Guide
When should a press fit become a shrink fit, and how do you calculate the interference and temperature?
A shrink fit is an interference fit assembled by changing the temperature of one part instead of forcing it: the hub is heated until its bore grows past the shaft, or the shaft is chilled until it shrinks below the bore, the parts slide together with clearance, and the grip forms as both return to room temperature. Switch from a press fit to a shrink fit when the press force becomes impractical, when the mating surfaces gall or score during pressing, or when the fit class the joint needs is one the standards describe as a shrink or force fit.
By Saola Manufacturing engineering team · Published · Updated
The problem usually arrives on the assembly floor. A shaft that was drawn with a heavy interference stalls halfway into its hub, the press runs out of tonnage, the bore comes out scored or galled, or the hub splits at a thin wall or a keyway. The tempting fix is a bigger press or a smaller shaft. Both can be wrong: a bigger press does nothing for galling, and a smaller shaft may no longer carry the torque. This guide shows how to recognize the problem, rank its causes, calculate the interference, the heating or cooling temperature and the hub stress, choose between pressing and thermal assembly, and write the drawing so the supplier and the assembler cannot misread it.
On this page
How do you know a press fit has outgrown the press?
A press fit has outgrown the press when the force needed at maximum interference is close to the capacity of the press or the strength of the parts, when the bore or shaft comes out scored, galled or with metal pushed ahead of the shaft, or when hubs crack during or soon after pressing. Any one of these signs means the joint should be recalculated before anyone tries harder.
The fit standards already anticipate the switch. In the inch system, ASME B4.1 groups interference fits into force and shrink fits, FN1 to FN5, and describes the two heaviest classes as suitable for parts that can be highly stressed or for shrink fits where the heavy pressing forces required are impractical. The metric preferred fits in ASME B4.2 use the same wording for the H7/u6 force fit, and describe the H7/s6 medium drive fit as suitable for ordinary steel parts or shrink fits on light sections. If your drawing carries an FN4, FN5 or H7/u6 callout and the work instruction says “press in”, the two documents disagree with each other before the first part is made.
Watch for these symptoms at first article or on the line:
- Force creep. Press force rises steadily along the stroke instead of leveling off, or varies widely from part to part. That points to interference at the top of its band, taper, or friction rising as the surfaces tear.
- Scored or galled bores. Axial scratches, smeared metal or a ring of material pushed ahead of the shaft. The holding power of that joint is already lower than calculated, because the high spots that should carry the grip have been sheared away.
- Stalled parts. A shaft that stops short of its shoulder and cannot be moved either way. Forcing it on usually ruins both parts.
- Cracked or bulged hubs. Splits starting at a keyway, a cross hole, a thin section or a sharp internal corner, sometimes appearing hours after assembly.
Before changing the method, run these checks on the parts that failed and on a few that did not:
- Measure the pair. Measure the bore and the shaft at several positions along the engaged length and calculate the actual interference of each pair. If it is outside the band the fit class gives, the problem is a size or taper nonconformance, not the assembly method.
- Compare with the calculated maximum. Calculate the press force and hub stress at the largest interference the drawing allows, as in the worked example below. If the press is near its rating at that value, or the hoop stress is near the hub’s yield strength, the drawing is asking for more than the method or the hub can give.
- Look at the damaged surfaces. Axial scoring and smeared metal point to galling or a missing lead-in; a crack from a keyway or cross hole points to hub stress.
- Check finish and chamfers. Compare the measured roughness and the entry chamfers with the drawing; a rough or burred entry explains force scatter and lost grip.
What makes a press fit need too much force, gall or crack the hub?
The usual causes are an interference that is heavier than the joint needs, a hub wall that is too thin for that interference, a galling-prone material couple, rough or unchamfered surfaces, and a long engaged length. Thermal assembly helps with the last three; it does nothing for the first two, which are design problems.
1. Too much interference. The interference is set by the fit class and the size tolerances on both parts, and the worst case is the largest shaft in the smallest bore. Designers often choose a heavy class “to be safe” without checking the torque or axial load the joint really carries. Pressure, press force and hub stress all rise in proportion to interference, so an unneeded step up in class costs force and stress at every part.
2. A hub that is too thin. The contact pressure loads the hub like a thick-walled cylinder, and the hoop (circumferential) stress is highest at the bore. A keyway, a set-screw hole or a cross drilling at the bore multiplies that stress locally, which is why cracks so often start there. Cast iron, die-cast alloys and hardened steels have little ductility to tolerate the peak.
3. A galling-prone couple. Galling is adhesive wear that happens when surfaces slide under high contact pressure. The welding institute TWI describes it as most common when tightening bolts made of stainless steel, aluminum or titanium, materials whose corrosion resistance comes from a passive oxide film that heavy sliding contact can tear away (TWI, what is galling). A press fit made from those materials puts the same sliding contact over the whole engaged length. A shrink fit removes the sliding under load, because the parts go together with clearance and only grip once they stop moving.
4. Rough, burred or unchamfered surfaces. Under contact pressure the surface peaks flatten, so part of the measured interference is lost and the effective grip is lower than the drawing promises; when the parts are pressed, the peaks are also sheared and dragged along the bore. The German calculation standard DIN 7190-1:2017 accounts for this with a smoothing factor on the roughness, reduced from 0.8 to 0.4 in the 2017 edition, and NASA’s bearing interference study, following Harris, deducts 4 µm (2 µm per surface) from the measured interference even for very accurately ground surfaces (NASA/TM-2010-216913). On a turned or milled finish the loss is larger. A sharp bore entry or a shaft without a lead-in also starts the shaving and the misalignment.
5. Long engagement. Press force grows with engaged length, while holding torque needs only enough length to carry the load. A hub much longer than the diameter makes pressing harder and straightness more critical without adding much useful grip.
Notice what thermal assembly does not change. Once a shrink-fitted joint is back at room temperature, its contact pressure and hub stress are the same as a pressed joint with the same interference. If the hub cracks because its stress is too high, the fix is less interference, a thicker hub or a stronger material, not a different assembly method.
How do you calculate the interference, the heating or cooling temperature and the hub stress?
Read the minimum and maximum interference from the fit tables, then size the temperature change so the bore grows (or the shaft shrinks) by the maximum interference plus an assembly clearance. Check the hub stress and the press force at maximum interference, and the holding torque at minimum interference.
Step 1: interference. For a metric hole-basis fit, ISO 286-2 lists the limit deviations of the hole and the shaft for each size range. Maximum interference is the largest shaft minus the smallest bore; minimum interference is the smallest shaft minus the largest bore. Always carry both numbers through the calculation, because they drive different checks.
Step 2: temperature. The change in temperature needed is the required growth divided by the product of the expansion coefficient and the diameter: ΔT = (δmax + c) ÷ (α × d), where δmax is the maximum diametral interference, c is the assembly clearance you want while the parts are being joined, α is the linear expansion coefficient and d is the bore diameter. Without some clearance, the parts touch, heat flows from the hub into the shaft, and the joint can freeze before the shaft reaches its shoulder. There is no single standard value for c; the worked example below uses an assumed clearance, not a standard figure, and a long hub or manual handling needs more. Agree the value with whoever assembles the joint.
Step 3: cooling instead of heating. Handbook expansion coefficients are room temperature values, and they overstate contraction at cryogenic temperatures because the coefficient falls as the metal gets colder. Use integrated contraction data such as the NIST cryogenic material properties curves, which give the total change in length from room temperature down to a given temperature for common alloys.
Step 4: pressure and stress. For a solid shaft in a hub of the same material, the contact pressure is p = Eδ(D² − d²) ÷ (2dD²), with D the hub outside diameter and δ the diametral interference; this is the solid-shaft result in Auburn University’s shrink fit stress notes, which write it with radii and radial interference. The hoop stress at the hub bore is p(D² + d²) ÷ (D² − d²); compare it with the yield strength of the hub with margin, and remember that a keyway or cross hole raises it locally. Holding torque is roughly friction coefficient times pressure times contact area times radius; press force is friction times pressure times contact area. The worked example below runs those numbers for one joint.
| Quantity | How it is found | Result |
|---|---|---|
| Bore limits in mm, H7 (size range over 30 up to 40 mm) | ISO 286-2 deviations 0 / +25 µm | 40.000 to 40.025 |
| Shaft limits in mm, u6 | ISO 286-2 deviations +60 / +76 µm | 40.060 to 40.076 |
| Interference, min to max | Shaft minus bore at the limits | 35–76 µm |
| Contact pressure at max / min interference | Lamé, E = 200 GPa for both parts, D = 80 mm | 142 / 66 MPa |
| Hoop stress at hub bore, max interference | p × (D² + d²) / (D² − d²) | 238 MPa |
| Same, if hub OD is cut to 60 mm | Thinner wall, same interference | 274 MPa |
| Press-in force at max interference | μ = 0.10 to 0.20 assumed | 72–143 kN |
| Holding torque at min interference | μ = 0.15 assumed | about 990 N·m |
| Hub heating, no assembly clearance | 76 µm ÷ (12 × 10⁻⁶/K × 40 mm) | 158 K rise |
| Hub heating, 40 µm assembly clearance (assumed) | 116 µm ÷ (12 × 10⁻⁶/K × 40 mm) | 242 K rise |
| Shaft in dry ice, about 195 K | NIST 304 curve, 293 K to 195 K | 58 µm shrink: not enough alone |
| Shaft in liquid nitrogen, 77 K | NIST 304 curve, 293 K to 77 K | 112 µm shrink: 36 µm clearance |
| Liquid nitrogen shaft plus hub heated 100 K | 112 µm + 48 µm growth | 160 µm: 84 µm clearance |
Read the example as a method, not a recipe. Pressing that joint takes the force of a large hydraulic press, and on a stainless shaft that force is applied while the surfaces slide, which is the galling case. Heating the hub alone needs a large temperature rise once a usable clearance is added, which may exceed what the hub’s heat treatment allows. Chilling the shaft alone leaves only a thin margin. The combination, a modest hub temperature plus a chilled shaft, gives a comfortable clearance without overheating anything, which is why combined heating and cooling is common for heavy fits.
When should you switch from a press fit to a shrink fit?
Switch when the fit is in the heavy force-fit classes, when the couple galls, when the engaged length makes the press force impractical, or when the surfaces must keep their full grip. Stay with pressing, or redesign, when the hub cannot tolerate the temperature, when the joint must come apart in service, or when the real problem is hub stress.
Use the table as a first pass, then confirm with the calculation above. Where the answer is “redesign”, the usual levers are a lighter fit class, a thicker hub, moving a keyway or cross hole away from the bore, or a different joint such as a keyed, splined, clamped or bonded connection.
| Situation | Lean toward | Why |
|---|---|---|
| Light or medium drive fit (FN1, FN2, H7/s6) on small steel parts | Press fit | Force is moderate; a press with force monitoring is fast and repeatable |
| Heavy drive or force fit (FN3 to FN5, H7/u6) on medium or large sections | Shrink fit | ASME B4.1 lists FN3 for shrink fits in medium sections and FN4 and FN5 for shrink fits where the heavy pressing forces are impractical; ASME B4.2 says the same of H7/u6 |
| Stainless, aluminum or titanium on itself, or any couple that scores on trial | Shrink fit | No sliding under contact pressure, so no galling and no loss of grip from shaved peaks |
| Engaged length long compared with diameter | Shrink fit | Press force grows with length; thermal assembly does not |
| Hub is quenched and tempered, case hardened or heat-treated aluminum | Chill the shaft, or combine | Hub heating must stay below the temperature that changes its temper or hardness |
| Hub is cast iron, a die casting or has a thin wall at a keyway | Redesign first | Neither method lowers the final hoop stress at the bore; ASME B4.2 describes H7/s6 as the tightest fit usable with cast iron |
| Joint must be taken apart in the field | Press fit, lighter class, or a keyed or clamped joint | Shrink-fitted parts are hard to separate without heat or damage |
| Shaft carries a finished coating or a bearing with seals | Follow the component limit | Coatings, seals and lubricants set their own temperature limits |
Heating the hub. Heat evenly in an oven or with an induction heater rather than a torch, which creates hot spots, scale and distortion. Measure the bore temperature with a probe, not by color. Keep the hub below its tempering temperature if it is quenched and tempered, and below its aging temperature if it is heat-treated aluminum; the material specification or heat-treat record gives those limits.
Chilling the shaft. Dry ice and liquid nitrogen are the usual media. Cold makes many carbon and alloy steels less tough, and sudden immersion creates steep temperature gradients. University of South Florida research on bascule bridge trunnion-hub assemblies, where a hub cracked while the assembly was being cooled in liquid nitrogen, traced the cracking to high interference stress, low fracture toughness at cryogenic temperature and steep thermal gradients from sudden cooling, and found that staged cooling raised the critical crack length substantially compared with direct immersion in liquid nitrogen (Nguyen, Kaw and Paul, 2007; Collier, 2004). Stage the cooling and handle liquid nitrogen with ventilation and cryogenic protective equipment.
Joining. Give the shaft a lead-in chamfer, fixture both parts so the shaft goes straight to its shoulder in one movement, and hold the hub against the shoulder while it cools, because a hub also shrinks along its length and can pull away from the shoulder and leave a gap.
What should the drawing and RFQ say so a shrink fit is made and assembled correctly?
Give both diameters as a fit class with limits, state the condition and temperature at which they are measured, and add an assembly note that names the method and the temperature limits. The word “shrink fit” on its own tells the machinist nothing about size and the assembler nothing about how hot is too hot.
- Fit and limits. For example, “Ø40 H7” on the hub bore and “Ø40 u6” on the shaft, each followed by its upper and lower limits in figures, as in the worked example table. If the fit is chosen from a calculation, keep the calculation with the drawing revision.
- Measurement condition. Sizes apply at the reference temperature per ISO 1, which is 20 °C, and before or after any coating, stated explicitly.
- Form and finish. Add cylindricity or the envelope requirement to both fit diameters, a surface texture limit, and a lead-in chamfer on the shaft and a break on the bore entry.
- Hub features. Keep keyways, cross holes and set-screw holes out of the fit zone, or show that the stress has been checked with them.
- Assembly note. “Assemble by heating hub uniformly to not more than [limit] or by chilling shaft; do not press. Hold hub against shoulder until at room temperature.” Put the limit you calculated in the brackets.
- Verification. Ask for measured bore and shaft diameters on the inspection report for the fit features, so the actual interference of each pair can be calculated before assembly.
If the parts will be assembled at your plant, the RFQ only needs the machining requirements and the inspection report. If the supplier is to deliver the assembly, say so separately, because thermal assembly is a quoted operation with its own equipment, fixture and safety requirements.
Who does what on a Saola order with a shrink fit?
The shaft and the hub bore are machined at our lead CNC plant in Vietnam, and any finishing goes to a partner plant we manage. Assembly is not assumed: if you want the joint delivered assembled, say so in the RFQ, and the quote states whether and where that step is done.
Shafts, pins and sleeves are produced by CNC turning, and hub and housing bores by CNC milling, at T.K Viet Nam Precision Company Limited, our lead CNC plant since 2017. Its inspection equipment includes a 2D vision measuring machine, a height gauge and a hardness tester; when a fit needs a dedicated bore gauge or plug and ring gauges, that is agreed at the quote stage rather than assumed. Before quoting, our engineers flag a drawing that says “shrink fit” or “press fit” in words only and ask for the class or limits, and they point out heavy fits whose assembly method is not stated.
If a fit diameter is coated, the coating is applied by a partner plant we manage through surface finishing, and the drawing note on the coating condition tells both plants what size to hold. Each quote names the production site and the country of origin. Our quality page describes how inspection records are supplied. Send both mating parts and the assembly note through the quote form.
Frequently asked
Is a shrink fit stronger than a press fit with the same interference?
It often holds better, because the surface peaks are flattened but not sheared and dragged along the bore during assembly, so more of the measured interference survives. The contact pressure and hub stress at room temperature are otherwise the same, so a shrink fit is not a cure for an overstressed hub.
Is it better to heat the hub or cool the shaft?
Heat the hub when it has no heat treatment, coating or seals that the temperature would damage. Cool the shaft when the hub is sensitive to heat, and combine both for heavy fits so neither part has to go to an extreme. Cooling needs integrated contraction data, because room temperature coefficients overstate the shrink.
What happens if the shaft sticks halfway during a shrink fit?
The joint has usually seized because the parts touched and equalized in temperature. Pressing it the rest of the way risks galling and cracking; separating it may need the hub reheated or the assembly cut. Prevent it with more assembly clearance, a lead-in chamfer and a fixture that drives the shaft to its shoulder in one movement.
Can a shrink-fitted joint be taken apart?
Sometimes, by heating the hub quickly while keeping the shaft cool, but the grip often holds until both parts are hot and the surfaces may be marked. If the joint must be serviced, design for it with a lighter fit, a keyed or clamped connection, or provisions for hydraulic removal.
Do I need to tell the supplier the parts will be shrink fitted?
Yes. It changes how the bore entry and shaft chamfer are finished, which surface texture is acceptable and which sizes go on the inspection report, and it lets the supplier flag a hub feature that will crack under the fit stress.
Sources
- ASME B4.1, Preferred Limits and Fits for Cylindrical Parts, ASME, 1967 (reaffirmed 2020)
- ASME B4.2, Preferred Metric Limits and Fits, ASME, 1978 (reaffirmed 2020)
- ISO 286-2:2010, ISO code system for tolerances on linear sizes — Part 2: Tables of standard tolerance classes and limit deviations for holes and shafts, ISO, 2010
- ISO 1:2022, Geometrical product specifications (GPS) — Standard reference temperature for the specification of geometrical and dimensional properties, ISO, 2022
- DIN 7190-1:2017, Interference fits — Part 1: Calculation and design rules for cylindrical self-locking pressfits (scope), DIN, 2017
- Material properties: 304 stainless (UNS S30400), linear expansion, NIST Cryogenics Technologies Group, accessed October 2026
- Shrink fit stresses, MECH 6300 course notes, Auburn University, accessed October 2026
- Interference-Fit Life Factors for Ball Bearings, NASA/TM-2010-216913, F. B. Oswald, E. V. Zaretsky and J. V. Poplawski, NASA Glenn Research Center, 2010
- Sensitivity analysis of cooling methods and geometric parameters in the assembly procedure of bascule bridge fulcra, C. Nguyen, A. Kaw and J. Paul, Journal of Strain Analysis for Engineering Design, 2007
- Benefit of staged cooling in shrink fitted composite cylinders, N. O. Collier, University of South Florida thesis, 2004
- What is galling?, TWI, accessed October 2026