urban-sim-flow

Urban Simulation Framework

WIP

src folder structure

file name	Usage
adaptors.py	adaptors for converting cbecs data to processor input
construction_meta_validator.py	validate construction meta json file (../resources/construction_meta.json)
constructions.py	construction processor
generate_hvac.rb	ruby osstd hvac generation call file
geometry.py	geometry processor
geometry_settings.json	geometry processor config file
hvac.py	hvac processor
hvac_settings.json	hvac processor config file
loads.py	loads processor
loads_settings.json	loads processor config file
new_hvac.py	temperary hvac testing file for new hvac systems, not used in production workflow
preprocessor.py	proprocessor used for organizing adaptors
processed_loads_sample.json	temporary file to show loads input format
processed_schedule_sample.json	temporary file to show schedule input format
rbcalls	temporary folder for ruby calls testing
recipes.py	utility functions imported and used in other places
reference-2020-04-30.bib	accidentally added ? lol ;D
schedule.py	schedule processor
schedule_database.json	used by schedule processor, probably should be put in resources folder
schedule_preparation.py	used by schedule processor
schedule_settings.json	schedule processor config file
standard_excel_processor.py	generate raw input json files from excel spreadsheet
workflow.py	python cells to execute the whole workflow

Parallel simulation workflow

• Insure you are using constance7a since normal constance doesn't have singularity.
• Currently, the source data comes from input/cbecs-standardized-200715.xlsx. Make any edits to input data here.
• cd into src/ subdirectory
• Run "sbatch run.sbatch"
• Track any output/error messages in the run.out and run.err text files

Usage

(Airflow adoption on the new implementation is on hold.)

• Running the code requires the following environments:

  • Python 3.5+ (The code is tested with 3.7.5)
  • Apache-Airflow (The code is tested with Airflow 1.10.7 and its defautl DB of SQLite3)
  • Linux (I had some issues setting up Airflow on Windows. Running Airflow using Docker or WSL on Windows 10 are tested and good)

• Follow the quick start guide to start the airflow service. Make sure to have environment variable AIRFLOW_HOME set up before running the the airflow db (airflow scheduler) and webserver(airflow webserver -p 8080). The database only needs to be initialized (airflow initdb) once, unless there are running errors. Before initializing the database, also make sure that AIRFLOW_HOME is set properly.

• If AIRFLOW_HOME is empty, then the airflow running folder is default to ~/airflow.

TO-DOs

• schedule change to non stochastic
• occupancy control: add occ based lighting control through schedule factor
• (code change) exterior fixtures: add one object (exterior lighting), with two inputs, schedule and design level.
• control electrical equipment schedule: similar to occupancy based lighting control
• (code change) improve electrical/gas equipment efficiency: add obe object for gas equipment
• (code change)shading overhang: add one object. one std input (overhang depth)
• (potential code change) vestibule / improve door: don't add door, but change infiltration rate and schedule. add std input: nodoor / door w/wo vestibule
• advanced control operation: change operation schedule
• HVAC setpoints
• (osstd code review before code change) economizer: check if economizer operation can be switched.
• (osstd code review before code change) DCV
• (osstd code review before code change) heat recovery: check ERV code in OSSTD
• (osstd code review before code change) SWH: check OSSTD. std input: swh efficiency, water temp setpoint.

Updates

• 05/04/2020:
  • modify geometry processor to accomodate no core zone cases. Tested with CBECS 2 (w/ core) & 3 (w/o core).
• 05/03/2020:
  • Schedule preprocessor in place.
• 04/29/2020:
  • Construction processor and preprocessor in place.
• 04/25/2020:
  • Loads preprocessor in place.
• 04/20/2020:
  • Add standard input loading
  • modify unit conversion to include post-conversion units and addtitional conversions
• 04/21/2020:
  • Add multi-storey capacity
  • Add schedule generator interface.
  • Add loads generator interface.

Dependencies

The following python pakcages need to be installed for using the framework

• eppy
• geomeppy
• esoreader
• apache-airflow
• notebook
• pandas

Developer tools utilized

We use black to format python code of this repository

Note

• Apache-airflow is used for workflow implementation and management. This modulize the framework and provides lots of features to run and manage the process. I also tried Metaflow and decided to go with Airflow because:

  1. Airflow has a feature-rich web-based UI, while Metaflow is purely CLI
  2. Metaflow has better documentation and community support, while Metaflow is just open-sourced
  3. Metaflow is AWS embeded, while Airflow are less oppinioned
  4. I tried to build AWS service for Metaflow and it is very non-trivial with bad documentation, while Airflow has AWS Batch operator already included (not tried yet)

• Sample I/O files are included in the repo at this moment for development purpose.