Hit one operation problem at a time, in the head.

Many industries have several players involved in a certain operation. Oil, Telecom, Banking, Manufacturing: one player is the owner of operations, and then you have one or many contractor performing tasks, consultants and other vendors supporting the operation. Each player might have many challenges to tackle - and many problems to solve. If you want to improve the outcome of a certain operation, what should you attack first?

In this article, I pick the Oil &Gas Industry, where a certain company is drilling for new wells, using rigs and multiple services from vendors. Your goal is to complete the drilling as soon as possible, at the lowest cost. My proposition is that you might need to solve one, two or more PLC/automation, data connection and aggregation, SCADA, ERP and IIoT (Industrial Internet of Things) or AI(assisted decision making, Machine Learning, Deep Learning, ) challenges at once to get this drilling operation to be best ever!

Yes, why not? We are in a world where computing power is incredibly cheap and potent. Tools and knowledge are abundant. Why not create a well of the future? RIght now.

Well, its tempting, but resources are scarce and change management says you should focus on small improvements instead of " big bangs" . My take: focus on one need of the Operator - the company that wants the oil &gas taken out for a profit.

For example: Is it possible to replicate best practices (your best performance ever, with a drilling rig, past or present, in any location), so that you can boost your team’s capacity, as if they gained super-intelligence and super-experience?

Humans cannot deal effectively with multiple real-time inputs to make the best decisions instantly. However, machines can ponder multiple scenarios in milisseconds, weighing alternatives and risks. It has been a while since Deep Blue beat Kasparov , and today you can have a chess-engine , running on your phone, showing you the best possible move at each moment. Artificial Intelligence applied in the here-and-now, was simplified by developers, so you dont have to study programming - just play the game.

How does AI apply to Oil Drilling?

As the drilling job progresses, the moves you can make are several: to increase or reduce weight, pressure, speed, or even stop and replace a drilling bit. How would that be assisted by a computer mind? One way is to build a model. In this case presented, a cause-and-effect network (Bayesian Network) building a belief system. This is your situational evaluation support. Based on data, the trends and consequences of certain parameters (i.e.pressure, depth, vibration, etc), the system might say: "We believe there is a 92% chance of slippage happening in 10 minutes from now" (example).

We wanted to explain more about it to the audience. So we did.

Darcy Partners innovation workshops and how we got the word out

In the past two weeks we packed a small suitcase, hopped on a plane, and joined our partner Intellicess for workshops in Midland,Texas and Calgary,Alberta(Can). The workshops represent an opportunity for innovation leaders from Oil & Gas companies to scout and validate breakthrough technologies. Sentinel RT™ and FactoryStudio were shown combined as a complete solution to potential new users.

To them we explained how a complete package, from drill rig data aggregation and SCADA, up to the guidance and advisory systems (powered by AI and Sentinel's beliefs model) with intuitive visualization can improve performance. Not uncommon, operators found gains with double digits. 

"How fast and when should I increase the RPM? Right now or in a few minutes?"

"Should we reduce or augment the WOB?"

Some of your professionals can "feel it" and do a wonderful job in monitoring the drilling at each moment. Some might struggle at times. 

It is indeed possible to replicate best practices, if you can capture the data, trends, put them into context and have a good Bayesian model. That is one operational problem, but it triggers several automation, data and network problems. Or challenges, shall we say.

-         Install or improving their systems to collect real-time data from the drilling operations;

-         Establish an interface at the rig that also forwards data to the central office;

-         Run the analytics, comparing information between wells, with a good model;

-         Enable their team to use it in real-time, to make decisions with reliable data.

One Operator might say: "I already have an EDR, so I dont need a platform for more data capture." Well, most if not all EDRs (Electronic Drilling Recorder) have many common limitations:

1. Daily Cost to acquire data, plus cost to store data on server
2. Set Design, No Flexibility
3. 1-10 second data
4. Fixed storage options and closed platform
5. Lack of real-time analytics integration

Also most EDR are not an open and extensible application. They do not allow for major changes in user displays, real-time calculations, and storage and analytics integrations are not open and not customizable. Failure to show the data’s true variation and data-time signify problems for data consistency and reliability. 

To enhance the legacy EDR data collection methods, Tatsoft provides an additional data feed from the rig PLC and control systems, capable of 10hz data output in real time. Tatsoft RM supports data acquisition from WITS0, WITSML, Omron PLC’s, ControlLogix PLC, and various other field controls systems and sensors.

On the data aggregation and automation side, Tatsoft’s SCADA platform is state of the art. And combined with Intelicess Sentinel RT, we are providing operators a complete and economically viable data aggregation, advisory and automation solution.

So you solve the data problem with infra-structure provided by Tatsoft, and you separate the other challenge that is the analytics.

Without worrying on aggregation, the ANALYTICS solution can focus on the algorithms engines itself. It can be faster and more precise, not having to deal with the peripheral tasks of data gathering, archiving or visualization.  

The data analytics tools we are talking about, integrated with Tatsoft RM, is Intellicess Sentinel RT™

Sentinel RT™ is an A.I. backend engine able to perform real-time streaming-data analysis at the rig or in the office. Sentinel RT™ has a highly sophisticated data-cleansing algorithm and uses a Bayesian network to differentiate between sensor and process faults.

Tatsoft RM has built-in displays and logic, to allow integration with Sentinel RT™, to add the following functionality:

• Intuitive and customizable Cone Drilling™ advisory
• Advanced rig state detection
• Real time torque and drag analysis
• Real time event detection

1. Washout in drill string and/or pump failure
2. Kick & lost circulation
3. Drilling dysfunctions (stick slip, whirl, bit balling, bit bounce, etc.)
4. Overpull and underpull

• Generate and email / text alerts based on events detected
• Consume contextual data from software like OpenWells®, WellView®, etc. to be provided as input data for Sentinel RT™.

The website for the combined offering is www.rigmonitoring.com

We invite you to check it out! We will happily address your questions here or in direct messages to the author. Suggestions on how our communications could be improved are also welcome.