You are Home   »   News   »   View Article

Intelie - analytics with major drilling company

Tuesday, September 28, 2021

Intelie, a drilling analytics company, developed an interesting way to work with an offshore drilling company client - providing partly built models which could then be finished off.

Analytics company Intelie, which specializes in drilling, has developed an innovative way to work together with drilling company customers - providing partly built models which can then be extended.

Intelie has a main office in Rio de Janeiro, and offices in Houston, London, São Paulo, and Dubai.

The offshore drilling company (whose name cannot be mentioned in this article) had been looking for ways to use data to optimise maintenance, and then to improve equipment uptime. The work started in 2016, with the first 2 years spent mainly finding better ways to acquire the data on shore.

By the end of 2019, the company realised that it could also use the data to better manage emissions, including to find better ways to understand how it generates power and emissions being produced. So, it developed a parallel project with Intelie to look at that.

The project covered data from all the equipment involved with drilling, including the blow out preventer on the seabed, and the drilling package, such as mud systems, draw works, control rooms, derricks, pipe handling systems, rotary tables, power tongs. There is data from engines and the dynamic positioning system, as well as cranes, power generation and electrical distribution.

Drilling operations have gathered drill bit data for many years (logging while drilling / measurement while drilling), although that data is generally handled directly by oilfield service companies, rather than being handled by the drilling operator.

The equipment already had sensors fitted on them, but the data processing and visualizing was designed for the purposes of the operators of the equipment, not for analytics. So, analogous to a car dashboard, which is designed to tell you what you need to know to drive a car, including to be alerted about problems, but not to (for example) monitor long term trends in engine performance.

To do this, the drilling company needed to get access to the data in the back of that system.

So, there was a process of aggregating data offshore, compressing it, and consolidating it on shore.

It ends up as an enormous amount of data, perhaps thousands of sensor points every second, from the fleet of vessels being monitored in this way.


The problem was integrating the data together, or 'modelling' it, in a way which could provide the drilling company staff with the insights they wanted.

A common approach with analytics is to take lots of data and try to model it, so a data first modelling approach.

But the drilling company took the approach that, since it already knew what sort of things it wanted to find in the data, it only needed data models to be built to look for them automatically. For example, when staff already know what the common 'failure modes' of equipment are, they just want automatic ways of looking through the data for them.

Intelie was chosen because it was happy to let the drilling company maintain ownership over its domain knowledge where many data analytics companies wanted to own the algorithms they created.

Intelie also had building block tools, which could be defined as building blocks or partly built models, that the drilling company could use to make the full models themselves.

Intelie had some staff sitting in the drilling company's offices, so they could learn how the business worked, and also teach the drilling company about analytics. And it had other staff in its main office in Rio de Janeiro.

A drilling company's approach to data is very different to how a large scale manufacturing company might approach data. A drilling company has a diverse range of equipment, with a few years of data from each piece of equipment. A large scale manufacturing company might have data from hundreds of thousands of products. This means that the conventional machine learning approach, which is designed for enormous data sets, may not be applicable to drilling.

Maintenance and emissions

The first objective for the drilling company was to use the data to improve how it does maintenance.

A big target from the project is learning about the best time to do maintenance, getting a better prediction of the lifetime of equipment, and being able to adjust the maintenance interval of equipment used more or less than expected.

The data can give advice if there is an anomaly in some equipment, so you can stop and investigate at a time which does not impact operations.

The second objective was to better understand emissions.

Drilling operators did not consider emissions and fuel consumption much in the past, because they did not see that they could do drilling without using fuel, and so emitting.

But a surprising outcome of the analysis was that one of the biggest sources of emissions in offshore drilling is from fuel used to power the dynamic positioning system. A DP system is not something offshore drillers can manage without. But this is still a useful insight to have.

The amount of data needed for studies on how to optimise and reduce the carbon footprint is small compared to the amount of data being used to optimise maintenance.

Much of the equipment related to emissions is made by different manufacturers, and reconciling it was a big challenge.

Project challenges

The toughest problems with the project, from Intelie's perspective, were handling sometimes unstable data streams, prioritising the work, and developing tools which would make the work easier, says Ricardo Clemente, co-founder and co-CEO with Intelie.

The operational world 'has complicated data coming in, he says. 'Links go down, sensors are replaced, the mapping is not 100 per cent, you have different equipment, hundreds and thousands of data points to manage.'

In any analytics project, 80 per cent of the work goes on data engineering preparing the data for the analytics, with 20 per cent of the work doing the models.

But once you have built a data platform to handle the data, and a platform to do analytics with it, new opportunities to get insights from the data may emerge. 'You see a problem that was not on the radar at first,' he says. 'You have a long tail [of opportunities] that emerges.'

Intelie has developed what it calls an 'operational AI platform,' called Intelie LIVE, a software system designed specifically for handling operational data and doing analytics on it.

The operational AI platform uses a mixture of machine learning models, rule based 'if / else' logic models, theoretical physics models, observed / empirical models, and signal processing, looking for trends in the data.

The drilling industry has many specific examples of how equipment fails. These failure modes can be programmed into the system as rules, so they can be searched for in the data.

The platform has been designed to be flexible, with a number of 'solution bricks' - pre-built components such as engineering models, visualization tools, and other 'widgets' which customers can use.

These widgets are 'like Lego blocks,' he said. You can literally break down the pieces and build something new on top. If you don't like the spaceship, you can put the pieces back together differently.

Intelie also has its own query language 'Intelie Pipes,' which customers can use to build their own models, working with data from time series sensors.

It has an event processor which can aggregate and normalize real time data from a range of sources.

It provides prebuilt functions, such as a rig state detection macro, and a WITSML connector, to gather drilling data provided in WITSML standard format.

Customers can edit these existing functions as if they are editing 'Excel macros.' As an analogy from the operational world to the finance world 'They [customers] can say, 'I didn't like the way he calculates NPV, I want to change the discount rate from this to that.''

Customers have a choice of using Intelie's tools, or developing their own tools, and then owning the IP themselves, he says.

Business model

Intelie's business model is to provide its platform on a subscription basis, and then charge additional fees for its pre-built apps, such as its hydraulics model and a rig performance solution.

Typically, customers will use a lot of the pre-built apps, particularly when they are smaller and don't have the budget or capability to build in house.

'Big operators have a lot of engineering, and R+D capacity. They may say, 'I think my performance model is better, and I don't want to use yours. I want you to code my model. I want to make this production-ready, to be able to handle hundreds of thousands of measurements a day. But you can't share that with anyone else.'' 'Our business model recognized that, it gives customers that flexibility.'

Intelie has been working with Petrobras since 2011, and the software is now used on all its drilling operations. It also works with several other majors, like BP, and drilling operators.

The company was acquired by Houston oil and gas satellite communications and networking company RigNet in January 2018.

Associated Companies
» Intelie
comments powered by Disqus


To attend our free events, receive our newsletter, and receive the free colour Digital Energy Journal.


Latest Edition Jul-Aug 2021
Jul 2021

Download latest and back issues


Learn more about supporting Digital Energy Journal