Skip to content

README.md

Latest commit

 

History

History
93 lines (74 loc) · 5.33 KB

README.md

File metadata and controls

93 lines (74 loc) · 5.33 KB
Raw

OCHRE Testing Scripts Using CTA-2045 HPWH Settings:

  • These scripts can be used to simulate HPWHs in residential buildings.
  • These scripts were developed to generate a CTA-2045 Implementation Guide for residential HPWHs.

Requirements:

  • OCHRE

Working Multi-Home Simulation Run Scripts (Dana):

  1. C5_Control_Sweep.py

    • This script can run as an uncontrolled baseline, where HPWHs are acting as they normally would.
      • To do this, you set Tset = Tshed = Tloadup and each of their deadbands to your baseline deadband width.
    • This script acts as an Energy grid service, using PGE's commercial ToU schedule, or using user input.
    • You can generate a list of deadband sweeps and it will aggregate results into a parquet file.
    • Users can set Tset, Tset_DB, Tshed, Tshed_DB, Loadup_DB, and mode durations.
    • This script coordinates loads to avoid impulse-like cold-load-pickups.
    • Coordination is currently randomized using weighted start-times/bins and time jitter.

  2. C6_LoadupSweep.py

    • This script performs a very long LoadUp or Advanced LoadUp and sweeps through deadbands.
    • You can generate a list of deadbands to iterate through and it will aggregate results by deadband and one with all data into a parquet file.
    • This script requires updated WaterHeater.py and Water.py files to avoid crashing simulation and missing house data.

  3. C7_longShed_Sweep.py

    • This script sweeps Shed-type temperature deadbands in a loop and aggregates data into one parquet file.
    • This is not analogous to an Energy grid service. It is an extended Shed period from 6am - midnight.
    • This script was used to stress-test configurations to measure approximate cumulative energy shifted at hour t.

  4. C8_Blackstart.py

    • This script simulates a Blackstart Support grid service.
    • WHs will have a very high cold-load-pickup and need to be carefully energized over a long period of time.
    • This script assumes a very long Grid Service event, and energizes over many hours by staggering Normal modes.
    • Staggered energization is randomized using weighted start-times/bins and time jitter.

  5. C9_Efficiency_Sweep.py

    • This script will sweep through Efficiency Levels 1-9.
    • Users can set efficiency coefficients in the LVL library.
    • Efficiency coefficients are used to change the HP-operating deadband of the HPWH
    • HP deadband:
      • Upper temperature threshold: Tset
      • Lower temperature threshold: Tset - Tset_DB * efficiency_coefficient
    • As the efficiency coefficient increases, the HP operating deadband widens, making the unit more efficient.
    • Embedded is a sweep through Shed deadbands, as well.
    • Each Efficiency Level run will iterate through all Shed deadbands and then aggregate deadband list into one master Efficiency Level parquet file.
    • The current script iterates through all Efficiency Levels

  6. C10_Reserve.py

    • A Reserve service can be simulated by asking a small percentage of aggregated loads to anticipate...the unaticipated!
    • Most loads will run as normal for the full simulation period.
    • Some loads will consume (Load Up, ALU) during a midday, hypothetical, overgeneration scenario.
    • This code needs work to better coordinate participating loads as specific call times...

  7. CX_EfficiencyExperiment.py

    • This script works similar to C9. It does not use sophisticated ramping--it only randomizes start times about the Shed region according to the jitter variable.
    • This script allows for individualized efficiency settings for Baseline, Shed, Load Up regions.
    • It sweeps from the LVL library. Variable L gets assigned by the user the efficiency coefficient that initializes a baseline for uncontrolled regions.
    • This is kind of a one-and-done script, and I don't expect it to be very useful for future testing.

ADMD scripts (Joe), Results plotter (Dana)

These scripts are used to run a Monte Carlo Simulation for any aggregated load CSV file (can be changed to parquet). - D0_Parse_Ochre_data.py - D1_Get_ADMD_data.py - D2_Plot_3D_ADMD.py or - D3_Plot_3D_ADMD_LtdTime.py or - D3_Plot_MCS_N.py or - D3_Plot_Time_N.py

  1. Upload original results CSV file to D0
  2. Upload results of D0 and run it in D1. This will take a while (10s of minutes) depending on Unit Runs (N-loads) and MCS Runs (currently set to 1000).
  3. The resultant file can be used in any of the D2/D3 files for plotting the results. Plotting files could use some work to make data input less hardcode-y.

Plotting jupyter notebooks (Dana)

These are some plotting jupyter notebooks I used to make figures for reports/thesis.

  1. X1_PlotAggregated
  2. X2_PlotViolin
    • Violin plot to show % Difference density across a 95% confidence interval
    • Separated results from morning, evening, and full day results
  3. X3_3DPlot_95CI
    • 3D plot script for Joe's ADMD outputs.
  4. X4_analyzeDB
    • Workhorse script used to disaggregate curves from aggregated results.
    • Can plot individual day curves from non-aggregated results.
    • Includes narrow bar-plot above the main subplot surface to show coordination intervals
    • iterates through label names, widths, styles, etc.
  5. X5_singleHouse_withEfficiency
    • This script runs a simulation for a single home, with efficiency capabilities.