Dataset for the metamodeling of naturally ventilated Brazilian low-cost houses to assess thermal performance

Research objective: This study aimed to develop metamodels to assess the thermal discomfort in naturally ventilated Brazilian low-cost houses during early design as a decision-making support framework, and for educational purposes. Method overview: The method encompassed a large number of simulations of the software EnergyPlus [EP] 8.1 using the Monte Carlo method. These simulations were used to develop a set of regression-based mathematical relationships between the inputs and the outputs. The Monte Carlo method was selected to help sample the many independent input variables, each with its own range of values, and form equally likely random combinations of input conditions for the energy simulation. The EnergyPlus outputs were post processed to assess the thermal comfort by means of the degree hours of discomfort by heat and by cold. Files description: The following data items were made available. (a) Parameter_Domains: Curitiba_parameterdomains.csv; Manaus_parameterdomains.csv and Sao_Paulo_parametersdomains.csv: CSV files created for each location. They contain a list of the 24 key parameters and random combinations of their values to create the input data for the 10,000 simulations. (b) Performance_Metrics: Curitiba_performancemetrics.csv; Manaus_performancemetrics.csv and Sao_Paulo_performancemetrics.csv: CSV files created for each location. They contain output values (outdoor and indoor discomfort by heat and by cold) for 10,000 simulations. (c) Sandbox: Sandbox.xlsx: Excel file for the application of the metamodels. It enables a quick and easy assessment of discomfort by heat and by cold for specific combinations of parameters, for each location. (d) Python_Codes: Python_Codes.zip: Compressed file consisting of the codes used to 1) run the simulations randomly combining values for each selected parameter within their specified ranges, and 2) to calculate the hours of discomfort by heat and by cold for each of the parameters’ combinations (10,000 simulations for each studied climate). (e) IDFs_Base: Curitiba_idfbase.idf; Manaus_idfbase.idf; Sao_Paulo_idfbase.idf : Input Data File (IDF) created for each location. They contain the description of all input data considered in an annual building performance simulation.

研究目标：本研究旨在构建元模型（metamodels），以在早期设计阶段评估巴西自然通风低成本住宅的热不舒适状况，以此作为决策支持框架，同时兼顾教育应用需求。
方法概述：本研究采用蒙特卡洛方法（Monte Carlo method），依托EnergyPlus [EP] 8.1软件开展大量模拟实验，通过模拟结果构建输入变量与输出变量之间的一系列基于回归分析的数学关系式。选择蒙特卡洛方法的目的在于，对众多独立输入变量进行采样——各变量均有专属取值范围，并生成等概率的输入条件随机组合，用于能耗模拟。随后对EnergyPlus的输出结果进行后处理，通过热不舒适度日小时与冷不舒适度日小时指标，评估热舒适状况。
数据集文件说明：本次公开提供以下数据项：
(a) 参数域（Parameter_Domains）：包含Curitiba_parameterdomains.csv、Manaus_parameterdomains.csv及Sao_Paulo_parametersdomains.csv三个文件，均针对对应城市生成。文件内收录24个关键参数列表及其取值的随机组合，用于为10000次模拟生成输入数据。
(b) 性能指标（Performance_Metrics）：包含Curitiba_performancemetrics.csv、Manaus_performancemetrics.csv及Sao_Paulo_performancemetrics.csv三个文件，均针对对应城市生成。文件内收录10000次模拟的输出值，涵盖室外与室内的热不舒适、冷不舒适相关数据。
(c) 沙盒（Sandbox）：Sandbox.xlsx，用于元模型的应用，可针对各城市的特定参数组合，快速便捷地评估热不舒适与冷不舒适状况。
(d) Python代码（Python_Codes）：Python_Codes.zip，为压缩代码包，包含两类代码：1）在各选定参数的指定取值范围内随机组合参数值以运行模拟；2）针对每组参数组合（每个研究气候场景对应10000次模拟）计算热不舒适与冷不舒适的小时数。
(e) 基础输入数据文件（IDFs_Base）：包含Curitiba_idfbase.idf、Manaus_idfbase.idf及Sao_Paulo_idfbase.idf三个输入数据文件（Input Data File，简称IDF），均针对对应城市生成，文件内收录年度建筑性能模拟所需的全部输入数据描述。