Koala, the faster Excel

open-source / consulting
CHALLENGE

The Center of Expertise in Modeling and Economic Studies of Engie, simulates continuous assessment of a natural gas fields wide range, along several physical and economical dimensions.

Therefore, thousands of complex Excel files need to be recomputed on the flight. Using several Excel processes is neither fast enough nor reliable, as one loop in the computation takes at least a week.

SOLUTION

Inspired by 2 open-source libraries, OpenPyXL and Pycel, we developed a python tool named Koala able to read an .xslx file, build a graph from its Abstract Syntax Tree, and perform calculations .

RESULT

Computations are about 6 times faster than an Excel-based solution, and can be parallelized. Using a cloud plateform, we were able to achieve the computation in 20 minutes.