Evaluation of Energy Systems for Distributed Green Data Centres using Lifetime Cash Flow Analysis
MetadataShow full item record
Distributed Green Data Centres (DGDC) are micro-data centre collocated with renewable energy generation sources. Multiple DGDCs are networked together to share information and computing processes. In this thesis, a year-long DGDC simulation balances the computing demand and supply each hour with variable renewable energy production and electricity prices. A five-year Net Present Value (NPV) analysis determines the optimal energy system with the lowest lifetime cost of computing (LCC), equal to the net present value divided by the amount of electricity used for computing. The LCC fairly compares alternative energy systems by the ability to provide lowest cost computing. Grid-isolated DGDCs with 100kW of solar panels have an optimal computing capacity of 20kW and a 20 hour battery capacity. When the optimal grid-isolated DGDC is connected to the Ithaca electricity grid the battery storage is 10 hours and the LCC is $10/MWh less. The NYSERDA Solar PV Program applied to a grid-tied DGDC reduces the panel cost by $76,806 and reduces the LCC by 8.5%. Grid-tied DGDCs in Ontario are not feasible due to the high Feed-in-Tariff incentive to sell electricity. A grid-isolated DGDC is optimal in Ithaca when the cost of transmission infrastructure upgrades are greater than $8291 or when the value of a generated Renewable Energy Credits (REC) are higher than $38/MWh. Both on- and off-grid DGDCs use equal amount of energy for computing but a grid-tied DGDC is able to sell 26% of electricity from a 100kW solar system with only 10kW of transmission capacity. Grid-tied DGDCs are active market participants responding to electricity price signals. DGDCs buy low-cost electricity and sell electricity when prices are high. On average the cost of energy used by the DGDC is $19/MWh less than grid electricity. DGDCs provide a reliable source of computing for high-priority applications such as video streaming as well as a low-cost option for time independent tasks such as batch processes or cloud storage.