⚡ electrical engineering

A Corrosion-to-CFD Framework for Quantifying Surface-Roughness Effects on a 316L Stainless-Steel Centrifugal Pump in Saline Service

This paper proposes a reproducible methodology linking temporal corrosion-induced surface roughness growth in saline-service 316L stainless-steel centrifugal pumps to CFD simulations, enabling the quantification of roughness-specific head loss to distinguish it from other degradation mechanisms.

Original authors: Nsini Ignatius Udo

Published 2026-10-29

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Nsini Ignatius Udo

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Pumps are the silent workhorses of modern industry, moving vast quantities of water through power plants, desalination facilities, and offshore oil rigs. When these machines handle saltwater, they face a relentless enemy: corrosion. Saltwater is aggressive; it eats away at the smooth metal surfaces inside the pump, creating tiny pits and rough patches over time. For engineers, this roughening is more than just a cosmetic issue. Just as a rough road slows down a car by creating friction, a roughened pump surface disrupts the smooth flow of water, causing the machine to work harder and deliver less power. The challenge has always been knowing exactly how much of a pump's failure is due to this surface roughness versus other mechanical problems, such as worn-out seals or a bent spinning wheel. Without this distinction, maintenance teams cannot know whether to simply smooth the surfaces or replace entire components.

A researcher named Nsini Ignatius Udo has proposed a new way to solve this puzzle, not by building a new pump, but by creating a precise digital map of how corrosion affects performance. The study focuses on a specific type of stainless steel pump that was tested in a laboratory setting, where it was exposed to synthetic seawater for 200 hours. During this time, the metal surface did not stay smooth; it became increasingly rough, with the average height of its tiny bumps growing from 0.8 micrometers to 3.4 micrometers. As the surface got rougher, the pump's performance declined noticeably. The amount of water it could push dropped, the pressure it could generate fell, and the energy required to run it became less efficient. The researchers also tracked the vibration of the machine, which increased as the corrosion worsened, serving as support evidence of the progressive surface deterioration.

The core of this new work is a framework designed to translate these physical measurements into a computer simulation. In the past, engineers might have tried to guess how a rough surface would behave in a computer model, often assuming a direct, simple link between the roughness they could measure with a tool and the roughness the computer needed to calculate fluid flow. Udo argues that this assumption is too risky. The roughness a tool measures is a statistical average of a surface, but the roughness that matters to the flowing water depends on the specific shape and arrangement of those bumps. To address this, the proposed method introduces a flexible factor that acknowledges the uncertainty in converting a physical measurement into a computer input. Instead of forcing a single, rigid number, the framework allows for a range of possibilities, ensuring that the computer model reflects the real-world ambiguity of corrosion.

The study breaks the 200-hour experiment into five distinct stages, from a brand-new condition to a heavily corroded state. For each stage, the researchers outline how a future computer simulation should be set up to test the effects of that specific level of roughness. They plan to run these simulations across different operating speeds and flow rates, not just at the pump's standard setting, to see how the roughness impacts the machine under various conditions. A critical part of this process is recognizing a major potential error in how performance is measured. The experiment recorded "electrical-input efficiency," which measures how well the pump converts electricity into work, including losses from the motor and mechanical friction. However, the computer simulation calculates "hydraulic efficiency," which measures only the fluid's movement. Because these two numbers are not the same, the framework insists on separating them. Engineers must either account for the motor and mechanical losses separately or convert the experimental data to a shaft-power basis before comparing it to the computer model. This ensures that the comparison is fair and that the results aren't skewed by mixing different types of efficiency measurements.

The goal is to isolate the exact amount of performance loss caused solely by the rough surface. By comparing the computer's prediction of roughness-induced loss against the actual total loss measured in the lab, engineers can determine what fraction of the problem is due to the surface texture and what fraction is due to other hidden issues, like gaps in the seals or mechanical imbalances.

This approach does not claim to have solved the problem with a final answer, as the actual computer simulations have not yet been run. The paper explicitly states that no solver-dependent CFD results are included because the original pump CAD geometry and CFD data were not available. Instead, it provides the rigorous blueprint for how to run them correctly. It establishes a clear path from the physical evidence of corrosion to the digital prediction of performance, ensuring that no assumptions are made about how roughness translates into fluid friction. By separating the known measurements from the uncertain computer inputs, the framework offers a way to build a "digital twin" of a pump that can accurately predict its future health. This method ensures that when a pump begins to fail, engineers will know exactly how much of that failure is caused by the roughness of the metal and how much is caused by something else, allowing for smarter, more targeted repairs.

Technical Summary: A Corrosion-to-CFD Framework for Quantifying Surface-Roughness Effects on a 316L Stainless-Steel Centrifugal Pump in Saline Service

Problem Statement
Centrifugal pumps operating in saline environments, such as seawater cooling or ballast systems, suffer from performance degradation due to corrosion. While it is established that surface roughening alters wall shear, turbulence, and hydraulic losses, a critical gap exists in tracing the temporal evolution of corrosion states directly to Computational Fluid Dynamics (CFD) models. Previous studies have linked roughness to performance loss but have failed to decouple the specific hydraulic losses caused by surface roughness from other degradation mechanisms, such as clearance growth, leakage, pitting geometry changes, or mechanical imbalance. Furthermore, the conversion of measured arithmetic surface roughness ( $R_a$ ) to the equivalent sand-grain roughness ( $k_s$ ) required for CFD wall treatments is often treated as a universal constant, ignoring pump-specific uncertainties and morphological factors.

Methodology
The paper proposes a methodological framework to bridge experimental corrosion data with CFD simulations without generating new solver-dependent results. The approach is anchored on a published experimental study of a 316L stainless-steel pump exposed to 30-ppt synthetic seawater for 200 hours.

Exposure-Indexed States: The experimental trajectory is decomposed into five discrete states ( $S_0$ to $S_4$ ) corresponding to 0, 50, 100, 150, and 200 hours of exposure. These states track the evolution of flow rate, head, electrical-input efficiency, vibration, and arithmetic roughness ( $R_a$ ).
Roughness Transfer Parameter ( $\alpha$ ): A core methodological innovation is the treatment of the relationship between measured $R_a$ and CFD input $k_s$ as uncertain and pump-specific, defined by $k_s = \alpha R_a$ . Instead of assuming $\alpha = 1$ , the framework proposes a sensitivity analysis where $\alpha$ is swept (e.g., 0.5 to 4.0) to account for the lack of a universal conversion factor.
CFD Protocol: The framework outlines a rigorous CFD protocol for future execution, including:
- Governing Equations: Incompressible RANS with SST $k-\omega$ turbulence models.
- Domain: 3D rotating/stationary interfaces with mesh and time-step independence verification.
- Regime Assessment: Calculation of the local roughness Reynolds number ( $k_s^+$ ) to determine if the flow is hydraulically smooth, transitional, or fully rough, rather than assuming a regime a priori.
- Multi-Point Analysis: Evaluation across a characteristic curve (0.60 to 1.40 times design flow) to capture off-design behavior.
Attribution Metric: The framework introduces a head-based attribution metric ( $F_{H,rough}$ ) to quantify the fraction of total measured head loss ( $\Delta H_{exp} = 1.30$ m) that can be attributed solely to roughness-induced hydraulic losses in the CFD model. The residual loss is assigned to other mechanisms (clearance, leakage, geometry distortion).

Key Contributions
The paper makes three primary methodological contributions:

Time-Indexed Boundary States: It converts a continuous experimental corrosion trajectory into five reproducible, exposure-indexed CFD boundary states ( $S_0-S_4$ ), providing a structured pathway for future simulations.
Explicit Uncertainty in Roughness Transfer: It separates surface metrology ( $R_a$ ) from hydraulic roughness ( $k_s$ ) by introducing $\alpha$ as a tunable, pump-specific parameter rather than a fixed constant, acknowledging that $R_a$ and $k_s$ are not interchangeable.
Residual Decomposition: It establishes a framework for attributing performance degradation. By comparing the "roughness-only" CFD head decline against the total experimental head decline, the method isolates the roughness contribution and explicitly defines a "residual degradation" term for other physical mechanisms.

Results and Limitations
The paper explicitly states that no CFD results are reported. The study is formulated strictly as a methodology.

Experimental Data: The paper utilizes existing experimental data showing a drop in electrical-input efficiency from 70.6% to 58.0%, a head decline from 12.0 m to 10.7 m, and an increase in $R_a$ from 0.8 $\mu$ m to 3.4 $\mu$ m over 200 hours.
Modest Claims: The author emphasizes that the current work does not fabricate CFD data or reconstruct the specific pump CAD geometry. Consequently, no solver-dependent values (head, torque, efficiency) are generated. The paper serves as a "work-in-progress" blueprint, defining the necessary inputs, verification steps (mesh/time-step independence), and validation metrics (MAE, RMSE) required for a subsequent study to execute the simulations.

Significance and Scope
The significance of this work lies in its structured approach to traceability and attribution. It addresses the research gap of linking specific corrosion states to hydraulic performance while rigorously accounting for the uncertainty in translating physical roughness measurements into CFD inputs.

The framework is intentionally limited to a single 316L pump in a controlled saline environment and does not claim to provide a universal degradation law for all pumps or materials. Its primary value is establishing a reproducible pathway for future researchers to quantify how much of a pump's performance decline is due to surface roughness versus other mechanical or geometric degradation, thereby enabling the development of more accurate, corrosion-aware digital twins. The paper concludes that while roughness-sensitive CFD is a mature field, the novelty here is the integration of corrosion time-histories, explicit uncertainty modeling in roughness transfer, and a residual-based attribution framework.

Technical Summary: A Corrosion-to-CFD Framework for Quantifying Surface-Roughness Effects on a 316L Stainless-Steel Centrifugal Pump in Saline Service

More like this