Most seal failures I investigate on centrifugal pumps trace back to a mismatch between the seal type and the operating conditions — not a manufacturing defect. Engineers who pick seals by catalog photo instead of operating envelope end up replacing them every six months. Centrifugal-pump service has its own selection logic because the bearing-housing layout, ANSI B73.1 / API 610 seal-chamber geometry, and pressure-pulse profile differ from positive-displacement and agitator pumps. The six configurations below cover roughly 95% of centrifugal-pump installations in the field.
The selection hierarchy that holds up across centrifugal service starts with arrangement (single vs dual), narrows by seal design (pusher vs bellows), and finishes with packaging format (cartridge vs component). Balance ratio is a secondary optimization, not the starting point. The upstream arrangement question — how single, dual, tandem, and back-to-back arrangements relate across all pump types — sits outside this page; this page assumes the arrangement is already decided and walks through the six type-level options a centrifugal pump engineer actually picks between.
Pusher Seals (API 682 Type A)
Pusher seals use a dynamic O-ring or elastomer secondary seal that slides axially along the shaft sleeve to maintain face contact as the faces wear. This pusher design handles roughly 80% of standard centrifugal pump applications across water, light hydrocarbons, and general chemical services.
API 682 classifies these as Type A. Under Category 1 conditions (non-API 610 pumps), they operate from -40 F to 350 F at pressures up to 300 psi. Category 2 service on API 610 pumps extends the pressure envelope to 600 psi, but Type A stays capped at 350 F because the elastomer secondary seal, not the face material, is the limiting factor.
Their weakness shows up in three scenarios: fluids that attack elastomers, temperatures that harden O-rings, and services where crystallization locks the dynamic seal in place. I’ve pulled dozens of pusher seals that failed not from face wear but from coked product jamming the O-ring groove. For clean water, cooling tower service, and light-end hydrocarbons below 350 F, Type A is the right first choice.
Metal Bellows Seals (API 682 Type B and Type C)
Metal bellows seals replace the sliding O-ring with a welded or formed metal bellows element that provides spring force and secondary sealing in one component. In the Type C version, flexible graphite replaces the O-rings, so no elastomer touches the process fluid.
That single change opens up the entire high-temperature and chemically aggressive envelope. Bellows seals handle cryogenic services down to -320 F, and Type C handles high-temperature duties above 500 F where elastomers degrade within weeks. They are the default for hot oil circulation and any fluid that crystallizes or polymerizes in crevices.
The cost premium runs two to three times a comparable pusher seal, and bellows are more sensitive to pressure reversals — a cracked bellows plate means a catastrophic leak, not a gradual weep. Below 350 F with clean, non-crystallizing fluids, that premium rarely pays for itself. Above 350 F or in any coking service, bellows seals are not optional.
Balanced vs Unbalanced Seals
An unbalanced seal exposes the full face area to system pressure, generating high closing forces and face heat. A balanced seal reduces hydraulic loading through a stepped sleeve or reduced-diameter design, cutting heat generation at elevated pressures.
The practical threshold sits around 200 PSIG. Below that, unbalanced seals run cooler than most engineers expect and offer a simpler, cheaper design. Above 200 PSIG, face heat cuts seal life in half or worse — I have seen carbon faces glaze over in under three months on a 350-psi booster pump running an unbalanced seal that should have been balanced from day one.
Most balanced seals target a ratio between 0.65 and 0.85. For high-pressure applications above 400 PSIG, specifying both a balanced design and a dual arrangement is standard practice.
Cartridge Seals
Cartridge seals arrive as a pre-assembled, pre-set unit: gland plate, sleeve, seal faces, springs, and hardware all factory-aligned. Component seals ship as loose parts the technician must measure, set, and assemble on the shaft.
The unit price runs 30-50% higher than the equivalent component seal. But total installed cost often favors the cartridge. Setting a component seal correctly requires measuring seal chamber depth, calculating compression height, verifying face parallelism, and checking spring load. One missed measurement and the seal either leaks immediately or overheats within weeks. Cartridge seals eliminate every one of those errors.
For teams with trained seal technicians who replace seals regularly, component seals still make economic sense. For everyone else — mixed skill levels, remote sites, or applications where downtime exceeds $5,000 per event — cartridge seals pay for themselves on the first installation. If your team has any turnover at all, default to cartridge.
Split Seals
Split seals clamp around the shaft in halves without removing the pump from piping, coupling, or driver. Every other seal type requires pulling the pump back to slide the seal over the shaft end.
That makes split seals the only practical option for large between-bearings pumps, vertical turbine pumps, and equipment where disassembly costs more than the seal itself. A split seal changeout on a large cooling water pump takes four to six hours. The same job with a cartridge seal can take two days after coupling removal, alignment, and piping disconnection.
The split joint introduces a potential leak path that solid seals do not have. Modern precision-machined mating surfaces have narrowed that gap, but for high-pressure or hazardous services I still prefer a solid cartridge when disassembly is feasible. Reserve split seals for applications where equipment geometry makes conventional installation impractical.
Dual Seal Arrangements
Single seals (API 682 Arrangement 1) handle the majority of centrifugal pump services. Dual seals become necessary when single-seal leakage of a toxic or flammable fluid would exceed fugitive emission limits, pose a fire risk, or create an exposure hazard.
API 682 defines two dual configurations. Arrangement 2 uses unpressurized buffer fluid between inboard and outboard seals — any leakage crosses into the buffer system rather than the atmosphere. Arrangement 3 pressurizes a barrier fluid above process pressure, ensuring zero process emission under any condition.
For Arrangement 3, the barrier fluid system is just as critical as the seal itself. Plan 53A uses direct nitrogen pressurization. Plan 53B adds a bladder separator. Plan 53C uses a piston accumulator referenced to seal chamber pressure, so barrier pressure tracks the process. All three target a minimum 28-day interval between barrier fluid replenishment.
The decision between Arrangement 2 and 3 comes down to consequence of leakage. If small amounts of process fluid in the buffer are acceptable and detectable, Arrangement 2 is simpler. If any emission is unacceptable — H2S, ethylene oxide, concentrated acids — Arrangement 3 with Plan 53B or 53C is the only defensible choice.
How to Match Seal Type to Your Application
Every seal selection I have done in 15 years follows the same three-step logic.
Step 1: Choose the arrangement by fluid hazard. Non-hazardous fluids get Arrangement 1 (single seal). Hazardous or regulated fluids get Arrangement 2 or 3 (dual seal) when single-seal emissions would exceed limits or pose a fire or exposure risk. This is always the first decision.
Step 2: Choose the seal design by operating conditions. Below 350 F with clean, non-crystallizing fluid: Type A pusher seal. Above 350 F, cryogenic, or crystallizing fluids: metal bellows (Type C above 350 F). Above 200 PSIG: balanced. Below 200 PSIG with stable conditions: unbalanced is acceptable.
Step 3: Choose the packaging by maintenance capability. Trained technicians with proper tooling: component seals save per-unit cost. Mixed skill levels or high downtime cost: cartridge. Equipment that cannot be disassembled: split seals.
Running these three filters narrows the choice from six options to one in under five minutes for almost every centrifugal-pump service. The remaining vendor conversation is then about model number and material codes, not about which family of seal applies.
The Selection Hierarchy That Prevents Failures
API 682 targets three years — 25,000 operating hours — of continuous operation for properly specified seals. Most seals that fall short were selected with the wrong starting point: picking a balance ratio or a brand before defining arrangement and operating envelope.
Start every specification with arrangement, then seal design, then packaging. Match your conditions to the API 682 category that fits your pump and document the full seal code — Category, Arrangement, Type, containment device, gasket material, face material, shaft size, piping plan — before contacting a vendor. That eight-field code forces the discipline that keeps seals running three years instead of nine months.