What determines Uptime Institute Tier classification?

Tier classification is a prescriptive topology check, not a calculation. The weakest subsystem defines the tier. Tier I = single non-redundant path. Tier II = redundant capacity components but single distribution path. Tier III = multiple independent distribution paths (concurrently maintainable; only one path active at a time). Tier IV = active/active redundant paths — a single fault anywhere cannot cause downtime (fault tolerant).

How does the Tier Advisor score power infrastructure?

The power subsystem score is a weighted sum: utilityFeeds (20%), genConfig (25%), upsConfig (20%), upsTopo (10%), atsConfig (10%), pduRedundancy (10%), fuelAutonomy (5%). Each input maps to a 0-100 raw score via a lookup table. For example, 2N UPS = 80, 2(N+1) = 100, N+1 = 50. Fuel autonomy uses a continuous ramp: 0-8 hrs maps 0-30, 8-24 hrs maps 30-60, 24-72 hrs maps 60-85, above 72 hrs asymptotes to 100.

What are the floor constraints that prevent over-classification?

Even if the composite score reaches 75+, Tier III requires: utility feeds must be dual (dual_same, dual_diverse, or on-site generation), generator N+1 or better, UPS N+1 or better, and cooling distribution must not be single loop. Tier IV additionally requires: dual diverse utility feeds from separate substations, 2N/2(N+1) generators, 2N/2(N+1) UPS, dual or triple PDU cords, 2N/2(N+1) cooling redundancy, and dual diverse or 3+ network entry paths. A composite score is capped at 74 if Tier III prerequisites are unmet, and capped at 89 if Tier IV prerequisites are unmet.

Tier Advisor — Technical Manual & Methodology

01 Purpose & engineering basis

The Tier Advisor maps a data center's actual infrastructure topology against four concurrent classification frameworks: Uptime Institute Tier Standard (Tier I–IV), ANSI/TIA-942-B (Rated 1–4), EN 50600-2-1 (Availability Class VK1–VK4), and BICSI 002 (Class F1–F4). It answers three operational questions: what tier does the current infrastructure support, which subsystem is the limiting factor, and what specific changes are required to reach the next tier.

Classification is user-driven and prescriptive: you supply the infrastructure topology; the engine applies the published requirement tables. There are no back-solved coefficients and no probabilistic sampling in the classification path — the tier result is deterministic from inputs. The governing authority for topology requirements is the Uptime Institute Tier Standard (2022), cross-referenced to ANSI/TIA-942-B (2017) and EN 50600-2-1:2021. Regional compliance overlays reference jurisdiction-specific frameworks (MAS TRM, EU DORA, FISC, APRA CPS 234, OJK/ESDM, BNM RMiT, and others) for the 12 supported regions.

Prescriptive, not probabilistic: Uptime Institute Tier is a binary topology check — a facility either satisfies the structural requirements for a given tier or it does not. The scoring model (0–100) is a continuous proxy that measures how completely inputs satisfy those requirements, providing nuanced guidance for partial compliance. The hard floor constraints enforce the binary pass/fail logic of the published standards on top of the continuous score.

02 Inputs

Twenty inputs across six groups — facility profile, power, cooling, network, physical security, and optional regional compliance (PRO). All inputs are enumerated categorical or continuous numeric values; ranges are enforced client-side.

Field (DOM id)	Type	Options / range	Group	Meaning
`facilityType`	enum	enterprise / colocation / hyperscale / edge / telecom / financial	Facility	Facility class — used in narrative generation and PDF export; does not affect the tier score directly.
`itLoad`	number	50–500 000 kW	Facility	Total IT load in kW — used in cost estimate scaling and narrative only.
`numFloors`	enum	1 / 2 / 3 / 4+	Facility	Building floor count — narrative context only.
`buildingAge`	number	0–50 years	Facility	Age in years — narrative context only; no score impact.
`utilityFeeds`	enum	single / dual_same / dual_diverse / onsite	Power	Number and diversity of utility supply paths. Tier III floor requires dual; Tier IV floor requires dual_diverse.
`genConfig`	enum	none / n / n1 / 2n / 2n1	Power	Generator redundancy topology. Tier III floor: N+1 minimum; Tier IV floor: 2N or 2(N+1).
`upsConfig`	enum	none / n / n1 / 2n / 2n1 / distributed	Power	UPS redundancy topology. Same floor thresholds as generators. Distributed = scored at 90.
`upsTopo`	enum	standby / line_interactive / double_conversion / rotary	Power	UPS conversion topology. Double-conversion (online) required for mission-critical; scores 90.
`atsConfig`	enum	none / single / dual / sts	Power	Transfer switch type. STS (static transfer switch) provides sub-cycle switching for IT loads.
`pduRedundancy`	enum	single / dual / triple	Power	PDU cord configuration at rack level. Tier IV floor requires dual or triple.
`fuelAutonomy`	number	0–720 hours	Power	Generator fuel storage in hours at rated load. Continuous ramp function (see §03).
`coolRedundancy`	enum	n / n1 / n2 / 2n / 2n1	Cooling	Cooling plant redundancy. Tier III floor: N+1 minimum implied; Tier IV floor: 2N or 2(N+1).
`coolDistribution`	enum	single / dual / n1_piping	Cooling	Cooling distribution loop count. Tier III floor: not single. N+1 piping scores highest.
`coolType`	enum	dx / chilled_water / free_cooling / rdhx / dlc / immersion	Cooling	Cooling technology. Higher sophistication (DLC, immersion) scores higher — reflects operational resilience.
`netEntry`	enum	single / dual_same / dual_diverse / three_plus	Network	Network entry path count and conduit diversity. Tier IV floor: dual_diverse or three_plus.
`carrierDiv`	enum	single / two / three_plus	Network	Number of distinct telecommunications carriers.
`meetMeRoom`	enum	none / single / redundant	Network	Meet-me room (MMR) presence and redundancy.
`fireSuppression`	enum	none / wet / preaction / clean / vesda_clean	Physical	Fire suppression type. VESDA + clean agent = highest detection latency + suppression effectiveness.
`accessControl`	enum	key / card / biometric / mfa_mantrap	Physical	Physical access control layer. MFA + mantrap = maximum depth of authentication.
`monitoring`	enum	none / basic / bms / full_dcim	Physical	Monitoring capability. Full DCIM + BMS = comprehensive telemetry, alarming, and capacity planning.

Regional compliance inputs (PRO): regionSelect (12 jurisdictions) and verticalSelect (5 industry verticals) are evaluated only in PRO mode and contribute to the regional score (weight 15%) via the calcRegional() function documented in §03.

03 Calculation methodology

The engine runs five sequential steps: (1) raw subsystem score computation, (2) weighted composite aggregation, (3) floor constraint enforcement, (4) tier and grade assignment, and (5) confidence computation. All function names map one-to-one to inline JS in tier-advisor.html.

Step 1 — Subsystem scores (in `calculate()`)

powerScore = gs('utilityFeeds')·0.20 + gs('genConfig')·0.25 + gs('upsConfig')·0.20 + gs('upsTopo')·0.10 + gs('atsConfig')·0.10 + gs('pduRedundancy')·0.10 + fuelScore(fuelAutonomy)·0.05 coolingScore = gs('coolRedundancy')·0.50 + gs('coolDistribution')·0.30 + gs('coolType')·0.20 networkScore = gs('netEntry')·0.45 + gs('carrierDiv')·0.30 + gs('meetMeRoom')·0.25 physicalScore = gs('fireSuppression')·0.35 + gs('accessControl')·0.35 + gs('monitoring')·0.30 monitorScore = SCORE_MAPS.monitoring[monitoring] gs(id) looks up the enum value of input id in SCORE_MAPS (see §04 constants table). fuelScore(h) is a piecewise ramp (see below). Monitoring is double-counted: as a component inside physicalScore and as a standalone subsystem — this emphasises its importance to overall operations confidence. TIER ADVISOR · ENGINE

Fuel autonomy ramp — `fuelScore(h)`

h ≤ 8 → score = h / 8 · 30 (0–30, fast initial ramp) h ≤ 24 → score = 30 + (h − 8) / 16 · 30 (30–60, moderate ramp) h ≤ 72 → score = 60 + (h − 24) / 48 · 25 (60–85, slow ramp) h > 72 → score = min(100, 85 + (h − 72) / 648 · 15) (85–100, asymptote) Reflects diminishing returns beyond 72 hours. The Uptime Institute does not specify a minimum autonomy for Tier III/IV; FISC (Japan) requires 72 hours; EU DORA guidance references 72 hours. The ramp captures this industry consensus. In fuelScore(). UPTIME · FISC · ENGINE

Step 2 — Weighted composite (in `calculate()`)

WEIGHTS = { power:0.35, cooling:0.20, network:0.15, physical:0.10, monitoring:0.05, regional:0.15 } compositeRaw = powerScore·0.35 + coolingScore·0.20 + networkScore·0.15 + physicalScore·0.10 + monitorScore·0.05 + regionalScore·0.15 Power carries the largest weight (35%) reflecting that electrical failure is the leading cause of data center outages (Uptime Institute Annual Outage Analysis). regionalScore defaults to 50 (neutral baseline) in free mode; in PRO mode it is computed by calcRegional(). UPTIME OUTAGE ANALYSIS · ENGINE

Step 3 — Floor constraints (prescriptive topology checks)

canT3 = (utilityFeeds ∈ {dual_same, dual_diverse, onsite}) AND (genConfig ∈ {n1, 2n, 2n1}) AND (upsConfig ∈ {n1, 2n, 2n1, distributed}) AND (coolDistribution ≠ single) canT4 = (utilityFeeds = dual_diverse) AND (genConfig ∈ {2n, 2n1}) AND (upsConfig ∈ {2n, 2n1}) AND (pduRedundancy ∈ {dual, triple}) AND (coolRedundancy ∈ {2n, 2n1}) AND (netEntry ∈ {dual_diverse, three_plus}) if compositeRaw ≥ 75 AND NOT canT3 → compositeRaw = min(compositeRaw, 74) if compositeRaw ≥ 90 AND NOT canT4 → compositeRaw = min(compositeRaw, 89) composite = round(compositeRaw) The floor constraints implement the binary "must have" topology requirements from TIA-942-B §6 and the Uptime Institute Tier Standard §4. No amount of high scores in monitoring or fire suppression can substitute for missing power or cooling path redundancy. In calculate(). TIA-942-B §6 · UPTIME TIER STD §4 · ENGINE

Step 4 — Tier and grade assignment

composite ≥ 90 → Tier IV · grade A+ · color #10b981 · "Fault Tolerant" composite ≥ 75 → Tier III · grade A · color #3b82f6 · "Concurrently Maintainable" composite ≥ 60 → Tier II+ · grade B · color #f59e0b · "Redundant Components+" composite ≥ 45 → Tier II · grade C · color #f59e0b · "Redundant Components" composite ≥ 25 → Tier I+ · grade D · color #ef4444 · "Basic Capacity+" composite < 25 → Tier I · grade F · color #ef4444 · "Basic Capacity" Framework mappings: uptimeTier = tier (strip "+") | tia942 = "Rated " + tierNum en50600 = "Class " + tierNum | bicsi = "Class F" + tierNum Tier II+ and Tier I+ are non-standard sub-classifications used by the tool to represent facilities that exceed the named tier's minimum but fall short of the next full tier's requirements. The official framework output strips the "+" suffix when reporting to Uptime Institute / TIA-942 / EN 50600 / BICSI. UPTIME TIER STD · TIA-942-B · EN 50600 · BICSI 002 · ENGINE

Step 5 — Confidence score

avg = mean([powerScore, coolingScore, networkScore, physicalScore, monitorScore]) variance = mean([(s − avg)² for each subsystem score s]) confidence = max(60, min(98, round(95 − sqrt(variance) · 0.5))) [%] Confidence measures internal consistency: a facility where all subsystems score similarly to the composite carries high confidence; one with extreme outliers (e.g. excellent power but critically deficient network) carries lower confidence. In calculate(). ENGINE

Regional compliance overlay — `calcRegional()` [PRO]

regionalScore starts at 50 (neutral baseline) Each satisfied jurisdiction-specific requirement adds a bonus (typically +8 to +15 pts) regionalScore = min(100, 50 + sum of passed-check bonuses) Jurisdiction-specific checks per region (example — Singapore MAS TRM): 2N UPS (+15), diverse fiber paths (+10), ≥48 hr fuel (+10), MFA access (+5), full DCIM (+10). Each jurisdiction has 4–5 checks drawn from the published regulatory guidance. See §04 regional table for the full check matrix. MAS TRM · DORA · FISC · APRA CPS 234 · OJK/ESDM · ENGINE

Gap analysis — `buildGapAnalysis()`

Gap analysis generates a specific action list by inspecting current input values against the next tier's floor requirements. For Tier I/II → Tier III gaps it checks: utilityFeeds, genConfig, upsConfig, coolRedundancy, coolDistribution, netEntry, atsConfig. For Tier III → Tier IV gaps it checks the stricter 2N constraints on all the same dimensions plus pduRedundancy and atsConfig type (STS or dual).

Recommendations — `buildRecommendations()`

All five subsystem scores are sorted ascending. The three lowest-scoring subsystems with scores below 85 are flagged. Priority is set by score bracket: <40 = P1-Critical, 40–60 = P2-High, 60–85 = P3-Medium. Each recommendation card includes a subsystem-specific upgrade action description and priority badge.

04 Score maps, constants & sources

Every raw score lives in the deep-frozen SCORE_MAPS object. The table below is the complete lookup — no hidden values exist outside this object.

Input field	Value	Raw score (0–100)	Rationale
`utilityFeeds`	`single`	10	Single point of failure at utility level — Tier I topology.
	`dual_same`	40	Two feeds from same substation: protects against cable fault, not substation failure.
	`dual_diverse`	80	Feeds from separate substations: single-source grid fault is isolated. Required for Tier IV.
	`onsite`	60	Self-generation eliminates utility dependency but adds fuel-chain risk.
`genConfig`	`none`	0	No backup generation — Tier I (no redundancy).
	`n`	30	Exactly enough capacity — zero margin for maintenance or fault.
	`n1`	60	One redundant unit — concurrent maintenance possible. Tier III floor minimum.
	`2n`	85	Full dual bus — single generator fault does not interrupt IT load. Tier IV floor.
	`2n1`	100	Dual bus plus spare — highest resilience; tolerates simultaneous maintenance and fault.
`upsConfig`	`none`	0	No UPS — no ride-through during utility-to-generator transfer.
	`n`	20	Exact capacity — maintenance requires outage.
	`n1`	50	One redundant module — concurrent maintenance; Tier III floor minimum.
	`2n`	80	Dual independent UPS systems — active/standby redundancy; Tier IV floor.
	`2n1`	100	Dual systems plus spare module — maximum tolerance.
	`distributed`	90	Distributed redundant UPS (e.g. modular delta conversion) — equivalent to 2N in reliability.
`upsTopo`	`standby`	20	Off-line UPS — brief transfer gap on utility failure; not suitable for Tier III+.
	`line_interactive`	40	Minimal conditioning; transfer time still measurable.
	`double_conversion`	90	Online double-conversion — continuous power conditioning, zero transfer time. Industry standard for DC.
	`rotary`	80	Rotary UPS/DRUPS — flywheel ride-through, no chemical batteries, good resilience.
`atsConfig`	`none`	0	Manual switchover — extended outage during source change.
	`single`	40	Single ATS — automatic but single point of failure in transfer path.
	`dual`	75	Dual ATS — each bus has its own transfer switch; ATS failure does not collapse both buses.
	`sts`	100	Static Transfer Switch — sub-cycle (<4 ms) switching; no IT load interruption during source fault.
`pduRedundancy`	`single`	20	Single-cord feed to rack — PDU failure = rack outage. Not acceptable at Tier IV.
	`dual`	70	A+B dual-cord — PDU or path failure does not interrupt dual-corded servers. Tier IV floor minimum.
	`triple`	100	Triple-corded distribution — supports highest-criticality loads with three independent paths.
`coolRedundancy`	`n`	15	No cooling redundancy — any CRAC/CRAH failure causes thermal runaway. Tier I.
	`n1`	50	One redundant cooling unit — concurrent maintenance possible. Tier III floor implied.
	`n2`	70	Two redundant units — higher resilience margin but not fully concurrent on all paths.
	`2n`	90	Dual cooling systems — full active/standby; Tier IV floor minimum.
	`2n1`	100	Dual systems plus spare — maximum cooling resilience.
`coolDistribution`	`single`	20	Single pipe loop — distribution fault is single point of failure. Tier III floor: must not be single.
	`dual`	70	Dual cooling loops — loop maintenance or fault isolates to one bus.
	`n1_piping`	100	N+1 pipe segments with isolation valves — highest distribution resilience.
`coolType`	`dx`	55	Direct expansion — simple but limited scalability and efficiency.
	`chilled_water`	65	Central chilled water plant — industry standard for medium–large DC.
	`free_cooling`	75	Economiser / adiabatic — reduced mechanical complexity in partial load conditions.
	`rdhx`	80	Rear-door heat exchanger — close-coupled; removes heat at source.
	`dlc`	90	Direct liquid cooling — very high density; eliminates raised-floor airflow dependency.
	`immersion`	100	Immersion cooling — no airside cooling path; maximum heat transfer efficiency.
`netEntry`	`single`	15	Single conduit — physical cable cut = total network loss.
	`dual_same`	40	Two cables entering same conduit or manhole — simultaneous backhoe vulnerability remains.
	`dual_diverse`	80	Separate conduit routes entering at separate physical locations. Tier IV floor minimum.
	`three_plus`	100	Three or more diverse entry paths — maximum network resiliency; common in carrier-neutral facilities.
`carrierDiv`	`single`	20	Single carrier — carrier peering or backbone issue affects entire facility.
	`two`	60	Two carriers — commercial standard for Tier III; protects against single-carrier outage.
	`three_plus`	100	Three or more carriers — highest provider diversity; carrier failure is isolated.
`meetMeRoom`	`none`	0	No dedicated MMR — cross-connects routed through data hall.
	`single`	50	Dedicated meet-me room — carrier cross-connects isolated from data hall.
	`redundant`	100	Redundant MMRs at separate locations — MMR fire or flood does not affect all carriers.
`fireSuppression`	`none`	0	No suppression — regulatory non-compliance in most jurisdictions.
	`wet`	50	Wet pipe — fast response but water damage risk to IT equipment.
	`preaction`	65	Pre-action (double interlock) — requires fire + flow signal before water release; reduces false activation.
	`clean`	85	Clean agent (FM-200 / Novec 1230) — no residue, no water damage, occupant-safe at design concentration.
	`vesda_clean`	100	VESDA (Very Early Smoke Detection) + clean agent — earliest possible detection + safest suppression.
`accessControl`	`key`	15	Physical key — no audit trail, no remote revocation.
	`card`	40	Card access — audit trail present but single factor (lost/stolen card = breach).
	`biometric`	75	Biometric (fingerprint/iris) — cannot be shared or stolen, but single layer.
	`mfa_mantrap`	100	Multi-factor (card + bio) with mantrap — maximum depth; TIA-942-B Rated 4 expectation.
`monitoring`	`none`	0	No monitoring — no visibility into thermal or power events.
	`basic`	30	Basic temperature/humidity sensors — minimal instrumentation.
	`bms`	65	Building Management System — plant-level monitoring and alarming; industry standard for Tier III.
	`full_dcim`	100	Full DCIM + BMS — capacity management, predictive analytics, asset tracking; Tier IV expectation.

Per-tier requirements summary (Uptime Institute Tier Standard + TIA-942-B convergence):

Tier	Availability target	Annual downtime allowance	Power redundancy	Cooling redundancy	Network	Maintainability
Tier I	≥ 99.671%	28.8 hrs/yr	N (non-redundant)	N	Single path	Outage required for maintenance
Tier II	≥ 99.741%	22.0 hrs/yr	N+1 components, single distribution path	N+1 components	Single path	Planned maintenance requires outage
Tier III	≥ 99.982%	1.6 hrs/yr	N+1 generators + UPS, dual distribution paths	N+1 with dual distribution	Multiple diverse paths (one active)	Concurrently maintainable — no outage for planned maintenance
Tier IV	≥ 99.995%	0.4 hrs/yr	2N generators + UPS, active/active dual systems	2N with independent loops	Dual diverse independent paths (both active)	Fault tolerant — single fault anywhere does not cause downtime

Regional compliance bonus matrix (PRO mode — partial excerpt; each region has 4–5 checks, starting baseline 50):

Region / Framework	Key requirement checks	Bonus per check	Governing document
Singapore — MAS TRM	2N UPS; dual diverse fiber; ≥ 48 hr fuel; MFA mantrap; full DCIM	+5 to +15	MAS Technology Risk Management Guidelines §7
Indonesia — OJK/ESDM	2N UPS; full DCIM; ≥ 48 hr fuel; biometric/MFA; 2N generators	+10 each	OJK Circular SEOJK No. 29/SEOJK.01/2022
Malaysia — BNM RMiT	2N UPS; diverse network; full DCIM; ≥ 24 hr fuel; biometric/MFA	+8 to +12	BNM Risk Management in Technology (RMiT) 2020
Thailand — BOT IT Risk	N+1/2N UPS; BMS/full DCIM; ≥ 24 hr fuel; biometric/MFA; diverse network	+10 each	BOT IT Risk Management Circular 2021
Japan — FISC	≥ 72 hr fuel; 2N UPS; biometric/MFA; full DCIM; seismic 2N cooling	+10 to +20	FISC Security Guidelines for Computer Systems §3.2
South Korea — FSS	2N UPS; diverse network; full DCIM; ≥ 48 hr fuel; biometric/MFA	+8 to +12	FSS Electronic Financial Infrastructure Standards
Australia — APRA CPS 234	biometric/MFA; full DCIM; 2N UPS; diverse network; ≥ 48 hr fuel	+8 to +12	APRA Prudential Standard CPS 234 §28–33
EU — DORA	Diverse network paths (Art. 11); full DCIM (Art. 12); 2N UPS; biometric/MFA; ≥ 72 hr fuel	+5 to +15	EU Regulation 2022/2554 (DORA) Art. 11–12
UK — FCA/PRA	Diverse network; full DCIM; 2N UPS; biometric/MFA; ≥ 48 hr fuel	+8 to +12	FCA PS21/3 Operational Resilience; PRA SS6/21
USA — SOC 2/HIPAA/PCI	biometric/MFA; full DCIM; 2N UPS; diverse network; ≥ 48 hr fuel	+8 to +12	AICPA TSC 2017 / HIPAA §164.310 / PCI DSS v4.0
India — RBI/SEBI	2N UPS; diverse network; full DCIM; ≥ 48 hr fuel; biometric/MFA	+7 to +15	RBI Master Direction IT 2021; SEBI Circular CIR/MRD/2012
Global (default)	No overlay — regional score fixed at 50 (neutral)	—	—

05 Outputs

Output	DOM id	Computation	Interpretation
Tier classification	`tierDisplay`	Composite score threshold + floor constraints → Tier I/I+/II/II+/III/IV	Primary result. Uptime Institute tier designation.
Composite score	`scoreValue`	Weighted sum of 6 subsystem scores (0–100)	Continuous measure of infrastructure completeness.
Grade	`gradeValue`	A+/A/B/C/D/F mapped from tier	Letter-grade proxy for tier level.
Confidence	`confidenceValue`	`max(60, min(98, 95 − √variance · 0.5))`	Subsystem score consistency; low confidence = one subsystem is far below the composite.
Uptime Institute tier	`uptimeTier`	Tier N (stripping "+")	Official Uptime Institute classification.
TIA-942-B rating	`tia942Rating`	"Rated " + tierNum	TIA-942-B equivalent rated level.
EN 50600 class	`en50600Class`	"Class " + tierNum	EN 50600-2-1 availability class.
BICSI 002 class	`bicsiClass`	"Class F" + tierNum	BICSI 002 data center design and implementation classification.
Subsystem scores (×5)	`powerVal` …	Per-subsystem weighted scores (0–100)	Bar chart + radar chart source. Identifies the limiting subsystem.
Gap analysis	`gapContent`	Conditional inspection of input values vs next-tier floor requirements	Specific action list: what must change to reach the next tier.
Upgrade cost estimate [PRO]	`upgradeCostVal`	Per-upgrade cost bands summed, ×0.8–1.5 range	Indicative CapEx range in USD thousands to reach next tier.
Regional compliance score [PRO]	`regionalScoreVal`	`calcRegional()` — 50 + jurisdiction bonuses	How well the facility meets the selected jurisdiction's specific requirements.

Upgrade cost bands (PRO cost estimator):

Upgrade action	Trigger condition	Cost band (USD)	Score impact
Upgrade UPS to 2N	upsConfig = n or n1	$200K – $800K	+15 pts
Upgrade generators to 2N	genConfig = n or n1	$300K – $1 200K	+12 pts
Add diverse utility feed	utilityFeeds ≠ dual_diverse	$150K – $600K	+10 pts
Upgrade cooling to 2N	coolRedundancy ≠ 2n or 2n1	$250K – $900K	+10 pts
Add dual-cord PDU (A+B)	pduRedundancy = single	$50K – $200K	+8 pts
Install STS / dual ATS	atsConfig ≠ sts and ≠ dual	$80K – $300K	+6 pts
Deploy full DCIM + BMS	monitoring ≠ full_dcim	$100K – $400K	+5 pts

06 Worked example

Profile: Colocation, 2 000 kW, dual feeds (same substation), N+1 generators, N+1 double-conversion UPS, single ATS, dual PDU, 48 hr fuel, N+1 cooling with dual loop + chilled water, dual diverse network entry, 2 carriers, single MMR, pre-action suppression, card access, BMS, global region. This is the tool's default state. Trace every number against the score maps in §04.

Power subsystem score:
utilityFeeds(dual_same)=40·0.20=8.0 + genConfig(n1)=60·0.25=15.0 + upsConfig(n1)=50·0.20=10.0 + upsTopo(double_conversion)=90·0.10=9.0 + atsConfig(single)=40·0.10=4.0 + pduRedundancy(dual)=70·0.10=7.0 + fuelScore(48)·0.05
fuelScore(48): h=48 → 60 + (48−24)/48·25 = 60 + 12.5 = 72.5 → ×0.05 = 3.625
powerScore = 8+15+10+9+4+7+3.625 = 56.6
Cooling subsystem score:
coolRedundancy(n1)=50·0.50=25.0 + coolDistribution(dual)=70·0.30=21.0 + coolType(chilled_water)=65·0.20=13.0
coolingScore = 59.0
Network subsystem score:
netEntry(dual_diverse)=80·0.45=36.0 + carrierDiv(two)=60·0.30=18.0 + meetMeRoom(single)=50·0.25=12.5
networkScore = 66.5
Physical subsystem score:
fireSuppression(preaction)=65·0.35=22.75 + accessControl(card)=40·0.35=14.0 + monitoring(bms)=65·0.30=19.5
physicalScore = 56.25
Monitoring standalone score:
monitoring(bms) = 65
Regional score (global, free mode):
regionalScore = 50 (neutral baseline)
Composite (raw):
56.6·0.35 + 59.0·0.20 + 66.5·0.15 + 56.25·0.10 + 65·0.05 + 50·0.15
= 19.81 + 11.80 + 9.975 + 5.625 + 3.25 + 7.50 = 57.96 → 58
Floor constraints:
canT3 check: utilityFeeds=dual_same ✓ · genConfig=n1 ✓ · upsConfig=n1 ✓ · coolDistribution=dual ✓ → canT3 = true
canT4 check: utilityFeeds ≠ dual_diverse → canT4 = false
composite=58 < 75 so neither cap applies. Composite remains 58.
Tier assignment:
composite=58 falls in [45, 60) → Tier II · grade C · "Redundant Components"
Framework map: Uptime Tier II · TIA-942 Rated 2 · EN 50600 Class 2 · BICSI 002 Class F2
Confidence:
scores = [56.6, 59.0, 66.5, 56.25, 65] · avg = 60.67
variance = [(56.6-60.67)²+(59-60.67)²+(66.5-60.67)²+(56.25-60.67)²+(65-60.67)²] / 5
= [16.56 + 2.79 + 33.98 + 19.53 + 18.75] / 5 = 91.61 / 5 = 18.32
confidence = max(60, min(98, round(95 − √18.32 · 0.5))) = max(60, min(98, round(95 − 2.14))) = 93%
Gap to Tier III (canT3 = true, so score-based gap applies):
composite=58 < 75 → gaps checked: single ATS (not none/single → gap flagged), coolRedundancy=n1 (gap: add N+2 or 2N), power score limited by N+1 UPS and same-substation feeds.
Primary gap: upgrade utilityFeeds to dual_diverse, UPS to 2N, and cooling to 2N — these collectively add ~30 pts and push composite above 75.

Engineering reading: The default 2 MW colocation profile with N+1 power and cooling is Tier II by score (58). It passes the Tier III structural prerequisites (canT3 = true) but falls short on score because dual-same utility feeds, single ATS, and N+1 UPS/cooling keep the weighted composite below 75. Upgrading to dual_diverse feeds (+24 pts on utilityFeeds), 2N UPS (+30 pts on upsConfig), and 2N cooling (+40 pts on coolRedundancy) would push the composite to approximately 76 — just over the Tier III threshold. The floor constraint check would also need dual cooling distribution (already met) and N+1 generators (already met).

07 References & standards

ANSI/TIA-942-B (2017) — Telecommunications Infrastructure Standard for Data Centers: topology requirements for Rated 1–4, redundancy definitions, telecom room requirements, physical security.
Uptime Institute Tier Standard: Topology (2022) — Definitive Tier I–IV topology requirements: concurrently maintainable (Tier III) and fault tolerant (Tier IV) definitions, single points of failure analysis, availability targets (99.671% / 99.741% / 99.982% / 99.995%).
EN 50600-2-1:2021 — Information technology — Data centre facilities and infrastructures — Part 2-1: Building construction; defines Availability Classes (VK1–VK4) aligned with the Uptime Institute tier concepts under the European framework.
BICSI 002 (2019) — Data Center Design and Implementation Best Practices: Class F1–F4 classification; telecom and network redundancy guidance.
IEC 62040-3 — UPS performance requirements: classification of standby, line-interactive, and double-conversion topologies.
IEEE 493 (Gold Book) — IEEE Recommended Practice for the Design of Reliable Industrial and Commercial Power Systems: MTBF/MTTR basis for component-level reliability data.
NFPA 70 (NEC) / IEC 60364 — Electrical installation standards referenced by TIA-942-B for wiring methods and distribution.
MAS TRM (2021) — Monetary Authority of Singapore Technology Risk Management Guidelines §7: data centre and recovery site requirements.
EU DORA (Regulation 2022/2554) — Digital Operational Resilience Act, Art. 11 (ICT continuity) and Art. 12 (response and recovery): network diversity and monitoring requirements.
FISC Security Guidelines (2022) — Center for Financial Industry Information Systems (Japan): §3.2 backup power 72-hour requirement, seismic considerations.
APRA CPS 234 (2019) — Prudential Standard on Information Security §28–33: physical security and monitoring expectations for APRA-regulated entities (Australia).
Uptime Institute Annual Outage Analysis (2023) — Source for power-subsystem weight (35%): electrical infrastructure failure cited as leading cause of major outages.
Engine source — Inline JS in tier-advisor.html: SCORE_MAPS, WEIGHTS, calculate(), fuelScore(), calcRegional(), buildGapAnalysis().

08 Assumptions & limitations

The Tier Advisor classifies infrastructure topology only — it does not assess the quality of installed equipment, commissioning state, operational procedures, or maintenance history. A data center can be architected as Tier IV on paper but operate at Tier I reliability if maintenance procedures are poor. The Uptime Institute Tier Standard explicitly states that topology defines the design intent; actual performance certification requires on-site audit by Uptime Institute-accredited personnel.

Regional compliance overlays are derived from publicly available regulatory guidance and represent a summarised assessment of key IT infrastructure requirements within each framework. They are not exhaustive compliance audits; full compliance with MAS TRM, DORA, FISC, APRA CPS 234, or any other jurisdiction requires formal assessment against the complete regulatory text by qualified professionals.

Cost estimates in the PRO cost-to-upgrade panel are indicative ranges based on industry benchmark data and are expressed in USD. Actual costs depend on facility size, geographic location, existing infrastructure state, contractor pricing, and scope definition. The range provided (±30% around the midpoint) reflects typical estimation-phase uncertainty (Class 5 estimate per AACE International). This tool and manual are an engineering education and pre-design aid; final infrastructure design and tier classification must be performed by qualified data center engineers and verified by the relevant accreditation body.