Yield Strength vs Tensile Strength in Bolts: Which Governs Preload, Which Governs Rupture

Which number is the design limit and which is the rupture limit

A grade standard does not give you one strength; it gives you two, and they sit at different levels. For ISO 898-1 classes the nominal Rm is 800, 1000 and 1200 MPa for 8.8, 10.9 and 12.9, while nominal Rp0.2 is 640, 900 and 1080 MPa for the same three classes ^{20, 24}. The minimum values follow the same order: 8.8 is 800 MPa tensile against 640 MPa yield, 10.9 is 1040 against 940, and 12.9 is 1220 against 1100 ²⁰. That gap is the whole point. Rp0.2 marks where the bolt starts to deform permanently, so it is the ceiling for anything you do to the bolt during tightening — preload lives below it. Rm marks the stress at which the bolt actually breaks, so it is the limit for the load the joint can survive in service ^{20, 24}. Size a joint on Rm alone and you have sized normal service against the wrong number: the bolt will yield long before it reaches the value you designed to. One caution before you use either figure. The nominal and minimum columns are not the same number, and a design that quotes "1000 MPa" for a 10.9 bolt is quoting the nominal value while the standard guarantees 1040 MPa minimum ^{20, 24}. Decide which column your calculation uses and say so on the drawing.

Comparing grades on the full property set, not tensile strength alone

Reading down the table shows the trade you are making. Each step up in class buys roughly 200 MPa of tensile and 300 MPa of yield, and gives back minimum elongation and reduction of area ²⁰. Elongation falls from 12% at 8.8 to 8% at 12.9, and reduction of area from 52% to 44% ²⁰. Those two numbers describe how much the bolt stretches and necks before it breaks — the reserve that lets a joint redistribute load instead of cracking. A 12.9 bolt is stronger and less forgiving, and the standard says so in the same table that gives its strength ²⁰.

Setting the preload target when torque is mostly friction

Preload is what holds the joint together, and it is the one quantity you cannot read off the grade table. The reason is mechanical: as the bolt turns down the thread helix, rotary motion becomes linear stretch, but most of the torque you apply is spent overcoming friction rather than stretching the bolt. One breakdown puts only about 10% of input torque into preload, with underhead friction taking roughly 50% and thread friction roughly 40% ¹³. That split is why the same torque figure produces different preload on two nominally identical joints. Change the lubricant and you change the friction, and the preload moves with it — a low-friction lubricant reduces the torque needed to reach the same preload ¹³. Where clamp load is specified, the friction coefficient becomes a controlled parameter rather than an accident of surface finish. Torque-tension testing per ISO 16047 measures the K-factor, with a typical target of µ = 0.12–0.18 for lubricated systems ¹⁴. So treat the preload target as a friction-controlled torque problem. Ask your supplier what friction coefficient the torque figure assumes, and whether it was validated on the actual joint rather than a generic table.

Choosing the design basis for preload and for ultimate load

You need two separate checks, because preload and rupture are limited by different values. Preload must stay under yield, so Rp0.2 is the number that governs it; the ultimate load the joint can carry is limited by Rm ^{20, 24}. Taking a safety factor against the wrong one is the common error — a factor applied to Rm looks generous while the bolt is already yielding. One widely circulated method sets preload as tensile strength divided by a safety factor of 2–3, giving 400 MPa for an 800 MPa bolt at a factor of 2 ⁷. That is a manufacturer's tensile-stress method. And it is worth knowing that it conflicts with the friction-based approach in the previous section: it derives a preload from Rm, while the torque evidence says the preload you actually get is set by friction ^{7, 13}. The two are not interchangeable, and the sources do not reconcile them. What you can do is keep the bases explicit. State whether your preload figure is a fraction of yield, a fraction of proof load, or a tensile-strength division, and state which value your ultimate-load check uses. If your customer's specification names a preload rule, that rule governs; the sources here do not establish a default one.

Whether a higher grade buys fatigue life

Fatigue is where the grade step stops paying. The endurance limit for rolled-thread alloy steel bolts is reported at roughly 7–12% of tensile strength, with a thread-root stress concentration factor of Kt = 3.5–4.5 ¹⁷. That is a guide-tier figure rather than a standard-mandated value, but the direction is clear: the stress the bolt tolerates in cyclic service is a small fraction of the number on the head. Thread form moves fatigue life more than class does. Rolled threads are reported to give roughly 2–3 times the fatigue life of cut threads, because rolling cold-works the thread root and leaves it in compressive residual stress ¹⁷. Preload helps too, by a different route: a higher mean stress from preload reduces the alternating stress amplitude the bolt sees. So the cyclic load is shared with the clamped parts instead of cycling the bolt ¹⁷. Higher classes also become more notch-sensitive as they get stronger, which works against fatigue rather than for it ²⁴. If your joint is fatigue-critical, the questions worth asking are about thread production route and preload control, not about whether 12.9 beats 10.9.

Ruling out the top classes in hydrogen or corrosive service

In hydrogen-charged service the usable stress is a fraction of yield, and that fraction shrinks as the class rises. The threshold stress is reported at 60–85% of yield for class 10.9 against 40–65% for class 12.9 under the same hydrogen conditions ⁹. The stronger grade therefore has the smaller safe window, which is the opposite of what the strength table suggests. The mechanism reported is hardness: class 10.9 sits at 32–39 HRC, in what is described as a transitional susceptibility zone, while class 12.9 is harder and tolerates less internal hydrogen before crack initiation ⁹. Class 12.9 is also described as more sensitive to notches, stress corrosion and hydrogen embrittlement, restricting it to specific designs ²⁴. Class 14.9 goes further still — it falls outside the ISO 898-1 scope, which stops at 12.9, and electroplating is typically prohibited for it, leaving only non-electrolytic coatings ²⁷. For a hydrogen-exposed joint, the practical move is to check whether a larger-diameter 10.9 bolt carries the load before reaching for 12.9. That trade is presented as a risk-reduction strategy by the source making it, not as a standard requirement ⁹.

What the certificate must show before you accept the quote

The design rests on minimum values, so the paperwork has to show minimum values. Proof load, yield strength and tensile strength are separate standard-set values, and not every standard specifies all three — SAE J429, for example, omits yield in favour of proof load ¹¹. If your calculation uses Rp0.2, a certificate that only reports proof load does not verify it.

• Confirm the certificate reports the minimum Rm and the minimum Rp0.2 for the class, not nominal values ^{20, 24}.
• Check which standard the certificate is issued against, and whether that standard specifies yield at all ¹¹.
• For hydrogen-exposed high-strength fasteners, require sustained-load testing at 75% of notched fracture strength or proof load for 200 hours minimum, with no failures permitted ^{10, 14}.
• Where the joint is critical, ask for the embrittlement ratio from incremental step loading — above 0.60 for general use, above 0.75 for critical applications ¹⁰.
• For high-temperature or corrosive service, ask what preload loss the supplier expects over the service interval and what inspection interval is recommended ²².

Where the sources disagree

Two disagreements matter to the design basis, and neither is resolved by the sources in this set. The first is how preload should be derived at all; the second is whether the numbers you are designing to are nominal or minimum.

FAQ

Related pages

• Bolts product page
• What Bolt Property Classes 8.8, 10.9 and 12.9 Encode — and When High Strength Stops Being the Right Choice
• Bolt Head Markings: What the Lines, Numbers and Letters Claim — and How to Verify the Grade Before Accepting a Lot
• Proof Load vs Yield and Tensile Strength in Bolt Selection: What the Evidence Settles
• Bolt Grade Selection: ISO 898-1 8.8/10.9/12.9 vs ASTM A307 and A325