HVAC
AMCA field guide to proving an AC-to-EC CRAC fan retrofit
A paired-duty field record for baseline airflow, installed fan watts, control compatibility, failure tests and normalized savings.
Direct answer
Prove an AC-to-EC CRAC fan retrofit by comparing measured electrical input at the same delivered airflow and defined pressure boundary, then demonstrating that the new controls maintain the required cooling through normal operation and specified failures. Use AMCA Publication 203-90 (R2011) for the field-measurement framework, record density and operating conditions with each result, and set project acceptance limits before removing the original fans. Report matched-duty fan savings separately from savings produced by reducing airflow or changing control schedules.
On this page
What proves an AC-to-EC CRAC fan retrofit?Which cooling duty must survive the retrofit?What must the AC-fan baseline record contain?How do you measure installed airflow and pressure?How do you measure installed EC fan watts?What counts as matched duty, and how long should the test run?Will the existing controller operate the EC fans correctly?Which failure tests must the retrofit pass?How do you normalize annual savings?Worked example: what does a published EC retrofit prove?How much retrofit spending do the savings support?Field checklist and turnover recordReview and citationOpen the related tool for the calculation, checklist, or packet.
What proves an AC-to-EC CRAC fan retrofit?
Acceptance requires three results: delivered cooling duty, the measured input reduction promised by the contract, and successful normal and fault control operation.
An AC induction motor can drive the wheel through belts or operate through a variable-frequency drive. An electronically commutated, or EC, motor uses electronic commutation and commonly incorporates its drive electronics. EC fans still accept their specified AC mains supply; this retrofit changes the motor and control arrangement, not the building to DC. See the fan-array and fan-wall guide for assembly architecture. S4, S8, S9.
What belongs to AMCA, the manufacturer and the project?
AMCA Publication 203-90 (R2011) addresses field fan-performance measurement, while ANSI/AMCA 210-25 / ASHRAE 51-25 addresses laboratory aerodynamic rating; a catalog rating does not by itself accept an installed CRAC retrofit. AMCA lists Publication 203 as under review; using its field framework does not confer AMCA certification on the retrofit. S1, S2.
Apply the same field method to chilled-water CRAHs. Equipment manuals govern operating limits; the project defines cooling duty and recovery.
Which cooling duty must survive the retrofit?
Define required airflow, external pressure, cooling mode, active neighboring units and rack-inlet limits before selection. Correct an existing cooling shortfall before accepting the new duty. DX CRAC testing must preserve evaporator airflow and compressor protection; CRAH testing must preserve coordinated chilled-water valve, fan and supply-temperature operation. EPA ENERGY STAR’s undated variable-speed-fan guidance identifies coil condensation and freezing concerns when reducing air-conditioner fan speed. S6, S7.
Map rack inlets at top, middle and bottom using the in-row cooling commissioning method, and preserve the containment air path. Use the installed IT class, humidity/dew-point envelope and thermal ride-through to set operating and abort limits. ASHRAE inlet recommendations in the acceptance table are neither discharge setpoints nor automatic fault-test abort thresholds. S13.
Fix the comparison boundary before changing the air path
Draw fixed supply and return measurement planes outside the replaced assembly, labeling static or total pressure and external or internal losses. Compare useful delivered duty there. Record changed internal losses separately from delivered service. This measures assembly/installation savings; motor-only efficiency requires mechanical-output measurement. S2, S4.
Keep filters, doors, tiles and neighboring-unit operation comparable. Record intentional changes separately, using the CRAC/CRAH airflow guide for plenum and tile delivery. S5.
Will the retrofit fit the cabinet, cooling unit and electrical supply?
Retain fan selections at actual density, inlet temperature, normal duty and degraded duty, including final design filter resistance and wet-coil operation. Check mounting, clearances, wheel/inlet-ring overlap, guards, replacement access, discharge area and drain arrangements against the selected assembly drawing. Obtain the cooling-unit manufacturer’s disposition of the modification and record warranty and listing implications.
Verify mains voltage, phase, frequency, permitted earthing, branch protection, isolation, aggregate inrush and protective-earth/leakage requirements for the exact fan. ebm-papst warns against PWM supply voltage to its fans: remove the old VFD output from their power path and use the approved mains connection and dedicated control input. Control-input PWM is a different signal. Include spare hardware and parameter restoration in the scope. S9, S15.
What must the AC-fan baseline record contain?
Complete the paired record before dismantling, using actual volume in CFM or cubic metres per second. Photograph the old assembly; the motor-nameplate guide explains the ratings. S2, S5.
| Record field | AC baseline entry required | EC retrofit entry required | Comparison/retention rule |
|---|---|---|---|
| Identity | Unit, fan, motor, drive and controller IDs; plate photographs | Replacement fan/controller IDs, firmware and wiring | Same cooling unit and documented assembly changes |
| Duty | Actual airflow in CFM or m³/s; pressure in Pa at labeled planes | Same quantities at the same boundary | Satisfy duty and pair-matching criteria separately |
| Air state | Inlet dry-bulb, humidity, barometric pressure, density | Same measurements and calculation basis | Retain raw and corrected quantities; density match band |
| Electrical input | True kW upstream of motor and any VFD; auxiliaries identified | True kW upstream of EC electronics and new auxiliaries | Same electrical boundary; no nameplate substitution |
| Operating mechanism | Measured wheel RPM, belt/sheave condition, active fan count | Each fan's RPM, active count, min/max settings | Equal RPM is not a match criterion across fan designs |
| Air-path condition | Filter type and differential pressure, coil state, doors, tiles, containment | Same list plus recorded alterations | Changed resistance is a separate contribution |
| Cooling/load | IT load, supply/return temperatures, rack-inlet map, DX stage or water-valve state | Same signals under corresponding mode | Cooling limits remain satisfied through test |
| Time | Time zone, sample interval, start/end and stable windows | Synchronized values and documented exclusions | Preserve raw files and one-minute steady-state means |
| Metrology | Instrument model/serial, range, calibration date and uncertainty | Same or traceably equivalent instruments | Calculate combined comparison uncertainty |
| Failure/return | Existing sequence and relevant faults | Injected event, observed response, alarms and restored settings | Attach fault traces and parameter backup |
Capture an as-found state and a corrected service state when cleaning, belt repair or filter replacement changes performance. Name which state defines contractual savings. A comparison between dirty old equipment and clean new equipment combines maintenance and retrofit effects.
Trend ordinary load variation as well as the peak snapshot; a single operating point cannot establish a variable-load annual profile.
How do you measure installed airflow and pressure?
Use an accessible traverse plane or a validated installed flow-measuring arrangement. AMCA 203-90 (R2011) covers ducted and nonducted fans. Record geometry, orientation, taps, density and instrument uncertainty. S2.
Where cabinet turbulence prevents a repeatable traverse, arrange a temporary measurement duct or calibrated flow station. Include its resistance in the test configuration and reproduce the accepted external duty. Hold the energy comparison until an independent flow reference is credible; adding velocity readings at turbulent cabinet openings does not establish unit airflow.
For inlet-ring measurement, retain the exact fan coefficient, units, density convention and tubing arrangement. The CN1127 V05 manual distinguishes the ring differential used to calculate volume from supply-to-exhaust differential used for pressure control. S8 §6.0.
| Observed pattern | Competing cause | Discriminating check | Action and basis |
|---|---|---|---|
| Low airflow and high filter/coil pressure drop | Added system resistance rather than weak motor output | Compare component drops with baseline at the same flow; retain coil wet/dry state | Restore intended resistance, then retest; S3, S4 |
| Low airflow with distorted inlet/discharge flow | Installation system effect or wrong selection | Inspect geometry and verify total/static pressure and RPM against the selected curve | Correct geometry or select the required duty; S2, S4 |
| Low reported flow while independent flow is adequate | Wrong inlet-ring coefficient, units, sensor range or tubes | Compare calibrated flow with ring signal and exact coefficient | Correct scaling/connection before comparing watts; S8 §6 |
| High input after replacement | More delivered airflow, unequal boundaries or extra resistance | Match external duty and audit fan/auxiliary meter boundaries | Separate service increase from input saving; S5, S10 |
| Surviving fans reach the cap after one failure | Missing reserve, capped capacity or reverse flow | Measure degraded flow, individual RPM, reverse flow and inlet temperatures | Restore specified reserve or reject the redundancy claim; S8, S9 |
| Controller appears healthy during a fault | Disabled warning/relay or stale data | Witness physical fault and remote alarm receipt together | Enable, map and test the required indication; S8 §§7.11-7.12 |
| Cycling after turndown | Competing fan/cooling loops or insufficient DX coil airflow | Trend speed, compressor/valve state, pressure and inlet temperatures together | Restore permitted airflow and coordinate the loops; S6, S8 |
Use the external-static-pressure guide to distinguish the pressure quantities. Raised-floor plenum pressure is not fan total pressure, and tile-flow totals require accounting for bypass and leakage. Component drops and installation geometry separate excessive resistance from inlet/discharge system effects. S2, S4.
What does density normalization change?
At equivalent geometry and speed, normalize aerodynamic pressure using the density ratio and retain both raw and corrected readings. Keep actual volume distinct from mass flow. The fan laws describe these relationships; they do not justify density-correcting mains watts as though electronics and motor losses were aerodynamic. Use comparable inlet conditions or a documented electrical model, with measured true power at each accepted condition. S4, S11.
How do you measure installed EC fan watts?
Measure active input kW upstream of every fan’s electronics and include incremental controls consistently. The old boundary must include its VFD where present. Separately meter fan power when compressors, heaters or humidifiers share the cooling-unit feeder. FEMP M&V Guidelines 5.0, September 2024, uses a defined measurement boundary for baseline and reporting-period energy. S5.
| Quantity/boundary | AC measurement | EC measurement | Use in acceptance |
|---|---|---|---|
| Fan electrical input | Mains active kW including VFD loss if present | Mains active kW including integrated drive loss | Primary paired-power comparison |
| External controls | Identify included/excluded old controls | Include additional controller supply consistently | Add any incremental auxiliary power outside fan meter |
| Shared unit feeder | Separate compressor, heater and humidifier consumption | Same separation | A whole-unit change is not automatically a fan saving |
| Shaft horsepower | Record rating and actual load only if separately measured | Record relevant rated output if provided | Not a measured electrical saving |
| Amps and power factor | Actual phase currents and true PF | Actual phase currents and true PF | Diagnostics; current alone cannot establish kW |
| Fan/controller telemetry | Compare against portable analyzer | Compare at low, ordinary and high duty | Trend only after checking scaling and agreement |
| Stopped but energized state | Record remaining control/drive watts | Record remaining electronics/controller watts | Include in annual bins where the mode occurs |
For US general-industry electrical work, OSHA 29 CFR 1910.333(b)(2) requires isolation from all electrical sources, lockout/tagging and verification of deenergization before treating exposed parts as deenergized. A BAS stop or enable input is not isolation. Qualified personnel must follow the employer’s procedure for meter connections; energized testing falls under 1910.333(a) and (c), with suitable protective practices and instruments. S16.
Use a calibrated analyzer with the correct wiring configuration, voltage/current channels, range and sensor polarity. For a balanced three-phase estimate, multiply line-to-line RMS volts, line RMS amps and true power factor by the square root of three, then divide by 1,000: P = √3 × VLL × I × true PF / 1000. For acceptance, sum active watts using the analyzer’s supported three-wire or four-wire connection, with bandwidth suitable for the current harmonics. For distorted current, displacement power factor is not true power factor for distorted current. S11, S12.
Can a small difference be called a saving?
Combine power uncertainty and repeatability on a standard-uncertainty basis, account for correlations, then state coverage and remaining duty mismatch. Convert expanded limits before root-sum-squaring. The same meter does not automatically cancel setup errors. Anvilfield accepts a saving only when its sign survives this treatment; improve unresolved measurements. S5, S12.
Anvilfield calculates that 3% less speed and airflow means 8.73% less aerodynamic power along an unchanged similarity/system relationship. Tighten the match or compare documented local power-versus-duty curves; do not apply the cube relationship as a mains-kW correction between different fans. S4.
What counts as matched duty, and how long should the test run?
Anvilfield’s proposed schedule totals three hours of accepted AC/EC data across the three duty points in the table. Add setup, stabilization, load changes and fault tests separately. Allow separately for access, lifting, wiring, mechanical work and rollback.
Test minimum permitted cooling duty, ordinary duty and peak design duty, then the specified degraded state. Match delivered airflow and the defined external pressure at comparable density, rather than equal RPM or equal command percentages. Stabilize through the relevant control cycle before collecting windows; extend runs when cycling, hunting or load variation prevents comparison.
Agree these Anvilfield proposed criteria, dated November 5, 2026, before baseline collection; they are project proposals, not AMCA tolerances.
| Item | Proposed value or governing rule | Pass / hold action | Authority category |
|---|---|---|---|
| Required duty | Meet specified minimum airflow and defined pressure duty | Fail measured duty shortfalls; hold missing evidence | Project design; S2 framework |
| Pair airflow match | Difference inside ±3% of agreed reference airflow, tightened where needed for the claimed saving | Include uncertainty and duty bias; rerun unresolved comparisons | Anvilfield proposal; S4 sensitivity |
| Pair pressure match | Difference inside ±3% at the fixed external pressure boundary, or a pre-agreed absolute Pa band near zero | Record pressure type, planes, uncertainty and internal losses | Anvilfield proposal |
| Density match | Within ±1% as an initial direct-comparison band | Preserve inlet conditions; aerodynamic corrections do not normalize mains watts | Anvilfield proposal; S4 physics |
| Stable comparison | Three ten-minute windows per operating point after stabilization | Extend for cycling, hunting or load changes; preserve each window | Anvilfield proposal |
| Steady-state logging | One-minute aligned averages, retaining underlying instrument data | No comparison of unsynchronized windows | Anvilfield proposal |
| Failure logging | One-second sampling or faster where the event demands it | Retain peaks and timestamps; minute averages conceal transients | Anvilfield proposal |
| Fan-loss alarm | Approved injection-to-remote-alarm maximum | Record elapsed time; define limit before testing | Project thermal ride-through and alarm sequence |
| Degraded-duty recovery | Approved injection-to-sustained-duty maximum | Maintain specified thermal limits throughout; test airflow and temperature rather than RPM alone | Project thermal ride-through and recovery sequence |
| Server inlet | ASHRAE’s fifth-edition thermal reference card recommends 18 to 27°C inlet air for A1-A4 and 18 to 22°C for H1. | Apply IT class/project limits; abort before threshold | S13, fifth edition reference card |
| Controls energy review | Recommends 15 or 30 days of post-installation measurements covering cooling-load variation | Apply to networked CRAC/CRAH/AHU controls; extend for missing conditions | S14 §6.3.4.2, Addendum 1.0, August 2025 |
| Annual controls savings | Recommends at least 9, preferably 12, months of performance-period data | Use common load/conditions and a stated annual basis | S14 §6.3.4.2; distinct from a short fan acceptance test |
Who sets the numerical limits?
The project sets minimum cooling duty, pressure duty, fault alarm time and recovery time. Pair matching does not permit operation below a required minimum. Put the uncertainty interval for each AC-minus-EC difference inside the agreed match band; a center reading inside the band is insufficient when its interval extends outside. Set an absolute Pa band before testing near zero pressure.
For a 10,000 CFM reference, Anvilfield calculates that a 200 CFM pair difference with an entered 95% expanded difference uncertainty of 150 CFM spans 50 to 350 CFM. The proposed 3% band permits 300 CFM: hold this pair because its uncertainty interval crosses the band, despite the center reading being inside it.
FEMP’s August 2025 Data Center Addendum §6.3.4.2 also recommends at least twelve, preferably twenty-four or more, months of baseline utility data for networked-controls M&V. Separate that annual verification from fan acceptance: use retrofit isolation or a documented load/weather model appropriate to the claim. Assumed annual hours remain calculation inputs. S5, S14.
Will the existing controller operate the EC fans correctly?
Record exact fan part number, firmware, controller mode, analog loading/reference, enable polarity, limits and status mapping. Prove command, feedback and physical response at start, low duty, ordinary duty, peak duty and stop. Verify the documented load and voltage at the farthest fan. S8, S9.
For the ECblue versions below, the command-loss behavior assumes mains supply is present and the fan is enabled. Record the installed version before disconnecting its signal.
| Device/control layer | Verified signal or setting | Acceptance action | Source/edition |
|---|---|---|---|
| ZIEHL-ABEGG ECblue Basic D/G, normal input version | 0-10 V or 0-100% PWM; 10 V means maximum; no signal stops fan | Record exact version, prove normal command and disconnected command | S9, L-BAL-F053-GB 2021/22 Index 013, §5.8 |
| Same family, inverse input version | 10-0 V or 100-0% PWM; 10 V stops; no signal means maximum | Test within cooling and speed limits; verify alarms independently | S9, same revision, §5.8 |
| Same family, zero minimum-speed configuration | Starts at at least 6% rated speed; stops below 4.5% | Verify minimum delivered airflow against CRAC requirements | S9, §5.8 |
| ebm-papst CN1127 MMCU fan interface | Supports ebm-papst-enabled Modbus fans version 5.0 and later; LITE variants omit some parameters | Inventory fans and prove each required readable/writable point | S8, OMI-CN1127-2025-06-V05, p.29 |
| CN1127 speed cap | Minimum cap range 0-50%; maximum from minimum to 100%; minimum cap unavailable in constant volume/pressure mode | Record selected mode and verify cap behavior under demand and fan loss | S8, §7.4 |
| CN1127 enable input | Default open/high enables; closed/low disables; polarity configurable | Prove both states and restoration at fan and BAS | S8, §7.6 |
| CN1127 alarms | Relay and controller-detected warnings disabled by default | Enable required warnings and prove actual remote indication | S8, §§7.11-7.12 |
| Existing BAS analog output | Output loading and reference must match all connected fan inputs | Verify documented electrical ratings and voltage at the farthest fan under load | Anvilfield integration test; S8, S9 |
| BAS-to-controller communications | Timeout, fallback command and restart values are configuration/model specific | Record installed timeout; test both network segments and stale-data behavior | Anvilfield integration test; S8 |
Witness configured alarms at the receiving system and verify stale-data indication. A last-good RPM display does not establish current fan response.
Minimum speed is a cooling constraint as well as a setting
Distinguish motor start thresholds, controller caps and minimum coil airflow. Schedule active configuration with cooling reserve: CN1127 V05 §7.0 stops fans on entering Advanced Settings outside Monitor mode, and §6.4 sensor mapping starts fans at potentially high speed. Save parameters and verify restoration. S8.
Sweep the permitted range for vibration, rubbing, current and sound against equipment limits. CN1127 §7.13 resonance avoidance is limited to compatible Gen3 fans with vibration sensors and disabled for Gen2; AMCA 203 excludes vibration, sound and stress measurement. When cap release is selected, prove its fan-failure response against current, sound and airflow limits. Include an AC-with-VFD alternative when it meets the same service requirement. S2, S8.
Which failure tests must the retrofit pass?
Anvilfield proposes the following functional tests against the approved project sequence, using a written method of procedure with cooling reserve available. Establish thermal abort limits first. Use approved isolation or signal simulation, never a mechanical jam, deliberate overheating or live conductor fault. ECblue Basic Index 013 §2.2 prohibits blocking the fan with objects; §2.6 warns of automatic restart and voltage generated by a windmilling rotor. S9. Restore settings and prove normal operation after every event.
Define alarm time from fault injection to receipt at the designated BAS/operator point, and recovery time from injection to sustained acceptable duty. The project’s thermal ride-through sets both limits. Record airflow and inlet-temperature traces as well as RPM; a speed response alone does not establish cooling recovery.
| Injected event | Controlled method | Signals to record | Required result |
|---|---|---|---|
| Speed-command loss | Open/simulate loss at approved signal interface | Actual input, RPM, airflow, rack inlet, alarm | Exact configured fail state, cooling maintained or approved protective response |
| BAS communications loss | Isolate supervisor link without removing local controls | Timeout, stale-data flag, local command, RPM | Documented autonomous fallback and alarm; correct return on restoration |
| Fan-bus loss | Isolate an approved test segment | Each affected fan's feedback, local status and array capacity | Expected fan behavior identified separately from supervisor loss |
| Pressure sensor open/out of range | Approved input simulator or test function | Raw input, interpreted pressure, command, alarm | Configured diagnostic, bounded demand and approved fallback |
| Pressure sensor frozen/plausible bias | Hold an approved simulated value while physical duty changes | Independent pressure/flow or temperature, sensor disagreement, command, alarm | Required independent plausibility logic detects fault; add detection when existing controls cannot meet the sequence |
| One fan disabled | Disable selected fan through approved means | Surviving speeds, airflow, backflow, rack-inlet peak | Specified degraded duty and alarm/recovery time |
| One fan loses power | Approved branch isolation under the method of procedure | Backflow, remaining input kW, controller detection | Prove capacity when powered braking is unavailable |
| Common controller power loss | Isolate only after cooling reserve is verified | All fan responses, remote alarm, rack inlets | Actual common-mode exposure matches accepted design |
| Unit power restoration | Approved supply interruption/restoration | Restart order, peak demand, command retention, alarms | Fans, cooling and BAS recover without lost settings or unsafe cycling |
| Load increase/decrease | Approved staged thermal load or test mode | Airflow, fan/valve/compressor states, inlet temperatures | Stable coordinated control; no sustained hunting or repeated alarms |
| Cooling-protection response | Approved simulation of airflow proof/protection input | Compressor inhibit, fan status, alarms, restoration | OEM-required cooling protection operates; never freeze a coil to prove it |
A frozen but plausible pressure signal needs independent airflow, temperature or cross-sensor plausibility detection. CN1127 V05 §7.12 low-pressure-versus-speed warning logic does not identify every frozen or high-biased sensor. Add the required detection or retain the unmet acceptance item. S8.
How do you prove N+1 and restart without concealing the transient?
Test each relevant fan position or the documented worst case at required duty. Measure surviving speeds, backflow, airflow, highest rack-inlet temperature and recovery. Disable and supply loss are separate events because powered braking and diagnostics can disappear with power; a holding brake does not establish an airtight seal. A full-duty N+1 claim requires the specified design cooling duty with any one required fan unavailable. Label reduced-duty survival separately; record common controller and supply failures beyond that single-fan claim. S8, S9.
Follow the installed model’s discharge and restart instructions. ECblue Basic Index 013 §2.6 requires at least three minutes after supply disconnection before opening its controller housing; elapsed time alone does not verify absence of voltage. Its §5.5.1 requires single-phase types to remain off at least 90 seconds before re-energization to reactivate inrush-current limitation. Coordinate compressor restart and group-control return with the CRAC sequence and the zone failover test. S9 §§2.6, 5.5.1.
How do you normalize annual savings?
Subtract EC input from AC input at each common duty, multiply each difference by that duty’s annual hours, then add the results: annual fan saving (kWh) = Σ[(P_AC,i − P_EC,i) × h_i]. Include standby, identify extrapolated bins and make hours sum to the reference year. Keep IT load, weather and cooling service on a common basis. S5, S14.
Meter original and new equipment at required duty for hardware/installation savings, then new automatic operation for additional controls savings. Separate airflow-management corrections from the motor change.
Using the published LBNL full-speed inputs and an assumed 8,000 running hours, Anvilfield estimates 14,800 kWh/year saved. The table weights those reported operating conditions by hours; a project’s normalized bins pair measured airflow, pressure and watts at common duty. S10.
| Reference-year mode | Hours | AC fan input, kW | EC fan input, kW | AC electricity, kWh | EC electricity, kWh | Saving, kWh |
|---|---|---|---|---|---|---|
| Running at the reported full-speed conditions | 8,000 | 6.57 | 4.72 | 52,560 | 37,760 | 14,800 |
| Fan supply isolated | 760 | 0 | 0 | 0 | 0 | 0 |
| Reference-year total | 8,760 | 52,560 | 37,760 | 14,800 |
Include energized standby watts whenever that mode occurs. Report fan electricity separately from compressor/chiller interactions; additional cooling savings need measured or documented marginal cooling efficiency. Site PUE is not that marginal efficiency. Apply sequential percentage savings to remaining energy rather than adding percentages. S5, S10.
Why the fan cube law is a screen, not a cross-motor acceptance test
Under the DOE/AMCA sourcebook’s similarity assumptions, Anvilfield calculates 48.8% less aerodynamic power for 20% less speed because 0.8³ = 0.512. Changing geometry, drive losses, electronics or the system curve breaks a direct command-percentage comparison. Measure installed watts at the new operating point. S4, S11.
Worked example: what does a published EC retrofit prove?
LBNL-6007E, November 2012, reports a full-speed input change from 6.57 to 4.72 kW per CRAH after an EC plug-fan retrofit, with greater upgraded airflow; Anvilfield calculates a 28.16% reduction in that reported full-speed input. S10, printed pages 15-16.
| Input or result | Filled value | Provenance and interpretation |
|---|---|---|
| Equipment | Liebert FH740C-AA005291 CRAH | S10, printed p.15; chilled-water case, not a DX coil test |
| Original fan arrangement | One 10 hp motor driving three fans by belt/shaft | S10 historical description |
| Retrofit arrangement | Two Rosenberg/Continental 630 mm EC plug fans | S10 historical description |
| Original full-speed input | 6.57 kW per CRAH | Facility records reported in S10 |
| Retrofit full-speed input | 4.72 kW per CRAH | Measurement reported in S10 |
| Difference | 1.85 kW per CRAH | Anvilfield calculation: 6.57 - 4.72 |
| Input reduction | 28.16% | Anvilfield calculation: 1.85 / 6.57 × 100 |
| Airflow comparison | Upgraded fans supplied more airflow at full speed | S10 printed p.16; equal-duty airflow/pressure pairs not supplied |
| Annual operating basis | 8,760 hours at these input levels | Annualization assumption |
| Baseline annual electricity | 57,553 kWh | Anvilfield estimate: 6.57 × 8,760 |
| Retrofit annual electricity | 41,347 kWh | Anvilfield estimate: 4.72 × 8,760 |
| Annual fan electricity reduction | 16,206 kWh | Anvilfield estimate: 1.85 × 8,760 |
| Energy-price input | $0.12/kWh | Assumed energy rate |
| Annual energy-cost reduction | $1,945 | Anvilfield estimate: 16,206 × $0.12 |
Anvilfield estimates annual energy and cost by multiplying the November 2012 report’s input difference by the table’s assumed hours and rate. Demand charges, incentives, maintenance and cooling-plant interactions are excluded. Round dollars after calculating from unrounded kWh.
The report used facility records for original input and demonstration measurements for upgraded input, and corrected telemetry with portable instruments. Retain that provenance with the result; collect paired flow, pressure and power for a new job’s equal-duty acceptance. S10.
How much retrofit spending do the savings support?
Anvilfield estimates a $5,834 energy-only net installed-cost ceiling for a three-year simple payback using the worked example’s continuous schedule and $0.12/kWh rate. Multiply unrounded annual dollar savings by the owner’s target payback to obtain the ceiling.
| Operating basis | Energy-price assumption | Annual saving | Two-year net-cost ceiling | Three-year net-cost ceiling |
|---|---|---|---|---|
| 8,760 h/year | $0.08/kWh | $1,296 | $2,593 | $3,889 |
| 8,760 h/year | $0.12/kWh | $1,945 | $3,889 | $5,834 |
| 8,760 h/year | $0.20/kWh | $3,241 | $6,482 | $9,724 |
| 8,000 h/year | $0.12/kWh | $1,776 | $3,552 | $5,328 |
| 4,000 h/year | $0.12/kWh | $888 | $1,776 | $2,664 |
Net installed cost is the total quoted work minus approved incentives. Scope the fan assembly, supports, air path, electrical work, controls, metering, testing and balancing, failure tests, outage/temporary cooling, disposal and turnover. State maintenance and demand-charge savings separately and deduct incentives once.
Past HVAC awards and contract amounts in PorchRocket Bid Control give estimators a procurement reference when separating fan hardware, metering, controls integration and witnessed acceptance in a public cooling bid.
Recalculate when duty hours, energy rate or accepted power changes. EPA’s historical VSD paybacks are project results, not current EC installation prices. S6.
Field checklist and turnover record
Give duty, energy, controls and degraded operation separate outcomes. Pass when the measured result and decision uncertainty meet the requirement; hold when a required condition or evidence is missing; fail when a measured result violates it. A cooling pass cannot erase a failed energy guarantee, and saved watts cannot erase a cooling failure.
Contractors retain equipment/power records, TAB retains airflow evidence, and controls integrators retain settings/fault traces. The owner’s commissioning lead accepts them against the approved criteria. Deliver raw files, excluded intervals, calibration records, equations, photographs, wiring drawings, timestamps and parameter backups. S5, S8, S9.
Anvilfield recommends a 30-day operating review, an annual comparable-duty check and retesting after fan, firmware, control, filter-specification or air-path changes. Follow shorter model-specific service intervals where required.
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Review and citation
Sources checked and reviewed by Anvilfield: 2026-11-05. Update log, 2026-11-05: verified source editions, clarified electrical isolation and N+1 acceptance, and checked the worked calculations.
How to cite this page: Anvilfield, AMCA field guide to proving an AC-to-EC CRAC fan retrofit, 2026, https://anvilfield.com/field-guides/hvac/crac-ec-fan-retrofit-acceptance/
FAQ
How do you prove a CRAC AC-to-EC fan retrofit using AMCA methods?
Compare installed airflow, defined pressure and electrical input under matched conditions using AMCA 203-90 (R2011)'s field framework, then prove the project's cooling and control sequence.
How are EC fan retrofit installed watts measured?
Measure active input power upstream of the EC electronics and include all fans and incremental controls within the stated boundary.
What belongs in a CRAC fan baseline measurement?
Record airflow, pressure planes, density, active kW, RPM, equipment identities, filter/coil condition, cooling load, settings, timestamps and measurement uncertainty before removal.
Which retrofit controls failure tests are required?
The project sequence defines acceptance, with command, sensor, network, individual-fan, controller-power and restoration tests covering the main failure paths.
Does 0 V always stop an EC fan?
No: with mains supply present and the fan enabled, the cited ECblue Basic normal-input version stops without a signal, while its inverse-input version runs at maximum.
Can I use amps or motor horsepower to calculate the saving?
Neither alone measures installed input power; use active watts with the correct voltage, current and true-power-factor measurement.
Can the new EC motor stay on the old VFD output?
ebm-papst warns against PWM supply voltage to its fans; power the selected fan through its approved mains arrangement and use the dedicated speed-control input, not the old VFD output.
Does a two-fan retrofit automatically provide N+1?
No. A full-duty N+1 claim requires the specified design cooling duty with any one required fan unavailable; test airflow, temperature, backflow and recovery, then record common controller and supply exposure separately.
Is a 30% lower power reading proof of a 30% efficiency improvement?
No: at the same useful air power, 30% less electrical input means a 42.86% increase in wire-to-air efficiency, calculated by Anvilfield as 1 / 0.70 - 1; motor-only efficiency requires mechanical-output measurement.
How much can I spend for a three-year payback?
At 1.85 kW saved for 8,760 hours and $0.12/kWh, Anvilfield estimates an energy-only net-cost ceiling of $5,834.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.
Sources checked
- S1. AMCA publications catalog: current editions Publisher listing; 203 under review.
- S2. AMCA 203-90 (R2011): public field-measurement scope Public description; field framework.
- S3. AMCA: Minimizing & Troubleshooting Fan System Effects, 2023 Educational installation/system-effect examples.
- S4. DOE/AMCA: Improving Fan System Performance, April 2003 Fan laws and drive losses.
- S5. DOE FEMP: M&V Guidelines 5.0, September 2024 Baseline, boundaries and uncertainty.
- S6. EPA ENERGY STAR: Replace Standard Fans with Variable Speed Fans Undated resource; DX airflow constraints.
- S7. EPRI 1024339: DX CRAC Airflow Management, December 2012 Chapter 6, Case Study 2, pp.6-4 to 6-5.
- S8. ebm-papst CN1127: OMI-CN1127-2025-06-V05 Firmware 1.3.0; §§6-7.
- S9. ZIEHL-ABEGG ECblue Basic: L-BAL-F053-GB, 2021/22 Index 013 D/G motor sizes; §§2.6, 5.5, 5.8, 7.3.
- S10. LBNL-6007E: Intelligent Control and Fan Improvements in CRAHs, November 2012 Printed pp.10-16; DOI 10.2172/1168605.
- S11. Yokogawa: Electric Motor Testing, Power Measurement and Analysis Undated; parts 2-3.
- S12. Fluke 433/434 Users Manual, August 2004 Connections, watts and measurement ranges.
- S13. ASHRAE Thermal Guidelines, fifth edition revised and expanded: 2021 reference card, ©2021/2024 Class-specific inlet conditions.
- S14. FEMP Data Center Addendum 1.0, DOE/FEMP-0037, August 2025 §6.3.4.2 controls M&V.
- S15. ebm-papst US FAQ: PWM supply versus control input Undated manufacturer resource; dedicated input, not PWM supply.
- S16. OSHA 29 CFR 1910.333: electrical work practices Current regulatory text checked 2026-11-05; (a), (b)(2), (c)(2).