CFD + HTRI + Physical Test: Our Three-Layer Validation Process
Most heat exchanger manufacturers run one validation pass: thermal sizing software produces a result, and the design goes to manufacturing. For commodity geometries in well-characterized service conditions, that single pass is usually sufficient. But for non-standard geometries — novel fin patterns, elliptical tubes, custom flow conditioners — the correlations embedded in standard software were not derived from those geometries. Predictions made outside a correlation’s derivation range are extrapolations, and extrapolations fail in the field in ways that are expensive to diagnose.
Dongrun runs three layers of validation on every custom design before signing off on a purchase order. Here is what each layer does and why all three are necessary.
Layer 1: In-house thermal design software
Dongrun’s primary sizing tool is proprietary software developed and refined over decades of project work and laboratory testing. The distinguishing factor is calibration: every correlation embedded in the software has been checked against data generated on Dongrun’s own heat exchanger testing platform. The platform characterizes air-air, air-water, air-oil, and air-steam heat transfer across a range of geometries and operating conditions, with full instrumentation for inlet and outlet temperatures, flow rates, and pressures.
When the software predicts a heat transfer coefficient or pressure drop for a given geometry, that prediction has a traceable relationship to measured data from similar geometries run on the same platform under controlled conditions. This is the foundation layer — it produces the candidate design and gives a baseline confidence interval on performance.
Layer 2: HTRI cross-check
Dongrun is a member of HTRI (Heat Transfer Research, Inc.), the industry consortium that maintains the most comprehensive database of experimentally validated heat transfer correlations in the world. HTRI’s software is used as a sanity boundary on the Layer 1 result. If in-house software and HTRI agree within a few percent, the design proceeds. If they diverge by more than the expected tolerance, the discrepancy is investigated before anything is built.
HTRI is not a replacement for Layer 1 — it has the same limitation that all correlation-based tools share: results are reliable only within the geometry range from which the correlations were derived. HTRI does, however, provide an independent check on the most common failure mode of in-house tools: systematic bias from a calibration error or an outdated coefficient. Two independent tools agreeing is meaningfully stronger evidence than one tool run twice.
Layer 3: Physical prototype test
For designs that involve novel geometry — Dongrun’s patented elliptical tubes, for example, sit outside the derivation range of most published correlations — or for projects where field failure would be costly, a physical prototype is built and tested before the production unit is manufactured. The test protocol measures air-side and tube-side resistance, confirms natural frequencies through vibration analysis (critical for fan-driven equipment that must avoid resonance with the drive system), characterizes fan performance and acoustic output, and in some cases applies strain gauges to confirm that thermal stress predictions match measured deflections.
A prototype test can surface problems that no amount of calculation will catch: manufacturing tolerance stack-up that shifts the pressure-drop curve, fin damage from assembly handling, a resonant frequency that sits closer to the excitation range than the modal model predicted. These are found on the test rig, not on the client’s site.
Why three and not two
Each layer catches a different class of error. Layer 1 catches geometry-specific effects that generic correlations miss. Layer 2 catches systematic bias in the in-house tool. Layer 3 catches manufacturing and assembly effects that no software sees. Removing any layer narrows the coverage and increases the probability that a problem passes through undetected.
The case for three layers is clearest for novel geometries. Dongrun’s elliptical tube designs deliver measurably better heat transfer per unit pressure drop than round tubes in certain service conditions — but that advantage was established through test data generated at Layer 3, not through calculation alone. The calculation pointed to the expected improvement; the test confirmed it was real and repeatable.
Talk to a thermal engineer
We can show test outputs from past projects — on-site or under NDA.