How to Take a Water Sample That Means Something

Why this matters

A lab can only measure what arrives. It cannot recover a parameter that changed on the drive back, and it will report the changed value with the same confident decimal places as a good one. That is the trap: a badly drawn sample does not come back flagged, it comes back wrong and plausible.

Several of the parameters you most want start moving the instant the water leaves the pipe. So the useful way to organize this job is not by sequence but by the clock on each parameter - what dies in minutes gets handled first and in the field, what survives days can wait for the courier. Order the visit that way and the sample means something. Order it by convenience and you will measure your own bucket.

Before you crack a sample point

Isolate the sample point, relieve pressure until a gauge you trust reads zero, and let the line cool or route it through a sample cooler before you open a fitting on a heated or pressurized system. Hot water flashing out of a loosened fitting scalds through clothing and does it at hand and face height.

Sampling from a treated sump, a chemical day tank or a feeder line puts your face over the chemical. Wear the eye and face protection and the specific glove class named in Section 8 of that product's safety data sheet, which your employer must make available (29 CFR 1910.1200). Where the sheet calls for a face shield in addition to goggles, wear both.

Do not draw a sample by disturbing standing water in a basin, sump or long-idle branch without addressing the aerosol. The exposure route is inhalation, so the control is respiratory protection issued under a written program with fit testing (29 CFR 1910.134, fit-test interval at 1910.134(f)(2) at least annually), or not disturbing it. In a building operating a water management program, whether that equipment gets opened at all is that program's decision (ASHRAE Standard 188).

Where sampling requires isolating a pump or a vessel, isolate the stored energy and lock it out under 29 CFR 1910.147 before you open anything, and treat the electrical disconnect for that pump as a separate matter under 29 CFR 1910.333(b)(2), since 1910.147 excludes work on electric utilization installations at 1910.147(a)(1)(ii)(C).

If you connect a hose to fill, flush or feed, that is a cross-connection to a potable system and the backflow protection required by the local plumbing code for that chemical goes in before the connection is made.

The clock, which sets the order

Those groupings follow the preservation and hold-time conventions the analytical methods carry, the most commonly mirrored published version being the table at 40 CFR Part 136 Table II, which is written for Clean Water Act analyses; a drinking-water compliance sample follows the method cited in 40 CFR Part 141 instead. The lab's own instructions govern, and they will send them if you ask. That request is the step most often skipped.

The procedure, ordered by that clock

1. Write the question first. One sentence: "is the closed loop taking air," "is the hot side depositing," "does this makeup water need softening." A sample without a question generates a report nobody can act on and a second trip. The question decides the point, the type and the parameter list, and everything after this step is downstream of it.

2. Pick the point that can answer it. Makeup line for what is entering. Recirculating water for what the system has become. As close to the failure as you can reach for what is attacking. The meter or the utility's published average tells you what arrived at the property, which is a different question from all three.

3. Get containers, preservatives and the hold-time sheet from the lab before the trip. Some analyses need an acidified bottle, some need no headspace, some need chilling from the moment of filling. Arriving with the wrong container converts a site visit into a screening exercise.

4. Decide deliberately whether to purge, and record which you did. Flushing gives you the source. Standing water gives you what the building's own piping contributes. They are different waters and neither is the correct one in general - only for a specific question. The tap protocol for lead and copper requires a first-draw sample after a minimum six-hour stagnation (40 CFR Part 141, Subpart I) precisely because a flushed sample measures the main rather than the plumbing, and flushing before that sample produces a clean result on a building with a real problem.

5. Measure the minutes-long parameters at the tap, before you fill anything. Temperature, dissolved oxygen, chlorine residual and pH, with the time recorded next to each. This is the step that separates a sample that means something from one that does not, and it is the step that requires you to have brought a calibrated meter rather than hoping the lab covers it.

6. Fill in the order the container list requires, and respect headspace. Anything measuring a dissolved gas is filled from the bottom with the tube submerged, overflowed, and capped with no bubble. A bubble in a dissolved-oxygen bottle is not a small error, it is the measurement.

7. Label with the point, the date, the time, the temperature, whether the system was running, and whether the sample was flushed. An unlabeled bottle is a bottle of water.

8. Chill and ship the same day. The hold times above start at the moment of collection, not at the moment the courier collects.

Worked example: the sample that could only give one answer

A closed hydronic loop, black water, magnetite sludge in the strainers. The question is whether it is taking air.

First attempt. The tech opens a drain valve, runs it into an open bucket, dips a probe, and reads 7.0 mg/L dissolved oxygen at 25 C. Everyone concludes the loop is ingesting air, and the search starts for a leak or a failed air separator.

Why the number could not mean that. Fresh water in contact with air at sea level saturates at about 8.3 mg/L at 25 C (and about 9.1 at 20 C, and near 14.6 at 0 C, so the figure is temperature-specific and the sample temperature has to be recorded with it). A sample that has fallen through air into an open bucket equilibrates toward saturation within minutes. 7.0 against a saturation value of 8.3 is 84 percent of saturation - which is what an exposed sample reads regardless of what the loop contained.

Check the direction at both ends, because this is what makes the point general rather than a story. A loop genuinely full of air-saturated water, sampled into a bucket, reads near saturation. A loop that is properly anoxic, sampled into a bucket, also reads near saturation. The open-bucket method can produce a false positive for oxygen and cannot produce a false negative. It is not a weak test, it is a test with only one possible answer.

Second attempt. Same valve, a tube pushed to the bottom of a bottle, the bottle overflowed several volumes and capped underwater with no bubble, probe reading taken immediately: 0.3 mg/L, about 4 percent of saturation at that temperature.

What that changes. The loop is not taking air. The magnetite is a record of oxygen that was consumed at some point in the past and it is not being resupplied now, so the fix is filtration and cleanup rather than a hunt for an ingress point. Note what is still unproven: a loop can read low on oxygen at the moment of sampling and still take makeup, because oxygen is consumed within a short distance of where it enters. Confirming that takes a makeup volume, not a chemistry test, and a separate article covers how to get one without a meter.

What would flip the conclusion: if the loop had been sitting idle and cold for a week before the visit, oxygen entering during that idle period may already have been consumed, and a single low reading taken cold does not represent the running system. Take the second sample with the loop circulating at temperature.

The failure mode: the first result stands, a crew spends a day pressure-testing a tight system, an air separator gets replaced on suspicion, and the sludge comes back, because nobody removed it and nothing was ever wrong with the air side.

How to verify a sample was good

Take a duplicate at the same point at the same time and send both. Two results that disagree beyond the method's stated precision tell you the sampling was the variable, not the water.

Carry a field blank on any trip where contamination is plausible - lab-supplied clean water, opened at the site, handled exactly like the sample, sent in the same cooler. A blank that comes back with your parameter in it means the bottle, the hands or the site air contributed it.

And call the lab when a result surprises you, before you act on it. Ask what the reporting limit was, whether the sample arrived within hold time, and whether it arrived at the required temperature. Those three answers resolve most surprising results without another trip, and a lab will tell you plainly when a sample should not have been run at all.

References

• 40 CFR Part 141, Subpart I, for the first-draw tap protocol and its six-hour minimum stagnation
• 40 CFR Part 136 Table II for the published preservation and hold-time conventions most laboratory instructions mirror, noting it is written for Clean Water Act analyses
• 29 CFR 1910.1200 (Hazard Communication) and Section 8 of the product safety data sheet for glove class and eye and face protection when sampling treated water
• 29 CFR 1910.134, including the fit-test interval at 1910.134(f)(2), for respiratory protection where sampling generates aerosol from standing water
• 29 CFR 1910.147 for isolating stored energy in a pump or pressure vessel, and 29 CFR 1910.333(b)(2) for the electrical disconnect, which 1910.147 excludes at 1910.147(a)(1)(ii)(C)
• See related: How to Read a Water Test Report; How to Find Out How Much Makeup Water a System Is Really Taking