When analysing a potential site for a new building, there are a range of metrics that can be used to quantify the amenity and resources available, and therefore the potential performance of a building if it were designed to take full advantage of them. As sites can be quite large and typically surrounded by obstructions such as other buildings, engineering works and vegetation, there can be significant variation in the availability of these amenities and resources at different points within it.
As a result, it can be insightful for a designer to be able to visualise the spatial variation of many different metrics across a site when starting to design a new building. Even more important in many cases is the ability to quantify the potential impact of a new building on any existing buildings within or near the site. In many urban environments, the designer’s role is often a delicate balance between maximising the amenity and resources available to their building whilst at the same time minimising its impact on the existing amenity of the buildings around it. Many cities and local authorities have quite specific regulations that govern both sides of this balancing act.
The calculation and visualisation of site analysis metrics is therefore a particularly important tool for design refinement and demonstrating compliance with building codes. However, with the increasing role of generative design techniques and performance-driven design, it can also be an important tool for actually driving the design process and optimising a potential solution. As with all things architectural, there is no single number to measure performance by or single target metric to aim for. Design is usually a matter of making the best set of compromises, so being able to understand and assess the widest range of performance metrics is key to achieving the best balance and most satisfactory solution.
The aim of this article is to list and describe in detail the various site analysis metrics that can be calculated, as well as what they measure and under what assumed conditions. It is also intended as my own aide-mémoire, created as a way to organise and formalise the calculations implemented in the site designer web app.
Incidence vs Flux
Before detailing specific metrics, it is important to point out that many of them can give completely different results depending on how they are measured.
For most site analysis calculations there are two primary ways to measure a value:
Incident on a surface
Buildings are primarily composed of surfaces, which are used to separate spaces from each other as well as the outside environment. How these surfaces heat up, cool down, transmit sound, collect radiation, reflect light and generally interact with the environment is critical to the performance of each space and the building as a whole.
When measuring how much of something is incident on a surface, the orientation of the surface relative to the direction of emission is very important. This is because the amount falling on the surface, per unit area, is directly proportional to the cosine of the angle between the emission direction and the surface normal, as per Lambert's cosine law. This effectively states that anything arriving at near normal incidence will contribute significantly more than that which arrives nearer to parallel with the surface.
At a point in space
Some metrics depend on the measurement of a resource arriving from any direction. These include sound when measuring noise levels, radiated heat when measuring comfort and air when measuring wind speed for example. There are also many metrics that can be measured both at a point and on a surface. These include shading, skylight, sunlight, pressure and even the quality of views.
When measured at a point, there is no dependence at all on the incoming direction of the emission - just the amount of flux that is passing that point. Flux values can still be given per unit area, but that measures the amount of resource that is either passing through a spatial region or present within it.
It is not always obvious which of these two measures is the most appropriate in every circumstance. Take for example the analysis of view. Whilst you may be able to see a picturesque river from a particular window, having to do so by looking sideways with your face against the glass is not as desirable as being able to look straight out at it.
Also views are typically enjoyed from some distance inside a room looking out through a window. In this case, you may want to use a grid with a normal in the same direction as the surface normal of the window pane to measure it rather than a grid of directionless sensor points.
However, if you are measuring the view from points in a garden or open space where the viewer is free to turn around and look in any direction, then using a directionless sensor point would make sense.
Thus, it is important to know and understand when surface grids are more appropriate and when to use a grid of independent sensor points.
Geometry-dependent metrics refer to values that are affected only by the size and shape of the site and the nature of surrounding buildings or obstructions. Such metrics are particularly useful when directly comparing buildings located on different sites and in different locations as they do not include the variable effects of localised weather or climatic conditions.
The following is a list of geometry-dependent site analysis metrics:
The term Visible Sky refers to that portion of the sky dome that is directly visible without reflection or obstruction from a point within the site. This metric does not consider the luminant or radiant distribution of the sky, but assumes that all parts of the sky carry equal weight. Visible Sky is typically expressed as a percentage of the total sky dome area, but may also be given as a solid angle.
The visible sky calculation is dependent only on the geometry of the site and its surroundings, and to some degree on the number and nature of patches in the selected sky subdivision. It is not affected by terrestrial location, weather data, sky distribution or solar geometry.
When considering access to sky views from windows or when mapping the variation in visible sky over the facades of a building, it is best to use a surface grid with the normal of each point aligned to the facade on which they sit. This is because the building itself will obstruct the visibility of sky patches that are behind the facade, so they must be excluded. Also sky views are more typically enjoyed from some distance inside a room looking out through apertures in the facade. Thus, more weight should be given to visible sky patches that are directly in front of the facade and less weight to those off to the side as they would only be visible when the observer is very close to the window pane and looking sideways. Applying Lambert’s cosine law does this automatically as this is basically the reverse of surface incidence.
When calculated and mapped on building surfaces, this value is sometimes also referred to as a Sky View Factor (SVF) and is essentially the same as a Sky Component (SC) when that is calculated using a CIE uniform sky distribution.
Vertical Sky Component
The Sky Component (SC) is a relative measure of the indirect sky illuminance from a standard overcast sky that reaches a surface either inside or outside a building, compared to that which would be received by a horizontal surface under the same sky in a completely unobstructed environment. Being fundamentally a measure of surface illuminance, this metric considers both the spatial distribution of skylight over the sky dome and the effects of surface incidence by applying Lambert’s cosine law to the relative contribution of each sky patch. Direct sunlight and any potentially reflected illuminance are not included.
The Vertical Sky Component refers to a sky component calculated for a vertical surface and is more widely known as it can be used as a comparative measure in some rights-to-light regulations, particularly in the United Kingdom.
As both measures use as their reference a completely unobstructed horizontal surface, the maximum possible Sky Component for a flat roof or other horizontal surface is 100%. For a vertical surface, the maximum possible Vertical Sky Component is only about 40% (varying between 39.5% and 39.7% depending on the number and distribution of sky patches). This is because the CIE traditional overcast sky distribution assigns sky patches at the zenith three (3) times more illuminance than those at the horizon. For a vertical surface, sky patches near the zenith make little or no contribution as their illuminance arrives almost parallel to the surface (nearly 90 degrees from its surface normal).
The Vertical and other Sky Component calculations are dependent only on the geometry of the site and its surroundings, and to some degree on the number and nature of patches in the selected sky subdivision. Despite being based on an overcast sky distribution, this is a relative measure so is the same for all locations. As such it is not affected by terrestrial location, weather data or solar geometry.
The Daylight Factor is a relative measure of sky illuminance arriving on a surface either inside or outside a building, compared to that which would be received by a horizontal surface under the same conditions but in a completely unobstructed environment. The Daylight Factor is the sum of both Sky Component and any reflected components from internal or external surfaces. Direct and reflected sunlight are not included.
The aim of the Daylight Factor is to quantify a worst-case scenario, which is typically under a CIE traditional overcast sky distribution which assigns sky patches at the zenith three (3) times more illuminance than those at the horizon, but is otherwise symmetrical. The daylight regulations in some locations may stipulate the use of a CIE uniform sky in cases where their worst-case condition is better described by more uniform conditions which are entirely symmetrical.
Whilst it is possible to use any of the Perez All-Weather Sky distributions or those described in ISO15469:2004(E) for daylight factor calculations, these should only be done as part of design feedback as they are not true worst-case conditions and therefore cannot be used to demonstrate compliance or anywhere actual Daylight Factor values are required.
Daylight Factor calculations are dependent on the geometry of the site and its surroundings, the reflectance of internal and external surfaces, and to some degree on the number and nature of patches in the selected sky subdivision. Despite being based on either overcast or uniform sky distributions, this is a relative measure so is the same for all locations. As such it is not affected by terrestrial location, weather data or solar geometry.
Daylight Factors are an older metric, but are still widely used as they are much simpler to calculate than dynamic or climate-based daylight metrics and can be used to directly compare the daylight performance of different buildings or spaces in completely different locations. This makes them a useful reference metric for design teams. For example, when designing a new classroom to meet a specific average daylight factor, a design team can draw from other example classrooms that have met or exceeded that target value regardless of whether they are from Norway or Singapore. However, daylight factors notably ignore the effects of direct sunlight, so the final design must still adapt to the specifics of its local climate.
Isovist View Analysis
An isovist refers to the three-dimensional (3D) volume of space that is directly visible from a point on the site. It can be thought of as a hemisphere centred at a selected point and projected outwards to some maximum view radius or until it hits an opaque obstruction. The internal volume of the resulting shape is the view isovist for that point.
Visualising the complete isovist for a point is typically not that useful as it can be difficult to appreciate its true 3D shape. This is because most of the obstruction typically occurs at the bottom of the hemisphere, which is not that easy to see. As a result, they are usually visualised as a single two-dimensional (2D) horizontal slice through the resulting shape to represent the horizontal field of view for an observer standing at the centre point.
Another useful visualisation is to take the negative of the isovist shape and project each of the surrounding obstructions out hemispherically from the selected point. This has the benefit of quite clearly revealing what is visible and what is obstructed, especially when the model is viewed from a higher altitude.
Isovists can also be particularly useful for planning and space syntax analysis as they allow a range of single-value spatial perception metrics to be extracted for each point on a grid, that can then be mapped and compared. These include basic geometric information such as volume (3D) or area (2D), more complex information such as jaggedness, circularity or elongation, right through to potential perceptions of enclosure, mystery or enticement to move in a particular direction. For more details on the wide range of isovist metrics that can be extracted, see the paper Isovists: Spatio-Visual Mathematics in Architecture.
Grid View Analysis
Whilst isovists deal with perceptions of space, Grid View Analysis deals with how well one set of spatial points can ‘see’ other points within the model. For example, if there is a particularly desirable feature near the site, it is possible to spatially map and visually identify those areas of each facade with the best view of it. Alternatively, this can be used to quantify and map the potential visual impact of a new development on the surrounding buildings or environment.
To set up a grid view analysis, simply select one spatial grid to be the desirable feature to view and then orient it appropriately to represent that feature. Then, the visibility of each point on the feature grid will be calculated at each point on all other grids. The results will be shown as the percentage of points that are directly visible.
View calculations are dependent only on the geometry of the site and its surroundings, and to some degree on the number and distribution of points on each grid in the analysis. They are not affected by terrestrial location, weather data, solar geometry, sky subdivision or luminance distribution.
Location-dependent metrics are those that are dependent on the terrestrial location of the site, as well as the size, shape and nature of surrounding buildings or obstructions. The dependence on location is typically due to the annual path of the Sun through the sky varying with latitude.
Such metrics are particularly useful when directly comparing buildings located on different sites and in different locations as they do not include the variable effects of localised weather or climatic conditions.
Time in Shade
Being in shade refers to the blocking of direct sunlight by an opaque obstruction that sits between the reference point and the position of the Sun. As the Sun moves across the sky each day and its altitude varies over the year, being in or out of shade can be highly dynamic and varies with both date and time. Thus shading is typically quantified as an amount of time, expressed either as an absolute number of hours or the percentage of a given period. It may also be expressed as the average number of hours per day over the given period, but this is only really useful when comparing seasons or months.
Time in shade calculations are dependent on the geometry of the site and its surroundings, as well as the terrestrial location as this determines the path of the Sun through the sky at different times of the year. The calculations themselves involve tracking the position of the Sun as it travels through the sky over the given date/time period and testing for obstruction at each point for each time interval. As such, this metric is not affected by weather data, the selected sky subdivision or sky luminance distribution.
Climate-dependent metrics refer to calculations that are dependent on hourly weather data and terrestrial location as well as the size, shape and nature of surrounding buildings or obstructions.
Probable Sunlight Hours
Probable Sunlight Hours (PSH) refer to the number of hours that a surface can expect to receive direct sunlight over a given period. To properly reflect the nature of different climates, this metric depends on recorded levels of cloudiness and direct solar radiation within the hourly weather data for a location to determine the total available sunlight hours.
This metric is typically given as a percentage of the total available sunlight hours over a given period. Some building regulations and published design guidance refer specifically to Annual Probable Sunlight Hours (APSH) measured over the entire year and Winter Probable Sunlight Hours (WPSH) measured over the coldest three months. This metric can also be given as Solar Exposure, typically quantified as an absolute number of hours or the average number of hours per day over the given period.
Guidance from the Building Research Establishment (BRE) in the UK recommends that the APSH received by equator-facing windows should be at least 25% of the total available, and WSPH should be at least 5%. When used to quantify the impact of a new development on surrounding buildings, the guidance recommends that, where any windows fall below these values and the loss is greater than 4%, then the impacted APSH and WPSH should not be less than 0.8 times their previous values.
The BRE guidance recognises that sunlight is highly dependent on orientation, so states that only windows with an orientation within +/-90 degrees of South need be assessed. In the Southern Hemisphere this would be within +/-90 degrees of North.
The Probable Sunlight Hours calculation involves tracking the position of the Sun as it travels through the sky over the given date/time period and recording the number of hours the recorded direct solar radiation value is above a given threshold, usually 125W. If the value is above the threshold at any time interval, each grid point is tested for obstruction and its percentage updated accordingly. As a result, this metric is dependent on hourly weather data, the geometry of the site and its surroundings, and its terrestrial location as this determines the path of the Sun through the sky at different times of the year. It is not affected by the selected sky subdivision or sky distribution.
Incident Solar Radiation
Incident Solar Radiation (insolation) is a measure of the short and long wave energy incident on a surface directly from the Sun and diffusely from the sky. Whilst solar energy can be measured as flux at a point in space, this is not usually a practical value to work with as the primary interest in solar radiation is its collection and/or its potential to heat surfaces.
Insolation can be measured at both an instant in time, expressed in W/m2 (or BTU/ft2.hr), or accumulated over a given period, expressed in Wh/m2 (or BTU/ft2). Calculating insolation involves both tracking the position of the Sun as it travels through the sky over the given date/time period and simulating appropriate diffuse sky conditions at each time interval. As a result, this metric is dependent on hourly weather data, the selected sky subdivision and dynamically calculated sky distribution, the geometry of the site and its surroundings, and its terrestrial location as this determines the path of the Sun through the sky at different times of the year.
Room-Based Internal Site Metrics
There are a number of metrics used for site analysis that can be calculated inside a building once its internal space layout and fenestration details have been resolved. Most are essentially a room-based statistical analysis of some of the previously described metrics, but provide additional single-number values for compliance analysis and performance targeting.
It is envisaged that these room-based metrics will be implemented in a subsequent phase of this project. The first phase will implement the external site metrics that are selected for prioritisation (1 to 8).
Horizontal Sight Angle
Horizontal Sight Angles are a measure of the width of exterior views available within a room. These can be calculated for any point in a space by summing the horizontal angle of the transparent areas in each aperture that are located within the external boundary of that space.
European standard EN 17037 (UK: BS EN 17037:2018 Daylight in Buildings) defines a set of horizontal sight angle bands that define the quality of exterior view - being 14˚ for minimum, 28˚ for medium and 54˚ for high quality views. It also specifies the minimum view distance to the first obstruction for each quality band - being 6m for minimum, 20m for medium and 50m for high. The aim is that these values are calculated across each room and each area of the space is apportioned to the appropriate quality of view band, with specific threshold area requirements for different room types and activities.
The horizontal sight angle calculation is dependent on the detailed building geometry, the spatial layout of rooms within the building, the type and geometric detail of apertures and fenestration, and the size and shape of the site and its surroundings. It is not affected by terrestrial location, weather data, solar geometry, sky subdivision or luminous distribution.
No Sky Line
The No Sky Line refers to a virtual line within a room that delineates between those areas that have some direct view of the visible sky and those that do not. This is typically calculated on a horizontal working plane 0.85m above floor level. As a rule of thumb, if less than 50% of the room area has no visible sky (≤0.2%), then it will be considered inadequately lit and appear gloomy.
The no sky line can also be used to quantify the impact of a new development on surrounding buildings and is referenced in some right-to-light regulations, particularly in the UK. Guidance from the Building Research Establishment (BRE) in the UK recommends that, where the impact of a new development would result in the no sky line moving deeper into a habitable room within an adjacent building, the area behind the no sky line should not fall below 0.8 times its previous value.
The no sky line is calculated by applying a grid of points over the working plane within a room and calculating the Visible Sky value at each one. A contour line or band is then generated at 0.2% to visualise the No Sky Line, and the areas above and below this threshold determined.
The no sky line calculation is dependent on the detailed building geometry, the spatial layout of rooms within the building, the type and geometric detail of apertures and fenestration, and the size and shape of the site and its surroundings. It is not affected by terrestrial location, weather data, solar geometry, sky subdivision or luminous distribution.
Average Daylight Factor
The Average Daylight Factor (ADF) is a room-averaged single-number value that measures the adequacy of diffuse daylight within a space. It is generated by summing and area-weighting multiple daylight factor calculated at points distributed across the working plane within a room.
The average daylight factor is another metric that can be used to quantify the impact of a new development on surrounding buildings. It too is referenced in some right-to-light regulations, particularly in the UK. Guidance from the Building Research Establishment (BRE) in the UK recommends minimum values of 1% for bedrooms, 1.5% for living areas and 2% for a family kitchen. Where the impact of a new development would result in the average daylight factor of rooms within an adjacent building falling below these minimums, then resulting value should not fall below 0.8 times its previous value.
The average daylight factor calculation is dependent on the detailed building geometry, the spatial layout of individual rooms within the building, the type and geometric detail of apertures and fenestration, the colour and reflectance of interior and exterior surfaces, the size and shape of the site and its surroundings, and to some degree on the number and nature of patches in the selected sky subdivision. It is not affected by terrestrial location, weather data or solar geometry.
Code Modules Required
Given the aide-mémoire role of this article, I am also going to put down some formalisation of the software requirements so I can keep clarity. In order to calculate these site metrics from building geometry, there are a number of inter-dependent software components or modules that need to be developed:
Geometry Import Module
A library to allow parsing of a range of CAD and/or BIM file formats and processing of the resulting geometry in a way that it can be efficiently ray-traced. Support for relatively simple OBJ, STL and PLY file formats can be readily provided, however more complex formats such as FBX, Collada, USD and glTF will require additional work to adapt open-source libraries from other tools, such as THREEJS or BABYLONJS for example.
Efficient Ray-Tracing Module
Building geometry has to be processed in order to allow for fast and efficient ray-tracing. This typically involves performing spatial subdivision using a KD-Tree or octree structure, which can then be traversed and tested relatively quickly. The exact architecture of this library will be dependent to some degree on the type of geometry formats selected for support and the model information provided by the geometry import module.
Solar Geometry Module
Once a site is assigned a terrestrial longitude, latitude and time-zone, a library is required to accurately calculate the position of the Sun at any date and time. There are several code bases available on the web, however they tend to be unoptimised and their licenses are either restrictive or unclear. For a more detailed explanation of this, see the section titled “Available Equations and Code” in the Solar Position article, which also gives some background to my own optimised solar position calcuation code that I have been working with and refining over many years.
Without my code, there are basically two options for this library:
Base its core on Mike Bostock’s implementation of the NOAA Solar Position Calculator code available here, and assume a permissive license, or
Sky Distribution Module
This is quite a heavy library that must deal with dividing the sky into multiple sample patches and then calculate the relative luminance and/or radiance for each patch using dynamic weather conditions to generate many different types of sky.
The core of this library must do the following:
Parametrically subdivide the sky into patches based on industry accepted techniques and maintain accurate angles, areas and weightings for each patch.
Implement the Perez All-Weather Sky model and be able to calculate the five (5) required coefficients from sky clearness and brightness data interpreted from annual hourly weather data.
Efficiently calculate the relative illuminance/radiance of each sky patch based on dynamic sky coefficients. Efficiency is a priority as a single annual cumulative sky calculations requires this to be done 43,800 times for each patch.
Consume data from the weather data module, generate the required sky distributions and provide the resulting sky patch data in a format easily consumed by the ray-tracing module.
This process involves a complex sequence of quite detailed equations, so a lot of testing and validation is required on each implementation as it is very easy for small mistakes to create large errors in the results.
Weather Data Module
As the calculation of some metrics require detailed annual hourly weather data, a module is required to import, parse, process and store weather data files. The most common format for this is the EnergyPlus Weather File (EPW) as these are widely available for most locations in the US, Europe and Australia. Support for other formats such as TMY2, TMY3 and BOM data files are desirable, but EPW files are a good basis to start with.
Click here to comment on this page.