Cloud computing for complex geothermal simulations using the parallel software Waiwera

Abstract

Numerical modelling of geothermal fields has been around for a few decades and has proven to be very insightful for sustainably managing geothermal projects. It is a great, cost-effective tool for evaluating various development plans using a realistic model of the field.
Simulating geothermal fields, however, is a time-consuming process, especially when the model become more complex and refined, and the simulation time increases. The geothermal simulator Waiwera, recently developed by the University of Auckland (UoA) in collaboration with GNS Science, has proven to be able to reduce simulation timedrastically by using parallel computing on standard desktops, and even more so when using high-performance (cloud) computing. Simulation times reduce by a factor of 10-50 depending on the complexity and size of the model. Additionally, uncertainty quantification of a model, which requires running thousands of sample models, benefits hugely from high-performance cloud computing.
In this paper we explore three cloud computing platforms: NeSI (New Zealand eScience Infrastructure), AWS (Amazon Web Services) and Azure (Microsoft). Different options of user interaction with the cloud computing platforms will be discussed. Moreover, the simulation speed and costs will be compared for these three platforms.