20.2 Mechanical Integrity, Reliability, and Failure Mechanisms
Key Takeaways
- PSM mechanical integrity applies to **pressure vessels and storage tanks, piping, relief and vent systems, emergency shutdown systems, controls including alarms and interlocks, and pumps**, and requires inspection and testing following recognized and generally accepted good engineering practice (RAGAGEP).
- Remaining life is \((t_{actual} - t_{required}) / \text{corrosion rate}\); under API 510 and API 570 the internal or thickness inspection interval is the **lesser of one-half the remaining life or 10 years**.
- Corrosion rate must be evaluated on both a **long-term** and a **short-term** basis, and the **greater** of the two governs, so that a recent acceleration is not masked by years of benign history.
- For a constant failure rate, reliability is \(R(t) = e^{-\lambda t}\) with \(MTBF = 1/\lambda\), and steady-state **availability** is \(MTBF/(MTBF + MTTR)\); redundancy multiplies unreliability, so \(n\) parallel units give \(R = 1 - (1-R_i)^n\).
- **Corrosion under insulation** is most aggressive where the metal sits between roughly \(-12^\circ\text{C}\) and \(175^\circ\text{C}\), is invisible from outside, and is a leading cause of unexpected piping failures in insulated service.
20.2 Mechanical Integrity, Reliability, and Failure Mechanisms
The NCEES specification lists Process equipment and reliability (e.g., testing, maintenance, mechanical integrity, failure mechanisms) under Operation and Maintenance. Section 16.2 covered the electrochemistry of corrosion. This section covers the program that finds damage before it causes a loss of containment, and the arithmetic that decides how often to look.
1. The Mechanical Integrity Element
Under (1910.119(j)), mechanical integrity applies to:
- Pressure vessels and storage tanks
- Piping systems, including components such as valves
- Relief and vent systems and devices
- Emergency shutdown systems
- Controls, including monitoring devices and sensors, alarms, and interlocks
- Pumps
The program must include written procedures, trained maintenance personnel, inspection and testing at frequencies consistent with manufacturer recommendations and good engineering practice, documentation of every inspection, correction of deficiencies before further use (or in a safe and timely manner with interim measures), and quality assurance that new equipment is fabricated and installed to specification.
RAGAGEP — recognized and generally accepted good engineering practice — is the benchmark. The standards a chemical engineer is expected to name:
| Standard | Scope |
|---|---|
| API 510 | Pressure vessel inspection, rating, repair, alteration |
| API 570 | In-service piping inspection |
| API 653 | Aboveground storage tank inspection and reconstruction |
| API 576 | Inspection of pressure-relieving devices |
| API 580 / 581 | Risk-based inspection methodology and quantitative basis |
| ASME BPVC Sec. VIII | Pressure vessel design and construction (Section 16.1) |
2. Corrosion Rate, Remaining Life, and Inspection Interval
Thickness is monitored at fixed thickness measurement locations (TMLs) so that successive readings are comparable.
The greater of the two governs. A component that corroded at (0.05\text{ mm/yr}) for fifteen years and then at (0.60\text{ mm/yr}) since the last inspection has a long-term average that looks harmless; using it would set an interval that lets the component fail. The short-term rate is what detects a process upset, a change in feedstock, or a failed corrosion inhibitor.
where (t_{required}) is the minimum thickness for the design pressure and temperature (Section 16.1), excluding the corrosion allowance — the corrosion allowance is what is being consumed, not what must be retained.
The interval rule. Under API 510 for vessels and API 570 for piping, the internal or on-stream thickness inspection interval is the lesser of one-half the remaining life or 10 years (shorter for high-consequence piping classes). External visual inspection is typically on a 5-year cycle.
Worked example. A vessel course has (t_{nominal} = 0.500\text{ in}), (t_{required} = 0.300\text{ in}), current measured thickness (0.380\text{ in}), and a governing corrosion rate of (0.008\text{ in/yr}).
The half-life rule exists so that the component is examined at least once more while enough wall remains to act on the finding. Inspecting at the full remaining life would mean the next look happens at the moment of retirement thickness.
3. Damage Mechanisms
| Mechanism | Conditions | Characteristic signature |
|---|---|---|
| General corrosion | Uniform attack in an aggressive bulk fluid | Even wall loss; predictable, manageable by allowance |
| Pitting | Chlorides, stagnant zones, deposits | Deep local penetration with negligible weight loss; not caught by average thickness |
| Chloride SCC | Austenitic stainless, chlorides, tensile stress, (> \sim 60^\circ\text{C}) | Branched transgranular cracking; catastrophic and fast |
| Corrosion under insulation (CUI) | Insulated carbon steel, roughly (-12) to (175^\circ\text{C}), water ingress | Hidden beneath the jacket; a leading cause of unexpected piping failure |
| HTHA | Hydrogen at elevated temperature and partial pressure | Methane formation at grain boundaries; governed by API 941 Nelson curves |
| Sulfidation | Sulfur compounds above (\sim 260^\circ\text{C}) | Scaling of carbon and low-alloy steel in hot hydrocarbon service |
| Creep | (T > 0.4)-(0.5 T_m) | Time-dependent deformation; Larson-Miller (Section 16.1) |
| Brittle fracture | BCC steel below its DBTT | Sudden cleavage, often on a cold startup or an autorefrigeration event |
| Erosion-corrosion | High velocity, entrained solids, two-phase | Grooved, horseshoe-shaped loss at elbows and downstream of restrictions |
| MIC | Stagnant water, sulfate-reducing bacteria | Localized pitting under deposits, often in hydrotest water left in place |
Two of these deserve emphasis because they defeat ordinary inspection. CUI is invisible without removing the insulation, which is expensive, so it is inspected on a risk-prioritized basis targeting known water-ingress points: penetrations, supports, damaged jacketing, and lines that cycle through the damp temperature band. Pitting and SCC produce essentially no measurable general wall loss, so a thickness-only inspection program will report a healthy component right up to the through-wall leak; detecting them requires appropriate techniques such as visual examination, dye penetrant, or ultrasonic shear wave.
Risk-based inspection (API 580/581) allocates inspection effort by (\text{risk} = \text{probability of failure} \times \text{consequence of failure}), concentrating resources on the small fraction of components that carry most of the risk rather than inspecting everything on a uniform calendar.
4. Reliability Mathematics
During the flat middle of the bathtub curve the failure rate (\lambda) is approximately constant, and:
Failure rate lambda
^
|\ /
| \ infant / wear-out
| \ mortality /
| \___________________________________/
| useful life (constant lambda)
+----------------------------------------------> time
Availability accounts for how long repairs take, not only how often failures occur:
A pump with (MTBF = 8{,}000\text{ h}) and (MTTR = 40\text{ h}) has (A = 8{,}000/8{,}040 = 0.995). Cutting (MTTR) to (10\text{ h}) by stocking a spare rotating assembly raises (A) to (0.99875) — often cheaper per point of availability than extending (MTBF).
System architecture. For components in series (all must work), reliabilities multiply:
For components in parallel (any one suffices), unreliabilities multiply:
Three parallel pumps each at (R = 0.90), with any one able to carry the duty, give (R_{sys} = 1 - (0.10)^3 = 0.999). The same three in series — as in a train where every unit must run — give (R_{sys} = 0.90^3 = 0.729). This asymmetry is why a long unspared chain of equipment has poor availability even when every individual item is good, and why sparing is applied to the specific items with the lowest (R).
Maintenance strategy follows from consequence:
| Strategy | Basis | Use when |
|---|---|---|
| Run to failure | React after breakdown | Non-critical, cheap, redundant items |
| Preventive | Fixed time or run-hours | Known wear-out behavior |
| Predictive / condition-based | Vibration, thermography, oil analysis, thickness | Failure develops detectably over time |
| Proactive / RCM | Eliminate the root cause | Chronic repeat failures |
Time-based preventive maintenance on an item with a constant failure rate is wasted effort: a memoryless failure process is no less likely to fail just after an overhaul than just before one, and the intervention itself introduces infant-mortality risk. Preventive replacement only pays where there is a genuine wear-out mechanism.
A piping circuit has a required minimum thickness of 0.220 in. Readings at the governing TML are: 0.500 in at installation 20 years ago, 0.390 in at the last inspection 4 years ago, and 0.350 in today. What corrosion rate governs, and what is the maximum inspection interval under the API 570 half-life rule?
A critical service is supplied by three identical parallel pumps, any one of which can carry the full duty. Each pump has a reliability of 0.92 over the mission time. What is the system reliability, and how would it change if the same three pumps were arranged in series so that all three must operate?
An insulated carbon steel line operating at 95 degrees C has passed every ultrasonic thickness survey at its designated TMLs for twelve years, then fails through-wall beneath the insulation at a pipe support. What is the most likely explanation?