DATACENTER TWIN LAB

Infrastructure Engineering · Lesson 24 of 24 · 40 minutes

Audit the correction before releasing the design

The final lesson asks for a justified release decision, not the largest possible green score. Bring capacity, connectivity, protection, thermal behaviour, controls and evidence together. Then audit a proposed correction from the manuscript’s separate campus example: a remedy is another engineering assertion and must be recalculated.

Run this lesson interactively

Learning objectives

Source statement

A release condition is an intersection

The twenty-megawatt case uses one consistent load ledger across utility, transformers, generator banks, UPS, batteries, distribution, cooling and station DC. A failed necessary condition is sufficient to reject the corresponding design claim. Passing every simplified screen still does not replace the full project studies. The source generator-bank claim fails, while battery sizing, distribution studies and cooling demonstration remain open. Proposed PV has no established firm contingency credit.

Editorial derivation

Recalculate a proposed campus remedy

The separate campus review assumes 560 MVA demand per independently full-capacity side. Three 250 MVA transformers leave only 500 MVA after one loss, so the suggested N+1 remedy fails. Four units are required by the elementary count. The same 250 MVA transformer at 22 kV has approximately 6,560.80 A full-load current; a 6,300 A incomer, or two nominal 3,150 A incomers, does not reach that total. A larger replacement number is not enough unless it meets the declared duty.

Unresolved

Close only what the evidence closes

A high current can expose unsupported equipment feasibility without proving universal physical impossibility. Preserve source-qualified identities: S01:SLD-002 is not interchangeable with the design-series DCD-002. The manuscript describes twelve reference families, but its accompanying 525-entry atlas and original corpus were not supplied with this task. The capstone can teach selection criteria and namespace discipline; it cannot claim to expose that missing archive.

Predict → run → explain

  1. Read the declared boundary and units. Predict the selected result before calculating.
  2. Run the starting case and inspect the ledger, assumptions and model limits.
  3. Apply the challenge, run again, and explain the difference. Export only the last completed inputs and results.

Available MVA per side = (units per side − unavailable) × unit MVA; full-load A = unit MVA × 1,000/(√3 × kV); margin A = incomer A − full-load A. Capacity and current are independent necessary screens; design release remains HOLD.

Declared starting inputs

{
  "demandMVA": 560,
  "unitMVA": 250,
  "unitsPerSide": 3,
  "unavailable": 1,
  "voltageKV": 22,
  "incomerA": 6300
}
Worked default — unrounded educational outputs
QuantityValueUnit
availableMVA500MVA
marginMVA-60MVA
minimumNPlusOne4units
fullLoadA6560.7985135184745A
incomerMarginA-260.79851351847446A
capacityStatusscreen-failedscreen state
currentStatusscreen-failedscreen state

Challenge

Does correcting the transformer count close the whole design?

{
  "unitsPerSide": 4
}

Surviving capacity becomes 750 MVA, but the 6,300 A incomer still falls below approximately 6,560.80 A. Other required studies remain unresolved, so the design-release outcome remains HOLD.

Worked challenge
QuantityValueUnit
availableMVA750MVA
marginMVA190MVA
minimumNPlusOne4units
fullLoadA6560.7985135184745A
incomerMarginA-260.79851351847446A
capacityStatusscreen-satisfiedscreen state
currentStatusscreen-failedscreen state

The evidence-controlled release gate — static equivalent

  1. Open the twenty-megawatt review matrix without mixing it with the separate campus case.
  2. Switch to the campus remedy, remove one transformer and expose the remaining shortfall.
  3. Recalculate a changed unit count while keeping the independent current issue visible.
  4. Assemble six explicit gate states; unresolved or failed evidence leads to HOLD.

Six-condition acceptance matrix with blocking reasons and a source-qualified case register.

Choose pass/fail/unresolved for six justified criteria; require a reason and source locator before exporting the review.

Narrative steps are not physical seconds. Essential states and quantities remain available without animation.

Assessment and answer criteria

12.1 · Find the elementary N+1 count per side for 560 MVA using 200 MVA units.

Read answer and marking criterion

Four installed per side; three remain with 600 MVA.

Criterion: Do not combine independent sides to satisfy one side’s duty.

Source: DCIE-R1 Appendix A, PDF pp.59-61

12.2 · Find the nominal full-load current screen for 100 MVA at 33 kV.

Read answer and marking criterion

Approximately 1,749.55 A before site and operating allowances.

Criterion: Use total three-phase apparent power and line-to-line voltage.

Source: DCIE-R1 Appendix A, PDF pp.59-61

12.3 · Why does a corrected equipment count not close a Tier or fault-tolerance claim?

Read answer and marking criterion

Connectivity, protection, dynamics, cooling, controls, maintenance and the applicable assessment criteria still need evidence.

Criterion: Acknowledge all relevant boundaries; do not award certification.

Source: DCIE-R1 Appendix A, PDF pp.59-61

Source and model boundaries

Data Centre Infrastructure Engineering, reference manuscript revision 1 (DCIE-R1), 11 October 2026, chapter 12, PDF pp.54–58. PDF page 1 includes the cover. manuscript-derived; not independent source verification.

Original educational adaptation. The private PDF and missing archive are not public downloads. Source statement, Editorial derivation, Illustrative assumption and Unresolved remain distinct.

Result schema: infrastructure-screen-1; content 2.0.0; model 1.0.0. Raw inputs, outputs, units, assumptions, source locators and canonical hashes. No supported in-browser Python verification protocol exists for this result family.