## PV Model Settings

Through LocusNOC, users can specify which model to use (Temperature Derate or Regression), as well as fully customize data inputs and related settings.

At a high-level, this provides the following features:

- Improved flexibility and transparency in PV modeling configuration
- Release of Regression model
- Ability to clip modeled AC Power at a user-defined value
- Performance Index alerts based on new modeling capabilities

### Temperature Derate Model

The Temperature derate model is an updated version of the Locus simple model. It adds user- configurability and the ability to clip at specified AC outputs. Users can also specify inputs from different measured weather stations and adjust the model coefficients.

### Methodology

Our implementation of the temperature derate model is based off of the

PV Watts Technical Reference. The Locus Model makes minor modifications to PV Watts model to better adjust for as built system losses on a daily basis.* The equation used is:

Modeled AC Power =*(1-S)*(I/1000)*Pdc0*(1+*(Tpanel-25))

Where:

- I = irradiance in the plane of array
- Tpanel = Panel temperature
- γ = temperature coefficient for how panel efficiency changes per degree C (default -0.47%)
- η = Inverter efficiency (default 96%)
- S = Percent system losses (default 4.5%)
- Pdc0 = array DC size

Default model parameters will be used unless specific values are provided. If AC capacity is provided, the model AC power is clipped at the specified AC capacity.

*Note: The PV Watts Technical Reference defaults to 14% for system losses. However, these losses include:

Constant Losses

- Mismatch 2 %
- Wiring 2 %
- Connections 0.5 %
- Nameplate rating 1 %
- Total Constant Losses 5.4%

Variable Losses

- Light-induced degradation 1.5 %
- Soiling 2 %
- Shading 3 %
- Availability 3 %
- Total Variable Losses: 9.2%

These factors are more useful on an annual basis rather than shorter time scales, which is the use case for our PV model. Therefore, we omit them from our default losses and instead default to 5.4%. (Total losses are calculated by multiplying the reduction due to each loss)

### Configuration

The model requires a user input DC size at the component level and configured weather data fields to I and Tpanel. Follow the instructions below to enable and configure the Temperature derate model.

1. In the LocusNOC config page, select a component or site and navigate to the “PV Model Settings” tab.

2. Enter a DC size for the components and check “Enable PV Model Settings.”

Note: The site level has a read only DC size and is taken from the sum of the subcomponents. Once you have entered DC size at a component level you can enable the model for the site.

3. Select the “Temperature derate model.”

4. Select the desired input for POA irradiance.

5. Select the desired input for Panel temperature.

6. If Max AC Output is provided, the model power is clipped at the registered AC value.

7. Click “Save PV Model Settings.”

Note: If you need to perform batch updates of PV model settings, please contact our support team for help.

### Regression Model

The Regression model is an entirely new release to the LocusNOC platform. The Regression model allows for accurate as-built modeling based on derived coefficients.

### Methodology

Our implementation of the Regression model is based off of Section 4 of the ASTM 2848 standard. This model is useful when a capacity test has been conducted to generate as-built modeled values. The equation used to generate modeled values is:

Modeled AC Power =a1*I + a2*I2 + a3*I*Ta + a4*I*v

Where:

- I = irradiance in the primary plane of array (time series data input)
- Ta = ambient temperature (time series data input)
- v = wind speed (time series data input)
- a1 = coefficient 1
- a2 = coefficient 2
- a3 = coefficient 3
- a4 = coefficient 4

If AC capacity is provided, the model power is clipped at the registered AC capacity. The coefficients for this model should come from running an ASTM 2848 capacity test, as output from PV system design modeling software, contract documents, or other sources. Typically the Regression model is run at a site level. In order to apply to a component level, each component must have its own set of unique coefficients.

### Configuration

The model requires a user to input the coefficients and configure weather data fields I, Ta, and v. Follow the instructions below to enable and configure the Regression model.

1. In the LocusNOC config page, select a component or site and navigate to the “PV Model Settings” tab.

2. Check “Enable PV Model Settings”. (DC size is not required)

3. Select the “Regression model.”

4. Select the desired input for POA irradiance.

5. Select the desired input for Ambient Temperature.

6. Select the desired input for Wind Speed.

7. Enter coefficients a1, a2,a3,a4

8. If Max AC Output is provided, the model power is clipped at the registered AC value.

9. Click “Save PV Model Settings.”

### Model Applications

Once the model is configured in the platform the output can be leveraged in a variety of applications.

### Charting tool

Plot measured and modeled power in the charting tool for a quick comparison of site or component performance. Modeled Power and Modeled Energy are available in the charting tool under the Modeled Data section.

### Reporting UI

Export Modeled values for a site, component, or the fleet. Modeled AC Power and Modeled AC Energy are available in the Reporting UI under the Modeled Data section.

### Performance Index Alerts

Performance Index(PI) Alerts are a new feature of the platform that allows the user to trigger alerts based on the ratio of measured to modeled power. Users can easily set value, time, and buffer thresholds to enable alerts on underperforming meters and inverters. Please see additional documentation on PI and PI alerts.

### FAQ

**Q:** Can I select a weather component from a different site for the PV model?**A:** It is also possible to configure the PV model for a given site or component to use a weather station from another site. This is useful in some cases where two sites are located very close to one another and share the same plane of array. Note, however, that the model may be less accurate if the selected weather station does not accurately represent the weather conditions in which the modeled PV system is operating. This feature is currently not available through the LocusNOC UI. If you would like to make this configuration, please contact our support team with the following information:

- Site name

If configuring component settings, the component node ID(s) to configure

- Irradiance:
- Node ID of weather station
- Data field
- Temperature:
- Node ID of weather station
- Data field

**Q:** Can model values rollup?**A:** Model values will rollup as long as:

- The component/site has sub-components that have the PV model enabled. The sub-components need the current version of the model enabled, legacy versions will not rollup.
- The PV model is not enabled for the component/site for which rollup is desired.

**Q:** Which Model Is More Accurate?**A:** Both PV models can be very accurate when configured with appropriate coefficients and input weather data. The regression model is sometimes more accurate when configured with regression coefficients from a capacity test, because that ensures the coefficients are well tuned to the as-built PV system’s performance.**Q:** Why does the regression model use Ambient Instead of Panel Temperature?**A:** The ASTM 2848 capacity test specifies the use of ambient temperature because it reduces the complexity and issues associated with attempting to measure temperature across an entire PV system. There tends to be variability in panel temperature across a large PV system, and panel temperature sensors can be challenging to maintain (e.g., sometimes detaching). Using ambient temperature in a regression model can lead to a more consistently accurate model.**Note, however, that with the Locus regression-based PV model, you can configure the model to use a panel temperature sensor as input. If you ran a capacity-test regression using the panel temperature sensor, such that you have appropriate regression coefficients to use with the panel temperature sensor, it is possible to configure the model this way (e.g., set Ta to be the panel temp sensor, set a4 = 0, and set v to us any data api field as a dummy placeholder field).****Q:** Are the models backwards compatible?**A:** Taking advantage of the new PV model features is backwards compatible with our previous version of the PV model. Modeled AC Power and Modeled AC Energy will appear in charting, reporting and other pages just as they have before. If no PV model configuration is provided, the LocusNOC platform will fall back to previous versions of our PV model that attempt to determine modeling configuration from the site hierarchy. Only the rollup feature is not backwards compatible.**Q:** Are the new Models available through the API**A:** The new fields are available through the External API. If you are an External API user, the updated PV model does not need the fieldParameters required with the singleDiode and simple models. The API will return results based on the configuration specified on LocusNOC.**Q:** How is the model aggregated at low granularity?**A:** At 1min, 5min, 15min and hourly granularity, the model uses weather data at the same requested granularity to calculate the model and return values. At daily, monthly and yearly granularity, we use weather data at hourly granularity, perform the modeling calculations at hourly granularity and aggregate it up the user request granularity.