A holistic numerical model to predict total utility demand such as thermal requirement, various energies, domestic hot water, and city water of a residential house or a set of dwellings like a residential building, a residential area and even a city was established, which we call Total Utility Demand Prediction System (TUD-PS). The system based on the methodology for generating actual inhabitants' behavior schedules with 15 minutes time-resolution, previously reported, and a dynamic thermal load calculation. The latter part of the model takes account into the probabilistic model for HVAC turning On/ Off events derived from the Markov Chain, also previously developed, which can realize to obtain probabilistic thermal requirement impacted by inhabitants' On/ Off behaviors. Simulation concerned on seasonal and peak loads for a single dwelling reveals that the so-called maximum load is phenomenally influenced by the assumption whether HVAC turning On/ Off events are probabilistic or deterministic. Hence, a spatial accumulated time-series of utility demands of respective dwellings must be predicted by the proposed model, where simultaneous dynamics of respective dwellings can be reproduced, at least, must not be applied a conventional and practical method where you predict a holistic demand by superposition of a demand at a typical and ideal dwelling.