For at least the last decade, the gold standard for daylight analysis of architectural projects has been DIVA Radiance. Cove.tool uses a full-floor daylighting analysis method taken from "LEED v4.0 - IEQ c7 Daylight, Option 1. Simulations: Spatial Daylight Autonomy" and "IES LM-83: Approved Method: IES Spatial Daylight Autonomy (sDA%) and Annual Sunlight Exposure (ASE)." This article presents the resulting comparison study between the two methods. It is important to note that when evaluating two simulation methods they should both allow users to make the same decision given the same inputs. The percentage difference between design options is the measure of accuracy in a simulation method.
Cove.tool runs a full ray-tracing simulation and is calibrated to within 1% to 5% of a Radiance simulation and is 1500x faster. The comparison charts below are a snapshot summary of the alignments between DIVA and cove.tool annual analysis for both Daylight (sDA) and Solar Exposure (ASE).
Let's breakdown a clear comparison between cove.tool and DIVA.
To make sure we were not seeing anomalies due to a specific weather file we tested the same building geometry for a sample U-Shaped Demo project in multiple city locations with both platforms.
Both models use a 2 ft grid set 30” above the floor and are looking for sDA 300 lux with a schedule between 8 am to 6 pm per the recommendations of IES LM-83 and ASE 1000 lux for glare. The 1500x speed improvement comes from employing statistical methods and parallelizing the calculations on our server environment. We are using a Tregenza sky with 576 subdivisions and keeping the reflectivity fixed for now with the following assumptions: Floors 40%, Interior Walls 70%, Ceiling 70%, Exterior Facade 35%, and Walls 35%, Outside Ground 40%. This allows any user to construct a realistic simulation and avoids "cheating" the simulation with incorrect inputs.
Floor Plan Studies
The middle floor of a 3 story U-shaped building available from Autodesk Revit is used as a test case. In general, cove.tool and DIVA agree, We tested in four different locations across the United States; Atlanta, GA, Denver, CO, Seattle, WA, and Anchorage, AK in order to show that it was not latitude dependent. All models were run at the airport location 500 m from any surrounding buildings or context. The following analysis diagrams present the results that graphically align based on the 3d analysis and grid color legends. DIVA studies took 10 h on average to complete with a standard i7 laptop, while the cove.tool studies each rendered within 5 mins. Use the legend for the cove.tool studies to see how they compare in the diagrams below.
Note that cove.tool has a split in the daylight grids at 50% so that it is easy to see what is passing (yellow to red) and not passing (blue to dark blue).
Both software perform accurate daylight and solar exposure analysis in compliance with international standards and common daylight certification requirements like LEED v4.0 - IEQ c7 Daylight credit. The average resulting full-floor analysis between the two is under 6.9% for sDA and under 1% for ASE. DIVA seems to slightly over estimate daylight on the Denver example and further study is warrented to determine the reason for this outcome.
The cove.tool analysis time with all studies complete in under 5 minutes for all 3 floors is drastically different from the DIVA models taking 30 hrs to complete on average. Thus, our models can run at the speed of design and can be run affordably without specialized training for every project.
We recommend testing cove.tool for yourself!