The Congolese Offshore has 9 oil fields with 65 wells of which 21 wells belong to the Mibale field. Of these 21 wells, we have 14 producers, 6 injectors and 1 appreciation well, which is equivalent to 32.3% of the total oil fields. As of today,

the field is producing an average of 5905.54 barrels of oil per day (BOPD) with a water-oil ratio (WOR) of over 60% (September 2021 average). Starting from the average official production of 25,000 BOPD for this company, we observe that this field contributes to the extent of 23.6%. Therefore, we can conclude that this field represents the largest offshore oil platform. It is worth noting that the Mibale field was discovered in 1973 by the oil company CHEVRON with the Mibale 1X well in the Upper Pinda formation, which is part of a fault-related structure with three branches. With three wells, MIB-01, MIB-02, and MIB-03, this field was put into production in 1976 [12] [13] .

A good quality oil (32˚ API) has been discovered in a multi-layer complex in the upper part of the Albien-age Pinda formation, also known as the Upper Pinda reservoir. The Upper Pinda in the Mibale field is a reservoir with high permeability intervals, sand-rich layers, and dolomitic layers. It is divided into eight layers (LP or UP) ranging from the transition layer to L7 (UP-7). The upper transition layer is often described as poor-quality limestone and clays, while the underlying UP-1 layer consists of relatively low-quality limestone and less developed carbonate sands [8] .

The Mibale field is located in the coastal basin of the Democratic Republic of the Congo, 5 km wide and 3 to 7 km southeast of the DRC’s offshore border with Cabinda (Angola), and covers an area of around 11 km² ( Figure 2 ).

The studies conducted in 2016 provided the STOIIP (Stock Tank Oil Initially in Place) of the different layers of the Upper Pinda reservoir, as summarized in Table 1 below, comparing them with the geological STOIIP values from 2006.

L1 layer (34%), L2 layer (22%) and L4 layer (25%) are the main STOIIPs. The L1 and L2 carbonate layer STOIIPs account for 56% of the Pinda field SToIIPs as a whole. The difference with the geological model lies mainly in the way water

saturation is used relative to depth, a relationship that introduces a large area of transition.

Table 2 shows the layers of the reservoir with the wells crossing them.

It is observed that well MIB-14ST2 penetrates layers L1, L2, L4, and L5. The amount of hydrocarbon reserves in these five layers amounts to 325.9 million stock tank barrels (MMstb).

The evolution of annual production in the Mibale field from 2010 to 2021 is illustrated in Figure 3 . In 2010, the production was good, but in 2015, it was too low, indicating a rapid decline in production.

The original MIB-14 well, which was drilled in July 1991, was completed in double sequence with a short sequence in the transition layer and a long sequence on the layers L1, L2 and L4. The MIB-14ST well was laterally followed to a TD of 7005 ftMD in May 1994 and was doubly completed in the TL layers (short sequence), L1, L2, L4, L5 and L6 (long sequence). In June 2007, total oil production was 1.7 MMSTB. Current production has 70 BOPD and a WOR of 68%.

Note:

Table 3 shows the reservoir and fluid properties of the Mibale field, in which all wells drain oil to surface.

Completion data for the MIB-14ST2 well are presented in Table 4 below.

It should be remembered that the Prosper IPM software is a good simulator for converting from activation mode to ESP [18] . Data from Table 4 (geothermal gradient, hole bottom equipment and tube and casing size data) were used as input data in IPM Prosper software to generate the MIB-14ST2 well architecture ( Figure 4 ).

Figure 5 below shows us the perforation depths of MIB-14ST2 well.

To assess the well’s productivity, the IPR and VLP curves must be analyzed to determine the well’s production capacity and operating point. To establish these curves, certain correlation choices must be made:

1) Choice of correlation to establish the IPR and VLP curves

a) Choice of gas solubility correlation (R_s)

To establish the IPR curve, we need to choose the dissolved GOR correlation. By replacing the fluid property values from Table 3 in mathematical Equations (1)-(5), we estimate the gas solubility (R_s) at bubble point pressure and compare this with the experimental value in terms of the absolute average error (AAE). The Rs’s correlations expressed as mathematical equations are given as follows [16] [19] :

i) Standing’s (1981) correlation is expressed in the following mathematical form:

$R_s=c{\left[\left(\frac{P}{18.2}+1.4\right){10}^x\right]}^{1.2048}$ (1)

$x=0.0125{\gamma }_o-0.00091\left(T-460\right)$

ii) Vasquez-Beggs correlation is given in the following form:

$R_s=C_1{\gamma }_{gc}{P}^{C_2}{e}^{\left[C_3\cdot \frac{°API}{T+460}\right]}$ (2)

with coefficient values C₁, C₂ et C₃.

iii) Glaso correlation is proposed according to this relation:

$R_s={\gamma }_g{\left[\frac{AP{I}^{0.989}}{{\left(T-460\right)}^{0.172}}\cdot P_b^{*}\right]}^{1.2255}$ (3)

where $P_b^{*}={10}^x$ and $x=2.8869-{\left[14.1811-3.3093\log P\right]}^{0.5}$

iv) Marhoun’s correlation proposed the following equation:

$R_s={\left[a{\gamma }_g^b{\gamma }_o^c{T}^{d}P\right]}^e$ (4)

The coefficient values a, b, c, d and e are: a = 184.843208; b = 1.8778480; c = −3.1437; d = −1.32657 and e = 1.398441.

v) Petrosky-Farshad correlation is given by the following expression:

$R_s={\left[\left(\frac{P}{112727}+12.34\right){\gamma }_g^{0.8439}\cdot {10}^x\right]}^{1.73184}$ (5)

$X=7.916\times {10}^{-4}\times AP{I}^{1.541}-4.561\times {10}^{-5}{\left(T-460\right)}^{1.3911}$

The following formula makes it possible to find the absolute average error (AAE):

$AAE=\left|\frac{Vr-Vm}{Vr}\right|\times 100$ as a percentage (6)

With Vr: Real Value (Field Data) and Vm: Measured Value (correlation values).

This allows us to summarize the R_s correlation values found with AAE ( Table 5 ).

We note that the value of GOR dissolved in oil (R_s) calculated by the Glaso correlation is a little close to that of the Mibale Field determined by PVT analyses with a deviation or AAE of 13.12%, which allows us to choose this correlation as the one applicable in the optimization of this well. Referring to Table 3 , the value of Rs is 300 scf/stb and using the Glaso correlation, R_s is 339.37 scf/stb with 13.12 deviation factor.

b) Choice of flow law correlations

Several correlations on the flow law in the reservoir have been established with their mathematical expressions [5] [20] [21] . These are the correlations of:

Given the available data and the characteristics of the Mibale field reservoir, Darcy’s law correlation was used to establish the IPR curve. The mathematical expression of Darcy’s law is as follows:

$Q_o=\frac{0.0078\times k\times h\left(P_r-P_{wf}\right)}{{\mu }_o{B}_o\ln \left(\frac{r_e}{r_w}\right)}$ (7)

c) Choice of upward two-phase flow correlation

Several correlations on the law of flow in the reservoir have been established with their mathematical expressions. Beggs and Brill summarized the correlations of the law of flow into three main categories, each of which varies in complexity and technique. These are the following categories [22] :

Category A: No slip effect or flow regime is considered—Poettmann & Carpenter, Fancher & Brown;

Category B: Slip effect is considered, no flow regime is considered—Hagedorn & Brown, Gray;

Category C: Both slip effect and flow regime are considered—Beggs & Brill, Orkiszewski, Duns & Ros.

However, no single correlation was found to be the best over the others for all flow conditions. Individual well tests and experience can be used to obtain the correlation that best suits the characteristics of each well [23] [24] . Some of these correlations are given in the following mathematical form:

i) Poettman and Carpenter’s correlation is a semi-empirical method using the general energy equation and considering the mixture of oil, gas and water as single-phase.

ii) Hagedorn and Brown’s correlation is an extension of Poettmann and Carpenter’s.

iii) Beggs and Brill’s correlation studied head losses in tubing using the same technique, as for horizontal pipes, with the introduction of a factor that takes into account the inclination, which changes from −90˚ to +90˚. This method is based on determining the flow regime, which depends on Hold up, the pressure gradient, the Froude number.

After inputting values in the software for the pressure at the wellhead, water cut, oil or liquid flow rate, gas-oil ratio (GOR), various depths of the well down to the tubing, and the pressures prevailing in the well, the Prosper software calculates correlation parameters by determining the deviation factor for each correlation. The obtained values are depicted in Figure 6 and Figure 7 .

In view of Figure 7 below, it is noted that each correlation has a relative error but it is found that the Beggs and Brill correlations are the most appropriate for calculating outflow load losses because the values of two parameters have a small value (0.74) with a small deviation factor of 212.174 compared to that of other correlations that are greater than 0.76.

2) Establishing the IPR and VLP curves

Unlike manual calculations and other optimization software, IPM Prosper was used to model the well in different scenarios to obtain accuracy on the actual well performance. This modeling was based on Darcy’s law and the data from Table 3 to establish the IPR (Inflow Performance Relationship) and the Beggs and Brill correlation, combined with data from the pressure at the upper node, GOR (Gas-Oil Ratio), and a 98.5% water cut.

Figure 8 illustrates the pressure and fluid flow rates values found after the calculations.

After plotting the IPR and VPL curves for the MIB-14ST2 well, it can be seen in Figure 9 that an absolute maximum flow rate “AOF” of 30998.8 stb per day should be produced with the IP of 12.29 stb per day per psi. Unfortunately, the IPR curve is a steep slope, demonstrating that this well was producing absolutely nothing, and there is no point of contact between the two curves.

3) Well sensitivities

Considering that the variables for the MIB-14ST2 well are the water cut, the pressure at the upper node, and the GOR, we varied the GOR from 0 to 1000 scf/stb the first node pressure from 500 to 2600 psig and the water cut from 0 to 100%.

Figure 10 illustrates the calculations performed taking into account the values of the variables listed above.

As we described in the problem, the production of the MIB-14ST2 gas-lift activated well is almost zero or too low so according to our study case, we do the sensitivity of variables such as the first node pressure, the water cut and the GOR to see if there will be a possibility to convert the activation mode.

After the sensitivity of these variables, we notice that the IPR and VLP curves intercept at several points between which we can get a realistic point of operation ( Figure 11 ).

Optimization by activation mode conversion of the MIB-14ST2 well is considered in this type of well to be the best activation mode capable of compensating

for the loss of pressure due to the high liquid rate and low gas flow rate according to the evaluation made in this well. The pump also has operational constraints for its functioning, namely [25] [26] :

Taking into account these different constraints, the dynamic water level in the well, the desired oil flow rate, the number of stages, engine power and cable length, we have the well design as shown in Figure 12 below.

Taking into account Figure 12 of the design, the IPM Prosper software performs calculations using integrated equations, giving precision on the number of stages, pump efficiency and real power (Equations (8)-(11)). If the flow rate produced exceeds the maximum flow rate, pump efficiency becomes low. Also, a high power factor in a pump is important because it improves energy efficiency, maintains the stability of the electrical network, improves pump performance and reduces operating costs, as is the case with the MIB-14ST2 well. The mathematical formulae for the pump parameters mentioned above are as follows [25] :

$TDH=H_d+F_t+P_d$ (8)

$PDP=THP+GAVG\cdot TVDP,psia$ (9)

$P_{HYD}=\frac{Q\times h\times \rho \times g}{n_h}$, kilowatts (10)

With:

$n_h=\frac{\Delta P\cdot q_{inlet}}{P_h}$ (11)

With:

${\eta }_{TOT}={\eta }_{PUMP}\cdot {\eta }_{SEAL}\cdot {\eta }_{MOTOR}\cdot {\eta }_{CABLE}\cdot {\eta }_{TRANS}\cdot {\eta }_{DRIVE}$, in decimal (12)

With:

$P_{SYS}=\frac{P_{HYD}}{{\eta }_{TOT}}$ in horsepower (13)

$PF=\frac{kW}{kVA}$ (14)

In decimal (Between 0 and 1)

With:

The results obtained for the entire pumping system are as follows ( Figure 13 ):

Using Equations (12)-(14), we find:

With these different values obtained for total efficiency, power factor and system power, we can see that the system’s total efficiency is good, as it exceeds 60% according to the criteria for good operation. Also, the power factor is above 80%, which proves that there will be no considerable or high voltage requirement during production, and that friction between fluids is reduced, thus reducing load losses. Considered as the surface electrical power, the value of the system power obtained is higher than that which we want to apply to the fluid in order to compensate for dissipation, thus increasing fluid velocity and reducing head losses by elevation.

The pump’s performance is delimited in a zone called the operational zone, in which the pump can operate with minimum and maximum frequency [17] .

The performance curve of the ESP pump can change based on the molecular weight of the gas and the inlet fluid temperature. We note that the molecular weight of the gas can influence the overall viscosity of the fluid which will lead to a change in fluid flow properties through the ESP pump, which can affect its overall efficiency and performance. And also if the gas in the pumped fluid reaches steam pressure, it can create gas pockets that can cause cavitation of the pump which could damage the pump and reduce its efficiency.

To avoid making corrections to this curve, the following formula is used:

$\Delta h=\frac{P_{out}-P_{in}}{{\rho }_{mix}\times g}$ (15)

When expressing performance in terms of head (∆h), the unit used is feet. The density of the fluid mixture (ρmix) is also an important parameter to consider in calculating head.

Thus, the pump’s head required is based on the depth of its location, while that of well MIB-14ST2 is at 1877.38 feet after simulation and would give 5083.94 stb/d as liquid flow. This pump will operate with a frequency of 50 Hertz and on the best efficiency line (BEL), we obtain the pump’s best efficiency point at 64.72% ( Figure 14 ).

Referring to the formula for discharge pressure, we derive the gradient average of the fluid at the pump location:

$GAVG=\frac{PDP-THP}{TVDPUMP}$ (16)

With:

Using the Equation (16) and the pressure at the head of the MIB-14ST2 well at 90 psig, we obtain 0.499 psi/ft as the average fluid gradient at the depth of the pump installation; this also confirms the dynamic level of the water according to the filling data of this well because the average water gradients are equal to or greater than 0.4 psi/ft.

Concerning sensitivities to frequency and number of stages, we have performed several scenarios by plotting the IPR and VLP performance curves with variables such as pump frequency from 30 to 70 Hertz, number of stages from 10 to 100 stages and water cut from 10% to 100%.

After several scenarios, the results are presented in 2 ways in a graphical form ( Figure 16 ) showing the well operating points from the intersection of the IPR and VLP curves and in a numeric form ( Figure 15 ) showing each scenario with its fluid production rates. Considering variable values, several tables have been presented but in this work, we present only one to illustrate this.

We note that there are several operating points in this well, demonstrating that this well can still produce after conversion of the activation mode ( Figure 16 ).

From the analysis and interpretation of Figures 9-12 we obtained the operating

point from the IPR and VLP plot. This new point was used to determine the new production of the MIB-14ST2 well ( Figure 17 ):

This result then leads to a study on the re-evaluation of the fluids in place in the reservoir and the reconditioning of the well (revising the perforation depth and the quantity of water in the different layers crossed).

Looking at Table 1 and Table 2 , we can see that the MIB-14ST2 well crosses the L1, L2, L4 and L5 layers, the last of which has a high permeability but a very

low STOIIP oil content and an interesting amount of water, so it would be worth reviewing the perforations at this level, or even at layer L4.

With regard to the diphasic flow in a well, the researchers were able to identify 7 flow structures that can be localized separately in the tubing. This flow regime is determined by calculating the hold up of fluid. These structures are presented as follows [27] :

Note: In the case of vertical tubing, we find flow patterns such as bubble flow, stratified flow, annular flow, and slug flow.

Knowing the negative impact of gases in the ESP pump, Figure 18 shows that the hold-up in the well is equal to 1 and the flow is in bubble flow (i.e. the GOR values are reduced so the presence of liquid fluids is high throughout the well.

Figure 18 shows that the hold-up in the well is equal to 1 and the flow is

bubble flow, i.e. GOR values are reduced, so the presence of liquid fluids is high throughout the well.

As the water requirement for the pump is very high, this does not adversely affect pump performance, whereas the viscosity of fluids in the well is considered a variable that can affect well performance. This significantly reduces pump performance in the case of viscous or overly viscous fluids. When fluid viscosity is high, the number of rotating parts (rotor) is high and creates a lot of friction surfaces, which results in high energy losses in the well. In the case of the well under study, we’ll try the sensitivities of frequency and number of pump stages, as the oil in the upper Pinda reservoir has a good viscosity of 1.1 Cp.

Once the liquid level has been obtained in TVD, we need to relate it to the depth measurement and according to the well deviation in order to measure the tubing fill volume [21] . This gives:

$t_{SURF}=\frac{V_{TUBING}}{Q_{PUMP}}$ (17)

With:

To calculate the tubing volume of an oil well, you need to take into account the dimensions of the tubing, i.e. its inside diameter and length ( Table 4 ). Tubing volume an oil well can be calculated using the volume of the cylindrical tubing:

$V_{tubing}=\frac{\pi \ast I{D}^2\ast l_t}{4}$ in cubic feet (18)

With:

Equation (17) and Equation (18) were used to calculate the pumping time. For the MIB-14ST2 well, the tubing volume is 232.19 ft3 and the surface fluid pumping time is 11 minutes and 9 seconds.