How to Specify a Fastener: A Six-Step Method from Property Class to Surface Finish

1. Background: the cost of a wrong screw hides in the 85 %

The trade often cites the 15/85 rule from the iceberg model of total cost of ownership: the direct purchase price of a fastener is typically only about 15 % of what the joint costs, and the other 85 % hides in design selection, assembly efficiency, quality rework, logistics and warehousing, and after-sales. Specify the wrong screw and you may pay with the cycle time of an entire production line.

Selection is nonetheless the step most often compressed. A drawing says "M6 screw", the supplier ships the cheapest option that satisfies it, and the investigation only starts when the line reports stripped threads, galling, loosening or rust.

The order of the six steps matters: size first, then strength, then length, and only then torque, locking and corrosion protection. Reverse it and the usual result is a cheap finish chosen first, on a bolt whose property class was never needed, tightened to a torque that exceeds the allowable surface pressure of the base material.

2. Step 1: nominal diameter, worked back from the load

Nominal diameter follows the force, not the previous drawing. The VDI 2230 estimation method works backwards: take the working force closest to the joint load, convert it to the minimum required preload for the load type, then look up the thread size for the chosen property class.

Estimated bolt diameter for a single bolted joint at 20 °C, from VDI 2230. Rows are the required maximum preload, columns the property class:

The conversion path: first step down the table by load type — four rows for transverse shear loading, two for dynamic eccentric axial, one for dynamic concentric or static eccentric, none for static concentric axial — then step down again by tightening method: two rows for a simple nutrunner set by repeated torque, one for a torque wrench or precision nutrunner, none for angle or yield control.

Worked example: a dynamic eccentric axial force F_A of 8,500 N, class 12.9, hand torque wrench. Round up to 10,000 N, step down two rows to F_{M min} = 25,000 N, step down one more to F_{M max} = 40,000 N, which gives M10.

Note: an estimate is not a design basis. It has to be verified by calculation.

3. Step 2: property class, where stronger is not better

The class marking carries two pieces of information. The number before the point is one hundredth of the tensile strength; the number after it is the yield-to-tensile ratio. So 8.8 means 800 MPa tensile and a ratio of 0.8, giving about 640 MPa yield. Per ISO 898-1:

Three rules for choosing a class:

• Higher class means lower toughness. Class 12.9 offers 8 % elongation against 12 % for 8.8, and absorbs less impact. For joints under impact or alternating load, 8.8 is often better than 12.9.
• The class has to match the base material. Against aluminium, magnesium or die castings, a higher class fastener simply embeds itself deeper during tightening. The answer is more bearing area — flange heads, washers — or lower assembly torque, not a stronger bolt.
• Higher class means more hydrogen embrittlement risk. Parts at 1,000 MPa and above, roughly 320 HV, can pick up hydrogen during pickling and plating. They must be baked per ISO 4042 or they may fail by delayed fracture days later.

4. Step 3: engagement length

Insufficient engagement is the most common hidden failure source in high-strength joints. With too few threads the load concentrates in the first few, and the screw shears at the thread while the rest of the structure is still intact.

When a bolt threads into a tapped hole rather than a standard nut, the hole wall is less elastic than a nut, so the minimum engagement has to be set from the tensile strength of the component material. The table gives the minimum engagement as a multiple of nominal diameter d, excluding countersink depth:

Note: these are starting reference values in the VDI 2230 system for steel bolts into tapped holes. The standard itself, VDI 2230-1:2015 §5.5.5, presents minimum effective engagement m_eff/d as curves against the shear strength of the internal thread material, read separately for classes 8.8, 10.9 and 12.9: about 0.8 d for steel, and clearly above 1 d for aluminium and magnesium. The standard gives curves, not fixed numbers, so light alloy joints should be calculated or established by a thread stripping test.

Two traps:

• Excessive engagement can seize. Beyond about 1.5 d, if both internal and external threads happen to sit at their tolerance limits, the assembly can gall and refuse to turn. ISO 965-1 defines the tolerance classes that exist to keep different batches interchangeable.
• "1.5 × diameter" is not universal. It is conservative in steel-to-steel and clearly inadequate in light alloys. Soft aluminium needs 2.0–2.5 d, high-strength aluminium 1.4–2.0 d and magnesium 2.0–2.5 d to approach a joint equal in strength to the bolt. Class 12.9 should not be tapped directly into light alloy at all; a wire thread insert is the reliable answer.

5. Step 4: tightening torque and the friction coefficient

Tightening torque is not a fixed number from a table, it is a range. What actually decides assembly accuracy is friction and the tightening factor α_A: the same torque on a zinc-plated part and on an anti-friction coated part can produce preloads that differ by a factor of two.

Friction coefficient classes per VDI 2230:

The tightening factor α_A is the ratio of the maximum preload that may occur during assembly to the minimum that is required. Even with a tool holding ±2 %, actual preload scatter runs from ±9 % to ±60 %, and all of that difference comes from the tightening method:

One counter-intuitive conclusion: rather than chasing torque accuracy on the tool — ±2 % is already excellent — control the friction coefficient. Preload error from friction scatter is routinely ten times the tool tolerance. That is exactly why anti-friction dry-film coatings have spread so quickly in EV and precision assembly: they lock friction to a stable value.

6. Step 5: anti-loosening, after separating the two mechanisms

"The screw came loose" describes two entirely different failures with different answers:

• Loosening from relaxation: the contact faces embed and the material creeps, so preload drains away. The answer is preserve elasticity — remove soft or plastically deforming parts from the joint, use a long bolt (k₀ > 4d) or a reduced-shank bolt, and use a spring washer to compensate preload loss.
• Rotational loosening: dynamic transverse force makes the joint faces slip, spinning the thread until preload reaches zero. The answer is resist transverse movement — go up a size or class, use a fitted bolt or dowel, use ribbed or toothed flange faces, or a wedge-locking washer.

Which element suits which mechanism:

Legend: ● very effective, ○ effective, — not applicable. Effectiveness depends on the actual joint, so verify on a vibration rig before committing.

Whether locking is needed at all has its own criterion, based on the ratio of clamp length L to nominal diameter d and on the load direction:

• Short joint (L < 2d) with transverse dynamic load: locking is mandatory, no exceptions.
• Medium joint (2d ≤ L < 5d) with transverse dynamic load: verify the locking effect.
• Long joint (L ≥ 5d) with transverse dynamic load: decide from the verification result.
• Static axial load only: generally no additional locking required.

7. Step 6: surface finish by environment, not by price

The finish governs corrosion resistance, wear resistance, lubrication, insulation and galling. The boundaries of the nine mainstream processes:

Three rules of thumb:

• High-strength parts go to hydrogen-free processes first. At 1,000 MPa and above, prefer zinc flake or a non-electrolytic zinc flake coating and avoid electroplating. If plating is unavoidable, bake within four hours of plating, the sooner the better, at 200–230 °C and below the tempering temperature, typically for 4–24 hours.
• Check potentials when joining dissimilar metals. On aluminium or magnesium, corrosion protection outranks strength class. An aluminium bolt of the same family, such as a ZT68 high-strength aluminium bolt, removes the galvanic couple at source.
• For torque consistency, add a lubricating coating. A dry-film lubricant evens out surface roughness and locks the friction coefficient, giving the tightest torque scatter available on an automated line.

8. The check everyone skips: allowable surface pressure

After all six steps there is one verification that is routinely missed: the contact pressure under the head or nut must not exceed the allowable surface pressure of the base material. Once it does, the material embeds, preload drains away, and the joint fails with no visible sign of loosening.

Limit surface pressure p_G at room temperature with unchamfered holes, from VDI 2230 table A9, where the check requires p_max ≤ p_G:

Note: p_G values differ between editions of VDI 2230 — 42CrMo4 was raised in the 2015 edition of table A9 — and some entries are typical experimental values that do not rise monotonically with tensile strength. Use the edition you are working to, or measured data.

Three effective ways to reduce contact pressure:

• Use flange-head screws or nuts. More bearing area means less embedding, more stable clamp load, and a larger hole tolerance is permissible.
• Chamfer the hole. On steel, chamfering the contact face at the hole can raise allowable surface pressure by up to about 25 % through the supporting effect.
• Use a precision clearance hole. Specify H13 per ISO 273 so that an oversized hole does not quietly shrink the effective bearing area.

One more caution: when tightening steel parts with a power nutrunner, reduce the allowable surface pressure by 25 %.

9. FAQ

Q1: How do I choose between 8.8 and 10.9?
A: Start from load and toughness. General structural joints are well served by 8.8. Move to 10.9 only when the load and the required preload demand it. If the joint sees impact or alternating load, the lower elongation of 10.9 and above can work against you.

Q2: I tightened to the torque table and it still came loose. Why?
A: Most likely the actual friction differs from what the table assumed, so the real preload was low. Check the finish and lubrication state, recalculate torque from measured friction, and confirm the tightening factor was not taken too small.

Q3: Can I use class 12.9 bolts into aluminium?
A: Yes, but you must check the allowable surface pressure of the base material and increase the bearing area or reduce assembly torque. The safer route is a high-strength aluminium bolt, which also removes the galvanic couple.

Q4: Why do stainless screws rust outdoors?
A: A2 (304) pits in chloride environments — coastal air, de-icing salt, sweat. Specify A4 (316) or a higher surface finish for those conditions.

10. Quality and certification

All Toprecision products pass strict quality inspection and meet REACH and RoHS compliance requirements. Batch traceability is provided and critical dimensions are inspected 100 %.

The company holds ISO 9001 certification, and material certificates, salt spray reports and torque test reports are available for critical products.

Direct from the factory: more than twenty years of in-house production experience, serving humanoid robotics, AI server hardware, new energy and consumer electronics, with non-standard customisation, build-to-drawing, rapid prototyping and volume supply.