Guangruizi Wang

THE USE OF AGENT-BASED MODELLING IN DETERMINING BUILDING OCCUPANT BEHAVIOUR AND THE APPLICATION IN CITY-SCALE ENERGY SIMULATIONThe influence of building archetypes on urban thermal energy simulation

Occupant behaviour and over-simplifications of building thermal properties are perceived primary explanatory factors accounting for the deviation of building energy predictions from real-life metered data. Despite considerable efforts made by predecessors, it is widely recognised that the conventional approaches aggregate occupant behaviours at a low spatial resolution. Besides, seldom do researches incorporate detailed building physical properties with bottom-up occupancy-based energy models, and the current building stock models tend to focus on ‘reference buildings’ that could represent the entire urban building stock whereas the level of ‘representativeness’ is hardly questioned.

As a mean to bridge the first gap regarding uncertain occupancy, the Energy Efficiency City Initiative (EECI) group of University of Cambridge have developed a domestic thermal energy simulation model based on individual occupancy times series generated from the UK Census 2011 and the UK Time Use Survey 2000. Provided that the model is built on hypothetical standardised dwellings with fixed floor areas and thermal properties, this research targets at improving the previous work and filling in the second gap by linking the original simplified modelling approach with real-world complexities via the deployment of real-life dwelling data.

The key research question awaiting to be answered is: what influences would the introduction of realistic dwelling archetypes have over city-scale heating energy demands both spatially and temporally. As a mean to respond to the question, the research substitutes actual building information retrieved from the EPC - UKBuildings dataset for the fixed dwelling modelling inputs. Also, the study further utilises the refined model to explore the energy and carbon implications of a range of passive system retrofitting options.

Firstly, based on the UK Census 2011, seven dwelling categories have been classified: detached/semi-detached/terraced whole house/bungalow, purpose built/converted flat/maisonette, temporary structures as well as flat/maisonette in commercial buildings. After adding the ‘dwelling type’ category to the synthetic population under the ‘household’ feature, for each resident/household in the case study area (London Borough of Haringey), the dwelling type could be matched precisely. Next, the EPC and UKBuildings datasets are spatially joined and cleared to pool together the six building physical inputs (window/roof/wall/floor u-values, window-to-wall ratios, floor areas) for all the dwellings within Haringey.Meanwhile, for each entry in the merged dataset, the corresponding building property values in the numerical form
are determined based on pre-defined conversion matrices modified from the RdSAP 2009. Moreover, the dwelling types of entries within the merged dataset are obtained by either examining the built form or the year of construction. During the following modelling stage, provided the dwelling type of a particular household of interest, the set of thermal parameters as inputs are randomly sampled from the pool of entries under the same dwelling category.

After the incorporation of building archetypes, no evident spatial pattern in average thermal energy demands could be identified, owing to the increased heterogeneity introduced by various building features. The fluctuations of temporal consumption pattern tally with the occupancy time series generated in the previous research, whereas higher variations could be observed during both weekday and the weekend. Also, single socioeconomic factor (age, household size, economic activity) has less influence over regional energy demand than building thermal properties, and those window-related factors (WWR and glazing u-values) are found to be especially powerful at determining thermal energy usage.
When the building envelopes are upgraded to the highest insulation level specified in CIBSE ‘Good Practice Guide’ and other sources, 58% of carbon (398 kgCO2/annum) and energy savings (1,000 kWh/annum - 3,000 kWh/annum) could be achieved on per household level. Moreover, window and roof insulation upgrades are identified to be more influential at offering domestic thermal energy savings.

Through examining the impacts of building archetypes on urban thermal energy usage, the individual/city-wide decision-making (e.g.,, building retrofit) could be better informed. The improved residential thermal energy simulation model could also be deployed by future researchers as the basis to identify the relationships between occupancy and alternative energy consumptions (e.g.,, electricity, gas, domestic hot water usage). In addition, owing to the modularity of the model, the spatial scope of the model structure could be scaled either up or out for other urban regions worldwide, incorporating the population and housing data of the particular area of interest.

Keywords: agent-based model, city-scale energy simulation, building stock modelling, retrofitting option evaluation, building archetype, bottom-up thermal energy simulation, domestic energy simulation, population, micro-simulation