Accuracy of VSPAero in estimating Neutral Point

[Jess Terr]

Hi,

I am trying to compare the neutral point estimate that VSPAero provides compared to a lower order approximation method uses here B747_Cnb_locus.eps (cornell.edu)

I have been finding some differences of about 6ft which is quite significant when it comes to static stability calculations. I would expect a lower difference between both methods. I know that VSPAero computes the neutral point based on the result of running two alphas and examining the slope of the CM vs CL curve. The empirical method uses aircraft geometry/aero parameters to estimate the NP (see ref.). For VSPAero, I am using the reference areas/lengths of the wing (including an estimated MAC value), the estimated CG based on the weight and balance of the vehicle, and the cruise flight conditions.  I am not completely sure about the accuracy of vspaero in providing the neutral point. Has anyone else had a good match in calculating the neutral point/ static margin between vspaero vs empirical method?

Also, where can I find VSPAero calculations of the neutral point and static margin in the original/build code? 

Thanks,

Jess

[Rob McDonald]

If you would like someone to comment on a reference, please don't post a link to a 153 page document.  At least tell us where in that document is the method that you want to compare with.

Does the VSPAERO solution look good -- do you believe the load distributions, have you looked at the Cp distribution and the wake?  Do you have good spanwise and chordwise resolution?  Does it pass all of your sanity checks?

If so, then the VSPAERO neutral point calculation should be just as believable.  It will be consistent as it is simply performed by looking at the two cases and computing the slope of the CM vs. CL curve.

The calculation  you ask for is here -- but that really is not very relevant.  It is a trivial calculation performed after a very complex aerodynamic calculation (the rest of VSPAERO).

I believe the method in the textbook you cite does not include sophisticated aerodynamic calculations -- it just uses simple MAC and C/4 arguments to build up an estimate.  If this is correct (and if you have a quality VSPAERO solution) then the VSPAERO solution should be superior.

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/b871852b-3c22-4ca6-a7be-c9470c823c51n%40googlegroups.com.

[Jess Terr]

Hi Rob,

Thanks for the reply. You are right- I totally forgot to include the reference equation from the link.  The referenced NP equation is 3.18 in the link. 

I have checked the model to make sure things look good. Load distributions seem reasonable, and an expected CL value of 0.53 is reasonable, there are no random Cp locations in the model when run, and the wake looks fine to me.  The spanwise and chord distribution seem ok but might need to be adjusted (for wing- NumW=49, numU=8, for Horizontal Tail-  NumW=41, numU=15, Vertical Tail-  NumW=33, numU=11, and fuselage-  NumW=33).  I am running the model using the fuse, horizontal tail, vertical tail, and wing components to compute the aero estimates and NP.  I am also running separately only the fuse to get CMy_aoa and also separately the horizontal tail to get CL_aoa and CMy_aoa so I can use those values in the low order method.

Correct, the method in the book includes low-order equations to approximate the aero coeffs. I have also used the aero coeffs from vspaero to estimate the NP using eqn 3.18, as mentioned above, but I still get about 7ft difference from both methods. I also agree that vspaero should provide a 'more' accurate NP/SM estimate since this is a higher-order approximation than eq 3.18 in the tex,t but I think they should be less than a couple of feet apart anyway.

Since the NP and SM change with a change in AoA, I am thinking the cruise flight AoA I am using might not be correct and might need to be modified. I have increased the AoA value I used when running vspaero and it matches closer to the low order estimate.... I still need to investigate this hypothesis.

Thanks,

-Jess

[Rob McDonald]

The neutral point in fact does move with angle of attack -- it is only a convenient artifice that it is a fixed point.  This is particularly true when all of the force and moment generating items are not aligned with the CG in the Z direction.  There is much that the elementary analysis leaves out.

I have much lower expectations than you do for the textbook method.  It has no way to know about the chordwise or spanwise load distribution on the wing -- it does not know how to model sweep in any detail -- it is not a sophisticated aerodynamic analysis.  Its buildup is primarily useful for understanding the main contributors and how they combine together -- not for a detailed analysis.

If you want to make relevant comparisons, then build an AVL model - or perhaps a Tornado model.  Or write your own VLM method from scratch and look at the results you get from there.  All will be more relevant for the comparison.

Rob