Distributed Energy

Optimizing Battery Storage to Improve Efficiency

  • Email This Post Email This Post

Credit: iStock/D3Damon
Improving return on investment by reducing the payback period, increasing efficiency, and lowering costs are common business goals. One method of enhancing the economics of distributed energy resources technology investments involves bundling grid applications to stack up multiple value streams.

Value stacking means having the capability to perform multiple energy services at the same time. For example, Geli systems can provide peak shaving at a commercial facility, as well as demand response services when the grid operator sends a request from the same system, explains Andrew Krulewitz, director of marketing and strategy. “Sometimes the peak shaving and demand response operations are coincident, and sometimes they are not. Our software has to—to the best of its ability—predict how a building’s load will change and schedule battery operations accordingly.”

Are you subscribed to Distributed Energy magazine? Click here for a free subscription!

The benefits of value stacking are twofold:

  • It boosts profitability by combining multiple sources to maximize returns.
  • It adds resilience by avoiding dependence on a single source of income, providing limited protection from the risk of market changes.

To enhance value streams, buildings are able to use the battery at different times for different applications to increase value for the stakeholder. Typically, they run the highest value application first, and continue on down the line. The software determines the application’s value and optimizes usage, while an algorithm determines its priority.

Behind the meter you achieve demand reduction and peak shaving, says Olaf Lohr, director of business development at Sonnen Inc. “The streams don’t exist at the same time. Backup power is a resource available at all times.” In front of the meter, the battery system for demand response provides voltage support.

In the past, most cost-benefit analyses of energy storage focused on a single purpose: demand charge reduction, grid services, backup power, or increasing renewable energy (usually solar PV) self-consumption. Jayesh Goyal, chief commercial officer of Younicos, says that business models based on only using batteries for a minority of the time neglected significant added value. “By leveraging a storage resource to address other services at the appropriate time (e.g., provide frequency regulation when batteries might otherwise be idle), or in parallel, owners can ‘stack up’ multiple value streams to deliver a total return that exceeds the cost of energy storage.”

The different value-add services provided by energy storage, often referred to as applications, include “demand charge reduction” or “frequency regulation.” According to Stem, an energy storage system operated for only one of these applications will likely be idle most of the time, leaving an opportunity to provide additional services.

Add Distributed Energy Weekly to your Newsletter Preferences and keep up with the latest articles on distributed power, fuel cells, HVAC options, solar, smart energy systems, and LED lighting retrofits.    

The fact that there will be large differences in the total amount of value generated by different energy storage projects of similar sizes, based on how many applications they can successfully serve, is implied. The ability to value stack is determined by the intelligence of the decision/optimization algorithms that control the energy storage system, based on the software’s ability to interface with markets and determine in real time the most valuable use of the system at any given time. As Stem says, unlike other energy technologies, intelligent energy storage can generate value for businesses and the electric grid in many different ways.

Demand Energy Inc.’s Mike Fife states it plainly: “Value stacking is simply leveraging the same equipment, system, or process to deliver multiple benefits (most of which can be valued economically).”

Some examples he provides include:

  1. Demand charge reduction
  2. Time-of-use supply charge reduction
  3. PV utilization (self-consumption), especially with
    zero-net export
  4. Incented maneuvers
  5. Demand response
  6. ESS-based demand response without impact to comfort
  7. Backup power
  8. ITC
  9. Frequency regulation
  10. Voltage support/Var injection
  11. Stickiness with the customer (future energy management services)
  12. Carbon offset
Credit: Younicos
A parking lot solar array at Panasonic’s Pena Station NEXT headquarters

As quoted in https://networks.online, Julian Jansen, manager of energy service research at Delta-ee, believes that the key to successful distributed storage businesses will be value stacking. As he explains, the energy system is undergoing a transformation that will allow energy storage to play an important role.

Delta-ee’s research indicates that the business model innovation required to make energy storage attractive to both customers and suppliers is the main challenge facing the industry.

The current market for distributed energy storage is either driven by self-consumption (PV and storage, on the customer side of the meter) or grid support (larger batteries further up the energy system on the utility side of the meter). But new business models are emerging.

Networks.online describes behind-the-meter storage as trying to give a free lunch to customers while providing value to the energy system, and energy banks that act as central stores for local communities, with consumers able to pay for energy in and out, depending on their self-generation and consumption patterns.

Many options and values can be created through distributed energy storage, but the most successful business models will create a value stack that enables them to access different values and diversify their risks.

Battery Storage Optimization
Battery storage plays a large part in efficiency. Optimizing battery systems enables both longer lifetimes and improved performance within a storage resource’s capabilities. As explained by Goyal, intelligent control software extends battery lifetime by maintaining certain state-of-charge zones as much as possible.

Optimizing usage increases cost efficiency, not the efficiency of electricity conversion through charging or discharging the battery. In other words, operating the energy storage system for the right purpose at the right time increases the value generated by the project per kWh installed.

Optimizing battery storage involves knowing when to use the battery and when not to. Fife thinks some benefits are not worth the ohmic and coulombic energy losses associated with the maneuver; these are typically lumped into what is known as “roundtrip efficiency.” If the energy lost during the operation exceeds the benefit because the roundtrip efficiency is too low, why do it? Intelligent control and real-time optimization are key to making the right control decision.

Intelligent optimization of battery storage is important for two main reasons, Fife says. The battery is typically the most expensive component in a battery-based energy storage system and it degrades over time. Each battery chemistry and configuration has different degradation properties as a function of its usage profile. For example, lithium-ion batteries tend to degrade faster at very high states of charge, while lead-acid batteries tend to degrade faster at low states of charge. Each battery chemistry and configuration has different relationships between discharge efficiency and usage profile.

Battery degradation can be a very complex function of battery charge and discharge rate, temperature, and state of charge. Whether to keep the battery at a low or high state of charge depends on the battery. Fife explains that Demand Energy’s technology treats battery degradation as a negative value stream, and accordingly manages the battery degradation simultaneously with all of the other opportunities, or value streams. It won’t undertake a maneuver or usage profile using the battery unless the benefit exceeds the battery degradation cost. This process is automatic and scalable. “DEN.OS will automatically use lithium batteries at lower SoC than lead-acid batteries. We just need to make sure the customer’s DEN.OS is knowledgeable of the type of battery connected via a configuration setting.”

“Maximizing asset life—and enabling stacked-service utilization—requires a deep understanding of the chemical properties of different battery types from various manufacturers, as well as how batteries react when exercised under different conditions,” Goyal says. Although achieving this adds another level of complexity to managing a system, it does pay off with higher performance and greater return on investment.

Davion Hill, energy storage leader for DNV GL, cautions that optimizing revenue streams depends on geographical location. For example, he cites storage projects started in Pennsylvania, New Jersey, and Maryland under PJM. When the market underwent changes, the opportunities weren’t as good. “PJM restructured the signal for frequency regulation because the regulation D signal lagged behind the real-time signal. Unfortunately, the new signal stresses the batteries, so they have a shorter life.” Relying on batteries to respond faster means that they cycle more frequently; that results in recharging at higher rates.

Credit: Younicos
The control center at Panasonic’s Pena Station Campus

Battery Degradation
Software is really the key here, Krulewitz believes. Being able to integrate inputs from multiple sources—building load, solar PV forecasting, utility signals—is what will drive maximum utilization of energy storage assets. Another major contributor to increased proliferation will be the ability to access new market designs for fast-responding resources like batteries. However, without clear guidelines for how these systems can interact with existing electricity infrastructure, the ability for batteries to provide widespread grid value is significantly diminished.

California uses a unique application of value stacking based on regional dependence, Hill says. “The utility pays for storage to be available/on call. You must have at least four hours of availability at any time to reduce demand charges in a power purchase agreement. Stem sells it to a capacity market. Now you’ve got two revenue streams.” It’s a creative way of monetizing backup power, although it doesn’t pay for all storage.

Each regional market is different. New York has high demand charges—enough to make the call for storage, Hill says. But resiliency is not monetized there like it is in California.

Credit: Sonnen
The sonnenBatterie eco

Many utilities are trying to reform their energy visions by restructuring markets to monetize independent value streams. First, of course, they must identify those streams.

Even with only one value stream, battery degradation is so important that it must be considered simultaneously with the value stream, Fife states. DEN worked initially with rule-based and time-partitioned approaches that attempt to use a set of logic rules for determining control setpoints and switch back and forth between “apps” to take advantage of multiple value streams. Unfortunately, he reports that both of those control techniques came up short when compared with true optimal control, and were not as scalable. “Only true real-time, optimal control can facilitate effective energy storage optimization.”

Goyal believes that intelligent control software based on real-world installations, diverse battery types and configurations, and a range of use cases is key to optimizing a storage asset. “In effect, the right software becomes the ‘brain cells behind the battery cells,’ optimizing system lifetime and extracting maximum value and resiliency.”

The Value of Stacking
The challenge for everyone in the business of energy management (especially with energy storage) is how to provide the greatest total value from all of these opportunities simultaneously. As the industry grows and matures, storage software operations will become increasingly complex to handle multiple behind-the-meter and grid-support functionalities.

“The solution Demand Energy has developed is a real-time optimal control system that we have taken from theory to real-world implementation,” says Fife. “We branded our solution DEN.OS (Optimization System). It incorporates machine learning and artificial intelligence elements.”

Demand Energy configures each customer’s DEN.OS to be knowledgeable of all aspects of the customer’s energy situation (tariffs, incentives, battery chemistry, sizing, etc.), lets it watch and learn a customer’s energy consumption patterns, and plan and execute on a control strategy that delivers the maximum total economic value from all of the value streams together.

Fife provides an analogy as explanation: determining the best altitude to fly a passenger jet from Seattle to New York while balancing passenger ride quality, fuel consumption, and time of arrival. “Of course, the winds aloft are changing throughout the flight path and must be considered, just as the changing building load must be considered in an optimal control system for a building’s electrical system.”

Technology
The basics begin with a reliable system. “You want to put your resources in place with longevity and uptime,” says Lohr. Sizing is important. Sonnen offers seven sizes with enough power to sustain loads and the capacity to cover outages and provide energy to cover peak times.

Proper sizing is half of energy storage. “You have to understand consumption patterns—maximums and surges—and what’s critical to backup. You can reduce costs during installation if you know what materials to bring. It’s difficult to coordinate the batteries, converters, and switches—the magic behind the scenes. Sonnen takes different parts off the installer’s hands.” The trade-off, he says, is to maximize utilization with optimization.

Once you have a reliable system, you need to consider programming—how to optimize energy use. Lohr advises measuring the whole-house consumption to maximize usage of the battery to achieve a high level of energy independence. When it comes to the battery storage system, a good backup system provides efficiency, says Lohr.

Ultimately, however, it’s not about choosing the best battery, Hill insists. It’s about how you control it.

Demand Energy based the economics on a spreadsheet, Fife says. “We built controls to leverage value streams simultaneously.” He adds that they can apply the controls to simulate the economics.

None of this is possible without the right software, Hill concludes. The software enables innovation. He says the market is headed toward creative risk-taking in designing algorithms and compares the goal with the Internet situation. “There are few errors due to ubiquitous storage. The web has 99.9% uptime—if the grid had that . . . ”

Providing efficiency, resiliency, simplicity—such as trying to make storage plug-and-play—and finding more ways to monetize is the future of DER, he believes.

Overcoming Risks
In some areas of the country, DER value stacking complicates the interconnection and operation of assets, especially when combining retail-level value with wholesale market services. Value stacking introduces “revenue interface risk,” a series of technical and commercial challenges that result from accessing multiple revenue streams. Some in the industry are concerned that there is too much emphasis on the total revenue stacked at the expense of the interfaces in how those revenue streams are stacked.

Everoze offers advice in light of the risk of revenue interface that leaves a storage project unable to deliver:

  • Recognize the complexity of value stacking. There’s a significant possibility of techno-commercial risks, and there is the likelihood that the revenue stack will change over time.
  • Start with just 2–3 revenue streams. For early projects, target 2–3 high-value revenue streams to reduce complexity. This will reduce the total possible revenue, but will maximize the likelihood of reaching the goal.
  • Conduct thorough due diligence. Conduct a thorough technical and commercial due diligence process, ensuring that all risks are clearly identified, allocated, and reflected in the financial model.
Credit: Younicos
Panasonic’s Pena Station rooftop solar array under construction.

You think you know what will happen, Fife reflects, but conditions can change. “The hardest challenge is predicting. This industry is still a frontier. We’re on the frontier of distributed energy and machine learning. We’ve solved the problem of energy economics. What’s next is prediction of a periodic load behavior.”

Obviously, when the equipment gets turned on, there is load. But it varies with the customer and with the weather. HVAC is a primary consumer of power; there’s a definite correlation between weather and building load. “We’ve studied it over time with algorithms,” says Fife. “The problem has been solved.”

Climate and season impact the algorithm, Lohr reasserts. The question for a building manager is how much they want to reserve for backup. Some want energy security for winter storms. “The higher capacity they use from battery, the higher the revenue. It’s a give and take to try to maximize the value of your resources. The interesting part is that there are so many different value streams and applications based on temperature, location, season, and type of utility.”

AI
Optimization of energy storage is facilitated by a host of artificial intelligence (AI)-focused advanced software technologies, such as predictive analytics, machine learning, distributed low-latency computing, and real-time economic optimization algorithms. Stem says it’s basically all about the intelligence of the software controls.

Demand Energy uses machine learning strategies to improve value stacking. “It’s important because without forecast of load, you can’t optimize electrical consumption,” explains Fife. For residential applications, they have a machine learning system to manage intermittent loads. The granularity of load and statistics tends to average out in residential applications, but industrial applications experience a similar number of large loads at a different percentage. Retail lands somewhere in between.

By watching the building load continuously, Fife says that they have data history from which to formulate their algorithms. “We leverage experience to make better predictions. The challenge is to use local data in the algorithm and incorporate weather information.”

Additional air conditioning units affect building load. So does climate change. A prediction must include an adaptive piece for seasonality and climate change. Because of changing conditions, Fife says, “You can’t set up models forever; we continuously learn and test prediction algorithms on a wide range of buildings for quantitative results.”

An electrical system is not static, either. Systems can be added, lighting can be changed, other adjustments and alterations can be made. There are so many ways to use energy systems to provide value, Fife says. The challenge is balancing value streams and knowing which to go for, when—and whether or not to go for more than one. “We solved that problem with optimal control. The software navigates changing conditions and finds the greatest economic benefit. It’s totally adaptable without human intervention.”

DEN.OS is a single piece of software, not multiple apps. “We built intelligence into DEN.OS and let it optimize,” elaborates Fife. “It can do things humans can’t; it figures out optimization better than we can, thanks to machine intelligence. It comes up with moves we hadn’t thought of.”

Computers are diligent, looking at every possible solution. “Teach it to learn, enable it to forecast, and make it knowledgeable about all other factors, such as load, outages, etc.,” continues Fife. This remarkable machine intelligence is scalable. “It’s extendable to additional electric systems and can optimize networks simultaneously.”

Credit: Younicos
Y.Cube systems at Panasonic headquarters Pena Station in Denver, CO.

The Continuing Evolution of Battery Storage
Many of the energy storage systems in use today do one thing, whether that’s peak shaving, solar self-consumption, or frequency regulation. Mono-objective operations are inherently inefficient, Krulewitz states. “A typical commercial peak-shaving energy storage system in California completes fewer than 100 full cycles per year. There’s huge potential to use these systems to deliver other energy services, either to the end-customer or to the grid operator. What’s missing are the markets, but utilities and regulators are quickly catching up, enabling energy storage to participate in a more meaningful way.

Hill mentions a recent proposed rule-making that recommends all ISOs in the US to consider market models for storage to be an aggregate resource. This would create vertically integrated utilities. “It’s a big deal that took years to develop.” ISOs value storage as a generation asset and want to incorporate more. If you integrate with DER, that makes it “dispatchable” and reliable. It makes it more efficient.

Because the market is saturated with too much storage, Hill says prices are going down. “It’s a shallow market that makes up only 1% of the total available generation.” Looking three years down the road, he says that if solar gets cheaper, storage could replace gas plants. “There’s discussion about storage as the industry matures. Policy is not the only thing driving the market. The cost of storage is cheap, so we can compete with energy generation. We might win on cost.”

It’s a game-changer made possible because the technology makes it flexible and resilient on the grid. It’s flexible; more values can be added once a system has been installed. “It took seven years for product generation,” notes Lohr. The biggest achievement is to make it simple.

In the end, Lohr says, it’s a business decision. BE_bug_web

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *

FORESTER