Biofuel research potential grows with updated software

The impact

The biofuel crop of the future is likely to be a woody biomass – trees such as poplar and willow, but grown in a very ancient, traditional style called short rotation coppicing. On a two to five year cycle, the trees are cut right back down to ground level, only to sprout again with healthy new growth. Unlike other biofuel crops that take space from food crops, this can be grown on land that is unsuited to arable farming.

Professor Gail Taylor, Director of Research and Professor of Plant Biology at University of Southampton, is responsible for a modelling project that predicts the potential yield of such crops, in different locations across the UK and at different levels of climate change.

To do such modelling well, the team needs software that can cope with the increasing demands being put on it. Unfortunately, Forest Growth SRC, the system being used, was slow and holding back what the researchers could do.

After working with the Software Sustainability Institute on a project to improve the software, however, the team has seen an 18-fold increase in the speed with which they can do their modelling. From a Windows PC-based system that struggled to keep up with the research being done, the team now has a system that can map the whole of the UK in two and a half hours.

The problem

"This is likely to be the main biofuel of the next few decades, it’s very cost effective and a sustainable way to decarbonise land," Professor Taylor says.

"But unlike wheat and barley, which we’ve been growing for hundreds of years and so have effective predictive techniques, this crop is a novel one, and we need models to identify which areas to target, and which to rule out," she says.

"We look at low, medium and high climate change scenarios out to 2050, with more than 30 input variables and covering every 1km2 of the UK. That’s over 200,000 squares. The system used to take 24 to 36 hours, with two PCs running back to back, and it would often crash and have to start again," Professor Taylor says.

When Professor Taylor got in touch through Open Call, the Institute began to look at what was possible. The answer to increased speed was clear: University of Southampton has its own supercomputer, IRIDIS4 – why not take advantage of that?

"I’d never have been able to use that on my own," says Professor Taylor, "but now we have time on it, and software that can run on it."

The solution

The Institute team ported the Windows-targeted Fortran and C++ source code to run on Linux, and developed a framework to manage large runs of parallel executions of the model.

Porting the legacy source code revealed the existence of several, potentially conflicting, versions of the Forest Growth SRC source code - a classic academic research problem, as software has evolved over time.

The Institute therefore established a subversion repository to bring together the existing, disparate Fortran, C++ and Java codebases under version control. This allows all changes to be tracked and attributed, and ensures that the precise version of the code used in a publication is known and accessible. New, user friendly, build and deployment processes were also put in place to make future development easier.

The software remains "very geeky", says Professor Taylor, and the next round of funding will be used to create a more user friendly version that can be used by other interested parties – government bodies, policy makers, energy companies and researchers in other universities, she says.

The team has also won a small contract to do some work with Energy company EON, and an EPSRC grant to work with researchers looking at other technologies and how they might work with biomass.

As Professor Taylor explains, "they’re interested in feeding our data into different technology scenarios to see what might work best. Biomass could be used in co-firing coal stations, for example, or in making bio-oil. So the outputs of our model – potential crop yield and timescales - feed into that."

Meanwhile, the new, faster code opens up research possibilities including modelling at a continental level, at higher resolutions and for a wider range of climate scenarios.

"SSI has done a great job in making the Forest Growth SRC software run smoothly and quickly", Professor Taylor says, and she is excited to see the next stage of development.