SPE Denver ACTE 2003

The message from the keynote addresses was ‘we need oil and will do for a few decades to come’. Oil supply is OK for the next decade or so—in fact global oil production may not yet have peaked. An aging workforce is a concern as are gas supplies to North America. The environment was practically absent from the debate—surprising, since CO2 sequestration was a popular topic in the paper sessions. We noted a growing interest in real time production control and optimization through simulation. Here techniques used in the oil refinery are being applied to modeling wells, flowlines and production facilities. New technology is also successfully changing the game for monitoring of drilling and production operations. ConocoPhillips’ account of its Norwegian Real Time Operations Center left no doubt that remote, real time operations are here to stay. Roxar has moved simulation into the geological model with its RMS FlowSim breakthrough technology.


Kemble Bennet (Texas A&M) believes that technologies such as 3D/4D seismics, artificial lift and the digital oilfield of the future are contributing to breathe new life into old fields—reversing natural decline. Demographics is a concern. US enrollment in petroleum engineering courses is currently around 200—way below its 1984 peak of 12,000. As many workers quit the industry, the only growth segment is of ‘free agents’—independent consultants. Bennet’s points out that as oil plays a major role in energy supply over the next several decades, a large proportion of the workforce will vanish. Mark Sikkel, (Exxon and NPC Council) observes a ‘fundamental shift’ to a tight gas market and high prices. Two alternative scenarios are suggested. ‘Conflict’—restricting supply and encouraging demand—will push gas prices to $6 per BTU. ‘Alignment’—of policies on supply and demand—will lead to moderate prices of around $3. A maturing conventional N. American resource base will only supply 75% of demand. Non conventional resources like coal bed methane, tight gas and shales will partly fill the gap. Halliburton Energy Services president John Gibson has been looking at the statistics to discover that, for wells in the 10,000-15,000 ft range, there has been no change in drilling efficiency over a 20 year period. A rig still drills about 30 wells per year. While marketing departments tend to focus on successes and exceptional achievements, in the ‘middle of the Gaussian distribution, technology is just not impacting drilling’.

Real time

We noted a lot of buzz around coupling surface facilities with reservoir models. This can be just at the design phase or taken a step further to real time optimization during production. Techniques borrowed from the refinery are touted as having widespread application to the oilfield. But there is some reticence to putting sophisticated and potentially delicate equipment down hole. Real time optimization (RTO) is usually associated with the high end of the production business—high-tech, fiber-equipped wells and sophisticated downhole instrumentation. But RTO also has a role to play in the much larger market of ‘dumb wells’ and old fields where surveillance and alarms may be of critical importance to safety.

Sim/Opt software

Petroleum Experts’ software can now simulate aggregations of fields and production facilities. Simulator outputs from multiple fields can now connect to ‘arbitrarily complex’ surface facilities. Applied Flow Technology’s Mercury provides ‘intelligent sizing’ of pipelines by coupling network flow modeling with optimization. Scandpower’s Olga 2000 system models wellbore to pipeline fluid flow. An ‘APIS’ real time add-on couples calculations to SCADA or DCS controls to provide a ‘window into the pipeline’. Caesar Systems PetroVR models the big picture. A demo modeled gas demand and deliverability to identify and mitigate risks such as schedule slippage and HSE issues in the fabrication yard. New collaborative business simulation software, ‘Team VR’ extends PetroVR to reservoir, well and facility modeling. United Devices GridMP can also be deployed to distribute computation. A new release of Decision Team’s integrates information from SCADA, simulators, FieldView and TOW into a central database. A ‘hybrid artificial intelligence’ approach makes the system ‘self-learning’. SimSci-Invensys’ PipePhase was originally developed by Chevron. The latest release includes copy/paste of pipeline profiles and other tabular data from Excel. A vertical flow performance generates profiles for well characterization.


Wellsite hardware is a very busy area. Innovations we spotted cover data acquisition, surveillance and communications. Aks Labs' new ‘Data Trap’ is a bolt-on data logger for gas well test and measurement. Data is recorded on to a Compact Flash card. The technology is used by Marathon and was the subject of an SPE paper (62881) on the field application of production analysis. E2 Business Services’ ‘PhoneHome’ intelligent remote video surveillance provides a virtual rig site ‘presence’. Motion detectors trigger recording and the technology can read car number plates as they drive on site. The big news in data acquisition was the acquisition of startup Luna iMonitoring by IHS Energy. iMonitor featured in our report from last years SPE ACTE. iMonitoring’s solar powered well data collection systems and remote automation technology will now integrate IHS’ FieldDirect service. Another company working this space is vMonitorwhich provides web-based automation software for data acquisition, data management and integration. vMonitor has received an equity investment from Baker Hughes.

Well reporting

Epoch’s myWells.com is a single point of information on operational activity including electronic tour sheets, drilling and mud logs. Logs can be plotted and data queried with secure user access to the data hosting service. Geoservices was showing its new ‘gWeb’ real time web-based mud logging platform. On-site data and documents from multiple vendors can be collated and transmitted to client sites. gWeb uses Visean’s Secure Web solution. Access control is secured with RSA SecurID one-time pad smart cards.

Reservoir simulation

Roxar is claiming a first for a new ‘Eclipse/VIP’-class simulator, now embedded in its flagship RMS geological model. By bringing dynamic flow modeling into the geological model, Roxar hopes to bring simulation to a wider user base. More scenarios can be screened before building the full-scale model. Static and dynamic data is kept in the same database. Decisions made are tracked and stored. Geologists and engineers ‘speak the same language’. A demo showed how multiple realizations of the geological model can be conditioned to dynamic data and matched with well test data. Pressure pulse data was analyzed with well-to-well animation. The new dynamic options integrate the RMS tree-view workflow manager and support data mining of production data and infill well planning. Optimization Petroleum Technologies PE Office is described as ‘the Petroleum Engineer’s Microsoft Office’. PE Office performs production data statistics and analysis, performance analysis, optimization and forecasting. We saw a preview of a new 3D module for ScienceSoft’s S3Graph offering data mining and graphing for reservoir simulation and production data. An OEM version is embedded in Landmark’s SimResults tool. The software uses Microsoft DirectX graphics and ScienceSoft expects to leverage cheap gaming cards to allow high end visualization on a PC. S3Graph works with Eclipse, VIP, MORE, SURE, FrontSim etc. 3DSLnet is StreamSim’s new hosted service running 3DSL remotely from any web browser. Once the job is complete, output is compressed, encrypted and transferred back to the client machine. All data is immediately destroyed on the server as soon as it is transferred back to the client. 3DSL interfaces with Scandpower Petroleum technology’s MEPO optimizer.

Data mining

Spotfire is now sold bundled with a data set IHS Energy’s Gulf of Mexico deepwater study. The data set can be used to compare different estimates of oil in place with decline curve analysis. Acquisitions can be evaluated by looking for inactive wells with sizable OOIP – as determined by statistical analysis. SpotFire is a bit bewildering—but definitely powerful. If there are determinant statistics hidden in your data—this is the tool to weasel them out.

Papers of note

If you ever doubted vendor’s hype on real-time remote operations you should have heard Mike Herbert’s (ConocoPhillips) report (SPE 84167) on its use of a remote visualization and control center on Ekofisk. The remote operations center—developed with Halliburton/Sperry-Sun, InSite and Sense Technology—supports drilling, geosteering and cementing operations— all carried out remotely from the Stavanger-based center. Saudi Aramco will require 50 large scale (1 million cells and up) reservoir simulations in 2006—a 20 fold increase over 2000 as Walid Habiballah related in SPE 84065. Aramco opted for PC clusters with a high speed MPI switch and OpenMP to run the Powers simulator, a dual parallel scheme with grid partitioning. A 9.6 million cell model of the Gahwar field matched 50 years of history from 3,200 wells in 8.3 hours. Parallel scalability was tested up to 125 processors and found to be 90% linear (computation was ‘super-linear’ but the network slowed things down). Grid partitioning schemes are critical for performance. 4 clusters, each with 128 processors are now installed. Luigi Saputelli (University of Houston and Halliburton) described (SPE 84064) how ‘real-time’ operates on different time scales: from fast (flow control at surface and SCADA) through planned well shut in, injection planning to slow (asset management). Optimization scope varies—from a single well focus to field-wide. In general the different numerical models used in optimization are not coupled. University of Houston has developed a hierarchical ‘data-driven’ model of a self-learning, self-optimizing process. The control system uses Model Predictive Control, a ‘very mature technology’ taken from refinery control systems, to optimize parameters such as the bottom hole pressure needed to maximize production. In multivariate optimization the least work path from existing to desired performance is determined by ‘exciting’ the system with small random perturbations. Multi-level (short and long term) optimization is reduced to a linear optimization problem of net present value.

This article has been abstracted from a 25 page illustrated report produced as part of The Data Room’s Technology Watch report service. For more information on this service please contact tw@oilit.com .CD-ROM proceedings and SPE papers available from www.spe.org.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.