Methods for integrating parametric design with

building performance analysis

1
Ajla Aksamija, PhD, LEED AP BD+C, CDT
1
University of Massachusetts Amherst, Amherst, MA

ABSTRACT: This paper discusses methods for integrating parametric design with building performance
analysis procedures, specifically presenting tools and design methodologies that are suitable for whole
building design. In this research, an ideal framework for integration of parametric and performance analysis
procedures was developed. Then, the framework was tested using existing software applications, including
building information modeling (BIM), non-BIM, parametric design and building performance analysis
applications. Current applications that can integrate some form of building performance simulation with
parametric modelling include Rhino 3D (non-BIM), Revit (BIM), and SketchUp (non-BIM). Revit and Rhino
each have visual programming plugins to aid in the creation of parametric forms. In this research, three
different workflows were tested. Specifically, Honeybee and Ladybug (for Rhino 3D), Insight 360 (for Revit)
and Sefaira (for Revit) were evaluated. A case study building was used to test and evaluate the workflows,
interoperability, modeling strategies and results. Three different building performance aspects were analyzed
for each workflow: 1) energy modeling, 2) solar radiation analysis, and 3) daylighting. Simulation results from
energy modeling, solar radiation and daylight simulations were recorded and analyzed. However, besides
simulation results, the paper compares modeling procedures, parametric capabilities of investigated
applications, ease of integration and interoperability. The results show a promising course for integrating
parametric design with building performance simulations.

KEYWORDS: Parametric design, building performance analysis, high-performance buildings, BIM, energy-
efficiency

INTRODUCTION
Advances in building performance simulations have enabled designers to better understand how
environmental factors affect building performance (Aksamija, 2013). Parametric design methods, on the other
hand, allow designers to generate and explore geometries of building elements by manipulating certain
parameters. There are a number of software platforms that focus individually on environmental analysis or
parametric design, but few integrate both. Common parametric design tools include Grasshopper (Rhino 3D),
Dynamo (Autodesk Revit) and GenerativeComponents (Bentley MicroStation). Environmental and energy
analysis tools include Ladybug and Honeybee (Rhino 3D), DIVA (Rhino 3D), Insight 360 and Green Building
Studio (Autodesk Revit), Sefaira (Autodesk Revit + Trimble SketchUp), Radiance, OpenStudio, EnergyPlus,
eQuest, DesignBuilder, IES VE, and many others. The most common method of integrating building
performance simulations (BPS) into early design work has been by exporting the design model (whole building,
partial building, or model of a building component) to a dedicated analysis tool to generate an analysis model,
assign inputs necessary for calculations and simulate energy usage, daylighting, or solar radiation. The results
from these simulations would be used to adjust the design model in the original design program, and then the
model would be exported again, thus repeating the process. By integrating the capabilities of parametric
design and building performance simulations, multiple design variables could be tested rapidly, creating a
more cohesive design process. These following research objectives were addressed in this study:
• Investigation of methods for integrating performance-based design with parametric modelling,
focusing on whole building design.
• Investigation of tools and software programs that can seamlessly integrate performance-based
design with parametric modelling, particularly focusing on energy analysis, solar radiation analysis
and daylight simulations.
• Testing the procedures on a specific case study and documenting the results.

1.0 BRIEF LITERATURE REVIEW

Typical energy modeling programs are often complex for architects to use during the early stages of design,
resulting in building performance analysis being performed at later stages (Bazjanac 2008; Schlueter and
Thesseling 2009). The architectural profession lacks established methodologies and protocols that incorporate
performative analyses into the early stages of design (Pratt and Bosworth, 2011). However, the most important
design decisions that have significant impacts on building performance are made at the conceptual stage of
a project, such as building massing, orientation, volume, shading, daylight strategies, etc. Tools that shift the
building performance assessment back into conceptual stages of design will have a bigger effect on building
performance (Rahmani et al., 2013). This introduces the concept of performance-based design, where the
environmental performance becomes the guiding factor in the design process. However, most current design
software applications are not capable of integrating the results from performance-based simulations back into
the design model. It is the designer who interprets the results and optimizes the model based on simulation
results (Oxman 2008).

Parametric modeling relies on geometric representation of a design with components and attributes that can
be parametrically varied, where each geometric configuration that derives from parametric variations is called
an instance. Instances represent a unique set of transformations based on parametric inputs, generating
design variations and different configurations (Turrin et al., 2011). Parametric modeling has the potential to
overcome current design process limitations and to facilitate the revelation and comparison of performative
solutions. Parametric modeling initially lacked applications in architectural design; however, new architectural
tools have been developed, allowing for new directions. The ability to produce many instances that result in
unique configurations of the same geometric component is the main advantage of parametric modeling.

Integrating parametric modeling with building performance analysis procedures could enable architects and
designers to analyze impacts of design decisions on building performance from the earliest stages of the
design. Testing multiple design strategies in an efficient way, and reducing the time necessary for modeling
and analysis procedures are the main benefits of integrated performance-based and parametric design.

2.0 FRAMEWORK FOR INTEGRATION OF PERFORMANCE-BASED AND PARAMETRIC

DESIGN METHODS

The ideal framework for integrating parametric and performance-based design is shown in Figure 1.
Parametric modelling, geometric preparation and analysis preparation should be streamlined and connected
to performance analysis. This would combine parametric control of building geometry and building systems
with analysis and visualization of results.

Figure 1: Ideal framework for parametric and performance-based design. Source: (Author 2017)

Current software applications that integrate some form of BPS with parametric modelling include Rhino 3D,
Revit, and SketchUp, as shown in Figure 2. Revit is a BIM-based design tool, while Rhino and SketchUp are
non-BIM based. Revit and Rhino each have visual programming plugins to aid in the creation of parametric
forms. And, a variety of BPS tools are available that address different aspects of performance analysis,
including solar radiation analysis, energy modeling and daylight analysis. This research investigated three
different workflows, including integration of both BIM and non-BIM design platforms with parametric modeling
and BPS. Specifically, Rhino 3D (with Grasshopper, Honeybee and Ladybug plugins) and Revit with Insight
360 and Sefaira workflows were investigated in this research. Figure 2 shows details of all three investigated
workflows, and relationships among software applications. The parametric and performance tools discussed
are run within Rhino (Ladybug and Honeybee) or Revit (Insight 360 and Sefaira). Honeybee and Sefaira use
the same simulation engines for running analysis (Daysim, Radiance, EnergyPlus and OpenStudio). Insight
360 also uses EnergyPlus for energy analysis, but Revit has its own engine for lighting simulations, which has
been validated against Radiance. Since all plugins use EnergyPlus as the energy simulation engine, any
variations in results are caused by geometry and input differences between the three plugins.
 Figure 2: BPS and parametric workflows for Rhino, Revit and SketchUp. The workflows and software applications
explored in this research are shown in red. Source: (Author 2017)

The next section reviews a case study, which was used to evaluate this framework.

3.0 EVALUATION OF THE FRAMEWORK AND CASE STUDY

A prototype building located in the financial district of Boston was used as a case study to evaluate the above
discussed framework, and the building site is shown in Figure 3a. The evaluated workflows included Honeybee
and Ladybug (Rhino 3D), Insight 360 (Revit) and Sefaira (Revit) software applications. The building is divided
into a low-rise portion of five stories, and a high-rise portion of 15 stories. Floor-to-floor height was 3.7 m (12
ft), with a total height of 73.2 m (240 ft). The perimeter zone depth was 14.6 m (48 ft). The surrounding
buildings were modelled from GIS data, and were included in the site model. Figure 3b shows Rhino model
with surrounding buildings, and Figure 3c shows Revit model. Energy, solar radiation, and daylight analysis
were run for the case study building using the three investigated workflows.
 a)
a) b) c)

Figure 3: a) Case study building site outlined in red; b) Rhino model of the case study building, with surrounding buildings;
and c) Revit model of the case study building, with surrounding buildings. Source: (Author 2017)

3.1 Rhino, Grasshopper, Ladybug and Honeybee Workflow

Figure 4 shows the Grasshopper definition used for the analysis and simulations within Rhino. Due to the
graphic nature of Grasshopper, the organization of components is vital for understanding the user's own
definition. As more components and connections are added to the canvas, the file becomes increasingly
visually complex. The components are divided into four stages: geometry preparation, analysis preparation,
analysis simulation and visualization of results.

Figure 4: Grasshopper definition used for the analysis. Source: (Author 2017)

Honeybee exports model geometry and settings to EnergyPlus, which performs the energy simulation. The
simulation options that were parametrically controlled in Honeybee included window-to-wall ratio (WWR),
temperature set-points, wall and roof R-values, glazing U-values, SHGC and Vt, infiltration rates, HVAC
systems, and lighting power density. Since the building geometry was complex, each run ranged from taking
20 to 60 minutes to complete. The number of windows had a direct impact on simulation time. Table 1 shows
properties of investigated models and respective EUI results.

Table 1: Energy simulation results (Rhino and Honeybee).

2 2
Run Name Variables EUI kWh/m (kBTU/ft )

Baseline 50% WWR; VAV w/ Reheat, 72°F/78°F Set-points; Wall: R-19; 216.7 (68.7)
Roof: R-30; Window: .45/.38/.42 (U/SHGC/VT); Infiltration: .8
2
ACH; Lighting power density: 10.54 w/m

Low WWR 30% WWR 205.1 (65)

High WWR 80% WWR 237.8 (75.4)

Setpoints 68°F/82°F Set-points 161 (51)

Higher R-Values Wall: R-40; Roof: R-60; Window: .2/.38/.42 (U/SHGC/VT) 185.7 (58.9)

Infiltration Infiltration: .2 ACH 129.5 (41)

 2
Lighting Power Lighting power density: 3 w/m 212.4 (67.3)
Density

HVAC Alternate Fan coil units + DOAS 127.3 (40.4)

Best Case 50% WWR; Wall: R-40; Roof: R-60; Window: .2/.38/.42 45.3 (14.4)
(U/SHGC/VT); Fan coil units + DOAS, 68°F/82°F Set-points;
2
Infiltration: .2 ACH; Lighting power density: 3 w/m

The Ladybug Radiation Analysis Tool uses the location of the sun for every hour of the year to determine how
much radiation the exterior surfaces receive. Surrounding buildings are taken into account during this analysis.
Simulation options that were parametrically controlled in Honeybee included analysis period, sky type, grid
size, grid distance off surface, and legend (low and high bound). Results are shown in Figure 5a. Honeybee
exports model geometry and settings to Radiance, which performs daylight simulations. The simulation
options that were parametrically controlled in Honeybee included analysis period, sky type, grid size and
distance off surface, radiance rendering parameters, and simulation type (illuminance, radiation, luminance).
Typical results are shown in Figure 5b.

a) b)

Figure 5: a) Solar radiation analysis results (cumulative, June 21); b) Daylight simulation results (June 21, WWR 80%).
Source: (Author 2017)

3.2 Revit and Insight 360 Workflow

Two Revit models were built to simulate the same conditions that were set up in the Rhino model, primarily
for the lighting analysis. One contained individual windows to simulate WWRs between 20% and 50%, while
one large window per facade orientation was needed for a WWR ratio of 80%. The solar radiation analysis
model utilized the massing model, as radiation results were desired for multiple facades, and not detailed
elements. Revit exports the model geometry and settings to the cloud, where the simulations are run through
EnergyPlus. Alternative design factors can be simulated by varying building loads, model geometry and
systems. Table 2 includes a series of options that represent all of the alternate design factors that were
simulated.

Table 2: Energy simulation results (Revit and Insight 360).

2 2
Run Name Variables EUI kWh/m (kBTU/ft )

Baseline 50% WWR, no shading; ASHRAE VAV; Wall: R-19; Roof: R-30; 225 (79.1)
Slab: R-23; Window: Double Low-E; Infiltration Rate: .8 ACH;
Lighting: 1.1 W/SF; Daylighting: None; Building Type: Office,
Schedule: 12/6

Low WWR 30% WWR 245 (77.6)

High WWR 80% WWR 271 (86)

Higher R-Values Wall: R-38; Roof: R-60; Slab: R-23; Windows: Triple Low-E 241 (76.4)

Infiltration Infiltration Rate: .17 ACH 233 (73.7)

Daylighting Daylighting and occupancy Controls; Horizontal shading south 232 (73.7)
2
and west (¼ window height); Lighting: 3 w/m

HVAC High Efficiency VAV 218 (69.2)

 HVAC alternate High Efficiency Package System 144 (45.7)

Best case 30% WWR; Horizontal shading south and west (¼ window 87.5 (27.7)
height); HVAC: High Efficiency Package System; Wall: R-38;
Roof: R-60; Slab: R-23; Windows: Triple Low-E; Daylighting and
2
occupancy Controls, Lighting: 3 w/m

The solar analysis tool was used to study the south facade of the building for average, cumulative and peak
insolation. Figure 6a shows cumulative insolation values for June 21, where the effects of adjacent buildings
can be seen on the low rise portion of the building. Insight 360 uploads the Revit model to the cloud, where
daylight simulations are run. The 10th floor was simulated according to the LEED v4 EQc7 opt 2 analysis type,
which measures the percentage of floor area that is between 300 and 3,000 LUX at 9am and 3pm on the
equinox averages. Figure 6b shows the results of daylight simulation.

a) b)

Figure 6: a) Solar radiation analysis results (cumulative, June 21); b) Daylight simulation results (June 21, WWR 80%).
Source: (Author 2017)

3.3 Revit and Sefaira Workflow

Sefaira is able to use the same Revit model as Insight 360, with some changes. The total number of glazing
planes cannot exceed 1,500. The window design for WWR of 30% and 50% had to be changed to one window
per orientation, instead of multiple windows, so that Sefaira could process and run the analysis. When an
energy analysis is run, the model is uploaded to the cloud, where it is run through EnergyPlus. The results are
either displayed in the web-based application, or within the plugin in Revit. On average, it takes three to five
minutes for each energy analysis run. Runs utilizing thermal comfort or natural ventilation factors take
significantly longer to process. Runs can be cloned to use as alternates with different design options within
the same model. Results are shown in Table 3. Solar radiation analysis cannot be performed in Sefaira.
Radiance and DAYSIM are used to perform daylight simulations, and an example is shown in Figure 7.

Table 3: Energy simulation results (Revit and Sefaira).

2 2
Run Name Variables EUI kWh/m (kBTU/ft )

Baseline 50% WWR, no shading; VAV - Return Air Package; Wall: R-19, 290 (92)
Roof: R-30; Slab: R-23; Window: .45/.38 (U/SHGC); Infiltration
2
Rate: .8 ACH; Lighting: 3 W/m

Low WWR 30% WWR 290 (92)

High WWR 80% WWR 296 (94)

Higher R-Values Wall: R-40; Roof: R-60; Slab: R-23; Window: .2/.38 (U/SHGC) 271 (86)

Infiltration Infiltration Rate: .17 ACH 148 (47)

Shading .3 m horizontal shades; Internal blinds 296 (94)

HVAC Option 1 Fan coil units and central plant 243 (77)

HVAC Option 2 Radiant floor 180 (57)

 HVAC Option 3 Active Chilled Beams 252 (80)

Best Case 50% WWR, no shading; Wall: R-4; Roof: R-60; Slab: R-23; 76 (24)
Window: .2/.38 (U/SHGC); Infiltration Rate: .17 ACH; HVAC:
Radiant floor

Figure 7: Daylight simulation results showing daylight factor. Source: (Author 2017)

4.0 DISCUSSION OF RESULTS

All simulations followed similar procedures for geometry preparation and visualization of results. Rhino,
utilizing Grasshopper, Ladybug and Honeybee, provides significant customizability for parametric control of
geometry, and offers different simulation types. The parametric nature of Grasshopper means that there is an
infinite number of forms and strategies that can be investigated. One major drawback of this workflow is that
the components have to be configured before the initial use. Once they are configured, the Grasshopper
definitions can be repeatedly used on the same or different projects. Another drawback is the learning curve
required to use the software. While other software programs utilize a series of dialog boxes to configure the
parametric and simulation options, this workflow requires the user to set up all the components before
visualizing the results. In terms of BIM-based workflow, Revit is the widely adopted BIM software, but both
Insight 360 and Sefaira have certain benefits and drawbacks. Learning curve for Insight 360 is fairly light. With
the energy and daylight simulations being run in the cloud, the results are quickly generated. The
customizability of the daylight and solar radiation analysis is adequate, but the energy simulation parameters
are limiting. Further, the lack of parametric tools to generate different WWRs or shading methods based on
building orientation is a shortcoming. Additionally, the parametrically generated energy model cannot be used
for daylight or solar radiation analysis. Sefaira is easy to install, as well as to use. The organization of energy
analysis options within the web-based application makes tracking model changes straightforward. Sefaira’s
use of the cloud for simulations decreases the time necessary for calculations. The daylight analysis tool is
excellent for analyzing the overall light levels and daylighting metrics, although its capabilities for showing
specific values on a floor plan are limiting. Additionally, the plugin lacks support for solar radiation analysis.

Although the accuracy of the simulations was not the focus of this research, the energy analysis results were
compared to each other, as shown in Table 4. All three program use the same engine, EnergyPlus, for the
energy calculations. The only differences between the three tools are variation in inputs and geometry
discrepancies between Rhino and Revit. The EUI results from Honeybee and Insight 360 were compared to
Sefaira, and the differences were expressed as a percentage. Insight 360 and Sefaira results were the closest
to each other, with the EUI results for Insight 360 baseline run being 16% lower than the Sefaira run, and the
best case run 15% higher. The Honeybee baseline EUI result was 34% lower, while the best was 66% lower.
The parameters for daylight and occupancy controls cannot be set in Sefaira. Insight 360 does not allow for
control of the temperature set-points. Ladybug and Honeybee have support for both of these variables, but
daylight and occupancy sensors were not simulated. The results from the daylight analysis are difficult to
compare since detailed illuminance values were not collected for Sefaira. In addition, the reflectance values
of floors, walls and ceilings cannot be set in Sefaira, so the illuminance values would not match even though
the simulation engine is the same. Further testing would be needed to compare the accuracy of the daylighting
analysis for Honeybee, Insight 360 and Sefaira.
Table 4: EUI results comparison (Honeybee, Insight 360 and Sefaira).
2 2
Run Name EUI (kBTU/ft , kWh/m ) Difference (%)
Honeybee Baseline 68.71 (216.74) -34%
Insight 360 Baseline 79.1 (225) -16%
Sefaira Baseline 92 (290) 0%
Honeybee Best Case 14.4 (45.3) -66%
Insight 360 Best Case 27.7 (87.5) +15%
Sefaira Best Case 24 (76) 0%

CONCLUSION
The objective of this research was to investigate methods for integrating parametric design with building
performance analysis. An ideal framework for integration of parametric and performance analysis procedures
was developed. Then, the framework was tested using existing software applications, including BIM-based
and non-BIM design software, parametric design and building performance analysis applications. Specifically,
Honeybee and Ladybug (for Rhino 3D) were evaluated as a non-BIM workflow, while Insight 360 (for Revit)
and Sefaira (for Revit) were evaluated as BIM-based methodologies. A case study building was used to test
and evaluate the workflows, interoperability, modeling strategies and results. Three different building
performance aspects were analyzed for each workflow: 1) energy analysis, 2) solar radiation analysis, and 3)
daylighting. The framework applied to Rhino, Grasshopper, Ladybug and Honeybee offers a lot of options and
customization for the parametric and simulation options. The lack of BIM integration in this framework is a
drawback, which means that many designers may use it for conceptual and/or schematic design, but will
migrate to a BIM-based software for schematic and design development phases. Insight 360 is able to
integrate building performance simulations within a BIM environment. However, Insight 360 has only been
available for a short time, and the functionality of the tool has its limits. Sefaira takes the customizability of
Ladybug and Honeybee and the accuracy of Insight 360 and integrates it into a BIM environment. The energy
and lighting analysis can be simulated quickly. The daylighting simulations have a few drawbacks, which
include lack of support for detailed illuminance values at specific points on the floor plan, an analysis grid that
cannot be adjusted by the user, as well as the inability to modify reflectance values of materials. Solar radiation
analysis is also not included in Sefaira. However, the overall results show a promising course for integrating
parametric design with building performance simulations. This would allow designers to evaluate the effects
of design decisions earlier in the design stages. Moreover, by integrating the capabilities of parametric design
and building performance simulations, multiple design variables can be tested rapidly, creating a more
cohesive and effective design process.

REFERENCES
Aksamija, A. 2013. "Building Simulations and High-Performance Buildings Research: Use of Building
Information Modeling (BIM) for Integrated Design and Analysis", Perkins+Will Research Journal, Vol. 5, No.
1, pp. 19-38.
Bazjanac, V. 2008. “IFC BIM-based Methodology for Semi-automated Building Energy Performance
Simulation”, Proceedings of the CIB-W78 25th International Conference on Information Technology in
Construction, Santiago, Chile, pp. 292–299.
Oxman, R. 2008. “Performance-based Design: Current Practices and Research Issues”, International Journal
of Architectural Computing, Vol. 6, No. 1, pp. 1-17.
Pratt, K. and Bosworth, D. 2011. "A Method for the Design and Analysis of Parametric Building Energy
Models", Proceedings of Building Simulation: 12th Conference of International Building Performance
Simulation Association, Sydney.
Rahmani, M., Zarrinmehr, S. and Yan, W. 2013. “Towards BIM-Based Parametric Building Energy
Performance Optimization”, Proceedings of the Association for Computer-Aided Design in Architecture
(ACADIA): Adaptive Architecture, Ontario.
Schlueter, A. and F. Thesseling. 2009. “Building Information Model Based Energy/Exergy Performance
Assessment in Early Design Stages”, Automation in Construction, Vol. 18, No. 2, pp. 153–163.
Turrin, M., von Buelow, P. and Stouffs, R. 2011. "Design Explorations of Performance Driven Geometry in
Architectural Design using Parametric Modeling and Genetic Algorithms", Advanced Engineering Informatics,
Vol. 25, No. 4, pp. 656-675.