Designing for Value in Large-Scale PV Systems
Inside this Article
Designing a high-value system is rarely about maximizing total production or specific yield, but is instead a process of optimizing economic performance based on customer expectations.
Booming demand for large-scale PV power plants in the US has required an evolution in solar project developers’ and contractors’ approaches to system design and engineering. Today, not just pure-play solar enterprises but, in growing numbers, power plant construction companies are building the world’s largest power plants in the southwestern US. These companies are outfits with massive balance sheets that have built thousands of megawatts of traditional and thermal power plants but are somewhat new to solar. Designing a PV power plant presents unique opportunities, challenges and risks compared to traditional power generation.
In this article, we discuss some of the strategies and approaches we have found useful in helping EPC contractors and project developers maximize the value of their utility-scale solar projects. Ultimately, our goal is to ensure that they view solar power as a mainstream solution for meeting future energy demand.
Evolution of Value-Based Design
Not even a decade ago, we would crack open a bottle of champagne to celebrate the completion of a 100 kW rooftop project. Projects like the 1,900 kW installation on the Googleplex rooftop, completed in 2007, were really big deals—not just because of their size, but because for the first time solar projects in the US were incorporating complex design considerations to maximize value for customers. Until then, most solar projects were too small to justify significant project design consideration.
For example, if you were going to build a 5 kW rooftop system at $10 per watt, it made little sense to invest much time in analyzing product performance data and tweaking design levers. You simply would not get a decent return on your time and energy, even if you did manage to improve the system’s energy yield by 10% with no additional equipment costs. Consequently, engineers at the time kept it simple by following some crude rules of thumb to maximize the energy output of their systems, such as installing fixed-tilt arrays facing due south with a tilt angle equal to the site latitude. Design flexibility was limited because the homeowner’s roof dictated the size and shape of the solar array. Installers could do little (short of building a new roof altogether) that would allow for a more sophisticated and productive design.
With the development of the commercial market, where rooftops were often a hundred times larger than those of private homes, for the first time design became a game worth playing. The prospect of increasing return on these larger projects, by even a few percentage points, warranted investment in analytics, equipment research and project design. This was an exciting time to be a project integrator. Against the backdrop of emerging solar-friendly utility tariffs featuring unique time-of-day or time-of-delivery (TOD) rate structures, to maximize project value PV designers started tweaking design levers such as rooftop orientation, power density, module tilt and inverter efficiency. Still, space and form constraints meant they could pull a limited number of design levers to improve project economics. It was a bit like playing checkers: The game involved some strategy, but it did not allow the great players to really distinguish themselves.
The game changer was the advent of the utility-scale solar market, characterized by megawatt-scale projects, which forced designers to give up checkers and start playing chess. Maximizing value for a massive, multimegawatt, open-field PV project requires designers to consider a much wider range of design levers, variables and permutations. For one, the game board itself is not defined at the outset. Unless virtual net metering is an option, designers working on a commercial solar rooftop project know exactly where they must build the array. However, utility-scale project developers have to start with site selection and control. The ability to choose and secure the right location is one of the single biggest drivers of project success. To a greater extent than for a gas power plant or other conventional generators, a move of as little as 10 miles in any direction can significantly impact the success of a solar generation facility. Whether due to microclimates, topography, the presence of endangered species, or proximity to transmission infrastructure, ROI can vary dramatically over a mere few miles.
How to Design for Value
Major design challenges begin once the developer has secured and permitted the site and signed a PPA with an off-taker, the entity that will purchase the solar-generated energy. At this point, EPCs huddle with their equipment vendors, especially experienced module manufacturers, and develop a game plan to construct the highest-value solar power plant possible. This design team must also assess owner requirements and account for the risk appetites of project stakeholders.
Before embarking on the design of a solar power plant in a competitive bidding situation, the designer must ask two fundamental questions:
- Is this customer the long-term owner of the plant, or will the customer sell the asset as soon as it is commissioned?
- What metrics will this customer use to compare bids?
Know your customer. Developers looking to flip the project soon after it is built are like homebuilders who buy IKEA cabinets. They are looking to make a good first impression based on providing decent quality at minimal cost. Although these customers do have quality expectations, durability and craftsmanship are not their highest priorities because they will own the asset for only a few years. Long-term owners, however, tend to invest in cherrywood cabinets and high-quality hardware that will stand up to years of use. This approach costs more up front, but provides greater value over time.
From a system design perspective, the type of customer should inform the equipment selection process. Long-term owners are more willing to pay for features that improve durability and performance. On a module, these features might include a lower long-term degradation rate; a junction box with a more robust ingress protection rating, such as IP67 (waterproof) versus IP65 (water resistant); and higher wind- and snow-load ratings that reduce module flexing and microcracks in the cells. For inverters, a long-term owner may choose to purchase an uptime guarantee and an extended O&M service plan, and may require an established, big-balance-sheet brand with low default or bankruptcy risk. Absent a large population of projects in operation for 25 years, the present value of more durable, longer-lasting equipment is difficult to quantify. In general, 25 years is a long time to withstand exposure to the elements, and cheap stuff will fail.
Know the PPA. Key project design decisions flow down from the structure of the PPA. In many regions, particularly in the desert Southwest, utilities apply different multipliers to the base PPA rate based on the time of day and year a generating asset is delivering energy to the grid. For example, the payment allocation factors in Table 1 are based on a TOD schedule from Southern California Edison (SCE) for weekdays, excepting holidays. (Note that SCE may be amending these TOD factors for future projects.) Generally, the daily and seasonal production profiles of solar assets correlate well with the rise and fall of market demand for electricity, so TOD rate structures benefit both utilities, which need to meet consumer peak demands, and the solar industry, which provides a solution well suited to meeting peak demands.
While some customers still compare bids based on cost per watt installed, most now focus more on value-based metrics like cost per kilowatt-hour or levelized cost of energy (LCOE). Sophisticated customers use investment models that analyze internal rate of return (IRR) and net present value (NPV) of cash flows. As described in “Value-Based Design Metrics”, the advantage of these investment models over LCOE is that they consider revenue.
The design optimization process is like solving a puzzle in which the design team tries to maximize revenue from the PPA while controlling costs. This means that designing a high-value utility-scale power plant is rarely about maximizing production efficiency (kWh/kWp) or even total energy production alone. It is not difficult to imagine a solar power plant designed by brilliant engineers that achieves incredible efficiencies and maximizes energy yield—think smart trackers, oversized copper wiring, super-efficient modules and wide row spacing—but loses money for investors because it ignores cost considerations and the time value of energy.
Once utility-scale PV system designers understand customer expectations and metrics, they have three key design levers to maximize the economic performance of a PV project: dc-to-ac sizing ratio, tracker versus fixed-tilt mounting, and row spacing and tilt angle. While the result of adjusting an individual design lever may be minor—perhaps increasing the system production by a fraction of a percent—when taken together, these minor enhancements often have a compounding impact and can make all the difference in a competitive bidding process. Fractions of a percent matter, especially over the life of a 20-year PPA.
Sizing ratio. Five years ago, when a 1 MW system was considered huge and PV modules were priced at $3.60 per watt, systems were routinely designed with dc-to-ac ratios in the 1.1 to 1.2 range, meaning 1.1 to 1.2 MWp of dc nameplate capacity feeding into 1 MW of inverter ac nameplate capacity. The California Solar Initiative incentive program prescribed a low ratio for system design. With modules accounting for 50% or more of a project’s construction costs, capturing every possible kilowatt-hour of generation from those modules made sense. Therefore, engineers designed systems that avoided any kind of shading or power limiting that would rob the system of production efficiency.
Today, power plants operating under PPAs with TOD rate structures are designed with much higher dc-to-ac ratios, up to and exceeding 1.4. In cold, sunny weather, the dc system is capable of generating more power than both the inverters and ac system are designed to handle. Whenever this is the case, the inverter restricts dc power output by simply moving the array off its maximum power point. Since inverter power limiting results in a clipped, flat-topped power curve—as shown in Figure 1—this phenomenon is often referred to as clipping. While the same term also describes distortion in an audio waveform, there is no distortion in the voltage or current waveforms during power limiting, as an interactive inverter must always adhere to strict power-quality requirements. What suffers instead is the PV system’s production efficiency.
Why would developers spend money on extra PV modules, only to have the extra power output from those modules wasted? The first reason is the need to increase utilization of all fixed development costs and system structural costs. Developers have invested a lot of money in land, interconnection fees, lawyers and personnel to create the project opportunity. They have also built a substantial ac system infrastructure—one that includes inverters, transformers, switchgear and a substation—and they want to push as much energy as possible through that fixed investment over the life of the PPA, even if that means sacrificing production efficiency. The second reason for increased dc-to-ac ratios is that developers want to deliver the greatest possible quantity of highest-value energy, as defined in the PPA’s TOD rate structure. To capitalize on energy values that are two to three times the baseline rate, designers oversize the dc-to-ac ratio so that inverters run at full power when energy is the most valuable. The general idea is that you are willing to give away (via clipping) 2 MWh of energy at $100/MWh to get 1 MWh of energy at $250/MWh, because this nets you $50.
DC-to-ac optimization is based in part on the premise that increased temperature negatively affects PV module performance. As cells heat up during operation due to internal resistance and ambient weather conditions, operating voltage decreases, thereby dragging down performance. For example, a 300 W– rated module at 25°C and an irradiance of 1,000 W/m2 generates 300 W of power. If irradiance is constant but ambient temperature increases, causing cell temperatures to reach 50°C, the same module (assuming a -0.43%/°C temperature coefficient) now produces only 268 W of power (300 W x [1 + (50°C - 25°C) x -0.43%/°C] = 268 W).
Fortunately, module manufacturers can improve the module power and temperature relationship. For example, they can reduce the nominal operating cell temperature through the use of advanced materials designed to more quickly dissipate the module’s internal heat to the atmosphere, allowing it to run cooler. In addition, sophisticated cell technologies can improve the temperature coefficient of modules. Modules using standard crystalline PV cells have a temperature coefficient of about -0.45%/C, meaning that a 1°C increase in operating cell temperature decreases power output by 0.45%. Use of these advanced cells can reduce the coefficient to -0.43%/°C, or even -0.41%/°C.
In hot climates where modules are consistently operating at 50°C or higher, module temperature coefficient is a critical driver of system performance. As shown in Figure 2, a 0.04%/°C difference in module temperature coefficient can result in more than a 1% difference in annual energy yield. For an EPC to take advantage of a module’s improved temperature coefficient, it must collaborate closely with the module manufacturer. Most manufacturers continually update their own PAN files, which include temperature coefficients and are used in PVsyst to characterize a module’s performance parameters. However, it is important for EPCs to understand the assumptions behind the file inputs and to ensure that their module suppliers can support these inputs with real-world and statistically significant performance data.
Although smart module selection and testing can mitigate heat resistance effects, the power output of PV modules is always higher at low operating temperatures if irradiance is constant. On cold but sunny days—think crisp spring mornings—solar power plants are at peak dc output. These are the days when you are most likely to see power-curve clipping resulting from inverter power limiting. However, this often pays off later. Consider a hot afternoon in July when high temperatures reduce the dc power output. When the available power falls well below the ac system’s nameplate capacity, two consequences follow: First, you are not making good use of your investment in ac equipment; and, second, you are missing out on the most valuable revenue opportunity of the year, that summer afternoon high-TOD multiplier. To maximize economic performance, you need to increase the dc-to-ac ratio to capture more of this peak revenue opportunity, even though that will reduce system efficiency.
Tracker versus fixed-tilt mounting. When making the decision between installing a fixed-tilt racking system or a tracker, you must consider several different factors, including cost differences, land use, energy output and TOD rates. Many EPCs use an LCOE model for analysis, but this approach, without proper consideration of TOD rates, does not lead to the best design. In simplified terms, an LCOE model is the project’s all-in price divided by total energy generated over the life of the power plant. The all-in price includes up-front costs for land, construction and interconnection of the plant, plus annual O&M costs discounted back to the present day. Total energy output incorporates annual energy production estimates, which are also discounted back to the present. An LCOE model is great for finding the most cost-efficient form of energy generation, but it fails to consider the revenue side of the equation and TOD rate structures.
Consider a 100-acre piece of land in the desert Southwest that a developer has permitted for a solar project. A request for proposal (RFP) goes to the EPCs, asking for design options for the PV system up to the medium-voltage connection point at the substation. The developer will issue a separate contract for the substation and interconnection work. Should the EPCs propose a fixed-tilt system or a tracker?
At today’s prices and module efficiencies, a tracker in areas of high direct irradiance almost always has a lower LCOE, meaning it generates energy more cost efficiently than does a fixed-tilt system. In short, this is because for systems of equal capacity, a tracker produces at least 20% more energy than a fixed-tilt system, but costs far less than 20% more to build. However, if you dig a little deeper—as illustrated in Table 2—you see that the LCOE model fails to account for the marginal value of additional revenue these systems generate. On 100 acres of flat land, you might be able to fit 23 MWp on a tracker versus 37 MWp using a fixed-tilt system. Even though the production efficiency (kWh/kWp) and cost efficiency ($/annual kWh) for the fixed-tilt installation are not as good as for the tracker, fitting more capacity on that piece of land yields more total energy.
From the developer’s perspective, additional energy—and added revenue, by extension—is crucial to offsetting the high fixed costs of the overall project. Remember, the developer has already sunk a lot of money into the land, environmental reviews, interconnection studies, substation and other development costs, which an EPC’s LCOE models often do not consider. Even if it does consider those costs, the LCOE approach searches for only the most cost-efficient generation option. Instead, developers search for the option that generates the highest investment returns—usually represented as the best IRR or highest NPV—as opposed to merely the lowest LCOE. EPCs and design firms that understand how these investment models work do better in competitive bidding situations than those relying on LCOE models.
Row spacing and tilt angle. Assuming you have determined that a fixed-tilt system is optimal for a particular project and that a high dc-to-ac ratio will yield better economic performance, what can you do regarding row spacing and tilt angle? For purposes of constructability and O&M, spacing should be at least wide enough for a 4x4 utility vehicle (like a John Deere Gator) with a trailer to drive between rows to deliver modules, materials and personnel to all locations on the project site. Beyond this, tighter is usually better. Although row-on-row shading can be a concern, this typically happens in the early morning and late afternoon when the sun’s angle to the modules, and thus their power output, is very low anyway. In addition, more shading occurs in the winter when TOD multipliers and irradiance are low, so the value of lost energy is negligible.
Lowering the tilt angle can be helpful in several ways: It allows for tighter row spacing without increasing row-on-row shading; it lowers the height of the top module (the installers will thank you); and it reduces wind loading on the structure, which could lead to cost savings in the mounting system and support posts and footings. Yes, lowering the tilt angle reduces the system’s overall production efficiency and annual energy yield, but it actually increases these metrics during the summer months when the sun is higher in the sky and the energy is generally most valuable.
Design Process Recommendations
In today’s market, collaboration and compromise are essential to the design process. We recommend following these two rules of thumb: First, put together an experienced team and share information among stakeholders early in the development process. Second, do not let engineers or accountants go at it alone.
Collaboration. It is crucial that project stakeholders share information. In our experience, developers who run an RFP process by holding a group of qualified EPCs at arm’s length and sharing a minimum of information about the project site and PPA end up with average-performing systems. It is imperative to define the project’s revenue objectives and financial parameters at the outset so that all stakeholders are working in the same direction. If EPCs do not have relevant information related to TOD rates and the developer’s fixed costs, they cannot optimize their system designs. Keep in mind that equipment providers know how to squeeze the most energy out of their products, so integrating them into the process early and often is critical.
Compromise. Designing a system for maximum value is a process of finding a healthy compromise between your best engineer and your best cost accountant. Do not let your engineer design the system alone. You may end up with the most efficient PV system and not the one with the best ROI. Similarly, do not give your procurement managers free rein. Commoditized purchasing may secure the lowest-priced modules and BOS equipment and provide the best short-term ROI. However, getting the lowest price or aggressively value-engineering the system may come at the expense of quality, which could increase supply and performance risk and leave millions of dollars in unrealized generation value on the table.
How Is Value Changing?
TOD rates have been the single most important driver to date when trying to extract value from a PV power plant in the desert Southwest, which is the area where EPCs are constructing most such projects. However, designing for TOD rates is not without its inefficiencies. Right now, thousands of megawatt-hours all over the country are literally thrown away in search of peak megawatt-hours, mainly because there are limited low-cost options for large-scale energy storage. This efficiency problem is not unique to solar, as some gas-powered “peaker plants” run close to idle most of the time, waiting for the opportunity to quickly ramp up during peak hours. All of this feels a bit like throwing away perfectly good food just because your refrigerator is full or because you did not properly plan a trip to the grocery store.
Imagine if you could reroute all of that clipped dc energy in the spring from the trashcan to somewhere more valuable. With advancements in low-cost storage, you could immediately capture clipped dc power and push it through the inverter when the system is operating well below its maximum capacity due to weather variation, and do so based on grid demand. The inclusion of low-cost storage in PV system designs will likely further increase dc-to-ac ratios, inverter utilization and PV plant output reliability, allowing PV project developers to offer their customers better value.
In the future, developers will still need to design projects for value, but value will mean something different next year and each year following. Rate structures will certainly change over time as utilities and off-takers assign different values to energy produced at different hours and seasons. In addition, technological improvements, such as higher-voltage systems and higher-efficiency modules, also present opportunities for redefining project value. Clearly, the inclusion of low-cost storage would be a game changer, influencing the way you think about kWh/kWp ratios and assigning more value to clipped power.
Graham Evarts / Suntech Power / San Francisco, CA / suntech-power.com
Matt LeDucq / Brisbane, CA