Modelling uncertainty in district energy simulations by stochastic residential occupant behaviour

Abstract

Occupant behaviour has since long been of main interest in the domain of building energy savings and indoor air quality; and its importance is recognized by its wide coverage in literature. In the recent developments of detailed transient building energy simulations, including the occupant behaviour as boundary condition for the thermal comfort and system efficiency calculations has been a major research topic given its significant impact. Simultaneous growing interest in district energy simulations raises similar questions at the aggregate level, where upscaling from the building to an aggregate neighborhood level at the spatial scale of a low-voltage feeder results in a natural regression to the mean lowering uncertainty compared to the level of the household.The presented work starts with the description of StROBe, a stochastic residential occupant behaviour for district energy simulations; integrating the modelling of receptacle loads, internal heat gains, thermostat settings and hot water tappings based on occupancy and activity prerequisites. Given this model, the uncertainty for district energy simulations is addressed. The epistemic uncertainties are aleborated first comparing model results with reference values and denoting local disaggregation of demographic statistics as possible main hiatus of general modelling methods for building energy occupant behaviour used at the neighborhood level. To conclude, the aleatory uncertainty caused by stochastic residential occupant behaviour in integrated district energy simulations are quantified. Here, the expected value of the objective functions have to a large extend the same minimizers as the measures of the proposed robustness. As such, optimizing an objective value for its expected value generally seems to result in a optimum near the optimum of robustness. However, 95 percent of the observed objectives lay between 0.81 and 1.6 times the expected value for a feeder larger than 10 houses and between 0.88 and 1.3 times the expected value for more than 20 houses denoting an overall 'rather small' uncertainty on the possible objective functions caused by user behaviour. Furthermore, we show that the design of the building energy system has its impact on the robustness of the objective criteria and it could thus be minimized as part of an optimiation exercise.

Occupancy Based Thermal Energy Modelling in the Urban Residential Sector

“…Further work must be spent on understanding the uncertainty introduced through this approach. Probabilistic methods as proposed by [19], [40], should be compared to this deterministic approach.…”

“…The methodology and simulation framework allow to analyse more aspects like the spatial distribution of heating patterns, set point preferences [40], fraction of dwellings heated [40], and electricity demand [19] and incorporating those will allow to create a more comprehensive picture of urban energy demand.…”

A Markov Process Approach to Ensemble Control of Smart Buildings

“…We construct a physical building model, described in Section II-B, using parameters for individual buildings, their heating systems, occupancy and behavioral choices of occupants reported in [16], [21], [22], [18], [23]. The key parameters are also summarized below.…”

“…When unoccupied, the lower bound is set to 16 • C and the upper bound is unchanged. To account for the stochasticity of internal gains, we generate 52 user behavior profiles using STroBE (https: //github.com/open-ideas/StROBe), a statistical tool described in [22], which allows to capture different levels of consumers' presence (e.g. occupied or not).…”

A theoretical benchmark for bypass controllers in a residential district heating network

“…In this case, which is limited to summer period simulations, only the DHW use is of importance. To set up these profiles, use was made of StROBe [25], a Python toolbox that generates stochastic user behaviour profiles. These profiles represent typical Belgian households, and feature several types of users: e.g.…”