Operating a data center in today’s dynamic environment is getting increasingly complex.
From ever-evolving cooling systems and shifting IT workloads to extreme weather conditions and new PUE regulations. Factor in rising energy costs, and running an efficient data center has become a competitive edge. So, how can you ensure 99.999% uptime while simultaneously reducing energy consumption? AI (Artificial Intelligence) is the key.
Unlocking Intelligent Efficiency, Without Additional Hardware
Their approach involves harnessing the wealth of available facility data into holistic recommendations to optimize cooling systems, harmonize facility assets and UPS systems from roof to room.
With Coolgradient, there is no need for additional hardware or complex IoT (Internet of Things) projects. They have not only reduced energy consumption but also help extend the lifetime of mission-critical assets. Not to mention, your teams will have peace of mind knowing their systems are fully optimized to deliver site reliability.
The Coolgradient AI Platform
So, how does Coolgradient achieve these impressive results? Their in-house experts have trained their machine learning models with over 240 billion data points and collaborated closely with data center operations teams to validate our models.
Their process starts with collecting data from your cooling and energy systems. They then use their SaaS platform to analyze the data, automatically identify inefficiencies, quantify potential savings, and validate the findings. If required, they work with their operations teams to co-create implementation plans to ensure prompt implementation.
After implementations, your teams can use their platform to monitor any impact of the adjustments continuously. their goal is to ensure your entire infrastructure remains harmonized with the ever-changing dynamics of your data center.
Coolgradient Success Stories
Let’s explore how they collaborated with their customers to achieve these remarkable energy savings:
1. Cooling Efficiency
The traditional “set and forget” approach is no longer adequate as cooling systems get more complicated, compute configurations become denser and PUE targets get more challenging.
Their machine learning models have been trained to find optimization opportunities, inefficiencies and malfunctions in (multi-mode) hybrid chillers, drycoolers, cooling towers, IAC and water source cooling systems. With their customers they found many energy saving areas. Some frequently occurring problems they discovered are with chilled water rings, malfunctioning dry coolers, blocked filters, and malfunctioning 3-way chiller valves, to name a few.
Sometimes the teams had an idea that there was a problem but had not validated and quantified the impacts. Their tool did both and provided prescriptive recommendations. Other times they discovered inefficiencies that the teams were unaware of by using AI’s power. It was not only legacy data centers but many new data centers that were not operating exactly as they were designed to run.
Combined with their AI models and in-house experts, they delivered advisory services and collaborated with the data center operational and design teams. They carefully examined the findings, quantified their impacts, and provided clear recommendations to address these issues. Subsequently, they devised an implementation plan to realize the optimizations. After changes were made their data center software delivered enhanced monitoring which assured that savings were realized without jeopardizing customer SLAs.
Up to now, they have achieved substantial energy savings, conserving approximately 6,200 MWhs, specifically in data center cooling equipment.
In addition, these efforts have significantly reduced wear and tear on mission critical assets, enhancing their reliability and potentially prolonging their operational lifespan.
2. Data Center Cooling Harmonization
To enhance data center efficiency, their machine learning models analyzed the entire cooling system from roof to room to harmonize the complete data center cooling ecosystem. If some components of your infrastructure are not optimized, the rest of your cooling systems are working harder than needed, which consumes more energy and increases equipment wear and tear. Not to mention stress for the operational team, who may know it is not optimal, but are not sure where to adjust and what other impacts that will have on the entire system.
With this holistic approach our SaaS platform consistently found inefficiencies and our AI models recommended optimization solutions across the whole data center infrastructure. The most common optimizations identified were CRAC system tuning, CWR (Chilled Water Rings) setpoints, CRAH fan speeds, too-high modulating valve openings, optimal floor pressure and differential pressure setpoints.
Their tools reduced complexity and increased visibility by showing what the operation teams should focus on to have the most impact. After delivering recommendations, they collaborated closely with local operations teams and crafted implementation plans. They then used coolgradient’s SaaS platform to enhance monitoring to make sure that the optimizations they implemented had the desired effect and did not adversely impact other parts of the cooling system to ensure they met their ASHRAE standards.
Across multiple data centers, coolgradient achieved remarkable energy savings of 4,300 MWh by optimizing from roofs to rooms. These enhancements not only reduce energy consumption but also improve uptime and asset reliability, offering peace of mind to operations teams in delivering energy savings without compromising uptime.
3. UPS Efficiency
In 2023, they started focusing on UPS systems for our customers and integrated UPS data into our platform.
Their SaaS platform found many of the data centers where the UPSs were not at optimum efficiency. Some were even showing alarming health deterioration over time. Working with the local ops team we changed some settings, resulting in an energy savings of 3,650 MWh per year. Not to mention some peace of mind for the teams as we discovered a potential UPS risk.
Read the full article here.