The Filter Alarm Cleared. Did the Filter Improve?

A filter alarm can disappear while the material caught in the element stays exactly where it was. Less coolant passing through the same restriction can produce a lower pressure drop. The display looks better; the filter may be unchanged.

Before closing the job, put measured flow beside the pressure difference. Then find the event that separates the two readings: a change in operating conditions, a change of flow path, or actual work on the element.

Inside the filter

The alarm clears. The material stays.

A loaded filter at half flow can read like a clean filter at full flow. Compare the snapshots, then explore what happened between them.

Housing cutawayPleat detail

Clean outer pleatsFlow100% · illustrated detail

P1 upstream

P2 downstream

Flow 100%

Coolant in

Coolant out

Retained material

Leaves with the element

P1 − P2 vs reference · example alarm at 1.5×

1.00×No alarm

PlayResetReady

Clean filter Material builds Flow falls Replace element

Explore the sequenceClean filter, reference flow · 0%

Two states, one 1× reading

Loaded at half flow, clean at full flow. Step through it to see why.

Two states, one 1× readingSame 1× reading. Different filter.Loaded element, different flowPart-loaded, different flowClean element, different flowLoaded element, reference flowPart-loaded, reference flowClean element, reference flowHousing closed and refilledLoaded element at this flowClean element at this flowNo flow through the elementMaterial loads the upstream pleatsHalf flow, example alarm clearedLess flow, same elementServicing the elementLoaded at half flow, clean at full flow. Step through it to see why.Only the flow trace and the service record can tell these two apart.Only the flow is moving. The material stays; the reading follows the flow.Only the flow is moving. What is on the element stays.A clean element reads lower at less flow too. The reading alone cannot tell.At reference flow, the loaded element reads 2×.At reference flow the reading shows the element's resistance.This is the reference point: a clean element at 100% flow reads 1×.Reassembled for this comparison. Only the flow changes from here.Reassembled for this comparison, with the loaded element in place.Reassembled for this comparison, with the clean element in place.The flow has not changed. Twice the resistance gives twice the reading.The flow has not changed. Half the resistance gives half the reading.This element is already in the housing. Nothing has changed.The flow has not changed. The reading follows the resistance.With no flow there is no pressure drop, whichever element is installed.No flow, so no pressure drop. The material stays on the element.Coolant crosses the pleats from the outside in. P1 − P2 reads 1×.Resistance has doubled. The reading is 2× and the example alarm has tripped.Flow holds. As material collects, resistance rises and the reading climbs.Nothing on the element has changed. Half the flow reads 1× again.Less coolant moves through the same loaded element, so the reading falls.Stopping the pump.Isolating and draining the housing.Removing the cover.Withdrawing the loaded element. The material leaves with it.Installing a clean element.Refitting the cover.Refilling the housing.Restarting the pump. Flow returns to reference.Pump stopped and housing drained. The material leaves with the element.The alarm clears. The material stays.Play the sequence, or compare a loaded filter at half flow with a clean filter at full flow. Both read 1×.Less flow, same elementFlow falls through the same loaded element. The example alarm clears while all the retained material stays in place.Half flow, example alarm clearedThe loaded element is unchanged. At half flow it reads 1×, exactly like the clean reference at full flow.Loaded filter · 50% flow · 1×The example alarm is clear. The retained material is still on the element. This is a comparison snapshot, not a service action.Clean filter · 100% flow · 1×The clean reference gives the same reading at twice the flow. This is a different comparison snapshot; switching views does not clean the element.

Compare the 1× snapshots

Loaded · 50% flowClean · 100% flow

Two different states, not an instant cleaning action.

Flow100%1.00×

Element

LoadedClean

Values at each step

Normalized values for the four states

State	Flow	Resistance	P1 − P2	Example alarm
Clean element	100%	1×	1.00×	No alarm
Loaded element	100%	2×	2.00×	Alarm
Loaded element, less flow	50%	2×	1.00×	No alarm
Clean element after service	100%	1×	1.00×	No alarm

Follow the change, not just the cleared warning

The sequence above holds one thing steady at a time. Material first raises the element's resistance at reference flow. The pressure difference rises. Flow then falls while the same material remains on the pleats, and the pressure difference falls with it. Only the service sequence removes the loaded element.

That is the distinction the maintenance record needs to preserve. A pump slowing down can explain a lower reading. It does not explain an improvement in the element. A documented element change gives a reason to expect a different filter condition, but the result still needs checking at an appropriate operating point.

Bring the alarm history, measured flow and work history onto the same timeline. Look just before the warning cleared. A change in pump speed, control mode, valve position or the path in service is useful context. A pump-speed command alone cannot tell you how much coolant actually passed through the filter.

First, find the two pressure taps

The name on a dashboard tile is not enough. Filter differential pressure is the pressure before the filter minus the pressure after it. A difference across the entire CDU answers a different question.

The Vertiv CoolChip CDU 121 manual gives a practical example. Its filter and unit pressure differences use different combinations of pressure channels. Follow the named taps below before interpreting a lower reading.

Check the measurement boundary

Filter ΔP and unit ΔP use different taps.

Follow the named pressure channels from return to supply.

Secondary returnSecondary supply

PS1Return pressure

FilterElement and housing

PS2Pump-inlet pressure

CDU circuitPump and other components

PS3Supply pressure

Filter ΔP PS1 − PS2Unit ΔP PS3 − PS1

Filter differential pressure: PS1 minus PS2.

Unit differential pressure: PS3 minus PS1.

Measured secondary flow is separate. A pump-speed command does not replace that channel.

Those names belong to that unit. For the installed equipment, trace the actual taps and signal channels. Record the filter path, units and subtraction order. If the display is showing rack or unit differential pressure, do not use it as a substitute for the filter measurement.

Compare at the right flow and fluid condition

A clean-condition reference is useful only if the team can reproduce or interpret the conditions behind it. Keep the measured flow, fluid temperature, coolant identity and concentration with the pressure reading. If the filter path or element part number changes, record that too.

Pall's Principles of Filtration, in its discussion of flow rate, temperature and pressure drop, explains the connection. At constant flow, greater resistance produces greater differential pressure. Viscosity also contributes to the resistance. The housing and element hardware add losses alongside the medium and retained material.

The figure deliberately simplifies that behaviour. A clean element at reference flow reads 1×. Doubling resistance gives 2× at the same flow. Halving flow through the loaded element gives 1× again. A clean element at reference flow also gives 1×. Matching pressure readings can therefore describe different filter conditions.

These are relative values in a linear teaching model, at one fixed fluid condition. The 1.5× alarm line is an example, not a maintenance limit. Real loading is not a steady animation, and real assemblies can behave nonlinearly. For an installed filter, use the manufacturer's clean-assembly flow and viscosity data, along with the site's accepted baseline. Do not turn the model's convenient arithmetic into a universal correction factor.

Check the other ways a warning can disappear

The fluid warmed up

A cold reading and a warm reading are different tests. More viscous fluid can produce a larger pressure difference without a change in retained material. As temperature rises and viscosity falls, the pressure difference can fall too.

Pall's differential-pressure guidance describes electrical switches that reset automatically as a system warms. It also describes visual indicators that retain a warning until manually reset. Confirm the behaviour of the fitted device. A cleared electrical signal, a reset visual indicator and an acknowledged alarm are different events to record.

The flow path changed

Some assemblies include a bypass that allows fluid around the element at high differential pressure. In that arrangement, the reading has to be interpreted alongside the filtration path. A pressure difference limited by a bypass does not establish that all the coolant is crossing the media.

The Pall UR610/690 service instructions warn that continued operation of its bypass-equipped variants with the indicator actuated can allow unfiltered downstream flow. That is a reason to identify the protection fitted to the installed housing, not to assume that every CDU has a bypass or the same limits.

The measurement is not comparable

Check whether the two readings refer to the same taps, units and path. Establish whether the instrument or channel was changed or reset. If the reading does not fit the known operating state, keep the measurement question open rather than assigning the change to the element.

The next step is governed by the unit's procedure. Do not raise flow, defeat protection or open a pressurised housing to reproduce a comparison. If an equivalent operating point cannot safely be reached, retain the available evidence and the reason the comparison remains incomplete.

Close the job with a comparison the next shift can use

The useful record connects a before reading, the work or operating change, and an after reading. A screenshot of the final gauge preserves only the last part of that story.

Keep one event together

A service result needs a before and an after.

Before the change

• Pressure difference, tap names and units
• Measured flow and fluid temperature
• Coolant identity, concentration and path

The change itself

• Time and the actual task performed
• Element part number or cleaning work
• Any valve, bypass or path change

After the work

• The same taps, units and intended path
• Comparable measured flow and fluid condition
• Clean-condition reference and any remaining difference

After service, compare the result with the expected clean-condition behaviour at a comparable measured flow and fluid condition. Confirm the correct element and the intended path are in service. If the result is still different, retain that difference in the handover rather than describing the filter as restored on the strength of one lower number.

An equipment or filter supplier can help resolve an uncertain reference. Send the element part number, pressure-tap locations, measured flow, coolant and temperature, together with the before and after readings and the work performed. Ask which clean-assembly curve and viscosity adjustment apply to that exact arrangement. A curve for another element or fluid leaves the comparison unresolved.

Reliability Engine's coolant-monitoring approach brings fluid condition and loop measurements into the same review. For this decision, the value is keeping the evidence connected: what the pressure did, what the flow did and what anyone actually changed.

A cleared alarm is a useful event. A repeatable comparison is what makes it useful evidence of the maintenance result.

References

1. Pall, Principles of Filtration, flow rate, viscosity and pressure-drop discussion.
2. Pall, Differential Pressure Indicators and Switches.
3. Pall, UR610/690 Service Instructions, fitted indicator and bypass variants.
4. Vertiv, CoolChip CDU 121 Operation and Maintenance Guide, named pressure/flow channels.