Gazprom Neft: Clustering Design Wells for Well Pad Placement in a Multi-Layer Field

This article suggests an approach to improving the design and development of multi-layer fields, enabling the project to mirror real-life conditions of the asset under development, thus making it possible to reduce capital expenditures. The key task to achieving this objective is to find the optimal well cluster designs based on maximizing production from the target in line with the surface production facilities. The well’s purpose and operational conditions are all factored into the cluster pattern. Such well pad placement schemes should ensure the minimum of capital expenditures and technical risks during the project’s implementation through multi-factor scenario optimization. For this purpose, the length of the well and its trajectory are taken into account, as well as, the infrastructure, natural and environmental restrictions which exist in the field and in any well construction constraints. In multi-layer fields, such tasks cannot be solved using the existing software programs.

A phase of well pad distribution based on the modified method of “k-mean clustering”, is considered in this article. The clustering algorithm was adapted for minimizing the cumulative footage length of the directional and the horizontal wells, whilst accounting for their complex trajectories.

Following the clustering of the drilling targets, a higher design accuracy can be achieved by analyzing the existing well stock – analyzing past well trajectories; taking account the field geology, including structural geological surfaces, a fault model, oil reservoir boundaries, production zone oil saturation models; the placement of ground infrastructure and hydrography – to place well pads. Such a high degree of detail makes it possible to maximize the design to real world conditions, and for the modern software – to unify all of this information into a singular project.

Project economics is one of the key criteria in the field development plan to enable optimal field development. Up to and over 80% of the total volume of capital investments can be spent on well construction. Obviously, a great deal of attention should be paid to finding and designing the optimal well development program for the field[1]:

• Placement of well pads, considering the field ground infrastructure (pipelines, roads, and field site facilities), the surface topography, terrain relief, sections (slots) at the existing well pads, environmental restrictions.

• Planning well trajectories by considering the histiorically drilled wells, field geologic cross-sections, engineering constraints.

To achieve the maximum economic effect, the clustering of the wells must be carried out at an early development phase, and further on it must be adjusted and updated as a more complete picture of the field’s geological structure is obtained.

At present, this task lies upon the shoulders of the subsoil user, since no regulations have been established about the implementation of the well pad clusters in any phase of the reservoir and production engineering, field infrastructure development or well construction. The well pad design phase is often practically missing, at least as far as the whole of a field is concerned. The design is carried out on site – from pad to pad, by clustering the design wells in accordance with their offset from the wellhead, depending on the derrick load capacity. Designing trajectories in the boundaries of a well pad is designed for a group of “confident” geological drilling targets, or even for that of a single well.

Such an approach results in more complex trajectories from well to well, which also increases the drilling footage. For the whole field, the element of well pad placement optimization is missing, or just conducted locally “on site”, which is limited by the total footage length within the pad and, as a rule, the size of the available drilling rigs. Such nearsighted planning has a negative impact on the project’s economics.

Regarding multi-layer fields, another consequence of the described approach lies in potentially missing the possible development of multi-zone production targets whose drilling, if an individual well spacing pattern is applied, would normally be economically inefficient. The economic efficiency of such targets (formations) can be increased by connecting them with the target (main) production facilities using multi-zone drilling or by designing wells to create a migration system from the producing and the injection wells.

Therefore, a lot of factors must be considered while designing well pads and it is important to find the optimal solutions which will ultimately reduce capital investments.

To optimize the design cluster, as one of the well pad tasks, various criteria can be applied: for instance, the resultant bottom hole kick-off, cumulative footage of the whole design well stock, trajectory complexity (cumulative angles of the azimuthal or dimensional deviations), total well cost. The values of relevant criteria or a group of criteria need to be minimized. To be able to construct small or large well clusters, using drilling rigs with various capacities must be factored in. The quantity of well pads and the field ground infrastructure, in general, depend on this.

The task is hardly solvable using any existing software, not in an autonomous mode any way, therefore, well pads or the well cluster for multi-layer fields will require manual design work which is dependent on the qualification and skill of the design engineer, i.e. it is an intellectually demanding task. The task is to search and find the optimal well padding scheme based on production target development ensuring the development of multi-zone formations. The well objectives and operational conditions should be considered as well.

The well clustering process is considered as a point clustering task, i.e. clustering drilling targets located in the production target areas. Thus, the task turns into one of searching for the zones of densely located points, to combine them into well pads. The well pads serve as the created cluster’s center.

Points can be clustered into spaces using various algorithms. One of them is the method of “k-mean clustering”, also known as the fast cluster analysis method. Its algorithm tends to minimize the total squared error of the cluster points from the cluster centers. The Euclidean distance is used in this example, and namely, the well trajectory length. The algorithm presumes a hypothesis on the most probable cluster number and the clusters initial position. Since the well pads number is initially not known, the calculation uses various well pad numbers, and the derived figures are compared. Each calculation results in an optimal well distribution by well pads. Having a series of computations, it is possible to define an optimal value for a required criterion, such as: economic indicators, drilling rig capacities and the number of risky wells to be drilled.

The clustering outcomes may depend on the well pads initial, selected position, therefore the enhanced k-means algorithms can offer better initial centroid values for future clusters. Using the k-means method does not solve the well clustering design task immediately. It is applicable for directional wells, while having some constraints with horizontal wells [2].

The k-means algorithms can be adapted to the clustering of both directional and horizontal wells. Firstly, this can be achieved by approximating the designed well trajectory to the existing one, considering the straight hole minimal depth, the horizontal section length and its positional coordinates in the formation. Secondly, the total azimuth value varying along the borehole should be considered until the production horizon is reached. The computed, weighted well length data sets are appropriate to use when the azimuthal angle variation values along the wellbore, including the inclination angle value exceeding 90⁰, are replaced with the equivalent length trajectory value. This helps minimize the trajectory curvature in the process of searching for an optimal well pad placement pattern, and it makes the project implementation much simpler.

To find the well equivalent trajectory length value, we can proceed from the assumption of the constancy of the Drilling Difficulty Index (DDI) [3]:

where, TD is borehole measured depth, AHD is the bottom hole kick-off value, VD is the true vertical depth, TORT is the trajectory cumulative curvature.

Then the weighted trajectory length value is changed by the value of (Δφ + Δα1)/(Δφ + Δα)‧TD, where Δφ  is variation of the azimuthal trajectory angle value, Δα is the varying inclination angle, Δα1 is the value exceeding 90⁰. Thus, the “weighted footage length” is a complex value and it enables optimization based on the total length value and the well trajectory angle varying values at the same time.

The design well clustering task was approached by the authors using a multi-layer field example, comprising of 52 productive formations (16 development targets) in the northern dome, and 53 formations (18 development targets) in the southern part. The Pokur, Tangal, Sortym, Vasyugan and the Tyumen Suites represent oil and gas saturated layers occurring in the depth of 1158 to 3239m. The most prospective development targets were drilled using individual spacing pattern for the targets. Nine such targets were delineated in the northern and southern domes, correspondingly. Drilling for the other targets was economically inefficient, those targets can be developed using the multi-zone wells. The drilled well stock amounted to 247 producing wells, 68 injection wells, 20 gas wells and 13 water supply wells at the start of the reservoir development. Further on, it was decided to drill 207 producing wells, 97 injection wells and 485 sidetracks. Four complexes were outlined in the southern dome and three complexes – in the northern dome; the development targets were combined into an integrated well stock, to implement the wells returning from the target well stock to the multi-zone wells.

The principal task therefore was to justify the possibility of commingling the multi-zone development targets to develop them using the well placement pattern that was designed for the target reservoirs. The calculations determined the well design placement ensuring the penetration of all the drilling targets. An integrated field model was built up using the modern engineering design software comprising of the structural geological surfaces, the fault model, the existing well stock, the infrastructure and the topography of the field.

The engineering design phase included:

1. Design well clustering using the k-means adapted algorithm with varying preset well pad numbers;

2. Comparing the computed variants and selecting a well spacing pattern;

3. Optimizing the well trajectories and the well pad locations, considering the criteria preset for a selected well placement pattern.

Figure 1 presents the design well clustering patterns in the southern dome of the multi-layer field matching with the minimum of capital investments. Both the directional and horizontal well pad clustering was carried out using the cumulative trajectory length optimization (Fig.1,a) and the cumulative weighted well length (Fig.1,b), to compare the optimization criteria. The drilling in this case was designed for six target formations, overlapping in the map, whose wells are located at different levels, and their trajectories were separated during further planning.

Figure 1 illustrates that clusters tend to occupy the areas of densely packed points. The trajectories of most wells are adjusted along the 90⁰+ azimuth. Such wellbore paths are complex to implement; they demand higher equipment standards, the drilling equipment endurance capability, and that of production strings as well, the circulation fluid quality, its lubricating properties and the capability to support borehole walls. Besides, accident risks are higher, the service contractor’s experience and qualifications also need to increase. In the second case (Fig.1, b), the clusters are built-up using the minimum of the trajectory curvature value.

Similar well clustering design patterns were built up for different numbers of well pads. Their comparison resulted in the derived relationship between the total investment values and the number of well pads (Fig.2). Capital investments include the well construction costs, the well pad filling and engineering costs. The well’s length is at its maximal with the minimal well pad number which increases project costs. When the well pad number is at a maximum, the well pad building and field construction costs go up.

When optimizing with the cumulative trajectory length, the minimum capital investments are obtained with 11 well pads in the variant “a” (Fig.2, a), and with 10 or 18 well pads in the variant “b” (Fig.2, b), using the weighted well length. In the first case, it is possible to drill all the planned wells using the drilling unit БУ 4000/250, when clustering by 6 well pads; in the second case it can be done using the clusters of more than 11 well pads – therefore, a more acceptable variant is the one with 18 well pads (Fig.18, b).

Excluding the curved trajectories is instrumental for building longer length trajectories. The trajectories become simpler to perform but their lengths extend by 42m in an average of 182 wells. However, additional expenses are neglected while performing more complex trajectories, with significant adjustments made by the azimuth, which increases the project implementation cost if we perform the optimization via the trajectory length criterion.

In the other field section (Fig.3), the well clustering pattern is presented, prepared for the drilling of 13 formations, overlapped in the map. The delineated drilling targets can be commingled with other wells, i.e. wells can be drilled through the nearest drilling targets, however the drilling of a sidetrack can be suggested if this is going to make the trajectory more complex. The commingled target exploration will be carried out until the reserves in the base (underlying) development targets are depleted; further on the transition to the overlying formation or the drilling of a sidetrack is going to takes place. 

The wells clustered and designed for commingling with the drilling targets, can be used for the further optimization of the well trajectories within a well pad, as well as for further adjustments in the well pad positions, considering other factors, like: the geology, the existing boreholes trajectories, the hydrography and infrastructure. If some nature reservation zones or water bodies happen to be there, it is not appropriate for well pad construction, this should be taken into account in the clustering phase, and an official ban is to be issued for placing pads in such areas.

An adequate design model of a modern software package was used in the course of the multi-layer field development, to design the drilling trajectories using the actual well stock and the available ground infrastructure as the input data. The designed well trajectories were built considering the technical constraints and the hazards of approaching the already drilled wells. The knowledge of the geologic conditions enabled a higher accuracy for the well pad placement and well trajectory design to be achieved. The following geological information was used to make up the project:

• Structural geological surfaces (the top and bottom of productive formations) and the fault model as part of the field’s uniform structural framework;

• Reservoir limits;

• 3D model of the hydrocarbon saturation of the development targets.

Such a level of detail enables maximum reality to be reflected in the design. The structural surfaces determine the intersection points between the design wells and the multi-zone development targets, offsetting them if needed, creating a development strategy using the multi-zone well stock, taking into account the well drive mechanism (production well, injection well etc.) The fault model clearly illustrates that the well trajectories pass through structural failures, and, depending on the fault permeability, a decision can be made if it is necessary to offset a borehole.

The reservoir boundaries and the hydrocarbon charge model help monitor the drilling targets and the intersection points between layers along the formation and adjust the well positions according to the residual oil saturation values.

Figure 4 presents the oil field’s structural framework with the topography map and the ground infrastructure (roads, pipelines, buildings, well pads, rivers, water bodies). Such information enables obtaining more accurate information about the well pads positions considering the natural obstacles, the infrastructural assets, the environmentally restricted areas. Figure 5 presents the existing well stock distribution pattern and the design well stock with sidetrack in the development targets hydrocarbon charge model.

When the reserves are estimated, the field unitary structural frame work should be developed and the fault model prepared. To enable a collaborative work between the drilling and engineering functions, a strategy should be developed on how to operate the target production zones, considering the designed well trajectories.

Designing well pads in the course of project planning provides the following advantages:

1. The implementation of a multi-factor optimization of the clustering patterns, considering the “easy” and the “risky” wells.

2. The possibility of a more detailed calculation of the project’s economics, considerable reduction in the capital investments for the drilling of the design wells.

3. Maximizing the project’s reflection of the field’s real-life conditions.

4. The design’s improved quality in all the phases: from the technological development plan to the well construction project and the program/plans for drilling each individual well bore.

5. The exclusion of technical risks of the project implementation at an early stage of field development.

List of References

1. V.A.Karsakov, S.V.Tretyakov, S.S.Devyatyarov., A.G.Pasynkov «Well Construction Capital Investment Optimization during Field Development Conceptual Engineering” Oil Industry Journal. 2013. – Issue 12. – pp. 33–35.

2. A.F. Mozhchil, S.V.Tretyakov, D.E.Dmitriev [and others]. “Technical and Economic Optimization of Well Padding in the Integrated Conceptual Design // Oil industry Journal. – 2016. – Issue No. 4. – pp. 126-129.

3. Mark J. Kaiser “A Survey of Drilling Cost and Complexity Estimation Models” // International Journal of Petroleum Science and Technology. – 2007. – V. 1. – No. 1. – 2007. – pp. 1–22.

A.G. Shatrovskiy, A.S. Chinarov, M.R. Salikhov (LLC Gazpromneft NTC)