Santos - cut production data time with Oracle's Exadata
Friday, December 02, 2011
Australian oil company Santos has managed to cut the amount of time it takes to process production data from 120 hours to 10 hours, using Oracle's new 'Exadata' computer, said Steven Benn, manager of surface information systems, Santos.
He was speaking at the Oracle OpenWorld event in San Francisco in October.
Production allocation is managing the data about production history, calculating and tracking the hydrocarbon products from the well head to the point of sale, to report to the joint venture partners and to regulators, and also for managing invoicing and taxes, forecasting and budgeting.
Santos' production allocation system is very complex because there are 17 different joint venture projects with different companies in each one. Each field has different hydrocarbons (including hydrocarbons changing from gas to condensate within the system), different wells, different flows and different partnerships involved. The calculation needs a number of iterations.
There are 400 oil wells, 820 gas wells, product from other companies and gas storage in the system. There are 3 plant sites.
Santos has 190 gas fields and 115 oil fields in production, 5,600km of pipelines and flow lines via 15 satellite facilities.
There are complex commercial arrangements (swaps, first in first out agreements).
The allocation is calculated on the basis of a mass balance from various flow path meeting points, where the amount of flow into the node has to be equal to the amount of the flow out (unless any is stored).
The company was missing reporting deadlines, due to the time taken to process the data. The quarterly reporting to the Australian Stock Exchange was going out to the market with 'preliminary' production data.
This meant that it was using old historical data to make major business decisions, such as for new capital projects.
There was incorrect invoicing, delayed invoicing (leading to reduced revenues), minimal quality control.
'Our production allocation system is incredibly complex system with a large amount of variables,' he said. 'There is extremely large processing, a lot of rework, having to do stuff consecutively,' he said.
So the company was under a lot of pressure to find a faster computer system which could be used for processing the data.
After managing to cut production data processing time from 120 hours to 10 hours using Oracle's Exadata, Mr Benn tried some other software packages on its Exadata computer and saw reduction in data processing time for other packages it is using, including Primavera P6, Spotfire, Fekete, PEEP, Oil Field Manager (OFM).
The Babelfish tool, which displays production and operational data, could process data in 2 minutes, when it previously took 6.5 hours.
Now Santos is considering using the system for other applications, and particularly for spatial data. It would also like software vendors to look at ways to make their products run more optimally over the hardware.