Real time flow profiling in hydrocarbon wells

Articles
Prad Thiruvenkatanathan, June 11 2020

With the world gravitating towards cleaner energy, the importance of improving production efficiency within the oil and gas sector has never been more important. Enhancing efficiency in hydrocarbon production operations however, does not come easy – it requires not just an adoption of improved production technologies, but also embedding new advanced surveillance technologies that provide the much needed visibility on flow dynamics downhole that enable better decisions to be taken at the right time. Today’s production logging solutions are run through intervention operations on an ad-hoc basis. Whilst they provide good quality insights into production dynamics in the well, they are just snapshots in time in what is a complex and often dynamically varying flow environments. As a consequence, operators frequently are compelled to rely on estimates and extrapolation, rather than robust and up to date data.

But LYTT is changing all that. Our real-time data analytics technology now offers the ability to monitor production operations 24X7 using non-intrusive fibre optic cables that are embedded within the completion string. We do this through the use of our Cloud native Real time pattern recognition software that embeds LYTT’s proprietary hybrid analytics models within stream and batch processing microservice infrastructures to transform large volumes of streaming Distributed Acoustic Sensing (DAS) and Distributed Temperature Sensing (DTS) data into unique and actionable insights – to enable real time visualisation and logging of flow dynamics downhole. The outputs flow profiles answer three fundamental questions in real time:

  • Where the fluids are flowing from
  • What the fluid types are
  • How much of each fluid type do we see flowing from each production zone

In fact, one of our major oil and gas clients has now been able to use the software solution to perform the first ever fibre data informed water shut off operation adding several thousand barrels of oil a day of daily production increments .

To explain more about our technology, plus some of the challenges we face, the LYTT team has created three webinars, each exploring the world of inflow profiling.

Webinar 1: Cracking the code - 24/7 real-time multiphase flow profiling capability

In the first session, our co-founder, Prad, was joined by a lead Petroleum engineer from bp, Teymur Sadigov, to discuss the journey towards the development and delivery of our real time flow monitoring technology deployment to bp operations. Teymur and Prad also provide a brief summary of some of the exciting field results from the use of the technology to implement the industry’s first ever water shut off operation purely informed by data insights from fibre based surveillance.

LYTT’s offer: Integrated end-to-end cloud native software for pattern recognition and event classification of DFOS data

Webinar 2: How LYTT makes it real - delivering real-time subsurface analytics with a robust, scalable system

Our second session was hosted by LYTT data scientist James Ramsay and cloud service engineer Vishnu Ramalingam who walked through the technology infrastructure that we use to create our real-time inflow profiling tool.

The scale of the challenge was large: how to shift from PLs that record snapshot data every couple of days to a real-time system that can sift through terabytes of data to deliver production insights in a matter of seconds. And all that via a robust platform that can be easily scaled up and maintained. James and Vishnu demonstrated how LYTT made that possible.


LYTT’s real-time system architecture

 

Webinar 3: Turning on the LYTT – real-time flow allocation with hybrid analytics.

Our third and final session was presented by LYTT data scientist Alessandro Delfino and operate and sustain engineer Cagri Cerrahoglu. Together, they took participants behind the scenes to demonstrate how our unique blend of first principles modelling and data science skills has helped us create a portfolio of real-time analytics products.

As well as demonstrating how our workflows support our “hybrid” approach, they shared a live example of how temperature and acoustic data reveals a fluid’s acoustic fingerprint as it moves from the reservoir into the well.

All three webinars threw up really interesting questions from our audience. Here’s a flavour of what we were asked and our replies:

How do you validate the results?

Accuracy is key to what we do. Last year, we had the opportunity to run some blind tests with one of our clients. We acquired DAS (acoustic) and DTS (temperature) data during a scheduled PL run. The client gave us the DAS and DTS data but held back the PL. We then ran the DAS/DTS data through our algorithms and only compared with the PL after we got our results. We were thrilled to get a great match.

Do you store the raw acoustic data?

DAS technology turns fiber optic cable into a long microphone array, which generates thousands of acoustic data traces a second at different depths along the length of the fibre. As you can imagine, this creates huge amounts of data –the equivalent of downloading thousands of Netflix films every hour. Our unique feature extraction algorithms extract the most relevant strands of information needed to generate flow logs at site, enabling a reduction in data volumes by a massive 1000X without losing information – the process we call “intelligent extraction”. This would mean we don’t necessaril;y need to store the raw data for processing and also ensures a scalable digital architecture for insight generation in real time. We could however, also remotely store the raw data for certain depth-time windows and trigger this remotely if a client specifically requested for this

How important is dynamic measurement for well optimisation?

Many of today’s productive reservoirs comprise of highly complex reservoir structures. This would imply that mapping changes in well flow dynamics is key to making good decisions for field optimisation. This challenge has been a key area of focus in LYTT and has been one of the prime drivers for the development and delivery of our real time dynamic flow profiling capability.

What if I only have DTS data?

Our models work best when used with both DAS and DTS data – this gives us the richest information set and help set better constraints to the predictions. However, we do have a tool that allows us to switch between DAS or DTS to record data. In essence, we drive using one form of data and then complement the result using the other.

How closely must your flow loop simulate a specific completion type in order to provide effective acoustic fingerprint information?

We don’t just look at the pattern of fluids under different flow conditions, we try and understand why we see what we see so that we can then calibrate for different well scenarios. We’ve gone a long way with our sand monitoring technologyin understanding the physics behind the acoustic signatures and are beginning to get a clearer understanding for inflow profiling, but it is something we continue to work on. This helps better calibrate the variabilities between lab and field environments and improve the accuracy of our predictions.

Does your algorithm require other open hole petrophysical measurement inputs, such as porosity / permeability?

No, our models only use fiber optic cable to generate either DAS or DTS data..

Do you use different predictors depending on location of the fiber optic cables in the well?

Yes, we use different algorithms / apply calibrations depending on whether its permanent or intervention cable and from one completion type to another.

Does the technology work on horizontal wells?

The technology has validated to work in deviated wells too.

What is the resolution of your inflow product? Can your analysis detect inflow from thin sands?

This would very much depend on the spatial resolution of the instrument being used for acquisition. Typical spatial resolutions to achive reasonable signal to noise performance in conventional wells that have measured depths of about 5 to 10km would range between 3 and 10 m, whilst the DTS resolutions can be better.

Visit our webinar page to watch all three sessions. And if you have any questions or would like to find out more about how our tools and skills can help you, please do get in touch.

x
This website uses cookies. Click here to learn more. That's Fine