A non-uniform onset flow may be prescribed. This support is useful
in a number of applications, especially during the design of
isolated components forming part of a larger configuration.
Propeller/airframe interactions may for example be modelled
by imposing the swirl induced downstream of a propeller as
an onset flow condition to a panel method model of the airframe.
Another typical example, presented here, is the case of the design
of the lower beam element of a Formula One rear wing.
A technique is required in which meaningful
studies can be performed without needing necessarily
to run a model including the whole of the car ahead of the wing
every time.
Furthermore, there may be situations where a panel method alone
has difficulty with solutions in a region of the model, for
example where gross flow separations exist. NEWPAN
is unquestionably the tool of choice for wing element design,
but in this case the element is sitting in very "dirty" air
behind the car. With support for non-uniform onset flow,
you have the opportunity to use experimental
or Navier-Stokes data to prescribe the flowfield that these
components sit in.
There is an additional complication. Consider again the case of a wing
assembly behind a racecar. The experimental or CFD complete car model
can provide you with the local velocities at points in the flowfield
surrounding the rear wing. But these velocities are of course a
combination of the onset flow induced by the car, and the perturbation to
the flow induced by the presence of the wing itself! The
model self-perturbation effect needs to be removed to leave the
actual non-uniform onset flow. (Taking the wing off the car and
recording the results is not a solution, since the car and its
wing are closely coupled - the flow about each affects the other).
Some codes other than NEWPAN claim to support non-uniform onset flow.
However this is of no use unless they also provide a means of
evaluating what the required onset flow should be! NEWPAN indeed
provides this capability - in the case we are describing, the key
is to iteratively remove the model self-perturbation effect during
the solution, thereby leaving the actual non-uniform onset flow.
Here we show example results from the case pursued in the side-bar.
We can compare the results achieved from the full car
model with those from the rear wing in isolation, with and
without a nonuniform onset flow. The magenta line shows the
pressure distribution from a section from the complete car.
The first image shows the poor agreement provided by a rear
wing in isolation in uniform flow. The second image shows the
close correspondence of results to the full car solution given
by the rear wing plus nonuniform onset flow case.
We shall now be able to perform rear wing design studies using
this model, without the rest of the car.

full car(magenta) v/s isolated rear wing in uniform onset flow(red).
Note the (not unsurprising)
poor agreement.

full car(magenta) v/s isolated rear wing in non uniform onset flow(red).
We now have much improved accuracy with significantly reduced run time
|
|
Non-uniform onset flow, an example:

Step 1. Using a complete car model, extract local velocities
at points in the flowfield surrounding the rear wing.

Step 2. Use these velocities to provide initial values for the
onset flow perturbations to our NEWPAN isolated rear wing model.
Step 3. Run NEWPAN. In the course of its run, NEWPAN evaluates
the self-influence velocity perturbations of the wing itself
at the given points and iteratively removes these.


On output, NEWPAN provides results for the rear wing sitting
in the actual non-uniform onset flow, without the self-influence,
and also returns the values of the onset flow perturbations at
the input points. These can be used for the next run directly.
Shown above are vectors of velocity perturbation
(i.e. with uniform onset flow removed) before and after the
influence of the rear wing itself has been removed.
|