New Geosteering Work Flow Integrates Real-Time Measurements With Geomodels

This work presents a systematic geosteering work flow that automatically integrates a priori information and real-time measurements to update geomodels with uncertainties and uses the latest model predictions in a decision-support system (DSS).

Fig. 1—Proposed geosteering work flow. The top part contains inputs to the work flow. The left part of the figure depicts the update loop. The part of the figure to the right contains the decision system that is based on the updated Earth model. The “drill ahead” decision results in new measurements that trigger another update and complete the full loop of the work flow.

To place a well in the best possible reservoir zone, operators use geosteering to support real-time well-trajectory adjustments. Geosteering refers to the process of making directional well adjustments on the basis of real-time information acquired while drilling. This work presents a systematic geosteering work flow that automatically integrates a priori information and real-time measurements to update geomodels with uncertainties and uses the latest model predictions in a decision-support system (DSS). The DSS supports geosteering decisions by evaluating production potential against drilling and completion risks.


This paper presents a consistent, systematic, and transparent work flow for geosteering. The starting point is a priori information, for example a probabilistic geomodel representing a geological interpretation based on surface seismic and logs from offset wells, including relevant interpretation uncertainties. Multiple geomodel realizations of the possible geological scenarios span the space of interpretation uncertainties. The real-time measurements obtained while drilling are continually integrated by updating the realizations automatically using an ensemble-based filtering method. The real-time update of the realizations leads to a reduction in interpretation uncertainty, providing up-to-date predictions of the geology ahead of the bit consistently. The update work flow is linked to a DSS. The DSS applies the probabilistic up-to-date geomodel to support geosteering decisions under uncertainty by evaluating the chosen value function of the well. The value function commonly includes multiple objectives, including production potential, costs for drilling and completion, and risks associated with the operation. The DSS presented here is optimized specifically for use with ensemble-based update work flows that are increasingly popular in the oil industry.

The focus of this paper is the DSS. The DSS suggests steering correction or stopping, optimizing well trajectories over the ensemble of up-to-date geomodel realizations. A graphical user interface (GUI) enables geosteering experts to control the input to the DSS by means of interactive selection and adjustment of value functions and constraints (e.g., dogleg severity). The adjustments are applied in a matter of seconds using advanced dynamic programming algorithms that yield consistently updated decisions. The proposed steering decision is communicated through the GUI, which contains a visual representation of the current uncertainty and trajectory possibilities that give the best value for each realization.

Real-Time Update Loop

In the proposed geosteering work flow (Fig. 1 above), real-time decision support is based on a probabilistic geomodel that is updated in a loop. The geomodel is represented as an ensemble of realizations that captures key geological uncertainties. The predrill realizations are created on the basis of seismic, logs from offset wells, production measurements, and additional knowledge about geological uncertainties provided by experts. While drilling, all realizations are updated continuously and automatically by incremental assimilation of the measurements acquired during the ongoing drilling process.

Measurements. By design, the ­ensemble-based methods perform incremental updates that can handle any number and any type of measurements simultaneously (see Fig. 1, Step 1a). A corresponding simulation model, however, is required that can transform the realizations and the measurements to a context in which they can be compared adequately to compute the mismatch.

Forward Modeling. The main contribution of this paper is the DSS and not the modeling of the measurements. Therefore, the authors use a simple integral model for electromagnetic measurements in their examples in the complete paper.

Ensemble-Based Update Algorithm. The update loop used is compatible with a number of ensemble-based methods that have been used previously for reservoir data assimilation, including the ensemble Kalman filter (EnKF), the ensemble smoother, the particle filter, and more-sophisticated combinations of the three such as the adaptive Gaussian mixture filter.

When new measurements are received, they are compared with the simulated measurements generated by the corresponding forward models for each realization.


The update loop results in an up-to-date ensemble of model realizations that contains both the prior knowledge and the latest measurements. The realizations are the input to the DSS. The DSS is based on optimization algorithms that take into account all realizations as well as multiple objectives, such as following the reservoir top while minimizing drilling cost and reducing the tortuosity of a well for easier completion. The objectives often represent contradicting interests, and, for each realization, the algorithm calculates a well path that is optimal with respect to the weight of each objective.

The DSS presented in this paper differs dramatically from traditional decision systems that were designed for strategic decisions; the geosteering decisions are operational, which means that they are sequential and need to be made in a relatively short time. Because several sequential decisions are handled by the DSS, the possibility of tweaking the weights of the objectives and previewing the outcomes in real time was introduced, allowing the user to understand how the choice of objectives influences the suggested decisions and provides a possibility to re-evaluate the trade-offs between objectives as the drilling progresses.

Objectives. A natural requirement for any DSS is the possibility to take into account multiple objectives. The objectives used in modern geosteering operations include placing the well in a specific position in the reservoir, reducing costs, and ensuring safety. For use in a DSS, the objectives need to be converted into objective functions that are defined in terms of a common metric (e.g., the estimated profit in US dollars or produced equivalent barrels of oil).

Ensemble-Based Optimization.

Ensemble-based work flows represent the model (parameter) uncertainty as a set of Earth-model realizations. At the same time, the outcome of the optimization should be a single decision for each decision point. An ensemble-based optimization approach works directly with all realizations to optimize the trajectory in a robust way.

The approach in which only the next segment of the well is considered for the optimization is called “myopic” in decision analytics or “greedy” in computer science. Greedy approaches do not provide an optimal solution for many optimization problems. In the DSS framework, the aim is to handle any type of objective function provided by the geosteering decision team. Hence, applying a global optimization algorithm is necessary. A global optimization algorithm in this paper is defined as one that optimizes the complete well path ahead of the bit against the currently available representation of the geological uncertainty (as represented by the set of realizations) and does not fall into local minima.

Dynamic Programming for Ensemble Optimization. This paper presents a new dynamic programming discrete optimization strategy that operates directly on the up-to-date ensemble of geomodel realizations obtained from the EnKF update loop. For simplicity, near-horizontal drilling is assumed and, therefore, the decision points are distributed along the horizontal axis ahead of the current decision point. At every decision point, the trajectory alternatives are discretized. The ­density of points (and hence decisions) can be decided by the user and affect the trade-offs between the optimization accuracy and the computational time. The optimization algorithm evaluates different trajectories that are represented as ­piecewise-linear curves that go through one possible depth for each decision point.

The optimization algorithm presented here extends the classical robust optimization to include the up-to-date knowledge for the full trajectory ahead of the bit. It is essential that only the first point is chosen through the ensemble-based optimization while the rest of the trajectory is deterministic within each realization. The full well path joint optimization for the whole ensemble is costly and unnecessary for a work flow in which updates of the realizations are performed sequentially in time when new measurements arrive during drilling. Instead, the decision for the next step (the next decision point) is recomputed once the new measurements become available and the ensemble is updated. This strategy allows for real-time reaction to new information while considering the prior information at every timestep.

Visualization of the Real-Time Modeling Results. Adoption of any DSS requires that the system can be trusted by its users. Therefore, the communication to the user of the reasoning behind the proposed decisions is essential. In the user interface, the proposed decision is visualized and the basis for the decision is explained.

At all times, the interface highlights the consequence of the immediate decision in thick red type and with a written communication of the decision. The display is fully updated when the realizations are updated or if the user adjusts the objectives as explained in next subsection.

Usage of the DSS. The possibility of updating the objectives and constraints of the decision optimization in real time is important for understanding the decision context and the decision outcomes. The flexible design allows for adjustment of both the weights and the parameters of the objective functions.

The dynamic programming algorithm has a relatively low computational complexity. When the parameters for an objective function are altered, or different objective functions are applied, the suggested decision can be recomputed and visualized in a few seconds. This enables real-time experimentation and helps in understanding how the selection of specific objectives and their parameters influences the suggested decisions and decision outcomes. This helps geosteering experts to formulate the objectives in a more-precise manner for current and future operations, providing even better decision support during the operation.

This article, written by Special Publications Editor Adam Wilson, contains highlights of paper SPE 191337, “An Interactive Decision Support System for Geosteering Operations,” by Sergey Alyaev, IRIS; Reidar Brumer Bratvold, SPE, University of Stavanger; and Xiaodong Luo, SPE, Erich Suter, SPE, and Erlend H. Vefring, SPE, IRIS, prepared for the 2018 SPE Norway One Day Seminar, Bergen, Norway, 18 April. The paper has not been peer reviewed.