The problem
At urban speeds rolling resistance dominates a scooter's losses, but above about 40 km/h aerodynamic drag grows quickly and the rider contributes most of it. Students rarely quantify this. A clean, repeatable CFD study shows how much a windscreen, leg-shield shape or rider posture changes drag area, and what that means for cruise power and range.
Objectives
- Build a simplified, watertight scooter and rider assembly suitable for external CFD.
- Set up an external Flow Simulation study with a properly sized computational domain and force goals.
- Compute drag force, drag coefficient and CdA at two example speeds, with a mesh-convergence check.
- Compare at least three design variants (baseline, windscreen, tucked posture) and convert drag into cruise power.
What you need
Tools and toolboxes
- SOLIDWORKS Part and Assembly — Simplified scooter body, wheels and a mannequin-style rider built from lofts and revolves, combined into one watertight assembly.
- SOLIDWORKS Flow Simulation — External analysis with global force goals, equation goals for Cd and CdA, local mesh refinement and a mesh-convergence series.
- Flow Simulation Parametric Study — Sweeping inlet speed and switching between configurations to compare variants on the same settings.
- Configurations — Baseline, windscreen and tucked-posture variants held in one assembly so every run uses identical geometry outside the change.
- Excel — Hand calculation of cruise power from CdA, comparison tables and plots.
Software
- SOLIDWORKS 2023 or later
- SOLIDWORKS Flow Simulation
- Excel
Extra parts and data
- Measured rider anthropometric data — Shoulder width, seated height and arm position set the frontal area, which drives the drag result more than any other input.
Key concepts
- Drag coefficient and CdA
- Cd = Fd / (0.5·ρ·v²·A). For a scooter with rider, the frontal area is hard to define, so CdA (drag area in m²) is the more useful comparison figure.
- External analysis
- Flow Simulation solves the flow around the body inside a box-shaped computational domain. The domain must be large enough that its walls do not influence the result.
- Mesh convergence
- Refining the mesh until the drag changes by less than a set amount (for example 2 per cent) shows that the answer depends on the physics, not on cell size.
- Cruise power
- Power to overcome drag is P = 0.5·ρ·CdA·v³, so a small cut in CdA saves a lot of power at highway speed.
Step-by-step guide
- 1
Define the study and example inputs
Write a one-page study plan. Choose example speeds of 40 km/h and 70 km/h, air at 20 °C and sea-level pressure, and an example upright rider of 1.70 m height. List outputs: drag force, Cd, CdA, lift force and cruise power at each speed. Decide the variants to compare: baseline body, a tall windscreen, and a tucked rider posture. State that the model will be simplified, with no spokes, cables or mirrors, and that results compare variants rather than predict an exact road figure.
Checkpoint: A signed-off plan listing speeds, air properties, outputs and three variants, with assumptions written down.
- 2
Model a simplified scooter body
In SOLIDWORKS Part, build the body from lofts and boundary bosses: front apron, leg shield, floorboard, seat and rear body. Model wheels as solid discs with a tyre profile revolved from a sketch, and close every gap so the body is watertight. Use a global variable for wheelbase and seat height so variants stay consistent. Fillet sharp edges of 3 mm or more only where the real part is rounded; leave deliberate sharp trailing edges sharp, because they fix separation lines.
Checkpoint: Tools > Evaluate > Check reports no open or invalid faces, and the body is a single solid.
- 3
Build the rider and assemble
Make a simple rider from revolved and lofted solids: head, torso, upper and lower arms, thighs and shins, sized from the anthropometric data. Assemble the rider on the scooter with coincident mates at the seat and footboard, and angle mates at hips and shoulders. Create configurations for upright and tucked postures by changing the angle mates. Add a windscreen part with its own configuration. Check that rider and scooter touch rather than overlap; small overlaps confuse the fluid volume detection.
Checkpoint: Interference Detection shows no interference other than intended touching faces, and three configurations switch cleanly.
- 4
Set up the external Flow Simulation project
Run the Flow Simulation Wizard: SI units, External analysis, exclude internal spaces, air as the fluid, laminar and turbulent flow, and the inlet velocity along the vehicle axis set as the free-stream velocity. Leave gravity off for a pure drag study. Use a half model with a symmetry plane only if the rider and body are truly symmetric; otherwise keep the full model. Set the computational domain to roughly three body lengths ahead, eight behind and three to the sides and top.
Checkpoint: The domain box is visible around the model, and a trial run reports no warnings about the domain cutting through the body.
- 5
Model the ground
A scooter sits close to the road, so the ground affects the wake. Set the bottom boundary of the domain just below the tyre contact patch. Add a thin solid plate representing the road and apply a Real Wall boundary condition with a translational wall velocity equal to the free-stream speed, so the road moves relative to the scooter as it does in reality. Note that a stationary ground gives a thicker boundary layer and a different drag, and record which you used.
Checkpoint: A velocity cut plot at the ground shows near-zero relative velocity between air and moving road beneath the scooter.
- 6
Add goals and mesh controls
Insert Global Goals for force in X (drag) and Z (lift). Add an Equation Goal for Cd using the frontal area, and another for CdA as drag divided by dynamic pressure. Measure frontal area by projecting the silhouette onto the front plane with Silhouette Edges and reading the area from Section Properties. Start with automatic global mesh level 4, then add a Local Mesh region around the rider and wake with one or two extra refinement levels.
Checkpoint: Goals appear in the Goal Plot during the first run and converge without oscillating at the end of the solve.
- 7
Run a mesh-convergence series
Solve the baseline at 70 km/h with global mesh levels 3, 4, 5 and, if time allows, 6, keeping the same local refinement rule. Record cell count, run time and drag force for each. Plot drag against cell count. Choose the mesh where the change to the next level is under 2 per cent. Check that Cd is nearly the same at 40 km/h and 70 km/h; a big difference may point to a mesh problem rather than Reynolds-number physics.
Checkpoint: A convergence plot where drag changes by less than 2 per cent between the two finest meshes.
- 8
Read the flow results
Make Cut Plots of velocity and pressure on the symmetry plane and a horizontal plane at knee height. Add Surface Plots of pressure on the rider and body to find the high-pressure stagnation zones on chest, knees and leg shield. Draw Flow Trajectories from a plane ahead of the vehicle to show separation behind the rider's back and helmet. Identify the two or three regions that contribute most drag and explain why.
Checkpoint: Annotated plots that identify the main stagnation and separation regions, consistent with the force results.
- 9
Compare the design variants
Set up a Parametric Study or clone the project for each configuration: baseline, windscreen, tucked posture, and windscreen with tuck. Keep mesh settings identical. Collect drag, lift, CdA and the difference from baseline. A windscreen may cut drag on the rider but add frontal area, so its net effect can be small or negative; explain what you see from the pressure plots rather than assuming it helps.
Checkpoint: A table of CdA for every variant, with each change explained by a matching pressure or trajectory plot.
- 10
Convert drag to power and check by hand
Compute cruise power for each variant as 0.5·ρ·CdA·v³ at both speeds, then add an example rolling resistance for a 120 kg rider plus payload and an example scooter mass. Show the share of drag in total resistive power at 40 and 70 km/h. Sanity-check CdA against a hand estimate using frontal area and a bluff-body Cd of roughly 0.6 to 1.0. State that a coast-down or wind-tunnel test would be needed to validate the absolute value.
Checkpoint: Cruise power figures for each variant and a hand estimate that is in the same range as the CFD result.
- 11
Write the report
Document geometry simplifications, domain size, boundary conditions, mesh study, results for each variant and the power comparison. Include the convergence plot, pressure and trajectory images, and a table of CdA. List limitations: steady flow, simplified rider, no crosswind, no rotating wheels unless you added them. Recommend one change with its expected saving and the test that would confirm it. Save the Flow Simulation project with its results so a reviewer can open the plots.
Checkpoint: A report a reviewer can reproduce from the stated settings, with limitations and a clear recommendation.
Testing & validation
- Mesh-convergence series shows drag changing less than 2 per cent at the chosen mesh.
- Goals converge and the domain walls are far enough away (a larger domain changes drag by under 1 per cent).
- CdA lies within the range of a hand estimate from frontal area and a bluff-body Cd.
- Variant differences are larger than the mesh uncertainty before being reported as real.
Expected results
A watertight scooter and rider model, a converged external CFD set-up, CdA for four variants with flow images that explain the differences, and cruise power at two speeds showing how much each change saves.
Take it further
- Add rotating wheels with a rotating wall boundary condition and compare the drag.
- Run a crosswind yaw sweep at 5, 10 and 15 degrees and report side force.
- Feed CdA into a drive-cycle model to estimate range change on a city and highway route.
- Prototype a 3D-printed windscreen scale model for a smoke or tuft test.
Viva and review questions
Why report CdA rather than Cd for a scooter with rider?
The frontal area of a scooter and rider is irregular and depends on posture, so different people measure it differently. CdA combines both and directly sets drag force and power, which makes variants easier to compare.
How did you decide the computational domain size?
I started from about three lengths upstream and eight downstream, then enlarged it once and confirmed drag changed by less than 1 per cent, so the walls were not affecting the flow.
Why did you use a moving ground?
On the road there is no relative motion between air and road, so a stationary floor in the model would grow an unrealistic boundary layer under the scooter and change the wake. A moving wall removes that error.
Is your drag value correct for a real scooter?
It is reliable for comparing variants on the same settings. The absolute value needs validation by a coast-down or wind-tunnel test because the model omits small parts, rider clothing and real road turbulence.
Learn it at DIYguru
These DIYguru courses and programmes teach what this project needs.