DCEP common mistakes
DCEP · Understand the traps and how to avoid them.
Computing PUE as IT load divided by total facility load, or using the inverse ratio, when asked to report a PUE value.
Candidates memorize the abbreviation 'Power Usage Effectiveness' and reason from the word order rather than the definition; PUE is total facility energy divided by IT equipment energy, always greater than or equal to 1.
What to do instead
Write the ratio explicitly before calculating: PUE = total facility power / IT power. Sanity-check that the result is at least 1.0; if it is below 1, the ratio is inverted.
Treating a single instantaneous PUE reading as the annualized or reported PUE for a facility.
Spot readings are easy to obtain from a BMS dashboard, and candidates overlook that PUE varies strongly with outdoor conditions and IT load, so a snapshot misrepresents yearly performance.
What to do instead
For reporting or benchmarking, use time-weighted energy totals (kWh) over the full period, or clearly label the value as instantaneous. Confirm the measurement interval and averaging method before answering.
Excluding cooling, UPS losses, or lighting from the total facility energy when calculating PUE, counting only the IT load and chillers.
Candidates picture 'the data center' as servers plus cooling and forget that the definition of total facility energy includes power distribution losses, UPS inefficiency, and all supporting infrastructure.
What to do instead
Enumerate every energy stream on the facility side before summing: IT, cooling plant, fans, UPS/PDU losses, lighting, and auxiliary loads. If a stream is metered separately, add it in.
Assuming a higher UPS or PDU efficiency rating at full load applies at the actual part-load operating point.
Efficiency curves are often quoted at or near 100% load, but data center UPS units frequently run at 30-50% load where efficiency and losses differ significantly.
What to do instead
Use the manufacturer's part-load efficiency curve and the actual load fraction; compute losses as input minus output at that operating point rather than applying a nameplate percentage.
Calculating airflow (CFM) from heat load without converting units consistently, mixing kW, BTU/hr, and tons of cooling.
The sensible heat airflow equation appears in several unit systems, and candidates plug kW into a formula expecting BTU/hr, producing answers off by a factor of roughly 3,412.
What to do instead
Convert all heat loads to a single unit (e.g., BTU/hr) before applying CFM = sensible heat / (1.08 x delta-T), and state the conversion factor used.
Confusing return air temperature with supply air temperature, or using the wrong delta-T, in cooling capacity and airflow calculations.
Racks are described by return temperature while CRAC units are set by supply temperature, and candidates grab whichever number appears first in the prompt.
What to do instead
Label each temperature explicitly as supply or return before computing delta-T, and use the difference across the coil or the rack as the problem specifies.
Ignoring bypass airflow and recirculation when assessing whether rack inlet temperatures meet the recommended envelope.
Candidates assume CRAC supply temperature equals rack inlet temperature, but bypass air and hot-air recirculation create a gap that is the core of airflow management analysis.
What to do instead
Account for bypass fraction and return temperature; use the rack inlet temperature, not the CRAC discharge, when evaluating compliance with inlet temperature guidelines.
Recommending containment or airflow changes without first checking that the cooling plant has capacity, treating airflow fixes as free capacity.
Airflow improvements raise return air temperature and can reduce cooling energy, but candidates jump to containment as a universal fix without verifying the chilled water or DX capacity.
What to do instead
Sequence the analysis: verify cooling capacity and delta-T first, then evaluate airflow management. State assumptions about plant capacity before recommending containment.
Treating nameplate server power as the actual IT load when sizing power or computing efficiency metrics.
Nameplate ratings are conservative maxima, and candidates use them for convenience, overstating load and skewing PUE, UPS loading, and capacity calculations.
What to do instead
Use measured or metered power where provided; if only nameplate data is available, apply a realistic utilization factor and state it as an assumption.
Comparing a facility's PUE against a benchmark without matching the measurement boundary, period, or climate.
Published PUE benchmarks come from different climates, IT loads, and measurement conventions, and candidates compare raw numbers as if they were interchangeable.
What to do instead
Before benchmarking, confirm the boundary (does it include the same loads?), the averaging period, and the climate/load context; normalize or caveat the comparison in the answer.