Redundancy must be tested and verified. It is not enough to include it in drawings.
Rising Power Demand in Data Centers
Power demand in data centers is growing rapidly, especially in AI and high-performance environments. Rack densities above 100 kW are now common in these installations.
This leads to a high concentration of power in limited space.
The consequences are clear:
- higher load on switchboards and components
- stricter requirements for selectivity and protection
- less margin for design errors
Mistakes in early phases lead to greater impact during operation.
Redundancy Must Work in Practice
Redundancy is part of the design in modern data centers. It is not an add-on.
Common principles in UPS systems:
- N – covers demand without backup
- N+1 – one additional unit secures operation in case of failure
- 2N – two independent systems provide full backup
The difference lies in what happens during failure.
In an N system, protection is lost during failure or maintenance.
In an N+1 system, operation continues even if one unit fails.
Typical issues in practice:
- incorrect load sharing between UPS units
- lack of selectivity
Redundancy must be tested and verified. It is not enough to include it in drawings.
Larger Installations Bring New Challenges
Data centers are built with higher total power than before.
This introduces new technical challenges:
- higher short-circuit levels
- stricter selectivity requirements
- more complex interaction between systems
The key issue is how power supply, protection, and distribution work together.
Failures in this interaction directly impact operations.
Location and Use Define Requirements
Data centers are not built the same way.
- Latency-critical solutions are placed close to network hubs
- AI and processing environments have more flexible placement
This leads to different needs:
- AI with decentralized solutions and high power density
- Traditional hyperscale with focus on low latency
- Both with strong focus on access to nearby energy
Power supply must match the application.
Delivery Does Not End at Installation
Power supply is not complete when the system is delivered.
It requires follow-up in operation:
- monitoring of load and status
- battery testing and maintenance
- documentation and follow-up
Lack of follow-up is a common reason for reduced reliability over time.
Service and Aftermarket Define Availability
Availability is not defined by design alone. It is defined in operation.
Anda-Olsen works across the full lifecycle of power supply solutions, including:
- UPS service and maintenance
- battery monitoring and testing
- troubleshooting and upgrades
- spare parts and remote support
This ensures long-term performance.
Relevance Across Segments
This development affects multiple types of installations:
- critical infrastructure
- data centers
- hospitals
- power and grid systems
- industry and automation
- maritime installations
All share the need for continuous operation.
Anda-Olsen’s Experience
Anda-Olsen delivers power supply solutions for installations with strict uptime requirements.
Experience from maritime, industry, and infrastructure provides a solid foundation for:
- handling high power demand
- designing redundant solutions
- system integration
- deliveries with high precision
This experience is directly relevant for modern data centers.
What This Means for You
The trend is clear:
- power demand is increasing
- redundancy is becoming standard
- availability requirements are tightening
- consequences of failure are growing
For you, this means power supply must be included early in planning.
One Question to Ask
What happens if one part of the power supply fails?
If it is not tested and documented, the solution is not robust enough.
FAQ: Power Supply in Data Centers and Critical Infrastructure
- What does redundancy mean in UPS systems?
- Redundancy means the system tolerates failure without stopping operation. This is achieved through solutions such as N, N+1, and 2N.
- What is the difference between N and N+1?
- N covers demand without backup. N+1 includes one additional unit that maintains operation in case of failure.
- Why is power demand increasing in data centers?
- Increased use of artificial intelligence and high-performance computing leads to higher power per rack.
- What is typical power per rack?
- Traditional data centers: 5–20 kW
AI/HPC environments: 100–150 kW or more - Why is service important?
- Because long-term reliability depends on maintenance, testing, and monitoring.
Would you like a review of power supply, redundancy, or availability in your facility?
Contact Anda-Olsen for a professional assessment based on your requirements for uptime and reliability.
