xu_etal_2022_sdata

A multi-scale time-series dataset of anthropogenic heat from buildings in Los Angeles County

Yujie Xu¹, Pouya Vahmani¹, Tianzhen Hong*, Andy Jones¹

¹ LBNL

* corresponding author: thong@lbl.gov

Abstract

Anthropogenic heat (AH) from buildings can increase urban temperature and contribute to the urban heat island effect. The increased urban temperature leads to more cooling demands in buildings during summer thus more air-conditioning energy use and higher associated waste heat to the urban environment, forming a positive feedback loop. This paper presents a dataset of AH from buildings in Los Angeles (LA) County. The dataset is created with the physics-based EnergyPlus building energy models to calculate individual buildings’ AH considering WRF-UCM simulated microclimate conditions. The dataset contains 8760 hourly AH in 2018 from buildings aggregated at three spatial resolutions of 450m x 450m grid, 12km x 12km grid, and census tract for the entire LA County. The AH is broken down into three components: building envelope surface convection, heating, ventilation and air conditioning (HVAC) system heat release, and zone exfiltration and exhaust air heat loss. The high-resolution AH dataset can be used for research on building AH and its interaction with urban microclimate, as well as informing effective AH reduction policy interventions across LA County.

Journal reference

Scientific Data

Code reference

The following code files are adapted from the repository im3-wrf

0_mv_inputs.py, 2_wrf_to_epw.py, 3_write_baseline_idf.py

Data reference

The data set is held in MSDLive https://doi.org/10.57931/1892041

Input data

Held in the input_data folder, including the external data files referenced downstream in the analysis. The data sources are shown in the following table

input data	source
Assessor_Parcels_Data_-_2019.csv	[link] (https://data.lacounty.gov/search?categories=open%20data)
LARIAC6_LA_County.geojson	[link] (https://lariac-lacounty.hub.arcgis.com/)
WRF data in annual_WRF and M02_EnergyPlus_Forcing_Historical_LowRes folder	Generated from The Weather Research and Forecasting (WRF) model (version 4.2.1) [Link] (https://github.com/wrf-model/WRF)
Energy Atlas data set	[link] (https://www.energyatlas.ucla.edu/)
CEC county-level electricity and gas consumption in CECdata folder	[link] (http://www.ecdms.energy.ca.gov/)
2018 Manufacturing Energy Consumption Survey (MECS), in MECS folder	[link] (https://www.eia.gov/consumption/manufacturing/)
LA design day data in la-design-day	[link] (https://energyplus.net/weather-location/north_and_central_america_wmo_region_4/USA/CA/USA_CA_Los.Angeles.Intl.AP.722950_TMY3)
building prototype building models in "LA Prototypes" folder	Various data sources see the "sources of the prototype building models"
ResStock and ComStock data	[link] (https://data.openei.org/s3_viewer?bucket=oedi-data-lake&prefix=nrel-pds-building-stock%2Fend-use-load-profiles-for-us-building-stock%2F2021%2F)

• LA buildings
  • LA building characteristics: Assessor_Parcels_Data_-_2019.csv, from Dropbox
  • LA building geometry: LARIAC6_LA_County.geojson (this data file is not included for sharing restrictions. Please contact us if you need this data)
• Climate
  • Annual WRF data: held in annual_WRF in a 12km x 12km grid system. The two .tar files are original data from Pouya. The tar balls contain weather data in the following format: Variable_<variable_name>.txt and grid latitude longitude data (Fixed_XLAT.txt, Fixed_XLONG.txt) The annual_WRF/M02_EnergyPlus_Forcing_Historical_LowRes_ann folders holds the grid weather data for each variable in separate folders (GLW, PSFC, Q2, RH, SWDOWN, T2, WINDD, WINDS). The grids_csv holds weather data compiled for each grid cell, time_series holds weather data for each WRF variable. The wrf_epw folder contains the .epw files for each grid cell. These epw files are used in the heat and energy simulation.
  • July WRF data: held in M02_EnergyPlus_Forcing_Historical_LowRes. The folder structure is similar to the annual_WRF/M02_EnergyPlus_Forcing_Historical_LowRes_ann_ folders
• geojson or shapefiles of regions of analysis
  • coarse WRF grid: M02_EnergyPlus_Forcing_Historical_LowRes/meta/wrf-grids-origin.geojson. This grid corresponds to the epw files used in EnergyPlus simulations.
  • fine WRF grid: high res grid for reporting/wrf-grids-origin.geojson. This grid system is used in reporting
  • census tract: domain/tl_2018_06_tract/tl_2018_06_tract.shp. This is used in reporting.
  • city boundary: input_data/domain/City_Boundary.geojson
  • county boundary: input_data/domain/la-county-boundary.geojson
• Prototype Building Models: held in folder "LA Prototypes"

  • commercial buildings: most commercial building models are held in Com (OS_Standards). The Com need mod folder contains models for Religious buildings in different climate zones. The study uses the one corresponding to 3B. "​​NursingHome source" contains the nursing home models from Sun et al. 2020 The blackout incident is removed and the systems are changed to autosize. The adjusted model is in "NursingHome mod sched autosize"

  • residential buildings: held in "Res (CBES)" folder. "bldg_11" are single-family buildings. "building_13" are multi-family buildings. "vin_1" is pre-1980. "vin_5" is 2004 and "vin_8" is 2013. "res_schedule" contains the schedule files used in residential models.

  • sources of the prototype building models:

    Model	Source
    single-family, multi-family	CBES
    heavy and light manufacturing facilities	adapted from the warehouse model from OpenStudio Standard Gem
    nursing home	[Sun et al. 2020]
    others	OpenStudio Standard Gem

  • final simulation input data files:

• Lookup tables to re-map types between assessor data, prototype building models, and Energy Atlas building types in verification
  • building_type_recode.csv maps the building types in Assessor_Parcels_Data_-_2019.csv and EnergyPlus prototype building type
  • type_vintage_to_idf_mapping.csv maps building type and vintage to the idf file used to simulate the building
  • prototype_bldg_area.csv maps prototype building models idf file names to their building size in m2
  • maps the idf key words (filename removing the ".") to the usetypes defined in Energy Atlas for later comparison
• MECS survey tables are held in "MECS" folder. It is used for extracting summary statistics to model heavy and light manufacturing facilities
• Energy Atlas: annual electricity and gas consumption. "usage_bld_kwh.csv" is electricity data. "usage_bld_therm.csv" is gas data. "usage_bld_btu.csv" is electricity + gas data.
• ResStock data: simulation data set of residential buildings, downloaded from this link, resstock_tmy3_release_1/ and comstock_tmy3_release_1. The downloaded data are in "input_resstock" and "input_comstock" folder

Intermediate data

Held in the intermediate_data folder, including the data files in the intermediate data analysis or simulation steps

• EnergyPlus input idf files in various processing stages
  • idf_to_modify: input idfs in their original state
  • idf_change_design_day: idf with design day and location changed to LA
  • idf_add_sim_period_output: idf with RunPeriod adjusted to Jan. 1st to Dec. 31st and with heat emission and energy consumption outputs added
  • warehouse_model_modify: warehouse models and light and heavy manufacturing facility models derived from them.
• Summary statistics: held in "summary" folder
• compiled MECS data: energy_intensity_per_type.csv, used in compiling the weighted quantile for manufacturing facilities
• epw_idf_to_simulate.csv: epw-idf combination to be simulated with EnergyPlus, referenced in run_sim.py
• weather_2018.csv: data file that generates the figure "compiled_epw_weather.png"

Output data

• simulation results for prototype-building-wrf-grid-combination: held in EP_output/result_ann_WRF_<year>, for 2016 and 2018: Each subfolder contains a simulation output of a prototype model and WRF grid as follows <prototype model key word>____<WRF grid ID>. The eplusout.csv in each subfolder holds the hourly energy and heat emission results.
• Building metadata: building_metadata.geojson file holds the type, vintage, building size, and centroid geometry of the compiled

Column name	Column definition
OBJECTID	Unique building ID inherited from LARIAC6_LA_County.geojson
GeneralUseType	Building type, inherited from Assessor_Parcels_Data_-_2019.csv
SpecificUseType	Building type, inherited from Assessor_Parcels_Data_-_2019.csv
EffectiveYearBuilt	Built year, inherited from Assessor_Parcels_Data_-_2019.csv
building.type	Prototype building type
vintage	Prototype building vintage
idf.name	Prototype model filename used to simulate the building
idf.kw	Prefix of folders holding EnergyPlus simulation results
usetype	Corresponding EnergyAtlas usetype
FootprintArea.m2	Total building footprint area [m2], inherited from SQFTMain column in Assessor_Parcels_Data_-_2019.csv
building.area.m2	Total building total floor area [m2]
id.grid.coarse	The grid cell in the 12km x 12km grid system containing this building
id.grid.finer	The grid cell in the 450m x 450m grid system containing this building
id.tract	The census tract GEOID containing this building
geometry	Point of building centroid

Building height info is also available upon request.

• aggregated heat emission and energy data for the three spatial resolutions.
  • finer grid 450 x 450m
  • coarser grid 12km x 12km
  • census tract

All files have the same column structure. The hourly_heat_energy folder contains the compiled hourly heat emission and energy consumption data in the following format. The "annual_2018.csv" holds data at the 12x12km grid level. The "annual_2018_finer_01.csv" through "annual_2018_finer_12.csv" holds the data at the 450m x 450m level. Each data file corresponds to a month. The "annual_2018_tract.csv" holds the hourly heat energy data at census tract level. The files with "2016" in the file names corresponds to 2016 energy and heat emission data.

Column name	Column definition
geoid	WRF grid ID or census tract GEOID. Use the geojson for the corresponding spatial resolution to look up the location and shape of the GeoID
timestamp	Hourly, local time of LA county.
emission.exfiltration	Zone exfiltration heat loss [J]
emission.exhaust	Zone exhaust air heat loss [J]
emission.rej	HVAC system heat rejection [J]
emission.rel	HVAC system relief air heat loss [J]
emission.surf	Surface heat emission [J]
emission.total	Total heat emission [J]
energy.elec	Total electricity consumption [J]
energy.gas	Total gas consumption [J]
energy.total	Total electricity and gas consumption [J]

• Aggregated geographical data referenced in heat emission and energy consumption. The "geo_data" folder contains the grid and census tract polygon shapes and associated area data with column layout as shown in the following table.

Column name	Column definition
geoid	WRF grid ID or census tract GEOID
geometry	Polygon shapes of the WRF grid points bounding box or census tracts
FootprintArea.m2	Total building footprint area [m2]
building.area.m2	Total building total floor area [m2]
area.m2	Grid or census tract polygon size

Contributing modeling software

Model	Version	Repository Link	DOI
EnergyPlus	22.1	https://github.com/NREL/EnergyPlus	https://doi.org/10.1016/S0378-7788(00)00114-6

Reproduce the data set

The following is an overview of the workflow

Following the steps to reproduce the analysis

1. Compile a LA county geojson file with building footprint, type, vintage, number of stories, and footprint area using the geometry and assessor data files from the Dropbox folder "City Data/LA".
2. Acquire WRF climate data (in a 12 km x 12 km grid system)
3. Convert the WRF climate data of the historic forcing to epw.
   1. Untar the "M02_EnergyPlus_Forcing_Historical_LowRes*" folder. This creates a folder to hold all the files from the tar ball
   2. Run 0_mv_inputs.py to create folder structure and move files to corresponding folders. "forcing_folder" is set to be where the WRF data is extracted, i.e. from above.
   3. Copy the "USA_CA_Los.Angeles.Intl.AP.722950_TMY3.epw" into folder
   4. Run 2_wrf_to_epw.py to before this line. Note that the WRF_FOLDER at the beginning should be set to as well. df_wrf_data = pd.read_csv(os.path.join(WRF_FOLDER, 'time_series', 'LA-SOLAR.csv'), sep=',', encoding='UTF-8')
   5. Get solar radiation input with get_solar_input.R. This creates a file "LA-SWDOWN_input_to_excel.csv" in the /time_series folder. The file looks like this
   6. Open the file and copy data to the excel tool Los_Angeles_TMY_2010s Solar irradiance.xlsx. Note that for leap year, we should not copy in the Feb 29 data, as the excel tool won't accept that. Also due to UTC to local time conversion, there is a time shift, the UTC time 2018 does not cover the whole local time 2018. We’ll use the year-end data of the previous year to fill in for the missing hours of the current end of year. Then paste the W, X, Y column in sheet "Output – Isotropic sky" to "LA-SWDOWN_input_to_excel.csv". For Feb 29 data, paste it in the "input" sheet in place of Feb 28, then get the output three solar component of Feb 28 from "Output – Isotropic sky", and paste them in the Feb 29 slots in the csv. Change the column names of the three solar component to "sw_normal", "sw_dif", and "sw_dir". Save the csv file as LA-SOLAR.csv
   7. Go back to 2_wrf_to_epw.py, now the previous line should run through and read in solar data to df_wrf_data.
   8. Check whether there are missing data in the generated .epw files using check_epw_err.R. Missing values are most often in Dew Point. If there are missing values, use fill_na_in_epw.R to fill the missing value with the previous non-missing record.
4. Assign the nearest grid point to each building. The epw files for the assigned grid point will be used in the simulation of the target building. This is documented in the rmd/1_match_building_to_grid.Rmd file
5. For each building type-vintage combination in each grid cell, simulate the historic forcing. There are xx possible prototype buildings and 3 possible vintages, but we will only simulate the type-vintage combination appearing in each grid cell (see xx for the mapping from grid cells to type-vintage combination). Use "3_write_baseline_idf.py" adapted from im3 repo from Xuan to create EnergyPlus models.
6. Adjust prototype models
   • Change the design condition, using 3_idf_preprocess.R. The script first copies the idf files from input_data/annual_WRF into intermediate_data/idf_to_modify, then change the design day and location of the idf files then output them to intermediate_data/idf_change_design_day
   • Add run period and output variables using 3_write_baseline_idf.py: takes idf from intermediate_data/idf_change_design_day and add runperiod and output variables, outputs to intermediate_data/idf_add_sim_period_output
   • Remove un-used dependencies from files: 3_replace_schedule_csv_path.R
   • Create heavy and light manufacturing facility models by adjusting the electric and gas equipment to match the EUI of the MECS 75th and 25th percentile
     • retrieve the 25th and 75th percentile using get_manufacturing_energy_stats.R
     • adjust the model electricity and gas equipment using 3_get_manufacturing_idf.R
   • Update model version using idfVersionUpdater.exe
   • fix errors in the Religious model by running 3_fix_religious.R. The original model have some errors of missing objects and wrong value for the start day in the RunPeriod object.
   • fix a field in nursing home model using 3_correct_nursingHome.R
   • change the year and "Day of Week for Start Day" in the RunPeriod object to match the actual simulation year
7. Use run_sim.py to run simulations:
   • evaluate run_multi_thread(k) function with the desired number of thread to run the simulation combinations specified in "df" at the beginning read from intermediate_data/epw_idf_to_simulate.csv.
   • use test_run_all_model() to simulate all idf files in the same folder using a certain .epw file. This function is used in generating the simulation result in the verification against ResStock ComStock, and various scorecards
8. Validation with measured and other data source
   • county level: see details in verify_county.R
   • county neighborhood: verify against Energy Atlas neighborhood data in 2016. See details in verify_county.R
   • verify with ResStock and ComStock data for LA county. See details in verify_resstock_comstock.R.
   • verify with PNNL prototype model and NREL reference model scorecards. See details in verify_scorecards.R
9. Produce grid-level heat emission data. The hourly grid-level or census tract level heat and energy data is compiled in aggregate_to_grid.R, by running compile.grid.data() for certain grid level (or census tract) and year.
10. Produce building metadata geojson: use join_building_to_tract_finer_grid.R

Reproduce my figures

Refer to rmd/figures.Rmd

Comparison of 2018 and 2016 results

2016 results is also available for hourly AH and energy consumption. Comparing with the 2018 results, the county total AH is 1.2% lower and energy consumption is 0.3% higher, with an annual average dry bulb temperature difference of 0.03°C. Monthly difference between the two years are more substential, with as high as 32% difference in AH in January with a 3.3°C difference in average dry bulb temperature.

The following figure plots the average monthly dry bulb temperature, relative humidity, and wind speed. 2018 has hotter weather from July to September and in January. 2018 is colder in February and March.

This figure compares the total heat emission and energy consumption for the whole county. We can see the heat emission between the two years is pretty similar. Energy consumption differs a bit more. The difference generally follows the temperature trend.

The following figure shows the AH without the surface component. The AH from non-envelope components is more pronounced between the two years and still follows the temperature trend.