TDO and TDI both handle combined radial and axial loading in two directions, but the asymmetry of the inner versus outer ring geometry decides which side fixes, which side floats, and how the bearing handles thermal shaft growth [S2][S4].
SKF, Timken, NTN, and AEC all catalogue both designs; the choice between them is a mounting-and-maintenance decision, not a capacity one, with size ranges from roughly 25.4 mm bore to over 1200 mm bore depending on series [S5].
What each design actually is
A TDO bearing has one double cup (double row outer race) and two single cones (inner rings with rollers and cage), usually with an intermediate ring between the cones [S2][S4]. Timken describes it as a Tapered Double Outer: one double cup, two cones, providing bidirectional axial capacity with a wider effective bearing spread [S7][S8].
A TDI bearing inverts that: one double cone (one-piece double inner race) and two single cups (outer rings), typically supplied as a matched assembly with an outer race spacer already preset [S5]. The TDIT suffix adds a 1:12 tapered bore to the double inner ring for interference-fit applications where the bearing must be removed repeatedly [S5].
Load path, contact angle, and effective spread
Contact angle in double row tapered bearings usually sits between 10° and 30°, and a larger angle (signalled by a higher calculation factor e in the data tables) raises axial load capacity at the cost of higher friction [S4]. Both TDO and TDI share that physics; the structural difference is whether the two rows of rollers share one outer race or one inner race.
Because the TDO pairs two cones under a single double cup, the two load lines diverge as they pass outward; the effective load spread is wider than the bearing envelope itself, which raises moment stiffness and resists shaft deflection [S4][S7]. The TDI loads lines converge outward through two single cups, which keeps the effective spread narrower and concentrates reaction forces closer to the shaft centreline [S5].
In an O-arrangement (lines of action converge toward the bearing axis) versus X-arrangement (lines diverge), the design community applies the same terminology to single-row matched pairs: a wider effective spread (X-type) tolerates more misalignment, while a narrower spread (O-type) carries higher tilting moment with stiffer response [S3]. A TDO behaves closer to the X-pattern on the outer side; a TDI behaves closer to an O-pattern on the inner side, and that is the practical distinction maintenance engineers feel.
Fixed-versus-floating position and thermal growth
TDI bearings are typically specified at the fixed (locating) position on a rotating shaft, because the double inner ring gives a positive axial location, while the two single cups with a preset outer race spacer allow controlled axial preload [S5]. For a rotating housing where the shaft is stationary, a TDI can be set to float by providing a loose fit on the shaft, accommodating thermal expansion of the shaft [S5].
TDO bearings are more commonly used where the housing is the fixed reference and a wider spread improves moment-carrying; they are also the standard for many gearbox and rolling-mill positions where the through-bore of the shaft is the locating element and the double cup provides the wider, stiffer footprint on the housing side [S2][S7][S4]. When overlapping load lines are used, radial thermal expansion has a larger effect on internal clearance than axial thermal expansion, which is why preload must be specified deliberately, not assumed [S3].
Mounting, fit, and field service
TDI ships as a matched assembly with the cup spacer already ground to set clearance or preload, so installation is single-piece onto the shaft; the trade-off is that removal requires pulling the entire assembly, which is exactly why TDIT (tapered bore, 1:12) exists for hydraulic dismounting on large bores [S5]. Timken catalogue coverage for TDI runs from 25.400 mm to 1200.150 mm bore and 30.480 mm to 2460.000 mm OD, which covers the heavy-industrial gearbox and mill span where this design earns its keep [S5].
TDO is often assembled in the housing, with the cones and the intermediate ring set onto the shaft first, then the single double cup dropped over both; this makes individual cone inspection easier during overhaul but means the cup-to-cone preload is set in the field with shims or spacer rings, not at the factory [S2][S7]. For interference fits, radial expansion of the inner raceway (TDI) or contraction of the outer raceway (TDO) must be calculated into the operating clearance, because fits change geometry before the load even arrives [S6].
Decision matrix: TDO vs TDI on four criteria
(1) Effective load spread and moment stiffness: TDO wins, wider footprint, better for tilting loads and shaft deflection [S4][S7]. (2) Fixed-position locating on a rotating shaft: TDI wins, double inner race gives positive axial location and factory-set preload [S5]. (3) Field dismount on large interference-fit bores: TDIT (tapered bore) wins, 1:12 taper allows hydraulic withdrawal [S5]. (4) Housing-side assembly and individual cone inspection: TDO wins, cones mount on the shaft, cup drops over [S2]. For gearboxes, hoists, and rolling mills (the typical double row TRB applications), TDO dominates where the housing carries the moment and TDI dominates where the shaft is the locating element [S4].
Selection criteria that drive the choice
Three checks decide the call: which member (shaft or housing) is the locating element, whether thermal growth runs predominantly through the shaft, and whether the bore is large enough to make field dismounting with a standard cylindrical fit impractical. If the shaft locates and growth is significant, specify TDI; if the housing locates and the bearing sees high tilting moment, specify TDO; if both, use a matched pair of single-row bearings in an X- or O-arrangement rather than a double row [S3].
Lubrication and contamination also feed in: variants with annular grooves and lubrication holes (suffix D on the outer ring for TDO, on the inner ring for TDI) and retention sleeves over the outer rings (suffix .2) are catalogue options across both designs, so grease path requirements do not, by themselves, force one design over the other [S4]. When selecting sealing and lubrication strategy, reference the broader tapered roller bearing page is not relevant, but the geometry logic above is consistent with general industrial bearing mounting practice and rotating-equipment engineering covered across the catalogue.
Limits, common mistakes, and failure modes to avoid
Specifying the wrong design for the locating function is the single most frequent error: a TDI used at a floating position on a rotating shaft without a controlled loose fit will overheat as the shaft grows and tries to displace the double inner race, while a TDO at a fixed position with no shim control will lose preload and skew under moment load [S3][S5]. Misalignment tolerance is set by the effective spread, so a TDO tolerates more housing-to-shaft squareness error than a TDI in the same envelope, but neither design forgives a bent shaft [S3].
Interference fits shift operating clearance before load is applied: inner-ring expansion tightens internal clearance, outer-ring contraction does the same, and on a TDI with a heavy shaft fit, the resulting negative operating clearance can spike contact stress past the rated value; always re-derive operating clearance from the chosen fits, never use the unmounted values from the table [S6]. For heavy-shock or oscillating loads, TDO with retention sleeves (variant .2) gives the outer rings a positive anti-rotation lock in the housing and prevents creep [S4].
For rotating-equipment selection beyond bearings, the same fixed-versus-floating logic shows up in process valve trim specification and in pressure transmitter manifold choices, where one side takes the locating reference and the other accommodates growth.
Next step: pull the existing bearing arrangement drawing, mark which member (shaft or housing) is the locating element, and check thermal growth direction; if the shaft locates, run a TDI (or TDIT for bores above ~300 mm with frequent removal), and if the housing locates under high moment, run a TDO with shim-set preload. A useful cross-reference for the taper-bore and wedge-clamping logic that handles large-bore mounting is the Taper-lock bushing wedge clamping article, which covers the same hydraulic-mounting principle on smaller bores.