Aug 23, 2024
Solar Availability Series Part 2: Measurements and Metrics
Welcome back for Part 2 of Camelot’s series on solar availability, which is an appropriately hot topic as the industry continues to mature. If you’re just joining us for the series, Part 1 can be found here, and it includes some background on the current state of industry assumptions. Today we’ll cover the not-so-simple task of calculating and reporting downtime, along with some implications. Subsequent parts will describe ways of maximizing availabilities and Camelot’s official stance as an IE. Thank you for joining us!
Introduction
As expressed in Part 1, availability is a way of quantifying lost generation potential due to outages; it measures whether a component or system is operating when it ought to be. An availability of 100% at any given time means everything is operating when it should, whereas an availability of 0% means the entire component or system is offline. The plot below illustrates a case where the entire site stopped producing power and was restored the following day. There will be more on this figure later.
To better summarize the operations at a project based on high-resolution data collected at a site, production and availability data are typically aggregated and reported into monthly operating reports (MORs) which are shared with key stakeholders on a project. Monthly numbers are also aggregated into quarterly and annual reports. Because there is typically some seasonal variation in downtime, most folks will refer to annual availability numbers when benchmarking against expectations, and so when we talk about availability assumptions, we are referring to annual averages.
A Deeper Dive Into Metrics
The simplest but less useful measure of availability is time-based. It’s calculated as Uptime/(Uptime+Downtime), so it only considers the time it takes to bring the system back online over the period. However, the most useful measure of availability in most contexts is energy-based. It uses an estimate of the energy lost during the period, and is calculated as Actual Production/(Actual Production+Lost Production). We care more about lost production than anything; when building out a financial model, we multiply pre-downtime production by the assumed availability to arrive at post-downtime production, so we want to use energy-based availability if possible. This is often why, despite PVSYST’s ability to model downtime, the loss factor is most commonly applied outside of PVSYST; the software interprets the loss as time-based and will apply random downtime throughout the modeled year, resulting in an unintended energy-based loss. Time-based availabilities are not well suited for financial modeling, and we recommend time-based metrics only be used if they are defined and used in O&M contracts, as we’ll touch on below.
How are uptime, downtime, actual production, and lost production determined?
Uptime and downtime are relatively easily defined on a site-level. SCADA systems will typically flag periods when the site or major components are down, and the duration of these events will sum to be the downtime for the site. In cases when a portion of the site is offline, uptime is often weighted by the portion of the affected site (ideally on a production-potential basis).
Actual production comes directly from the power meter, typically at the point of interconnect (POI). Calculating lost production usually involves several steps which are all built into the software used to log and report operational data:
Determine “expected production” for each timestep based on the energy model for the site and the existing, measured site conditions (eg irradiance). The model should be validated as an accurate representation of the relationship between measured inputs and production. Referring to the plot above, expected production is the red line, which is based primarily on the plane-of-array irradiance (green line).
Calculate the energy lost for each timestep, which is represented by the “Δ” in the plot above.
Sum energy lost at each timestep across the entire reporting period.
The same calculations hold for any reporting period. To calculate an annual availability number based on monthly data, you can sum the monthly time or production values before doing the same math, or take an energy-weighted average of the monthly availability numbers.
What about data gaps or QC? Unfortunately, we see data concerns very often at operating sites, and garbage in equals garbage out. Some meters and sensors will have redundancy onsite in case one fails, but if we run into data concerns due to whatever issues arise, all may not be lost. Even in a system-wide SCADA outage or memory failure, some form of data are always being collected or modeled onsite, and inferences can be made. As a couple examples:
If an inverter power meter at a site with 5 central inverters starts to fail, but the inverter should still be online, an operator can verify the inverter’s availability using the POI (revenue) meter. The total power at the POI meter minus the power from the other inverters should roughly equal the power from the fifth inverter (“roughly” because of electrical losses and measurement uncertainties, which can generally be determined from operational data anyways).
Even if the entire site goes offline for a period of time and no actual measured data is available, besides the power flowing to the grid at the POI, high-resolution meteorological satellite data can be used. Operators can observe the relationship between the solar resource and production during a fully-operational period to fill in the gaps and define expected production.
Admittedly, many O&M providers will not go to the effort to fill in data gaps when they occur, which can lead to missing or inaccurate data. This, in turn, can lead to an inaccurate understanding of overall system performance, which in some cases can even impact a project’s valuation: availability is a key factor when reforecasting a project’s future production, and we have seen cases where missing data makes a significant difference in the uncertainty (leading to lower P99s). This is where Technical Advisors such as Camelot Energy Group can help ensure you are working with the most accurate data you can.
Not only can availability be calculated based on a fundamentally different basis (time vs energy), but we need to be careful to scrutinize what is included in the definition as well. Until now, we’ve focused on System Availability, but you might find other metrics floating around and serving other purposes. A few common terms and measures are:
System Availability - Captures all quantifiable downtime over the entire site for the entire period, with no carveouts. The following is a list of possible synonyms, noting that the definition of every availability metric should be scrutinized because they can be inconsistent:
Plant Availability
Project Availability
Operational Availability
Total Availability
Overall System Availability (OSA)
Component Availability – Captures only the availability of an individual component over a given time. These commonly include inverter availability or module availability, but can be broken into any components, including trackers. Sometimes referred to as Manufacturer Availability.
Contractual Availability – Sometimes also referred to as Guaranteed Availability, this metric is the most commonly-confused one of them all. It should be clearly defined in an O&M agreement, and the downtime it includes can vary. The denominator in the calculation is often more complicated than simple “total time” or “total production” during the period, and both parts of the equation can include carveouts for periods which are often deemed outside of the operator’s control. This is the most commonly-reported time-based availability, but we are seeing an increase in contracts which define Contractual Availability on an energy basis. This incentivizes operators to perform maintenance at more optimal (lower resource) times.
Balance of System (BOS) Availability – Includes the availability of all components other than the modules and inverters, such as wiring, mounting structures, and monitoring equipment. Sometimes also termed Balance of Plant (BOP) Availability, but as always, the definitions must be scrutinized.
Grid Availability – Captures downtime when the grid is not available to accept power generated by the project. This is the most common carveout for contractual availabilities, as it is almost always outside the control of the operator.
We hope this moderately deep dive into solar availabilities helps to put the numbers into perspective and emphasize the importance of understanding what metrics you are looking at when evaluating a project’s uptime. We can always go deeper into the topic, and we’d be happy to support with any questions you may have. The next article in this series will cover a number of ways of maximizing availability and improving your metrics. In the meantime, for questions and more details about Camelot Energy Group and our distinct attitude towards these issues, please reach out at info@camelotenergygroup.com.
About
Camelot Energy Group is a technical and strategic advisor to owners and investors in clean energy and energy storage projects, programs, and infrastructure. Guided by our core values of courage, empathy, integrity, and service we seek to support the energy needs of a just, sustainable, and equitable future. Our team has experience in supporting 7+GW of solar PV and 10+ GWh of energy storage and offers expertise in technology, codes and standards, engineering, public programs, project finance, installation methods, quality assurance, safety, contract negotiation, and related topics. Our services are tailored to a providing a different kind of consulting experience that emphasizes the humanity of our clients and team members, resulting in a high quality bespoke service, delivered with focus, attention, and purpose. Key services include: -Technical due diligence of projects and technologies -Owner’s representative and engineer support -Strategic planning -Training and coaching -Codes and standards consulting -Contract negotiation and support