High-Traffic Format · Technical Perspective · Published: October 29, 2026
Table of Contents
Key Takeaways
- A pump or valve flow figure is a promise made under specific conditions, usually clean air-free water at a reference temperature, and every published curve carries a tolerance band. ISO 9906 Grade 1B permits ±5% on flow and ±3% on head at the guarantee point [1].
- The loosest common grade, 3B, allows ±9% on flow and ±7% on head, so the same “rated flow” can describe very different machines depending on which grade the purchase order accepts [1].
- ISO 9906 and ANSI/HI 14.6 define six acceptance grades: 1U, 1E, 1B, 2U, 2B and 3B. The Hydraulic Institute published a white paper on choosing between them precisely because buyers who stay silent inherit a default [2].
- An electromagnetic flowmeter rated at ±0.5% needs roughly 5 diameters of straight pipe upstream and 2 downstream. Feed it a distorted profile and it can contribute 2–5% error, rising sharply when the upstream fitting is a partly closed valve.
- Volumetric verification costs almost nothing. A 10-litre bucket resolves flows up to about 4 L/s, and a 200-litre drum reaches about 50 L/s [3].
- Tracer dilution with salt or dye measures flow in channels and outfalls where no meter can be installed, by injecting a known concentration and reading the diluted result downstream [4].
Every purchasing dispute over flow starts the same way. The datasheet said one number, the plant delivers another, and both parties are describing something true.
The vendor is quoting a test-bench curve taken under controlled conditions. You are running warm, dirty, viscous or aerated water through real pipework against a real static head. The gap between those two worlds is not necessarily dishonesty. It is tolerance, test conditions and installation effects that nobody wrote down.
What follows is how to close that gap before money changes hands, using the standards the pump test labs work to and the verification methods water utilities have trusted for a century.
1. Every Datasheet Number Has an Address
A flow figure without its conditions is an address without a city. Before comparing two vendors, find out where each number lives.
- Test medium and temperature. Hydraulic performance curves are conventionally established with airless water at about 20 °C, because viscosity and density move the curve. If your service is 60 °C brine, the cold-water curve is an optimistic starting point, not a commitment.
- The point on the curve. Maximum flow at zero head, best-efficiency-point flow and flow at rated duty head are three different numbers. Marketing documents tend to quote whichever is largest.
- Speed and impeller trim. A curve valid at rated speed with a full-diameter impeller says very little about the machine that arrives with a trimmed impeller, or about the same machine on a variable-frequency drive at reduced speed.
- Tolerances. Covered next. Even a laboratory-correct test result is allowed to miss the guarantee point by a defined band.
Vendors who publish conditions and grades openly make this step trivial. Shanghai ChiMay’s rule on its softening and filtration valves is to publish the numbers other suppliers make you ask for, the same discipline behind a procurement specification note for softening-and-filtering valves that states inlet tolerances, regeneration control modes and duty limits in contract-ready language. When a vendor does not, the four questions above are your script.
2. Curves Come With Tolerances — Negotiate the Grade
Pump acceptance testing is standardized on both sides of the Atlantic. ISO 9906 and its US counterpart ANSI/HI 14.6 define how the performance test runs and how far the result may sit from the guarantee point. Six grades are available, from Grade 1 (tightest) to Grade 3:
| Acceptance grade | Flow tolerance at guarantee point | Head tolerance at guarantee point |
|---|---|---|
| 1B (bilateral, Grade 1) | ±5% | ±3% |
| 2B (bilateral, Grade 2) | ±8% | ±5% |
| 3B (bilateral, Grade 3) | ±9% | ±7% |
Values from the ISO 9906 / ANSI/HI 14.6 acceptance-grade tables [1]. Unilateral grades (1U, 2U and 1E in its own way) trade the negative side of the band for a wider positive side, protecting minimum performance only. In China the same standard text arrives as GB/T 3216-2016, which is identical to ISO 9906:2012, so a grade quoted locally means the same thing as one quoted in Europe.
Two practical consequences follow.
First, a pump that tests at −5% flow in Grade 1B has passed. If you need the full design flow for cooling capacity or for backwash velocity, a Grade 3B purchase with a −9% leg can leave your process short while the machine has technically failed nothing.
Second, the grade is negotiable and belongs in the purchase order. The Hydraulic Institute’s white paper on specifying acceptance grades exists because buyers who stay silent inherit whatever grade the vendor’s quality system finds convenient [2].
Write into the contract: the guarantee points (flow, head, power at duty), the acceptance grade, the test standard and edition, whether the test is witnessed, and whether the certificate reports measured points or simply states “conforming.”
3. Prove It in the Field: Three Methods That Don’t Take the Vendor’s Word
3.1 Meter comparison — done properly
Comparing delivered flow against a plant flowmeter is the fastest check, but the meter is on trial too. An electromagnetic flowmeter specified at ±0.5% of rate can easily contribute 2–5% error if the profile reaching it is distorted. The minimum straight-run convention is about 5 pipe diameters upstream and 2 downstream, climbing to 25 diameters or more when the upstream element is a partly open valve or a double-plane elbow.
Get full-pipe conditions and correct grounding right, and the comparison means something. Skip them and you have swapped the vendor’s tolerance for your own plumbing’s.
3.2 The bucket method — primitive and almost unimpeachable
Volumetric measurement is the oldest trick in hydraulics: divert the whole stream into a container of known volume and time the fill. Field guidance is refreshingly concrete. A 10-litre bucket covers flows up to about 4 L/s, a 200-litre drum covers up to about 50 L/s, and the fill time needs care when it is only a few seconds [3]. Repeat five to seven times and average.
3.3 Tracer dilution — where no meter can go
For open channels, mixed-flow conduits or outfalls, tracer dilution is the accepted method. Inject a known, strong concentration of salt or dye at a constant rate, measure the fully mixed diluted concentration downstream, and the dilution ratio gives you the flow [4]. It needs a proper mixing length and an inert tracer, but it turns an unmeasurable stream into a measurable one using laboratory-grade chemistry.
Judge results against the right yardstick. A field check inside the contractual tolerance band is a pass. Landing exactly on the datasheet number was never the standard you bought. A shortfall beyond the band, on the other hand, is a documented finding with a remedy attached, which is the whole reason the grade matters.
4. Put the Test in the Contract: FAT and SAT
Verification is cheapest before shipment and most expensive after handover, which is why acceptance testing has two gates.
- Factory Acceptance Test (FAT): the package (pump skid, dosing system, treatment train) is proven at the works before dispatch, on water or a simulant, with the purchaser present or represented. A dosing-skid FAT typically covers construction against approved drawings, instrument calibration, sequence operation and performance against specification [5].
- Site Acceptance Test (SAT): the installed plant is proven on site with real power, real water and real interfaces, confirming installation quality, I/O and control sequences before start-up [5].
For flow claims, the FAT proves the machine against the curve; the SAT proves the system against the process. A pump can pass its FAT and still disappoint on site, because friction losses, static head or suction conditions differ from the datasheet’s assumptions. The SAT is where that mismatch surfaces while it is still the vendor’s problem.
Clauses that earn their keep:
| Commitment on the datasheet | Verification method | Acceptance criterion | Contract clause to include |
|---|---|---|---|
| Rated flow at duty head | Works test per ISO 9906 / HI 14.6 | Flow and head within the agreed grade band at guarantee point [1] | Test standard, grade, witnessed FAT, measured-point certificate |
| System flow at process conditions | SAT with real water and a verified meter (5D/2D straight run) | Delivered flow meets process requirement within tolerance | SAT acceptance values; remediation if short |
| Flow accuracy of bundled metering | Bucket test against meter [3] | Meter within its stated class | Meter class, verification procedure at SAT |
| Open-channel or outfall delivery | Tracer-dilution measurement [4] | Flow within agreed uncertainty | Method and mixing length agreed pre-order |
| NPSH margin at duty | NPSH test (3% head-drop criterion) | Margin above required NPSH stated and met | Minimum suction conditions guaranteed |
| Long-term performance | Post-commissioning trend review | Performance retained at agreed retest interval | Retest at 3–6 months; remedies defined |
5. Common Inflation Patterns — and the Questions That Deflate Them
- Best-efficiency-point flow quoted as duty flow. Ask which point on the curve is your duty point, and what flow and power look like there.
- Cold-water curves sold into hot or viscous service. Ask at what temperature and viscosity the curve is valid, and what the corrected prediction is for your fluid.
- “Up to 30 m³/h” doing the work of a duty point. Ask for the head, the tolerance grade and the impeller diameter behind that number.
- Curves without grades. A curve without an acceptance grade is a wish. Ask which grade the works test will certify (1B, 2B or 3B) and whether you may witness it [1][2].
- A meter bundled free with the skid. A flowmeter that cannot hold its class in your pipework is decoration. Ask for the straight-run and grounding requirements in writing before you accept the instrument.
- NPSH quoted at test conditions only. Ask for the margin between available and required NPSH at your duty point, at your water temperature.
None of these questions is adversarial. Good manufacturers welcome them, because a buyer who verifies claims is a buyer who never discovers a shortfall during commissioning. The pattern worth avoiding is paying for a datasheet and operating a different machine, which is what you have until somebody tests one against the other.
References
- ISO 9906:2012 Rotodynamic pumps — Hydraulic performance acceptance tests — Grades 1, 2 and 3 (adopted in China as GB/T 3216-2016); acceptance-grade tolerance table, including 1B ±5% flow / ±3% head, 2B ±8% / ±5%, 3B ±9% / ±7%: https://madantec.com/wp-content/uploads/2025/05/ISO9906.pdf and Grundfos KSN documentation https://api.grundfos.com/literature/Grundfosliterature-6911819.pdf
- Hydraulic Institute white paper on selecting a pump performance test acceptance grade, per ANSI/HI 14.6 and ISO 9906 (grades 1U/1E/1B/2U/2B/3B). Summary published by Pumps & Systems: https://www.pumpsandsystems.com/article/specification-pump-performance-test-acceptance/
- FAO Irrigation and Drainage Paper — Chapter 4, Streamflow: volumetric method guidance, a 10 L bucket for flows up to about 4 L/s and a 200 L drum up to about 50 L/s. https://openknowledge.fao.org/server/api/core/bitstreams/c71ceb1f-01e4-4b3a-be63-65a493963e92/content/t0848e-09.htm
- U.S. Bureau of Reclamation — Water Measurement Manual, Chapter 12: tracer-dilution methods with salts or dyes for open-channel and closed-conduit flow. https://www.usbr.gov/tsc/techreferences/mands/wmm/chap12_07.html
- Verito Engineering — “FAT, SAT, and Commissioning of Chemical Dosing Systems”: what factory and site acceptance testing cover and why both gates matter. https://www.veritoengineering.com/blogs/dosing-system-fat-and-commissioning/
- Shanghai ChiMay — procurement specification note for softening and filtering valves: inlet tolerances, regeneration control modes and duty limits.
About the author: Written by the Shanghai ChiMay Technical Editorial Team — engineers who publish hydraulic and control specifications for softening and filtration valves and water quality instrumentation, and who support buyers through factory and site acceptance testing.
