This article is part of a VB special issue. Read the full series here: Intelligent Sustainability.
As enterprises look to operate more sustainably, they are demanding more from their digital infrastructure – not only from a cost and efficiency perspective, but from an environmental one, too.
Arno van Gennip, vice president of global IBX operations engineering at Equinix, told VentureBeat, “Digital twins are becoming key to improving data center efficiency and reducing our customers’ carbon footprint at every stage – from design to construction to facility management.”
Digital twins help to centralize data from across different areas of concern into a shared environment. This allows IT, engineering, finance, procurement and construction teams to explore and simulate the performance, financial and environmental tradeoffs much earlier in the process. Various efficiency gains in equipment and space utilization directly reduce energy and carbon footprint. Digital twins can also help improve construction and operational efficiency to reduce waste, staffing requirements and the associated environmental footprint of these activities.
Enterprises and data center operators like Nvidia may cobble together a digital twin workflow from an assortment of simulation and modeling tools that combine engineering, CAD and data center information management (DCIM) capabilities. Increasingly, DCIM vendors like Schneider Electric are introducing digital twin capabilities directly into their tools. Vendors like Dassault Systèmes and Future Facilities provide more integrated digital twins for data centers. And companies like Nvidia are starting to roll out new tools like Nvidia Air for optimizing data center physical and logical layout.
Putting it into operation
Equinix works with Future Facilities to build digital twins for several of the company’s data centers. The digital twin helps engineers ensure cooling systems and connected ecosystem components are designed to deliver the required capacity and optimal efficiency. Engineers can compare the expected and actual behavior and energy use of data centers.
“This provides us great insight into required maintenance and possibilities to optimize energy efficiency,” van Gennip said.
Equinix engineers work with partners to create a 3D model of the physical data center. The data center twin is modeled based on various factors, such as the capacity and density of computing equipment within the data center and cooling system paths. A centralized digital twin platform helps engineers predict proposed changes’ impact on power distribution, space utilization and cooling paths using live data, like power and temperature. This real-time data is merged into the existing model for accurate analysis and predictions, allowing the data center’s twin to boost efficiency by forecasting energy needs.
Dassault Systèmes works with leading hyperscale data center companies to design and construct next-generation data centers.
“Their biggest challenges are how to reduce project standup time to keep up with the growing demand and how to make data centers more sustainable by reducing energy, water consumption and waste during construction and operations,” said Marty Rozmanith, sales strategy director for the architecture, engineering and construction (AEC) industry at Dassault Systèmes.
Making it easy
Historically, data center management has been split into silos that each focus on one aspect of managing a facility, Kasper Dessing, director of global building management optimization at Digital Realty, a data center real estate investment trust, told VentureBeat.
As a result, managers of different areas can miss the bigger picture. This becomes particularly important when looking at facility maintenance, now and in the future. Data centers produce inordinate amounts of data that is impossible for humans to capture, aggregate and manage. And this will only get worse as digital services continue to become more sophisticated.
“With digital twins, we’re able to take a virtual representation of the elements and dynamics within our facilities and simulate their actual behavior, in real time, under any operating scenario,” Dessing said.
Digital Realty has found that the generic data centers of their operations are not good enough because of the amount of data and interdependencies between components. Because of this, Digital Realty integrated digital twins of its facilities with its own proprietary artificial intelligence (AI) and machine learning (ML) platform to analyze thousands of data streams. This allows them to track all the components within facilities and make real-time adjustments. It can also help predict behavior in the future for predictive maintenance, saving time and cutting costs.
This visibility into facilities and the relationships between individual components also helps improve new facility designs to make them more efficient. Digital Realty also uses digital twins and its AI platform to optimize energy consumption.
“Sustainability is a priority for us, and optimizing energy consumption at each of our facilities helps us to cut costs and reduce our impact on the environment simultaneously,” Dessing said.
Not every employee has the technical expertise to run a simulation when making a decision. So Digital Realty integrated a recommendation engine into its digital twin platform.
“This allows us to make the technology accessible to a much broader range of colleagues, so we don’t have to rely on the experts all the time,” Dessing said.
How the pieces fit together
The process of designing, building and operating a data center generates a lot of data, which is stored in different formats and disparate systems. Managing and organizing the data with appropriate access control and change management is very challenging, said Rozmanith. Digital twins can bring data from multiple disciplines, levels of development (LOD) and multiple scales. This allows multiple stakeholders to collaborate on a single source of truth in real time. More sophisticated digital twins combine various techniques for simulating thermal, structural, electrical, control and monitoring, manufacturing and assembly using an integrated digital twin.
“With a common platform, everyone will be working from a single source of truth, resulting in time savings, quality improvements and overall data center delivery,” Rozmanith explained, “The platform is a change agent.”
Enterprises are increasingly investigating how to bring together multiple digital twins.
“As we’ve started incorporating more data and simulations that connect engineering designs, construction scheduling and operations, interoperability across twins has become a challenge,” said Teresa Tung, cloud first chief technologist at Accenture.
Tung’s team is working with data center providers to apply data and domain expertise to analytics to determine the number and configurations of simulations needed to drive what-if predictions. They use domain knowledge graphs, the same technology that underpins internet search, to capture these requirements and map their relationships between elements.
Carsten Baumann, director of strategic initiatives and solution architect at Schneider Electric, said that providers are increasingly adding digital twin capabilities to DCIM tools to simulate the impact of infrastructure upgrades before the actual physical deployment. He believes that open standards that lead to simpler integration between data center equipment and management tools make it easier to use digital twins as part of everyday data center workflows.
Here are 19 ways digital twins can improve sustainability across design, construction, operations and planning:
Placing new servers
“Perhaps the biggest impact of using digital twins in the data center industry is in airflow management and IT equipment placements,” said Baumann.
The rapid and increased demand to deploy compute, storage and network resources comes with significant infrastructure challenges. Just because there is physical space in a particular rack or location does not mean that there is sufficient power, connectivity and heat removal capacity available.
Digital twins can help when a seemingly simple installation might require a significant power upgrade and recommend a better alternative.
Increasing the density of equipment in a data center can reduce the climate impact of constructing new facilities.
Loren Absher, director of enterprise agility with Information Services Group (ISG), said digital twins can help optimize data center designs to improve all the associated elements of power, cabling, cooling requirements, airflow and even raised floor integrity to prevent catastrophic failure. They can also help plan physical workflow changes required for the increased density.
Improving thermal performance
Cooling is the second-largest energy consumer in data centers, behind the equipment itself. A modern data center includes cooling systems composed of chillers, piping and HVAC equipment.
Digital twins can use thermal simulation to understand the cooling system’s behavior and improve its performance.
Rozmanith said teams often combine 1D simulation of the chain of equipment representing variation in the number of chillers and size of pipes with 3D computational fluid dynamics (CFD) analysis of air flows to find the best balance between the cold air generated and the cooling of equipment to optimize energy consumption.
Assessing seasonal impact
Digital twins can also help data center designers better plan for seasonal climate change, said Dan Kirsch, managing director and cofounder of Techstrong Research. In this case, designers can also plan for the impact of outside seasonal climate variations to reduce overall operational cost and energy footprint.
“Digital twins allow a truly customized and optimized design based on the specific needs and on-site conditions of a client without the need for on-the-ground experimentation,” Kirsch said.
Creating modular components
Dassault works with large-scale data center operators to create modular components that can be reused across different data center designs.
Rozmanith said digital twins help enterprises define and configure the properties of these modules to reduce the design, procurement and installation time with a configure-to-order approach. This can help reduce the environmental footprint of new data center builds.
Testing and validating equipment
Bruno Berti, senior vice president of product at NTT Global Data Centers Americas, said they are using digital twins to test and validate equipment before deploying it into data centers.
These new workflows allow them to build and test electrical and generator modules, so engineers can identify any process failure before the product goes into production. This reduces the environmental impact of waste and improves risk assessment, accelerates new product development and enhances data center reliability and resiliency. The digital twin also helps schedule predictive maintenance to lower maintenance costs.
Optimizing battery performance
Digital twins can also be used to model and design systems to improve battery health and expected lifespan, according to Greg Ratcliff, chief innovation officer at Vertiv, a company that makes data center equipment. This could help reduce the environmental footprint associated with creating new batteries. In this case, digital twins help teams simulate different design choices using battery health measurements and facility details to predict every battery’s health and service life in a string.
“If a single battery in a string fails, the entire string fails, so monitoring the health of each battery is critical,” Ratcliff said.
Assessing environmentally friendly alternatives
Data center operators can assess the performance, environmental benefits and downsides of new approaches.
For example, Kao Data used Future Facilities digital twin tools to virtually test and deploy a refrigerant-free indirect evaporative cooling (IEC) system that uses water evaporation in place of mechanical systems to cool air on hot days. This helped Kao improve its power utilization effectiveness and reduce its environmental footprint.
Digital twins can simulate complex tasks, assembly, equipment usage and human safety. They can also improve collaboration across the design and construction ecosystem of suppliers, integrators and contractors to remove the process friction.
Rozmanith said the combination of better simulation and collaboration could reduce construction time, problems, rework, requests for information and the number of safety incidents. This has helped Dassault customers reduce time to market by an average of 10-15% and reduced the environmental impact associated with longer construction times.
Reducing construction waste
Data center designers are using digital twins to better plan for construction so that crews can work more efficiently, creating less waste and reducing the time between different stages of construction.
“By creating a virtual model of the data center along with the full bill of materials, designers can fine-tune how a construction crew will assemble the data center down to every last detail,” said Kirsch.
This planning helps to reduce the need for teams to wait on standby as other teams complete their portion of the build. Reducing waste during the construction of a data center is not trivial. Many of those components cannot be reused or recycled and instead end up in landfills, he said.
Digital twins can help identify the root causes of issues and make maintenance recommendations for quick fixes that reduce energy.
For example, a digital twin model of Equinix’s Amsterdam facility pointed out that they had to clean the cooling towers and tune the fans, which both used more energy than the model expected. This led to an further 10% improvement in energy efficiency in an already efficient data center IBX, said van Gennip.
Extending asset life
Dassault’s virtual twins can contextualize operations data for AI and ML algorithms to improve predictive maintenance. Rozmanith said this extends the life of equipment, which reduces e-waste. Virtual twins can also optimize energy and water use by improving the efficiency of cooling and power systems.
More efficient maintenance and repairs
Digital twins can simplify access to all the information required to do maintenance, repairs and refurbishments more efficiently such as documentation, user manuals, maintenance manuals, material supplier information and spare parts lists. Lorenz Hofmann, vice president of custom air handling and modular solutions at Vertiv, said this saves time and work and reduces the CO2 footprint.
Automating data center processes
Improvements in process mining capabilities can help data center leaders understand how their teams interact with applications and react to changes in the data center environment.
Ryan Raiker, senior director of process intelligence at ABBYY, said that the ability to understand and document procedures with digital twins could help data center teams identify automation candidates. They can also implement protocols to take action when failures happen to ensure data center uptime and reduce failure and waste.
Improving collaboration between colocation providers and enterprises
Colocation data centers allow multiple enterprises to share the same data center. But when an enterprise client decides to install new equipment, this could have power, thermal and weight impact on nearby equipment owned by others.
Thésée DataCenter in France worked with Future Forward to deploy a digital twin of each facility in the cloud. This digital twin enables customers to explore the impact of proposed changes from their own or nearby equipment through a web service portal. This helps Thésée engineers collaborate with customers to improve their data center space usage and reduce the need for new data center builds.
Ensuring compliance with data twins
NTT is working on a concept for data twins to help enterprises collect and standardize data relating to all aspects of the business. The data twin replicates enterprise data sources and their interrelationships into a standard format to provide a centralized location for analysis and reporting.
Bennett Indart, vice president of SMART world solutions at NTT Data Services, said this would help data centers report their progress in complying with sustainability goals. It can also identify new opportunities for improvement.
Improving finance decisions
NTT’s Berti said that NTT has started integrating financial data into its digital twin. This helps NTT review the cost of materials and labor using real-time data and advanced analytics as part of their planning.
It can also help determine whether adjustments to the manufacturing value chain are financially sound and if the projected outcomes reduce data center operational costs.
Assessing data center migration impact
Accenture worked with Carnegie Mellon University on a digital twin that allows enterprises to measure the sustainability impact of migrations between data centers and cloud providers, called myNav Green Cloud Advisor.
Accenture’s Tung said it starts with a twin that baselines the current data center energy consumption, computing requirements and sustainability goals. The tool allows enterprises to plan and compare various cloud solution options that include carbon emissions goals, locations, energy sources and readiness to transition to clean energy.
Understanding material impact
Kirsch said it’s often difficult to know the actual bill of materials within a data center until construction is complete. During data center construction, teams encounter site-specific conditions that require deviations from the original design. Design teams can plan for all on-site conditions with digital twins and specify the materials needed.
“By creating an accurate bill of materials, data center creators and end-users have a full understanding of the materials to be used and their impact on overall sustainability goals before construction begins,” Kirsch said.