Finally another week in 4D!
Sorry for the radio silence, but nothing noteworthy has come to my attention for the last two round-ups of the week.
This week we have two 4D announcements that are not directly related to publications. The EAGE announced a workshop on Practical Reservoir Monitoring, explicitly asking for 4D submissions. The event will be located in beautiful Amsterdam on 6 – 7 of March 2017. Registration is open on November 1st, however, the Call for Abstracts will close on November 20th. Get writing and see you there!
The aim of this workshop is to investigate how reservoir monitoring processes – utilising geophysical, geomechanical and reservoir engineering data – can be practically optimised in order to realise the full recovery potential of any oil and gas field, with the emphasis on practical applications of integrated studies incorporating use of reservoir monitoring data by reservoir engineers, production engineers and others. The benefits of different approaches to incorporate geophysical and engineering data for reservoir monitoring purposes will be investigated through case studies and discussion.
ION will be responsible for optimization of five 4D projects in the North Sea. Stuart Darling, VP of ION’s Optimization sector has been quoted:
“Towed streamer 4D acquisition remains one of the most complex operations for the seismic industry. 4D surveys require precise marine current predictions to accurately match past receiver positions and focused simultaneous operations management to ensure the vessel and equipment navigates around field infrastructure in a safe, efficient manner.
The AAPG and SEG held their International Conference & Exhibition in Cancun, Mexico in September 2016. Unfortunately, I am unable to retrieve their papers, however, after a thorough analysis of the abstracts, I will give you the gist (you can really see the sarcasm drip from that statement, right?).
I love the fact that a Pyrenees paper was presented in Cancun. Those places could barely be further apart. Additionally, I am intrigued as I have just done a geological field course including play analysis in the Southern Pyrenees.
Seismic Monitoring of the Pyrenees Field: A 4-D Success Case Study
R. Meza, G. Duncan, K. Kostas, S. Kuzmin, M. Florez, T. Perret and J. Stewart
They tell a success story, where 4D analysis could identify bypassed zones of oil to accurately place infill wells. They had the common problem of gas coming out of solution and expanding the gas cap.
Hubans et al. from Total have published an overview paper:
Recent improvements for Quantitative 4-D Seismic
It seems like a nice overview paper that highlights common 4D practices like image warping, geomechanical analysis, and quantitative interpretation. Several case studies would be presented to back their claims and provide a shameless plug for their technology solutions.
Kenedy Fikeni has produced a report for their Master thesis that gives a close analysis of history matching with the use of 4D seismic data, titled:
History Matching with the use of Qualitative 4D Seismic Data Application to Norne Field
The Norne field is located in the Norwegian North Sea and operated by Statoil. The thesis itself focuses on manual history matching of qualitative 4D seismic. Quantitative 4D, as well as matching on a well level, have not been performed, as they would exceed a Master thesis. The main steps undertaken were:
- Locate seismic line in model
- Slice simulation model
- Interpret OWC
- Wiggle parameters to match OWC of model and data
Overall the model could be adjusted to better fit the qualitative data. I like how they clearly outlined the changes, the improvements and the shortcomings of this study.
The Society of Exploration Engineers has held their annual technical conference in Dubai from September 26 to 28. They have pushed a few conference papers out.
Turkish Petroleum has published the nice sounding overview:
Turning Data into Knowledge: Data-Driven Surveillance and Optimization in Mature Fields
In this strategy, they place the value of 4D seismic at 10 million barrel per day of production in 15 years. They point out however that resolution is poor and additional InSAR and microseismic data has to be used in the data mining process to obtain a good estimate of recoverable resources.
Chris Riley of Shell tells a story of success in
Perdido – A Five Year Look at the Technologies that have Enabled Ultra-Deepwater Success
The 2002 discovery in the Gulf of Mexico was once the deepest producing well worldwide. At these depths, a 4D survey is comparably cheap, compared to creating a shortcut by waterflooding. Therefore it has been employed early on and proved to be invaluable in the reservoir and especially water injection management. Riley summarizes the benefits as follows:
- Process Safety
- Out of zone injection (OOZI) monitoring
- Waterfront tracking
- Frio compaction and overburden effects due to production
- Reservoir management
- Scale squeeze timing
- GOR management
- Water breakthrough and injector rate management
- Future Field Development
- Well count
- Target Placement
- History Matching
- Business Planning
- 4D Surprises
- Reacting to any sort of unexpected 4D signal in the data
Particularly the 4D surprises made me chuckle a bit. Working with field data will obviously always have some surprises in store.
A team of ZADCO is using seismic attributes to study naturally fractured carbonates.
A New Method to Calculate Effective Permeability of a Naturally Fractured Carbonate Reservoir from Seismic Attributes
They use a binary variant of ant-tracks coupled with an isotropic fracture distribution to obtain a fracture network. This assumes purely vertical fractures. Subsequently, the effective permeability is calculated. This will be matched to the water cut and reiterated in a 10 step workflow. It seems rather simplistic but it is definitely a good place to start for an initial model.
BGP and CNPC make a thorough reservoir architecture analysis that is then used to constrain interpretation of time-lapse seismic. It is quite fascinating to see how they construct the case of bypassed oil in this one. A very real example of data integration in process:
Integrated Analysis of Reservoir Architecture and Time-Lapse Seismic for Remaining Oil Prediction: A Lithologic Reservoir in Pearl River Mouth Basin, China
A contribution from the Campinas university addresses the problem that deterministic data integration of 4D and simulation is ignoring uncertainties. Dynamic changes in pressure and water saturation are inverted for stochastically. They use probability density functions to match the inversion to simulation products.
Using a moderately complex synthetic model, they compare 500 inversion products to 500 simulation products. This comparison has four possible results for areas:
- 4D and Simulation are correct
- 4D outperforms Simulation
- Simulation outperforms 4D
- Both are wrong
I love the publication already for the proper use of statistical analysis. If you read any of the papers I present today, take this one. I hope they study this on field data as well to evaluate the method. An interesting aspect I want to mention is the possibility of “unknown unknowns” in areas of high mismatch of both data types when simultaneously considering 4D interpretation and simulation data to have given uncertainties.
Reservoir Evaluation and Engineering
Another publication of SPE has a 4D paper in store for us, this time, it’s not a conference contribution but a full journal article penned by the university of Alberta:
A Practical Methodology For Integration of 4D Seismic in Steam-Assisted-Gravity-Drainage Reservoir Characterization
Last but not least a history matching paper has been published on ArXiv:
Efficient big data assimilation through sparse representation: A 3D benchmark case study in seismic history matching
It is an extension to Luo et al (2016) that proposed the use of angle stacks instead of inversion products in the seismic history matching process. To address the big data problem in the process they chose a wavelet-based sparse inversion. While the predecessor was a proof of concept on 2D data, this paper applies the complete method on the Brugge field in 3D. The authors acknowledge that ensemble collapse is a problem in the history matching process that proves to be hard to address.
Latest posts by JesperDramsch (see all)
- Dipping your Toes – Machine Learning for Geoscientists - 2017-08-03
- Geoscience and Machine Learning – EAGE 2017 Workshop - 2017-06-13
- Lavafalls – Waterfalls are so last year! - 2017-02-13