GNY Climate Change Use Case
Updated: May 14
GNY has built neural nets that are now predicting electricity demand in California 5% more accurately than the U.S. Energy Information Administration.
To run our ML model predicting California electrical demand CLICK HERE
To run our ML model that predicts California’s energy supply CLICK HERE
We have also built off of these powerful neural nets to predict California’s exact date of “peak” fossil fuel consumption.
As well as this we are providing a vision for how a machine learning powered blockchain could revolutionize renewable-only electric grids. Our hope is that these datasets may advance conversations that investors are having about divesting from fossil fuels.
Our team is eager to connect with climate change focused NGOs, and with readers interested in collaborating. Connect with us on Telegram, or Twitter.
Introduction- How energy data, machine learning, and blockchain can fight climate change by hastening the divestment from fossil fuels
GNY set out to build a powerful platform for developers and businesses that combined the power of machine learning with a blockchain. Our goal was to provide a platform that delivered high transaction rates, and easy to use tools through a secure on-chain machine learning. We did this because we believed in the capacity of these two technologies to work together to solve large and complex problems. Climate change is the perfect example of such a problem. To fight it one needs to collaborate, successfully incorporate both private and public data, and access to the most cutting edge machine learning.
A growing movement among NGOs fighting climate change is to target the producers of fossil fuels where it hurts them most- their bottom lines. The extraction of fossil fuels is expensive, dangerous, has huge environmental costs and demands a constant flow of capital. NGO’s are starting to convince investors that fossil fuels are a poor investment when compared to renewable energy. In order to succeed at this effort these groups need convincing data-driven arguments for investors. When will peak fossil fuel consumption occur? What are the true all costs (environmental, fiscal, societal) associated with different energy sources? Our team decided to investigate how our technology might be of service.
Our goal in producing a GNY use case for climate change was two-fold. First, we wanted to demonstrate how our ML and BC technologies could be applied to combat climate change. Secondly, we hope that these demonstrations might attract the attention of NGOs for a real-world application and partnership. We are NOT climate experts, but we have built a powerful tool that we think might be impactful.
In order to show powerful results in a short period of time we narrowed our scope to focus on the electric generation market in California. Production of electricity accounts for approximately 35% of all fossil fuel consumption, and 63% of all electricity is generated from fossil fuels. Additionally, electricity generation represents a market that is actively being disrupted by renewable energy. How could superior predictions, smart-chain technology, and machine learning advance that disruption? That is where things got interesting.
In this live and ongoing GNY use case we will demonstrate how GNY’s proprietary machine learning technology and future integration of our blockchain technologies can be applied to:
- Outperform Near-Future Supply and Demand Energy Predictions from the U.S. Government
- Predict when the “Peak” will be for Fossil Fuel Generated Electricity in California.
- Demonstrate Renewable Energy Can Provide As Stable Energy at Lower-Cost Than Fossil Fuels to Encourage Divestment from Fossil Fuels
- Demonstrate how Fossil Fuel Generated Energy has higher lifetime cost and should be divested from
- Model how moving these models onto the blockchain could provide increased collaboration, impact, and faster transition to a greener world.
This work will be published in stages, and then the full use case will be published when the work is complete. We will also provide models for how blockchain can be utilized to build systems to support a smart grid of renewable electricity and other financial incentives to speed the transition to greener energy.
How does prediction of peak fossil fuel consumption influence the fight against climate change?
Peak consumption is not just a buzz worthy term- it is a key indicator of change between a legacy technology and a disruptive innovator. This “peak” is the moment when supply starts to overshadow demand, prices plummet, and the entire fossil fuel industry starts to contract. Our team set out to make a process to define “peak” that could be applied to various geographies. The below chart from the excellent organization Carbon Tracker displays it beautifully.
Calculating the peak fossil fuel consumption helps to analyze how competitive renewable energy is cannibalizing demand for fossil fuels. This process is easily achieved in hindsight, but predicting it is a much more difficult task. This peak spells the beginning of the end for the legacy tech for a number of reasons. In this particular case here are some of the biggest contributing factors:
- In the US the price of solar has decreased by 70% in the last decade and the home solar market has grown by 49%. The price of producing sustainable solar energy will continue to decline ensuring that solar costs become more and more competitive in the marketplace.
- The cost for sourcing fossil fuels are incredibly high and shows no sign of slowing down
- As the effects of climate change become more and more severe there will be an increase in restrictions and limitations and tariffs of carbon f combustion of fossil fuels. For example: California has already outlawed the sale of combustion vehicles after 2035. The trend to renewable electricity is growing and unstoppable.
This use case also provides GNY the opportunity to demonstrate how we approach complex data problems, and the bespoke data solutions we offer businesses of all sizes. Our goal is to share our logic process, technology application, and the results as they become available.
Why We Selected California As Our Geographic Target
To help advance the understanding of the peak fossil fuel consumption in the US we decided to start looking at a smaller segment of the overall challenge that had great public data. After looking into several potential applications, the GNY technology team landed on analyzing California’s electricity supply for a number of key reasons:
- California is the largest economy in the US with a GDP of $3 trillion.
- Leadership in California has already signaled that they want to move to renewable power and away from fossil fuels.
- California over the past several years has seen some of the devastating effects of s changing climate from the biggest fires in the state’s history to record temperature and drought. So, ideally, our work here may actually help inform policy and future legislation.
Phase 1- Generate the Most Accurate 24-Hour Predictions of Supply and Demand for Electricity in California
Before we could predict the future of electricity supply and demand in California, our team needed to build a complex and powerful multi-factor ML model that could explain the present. Building this predictive model has two additional potential benefits:
- In order to effectively transition to renewable energy without destabilizing the power grid, accurate predictions for supply and demand are essential. This will empower utility companies to freely move more resources to renewable energy. This is often referred to a “smart grid”.
- As electricity suppliers feel secure in moving to renewable energy they help advance the divestment from fossil fuel generated power.
It took our proprietary GNY ML engine a month of training to start to provide accurate next day predictions. The graph below shows how our predictions improved as the neural nets continued to self-learn and adjust the weights of the hundreds of parameters we used to correctly predict the next day’s supply and demand. A full excel file of our predictions vs actuals is available for download below.
Our team also compared our neural net’s predictions to those made publicly available by the U.S Department of Energy. Our team analyzed the energy predictions that the EIA was sharing and concluded that they were generated using linear regression techniques. When compared to the predictions generated on our neural net (after it’s one-month training period) GNY delivers an improvement of approximately 5% for next day predictions.
This 24-hour improved prediction let us know that we had constructed neural nets that were sufficiently complex and accurate to start predicting further into the future to start to predict the “peak” and other compelling data points that will hopefully erode investor confidence in fossil fuel’s long term potential profitability.
Our excel downloadable file shows our training period, predictions, actuals, and the predictions made by the California Department of Energy.
This is just the first step on our multi-segment use case. In the following weeks we will continue to publish content including, but not limited to the following:
- Predicting the “Peak”
Now that we have the neural nets accurately predicting 24 hours in the future we can start to predict more impactful information that will help influence investors make more accurate decisions about continued investment in fossil fuels. There is a chance that the peak has already occurred, but showing an accurate prediction of the upcoming energy transition will assist investors in making investments that reflect a rapidly changing market.
- Predicting Supply Needed for Stable Electric Grid
In order to effectively transition to renewable energy without destabilizing the power grid, accurate predictions for supply and demand are essential. This will empower utility companies to freely move more resources to renewable energy. This is often referred to as smart grids.
- Blockchain Integrations
The concept of a “smart grid” demands a “trust-less” system, and one that can be constantly learning and improving regardless of the preferences of individual energy producers. We will model out how a GNY decentralized smart-chain could be the solution everyone has been waiting for.
***With an NGO partner we could do much more. As we stated we are not climate experts, but our entire team is extremely dedicated to this issue and committed to being a part of developing solutions.
Our team is eager to connect with climate change focused NGOs, and with readers interested in collaborating. Connect with us on Telegram, or Twitter.