When you hear about digital transformation in oil and gas, you probably think words like “complex” or “slow” — but Bluware is already charting new territory in E&P. Our CEO, Dan Piette, recently spoke on cloud, artificial intelligence, and more at the D2 Upheaval Energy Disruption Conference in Houston, sponsored by Tudor, Pickering, Holt & Co. Here are the highlights from his talk in this video.
See Dan’s full talk courtesy of Tudor, Pickering, Holt & Co.’s Vimeo page.
Technology drives us forward. It enables things we never thought possible. It has a massive impact on how we work and gives us access to a never-ending stream of information 24 hours a day. So why is it that the rest of the world moves at incredible speed, reaping the benefits from artificial intelligence, while upstream oil and gas is seemingly stuck on problems like “our data is too big.” It’s like trying to lift an elephant into the air – it’s not impossible, but definitely not easy.
Let’s look at actual examples of how we, as individuals, have adopted advanced technology that we now take for granted in our everyday life.
Everyone Uses the Cloud Every Day
Google Maps, E-mail, WhatsApp, Waze – these are all apps and functionality we use to find our way or communicate with clients, friends and colleagues.
Everyone Uses Artificial Intelligence (AI) Every Day
“Siri, is it going to rain today?” Asking your virtual iPhone assistant about the weather, telling Alexa to order you pizza, or having Google Assistant to share what’s on your calendar for today are all examples of AI.
Everyone Uses Machine Learning (ML) Every Day
If you like to relax with a Netflix show, you have surely received tons of recommendations from Netflix on what to watch next. If you’re on Facebook, you have probably discovered an old friend from school you haven’t seen in twenty years through Facebook Friends. Have you thought about how this is actually an example of machine learning?
Most of us Use Deep Learning (DL) Every Day
In the early days of deep learning, training the machine to know the difference between cats and dog was very impressive. Now, the machine can even tell the difference between you and your brother, your wife and your mother-in-law.
Where is upstream oil and gas in this scenario? At the forefront of innovation and digital transformation? NOPE! We use “data”. We process “algorithms”. We “interpret” seismic data. The volume of seismic data, the sheer size of the files associated with seismic interpretation, is a massive barrier to the use of the cloud.
Why is This?
Well, first of all it is expensive to go to the cloud. Let’s check some math examples.
– $22k/mo/PByte S3 (active) storage
– $4k/mo/PByte Glacier (cold) storage
Secondly, it is slow to go to the cloud.
– Data movement isn’t simple – office gigabit lines allow for at most 10 TByte per day
– Getting seismic data, which can easily blow up to a petabyte (PB) size (1024 terabytes), is almost impossible online
Third, it is cumbersome to get data out of the cloud into a format your legacy applications can consume; moving data to a Petrel project from the cloud drops through performance to about 64 Mbps – slower than most PCs.
This is probably why according to Gartner, 86% of O&G companies have digitalization efforts underway, but only 10% are scaling them, or are harvesting benefits.
Bluware Technology Making the Impossible, Possible.
We’ve developed a cloud-native, object storage architecture for seismic data, and indeed, all subsurface data. All this data is available via transcoding directly to legacy applications such as Kingdom, Decision Space, and Petrel at least 10 Gbps – over 100 x faster.
We look at data the way you want to look at data – compressed appropriately (lossless, virtually lossless, lossy – but with no decrease in dynamic range or information content), streaming, true random access (which is not possible with SEG-Y files), lossless compression in “cold storage” at 40:1 compression for interpretation, and various options in between for machine learning and deep learning.
Furthermore, the data is actually ML/DL ready – up to now most companies applying these techniques to seismic data would need about 95% of the time to condition the data, and only 5% of the time to apply the ML/DL process, which allows for many more scenarios to be evaluation. This means that once the data gets to the cloud, it is immediately useful for any number of data analysis techniques.
Instead of creating XYZ feature in a closed application, we have focused on developing a completely open platform, with self-propagating user defined data types.
Why Should You Care?
1. You are able to get seismic data in the cloud – This is not an end in itself, but it will help save costs and eliminate the need for hardware upgrades and purchases. No expensive infrastructure burden to the organization.
2. True ML/DL capabilities – We’ve seen 12,000 sq km interpreted in 20 minutes to pick top and bottom of salt. 6,000 sq km that picked 513 faults and 90 horizons in 1 hour.
3. The ability to run thousands of interpretations and models and evaluate them stochastically. This brings a much higher confidence level to all interpretations. It means that you no longer have to rely on one interpreter to understand what is going on subsurface.
4. All this means a much faster cycle time, decreasing time to first oil, with an inherently lower risk. Would you rather have 50 geoscientists looking at 1 PB, or 1 data scientist looking at 50 PB?
We all use the cloud every day. At Bluware let you use the cloud for the most advanced exploration possible as well.