Over last summer, I had a research internship at the Department of Earth Sciences, Durham University. The placement was sponsored by BP and the Natural Environment Research Council. Based in the academic offices at the University, I worked with a team of highly knowledgeable PhD students, lecturers and postdocs on work in Enhanced Oil Recovery (EOR), a process to extract more oil from existing oil wells. The internship was for ten weeks.
Imagine an oil well. Drilling a pipe into the earth will result in about 5% of the well’s oil flowing up due to natural pressure. Drill a second hole and pump in cold water and you can get about 30% out. A new process known as Enhanced Oil Recovery can increase this number by about two to three percentage points. The translation of a couple of percentage points in extra oil from every oil well in the world is not a trivial thing and represents billions in saved time and effort, not to mention the financial importance of these processes to companies and the economy. In simple terms, more supply leads to a fall in prices. This translates into lower raw material costs for firms in the hydrocarbons industry and for the general economy in fuel costs.
During my time, I consulted, designed, implemented, tested and published a software application to automate submissions to HPCs, creatively called the High Performance Computing Automator or HPCA. Built in Java, the software takes a file for submission to an HPC, asks the user for parameters for their simulation and sends it for computation. Trivial as this sounds, it is no easy feat for chemists with no experience of parallel and cluster computing, never mind Unix. The software runs completely in a window and no command lines are necessary – a vital feature for easy human-computer interaction. Gre
en typefaces and black backgrounds don’t make for the most welcoming of experiences to newcomers in computational chemistry.
The program flow is as follows:
- Launch window system
- Offer option to log in
- On successful login, allow capturing of simulation parameters
- On submission, set up SSH connection in background
- Send necessary commands
- Perform a secure copy of the simulation file to the specified directory
- Inform user of submission success outcome
The program was a success and is estimated to save approximately five hours of learning time per new chemist.