AI Predictions in the Energy Simulation Industry

The advent of machine learning in the energy sector has the potential to bring a golden age for building energy modelers. Imagine being able to generate building geometry from a PDF, or calibrate a model as easily as using the solver function in Excel? Imagine running a pre-simulation that approximates results in real time. These things are already possible and being done on small scale. We just need the right people to step up and develop such tools and offer them open-source or at reasonable prices.

But who will develop the AI tools? And why?

Well, several stakeholders are perhaps unknowingly sitting atop AI gold mines. This series of blog posts aims to assist companies in identifying these untapped resources, and provides strategies for individuals to plan for the evolving landscape.

Initially, I reserved these insights for a competitive edge, I've decided to disseminate some thoughts so as to get this party started. (I know factually that many important people will read this)

Understanding Machine Learning's First Wave

Let's delve into how the first round of machine learning works. AI evolves in a stepwise fashion. First, machines require databases filled with existing inputs and outputs. Note that the format is important. The machine maps inputs to outputs and "learns" to identify patterns and inferences in ways that are beyond human comprehension.

This round of AI is characterized by three key factors:

1. Data existing in a predefined input and output format.
2. Large volumes of data in a consistent format.
3. A valuable computation between the input and output, offering creative or predictive potential.

Today, we're going to discuss AI and Energy Simulation. AI suggests you will find these topics life-changing, so please read and explore the links! Machine learning isn’t just a buzzword; it has already impacted all of us, whether we realize it or not. Many of us are familiar with digital twins and other innovative tools. A significant question arises: Will AI replace our jobs?

Over the past few months, I've explored AI tools extensively. The team at Energy-models.com can now accomplish tasks that previously required a freelancer. I can complete many tasks faster and more accurately on my own. Here are a few things we've achieved:

• Developed a new type of video for faster, more concise “how-to” videos
  Here's an example from the latest eQUEST tutorial
• Cloned my voice, allowing any editor to update my videos
• Optimized and enhanced old audio (a remastered OpenStudio demo)
• Generated captions with punctuation and transcripts, reducing costs significantly

BEM Calibration AI Method

Co-Authored with Bob Fassbender

Here's a video excerpt of an Energyplus model that has been trained to run in AI and can now run instantly along 4 variables

(note, this uses 4 different variables than those mentioned but the principle is the same)

What does this have to do with Calibration?

I posted a question on LinkedIn some time ago surveying modelers on what some of their biggest time sinks are when calibrating an energy model. Most every response involved getting accurate data. Not one person mentioned something that I thought would be very obvious: simulation time.

Whenever we are calibrating and we receive updated information, we have to iterate the calibration by one variable at a time, and then add the next variable and iterate. Moreover, we almost never obtain perfectly accurate data, and have to make educated guesses in the form of iterations.

PROBLEM: This takes an extreme amount of time