CDi in VSPaero

507 views
Skip to first unread message

hugh.bl...@gmail.com

unread,
Sep 6, 2021, 5:00:34 AM9/6/21
to OpenVSP
What is the exact meaning of CDi in VSPaero calculations?  Is it (as I have been assuming) inviscid lift-induced drag coefficient, or is it a more general lift-induced drag coefficient, which could include CL effects on viscous drag estimates used in VSPaero?  I generally like to use the Parasite Drag tool to find viscous drag and VSPaero to find inviscid drag, since I believe that the Parasite Drag tool is probably more accurate for the viscous drag, but want to be sure there is no viscous contribution to CDi.

Thanks
Hugh

Rob McDonald

unread,
Sep 6, 2021, 1:07:53 PM9/6/21
to ope...@googlegroups.com
You're on the right track.

The CDo column contains all the viscous estimates.  These can vary with lift.

The CDi column is a fully inviscid calculation based on summing loads on all panels.

CDtot is the sum of CDo and CDi.

The CDTrefftz is an inviscid calculation based on a trailing edge analysis.

My general opinion of the viscous model built-in to VSPAERO is similar to yours....  I would generally rather use the parasite drag tool.  At a minimum, it is common practice to eliminate various components (certainly excrescences) -- these can be captured in the parasite drag analysis.

The place where the VSPAERO viscous model is essential is when calculating rotating blades.  It provides an estimate of prop / rotor profile power that would not be easily obtained otherwise.  

The TODO list for VSPAERO is long and ambitious....  There are intentions to improve the viscous model on that list (increasing transparency and tweakability), but I have no idea when it will percolate to the top of the priority stack.

Rob



--
You received this message because you are subscribed to the Google Groups "OpenVSP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openvsp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openvsp/63c5c3ca-fcf0-430d-b7fb-0b769b5bcce2n%40googlegroups.com.

hugh.bl...@gmail.com

unread,
Sep 6, 2021, 10:47:06 PM9/6/21
to OpenVSP
Thank you.  I agree with all that.  

But then: what is the difference between CDi and CDTrefftz?  Shouldn't they be much the same?  I'm guessing the calculation method differs - and that generally one (i.e. me) should prefer CDTrefftz over CDi.  But when I check, CDTrefftz seems to be identically zero over a CL sweep, so either it's non-functional in 3.25.0 or I don't understand what it is meant to compute.

thanks again...
Hugh

Rob McDonald

unread,
Sep 7, 2021, 12:41:25 AM9/7/21
to ope...@googlegroups.com
CD is obtained by an integral of the load on all the panels.

The Trefftz version is obtained through an integral on just the trailing edges.

The load on a panel is obtained by summing the loads on the vortex loop edges of the panel.  'Before', the velocities at the edge centers were interpolated from known values.  In the version bundled with 3.25.0, this was changed to use velocities directly calculated at the edge centers - this improved accuracy, but is computationally more expensive.

In this version, Dave also recognized that the formulation for the Trefftz analysis was not strictly correct for unsteady cases.  I thought he had made another change to fix this, but if it is showing up as zero, I suspect he decided to zero it out when it is not correct.  If he can sort out a formulation that works generally, I think it will come back.  So long as it is zero, I clearly wouldn't use it.

Yes, in general the Trefftz analysis should match the panel integration version.  Yes, it will generally converge before the panel integration version converges.

Rob


Reply all
Reply to author
Forward
0 new messages