Bearings and structure
Follow each load path through the structural loop, and check every element on it [1]. For each joint, a designer must do these.
- Choose a bearing arrangement that can carry the moment [2].
- Check life if it rotates, and static capacity if it does not [3], [4]. Slow robot joints usually need the static check.
- Check the mount: over-constraint, preload and misalignment [5], [6].
- Check structural strength, deflection and resonance with a stated factor [7], [8], [9].
- Choose a material from what the part must do, then check the process in chapter 8.
Most bearing parts have a price row, but none comes from a store whose terms allow copying, so this site does not show them. The THK and IKO catalog rows have no price. Stock materials such as aluminium and steel have no price rows in this edition.
Bearing classes
| Class | Choose it for | Main costs | Parts |
|---|---|---|---|
| Ball bearing | Radial and moderate axial load, high speed | Point contact; limited moment unless paired | Deep groove M10-26-8 |
| Roller bearing | Higher radial capacity | Lower axial capacity; sensitive to misalignment | Cylindrical and tapered tables in the category |
| Crossed-roller | Radial, axial and moment load in one slim bearing | Needs a stiff, accurate housing; more friction than ball bearings | THK RB 5013, IKO CRB 5013 |
| Thin-section | Large bore, hollow joints | Thin rings deform with the housing; lower capacity | The class note names IKO CRBS, but no CRBS row is in the database; thin-section ball families are not collected |
| Spaced pair or 4-point | A moment load, with two bearings spaced on a shaft or one 4-point bearing | Spaced pair needs length; 4-point has substantial slip | [2] |
Catalog ranges are minimum / median / maximum.
| Subcategory | Dynamic rating C | Static rating C0 | Other | Note |
|---|---|---|---|---|
| Crossed-roller, THK | 3.23 / 66.9 / 494 kN | 3.1 / 128 / 1490 kN | Mass 0.04 / 3.14 / 120 kg | 51 rows |
| Crossed-roller, IKO | not shown | not shown | Bore 30 / 60 / 90 mm, mass 0.12 / 0.33 / 0.81 kg | 108 rows; 7 shown |
| Deep groove ball | 0.423 / 1.46 / 8.52 | 0.116 / 0.465 / 3.4 | Speed limit 32000 / 53000 / 85000 | Units not stated in the data file |
| Angular contact ball | 7.02 / 11 / 26.5 | 3.35 / 5.85 / 15.3 | Speed limit 15000 / 22000 / 30000 | Units not stated in the data file |
| Cylindrical roller | 12.5 / 28.5 / 47.5 | 10.2 / 22 / 38 | Speed limit 16000 / 19000 / 26000 | Units not stated in the data file |
| Tapered roller | 19 / 35.8 / 76.5 | 18.6 / 43 / 85 | Speed limit 10000 / 13000 / 18000 | Units not stated in the data file |
The ball and roller rows come from an open CAD table whose file states no units. The usual catalog units are kN and rpm, but the file does not confirm them. Check the maker catalog before you size from these rows.
IKO lists CRB as full complement and CRBC as with cage, each open or sealed (UU). At one size the full complement type rates higher than the caged type: compare CRB 8016 with CRBC 8016 on the IKO sheet. The database holds both ratings, but this site does not show them. IKO mass is one reference value per size.
Rating life
Use this for bearings that rotate under load.
- Find the equivalent load
P. Add misalignment forceF = k d[6]. - Check the equation applies:
P_r <= min(C0r, 0.5*Cr). Above this, NSK says to consult it on whether the rating fatigue-life equations apply [10]. - Compute
L = a1 * aSKF * (C/P)^p, withpof 3 for balls and 10/3 for rollers [11]. - Convert to hours:
L10h = 10^6 / (60 * n) * L10[12]. - Compare to the required service hours [3]. Then run the checklist: minimum load, speed limit, misalignment and stabilisation class [13].
L10 is a statistical value: 90 percent of a large group reach it [14]. For higher reliability, apply a1.
| Reliability, percent | 90 | 95 | 96 | 97 | 98 | 99 |
|---|---|---|---|---|---|---|
a1 [15] | 1.00 | 0.62 | 0.53 | 0.44 | 0.33 | 0.21 |
| Temperature, C | 125 | 150 | 175 | 200 | 250 |
|---|---|---|---|---|---|
ft on the load rating [16] | 1.00 | 1.00 | 0.95 | 0.90 | 0.75 |
Duty matters. A cross-roller load factor fw is 1 to 1.2 for steady operation and 1.5 to 3 with impact and vibration [17], [18].
Sources differ on the reliability basis. Commercial makers rate at L10 [14], [19]. NASA spaceflight mechanisms use L0.05 under maximum time-varying loads [20]. State which one you used.
Static capacity
Use s0 = C0 / P0 at the maximum load combination when the bearing is stationary under load, oscillates slowly under load, or sees temporary peak loads [4], [21]. Exceeding the rated static load makes craters that quickly destroy the bearing [22].
The lower limit depends on the source and duty. These do not agree.
| Source and use | Ball | Roller or guide | Source |
|---|---|---|---|
| NSK, low noise | 2.0 | 3.0 | [23] |
| NSK, vibration and shock | 1.5 | 2.0 | [24] |
| NSK, standard conditions | 1.0 | 1.5 | [25] |
| Harmonic Drive cross-roller, not rotating, slight oscillation | not listed | 0.5 | [26] |
| THK LM Guide, no vibration or impact | not listed | 2 | [27] |
| THK LM Guide, vibration or impact | not listed | 5 | [28] |
Small oscillation has its own failure. Under 5 degrees on a cross-roller, grease may not circulate and fretting can result [29].
Minimum load, mounting and layout
Bearings can fail from too little load. Apply 0.01 C on balls and 0.02 C on rollers [30], with extra care above 50 percent of limiting speed [31].
| Level | Statement | Check | Source |
|---|---|---|---|
| must | Avoid mounting that over-constrains bearings; tightening bolts or nuts can exceed bearing capacity. | review: one locating and one floating bearing, or a calculated preload | [5] |
| should | Add force F = k d from mounting misalignment d to the bearing load in the life calculation. | calc: P_total includes k * misalignment | [6] |
| should | Preload rolling element bearings to raise stiffness and stop skidding, but check heat from preload. | calc: preload force noted with stiffness and bearing heat estimate | [32] |
| must | Preload all ball bearings, except four-point, deep groove where absence is shown harmless, and wave generator bearings. | review: preload record per ball bearing | [33] |
| should | Use a spring preload rather than a rigid one when wear or dimensional variation could lose preload. | review: preload method notes wear and dimensional variation | [34] |
| should | Use bearing spacing L/D above 3 if bearings must constrain a shaft like a cantilever; below 3 check edge loading from shaft bending slope. | model-lint: bearing spacing / shaft diameter >= 3 | [35] |
| should | To support a moment load, use two bearings spaced apart on a shaft. One four-point contact bearing can also carry moment loads, but with substantial slip. | review: joint moment load path uses two spaced bearings or a moment-rated bearing | [2] |
| must | For NASA mechanisms, bearings shall not carry electrical current; provide other current paths and bonding/ground paths between moving and stationary parts. | review: moving joints have a slip ring or flexible bond strap | [36] |
| should | Check shaft forces when deciding whether ball bearings or less expensive sintered bearings are sufficient. | calc: shaft forces <= motor rating; ball bearing vs sintered per load | [37] |
Linear guides
Life numbers for balls and rollers use different definitions, so do not compare them directly: 50 km and 100 km [38]. THK says its nominal-life formulas may not apply when stroke is at or below twice the block length [39].
| Level | Statement | Check | Source |
|---|---|---|---|
| must | Compute linear guide nominal life from C based on 50 km for balls or 100 km for rollers, and do not compare life numbers across the two definitions without converting. | calc: L10 from C and calculated load by the maker formula, with a stated 50 km or 100 km rating basis | [38] |
| guide | THK nominal-life formulas may not apply when stroke length is at or below twice the LM block length. | calc: stroke > 2 * block length, otherwise ask the maker | [39] |
| must | Keep linear guide raceway hardness between 58 and 64 HRC; lower hardness lowers the load ratings. | review: rail hardness in the specification | [40] |
| should | Keep a standard LM Guide at 80°C or below; above that, use a high-temperature type and ask THK. | measure: rail temperature <= 80 C | [41] |
| should | Reduce load capacity with a contact factor when blocks sit close together: 0.81 for two, 0.72 for three, 0.66 for four. | calc: C_effective = fC * C | [42] |
| should | Keep the length-to-spacing ratio of a linear bearing carriage above 1 to avoid jamming; 1.6:1 is very good and 3:1 is excellent. | model-lint: carriage L/D > 1, prefer 1.6 or more | [43] |
| should | Actuate both sides of a slide if L/D is below 1. | review: wide slides have two driven sides or central guiding | [44] |
| should | Check the linear guide static safety factor per load direction: radial, reverse-radial and lateral each have their own rating. | calc: fS per direction using C01, C0L1, C0T1 | [45] |
Structure
Factors of safety
State the factor and the basis together. These sources serve different purposes.
| Source | Case | Factor | Source |
|---|---|---|---|
| NASA spaceflight metal | Prototype verified: ultimate, qualification | 1.4, 1.4 | [46] |
| NASA spaceflight metal | Protoflight verified: ultimate, yield, test | 1.4, 1.25, 1.2 | [47] |
| Slocum | General, for load uncertainty | about 2 | [48] |
| Slocum | Severe impact load factor, early feasibility only | up to 3 | [49] |
| Slocum | Hot, humid or corrosive environment | up to 3, otherwise 1 | [50] |
| Slocum | Stress concentration | up to 3, usually under 2 with good design | [51] |
| NASA mechanisms | Life test if failure could cost a life | cycles x 4.0 | [52] |
| NASA mechanisms | Life test, other mechanisms | cycles x 2.0 | [53] |
| NASA structures | Fatigue and creep life factor | 4.0 | [54] |
Slocum's factor of about 2 is for uncertainty in actual load magnitudes [48]. The NASA factors are for prototype or protoflight verified spaceflight metal [46], [47]. If you do not test, do not use the NASA number. Minimum factors are a floor, and stiffness or safety can override them [55].
Strength, fatigue and buckling
Report MS = allowable / (limit * FS) - 1 on both yield and ultimate [7]. Do not assume an endurance limit for aluminium: it needs a finite design life [56]. For infinite life in steel, keep stress below half of yield [57].
| Level | Statement | Check | Source |
|---|---|---|---|
| should | Use preload so stress does not alternate between positive and negative in fatigue-loaded parts. | calc: stress ratio R >= 0 on cycled members where possible | [58] |
| must | Analyse any member with significant in-plane compression or shear for buckling at ultimate design load. | calc: P_cr > ultimate design compression load | [59] |
| should | Evaluate general instability, local instability, crippling and creep in buckling checks. | review: buckling sheet lists the four modes | [60] |
| should | Do not use a factored load if a loading condition relieves buckling; take the unfactored load. | calc: alleviating loads are not multiplied by FS | [61] |
Deflection and resonance
Robot links are not rigid. Sum compliances along the path, 1/k_system = sum(1/k_i) [62], and keep deflection 3 to 5 times below critical clearances [8]. Machine structures can have a Q of 20, and the added deflection can overload bearings and reduce their life [63]. FEA predicts frequency, but do not trust it for amplitude [64].
| Level | Statement | Check | Source |
|---|---|---|---|
| guide | Stay several characteristic dimensions, taken as 3 to 5, away from a load to avoid its local deformation. | model-lint: critical feature >= 3 characteristic dimensions from a local load | [65] |
| should | Include shear deformation for beams with length-to-thickness ratio below about 3 to 5. | calc: Timoshenko check when L/h is under 5 | [66] |
| should | Treat Abbe error as Abbe offset times the sine of the angular error, and minimise the offset. | calc: e = offset * sin(angle error) | [67] |
| should | Model link and joint flexibility if the mass cost of rigid design is not acceptable; flexible joints and links add states and complicate control. | sim: flexible link and joint model when stiffness is limited by mass | [68] |
| must | Avoid forcing at a structural natural frequency or an integer multiple; only good damping prevents build-up. | calc: excitation frequency list vs natural frequencies, with separation | [9] |
Mounting and stops
| Level | Statement | Check | Source |
|---|---|---|---|
| must | Analyse mounting stiffness, alignment tolerance, thermal and load distortion, and interface friction. | review: analysis lists each boundary term | [69] |
| should | Design exposed mechanism parts and covers for inadvertent contact where manipulators or people can touch them. | calc: covers take the stated accidental load | [70] |
| should | Do not bolt a component onto an imperfect surface without allowing for congruence; bolting can warp a structure and misalign critical parts. | measure: flatness and critical alignment after bolting | [71] |
| should | Space bolts so strain cones overlap; the cone projects at 30 to 45 degrees from vertical, and non-overlap causes waviness that can make bearings fail. | model-lint: strain cones of adjacent bolts overlap at the interface | [72] |
| guide | Keep material under a bolt head one to three bolt diameters thick, and bolt centres three to five diameters apart. | model-lint: flange thickness 1 to 3 d; pitch 3 to 5 d | [73] |
| should | Thread engagement equal to one bolt diameter gives thread shear strength equal to bolt tensile strength; more adds little. | model-lint: engagement >= 1 d in the same strength material; recompute for weaker material | [74] |
| must | Do not thread both parts of a bolted pair; threads will not be in phase. | review: bolted pair has one clearance hole | [75] |
| must | Fit non-jamming mechanical stops wherever overtravel harms the mechanism or system. | review: hard stops per joint | [76] |
| must | Keep positive margin with full design factors under worst-case transient loads from stop impact. | calc: impact load at stop within margin at max speed | [77] |
| must | Keep the mechanism functional after stall at any point in its travel. | measure: stall test at several positions, then a performance test | [78] |
Materials
Classes are on the material category. Values below are from part pages. Most are encyclopedia or open-card values, so confirm them on the supplier sheet before a critical calculation.
| Material | Density | Modulus | Yield | Ultimate | Conductivity, W/m K |
|---|---|---|---|---|---|
| Aluminium 6061-T6 | 2.7 g/cm3 | 69 to 70 GPa | 240 to 270 MPa | 260 to 290 MPa | 152 to 166 |
| Aluminium 7075-T6 | 2.81 g/cm3 | 71.7 GPa | 430 to 480 MPa | 510 to 540 MPa | 130 to 150 |
| Aluminium 6063-T6 | 2.7 g/cm3 | 69.5 GPa | 170 MPa minimum | 215 MPa minimum | 201 |
| Aluminium 5052-H32 | 2.68 g/cm3 | 70 GPa | 130 MPa minimum | not listed | 138 |
| Aluminium 2024-T3 | 2.78 g/cm3 | 73 GPa | 270 MPa | 400 MPa | not listed |
| Magnesium alloys | 1.8 g/cm3 | 42 GPa, casting alloys | not listed | 135 to 285 MPa cast, 180 to 440 MPa wrought | not listed |
| Steel S235JR | 7800 kg/m3 | 210000 MPa | 235 MPa | 360 MPa | not listed |
| Steel 34CrNiMo6 | 7800 kg/m3 | 210000 MPa | 1000 MPa | 1200 MPa | not listed |
| Stainless 304 | 7900 kg/m3 | 183 to 200 GPa | 210 to 1050 MPa | not listed | not listed |
| Ti-6Al-4V | 4.429 to 4.512 g/cm3 | 104 to 113 GPa | 880 to 920 MPa | 900 to 950 MPa | 6.7 to 7.5 |
| PEEK | 1.32 g/cm3 | 3.6 GPa | not listed | 90 to 100 MPa | not listed |
| POM copolymer | 1410 kg/m3 | 2660 MPa | 59.8 MPa | not listed | 0.304 |
How the classes differ:
- Aluminium is the default structure metal, and a good heat path for motors. 6061-T6 is the all-rounder. 7075-T6 is much stronger, but it is costlier and welds worse. 5052 bends well in sheet.
- Steel is right for shafts, pins, gear blanks and fasteners. It is about 2.9 times the density of aluminium, with higher modulus. Stainless 304 cannot be heat treated, only cold worked. Its yield row spans 210 to 1050 MPa.
- Titanium suits weight-critical, high-load joints where aluminium and steel both fail. It is costly and slow to machine, and low conductivity makes machining hot.
- Magnesium suits hand-held or wearable housings where each gram counts. It needs a coating and careful chip handling, since it burns once ignited.
- Carbon fibre is stiff along the fibre only. A unidirectional laminate can be much stiffer than aluminium in that direction, and far less across it. The catalog lists Toray T700S by name only. It is conductive, it fails without warning, and it is poor in bearing and clamping loads without inserts or bonded bushings.
- Engineering plastics have modulus of 0.7 to 4.5 GPa, far below metals. They creep under constant load. Use POM for gears and bushings, UHMW for sliding wear pads, PEEK and PPS for hot parts.
- T-slot extrusion makes a frame you can reconfigure. Slot nuts can slip under shock. The extrusion category shows profile size and slot width. Mass per metre and second moment of area come only from Alufab sheets such as the TS25-25M, which the catalog lists by name only. Take them from the maker's sheet before you size a beam.
Sources
Each guideline in this chapter with its source, the date it was read and the exact words of the source. Level: must means breaking it causes failure or harm, should is good practice, guide is a rule of thumb.
- Follow each load path through the structural loop and evaluate each element's failure risk. Level: should. pergatory.mit.edu 2026-11-08
- To support a moment load, use two bearings spaced apart on a shaft. One four-point contact bearing can also carry moment loads, but with substantial slip. Level: should. pergatory.mit.edu 2026-11-08
- Compare the bearing L10 rating life with the specification life of the application. Level: should. skf.com 2026-11-08
- For a bearing that is stationary under load, oscillates slowly under load, or sees temporary peak loads, calculate s0 = C0/P0 using the maximum load combination. Level: should. skf.com 2026-11-08
- Avoid mounting that over-constrains bearings; tightening bolts or nuts can exceed bearing capacity. Level: must. pergatory.mit.edu 2026-11-08
- Add force F = k d from mounting misalignment d to the bearing load in the life calculation. Level: should. pergatory.mit.edu 2026-11-08
- Report margin of safety as allowable load divided by the product of limit load and factor of safety, minus 1, for yield and ultimate. Level: should. standards.nasa.gov 2026-11-08“Margin of Safety (MS): MS = [Allowable Load (Yield or Ultimate)/Limit Load*Factor of Safety (Yield or Ultimate)] - 1.”
- Make structural deflection 3 to 5 times less than critical clearances. Level: should. pergatory.mit.edu 2026-11-08
- Avoid forcing at a structural natural frequency or an integer multiple; only good damping prevents build-up. Level: must. pergatory.mit.edu 2026-11-08
- If radial load Pr exceeds the smaller of C0r and 0.5Cr, consult NSK to establish applicability of the rating fatigue-life equations. 0.5 Level: should. res.cloudinary.com 2026-11-08“or 0.5 C r, whichever is smaller, for radial bearings or”
- Use exponent p = 10/3 for roller bearings and p = 3 for ball bearings in the rating life equation. Level: guide. skf.com 2026-11-08
- At constant speed, convert rating life to hours by multiplying the life in millions of revolutions by 10^6 / (60 n), with n in r/min. Level: guide. skf.com 2026-11-08
- After sizing a bearing, check allowed axial and radial load, minimum load, speed limit, misalignment and stabilisation class. Level: should. skf.com 2026-11-08
- Treat bearing L10 as the life that 90 percent of a large group of identical bearings reach; it is a statistical value, not a promise for one bearing. 90 percent Level: guide. skf.com 2026-11-08
- For reliability above 90 percent, multiply rating life by the a1 factor: 1.00, 0.62, 0.53, 0.44, 0.33, 0.21 for the 90, 95, 96, 97, 98, 99 percent columns. 0.21 Level: should. res.cloudinary.com 2026-11-08“Reliability (%) 90 95 96 97 98 99 Temperature oC Temperature Factor f t 1.00 1.00 0.95 0.90 0.75 a1 1.00 0.62 0.53 0.44 0.33 0.21”
- Correct the basic load rating by temperature factor ft for hot bearings: 1.00, 1.00, 0.95, 0.90, 0.75 across the five temperature columns 125, 150, 175, 200, 250 C. 0.75 Level: should. res.cloudinary.com 2026-11-08“125 150 175 200 250 Reliability (%) 90 95 96 97 98 99 Temperature oC Temperature Factor f t 1.00 1.00 0.95 0.90 0.75”
- Multiply the equivalent load by a load factor fw in the cross-roller life calculation: 1 to 1.2 for steady operation without impact and vibration. Level: should. harmonicdrive.net 2026-11-08
- Use a load factor fw of 1.5 to 3 for cross-roller bearing life when operation includes impact and vibration. Level: should. harmonicdrive.net 2026-11-08
- Timken defines L10 as the hours 90% of identical bearings reach before fatigue spalling and associates a single bearing's L10 with 90% reliability under a particular load and speed. 90 percent Level: should. engineering.timken.com 2026-11-08
- For NASA spaceflight mechanisms, base bearing fatigue-life calculations on L0.05 life under maximum time-varying loads consistent with its determination conditions. Level: must. standards.nasa.gov 2026-11-08“[DDMR 40] Bearing fatigue life calculations shall be based on the L0.05 life when subjected to”
- Verify bearing size on static load when it is not rotating under load, oscillates slowly under load, or rotates with temporary high peaks. Level: must. skf.com 2026-11-08
- Do not exceed the rated static load, because exceeding the maximum Hertz stress creates craters that quickly destroy the bearing. Level: must. pergatory.mit.edu 2026-11-08
- For low-noise use NSK gives a lower limit of static load factor fs of 2.0 for ball and 3.0 for roller bearings. 3.0 Level: should. res.cloudinary.com 2026-11-08“Low-noise applications 2.0 3.0”
- For bearings under vibration and shock NSK gives a lower limit of fs of 1.5 for ball and 2.0 for roller bearings. 2.0 Level: should. res.cloudinary.com 2026-11-08“shock loads 1.5 2.0”
- Under standard operating conditions NSK gives a lower limit of fs of 1.0 for ball and 1.5 for roller bearings. 1.5 Level: should. res.cloudinary.com 2026-11-08“Standard operating conditions 1.0 1.5”
- For a Harmonic Drive cross-roller output bearing normally not rotating with slight oscillation, the lower static-factor limit is 0.5. 0.5 Level: should. harmonicdrive.net 2026-11-08
- Use a static safety factor of at least 2 for an LM Guide without vibration or impact. 2 Level: should. thk.com 2026-11-08
- Use a static safety factor of at least 5 for an LM Guide with vibration or impact, including acceleration and sudden starts. 5 Level: should. thk.com 2026-11-08
- On a cross-roller bearing, oscillation angles below 5 degrees may cause fretting corrosion because lubrication may not circulate properly. 5 degrees Level: should. harmonicdrive.net 2026-11-08
- Apply a minimum load of 0.01 C on ball bearings and 0.02 C on roller bearings to avoid skidding and raceway smearing. 0.01 Level: should. skf.com 2026-11-08
- Check minimum load with extra care where speeds exceed 50 percent of limiting speed, or where acceleration is rapid. 50 percent of limiting speed Level: should. skf.com 2026-11-08
- Preload rolling element bearings to raise stiffness and stop skidding, but check heat from preload. Level: should. pergatory.mit.edu 2026-11-08
- Preload all ball bearings, except four-point, deep groove where absence is shown harmless, and wave generator bearings. Level: must. standards.nasa.gov 2026-11-08“[DDMR 42] All ball bearings shall be preloaded with the following exceptions:”
- Use a spring preload rather than a rigid one when wear or dimensional variation could lose preload. Level: should. pergatory.mit.edu 2026-11-08
- Use bearing spacing L/D above 3 if bearings must constrain a shaft like a cantilever; below 3 check edge loading from shaft bending slope. 3 Level: should. pergatory.mit.edu 2026-11-08
- For NASA mechanisms, bearings shall not carry electrical current; provide other current paths and bonding/ground paths between moving and stationary parts. Level: must. standards.nasa.gov 2026-11-08“[DDMR 20] Bearings shall not be used to carry electrical current.”
- Check shaft forces when deciding whether ball bearings or less expensive sintered bearings are sufficient. Level: should. maxonmotor.com 2026-11-08
- Compute linear guide nominal life from C based on 50 km for balls or 100 km for rollers, and do not compare life numbers across the two definitions without converting. 50 km Level: must. thk.com 2026-11-08
- THK nominal-life formulas may not apply when stroke length is at or below twice the LM block length. Level: guide. thk.com 2026-11-08
- Keep linear guide raceway hardness between 58 and 64 HRC; lower hardness lowers the load ratings. 58 HRC Level: must. thk.com 2026-11-08
- LM Guides not designed for high temperatures should be used at 80°C or less; contact THK if requirements exceed 80°C. 80 C Level: should. thk.com 2026-11-08
- Reduce load capacity with a contact factor when blocks sit close together: 0.81 for two, 0.72 for three, 0.66 for four. 0.81 Level: should. thk.com 2026-11-08
- Keep the length-to-spacing ratio of a linear bearing carriage above 1 to avoid jamming; 1.6:1 is very good and 3:1 is excellent. 1.6 Level: should. pergatory.mit.edu 2026-11-08
- Actuate both sides of a slide if L/D is below 1. Level: should. pergatory.mit.edu 2026-11-08
- Check the linear guide static safety factor per load direction: radial, reverse-radial and lateral each have their own rating. Level: should. thk.com 2026-11-08
- For prototype verified metal spaceflight structure use ultimate design factor 1.4 and qualification test factor 1.4, with no detrimental yield. 1.4 Level: must. standards.nasa.gov 2026-11-08“Prototype 1.4 1.0* 1.4 N/A or 1.05**”
- For protoflight verified metal spaceflight structure use ultimate 1.4, yield 1.25, and qualification test factor 1.2. 1.25 Level: must. standards.nasa.gov 2026-11-08“Protoflight 1.4 1.25 1.2 N/A or 1.05**”
- Use a general safety factor of about 2 for uncertainty in actual load magnitudes. 2 Level: guide. pergatory.mit.edu 2026-11-08
- Use a load factor up to 3 for severe dynamic impact only for initial feasibility; large impacts need simulation. 3 Level: guide. pergatory.mit.edu 2026-11-08
- Add an environmental factor up to 3 for hot, humid or corrosive environments, otherwise 1. 3 Level: guide. pergatory.mit.edu 2026-11-08
- Include a stress concentration factor; it can reach 3, but with good design is usually under 2. 3 Level: should. pergatory.mit.edu 2026-11-08
- Life test mechanisms whose failure could cost a human life to total operational and ground cycles times at least 4.0. 4.0 Level: must. standards.nasa.gov 2026-11-08“maintenance cycles) by a minimum life test factor of 4.0.”
- Life test other mechanisms to total operational and ground cycles times at least 2.0. 2.0 Level: must. standards.nasa.gov 2026-11-08“maintenance cycles) by a minimum life test factor of 2.0.”
- Apply a minimum service life factor of 4.0 to fatigue and creep assessment in NASA spaceflight structures. 4.0 Level: must. standards.nasa.gov 2026-11-08“factor of 4.0 shall be applied to the service life for fatigue and creep life assessments.”
- Treat minimum factors of safety as a floor; stiffness, pointing accuracy and safety requirements can override them. Level: must. standards.nasa.gov 2026-11-08“requirements, such as dimensional stability, pointing accuracy, stiffness/frequency constraints,”
- Do not assume an endurance limit for aluminium: give it a finite design life. Level: must. pergatory.mit.edu 2026-11-08
- For infinite fatigue life in steel keep applied stress below half of yield stress. Level: should. pergatory.mit.edu 2026-11-08
- Use preload so stress does not alternate between positive and negative in fatigue-loaded parts. Level: should. pergatory.mit.edu 2026-11-08
- Analyse any member with significant in-plane compression or shear for buckling at ultimate design load. Level: must. standards.nasa.gov 2026-11-08“[FSR 53] Design loads for buckling shall be ultimate loads.”
- Evaluate general instability, local instability, crippling and creep in buckling checks. Level: should. standards.nasa.gov 2026-11-08“[FSR 55] Buckling evaluation shall address general instability, local or panel instability,”
- Do not use a factored load if a loading condition relieves buckling; take the unfactored load. Level: should. standards.nasa.gov 2026-11-08“[FSR 54] If a loading condition tends to alleviate buckling, then the unfactored load shall”
- Sum structural and element compliances to obtain system compliance. Level: should. pergatory.mit.edu 2026-11-08
- Machine structures can have Q of 20; increased deflection can overload bearings and reduce their life. 20 Level: should. pergatory.mit.edu 2026-11-08
- Use FEA to predict resonant frequencies but do not rely on it for amplitude, since damping is hard to predict. Level: should. pergatory.mit.edu 2026-11-08
- Stay several characteristic dimensions, taken as 3 to 5, away from a load to avoid its local deformation. Level: guide. pergatory.mit.edu 2026-11-08
- Include shear deformation for beams with length-to-thickness ratio below about 3 to 5. Level: should. pergatory.mit.edu 2026-11-08
- Treat Abbe error as Abbe offset times the sine of the angular error, and minimise the offset. Level: should. pergatory.mit.edu 2026-11-08
- Model link and joint flexibility if the mass cost of rigid design is not acceptable; flexible joints and links add states and complicate control. Level: should. hades.mech.northwestern.edu 2026-11-08
- Analyse mounting stiffness, alignment tolerance, thermal and load distortion, and interface friction. Level: must. standards.nasa.gov 2026-11-08“[DDMR 28] Engineering analyses shall account for the structural mounting boundary conditions,”
- Design exposed mechanism parts and covers for inadvertent contact where manipulators or people can touch them. Level: should. standards.nasa.gov 2026-11-08“other situations presenting a risk of inadvertent contact are present, exposed mechanism components,”
- Do not bolt a component onto an imperfect surface without allowing for congruence; bolting can warp a structure and misalign critical parts. Level: should. pergatory.mit.edu 2026-11-08
- Space bolts so strain cones overlap; the cone projects at 30 to 45 degrees from vertical, and non-overlap causes waviness that can make bearings fail. Level: should. pergatory.mit.edu 2026-11-08
- Keep material under a bolt head one to three bolt diameters thick, and bolt centres three to five diameters apart. Level: guide. pergatory.mit.edu 2026-11-08
- Thread engagement equal to one bolt diameter gives thread shear strength equal to bolt tensile strength; more adds little. Level: should. pergatory.mit.edu 2026-11-08
- Do not thread both parts of a bolted pair; threads will not be in phase. Level: must. pergatory.mit.edu 2026-11-08
- Fit non-jamming mechanical stops wherever overtravel harms the mechanism or system. Level: must. standards.nasa.gov 2026-11-08“[DDMR 30] Non-jamming mechanical stops shall be incorporated into all mechanisms where”
- Keep positive margin with full design factors under worst-case transient loads from stop impact. Level: must. standards.nasa.gov 2026-11-08“factors of safety applied when subjected to worst-case transient loads from mechanical stop impact.”
- Keep the mechanism functional after stall at any point in its travel. Level: must. standards.nasa.gov 2026-11-08“[DDMR 29] The mechanism shall remain functional after exposure to stall conditions at any point in”