Development of prediction model of hourly water consumption in water purification plant

In a water purification plant, a precise prediction of hourly water demand is needed for supplying water stably to consumers and operating the plant efficiently. Hourly water consumption data per day, which we call a day pattern, reflects the style of our living. The day patterns in weekdays are influenced by fluctuation factors, such as the day of the week, weather and atmospheric temperature. By clustering and analyzing a time series data of hourly water consumption gathered in a water purification plant located in a metropolitan area in Japan for a decade, we can find a regularity of patterns in weekdays. But the day patterns in New Year´s holiday, Golden week (long term holidays in spring) and Bon season (the Feast of Lanterns)-those days are called unique days-show irregularity compared to other usual days. Gathering data of unique days for a decade and observing them from various points of view led us to reveal some regularity, in other words, a reflection of fundamental behavior of the Japanese in these days. These results can be applied in developing a prediction model of water demand through a year including unique days