Part 1 of 2
Renewables are becoming a major source of electricity in the U.S. and many other countries and can no longer be considered “alternative” energy. Now about 7 percent of U.S. electricity comes from wind and solar annually, up from less than 1 percent a decade ago, and all signs point to this proportion continuing to grow. But the intermittent nature of renewables will increasingly affect power grid performance and require new technologies to manage grids. What emerging technologies are on the horizon?
Oilprice.com interviewed Steven Griffiths, Ph.D., who is Senior Vice President for Research and Development and Professor of Practice at Khalifa University of Science and Technology in Abu Dhabi. He has led numerous studies on integrating renewables into electrical grids and his research at Khalifa University covers this topic in detail. His work in the UAE sheds light on challenges now facing the U.S.
Oilprice.com: Steve, when we’re looking at the growing share of renewables in power grids, where can we expect them to be in 30 years?
Steve Griffiths: If you look right now where the energy sector is headed, particularly power generation, in most outlooks the share of renewables in electricity is going to grow pretty strongly between now and 2050. If we do well incorporating renewables into the power sector, then somewhere around 70 percent of electricity from the power sector could be from renewable energy, and maybe 50 percent from intermittent renewables – solar and wind – in 2050.
So looking out 20-30 years, it’s really going to be a story about solar and wind for renewable power. Other forms of clean energy such as nuclear, bio-based and hydro will complement further.
OP: It seems fair to say that intermittent renewables will become significant when they compose about 20 percent of national grids, and very significant when they compose over 40 percent of national grids. Are these useful thresholds?
SG: I put the number out there of 20 percent being a hurdle, but it’s very context dependent. Nonetheless, 20 percent is a reasonable ballpark figure when you consider the progression of incorporating intermittent renewables into the grid at increasingly higher shares. You go from a situation where renewables come into the power sector but the sector doesn’t have to operate much differently, toward a situation where you start to see changes in the way current power plants and the electricity grid have to operate. Finally, as you start to get to very high shares of intermittent renewables, the way in which you run the power sector will need to change in significant ways and new technologies will be needed to ensure grid balancing and stability.
It’s not a kind of stair step process, but rather a general progression that many countries are going through now. We’re in somewhat of an early phase in most countries where it’s not difficult to incorporate intermittent renewables into the power sector, but we’re headed globally toward a situation where it starts to be a consideration. The key issue is how you’re going to build out and manage the power sector because of the intermittency issues around the rapidly growing amount of solar and wind renewable energy technologies being installed.
Of course, geographical location and the current organization of the power sector are critical contextual factors. But from a broad perspective, I think 20% is a reasonable number, between 20 and 30% is what I would say. For instance, if you look at the SunShot Initiative in the US and their targets for PV levelized costs, the targets are achievable to between 20 and 30% of electricity from PV without energy storage being required.
When you’re going beyond 30% intermittent renewables, then it’s definitely going to be a situation where storage has to increasingly come into play. We’ll talk about the different modes of storage that are evolving, but it’s not going to be business as usual. You have to go into a different way of thinking about how you operate.
Then when you’re getting to 40-50% and beyond, it’s a whole new game, and we’ll talk about that. That’s when you start to really consider long-term storage and the need to balance supply and demand between seasons. With very high shares of intermittent renewables, you try to optimize the amount of renewables you have on the grid and not curtail a substantial amount due to the fact that one season has much higher electricity demand than another.
OP: What are the main concerns as greater shares of renewables come into grids?
SG: When you start to get to high shares of intermittent renewables, then you start to become more concerned about being able to utilize the energy that is not dispatchable when you need it. It is available when it is, and if there are very high shares you may not be able to use all of the electricity produced at one point in the day and that’s where we start talking about how you achieve a more challenging supply and demand balance. This is where there’s some pretty interesting technologies that become relevant and I’ll discuss a few later.
When you finally get to much higher shares of intermittent renewables, you start worrying about the seasonal balancing of supply and demand if there’s a relatively different profile of demand seasonally and/or your renewable energy resources are much different between seasons. Also, you have to worry about the very short term, because of the way power generation and consumption works on the grid. Today’s electricity grids have lots of power generators that are able to maintain a stable frequency of electricity supply even when short term imbalances arise between power generation and consumption. The inertia that these big rotating machines have is very helpful for achieving stability. In a power system with generation mostly from solar and wind technologies that do not inherently produce alternating electrical current in the same way as traditional power generators, it’s important to ensure that such very short-term frequency regulation is achievable.
OP: At what points will new technology be needed as renewables increase their share?
SG: So, the way in which you think about technology progression, as you get from low shares to very high shares of renewables, you need to accommodate the ramping up and down of your current power generation fleet’s electricity production to accommodate production from renewables. Then as the amount of renewable energy produced at certain times exceeds your need for it, you start to think about how you can shift the electricity produced from the renewables to periods in which they can be effectively utilized or alternatively you shift demand to periods when renewable electricity is available. Then as renewables begin to dominate electricity supply, you need to think about seasonality and also very short-term power grid management and stability.
4 ‘Buckets’ To Think About
OP: What can be anticipated in a 30-year horizon in terms of technologies being actively researched now?
SG: Right now, I think there are four areas or four buckets in which you can start to think about how you’re going to manage renewable energy as you bring it into the grid. These areas are supply side technologies, demand side technologies, energy storage and electrical grid infrastructure. For the near term, Concentrated Solar Power (CSP) with storage is an interesting supply side option for locations with the appropriate solar resources. At the same time, we need to be working on the demand side and grid infrastructure, which concerns sensors, software systems and new power electronic devices to achieve optimal power flows as the power grid evolves to accommodate both a higher share of renewables as well as distributed generation. Further, the growing share of renewables should also turn our attention to storage technologies for very long duration, which calls into question the case for lithium ion batteries, which really are only optimal for intraday storage operations, and opens the way for a new class of technologies, such as flow batteries. Finally, we have to be concerned with seasonal storage, which calls potentially for hydrogen storage and production via electrolysis.
I’ll talk about what we might expect to see, in terms of a few key emerging technologies in these four areas.
CSP + Storage
OP: Let’s start from where we’re at now, at just a few percentage points of renewables in the grid. The US is at 7 or 8% solar and wind, while the UAE is just starting out with renewables generation. On the technology horizon, what is most needed now?
SG: On the supply side, in geographies like the Southwest U.S. and the Middle East, I would highlight an opportunity for concentrated solar power with thermal energy storage. That’s a technology in which there is quite a bit of R&D activity and opportunity for cost reduction. It’s not quite as in vogue as solar photovoltaics right now, but there are a number of technical targets to try and bring the levelized cost of electricity from CSP with storage to much lower levels in the next decade.
OP: That’s about 8-12 hour storage with CSP?
SG: The SunShot program in the U.S. has two different targets, one is for 6-hour storage, which they call the “peaker” duration because with 6 hours of storage you’re most often competing against the cost of electricity from power plants that are designed to meet peak daily electricity demand. 12 hours-plus storage is aimed at competing with the cost of electricity from power plants that operate with much higher utilization, if not continuously. This mid and baseload electricity is much less expensive per unit of electricity produced and so the competing cost target for CSP with storage is 5 cents per kilowatt hour on a levelized basis.
Concerning technology, there are some very interesting investment opportunities today. The main mode currently for CSP with storage, the one with which people are very familiar with, is molten salt storage, typically coupled with the power tower, the solar tower. When those salts heat up they can store heat that can later be used to heat a working fluid to run a traditional steam-based power generator. Today’s salts tend to degrade at temperatures between 500 and 600 degrees C but much work is being done to design salts stable well above 700 degrees C. This provides for a higher efficiency power cycle by having a higher temperature for the steam or gas that passes through a turbine to generate power.
Beyond molten salts, there’s some very interesting new thermal storage technologies being worked on now. One is falling particle technology, which is dropping particles that you want to heat through a tower, heating them and then storing them in the bottom of the tower, then running a fluid through a tube that goes through that hot media and is heated to become the working fluid for running a power cycle.
Another technology is the gas phase receiver, which is interesting in the sense that you’re heating up a gas directly, which can then be run through solid-state storage media in which you can store the thermal energy for use later. The high-temperature gas can also be run directly through turbines to produce power.
These are a couple of the technologies that I think are quite interesting for CSP. They’re aimed at incorporating much higher temperatures into thermal storage to ultimately achieve higher efficiency in power generation. When combined with other CSP advances, the levelized cost of electricity from CSP systems with high-temperature storage can become very competitive.
A company that we’re working with at the Khalifa University’s Masdar Institute is 24/7 Solar. This company has created gas phase receivers that run at low pressure, and so they decouple the pressurized hot air that goes to the turbine from the air that they’re heating up. With this approach, you don’t have to worry about the cost of a pressurized receiver. All that is required is a heat exchanger to transfer the thermal energy from the low-pressure hot air to high-pressure air. Air is heated in the receiver to just under 1000 degrees C, and then some of that energy is stored in solid-state media for use later while another fraction of the air is used to run a high-temperature power cycle. With this approach, you never have to deal with high pressures in the receiver, which I think is economically quite useful. The tradeoff is that the heat exchanger must be able to withstand extremely high-temperature gases.
CSP works and it’s becoming low cost. I think it’s an important area of research for a place like Dubai, which will certainly need dispatchable energy if they’re going to get to their target of 75% renewable power by 2050 in a rather small power grid. Of course, CSP is amenable only to certain environmental conditions. You can’t use CSP in a very cold, cloudy climate or a climate where sunlight is easily scattered by particles in the air. But where it works, it’s a very useful technology because of its relatively long-duration storage capability. Based on all this, I think as an R&D focus area, CSP with storage is a supply side technology that makes a lot of sense for continued work.
OP: Is CSP with storage equally important for the US at this point?
SG: Yes, it is important in the U.S. in locations where you have appropriate solar resources.
Demand Side Developments
OP: So CSP is needed on the supply side. What about the demand side?
SG: I think this is the next area to look at closely. It’s actually a pretty hot topic, when you’re considering how to balance supply and demand. You can always worry about how you’re going to modify electricity supply, have more or less of it, make it ramp up and down. But you can also better control demand and make it more dispatchable.
I think that controlling the demand side is going to be an increasingly important part of the future electricity system. And I’d say, if you look at a collection of energy system modeling studies that consider demand side management, by 2050 15 to 20% of all electricity demand could be met through demand side management in support of intermittent renewable electricity reaching very high levels.
Now, given the strong growth in artificial intelligence, data analytics and computer science, I think demand side management ties in very well with what some refer to as the fourth industrial revolution. Artificial intelligence and data science techniques can work very well when you’re trying to balance intrinsically variable renewable energy supply with uncertain but potentially modifiable electricity demand.
Achieving the right electricity supply and demand balance is not necessarily a deterministic, algorithmic matter. Rather, it can benefit from a holistic artificial intelligence approach like machine learning, which can figure out how you best modulate load, taking into consideration parameters like weather, grid status, available generation, historical demand profiles and other such factors.
A hot area now, where there’s quite a bit of investment going on, is distributed energy resource management. It especially concerns the distribution grid and distributed assets whether they be batteries, charging electric vehicles, flexible heating and cooling loads or PV with storage. All of these technologies will grow in importance since they can be coordinated and utilized very well to support the supply and demand balance for the electrical grid.
Companies that are working in distributed energy resource management aim to help business and home owners monetize their distributed assets and they help utilities optimize distribution grid power flows and grid voltage. Hence, this is a very important area of R&D and investment. I should note, however, that distributed energy resources are likely to be most important in places like the US where distributed assets, like rooftop PV and electric vehicles, are expected to have significant impact. In places like Abu Dhabi, centralized generation is the paradigm and the adoption of electric vehicles is rather uncertain.
The DOE and specifically ARPA-E (Advanced Research Projects Agency-Energy) have some very interesting R&D programs in this area. They give acronyms to programs like the one called NODES (Network Optimized Distributed Energy Systems), which is focused on how distributed resources can help with frequency regulation on the grid, and also provide load shaping through modulation of flexible demand side loads. An ambition of the program is to achieve a virtualized means of energy storage through intelligent coordination of flexible loads and distributed energy resources.
In general, intelligent systems for demand management is an area where we’ve got to keep pushing ahead as sensing capabilities, artificial intelligence and other computational techniques allow us to tie in consumers of electricity, whether they’re businesses or residents, to the challenge of grid management. That is going to be critically important.
OP: So it’s about more than just lessening the demand for energy?
SG: Yes, the future of electricity is about looking at the entire system, at the transmission grid and the distribution grid together to then trying to deal with the way power flows are managed. Increasingly you’ll have not just one way electricity flow from generator to consumer, but the other way as well, with consumers providing electricity back into the grid through the distributed generation I talked about.
There’s something called the optimal power flow challenge, which is about trying to deal with uncertain supply and demand in a power grid connected with millions of assets and dynamically updated power flows. The issue is figuring out how you take all these generators, all these consumers, all these different constraints on the grid, and figure out how to manage the whole thing optimally. The reality is, it’s not possible to use a deterministic algorithm that has a precise end solution. Actually right now people use a lot of heuristics, they use a lot of statistical approaches to figure out how to optimally manage the grid.
The challenge is only getting worse. I mean, a reasonable size electricity grid will have probably more than 500 million different variables to manage in order to optimally dispatch electricity and deal with balancing of supply and demand. I think software systems that can manage the grid in an effective and robust way at lowest cost while ensuring reliability and not violating grid constraints are going to be extraordinarily important.
OP: Can you talk more about optimal power flow?
SG: At the core, it’s a software and computing challenge. However, there is another aspect that is directly related, which is managing electricity flow in the grid, specifically via hardware like transformers.
Right now, transformers on the grid allow electricity to go from high voltage to medium to low voltage. Modern transformers are very big devices that are robust and relatively efficient but they’re pretty much set up for electricity to flow in one direction. They’re not very amenable to significant power flow changes running at very high frequencies and very high voltages. I think as you start to look at trying to work with distributed generation at a large scale and deal with a grid that is more dynamic and carrying more electricity, solid-state electronics, especially power semiconductors, will come into play in an important way.
Materials like gallium nitride and silicon carbide are poised to replace silicon in power semiconductor devices and open up a new realm of possibilities in power electronics. These devices can help manage the grid, voltages and current flow, in a way that will allow it to run at very high voltages, very high frequencies and with very high efficiency and flexibility. Further, they are robust to the high temperatures generated when transferring high amounts of electrical current.
These devices will be used increasingly in many different sectors because of their efficiency in power management. They’ll be used in industry, data centers, renewable energy technologies, the automotive sector and others. When it comes to investment, and where I’m seeing lots of research opportunity and lots of R&D challenges, is in trying to bring solid state electronics to the electricity grid for power conversion and control.
So regarding what I talked about with managing the power flow, being able to connect distribution grids to transmission grids and have that connectivity be seamless with two-way power flow, I think these devices are going to be quite revolutionary. There are some really interesting companies now working with gallium nitride for important power semiconductor applications. I mean you’ve got basically an opportunity here to do some new things with power which haven’t been possible before. In fact, demand side management and the grid collectively encompass a completely new realm of opportunity for research and also for investment.