Frequent seal failures occur on a batch-service gear pump delivering hydrogen peroxide at 0.5-1 gpm and 25-27 psi. The failures follow starts made after hydrogen peroxide has stood in the pump and piping and released oxygen. The immediate problem is loss of liquid inventory at the seal faces, not insufficient discharge-side seal flushing.
Where does the liquid path stop?
Follow the process path from the storage source through the suction line, pump cavities, seal chamber, and discharge line. During the idle period, hydrogen peroxide decomposition produces oxygen and water. Oxygen can collect in the pump or at piping high points, displacing the liquid that normally lubricates and cools the John Crane Type 9 component seal.
At startup, the gear set and shaft begin moving before a continuous liquid phase reaches the seal faces. A mechanical seal depends on a controlled fluid film between its faces. Gas does not provide the same lubrication or heat removal. Repeated dry starts can damage the faces even when the pump subsequently reaches its normal flow and pressure.
Map the liquid boundary before changing equipment. Record static suction pressure before startup, identify high points that can retain oxygen, and determine whether the pump casing and seal chamber remain liquid-filled after the normal batch dwell. A storage tank that maintains positive suction head helps move liquid toward the pump, but positive pressure alone does not remove a trapped gas pocket that has no vent path.
Which symptoms distinguish the likely causes?
| Observed condition | Likely mechanism | Deciding check |
|---|---|---|
| Seal failures correlate with startup after an idle period | Oxygen has displaced liquid at the pump or seal faces | Check pump inventory immediately before the start and vent the high point through an approved closed path |
The pump later reaches 0.5-1 gpm at 25-27 psi
|
The damaging interval occurs before normal operation | Trend suction pressure, discharge pressure, and motor power from the start command onward |
| A discharge recirculation line contains no liquid during the dry start | The recirculation source depends on discharge flow that does not yet exist | Confirm whether liquid reaches the seal chamber before shaft rotation |
| Gas repeatedly returns after each batch dwell | The idle-state piping geometry retains decomposition gas | Locate unvented high points and compare gas accumulation with dwell conditions |
| Motor power changes when liquid is lost | Pump load changes with pump inventory | Establish installation-specific wet and dry signatures before selecting a power-monitor trip point |
Inspect the failed seal faces as a secondary check. Dry-running damage, chemical attack, misalignment, and inadequate face loading require different corrections. Correlation with the startup event is necessary before treating dry running as the sole failure mode.
Which corrective approaches fit this service?
| Approach | What it addresses | Heat implication | Main limitation |
|---|---|---|---|
| Vent oxygen upstream to the storage tank while inventorying the pump | Removes startup gas and restores liquid before rotation | Does not depend on continuous recirculation | Requires a compatible, controlled return path and suitable piping geometry |
| Maintain positive suction head from a storage tank | Promotes liquid inventory at the inlet | No recirculation heat load | Cannot clear an isolated gas pocket without a vent path |
| Install a motor power monitor and shutdown interlock | Detects a repeatable loss-of-load condition | No process heat addition | Protects after detection; it does not remove the source of gas |
Use discharge recirculation or API Plan 21
|
Can cool and lubricate a seal when discharge liquid exists | A heat exchanger can remove recirculation heat | Cannot supply liquid when the pump has no liquid to recirculate |
| Use a double seal with pressurized flush | Supplies an independent lubricating barrier fluid | Adds a separate fluid and support system | Process must tolerate the selected barrier fluid and leakage direction |
| Use a gas seal or dry-running face combination | Changes the seal arrangement for periodic gas exposure | Avoids a liquid recirculation loop | Needs seal-vendor review of speed, pressure, geometry, materials, and gas compatibility |
| Replace the gear pump with a diaphragm chemical-feed or metering pump | Matches low-flow chemical feed and can eliminate a rotating mechanical seal | No centrifugal recirculation requirement | The selected design must be approved for hydrogen peroxide, required capacity, pressure, duty cycle, and dry-running service |
| Use a magnetic-drive pump | Eliminates the external shaft seal | Rejected here because added heat is prohibited | Still requires acceptable thermal behavior and dry-running capability |
The first choice for the stated failure sequence is source control: vent the accumulated oxygen back upstream while filling the pump, maintain positive suction head, and use power monitoring as backup protection. Evaluate a different pump or seal arrangement when the process cannot provide a reliable liquid-filled start.
Why does discharge recirculation fail during a dry start?
A discharge-to-seal recirculation line needs developed discharge flow. If oxygen occupies the pump at startup, the discharge branch has no liquid source. The shaft and seal can therefore rotate dry while the recirculation tubing remains empty or gas-filled.
API Plan 21 adds cooling to a discharge recirculation stream before returning that stream to the seal area. It can address seal heat when the pump is already moving liquid, but it cannot inventory an empty pump. Adding a heat exchanger to that line changes the temperature of available liquid; it does not create liquid circulation during the initiating dry condition.
The same distinction applies to the rejected centrifugal arrangement. A recirculation loop may stabilize minimum operating flow or remove heat during liquid operation, but it is not a substitute for venting decomposition gas before startup.
What configuration should be recommended?
Provide an upstream vent path that lets oxygen return to the storage tank as the pump and suction line are inventoried. Arrange the suction source to maintain positive head at the pump. The vent must remove gas from the actual collection point; a connection below an unvented high point can leave the gas pocket intact.
Route oxygen and hydrogen peroxide through materials and equipment approved for that service. Keep incompatible contamination out of wetted and vented components. Review the vent destination, pressure relationship, isolation method, and operating sequence through the plant chemical-safety process before commissioning.
Add a power monitor when the wet and gas-filled states produce a repeatable difference in motor load. Use it as a shutdown layer, not as permission to start an un-inventoried pump. Where oxygen cannot be returned upstream and the process can tolerate an inert gas, submit a gas-seal concept to the seal supplier for application review.
How should the vented startup be implemented?
- Document the normal shutdown state, batch dwell, suction arrangement, and every piping high point between the storage tank and pump discharge.
- Measure static suction pressure at the pump before startup. Confirm that the source provides positive head under the lowest normal tank level.
- Place the vent connection where decomposition gas collects. Route it through a closed, compatible path back to the upstream tank when the tank and process design permit that return.
- Before commanding rotation, open or actuate the approved vent path and admit liquid from the suction source. Establish liquid continuity through the pump and seal region using the installation's approved indication or measurement.
- Close or place the vent in its defined operating position. Do not start solely because suction pressure is positive; verify that the trapped-gas location has cleared.
- Start the pump while trending motor power, suction pressure, and discharge pressure. Confirm that pressure and load develop along the known liquid-filled startup signature.
- Stop the pump when the dry-running indication appears or when the expected liquid-loaded signature fails to develop. Diagnose the inventory path before another start.
- Repeat the sequence after a representative batch dwell, because an immediate restart may not reproduce oxygen accumulation.
Frequent batch starts make repeatability more valuable than operator judgment. If the vent sequence uses actuated valves, build the pump permissive from proven valve state and the selected liquid-inventory indication. Define failure states so loss of the indication blocks startup or trips the pump.
How should power monitoring be set and tested?
Measure motor power during known liquid-filled starts and during documented gas-affected events. Select the monitored variable and trip boundary from those traces. Current alone may not separate the states cleanly across startup acceleration and varying discharge conditions; a power monitor can provide a more useful load measurement, but its suitability must be demonstrated on this installation.
| Signal | Diagnostic value | Required action |
|---|---|---|
| Static suction pressure | Shows available inlet head | Block the start when the value does not meet the approved operating condition |
| Motor power | Detects a repeatable change in pump load | Trip on the validated dry-state signature |
| Discharge pressure | Shows whether the pump develops its normal load | Compare the startup trace with the normal path to 25-27 psi
|
| Flow | Confirms delivered capacity after startup | Verify operation within the required 0.5-1 gpm range |
| Vent or inventory indication | Confirms that the trapped-gas location has cleared | Use as a start permissive when the instrument and failure response are suitable |
Derive any startup bypass from measured acceleration and liquid-loading traces; an arbitrary delay can mask the exact dry interval being protected. Test sensor failure, valve-state disagreement, failed pressure development, and manual restart behavior before relying on the interlock.
When should the seal or pump be redesigned?
Move to seal redesign when gas exposure cannot be prevented by piping and sequencing. Give the seal supplier the shaft size, shaft speed, suction pressure, distance from the stuffing-box face to the nearest obstruction, process temperature range, operating cycle, and a seal drawing or complete part reference. State all material restrictions and whether the process can tolerate barrier liquid, buffer liquid, or inert gas.
A double seal with pressurized flush can maintain face lubrication independently of process inventory, but it introduces a support fluid and a defined leakage path. A gas seal may suit unavoidable gas exposure if the process accepts the sealing gas. A face-material combination rated by the seal supplier for periodic dry operation may reduce damage, but it does not remove oxygen accumulation or validate indefinite dry running.
Consider a diaphragm chemical-feed or metering pump when replacement is practical. The required 0.5-1 gpm flow and 25-27 psi discharge condition fall within the type of low-flow chemical-feed duty such pumps address, but select the actual unit from manufacturer curves and chemical-compatibility data. Confirm whether the proposed design is mechanically actuated, hydraulically actuated, or air powered; the pump category alone does not answer that question. Obtain written confirmation for hydrogen peroxide concentration, wetted materials, valve design, required turndown, batch frequency, and permitted dry-running duration.
How is the correction verified?
Test after an idle period representative of the batch process. Before rotation, verify positive suction head and clear the oxygen from its collection point through the approved vent route. Record the complete startup trace, not only the steady-state readings.
A successful test shows liquid inventory before shaft motion, a motor-power trace inside the validated wet-start band, discharge pressure rising normally to 25-27 psi, and delivered flow within 0.5-1 gpm. Repeat enough batch starts to expose the former failure condition, then inspect leakage and seal condition for a developing trend rather than relying on one clean start.
FAQ
What happens if oxygen remains in the gear pump at startup?
The shaft and Type 9 seal can begin moving without a liquid lubricating film. Vent the gas and establish liquid inventory before issuing the start command.
What happens if the discharge recirculation line is left as the only protection?
It supplies no liquid when the gas-filled pump has not developed discharge flow. Even a cooled API Plan 21 loop cannot lubricate the seal until liquid circulation exists.
What happens if positive suction head is available but oxygen is trapped at a high point?
Pressure can exist at the inlet while the pump or seal region remains gas-bound. Put the vent at the collection point and prove that liquid has replaced the gas before rotation.
What happens if a power monitor is used without changing the startup sequence?
The monitor may stop a detected dry event, but the seal can still experience dry rotation before the trip. Use the validated power signature as backup to venting and liquid-inventory control.
How do I verify that the H2O2 dry-start correction works?
After a representative dwell, prove liquid inventory before rotation, capture the wet-start power and pressure traces, verify 25-27 psi and 0.5-1 gpm, and complete the final repeated-start seal-leakage inspection.