Deploying Composite AI and Historical Know-How for Energy Management

clock-icon-white  5 min read

Advanced AI solutions shouldn't boggle the mind. The most effective approach is always based on what you already know.

You’ve been hearing for years about the coming AI-driven revolution in manufacturing. With a looming recession, record inflation, and skyrocketing energy costs, it couldn’t come any sooner. Energy management, in particular, seems like a perfect place to deploy advanced optimization algorithms. So why haven’t manufacturers fully leaned into deploying AI for energy management?

AI in Energy Management

Their hesitance is understandable. AI algorithms tend to be highly complicated black boxes, and failures are commonplace. And because energy-intensive processes are essential to production, attempts to reduce their energy use cannot be allowed to interfere with their effectiveness. The stakes are simply too high to entrust the future of your company to a technology you don’t yet have complete faith in.

But you can make good on AI’s potential for energy management. It just needs to be accessible to non-technologists. That means making it simpler and building it into existing operations, which hinges on software engineers synchronizing their solutions with the business and operational sides.

For AI to work for you, you should be able to trust it — the algorithm must adapt to your business.

Putting AI in its place

No one knows your operations better than you. So, you've probably already identified processes that could be made more energy efficient. The tricky part starts when you try to reduce energy consumption with advanced technologies.

Manufacturing AI Efficiency

That was certainly the case for one of our clients, a car manufacturer. Its paint booth required large volumes of air conditioned to a precise humidity and heated to a specific temperature. This energy-intensive, climate-controlled operation eats up a lot of energy. At the time, the air conditioning equipment was reactively set according to changes in the input air temperature and humidity. That meant that settings were frequently suboptimal and resulted in energy inefficiencies.

Thus, the goal was to move away from reactively calibrating equipment to using a proactive, dynamic approach. While a complex predictive AI algorithm would theoretically provide just that, rushing to deploy the technology would not have fully utilized our client’s knowledge. Moreover, the risk of failure — in the form of production stoppages or quality losses — would have been too high. So, we aligned on a different tactic.

How to integrate the algorithm into your production

To build proactive control into the pre-existing workflow, we first assisted our client to draw on historical data and deploy new sensors to gain visibility about the machinery settings and weather. By generating a data-based picture of the paint booth using system data analysis, our client gained actionable insights into the it's dynamic energy usage and inefficiencies.

Using those insights, the facilities manager was able to set the equipment with greater dynamism and precision. And thanks to those dynamically optimized equipment settings, management reduced the paint booth’s energy consumption by almost 10%.

At this point, our client had built a robust and intelligible data platform on top of their workflows and equipment. With a deep understanding of how this platform interacted with the heating and humidifying infrastructure, our client saw exactly how AI could be integrated into an energy management solution.

Manufacturing AI Efficiency

Making good on the promise of AI

To round out the tech stack, we collaborated with our client to create a composite AI solution using low-code methods. Drawing on the data generated by the paint booth, the algorithm anticipates the future optimal settings of air conditioning equipment based on weather and energy dynamics

AI Dynamics

After deploying an internet of things (IoT) control solution to automatically implement the algorithm’s recommendations, our client saw a further 10% reduction in energy consumption in the paint booth.

With a total reduction of energy usage of around 20%, our client will see substantial returns from the project. The AI intervention was minimally invasive and wholly understandable by virtue of being coordinated with the existing operations and equipment.

That resulting decrease in utility costs will bring up to €4 million in annual savings — representing an ROI of over four times after the first year.

Simplicity: A mature approach to AI solutions

Manufacturing AI Efficiency

The problem with many technology partners is that they don’t design their AI solutions around their clients’ existing expertise and infrastructure. That’s a shame. Manufacturers have deep knowledge of their operations, and software engineers should take advantage of it.

Because our client’s AI solution is based around established practices and knowledge, it scales easily and can be implemented in less than six months. With a potential deployment across multiple plants, our client’s energy use reductions will be rapid and affordable — and less energy-intensive — across the board.

Drawing on this AI solution’s high transferability and auditability, our client can quickly test and roll out further energy reduction initiatives throughout the production process. Together, we helped our client develop a recipe for repeatable success: Melding the power of composite AI with internal practices for maximized energy efficiency.

Using AI to reduce energy use doesn’t have to mean a leap of faith. If you are interested in acquiring a proven energy management solution for your manufacturing facility, SoftServe can help.

And if you want to learn even more about data-driven energy management, read our white paper.