Spatial microsimulation

Research articles · Apr 30, 2014

Constructing an urban microsimulation model to assess the influence of demographics on heat consumption

M. Esteban Muñoz H., Irene Peters

Technical Urban Infrastructure Systems Group, HafenCity University, Germany

Volume 7(1) Spring 2014

Cite as: M. Muñoz H., I. Peters; 2014; Constructing an urban microsimulation model to assess the influence of demographics on heat consumption; International Journal of Microsimulation; 7(1); 127–157. doi: 10.34196/ijm.00096

Abstract

We present ongoing work on the construction of a spatial microsimulation model to assess the influence of demographics on residential heat consumption for Hamburg, Germany. Demographics are important for urban energy planning as: (1) Buildings are becoming more energy-efficient and building occupant behaviour accounts for a growing share in the variation of consumption; (2) building occupant needs are changing along with demographic change; and (3) the share of small decentralized district heating grids, in which fewer customers mean less averaging out of heterogeneous occupant profiles, is set to play a bigger role in the country’s heat supply. We construct a spatial microdata set for the city of Hamburg (of roughly 1.8 million inhabitants and 370 000 buildings), with households populating geo-referenced buildings, in three steps: (a) Synthesizing the population of small scale “statistical areas”, comprising up to around 2000 people (we do this by selecting households recorded in the German microcensus and fitting them into the statistical areas); (b) assigning energy relevant properties to the geo-referenced buildings from the Hamburg digital cadaster (we do this by making use of a well-established building typology developed for energy assessment) and constructing dwelling units in these buildings; and (c) matching households to the dwelling units in these buildings (which we do again by using household data from the microcensus). This last step – allocating households to buildings – may be the most interesting and challenging task. As of to date, we use a combinatorial optimization algorithm to achieve this. Once we have a microsimulation model of buildings and households living in them, including their demographic composition, the range of questions that can be explored is immense. The illustration presented here is a simple heat balance computation of individual buildings, using the constructed socio-demographic data and the digital cadaster data as input parameters.

This article contains 8 figures and 6 tables.

Figures

Figure 1. Model data structure and simulation steps.
Figure 2. Spatial data used in our model.
Figure 3. Occupant influence on different building typologies arrange by construction year.
Figure 4. Performance of the algorithm used for selecting households to fit aggregate values of statistical area 16004.
Figure 5. Geometrical simplification of selected buildings (left) original geometry; and (right) simplified geometry used as input for the estimation of heat demand.
Figure 6. Comparison between simulated and observed parameters.
Figure 7. Comparison between estimated heat demand and simulated heat consumption.
Figure 8. Histogram showing the frequency of construction year in statistical area 16004.

Tables

Table 1. Different types of models for energy simulation of buildings.
Model Type\Occupant Type Static heat balance model (at individual building level) Dynamic thermal simulation model (at individual building level) Statistic model relating heat consumption to occupants, at various levels of aggregation, no heat balance imposed
“Average Occupant” (household comprised of identical norm persons; only their number counts) Blesl, Kempe, Ohl and Fahl (2007), Deutsches Institut für Normung e. V (2011), Dascalaki, Droutsa, Balaras and Kontoyiannidis (2011), Loga, Diefenbach and Born (2011) EnergyPlus Development Team (2012), Henden and Lamberts (2011)
Household with individualized demographics and behaviour Chingcuanco and Miller (2012) Munoz & Peters (2014) Lüdemann (2001), Borgeson and Brager (2008), Page et al. (2008), Mahdavi (2011), Widén et al. (2011). Scott (1980), Bohi and Zimmerman (1984), Guerra Santin et al. (2009).
Table 2. Variables used to fit households from the Microcensus into statistical areas.
Constructed variables
Variable ID Variable Name
Average household size (No. of individuals per dwelling unit)
Number of single households
Number of households with children
Number of single parent households
Variables directly taken from the micro-census Variable Name
ef455 In what year was your home built?
ef451 How many dwelling units has the building in which you are living?

Source: MIKROZENSUS 2002 Statistisches Bundesamt (2002). Variable names: translation by the authors.

Table 3. Variables used for the allocation of the synthetic households into the semi-synthetic dwelling units.
Living standard of the household
Variable ID Variable Name
ef451 How many dwelling units has the building in which you are living?
ef453 What is the floor area of the dwelling unit?
ef455 In what year was your home built?

Source: MIKROZENSUS 2002 Statistisches Bundesamt (2002). Variable names: translation by the authors.

Table 4. Variables used to estimate the time at home for the single individuals.
Labor participation
Variable ID Variable Name
ef95 Employment status in the reference week
ef138 Full-time / part-time job
ef141 Full-time / part-time job
ef147 Work on Saturday (February until April)
ef148 Work on Sundays and public holidays (February until April)
ef149 Evening work (between 6 p.m and 11 p.m) (February until April)
ef150 Night work (between 11 p.m and 6 a.m) (February until April)
ef151 Night work hours (between 11 p.m and 6 a.m) (February until April)
ef163 Home office (February until April)

Source: MIKROZENSUS 2002 Statistisches Bundesamt (2002). Variable names: translation by the authors.

Table 5. Rules to select input variables to the heat balance model based on the average working hours in the building.
Computed work time (wk) qi[W/m2] Ti[C°] n[h−1] Possible occupant type
wt = > 10 3 18 0.3 single household, employed
wt = ≤ 9 ∧ > 8 6 19 0.4 both parents employed
wt = ≤ 8 ∧ > 4 5 20 0.5 “average” occupant
wt = ≤ 4 ∧ > 1 6 21 0.6 part time
wt = ≤ 1 7 22 0.7 unemployed

qi[W /m2] = Internal heat emissions; Ti[C°] = Internal temperature; and n[h−1] = Air exchange rate.

Table 6. Description of some parameters of statistical area 16004 in comparison of the average value for the city of Hamburg.
Statistical area 16004 Hamburg Δ
Size of household [# individuals per household] 2.0 1.8 0.2
Share of foreign nationals [%] 45.1 13.6 31.5
Share of household with kids [%] 26.7 16.9 9.8
Share of unemployed residents*[%] 12.2 4.8 7.4
Share of single person households [%] 49.8 51.2 1.4
Number of private cars per 1000 residents [cars] 180 353 172

*Residents between the ages of 15 and 65.