2. PLANTPREDICT: SOLAR PERFORMANCE MODELING MADE
SIMPLE
Reviewed
by:
• Generate quick, contract-grade
predictions via a streamlined user
interface
• Designed specifically for utility-scale
solar
— Sub-hourly and multi-year predictions
— Direct weather download
— Built-in spectral correction
— Cloud-based application
• Independently reviewed and
benchmarked against more than 1 GW
of operating facilities
3. THE BENEFITS OF PLANTPREDICT
• Reduce prediction time by up to 75%
— Easy to learn, short learning curve
• End-to-end utility-scale modeling
— Built-in MV and HV transformers, Tx lines
— Built-in availability and LGIA losses
— No need for pre- or post-processing
• Pre-loaded with industry standard weather,
module, and inverter files
• Optimize your design with “Clone” and
“Quick Edit”
• Cloud-hosted for ease of sharing and data
security
4. Solar position: NREL’s Solar Position Algorithm
Decomposition Model: Erbs, Reindl, or DIRINT
Transposition Model: Hay or Perez
Incidence Angle Modifier: ASHRAE, Sandia, or user defined
Spectral Correction: 1 and 2 Parameter (Pwat, AM), Sandia, or user defined
monthly
Shading: 2D trigonometric
Module Temperature: Heat Balance or Sandia
PV Module IV Curve: 1-diode model with and without recombination
Degradation Model: Linear DC option
Transformer and AC Losses: Up to 6 transmission lines or transformers
Other Losses: Availability, LGIA limit, and auxiliary losses included
Degradation Model: Stepped AC or Linear AC
PLANTPREDICT MAIN CALCULATION METHODS
Irradiance
Modeling
Effective
Irradiance
DC System
AC System
All algorithms are published. See Resource Center for more details.
5. K. Passow, L. Ngan, B. Littmann, M. Lee, and A. Panchula, “Accuracy of Energy Assessments in Utility Scale PV Power Plant using PlantPredict,” 42nd IEEE Photovoltaic Specialists
Conference (2015).
INTERNAL VALIDATION
PlantPredict vs. PVsyst:
Comparison of 51 Simulations
PlantPredict vs. Measured Data:
Comparison of 20 Plants (>1 GW)
Mean energy yield difference of 0.13% ± 0.52% Average energy meter error of 0.41% ± 2.01%
SITESBENCHMARKING PLANTPREDICT
6. RECENT UPDATES – WEATHER IMPORT
• New weather data sources
- NSRDB (PSM and SUNY)
- CPR*
- Meteonorm*
- SolarGIS*
• Auto-recognition of common
weather formats
• Visual parsing of data columns
• Data quality check
• Soiling time series
*Limited free trial (10 runs) unless vendor login provided
7. RECENT UPDATES – API RELEASE
Generate quick, easily altered predictions via the API service
%
Useful for O&M, optimization, or testing
Contact support@plantpredict.com for more information
ONE HUNDRED
predictions run in
less than an hour!
All modeling algorithms visible
Easy to clone and compare predictions
Regularly updates according to new industry technologies
Cloud-based for easy sharing and enhanced data security
Cloud security – data is safe if you lose your computer
Can share predictions within or outside company, as desired
Built-in weather download functionality
Industry supported
Validated by IEs
By running quick, accurate predictions, your engineers can use their time doing real work instead of spending hours clicking around in clunky software.
PlantPredict is so easy to use, by the end of this webinar I have full confidence that if you know how a solar plant works, you can run a model in PlantPredict.
With everything you need for a utility-scale prediction, there is no longer a need for Excel workarounds for spectral modeling, unavailability, transformers, transmission lines, or any complex plant architecture.
It comes pre-loaded with hundreds of industry standard files, and we have made it easier than ever to import files from your desktop or other tools
With the super-handy “Clone” and “Quick Edit” features, you can infinitely tweak and optimize your plant for tilt angle, row spacing, or any desired parameter
PlantPredict is hosted entirely on the cloud, so your data is completely secure and off your desktop. The added benefit is that you can easily share files and predictions within your company, without having to export and email any files!
How long did it take to run the 100 predictions?
This report was prior to the drop in module prices