✦ Ensmart BMS Academy Home About Us Products Solutions Case Studies eNews Blog Downloads Team Contact Design Engg Get a Demo →
Home › AI BMS Mentor › BMS Systems Design

Environmental Monitoring System in Data Centres — Rack-Level Temperature and Humidity

BMS Systems Design 43 views ✓ Reviewed by EnSmart engineer Updated 3 hr ago
Copied to clipboard ✓
Environmental Monitoring System in Data Centres — Rack-Level Temperature and Humidity — infographic

A Hyderabad Datacenter, the CTO's Question

Naveen's Tier-3 colocation operates well. PUE is at 1.42. SLA compliance is strong. The BMS reads zone-level temperature and humidity per cold aisle. The reports look fine. But the CTO asks a question Naveen has not heard before: "Customer Cage 7 had a server thermal-throttle event at 14:32 last Thursday. Their server logs show inlet air at 31 degC. Our BMS shows the cold aisle at 22 degC at the same time. Where is the gap?" Naveen investigates. The rack in question is at the end of a cold aisle, near a containment door that was held open for 90 seconds while a technician worked on a neighbouring cabinet. The zone-average reading was correct — for the zone average. The specific rack experienced a transient hot inlet that the zone sensors did not see. ``` Zone-level monitoring caught: nothing unusual Rack-level monitoring would have caught: 31 degC inlet on Rack 47B, lasting 8 minutes Customer impact: thermal throttle, brief performance degradation, SLA report point ``` Datacenters cannot operate on zone averages alone. Modern customer SLAs require rack-level evidence. Every single one of these problems has one solution — Environmental Monitoring System for datacenters with rack-level granularity.

Why Rack-Level Monitoring Differs From Zone-Level

``` Zone-level (BMS standard): 1-3 sensors per cold aisle Calculates aisle-average inlet temperature Drives PAC staging and setpoint resets Adequate for plant-level control What zone-level misses: Hot spot at end-of-aisle (near doors, walls) Hot spot at top-of-rack (heat rises in unsealed cabinets) Containment leakage at specific rack pairs Customer-specific server load spikes Rack-level (EnvMS): 2 sensors per rack: inlet bottom + inlet top Often: outlet temperature also Independent of zone-level instrumentation Provides per-customer evidence Identifies specific hotspots and trends ```

What EnvMS Records in a Datacenter

``` Per rack: Inlet temperature (bottom) AI every rack Inlet temperature (top) AI every rack Outlet temperature (optional) AI every rack Door-open status DI every rack Per aisle: Aisle differential pressure AI every aisle pair Aisle temperature trend AI multiple per aisle Per row: Containment door status DI every door Door-open duration trends calculated Per zone: Zone temperature AI backup to rack Zone humidity AI one per zone Air-handler discharge temperature AI correlate with rack inlet Raised floor: Leak detection cable continuous Floor void temperature AI multiple zones Static pressure AI multiple zones Per CRAC/CRAH: Discharge temperature AI independent of BMS Return temperature AI independent of BMS Humidity AI independent of BMS Per customer cage: Aggregated cage temperature calculated from rack-level Per-cage cooling delivery calculated Per-cage power use from PDU integration ``` For a 180-rack DC, this is 400-600 sensors. The infrastructure cost is meaningful, but the customer-evidence value is far greater.

ASHRAE TC 9.9 Reference

ASHRAE Technical Committee 9.9 publishes thermal guidelines for datacenters: ``` A1 Class: Recommended: 18-27 degC inlet Allowable: 15-32 degC inlet Application: most colocation, modern equipment A2 Class: Allowable: 10-35 degC Application: modern enterprise A3, A4: progressively wider ranges for newer equipment Recommended humidity: Lower limit: 5.5 degC dew point or 20 percent RH (lower of the two) Upper limit: 21 degC dew point or 80 percent RH (lower of the two) ``` For Naveen's facility, A1 class applies. The recommended inlet temperature range is 18-27 degC. Customer SLAs typically commit to "below 27 degC inlet" or "ASHRAE A1 recommended range."

How Rack-Level EnvMS Catches What Zone Cannot

``` Hot spot at end of aisle: Zone sensor reads 22 degC (aisle average) Rack-end sensor reads 28 degC EnvMS catches the 28 degC; zone alone misses it. Containment leakage: Specific rack pair has a gap in containment seal Zone-average sensor reads acceptable Rack-level sensor on the leaking pair reads 31 degC EnvMS isolates the problem to specific rack pair Door held open during work: Zone average shows brief blip Specific racks near the door drop to 30+ degC for several minutes EnvMS captures the per-rack impact Investigation links to door-open event automatically PAC discharge drift: PAC discharge sensor (BMS) reads correct Independent EnvMS PAC discharge sensor reads 2 degC higher Discrepancy flagged Investigation reveals BMS sensor drift; EnvMS sensor catches ``` In every case, EnvMS provides evidence that BMS alone could not.

Customer-Facing SLA Reports

EnvMS produces customer-specific reports: ``` Monthly customer environmental report: Period: 1-31 March 2026 Customer: Cage 7, 28 racks Inlet temperature compliance: SLA target: ≤27 degC at all times Compliance: 99.94 percent of measurement intervals Excursions: 2 events Event 1: 27.3 degC for 4 minutes on 14 March Event 2: 28.1 degC for 11 minutes on 26 March Humidity compliance: SLA target: 20-80 percent RH Compliance: 100 percent PUE for the customer's load: Average: 1.43 Investigation reports: Event 1: Containment door held open for service work. Procedural review completed. Event 2: PAC-3 brief stage transition during DG test. Sequence adjusted; not expected to recur. ``` These reports build customer trust. They are produced automatically from EnvMS data.

Architecture — EnvMS Independent of BMS

``` The EnvMS layer: Separate sensors (validated) installed in every rack Separate network (often a dedicated VLAN) Separate controllers (or BACnet IP appliances dedicated to EnvMS) Separate database Separate user access (DC operations + customer-facing reporting team) The BMS layer continues to: Control PAC units, fans, dampers Manage chiller plant, UPS, DG Drive setpoint resets Generate operational alarms The two systems interconnect at the integration layer: Each can read the other's data via BACnet IP read-only Cross-validation at zone level Combined alarms (e.g., "rack inlet >27 degC AND CRAC running normally" = containment problem) ``` The independence is the same principle as in pharma — one controls, the other records.

Leak Detection — A Distinct Critical EnvMS Function

``` Datacenters have water in unexpected places: Chilled water under raised floors (CRAH coils) Condensate from CRAC units Humidifier supply water Sprinkler system charge A leak in the white-space is a critical event: Water + electricity = catastrophic risk Even small drips can damage equipment Customer SLA breach is severe Leak detection cable: Sensitive cable run under the raised floor Detects water contact along the entire length Triggers immediate alarm to EnvMS and to operations Maps the leak location to within 1-2 metres EnvMS records: Every leak event Investigation outcome Resolution time Compliance with response SLA ``` Leak detection is one of the highest-priority EnvMS functions in a DC.

Naveen's Deployment

``` Phase 1 Per-rack inlet temp sensors (180 racks × 2 = 360 sensors) Phase 2 Containment door-open contacts on all aisle doors Phase 3 Independent CRAC discharge sensors Phase 4 Leak detection cable refresh and full coverage Phase 5 Customer-facing report templates and data exports Phase 6 DCIM integration of EnvMS data Six months later: Hot-spot identification: 4 specific racks identified for containment improvement Customer report adoption: standard for top 30 customers SLA dispute reduction: significantly fewer disputes (customers see their own data) PUE improvement: rack-level data revealed two PAC units running at higher load than necessary; sequence adjusted, PUE dropped to 1.39 ```

Why DC EnvMS Matters Now

Datacenter customers are increasingly sophisticated. Hyperscalers run their own analytics on inlet temperature data they request from colocation providers. Per-rack environmental data has become a contractual deliverable, not just a nice-to-have. For Indian colocation, the competitive landscape requires this level of evidence. The provider who can deliver granular customer reports wins on quality. The provider who can only show zone-averages competes only on price. Datacenter EnvMS is rack-level evidence at customer-SLA precision. The zone average is the BMS comfort zone. The rack-inlet temperature is the customer's experience. The two diverge at the moments that matter — and the EnvMS that captures the divergence is what makes the SLA report defensible. Every rack, every minute, every degree — all recorded, all retained, all customer-visible.

Related Topics

Related Topics

Related Topics

Related Topics

Was this answer helpful? ✓ Thanks — your feedback was recorded.

Related questions

Have a different question?

✦ Ask the AI BMS Mentor → More from BMS Systems Design →