WindTunnel

I built an open-circuit desktop wind tunnel for model-car aerodynamic experiments and smoke-flow visualisation. V1 exposed weaknesses that should have been found before construction. It is still unfinished, and I am evaluating whether completing it is the right use of further work.

Original architecture

The original tunnel was approximately 1200 × 400 × 600 mm, with a clear test section of about 500 × 350 × 250 mm. Nine 120 mm Arctic P12 fans pushed air through an upstream plenum and flow-straightening region into the working section.

Approximately 1,300 paper straws formed a flow straightener intended to reduce lateral velocity components and improve flow uniformity. This was not a demonstration of laminar flow.

Construction combined MDF, PLA 3D-printed parts and acrylic supplied with support from Waveney Precision. Integrated lighting made the test section easier to observe.

What flow testing exposed

Smoke/fog visualisation and tuft testing showed that the airflow was less uniform and controlled than I had intended. These observations were useful, but they were not measurements of velocity, turbulence intensity or pressure loss.

The test hardware was also part of the problem. Supports could produce a substantial wake of their own, interfering with the flow I wanted to observe. I designed thin, dedicated fixtures to reduce that interference rather than treating the rig as invisible to the experiment.

Reassessing the flow path

Research after construction led me to reconsider the architecture, not just individual components. In the current V1 redesign, six fans replace the original nine and move downstream of the test section. They are being repositioned to pull air through the working section rather than push air into it.

The original upstream plenum concept remains, now open at the top, along with the existing flow-straightening approach. Moving the fan bank downstream is intended to keep its strong disturbance away from the immediate inlet to the working section; the benefit still needs testing.

The downstream expansion

After the test section, a diffuser expands outward to meet the wider six-fan bank. I designed the geometry with a diffuser half-angle below 7 degrees, intending to reduce the risk of separation through the expansion.

That is a design choice, not a validated result. I have not demonstrated that separation is eliminated or established the resulting flow quality.

Lesson: The largest weakness in V1 was a process failure. I committed to construction before doing enough aerodynamic research and low-cost validation. Several architectural problems only became obvious once the tunnel existed physically.

Is V1 worth completing?

The redesign has changed V1 substantially, but some limitations are fundamental. I may choose not to finish it. The useful question is whether more work on this prototype would teach me more than starting again later with clearer requirements.

The project has changed how I approach physical builds: investigate the system architecture, define what a useful test would show, and challenge assumptions before committing material and time.

What V2 needs to solve

V2 is a future project, not a build in progress. It will probably begin after I acquire a new 3D printer; there is no fixed date, final geometry or implied component purchase.

The likely direction is a smaller test section of roughly 100 × 100 mm, a more deliberate open-circuit layout, and a properly designed contraction—potentially using a fifth-order profile. Better fan selection and arrangement would support the main aim: controlled flow quality rather than the largest possible test section.

Smoke visualisation, load-cell-based force measurement and more quantitative testing are possible parts of that next iteration. They remain requirements to investigate, not capabilities of the current rig.

My experience around the Greenpower F24+ car and its aerodynamic development helped motivate the tunnel: I wanted a physical way to examine airflow rather than rely only on a plausible-looking shape.