Climate-Based Daylight Modelling for compliance verification: Benchmarking multiple state-of-the-art methods
Climate-Based Daylight Modelling for compliance verification: Benchmarking multiple state-of-the-art methods
复制标题
DOI：
10.1016/j.buildenv.2019.04.051
复制
发表时间：
2019-07
影响因子：
7.4
通讯作者：
E. Brembilla;J. Mardaljevic
中科院分区：
文献类型：
--
作者：
E. Brembilla;J. Mardaljevic
摘要基于气候的日光建模（CBDM）为设计师提供了评估复杂的长期光照环境动态的可能性。这种复杂性可能很难模拟，甚至更难通过使用性能指标进行有效的通信。在过去的二十年中，已经开发了多种CBDM技术和指标，但这些技术和指标很少相互评估。本文回顾了四种基于Radiance光线跟踪引擎的最新技术，并将其与基准CBDM方法进行了系统的比较。四个教室空间用于进行模型间的性能指标之间的比较，通常用于从所有分析技术的合规性验证。额外的敏感性分析评估了输入变量的变化如何影响这些指标。从模型间比较的结果表明，直射阳光的表示是显着不同的各种CBDM技术之间，和基于水平直射阳光的指标是非常敏感的模拟方法的选择。这导致预测的年度阳光照射差异高达39个百分点。同时考虑直射光和内部反射光的方法更稳健，与基准结果的偏差在± 15%以内。对输入变量的分析表明，传感器网格间距和时间步长插值不会显着影响任何这些指标。方向和天空离散化方案的变化有不同的影响，这取决于所考虑的度量和技术。还讨论了在引入新的性能指标进行合规性验证或新的模拟方法时，对权威基准测试系统的需求。
Abstract Climate-Based Daylight Modelling (CBDM) gives designers the possibility to evaluate complex, long-term luminous environment dynamics. This complexity can be challenging to simulate, and even more challenging to communicate effectively through the use of performance metrics. A multiplicity of CBDM techniques and metrics has been developed over the last two decades, but these were rarely assessed against each other. This paper reviews four state-of-the-art techniques based on the Radiance raytracing engine and systematically compares them against a benchmark CBDM method. Four classroom spaces are used to carry out an inter-model comparison between performance metrics commonly used for compliance verification obtained from all analysed techniques. Additional sensitivity analyses assessed how changes in input variables influence such metrics. Results from the inter-model comparison showed that the representation of direct sunlight is markedly different between the various CBDM techniques, and that metrics based on horizontal direct sunlight are very sensitive to the choice of simulation method. This led to differences in predicted Annual Sunlight Exposure up to 39 percentage points. Metrics that consider both direct and inter-reflected light were found to be more robust, with variations from benchmark results within±15%. The analysis of the input variables showed that sensor grid spacing and time-step interpolation do not significantly affect any of these metrics. Changes in orientation and sky discretisation scheme had different effects depending on the metric and technique considered. The need for authoritative benchmarking systems when introducing new performance metrics for compliance verification or new simulation methods is also discussed.