A successful straightening sample is not simply a part that looks straight after one press. The test must prove that an agreed workpiece family can be measured, corrected and accepted without unacceptable damage, under conditions that can be reproduced in production.
This guide separates early feasibility trials, engineering sample validation, factory acceptance testing (FAT) and site acceptance testing (SAT). It also explains what data is needed before a supplier can convert a promising sample into a defendable machine guarantee.
*Engineering concept illustration.* It demonstrates a possible measurement role; the approved gauge and support condition belong in the test protocol.
Four Different Validation Stages
| Stage | Main Question | Typical Output | What It Does Not Prove Alone |
|---|---|---|---|
| Feasibility trial | Can this workpiece be measured and corrected without obvious damage? | Initial method, 基準, tooling and risk assessment | Production capability or final cycle time |
| Engineering sample test | What process window works across representative parts? | Sample matrix, recipes, before/after data and limits | Complete production machine readiness |
| Factory Acceptance Test (FAT) | Does the built machine meet the agreed factory test specification? | Witnessed FAT protocol, 結果, open-item list and release decision | Performance after transport and site installation |
| Site Acceptance Test (SAT) | Does the installed machine perform in the customer’s real environment? | Site results, gauge correlation, training and final acceptance | Conditions outside the agreed site and product scope |
Do not use one successful feasibility sample as evidence that every model, batch, incoming bend and production condition is guaranteed.
Freeze the Measurement Definition First
The test cannot be valid until both parties agree what “straight” means. 直率, 圓跳動, total runout and shop-floor TIR are different characteristics and may produce different values.
Standards and Project Acceptance Criteria Have Different Jobs
ISO 14253-1 and ASME B89.7.3.1 provide frameworks for making conformity decisions when measurement uncertainty matters. They do not create the workpiece tolerance, choose the controlling gauge, set a universal sample count or Gage R&R limit, or decide which party accepts a project deviation.
Those project-specific criteria must come from the approved drawing, contract and signed test protocol. Before any result is classified as pass or fail, the protocol should identify the requirement source, 測量方法, acceptance limit, uncertainty-aware decision rule, sample scope and authorized disposition path.
Before testing, define:
- controlled characteristic and drawing tolerance;
- 適用的圖紙標準和版本;
- datum features or support reference;
- probe surface, axial stations and probe path;
- free-state or clamped-state condition;
- rotation speed, filtering and reading rule where relevant;
- excluded features such as keyways, 凹槽, teeth or holes;
- customer reference gauge and machine gauge;
- acceptance decision near the tolerance limit.
For a long or flexible workpiece, freeze the complete support and release condition rather than recording only “two supports” or “measured on centers.”
| Measurement-Control Item | 所需記錄 |
|---|---|
| Support arrangement | Type, quantity, axial coordinates, height, end overhang and permitted degrees of freedom |
| Workpiece state | Under correction load, press retracted on machine supports, removed and repositioned, released-state or customer functional setup |
| Characteristic and reference | Drawing characteristic, datum or functional reference, measured surface and axial stations |
| Measuring system | Gauge, 探測, contact force, rotation/sampling method, filtering and evaluation rule |
| 規格相關性 | 偏見, variation, qualified product range, decision rule and conditions requiring requalification |
Self-weight deflection under a defined support condition must not be silently treated as permanent bend. Additional supports or a calculated compensation method are acceptable only when the test protocol defines them and representative samples validate that they do not mask the geometry being accepted.
看 軸直線度、跳動和 TIR for the measurement distinctions that must be resolved before the test protocol is signed.
Build a Representative Sample Matrix
Sending a few convenient parts creates weak evidence. The sample set should cover the real product envelope and the conditions most likely to fail.
| Sample Dimension | Minimum Coverage Question |
|---|---|
| Geometry | Shortest/longest, smallest/largest diameter, thinnest wall, most complex feature set |
| 材料 | Lowest/highest strength or hardness, relevant alloy and heat-treatment batches |
| Incoming deformation | Normal distribution, worst expected bend, multiple-bend and torsion cases |
| Surface condition | Raw, turned, hardened, 地面, coated or otherwise finished states |
| Functional features | Gear, 樣條線, thread, 鍵槽, flange, radial hole, weld or thin-wall section |
| Tolerance | Normal parts plus samples near the final acceptance boundary |
| Production model | Every recipe/tooling family or justified worst-case representatives |
| Handling | Loading orientation, mixed models and parts most difficult to separate or locate |
The matrix should record how each sample was selected. A “worst case” without a drawing, material record or measured incoming value is not traceable evidence.
A representative engineering sample matrix is not automatically a statistical lot-acceptance plan. If the result will be used to accept a production lot, estimate a nonconforming fraction or make a capability claim, the protocol must define the population, selection method, statistical objective, required precision and decision risk. ASTM E105 and ASTM E122 provide general sampling frameworks, but neither supplies one universal sample count for straightening projects.
Pre-Test Readiness Checklist
Acceptance Gate Before the First Sample
| Gate Item | Must Be Frozen in the Protocol |
|---|---|
| Requirement source | Drawing, specification or contract clause, including revision |
| Measurand | Characteristic, 基準, 表面, stations and part state |
| Acceptance limit | Project tolerance or other contractual criterion |
| Measurement system | Controlling gauge, fixture, 範圍, resolution and method |
| Metrological evidence | Calibration scope/status, traceability chain and relevant uncertainty |
| Decision rule | How uncertainty and results near a limit are handled |
| Sample scope | Product family, sample selection basis and any statistical objective |
| Part integrity | Prohibited marks, cracks, ovality or functional damage |
| Data completeness | Required raw readings, setup records and retained evidence |
| Authority and disposition | Who can pass, reject, waive, review or require retest |
If a required gate item is still open, the result may support engineering learning, but it should not be presented as contractual acceptance.
Workpiece Information
- Approved drawing and revision are available.
- 材料, 硬度, heat-treatment state and process stage are recorded.
- Incoming bend or runout is measured before correction.
- Protected support, measuring and pressing/rolling zones are marked.
- Samples are identified individually and cannot be mixed after testing.
Measurement Readiness
- Customer and supplier agree the datum and characteristic.
- Calibration evidence covers the relevant measurand, range and test date.
- The measurement result has a documented traceability chain and an uncertainty evaluation appropriate to its intended acceptance use; a calibration sticker alone is not sufficient evidence.
- Fixtures, 中心, contacts and probe surfaces are clean and repeatable.
- Repeat readings have been taken to expose unstable setup or operator influence.
- A correlation method is defined before comparing machine and customer results.
Process Readiness
- The selected press or roller concept matches the workpiece geometry.
- Initial stroke, 力量, roll gap, angle or other limits are conservative.
- Safety, guarding and abnormal-part reaction are ready for the test.
- The data sheet and pass/fail rule are agreed before the first result is seen.
For method selection, review 壓力矯正與滾筒矯直.
Recommended Sample-Test Sequence
1. Identify and Inspect the Sample
Record sample ID, drawing revision, material/batch, process stage, dimensions, hardness where relevant and visible surface condition. Photograph or document pre-existing damage so it is not confused with straightening marks.
2. Measure the Incoming Condition
Use the agreed fixture, datum and stations. Repeat at least enough measurements to identify unstable seating or obvious gauge variation. Record the full result, not only the single worst point.
3. Apply a Controlled Initial Correction
Begin inside a conservative process window. For point press straightening, record support positions, press location, angular position, stroke and force where available. For roller straightening, record roll set, gap, angle, 速度, guide and pass information.
*Engineering concept illustration.* It depicts a correction stage; proof of process limits requires recorded sample-test results.
4. Remeasure with the Same Method
Use the same datum, fixture and measuring stations. If a second correction is allowed, the acceptance plan should define maximum iterations, stroke/force or pass limits instead of letting the operator continue until a favorable number appears.
5. Inspect Part Integrity
Check contact marks, cracks, 凹痕, 橢圓度, thread/tooth damage, coating condition and other functional risks. A part that meets runout but has a damaged bearing journal is not acceptable; the same principle is critical for a thin-wall tube straightening application, where collapse or section damage can invalidate an apparently improved geometry result.
6. Record Cycle and Handling Evidence
Separate automatic machine time, loading/unloading, 測量, correction iterations, model change and manual inspection. A short edited video is useful visual evidence but is not a statistically valid cycle-time study.
The video shows one automatic small-shaft loading and press-straightening sequence. It demonstrates equipment flow only; a final cycle guarantee requires a defined start/end point, representative parts and repeated timed observations.
7. Classify the Result
Use one of the pre-agreed outcomes: pass, fail, engineering review or retest after a documented correction to the setup. Do not silently remove failed samples from the report.
*Engineering concept illustration.* Actual sorting logic, traceability and acceptance routing must be defined by the approved test protocol.
Minimum Per-Part Data Record
| Record Field | 為什麼它很重要 |
|---|---|
| Sample ID and drawing revision | Preserves traceability |
| 材料, batch and hardness/process state | Explains springback and crack-risk variation |
| Incoming values at all stations | Defines the actual challenge |
| 基準, fixture and gauge | Makes the result reproducible |
| Recipe/tooling version | Connects the result to machine settings |
| Correction count, positions and limits | Shows how the result was achieved |
| Final values at all stations | Prevents selective reporting |
| Surface/integrity inspection | Confirms the part remains usable |
| Automatic and total cycle observations | Separates machine speed from complete handling time |
| Final disposition | Pass, fail, review or retest with reason |
Gauge Correlation and Decision Rules
Two gauges can be individually repeatable and still disagree because they realize the datum differently or evaluate different surfaces. Correlation should include good, borderline and bad samples across the model range.
| Correlation Item | Agreement Needed |
|---|---|
| Datum realization | 中心, journals, V-blocks, functional gauge or CMM construction |
| Measuring points | Same axial stations, surfaces and excluded zones |
| Part condition | 溫度, 清潔度, free/clamped state and elapsed time |
| Reading method | 峰峰值, fitted axis, 過濾和舍入 |
| Repeatability | Repeated readings on both systems |
| Borderline result | 保護帶, recheck or engineering disposition rule |
| Reference method | Which gauge controls FAT, SAT and final contractual acceptance |
When a measured value is close to the tolerance limit, measurement uncertainty and the agreed conformity decision rule matter. The quotation and protocol should state the rule instead of assuming every displayed value can be treated as exact.
Run paired comparisons on identified parts across the qualified product range, including results near the acceptance boundary. Predefine the comparison metric, the project limit for disagreement and the disposition when the two systems do not agree. There is no universal correlation or Gage R&R percentage that can be inserted into every straightening project without considering tolerance, measurement risk and intended use.
The machine gauge may control the correction loop, but it should not replace a different customer released-state method by assumption. If the machine result is intended to control contractual acceptance, the correlation record must define its bias, variation, qualified range and project-specific requalification triggers. Correlation with another gauge does not by itself establish metrological traceability for either measurement result.
From Sample Test to Machine Specification
A validated sample test should change the proposal from generic marketing language into an engineering specification.
| Sample Evidence | Specification Output |
|---|---|
| Proven workpiece envelope | Covered model/size/material family |
| Agreed datum and gauge correlation | Machine measurement and reference acceptance method |
| Correction response | Press/roll capacity, resolution and process limits |
| Protected feature study | 支持, measuring and correction tooling design |
| Cycle observations | Defined takt basis and excluded manual activities |
| Failure behavior | NOK route, stop limits and operator/engineering review |
| Data needs | Traceability fields, storage and interface scope |
| Changeover test | Tooling, recipe and setup-time requirements |
Anything not tested should remain an assumption, option or open engineering item—not a guaranteed performance statement.
Factory Acceptance Test (FAT)
FAT is performed at the supplier’s factory against an agreed protocol before shipment. The exact scope must be adapted to the machine and contract.
The applicable standard or guidance can structure measurement and conformity decisions, but the signed contract and FAT protocol supply the actual machine requirements, acceptance values, witness responsibilities and shipment gate.
FAT Scope
- machine configuration matches approved drawings and purchase specification;
- guards, interlocks, emergency stops and defined safety functions are tested;
- utilities, axes, 感應器, press/roll functions and abnormal alarms operate;
- agreed workpiece models and changeovers are demonstrated;
- measurement correlation and acceptance logic are checked;
- representative production run and cycle measurement follow the protocol;
- 挪威克朗行動, data recording and interfaces are exercised;
- manuals, drawings, backups, spare parts and training records are reviewed;
- deviations are listed with owner, action and closure evidence.
A FAT should not be reduced to “the machine ran successfully.” Each acceptance point needs an observable test, measurable criterion, recorded result and disposition.
Site Acceptance Test (SAT)
SAT verifies the installed system after transport, installation and connection to the customer’s real utilities, material flow and quality process.
The SAT acceptance boundary is likewise project-specific. It should verify only the agreed site-dependent requirements and approved FAT carry-over items unless the parties formally change the contract.
SAT Adds Site-Specific Evidence
- foundation, level, utilities and environmental conditions;
- upstream/downstream equipment and real loading logistics;
- customer production material and approved recipes;
- customer reference gauge and local quality workflow;
- network, MES or traceability interfaces where included;
- 操作員, maintenance and quality training;
- production-like run under the agreed site conditions;
- closure of FAT open items affected by installation.
SAT is not a second opportunity to introduce new requirements that were absent from the contract. Changes should follow a documented variation process.
FAT/SAT Acceptance Table Template
| Test Item | 方法 | Acceptance Criterion | Evidence | 結果 / Disposition |
|---|---|---|---|---|
| Workpiece measurement | Agreed gauge and stations | Defined tolerance and decision rule | Before/after data sheet | Pass/fail/open |
| Surface integrity | Visual or specified inspection | No prohibited damage | Photos/inspection record | Pass/fail/open |
| Cycle time | Defined start/end and repeated runs | Contract value under stated conditions | Timestamped run log | Pass/fail/open |
| Model change | Approved sequence and tooling | Required setup and correct recipe | Witness record | Pass/fail/open |
| 挪威克朗行動 | Challenge part or simulated fault | Correct stop/sort/alarm response | Event log/video | Pass/fail/open |
| Safety function | Approved safety test procedure | Required response achieved | Signed checklist | Pass/fail/open |
| 可追溯性 | Test part and data export | Required fields stored/transferred | File/interface record | Pass/fail/open |
Managing Deviations and Open Items
Every deviation should have:
- unique ID and requirement reference;
- observed condition and objective evidence;
- impact on safety, quality, delivery and production;
- temporary disposition if operation continues;
- responsible owner and target date;
- corrective action and retest method;
- closure approval by the authorized parties.
Do not convert an unresolved performance failure into a vague “optimization after shipment” note. Shipment and final acceptance gates should state which open items are allowed and which are blocking.
Common Testing Mistakes
- Testing only nominal parts and excluding worst-case samples.
- Changing the datum or gauge after seeing an unfavorable result.
- Reporting only successful parts or the best final value.
- Mixing machine cycle time with complete operator handling time.
- Using customer tolerance as if it were already demonstrated capability.
- Ignoring contact marks, cracks, ovality or feature damage.
- Treating one video or one sample as production validation.
- Leaving FAT/SAT criteria until the machine is already built.
- Accepting displayed values near the limit without an agreed decision rule.
Information to Send for a Sample Study
請提供:
- drawings, revisions and model matrix;
- 材料, hardness and heat-treatment condition;
- incoming bend/runout distribution and known worst cases;
- final characteristic, 寬容, datum and inspection work instruction;
- protected surfaces and allowed support/correction zones;
- production stage, volume, takt and loading requirements;
- representative samples covering the full envelope;
- required traceability and data interface;
- target FAT/SAT location, witness needs and documentation language.
常問問題
How many samples are enough?
There is no universal number. For engineering validation, the set must cover geometry, 材料, incoming deformation, tolerance boundary and production variation. For statistical lot acceptance or estimation, sample size must follow a documented objective, population, selection method, precision and risk basis. A small feasibility set can select a concept, but it cannot establish broad production capability.
Can the supplier use its own gauge?
是的, if the method is agreed and correlated with the customer’s reference method. The protocol must state which result controls acceptance and how borderline disagreements are handled.
Does a passed FAT mean the machine is finally accepted?
Only according to the contract. FAT proves the agreed factory scope. SAT may still be required to verify installation, utilities, interfaces, customer material and site operation.
Should failed samples stay in the report?
是的. They define the process boundary and help distinguish machine limits, sample defects, setup problems and measurement disagreement.
Can cycle time be guaranteed from one run?
不. Define the cycle start/end, part condition, correction distribution, loading and inspection scope, then use repeated observations across representative samples.
What happens if the customer changes the drawing after testing?
The new revision requires an impact review. Changes to tolerance, 基準, 幾何學, material or protected zones may require new samples, tooling, software or commercial scope.
Turn Sample Evidence into an Acceptance Plan
The purpose of a sample test is not to create a favorable demonstration. It is to discover the reproducible process window, measurement agreement, risks and limits before they become contractual obligations.
Technical references used for the measurement and sampling boundaries in this guide:
- ISO 14253-1:2017 — Decision rules for verifying conformity or nonconformity with specifications
- ISO/IEC Guide 98-4:2012 — Role of measurement uncertainty in conformity assessment
- ASME B89.7.3.1 — Decision rules considering measurement uncertainty
- NIST — Metrological Traceability: Frequently Asked Questions and NIST Policy
- NIST Technical Note 1297 — Evaluating and Expressing Measurement Uncertainty
- ASTM E105 — Standard Guide for Probability Sampling of Materials
- ASTM E122-17(2022) — Sample Size With Specified Precision
Contact StraighteningTech 和你的圖畫, measurement method and representative sample matrix. Our solution team can prepare a feasibility study and project-specific sample, FAT and SAT plan.