Skip to the content

Where the heat goes, and what the air does about it

DraftSolver models air-cooled condensers, dry coolers, air-cooled heat exchangers and natural-draft towers from one setup and one set of correlations: a rating, a sweep, a design, an annual run, a transient, the ASME PTC 30.1 curves and acceptance test, and CFD of the same plant, each with a datasheet.

Request early accessBook a demo

DraftSolver opens in 2027. Early access is granted by application.

Steam headerFinned-tube bundlesFanPlatformSteelwork010 mExhaust steamHeated airAir in
Figure 1. Air-cooled condenser, A-frame forced draft. Condenses turbine exhaust steam in A-frame bundles, with fans below pushing the air through them. Drawn at the proportions of the app's default case. The air is the app's own CFD of that case, in still air on its coarsest mesh: paths traced through the solved field from the surrounding air, through the fans and the bundles and into the plume, blue as the air arrives and orange as it leaves the bundles warmed.In elevation: the air in blue going in and dashed red coming out heated, the hot fluid in amber. More about this system

What sets it apart

One workflow from the 1-D model to CFD. A plant is set up once, and the same case is rated, swept, designed, run through a weather year, simulated in time and tested to ASME PTC 30.1, then ported to CFD of the same plant and read back beside the rating, or solved together with it in a co-simulation.

  • Crosswind in the 1-D modelCrosswind and plume recirculation are in the one-dimensional model, not left to CFD.
  • Every number shows its workingA result names the equations that produced it, each opening its Handbook page with its source, its fitted range and its stated accuracy.
  • A stated uncertaintyEvery rating carries its uncertainty band, broken down by the closure it comes from.
  • A public validation recordIt lists 47 cases, each naming what it was compared with: a published calculation, a measurement or another model.
  • Independent of the vendorAdiabat builds no cooling plant, so an owner can judge a vendor's guarantee on a model that is not the vendor's.

One setup, seven analyses

A case is set up once: the plant, its finned tube, the ambient and the models. Every analysis then runs on that setup. A rating, a sweep, a design and a year of hourly ratings of one plant therefore report the same quantities, and each run comes with a datasheet that lists the equations which produced it.

The system and the analysis. A case is saved as one file and opened again with its setup and its run settings.

DraftSolver's workspace with the analysis menu open, listing Rating, Design, Parametric sweep, Performance curves, Acceptance test and Annual run, each with a one-line description.
Figure 2. The case. The analysis is chosen beside the system, and every one of the seven runs on the same setup. From DraftSolver 0.1.0.

The plant node by node: the tower shell or the cells and fans, the finned tube, the bundles, the fluid, the ambient and the models. Each node shows whether its fields are valid before anything runs.

The Cells and fans node of an A-frame air-cooled condenser's setup: the fan and the counts of fans, rows and units on the left; a 3-D drawing of the A-frame cell and the fan's static pressure and shaft power against volume flow rate on the right, with the setup nodes listed in the left rail.
Figure 2. The setup of the A-frame condenser of Kröger's Example 8.1.4, one of the worked examples the app ships: one node of inputs, each validated as it is typed, beside the plant drawn from them and the fan's own curves. From DraftSolver 0.1.0.

The analysis's own settings: a sweep's axes, a design target, a test's data, a year's weather, a transient's scenario.

The Run step of a parametric sweep, with an axis of ambient temperature from 0 to 40 °C in steps of 5 °C, the option to give every point its uncertainty band, and the plant beside it.
Figure 2. A run's own settings: a parametric sweep of the ambient temperature from 0 to 40 °C in nine points, each with its uncertainty band. From DraftSolver 0.1.0.

The plant's figures, the findings graded by how much they matter, the equations that produced the result, and the datasheet.

The results of a rating of an A-frame air-cooled condenser, Kröger's Example 8.1.4: heat transfer rate 19.63 MW, steam condensed 8.316 kg/s, back pressure 19.93 kPa, air mass flow 604.5 kg/s, followed by the findings on the result and a table of the example's published values beside the computed ones, within tolerance.
Figure 2. A rating's results: the plant's figures, then what to know about the result, graded by how much it matters, and on a worked example the published values beside the computed ones. From DraftSolver 0.1.0.
Table 1. The analyses each of the nine systems offers. A system's page lists what it predicts and what each prediction was compared with.
SystemRatingDesignParametric sweepPerformance curvesAcceptance testAnnual runTransient simulation
Air-cooled condensers
Air-cooled condenser, A-frame forced draft●●●●●●●
Air-cooled condenser, A-frame induced draft●●●●●●●
Dry coolers
Dry cooler, plate-fin coil●●●––●●
Dry cooler, plate-fin coil, induced draft●●●––●●
Air-cooled heat exchangers
Air-cooled heat exchanger, forced draft●●●––●●
Air-cooled heat exchanger, induced draft●●●––●●
Natural-draft towers
Direct natural draft, vertical bundles●●●––●●
Indirect dry cooling, natural draft, horizontal array●●●––●●
Indirect dry cooling, natural draft, vertical bundles●●●––●●

Nine systems, from the condenser street to the tower

Each system is solved from its own draft equation, its loss coefficients and the heat transfer on the air side and the fluid side. Crosswind and recirculation enter the one-dimensional model wherever the literature states a method for them.

Air-cooled condensers

Dry coolers

Air-cooled heat exchangers

Natural-draft towers

From the rating to CFD of the same plant

Any rating is ported to OpenFOAM as it stands: the plant's geometry, each bundle as porous zones on the rating's own tube law, each fan on its own curve, and the steam at its own temperature. The case is meshed and solved on Adiabat's compute service, with nothing to install, and the solution is read back beside the rating it came from, per fan, tube row and bundle.

DraftSolver's CFD page for a natural-draft tower with vertical bundles round its base: on the left the CFD's air mass flow rate and heat transfer rate beside the rating's, with the difference in per cent, and the air through each bundle; on the right the air velocity on a vertical plane and the streamlines drawn on the plant in 3-D.
Figure 3. The CFD page after a run: the solution read back beside the rating it came from, with its convergence stated, and drawn on the plant. The run is the app's own CFD of the direct natural-draft tower's default case in still air, at the medium resolution. From DraftSolver 0.1.0.
Run in CFD
The rated plant ported to OpenFOAM as it is; the user chooses the mesh resolution, the turbulence model and the discretisation order, sees the mesh and its statistics before the solve, and reads the solution back beside the rating.
Plans: Expert, Enterprise, every academic licence, the Trial, Pro with its add-on.
CFD lab
The whole case open in an outline from setup to mesh, solution and results: windbreak walls, bodies from STL, the wind's direction, the domain, refinement and layers, the turbulence, schemes, solvers and relaxation, monitors and probes.
Plans: Expert and Enterprise, with every academic licence and the Trial.
Co-simulation
In the CFD lab, on every system: the CFD coupled to the rating's own model, in two forms. The steady co-simulation solves the CFD in blocks and, between them, rates every bundle's steam or fluid side again at the air the CFD gave it, splitting the flow from the header among the bundles until the split settles; it runs in a crosswind as it does in still air. The transient co-simulation is a fully transient simulation of the plant: the CFD runs time-accurately, window by window, beside the plant's own transient, as the steam or fluid conditions change.
Plans: Enterprise, every academic licence and the Trial; an add-on to Expert.
A 2-D map of air velocity on the vertical plane through a natural-draft tower, height to 260 m against distance across, with contours at 5 and 10 m/s, the faster air inside the shell and in the plume above it.
Figure 4. The air velocity on the vertical plane through the tower's axis, resampled from the solution, with its contours: the inflow to the bundles at the base, the air rising through the shell and the plume above. From DraftSolver 0.1.0.

One credit is one core-hour of CFD compute. Ratings, sweeps, designs, annual runs and transients use no credits.

A CFD solution and the rating are two models of one plant, and the comparison validates neither.

CFD in DraftSolver

Every claim names what it was compared with

The validation catalogue holds 47 cases: 13 textbook worked examples, 2 measurements on operating plants, 13 measurements on controlled rigs and 19 comparisons with CFD, a co-simulation or another model. Kröger's worked examples are published calculations, not measurements. Each case states what was compared, against what, the agreement found and what that agreement shows.

Table 2. The headline claim of each system's steady rating, as the app's capability statement prints it.
SystemCompared withAgreement

Kröger publishes no worked example for an induced-draft A-frame, so this system has no textbook verification case of its own; its parts are verified as Examples 8.1.2 and 8.1.4.

Kröger's worked Example 8.1.4Air mass flow rate, heat transfer rate and fan power within 0.011%; every loss coefficient to five significant figures
Wang, Chi and Chang's (2000) coil database; the draft side, fan and platform on Kröger's Examples 8.1.1 and 8.1.288.6% of the coil database within 15%; no measured dry cooler is reproduced
Kröger's worked Examples 8.1.1 and 8.1.2, and 9.3.1 and 9.3.2 under windAir mass flow rate within 0.10%, heat transfer rate and fan power within 0.04%, temperatures within 0.04 K; under wind within 1.5% and 0.5%
The published design point of the reference towerHeat transfer rate within 0.5%; air mass flow rate 3.5% above the published value
Kröger's worked Example 7.3.1Heat transfer rate within 0.024%, air mass flow rate within 0.010%, both outlet temperatures within 0.005 K

Built, but not yet verified against a worked example.

An independent implementation of the same specification (another model, not a measurement)Air mass flow rate within 0.7%, heat transfer rate within 0.2%

The validation record, case by case

The equations behind a result

Each result lists the equations that produced it. The Handbook in the app gives every equation's source, its coefficients, its fitted range and its stated accuracy, and a result outside a fitted range is reported as an extrapolation.

An uncertainty band on every rating

The band re-solves the plant with each correlation at the 95% bound of its stated accuracy and reports the interval each one contributes. It is included in every plan; Expert, Enterprise and Academic add a Monte Carlo over all of them at once.

Worked examples that check the build

The app ships complete cases with their expected results. When a loaded example is run, the results page reports the agreement of the build with the recorded and published values.

Published methods, cited

DraftSolver is an independent implementation of published methods: Kröger's one-dimensional methods for air-cooled heat exchangers and cooling towers, and the published literature on natural-draft dry cooling under crosswind and through start-up, including its author's two papers below. Both papers are open access, and each validation case cites its source.

  1. W. Strydom, J. Pretorius and R. Laubscher

    Co-simulation of transient start-up performance in natural draft direct dry cooling system under varying wind conditions

    Applied Thermal Engineering 300 (2026) 131471. doi:10.1016/j.applthermaleng.2026.131471

  2. W. Strydom, J. Pretorius and J. Hoffmann

    Natural draft direct dry cooling system performance at various application scales under windless and windy conditions

    Applied Thermal Engineering 248 (2024) 123181. doi:10.1016/j.applthermaleng.2024.123181

DraftSolver opens in 2027

Access is granted by application to owner's engineers and test agencies, consultancies, OEM and EPC bid engineers, data-centre engineers and universities. Each application is read by the engineer who builds the product.

Request early accessBook a demo