The auditor asked a simple question: how do you know this meter is still accurate? The site had a calibration certificate from three years ago, a loop that ran continuously, and no in-place verification. The answer was, in effect, “we assume so”.
That answer is fine for a cooling water line. It is not fine on a WFI distribution loop, where the flow signal is part of the contamination control strategy and the number is cited in a validation document.
Pharmaceutical water is the cleanest fluid in the plant and one of the hardest to meter. It is deliberately free of ions — 0.5 to 5 µS/cm — which puts it well below what a standard electromagnetic meter needs. Then the loop gets sanitised at 80°C or steamed at 121°C, which rules out a large part of the material list.
The conductivity problem, stated plainly
A standard electromagnetic meter wants 20 µS/cm or better for a clean signal. Purified water at 25°C sits around 1 to 5 µS/cm. WFI can be lower. At that level a conventional meter will read, but the zero drifts and low-flow performance is poor.
Three ways through it:
- Low-conductivity electromagnetic meter. A specialised build with capacitive or non-contact electrodes, rated down to 0.1 µS/cm. The mainstream answer for WFI distribution loops. Expect 40% to 60% more than a standard sanitary meter.
- Coriolis mass meter. Independent of conductivity. Gives mass flow and density. More expensive at larger line sizes, but the natural choice for small-bore WFI take-off points.
- Clamp-on ultrasonic. Works on non-conductive fluids. Rarely acceptable inside a validated WFI loop on hygienic design grounds, but useful for engineering surveys and utility water outside the GMP boundary.
Sanitisation is the design case
Most pharmaceutical water loops are sanitised by circulating water at 80°C, or by steaming. That rules out EPDM and natural rubber liners on lifetime grounds, and it puts the electronics under thermal stress if the transmitter is mounted on the sensor.
Practical build for a WFI loop:
- PFA liner, 316L body, Ra ≤ 0.8 µm on wetted surfaces
- Hastelloy C or platinum electrodes
- Remote transmitter, mounted away from the hot pipe
- Clamp or DIN 11851 connections, 3-A and EHEDG certified
- IP66 or better on the sensor, IP65 on the transmitter
PFA rather than PTFE because the loop cycles thermally every week. PTFE is fine at 80°C steady, but the liner-to-flange joint sees the cycles and eventually works loose.
Where the meters go
| Point | Typical size | Recommended | Why |
|---|---|---|---|
| RO permeate / EDI outlet | DN25–DN80 | Low-conductivity electromagnetic | Very low conductivity, continuous duty |
| WFI storage tank outlet | DN50–DN100 | Low-conductivity electromagnetic or Coriolis | 80°C sanitisation, low conductivity |
| Distribution loop return | DN50–DN100 | Low-conductivity electromagnetic | Loop velocity must stay above 1 m/s |
| Point-of-use branches | DN15–DN25 | Coriolis | Small bore, low flow, density useful |
| CIP supply | DN40–DN80 | Sanitary electromagnetic | Caustic and acid at 80°C |
| SIP steam | DN25–DN50 | Vortex | Clean steam, temperature compensated |
| Clean steam condensate | DN25–DN50 | Coriolis | Low flow, two-phase risk |
Loop velocity is a compliance requirement
Pharmaceutical water loops are designed for turbulent flow at all times, typically 1.0 to 1.5 m/s, to prevent biofilm formation on the pipe wall. That makes the flow meter part of the contamination control strategy, not just a measurement device.
Two consequences. The meter must be accurate at the low end, because the loop runs at minimum velocity overnight and that is exactly when flow is hardest to measure. And the flow signal usually goes to the BMS with a low-velocity alarm, so the meter’s low-flow cut-off has to sit below the alarm threshold — otherwise the alarm and the meter disagree and nobody trusts either.
Validation documentation
This is where pharmaceutical projects differ from every other industry. The meter is a small part of the paperwork, and the paperwork decides whether the system passes.
Expect to supply, at minimum:
- Material certificates, 3.1 to EN 10204, for wetted parts
- Surface roughness certificate with the measurement method stated
- FDA-compliant elastomer declarations for gaskets
- Calibration certificate traceable to a national standard
- IQ/OQ protocol support, including zero-check procedure and as-found/as-left records
- 3-A and EHEDG certificates where the design claims compliance
Ask for these before the order is placed. Getting a material certificate retroactively for a meter already installed in a validated loop is an unpleasant experience for everyone involved.
Calibration and the audit question
In-place calibration of a pharmaceutical water meter is difficult: the loop cannot easily come out of service and the reference method has to be validated too. Most sites settle on a three-year cycle with removal and recalibration, plus an annual zero and conductivity check in place.
The audit question is always the same — how do you know it is still accurate? A documented drift history is the answer. Keep as-found and as-left data from every calibration, and keep the electrode resistance baseline from commissioning. When resistance has doubled, the meter needs cleaning whether or not the calibration says it is in tolerance.
Common mistakes
A standard meter on WFI because “it worked on the RO feed”. The RO feed has ten times the conductivity. Different fluid, different specification.
Transmitter mounted on the sensor in a hot loop. Electronics rated to 60°C ambient, installed beside an 80°C pipe in a plant room at 35°C. Failures start in the second summer.
Dead legs at point of use. Any branch longer than three pipe diameters is a microbial risk, and a meter on a dead leg is worse than no meter — it looks like monitoring while doing nothing.
Gasket chosen from the standard list. A sanitary clamp joint with an industrial EPDM gasket is not a hygienic joint. The gasket needs certification too.
Frequently asked questions
Can I use a standard sanitary electromagnetic meter on purified water?
On the RO feed and reject line, usually yes. On the WFI loop itself, no — conductivity is too low and the zero drifts outside the tolerance the validation assumes.
What is the minimum conductivity for a standard meter?
Nominally 5 µS/cm. Practically, 20 µS/cm for reliable low-flow performance. Below that, specify a low-conductivity build.
Is Coriolis acceptable in a GMP loop?
Yes, and common on point-of-use branches. Hygienic design requirements are the same; the advantage is independence from conductivity plus a density output.
How often should these meters be calibrated?
Every two to three years with removal, plus an annual in-place zero check. Shorten the interval if the loop is sanitised frequently, because thermal cycling is what moves the zero.
The hygienic design rules here are the strict end of what food plants also work to — sanitary flow meters for food and beverage plants covers the CIP side. And the low-conductivity problem is the same one that catches chemical engineers out on demineralised water: liner and electrode selection for corrosive chemicals covers the conductivity limits in more detail.