Oklahoma, United States [Renewable Energy World North America Magazine] In today’s popular solar publications the focus is unerringly on solar’s leading edge: technology, technique and policy, with occasional forays into finance and sales. What is often missing is cogent discussion on solar’s trailing edge: operations and maintenance.
The omission can be attributed to a pervasive misconception that solar is a static technology: build it and it will work error-free for 30 years. This philosophy works well in a dark vacuum. But in the unpredictable cycles of weather, solar facilities require a watchful eye, intuition, skilled hands and a budget. In a sense, operating a solar array of any size is similar to raising a child. They’re expensive and each one has its own personality requiring patience and considerable attention.
As director of Asset Management Services for Solar Power Partners, I oversee the health and wellbeing of over 40 solar electric facilities (SEFs) generating more than 21 GWh a year. Our principal business is in financing, owning and managing a constellation of commercial-scale SEFs including fixed rooftop arrays, fixed elevated arrays, tracking elevated arrays and single and dual axis ground mount arrays. System sizes range widely from 100 kW up to 2.4 MW. Our power purchase agreement customers include schools, universities, airports, retail stores and municipalities. All told, our portfolio of assets is a fine cross-section of distributed solar.
It is also a daunting exercise in O&M made manageable with accurate upfront performance modeling, a reliable and effective monitoring system, active monitoring by knowledgeable personnel, realistic preventative maintenance policies and budgets, solid warranty backing, efficient reactive repair policies and targeted analytics.
This is the third of three articles focusing on O&M issues for solar electric facilities. We began in the September/October issue by examining some of the factors that contribute to an effective O&M strategy. In the November/December issue, we looked at preventative maintenance strategies, warranties and budgets. In this final installment, we look at both critical and non-critical reactive repairs. (Access the first two parts at the bottom of this article.)
In commercial systems, downtime can mean the difference between a financially successful solar energy facility and a financial burden. When modeling down time into the financial model, we make assumptions based upon historical evidence and manufacturer’s uptime claims. Ideally, Solar Power Partners would like to see no more than 1 percent downtime for any solar energy facility (SEF). This means a system is allowed 3.65 days of downtime per year, a length of time easily consumed by one inverter failure episode.
In real world experience, some systems experience downtimes of up to 10 percent (36.5 days down) and others have no downtime. As a result, we judge downtime on a portfolio basis averaged over a dozen or more SEFs. Plus, we dilute downtime risk by designing systems with multiple smaller inverters rather than a single large unit, for example 4 x 250 kw inverters rather than a single 1MW inverter.
Mitigating system downtime is the number one goal of reactive repair. Most integrators and SEF managers recognize that reaching this objective requires a strong long-term relationship based on respect, reasonable expectations, fair demands and cooperation. It is easy for an SEF manager to place the entire reactive repair burden on the integrator, the justification being that the integrator built the system and therefore knows it best. (And, of course, they carry warranty obligation.) Unfortunately, the integrator whose primary business is to build systems, not to monitor them, will quickly resent this.
To be fair, the SEF manager must attempt to identify and solve system problems in-house before passing them fully to the integrator. SEF managers who oversee many systems will have a long list of contacts from which they can gather information. The monitoring provider, host customer, component manufacturers and system integrator are quite often the first to be consulted. Sometimes a fix can be as simple as resetting an inverter or turning on a communications connection that was inadvertently shut off. Solving the smaller problems without burdening the integrator is by far the best way to maintain a healthy relationship.
As with preventative maintenance, an effective approach to reactive repairs requires an SEF manager with a thorough understanding of the SEF and its components and how they behave in different situations. An example is the effect heavy overcast skies (or fog) has on tracking systems, which make the best use of direct sunlight. When the sunlight is diffuse, as in cloudy or foggy conditions, the production curve from a tracking system will assume a similar bell-shaped production curve as that of a faulty fixed system (Figure 1a-b). An untrained eye, ignorant of the foggy weather conditions at a remote tracking array, may look at the curve and conclude the system is not tracking properly. Bells will sound and soon a crew will be dispatched to inspect a perfectly functioning array.
There are two types of reactive repair: critical and non-critical. Critical repairs are required to stem production losses and are most likely due to inverter failure. Other causes can be faulty meters, communications failures and site disruptions. Critical repairs require immediate action to limit total downtime. At SPP we require an acknowledgment of our service request by the warranty provider within four hours and a plan of action submittal within eight hours.
Non-critical repairs are those that do not necessarily hinder effective energy production but do create degraded performance. These result primarily from a loss of monitoring communications. Less frequently, broken panels (one or two), shading issues, faulty tracking and unusual soiling will have a greater impact on energy production. SPP’s response requirements are acknowledgment of the service request within 24 hours and the submittal of a plan-of-action within 48 hours. Often, non-critical repairs can wait until a scheduled maintenance visit to the site.
Acts of God, theft and vandalism lie outside the scope of warranty repairs yet often require fast responses to limit production losses. Proper measures must be placed early in the life of a SEF to meet these unforeseen challenges. Most often an agreement can be reached with the warranty holder or preventative maintenance contractor to provide reactive repair services at a fixed hourly rate that is acceptable to the owner and insurance provider. If a contract cannot be placed early on, the SEF manager will be subject to fluctuating retail costs that may exceed insurance coverage levels.
We don’t like to contemplate devastating loss due to flood, fire, earthquake or theft. But given enough time and an ever-expanding portfolio of projects, the odds begin to rise. Proper insurance coverage must be maintained to provide for system replacement or, in the case of power purchase agreement providers, system liquidation or lost revenue recovery during reconstruction. In all cases, quick resolution is necessary to maintain a positive revenue flow.
Gleaning meaning from production data is the role of analytics. It is the big picture of SEF health that allows the SEF manager to measure the effectiveness of operations and maintenance. Also, by analyzing data, a short-term road map of future preventative maintenance needs can be created (as described in Part II in the November/December issue.)
Most energy production estimates are based on mid- to long-term goals of one month or longer due to the fact that predicting far-future weather in shorter time frames is extremely difficult. (The insolation on any given day of the year can fluctuate greatly from year to year, whereas the total insolation of any given month will appear nearly normalized from year to year.) The lack of reliable production estimates shorter than one month precludes most reliable data analysis in these shorter intervals. This realm is relegated to active monitoring.
The depth of analysis needs to be relatively shallow compared with active monitoring, otherwise the SEF manager will become swamped with irrelevant data. For example, string-level monitoring is terrific for active monitoring but provides little insight into system health over extended periods of time. If it does then there is a deficiency in active monitoring, allowing string problems to affect long-term production. The focus should be on actual monthly production relative to expected monthly production. From this top level of detail, the SEF manager can see the effects of soiling, photovoltaic degradation, heat degradation and weather trends. Deeper digging can help define more minute trends in cause-and-effect reports: current drop due to heat, voltage drop from degradation and so on.
Broader trends in time–quarterly and annual reports–help smooth production fluctuations further. This is important in reporting the financial performance of an SEF or a portfolio of SEFs to financiers who are more comfortable funding traditional power generation technology where production is constant and highly predictable. Ultimately, the goal is to fashion the analytics to the audience: minute details for the engineers, mid-level detail for the maintenance people and broad detail for the financiers.
As the United States begins the long march into a new, more environmentally benign future, commercial-scale solar is trending toward larger, utility-scale systems like those covering Spain. The transition will create new O&M concerns that dwarf those of today’s mid-market solar facilities in the U.S. and Spain’s behemoth installations. European-style feed-in tariffs, responsible for the growth of Spain’s solar fields, are so rich that many owners find no need to monitor their installations beyond basic functionality. We can be assured that the push into utility scale solar will not be a free lunch and every kilowatt-hour possible will be squeezed from these systems.
How best to do this is still up for discussion and raises questions about the viability of current systems to handle monitoring, maintenance and reactive repair. What we see will probably be large fields of photovoltaics broken up into smaller, more recognizable and manageable sub-arrays of 5 MW or less, complete with string-level or array-level monitoring. Permanent crews will manage the sites in an effort to maintain peak efficiency and up-time throughout the years. Costs will become ever larger and headaches will multiply. In the end, though, we will find the benefits far outweighing the effort. Systems will evolve, efficiency will rise and–before we know it–the U.S. will be producing 20 percent of its electricity from solar.
Bryan Banke is director if Asset Management Services for Solar Power Partners and has been with the company since November 2007. He was a private solar PV project analyst and feasibility consultant with Banke Energy. His career in energy began in the 1980s as an analyst with Southwest Gas Corp.