Long read: Delaying derating

Share

Fun fact: Camels have blood vessels posterior to the brain that are connected to a network of fine ventricles in their nostrils, where blood can be cooled by around 4°C, through respiratory evaporation, chilling its brain selectively in comparison with its body temperature.

The mechanism allows camels to sustain extreme heat without breaking a sweat, thus retaining their internal water reserves. Also, a camel can close its nostrils to keep out the fine dust blowing through the desert. Millennia of evolution have delivered a biological solution that power electronics suppliers increasingly require as distributed MENA solar markets emerge.

The insulated-gate bipolar transistor (IGBT), an inverter’s ‘heart’, produces considerable amounts of heat while switching. At the same time, IGBTs, printed circuit boards, and other electrical equipment start operating at lower efficiencies when they reach a certain temperature. An increase of just 1 or 2°C above the maximum operating temperature, which usually sits between 90° and 110°C, can cause equipment lifetime to halve.

To avoid this, inverters lower their power output through derating. This, however, has significant effects on a plant’s levelized cost of energy (LCOE). Therefore, the more efficient an inverter’s cooling mechanism the later derating occurs. A quick look towards inverter data sheets shows that most inverters can maintain operations at a maximum ambient temperature of around 60°C, but start derating at slightly above 40°C. Not all suppliers provide data on different outputs under different ambient temperature scenarios, but those which do admit power losses between 5 and 10% at an ambient temperature of 50-55°C compared to 30°C.

Additionally, different inverter models of the same supplier can have severe differences in derating behavior – with some severely derating at 35°C – suggesting that different heat management solutions had been deployed. Of course, inverters don’t need to sustain ambient temperatures of more than 45°C if they are being deployed in many parts of Europe or North America. It does make clear, however, that EPCs are well advised to bear this issue in mind when planning in the Middle East.

Beat the heat

Most high-power string inverters utilize forced air cooling, employing fans that are mounted onto a heat sink. Speaking to pv magazine, Andreas Schlumberger of TÜV Rheinland says that passive cooling systems for inverters above 10 kVA are rare. Higher powered string inverters dissipate the heat through heat pipes and convection zones to a larger heatsink on the outside of the inverter, where no fan is mounted. Huawei, for example, supplies inverters with this type of cooling.

Manufacturers such as Fronius or SMA rely on forced air cooling. This means that heat is dissipated through heat pipes to a smaller heatsink onto which a fan is mounted. The use of a fan increases the cooling power, which makes it the system of choice for most manufacturers of high-power string inverters.

SMA’s Andreas Tügel, Technical Product Manager, Product Group Inverter says that he believes that passive cooling does not provide the necessary flexibility to react to changing irradiation levels due to passing cloud cover. “It is absolutely crucial to implement an actively controlled cooling system, like speed controlled fans, which can adapt the cooling grade based on the currently available share of PV power,” said Tügel. For him, passive systems are not reliable enough to sustain changing heat patterns for over 25 years. Fronius holds a similar position and points passive cooling being prone to hot spots, which strain internal components.

On the other hand, Huawei says that it places heat generating components and heat sensitive parts in different compartments within the inverters while utilizing various thermal insulation strategies. Through this, the build-up of hot spots in sensitive areas can be avoided. Huawei says the benefit is better dust protection because there are no air inlets. It is important to note, however, that also with forced air cooled string inverters, the inside of the inverter is seldom exposed to dusty outside air.

Cooling methods for central inverters also differ between manufacturers. Siemens, for example, uses a combination of air-to-air heat exchangers and a liquid-to-air heat exchanger. Dorko Eliaszewskyj, Technology Director PV Utility-Scale, at Siemens AG explains that the liquid, a mixture of water and glycol is actively pumped past the power stacks, which generate the most heat. The air-to-air heat exchanger cools the remaining areas inside of the cabinet.

Through this process, outside air never really gets into contact with the inside air. Instead, a heat exchanging technology which is analogous to that used in the auto industry and air conditioning industry is deployed. Because the cooling systems are not purpose-built solely for the central inverters they are cost competitive, with over thousands of hours of validation across different applications. By contrast, SMA uses an architecture which doesn’t use liquid cooling for the power stacks. Instead, a powerful heat exchanger sucks in air two meters above ground to cool the circulating air on the inside of the inverter.

The company has deployed its systems in the hot conditions of Jordan, the Australian Outback, and the deserts of Mauretania, and holds that its central inverters are designed for operations above 50°C. These temperatures are surpassed extremely rarely in any geography.

Dust – only an O&M challenge

Inverters, it appears then, are as sealed as a camel’s nostrils. The interior of the inverter is hermetically sealed. If fans are used to cool the inverters, these are mounted onto a heat sink that dissipates heat from the interior to an outside cooling element. All inverters are marked with an IP number usually between 65 and 67. The first number shows its sealing level to dust and the second to water. Thereby, IP 67 is “dust tight” and protected against water jets and waves. The inside of the inverter may be a high-power string inverter, or a central inverter that never actually comes into contact with the air outside. Whatever heat dissipation system is deployed, it is a closed system working with heat sinks and heat pipes. Dust then only becomes an O&M challenge, insofar as dust must be removed from the outside of heatsinks or fans.

To demonstrate the effectiveness of sealing, most prominent manufacturers boast that their products have been tested under unnaturally extreme dust conditions, with particles considerably smaller than naturally occurring dust. As a result, the likelihood of an inverter being damaged through dust ingress is reasonably limited. Though, of course, O&M can account for a considerable part of a PV project’s LCOE. Manufacturers themselves claim that their systems are maintenance free with some claiming that competitors’ systems incur hefty O&M bills. TÜV Rheinland’s Schlumberger, on the other hand, believes that in principle no system is maintenance free, but also does not report any striking failures.

Liquid cooled systems, unless they work with high pressure, are usually low maintenance. However, if a problem arises it is likely more complicated to fix, not to mention the environmental issues that could arise if coolant leaks into the soil or the atmosphere. If the central inverter is containerized and located in a small building and uses a gas-based coolant, O&M staff could also be required to enter the compartment with gas mask equipment to avoid the health impacts of the fumes inside.

Though containerized solutions may be impractical in hot climates, shading the inverter is presented as a rather simple and effective solution. Both TÜV’s Schlumberger and some of the manufacturers advocate shading.

Next steps in inverter evolution

Siemens’ Eliaszewskyj sees that with the evolution of inverter topologies and materials significant progress is being made in reducing inverter operational temperature. The switch from a two-level architecture to a three-level architecture allowed for more efficient and higher frequency switching, without using higher voltage and higher heat IGBTs. In so doing, both power density and efficiency can be increased.

Some years ago, few inverters with efficiencies in excess of 98% were supplied. In this vein, Eliaszewskyj said that for a 5 MW inverter the losses, for example, have the same impact on power output as driving a 2 kW heater fan.

Eliaszewskyj believes that it is likely that the use of silicon carbide MOSFETs in inverters will be far more commonplace. These transistors allow for a faster and more efficient switching process, that generates far less heat. He said that MOSFETs themselves are also less sensitive to heat and can maintain switching properties through a wider temperature range.

Looking further ahead still, gallium nitride-based semiconductors demonstrate better performance under high heat, which could be of great use. Schlumberger of TÜV Rheinland believes that this could be an area in which activity could pick up within supplier ranks.

Eliaszewskyj adds, however, that while less heat will be produced through higher efficiencies, manufacturers are likely to offset the effect by increasing the power density of future inverters, to increase competitiveness. Hence, efficient cooling will continue to be a requirement, irrespective of efficiency gains. What is likely is that manufacturers will have to strike a delicate balance between cooling methods, materials used, derating, and power density. Getting that balance right, particularly as it pertains to component and project economics, will be no easy task.

SMA’s Tügel, for example, says that while it would be technically possible to design an inverter to start derating at higher heat, the production gains on a few occasions per year would be offset by the additional capex of the inverter.

It took the camel thousands of years to adapt to the climate of the deserts. Power electronics have only had a few years to go through this evolution. While great developments have been achieved already, the present shortcomings can easily be mitigated.

Aside from avoiding placing the inverter in the direct sun, project developers are well advised to study the temperature data of their proposed plant location. The data should be matched with the manufacturer’s derating curves, to identify the average losses throughout the year.