Mineral scale does more than insulate a heat exchanger. In a study published in Discover Fluid Mechanics, L. Rohwer, L. Augustin, J.-H. Evers, W. Augustin and S. Scholl describe how calcium sulfate fouling grown directly inside an experimental flow channel distorts velocity profiles and injects turbulence into the near-wall region, using a laser-based imaging technique fine enough to resolve individual crystal clusters.
The work extends an earlier experimental approach in which fouling layers were grown outside the channel, in a batch reactor, and then inserted as sample plates. This time the researchers generated the deposits in situ, inside the rectangular test channel itself, so the crystals grew under the same flow and temperature conditions experienced during measurement. To separate roughness effects from unwanted deposits elsewhere in the channel, they also printed artificial surfaces with additively manufactured structures that mimic the crystal topography, allowing controlled comparisons without interference from deposition outside the areas of interest.
The technique at the heart of the study is stereoscopic micro particle image velocimetry, or Stereo µPIV. The method extends conventional particle image velocimetry into three dimensions on a microscopic scale: laser-illuminated tracer particles are photographed at two closely spaced instants, and their displacements yield local flow velocities. Two cameras arranged stereoscopically capture all three velocity components simultaneously within a two-dimensional measurement plane, a configuration known as 2D3C. In this setup, a double-pulsed Nd:YAG laser, two high-resolution sCMOS cameras and fluorescent PMMA tracer particles roughly ten micrometers in diameter recorded flow fields at ten hertz across a field of view about fourteen millimeters in diameter. The tracers, with a Stokes number far below one, follow the fluid streamlines faithfully without disturbing the flow.
The test channel itself is a rectangular duct with a cross-section of 18 by 4 millimeters and a full length of 520 millimeters. An inlet zone 350 millimeters long ensures a fully developed flow profile before the measuring section. Bulk velocities between 1 and 4 meters per second correspond to Reynolds numbers between roughly 10,000 and 40,000 in the clean channel. For the fouling experiments, the team prepared 100 liters of solution containing 0.027 moles per liter of calcium sulfate at 40 degrees Celsius, circulated it at 1 meter per second, and held the heating circuit beneath the channel at 70 degrees Celsius while cooling water sat at 40 degrees. Only the run time varied, from 6 to 24 hours, so that deposit extent could be tied directly to duration.
Crystallization fouling of this kind follows a well-documented sequence. First, individual nuclei form on the hot surface. Growth and new nucleation then produce full surface coverage, after which crystals cross-link into clusters and finally build a compact layer. Heat-transfer measurements divide the process into phases: an initiation phase with no measurable change, a roughness-controlled phase in which the crystal roughness enhances turbulent convective heat transfer so strongly that the apparent thermal fouling resistance briefly turns negative, and a growth phase in which the insulating effect of the accumulating layer dominates. The in-situ experiments reproduced this pattern, with the roughness-controlled phase lasting around 12 hours before a significant layer growth phase and an asymptotic fouling behavior after 36 hours. Notably, fouling inside the channel affected heat transfer at a higher rate than the earlier ex-situ approach, shortening the required experimental run times.
The deposits were far from uniform. After 48 hours of growth, surface coverage and layer density varied strongly along the flow path, with the most developed crystal layer near the outlet of the product stream, where the heating medium enters in the counter-current arrangement. Deposits formed preferentially near the channel walls and grew inward toward the channel center, a pattern the researchers attribute to the lower local flow velocities in the corners of the rectangular cross-section. Microscopic characterization using a 4K digital microscope quantified surface parameters including the arithmetic mean height and maximum height of each fouled region, confirming that roughness increased both with growth time and in the downstream direction.
The velocity measurements reveal the consequences in detail. In the clean channel at 1 meter per second bulk velocity, the local mean velocity peaked at about 1.18 meters per second at a channel height of 2.75 millimeters, with the profile shifted slightly toward the smoother glass lid because the steel bottom plate is rougher. With fouling present, the picture changed dramatically. No tracer particles could be detected in the plane closest to the fouled bottom, indicating that crystals had grown into that region and blocked the flow there. The maximum detected velocity, at 3 millimeters height, rose to about 1.46 meters per second, a 24 percent increase over the clean channel, while the entire profile became more convex. Near the channel bottom, turbulent kinetic energy, a measure of velocity fluctuations calculated from the variances of the three velocity components, increased by a factor of three compared with the clean channel, with the effect fading toward the lid where no deposits form.
Correlating these flow measurements with the roughness data showed a consistent physical picture. At a channel height of 3 millimeters, mean flow velocity increased with the roughness of the fouled bottom, consistent with a constriction effect: deposits narrow the free cross-section, and the constant volumetric flow must therefore accelerate. At 1 millimeter height, close to the wall, mean velocity instead decreased as roughness grew, because roughness-induced turbulence and boundary layers forming around the crystal structures slow the near-wall flow. The researchers estimate that the clean channel sits in the hydraulically smooth regime, with the product of Reynolds number and relative roughness near 30, while fouling pushes the flow toward at least the transition zone, with estimated values of roughly 100 to 300. Increased roughness also raised turbulence across most of the channel height, suggesting the hydraulically rough regime may be reached.
Because deposits outside the measurement windows confounded the local analysis, the team turned to additively manufactured comparison surfaces. Printed by fused deposition modelling from polylactide, these samples replicated the height profiles of real fouling layers measured with the digital microscope, at mesh resolutions of up to 200 triangular facets per square millimeter. On average, the roughness of the printed surfaces fell within 20 percent of the natural deposits, and the turbulence enhancement was reproduced within 20 percent at 1 millimeter above the channel bottom. Because the artificial layers have smooth side walls, they isolate the effect of the bottom roughness from the confounding influence of side-wall deposits. The comparison confirmed that the printed structures capture the characteristic S-shaped velocity profile and near-wall turbulence induction of real fouling, though single crystal needles protruding from the natural layer appear to be underrepresented, leaving measured turbulence somewhat lower than in the in-situ case.
The authors caution that the current setup cannot yet fully disentangle all local effects, and that local heat transfer must be investigated with suitable methods, using dimensionless indicators such as the Nusselt number, to extend the fluid dynamic characterization into a thermal one. The fourth phase of crystal growth, in which the sublayer thickens without further roughening, could not be captured because of insufficient data at extended fouling times. Still, the study demonstrates that stereoscopic micro particle image velocimetry can resolve the microscale interplay between crystal deposits, near-wall turbulence and overall thermal performance, and the planned extension of the artificial-layer experiments, with finer measurement spacing and varying sublayer heights, aims to decouple cross-sectional constriction from turbulence enhancement and ultimately support site-specific predictions of heat transfer and pressure drop in fouled equipment.
Subject of Research: How calcium sulfate crystal fouling layers alter near-wall turbulence and velocity profiles in turbulent channel flow, measured with stereoscopic micro particle image velocimetry.
Article Title: Effects of crystal fouling on local fluid dynamics
Article References: Rohwer, L., Augustin, L., Evers, J. H., Augustin, W., & Scholl, S. (2026). Effects of crystal fouling on local fluid dynamics. Discover Fluid Mechanics, 2(1), Article 2. https://doi.org/10.1007/s44369-026-00006-6
Image Credits: AI Generated
DOI: 10.1007/s44369-026-00006-6
Keywords: crystallization fouling, calcium sulfate, Stereo µPIV, turbulence, heat exchangers, thermal fouling resistance, surface roughness, additive manufacturing, channel flow, turbulent kinetic energy, Effects, crystal
News Source: Denise Maddox. (October 29, 2026). Crystal Fouling Reshapes Turbulence in Channels, Micro-PIV Reveals. Scienmag.
