Standards & compliance

Uptime Tier compliance: what each level actually requires

Tier is not a rating you award yourself for having two of something. It is a set of requirements a topology either meets or does not, assessed at the weakest point.

The ladder is cumulative, and it is graded at its weakest point

Each level carries every requirement of the levels below it. A facility with fault-tolerant electrical distribution and a single cooling path is not “Tier IV on power” — it is held at whatever level its cooling satisfies. One unmet Tier I requirement caps the whole assessment at Tier I, however good everything above it is.

This is the single most common misreading. Tier is not an average, a score, or a per-system badge. It is the highest level at which every requirement is met.

What each level adds

The two sweeps, and why they are not the same test

Tier III and Tier IV are both assessed by removing things one at a time. What separates them is what you are allowed to assume while you do it.

The Tier III sweep is planned

Maintenance is scheduled. An operator is present, procedures are written, and a tie breaker may be closed before the work starts. A manually operated tie is a legitimate part of a Tier III scheme, because somebody will close it deliberately as a step in the method statement.

The Tier IV sweep is not

A fault does not wait for an operator. Only responses that happen on their own count — automatic transfer, automatic tie closure, static switching. Draw a manual tie and it earns you Tier III credit and no Tier IV credit, correctly.

The six checks designs most often fail

1. Cooling is not in the sweep

Power gets tested; cooling gets counted. A facility with N+2 cooling on the schedule can still fail concurrent maintainability if the units are distributed so that any one board carries more than the spare capacity covers. Take out each board and re-add the cooling that is still powered, against the IT load.

2. Generators are Standby rated

ISO 8528 Standby rating limits a set to roughly 200 hours a year with no sustained overload capability. Utility loss is treated as a normal operating condition at Tier III and above, so the generators are the availability source, not an emergency backstop. Prime or Continuous rating is what the requirement implies.

3. Ride-through is assumed rather than traced

Every cabinet has to be fed through something that holds the load up during transfer. In a telecom facility much of that path is a battery-backed DC plant rather than an AC UPS, and it counts — but only if the battery is actually there and connected. A rectifier with no battery is not stored energy.

4. Ties are drawn open and counted as closed

An open tie carries nothing. It earns credit for a path the facility is not running on, and only under a mode of operation where somebody closes it. Draw the tie’s position and its control — manual or automatic — and let the assessment use them honestly.

5. Continuous cooling is missing at Tier IV

Tier IV requires cooling that rides through the transfer, which means CRAH and CRAC units on UPS-backed power, or thermal storage sized for the gap. Ceiling-level temperatures in a dense hall can reach critical within about a minute of airflow stopping. Chillers restarting on generator in forty seconds is not fast enough on its own.

6. Compartmentalisation is never checked

Tier IV requires the complementary A and B systems to be physically separated, so one event cannot take both. This one cannot be read off a single line diagram at all — it is a floor plan question. Verify that no room holds more than the redundant number of like components, and that A and B distribution never share a space.

Where ANSI/BICSI 002 differs

BICSI 002 assesses Availability Class F0 to F4 across electrical, mechanical, telecommunications and fire protection, and the weakest subsystem sets the class. It reaches wider than the Uptime Tier Standard, which is deliberately a topology standard.

The two broadly line up — Class F3 sits alongside Tier III — but they are not interchangeable, and a design can satisfy one while missing the other. On a tender that names both, assess both.

What a defensible assessment looks like

1. Trace the topology from utility to load, rather than reading redundancy off a schedule.
2. Run the planned sweep across every capacity component and every distribution path, power and cooling together.
3. Run the automatic sweep separately, with consequential impact, for any Tier IV claim.
4. Record the mode of operation assumed for every tie and transfer switch.
5. Cite the clause behind each finding, so the verdict can be argued rather than believed.
6. State plainly which checks are not evaluable from the drawings and must be verified on the floor.