Size a repeatedly loaded ball screw from the applied axial-load history, not the peak load alone: THK defines nominal life as the revolutions achieved by 90% of identical screws without raceway or ball flaking [S3].
The calculation must combine basic dynamic load rating Ca, average or equivalent axial load, speed, vibration, shock, preload, lead, and installation condition, then check the resulting life against the specified duty cycle [S1][S3].
Build the load history before selecting a screw
For a repeated duty cycle, list every operating segment with axial force Fi, duration or travel ti, and rotational speed vi. The equivalent mean load Fm represents the constant load that produces the same fatigue exposure as the varying-load sequence [S1].
THK states that ball-screw nominal life is calculated using the basic dynamic load rating and applied axial load, and notes that vibrations, shocks, and fluctuating loads can occur during use.
For reference, the ball screw converts rotation into linear travel, so a force history can also be built from actuator thrust, moving mass, acceleration, and external process force.
Use L10 as a population statistic, not a guaranteed life
THK defines nominal life as the total revolutions until the first fatigue flaking on any raceway or ball. The L10 value is the life reached by 90% of identical screws operating under the same conditions without flaking, while the basic dynamic load rating Ca is based on an axial load that does not vary in magnitude or direction and a rated life of 10^6 revolutions [S3].
Use the rated-life equation with Ca and applied axial load Fa, then apply the modified nominal life L10m when vibration, shock, or fluctuating load requires a load factor fW. This is a screening calculation for design comparison, not a prediction of one individual screw’s failure date [S3].
Apply the speed and vibration load factor

THK’s published load-factor table links service severity to speed: fW is 1 to 1.2 for faint vibration at speed V ≤ 0.25 m/s, 1.2 to 1.5 for weak vibration at 0.25 < V ≤ 1 m/s, 1.5 to 2 for medium vibration at 1 < V ≤ 2 m/s, and 2 to 3.5 for strong vibration at V > 2 m/s [S3].
SHAC states that excessive speed may cause axial vibration or resonance, especially in longer ball screws, and that the manufacturer's recommended maximum safe speed must not be exceeded.
For a separate method to compare screw drives on cost, efficiency, speed, and maintenance, use the ball screw versus Acme screw actuator selection guide.
Compare candidate screws on four design criteria
The comparison should test each candidate against calculated life, operating speed, peak structural capacity, and installation quality. L10 and L10m are calculated from Ca and load factors, rpm converts life into hours, and lead converts revolutions into travel distance [S3].
Use the same load history for every candidate. A screw with a higher Ca may improve calculated fatigue life, but it cannot compensate for an unsupported length that exceeds its speed or buckling limits, incorrect end supports, or excessive preload [S3][S5].
For repeated CNC feed or positioning duty, compare candidate diameters, leads, and accuracy classes within the machine-tool architecture; the CNC ball screw selection guide provides a focused axis-level workflow.
Do not select from nominal static capacity alone. Fatigue is governed by repeated rolling contact, while compression can be governed by Euler buckling, and excessive speed can introduce resonance and thermal loading [S5][S6].
Account for preload and uneven ball loading

Preloaded or double-nut systems can improve stiffness and positioning consistency, but preload introduces a continuing load into the raceway and must be reflected in life calculations [S3]. Higher preload is not automatically better because the nut already carries an internal load before external thrust is applied.
Ball loads are not uniform in a real assembly. A 2024 study reports that machining and assembly error create uneven raceway loading, and that support-end movement and assembly error can add rotating torque; positive geometric error can produce interference contact, greater deformation, and increased load unevenness [S2].
The study states that machining and assembly uncertainties can produce uneven ball loading and additional rotating torque in a fixed-end, mounted-end ball-screw subassembly, affecting fatigue life.
Use a crossed roller guide or suitable support arrangement only when its load path and stiffness match the actuator design, but do not assume an external guide compensates for internal nut preload.
Verify critical speed, buckling, lubrication, and travel
Check four independent limits before release: rated speed, critical speed, maximum axial load, and calculated fatigue life. SHAC provides the critical-speed relation Nc = (k × D × √E / L²) and the buckling relation Pb = (π² × E × I) / (K × L)², where k and K depend on end conditions [S5].
The rated-life result is conditional on ideal mounting and good lubrication. Mounting-material precision and distortion can reduce achieved life, while poor lubrication increases friction, heat, and rolling-contact wear [S3][S6].
For speed selection, THK recommends good lubrication; an oil-lubricated or appropriately specified grease system should be matched to speed, temperature, contamination, and relubrication interval rather than selected by lubricant color [S3].
Use lead to convert revolutions to travel: travel distance equals revolutions multiplied by lead, with the effective lead including any gearing ratio between the motor and screw. Verify that the calculated L10 distance covers the required maintenance interval without assuming a fixed 50,000 km life [S3][S9].
Match the failure mechanism to the duty

Fatigue damage in a ball screw appears as surface pitting, raceway spalling or flaking, progressive ball wear, and eventual loss of positioning accuracy under cyclic loading [S3][S4]. The dominant check changes with duty: rotating machine-tool feed axes emphasize speed and lubrication, while vertical lifting axes require compressive capacity, braking, and retained-load safety.
A screw with a ball bearing support arrangement still requires separate support-bearing checks for reaction load, lubrication, sealing, and mounting stiffness; the support bearing does not perform the screw's primary axial translation function.
Do not use a ball screw for high-shock or dirty service without a documented lubrication and contamination-control plan. The lead screw family can be preferable where shock resistance and simpler maintenance outweigh the efficiency and positioning stiffness of a recirculating ball screw.
Selection sequence for a repeatable procurement decision
First, record force, travel, speed, duty fraction, shock class, support arrangement, preload, lubrication, and target life. Second, calculate Fm, select fW from the documented speed and vibration class, and calculate L10 and L10m from Ca and Fa [S1][S3].
Third, convert revolutions into hours and travel distance, then compare the result with the maintenance target. Fourth, reject any candidate that fails the manufacturer speed limit, the 80% critical-speed screening rule, or the buckling check [S5].
Final selection should require documented Ca, allowable axial load, maximum speed, critical speed, buckling load, preload, lead, accuracy class, lubrication method, and mounting tolerances. A model is ready for release only after its calculated repeated-duty life exceeds the stated target with the selected load factor and installation assumptions visible in the calculation sheet [S3].