We apply the process to a 70,000 square meter international school campus in the humid, tropical climate of Singapore. It was built in 2010 and includes a building management system (BMS) with over 4,000 measured data points taken at 5 minute intervals from the years of 2011-2013 - resulting in close to 800 million records of raw data. This collection includes 120 Power Meters and 100 Water Meters in the energy and water management system. The dataset for this building is a seed dataset in an open repository of detailed commercial building datasets \cite{Miller:2014vr}. The primary energy load for this campus is the chilled water plant which consumes almost /month in electricity.

We focus on the chilled water plant electricity consumption due to its importance in this climate and the potential savings opportunities available through chilled water plant optimization. Measured kilowatt-hour readings were taken from July 12, 2012 to October 29, 2013 with 474 total daily profiles analyzed. Figure \ref{fig:sankeyheatmap1} illustrates a sankey diagram with heatmap of the output of the DayFilter process with parameters set to \(A\)=3 and \(W\)=4. The discord and motif candidates are separated in this case according to a decision threshold which quantifies a discord as a day-type with a frequency count less than 2% of total days available. This distinction results in 39 days with patterns tagged as discord candidates which is 8.2% of the total days in the dataset.

In general, there are six main motif candidates with two candidates appearing to be typical weekday types, two holiday or half-capacity types, and two weekend types. Pattern \(aaaa\) and \(abaa\) are predominantly flat profiles common to non-occupied cooling consumption. Patterns \(abba\) and \(acba\) are representative of days in which school is out of session but the office spaces are still occupied by staff. Pattern \(acca\) represents a standard full-occupied school day and it is by far the most common with 202 days tagged out of 474. Pattern \(accb\) is similar to \(acca\) with slightly more use in the late afternoon and early evening. This phenomenon is due to extracurricular activities planned outside the normal operating schedule of the facility.