Stability of the Electrical Supply to Grid Computing Facilities in Tower B, Brunel University in 2009
Datasets usually provide raw data for analysis. This raw data often comes in spreadsheet form, but can be any collection of data, on which analysis can be performed.
During the latter half of 2009 a monitoring programme was undertaken so as to understand the power supply issues underlying the reliability and hence external availability of a tranche of Brunel University's GridPP Grid Computing resources.
The power supply voltage in room TB208 was sampled by querying one of the UPS units. The UPS (APC SmartUPS 3000XL) reports three values for each 60-second interval (defined internally by the UPS):
- the maximum supply voltage during that interval
- the minimum supply voltage during that interval
- the "instantaneous" supply voltage at the end of that interval
The granularity of the measurement is coarse, about 1.5 V. Voltage spikes will be seen as high values of the maximum voltage, and supply dips as low values of the minimum voltage.
The values were transmitted using the standard logging mechanism which removes duplicate values, so missing data points should be assumed to be all identical to the last provided value. They were then extracted from the logs into the text files provided in the accompanying zip archive (MainsVoltage_TowerB-2009.zip).
The text files have been imported into the spreadsheet (MainsVoltage_TowerB.xls) as one week per tab with accompanying chart of the values, starting about 4.00 a.m. each Sunday.
Both daily and weekly variations are clearly visible. Most of the noise is due to the building lift and workshop machinery starting and stopping. An extended brown-out can be seen on the morning of 1st December (week 48).
Raul Lopes wrote the UPS query and logging mechanism.
Henry Nebrensky did the data extraction and import into Microsoft Excel.
A similar approach was subsequently used within the MICE project at RAL, DOI: 10.17633/rd.brunel.9691628.