How to Read a Pressure Drop Across a Component

Why this matters

A pressure drop is the most available number in fluid work. Two ports, one gauge, thirty seconds. It is also the number most often read and then misused, because a drop is never caused by one thing. Every drop you measure is the product of how much fluid is moving and how hard the component makes it work, and a single reading cannot separate those two. Techs condemn clean coils on a low drop and pass fouled ones on a normal drop, both times because they read a number that was answering a different question than the one they asked.

Before you crack a port: what is behind it

A test port on a live system holds stored pressure and, on a heating loop, fluid hot enough to burn through a glove.

• Relieve or isolate before you break any connection. Close the isolation valves that bracket the component, open the drain or the port with the drain hose already in a bucket, and confirm the gauge reads zero before a fitting comes loose. On a heating loop, shut the heat source and let the fluid drop below 120 F first, or pipe the discharge to a floor drain and stand to the side of it with a face shield and heat-rated gloves.
• Water held above 212 F under pressure flashes to steam the instant the joint opens. Read the loop temperature before you touch a fitting, and if it is above 212 F, do not open anything until it has cooled below that.
• Do not open a refrigerant-bearing path to install a port. Use the service ports that are already there. Cutting into a charged circuit vents refrigerant into your breathing zone and can frostbite a hand at the fitting.
• Reading a gauge on a running machine means standing next to a coupling. Keep the guard on, keep sleeves and a lanyard clear of it, and route your hose so it cannot be pulled into the shaft.

The two unknowns in every drop reading

For a fixed piece of geometry with fluid moving turbulently through it, pressure drop rises with roughly the square of flow:

drop = k x flow squared

k is the resistance of that specific component in its current condition. Foul it, close it, or partly plug it and k goes up. So a measured drop has one equation and two unknowns. Halve the flow through a clean strainer and the drop falls to a quarter. Leave the flow alone and double the strainer's k and the drop doubles. The gauge cannot tell those apart.

Everything in the procedure below exists to pin one of the two unknowns so the reading resolves the other.

Step 1: decide which unknown you are pinning

Pick before you take the reading, because it changes where you put the taps.

• Pinning k means you are treating the component as being in known condition (new filter, a valve you just set to a known position, a coil with a clean-condition curve you trust). The drop then reports flow. This is the "pump as a flow meter" family of methods.
• Pinning flow means you have an independent flow number from somewhere else - a catch and time at a terminal, a heat balance, a traverse - and the drop then reports whether k has moved off its published value.

If you can pin neither, you can still read the drop, but write it down as a baseline and say so. A drop with no companion number is a data point for next visit, not a diagnosis today.

What breaks if you skip this: you take the reading, find it low, and announce the component is clean. Low drop is equally consistent with a clean component and with a flow shortfall upstream, and the second one is the more common fault.

Step 2: pick taps that bracket the component and nothing else

The reading covers everything between the two taps. A port pair that spans a coil plus a control valve plus two elbows returns the sum, and the coil's published curve no longer applies to it.

Walk the piping between your two candidate taps and list what is in there. If it is anything more than the component and a short length of straight pipe, either move a tap or accept that you are measuring an assembly and compare against an assembly value, not a component value.

Prefer taps at the same elevation. That removes an entire correction from the arithmetic in Step 4.

Step 3: use one instrument, not two

Gauge accuracy is specified as a percentage of full span, not of reading, and the accuracy grade on the dial is what sets it. ASME B40.100 defines those grades. The practical consequence is brutal on small differentials: say you use two gauges with a 60 psi span and a 1 percent grade. Each carries about 0.6 psi of allowable error, so a 4 psi differential taken as the difference of two of them carries up to 1.2 psi of instrument error, which is 30 percent of the answer.

Use one gauge moved between the two ports, or a differential gauge with a span sized to the drop you expect. A single gauge carries its offset into both readings, and the offset subtracts out. What is left is repeatability, which is far smaller.

Purge the hose at each port before you read. Trapped air in the hose compresses and the needle settles low and slow; purge into a bucket, not into your hand, because the first slug out is at system temperature.

Step 4: correct for elevation before you subtract

A static water column contributes pressure whether anything is flowing or not: 1 psi is 2.31 ft of water at ordinary service temperatures. If your two taps are at different heights, that column is inside your raw difference and has nothing to do with the component.

If the downstream tap is higher than the upstream tap by h feet, the raw difference overstates the friction drop by h / 2.31 psi. Subtract it. If the downstream tap is lower, the raw difference understates it by the same amount and you add it back.

Air is the exception that makes this easy: the column of air in a duct is negligible at the pressures duct systems run, so duct readings need no elevation correction.

Step 5: take the pair at the same operating state

Between your first port and your second, something can change: a stage cycles, a two-way valve modulates, a variable-speed drive ramps. Then you have subtracted two numbers taken from two different systems.

Lock the state. Put the equipment in a fixed manual mode if the controls allow it, or watch a flow-related indication (motor current, drive speed feedback, a differential you can see) and confirm it is the same when you take the second reading as when you took the first. If the system is modulating and cannot be held, use a differential gauge so both taps are read at the same instant.

The worked reading

A chilled water coil. The submittal gives its clean drop as 11 ft of water, which is 4.8 psi, at its design flow of 45 GPM. Complaint is weak cooling on that zone.

One 0 to 60 psi gauge, moved between the two coil taps. Inlet port reads 41.6 psi. Outlet port reads 36.3 psi. Raw difference is 5.3 psi.

The outlet tap is 3 ft above the inlet tap, because the coil is piped up and over. That is 3 / 2.31 = 1.30 psi of static column sitting in the raw number. Corrected friction drop is 5.3 - 1.30 = 4.0 psi.

Now pin an unknown. Nobody has cleaned this coil in three seasons, so k is not trustworthy. Treat the reading as a flow estimate anyway and see what it can and cannot support:

flow estimate = 45 x square root of (4.0 / 4.8) = 45 x 0.913 = 41 GPM

That is 41 GPM against a design of 45, about 9 percent below design. But the estimate assumed the coil is at its published k. If it is fouled, k is higher, and the same 4.0 psi is being produced by less flow than the clean-coil arithmetic reports. So 41 GPM is an upper bound on the true flow, never a lower one. The honest statement to the customer is "flow is at most 41 GPM, and if the coil is dirty it is less than that."

Compare that against what you would have concluded from the uncorrected 5.3 psi:

flow estimate = 45 x square root of (5.3 / 4.8) = 45 x 1.051 = 47 GPM

Both of those are clean-coil estimates, so they are comparable to each other. The uncorrected one reads about 15 percent higher than the corrected one and lands above design, which would have sent you looking at the air side for the rest of the afternoon. One 3 ft riser did that.

Confirming the number before you quote from it

The reading above is a bound, not a measurement, and that is not enough to sell a coil cleaning. Two independent confirmations, either of which closes it:

1. Re-read after isolating the suspect resistance. Pull and inspect the strainer element upstream of the coil, put it back clean, and take the same corrected pair again at the same operating state. If the coil drop rises, flow rose, which means the restriction was upstream and the coil is not your problem.
2. Get a flow number that does not come from a pressure reading. A heat balance across the coil using entering and leaving fluid temperatures and the load, or a catch and time where the system permits it, gives you flow independent of k. Put that flow into drop = k x flow squared and solve for k. If the solved k is well above the published value, the coil is fouled and you have the evidence in writing.

Then repeat the whole pair once. A drop reading that does not repeat within the width of your gauge's smallest division was taken during a state change, and the fix is Step 5, not a bigger number.

References

• ASME B40.100, gauge accuracy grades and installation practice for bourdon-tube pressure gauges
• Manufacturer submittal data for the component's clean-condition pressure drop at rated flow
• OSHA 29 CFR 1910.147, control of hazardous energy for mechanical isolation and stored pressure before breaking a joint
• See related: The Pressure Differential Tells the Story (the concept behind the reading); How to Find a Restriction From Upstream and Downstream Readings; How to Estimate Flow Without a Flow Meter