Producing hydrogen from renewable electricity (electrolysis) can provide long-duration storage critical to reliable power system operation in grids dominated by wind and solar power. While that need has been slow in materializing, other renewable hydrogen applications, such as transportation, are helping build the expertise and infrastructure important to scaling up the industry.

System Reliability and Renewable Resources

In 2015, the consulting firm I worked for was asked to deliver a report on how power systems can reliably operate using only variable resources such as wind and solar. I had worked as a hydro power analyst with the Bonneville Power Administration (BPA) for 15 years, and since hydro is also a varying and not entirely predictable source of renewable electricity, responsibility for the report fell into my lap.

Hydro power is highly variable both seasonally and annually. Northwest hydro power has relatively low output in the Fall when precipitation is low, and high output in Spring due to snow melt. Annual variation is also significant, with energy produced in high-precipitation years nearly twice that of drought years. The timescales of variability and predictability differ among hydro, wind, and solar, but the challenge is similar.

Power system operators are responsible for meeting the challenges of variability and uncertainty in real time. This is a struggle even on systems without wind and solar; power plants and high voltage transmission lines can suddenly fall out of service, and of course power demand is constantly changing, day-to-day, hour-to-hour, and minute-to-minute. To me, emerging wind and solar were simply a new source of a familiar challenge, not a new challenge.

I became something of an expert in this area in 2003 while working for PacifiCorp. Management wanted to know how much in additional “reserves” would be needed to accommodate the growing wind fleet. Reserves are sources of power (and increasingly today, demand) that have the ability to quickly adjust production (or consumption) levels as needed to counter unexpected and less-controllable changes in production or consumption. Surprisingly to me, the company didn’t have a fixed methodology for calculating reserves, nor did most other utilities. I wrote a book on doing such calculations in 2010.

Two Main Issues Emerge

Because reserves are traditionally provided by fossil-fueled power plants, running entirely on renewables would break new ground. We recruited an international panel of power system experts and engineers to help address the technical challenges. A first question was where reserves would come from without the conventional power plants that normally provide them.

We had done a paper on the reserves issue a year earlier, identifying largely untapped sources of power system flexibility that could be used as reserves to balance the grid. These included load control (e.g., pumps, air conditioning equipment, and electric water heaters) and generator control (wind and solar output limited and leveled electronically). That left two unresolved challenges:

• Maintaining stability on a system without rotating (spinning) equipment to supply inertia; and
• Bridging one or more weeks of low output from wind and solar, particularly in higher-latitude winters when demand is high.

Power System Stability

A common misconception is that power generation and consumption must be exactly matched at every moment, conjuring a picture of operators somehow instantaneously adjusting generation to exactly match consumption. That isn’t quite how it works. A BPA engineer once told me that they delay responses to mismatches in supply and demand[1] for several seconds, lest their response end up adding to a problem that will have reversed by the time they respond. He said they don’t bother doing anything about mismatches of less than 50 megawatts.

This isn’t to say that an over- or undersupply of power is not a problem. When there is a significant mismatch in power demand and production, there can be brownouts and/or equipment failures. Reserves are there to respond to mismatches that occur over several seconds, minutes, or hours.

Reserves don’t react quickly enough for large, sub-second disturbances such as those that occur in the first moment when a major transmission line or power plant suddenly fails. What largely holds the system together in the first instance is the energy stored in the huge mass of spinning equipment: the generators and turbines of conventional power plants. They act collectively like a huge flywheel that keeps the system going. This effect is referred to as system inertia. Without it, sub-second disturbances can grow and cause the grid to become unstable, risking widescale blackouts.

Power from solar and wind enters the system through electronic inverters that condition the power for injection into the grid. Inverters don’t provide the same stabilizing inertia conventional plants provide, although they can be programmed to simulate it (at the expense of some energy production). Engineers on our panel were concerned the system couldn’t operate reliably without actual rotational inertia, and worried simulated inertia  was unprecedented at such a scale; if not done right, the programming could conceivably make the problem worse.

After much discussion, we found a simple, inexpensive solution. Devices called “synchronous condensers” are essentially motors that can provide the needed inertia. Even the existing conventional power plant generators could be modified slightly to serve that purpose, providing the inertia without even requiring significant new capital investments.

The Bigger Problem: Long-Duration Storage

Back when I was a hydro system engineer, we were focused on the challenge of low production in drought years—years when precipitation is historically low and may only reach 70% of average. I had seen similar events with wind, where wind production would fall off almost entirely in the winter for a week across the entire Northwest. Meteorologists note that this happens within large-scale high pressure weather systems, and it can happen over areas larger than the entire Northwest.

Winter solar production at this latitude, even on sunny winter days, is less than half of annual average. The sun is up for fewer hours, and it is lower in the sky with rays striking the earth at a glancing angle. More solar can help, but can’t on its own take a system through a low-wind winter week without additional supply.

Some have argued that low-wind events can be countered with bigger and longer transmission lines importing power from other regions. Although that may help, it assumes other regions produce winter surpluses sufficient to supply the needs of almost the entire Northwest, which is highly unlikely. Others argue that we can rely on hydro system storage, however while reservoir storage is a huge asset, it cannot supply the entirety of the region’s need for a week, even if environmental and other constraints were relaxed to allow this.

The bulk of the solution must come from large-scale, long-duration (days and weeks) energy storage. Battery storage seems an obvious solution, but the cost of running the Northwest grid on hydro and batteries for a week of demand is astronomical, around four times the cost of renewables that would otherwise produce enough energy to run the system—about $295 billion for the Northwest’s system[2]. Double that to cover two weeks of storage. Long-duration storage is the greatest techno-economic hurdle that must be overcome to reach 100% renewables.

Enter Renewable Hydrogen

In my research for the consulting firm’s 2015 paper, I discovered that a 6-MW electrolyzer in Werlte, Germany, had begun producing hydrogen in 2013 at scale from water and wind power. A week’s worth of energy storage isn’t really that much energy compared with the annual production of a power system. It struck me that if we could make a relatively small amount of hydrogen over a long period of time, it could be stored until needed, and then turned back into electricity to bridge demand during low-wind events. The stored hydrogen could even be used to fuel existing gas power plants, releasing no carbon dioxide and minimizing new infrastructure investments.

For example, a 150-MW electrolyzer plant operating 40% of the time (to capture renewable electricity when available) could produce enough hydrogen in a year to fuel a thousand-megawatt combined-cycle gas plant for a week. Most of today’s gas turbines can  accept 30% hydrogen (by volume) without changes. Manufacturers have created retrofits making it possible to run 100% hydrogen on smaller turbines, and they expect to have retrofits soon for most of their models.

The Werlte plant is situated next to a manure digester plant and can combine recycled carbon dioxide from the digester with hydrogen to make methane. Methane is the main component of natural gas and completely compatible with today’s natural gas systems and power plants without modification. Although using the methane to produce power releases carbon dioxide, it was headed into the atmosphere anyway from the composting manure.

People often view storing hydrogen as expensive, but it is much less expensive for long-duration storage than today’s alternatives. If hydrogen or methane were stored in existing natural gas infrastructure, the cost would be very modest. The energy could be converted back to electricity in existing power plants—at a far lower cost than building new batteries and associated infrastructure.

We had a solution to offer in our paper: 100% renewable power systems could operate reliably, and renewable hydrogen presented an existing technology that provided the key.

Other Integration Benefits of Renewable Hydrogen

In many years, Spring snow melt causes the Northwest’s hydro system to produce more power than can be absorbed locally, so it is shipped outside the region, mostly to California, and sometimes “curtailed” (meaning water is released over spillways instead of through turbines). Such surplus or unused energy is characteristic of all systems significantly relying on variable renewable resources: there are times when production is low and other times when it exceeds market demand. Even today, BPA curtails regional renewable resources when the market price drops below zero[3].

In a system where most power comes from wind and solar, there are times when ability to produce low-cost power exceeds market demand and generation is curtailed. Electrolyzers can be used as opportunistic loads to make use of those surpluses. Today, the amount of curtailed power is not economically significant, but in systems where most or all power comes from variable renewables, it becomes significant. Electrolyzer loads can both bridge times of low renewable power production as previously described, and also capture energy that would otherwise be turned away—producing hydrogen, a high-value commodity.

Electrolyzers can also ramp up or down quickly, helping balance the grid while reducing the burden for other reserve units. Ramping power generation up and down on reserve power plants causes mechanical wear and tear, whereas adjusting electrolyzer consumption is done mainly by changing electric current, without mechanical torques or other stresses. Douglas County PUD’s electrolyzer program expects to shift the burden of holding reserves from their hydro turbines onto electrolyzers, saving significant maintenance costs from turbine wear and tear.

Slow Start

I founded the Renewable Hydrogen Alliance in 2018 when two things were clear: renewable hydrogen would be necessary for getting to a zero-carbon grid, and the fact was widely unknown in the utility industry. Eight years later we clearly are not on track to achieve the necessary renewable buildout on the timeline scientists have deemed necessary to avert the most cataclysmic effects of climate change. The impetus for renewable integration hasn’t materialized, and deployment of electrolyzers has similarly been delayed. We have nevertheless succeeded in vastly increasing awareness among the utilities of hydrogen’s role in a decarbonized system, if perhaps not yet among other important constituencies.

Policy Implications

While we aren’t immediately desperate for the storage solutions we will eventually need, now is the time to start gaining experience with electrolyzers. Implementing anything new is a slow process. Douglas County PUD has been working on a first electrolyzer for nearly ten years.

Despite the technology being a century old, there are things to work out. Designs are not standardized so plants are expensive. We need to create a workforce experienced in building those plants. We need to work out optimal storage options and decide whether hydrogen itself should be stored or some derivative such as methane, methanol, or ammonia. In short, we need to act now to have this technology available at scale and low cost when it does become critically important.

Meanwhile, hydrogen in other applications today comes almost exclusively from fossil fuels; those uses can be decarbonized with renewable hydrogen. Most notable is transportation, where hydrogen vehicles emitting only water are already attractive. For example, three Washington State transit agencies have ordered hydrogen buses, saving neighborhoods from noise and air pollution on routes too long to be served by battery buses. Such projects can act as stepping stones to building the hydrogen infrastructure that will be needed in a decarbonized world.

Endnotes

[1] I draw a distinction between “demand” and “consumption.” A nominally ten-watt light bulb may optimally use ten watts, but the consumption depends on the power supply and could be more or less than that. If less, the bulb may be dim. If supply pushes more than ten watts into the bulb, the bulb may burn up. Power that is produced gets consumed by virture of the theory of conservation of energy, not the quick actions of power system operators.

[2] Assumes battery cost of $100/kWh, supplying 51.9% of the region’s 33,800 MWa demand (the rest being existing hydro). The actual demand could be higher if the outage is during high-demand winter months. Wind is used as the proxy renewable resource, at a cost of $1,300/kWh and 33% average capacity factor.

[3] A negative market price means wholesale power producers must pay to send the power to another wholesale market participant. This can happen, for example, if a wind power plant owner is contractually obligated to deliver power when the wind blows and faces penalties for failing to do so. Thermal power plants (coal, gas, nuclear) experience shutdown and start-up costs causing operators to seek compensation if the power may be needed again in a few hours or days.

With thanks to AJ Perkins, David Brown, Gary Ivory, and Sam Lowry for their comments and contributions to this post.