Statistical and Numerical Investigation of the Effect of Wellbore Trajectories in Williston Basin Horizontal Wells and their Effects on Production Performance

Introduction

The Bakken and Three Forks oil-bearing formations, located in the Williston Basin, produced in North Dakota and Montana in the United States, and Saskatchewan and Manitoba in Canada. Most wells in this area are drilled in the Three Forks or Middle Bakken members. Currently, 65.8% of the existing wells are horizontal, the remaining are either vertical or directional [1]. Along with the development of hydraulic fracturing, horizontal drilling enabled the production of unconventional and tight formations basin- wide [1] Figure 1.

Click to enlarge

Click to enlarge

Oil production, after being dominated by vertical wells in the past, is now obtained at 98% from the horizontal hydraulically fractured wells as shown in Figure 2. The 22,867 horizontal wells have an average lateral length of

8,500 ft for a total depth of 19,200 ft with an MD/TVD ratio of 2.25 on average. Since 2006, wells in the Williston basin were drilled using extended-reach drilling. Hence, the lateral section length is longer than what is usually practiced for conventional wells [1].

With advancement of technology, especially the bit design, the lateral length continuously increased over time as it is seen from Figure 3. This will increase production due to higher reservoir contact and larger numbers of hydraulic fracturing stages (see Figure 4), given the lower productivity of unconventional wells.

Click to enlarge

Over time, not only the lateral length increased, but so did the total length to true vertical depth ratio, a consequence of the domination of horizontal wells, as depicted in Figure

4. The average number of hydraulic fracturing stages first gradually increased to 47 from 2009 until 2019, then decreased and stabilized at 40 stages.

Click to enlarge

Two parameters are usually used by drilling engineering professionals: undulations and tortuosity. While the undulations are two-dimensional in terms of true vertical depth variation versus measured depth, the tortuosity takes into account the dogleg severity deviation between the planned well path and the as-drilled path. The latter is defined by the average excess of dogleg severity over the path, the mean unwanted dogleg severity, and the average angle change (AAC) as described by Weijermans, et al. [2], the latter of which has been adopted in this study. The lateral well path of the Nordeng 34X-23A well is shown in Figure 5, where the as-drilled path oscillates above and under the target envelope and its center is designed to follow the middle Bakken.

Intentional undulations are usually performed to enhance the inflow performances from the reservoir to the wellbore by increasing the contact area and reducing the dependency on the vertical permeability while enabling the access of multiple thin layers of reservoirs [3].

Unintentional undulations result from the directional drilling control limitations caused by either formation characteristics, directional drilling technology used, or both [3]. Several technologies have been developed to enhance the directional drilling rotary steerable systems (RSS) control and steerability, enabling the undulation reduction (amplitude and index), such as the continuous proportional steering method (CPSM) which reduces the average angle change by up to 4 folds [4].

In the literature, undulations are usually considered a well-centric parameter, and no basin, field, or statewide statistical analysis has been previously reported. An overview of the undulations’ magnitude and their spread over a wide range of wells, fields, and zones located in North Dakota provides a quantified view of how well paths deviate from an ideal one. On the other hand, the effect of undulations on production performances has been studied by Tang, et al. [5] and can be looked at from different angles. These can be reservoir inflow, wells liquids loading, sand transport and accumulation, and unstable slug flow. Excluding a deliberate and specific undulated configuration that increases reservoir- to-well contact, all the undulation effects are negative and jeopardize the production performances of the wells.

Click to enlarge

The present study discusses the statistical analysis with regards to Williston wells trajectories distribution, and their related vertical depth variation amplitude, with an effort to correlate the production performances to the trajectory and undulations of the well. Its novelty relies on the methodology, which requires newly defined KPIs’ and the large database from which it extracts the trajectory and production data and calculates the key performances indicators; the production efficiency index is a parameter defined for the sole objective of this study, trying to capture trajectories and undulations effect.

First, the database is analyzed, and wells that correspond to trajectory and production history criteria are selected; then, the production efficiency index of oil, water, and gas is calculated for each well along with the wells’ AAC, AAA, MA, and SI trajectory characteristics. The dominant trajectories are identified using the distributions of each parameter for an assessment of the explored data ranges and dominant structures. The final step is to assess whether a quantifiable correlation or trend can be captured between production efficiency indexes and trajectory or undulations related parameters.

Trajectory Effect on Severe Slugging Tendency and Production Performances

The effects of undulations on production performances are briefly reviewed in this section with reference to the lab experiments, field data, and simulation works performed by researchers.

Tang et al. [5] studied the effect of sinusoidal horizontal well lateral undulations on the slugging tendency of the multiphase flow. The study showed that undulation can onset slugging flow regime through a flow chocking due to heavier phases (water and oil) accumulations in sumps. In this vein, higher undulations amplitude reduces the slugs’ frequency and increases the slug volume or liquid inventory change.

Norris [7] conducted transient flow simulations for different well geometries and flow conditions. They showed that, as the pressure declines below the bubble point, severe slugging is expected. It is suggested that the toe-down geometry generates fewer instabilities compared to the toe- up configurations. The suggested remedial recommendations included higher surface pressures, smaller lateral section diameter, and vapor- liquid separation at the heel. Veeken, et al. [8] found that the simulations performed for liquid loading of a gas well, the simulations did not show an effect of the undulation on the critical gas velocity required to lift the liquid to the surface, while experiments showed that the undulations affect the critical gas velocity and the onset of the liquid loading.

Malekzadeh and Mudde [9], through their simulation works, analyzed the effect of undulations using Vogel’s modified inflow performance relationship for horizontal wells [10], which concluded that severe slugging may be initiated by the trajectory undulations occurring at the bottom of the well, and the pressure fluctuations can influence the reservoir performances.

Brito, et al. [11], after experimenting with a toe-up, toe- down, one hump, and one sump undulations observed that the slug lengths were up to four times higher than the ones reported in the literature. Additionally, the lateral deviation angle and length affect the vertical sections of the slug flow development, and the lateral section slug flow development is affected by the liquid fall back from the vertical section and lateral length. Slug Merging between lateral and vertical sections occurs by a decreased frequency and an increased slug length, while dissipation can be observed if short unstable slugs are generated at the lateral section.

Suarez-Rivera, et al. [12] based their study on 212 wells from the Eagle Ford and observed that the 5% to 20% reservoir quality (RQ) variation does not explain the large variation of the production performances. Pankaj, et al. [13] analyzed the effect of the wellbore quality (WQ) and completion quality (CQ) on production performances. With completion quality being related to the geometry of the hydraulic fractures, the wellbore quality is dependent on the wellbore pressure loss driving factors, such as the connection to the hydraulic fractures, the undulations, and production technique and flow back practices. The author considered the following trajectories: horizontal, toe-down, toe-up, hold-toe-up, and hold-toe-down. Pankaj, et al. [13] concluded that horizontal and toe-up trajectories have the highest productivity index, whereas the hold-toe- down has the lowest. The GOR variation is the largest for the hold toa e-down, while the water cut variation is the smallest for the same case. The toe-up and hold-toe-up trajectories, showed the earliest onset of unstable production, hence severe slugging due to higher liquid accumulations at the heel.

Important to note that the flat horizontal trajectory shows the highest cumulative, whereas the toe-down and hold toe-down, are the most underperforming trajectories, due to production impairment of the stages close to the toe, by water accumulation.

Tebowei, et al. [14] using Computational Fluids Dynamics (CFD), showed that the undulations influence sand transport in the horizontal laterals undulated liner, where the solid transport regime is significantly altered, and sand deposition occurs downstream of a sump (V-inclined) at much higher fluid velocities.

Tran, et al. [15] performed a simulation study and showed that during early years the trajectory has minimal impact, and significant effect on slugs’ severity at late years.

With highest instability observed for toe- up trajectories, low undulations increase the flow instability while large undulations reduce the severity of slugging. Toe-down shows the lowest instabilities and higher cumulative, the study conclude that undulations do have a negative effect on toe- down trajectories and a positive effect on toe-up trajectories, but no quantification is performed.

Khetib, et al. [16] captured the geomechanical effect of the severe slugging pressure variations on the hydraulic fracture aperture through effective stress variation and showed that for a typical Williston basin well completion and flowing condition the hydraulic fracture is subject to large aperture variation and proppant dis-embedment is possible. It is worth noting that Khetib, et al. [17] demonstrated experimentally and numerically that it is possible to dampen the flow instabilities by using a dampening pipe volume, such as the annular space of the well.

Williston Basin Statistical Analysis Methodology

Considering a lateral section starts at an angle lower than ±5° from the horizontal (85-95° from the vertical), a statistical analysis was conducted on 10,428 wells where the angles of the lateral section deviation survey were considered for the analysis. The 10,428 wells contained exclusively horizontal wells and re-entry, discarding the vertical and directional wells from the 38,140 wells database. Further criteria were used to refine the database down to a selection of wells suitable for our study:

• Drain length: To capture sufficiently long lateral sections, we selected only wells with a minimum length of 5000 ft (1,524 m), which represent the typical length of the unconventional well’s development in 2006, (Figure 5).
• Production pool: Only wells that target the Bakken pool were selected.
• Data availability: Only wells that have a significant production history, a deviation survey, stimulation data and formation thickness data were selected.

After applying the aforementioned criteria, the number of statistically analyzed wells was reduced to 2180 well. The code used to perform the analysis is available through GitHub Allam & Khetib [18], an illustrative schematic of it.

Transient Multiphase Flow Simulation Results

In order to show the effect of the well trajectory and undulations on the flow performances transient multiphase flow simulation models are built and simulated for two wells from the analyzed database, well THORVALD 1-6H (NDIC # 17451) from Rattlesnake point field, and well HAWKINSON 10-22H1 (NDIC # 24285) from Oakdale field, both targeting the Middle Bakken formation. The well THORVALD 1-6H is a straight toe- down well with a depth variation ~50ft as shown in Figure 15.

Click to enlarge

In terms of Angles variation distribution, the well do show majority of negative angles along its toe-down trajectory as shown in Figure 16, with an Average angle of -0.19, and average angle change of -0.689, a sinuosity index of 1.000089.

Click to enlarge

The second analyzed well show a large undulation with a toe-down section followed by a toe-up one as shown in Figure 17, a configuration particularly prone to flow instabilities.

Click to enlarge

The well trajectory does show three peaks of angles at different values, negative, null and positive, corresponding to the downward section, the straight and the upward section respectively as shown in Figure 18, the well have an average angle change of 0.0044, an average angle of 0.226, and a sinuosity index of 1.000124.

Click to enlarge

Click to enlarge

Two models built on the same fluid PVT data, and well productivity, only well trajectory differs, when comparing the two wells liquid inventory results the well 24285 is highly unstable compared to the 17451, for a simulation run of 5 days, the well generate approximately one slug per day, while the 17451 although unstable oscillations exist, but the magnitude is very small. As shown in Figure 19.

The inventory variation, creates different vertical column heads from one moment to another, inducing relatively large pressure variations in the well bottomhole pressure as shown in Figure 20.

Click to enlarge

These instabilities translate also to wellhead flowrates, which intermittently produce liquids and gas as shown in Figure 21, contributing to surface equipment operational issues and control difficulties, promoting carryover and carry under for the three phase separation systems, and in case of compressions, increasing tripping risk.

Click to enlarge

Same observation for gas flowrate which flows intermittently at wellhead with different severity (Peak flowrate and cycle duration), the well with one undulation is showing a more unstable flow (Figure 22).

Click to enlarge

Conclusion

The Average Angle Change (AAC) was successfully used to identify the lateral section orientation (Toe-up, Toe- down, hold and Toe-Up, hold and Toe-down, Horizontal). The frequency distribution may be used to characterize more complex configurations such as hold-toe-up, hold-toe- down, hold toe-down, and toe-up. Williston basin horizontal wells drilled in North Dakota are dominated by a toe-down configuration. From a flow assurance point of view, and based on the reported literature, this configuration influences production performances by reducing severe slugging, increasing the frequency, and reducing the slug lengths. However, potential sand production will block the flow close to the toe region. Furthermore, according to the literature, as observed in Eagle Ford wells and as demonstrated by the transient multiphase flow simulations results, the toe- down trajectory is less prone to flow instability (less severe slugging). At the same time, it promotes water accumulation and exhibits the highest variation in GOR between the early and late-life periods of the well.

The dominant maximum amplitude varies between 20 and 50ft of vertical depth variation between the heel and toe of the wells. In contrast, the dominant lateral section sinuosity is less than 20% of the observed range.

As a trend, higher Average Angle Change and TVD Maximum Amplitude correspond to lower Normalized Production Efficiency. With no quantifiable correlation or direct relationship to be mentioned.

In terms of arithmetic average angle, the observed trend shows a bell-shaped form with lower production efficiencies at higher deviation angles.

The sinuosity index didn’t show any interpretable results with regard to the relationship with production efficiency; thus, the effect of the undulations could not be captured.

Finally, an aerial assessment of the production efficiency distribution was made with the heat maps. Still, no pattern follows the tier 1, tier 2, and tier 3 production performances as generally observed across the Williston basin in North Dakota.

Although it is tempting to confirm that the defined production efficiency does not depend on the location of the well in the basin, is not affected by the formation dip, and is not affected by the parent-child well effect. No solid conclusion can be drawn on the existence or not of a co-correlation between these factors. More refined data and methodology are required to state the effect of these parameters on production performances.

Acknowledgment

We acknowledge the financial support of the North Dakota Industrial Commission (NDIC). We also acknowledge Enverus Company for providing access to the Drilling Info database.