Negative drag due to elevator deflection #357

seanmcleod · 2020-12-11T16:16:49Z

While debugging a difference in trim results I noticed that the drag due to elevator deflection was negative for the 737 model.

    aero/coefficient/CD0 = 2801.679561
    aero/coefficient/CDi = 3813.129823
    aero/coefficient/CDflap = 5689.318388
    aero/coefficient/CDgear = 1928.582504
    aero/coefficient/CDsb = 0.000000
    aero/coefficient/CDsb = 0.000000
    aero/coefficient/CDsp = 0.000000
    aero/coefficient/CDbeta = 0.007089
    aero/coefficient/CDde = -181.721539

Taking a look at the 737 model's calculation for elevator deflection drag it's using elevator-pos-norm [-1, +1] without converting it to an absolute value, hence the reason you can see negative drag values.

<function name="aero/coefficient/CDde">
    <description>Drag_due_to_Elevator_Deflection</description>
    <product>
        <property>aero/qbar-psf</property>
        <property>metrics/Sw-sqft</property>
        <property>fcs/elevator-pos-norm</property>
        <value>0.059</value>
    </product>
</function>

In context that means at around the critical AoA of ~12.5 degrees the negative drag from a full elevator deflection will completely cancel out the total induced drag!

Also if the aircraft was trimmed at say ~2 degrees AoA and you put in a sudden step of full elevator deflection the aircraft would actually start accelerating with this extra negative drag 'thrust'.

In addition to the sign issue with the elevator deflection the use of a normalized position [-1, +1] is unusual since the coefficient is normally per radian. It's not impossible that it's been converted/scaled, however the value of 0.059 is very similar to the typical values published that are per radian.

The 737 model for example has a maximum elevator deflection of 0.3 rad so if the figure of 0.059 is per radian then that would equate to a value of 0.018 for a normalized range.

I did a search through all the JSBSim aircraft for /CDde to see whether the 737 model was an exception or not.

Unfortunately it's not. The following aircraft models follow the same setup as the 737 model.

Incorrect use of `elevator-pos-norm`

jsbsim/aircraft/sgs233/sgs233.xml

Line 321 in eccfd64

jsbsim/aircraft/F4N/F4N-jsbsim.xml

Line 403 in eccfd64

jsbsim/aircraft/L17/L17.xml

Line 378 in eccfd64

jsbsim/aircraft/sgs126/sgs126.xml

Line 363 in eccfd64

jsbsim/aircraft/OV10/OV10.xml

Line 502 in eccfd64

jsbsim/aircraft/MD11/MD11.xml

Line 495 in eccfd64

jsbsim/aircraft/A4/A4.xml

Line 541 in eccfd64

jsbsim/aircraft/C130/C130.xml

Line 477 in eccfd64

jsbsim/aircraft/pogo-jsbsim/pogo-jsbsim.xml

Line 614 in eccfd64

jsbsim/aircraft/B747/B747.xml

Line 533 in eccfd64

jsbsim/aircraft/T38/T38.xml

Line 570 in eccfd64

jsbsim/aircraft/XB-70/XB-70.xml

Line 604 in eccfd64

jsbsim/aircraft/DHC6/DHC6.xml

Line 324 in eccfd64

jsbsim/aircraft/737/737.xml

Line 632 in eccfd64

jsbsim/aircraft/T37/T37.xml

Line 469 in eccfd64

jsbsim/aircraft/f22/yf22.xml

Line 529 in eccfd64

jsbsim/aircraft/B17/B17.xml

Line 910 in eccfd64

jsbsim/aircraft/fokker50/fokker50.xml

Line 695 in eccfd64

jsbsim/aircraft/L410/L410.xml

Line 1187 in eccfd64

Correct use of `mag-elevator-pos-rad`

The following aircraft make use of the magnitude that the FCS provides - mag-elevator-pos-rad which is an absolute version and radian based.

            <function name="aero/coefficient/CDde">
                <description>Drag_due_to_elevator_Deflection</description>
                <product>
                    <property>aero/qbar-psf</property>
                    <property>metrics/Sw-sqft</property>
                    <property>fcs/mag-elevator-pos-rad</property>
                    <value>0.0500</value>
                </product>
            </function>

jsbsim/aircraft/t6texan2/t6texan2.xml

Line 327 in eccfd64

jsbsim/aircraft/pa28/pa28.xml

Line 431 in eccfd64

jsbsim/aircraft/c172x/c172x.xml

Line 911 in eccfd64

jsbsim/aircraft/f15/f15.xml

Line 594 in eccfd64

jsbsim/aircraft/fokker100/xf100.xml

Line 720 in eccfd64

jsbsim/aircraft/A320/A320.xml

Line 602 in eccfd64

jsbsim/aircraft/c172r/c172r.xml

Line 500 in eccfd64

jsbsim/aircraft/fokker100/fokker70.xml

Line 727 in eccfd64

jsbsim/aircraft/fokker100/fokker100.xml

Line 776 in eccfd64

jsbsim/aircraft/dr1/dr1.xml

Line 665 in eccfd64

Coefficient of 0

The following aircraft make use of mag-elevator-pos-rad but have a coefficient of 0.

jsbsim/aircraft/c182/c182.xml

Line 483 in eccfd64

jsbsim/aircraft/c172p/c172p.xml

Line 551 in eccfd64

jsbsim/aircraft/c310/c310.xml

Line 670 in eccfd64

Sort of correct use of `<abs> elevator-pos-norm </abs>`

Some aircraft make use of an absolute version of elevator-pos-norm using <abs /> but should more than likely be using pos-rad.

jsbsim/aircraft/p51d/p51d.xml

Line 960 in eccfd64

jsbsim/aircraft/Boeing314/Boeing314.xml

Line 1313 in eccfd64

jsbsim/aircraft/Concorde/Concorde.xml

Line 1526 in eccfd64

Correct sign but confusing `elevator-pos-rad` and `elevator-pos-norm`?

The following two aircraft use a table to get the sign correct, but they appear to be confusing elevator-pos-rad with elevator-pos-norm.

                          <independentVar>fcs/elevator-pos-rad</independentVar>
                          <tableData>
                              -1.0000	0.1140
                              0.0000	0.0000
                              1.0000	0.1140
                          </tableData>

jsbsim/aircraft/minisgs/minisgs.xml

Line 331 in eccfd64

jsbsim/aircraft/SGS/SGS.xml

Line 332 in eccfd64

<description>Drag_due_to_Elevator_Deflection</description>

Sign issue with `elevator-pos-rad`

This aircraft uses elevator-pos-rad but forgets about the sign issue.

jsbsim/aircraft/F80C/F80C.xml

Line 547 in eccfd64

Table based `elevator-pos-rad` that gets the sign correct

jsbsim/aircraft/f22/f22.xml

Line 1931 in eccfd64

Aeromatic

Lastly I took at look at the Aeromatic source code to see if it was possibly the reason for most of the incorrect cases, but at least currently, I didn't look at it's source code history, it does the correct thing, outputting - <abs><property>fcs/elevator-pos-rad</property></abs>

To Do

I'm busy preparing a pull request to correct all the examples listed above that I labelled problematic, unless someone tells me I've got it all wrong 😉.

The text was updated successfully, but these errors were encountered:

bcoconni · 2020-12-11T18:25:19Z

Looks good to me ! 👍

Out of curiosity, I have checked the official c172p model for FlightGear and they are using the exact same function than our c172p model: the drag is computed correctly with fcs/mag-elevator-pos-rad and cancelled out with a 0 value.

            <function name="aero/coefficient/CDDe">
                <description>Drag_due_to_Elevator_Deflection</description>
                <product>
                    <property>aero/qbar-psf</property>
                    <property>metrics/Sw-sqft</property>
                    <property>fcs/mag-elevator-pos-rad</property>
                    <value>0.0000</value>
                </product>
            </function>

seanmcleod · 2020-12-22T14:36:37Z

Example of the increase in IAS as a step input to the elevator is made with the 737 model.

Same 737 model except that the orange version /fdm/jsbsim/velocities/vc-kts uses <abs> for elevator deflection drag.

And how the IAS differs over the next 8s. This is in response to an elevator doublet.

Octal450 · 2021-01-30T15:44:57Z

Its definitely wrong. Negative drag should not. Either use the property that has absolute value or do it yourself.. for example in my MD-11:

	<function name="aero/coefficient/CDde_lo">
		<description>Drag due to left outer elevator</description>
		<product>
			<property>aero/qbar-psf</property>
			<property>metrics/Sw-sqft</property>
			<abs><toradians><property>hydraulics/elevator-out-l/final-deg</property></toradians></abs>
			<value>0.003625</value>
		</product>
	</function>

Kind Regards,
Josh

bcoconni · 2021-01-30T15:47:50Z

@Octal450 What's wrong ?

bcoconni · 2021-02-20T10:30:46Z

@seanmcleod it seems this issue has been fixed. Shall we close it now ?

seanmcleod · 2021-02-20T10:36:27Z

Yep, closing.

Octal450 · 2021-02-20T15:10:44Z

@Octal450 What's wrong ?

Just agreeing with that it was wrong, that's all.

Kind Regards,
Josh

bcoconni added the bug label Dec 11, 2020

seanmcleod mentioned this issue Dec 11, 2020

Fix negative drag due to elevator deflection #358

Merged

bcoconni mentioned this issue Dec 12, 2020

Update CDDe from 0 to 0.06 #359

Merged

seanmcleod closed this as completed Feb 20, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Negative drag due to elevator deflection #357

Negative drag due to elevator deflection #357

seanmcleod commented Dec 11, 2020

bcoconni commented Dec 11, 2020

seanmcleod commented Dec 22, 2020 •

edited

Loading

Octal450 commented Jan 30, 2021 •

edited

Loading

bcoconni commented Jan 30, 2021

bcoconni commented Feb 20, 2021

seanmcleod commented Feb 20, 2021

Octal450 commented Feb 20, 2021

Negative drag due to elevator deflection #357

Negative drag due to elevator deflection #357

Comments

seanmcleod commented Dec 11, 2020

Incorrect use of elevator-pos-norm

Correct use of mag-elevator-pos-rad

Coefficient of 0

Sort of correct use of <abs> elevator-pos-norm </abs>

Correct sign but confusing elevator-pos-rad and elevator-pos-norm?

Sign issue with elevator-pos-rad

Table based elevator-pos-rad that gets the sign correct

Aeromatic

To Do

bcoconni commented Dec 11, 2020

seanmcleod commented Dec 22, 2020 • edited Loading

Octal450 commented Jan 30, 2021 • edited Loading

bcoconni commented Jan 30, 2021

bcoconni commented Feb 20, 2021

seanmcleod commented Feb 20, 2021

Octal450 commented Feb 20, 2021

Incorrect use of `elevator-pos-norm`

Correct use of `mag-elevator-pos-rad`

Sort of correct use of `<abs> elevator-pos-norm </abs>`

Correct sign but confusing `elevator-pos-rad` and `elevator-pos-norm`?

Sign issue with `elevator-pos-rad`

Table based `elevator-pos-rad` that gets the sign correct

seanmcleod commented Dec 22, 2020 •

edited

Loading

Octal450 commented Jan 30, 2021 •

edited

Loading