Content uploaded by J. Alstan Jakubiec
Author content
All content in this area was uploaded by J. Alstan Jakubiec on May 21, 2018
Content may be subject to copyright.
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2439
https://doi.org/10.26868/25222708.2017.687
A Spatial and Temporal Framework for Analysing Daylight, Comfort, Energy and
Beyond in Conceptual Building Design
J. Alstan Jakubiec1, Max C. Doelling2, Oliver Heckmann1
1Singapore University of Technology and Design, Singapore
2Buro Happold Engineering, Berlin, Germany
Abstract
This manuscript describes a methodology for organizing
and displaying disparate sources of building performance
simulation data to form spatial displays of a myriad of
building performance outputs and drive design
understanding from an architectural point of view. The
results are based on annual simulation data, are spatially
discrete, and can be displayed and interpreted through a
variety statistical methods. A series of simple
visualizations to describe the performance of passive and
active architectural design strategies are presented in
addition to further suggestions for temporally varying
data display animations.
Introduction
The building simulation community is increasingly
generating greater amounts of raw data. This is most
obvious when the number of discrete simulations is large
as in the case of massively parametric calculations
(Lagios, Niemasz, and Reinhart 2010; Samuelson, et al.
2016) or performance optimization simulations (Attia, et
al. 2013; Nguyen, Reiter and Rigo 2014). It is less obvious
that complex thermal, lighting, or fluid dynamics
simulations produce staggering amounts of information
within a single simulation model. This is due to the
climatic, time-varying nature of the calculation (8760
hours / year), many spatial locations (sensor points) or the
plethora of possible outputs. Building performance
simulation (BPS) professionals and designers often hope
to extract much more from an individual model than
simple energy use intensity (EUI) to make intelligent
design decisions. The vision of this paper is that designers
and engineers should be able to learn as much as possible
from these complex models through intelligent spatial and
temporal data display and interaction with the results. To
achieve this vision, careful attention must be paid to the
reading, processing, interpretation and graphic display of
BPS results.
Unfortunately, the default modes of data communication
in environmental simulation graphical user interfaces
often focus on total energy in a single whole-building
number, EUI. This is sometimes broken down by energy
sources—such as electricity, gas and renewables—or by
heat transfer mechanisms: envelope conduction, solar
heat gains, ventilation, internal gains, or air conditioning
systems. Such methods are useful for assessing a design's
performance and establishing a detailed understanding of
said performance for an entire building or a single thermal
zone. However, it is extremely difficult to understand the
flows of energy, heat, air and light within a building based
on predominantly numerical, tabular or graph-based
outputs. While these outputs are very specific at a broad
level (building) or a narrow focus (a single thermal zone),
it is difficult for a designer to comprehend the relationship
between formal design features, spatial qualities and the
flow of energies throughout a building.
Architects and designers predominantly work with
graphical data to make design decisions, especially in the
early phases of the design process. Sketches and simple
drawings provide direct feedback about the appearance
and certain spatial qualities of a building in its entirety.
BPS metrics should be equally capable of this type of
graphical guidance to better support building designs
from the conception and design development stages of a
project. The authors propose that graphical
communication of BPS results should meet the following
goals,
(1) To be based upon annual, climate-specific BPS
results; therefore, to align with current best-practice
simulation methods in energy and daylight modelling.
(2) To be spatially and geometrically discrete as much as
is possible so that performance is related to specific
design features and spatial qualities.
(3) To enable navigation of performance results across
different time scales and schedules such as hourly, daily
and seasonally.
This paper details a technical approach to this data
communication problem. Fundamentally, the authors
propose a new method for how design or BPS
professionals interface with a set of complex
environmental data to drive design decisions. The authors
propose methods for creating and displaying high quality
BPS data starting from the common source of an
EnergyPlus (Crawley 2011) input data file (IDF).
Methods for displaying spatial energy, operative
temperature, thermal comfort, daylighting, and natural
ventilation at a high degree of spatial resolution are
discussed. Finally, two types of standard building
performance dashboards are suggested and prototyped—
for passive and active, mechanically conditioned
buildings. These new graphical, spatial and temporal
outputs are intended to aid environmentally responsive
passive design and low-energy building design in the
future.
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2440
Review of Communication Methods
Representation of BPS outcomes is always imperfect,
because buildings and their internal environments have no
holistic representation due to their complexity (Doelling
and Nasrollahi 2012). Vernacio and colleagues (2001)
suggest that to connect building performance simulation
outputs to design understanding and thought processes,
performance metrics must be synthesized in a way which
can address specific design dilemmas such as internal
layout, shading devices, openings, finishes, and material
selections. Agostinho (2005) hypothesized that designers
use visual and spatial representation as their prime mode
of communication and basis for reasoning; therefore, an
intuitive visual representation is key. Marsh (2004), in an
early paper on the Ecotect software, supports this
statement by suggesting that environmental performance
data is better understood when visualized in a 3D model.
Associating BPS data with 3D geometry within the design
environment, Marsh (2004) states, enables environmental
analysis to drive a design process.
Chen (2004) notes that a visual connection to a design's
performance is desirable for designers. In airflow
analysis, this often takes the form of 'smart arrows' that
indicate a presumed ventilation condition as a vector.
Chen continues, stating that such arrow-based
presumptions can be completely inaccurate; therefore,
visual and evidence-based methods are necessary to
meaningfully impact the design. Malkawi and Srinivasan
(2005) took the disp lay of such spati al information data to
an extreme by producing a virtual augmented reality
environment to explore computational fluid dynamics
(CFD) results. A similar process is often followed in
daylighting design; however, the vectors drawn are more
often representative of real solar angles to design shading
against direct sunlight or to map the depth of direct light
penetration into a space (DeKay and Brown 2013).
Thermal analysis interfaces predominantly make use of
graph-based displays which do not provide a visual
connection to the design of a space, although tools such
as Honeybee (Sadeghipour, Pak and Smith 2013) and
Archsim (Dogan 2016) allow spatial performance metrics
to be mapped at the thermal zone level. Honeybee can
also perform spatial thermal comfort calculations based
on air stratification models, view factors to surrounding
surfaces, and radiant adjustment for direct solar access
(Mackey 2015).
CFD simulations by their nature produce spatialized BPS
results for fluid flows, comfort, and thermal measures.
Unfortunately, CFD simulations are, at this moment, too
time-consuming to make annual simulations practical in
practice. The development of faster strategies for
simulating annual CFD outputs is an active field (Wang
and Malkawi 2015).
Several attempts have been made to synthesize BPS
information in a spatial manner. Reinhart and Wienold
(2011) proposed a dashboard view of simulation results
that provides comprehensive information for a perimeter-
space 'shoebox' model on daylighting and shade
operation, visual comfort, view, EUI, operation costs, and
carbon emissions. Their analysis was focused on the
impacts of daylighting on energy use in perimeter spaces
and cannot be easily extended to entire buildings. Sustain
(Greenberg, et al. 2013) is a private tool to display spatial
and temporal data with a focus on designer understanding
and ease of navigating large, parametric datasets.
Doelling (2014) developed a tool to explore thermal and
daylighting measures spatially based on a co-display of
thermal results from EnergyPlus (Crawley, et al. 2001)
and daylighting results from Radiance/Daysim (Reinhart
and Walkenhorst 2001) simulation results.
Figure 1: Flow chart of the methods described in this paper to translate EnergyPlus IDF input into spatialized output
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2441
Methodology for Spatial Performance
Metric Communication
This section presents a computational process through
which BPS data is created, organized, and displayed
beginning from the EnergyPlus IDF format for multizone
thermal building models. Spatialized thermal comfort and
annual climate-based daylighting metrics are created and
displayed across a grid of sensor points. In addition, every
EnergyPlus zone or surface-level output can be displayed
graphically overlaid on a building plan or 3D model. An
overview of the process can be seen in Figure 1.
Processing IDFs into Daylight and MRT Models
An IDF file is read in by a simple, command-line tool to
create Radiance format (Ward 1994) files for view factor,
solar adjusted mean radiant temperature (MRT)
calculations, and for annual climate-based daylighting
measures. Each object in the IDF is stripped of white
space, new lines, comment characters (!) and subsequent
comments. Surface material properties are recorded from
the WindowMaterial:Glazing, WindowMaterial:
SimpleGlazingSystem, Material and Material:
NoMass objects. Emissivity, visible absorptance, visible
transmittance (Tvis) and solar heat gain coefficients
(SHGC) are recorded for each material object. For each
Construction object, the reflectance of the front and
back surface materials are recorded, and from the
WindowConstruction object, the combined Tvis is
calculated by multiplying the values of all glazing layers
in the construction.
Opaque and transmissive surface objects (Building-
Surface:Detailed, FenestrationSurface:Detailed,
and Shading:Building:Detailed) are recorded based
on their construction materials and the coplanar points
defining their polygonal shape; however, transitioning
from an IDF representation of a thermal model to a
daylighting model, there are two major geometric issues
to overcome: (1) windows within opaque surfaces are
represented by two co-planar polygons, parent and child,
without an explicit hole in the parent polygon through
which light could pass, and (2) surfaces are infinitely thin.
When calculating view factors for MRT calculation,
another (3) issue is that the surfaces between two thermal
zones are coplanar; therefore, it is difficult to separate the
inside or outside faces of interior partition walls and
internal floors. To solve the first problem, windows are
treated as holes in wall surfaces. Figure 2 illustrates these
changes graphically. A Delaunay triangulation algorithm
(Shewchuk 1996) is used to triangulate the gaps while
maintaining an open area for window surfaces to occupy.
Regarding the second issue, utilizing infinitely thin
surfaces rather than true volumetric representations of
geometry can result in overpredicting daylight levels
within a space (Ibarra and Reinhart 2009). The authors
opted to apply a 0.8 reduction factor to account for
mullions and a 0.8 reduction factor to account for the
depth of walls. A 0.6 Tvis window transmits 0.6 · 0.82 =
0.384 percent of light to account for the lack of geometric
specificity in EnergyPlus models. Finally, to address the
third issue of coplanar surfaces, in the MRT view factor
calculation models: all floors are translated upwards by
0.5 mm, all ceilings are translated downwards by 0.5 mm,
and all walls are translated 0.5 mm opposite of the surface
normal direction. By slightly moving the surfaces in this
way, there is no ambiguity between whether a specific
spatial location has a view to the front or back of any
surface, which will have different temperatures.
Sensor grids over which to calculate MRT, thermal
comfort, and daylighting metrics are automatically
generated 0.85 m above each floor plane in the thermal
model at a user-settable uniform distance; however,
custom grids can also be defined. Sensors are also
associated with the thermal zone (or room) name they are
contained within to be associated with data useful in
calculating thermal comfort such as relative humidity and
dry bulb air temperature.
Performance Simulation Engines
Annual daylighting simulations are calculated using the
Daysim (Reinhart and Walkenhorst 2001) engine, which
employs a daylight coefficient method calculating the
contributions of 145 portions of the diffuse sky, 3 areas of
ground reflectance, and approximately 65 direct solar
positions. Hourly illuminance is predicted based on the
Perez sky model paired with climatic datasets. Active
shading systems such as blinds or electrochromic glazing
are not accounted for in the calculations; therefore, the
outputs are meant to be interpreted based on the potential
for useful daylight and excessive light exposure as per
contemporary standards and metrics such as Annual
Sunlight Exposure (IESNA 2012) and Useful Daylight
Illuminances (Mardaljevic, et al. 2012). To ensure
adequate distribution of light, Radiance/Daysim
simulation parameters are set as follows: ab=7, ad=2048,
ar=2048, as=512, aa=0.1, lw=0.0001.
EnergyPlus simulations are run with hourly output
utilizing the same climate data, and are forced to include
the following outputs necessary for MRT and thermal
comfort analysis in addition to those requested by the
simulator: Surface Inside Face Temperature, Zone
Air Temperature, Zone Air Relative Humidity, and
Zone Air Humidity Ratio. If an airflo w network (AFN )
(Walton 1989) is used to calculate natural ventilation, the
following outputs are also included such that volumetric
flows throughout openings and thermal zones can be
accounted for: AFN Zone Infiltration Volume, AFN
Zone Mixing Volume, AFN Linkage Node 1 to Node 2
Volume Flow Rate, and AFN Linkage Node 2 to Node
1 Volume Flow Rate.
In this manuscript, all performance simulations for
illustrative purposes utilize the Singaporean IWEC
weather file and the conceptual model diagrammed in
Figure 3. This model is based on the design of a retrofitted
house by Richard Hyde in 1998 located in Brisbane,
Australia (Hyde 2000) translated into a high-rise design.
The design is built in EnergyPlus using uninsulated
concrete floors, ceilings and walls as per typical
Singaporean constructions. Windows are single pane, 3
mm glass with a solar control coating applied
(SHGC=0.288; Tvis=0.65). Adiabatic adjacency
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2442
properties are applied to ceilings, floors and shared walls.
Schedules and internal loads are based on the Building
America House Simulation Protocols (Hendron and
Engebrecht 2010). Free-running variants of the design
with mechanical conditioning systems use an AFN with
windows open when outdoor temperatures are below 28.5
C. Air-conditioned variants of the design maintain a
sealed building envelope and are conditioned to 26 C
when a space is occupied.
Figure 3: Conceptual model of the building used for
data display examples in this manuscript.
Spatialization of Radiant and Comfort Information
Spatialization of thermal comfort results is valuable as it
can be directly related to other intuitive spatial measures
such as annual daylighting to better intuit building
performance. To achieve this, mean radiant temperature
(MRT) in the authors’ proposed framework is calculated
relative to surrounding surface temperatures before being
adjusted for direct solar heat gains for each sensor grid in
the analysis. The steps involved in this process are
documented visually in Figure 2C–E. As described
previously, thermal geometry and material properties are
read in from an IDF file (2A), coincident wall surfaces are
clipped by the windows, and fronts and backs are slightly
offset to avoid view collisions—resulting in a Radiance
format model (Ward 1994) (2B). Next, view factors are
calculated between each sensor point in the grid and all
surrounding building surfaces using a set of 2562 rays
distributed in an equal solid angle manner (2C). This is
achieved using the Radiance rtrace command with the
-oM option to report the material name of the surface at
each ray’s first intersection for every sensor point.
Finally, hourly inside surface temperature results from the
EnergyPlus simulation are paired with the view factor
calculations to derive hourly MRT (2D), and direct solar
calculations are used to derive a solar adjusted mean
radiant temperature at each hour attenuated by the solar
heat gain coefficient and angle of transmission through
the glazing (2E). The calculation of hourly solar adjusted
MRT takes the form in Equation 1.
0.250.7E1
T,
(1)
where E is hourly direct irradiation arriving at a sensor
(W/m2); is the Stefan–Boltzmann constant
(5.670367×10−8 W/m2K4); is the view factor (percent)
from a sensor location to an individual surface, and
is the hourly surface temperature (deg. K). 0.25 is a seated
projection factor and 0.7 is a typical solar absorption
factor for human skin. It should be noted that for sensors
not receiving direct solar irradiation (E0), the MRT
will simply be a function of the surrounding surface
temperatures. Once MRT is spatially determined in this
manner, it can be paired with associated thermal zone
temperature and humidity levels to predict thermal
comfort measures such as PMV, PPD, SET, and the
ASHRAE 55 standards. The authors utilize a translation
of the UC Berkeley CBE comfort tool calculations (Hoyt,
et al. 2013) for this purpose.
Annual-Spatial Data Structures and Data Display
Any annual data set of 8,760 hours is interpreted as a
conceptual ‘zone,’ which is derived from the thermal
modelling term relating performance measures with a
spatial component—typically an air volume; however, the
authors do not limit the zone concept to this spatial unit.
Zones can be a single point in space, a vector, a surface,
or an air volume (an actual thermal zone). Annual
performance data is required, because it facilitates a
holistic understanding of performance across a complete
weather cycle. For example, point zones can contain
thermal comfort or climate-based daylighting
(a) Read thermal model
geometry and material
properties from IDF
file
(b) Clip coincident
walls from window
surfaces for daylight or
view factor analysis
(c) Calculate view
factors from a sensor
grid to every surface in
the thermal model
(d) Derive hourly
radiant temperature
distribution from
thermal results and
view factors
(e) Adjust radiant
temperatures based on
hourly direct solar heat
gain adjustments
Figure 2: Diagram of the calculation steps involved in spatializing radiant temperature results for TCA calculation
across a sensor grid.
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2443
information; vector zones can contain visual comfort
information; surface zones can contain information about
heat gains, losses, temperatures and volumetric air flows;
and volumetric zones can contain air psychrometric
properties, energy use information or any EnergyPlus
zone-level output. Disparate simulation results from
thermal, ventilation and daylight performance are
organized within a single framework using the zone
concept.
Once organized in the zonal geometric and annual data
framework, results can be navigated and displayed in a
variety of ways. Outputs can be quantified for display
using several methods to assess temporal data:
Average—the mean value during a specified time
interval.
Frequency—the percentage of time a value with a
certain range is achieved.
Sum—the total of values over a specified time interval.
An example of these display methods is shown in Figure
4 using an annual thermal model for a free-running,
natural ventilated version of the building described
previously. Figure 4A displays the annual average of
mean radiant temperature for all 8760 hours in the year,
and 4B displays the frequency of operative temperatures
below 28.5 C. Finally, 4C illustrates the sum of zone-level
solar hear gains normalized per floor area.
Because ‘zone’ data sets are comprehensive annually,
outputs can be calculated and visualized across multiple
time scales and schedules. It is therefore possible to
switch between different seasonal, hourly, monthly, time-
of-day, and annual outputs to enhance understanding of
the design’s performance. In contrast to Figure 4, Figure
5 illustrates thermal performance across a single, sunny
day by displaying room-level normalized solar heat gains
(solid color of the floor) and spatialized operative
temperature calculations (colored dots) on an hourly
basis. Starting at 8:00 until 17:00, direct solar gains can
be seen to locally increase operative temperature in sunlit
areas. During these times, the northern LIVING_1 and
LIVING_2 thermal zones are negatively impacted by
excessive solar gains.
Likewise, the benefits of external shading and the
moderate glazed areas of BEDROOM_1, BEDROOM_2, and
TOILET_1 are visible. These spaces cool down most
quickly in the evening (from 18:00 to 21:00 in Figure 5),
and begin to heat up slightly as occupants utilize the space
for sleeping. Relatedly, generally lower operative
temperatures can be seen (4B) in these spaces throughout
the year, and higher average radiant temperatures can be
seen near unshaded windows than near more shaded
windows (4A). While these outputs are not
comprehensive and are mainly intended to illustrate
different forms of temporal and spatial data
communication, they show the potential for
communication methods to enable better understanding of
the performance of a design.
Suggested Performance Metrics
Passive, Free-Running Designs
In passive architectural design concepts, three issues are
typically under consideration: (1) daylighting, (2) natural
ventilation and (3) thermal comfort. Appropriate display
of performance metrics within the challenge of a passive
design should account for these three considerations
across space and time as well as for when they are within
a beneficial range or exceed established values.
A. Temporal average data display.
B. Frequency data display.
C. Sum data display.
Figure 4: Depictions of three types of temporal and
spatial thermal model output displays.
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2444
Figure 5: Hourly data display of predicted operative temperature and solar heat gains
from 7AM until 11PM on 10 July.
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2445
Daylighting results should be visualized as the co-display
of two metri cs to account for be neficial lighti ng as well as
overlighting. The authors suggest to use UDIa
300 lx–3000 lx
and UDIe
3000 lx
. UDIa
300 lx–3000 lx is
selected as it
approximately matches a lighting threshold that relates to
human perception of what is daylit (Reinhart, Rakha and
Weissman 2014) but also discounts excessive daylight.
UDIa
300 lx–3000 lx
can be displayed on a scale ranging from
0 to 100% of occupied, daylit hours in the year (8am–
6pm), and values of over 50% identify the daylit area.
UDIe
3000 lx
on the other hand identifies overlit areas that
indicate increased potential for glare and visual
discomfort (Mardaljevic, et al. 2012). Areas with a
UDIe
3000 lx
value greater than 15% of occupied hours are
colored pink to indicate undesirable visual conditions and
potential for excessive solar heat gains.
Ventilation information on an annual, hourly basis can be
calculated using the EnergyPlus airflow network (AFN)
model (Walton 1989). The AFN is a simplified model for
the prediction of bulk airflow rates throughout an
architectural design based on pressures at node outlets and
a series of linkages (windows, doors, grills, passages,
etc.). It is capable of accurately calculating total
volumetric airflow across openings and spaces; however,
it is not capable of detailed air velocity and distribution
information such as is found in CFD calculations. An
AFN is chosen because it is computationally limiting at
this time to generate enough CFD results in order to
populate an annual spatial zone as described in this
manuscript. Ventilation results therefore are only spatially
localized at the thermal zone and surface (opening) level.
At the thermal zone level this can be visualized as average
annual air change rates per hour (ACH). ACH represents
the number of times air is replaced in a space each hour
by fresh outdoor air due to leakage or intentional
ventilation. In addition, the average annual velocity of air
passing through windows while open can be displayed as
a surface output.
Mackey’s (2015) concept of TCA is selected to display
thermal comfort information, because it presents thermal
comfort in the same terms and using the same climate data
as climate-based daylighting metrics such as UDI. TCA
represents the percentage of hours in a year where thermal
comfort is achieved based on a specified comfort model.
The authors recommend to calculate and display TCA
(Mackey 2015) using the ASHRAE standard 55 adaptive
thermal comfort model 90% acceptance threshold
(DeDear, et al. 1998). Are as comfortable less than 35% of
time in the year are colored pink to indicate unacceptable
comfort, while values greater than or equal to 75% are
considered good. This value scale has been derived for the
Singaporean climate; however, such values may be
tailored based on other climates’ potential for passive
thermal comfort.
Fig ure 6 illustrates a concept for an at-a-gl ance das hboard
combining these three, passive metrics. Several
performance issues and successes become immediately
clear to a theoretical designer working with this building
design. The northern living spaces (#1 & #2) receive a
large amount of overlighting relative to their floor area,
visible from the pink coloration on the daylight display.
The ventilation concept of a connected core works well,
receiving annual average ACH rates of 50 or above;
however, the side bedrooms and bathrooms do not
achieve as much ventilation from the schema and could
perhaps benefit from greater connectivity to the main
living spaces. The dining room and kitchen is therefore
extremely well ventilated while having little direct
connection to the outdoors and daylight. A moderately
ventilated room at the building periphery without large
relative areas of overlighting performs the best across this
design, achieving TCA between 60 % and 70 %.
Active, Mechanically-Conditioned Designs
For mechanically-conditioned designs, design
investigations focus on energy reduction rather than
completely passive strategies; therefore, the authors
suggest that natural ventilation and thermal comfort are
not as critical when ventilation is handled mechanically
and when active systems ensure thermal comfort. Annual
EUI values suddenly become the most important results
to assess the success of an active design. The display of
thermal gains from external factors is also sensible, from:
windows (as in Figure 4C), infiltration, conduction, etc.
Interior loads due to lighting, equipment and occupancy
also id entify loc at ions that a re success ful v ersus th os e that
contribute to a higher overall energy utilization. These
outputs are plotted in Figure 7 for an actively conditioned
version of the design model.
Figure 7: Mechanically conditioned metric outputs
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2446
The display of daylighting and natural ventilation
information is still worthwhile for actively conditioned
buildings, especially if they are conditioned using a
mixed-mode system that relies on passive strategies some
of the time.
Discussion
The preceding sections describe a methodology for
organizing and displaying disparate sources of BPS data
to form spatial displays of building performance outputs
and drive design understanding from an architectural
point of view. The results are based on annual simulation
data, are spatially discrete, and can be displayed and
interpreted through a variety of temporal and statistical
methods through the utilization of the spatial ‘zone’ data
structure. A series of simple visualizations to describe the
total performance of a passive architectural design
strategies (Figure 6) was presented in addition to
suggestions for data display of buildings with active
conditioning systems. This section discusses some of the
impacts and limitations of this methodology.
Figure 6: Proposed passive dashboard view illustrating spatial ventilation, thermal comfort, and daylighting results
based on annual thermal, MRT and daylight performance simulations.
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2447
Relevance and Release of Tool
As covered in the introduction and review of
communication methods sections, spatial data
communication is important to enable intuitive design
understanding. The methods outlined in this paper are
implemented into a Grasshopper / Rhinoceros 3D plugin
which takes as an input a single IDF file. Therefore, the
goal is to create a large amount of actionable
environmental data while minimizing the need for
specialist modelling expertise. Referencing Figure 1, the
simulation model can be created in any GUI which
supports EnergyPlus such as DesignBuilder, DIVA-for-
Rhino, Honeybee, OpenStudio, or Simergy. The model
processing and generation of spatial data steps in Figure 1
are made possible through standalone Python programs
which are independent of any paid software. Only the
visualization and display framework reported herein is
tied to a specific commercial software package—
McNeel’s Grasshopper environment for Rhinoceros 3D.
An attempt has been made at creating a standardized
results format which may be in the future readable by
other display frameworks such as web-based tools or
other building performance simulation GUI’s.
The tool is based upon the Mr. Comfy plugin and will be
released simultaneously with the publication of this paper.
Users will be able to use spatially mapped thermal
outputs, thermal comfort, daylight and ventilation to
better comprehend, communicate and respond to a
design’s performance.
As with any method, time and effort required are
important to consider when assessing impact potential. To
generate the spatially specific thermal comfort and
daylight results depicted in this paper, the simulation time
was approximately 11 seconds per sensor using a single-
core of a 2.4 GHz processor. This equated to a simulation
time of approximately 3 and 1/2 hours for the display
shown in Figure 6. For rapid design iterations, the density
of sensors can be reduced, significantly lowering the time
required. Effort in preparing the simulations and
visualizations is often a larger concern than calculation
time, as it takes direct human effort to prepare multiple
simulation models rather than computer effort. Because
the tool presented herein generates data from a single
energy modelling input, the intent is to minimize time
spent building separate models. In addition, all the
building performance outputs displayed in this paper were
completely generated using the tool, sans-editing with the
exception of the axonometric building graphic at the
bottom of Figure 6, the addition of North arrows and scale
bars, and the CFD results displayed in Figure 8.
Limitations
One serious limitation of the tool is that AFN’s in
EnergyPlus do not actually allow for the extrapolation of
spatially localized air velocity within spaces, only bulk
volumetric flow through an air volume. CFD calculations
would allow a deeper level of understanding of the
ventilation performance of a space; however, as
mentioned earlier, this is computationally infeasible,
because a significant number of individual CFD
calculations would be needed in order to fill the ‘zone’
data structure with 8760 hours of data. While recognizing
this limitation, the authors did test a typical wind situation
using CFD for the test building, which is depicted in
Figure 8 below. It is compared against the annual average
ACH data shown earlier. This is not a direct comparison
because the result is annualized data (ACH) versus point-
in-time data (m/s); however, it seems clear that ventilation
trends do match between the two calculations. The
authors look forward to a time where annualized CFD
calculations are easily possible.
Acknowledgement
The authors acknowledge financial support from the
SUTD-MIT International Design Center (IDC) under
project no. IDG21500108. Any opinions expressed herein
are those of the authors and do not reflect the views of the
IDC.
Figure 8: Comparison Between a Single CFD Result and
Annual AFN Ventilation Simulation
References
Agostinho, FS. (2005). Architecture as Drawing,
Perception and Cognition. In Digital Design: The
Quest for New Paradigms, proceedings of the 23rd
Conference on Education in Computer Aided
Architectural Design in Europe, Instituto Superior
Tecnico, Lisbon (pp. 83-90).
Attia, S., et al. (2013). Assessing gaps and needs for
integrating building performance optimization tools
in net zero energy buildings design. Energy and
Buildings, 60, 110-124.
Chen, QY. (2004). Using computational tools to factor
wind into architectural environment design. Energy
and buildings, 36(12), 1197-1209.
Crawley, DB, Lawrie, LK, Winkelmann, FC, Buhl, WF,
Huang, YJ, Pedersen, CO, and Glazer, J. (2001).
EnergyPlus: creating a new-generation building
energy simulation program. Energy and buildings,
33(4), 319-331.
De Dear, RJ., Brager, GS., Reardon, J., & Nicol, F.
(1998). Developing an adaptive model of thermal
comfort and preference/Discussion. ASHRAE
transactions, 104, 145. IBPSA Board. Journal of
Building Performance Simulation 8(2), 39–43.
Proceedings of the 15th IBPSA Conference
San Francisco, CA, USA, Aug. 7-9, 2017
2448
DeKay, Mark, and GZ. Brown. Sun, wind, and light:
Architectural design strategies. John Wiley & Sons,
2013.
Doelling, M., & Nasrollahi, F. (2012). Building
Performance Modeling in Non-Simplified
Architectural Design. In Proceedings of the 30th
eCAADe Conference, Prague, Czech Republic.
Doelling, MC. (2014). Space-based thermal metrics
mapping for conceptual low-energy architectural
design. In Proceedings of BSO2014.
Greenberg, D, Pratt, K, Hencey, B, Jones, N, Schumann,
L, Dobbs, J, and Walter, B. (2013). Sustain: An
experimental test bed for building energy simulation.
Energy and Buildings, 58, 44-57.
Hendron, R and Engebrecht, C. (2010). Building America
house simulation protocols. National Renewable
Energy Laboratory,
http://www.nrel.gov/docs/fy11osti/49246.pdf.
Hoyt, T, Schiavon, S, Piccioli, A, Moon, D, and Steinfeld,
K. (2013). CBE thermal comfort tool. Center for the
Built Environment, University of California
Berkeley, http://cbe.berkeley. edu/comforttool.
Hyde, R. (2000). Climatic Responsive Design. E & FN
Spon Publications, New York, 97-112.
Ibarra, D, & Reinhart, CF. (2009, July). Daylight factor
simulations–how close do simulation beginners
‘really’get?. In Building Simulation (Vol. 196, p.
203).
IESNA. (2012). LM-83-12 IES Spatial Daylight
Autonomy (sDA) and Annual Sunlight Exposure
(ASE). New York, NY, USA: IESNA Lighting
Measurement.
Lagios, K, Niemasz, J, and Reinhart, CF. "Animated
building performance simulation (abps)-linking
rhinoceros/grasshopper with radiance/daysim."
Proceedings of SimBuild (2010).
Mackey, CCW. (2015). Pan climatic humans: shaping
thermal habits in an unconditioned society (Masters
dissertation, Massachusetts Institute of Technology).
Malkawi, AM, & Srinivasan, RS. (2005). A new
paradigm for Human-Building Interaction: the use of
CFD and Augmented Reality. Automation in
Construction, 14(1), 71-84.
Mardaljevic, J, Andersen, M, Roy, N, & Christoffersen, J.
(2012). Daylighting metrics: is there a relation
between useful daylight illuminance and daylight
glare probability. In Proceedings of the building
simulation and optimization conference (BSO12),
Loughborough, UK (Vol. 1011).
Marsh, A, (2004). Performance Analysis and Concept
Design: The Parallel Needs of Classroom and Office.
In Proceedings of Between Research and Practice
Conference.
Nguyen, AT, Reiter, S, and Philippe, R. "A review on
simulation-based optimization methods applied to
building performance analysis." Applied Energy 113
(2014): 1043-1058.
Reinhart, CF and Walkenhorst, O. (2001). Validation of
dynamic RADIANCE-based daylight simulations for
a test office with external blinds. Energy and
buildings, 33(7), 683-697.
Reinhart, CF and Wienold, J. (2011). The daylighting
dashboard–A simulation-based design analysis for
daylit spaces. Building and environment, 46(2), 386-
396.
Roudsari, MS, Pak, M, and Smith, A. (2013). Ladybug: a
parametric environmental plugin for grasshopper to
help designers create an environmentally-conscious
design. In Proceedings of the 13th International
IBPSA Conference Held in Lyon, France Aug.
Samuelson, H, Claussnitzer, S, Goyal, A, Chen, Y, and
Romo-Castillo, A. (2016). Parametric energy
simulation in early design: High-rise residential
buildings in urban contexts. Building and
Environment, 101, 19-31.
Shewchuk, JR. (1996). Triangle: Engineering a 2D
quality mesh generator and Delaunay triangulator. In
Applied computational geometry towards geometric
engineering (pp. 203-222). Springer Berlin
Heidelberg.
Venancio, R, Pedrini, A, Van der Linden, AC, Van den
Ham, ER, and Stouffs, RMF. (2011, November).
Think designerly! Using multiple simulation tools to
solve architectural dilemmas. In Proceedings of
Building Simulation 2011, BS 2011: 12th
International Conference of the International
Building Performance Simulation Association,
Sydney, Australia, 14-16 November 2011.
Walton, GN. (1989). AIRNET: a computer program for
building airflow network modeling. Gaithersburg,
USA: National Institute of Standards and
Technology.
Wang, Y, & Malkawi, A. (2014). Annual hourly CFD
simulation: New approach—An efficient scheduling
algorithm for fast iteration convergence. In Building
Simulation (Vol. 7, No. 4, pp. 401-415). Tsinghua
University Press.
Ward, GJ. (1994). The RADIANCE lighting simulation
and rendering system. In Proceedings of the 21st
annual conference on Computer graphics and
interactive techniques (pp. 459-472). ACM.
