Mechanical Impurities Carry-Over from Horizontal Heavy Oil Production Well

Abstract

Currently, a global attention has been paid to the development and exploitation of shallow depth heavy oil reservoirs. Such oilfields are rationally developed via a network of horizontal wells. However, the weakly cemented reservoir rock beds lead to the high sand production during well flow. Removing mechanical impurities is still challenging even with the application of sophisticated techniques and a variety of filters. In this study, we propose an analysis method for describing the removal of mechanical impurity particles from horizontal wells containing heavy oil. We employed a simulation model of typical well in OLGA program, and several calculations was made for different values of the flow rate. Moreover, deep sand samples were isolated from a well of Uchebny field to compare the real data with the estimated one. Calculations are used to estimate the quality of mechanical impurity removal for various diameters, and a relationship between the critical diameter and well flow rate is built.

1. Introduction

Nowadays, due to the depletion of “traditional” reserves, the share of oil extracted from weakly cemented reservoirs with high viscosity is gradually increasing [1,2]. Therefore, the main difficulty in the development of such reservoirs is the large number of failures associated with the penetration of sand into the deep well pumping (DWP) system. Most of the reserves of the Uchebny field are located in the deposits, where oil production is carried out by horizontal wells. In addition, the reservoir consists of weakly cemented sandstone with high viscosity of the produced oil. Note that the viscosity of the oil of the Uchebny field under the conditions of the deposit reaches 67 mPa·s.

The process of sanding is related to the development of fields with weakly cemented reservoirs, which results in sand being removed from the reservoir and entering the wellbore. In fact, reservoir failure can occur for a variety of reasons: Drilling, cyclic effects of stopping and starting the well, operating conditions, increasing effective pressure, high water cut, and others [3,4].

In general, the operation of a well with high sand content in the reservoir product directly affects the failure rate of submersible equipment. In addition, mechanical impurities, in the Uchebny field is the most significant aggravating factor, accounting for approximately 72% of all failures (Figure 1). These failures include failures due to clogging in the passage area of the working elements of the submersible pump, erosion of production system [5], erosive wear of tubing strings, etc.

The largest share in the structure of mechanical impurities entering the pumping unit at the field under study are products of rock destruction (54%) and process fluids pumped into the well (16%), which have not been sufficiently prepared and contain chemicals that complicate the cleaning process (leading to rapid destruction of filters). To solve the problem of operation of flooded horizontal wells at the field, it is also proposed to use a shank that is lowered into the well below the reception of the electric centrifugal pump and ensures the removal of mechanical impurities from the bottom of the wells. Thus, the task of optimizing the conditions for removing mechanical impurities from the borehole, selecting filters and justifying the calculation of the shank diameter is an urgent task that contributes to increasing the profitability of field operation.

The purpose of this work is to study the process of removing solid particles from a horizontal well by a highly viscous fluid flow. In addition, in this work, the accuracy of modeling the sand outflow process using the OLGA software was questioned. The results obtained are the basis for further studies of the sand outflow process.

In order to fulfill the goal, following tasks were implemented in the work:

• Selection of deep samples of sand from the field;
• Studying of the selected samples with a microscope;
• Analysis of the results;
• Comparison of calculated values with field data;
• Summing up.

Similar studies of the removal of mechanical impurities were already conducted in the works of Kamel Alsawi, Mahmoud Hegazy, Khaled Elhady and Mahmoud Sayed [6], M.G. Jaimes, Y.A. Quintero and G.Y. Contreras [7], A.N. Betekhtin, D.K. Kostin, E.V. Tikhomirov and M.N. Nikolaev [8].

Working of a screw pump in sandy conditions was considered in the work of Kamel Alsawi, Mahmoud Hegazy, Khaled Elhady and Mahmoud Sayed [6], where the main task was to optimize the operation of a well that was complicated by mechanical impurities. In order to do that, the author selected deep samples of sand, analyzed its particle size, and came up with the following recommendations: smaller tubing diameter, increasing the pump speed by more than 200 rpm, using flow turbulators where all this is necessary to prevent sedimentation of sand particles and carry it to the surface.

The measures allowed to increase the daily fluid flow rate of the well from 11 m³/day to 27 m³/day. Moreover, if previously the wells worked uninterruptedly for several weeks, the current operation period exceeds three months.

Influence of the drawdown value on the removal of mechanical impurities was studied in the work of M.G. Jaimes, Y.A. Quintero and G.Y. Contreras [7]. In order to justify the obtained patterns were used as a technological assessment, as well as economic. As a result of the carried-out measures for two years the frequency of sand occurrences on 116 wells was reduced by 100%. In addition, the productivity of wells increased, which allowed for three years to move from a negative present value of wells (−2.23 million $) to a positive (0.535 million $).

To prevent mechanical impurities from entering the well from the reservoir, various control systems were used. Such studies under the conditions of a weakly cemented reservoir and high-viscosity production were discussed in the article by A.N. Betekhtin, D.K. Kostin, E.V. Tikhomirov, and M.N. Nikolaev [8] discussed. These results can be used to select filters used to separate solid particles from the flow of highly viscous formation products. Based on the predicted permeability change coefficient (PCC) near the wellbore, suitable filters can be selected that provide the lowest PCC values, i.e., are most stable under these conditions. The permeability change coefficient is mainly determined by the degree of cementation of the reservoir, the fluid viscosity, the flow phase (presence or absence of gas), and the relationship between the size of the filter slots and the sand particles discharged [9].

In fact, there are more than 60 models that describe the behavior of particles in the flow [10], among the main ones stand out: the model proposed in 1980, Oroskar and Turian [11], the model of Salama (1998) [12], Danielson (2007) [13], Davies (1987) [14].

The main value characterizing the processes of sand removal/sedimentation in the borehole is the critical velocity parameter ${\upsilon }_c$ [15]. The critical velocity, the minimum velocity required to continuously maintain the moving particles in suspension, is a function of a number of parameters:

$${\upsilon }_c=f\left(u_{sl},u_m, d_p, D, {\rho }_P, {\rho }_L, {\vartheta }_L,{\mu }_L, C, x, \alpha , g\right)$$

(1)

where ${\upsilon }_{\mathrm{c}}$—is the critical velocity, m/s; $u_{sl}$—is the reduced fluid velocity, m/s; $u_m$—is the s of the velocity of mixture, m/s; $d_p$—is the average particle diameter, m; $D$—pipe diameter, m; is the density ${\rho }_P$—solid particles density, kg/m³; ${\rho }_L$—liquid density, kg/m³; ${\vartheta }_L$—is the kinematic viscosity of the liquid, mPa s; ${\mu }_L$—is the dynamic viscosity of the liquid, mPa s; $C$—is the concentration of particles, %; $x$—is the proportion of flow turbulence; $\alpha$—is the volume fraction of particles, %; $g$—free fall acceleration, m/s².

The main physical parameters considered by one or another method to determine the value of the critical velocity is considered. For this purpose, the results of the comparison are presented in

Table 1. Comparison of various methods for determining the critical velocity.

Parameter/ Methodology	$\mathbf{C}$	${\mathbf{d}}_{\mathbf{p}}$	$\mathbf{D}$	${\mathsf{\rho }}_{\mathbf{P}}$	${\mathsf{\rho }}_{\mathbf{L}}$	${\mathbf{u}}_{\mathbf{s}\mathbf{l}}$	${\mathbf{u}}_{\mathbf{m}}$	${\mathsf{\vartheta }}_{\mathbf{L}}$	${\mathsf{\mu }}_{\mathbf{L}}$	$\mathsf{\alpha }$	$\mathbf{x}$	Particle Shape
Oroskar and Turian (1980)	+	+	+	+	+	−	−	−	+	−	+	−
Salama (1998)	−	+	+	+	+	+	+	+	−	*	−	−
Danielson (2007)	−	+	+	+	+	−	−	+	−	−	−	−
Davies (1987)	+	+	+	+	+	−	−	+	−	−	−	−
OLGA software	+	+	+	+	+	+	+	−	+	**	−	+

«+» The parameter is taken into account in the method of determining the critical speed. «−» The parameter is not used when calculating the critical speed in this method. * Two equations for vertical and horizontal flows are given. ** The calculation of the angle of inclination of the pipe is carried out by multiplying the sliding speed by the cosine of the angle of inclination of the pipe.

. Computational modeling of the process of mechanical removal of impurities was performed using the software package «OLGA» for different values of the flow rate of the well fluid and sand fractions.

2. Materials and Methods

In order to investigate the sand removal process, downhole samples were taken from several wells in the Uchebny field where the wells were chosen in such a way that a holistic picture could be obtained for different fluid flow rates.

The design of the oil wells in the Uchebny field was similar which has a vertical borehole of about 600 m, a curvature set section and a horizontal borehole of 1200 to 2000 m. In addition, the system has three zones of solids movement and accumulation: zone 1—bottomhole of the well (horizontal section, liner), zone 2—tubing and zone 3—annular space. Sampling was performed from the bottom hole while flushing the well (zone 1, Figure 2).

Bottom hole sampling was performed using the hydro-vacuum bailer (HVB) technology [16]. The principle of hydrovacuum bailer operation is based on pressure difference in annular space and tubing string. Thus, as the weight is transferred from the tubing string to the perforator, the upper valve of the HVB opens and the fluid starts to flow with high velocity into the tubing string. At the same time the check valve is opened and together with fluid mechanical impurities start to be carried out from the bottomhole [17,18].

Method for Determining the Granulometric Composition of Sand Samples

The particle size distribution was determined using a Biolab 6T microscope where the pre-selected deep sand samples were cleaned from the hydrocarbon phase. This was done by washing with petroleum solvent Nefras-C4-155/200. The process of sample preparation is shown in Figure 3.

In order to perform the analysis, a 15 mL spoonful of sand was collected from the depth sample container and placed on a filter paper. Process was performed on a flat surface in order to obtain an even distribution of particles. Initially, the larger particles may be concentrated at the bottom and the smaller ones at the top.

The quarting process is the mixing of the sampled sand sample, its sequential separation into four equal parts (several steps), with only ¼ of the original mass remaining for further study (Figure 4), the remaining ¾ being returned to the original container with the depth samples where quartization was carried out in 2–3 stages.

After quarting, the obtained sand sample was washed with petroleum solvent. For this purpose, a filter paper was placed in a funnel, which was placed on a cylinder (Figure 5). The washing process was carried out accordingly until the color of the solvent did not change.

After washing, each sample was examined with a Biolab 6T microscope and images were taken, which were used to determine particle size.

From the images, the effective diameters of the sand particles were determined and differential and integral curves were plotted. In addition, particle size was determined using Toup View software [19]. Since sand particles have a complex irregular shape, circles were used to describe their size, and their diameters were determined in the program. The method for determining the particle diameter is shown in Figure 6, its results will increase the accuracy of the prediction of the removal of mechanical sand at the stage of development and selection of the appropriate set of technologies for the proper operation of wells, minimizing financial and time costs.

In order to obtain the final characteristics reflecting the particle size distribution, it was necessary to analyze the results of the studies to exclude outlier points.

For this purpose, we used the «MatLab» R 2010a software [20], which allowed us to determine the distribution law to which the obtained dependencies. Also, different assumptions were made regarding the distribution type: normal, lognormal and bimodal. It was also found that in all the sand samples studied, the distribution of particle size is normal [21].

In order to detect outlier points, regression analysis was used, and the initial distributions were normalized beforehand (Figure 7).

In regression analysis, fitted values and residuals are important concepts. Usually the straight line does not exactly pass through the available data, so the regression equation must include an explicitly defined residual term ei (residual) [22]:

$$Y=b_0+b_{1}X+e_i$$

(2)

where $b_0$ and $b_1$ are the regression coefficients.

Next, the regression analysis was performed using «MatLab» R 2010a software, and the results are plots of reliable intervals for the residuals (Figure 8). The red color indicates outliers, green—points that passed the test.

Analogous studies were performed for other wells. As a result, it was obtained that the average number of outliers for the sample was five points. Once the outlier points were removed, the processed data were used to obtain differential and integral curves.

In order to build the computer model, we used the following parameters as input data: horizontal well inclinometry, submersible equipment characteristics, geological and physical characteristics of the reservoir system, PVT properties of the fluid. Accordingly, in the modeling process, the value of gas factor was assumed constant and equal to 53 m³/t.

The calculations were performed for 5 different well rates (Figure 9). For each fluid flow rate, data were obtained for well operation with particles of a certain diameter (from 10 to 200 μm). The «OLGA-2017» software does not have the ability to simulate the flow consisting of particles of different diameters. Thus, each case is data on the behavior of particles of a selected diameter at a certain well flow rate [23,24,25].

The principle was that a series of calculations was performed, presented in Figure 9, according to the results of which the points of the critical diameter of the particles from the flow rate of the well through the liquid were established. Then the values were approximated and with the help of computer modeling, the dependence of the critical diameter of sand particles on the flow rate of the well in terms of liquid was obtained. After that, in accordance with the technological regime of the studied wells, the values of the critical diameter of the particles corresponding to the actual operating modes of the wells were established. According to the results of the study, the values of the calculated critical diameter were determined in accordance with the methodology previously described in the work. The actual values of the critical diameter were obtained and a cross-comparison with the model data was carried out.

3. Results

The final characteristics of the sand particle size distribution are shown in Figure 10. On these dependencies are plotted on the ordinate axis the percentage of particles of a given size (for the differential curve) and the percentage, i.e., the probability of finding in the sample particles equal to or smaller than the given size (for the integral curve). Accordingly, the size of the diameter intervals was chosen based on the number of points in the final samples [26].

Consequently, the resulting differential curves show the percentage of particles of a given size in the whole studied sample [27,28]. In contrast, the integral curves show the probability of finding particles in the sample whose size is equal to or smaller than the specified one (Figure 11).

Simulations for different groups of flow rates and sizes of sand particles entering the well resulted in the following dependence (Figure 12).

This dependence allows us to establish a relationship between the critical diameter of the sand particles and the flow rate of the well. Therefore, we understand the critical diameter as the minimum diameter of the solid particles that do not participate in the flow of the gas-liquid mixture [29,30]. For this purpose, the value of the critical diameter was chosen based on the determination of the minimum particle size at which the sand concentration at the inlet to the tubing was zero.

It can also be seen that the dependence is initially quadratic and then becomes linear with respect to the fluid as the well flow rate increases. This is due to the fact that the dependence between deposition velocity and particle diameter is quadratic at low flow rates, but linear at higher values. In addition, the particle velocity is determined by the slip velocity, which is the velocity of the particles relative to the carrier phase (gas, oil, and water). Thus, only the vertical component of the deposition velocity is considered in the modeling, which is calculated as a function of the value of the Reynolds number [31].

Similarly, the data obtained with the software «OLGA-2017» and other known methods were compared in Musi A. and Larrey D. [15]. It was found that the software «OLGA-2017» performs well in the case of a single-phase flow (error within 20%), but tends to underestimate the carrying capacity of a two-phase mixture of particles in suspension (according to the results of the simulation of settled particles, more than in actual laboratory tests). It is worth mentioning that in the study by Musi A., Larrey D. [15] the simulation results were compared with the data obtained in laboratory conditions under a number of assumptions. Ordinary water and a gas-liquid mixture of water and air were used as the system capable of transporting sand.

Accordingly, the results of comparison of the field data with the values obtained during modeling are presented in

Table 2. Comparison of the values of the critical particle diameter obtained using computer simulation.

Well	${\mathit{Q}}_{\mathit{F}},\frac{{\mathit{m}}^3}{\mathit{d}\mathit{a}\mathit{y}}$	$\mathit{G}\mathit{F},\frac{{\mathit{m}}^3}{\mathit{t}}$	Model, Micron	Fact, Micron	Deviation of the Model from the Fact, %
A	51	14.06	45	29.4	53.1
B	129	411.07	67	101	−33.7
C	218	2610.06	87	67.9	28.1
E	28	4835.46	36	34.2	5.3
D	90	136.55	58	67.4	−13.9

, where the actual value of the critical diameter $d_{cr.}$ was taken as the minimum particle size in the sample (at which the probability of finding a particle of a given diameter is zero) [32]. Moreover, the results of the computer model calculation were determined by the dependence shown in Figure 12. The comparison revealed both an underestimation of the calculated values in relation to the field values and an overestimation. This was justified by the fact that particle efflux is influenced not only by the flow rate, but also by a number of other parameters, including the gas factor, which was determined in this work.

In previous work by Ramin Dabirian [29,31,33], the influence of free gas on particle outflow was studied. It was found that the presence of gas in the flow of formation products helps to keep the sand in a suspended state. However, all studies were also conducted under laboratory conditions, which imposed a number of limitations. Apparently from the presented dependence (Figure 13), values of gas factor on candidate wells differ from the one accepted in modeling (about 53 m³/t), therefore the obtained deviations are acceptable.

It is also worth considering a number of factors that may have influenced the final value of the deviation:

• The error in determining the particle size distribution due to the use of the microscopic method [34,35,36,37]. In fact, this method is one of the most inaccurate due to the error in determining the size of the particles, which does not take into account the shape of the particles when viewed in different planes, inaccuracies are possible due to the setting of the device and its error (less than 5%);
• Particles deposited on the bottom of the borehole when the borehole is stopped for flushing. A certain amount of time elapses during which the small size particles removed have time to settle on the bottom of the well [38];
• The error in the oil model [39] used for the calculations, i.e., it is an error in setting the oil-gas saturation pressure of less than 5% and the dynamic viscosity of less than 10%.
• Introduction of the assumption that all particles have the same density.

Figure 14 shows images of well samples taken from the lower wellbore of the candidate wells. The particles in the sand are mostly quartz (most of it), feldspar, rock debris, and mica. Thus, as can be seen in the pictures, the sand particles have a complex individual shape.

4. Discussion

Considering the main complicating factors affecting the failure rate of the downhole pumping equipment, it was revealed that the majority of failures (about 72%) were related to mechanical impurities (EO erosion processes, clogging of flowing sections of the flowing equipment, etc.).

In this regard, in order to study the process of mechanical impurities carry-over, we examined several deep samples, taken from the bottom holes of the candidate wells. Also, we proposed a method to determine the particle size distribution of selected downhole samples, i.e., to obtain differential and integral curves that reflect the size distribution of sand particles.

Previously, the applicability of the «OLGA» software was considered only under laboratory conditions with a number of appropriate assumptions. As a result of the conducted research it was found that the software «OLGA» allows to simulate the behavior of particles of mechanical impurities in a real oil well.

It should be noted that the influence of the gas factor on the conditions of removal of mechanical impurities was not studied in the work, although it is assumed that the presence of expanding gas increases the flow rate (settling) of particles in parts of pumping equipment.

5. Summary

Thus, the paper formulates an analysis method for removal of particles of mechanical impurities from horizontal wells containing heavy oil using a simulation model in the OLGA program. The confirmation of its viability was proved by making calculations for five horizontal wells with different flow rates. Calculations were used to improve the quality of removal of mechanical impurities for different diameters, when constructing a relationship between the critical diameter and the flow rate of the well.

The practical significance of the study is expressed in providing the possibility of operational justification of the necessary equipment (filters, the diameter of the shanks (in flooded wells), etc.) and the parameters of its operation in horizontal wells during the extraction of high-viscosity oil, where the removal of mechanical impurities is observed. In particular, it is possible to determine the degree of purification of the separator from mechanical impurities by the minimum size of sand particles that rise to the level of the pump suspension.

The dependencies obtained in this work are currently being used at the Uchebny field to select the appropriate operating modes of production wells, having a positive impact on economic indicators. The economic indicators of well operation are improving due to the accurate prediction of the sand removal regime, which makes it possible to select technologies that prevent the breakdown of the electric center pump. Accordingly, the losses caused by stopping the well to extract and replace the pump, repair costs, capital costs for the purchase of pumps that are rapidly failing are reduced.

In addition, a hypothesis was put forward about the influence of the gas factor on the process of removing mechanical impurities., to substantiate which and obtain reliable results, additional studies of the sensitivity of sand particle removal to the value of the gas factor are needed. It should be noted that the presence of gas in the products affects the conditions of sand removal: firstly, the flow of gas released from oil and expanding gas as the well rises contributes to accelerating the removal of mechanical impurities—the settling rate in the pump; secondly, with an increase in the gas factor, the density and viscosity of oil decrease, which leads to an increase in the relative velocity of the phases.