Model-Based Damage Tracking: Engineering Turnaround Decisions

Turnarounds Are Decisions, Not Events

Turnarounds are not events to be managed. They are decisions — among the most consequential an industrial organization makes.

And the quality of that decision depends on one thing: whether you can see how your equipment is aging right now, not just at inspection time.

Most organizations rely on RBI reports, inspection snapshots, and calendar-based logic. But damage doesn't wait for inspections. In high-temperature systems, damage mechanisms like creep, fatigue, oxidation, and thermal cycling evolve continuously under stress and temperature. Equipment operating in the creep regime is degrading minute by minute — whether we observe it or not.

When we plan a turnaround based on averages or isolated data, we're not making a decision.

We're placing a bet.

Model-Based Damage Tracking Closes This Gap

Model-Based Damage Tracking (MBDT) closes this gap.

It is not a simulation. It is a formal method grounded in material science that reconstructs how damage accumulates over time using validated parametric models. It uses three inputs: operating history, material properties, and inspection results.

How Model-Based Damage Tracking Works

Operating History

The operating history comes from PI historian and DCS logs — temperature cycles, pressure profiles, run durations, upsets. For each cycle, the model calculates the time-at-temperature above critical thresholds.

Temperature has a dramatic exponential effect on damage accumulation due to activation energy (Q) — the energy barrier atoms must overcome to participate in thermally activated processes such as diffusion, dislocation climb, and oxidation. The rate of these processes follows an Arrhenius relationship:

For most engineering alloys, the activation energy for creep closely matches the activation energy for lattice self-diffusion, confirming that vacancy-controlled dislocation climb is the rate-limiting mechanism. Changes in measured activation energy signal mechanism transitions — for example, a shift from bulk diffusion control to grain boundary diffusion at lower temperatures. When such transitions occur, damage models calibrated for one regime may significantly mispredict behavior in another.

Material Properties and Parametric Models

Material properties come from NIMS, supplier certifications, and published literature. The method prioritizes sources in order of confidence: material supplier technical data, peer-reviewed research in metallurgical journals, industry-recognized databases like NIST or CINDAS, and relevant codes such as API 530.

MBDT employs multiple parametric models depending on the damage mechanism and material system. For high-temperature creep in low-alloy steels, the Larson-Miller parameter is one commonly used approach:

Alternative parametric methods include Sherby-Dorn, Manson-Haferd, and Orr-Sherby-Dorn — each valid for specific material systems and temperature ranges. For fatigue, Coffin-Manson and Paris law models are used. For oxidation, parabolic rate laws apply. For corrosion, electrochemical kinetics or empirical corrosion rates are employed depending on the environment.

The choice of model depends on the mechanism, material, and available data. The key is that each model is grounded in physical principles and validated against experimental data — not arbitrary empiricism.

Damage Index Calculation

For each critical component, MBDT calculates a Damage Index (D) — a unitless metric representing accumulated degradation relative to expected life. The calculation sheet documents all assumptions: source of inputs, temperature-dependent curves used, uncertainty ranges. D is calculated per mechanism — creep, fatigue, corrosion — and combined where applicable.

In high-temperature equipment, variations in refractory thickness or insulation condition directly affect local metal temperature. Even modest temperature increases — on the order of 30-50°C — can reduce rupture time by more than half due to the exponential nature of the Arrhenius relationship. These differences are invisible to traditional RBI but decisive in practice.

MBDT Doesn't Replace RBI — It Completes It

Let's be honest: no one likes being told their RBI program is "incomplete." And rightly so — when done well, RBI is a powerful framework. But too often, what gets called "RBI" is just a spreadsheet with generic failure modes and average process conditions. It assumes steady-state operation, ignores thermal cycling, and treats time-dependent damage as a footnote.

That's not RBI. That's theater.

MBDT doesn't replace good RBI. It completes it. Where RBI stops — at the edge of time-dependent mechanisms — MBDT continues. It brings physics into the room, so the model reflects not just what might fail, but how fast it's failing.

Once D Is Known, Everything Changes

Once D is known, everything changes.

You no longer ask: "What should we replace during the turnaround?"

You ask:

• •"When must we take the unit down?"
• •"Can we extend the cycle safely?"
• •"Which components define the critical path?"

The answer lies in the highest D value across the system.

This shifts planning from guesswork to condition-driven execution.

MBDT Accounts for Uncertainty

But MBDT does not ignore uncertainty. It accounts for it.

• •±15% on temperature if control system drift is possible
• •±10–20% on rupture strength due to material variation
• •±5–10% on thickness measurements

It performs sensitivity analysis to quantify impact. If the lower bound of D exceeds 0.75, action is required regardless of uncertainty. If the upper bound stays below 0.50, extension is safe. In between, engineering judgment applies — conservative by default.

Then, it links technical state to financial outcome.

Materials Aren't Perfectly Predictable — And That's Accounted For

Some might say: "But materials aren't that predictable." And they're right — if you treat steel like a single substance. But we don't. We know microstructure matters. Fine grains favor grain boundary diffusion. Precipitates pin dislocations. Brittle phases promote intergranular fracture under creep.

That's why the model doesn't assume idealized behavior. It incorporates these realities through conservative bounds and documented assumptions. When data is missing, it uses lower-bound properties. When inspection shows faster degradation than predicted, it revises the parameters — not because the model failed, but because now we know better.

It's not about being perfect. It's about being traceable.

MBDT Is a Living Process

MBDT is not a static report. It is a living process.

Every year, predicted damage accumulation is compared to actual field observations. When multiple inspections exist, actual damage rates are calculated and compared to projections. If actual exceeds predicted by more than 30%, model parameters are revised.

Inspection Methods

Inspection methods are standardized. Surface replication is performed after cleaning the component surface and applying plastic medium. The replica is examined under optical microscope at 50–1000× magnification. Observed cavity density and morphology are classified into levels: Level 2 (D ≈ 0.25), Level 3 (D ≈ 0.40). If two replications separated by Δt show progression from Level 2 to Level 3, the damage accumulation rate is established and remaining life estimated.

Post-Failure Analysis

For failed components, post-failure analysis validates or corrects the assumed mechanism. Root cause analysis asks: Was the failure mode predicted? Was D rising appropriately? Corrective actions follow: adjust material property data, revise operating envelope, change inspection frequency.

Industry experience is monitored. Peer company performance on similar equipment is tracked. Technical forums and user groups are followed. Improved models or parameters are adopted when published literature provides better basis.

This closes the loop.

The model learns. Confidence grows.

No One Trusts a Number Without Context

One thing we've learned: no one trusts a number without context. That's why every D value comes with its own biography — where the inputs came from, which assumptions were made, how uncertainty was handled. It's not just a result. It's an argument.

And like any good engineering argument, it's open to challenge.

MBDT Requires Structure

MBDT requires structure.

Documentation Requirements

Documentation is mandatory. Each equipment item has an Equipment Data Sheet: ID, description, location, design conditions, original fabrication date, service start date, applicable code.

• •Operating History Summary tabulates nominal and upset conditions
• •Material Property Documentation includes grade, heat treatment, source reference, temperature-dependent curves
• •Damage Calculation Summary shows detailed sheets per mechanism, total D, assumptions, input sources
• •Inspection History records date, method, results, inspector qualification, measurement data, trend analysis, limitations
• •Risk Assessment and Recommendation specifies predicted remaining life with confidence interval, recommended action with justification

Turnaround Scope Summary

MBDT operates as a continuous review cycle, not a pre-turnaround exercise. Damage indices are updated and reviewed every 3 to 6 months as part of ongoing asset management. This regular cadence ensures that equipment condition is continuously monitored and turnaround decisions are made with current data — not scrambled together weeks before shutdown.

Each review cycle produces a Turnaround Scope Summary Report with the following structure:

This regular review process eliminates the chaos of last-minute turnaround planning. By the time a turnaround date approaches, the scope is already known, justified, and integrated into long-term asset strategy. Training is maintained annually, and records confirm competency in MBDT responsibilities.

This Isn't Digital Hype — It's Engineering Discipline

It's true — some engineers still roll their eyes at "yet another digital initiative." But this isn't about dashboards or AI buzzwords. It's about something older and more reliable: a calculation sheet that tells you when to act, why, and what happens if you don't.

No drama. Just clarity.

The Only Sustainable Advantage Is Continuity

Because the only sustainable advantage in industrial operations is not speed — it is continuity.

• •Between data and decision
• •Between today's reading and tomorrow's failure
• •Between technical judgment and corporate outcome

When that continuity exists, you don't react.

You anticipate.