Waterflooding Surveillance and Monitoring: Putting Principles Into Practice

Terrado, Rodolfo Martin; Yudono, Suryo; Thakur, Ganesh

doi:10.2118/102200-ms

Cited by 16 publications

(1 citation statement)

References 7 publications

(5 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In recent years, the oil and gas industry began to be aware that the collection of large amounts of data coming from both gauges and specific events allows engineers, by applying artificial intelligence algorithms, to better manage the fields and reduce the maintenance cost. For example, predictive analysis of the reservoir surveillance data (e.g., production rate, following pressure) can help to identify in advance some possible failure of equipment and to intervene before an event occurs [1].…”

Section: Introductionmentioning

confidence: 99%

Predicting the Surveillance Data in a Low-Permeability Carbonate Reservoir with the Machine-Learning Tree Boosting Method and the Time-Segmented Feature Extraction

Zhao

Xiao

2020

Energies

View full text Add to dashboard Cite

Predictive analysis of the reservoir surveillance data is crucial for the high-efficiency management of oil and gas reservoirs. Here we introduce a new approach to reservoir surveillance that uses the machine learning tree boosting method to forecast production data. In this method, the prediction target is the decline rate of oil production at a given time for one well in the low-permeability carbonate reservoir. The input data to train the model includes reservoir production data (e.g., oil rate, water cut, gas oil ratio (GOR)) and reservoir operation data (e.g., history of choke size and shut-down activity) of 91 producers in this reservoir for the last 20 years. The tree boosting algorithm aims to quantitatively uncover the complicated hidden patterns between the target prediction parameter and other monitored data of a high variety, through state-of-the-art automatic classification and multiple linear regression algorithms. We also introduce a segmentation technique that divides the multivariate time-series production and operation data into a sequence of discrete segments. This feature extraction technique can transfer key features, based on expert knowledge derived from the in-reservoir surveillance, into a data form that is suitable for the machine learning algorithm. Compared with traditional methods, the approach proposed in this article can handle surveillance data in a multivariate time-series form with different strengths of internal correlation. It also provides capabilities for data obtained in multiple wells, measured from multiple sources, as well as of multiple attributes. Our application results indicate that this approach is quite promising in capturing the complicated patterns between the target variable and several other explanatory variables, and thus in predicting the daily oil production rate.

show abstract

Section: Introductionmentioning

confidence: 99%

Predicting the Surveillance Data in a Low-Permeability Carbonate Reservoir with the Machine-Learning Tree Boosting Method and the Time-Segmented Feature Extraction

Zhao

Xiao

2020

Energies

View full text Add to dashboard Cite

show abstract

References and Bibliography

2019

Economically and Environmentally Sustainable Enhanced Oil Recovery

View full text Add to dashboard Cite

Streamline-Based Integral Modeling for Waterflooding Design Optimization, Surveillance and Monitoring

Galacho¹,

Vazquez²,

Laiz³

et al. 2007

Latin American &Amp; Caribbean Petroleum Engineering Conference

View full text Add to dashboard Cite

The Dynamic Numerical Simulation in Flowlines (DNSFL) is an alternative tool adapted to handle Dynamic Models in Fine Scale. This feature has been particularly relevant for studying the current case, a huge multilayered waterflooding process developed in a giant field of great extension reservoirs with considerable facial and stratigraphical variations. The DNSLF develops these tasks suitably because uncouples reservoir geometry and heterogeneity from transport equations, solving the problems dominated by convective flows in a faster and computationally more efficient way. This allows to build models of greater space discretization and, therefore, to represent better the heterogeneity of the reservoirs. The analyzed field is constituted by fluvio-lacustrine deposits, nine sand-clay cycles of normal grading (only eight of them were modeled), partially connected reservoirs; with 250 actives wells in commingled production and water injection; and with a long and detailed history of simultaneous primary and secondary events. In a previous paper (SPE 94815) a Streamline-based Global History Matching of this field was presented. This process enabled to achieve the Geological Modeling Calibration, a clear conceptualization of the current primary and secondary production mechanisms, its productive behavior, and to evaluate the geostatistical and Upscaling procedures to apply for the definition of the Simulation Model. This paper illustrates how the Integral Model achieved, with a detailed Streamline-based History Matching, is used for Waterflooding Design Optimization, Surveillance and Monitoring, showing that these principles are key factors to understanding reservoir performance and identifying opportunities that will improve the ultimate recovery. During the detailed History Matching process CPU runtimes around 200 minutes were achieved using a 1225000 grid cells Model, with 190 timesteps, quarterly at the starting, and monthly after, based in a Pentium 4 PC, 3.2GHz CPU and 2GB RAM. It showed that it is possible to work with a big Streamline-based Model in relatively short processing time. Introduction The classical techniques of waterflooding surveillance, including methods like: mapping (gas-oil ratio, water cut, pressure, etc.) total liquid production vs. time, injected poral volumes vs. recovery factor, Hall plots, etc., jointly with the monitoring process have backed-up successful Exploitation Optimization processes[1]. Most of these techniques, as those mentioned above, are 0D; some recent practices made 1D or 2D spatial distribution of the relevant characteristics to the process, essentially productions and injections, but without integrating the pressure fields. Streamline simulation goes further and incorporates pressure fields, which determine streamline as the most probable way of fluid movement. In this sense, and always in the pressure field, there are two well differentiated work levels in the waterflooding projects[2] (Fig. 1):The first level, called "Production-injection surveillance" based on the historic analysis of the waterflooding, allows for the examination of injectors and producers, identifying well and poorly swept areas.The second level, "Streamline-based simulation" based on a detailed geological model, fluid physics and history matching, which allows, in addition to the previous one, for an integral redesign and forecast of the waterflooding towards its optimization. Both levels are sustained and fed by monitoring the field process; as it is clear, the development of the second level, as in the example of Puesto Hernández Rayoso Field- Block 4, implies achieving the first level.

show abstract

Waterflooding Surveillance and Monitoring: Putting Principles Into Practice

Cited by 16 publications

References 7 publications

Predicting the Surveillance Data in a Low-Permeability Carbonate Reservoir with the Machine-Learning Tree Boosting Method and the Time-Segmented Feature Extraction

Predicting the Surveillance Data in a Low-Permeability Carbonate Reservoir with the Machine-Learning Tree Boosting Method and the Time-Segmented Feature Extraction

References and Bibliography

Streamline-Based Integral Modeling for Waterflooding Design Optimization, Surveillance and Monitoring

Contact Info

Product

Resources

About