary caused by the nearby water
injection well, which corroborates the fact that flowing pressures observed during production varied as a
function of injection being turned on and off.
Figure 24—PTA for well NRZK-16, type curve shows that there is skin and there is a no-flow boundary
26 SPE-185144-MS
Figure 25—PTA for well WRZK-82, type curve shows that there is no skin and
there is a constant pressure boundary, most probably provided by the aquifer
Measuring PI
While the preceding analysis provides us with dynamic points on the IPR curve (flow rate and flowing
pressure), what we do not know is how the drainage area reservoir pressure evolves with time, which is
indispensable to calculating the well PI. This is because none of the shutdowns are long enough to capture
a buildup that can provide a measurement of the reservoir pressure. The classical technique for obtaining a
trend of drainage area reservoir pressure is to perform simulation as illustrated by Camilleri et al (2015 &
2016), however we will show how a rapid estimate can be obtained from the high resolution high frequency
rate data generated by the flowrate engine.
the calculated liquid rate using the power equation was less than 7 BPD for WKAL-A4 using the formulation
in Appendix A.
Figure 26—PI measurement during IARF for Well WKAL-A4. Multirate test as a result of one frequency
change and one choke movement provides PI with high confidence, PI = 1/0.0994 = 10 BPD/psi
Table 2—PI measurement during initial drawdown and using flowing material balance during BDF
28 SPE-185144-MS
Conclusion
This field trial demonstrated the following:
1. Flowrate calculation is feasible using commonly available gauge data.
1.1. Using commonly available ESP gauge data, it is possible to generate a liquid rate and water cut
trend that has sufficient frequency, repeatability, and resolution to perform reservoir analysis.
1.2. The algorithms are non-iterative and therefore can be programmed in a virtual engine that can
serve multiple clients and wells thereby making the process cost-effective i.e. post-processing
does not need to be updated once the model is built and calibrated as the engine proves real
time calculations. Recalibration is only required if the PHI deviates substantially from 1.
1.3. The liquid rate model can be calibrated without reference to a physical meter utilizing a novel
pump health indicator (PHI) where real time wellhead pressure is available
1.4. An excellent match was obtained between calculated and test separator rates with a single
calibration with the exception of one well (SIWA 3R6), which was due to high wear on
the pump, however the PHI did provide an indication of the wear and when calibration was
maintained.
1.5. The type curve feature of the liquid rate trends provides identification of test separator data
outliers i.e. where there is a lack of repeatability from the test separator.
2. ESP monitorin
2.1. The rate calculation provides a continuous measurement of the ESP operating point in real time
without the need to perform nodal analysis and well testing.
2.2. The power model enabled optimization of the motor voltage to minimize the current drawn
and thereby minimize consumed power and motor temperature without the need to run ESP
simulation software.
3. Inflow monitorin
3.1. Reservoir flow regime can be identified with the uncalibrated calculated liquid rate by plotting
rate normalized pressure difference versus elapsed time on a log-log scale, thereby identifying
in real time when the well is in BDF as well as identifying fracture communication.
3.2. The high frequency liquid rate calculations made it possible to obtain reserve estimates, or,
more precisely, connected liquid pore volume utilizing the flowing material balance equation,
which would not be possible with conventional monthly or even weekly surface testing.
3.3. PI can be measured in two ways:
▪ With very small changes in frequency and choke movements because of the excellent
calculated liquid rate
▪ With the flowing material balance plots to obtain a pseudo-steady-state PI.
3.4. PTA in drawdown mode is made possible with the knowledge of downhole liquid rate,
especially with regard to measuring skin.
3.5. Most of the inflow monitoring can be performed with just liquid rate calculation i.e. without
WC calculation. This is an important point as the liquid rate calculation is decoupled from the
WC calculation in the method proposed by the authors.
4. Increasing Productio
The methods available to the practicing production engineer for increasing production are:
– Maximise drawdown, which requires knowledge of PI and current average drainage area
pressure.
– Stimulation to remove skin, which requires the ability to measure skin.
SPE-185144-MS 29
– Water injection management, which requires knowledge of local depletion and interaction
between producers and injectors.
This trial has shown how high-frequency, high-repeatability and high-resolution flowrate measurement
enables these three functions and avoids the need for costly build-ups.
Nomenclature
References
API RP11S2. 1997. Recommended Practice for Electric Submersible Pump Testing, Second edition. Washington, D.C:
API.
API RP86 (2005) Recommended Practice for Measurement of Multiphase Flow, first edition. 2005. Washington, D.C.:
API.
Blasingame, T. A. and Lee, W.J., 1986, Variable-Rate Reservoir Limits Testing, presented at the Permian Basin Oil & Gas
Recovery Conference, held in Midland, Texas, 13 March.
30 SPE-185144-MS
Camilleri, L.A.P. 2013. System, Method, and Computer Readable Medium for Calculating Well Flow Rates Produced
with Electric Submersible Pumps. U.S. Pat. No. 8, 527, 219.
Camilleri, L.A.P. 2015. Centrifugal Pump Degradation Monitoring Without Flow Rate Measurement. International
Application No. PCT/US15/44241, August 7, 2015.
Camilleri, L.A.P., Banciu, T., and Ditoiu, G. 2010. First Installation of Five ESPs Offshore Romania—A Case Study and
Lessons Learned. Presented at the SPE Intelligent Energy Conference and Exhibition, Utrecht, The Netherlands, 23–
25 March. SPE-127593-MS. http://dx.doi.org/10.2118/127593-MS.
Camilleri, L., Brunet, L., and Segui, E. 2011. Poseidon Gas Handling Technology: A Case Study of Three ESP Wells in
the Congo. Presented at the SPE Middle East Oil and Gas Show and Conference, Manama, Bahrain, 6–9 March. SPE
141668-MS. http://dx.doi.org/10.2118/141668-MS.
Camilleri, L.A.P. and Zhou, W. 2011. Obtaining Real-Time Flowrate, Water Cut and Reservoir Diagnostics from ESP
Gauge Data. Presented at the SPE Offshore Europe Oil and Gas Conference and Exhibition, Aberdeen, UK, 6–8
September. SPE-145542-MS. http://dx.doi.org/10.2118/145542-MS.
Camilleri, L., El Gindy, M., and Rusakov, A. 2015. Converting ESP Real-Time Data to Flow Rate and Reservoir
Information for a Remote Oil Well. Presented at the SPE Middle East Intelligent Oil & Gas Conference & Exhibition,
Abu Dhabi, UAE, 15–16 September. SPE-176780-MS. http://dx.doi.org/10.2118/176780-MS.
Camilleri, L., El Gindy, M., and Rusakov, A. 2016. ESP Real-Time Data Enables Well Testing with High Frequency,
High Resolution, and High Repeatability in an Unconventional Well. Presented at the Unconventional Resources
Technology Conference, San Antonio, Texas, USA, 1–3 August 2016. URTEC 2471526.
Camilleri, L., El Gindy, M., and Rusakov, A. 2016. Providing Accurate ESP Flow Rate Measurement in the Absence of
a Test Separator presented at the SPE Annual Technical Conference held in Dubai, UAE, 26 – 28 September 2016,
SPE – 181663 – MS
Camilleri, L., El Gindy, M., and Rusakov, A. 2016. Testing the Untestable… Delivering Flowrate Measurements with
High Accuracy on a Remote ESP Wellresented at the Abu Dhabi International Petroleum Exhibition and Conference
held in Abu Dhabi, UAE, 7–10 November 2016 – SPE-183337-MS
Dake, L.P., 1978, Fundamentals of Reservoir Engineering. Elsevier.
Jones, J.R, and Britt, L.K., 2009, Design and Appraisal of Hydrauli Fractures, Society of Petroleum Engineers
Mattar, L. and Anderson, D.M., 2003, A Systematic and Comprehensive Methodology for Advanced Analysis of
Production Data, presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, USA, 5–8
October. SPE-84472-MS. http://dx.doi.org/10.2118/84472-MS.
Sultan, M., Al-Enezi, M., Warwick, L. M., Almusabeh, M. I., and Kaba, A.A. 2012. Forecasting and Monitoring Water
Cut Utilizing ESP Pump Discharge Pressures and Fluid PVT Analysis. Saudi Aramco Journal of Technology, Summer
2012
Theuveny, B. C. and Mehdizadeh, P. 2002. Multiphase Flowmeter Application for Well and Fiscal Allocation. Paper SPE
76766, presented at the SPE Western Regional/AAPG Pacific Section Joint Meeting, Anchorage, Alaska, 20–22 May.
SPE-185144-MS 31
(A-1)
(A-2)
DP This is the differential pressure across the pump in psi. The downhole gauge
measures the intake (Pi) and discharge (Pd) pressures in real time; DP is just the
difference between the two. Note that it is important to run a gauge that measures
discharge pressure in addition to intake pressure to enable flow-rate calculation.
ηp This is pump efficiency. Because the pump efficiency cannot be measured
directly, the solution requires that the pump efficiency is based on that of a new
pump taken from either the catalog or test curves.
I This is the motor current (amps) and is usually measured directly by the variable
speed drive (VSD) or switchboard controller. Often, the current is measured at
the VSD and needs to be multiplied by the transformer ratio to obtain downhole
current.
Vm This is the downhole motor voltage (volts). Because only surface voltage can be
measured, it is necessary to subtract the voltage loss in the power cable. This can
be estimated from cable resistance properties and the measured current. As with
the current measurement, the surface voltage is measured by the VSD controller
and therefore needs to be divided by the transformer ratio to obtain downhole
voltage.
ηm×PF This is the product of motor efficiency and power factor. For the current case
study, these values could not be measured. However, it is possible to take direct
measurements of power factor with additional surface electrical instrumentation,
which can be considered on other wells. A motor model based on motor
laboratory test data calculates the product of motor efficiency and power factor
32 SPE-185144-MS
for all loads and voltages, thereby ensuring that the liquid rate was valid at low
startup frequencies when the load factor is low.
(A-3)
(A-4)
where .
The formulation in Eq. A-4 assumes constant ηm×PF, which is not strictly true, but is a first approximation,
and the authors have found that a more complex form does not impact the calculation of resolution. The
gauge measurement resolutions used in the case study are:
– δPd = δPi = 0.1 psi is the gauge pressure resolution
– based on frequency resolution being at least 0.1 Hz
– δI = 0.1 amps is the current measurement resolution and is 1% of the full load rating of the VSD
– δV = 0.1 volts.
The resolution for nondimensional flow is then obtained by multiplying the resolution of DP/P by the
inverse of the derivative of the pump characteristic curve shown below
SPE-185144-MS 33
Appendix B: WC Calculation
The proposed method for calculating water is based on calculating the average mixture density in the
production tubing above the ESP as a function of the pressure difference between the pump discharge
and wellhead, as expressed by Eq. B-1. This technique is analogous to a gradiometer, with the difference
being that because of the length of the tubing, friction must be taken into consideration to achieve a single
calibration over a wide range of flow rates. In environments where the average tubing holdup change is
small, the change in density can be related to a WC change using Eq. B-2. Once such a model is calibrated
against measured WC, ideally measured with a multiphase flowmeter, trending WC with time becomes
possible. This technique requires both pump-discharge pressure and tubing head-pressure measurements in
real time, and when slugging is present in the tubing, the data are required at high frequency to capture
pressure oscillations. The relationship between mixture density and outflow pressures is:
(B-1)
WC is then obtained using Eq. B-2 after neglecting the pressure drop in the gas phase:
(B-2)
34 SPE-185144-MS
the initial month and even 2 months after installation. The need for recalibration is subsequently
flagged using a calculated quality factor called the PHI, which is explained further on.
4. There are potential errors in the calculation of the motor-generated power, which relies on modeling
of the motor to obtain the motor power factor and efficiency. These factors are effectively constant
when operating at motor load factors greater than 50%, which is typically the case, and therefore only
introduces a bias in the calculations as explained by Camilleri et al. (2010). However, to improve
the accuracy, the latest motor power model includes an algorithm that takes into account changes
in voltage (i.e., motor saturation) which can also be used to optimize motor current. This provides
the possibility of calculating power at load factors below 50%. Despite these recent improvements,
there are still some inaccuracies and calibration is required to take into account changes in motor
performance associated with bottom hole temperature.
The traditional process for calibrating calculated flow rates is to simply shift the whole flow-rate curve
(up and down) using a factor that is calculated by comparing the calculated and surface-measured flow rate
defined by Eq. C1:
(C-1)
Although this method is fine, it presumes that surface rates are available and reliable, which is not always
the case. For this reason, the trial with Khalda utilized an alternative novel technique which removes the
need to calibrate against a measured surface rate. This patented method has already been deployed numerous
times and the results have been documented by Camilleri et al. in URTEC and ADIPEC 2016, where the
difference in liquid rate between the measured surface test separator and the calculated results was less than
1% and 2.5% in both papers respectively. Furthermore, the calculated liquid rate had improved repeatability
when compared to the test separator measurements. Based on these results and those shown by Camilleri et
al. (2015), this new technique was deployed on this well. The method involves making an initial assumption
that the in-situ pump efficiency curve is performing as per the factory test curve, which is fair as long as
the following two conditions are met:
• Condition #1. The pump has not suffered any wear prior to calibration. This is a fair assumption if
the calibration is performed immediately after installation of the ESP or if there is a reliable surface
test before the start of the use of the PHI.
• Condition 2. There is no pump performance degradation caused by gas during calibration. To
ensure that this was the case, the real-time pump DP was plotted versus time to identify the start
of surging, which never occurred during the 18 months of production, which confirmed the low
GVF expectation based on PVT analysis discussed above.
This calibration method relies on the comparison of the ratio of pump DP to power, as described in Eq.
C-2, which is called the PHI (Pump Health Indicator)
(C-2)
A detailed explanation of the workflow can be found in the patent by Camilleri (2015); however, a
summary is as follows:
• (DP/Power)actual (numerator); the pump-absorbed power is calculated using the same method
employed for liquid rate calculation (i.e., using voltage, current, and frequency real-time
36 SPE-185144-MS
measurements as per Appendix A), whereas the pump DP is taken directly from the downhole
gauge intake and discharge pressures.
• (DP/Power)reference= ηp/Qp is obtained from the pump curve using factory test points.
• Both these values should be calculated at the same differential pump pressure