The Interactive Multi-Model (IMM) algorithm uses multiple motion models to simultaneously track the target, which effectively solves the problem of model mismatch when a single model tracks the maneuvering target, and is widely used in maneuvering target tracking tasks. However, the Interactive Multi-Model recognition motion model is not accurate enough, and there is a certain delay in the maneuver recognition of the target, which leads to a reduction in tracking accuracy. To solve this problem, considering that deep neural networks are very good at processing classification tasks, we introduce it into target tracking tasks, combining the respective of deep neural networks and traditional tracking filtering methods for maneuvering target tracking. we use the Recurrent Neural Networks to identify the motion model of the target and propose an improved LSTM-IMM model algorithm based on the interactive multi-model algorithm. Finally, we compare the traditional interactive multi-model algorithm and verify the algorithm using Monte Carlo simulation. The results show that the proposed algorithm has improved the recognition accuracy and recognition speed of the model, and the tracking accuracy has been improved.

Logging-While-Drilling (LWD) has incorporated almost all wireline-equivalent technology with the added advantage of logging high-angle and horizontal wells with reduced rig time, critical for cost optimization efforts. LWD measurements are affected by a rugged drilling environment, and logging interpretation with a wireline mindset leads to erroneous results. Identifying measurement artefacts from real formation information is critical for reliable log analysis. This publication discusses the most common effects of drilling dynamics and environments on LWD logs that were observed during logging and drilling wells in cretaceous carbonate reservoirs in an Abu Dhabi onshore field. Log data from more than one hundred wells are reviewed to identify several interesting effects due to bottom-hole-assembly (BHA) design, BHA driving mechanism (Rotary steerable system versus mud motor), tool eccentricity, well angle, mud properties, differential invasion, borehole condition, formation fluid properties and lithology. In a few instances, some of these effects occur simultaneously, complicating the log response. These phenomenons are discussed in detail with actual examples and compared to offset wells and response modellings. The rugged logging environment and limited formation damage due to invasion provide a unique opportunity to obtain additional insight about reservoir behavior, especially when compared to wireline data in an offset well or in the same well. Pre-job planning and modelling can use these phenomena for getting additional information about dynamic reservoir behavior. This paper highlights a few such applications. This paper explains the impact of a dynamic drilling environment on LWD measurements and serves as a ready reference to identify measurement artifacts from real formation information. It is helpful as a guidebook for log analysts, geologist, geo-steering engineers and other non-specialists to identify LWD measurement artefacts.

The Archie equation is the most common approach for calculating water saturation. The true formation resistivity that is derived from resistivity logs is an important component. In high-angle or horizontal wells (Ha/Hz) the commonly employed induction style tools and multi-propagation resistivity (MPR) tools employed in logging-while-drilling (LWD) have challenges. In particular, at bed boundaries, the formation-to-wellbore geometry affects deep-reading logs and generates artifacts such as so-called polarization horns on the logs. These effects become more significant with increases in the relative dip and the resistivity contrast between the beds. These conditions impair the use of resistivity for water saturation determination. An innovative modeling workflow to generate a true formation resistivity (Rt) from LWD MPR logs is presented. In addition, a number of case examples from Abu Dhabi reservoirs are portrayed. The workflow described in this study begins with the interpretation of borehole image data to build a structural earth model. For this, the picked boundaries are extended away from the trajectory within an investigative volume of the MPR responses and are used to constrain an inversion algorithm that solves for Rt. The inversion is conducted using eight short- and long-spaced apparent phase differences and attenuation data. Different starting models and inversion constraints are applied to evaluate the sensitivity of the inversion results. The inversion results are further qualified from the ‘misfit’ calculated by the inversion algorithm. This methodology was used to process data from multiple wells in a development field near Abu Dhabi. These high-angle wells are from carbonate reservoirs with varying characteristics (such as tight, layered, high-permeability streaks, etc) and all employ LWD measurements. The integrated data of vertical wells formation evaluation, dean stark and dynamic data (well test) were used to validate results of the case study. The processing results showed significant improvement in determining true resistivity that provided highly coherent saturation determination along the entire wellbore profile. It gave confidence in the effectiveness of the approach for an improved quantitative petrophysical evaluation in Ha/Hz wells. This new processing method is able to solve the existing issue associated with LWD measurement in high angle wells thereby improving the saturation calculation significantly.

This paper presents the successful use of LWD NMR and LWD resistivity image log technology to meet the challenge of placing wells in a thin reservoir with lateral facies variation without the use of radioactive sources and with simultaneous data acquisition to evaluate the wells and design their downhole inflow control devices (ICDs). A series of horizontal producer wells were planned in a thin reservoir with lateral facies variation. After drilling the wells they were completed with downhole ICDs. Optimum placement of the wells within the reservoir and data acquisition to evaluate the wells and design their completions was achieved without the use of radioactive sources, as these created an unacceptable drilling risk. Rapid and accurate processing of the data in real time and subsequent design of the ICDs was required to enable the completions to run in a timely fashion. The NMR permeability was normalized using the new calibration parameters that were developed by integrating NMR results with core data, and the same relationship has been tested in other lateral wells. Real-time NMR total porosity played a significant part in facilitating effective geosteering and well placement without the drilling risks associated with radioactive sources. In addition, the NMR provided a porosity distribution that was used to estimate a permeability index. This index was normalized using core permeability available from offset appraisal wells. The core and NMR log data in the offset wells were combined to derive the parameters for an NMR permeability relationship. The standard volumetric analysis results and the permeability index were used for identifying reservoir flow units using crossplots of normalized flow capacity versus normalized storage capacity (modified Lorenz Plots). These results were then used to develop the parameters for the ICD completions. High resolution LWD image log data was incorporated to select the best possible sections in the wells for isolation of the ICD segments. Following completion and stimulation, multiphase PLTs were run across the ICD compartments to evaluate the wells. These results were then compared against expectations and used in subsequent well completion designs. The results of the wells presented show that the chosen methodology enables the successful placement and completion of horizontal wells in this reservoir. Decisions about ICD completion design can be made in a timely fashion just after the drilling phase is complete, avoiding rig downtime. This approach has become the default procedure for the field and will be used for the bulk of the remaining producer wells in the reservoir.

Reservoir A is an Upper Jurassic reservoir in offshore Abu Dhabi, composing layers of dense anhydrite and porous mixed lithology of dolomite and limestone. Petrophysical study from multiple wells suggests that the rock quality within the reservoir has significant lateral and vertical variations that can result in different flow capacities. Consequently, it is crucial to identify the rock quality variations and the consequent flow capacity in horizontal wells to optimize development plan, ideally in real-time. However, these lateral and vertical variations are not visible from conventional porosity (density / neutron) logs, making identification of rock quality very challenging. This paper introduces an innovative magnetic resonance (NMR)-based real-time method of permeability prediction and rock typing. Wireline logs including NMR were acquired in a pilot well, providing porosity and extensive T2-based information (permeability index, irreducible and movable fluid volume and porosity partition). Routine core analysis was also available to calibrate the NMR data, achieving a suitable correlation for NMR permeability index calibration in this field. Several rock types could be identified with the Windland R35 technique using porosity and calibrated permeability from NMR. This identification was then validated by rock types from cores. The application of knowledge gained from the study led to advanced reservoir characterization solely based on the NMR log. The process was applied to high-angle and horizontal (HAHZ) wells where the NMR full-spectrum log while drilling was available. Several slanted wells were drilled with a fit-for-purpose logging-while-drilling (LWD) suite including NMR for geo-steering and formation evaluation. The real-time LWD NMR data helped trace a remarkable change of irreducible water level through certain layers, suggesting that the subzones of Reservoir A changed pore geometry and rock type laterally, resulting in variations of flow capacity and reservoir performance. In one example, this method indicated unexpected good rock quality in one of these subzones considering the experience from offset well. Subsequently, the LWD formation-testing tool confirmed the result with mobility measurements, proving the NMR-based methodology was valid. This process normally applies to memory data after drilling, playing a key role in designing completion strategy in a timely manner. The process is also available in real-time while drilling if full NMR data is transmitted to surface, serving as a safer logging-tool for identification of sub-zones with additional valuable information compared to regular porosity tools with chemical radioactive source.