Electrical Design for Data Centres: What Makes It Different
Data centres are among the most electrically demanding facilities being built in Australia today. On the surface, the electrical design brief looks familiar — transformers, switchboards, distribution, protection. But the assumptions underneath are unlike almost any other sector, and designing a data centre as if it were a standard industrial facility is a fast route to problems.
Here’s what actually makes data centre electrical design different — and what it takes to get it right.
Uptime is the whole point
In most facilities, a brief loss of power is an inconvenience. In a data centre, it’s a business-ending event. The entire electrical design is organised around a single principle: the load must never lose power — not during a fault, not during maintenance, not during a utility outage.
That reframes every decision. You’re not just designing a system that works; you’re designing one that keeps working while parts of it are broken, isolated or being serviced. Redundancy stops being a nice-to-have and becomes the governing constraint.
Redundancy changes the architecture
Data centres are designed to redundancy tiers — commonly N+1 or fully 2N — where the network has enough independent capacity and paths that any single component can fail or be taken offline with no impact on the load.
This drives design choices that don’t appear in standard industrial work:
- Multiple independent supply paths feeding dual-corded loads, so power comes from two directions at once.
- Segregated A/B distribution kept physically and electrically separate all the way to the rack.
- UPS and standby generation integrated so transfer is seamless and the model has to account for those sources.
- Concurrent maintainability — the ability to service any part of the electrical system without shutting anything down.
Every one of those decisions has to be verified, not assumed — which is where modelling comes in.
The design has to be proven, not just drawn
Because the redundancy claims are the product, they have to be demonstrable. A data centre electrical design leans heavily on power system modelling: load flow across normal and failure-mode configurations, fault studies for equipment rating and protection, and protection coordination that maintains discrimination across multiple parallel supply paths — a genuinely harder problem than a single radial feed.
Protection in particular is more complex than most sectors realise. With multiple sources and parallel paths, fault current can flow from several directions, and the protection scheme has to stay coordinated in every operating and failure state. We covered this in detail in power system protection for hyperscale data centres.
Scale and speed
Hyperscale developments are large and fast. The electrical design has to be modular and repeatable — a validated building block replicated across halls — and it has to be delivered on programmes that don’t wait. That puts a premium on design packages that are complete, coordinated and construction-ready: accurate single-line diagrams, cable schedules, switchroom layouts and equipment specifications that contractors can build from without constant clarification.
Getting it right
Good data centre electrical design comes from combining two things that don’t always sit together: the discipline of critical-infrastructure redundancy and the modelling rigour to prove it. At JMB Engineering we bring both — the same power system modelling and protection capability we apply to complex mine reticulation, directed at the uptime and redundancy demands of data centres. Our work on the DigiCo SYD1 data centre is a direct example.
Whether it’s a new build or an expansion, the goal is the same: an electrical design that’s not just drawn correctly, but proven to deliver the availability the facility is sold on.
Planning a data centre build or expansion? Contact JMB Engineering to discuss the electrical design.
