Understanding the X_resource_file parameters in the 03-wind-solar-battery example

[drewlwhitney]

Hello!

I'm a student at Colorado School of Mines. We're using HOPP to run an economic analysis comparing a system with only wind and solar to a system with wind, solar, and battery. We've never used HOPP before, so we're trying to understand the examples and documentation better.

Looking at the configuration file examples/inputs/03-wind-solar-battery.yaml, I want to know what the solar_resource_file, wind_resource_file, and grid_resource_file parameters are. I understand that the solar and wind files are related to national databases (and we need an API key to access them), but how do we know what those file names should be? I can tell the names are related to the latitude and longitude, but past that I do not know how to interface with the databases.

I could not determine what database the grid_resource_file is pulling from. Is it also public-access? What specifically does the grid resource entail (i.e. is that related to grid demand)?

Finally, in the data parameter, how can we acquire the urdb_label?

If there are areas in the documentation or other resources you can point me toward, I would appreciate it. Thank you!

[elenya-grant]

Thanks for this question! Below is some information that hopefully addresses your questions:

Wind and Solar Resource Files:

The solar_resource_file and wind_resource_file are not required inputs. If these parameters are not provided in the input file, then HOPP will call the API to download the resource data files. The directory to save resource files to is specified with the path_resource input parameter and defaults to /path/to/HOPP/hopp/simulation/resource_files (linked here):

site:
    data:
        lat:
        lon:
        year: 
    path_resource: /path/to/directory/with/resource/files

If wind_resource_file or solar_resource_file are not provided in the input file, HOPP will look in the path_resource directory to see if any files exist that match the naming convention described below:

• Solar resource data is saved to the folder path_resource/solar. Solar resource filenames are formatted as:
  • {lat}_{lon}_psmv3_60_{year}.csv
• Wind resource data is saved to the folder path_resource/wind. Wind resource filenames are formatted as:
  • {lat}_{lon}_windtoolkit_{year}_60min_{height1}_{height2}.srw or
  • {lat}_{lon}_windtoolkit_{year}_60min_{hub_height}.srw

If the resource filepath matches an existing file, then HOPP will load resource data from that file. If not, HOPP will download resource data using the API call and save it to the resource filepath.

If HOPP was installed following the Installing from Source instructions, then you can find the resource files you've downloaded in default folder. If HOPP was installed using pip install HOPP, then I'd suggest:

1. specifying path_resource in the input file
2. specifying the path to the .env file when calling set_nrel_key_dot_env to avoid potential errors with calling the API to download resource data (see below).

from hopp.utilities.keys import set_nrel_key_dot_env

set_nrel_key_dot_env(path = '/path/to/file/.env')

You can check your API set-up by running Example 8 or Example 9 to see if a new resource file is downloaded.

All that is needed to interface with the resource databases is:

1. get an API key, create the .env file following the README instructions, and run the set_nrel_key_dot_env() function
2. all that is needed in the input file for HOPP to download resource data is:

site:
    data:
        lat: # latitude of site within the continental U.S.
        lon: # longitude of site within the continental U.S.
        year: # use a year between 2007 and 2014
    hub_height: 80 # hub-height of wind turbine, used to determine what heights of wind resource data should be downloaded

Alternatively, you can make a python script to download the files manually, for example:

from hopp.utilities.keys import set_nrel_key_dot_env
from hopp.simulation.technologies.resource import SolarResource, WindResource

#update the path below to the full filepath of your .env file
set_nrel_key_dot_env(path = '/fake/path/to/file/.env') 

# example location for south table mountain
lat = 39.7560
lon = -105.1914
year = 2008 # random year between 2007 and 2014

#update path_resource with the folder you want to save resource files to
path_resource = "/fake/path/to/resource/file/folder" 

# random hub-height chosen
hub_height = 90

# download wind resource data
wind_resource = WindResource(lat, lon, year, wind_turbine_hub_ht=hub_height,path_resource=path_resource)
print(f"wind resource file is saved to: {wind_resource.filename}")

# download solar resource data
solar_resource = SolarResource(lat, lon, year, path_resource=path_resource)
print(f"solar resource file is saved to: {solar_resource.filename}")

Utility Rate Database (URDB) Label

The urdb_label is also not required in the hopp input file.
The urdb_label is a label specific to the Utility Rate Database that identifies a utility rate corresponding to a unique utility, rate type, sector, etc. More information is provided in the URDB API documentation. This feature is only used if running REopt within HOPP (this functionality has not been finalized and is not available at the moment.)

You can acquire a urdb_label by searching the Utility Rate Database to find a rate that you're interested in. Click the "View" link under the "Actions" column of the table for the rate you want to use, then copy the last part of the url.

For example, if the url is: https://apps.openei.org/USURDB/rate/view/67a6618c4a6d186c0c0c6870, then the urdb_label is '67a6618c4a6d186c0c0c6870'.

Grid resource file

The grid_resource_file is also an optional input. Grid resource data is not downloaded from anywhere and the grid_resource_file would likely have to be made manually and should look similar to files in this folder. There is no standard file naming convention for grid resource files in HOPP. If not input, the grid price profile defaults to the one in dispatch_factors_ts.csv.

The grid_resource_file should contain hourly grid prices in $/MWh for 1-year. The first number should correspond to the grid electricity price starting during the first hour of the year (between 12:01am and 12:59am January 1). The grid price profile impacts the battery dispatch for some battery dispatch strategies. The battery dispatch strategies that would be impacted by the grid price profile are: 'simple', 'non_convex_LV','convex_LV','one_cycle_heuristic'. The battery dispatch strategies that would not be impacted by the grid price profile are 'heuristic' and 'load_following_heuristic'.

Some examples on specifying the battery dispatch method are found in the:

• Example 4 input file, this uses the 'simple' battery dispatch method.
• Example 7 input file, this uses the 'load_following_heuristic dispatch method
• Example 8 input file and the Example 9 input file which both use the 'load_following_heuristic dispatch method.

[drewlwhitney]

You're incredible. The solar and wind resource files make complete sense now, we've been able to get HOPP to download them through the database APIs by setting the longitude and latitude in the .yaml file. I didn't realize the URDB was a linked database HOPP could access, we'll keep that in mind. I have some follow up questions concerning the grid.

I see how the grid resource file works now, I think we could get by with one of the example files. I want to know if we need to supply HOPP with information about the grid-side power demand (i.e. something in MW or kW over time). I ask because if I run something like

config_file = path.join("config", "without_battery.yaml")
interface = HoppInterface(config_file)
interface.simulate(project_life=20)

hybrid_plant = interface.system
total_missed_load = sum(hybrid_plant.grid.missed_load)
print(f"Total missed load (kW): {total_missed_load}")

with a configuration file like
without_battery.yaml

# SiteInfo
site:
  data:
    lat: 39.345795
    lon: -103.873505
    elev: 1743
    year: 2014
    tz: -7
    site_details:
      site_area_m2: 76890.27
      site_shape: "square"

  hub_height: 80.0
  solar: true
  wind: true
  wave: false

# Technologies
technologies:
  pv:
    system_capacity_kw: 2000
    dc_degradation: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
  wind:
    num_turbines: 6
    turbine_rating_kw: 1500
  grid:
    interconnect_kw: 8800
    ppa_price: 0.06229

we get Total missed load (kW): 0.0, so it seems like HOPP is not simulating the grid demand. If we do need to supply this file, where is a good place to find that information? Otherwise, what parameters do we need to change/include so that we can understand how reliable our system is in terms of meeting load requirements?

Thank you so much, y'all are the best!

[elenya-grant]

Glad to hear the previous response was helpful!

The grid demand/load profile is set in the site section of the input file under the variable desired_schedule. The desired_schedule should be an list containing the hourly load demand in MW over a year (a list with 8760 entries).

site:
  data:
     lat: 39.345795
     lon: -103.873505
     year: 2014
     site_details:
       site_area_m2: 76890.27
       site_shape: "square"
  solar: True  
  wind: True  
  desired_schedule: [100.0, 200.0, .... 100.0] #8760 length list of load demand in MW

If the desired_schedule is not input, then it defaults to an empty list, and the variable hybrid_plant.site.follow_desired_schedule is set to False (resulting in the hybrid_plant.grid.missed_load value to never be updated from its default value of 0 - see here). If you add the desired_schedule to the input file, then the hybrid_plant.grid.missed_load should be nonzero (even without a battery).

A quick tip: set the grid interconnect_kw variable to be larger than the desired_schedule or you'll get a lot of printed out warnings to the terminal:

config["technologies"]["grid"]["interconnect_kw"] = max(desired_schedule)*1e3

• Example 9 shows one potential method to analyze system reliability to meet load demand by seeing how missed load varies for wind and solar resource years (how robust is the design to changes in wind and solar resource data).
• Example 7 shows an example of setting the desired schedule in python
• Example 8 input file shows another options for setting the desired schedule in the input file (the load profile is provided in a separate file)

[drewlwhitney]

Brilliant!

I'll be more thorough looking through the documentation for finding those parameters next time. My last question: is there a good resource for getting hourly grid demand data, even if it is simulated? I've been looking at NREL's dsgrid toolkit, but I'm not sure if that's what it is designed for. If not dsgrid, does the URDB have that kind of information?

Thank you!

[elenya-grant]

In the past, I've referenced ResStock and ComStock, and it seems like dsgrid includes these datasets in some way. Please note that load profiles from these datasets are usually at 15 minute time steps, so some pre-processing is needed to convert those profiles into a HOPP-friendly format (at 60 minute intervals). More documentation on these datasets can be found here.

I'd also recommend looking at tools and datasets from NREL's Building research to see if anything stands out! Databases like EIA also has some hourly load data for each ISO region. If there's a specific location you're using, sometimes locally managed websites may have load demand information (it isn't common though).

I am not sure if URDB has that information, or at least on an hourly scale (just from what I've looked at it seems like most data is aggregated by month or the same for each day - but I am not super familiar with this database).

Finding or creating a reference load demand profile can be challenging, so I hope these resources can be tailored to your project!

[drewlwhitney]

Awesome! I'll look more into these and see what I can find. Thank you so much for helping me, my group will really appreciate it! Have an amazing rest of your day.