API 671 calls for a 1.5 service (or experience) factor on the equipment's nominal steady-state torque when selecting metallic disc and diaphragm couplings, a rule confirmed by multiple OEM technical references [S1][S2][S8].
For general industrial machinery outside API 671, service factors of 1.5 to 2.5 are common on disc coupling selection tables, with the upper end reserved for high-inertia starts, frequent reversing, and jam-prone loads [S3][S5][S6].
What the Service Factor Actually Multiplies
The service factor is a dimensionless multiplier, never a torque value in itself, and it is applied to the continuous (steady-state) torque the driven machine sees at the coupling stub [S2][S5].
Calculation order is: (1) compute running torque from power and speed, T = 9550 × P(kW) / n(rpm) for metric units or 63,025 × P(hp) / n(rpm) for imperial; (2) multiply by the application service factor from the coupling catalog; (3) compare the result against the coupling's rated torque; (4) independently check peak or jam torque against the coupling's maximum/peak rating, because the service factor only covers off-design running, not short transients [S1][S2][S5].
Typical service factors for disc couplings by application, per SKF's published disc coupling table, are 1.0 for an electric motor with standard torque, rising to roughly 1.5 for reciprocating equipment and 2.0 or higher where start/stop cycling or shock loads are routine [S3]. Other published worked examples place a frequently-stopping conveyor with jam risk at 2.5, yielding 362 N·m from a 144.9 N·m base running torque [S6].
Why API 671 Sticks to 1.5
API 671 is a special-purpose coupling standard for refinery, petrochemical, and high-speed turbomachinery service, and the 1.5 factor assumes the rest of the safety case is carried by tighter materials, balancing, and inspection regimes [S1][S8].
The factor is intended to absorb "off design" conditions, including minor vibratory torque excursions, rather than to substitute for a separate peak-torque check during starts, jams, or load rejections [S1][S2]. For context on adjacent rotating-equipment ratings, a related spec walkthrough of AC motor nameplate service factor and insulation class shows how the same 1.0/1.15 idea is handled at the driver side of the drivetrain.
Disc Couplings vs Other Metallic Couplings on Service Factor
Service factor is not coupling-type specific in API 671; what changes between disc couplings, gear couplings, and grid couplings is the catalog table that lists application-by-application multipliers [S1][S3].
Comparison of the three common metallic flexible families on torque derating practice:
Disc couplings (per API 671 / OEM catalogs): service factor 1.0 to 2.5, peak/torque-density typically highest in class, no lubrication required, sensitive to misalignment above roughly 0.5–1.0° per disc pack [S1][S3][S4].
Gear couplings: service factor 1.0 to 2.0, torque density high but requires lubricant, generous parallel misalignment capacity, backlash of roughly 20–60 arc-minutes depending on size and tooth finish [S3].
Grid couplings: service factor 1.0 to 3.0 for the most heavily shock-loaded applications, torque capacity grows with grid size, lubricant required, and visual wear inspection is part of routine service [S3].
For most general-purpose process lines the choice between these three reduces to three gates: lubrication tolerance, peak torque, and misalignment budget. If any of the three are not the binding constraint, service factor by itself is rarely the discriminator [S3][S5].
Applying the Service Factor to Disc Coupling Standard Torque
A disc coupling's "standard torque" in the catalog is a continuous-duty rating at a reference speed and alignment condition; multiplying it by the application service factor is the wrong direction. The correct direction is to multiply the application's running torque by the service factor, then select a coupling whose standard (rated) torque is at least that high [S2][S5][S9].
Procedure, in one line: T_select = SF × T_run, where T_run is the steady-state torque at the coupling, and the chosen coupling must have rated torque ≥ T_select and peak torque ≥ the maximum transient the drivetrain can deliver [S5][S6].
A published worked comparison plotted several disc-coupling models against a NEMA 449T frame at a 1.5 service factor and showed torque density in the 30–90 N·m per kg band for mid-size disc packs, with bore capacity, not torque, often becoming the binding limit at larger frame sizes [S7].
Who the 1.5 Service Factor Is For, and Who It Is Not For
The API 671 1.5 factor is built for documented turbomachinery trains: drivers and driven equipment covered by API 610 pumps, API 617 compressors, API 612 steam turbines, and similar standards, where torsional analysis, balance grade, and material traceability are part of the same spec package [S1][S8].
It is not a safe default for conveyors with frequent E-stops, mixers with high-density slurries, crushers, or any drivetrain where the peak-to-running torque ratio routinely exceeds about 1.5; for those loads, 2.0 to 2.5 from the general industrial table is the correct starting point [S3][S5][S6].
For pumps with stable, non-pulsating head, the 1.0 to 1.25 service factor is usually sufficient and over-spec'ing with 1.5 just inflates coupling size, bore, and balance class without buying useful life [S3].
Limits, Failure Modes, and Common Mistakes
Disc coupling failure modes most often traced back to service-factor misuse are: (a) fretting wear accelerated by sustained running above the corrected torque rating, (b) membrane fatigue from cumulative off-design excursions that the 1.5 factor was never intended to absorb, and (c) fastener loosening when peak torque was not independently checked against the coupling's maximum rating [S4].
A directly cited engineering observation: "Adhering to the manufacturer's recommended bolt tightening torque and minimizing misalignment will assure the longest service life" [S4]. In other words, the service factor buys margin against torque; it does not buy margin against sloppy installation.
Common mistakes worth flagging: applying the service factor to peak torque instead of running torque (double-counting), using the gear-coupling service-factor table for a disc coupling (different families have different reference baselines), and assuming a 1.5 service factor substitutes for a separate torsional analysis on a variable-frequency drive where critical speeds can be excited near running speed [S1][S5].
Standards, Sourcing, and What to Verify Before You Order
Three primary references govern the multiplier: API 671 for special-purpose couplings in process service, AGMA 9003 / AGMA 2004 for gear-coupling applications, and the individual OEM catalog service-factor table for the exact disc coupling series being quoted [S1][S8].
Before placing an order, confirm in writing: (1) the source of the 1.5 number, ideally a copy of the relevant catalog page; (2) whether the supplier's rated torque already includes an internal service factor, which would make the application multiplier additive and produce an over-rated selection; (3) that the maximum or peak torque row in the catalog has been checked against the worst transient in the duty cycle, including two-phase starting on conveyors and load-rejection torque on pumps [S1][S2][S5].
For drivetrains where the service factor is genuinely borderline, track the actual running torque with a torque sensor during commissioning to confirm the catalog assumption; for periodic re-verification, a torque wrench tester on the disc-pack bolts catches the loose-fastener failure path that service factor cannot address [S4].
Next signal to watch: AGMA's ongoing alignment of its service-factor tables with the API 671 framework, which has produced a small but visible convergence in published multipliers between 2024 and 2026, and any future change will most likely land first in the gear-coupling and gear coupling catalog tables rather than the disc-coupling tables [S8].