CAISO Issue Paper

Integration of Solar Energy into the Participating Intermittent Resource Program (PIRP)

DRAFT August 10th 2007

Revised February 8, 2008

Prepared by

Jim Blatchford

Sr. Policy Representative

CAISO

John Zack

AWS True Wind, LLC.

Integration of Solar Into the PIRP

ISSUE PAPER, Revision 1, February 6, 2008

Introduction

The California ISO (CAISO) established the Participating Intermittent Resource Program (PIRP) to ensure the successful integration of Intermittent Resources into the market and operations of the California grid. The CAISO tariff's original focus was on wind resources but also took into account other intermittent resources including solar energy production. The CAISO tariff states:

Eligible Intermittent Resources other than wind projects that wish to become Participating Intermittent Resources will be required to provide data of comparable relevance to estimating Energy generation. Standards will be developed as such projects are identified and will be posted on the ISO Home Page.[1]

Solar energy production is more predictable than wind energy production, but it does have distinct components of intermittency. For instance, energy production is obviously non-existent during the twilight to dark periods, but it will also be reduced due to cloud cover, dust storms or even high winds that may affect the focus of the solar beam into thermal troughs. Because of its intermittency, solar energy production is included in the California Renewable Portfolio Standard (RPS) and the PIRP.

The purpose of this paper is to establish guidelines for the integration of solar energy production (concentrated and photovoltaic) into the PIRP.

Eligibility

In order for a solar power producer to participate in the solar component of PIRP, they must meet ALL criteria set forth in the Appendix Q EIRP of the CAISO Conformed Simplified and Reorganized Tariff dated Apr 6, 2007, except for those paragraphs directly related to wind production. To participate in the PIRP, solar energy production cannot be augmented by fossil fuel generation devices. Although the California Renewable Portfolio Standard[2]allows for a de minimis amount of fossil fuels to augment the production, because augmentation of the hourly energy production for a solar Participating Intermittent Resource (PIR) could unfairly advantage the other PIRs.

Physical Site Data

A solar farm must provide the CAISO with an accurate footprint of the site before a forecast can be produced. The footprint must include (1) the location (latitude and longitude coordinates), and elevation of meteorological collection devices, (2) the location, elevation and orientation angles of arrays or concentrators, (3) the generation capacity of the facility and (4) the type of solar generation technology employed at the facility. For redundancy purpose, each solar farm must provide a minimum of 2 meteorological stations with an independent power source.

Meteorological and Production Data

As outlined in the PIRP, meteorological data must be provided to the CAISO via the Data Processing Gateway (DPG) for accurate power generation forecasting. The amount of global solar irradiance that falls on the solar array (either photovoltaic or solar thermal) accounts for about 90% of the variability of the power output of the solar array. So the measurement and forecasting of this parameter is essential for any solar power (MW) prediction system.

The total global solar irradiance that falls on a horizontal surface is from two sources, direct and diffuse solar irradiance. During clear sky conditions,the dominate source of solar irradiance is the direct form, which typically accounts for about 80-85% of the variability of the power output of the solar array. The remaining variability would be from the diffuse solar irradiance which is a measure of the solar energy arriving at the Earth's surface that is the result of scattering of the Sun's beam due to the various atmospheric constituents. Diffuse solar irradiance accounts for much more of the variability during cloudy sky conditions and under thick cloud it can account for 80 % or more of the variability.

The second most important factor in the performance of the solar array is the temperature. For a photovoltaic (PV) array, it is the back panel temperature that is particularly important. In general, temperature accounts for slightly less than 10% of the variability of output of the solar array.

The third most important factor is the wind, but it accounts for less than 1% of the variability of output from a solar array. The primary impact of the wind is the ventilation factor of removing heat away from the array.

Considering the elements and factors that influence the performance of the array, the required data from the production site should include:

  • Real Time MW production
  • Global horizontal irradiance in watts/ m2, accounts for ~90% of variability[3]
  • Diffuse horizontal irradiance in watts/ m2 accounts for ~10% clear sky and up to +80% of cloudy sky variability
  • Direct normal irradiance in watts/ m2 accounts for ~80% of clear sky and ~10% of cloudy sky variability [4]
  • Ambient temperature at the array height in ◦C accounts for slightly less than 10% of the variability
  • Back panel temperature for PV-type arrays in ◦C accounts for less than 1% of the variability
  • Wind speed and direction at the array height in m/s and degrees accounts for less than 1% of the variability

The Global horizontal irradiance, diffuse horizontal irradiance and direct normal irradiance measurements can be obtained with the use of two pyranometers. One pyranometer for the global horizontal solar radiation and one for the plane of array solar radiation. The diffuse horizontal irradiance can also be calculated from the measurements provided by these two devices.

The ambient and back panel temperature will require two separate temperature probes. In addition the ambient temperature probe will require a probe shield to protect it from the direct solar radiation. The wind speed and direction requires a mounting mast and standard cup anemometer and wind vane device.

There will be other miscellaneous pieces of equipment required such as masts mounting hardware and surge protectors. Table 1 gives the devices needed along with the units and costs associated with the installation of the equipment. It should be noted that the all cost estimates are based on current costs and are subject to change and the cost of communication equipment is not included in these estimates.

Table 1. Equipment needed and estimated costs to make required measurements[5].

Element / Device (s)
Needed / Units / Var / Equip Cost / Install
Costs
Global, diffuse horizontal irradiance & direct normal irradiance / pyranometers for global horizontal & for plane of array solar radiation / W/m 2 / ~90% / $1600 - $2000 / $800 - $1000
Ambient temperature (height to be determined) / temperature probe and shield for ambient temperature / ◦C / ~10% / $225 - $300 / $100 - $150
Back panel temperature for PV type arrays (height to be determined) / temperature probe for back panel temperature / ◦C / ~1% / $90- $120 / $100 - $150
Wind speed and direction (height to be determined) / Cup anemometer, wind vane and wind mast / m/s deg / ~1% / $460 - $500 / $460 - $500
Other Miscellaneous Equipment / Masts, mounting hardware, surge protectors etc. / N/A / N/A / $1050 - $1200 / $1000 - $1100

Note: It is recognized that the relevancy of each parameter will vary between CSP and PV application and will be addressed with further stakeholder input.

Production and meteorological data will be collected for a minimum of 60 days before the farm is considered in the PIRP. This data needs to be collected in advance in order to train the forecast models (e.g. artificial neural networks) responsible for producing the power production (MW) forecast for each site. The forecast service provider requires high quality, continuously streaming data to provide an accurate forecast.

Solar Power Prediction System

The PIRP solar power prediction system will use a short term forecasting approach based heavily on the use of satellite-based data and recent measurements from the solar generation facility for the 0-6 hour portion of the forecast period and a numerical weather predication (NWP) approach for hours six and longer of the forecast period. The statistical methods will be configured in a manner that ensures a smooth transition the predictions from those produced by the short-term prediction scheme to those generated by the longer-term approach. This will be done in a manner that is analogous to the transition to short to longer term approach in the PIRP wind forecast system.

The reason for the need for the different approaches is the result of the strengths and weaknesses of the forecasting methods. Short term forecasting methods are best for capturing short-term trends in the weather and power production rates. But such methods tend to have a large error after about 6 hours. NWP has a problem for approximately the first 4-6 hours due to the need for such physics-based models to have time to "spin-up", but such models tend to perform quite well from after 6 hours. A key component to both approaches is the incorporation of neural net and statistical forecasting methods that help to find and compensation for any NWP model or method-based systematic errors that may occur for any given solar plant.

Anticipated Forecast Error

Studies have shown that the accuracy of a solar power prediction system is highly depended on the types of local weather conditions and the time of day . [6] In general such solar irradiance prediction systems are quite accurate with an MAE of 3- 4 % of the solar irradiance for clear sky or consistent cloud cover that persist for more than an hour. However, the MAE increases to as high as 15% of actual irradiance for situations when small scale convective cloud elements develop that have life cycles on the order of 15 - 20 minutes. This type of condition is most likely to occur during mid to late afternoon during the warm season.

Outage Data

If the solar farm is reducing its production from its stated maximum production value (pMax), it is the responsibility of the solar farm (or its Scheduling Coordinator) to provide the CAISO with plant outage information via the CAISOs Scheduling Logging for the ISO of California (SLIC) reporting system. This data is needed to ensure the MW forecast does not exceed the plants derated capability.

Explanation of Terms

Global solar irradiance isa measure of the rate of total incoming solar energy (both direct and diffuse) on a horizontal plane at the Earth's surface.

Direct (normal) solar irradiance is a measure of the rate of solar energy arriving at the Earth's surface from the Sun's direct beam, on a plane perpendicular to the beam.

Diffuse solar irradiance is a measure of the rate of solar energy arriving at the Earth's surface that is the result of scattering of the Sun's beam due to the various atmospheric constituents.

Satellite-derived measurements of solar irradiance (both global and direct solar irradiance) are possible through the use of computer models. A model of solar irradiance uses radiation measurements from the visible-radiation channel and visible cloud imagery from geostationary meteorological satellites to estimate ground level global and direct irradiation. Diffuse irradiance can be calculated by using the relationship: diffuse = global – direct. For the current generation of geostationary meteorological satellites, the ground resolution is about one kilometer for the visible-radiation sensors. Studies have concluded that satellite-derived measurements of solar irradiance are more accurate to use than ground based observation if the ground based observing site is more than 25 km away from the site of interest.

C:\Documents and Settings\jblatchford\

My Documents\PIRP\Solar\

Integration of Solar into PIRP.docPage 110/8/2018

[1]Conformed Simplified and Reorganized Tariff as of April 6, 2007 Appendix Q

[2]Ca RPS defines de minimus as 2% for new facilities and 5% for existing facilities measured on an annual basis of electricity production.

[3]

[4]

[5] Costs are provided as estimates and are neither a minimum nor maximum requirement.

[6]