{{ result.summaryTitle }}
{{ result.primary }}
{{ result.summaryLine }}
{{ badge.label }}
Enter the usable rack height for this cabinet.
U
Use the rack PDU or branch-circuit voltage, for example 208V, 230V, or 240V.
V
Use the per-feed amp rating before the planning load limit.
A
Choose how rack load should be judged against the available feeds.
Percent of each circuit rating allowed for the plan.
%
Rows use device,count,U each,watts each,priority. Priority accepts critical, high, normal, or low.
Keep one device class per row; quoted CSV names are supported.
Add planned growth, spare slots, or measured peak uplift.
%
100% means every entered watt is counted at the same time.
%
Optional current conversion factor; 1.00 keeps watts and volt-amps equal.
PF
Used only for normal-operation feed current readouts in dual-feed modes.
%
Adjust visible tables and export precision without changing calculations.
Metric Value Readout Copy
{{ row.metric }} {{ row.value }} {{ row.readout }}
Device Count Rack U Watts Each Total kW Share Priority Status Copy
{{ row.device }} {{ row.count }} {{ row.rackU }} {{ row.wattsEach }} {{ row.totalKw }} {{ row.share }} {{ row.priority }} {{ row.status }}
State A Feed B Feed Limit Readout Copy
{{ row.state }} {{ row.aFeed }} {{ row.bFeed }} {{ row.limit }} {{ row.readout }}
Tags: Devops, Storage, Sysadmin
Export to PDF Fullscreen
Customize
Advanced
:

Rack power density expresses how much electrical load is concentrated in a server rack or cabinet. It is usually discussed as kilowatts per rack, watts per rack unit, feed current, and cooling heat load, because the same device inventory affects electrical capacity, rack space, airflow, and cooling at the same time.

The number matters most before equipment is moved, ordered, or powered on. A rack that looks half empty can still be electrically tight if the remaining slots are planned for dense servers, GPU nodes, storage shelves, or dual-cord devices. A rack that fits the branch-circuit rating during normal operation can still fail an A+B redundancy check if one surviving feed cannot carry the full load.

Rack power density path from rack units and device watts through feed limits to headroom.

Cooling follows the same electrical load. Nearly all IT power drawn inside the room becomes heat that the cooling system must remove. A 5 kW rack adds about 17,061 BTU/h of heat, and a 20 kW rack adds about 68,243 BTU/h before room, UPS, lighting, or distribution losses are considered.

Rack power density is a planning signal, not a commissioning result. It does not verify breaker coordination, PDU phase balance, cable ampacity, airflow delivery, inlet temperature, floor loading, or the real load profile of a live workload. Use it to find the racks that deserve a closer electrical and cooling review.

Technical Details:

Rack density combines three related quantities. The device inventory creates a planned watt load, the rack height converts that load into W/U, and the selected feed model determines how much usable kW counts as available capacity. The calculation is most sensitive to device wattage, the planning load limit on each circuit, and the redundancy rule used for the rack.

A+B redundant power is intentionally stricter than dual active capacity. In A+B mode, the rack can have two feeds during normal operation, but the planning capacity is still one feed because the surviving feed must carry the whole rack after a feed loss. Dual active mode adds the two feed capacities together, which can be useful for non-redundant capacity planning but does not prove single-feed failover.

Formula Core:

Raw device load = ∑i=1n(counti×watts eachi) Planning rack load = raw load×simultaneous load factor100×(1+growth allowance100) Per-feed planning limit = circuit rating×planning load limit100 Per-feed usable kW = voltage×per-feed planning limit×power factor1000 Rack density = planning rack load in Wrack units

Allowed rack capacity is one per-feed usable kW value for single-feed and A+B redundant plans. It is two per-feed usable kW values for dual active mode. Spare capacity is allowed rack capacity minus planning rack load. A negative value means the modeled rack is short against the selected feed rule.

Band Lower kW Upper kW Interpretation
Low density 0 5 inclusive Conventional air capacity is usually enough when airflow is managed.
Standard density More than 5 10 inclusive Blanking panels, cable paths, and return-air discipline matter more.
High density More than 10 20 inclusive Containment or row-level cooling review is usually warranted.
Very high density More than 20 40 inclusive Power and cooling design should be reviewed before deployment.
Extreme density More than 40 No fixed upper bound Treat as a specialist high-density or liquid-cooling design case.

The heat-load output uses the same planning rack load and converts watts to BTU/h with 1 W = 3.412142 BTU/h. That conversion is useful for quick cooling conversations, but it does not include room envelope gains, lighting, people, UPS losses, PDU losses, fan energy outside the rack, or any spare cooling unit assumptions.

Input Accepted value Why it matters
Rack size At least 1U, rounded to a whole rack unit Sets W/U density and rack occupancy checks.
Voltage and circuit rating Voltage at least 1 V; circuit rating at least 0.1 A Convert rack watts into current and usable kW per feed.
Feed model Single feed, A+B redundant, or dual active Determines whether one feed or both feeds count toward capacity.
Planning load limit 1% to 100% Derates the circuit rating before headroom is judged.
Device inventory Rows use device, count, U each, watts each, priority; three-column legacy rows are accepted Creates raw device load, occupied U, priority sorting, and device warnings.
Growth and simultaneous load factor Growth 0% to 500%; simultaneous load factor 1% to 150% Scales raw device watts into the planning rack load.
Power factor 0.50 to 1.00 Changes current and usable kW when watts and volt-amps should not be treated as equal.
Normal A-feed share 0% to 100% Splits normal-operation current between A and B feeds in dual-feed modes.

Device rows are rejected when count is not greater than zero, watts each is negative, rack U each is negative, or the inventory is empty. Unknown priority text is treated as normal priority. Display precision changes visible decimals and exports, not the underlying calculations.

Everyday Use & Decision Guide:

Start with measured watts when you have them. Nameplate watts are useful for a conservative first pass, but live PDU readings, vendor power calculators, and recent monitoring data usually make the headroom check more useful. Leave the simultaneous load factor at 100% when you are planning from nameplate or when the devices can peak together.

Use A+B redundant mode for dual-cord equipment that is expected to keep running after one feed is lost. Use dual active mode only when the design intentionally depends on both feeds being available. A rack that passes dual active can still show a failover shortfall in A+B mode because the capacity rule changes.

  • Capacity Ledger is the main planning view. It shows Planning rack load, Selected capacity rule, Spare capacity, Rack density, Rack occupancy, Loaded-U density, Cooling heat load, and Planning notes.
  • Device Load Ledger shows each device class, count, total rack U, watts each, total kW, share of raw load, priority, and row status.
  • Feed Failover Plan compares normal feed current with loss-of-feed cases when A+B mode is selected.
  • Rack Density Profile charts used kW, headroom kW, shortfall kW, and planning limit kW for the selected capacity scenarios.
  • JSON gives a structured record of inputs, totals, density band, device rows, warnings, and errors.

Check the warning count even when the headline status is green. A rack can have spare kW and still be physically wrong if the device inventory exceeds the rack size. Three-column legacy rows are accepted, but they leave rack U at zero, so Loaded-U density and occupancy can look better than the real cabinet will be.

Treat a negative Spare capacity, Failover shortfall, Feed split over limit, or a high density band as a stop-and-review cue. Before adding equipment, compare the device rows with the PDU plan, confirm the circuit load limit for the site, and ask facilities or the colocation provider whether the row cooling design can handle the resulting heat load.

Step-by-Step Guide:

  1. Enter Rack size in rack units, such as 42U or 48U, so Rack density and Rack occupancy use the right denominator.
  2. Enter Nominal voltage and Circuit rating per feed from the rack PDU or branch-circuit plan.
  3. Choose the Feed model. Select A+B redundant when one remaining feed must carry the rack after failover.
  4. Set Planning load limit. The default 80% is a common continuous-load planning value, but site policy and listed equipment ratings should control final use.
  5. Paste the Device inventory with one device class per row using device, count, U each, watts each, and priority.
  6. Use Normalize after pasting mixed CSV rows; the normalized inventory should show one clean row per parsed device class.
  7. Open Advanced only when you need future growth, a simultaneous load factor, power factor, normal A-feed share, or display precision.
  8. Review the summary badge and Spare capacity. If the red input alert appears, fix the listed row error before trusting any table or chart.
  9. Open Feed Failover Plan for A+B racks and verify that the loss-of-feed rows still show spare current on the surviving feed.
  10. Use Rack Density Profile to see which capacity scenario creates headroom or shortfall, then keep the current inputs consistent when comparing alternative inventories.

Interpreting Results:

Read Spare capacity before reading the density band. A low-density rack can still be over capacity if the feed limit is small, and a high-density rack can pass the electrical check if the rack has enough power and cooling design behind it. The selected feed model controls that judgment.

Rack density uses the full rack height. Loaded-U density uses only device rows with rack U values, so it rises when a small number of occupied units carry most of the load. If device rows omit U values, the loaded-U number can be misleading and the planning notes will call that out.

Status or cue Meaning Practical response
Failover-safe plan A+B mode has spare kW against one-feed capacity and normal feed current is within the planning limit. Still verify phase balance, actual PDU readings, airflow, and row cooling capacity.
Within feed plan The selected non-redundant capacity rule has spare kW. Do not read this as A+B redundancy unless A+B mode was selected.
Tight headroom Utilization is at least 90% of the modeled capacity. Recheck watts, growth allowance, and site derating before adding load.
Feed split over limit Normal-operation current exceeds the per-feed planning limit on at least one feed. Adjust load sharing, reduce rack load, or review PDU/circuit capacity.
Failover shortfall or Over capacity Planning rack load is greater than the allowed rack capacity for the selected feed model. Do not deploy the plan without reducing load, changing capacity, or changing the design assumption.

Cooling heat load is useful for facilities discussion because it translates IT electrical load into BTU/h. It is not a complete cooling design. Confirm inlet temperatures, containment, airflow direction, and cooling unit redundancy before treating a dense rack as ready.

Worked Examples:

Default A+B rack with mixed equipment:

The sample inventory has eight 420 W compute nodes, two 180 W top-of-rack switches, one 650 W storage shelf, and two 95 W firewalls. With a 42U rack, 230 V, 32 A per feed, A+B redundant mode, and an 80% planning load limit, Planning rack load is 4.56 kW. Selected capacity rule is 5.89 kW, Spare capacity is 1.33 kW, and Rack density is 108.6 W/U. The failover rows show about 5.77 A spare on the surviving feed, so the plan passes this calculator's A+B check.

GPU-heavy rack that fails failover:

Four 6U GPU nodes at 3,000 W each plus two 250 W switches create a 12.50 kW raw load. On the same 230 V, 32 A, 80% A+B plan, allowed rack capacity remains 5.89 kW, so Spare capacity becomes -6.61 kW and the status becomes Failover shortfall. Rack density lands at 297.6 W/U and the density band is High density. The result means both power and cooling design need review before the rack is treated as deployable.

Growth allowance with measured diversity:

The sample rack changes from 4.56 kW to 4.85 kW when Future growth allowance is 25% and Simultaneous load factor is 85%. The calculation warns that both factors are applied, Spare capacity falls to about 1.04 kW, and Cooling heat load rises to roughly 16,531 BTU/h. That is still electrically within the default A+B plan, but there is much less room for another device class.

Physically full rack with spare kW:

Fifty 1U microservers at 80 W each draw only 4.00 kW, so the default A+B capacity check still has about 1.89 kW spare. The same inventory occupies 50U in a 42U rack, so Planning notes reports that device height exceeds rack size by 8.0U. This is a useful false-confidence case: the electrical result is acceptable, but the cabinet plan is not.

These examples show why the headline kW is not enough. The feed rule, rack U, warnings, and heat output decide whether the inventory is only numerically tidy or actually ready for a deployment conversation.

FAQ:

Why does A+B mode use only one feed for capacity?

A+B redundant equipment is expected to survive the loss of one feed. The calculator therefore checks whether one remaining feed can carry the whole planning rack load after the planning load limit is applied.

Should I enter nameplate watts or measured watts?

Use measured or vendor-modeled watts when you have reliable data. Use nameplate watts for a conservative first pass, then keep the simultaneous load factor at 100% unless you have evidence that the devices will not peak together.

Why is the default planning load limit 80%?

The default matches a common continuous-load planning practice where a circuit is not planned to carry its full rating for sustained load. Final values should follow the site's electrical design, equipment listing, and local code review.

What does a high rack density band mean?

It means the planned kW in the rack is high enough to deserve cooling and airflow review. The band does not prove that the rack will overheat, and it does not replace inlet-temperature measurements or facility cooling checks.

Why did the calculator accept a three-column device row?

Legacy rows with device, count, and watts each are accepted. They leave rack U at zero, so the row status and planning notes should be checked before using occupancy or loaded-U density.

Are my device rows sent to a server for calculation?

The entered rack inventory is calculated in the browser. Be careful with copied rows, downloaded files, JSON records, chart images, and shared URLs because they can keep the values you entered.

Glossary:

Rack unit
A vertical rack measurement equal to 1.75 inches; rack size is entered in U.
Planning rack load
Raw device watts after the simultaneous load factor and growth allowance are applied.
Feed model
The rule used to judge available capacity: single feed, A+B redundant, or dual active.
A+B redundant
A dual-feed design where one surviving feed must be able to carry the rack after the other feed is lost.
Planning load limit
The percentage of each feed rating allowed for the plan before headroom is judged.
Power factor
The factor used to convert rack watts into current and usable feed kW when watts and volt-amps differ.
Loaded-U density
Planning rack load divided by occupied rack units from device rows that include U values.
Heat load
The cooling load associated with the planned IT electrical watts, shown in BTU/h.

References: