Discussion on common faults and design of communication power supply systems

Local time 19th, Nov, we luckily attended the Provincial Low-Voltage Electrical Industry Association Summit. During the meeting, we discussed “How to manage the power system in data center”.

Besides equipment failures in UPS (Uninterruptible Power Supply) systems, battery banks, and power distribution systems, improper configuration and design can also lead to daily operation and maintenance problems in data center power supply systems. We should pay attention to the following aspects:

I. Architectural Risks (Systemic Risks)

- Single-circuit power supply problem

Follow a "2N" or "N+1" redundancy architecture. Achieve end-to-end redundancy from dual mains power, dual transformers, dual UPS systems, dual power distribution circuits to dual server power supplies and dual-path cable trays. Draw clear single-line diagrams and regularly rehearse switchover procedures.

- False Redundancy/Logical Defects

Adopt a pure dual-bus architecture with "isolated parallelism" to avoid design logic errors. Generators and mains power must not be connected asynchronously.

Dual-bus systems should not share battery banks or bypasses. Cables, cabinets, and equipment on different routes should be physically isolated to prevent simultaneous failure of both circuits due to fire, water damage, etc.

- Capacity and Load Planning

1. Insufficient capacity will lead to UPS and distribution cabinet overload operation and frequent tripping. Low load operation will waste energy and damage equipment. Modular and Scalable Design: Utilizes modular UPS and pluggable power distribution modules, with sufficient reserved incoming line capacity, space, and heat dissipation capabilities to avoid forced overload operation later.
2. Introduces DCIM tools to monitor circuit and UPS load current and power in real time, performing end-to-end capacity simulation and prediction from cabinet to bus. Avoids three-phase imbalance and localized overheating.
3. Pay attention to harmonics and power factor: Install active harmonic filters at the UPS input or key power distribution nodes, or select UPS with built-in high input power factor correction. Avoid falsely labeled transformer and cable capacities and neutral line overload.

II. Equipment Selection and Configuration Potential Issues

- UPS System Configuration

1. Configure the UPS system appropriately, considering efficiency, space, expansion requirements, and operating costs when choosing between line-frequency and high-frequency UPS, and tower and modular UPS.

2. Set reasonable battery backup time according to fire protection strategies to ensure timely generator start-up (usually 10-15 minutes). 3. Limit the number of parallel units; generally, it is not recommended to exceed 8 units in parallel to ensure current sharing and the reliability of communication links.

- Power Distribution System Design

1. Conduct protection coordination studies using A-S curves of upstream and downstream circuit breakers (or fuses): Plot TCC curves to ensure the selectivity of protection devices at each level from the end to the source.

2. Use MESH-BN or equipotential bonding networks for AC working ground, DC ground, lightning protection ground, and equipment safety ground. All grounding points should converge to the main grounding busbar.

3. Standardize end-point configuration: Mandate the use of dual PDUs. The PDU output circuit breaker must be strictly matched to the server plug type; the use of adapter plugs is prohibited.

Additionally, we must pay attention to the lack of emergency design and physical safety. An excellent power system design is not only about drawing a perfect architecture without single points of failure on paper, but also about anticipating all the challenges that may be encountered in operation and maintenance over the next ten years during the design phase and preparing safe and efficient solutions for them.

Previous
Previous

The first 40HQ container of Charmlink