Evaluation of glider model realism in CRRCsim

1. Revisions of this document

2. Introduction

CRRCSim provides several glider models to fly with.

Some of them are included in the distribution package, others can be download from CRRCsim Wiki Models page.
An airplane model is defined by its 3D model, often an .ac file in folder objects, which is responsible for the appearance of the airplane, and by its FDM (flight dynamic model), descibed in a .xml file in folder models, which is responsible for the simulation of its (aero)dynamic behaviour.

This document analyzises most of the glider models looking in detail at a few of the many parameters which are part of the FDM and showing that (to the author's opinion) quite a few of the glider models available as of December 2012 have serious flaws which produce unrealistic simulations.

3. A table of FDM parameters

The table below shows some of the most important parameters of the FDM related to: wing size, airfoil and airplane drag, mass and inertias. The reliability of these parameters can often be judged based on some relatively simple considerations.

While many more parameters are also important, in particular those describing the stability of the airplane and the reaction to the controls, they shall best be derived from an analysis of the airplane through e.g. AVL or XFLR5 codes. In most cases it is not that easy to judge "by eyes" if they look reasonable or not. Thus I will not make any comment about them here, assuming they are all ok.

In the table, models from Allegro to Zipper were included in the standard package as of December 2012, except for Pilatus B4 2.6m and Generic F3F which have been added by the author on January 2013; Pilatus B4 New is just the revised version of the original Pilatus B4. The other models are representative of most of the glider models available for download on the CRRCsim Wiki Models page.

In the table above:

• RED marks wrong values.
  
• ORANGE marks likely wrong values.
  
• YELLOW marks likely inaccurate values.
  

Note that Crossfire, Erwin, Ceres, Vampire and Sniper Evo actually share most of the FDM parameters with the Skorpion, and in fact all the stability & control parameters (not shown in the table) are just an exact copy of those of the Skorpion while other parameters are equal or just slightly modified.
This means that apart from looking differently they will all fly exactely the same in CRRCsim.
Actually many more F3F/F3B gliders are available for download (in the CRRCsim-addon-models package): while having detailed and specific 3D models and very nice liveries, they all share the same FDM model of the Skorpion.
Unfortunately, this means that they all share the flaws of the Skorpion FDM, as will be shown in the following!

Similarly, Sovereign and Zipper actually uses the same FDM as the Allegro. In this case, at least, they share a good FDM although the real glider may behave differently.

3.1 Mass and inertias

The following picture shows for each model the radius of gyration of the mass moment of inertia Ixx, Iyy, Izz (respectively for roll, pitch and yaw axis) in percentage of the wing span.

From the chart above we notice that:

1. All the Skorpion-like models have significantly smaller radius of gyration, hence lower moments of inertia. Looking at the table above this is mostly due to an excessive mass. In fact, although F3F models may be ballasted, mass of the order of 5kg up to 11kg (!) seem really excessive to me.
2. The original Pilatus B4 (a 4.6m span glider) also has much too low moments of inertia. In this case the problem is not the mass, which is reasonable, but the values of the inertias which have been estimated wrong. This fact is promptly recognized when flying the model, which looks really unrealistically "light" and responds too sharply to control inputs (e.g rudder).
3. The somewhat low inertias of the Flexifly are justified being it a foamie, and are thus realistic.
   
4. A new Generic F3F model has been derived from the Skorpion, to be used with any of the available F3F 3D model. Wrong values have been corrected using reasonable guesses. Just setting the mass in the range 3-4kg brings the inertias in the expected range suggesting that, contrary to the mass, they were estimated right.
   
5. The new Pilatus B4 2.6m model uses the same 3D model (i.e. the .ac file) of the original Pilatus B4 (scaled and slightly revised), but all the FDM properties have been now redefined to reproduce as accurately as possible the 2.6m wingspan version of this glider made by Tangent (author's latest model :-).
   
6. Mass and inertias of the Pilatus B4 2.6m have then been scaled-up back to the 4.6m size, and used to correct the original Pilatus B4 data. The Pilatus B4 New model is thus back in line, providing a much more sensible and realistic simulation.

3.2 Drag

The following picture shows for each model the efficiency (L/D) resulting from the FDM parameters for a few values of the lift coefficient: -0.6 (inverted flight), 0.2 (speed), 0.6 (often close to max efficiency), 0.9 (often close to max lift).

In CRRCsim (Larcsim FDM) the airplane drag is estimated considering a few contributions, among which:

• Minimum airplane drag CD_prof, which is the drag at the CL specified by CL_CD0. Note that despite its name CD_prof shall include the drag contribution due to fuselage, tails, etc.. not just that due to wing airfoil.
  
• Drag variation with the square of the lift coefficient CD_CLsq (mostly due to the variation of airfoil drag with angle of attack).
• Effect of the Reynolds number based on the reference Speed and Uexp_CD: the larger the ratio between flying speed and the reference Speed, i.e. the larger the Reynolds number, the lower the drag; this effect is stronger for larger negative value of Uexp_CD.
• Induced drag, related to CL, aspect ratio (= Span^2/Area) and span_eff factor.
  

Additional drag contributions may come from control deflection and stalled condition but these effects are not included in the computation of the efficiency shown in the chart.

From the chart above we notice that:

1. All the Skorpion-like models have too high efficiency at low CL (0.2) and much lower efficiency at higher CL: this is not right. Looking at the table above we see it is caused by: firstly a much too small CD_prof (too small even if it was the airfoil CD alone); secondly a huge wing area (about 3 times the real value!) which results in an extremely low aspect ratio (about 5, instead of 15!), hence a large induced drag at increasing CL.
2. The original Pilatus B4 also has a similar behaviour due again to a much too small CD_prof. But now the aspect ratio is ok, so the glider has an incredibly high efficiency.
3. By comparison, the new Pilatus B4 2.6m, Pilatus B4 New and Generic F3F models with appropriate estimate of CD_prof, CL_CD0 and Uexp_CD have reasonable and realistic perfomances, with a proper variation of efficiency with CL.
4. Both Fennec and Crobe models have too small CD_prof, resulting in too high efficiency for such small gliders (should be close but lower than respectively Allegro and Apogee).
5. In most cases efficiency at CL=-0.6 (inverted flight) results equal to that at CL=0.6. This is almost always wrong since cambered airfoils usually have minimum drag at a CL greater than zero, the more cambered the airfoil the higher the CL at which minimum drag is achieved. Hence, they generate much less drag at CL=0.6 than CL=-0.6 (which in some cases might be even beyond the limit of negative stall).
   As we can see from the table for most models CL_CD0 has just been set to zero (or practically zero), only exception being: Pilatus B4 2.6m, Pilatus B4 New, Generic F3F, Zagi-xs, Fennec, Crobe and Riser. As a consequence only these few models realistically have low efficiency at CL=-0.6.
   Thus, I marked in yellow all the models whose value of CL_CD0 (and of CD_CLsq which is related to it) are most likely inaccurate, causing too good performance in inverted flight (although they may look realistic in normal flight).
6. The Wasabi model is a justified exception to the previous comment. In fact it has a symmetric airfoil so that with flap at 0° it flies equally well straight or inverted. However, when flaps are lowered to 4° the CL for minimum drag becomes positive, hence efficiency at large CL improves while efficiency at negative CL reduces, as it happens in real-life.

Although their effect are more subtle and would require a longer explanation, the specification of the reference Speed and of Uexp_CD is also likely not right in many cases.
Speed is often too low resulting in too low a reference Reynolds number, thus the drag computed while flying is lower than intended.
Uexp_CD is sometimes to large (e.g. -0.6 used for Skorpion-like models), which results in too low drag at high flying speeds. In my opinion Uexp_CD should be within -0.25 to -0.45

4. Conclusion

Many models are available for CRRCsim but some have serious flaws in their FDM parameters, significantly compromising the realism of the simulation. The worst cases are the various F3F gliders (which are actually all sharing the same problematic FDM) and the two large scale gliders Pilatus B4 and Fox (this last has not been discussed so far but its FDM is an exact copy of the flawed Pilatus one..).

To provide a reasonable FDM for F3F-like gliders a new Generic F3F model has thus been implemented, replacing Crossfire, Erwin and Skorpion in the standard distribution package; the various 3D models are now just different graphical appearence of the same FDM.
This FDM could also be applied to the many more 3D models of F3F gliders available as add-on (this requires editing of the relevant .xml files).

The FDM of Pilatus B4 has been revised, based on new estimate of aerodynamics, mass and inertias obtained scaling to the larger size those of a new Pilatus B4 2.6m small scale glider which reproduces the model made by Tangent.

Most other models are reasonably realistic except for what the drag variation with lift is concerned which is appropriately simulated only by a few models.