Blog: The Combo effect against the Butterfly effect
Artificial Intelligence to boost Energy Storage that will improve resilience and efficiency to the Renewable Energy production
In the chaos theory, the Butterfly effect is the sensitive dependence on initial conditions in which a small change in one state (of a deterministic nonlinear system) can result in large differences in a later state. As of today, we’ve already sent tons of “Butterflies” on the Globe, whether it’s the CO2 emissions for our electricity generation or for our transportation, whether it’s the forests that are disappearing in front of extensive (or excessive) agricultural production, or it’s the methane that melts out of the permafrost or the “islands” of plastic that decomposes in the ocean. For the “non-climatologist”, 2 or 3 degrees might not look big, but the effect of such a change in global temperature would be way larger than what we’ve seen so far. We already have learned last year that we have about a dozen years or so left before climate change revved up so much that it will look like the Butterfly effect’s so-called later state: the big storms ahead.
Whether or not you understand the supercharged hurricanes, record-setting hot summers, and disappearing glaciers are a result of climate change or not, you can’t argue against cleaning up the planet, especially if there’s a profit to be had in doing so.
We definitely need to find ways to fight this “exponentially” multiplying effect.
The Combo Effect
We can be inspired by the Combo Effect that gamers (Anthems, Mass Effect, Blizzards’ Hearthstone, etc.) or guitar players do know and use in many environments. For example, in “Anthem”, “Combo system creates a dynamic way of damaging enemies. By using certain abilities in conjunction with each other, you can create a combo attack that can inflict crazy amounts of damage to your opponents.” By using 2 or more “powers” in sequence, the combined effects of each power can result in a greatly boosted effect. And it’s quite the same “combination approach’ that is also used in the guitar world to generate exceptional or crazy effects.
I strongly believe in the power of science and technology not only to understand but also to tackle humanity’s biggest challenges. That ‘s why I suggest we use a scientific version of the Combo effect, combining major solutions to generate a powerful response to the situation. Here the challenges (we can say our opponents) are CO2 emissions that we need to “crazily” reduce in a short time period. It’s not just a gentle reduction that is needed, it’s clearly a reinforced or a boosted effect. Indeed, it would not be the only solution needed. But it could really become an efficient one in front of an “exponential” opponent.
Let’s focus on the constructive stuff and on how a Combo Effect exploiting Artificial Intelligence and Energy Storage to improve Renewable Energy (RE) impact. So that investment in RE, which are already outpacing “less clean” energy sources, would be no-brainer thanks to the fact that Combo-Effect-Boosted RE could become even more economically sound, but also because this could become one of the big advocates in the fight against climate change.
Renewable Energy’s Strong Fundamentals
Following the 2019 Renewable Energy Industry Outlook, there are “Strong fundamentals bolstered by three enabling trends”. Those trends include emerging policies that support renewable growth, expanding investor interest in the sector, and advancing technologies that boost wind and solar energy’s value to the grid, asset owners, and customers. And, perhaps most significant, was robust demand from most market segments. Utilities demonstrated strong “voluntary demand” (52% of utility-scale solar projects in development and 73% of projects announced in the first half of 2018, as opposed to the demand driven by policy mandates we’ve seen in the past). In 2017, new solar energy projects alone accounted for 157 gigawatts in 2017, more than double coal, gas and nuclear. In 2018, the voluntary demand was partly driven by corporations’ rapidly growing appetite for renewables.
A Rocky Mountain Institute study has shown that they had purchased nearly 6.5 gigawatts (GW) of renewables through a variety of procurement routes (http://ow.ly/XCrP50r1Ha3).
Saving the planet has never looked so profitable; another sign of the maturity of the industry. Thus, the conversation is now turning to how artificial intelligence and energy storage can improve renewable energy. Indeed, a combination of tools, the Combo Effect, could lead to substantial improvements. That’s why we not only need a set of tools but also to efficiently integrate it.
AI for Energy Efficiency
We’ve already seen some hints about the possible ways AI and Energy storage (through Intelligent Energy Storage or IES) might improve energy efficiency in general in a preceding article: https://towardsdatascience.com/artificial-intelligence-in-a-no-choice-but-to-get-it-smart-energy-industry-1bd1396a87f8. Many tools are available to use with artificial intelligence and energy storage to improve a building’s energy efficiency after crunching data from dozens or hundreds of sensors. Through the Internet of Things (IoT) and AI, buildings’ energy efficiency can be improved substantially. IoT essentially is the networking of smart devices, buildings and other items with electronics, which enable the collection and exchange of data. IoT has been shown to provide energy efficiency when utilized correctly. “Things” may be simple sensors (e.g. temperature sensor in a room), more complex sensors (e.g. electrical power measuring device), actuators (e.g. HVAC room controller, motor), or complex devices (e.g. industrial circuit-breaker, PLC providing home, building or industrial automation).
The IoT application may range from a simple monitoring application such as gauging the temperature in a building, to a complex application such as providing complete energy automation of a campus. It is when it becomes complex that AI can be of major impact. And it is when electric loads and peak are substantial that Energy Storage can be an important asset. So, what is the impact of AI and Energy Storage integrated with IoT on energy efficiency? It’s all about data.
That’s right, the Internet of Things — the network of devices connected to the Internet and among them — has led (and will continue to lead) to great volumes of data. Increasingly large volumes of information circulate faster and more independently within the systems, making it easier to optimize resources. How? A constant flow of data within the production plant enables the continuous monitoring of energy consumption. Moreover, in the IoT interconnected system, with AI, it becomes easier to identify any problem and to intervene before excessive waste occurs, and with Energy storage, it’s possible to control when we react to changes in the load. Not to mention the cut to maintenance times.
In large buildings and in manufacturing the link between the Internet of Things and energy efficiency is directly proportional, and it becomes essential to invest in IoT solutions and energy efficiency. Indeed, the combination of IES with the Internet of Things (IoT) promises to reduce utility bills in smart buildings or to save costs by predicting industrial-scale problems before they occur, such as a solar farm downtime or a wind turbine failure.
In the bigger picture, AI could be the brains behind a small (facility size) or large (community or region size) smart grid that will take the input of thousands or millions of sensors to make real-time decisions about where to allocate energy resources. One very famous example of an energy-saving algorithm involves Google which use an AI developed by DeepMind to stop its data centres around the world from overheating. The tech giant turned over energy management of its data centers to AI, which back in 2016 first helped reduce energy consumption for cooling all those computer servers.
AI to Improve Renewable Energy Operations
Wind and solar offer particularly rich datasets that algorithms can chew on in order to provide insights on current problems predict issues that might arise, improve energy storage or identify the optimal layout of the next mega solar farm.
Let’s start with some low-hanging fruit, such as how. As an example, Raycatch, the Tel Aviv 2015 start-up (which has raised $7.3 million) is developing algorithms that can diagnose problems and optimize solar plant operations. Raycatch provides an “AI-based diagnostics and optimization solution” that takes all of the data produced by a solar energy plant and turns it into a daily, real-time action plan without any additional hardware installations. It will even calculate ROI on its suggested improvements, as well as analyze actual performance against projected models.
AI to Design Renewable Energy Systems
Considering that there’s much more involved in designing a solar electric plant than just looking at the number of solar panels and their orientation toward the general direction of our huge star, the Sun. As we understand it, most of the design work today is still done manually, with an army of different engineers from various specialities — structural, electrical, etc. — often taking months to develop an optimal layout for big commercial projects. Why not let AI do it in a much faster and comprehensive way? That’s the business concept behind a Los Angeles high-tech start-up founded in 2013: HST Solar. Forbes has more details with an interview with the co-founder.
How is the AI platform exploited here: After the user inputs basic information, such as site location and details about the equipment to be installed, the algorithms go to work. The software then configures every piece of the solar farm puzzle, down to the specific orientation and tilt of each solar panel to maximize energy gain while minimizing problems from other factors such as strong winds. HST Solar has such a platform whose proprietary software can scan through hundreds of millions of possible designs for any potential site in a matter of hours, if not minutes. The objective is to identify which designs will yield the highest return on investment and the lowest cost of solar electricity.
HST Solar’s AI can identify designs that most engineers working manually often overlook. Its analysis can make projects more profitable and help attract investor capital. In emerging and developing markets, this approach can make sure that private and government investments in solar power are maximized. The company claims that AI-designed solar farms can reduce the cost of producing renewable energy by 10% to 20% compared to those systems designed by soon-to-be unemployed engineers. Using tools such as this AI platform can help the solar industry scale up from the 70+ gigawatts of new installed solar every year at present to the 1,000-gigawatt level the world will need each year by 2030 in order to address climate change in a meaningful way.
AI Weather Forecasts for Renewable Energy
Renewable energies, especially solar and wind, are obviously reliant on the weather. Proper design and operation are reliant on accurate forecasts and it is even more important if it is part of their AI-driven solar/wind and energy storage calculations. But instead of just getting a weather forecast, wouldn’t it be even better to get an energy forecast? That’s the concept behind Nnergix, a Spanish six-year-old start-up based in Barcelona that has raised about $1.7 million in disclosed funding. What Nnergix does is using weather data tech and machine learning to make energy forecasts. It recently launched a product called Sentinel Weather, which does “weather analytics” and provides access to historical data and weather forecasts in any location on the planet. Energy engineers will be particularly interested in the platform’s solar and wind power forecasts for individual plants at a 15-minute resolution for up to 18 hours. Hourly power forecasts are possible for one week.
AI for Renewable Energy Storage
Another emerging trend that was mentioned about in renewable energy is combining solar and storage at the outset, which can help deliver costs savings to both solar providers and users. Solar projects can be costly, especially given the volatility (and even elimination) of government subsidies for investing in renewable energy. One way to hedge against price fluctuations is by efficiently storing the energy, especially at low rates, and tapping it when energy prices spike. Stem is a startup sandwiched between San Francisco and Silicon Valley proper that has created an AI platform called Athena that says it can get the most value out of energy storage, including solar. Founded in 2009, the company has taken in a whopping $321.1 million.
The company says Athena can “future proof” a customer’s solar investment by giving it “greater control and flexibility on their energy decisions through automated, real-time, energy optimization.” Athena ingests reams of data per second, from solar generation and load behavior to electricity rates and even weather forecasts, according to an interview in Alt Energy Magazine. Algorithms then go to work to squeeze value out of energy storage, even going so far as to create virtual networks of power plants that can send or store power based on the most favorable economics. While the company concedes that adding storage to solar can increase the cost of the system up to 20%, those costs should be more than offset by up to 30% in savings from its intelligent storage solution.
Hybridization with Regulation Strategy Combining Thermal and Electric Capacities
Using this joint/collaborative approach and using their work in the hybrid energy storage for vehicles, Smart Phases (DBA Novacab) has developed the Hybrid Thermal and Electric Energy Storage System.
The control of the system is based on an anticipatory regulation strategy using fuzzy logic and a combined feedforward plus feedback control that can handle simultaneously the storage and retrieval of both electricity and solar energy. AI is required to get the full potential of the approach. It takes into account a multitude of operating conditions such as load, outside air temperature, and optimizes the off and the on-peak periods for electrical heating. The combined strategy can significantly improve performance over simple feedback control whenever there is fluctuations or disturbances. Electric Energy Storage (EES) and Thermal Energy Storage (TES) have been integrated into a Hybrid approach in order to optimize energy efficiency and load leveling. This integration is allowing for significant improvement and stability in the operation in critical applications such as Hospital, Datacenters, Military, Manufacturing Plants and other critical thermal + electric demand-side management.
The Novacab Hybrid Thermal & Electrical Energy Storage System maximizes the flexibility and the overall performance of the equipment on the grid. In a building set-up, it allows for performance improvement and more reliability in the operation. Generalizing for the grid, the impact would also be substantial: smoothing the load profile and optimizing demand-side management; improved Redundancy and Predictability of the energy distribution. The integration and combined outcomes of the hybrid systems is highlighted in a paper published by the DOE’s Sandia Lab after the EESAT conference, including on-site operational data, Power Usage Effectiveness, reliability, and performance. This “predictive” approach is allowing for better use of the Energy Storage but also for better integration of EES and TES to maximize Operational flexibility and stability, Performance improvement in the operation, Demand-Side Management with Predictability, Reliability in the operation.
AI for Renewable Energy R&D
As discussed, AI is more and more involved in designing renewable energy and energy storage systems, analyzing their performance, and predicting future production and storage needs. Some companies are leveraging machine learning from the very beginning in the research and development phase, a trend we’ve seen in other areas of research and AI.
SunPower (SPWR), a Silicon Valley-based company recently built a new solar research facility. One of the key tools at the new $25 million facilities involves quality control over the photovoltaic cell manufacturing process. AI is used to analyze the process to not only ensure a quality product but to provide new insights to improve the technology over time. Incidentally, SunPower is embracing all kinds of emerging technology, including the use of drones to survey potential solar power farm sites in order to create more efficient designs. Engineers might employ machine learning there as well.
Finally, a small startup out of Seattle called Energsoft, which has raised about $170,000 in disclosed funding, uses AI to build better batteries for renewable energy customers, particularly for electric vehicles and utility companies. The company’s AI platform automates data collection and identifies design problems, such as material choices or even the manufacturing process. Founded last year, Energsoft says its software can reduce development cost and time-to-market.
The long-term viability of the renewable energy sector requires economic scalability, meaning being able to build enormous energy systems that are more efficient and cheaper than non-renewables. So AI-accelerated R&D would be required to keep up the pace in the fight against climate change.
Coming back to our initial thoughts, we can be filled with the urge or ability to do all this with the Combo Effect. The smart combination of artificial intelligence, energy storage and renewable energy seems to be the perfect marriage of emerging technologies with a maturing industry: the energy sector.
That was why I suggested using an “engineer” version of the Combo effect, combining major solutions to generate a powerful response to the global challenge (the Butterfly-like booming CO2 emissions and the storms ahead) that we need to “crazily” reduce in a short time period.
By using the abilities of AI, IoT, Energy Storage in conjunction with each other, we can create a combo “powered” that can result in a greatly boosted RE. This way, we’ll maximize the chances to tackle humanity’s biggest challenges, climate change, and to create a bit of wealth along the way for the savvy stakeholders.
Until we start producing cheap, limitless nuclear fusion energy (and that would be far too late to address the urgent climate change fight), we’ll need to find ways to conserve resources and make renewable energy systems cheaper and more efficient to manufacture, install and operate. The development of AI-powered IES solutions in many facets of the renewables energy supply chain won’t just make it cheaper, it will allow us to afford new clean ways to use energy, such as to desalinate water or to electrify remote or very poor communities, on a mass scale. It would mean a brighter, if not too warm, future for all.
This article is part of series AI and Energy Storage by Stephane Bilodeau, ing., P.Eng, Ph.D, FEC. You can find another article here: https://towardsdatascience.com/artificial-intelligence-in-a-no-choice-but-to-get-it-smart-energy-industry-1bd1396a87f8