Integrating static and dynamic hierarchical clustering and its application to retail segmentation

Track and Trace: Integrating static and dynamic data in a hybrid graph-based traceability model

In this paper, we investigated the customer churn prediction problem in the Internet funds industry. We designed a novel feature embedded convolutional neural networks (FE-CNN) method that can automatically learn features from both the dynamic customer behavioral data and static customer demographic data and can utilize the advantage of convolutional neural networks to automatically learn features that capture the structured information. Our results show that our FE-CNN model outperforms the other traditional machine learning models with hand-crafted features, such as logistic regression (LR), support vector machines (SVM), random forests (RF) and neural networks (NN) in terms of accuracy, area under the receiver operating characteristics curve (AUC) and top-decile lift. Furthermore, we found that after adding the demographic data feature to the basic CNN model, the performance of the FE-CNN model improved. Overall, we found that the FE-CNN is the most powerful way to solve the problem of customer churn prediction in the Internet funds industry. Our FE-CNN method can also be applied to other fields that have both dynamic data and static data.

Mature and complex reservoirs are usually covered by many kinds of static and dynamic data—i.e. petrophysical, PVT, well testing, geologic, completion, and reservoir production information. This paper illustrates an effective methodology to develop a consistent simulation model, via integration of the available static and dynamic data sets, that allows better understanding of the subsurface reservoir flow characteristics. A complex gas condensate reservoir, referred to as reservoir X and first put into production 14 years ago, serves as the example in this paper. There is abundance of available data from different sources. However, the data can not be directly applied to build a reservoir prediction model because of the various measurement scales—e.g., core = about 0.1 ft long, log = tens of feet away from wellbore, and reservoir simulation = hundreds of feet for the grid size. In some cases, the same parameter is derived from different sources—i.e., permeability from core samples, logging data, and pressure transient test. Due to the extreme intrinsic complexity of reservoir X, actual subsurface flow characteristics were not well understood before this study. By using the proposed methodology, a reliable, coherent and improved reservoir simulation model was built. In the past decade, the multidisciplinary approach (usually including geology, geophysics, petrophysics, reservoir engineering, and production engineering) has gained popularity in industry's attempt to achieve more effective reservoir management. This paper focuses on a multidisciplinary approach that integrated PVT data, well tests, geologic, petrophysical, and dynamic reservoir production information to build a reservoir simulation model, which can be used to predict reservoir performance. Reservoir X, discovered in early 1980s in Tarim Basin (P.R. China), is a gas condensate reservoir without oil rim. Geologic formations of this reservoir are relatively flat with only a few degrees of inclination. The depositional environment is fluvial-delta with bedded sandstones. The southern boundary …

We describe a method for the development of the optimal spatial distributions of the porosity phi and permeability k of a large-scale porous medium. The optimal distributions are constrained by static and dynamic data. The static data that we utilize are limited data for phi and k, which the method honors in the optimal model and utilizes their correlation functions in the optimization process. The dynamic data include the first-arrival (FA) times, at a number of receivers, of seismic waves that have propagated in the porous medium, and the time-dependent production rates of a fluid that flows in the medium. The method combines the simulated-annealing method with a simulator that solves numerically the three-dimensional (3D) acoustic wave equation and computes the FA times, and a second simulator that solves the 3D governing equation for the fluid's pressure as a function of time. To our knowledge, this is the first time that an optimization method has been developed to determine simultaneously the global minima of two distinct total energy functions. As a stringent test of the method's accuracy, we solve for flow of two immiscible fluids in the same porous medium, without using any data for the two-phase flow problem in the optimization process. We show that the optimal model, in addition to honoring the data, also yields accurate spatial distributions of phi and k, as well as providing accurate quantitative predictions for the single- and two-phase flow problems. The efficiency of the computations is discussed in detail.

Clustering is an effective unsupervised machine learning method that can be used as a stand-alone heuristic or as a part of a data mining process. The goal of clustering analysis is to partition data into groups with high intra-cluster association, and low inter-cluster association. Hierarchical clustering requires minimal parameters, has flexibility with similarity measure, and has strong visualization power, all of which makes it ideal for exploratory analysis. Hierarchical clustering is a particular branch of clustering algorithms where the results are not given as partitions, but rather a nested structure, which can represent the ordering among elements within a dataset. The study of the performance of hierarchical structure on time series data is limited. In this paper, we examine the hierarchical structure of time series datasets. The most popular hierarchical structured clustering heuristics include the Hierarchical Agglomerative Clustering, which is distance based; and Ordering Points To Identify the Clustering Structure, which is density based. Both share many similar characteristics and are suitable for time series data processing. We examine the performance of different hierarchical clustering algorithms with time series both internally as well as externally.

Listed firms all over the globe are moving their focus from short‐term profit maximization to long‐term environmental, social, and governance (ESG) objectives. Most business leaders are now cognizant of the growing importance of ESG concerns, which may have an effect on an organization’s financial health as well as its reputation in the marketplace. A recent study found that countries with strong ESG performance may be able to improve their financial performance. Yet, in China, the subject of “how does ESG impact financial performance” has received little attention. In this paper, we examine the link between ESG operations and financial indices in China’s publicly traded firms using dynamic and static panel data analysis. We begin by gathering financial data and preprocessing it using z‐score normalization. The consequence of ESG variables on the financial performance of the company under the pandemic was investigated using statistical analytic techniques such as the Pearson correlation test, logistic regression model, and Fisher’s exact test. Due to the study’s dynamic and static data, comprehensive ESG has a considerable influence on corporate value and profitability per share. The impact of ESG variables on the financial performance of the company in the event of a pandemic was analyzed using analytical methods such as Pearson correlation, logistic regression, and Fisher’s exact test. Performance in ESG may boost financial performance, which could impact investors, business administrators, decision‐makers, and industry regulations.

A Practical probabilistic history matching framework for characterizing fracture network parameters of shale gas reservoirs

A pilot point guided pattern matching approach to integrate dynamic data into geological modeling

Surface roughness prediction based on fusion of dynamic-static data

This paper presents a new methodology to continuously update and improve fracture network models. We begin with a hypothetical model whose fracture network parameters and geological information are known. After generating the "exact" fracture network with known characteristics, the data were exported to a reservoir simulator and simulations were run over a period of time. Intelligent wells equipped with downhole multiple pressure and flow sensors were placed throughout the reservoir and put on production. These producers were completed in different fracture zones to create a representative pressure and production response. We then considered a number of wells of which static and dynamic data were used to model well fracture density. When new wells were drilled, historical and new data were used to update the fracture density using Artificial Neural Networks (ANN). More dynamic data will be provided as well as more static data when additional wells are drilled. The accuracy of the prediction model depends significantly on the representation of the available data of the existing fracture network. The importance of conventional data and smart data prediction capability was also investigated. A highly sensitive input data was selected through forward selection scheme to train the ANN. Well geometric locations were included as a new link in the ANN regression process. Once the relationship between fracture network parameters and well performance data was established, the ANN model was used to predict fracture density at newly drilled locations. Finally, an error analysis through correlation coefficient and percentage absolute relative error performance was performed to examine the accuracy of the proposed inverse modeling methodology. It was shown that fracture dominated production performance data collected from both conventional and smart wells allow automatically updating the fracture network model. The technique proposed helps in generating another -readily available at no cost- data source for fracture characterization to be used as supplementary to limited 1-D data obtained from well logs and cores.

When considering the task of creating reservoir models for fields under development, dynamic data measurements often have limited impact compared with static (geophysical and geological) data. This is not necessarily true for the Johan Sverdrup field offshore Norway, where exceptional reservoir properties make the data from eight drill-stem tests (DSTs) particularly interesting. For this reason, it is important to utilize the information found in the collected static and dynamic data in a consistent manner, to improve the understanding of the reservoir. This is especially true for the Avaldsnes High area, located in the southeastern part of the Johan Sverdrup field, where the observed thickness is below the seismic resolution, and the DST data from four wells indicate permeabilities in the range of 20 to 80 Darcy, with an overlapping radius of investigation. In this paper, we apply an ensemble-based approach to generate a large set of reservoir models for the Avaldsnes High area of the Johan Sverdrup field, all of which are plausible given the current observed static and dynamic data. We consider multiple modelling scenarios, introducing uncertainty in the sand thickness, facies (rock type) description and the permeability modelling. Unlike conventional pressure transient analysis (PTA), where we analyze the DSTs separately, and the non-uniqueness in the data interpretation is hard to address and quantify, this is not the case with the ensemble-based approach. Since we conduct the static and dynamic data conditioning simultaneously, we can consistently address possible ambiguities in interpreted permeabilities, thicknesses and flow barriers seen in the conventional PTA analysis. The study reveals that by conditioning the generated models to dynamic data we introduce clear spatial trends in both the sand thickness and permeability. In particular, we greatly reduce the potential downside with respect to the sand thickness in the Avaldsnes High area.

A giant Middle East Carbonate Field is producing since 47 years with more than 700 borehole penetrations. The reservoir is 400ft thick of Lower Cretaceous shallow water carbonates with large intra-shelf basin. The reservoir has complex internal architecture due to the formation and in fill of the Basin. Although connected to a large aquifer, with production the reservoir pressure declined by almost a 1000 psi within 10 years. Peripheral injection started to restore the reservoir energy and sustain production. Injected water in the periphery naturally went into the most permeable layers causing uneven sweep both aerially and vertically. Especially in the progrades with sequence of porous/dense intervals and typical permeability range between 0.1-10 md, oil production could not be sustained with peripheral injection, resulting in dedicated pattern injection schemes. The reservoir unit, subject of this paper, is geologically complex due ro-grading and oblique clinoforms intervened by dense subunits, overlain with high permeable unit. The unit was developed in the last 5 years using gas-injection line-drive in the northern flank (permeability less than 2 md) and inverted 9-spot WAG patterns. Production from the area has not reached the target rate based on simulation. In addition, some injectors are suffering from low injectivity and based on pressure response compartmentalization is an issue. This study integrates newly acquired data through additional wells, 3D seismic reprocessing, updated stratigraphy studies, better understanding of the porous/dense correlation, updated static model and dynamic performance data (well testing, transient surveys, saturation logging, production logging, etc…). Main conclusions of the study include: Lack of connectivity is largely attributed to completing wells in different clinoforms.Producers and injectors do not have uniform profile across the horizontal section.Injected gas in the WAG is not contained and has percolated to the upper unit. As result of integrating the static and dynamic data, the newly developed model will be utilized to optimize well trajectories, reduce pattern size and study the merits of continuing the WAG injection scheme.

The aim of this work is to present the effectiveness of a fully integrated approach for ensemble-based history matching on a complex real field application. We show that the predictive ability of the ensemble of models is greatly enhanced through an integrated workflow promoting multidisciplinary collaboration between all subsurface disciplines. Consistent integration of geological and engineering understanding within the dynamic data conditioning phase of a reservoir study is a challenging task. The ensemble-based approach offers an efficient solution to this challenge, especially when it is tied in with an appropriate reservoir modeling approach, suitable parameterization and repeatable workflows. All these are key factors to ensure geological consistency while improving the match quality and reliability of the generated reservoir models. One key feature of the ensemble-based method is that it overcomes the typical limitation of the traditional approaches where the number of uncertainty parameters often has to be reduced resulting from practical or algorithm constraints. This is especially important on complex reservoirs, where the subsurface uncertainties cannot be represented in a handful of scalar multipliers while honoring the static and dynamic data measurements. We organize the proposed methodology in four main steps: firstly, an initial ensemble of models able to capture the model uncertainties in all parts of the modeling processes are generated. Next, the match level for different observed data specifying the likelihood error is assigned. Then the uncertainty parameters that should be modified in the history matching process are identified. Finally, we perform a computational intensive step on large-scale computing facilities, where a state-of-art iterative ensemble algorithm, tightly connected with standard geo-modeling tools, is used to consistently update the full set of models. The result is an ensemble of reservoir models where we can quantify both the uncertainty in the production forecast and in the reservoir model parameters. The methodology has been applied on a challenging History Matching problem for a turbiditic channel complex reservoir. We demonstrate the efficiency of the method, by generating an ensemble of reservoir models that offer geologically consistent explanations of the currently measured static and dynamic data. The model forecasts reliability is confirmed through a validation process using dynamic data that have been left out of the history matching process. More importantly, however, is that the proposed workflow, and resulting models provides a good starting point for a more efficient update of the reservoir models as new static and dynamic data become available. With the proposed approach is possible to move from sporadic updates of reservoir models towards frequent model updates when the new data arrive. Utilizing all available data in a consistent manner - when the data are collected - is the key to making more robust reservoir management decisions, especially for complex reservoirs.

Dynamic and static feature fusion for increased accuracy in signature verification

Special core analysis techniques are often based on the assumption of uniform petrophysical properties of core samples. This assumption is obviously erroneous for heterogeneous core samples. This paper presents an integrated approach for characterizing permeability heterogeneity at the core level. The method combines the results of laboratory core flooding with already developed field scale history matching techniques to identify a 3D heterogeneous permeability field at the core scale. We focus on core samples with local heterogeneities. Their 3D porosity maps are preliminary obtained using conventional tomography techniques. Production data are measured when performing a viscous miscible displacement experiment: a high viscosity glycerin displaces a low viscosity brine at a constant flow rate. Concentration maps are collected at successive times during displacement using a medical CT device. A 1D permeability profile can be determined from the variations in the slope of the pressure signal against front location inside core samples. The 3D permeability fields are modelled using a relationship such as log(k(x)) = Aϕ(x) + b(x) where A is a constant and b(x)is a realization of a random function characterized by a mean, a variance and a variogram. We developed an assisted history-matching process, which takes advantage of all the available static and dynamic data to model permeability heterogeneity inside core samples. A first guess is initially generated honoring the measured Δp(t) variations along the flow axis. This first guess is subsequently modified until the mismatch between dynamic data and fluid flow simulated answers is minimized. To do so, we use a gradual deformation based optimization method, which allows us to vary any given distribution from one or a few deformation parameter(s) while preserving the overall spatial variability. The proposed method is fast and non-destructive. The final 3D permeability map honors the measured dynamic data and respects the spatial variability of the collected static data.