Cloud Computing Can Be Applied For Reservoir Modeling

?What do online shopping and fluvial geostatistics have in common? The initial answer might be “not a lot,” but the longer answer could be surprising. The rise of cloud computing means that the world’s most famous online retailer and reservoir modeling in the oil and gas industry might have a considerable amount in common.

While the idea of taking datasets away from physical servers and hard drives and placing them on multiple virtual servers on the Internet (the cloud) already has been implemented in sectors such as government and financial services, there also are clear applications for the oil and gas industry and, in particular, reservoir management.

The importance of uncertainty

There is an important need to reduce risk and better quantify uncertainty in reservoir management today – something that is becoming more challenging as reservoirs become more geologically complex and difficult to reach.

The reservoir modeler normally is confronted with sparse data and a need to generate countless realizations and stochastic models to generate a range of possibilities of what appears in the subsurface to reduce uncertainty. Such complex and intensive processes require immense amounts of computer power and many man hours. And while supercomputers today are very fast, they still are struggling to scale up to the growing amounts of work and data at the speeds a user requires.

Enter the cloud

This is where cloud computing comes into its own. Computers have been scaled up about as far as current technology allows within a central processing unit, so the next best option, as many operators already have discovered,

is to scale out through multiple clusters of computers, often termed “distributed computing.”

Multiple clusters, which are used regularly in seismic processing and reservoir simulation, have their downside. The cost of ownership is high, IT infrastructures can be complex and often not well integrated, and smaller operators and other organizations such as universities and think tanks likely will be left behind due to accessibility issues.

Roxar RMS runs on the .rox database. (Images courtesy of Emerson Process Management)

Cloud computing provides all of the scalability benefits of clusters and distributed computing with none of the downside. Through a thin client device such as an iPad or laptop, users can enjoy an elastic capacity of on-demand data and computer power, zero maintenance costs, and significantly reduced capital expenditure requirements.

Cloud computing also can lead to a much more integrated and seamless workflow. Gone are the days where huge datasets need to be transferred to different sites and time lags were common between different applications. With cloud computing, reservoir modelers can enjoy real-time collaboration across different projects and access information from a single truly scalable system.

A transparent and structured reservoir modeling workflow through cloud computing also can act as a repository for years of expertise and modeling advances (particularly important given the number of people leaving the industry over the next few years), help publicize and enforce best practices, and foster a uniform style and standard of work across the operating company and across physical locations. It also can ensure greater productivity from reservoir asset teams, which is crucial in today’s environment.

If the benefits are so compelling, why have all operators not yet adopted cloud computing? Security seems to be the number one barrier. To reach a level of trust, it is helpful to distinguish between public and private clouds. Whereas a public cloud entails the cloud being open to a largely unrestricted universe of potential users, the cloud also can be restricted to a single company with the same robust security as internal IT servers. In fact, while larger operators are likely to be most concerned about security, they also have the resources and expertise to put internal clouds in place.

Just as passing credit card details over the Internet was treated with great skepticism a few years back, this is largely an issue of educating users.

Cloud computing in reservoir modeling today

Recent research from IT analysts at IDC Energy Insights indicates that there is rapid growth in spending for virtual machines. A number of operators also are starting to develop internal clouds, spurred by the appeal of “an on-demand, elastic environment,” as Catherine Madden of IDC puts it.

Emerson’s portfolio of reservoir modeling, simulation, and history-matching products is positioned to capitalize on the cloud-computing phenomenon. There are several reasons for this. First, its software solutions run on the Linux operating system, making the transition to the cloud more seamless. Today, of the 500 fastest supercomputers in the world, 455 run on Linux, according to the biannual Top 500 supercomputer list.

Emerson’s software architecture is focused on being flexible and agile with an emphasis on “thin clients,” where computers are distributed over a network and models can be built up quickly and accurately. Other reservoir management software packages are more “fat-client” focused, where most resources are installed locally, leading to more data at the desktop. The fat-client approach is less well-suited to cloud computing today.

Emerson already has tested a number of its reservoir modeling tools on Amazon to positive effect. A system that relies on Amazon’s S3 and EC2 offerings has a distinct structure.

Emerson also has a number of cluster-enabled products that fit comfortably with cloud computing. For example, the Roxar Tempest simulator deploys simulations across multiple computer nodes, and the automated history-matching tool, Roxar EnABLE, generates multiple realizations and multiple simulator instances across computer nodes. Elastic cloud computing allows the reservoir modeler to scale the cluster according to the size of the problem.

Cloud computing also needs a coherent management strategy. It is in this context that Emerson has developed a common data management platform and architecture that is designed for scalability, can integrate all of its software functions, and help facilitate and navigate reservoir management tools as part of cloud computing.

The new platform, known as “.rox,” will consist of:

• Distributable applications, scripts, and plugins, enabling users to

access software;

• A domain model of the subsurface that covers geophysics, geology, and reservoir engineering objects (reservoir.rox); and

• An object database that exists for geological and production data (source.rox).

The reservoir modeling software, Roxar RMS, already runs on this database.

Do not forget Amazon

So what do Amazon and fluvial geostatistics have in common? Amazon today is much more than just an online book store. It also is one of the world’s leading providers of cloud computing space alongside companies such as Google, IBM, Microsoft, and Rackspace. According to a recent Business Week article, Amazon predicts that its cloud-computing efforts could one day surpass its retailing revenues.

With Amazon, it takes just a few commands to boot up new virtual machines in seconds. Capacity is available on demand and payable on usage. It also is important to note that the cost to carry out processing on its offering, Amazon EC2 (Amazon Elastic Compute Cloud), is up to 10 times cheaper on Linux than it would be on Windows.

Emerson already has tested a number of its reservoir modeling tools on Amazon. Using Roxar RMS on the .rox platform, jobs can be distributed transparently to the cloud. Data can be distributed using source.rox on Amazon S3, Amazon’s storage service. Computations can be performed on the correctly sized virtual cluster on Amazon EC2 with control of the job taking place through Amazon Web Services. Reservoir modelers also can choose to run the jobs locally or in the cloud, depending on their need.

Cloud computing has the potential to usher in a revolution in how the industry handles reservoir models and the data it generates. If a link can continue to be provided between cloud computing and improved reservoir management economics, the sky literally could be the limit.