EN 12845 Water Supply Test Arrangements and Test Procedure

How a full-flow pressure-flow curve is produced, which orifice sizes to use, why pitot and gauge readings diverge — and the annual test routine from field experience.

Opening the annual maintenance report on the OH3 sprinkler installation at a logistics centre, static pressure read 8.4 bar and "at full flow" 7.9 bar. The pump nameplate gave 5.2 bar at its rated point, so the site reading appeared to sit above the pump curve. The reason took an hour to find: the team had opened the drain-and-test valve only partly, and while the flow meter read 850 L/min, the pump was assumed to be running at its rated 2100 L/min. The full-load test requirement exists precisely to prevent that kind of self-confirmation.

EN 12845:2015+A1:2019 requires permanent test equipment on every sprinkler installation, and an annual full-flow test using it. This article covers how the test arrangement should be built, how a pressure-flow curve is properly produced, and the "ambiguous pitot reading" error we meet most often.

What the test arrangement must contain

To measure whether the water supply genuinely meets the hazard class, EN 12845 requires permanent test equipment. It is installed in one of two configurations — in the pump room, or at the control valve set:

• In the pump room: on systems fed by an automatic pump, the flow meter is permanently connected in the pump room. Gauges go on the pump suction and discharge flanges, on a straight length of pipe. Where several pumps run in parallel, the flow meter capacity may be selected for the largest pump.
• At the control valve set: on pumpless supplies such as a town main or gravity tank, the flow meter may be fixed or portable at the valve set. The hydraulically most remote set is tested, and where two hazard classes coexist, the higher flow class governs.

Three critical limits govern equipment selection, and they are routinely skipped on site:

• Flow meter accuracy: better than ± 5 % of maximum flow. Because a portable device may be affected in transit, no data may be recorded without verifying calibration before the test.
• Flow meter range: in fully calculated systems, the range must cover both the most and least favourable areas of the system demand, and must measure at least 140 % of maximum demand.
• Gauge accuracy: pressure data for the curve may only be taken from a calibrated test gauge accurate to ± 1.6 %. Producing a curve from a cheap analogue gauge screwed into the valve set is outside the standard.

Test orifices and drain-test line sizes

The drain-and-test valve exists to confirm, by putting flow on the ground, that both the alarm and the pump operate at full load. The lower limits:

Hazard class	Min. drain-test line size	Typical orifice
LH	40 mm	40–65 mm; about 380 L/min is sufficient
OH1 / OH2	50 mm	50–65 mm; demand 540–720 L/min
OH3 / OH4	50 mm	65 mm typical; demand above 1100 L/min
HHP / HHS	50 mm mandatory minimum; 100 mm in practice	100 mm orifice or a multi-outlet rosette; 2700–9000 L/min

A working rule from site: where the test discharge is free, orifices are kept at two main sizes, 65 mm and 100 mm. A 65 mm orifice opens the OH classes to full flow, and a 100 mm orifice does the same for HHP and HHS pumps without choking them. It is impossible to drive an 1800 L/min pump to full load through a smaller orifice; pressure never falls, the pump behaves as if against a closed valve, and the curve comes out biased.

Where test water is not discharged to waste — returned to the town main or the fire water tank — a return line of the same size must be provided. The standard requires an arrangement for disposing of test water; it is the item most often omitted.

The full-flow curve: a four-point method

The standard requires at least three pressure-flow points plus a static reading at zero flow. Taking four on site makes the curve readable:

1. Static point (Q = 0): drain-test valve closed. Record the churn pressure from the gauge. It also serves the maximum pressure check.
2. Low flow (Q ≈ 0.5 Q_rated): valve partly open. The left branch of the curve, before the pump enters overload.
3. Rated flow (Q = Q_rated): the nameplate figure. The pump must deliver at least its nameplate pressure here; deviations like the 7.9 bar in our report are caught at this point.
4. Overflow (Q ≈ 1.4 Q_rated): the EN pump operating condition. The right branch collapses; pressure must not fall below 65 % of the mid-point figure.

Allow at least 30 seconds at each point for the flow to settle. Where the gauge pulses, fit a snubber before reading; otherwise even a ± 1.6 % gauge swings within a ± 0.3 bar band.

When is a pitot tube useful?

EN 12845 does not require a pitot tube for the pressure-flow curve, emphasising that the primary data comes from the gauge and flow meter. A pitot helps only when measuring the jet from an open discharge orifice. But on site, pitot readings come out "ambiguous" for several reasons:

• Jet quality: a cracked or rusted orifice edge produces a scattered jet, and the pitot reads turbulence rather than pressure.
• Position: the pitot tip must sit half an orifice diameter away, on the axis. A gloved hand touching it shifts the reading by 0.5 bar.
• Entrained air: if suction air has not been purged when the pump first starts, the jet runs white and pressure reads falsely low.
• Wind: on outdoor discharges, wind can move the pitot gauge too much to read.

Our working rule: where a pitot is used to corroborate the curve, the gauge at the pump flange is the primary reading and Q = K√P serves only as a check. If the pitot reading diverges from the pump pressure by more than 10 %, stop the test and re-set the orifice and pitot position.

The annual test sequence

The annual routine treats installations with and without a supply pump separately. Both are carried out at full load:

Step	With pump	Without pump
1	Open the test line connected below the pump discharge check valve	Open the test line above the control valve set
2	Record static pressure, then bring the flow meter to full flow	Open the drain-test valve fully to reach the required flow
3	Compare with the pump nameplate figures and report any deviation	Check whether the gauge meets the minimum design pressure
4	Account for pipe and valve friction losses	Account for losses between the supply and the valve set

The same routine also includes the diesel fail-to-start alarm, the tank float valves and the suction strainers. Those are independent of the curve but done on the same day, and the report template should have a place for each.

Field error: what an ambiguous pitot reading was hiding

Back to the OH3 logistics facility. Because the pitot read 1.2 bar, the team calculated flow through the 65 mm orifice with K = 18.3 as Q = 18.3 × √1.2 ≈ 2005 L/min and declared it satisfactory. But:

• The drain-test valve was not fully open, so the jet sprayed sideways from the valve body rather than downward. The pitot therefore read a true dynamic pressure of 1.2 bar, while the flow at the orifice was about 30 % of that.
• The flow meter read only 850 L/min at the time, and the team had reconciled the two readings by blaming the meter.
• The flow meter calibration was checked as the standard requires; deviation was 2 %. The problem was not the meter — the valve was not open.

With the valve fully open, the flow meter read 2080 L/min and the pump discharge gauge 5.3 bar, meeting the nameplate. The only intervention was cleaning the drain-test valve spindle. Trusting the pitot reading is a very common site error; the primary readings must always be the calibrated gauge at the pump flange and the flow meter.

Comparison with NFPA 25

NFPA 25 also requires three points — churn, rated and peak — in the annual pump test. The differences:

• NFPA 25 explicitly defines a multi-outlet hose stream flow test with a pitot tube; EN 12845 makes the flow meter the primary method.
• NFPA 25 requires the pump curve not to fall below 95 % of the nameplate at the rated point; EN 12845 takes 100 % as the basis.
• NFPA 25 supports the test with a weekly no-flow churn run; EN 12845 achieves the same through its routine maintenance clauses.

Turkish practice

BYKHY requires sprinkler installations to be inspected at least annually but does not detail the test procedure. As a result, "I read the gauge and signed" reports are common. The EN 12845 test equipment and annual routine clauses provide the most robust framework for defining which instruments, which ranges and which curve points that report should contain — and a report referencing EN 12845 passes insurance assessment without difficulty.

My practical recommendation: always put four headings on the annual maintenance form — static pressure; at least three pressure-flow pairs; the calibration dates of the flow meter and gauge; and confirmation that the drain-test valve was fully open, photographed with the handle visible. With those four items, both the standard and the inspector are satisfied.

Summary

• EN 12845 requires permanent test equipment — a flow meter and a calibrated gauge — on every installation.
• The curve needs at least three points plus a static reading, with gauge accuracy of ± 1.6 % and flow meter accuracy of ± 5 %.
• The drain-test line is a minimum of 40 mm for LH and 50 mm for other classes; 65 or 100 mm orifices are effectively mandatory for OH3 and high hazard.
• The annual full-flow test has separate procedures for systems with and without pumps.
• A pitot reading is supporting evidence, not primary data; an ambiguous pitot reading usually hides a valve that is not fully open.

Frequently asked questions

Is full flow mandatory in the annual test?

Yes. Every water supply pump must be tested at least annually under full load, meeting its nameplate figures. Static or partial flow alone is not sufficient.

Why is a gauge preferred over a pitot tube?

Because the standard requires a calibrated test gauge accurate to ± 1.6 % for the test curve. Pitot readings vary significantly with nozzle axis, entrained air and observer, so the primary data comes from the pump flange gauges and the flow meter.

What accuracy must the test flow meter have?

Better than ± 5 % of maximum flow. A portable device must be verified again before use.

Does a pumpless supply need a different test?

Yes. The full-load test is carried out through a test line connected above the control valve set, with a drain-test valve of at least 40 mm for LH and 50 mm for other classes.

How many readings should be taken?

At least three pressure-flow points plus a static reading at zero flow. In practice, four points — static, low, rated and 140 % — produce a far more reliable curve.

SprinkCalc — Fire Sprinkler Design Across Three Standards

SprinkCalc covers hazard classification, design density and area, K-factor selection, water demand and hydraulic calculations for NFPA 13, FM Global and BS EN 12845 in a single iOS app, and exports a professional PDF report.

Download SprinkCalc on the App Store

MEP Calc — 86+ Engineering Calculators

MEP Calc bundles 86+ engineering modules in one iOS app: 21 fire calculations plus heating, cooling, HVAC, plumbing, steam and natural gas.

Download MEP Calc on the App Store