+What kind of alerts are possible?
Sentinel RT estimates beliefs for kicks, lost circulation, washout, pump failure, stick slip, whirl, bit bounce, bit balling, overpull and underpull. Alerts may be emailed, texted or embedded in a display monitor. Typically, emailed alerts are more detailed than texts or display messages and may include descriptive charts to help the user better understand why the alert was generated.
+Who sets the thresholds for the alerting system?
Intellicess usually sets belief thresholds for alert disseminations based on end-user preferences. Intellicess can work with users’ existing datasets to determine thresholds, which will minimize the occurrence of false and missed alarms. Ideally, all alerts include information indicating why the alert was generated and suggesting potential corrective action.
+Can an A.I. system be trusted?
Many software providers treat their A.I. applications with their customers like a mysterious black box that can’t be articulated. At Intellicess we have explainable A.I. and processes in place to sequentially introduce the numerous beliefs and capabilities within SentinelRT® to your organization. Our sophisticated, yet explainable, technology allows us to instill confidence and gain the trust of your team so they can utilize SentinelRT® for optimized decision making.
+When does SentinelRT learn?
SentinelRT® continually learns certain key parameters during run-time. However, the overall belief system is already highly developed and is not modified in real-time. When an end user identifies instances where SentinelRT®‘s belief system could be improved, the specific machine learning happens offline in a controlled and scrutinized manner under the supervision of our domain experts. This ensures that SentinelRT® belief system remains robust and highly intelligent.
+Do we need to provide historical data sets to train Sentinel RT before using it?
The genesis of SentinelRT® goes back over a decade. Over that time, tens of thousands of data sets have been used to refine and continually improve the belief system. Further, although incredibly sophisticated, SentinelRT® is ready to use out of the box.
+Is SentinelRT truly an A.I. system?
True AI is a very debatable topic. We believe Sentinel RT is an A.I. system due to fact that it mimics the human decision making process. Humans make decisions based on their beliefs. SentinelRT® acts similarly, and generates beliefs using physics and data based models. SentinelRT® keeps learning and adjusting its belief over time.
+Is the SentinelRT engine a black box?
No, It is based on published physics and well-documented data based models. Please click here for publications that will help understand what is inside SentinelRT®.
Back End Engine
+How much validation of the back end engine has occurred?
Below are some statistics on the current usage of our back end engine, SentinelRT®.
Time in commercial use: Jun 1, 2015 to current
Number of wells processed: 1,500
Drilled footage: 22.5 million feet
Number of unique rig deployments as an edge solution: 45
During this period, uptime for the backend engine was 100%. The software experienced no crashes caused by missing or bad data. Because the software uses a variety of physics-based models and pattern recognition algorithms, and a host of usable output, not all output can be validated equally. The validation of the output is performed to customer-set specifications on the customer’s own data sets, usually before committing to purchase the software.
+What is meant by “Real-Time”?
The backend engine is currently optimized for 1-sec and 10-sec data feeds. The analysis of the data takes less than 300 ms per data row on a computer having a 2.4 GHz processor speed and 8 GB of RAM. This includes running the models for data cleansing, rig state detection, unplanned event detection, drilling advisory, hydraulics monitoring, torque and drag estimation and cuttings transport. Disabling unwanted modules can result in faster run times. It is possible, with some ingenuity, to run the backend engine on a Raspberry PI.
+On what software platforms can the back end engine run?
The backend engine itself is built in Java. Because Java uses an operating system-independent virtual machine for its compiled code, the backend engine can run on Windows, Linux and Mac platforms.
+What is Cone Drilling™?
For real-time analysis to be useful, it must be used to provide actionable steps to drillers at the rig site. Intellicess has devised an innovative method to provide actionable suggestions to the driller using a cone graphic imposed on a WOB-RPM chart. The cone shows the driller the path towards better drilling.
The visual also provides an indication of what dysfunction the driller is moving away from, and thereby teaches better drilling practices.
+How does the software predict Cone direction?
The direction of the cone is dependent on the dysfunction detected by the software through pattern recognition algorithms. Based on the software’s belief in the various dysfunctions, it calculates the cone’s angle and radius.
+Why type of consulting does Intellicess offer?
The engineers at Intellicess have a varied history in data analysis. We understand techniques to reduce Non Productive Time (NPT) and Invisible Lost Time (ILT) that is responsible for as much as 50% additional costs during drilling. We also understand techniques to use drilling data to engineer completions and maximize long term production. We can train you on these techniques and work with you to create a game plan to reduce costs in your organization over the long term. Our goal is to make you self-sufficient in drilling and completing the well both safely and efficiently.
+What sensor data are validated?
Through use of the Bayesian network model, we validate the following sensor data: hook load, standpipe pressure, flow in, flow out, total mud volume, block position, torque and RPM.
+How is data cleansed?
The backend engine can detect and replace missing data, outliers, or biased (drifting) data with model-predicted values. Identification of missing data and outliers is elementary and is done first. We then use a physics-based Bayesian network model to validate the sensor values against the model-predicted values. When we determine drift or bias to be above a certain threshold, the model-predicted value may be used to replace the sensed value.
Some key points to note about data cleansing:
• The raw data are never discarded.
• The model-predicted value is an additional channel that may be used only carefully and when appropriate.
• The model used changes with the rig states and the model-predicted values are not always correct but overall the data do become cleaner.
Our publications provide more details on technique used.
+Is there a service component to the ILT analysis?
Intellicess provides suggestions and sets up customers systems for automated report generation. This process involves discussions of the customer’s needs and configuring reports to reflect customer preferences. We are currently working to make report setup and configuration easier.
+Can the generation of ILT reports be fully automated?
Generation of these reports can be fully automated and is enabled primarily by the layering of the rig state detection algorithm on top of the data cleansing algorithm. The backend engine detects rig state with extreme accuracy and various KPIs, such as S2S time, W2W times, and slide vs. rotary drilling times, can be automatically calculated. We have automated the process of generating pdf reports containing such information without the human involvement typically required for certain steps, such as data cleansing.
+Why do we focus on the back end engine and not the full software?
We understand that all operators, rig contractors and oilfield services companies have their own history of unique operations and cultures. The idea of creating a single standardized interface / workflow to enable action that fits all companies is inevitably counterproductive because all of our clients are different. We have seen many well-meaning projects fail because human factors were ignored, or a one-size-fits-all approach was leveraged. Our approach of working within our customer’s existing ecosystems and slowly changing the culture through improved drilling operations has a much higher probability of long-term success. Therefore, rather than provide a full system, we remain flexible and work to integrate and continually improve the backend engine within our customer’s distinct environment.
+Who builds the frontend UI required by the backend engine?
Often, our customers may already have a frontend user interface (UI) they’ve developed in-house and it is primarily a matter of modifying it to expose the analyses performed by our backend engine, SentinelRT®. We also have strong partnerships in place with industry leading visualization and UI firms to which we can introduce you. In either scenario, we can evaluate your needs and collaboratively develop a plan for integration with SentinelRT®.
+Who integrates the back end engine?
Intellicess can provide the software and technical support to integrate the backend engine into any system. For smaller projects, Intellicess can provide this service for free. When the scope of the project is more extensive, a fee will be involved.
+What core technologies are used in the product?
We use numerous technologies and techniques for the various features available in the backend engine. This includes four pending patent applications, one of which was licensed from the University of Texas at Austin. We have published extensively about the technologies we employ and a select list of those publications can be found in the publications page.