You are Home   »   News   »   View Article

Peter Breunig Joins Maana Advisory Board

Friday, December 11, 2015

Palo Alto based start-up Maana, which aims to revolutionise the way oil and gas companies work with data, has appointed Peter Breunig, former general manager of Technology Management at Chevron, to chair its oil and gas customer advisory board.

Maana, a big data start-up company aiming to revolutionise the way oil and gas companies gain insights from their technical datasets across multiple data silos, has appointed Peter Breunig, formerly with Chevron, to chair its oil and gas customer advisory board.

Mr Breunig is a former general manager at Chevron, having lead Technology Management and Architecture, Technical Computing, and Seismic Imaging Research for the company.

'Maana's search and discovery platform is unique and poised to change the way upstream oil and gas companies leverage their big data,' Mr Breunig says.

'Increased use of sensors, coupled with the industry's increasing success at collecting data is driving wider acceptance of data analytics tools that promote better decision making, greater reliability and improved safety, and uncover affordable ways to find new reserves, predict failures and drive cost savings,' he says.

'Maana's oil and gas advisory board will look at how oil and gas companies can use Maana to get more value from their various technical datasets,' says CEO Babur Ozden.

Maana's software aims to help oil and gas companies use their data to answer questions such as which completions techniques yield the fewest production failures, which rigs and suppliers are most efficient, and where are the best opportunities for well workovers.

More From Your Existing Data

The core aim is to help oil and gas companies to use more of their existing data in day-to-day decisions by thousands of their employees.

'Day-to-day decisions by subject matter experts such as drilling engineers, petro-technical professionals, and field service technicians, eventually end up making or losing money for oil and gas employers,' Mr Ozden says.

'Oil and gas companies generate and store vast amounts of data, but ironically, 95 per cent of the day-to-day decisions are based on only 15 per cent of that data,' Mr Ozden says. 'Something gets in the way.'

'Corporations have hundreds, sometimes thousands of data silos, but there's no one single software out there that can crawl these sources, as Google or Bing crawls websites,' he says.

The Internet has over 16 billion silos, called websites, and it can be searched easily. Any new data on a website, almost instantly, is searchable, discoverable, and hence useful and useable by anyone. Why is this not the case for data in corporate data silos for use by employees?

'Maana's algorithms and knowledge structure learns and adapts to the data that is crawled, no matter from which data silo it came,' says Mr Ozden. 'We believe that information locked in corporate data silos can be discovered and operationalised just as easily as the information on the Internet, once exposed to Maana's technology.'

Consider a subset of the upstream information systems that oil and gas companies use to store well data: drilling data in WellView or OpenWells; G&G data in OpenWorks and Petrel; log files in Recall or LAS files; financial data in SAP or JDE; and production data in Energy Components and OSI PI data historians. Each of these upstream applications has its own search tool and database. There is no unifying search tool that can index them all and run analytics across all these data sources. Maana wants to change this by approaching the problem of integrating technical datasets as a machine learning and search problem.

Maana leverages machine learning, information classification, information clustering and correlation clustering to discover relationships in the multiple technical datasets. Using these techniques, an emergent semantic graph develops that can be used to organize the petro-technical information. Specialized oil and gas search indexes, such as well names, drilling codes, business unit names and log types, can then be combined with the semantic graph to allow a very powerful integrated search to be performed across the multiple datasets.

'Operationalising' Data

The system is only useful if the results can be 'operationalised', or incorporated as part of people's day-to-day work.

To make this easier, Maana develops custom interfaces (or apps) for different industry roles, drawing data from the main Maana data analytics systems, so everyone has the data they need.

These interfaces can be built in-house by staff, or developed by third party system integrators or other organisations with domain expertise, Ozden suggests.

To write these interfaces 'requires very specific know-how about someone's job,' Mr Ozden says. However, these applications should be extremely simple to write, taking from two days to a week.

If you are a well planner looking for offset drilling performance information, a geologist looking for well logs or a field development engineer looking at risk factors in a certain basin, there will be an interface for you.

Maana aims to give you as much relevant information as possible. If you have a question, it gives you a framework you can build answers from, calculating what might be the most relevant to your question.

'Its goal is to find a set of search results that are very close to your business question,' Ozden says.

Data Processing

The most important aspect of Maana is the data analysis and processing framework.

The data does not need to be moved anywhere, it can be analysed from wherever it currently is.

Maana's algorithms look for relationships between unstructured data and perform statistical analysis to see how your datasets are related to one another. These same statistical algorithms can also discover data similar to the data you are searching so you can benefit from past experiences.

This way, Maana aims to create structure and order out of a company's disparate data sources and present it back to a user in a unified, analytical, searchable format.

Associated Companies
comments powered by Disqus


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


Clustering Considerations in the Machine Learning Workflow – Examples with Exploration Data
Philip Lesslar
from Precision DM


Latest Edition Jun-Jul 2020
Jun 2020

Download latest and back issues


Learn more about supporting Digital Energy Journal