As California's "dumb" grid gets smart over the next decade, a clever convergence between technology and PV systems is also required. Otherwise, attempts to harmonize the state's 33% Renewable Portfolio Standard are at risk of turning into a cacophony for policymakers, utilities and technologists.

California's utilities are busy hatching smart-grid plans as the state hurtles towards high renewables penetration of a predicted 12,334MW by 2020, and government agencies demand upgrades to the creaking grid system.

California's investor-owned utilities submitted smart-grid "roadmaps" for 2020 last year, and from October they will be required to file annual reports on their progress. The California Public Utilities Commission is due to publish its proposed decision in June, and adopt metrics and goals on smart-grid deployment a month later.

As of January 2012, there were 10,611 MW of signed power purchase agreements with a further 31,707 MW in the US PV pipeline, according to GTM Research.

Mj Shiao, solar markets analyst at GTM Research, told a recent SolarPraxis conference in Burlingame, California, that not all of this pipeline capacity will get built, but solar systems themselves need to get smarter and "play nicely" with the grid.

"The industry needs to push for standards so that we know how to deal with grid interconnection and grid interoperability," he said.

GTM Research's US PV figures due to be released tomorrow [March 14] puts US installations for 2011 above 1.8GW, representing another doubling of growth that follows from 2009–2010.

But challenges have also increased with this rapid growth and even bigger ambitions – and technology in some cases has not kept pace.

Fault behaviour in large-scale centralized PV systems had been a major technical challenge, said Shiao. "If we suddenly have low voltage throughout the system, a few years ago inverters would disconnect because it went under the low voltage threshold. Now we're a bit smarter. We want to test and see that it's not actually a blackout condition. We don't want to exacerbate the problem."

According to 2011 figures, 50% of the USA's residential market is in California – some 8 million homes, with 30,000 added last year alone.

But distributed generation has its own challenges, he said. "With DG we're worried about things like low voltage stability. We don't want a lot of DG to cause the voltage to rise on a feeder line and then take out all the other DG on that feeder. We're [also] worried about islanding. We don't want these systems to start generating power in a blackout.

"If there's too much distributed PV we have all this protective equipment on our grid that ensures power flows in a single direction and in order to make it flow in the other direction we have to perform costly upgrades. These are the technical challenges that we have to think about.

"Fundamentally, at the core of the challenges is this intermittency issue – we can't regulate the sun. There are things we can do to regulate the inverter on the PV system, but right now we haven't done that yet.”

Following Germany's lead could help solve these technical challenges, he said.

“We've looked to Germany for a lot of other things. Germany in December installed about 3GW of solar energy; that's almost as much as PV as the US has installed cumulatively.

"Germany figured out about three to four years ago that this was going to be a problem. So they developed these medium voltage guidelines – now inverters have to behave a certain way. They have to deal with over-frequency and active power control and they have to play nicely with the grid."

Power electronics can help harmonize renewables on the grid and balance variability or loss of power caused by dirt, temperature, arcing and module mismatch.

Degradation of PV modules over the lifetime of a solar project can lose around 0.4% performance per year, said Levent Gun, chief executive officer at Ampt, a Colorado power conversion electronics company.

"It may seem small, but those numbers add up over 25 years," he said. "Module mismatches magnify degradation. So if one module out of 20 is degraded significantly more than the others it drives the performance of the system.

"Module electronics decouple that correlation so each module is an individual power source. So as a result, you still get 20% less power from that module that was driving the performance of that string down. But you're not polluting the neighbourhood."

Levent, a veteran of the IT industry, said that it was a "mind-boggling" shift to the PV electronics to make technology decisions for the 25-year lifetime of a solar project.

He also claims that DC/DC power converters with Maximum Power Point Tracking on each PV module makes solar systems more intelligent by enabling fine-grain monitoring and remote modifications in real time to create "smart PV plants."

"Right now nobody knows what goes on at a utility-scale plant when you have 10,000 panels; you don't know if 100 of those have gone dead," he said.
"You only see if it gets to such a bad state that the inverter shuts down and modules break down – you lose the string, then you lose the inverter. But by then you've been losing a lot of energy, sometimes for days. It's a black box – now we're shedding some light so you can see what's going on and take action."

Sandia National Lab in New Mexico had yet to report Ampt's product tests, he said.

Jennifer E Granata, project lead in PV reliability at Sandia National Lab in New Mexico, said more companies were turning to optimization electronics to improve performance.

"The basic idea with DC to DC converters some of them will optimize the power coming of each individual module. Some of them will 'boost or buck' to optimize it in a certain way, for example, if you have localized shading.

"As these companies are getting better at optimization we're definitely seeing that over time modules can degrade unevenly in the field. At Sandia, we've definitely seen data that demonstrate it can make sense. You can see regularly 20–30% improvement of what you would have lost and up to 50% in some cases.

"We've seen losses in the power electronics and even in a brand new system you have some level of mismatch. We've seen that optimizers will recover that mismatch but they also have their own power draw.

"But as the system ages and you get to a greater module-to-module mismatch which can also lead to greater string-to-string mismatch which has an impact on the inverter's ability to optimize the output of the entire array. The older it gets, that's when you start to see greater payback in the optimization.

"But there are questions out there. Does it make sense to put those optimizers in the field at the beginning and assume a lifetime [cost] or to retrofit? Does it make sense in a large commercial field or utility-scale field?"  

But how clever does the PV system have to be to help smart-grid ambitions?

Granata said: "What's going to be more important for smart grid is communication with the inverter and being able to curtail output. It's not universal, but it's beyond prototype phase now."

Nameplate capacity isn't the same as actual power harvested, however. And if it strays too far from the rate agreed under Power Purchase Agreements, it could spell trouble for developers and policy makers' targets.

Initial higher costs of optimization on PV systems would eventually be balanced out over the lifetime of the project and could lower risk for investors, said Levent.

"This is what is scary to investors because they now say we can't really predict but there is an underlying statistical effect that could potentially cost you a lot more harvest. And you end up as a system owner or whoever financed the project ends up with an underperforming system not complying with the PPA. Things get nasty if you're contractually underperforming."