Step 1: Setup Automated POSTing of Solar Farm Power Output Data to our API

You can schedule a regular update every week, every day, or as regularly as every 15-30 seconds. The minimum is once a week.

  1. POSTing must be done using JSON format and must use a HTTP POST command to the correct resource ID.

  2. Power output data can be provided using the 'measurement' or 'measurements' fields ('measurements' is used for POSTing more than one interval of data), and should be provide as averages of power output in MW, as calculated to the end of the averaging period.

  3. You must use UTC timestamps to the ISO8601 standard. And should timestamp the 'period_end' field to the UTC time of the end of the period.

  4. The average 'period' is in ISO8601 standard for a time duration. For example, 'PT5M' is a 5 minute duration.

  5. Please refer to the API documentation for additional information.

Step 2: Edit PV model metadata for your Utility Scale Site

Basic steps

  1. Open the PV model settings

  2. Locate a day with no clouds present

  3. Using this day as a reference, adjust the settings in the PV model

  4. Review changes by comparing Solcast model actuals with actuals measures power

  5. Repeat steps 3 and 4 until there is a good fit between the Solcast model and actuals measurements

  6. Save your changes

Worked Example 1

  1. On the site details page, select the accuracy button and review the graph. The goal is to review the fit between Solcast actuals (in orange) and actual power measutments (in green). At this example site, we can notice the following items

    a) Actual power is significantly lower than the Solcast model on all days. This could indicate either: incorrectly specified AC and DC capacity, incorrectly set losses for the site, or some operational issue (ie the site was undergoing maintenance and was not generating at 100% capacity). Use your knowledge of your site to make a judgement call here.

    b) There appear to be some grid constraints on May 10 and May 11 - these days should be ignored in the following steps

    c) There are no measutments being sent at night, which is caused a graphing error that appears to show generation at night - this should be ignored in the following steps

  2. Open the site settings by clicking on the Edit Site Metadata button.

  3. This site had incorrectly specified DC capacity. After modifying the DC capacity to the correct value, use the Step 4 page to review and save the changes. The change has had a significant effect, with the updated Solcast model (in green) aligning with the actual power measurements (in orange). We can also see the MAE and RMSE error has reduced significantly (located in the red box).

Worked Example 2

  1. On the site details page, select the accuracy button and review the graph. The goal is to review the fit between Solcast actuals (in orange) and actual power measutments (in green). At this example site, we can notice the following items

    a) Actual power is slightly greater than the Solcast model. This likely indicates a minor adjustment in losses is required.

    b) There are no measutments being sent at night, which is caused a graphing error that appears to show generation at night - this should be ignored in the following steps

  2. Open the site settings by clicking on the Edit Site Metadata button.

  3. After reducing the Solcast model Other Losses Derating Coefficient, use the Step 4 page to review and save the changes. The change has had the desired effect, with the updated Solcast model (in green) better aligning with the actual power measurements (in orange). We can also see the MAE and RMSE error has reduced (located in the red box).

Worked Example 3

  1. On the site details page, select the accuracy button and review the graph. The goal is to review the fit between Solcast actuals (in orange) and actual power measutments (in green). At this example site, we can notice the following items

    a) There are no clear sky days, all days have some cloud activity making it difficult to locate any model error

    b) There are no measutments being sent at night, which is caused a graphing error that appears to show generation at night - this should be ignored in the following steps

  2. Make no changes. There is little evidence the model needs any adjustment. Review in a week when there is an additional 7 days of data.

Worked Example 4

  1. On the site details page, select the accuracy button and review the graph. The goal is to review the fit between Solcast actuals (in orange) and actual power measutments (in green). At this example site, we can notice the following items

    a) There is one clear sky days, which appears to indicate the Solcast model is too low. Alternativity, this could be a day where the Solcast model estimated clouds to be more opaque than reality.

    b) Every other day the model seems well aligned.

    b) There are no measutments being sent at night, which is caused a graphing error that appears to show generation at night - this should be ignored in the following steps

  2. Make no changes. There is not enough evidence the model needs adjustment. Review in a week when there is an additional 7 days of data.

Did this answer your question?