# Why High-Temperature Bolts Fail: 5 Failure Modes and How to Prevent Them

> Analysis of 340 returned bolts from 12 customer sites reveals five failure modes account for 97% of high-temperature bolt failures. Forensic breakdown with prevention protocols for stress relaxation, creep rupture, hydrogen embrittlement, galling, and thermal fatigue.


**We analyzed 340 returned bolts from 12 customer sites over three years. Five failure modes accounted for 97% of cases. None of them were material defects.**

Every bolt that landed on our failure analysis bench had passed its original outgoing inspection. The mill test certificates were in order. Chemistry checked out. Mechanical properties met the purchase specification. Yet somewhere between installation and the first scheduled outage, these fasteners lost preload, fractured, or seized, and in three cases the resulting flange separation caused unplanned shutdowns costing over $400,000 each.

The pattern was unmistakable. The bolts did not fail because they were made wrong. They failed because they were specified wrong, or installed without accounting for the mechanisms that dominate above 500°C.

> **TLDR:** Stress relaxation (38%), creep rupture (22%), hydrogen embrittlement (18%), galling during assembly (12%), and thermal fatigue (7%). These are not random failures. Each has a predictable root cause and a proven prevention protocol. Here is what we found.

## 1. Stress Relaxation: 38% of Cases

Stress relaxation is the silent killer of bolted joints at temperature. Unlike creep, which produces measurable elongation, relaxation occurs at constant strain: the bolt stays the same length, but the clamping force decays. At 600°C, a standard A-286 bolt torqued to 50% of yield loses approximately 28% of its preload within the first 200 hours, without any visible indication.

The mechanism is straightforward. At elevated temperature, elastic strain in the bolt shank converts to plastic (creep) strain at the thread roots and under-head bearing surface. The total bolt length does not change, so you cannot detect it with a wrench. But the joint clamping force, the only thing keeping the gasket seated, is bleeding away.

**What we found in the returned bolts:** 129 of the 340 bolts showed evidence of relaxation-driven leakage. In 84% of those cases, the end user had specified a standard flat washer. The remaining 16% had no washer at all. None of the 129 had a Belleville spring washer.

**The fix:** Belleville (conical spring) washers compensate for relaxation by maintaining elastic follow-up. A properly sized Belleville stack, typically two washers in series for travel and in parallel for load, can recover up to 0.5 mm of joint settlement and thermal relaxation before the gasket unloads. For service above 550°C, specify Inconel 718 or Waspaloy Bellevilles, not standard spring steel. The extra cost is approximately $3-8 per bolt. The alternative is a leak path.

**Key specification:** ASME PCC-1 Appendix O provides the accepted method for calculating required Belleville travel. Do not guess.

## 2. Creep Rupture: 22% of Cases

If stress relaxation is the slow bleed, creep rupture is the sudden fracture. At temperatures above approximately 0.4 times the alloy's melting point (in Kelvin), time-dependent plastic deformation becomes the dominant failure mechanism. A bolt that held 550 MPa at room temperature may sustain only 110 MPa at 750°C over a 100,000-hour design life, a factor of five reduction in allowable stress.

**The Larson-Miller parameter (LMP)** is the industry-standard tool for predicting creep life. LMP = T (20 + log t), where T is in Kelvin and t is time in hours. For a given alloy and stress level, the LMP is constant. If you know two of the three variables (temperature, stress, or time), you can calculate the third.

Among our 75 creep rupture cases, the dominant error was grade selection. Bolts ordered as "high-temperature fasteners" without a specific alloy designation were frequently supplied in AISI 310 stainless steel, a grade adequate for oxidation resistance to 1,000°C but with a creep-limited stress of only 15 MPa at 800°C. The correct grade for that service condition would have been Nimonic 80A (creep-limited stress ~85 MPa at 800°C) or Inconel 718 for service below 650°C where precipitation hardening remains effective.

**Grade selection above 700°C:**
- **650-760°C:** Inconel 718 (aged), A-286 (solution treated + aged)
- **760-870°C:** Nimonic 80A, Waspaloy, Inconel X-750 (AMS 5667)
- **870-980°C:** Hastelloy X, Inconel 625 (note: oxidation-limited, not creep-optimised)
- **Above 980°C:** RA330, 253 MA (for oxidation resistance only, consult a creep specialist)

**Critical check:** If your bolt specification lists only room-temperature tensile properties, you are designing blind. Demand elevated-temperature tensile data at the service temperature. ASTM E139 covers standard creep testing, and ASTM E292 covers stress-rupture.

## 3. Hydrogen Embrittlement: 18% of Cases

Hydrogen embrittlement is the most insidious failure mode because it leaves almost no macroscopic evidence. The fracture surface is typically intergranular, with "chicken-wire" microvoid coalescence visible only under SEM at 500x or higher. The bolt looks fine until it snaps, often at 30-50% of its rated tensile strength and often during the first thermal cycle after installation.

**Three hydrogen entry paths accounted for all 61 cases in our dataset:**

1. **Electroplating without post-bake (42 cases).** Cadmium and zinc electroplating generate atomic hydrogen at the cathode. Without a hydrogen embrittlement relief bake (190-220°C for 4-24 hours within one hour of plating, per ASTM B850), hydrogen diffuses to grain boundaries and stress concentrations. For fasteners above 36 HRC, this bake is mandatory, not advisory.

2. **Cathodic protection over-protection (11 cases).** Offshore platforms and buried pipelines often operate CP systems at potentials more negative than -1.05 V vs. Ag/AgCl. At these potentials, hydrogen evolution occurs directly on the bolt surface. For Monel K-500 and high-strength nickel alloys, limit CP potential to -0.90 V maximum.

3. **Sour service without NACE MR0175 qualification (8 cases).** H2S promotes hydrogen uptake by poisoning the hydrogen recombination reaction. NACE MR0175/ISO 15156 specifies hardness limits (typically 35 HRC maximum for most nickel alloys in sour service) and requires SSC (sulphide stress cracking) testing per NACE TM0177 Method A.

**Prevention protocol:**
- Electroplated fasteners above 32 HRC: mandatory post-plating hydrogen bake per ASTM B850, with batch traceability
- Cathodically protected joints: verify potential at the fastener, not the structure
- Sour gas environments: NACE MR0175-qualified grades only, with hardness verification on every heat
- Substitute mechanical plating (no hydrogen generation) or PTFE coating where possible

## 4. Galling During Assembly: 12% of Cases

Nickel alloys have a galling problem. Unlike carbon steel, whose surface oxides fracture and act as a solid lubricant, the chromium oxide layer on austenitic nickel alloys is thin, tenacious, and easily stripped under sliding contact. Once the oxide is breached, bare nickel-to-nickel contact produces instantaneous cold welding at the thread flank.

Of the 41 galling-related returns, 37 involved Inconel 718 nuts running onto Inconel 718 studs. The thread geometry was correct. The pitch diameter was within tolerance. The damage occurred because no anti-seize was applied, and the assembly speed exceeded the material's ability to dissipate frictional heat.

**Anti-seize protocol for nickel-alloy fasteners:**
- Use nickel-based anti-seize (e.g., Never-Seez NS-165 or equivalent) rated to the service temperature. Copper-based anti-seize is acceptable below 980°C but avoid in applications with sulphur-bearing atmospheres.
- Reduce assembly speed. For bolts above M20, limit to 10-15 RPM during the final 30% of the tightening cycle.
- Specify rolled threads, not cut threads. Rolled threads have 20-40% lower surface roughness (Ra 0.8-1.6 micron vs. 3.2-6.3 micron for cut threads) and a compressive residual stress layer that resists galling initiation.
- For critical joints, specify a dissimilar nut material. A Nimonic 80A bolt with an Inconel 625 nut is far less likely to gall than a like-on-like pair.

## 5. Thermal Fatigue: 7% of Cases

Thermal fatigue accounted for 24 cases, the smallest share, but disproportionately costly because the failures occurred in cyclic-service equipment where a single bolt fracture triggered cascading joint damage.

The mechanism is differential thermal expansion. A bolted flange joint is a system of three materials (bolt, flange, and gasket), each with a different coefficient of thermal expansion (CTE). During a heat-up from ambient to 700°C at 50°C/min, an Inconel 718 bolt (CTE ~13.0 x 10^-6/°C) in a 304H stainless steel flange (CTE ~18.5 x 10^-6/°C) experiences approximately 2.2 MPa of additional tensile stress per cycle, purely from CTE mismatch.

Over hundreds of cycles, this cyclic stress, superimposed on the static preload, initiates fatigue cracks at the first engaged thread root, the point of maximum stress concentration (Kt typically 3.5-5.0 for standard UN thread forms).

**Prevention:**
- Match bolt and flange CTE to within 2 x 10^-6/°C where feasible. For 304H flanges, consider Incoloy 800H bolts (CTE ~15.5 x 10^-6/°C) rather than Inconel 718.
- Reduce preload to 40-45% of yield for cyclic service, the lower mean stress extends fatigue life exponentially.
- Specify rolled threads with radiused root (UNJ or MJ profile) to reduce stress concentration factor.
- For applications exceeding 500 thermal cycles, require a fatigue analysis per ASME Section VIII Division 2 Part 5 (Design by Analysis).

## Prevention Checklist: 8 Actionable Items

1. **Specify the alloy by UNS number.** Never order "high-temperature bolts" generically.
2. **Require elevated-temperature tensile data** at the service temperature, not room temperature.
3. **Use Belleville washers** on every joint operating above 500°C, and calculate required travel per ASME PCC-1 Appendix O.
4. **Demand hydrogen embrittlement relief bake certification** for all plated fasteners above 32 HRC.
5. **Select anti-seize rated to service temperature.** Apply it to every nickel-alloy threaded assembly.
6. **Verify CTE compatibility** between bolt, flange, and gasket, because mismatch drives thermal fatigue.
7. **Limit CP potential to -0.90 V vs. Ag/AgCl** at the fastener surface for high-strength nickel alloys.
8. **Audit the MTC for heat-specific data**, covering chemistry, mechanicals, hardness, and NDT results per EN 10204 Type 3.1.