Independent study 01 · Data center cooling

Supplier Qualification & FAT Strategy for a 1 MW Liquid Cooling Skid

A complete supplier-quality approach for a hypothetical liquid-cooling skid: qualification scoring, manufacturing readiness, inspection and test planning, factory acceptance testing and shipment release — with every decision gate, its evidence, and the criteria that would have changed it.

Independent Engineering Study

This project uses a hypothetical supplier and synthetic equipment specifications. It does not represent professional data-center project experience. Its purpose is to show how I would structure supplier quality for mission-critical cooling equipment, using publicly available technical references — not to present design guidance or to imply a client relationship.

Project record

Independent Study

Project type: Independent engineering study
Domain: Data center cooling — liquid-to-liquid coolant distribution
Methods: Supplier qualification · Manufacturing readiness review · Inspection & test planning · FAT protocol design · Shipment release
Date: September 2026
Data: Synthetic. Supplier, equipment and all results are constructed for this study.
Confidentiality: No proprietary employer information used. All technical grounding is from public sources, cited at the end.

How this document set is produced

The tables on this page and the downloadable engineering documents are generated from the same structured source data, so the two cannot drift apart. The supplier qualification total is computed from the individual area scores rather than stated — change a score and the total, the decision band and the verdict all change with it. Configuration control applies to a portfolio as much as to a product.

Section 1

The equipment under qualification

Supplier quality is meaningless in the abstract, so the study starts by defining what is being bought. The Northstar Thermal Systems LC-1000 is a liquid-to-liquid coolant distribution skid, nominal 1 MW — a hypothetical unit created for this study.

Every parameter is classified as design context (traceable to a cited public source), study assumption (chosen here, not a real specification), or acceptance requirement (the criterion the equipment is tested against, derived from the assumptions above it). Nothing on this page should be read as a product specification.

Block diagram — LC-1000 (hypothetical)

Hypothetical arrangement created for this study. Values shown are study assumptions or acceptance requirements derived from them — see the parameter table below for the basis of each. Not a product specification and not suitable for design use.

Engineering Assumption

Nominal thermal capacity is assumed to be 1 MW, with a 10 K rise on the technology circuit. Those two assumptions set the design flow, and the design flow sets every hydraulic acceptance value in this study. On a production project all three would come from the approved equipment specification, and the acceptance values would follow the owner’s requirement rather than this derivation.

Equipment definition and the basis of each parameter

Hypothetical specification

Parameter	Value	Basis	Source or derivation
Nominal thermal capacity	1 MW	Study assumption	Chosen as a representative single-skid capacity so the quality problem is concrete. Not derived from any product.
Architecture	Liquid-to-liquid, separated facility (primary) and technology (secondary) circuits	Design context	A CDU transfers heat between dedicated liquid loops and provides pressure, flow, temperature, dew-point control, cleanliness and leak detection (OCP Cooling Environments).
Secondary (technology) supply temperature	32 °C maximum	Study assumption	Selected to sit at the ASHRAE W32 facility water class boundary. The class exists; selecting it for this skid is a study assumption.
Primary (facility) supply temperature	27 °C maximum	Study assumption	Selected at the ASHRAE W27 class boundary to give a workable approach temperature across the heat exchanger.
Secondary circuit temperature rise	10 K	Study assumption	Chosen to set the design flow. A different ΔT would change flow and therefore every hydraulic acceptance value below.
Secondary design flow	1,440 L/min (24 L/s)	Acceptance requirement	Derived: 1,000 kW ÷ (4.18 kJ/kg·K × 10 K) ≈ 23.9 kg/s ≈ 24 L/s. Rounded to 1,440 L/min, which returns a 10.0 K rise at capacity.
Secondary differential pressure at design flow	250 kPa minimum	Acceptance requirement	The head the skid must deliver to the technology loop at design flow. Assumed for this study; on a real project it comes from the downstream loop resistance.
Circuit design pressure	1,000 kPa (10 bar)	Study assumption	Chosen to make the pressure-test acceptance value derivable rather than asserted.
Hydrostatic test pressure	1,500 kPa	Acceptance requirement	Derived as 1.5 × design pressure, the minimum hydrostatic leak-test pressure in ASME B31.3, subject to not exceeding the rating of any non-isolated component.
Circulation pumps	Two, N+1, each rated for 100% design flow, VFD driven	Study assumption	Redundancy configuration chosen so that pump changeover becomes a testable acceptance criterion.
Pump control mode	Constant differential-pressure setpoint on the secondary circuit	Study assumption	Determines what the controls and performance tests have to demonstrate.
Heat exchanger	Gasketed plate-and-frame, 3 K approach	Study assumption	Approach temperature chosen to be consistent with the 27 °C primary and 32 °C secondary assumptions above.
Secondary-side filtration	50 µm	Design context	OCP reports typical technology-cooling-system loop filtration of 25–50 µm, set by the most particle-sensitive component in the loop.
Primary-side filtration	250 µm	Design context	OCP reports typical facility-side filtration of 150–250 µm depending on design and customer requirements.
Wetted materials	Stainless steel piping, compatible elastomers, validated against the specified coolant	Design context	Material compatibility between wetted materials and the cooling liquid must be validated for all liquid-cooling components (OCP).
Instrumentation	Pressure, temperature and flow on both circuits; differential-pressure transmitter across secondary supply and return	Study assumption	Instrument set chosen so every performance acceptance criterion can actually be measured on the skid.
Controls	PLC with local HMI; BMS interface over Modbus TCP; alarm and interlock set; fail-safe state defined	Study assumption	Defines the scope of the controls section of the FAT.
Leak detection	Skid-level leak detection with alarm to the BMS and defined fail-safe action	Design context	Leak detection is named as a CDU function in OCP Cooling Environments documentation.
Expansion & make-up	Bladder expansion tank on the secondary circuit with low-level alarm	Study assumption	Included so the pressure-control scope of the FAT is complete.

Scroll table horizontally →

Section 2

Supplier qualification assessment

The method is defined before any score exists, because a scoring system invented after the conclusion is just the conclusion with arithmetic attached.

Eleven control areas are weighted to 100 points. Each is rated on a 0–5 maturity scale, and the weighted contribution is (score ÷ 5) × weight.

Maturity scale

Score	Level	Definition
5	Effective and demonstrated	Control is defined, applied consistently, and objective evidence of effectiveness was available during assessment.
4	Minor opportunities	Control is effective; isolated gaps in rigour or documentation that do not put conformity at risk.
3	Acceptable with gaps	Control exists and is broadly applied, but has identifiable gaps that could allow a nonconformity to escape.
2	Major weakness	Control is defined but not reliably applied, or is applied without evidence that it works.
1	Control largely absent	Little more than intent. No systematic application.
0	No evidence	No control and no evidence of any equivalent compensating practice.

Scroll table horizontally →

Weighted assessment — Northstar Thermal Systems

MQA-SQA-001 · Rev A

Control area	Weight	Score	Weighted	Finding	Observation
Quality Management System	15	5	15.0	—	Certified QMS with evident management review, internal audit programme and process ownership. Audit trail was complete for the sampled processes.
Manufacturing Process Control	15	4	12.0	minor	Work instructions released and in use at the stations sampled. Two assembly operations relied on operator knowledge for component selection rather than a verification step in the instruction.
Welding / Fabrication	10	4	8.0	minor	Qualified welding procedure specifications supported by procedure qualification records; welders currently qualified. One WPS range was broader than the production joints required, which is permissible but was not risk-assessed.
Incoming Material Control	10	5	10.0	—	Receiving inspection applied by material class with certificate review, and rejected material physically and systemically segregated. Demonstrated on a live receipt during the assessment.
Calibration / Measurement	10	2	4.0	major	Calibration is performed, but the recall system is a spreadsheet without escalation, three instruments in the test area were past due at the time of assessment, and there is no defined reaction when a measurement is found to have been taken with an out-of-calibration instrument.
Traceability	10	3	6.0	major	Material and pressure-test traceability to skid serial is intact. Weld traceability is not: welds are recorded by joint type rather than by joint identifier and welder, so a specific weld on a specific skid cannot be traced to the welder and WPS that produced it.
Pressure / Leak Testing	10	5	10.0	—	Documented hydrostatic procedure with defined hold time, calibrated gauges of appropriate range, and per-unit records retained. The strongest area in the assessment.
Cleanliness Control	5	5	5.0	—	Flushing procedure with defined acceptance, controlled capping of open connections, and a clean assembly area separated from fabrication.
Nonconformance / CAPA	5	2	2.0	major	Nonconformances are raised and dispositioned, but closure is recorded against the corrective action rather than against evidence that it worked. No effectiveness verification step exists, and two sampled NCRs described recurring conditions closed independently of one another.
Document Control	5	5	5.0	—	Drawing and procedure revisions controlled, superseded revisions withdrawn from the floor, and the revision in use at the station matched the released revision in every case sampled.
Final Inspection	5	5	5.0	—	Final inspection against a checklist tied to the drawing set, performed by someone independent of assembly, with records retained per unit.
Total	100		82.0		Computed from the area scores above — Conditional Approval

Scroll table horizontally →

Decision bands

Score	Status	What it means
90–100	Approved	Qualified for production award. Normal surveillance.
80–89	Conditional Approval ←	May proceed with defined conditions; every major finding must be closed with verified evidence before production release.
70–79	Corrective Action Required	No award until a corrective action plan is accepted and a reassessment is passed.
0–69	Not Approved	Not qualified. Requalification only after systemic remediation.

Scroll table horizontally →

Thresholds are illustrative for this study. On a production programme they would be defined by the owner’s supplier-quality framework, along with the rules for which findings are automatically disqualifying regardless of total score.

Engineering Decision

Supplier status

CONDITIONAL APPROVAL — 82 / 100

Basis

The supplier’s manufacturing fundamentals are strong: quality management system, incoming material control, pressure testing, cleanliness, document control and final inspection all scored as effective and demonstrated. The weaknesses are all in the systems that prove the work was done correctly rather than in the work itself — which is a common and recoverable pattern for a competent fabricator, and a genuine risk for equipment whose failure consequences are operational.

Three major findings: the calibration recall system has no escalation and instruments were found past due in the test area; weld traceability cannot resolve a specific weld to a welder and WPS; and nonconformance closure is recorded against the corrective action rather than against evidence that it worked.

Conditions before production release

Calibration system gap — Recall system with automatic escalation, all test-area instruments returned to current calibration, and a documented reaction plan for measurements taken on an out-of-calibration instrument.
Incomplete weld traceability — Weld records keyed to joint identifier, welder and WPS, with the joint identifiers shown on the fabrication drawings.
NCR closure effectiveness not demonstrated — Effectiveness verification added to the nonconformance process, with a defined interval or sample after corrective action before an NCR can be closed.

MQA-SQA-001 Rev A · September 2026 · synthetic assessment

Conditions and the evidence required to close each

Finding	Required before production	Closure evidence
Calibration system gap	Recall system with automatic escalation, all test-area instruments returned to current calibration, and a documented reaction plan for measurements taken on an out-of-calibration instrument.	Revised procedure, calibration status report for the test area showing zero past due, and one worked example of the reaction plan applied retrospectively.
Incomplete weld traceability	Weld records keyed to joint identifier, welder and WPS, with the joint identifiers shown on the fabrication drawings.	Revised weld record format plus a completed record for one production skid, traceable joint by joint.
NCR closure effectiveness not demonstrated	Effectiveness verification added to the nonconformance process, with a defined interval or sample after corrective action before an NCR can be closed.	Revised procedure and two NCRs closed under it showing the verification step and its result.

Scroll table horizontally →

Note what the evidence column does not accept: a revised procedure on its own. A procedure is a statement of intent. Each condition requires the procedure and a worked application of it, because the finding in every case was that a defined control was not producing evidence of effectiveness.

Section 3

How conditional approval becomes a released shipment

A reasonable reader will ask why equipment was ever released from a supplier that had three major qualification findings. The answer is that the findings were closed with verified evidence before production, and each subsequent gate carried its own evidence requirement. The chain is the argument.

Evidence gates — qualification to shipment

Supplier qualificationConditional approval — 82 / 100
Three major findings: calibration system, weld traceability, NCR closure effectiveness.
Qualification conditions closedVerified
Revised procedures plus worked evidence for each of the three findings, verified rather than accepted on submission.
Manufacturing readiness reviewReady with actions
Three open actions, each with an owner and a gate it blocks — none of them blocking fabrication start.
MRR actions closedVerified
Configuration list completed, sub-tier flow-down confirmed, one full calibration recall cycle evidenced.
Production and in-process surveillancePer ITP
Surveillance and witness points executed as planned; hold points released individually.
Factory acceptance testPass
All acceptance criteria met with recorded results; no punch item affecting a tested characteristic.
Documentation turnoverAccepted
Turnover index signed complete and traceable to the skid serial number.
Shipment releaseReleased
Release checklist satisfied in full, including preservation and damage-prevention controls.

Section 4

Manufacturing readiness review

Qualification establishes that a supplier is capable in general. Manufacturing readiness establishes that they are ready to build this unit — released engineering, available material, qualified people, a working inspection plan, and a test capability that can actually measure the acceptance values.

Of 21 items, 18 are ready and 3 carry an open action. The verdict below is computed from those statuses.

Engineering release

Readiness item	Status	Basis
Drawings approved and released at a controlled revision	Ready	Full drawing set released; revision list issued to manufacturing.
Bill of material released and matched to the drawing set	Ready	BOM reconciled against the released drawings line by line.
Specifications and referenced standards controlled	Ready	Specification register issued with the revision of each referenced document.
Software / firmware revision defined for PLC and instruments	Action	PLC application revision is defined; instrument firmware revisions are not yet recorded in the configuration list.

Scroll table horizontally →

Supply chain

Readiness item	Status	Basis
Critical components sourced from approved sub-tier suppliers	Ready	Pumps, heat exchanger, instrumentation and valves sourced from the approved list.
Long-lead material on site or confirmed to the build schedule	Ready	Heat exchanger and pump deliveries confirmed ahead of the fabrication start.
Sub-tier quality requirements flowed down	Action	Certificate requirements are flowed down; the weld traceability requirement from the qualification conditions is not yet in the sub-tier fabrication purchase order.

Scroll table horizontally →

Manufacturing

Readiness item	Status	Basis
Work instructions released for every operation in the routing	Ready	Instructions released; the two operations flagged during qualification now include a component verification step.
Tooling and fixtures available and verified	Ready	Fit-up fixtures and test connections available and checked against the drawings.
Operators qualified for their assigned operations	Ready	Qualification records current for assembly, brazing and electrical operations.
Welding procedures approved and welders currently qualified	Ready	WPS and supporting PQR in place for the production joints; welder qualifications current.

Scroll table horizontally →

Quality

Readiness item	Status	Basis
Inspection and test plan released and accepted by the supplier	Ready	ITP MQA-ITP-001 Rev A issued and accepted, including the owner hold points.
Inspection and test equipment calibrated	Action	Test-area instruments returned to current calibration per the qualification conditions; the escalating recall system is implemented but has not yet completed one full recall cycle.
Nonconformance process defined including notification to the owner	Ready	Revised NCR procedure includes effectiveness verification and owner notification for hold-point nonconformances.
Traceability method validated to skid serial number	Ready	Weld records now keyed to joint identifier, welder and WPS; validated on the first article.

Scroll table horizontally →

Test

Readiness item	Status	Basis
FAT equipment and test loop available	Ready	Test loop capable of design flow with adjustable resistance to set the operating point.
Test instrumentation calibrated and of adequate range and resolution	Ready	Flow, pressure, differential pressure and temperature instruments calibrated, with ranges appropriate to the acceptance values.
FAT protocol approved by both parties	Ready	MQA-FAT-001 Rev A approved, including pass / pass-with-deviation / fail definitions.

Scroll table horizontally →

Documentation

Readiness item	Status	Basis
Turnover documentation index defined and agreed	Ready	Index issued listing every document required for release, with the responsible party for each.
Certificates identified for material, pressure test and calibration	Ready	Certificate list agreed and mapped to ITP record requirements.
Test record templates approved before test execution	Ready	Templates approved so results are recorded against acceptance criteria rather than transcribed afterwards.

Scroll table horizontally →

Open actions

MQA-MRR-001 · Rev A

Ref	Action	Owner	Due	Gate it blocks
MRR-01	Record instrument firmware revisions in the configuration list and freeze the as-built configuration.	Supplier controls engineering	Before controls testing	Controls section of FAT
MRR-02	Add the weld traceability requirement to the sub-tier fabrication purchase order and confirm acceptance.	Supplier procurement	Before fabrication release	Fabrication start
MRR-03	Complete one full recall cycle on the calibration system and provide the resulting status report.	Supplier quality	Before FAT	FAT hold points

Scroll table horizontally →

Engineering Decision

Manufacturing readiness

READY WITH ACTIONS

Basis

Engineering is released, material is available, people and procedures are qualified, and the ITP is issued and accepted. The three open actions are all traceability- and calibration-related consequences of the qualification findings, and none of them blocks fabrication start — each is tied to a specific later gate instead.

Gate conditions

MRR-02 closed before fabrication release — weld traceability requirement in the sub-tier purchase order
MRR-01 closed before controls testing — instrument firmware revisions recorded and configuration frozen
MRR-03 closed before FAT — one full calibration recall cycle evidenced
No hold point released without its required record present and reviewed

MQA-MRR-001 Rev A · September 2026

Section 5

Inspection & test plan

The ITP is the document the supplier actually receives, so it has to be usable: every row states the characteristic, the reference it comes from, how it is verified, what constitutes acceptance, the sampling, who performs it, the owner’s involvement, and the record that has to exist afterwards.

Hold points are the expensive ones, so they are chosen deliberately — 3 of 20 steps. The rule I use: a hold point belongs where the evidence becomes unavailable or unaffordable to obtain later. A pressure boundary that gets insulated and enclosed, a performance figure that cannot be re-measured once the skid is crated, and a documentation package that loses its leverage the moment the equipment ships.

Owner quality points

Code	Point	Definition
R	Review	Owner reviews the record after the fact. Work proceeds without waiting; a finding is handled as a nonconformance.
S	Surveillance	Owner may observe at its discretion without notification. Work is not gated on the owner being present.
W	Witness	Owner is notified and may attend. If the owner does not attend by the agreed time, work may proceed and the record is reviewed instead.
H	Hold	Work may not proceed past this point without owner release. Chosen only where the evidence becomes unavailable or unaffordable to obtain afterwards.

Scroll table horizontally →

Inspection & test plan — LC-1000

MQA-ITP-001 · Rev A

Step	Operation	Characteristic	Reference	Method	Acceptance criteria	Sampling	Supplier	Owner	Records required
010	Material receipt — piping, fittings, plate	Material specification and grade	Drawing / BOM, material specification register	Certificate review against BOM plus receiving inspection for identification and damage	Material grade and heat identification match the certificate and the BOM; no transit damage	100% certificate review; 100% identification check	Perform and record	R	Material certificates, receiving inspection record
020	Material receipt — pumps, heat exchanger, instruments	Configuration and nameplate conformity	Purchase order, approved component data sheets	Nameplate and data-sheet verification against the purchase order, including impeller trim on each pump	Nameplate data, model and impeller trim match the purchase order and the approved data sheet	100% of major components	Perform and record	W	Component verification record with nameplate data, pump impeller trim record
030	Instrument calibration verification	Calibration status of test and process instruments	Calibration procedure; instrument list	Calibration certificate review; range and resolution check against the acceptance values to be measured	All instruments within calibration; range and resolution adequate for the acceptance criteria	100% of instruments used for acceptance measurements	Perform and record	R	Calibration certificates, instrument list with range and due date
040	Weld fit-up	Joint preparation, alignment, root gap	Approved WPS; fabrication drawings	Visual inspection against the WPS and drawing	Fit-up within the WPS range; joint configuration per drawing	Per supplier procedure	Perform and record	S	Fit-up inspection record by joint identifier
050	Welding	Procedure and personnel qualification; weld quality	Approved WPS supported by PQR; welder qualification records	Record verification plus visual examination; NDE where specified by joint class	Each weld produced under an approved WPS by a currently qualified welder; visual and NDE results meet the applicable acceptance criteria	100% visual; NDE per the joint class on the drawing	Perform and record	W	Weld record by joint identifier, welder and WPS; NDE reports
060	Piping assembly and supports	Configuration, routing, support and flexibility provisions	Piping drawings and support details	Dimensional and visual inspection against the drawings	As-built routing and supports per drawing; no unsupported spans beyond the detail	100% visual; dimensional on defined features	Perform and record	S	Assembly inspection record, as-built markup
070	Hydrostatic pressure test	Pressure integrity and leakage	Test procedure; design pressure from the specification	Hydrostatic test at 1.5 × design pressure with a defined hold period and calibrated gauges	No visible leakage and no unexplained pressure decay over the hold period	100% of pressure-retaining circuits	Perform and record	H	Pressure test record with gauge identification, pressure trace and hold time
080	Cleanliness and flushing	Internal cleanliness of both circuits	Flushing procedure; filtration specification	Flush to the defined endpoint; filter inspection; capping of open connections after flush	Flush endpoint achieved; no residual debris on filter inspection; all connections capped	100% of both circuits	Perform and record	W	Flushing record, filter inspection record, capping confirmation
090	Pump installation and alignment	Alignment and rotation	Pump manufacturer installation requirements	Alignment measurement; uncoupled rotation check	Alignment within the pump manufacturer tolerance; correct direction of rotation	100% of pumps	Perform and record	W	Alignment record per pump, rotation check record
100	Electrical assembly and wiring	Drawing conformity, terminations, protection and labelling	Electrical schematics and panel layout	Point-to-point verification; insulation and continuity checks; visual workmanship inspection	Wiring per schematic; insulation and continuity results within limits; all terminations labelled	100% point-to-point on control circuits	Perform and record	S	Wiring verification record, electrical test results
110	Configuration freeze	As-built configuration and software revision	Configuration list; released revision list	Record the as-built configuration including PLC application and instrument firmware revisions	As-built configuration recorded and matches the released revisions; any difference covered by an approved deviation	100%	Perform and record	R	As-built configuration record with software and firmware revisions
120	Instrument loop check	Sensor scaling and indication	Instrument index; control narrative	Loop check with a known input; verify indication at the HMI and over the BMS interface	Indicated value matches the applied input within the stated instrument accuracy across the range	100% of loops used for control, alarm or acceptance	Perform and record	W	Loop check record per instrument including scaling and applied values
130	Controls functional test — alarms and interlocks	Alarm, interlock and emergency-stop function	Functional specification; control narrative	Induce each condition and verify the annunciation, the interlock action and the fail-safe state	Every alarm and interlock operates as specified and returns to a defined safe state	100% of alarms and interlocks	Perform and record	W	Functional test record listing each condition, action and result
140	BMS communication test	Interface points and behaviour on loss of communication	Interface point list; control narrative	Verify each point end to end; simulate loss of communication	All points readable and correctly scaled; defined behaviour on communication loss	100% of interface points	Perform and record	W	Interface point verification record
150	Performance test — flow and differential pressure	Secondary circuit hydraulic performance	Specification: design flow and minimum differential pressure	Instrumented test at design flow with the test loop set to the design operating point	Differential pressure at or above the specified minimum at design flow, sustained over the defined period	100% of skids	Perform and record	H	Performance test record with flow, differential pressure, pump speed and instrument identification
160	Performance test — thermal and control response	Heat transfer and control stability	Specification: capacity, supply temperature setpoint	Instrumented test at design conditions; step change in load to observe control response	Supply temperature held within the specified band at design conditions; control response stable with no sustained oscillation	100% of skids	Perform and record	W	Thermal performance record, control response trace
170	Pump redundancy and changeover test	N+1 changeover behaviour	Control narrative; redundancy requirement	Fail the duty pump and verify automatic changeover while flow is maintained	Standby pump starts and design flow is re-established within the specified time, with correct annunciation	100% of skids, both directions	Perform and record	W	Changeover test record with time to re-establish flow
180	Leak detection test	Detection, annunciation and fail-safe action	Control narrative; leak detection specification	Simulate a leak at each detection zone and verify the response	Leak detected and annunciated at every zone; specified fail-safe action executed	100% of detection zones	Perform and record	W	Leak detection test record by zone
190	Final inspection and preservation	Workmanship, labelling, valve identification, preservation for shipment	Final inspection checklist; preservation and packing specification	Independent final inspection against the checklist; preservation and packing verification	Checklist complete with no open items; preservation and protection per specification	100%	Perform and record	W	Final inspection record, preservation and packing record, photographic record
200	Documentation turnover	Completeness of the turnover package	Turnover documentation index	Document review against the index, item by item	Every indexed document present, legible, at the correct revision and traceable to the skid serial number	100% of the index	Compile and submit	H	Turnover package with index and completion sign-off

Scroll table horizontally →

Why step 020 exists

Component configuration verification at receipt — including pump impeller trim — is a witness point. In the original version of this ITP it was not, and the companion 8D investigation is what put it there: a skid failed its performance test because an impeller one trim size too small was ordered, received and installed without any control catching it. That is the honest relationship between an ITP and reality — the plan improves when it fails.

Section 6

Factory acceptance test protocol

Factory acceptance is the last point at which a supplier still has both the obligation and the leverage to fix something. The protocol is therefore structured so that the cheap evidence is gathered before the expensive evidence is trusted: documentation and mechanical integrity before controls, controls before performance.

Document control

Document: MQA-FAT-001
Revision: A
Equipment: Northstar Thermal Systems LC-1000
Date: September 2026
Prepared by: Mohammad Al-Araidah
Status: Issued for independent engineering study
Classification: Synthetic data — not for production use

Preconditions — FAT does not start until all are satisfied

Drawings, specifications and the control narrative issued at the revisions the skid was built to, and those revisions recorded on the test record.
All test and process instruments within current calibration, with range and resolution adequate for the acceptance values being measured.
As-built configuration frozen and recorded, including PLC application and instrument firmware revisions.
Hydrostatic pressure test completed and accepted; hold point released.
Cleanliness and flushing completed and accepted.
Test loop configured to the design operating point, with the means to set and hold design flow.
Nonconformance status declared: every open NCR listed with its disposition, and no open NCR affecting a characteristic under test.
Test record templates approved before execution, so results are recorded against acceptance criteria rather than transcribed afterwards.

1 — Visual and workmanship inspection

Establish that the skid as presented matches the released documentation and can be operated and maintained.

01Skid nameplate data verified against the purchase order and the specification.
02General workmanship: supports, fasteners, pipe routing, insulation, absence of mechanical damage.
03Valve identification and tag numbers present and consistent with the P&ID.
04Instrument tagging present and consistent with the instrument index.
05Enclosure condition, ingress protection provisions, and cable entry workmanship.
06Access and maintainability: filter, strainer and pump access without removing unrelated components.
07Drain and vent provisions present and operable.

2 — Mechanical verification

Confirm the pressure boundary and the rotating equipment are correct before any performance number is trusted.

01Hydrostatic test record reviewed: test pressure at or above 1.5 × design pressure, hold time achieved, calibrated gauge identified.
02No visible leakage at any joint under operating pressure after filling.
03Pump rotation direction verified.
04Pump alignment record reviewed against the manufacturer tolerance.
05Vibration measured at each pump at design duty and compared against the agreed limit.
06Filter and strainer elements confirmed installed to the specified rating.
07Expansion tank pre-charge verified and low-level alarm function confirmed.

3 — Controls verification

Prove that the protective functions work, including on failure — the section most often reduced to a demonstration rather than a test.

01Instrument loop check: applied input versus indication at the HMI and over the BMS interface, across the instrument range.
02Differential-pressure transmitter scaling verified against an applied reference at two points spanning the operating range.
03Every alarm induced and its annunciation verified, at the HMI and at the BMS.
04Every interlock induced and its action verified, including the resulting equipment state.
05Emergency stop operated and the resulting state verified.
06BMS interface point list verified end to end; behaviour on loss of communication verified against the narrative.
07Fail-safe state verified on loss of power and on loss of control signal.
08Parameter security verified: setpoints cannot be altered without the defined authorization.

4 — Performance verification

Demonstrate the skid meets its hydraulic, thermal and redundancy requirements at the design operating point.

01Design flow established on the secondary circuit and held for the defined period.
02Differential pressure recorded at design flow and compared against the specified minimum.
03Pump speed and absorbed power recorded at the design operating point.
04Thermal performance recorded at design conditions; supply temperature compared against the setpoint band.
05Control response to a step change in load recorded; stability assessed for sustained oscillation.
06Pump redundancy changeover tested in both directions, with time to re-establish design flow recorded.
07Leak detection tested at every zone with the fail-safe action verified.
08Test repeated for confirmation where a result falls within the measurement uncertainty of the acceptance limit.

5 — Documentation verification

Confirm the record the owner will actually receive is complete and traceable, before the skid leaves.

01As-built drawings issued at the revision matching the skid.
02Material certificates present and traceable to the skid serial number.
03Weld records present by joint identifier, welder and WPS; NDE reports attached where required.
04Calibration certificates for every instrument used in acceptance measurements.
05Pressure test, flushing, alignment and electrical test records present.
06Completed FAT record with results recorded against each acceptance criterion.
07Punch list issued with severity, owner and required closure evidence for each item.
08Operation and maintenance documentation and spare parts list provided.

Disposition definitions

Verdict	Definition	Consequence
PASS	Every acceptance criterion met, with results recorded. Punch list contains no item affecting a tested characteristic.	Proceed to documentation turnover and shipment release.
PASS WITH DEVIATION	All criteria met except one or more where an approved engineering deviation exists, granted before the test on a stated technical basis — not a result renegotiated afterwards.	Proceed with the deviation recorded in the release package and carried into the site records, so the as-accepted condition is known for the life of the equipment.
FAIL / HOLD	One or more acceptance criteria not met with no approved deviation, or the evidence required to judge a criterion is unavailable.	Shipment placed on hold. Nonconformance raised, root cause established, corrective action verified, and the affected tests repeated before the disposition is revisited.

Scroll table horizontally →

Measurement uncertainty

Where a measured value falls within the measurement uncertainty of an acceptance limit, the criterion is not met by default. The test is repeated, and if the result remains within uncertainty the disposition is escalated to engineering rather than resolved by rounding.

The distinction that matters most here is between pass with deviation and a renegotiated result. A deviation is an engineering judgement made in advance, on a stated technical basis, by someone with the authority to accept the consequence. Reinterpreting an acceptance criterion after seeing the measurement is not a deviation — it is the removal of the criterion, and it is how equipment that nobody approved ends up in service.

Section 7

Shipment release

Release is a decision, not an administrative step. Everything below has to be true, and each requirement is paired with the evidence that satisfies it — because a checklist without an evidence column becomes a list of things people believe.

Shipment release checklist

MQA-SRC-001 · Rev A

Requirement	Evidence
FAT completed with a PASS or an approved PASS WITH DEVIATION disposition	Signed FAT record with results against every acceptance criterion
All hold-point nonconformances closed	NCR records showing corrective action, objective evidence and effectiveness verification
All deviations formally approved before release	Approved deviation requests with the technical basis and the approving authority recorded
Punch list dispositioned — closed, or accepted for site closure with an owner and date	Punch list with per-item status and the evidence type defined at logging
Serial numbers recorded and traceable to the record set	Skid and major-component serial numbers on the turnover index
As-built drawings issued at the revision matching the skid	Drawing list with revisions, reconciled against the as-built configuration record
Operation and maintenance documentation complete	O&M manuals and spare parts list received and reviewed
Certificates present and traceable	Material, pressure test and calibration certificates traceable to the skid serial number
Test reports reviewed and accepted	Pressure test, flushing, alignment, electrical, loop check, functional and performance records
Preservation and protection complete for the shipping mode and duration	Preservation record: drained or treated as specified, connections capped, desiccant where required
Damage-prevention controls verified	Packing and bracing inspection, lifting-point marking, shock or tilt indicators where specified
Turnover documentation package accepted	Turnover index signed complete, item by item

Scroll table horizontally →

Engineering Decision

Shipment status

RELEASED

Basis

FAT passed with all acceptance criteria met and results recorded. All hold-point nonconformances closed with verified evidence. No open deviations. Turnover documentation accepted against the index, item by item, and preservation and damage-prevention controls verified for the shipping mode and duration.

What would have reversed this

Any acceptance criterion unmet without an approved, pre-existing deviation
Any hold-point nonconformance closed without effectiveness evidence
A performance result sitting within measurement uncertainty of the limit and unresolved on repeat
An incomplete turnover package — the leverage to obtain it disappears once the equipment ships
Preservation inadequate for the transit mode and duration

MQA-SRC-001 Rev A · September 2026 · synthetic study

Section 8

Engineering judgment

What decision had to be made?

Two decisions, and the second is the harder one. First: should this supplier be qualified at all, at 82 of 100 with three major findings? Second: can equipment from that supplier be released to a mission-critical site?

What evidence mattered?

For qualification, the pattern of the findings mattered more than the total. All three were in evidence systems — calibration recall, weld traceability, nonconformance effectiveness — rather than in fabrication capability, which is remediable within a programme timeline and does not require the supplier to become a different company. Had the low scores been in welding, pressure testing or process control, the same 82 would have been a different answer.

For release, the evidence is the closure chain: qualification conditions verified with worked examples rather than accepted on submitted procedures, MRR actions closed against the gates they blocked, hold points released individually against their records, FAT passed on recorded results, and a complete turnover package.

What would cause me to change the decision?

A total score is not a release criterion, so a change in it would not change this decision — but three specific things would. An unclosed calibration finding at FAT would invalidate the measurements the release depends on, which means the performance evidence does not exist regardless of what the numbers say. Weld traceability still open would mean the pressure boundary cannot be substantiated joint by joint, which is unacceptable for equipment whose failure mode is a leak in a data hall. And an incomplete turnover package holds shipment on its own, because it is the only requirement whose leverage vanishes entirely the moment the crate leaves.

More generally: I would decline to release on schedule pressure alone, and I would accept a deviation — with an owner engineering approval and a recorded technical basis — where a criterion is not met but the consequence is understood and bounded. The difference between those two is whether someone with authority made a decision, or whether nobody did.

Section 9

Engineering documents

Generated from the data on this page. Each carries a document-control header and is marked as synthetic study material.

MQA-SQA-001 · Rev A

XLSX

Supplier Qualification Assessment — Northstar Thermal Systems

Weighted maturity assessment across eleven control areas, with live scoring formulas and decision bands.