Total utility demand prediction system for dwellings based on stochastic processes of actual inhabitants

Abstract

This article describes a new methodology to calculate the likely utility load profiles (energy such as power, natural gas, space heating and cooling, and other thermal requirements, as well as city water) in a dwelling. This calculation takes into account the behavioural variations of the dwelling inhabitants. The proposed method contains a procedure for cooling load calculations based on a series of Monte Carlo simulations where the heating, ventilating and air conditioning (HVAC) on/off state and the indoor heat generation schedules are varied, time-step by timestep. A data set of time-varying inhabitant behaviour schedules, with a 15-min resolution, generated by the authors in previous studies and validated by a comparison analysis to several field measurement data sets, was integrated into the model. The established model, which is called the total utility demand prediction system, can be applied to, for example, likely estimation of an integrated space maximum requirement, such as the total load of a building or an urban area. In a series of numerical experiments, huge discrepancies were found between the conventional results and those considering the time-varying inhabitant behaviour schedules. In particular, deriving the dynamic state change, of having the HVAC on/off from the inhabitants' schedules, was found to be a significant factor in the maximum cooling and heating loads.

Occupant behaviour modelling in domestic buildings: the case of household electrical appliances

“…As a consequence of the challenges of modelling occupant behaviour, the occupants' use of household electrical appliances is often modelled as fixed static schedules in building simulation tools such as DOE-2, BLAST and EnergyPlus (Abushakra and Claidge 2001;Hoes et al 2009). However, such procedures can lead to overestimated peak values, as defining standard behaviour for types of households fails to consider the random variability of the occupant behaviour (Tanimoto and Hagishima 2010). This paper develops new insights into appliance behaviour modelling for household electrical appliance use.…”

Estimation of the contribution of the residential sector to summer peak demand reduction in Japan using an energy end-use simulation model

“…By replacing the RACs, the reduction effect is high (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17). The reduction effect is greater after 16:00, at which time the occupants start arriving home and using the RACs.…”

“…The model was validated by either the aggregate demand (200 detached houses and 200 apartments) or the individual demands of 14 households. Tanimoto et al [8,9] developed a methodology of generating timevarying inhabitant-behavior schedules that utilizes two published statistical data. By defining links between each behavior and an energy-consuming event, they simulated electricity, gas, water, and hot water demand with a 15-minute resolution.…”

State transition stochastic model for predictingofftooncooling schedule in dwellings as implemented using a multilayered artificial neural network

“…This requires accurate prediction of utility demands, such as electric power and hot water with high time resolution, because CGS provides power and hot water simultaneously. With this background, we developed a new methodology for accurately calculating time series utility loads (energy, power, city water and hot water) in a dwelling, called the total utility demand-prediction system (TUD-PS) (Tanimoto and Hagishima 2010 For the HVAC on/off state, we applied a stochastic model to deal with probabilistic events for turning the HVAC on/off, based on Markov chain theory (Tanimoto and Hagishima 2005; hereafter TH05). This model provides both state transition probabilities of the HVAC from the off to on and on to off states, as defined by the indoor globe and outdoor temperatures, respectively.…”