Solar power, specifically the kind that runs lights and television sets and keeps the milk in the refrigerator cold, has undergone a resurgence in the U.S. since 2000. That's when the IEEE released a standard—IEEE 929-2000, Recommended Practice for Utility Interface of Photovoltaic (PV) Systems—covering the design of dc-ac inverters for systems connecting PV systems to the utility grid.
Prior to that date, even if they wanted to allow homeowners and businesses to connect to the grid, the utilities had to certify each installation individually for safety and compliance with power standards. Consequently, U.S. solar power claimed a sort of outlaw status. Most installations were off the grid, charging battery banks that supported 12-V lighting and appliances.
The off-the-grid/battery model wasn't just limited to remote farmsteads, vacation cabins, cruising sailboats, and RVs, though. According to Dan Magni, director of engineering at RWE Schott Solar, which designs and implements large solar projects, most of the company's work on off-the-grid projects with battery backup (mostly in the government sector) ended a few years ago. The last two years have seen an "explosion" of grid-tied business in the private sector. (The company's largest grid-tied project to date is a 300-kW system for the Cache Creek Indian Casino in Yolo County, Calif.) Magni's observations are echoed by Clifford B. Schrock, whose company, Solar/Wind Power of Portland, Ore., designs small residential systems in the Pacific Northwest.
MAKE HAY WHILE THE SUN SHINES
There's a completely different mindset behind these new installations. The people who purchase the systems have an incentive to reduce their electric bills through net-metering. Essentially, their electric meters turn backward some of the time.
At that point, the economics get interesting. It's true that the home or business owner may be putting extra power minute by minute onto the grid, which the meter is recording. Yet at the end of a billing cycle, the home or business probably has used more energy than it has created. That means for most installations the model is more like using the grid in lieu of a battery bank, rather than being a profit-making venture. If one does manage to put net positive energy into the grid, the utility will pay for it—but not at its regular billing rates. (See "Entrepreneuring in Solar Power Generation," p. 48).
Because this means the utility is selling less of the commodity it is in business to sell, what economic incentive does it have to let customers connect solar systems to its meters? Ignoring political pressures, there wouldn't be any incentive if electrical demand were constant over the course of a day or if it were easy to add surplus generating capacity to meet peak demand. Yet demand does vary, summer brownouts are becoming too common, and it isn't easy to get permits to build new power plants. Solar installations supply extra generating capacity for the utilities precisely when most necessary without requiring them to build new plants. Add political pressures back into the equation and it's not hard to understand why utilities now tolerate and even encourage customers to hook up Underwriters Laboratories-approved, properly installed grid-tied solar systems.
There's another reason why utilities don't discourage solar power. According to RWE Schott senior field engineer Mike McGoney, the buzz in the power-generating community is that natural gas prices will continue to be volatile, with an upward trend. Thus, the more experience the utilities can get with all forms of unconventional generation, the better.
Note that the design standards discussed in this article relate to grid-tied solar power in the U.S. (Other important standards are described in "A Convocation of Standards" at www.elecdesign.com, Drill Deeper 9323.) But solar power is a global phenomenon, and demand in the U.S. is considerably less than it is in other places. For instance, because demand in Germany is so great, suppliers are rushing to increase capacity to feed a Central European market that's estimated to be installing 1 MW of solar capacity every day. But a discussion of worldwide standards is fodder for another article.
BATTERIES NOT INCLUDED
Grid-tie systems have no batteries, which reduces system cost and complexity. Cost is significantly less without batteries, and with no need for battery-charging circuitry, grid-tie inverters are comparatively simple. In fact, the whole system is simple from an installation standpoint (Fig. 1).
Other than safety disconnects, mounting structure, and wiring, solar modules and a grid-tie inverter are all that's involved in a grid-tie system. Inverters actually integrate three functions: converting the dc from the modules to ac and synchronizing it with the utility; tracking the maximum power point of the modules to operate the system at peak efficiency; and disconnecting the PV system from the grid if power from the utility is interrupted.
Daryl Chapin, Calvin Fuller, and Gerald Pearson developed the first practical solar cell at Bell Laboratories in 1953. It was 2.5 cm2 of crystalline silicon with a conversion efficiency of about 6%.
Cell efficiency determines how well solar power works in terms of economics. Irradiance, the sun's power available at the surface of the earth, peaks at about 1000 W/m2. Insolation is the energy that's actually available at any location. The map in Figure 2 depicts average insolation for the U.S.
Efficiency numbers for conventional crystalline-silicon cells are about 15% to 20%. In the commercial arena, BP Solar recently claimed the world's record for efficiency in 125-mm diameter cells at 18.3%. Kyocera claims 14% efficiencies for its cells.
Higher efficiencies are possible—in the lab and in applications where price is no object. In mid-2003, Boeing's Spectrolab subsidiary said the latest terrestrial versions of its Improved Triple-Junction (GaInP2/GaAs/Ge) space solar cells could convert at 36.9% efficiency. Each of the cells' three layers captures and converts a different portion of the solar spectrum.
These are high efficiencies. But obviously, three layers of III-V semiconductor built on a metal-organic CVD manufacturing process doesn't lead to a low-cost product.
Emcore Photovoltaics, which acquired Tecstar's PV division, also produces triple-junction PV cells for space applications. But it hasn't claimed efficiencies quite as high as Spectrolab.
A recent collaboration between the University of California at Berkeley, Cornell University, and Lawrence Berkeley National Laboratory observed that the bandgap energy of InN is approximately 0.7 eV at room temperature. Earlier work had reported a lower value. The significance of the discovery, if correct, is that multijunction PV cells fabricated of InN and GaN to form InGaN ternary alloys would have bandgaps of 0.7 to 3.4 eV. This would cover the entire solar spectrum and provide 50% to 70% conversion efficiencies. Fortunately, InGaN alloys are said to tolerate relatively large lattice mismatches, so manufacturing could be cost-effective, even given the exotic materials.
More in line with emulating Mother Nature, MIT's Shuguang Zhang and collaborators are exploring cells that use spinach to convert light into an electrical charge (see "Spinach-Powered Portables Take 'Green' To A New Level," electronic design, Oct. 18, 2004, p. 8). They integrated a protein complex derived from spinach chloroplasts with organic semiconductors to make solar cells that are thinner and lighter than conventional cells. But Zhang cautions that the peptides currently keep the protein complex stable for about three weeks, and the cells convert only 12% of light to electrical charge.
The alternative to crystalline-silicon cells is amorphous silicon (a-Si). It can be produced in a continuous-sheet manufacturing process, making it potentially much more economical than wafer-grown cells. The drawbacks are lower efficiency and, at this time, only modest manufacturing capacity.
Uni-Solar's a-Si panels employ a tri-layer technology like Spectrolab's (Fig. 3). The process is described in a 1997 Photovoltaic Specialists Conference paper titled "Triple-Junction Amorphous Silicon Alloy PV Manufacturing Plant of 5 MW Annual Capacity," which can be found at www.physics.utoledo.edu/~dengx/papers/guha97d.pdf.
The paper cites 7.5% efficiency, but the company later said it improved that figure by 30%. In looking at these cells for space applications, the NASA Lewis Center measured a 12% beginning-of-life conversion efficiency.
Efficiency, however, isn't the end of the story. Because Uni-Solar's manufacturing process is based on materials deposited on a continuous web, production costs are lower than wafer-based processes. If space isn't an issue, low cost becomes very attractive. Previously, Uni-Solar's production capacity was limited to 5 MW per year. But a new manufacturing facility with six times the capacity, which is bound to introduce economies of scale, is now online.
FROM CELLS TO MODULES
The voltage output from a single crystalline silicon solar cell is about 0.5 V. Current is directly proportional to the cell's surface area, and designers count on about 7-A maximum output for a 6-in.2 multicrystalline cell. The typical panel module has 30 to 36 cells that are wired in series to generate a nominal 12-V output, which is actually about 17 V at peak power.
Because the cells are connected in series, and cells that are partially or completely shaded by tree branches, chimneys, or even guy wires have a high internal resistance, shading is a problem for photovoltaic modules. If even one full cell is completely shaded, the output voltage of the array will drop to half of its unshaded value to protect itself. If it didn't, the cell would be destroyed by the need to carry the current produced by the rest of the array while in its high-resistance state. If enough cells are shaded by nearby objects that throw hard-edge shadows, the whole module will drop out.
To deal with the shading problem, some modules place bypass diodes across each cell to carry the current when the cells themselves cannot. Otherwise, they would dissipate power from the unshaded cells. A separate blocking diode in each array in a parallel configuration of modules could isolate that panel if it becomes severely shaded. It would prevent other panels' current from flowing through it and being turned into heat.
An inverter is only required to be totally disconnected from the utility for service or maintenance. At all other times, whether the power circuit to the utility is open or not, the control circuits remain connected to the mains to monitor that voltage. To completely disconnect the inverter from the utility for maintenance, one must open the ac-disconnect switch required by the National Electrical Code (NEC).
Modules and inverters can be configured in several ways. With a central converter, modules are connected in series strings, and the strings are paralleled and connected to a single converter. A more efficient approach uses separate string inverters for each series string of modules, with the outputs of the string inverters paralleled. Large systems that have shading concerns may include a small inverter for each module, with their outputs connected in parallel.
One of the most critical considerations in inverter design is what to do about "islanding." An island is what happens when there's a utility blackout and the distributed-resource system is still generating power. An unexpected island can kill a utility worker or damage equipment.
Most inverters today are certified as "non-island inverters." They cease to energize the utility line within 10 cycles if the real component of the power load seen by the inverter goes to less than 50% or more than 150% of its output, or if the islanded load power factor goes below 0.95, either leading or lagging.
POWER OUTPUT SPECIFICATIONS
An inverter must synchronize its phase with the utility, as long as the utility line frequency is within a range of 59.3 to 60.6 Hz. If the utility frequency exceeds those limits, the inverter must disconnect itself within six cycles. The standards allow anything down to, but not including, 0.85 PF. They also acknowledge special cases where an inverter's design may be greater to provide reactive power compensation.
Today's inverters switch at frequencies as high as 14 kHz, so their output is much closer to a pure sinusoid. However, the standards do require less than 5% total harmonic distortion and provide tables of distortion limits for harmonics even beyond the thirty-third.
IEEE 929-2000 informs readers that most grid-tie inverters are designed as current sources. They use the utility voltage as a reference and supply the current available from the photovoltaic array at whatever voltage and frequency presented by the utility.
One area not covered by the standards concerns the inverter's task of maintaining the module string that its connected to at the maximum power point (MPP). A PV module can be operated at any combination of current and voltage found on its characteristic curve. But the inverter determines the point on the curve at which it operates, and the target is generally the MPP (though it can sometimes be the maximum voltage point). To do this, the inverter self-calibrates the panels connected to it and determines the MPP empirically (Fig. 4).
METERING AND MONITORING
Although present incentives are based simply on installing PV systems, RWE Schott's Dan Magni says that incentives are about to move to algorithms based on meeting production quotas. Thus, measurement and monitoring become particularly critical in evolving designs. While some inverter systems monitor and record data, there aren't any consistent standards yet (Fig. 5). When asked for their wish list of future enhancements, Magni and McGoney both responded that such standards took the top spot, even ahead of higher cell efficiency.