GTS
03-17-2015, 11:44 PM
A quick note to teams preparing 2015 FSAE-A entrants regards aero; I’ve had a few teams reach out and go over preliminary work, and thought I’d offer a little early feedback.
I’ll stress (it’s been said many times) that CFD is not a prerequisite for aerodynamic development. Full-car CFD of an open-wheeled racecar is a difficult challenge even for CFD professionals. Done poorly it will not advance your team’s progress, and will actually impede it.
A few CFD supplier papers seem to have suggested (among other things) a peculiar domain configuration, or particularly tight domain dimensions against vehicle models.
The peculiar domain configuration in question requires some significant understanding to implement as intended, and even so its relevance is questionable. No university has yet shown me a replication of this configuration that is at least true to the originator’s work. Please check your boundary conditions at least. You know who you are.
Teams having shown rectangular domains need to reconsider domain size from a pressure-at-boundary perspective. I am yet to see a domain appropriately sized relative to an FSAE racer from teams having approached me so far.
An open-wheeled car simulated for relatively complete fidelity on a moderate turbulence model will far exceed the computational resources of current consumer-class desktop computers. Cutting corners indiscriminately is not a solution. Compromised results born of poor understanding only waste your and your team’s time.
Whilst CFD allows some impractical simulations, it is not feasible to run these. Slipping the floor without rotating wheels is not generally a simulation worth your time.
Inflation layers are essential to a relevant simulation. Please understand why and how to validate their appropriateness for your chosen turbulence model.
If you insist on CFD and high-performance aerodynamic configurations, please consider designing your car as to be easier to simulate and evaluate. If practical flows are difficult to simulate in CFD owing to the nature of their turbulence, chances are that what geometries determine them could be better designed. Remember above all that motorsport aerodynamics (as with most other aerodynamics) is an exercise in momentum conservation: to keep air as axially unimpeded as possible (low drag) for maximum vertical force imparted (high downforce) under all relevant conditions.
I would ask all teams to undertake to understand what they’re simulating and why.
A FSAE racer is not, under any circumstances, a suitable first CFD project. It is not the complexity of the shape that determines difficulty; it’s the complexity of inherent flows. Don’t be shy about starting with simple shapes for which much data already exists. Start with 2D wings, through 3D wings and Ahmed models if you’re getting started – start with known shapes with characteristic flows, and use published data and prior works to validate progress in your skills. If you can do this and then you get started on a car, your progress will be considerably more rapid.
Consider that for a winged design there are better computational tools for basic concept validation than high-fidelity RANS simulations. Think about it.
Pick your battles in development carefully. Start with performance targets, not with component design. Your first simulations do not require fluids simulation.
Remember above all that simulation is not an ideation tool; it’s an inspection and validation tool. You don’t need CFD at all to generate and iterate through good and better ideas. You need a solid understanding of first principles, butcher’s paper and pencils. Or clay. Or whatever it takes you to explore your ideas. Simulation comes much later, well after having solid ideas to evaluate.
If not so far ahead, seek out solid ideas first.
Good luck all in 2015.
I’ll stress (it’s been said many times) that CFD is not a prerequisite for aerodynamic development. Full-car CFD of an open-wheeled racecar is a difficult challenge even for CFD professionals. Done poorly it will not advance your team’s progress, and will actually impede it.
A few CFD supplier papers seem to have suggested (among other things) a peculiar domain configuration, or particularly tight domain dimensions against vehicle models.
The peculiar domain configuration in question requires some significant understanding to implement as intended, and even so its relevance is questionable. No university has yet shown me a replication of this configuration that is at least true to the originator’s work. Please check your boundary conditions at least. You know who you are.
Teams having shown rectangular domains need to reconsider domain size from a pressure-at-boundary perspective. I am yet to see a domain appropriately sized relative to an FSAE racer from teams having approached me so far.
An open-wheeled car simulated for relatively complete fidelity on a moderate turbulence model will far exceed the computational resources of current consumer-class desktop computers. Cutting corners indiscriminately is not a solution. Compromised results born of poor understanding only waste your and your team’s time.
Whilst CFD allows some impractical simulations, it is not feasible to run these. Slipping the floor without rotating wheels is not generally a simulation worth your time.
Inflation layers are essential to a relevant simulation. Please understand why and how to validate their appropriateness for your chosen turbulence model.
If you insist on CFD and high-performance aerodynamic configurations, please consider designing your car as to be easier to simulate and evaluate. If practical flows are difficult to simulate in CFD owing to the nature of their turbulence, chances are that what geometries determine them could be better designed. Remember above all that motorsport aerodynamics (as with most other aerodynamics) is an exercise in momentum conservation: to keep air as axially unimpeded as possible (low drag) for maximum vertical force imparted (high downforce) under all relevant conditions.
I would ask all teams to undertake to understand what they’re simulating and why.
A FSAE racer is not, under any circumstances, a suitable first CFD project. It is not the complexity of the shape that determines difficulty; it’s the complexity of inherent flows. Don’t be shy about starting with simple shapes for which much data already exists. Start with 2D wings, through 3D wings and Ahmed models if you’re getting started – start with known shapes with characteristic flows, and use published data and prior works to validate progress in your skills. If you can do this and then you get started on a car, your progress will be considerably more rapid.
Consider that for a winged design there are better computational tools for basic concept validation than high-fidelity RANS simulations. Think about it.
Pick your battles in development carefully. Start with performance targets, not with component design. Your first simulations do not require fluids simulation.
Remember above all that simulation is not an ideation tool; it’s an inspection and validation tool. You don’t need CFD at all to generate and iterate through good and better ideas. You need a solid understanding of first principles, butcher’s paper and pencils. Or clay. Or whatever it takes you to explore your ideas. Simulation comes much later, well after having solid ideas to evaluate.
If not so far ahead, seek out solid ideas first.
Good luck all in 2015.