I'm a senior post-doctoral researcher in the Insight Centre for Data Analytics, UCC. I received my doctoral degree for my dissertation "Identifying Sources of Global Contention in Constraint Satisfaction Search" in 2012, where I investigated generic adaptive methods to identify critical variables for branching in constraint programming. The method developed was shown to outperform the state-of-the-art commercial tool in solving a number of well-known scheduling problems.

Globally Optimised Energy Efficient Data Centres

My current primary research work involves optimizing workload allocation across a single / multiple data centres such that total energy cost is minimised, as part of the EU FP7-Transport funded project "GENiC". Information regarding the expected energy usage of VMs and the expected impact of VM assignments on energy consumption of the cooling system are considered.

Condition-based Maintenance Optimization for Trains

My previous post-doctoral research involved fleet maintenance optimization for trains, as part of the EU FP7-Transport funded project "MAXBE". The problem required scheduling planned periodic maintenance, while incorporating uncertainty regarding possible unplanned maintenance requirements due to operational failures, such that a time-variable fleet demand profile is met, and exam due dates are respected, where possible.

Energy Management Optimization

Prior to the work on condition-based maintenance optimization, I worked on optimization for energy management systems, in particular automation of energy consumption in the residential sector. There are a number of motivatory factors for this work (the increase in renewable energy generation on the supply-side, forecasted widescale influx of electric vehicles to the grid on the demand-side, etc.), which will require in a shift from the traditional approach of "matching supply to demand" towards an approach of "matching demand to supply".

In order to incentivize users to shift their energy consumption from peak to off-peak times, electricity prices which are more reflective of the true/predicted cost of electricity at a given time can be used. However this still requires the user to be in a position to react to these time-variable prices. Home energy management systems can remove this burden from the user, by automating the scheduling of flexible energy consumers in the home with respect to electricity price and user comfort objectives.

This research was funded by the Irish Research Council and Intel Labs Europe, and was joint work with Helmut Simonis of the Insight Centre for Data Analytics and Charles Sheridan, Annabelle Pratt, and Dave Boundy of the Energy and Sustainability Lab, Intel Labs Europe.


Detailed experimental results on problems in the XCSP format, summarized in the dissertation, can be found here; while detailed results for the job shop scheduling problem and a number of variants of this problem can be found here.


Copyright © Diarmuid Grimes     Design by Iron Spider