Article on glide strategy

On Feb 28, 4:13*pm, Max Kellermann wrote:
Ramy wrote:
This is why Safety MC, or polar degradation are better solutions,
since they are not depending on distance.

Safety MC is poor man's polar degradation. *It doesn't make a lot of
sense, and will eventually be removed from XCSoar. *We should have
implemented persistent polar degradation from the start.

Max

Max, first thanks for addressing those issues in 6.3.
I tried both options (polar degradation and safety MC) recently, with
exact same results, so am not completely sure why Safety MC is poor
man's polar degradation. Since MC is now decoupled, I don't see a
problem using it for polar degradation, from the reasons John
described in his article. I think many still prefer to use MC to
degrade the polar instead of polar degradation. I also noticed at
least one advantage in safety MC: when clicking on a waypoint, you can
see you arrival altitude both with safety MC and with MC=0, so you can
estimate your chances to make it in still air. Polar degradation does
not provide such best case scenario an option since MC=0 is already
degraded.
Please explain why safety MC doesn't make a lot of sense.
In any case, I would recommend against removing safety MC, as pilots
may be using it for various reasons.

Ramy

On Feb 27, 8:15*pm, John Cochrane
wrote:
This will probably end up in Soaring sooner or later, but I always get
a lot of help from early readers. If it's not clear or you see
problems etc. let me know. (john dot cochrane at chicagobooth dot
edu0

John Cochrane

John,

You asked for feedback.

The lift to drag ratio of a glider is not impacted by wind. I know
you know that so why use the term L/D out of context? Can you please
use a term such as (achieved) glide angle instead of L/D. Others have
coined the term "efficiency" but I have not taken to that yet.

Other that that nice article. I have been flying decoupled MC glides
for a long time. Even if don't split the speed director and glide
computer I alternate between the two MC values to sample "what glide
do I have" and "what speed should I fly right now". I suppose a
classic example of this is to climb in the last thermal to the
equivalent MC required glide altitude and then fly a lower MC until
clear of hostile terrain. I like to think of my final glides as x
feet over a y MC setting where how good x and y feel depends directly
on the terrain between myself and the goal. Where I fly having a bit
extra 5 miles out is worth quite a few contest points.

The biggest gotcha when believing a glide computer may be when there
is a strong tail wind at altitude. Easy to get suckered into a final
glide that doesn't work out when the tail wind decreases.

Andy (GY)

Ramy wrote:
Please explain why safety MC doesn't make a lot of sense.

Safety MC was implemented to degrade the polar, nothing else. It has
no effect on speed suggestions or anything else related to MacCready's
theory.

So what you probably want is a third polar degradation setting:

1. "polar degradation" adjusts the configured polar to match your
real-world glider (persistent, new XCSoar 6.3 feature)

2. "bugs" adjusts the polar to match the dirty glider (not persistent,
has always been there)

3. "safety degradation" to replace "safety MC"; basically the same as
"safety MC" just with a name and definition that makes more sense.

I'm not 100% sure about this; these are just my unfinished thoughts
after weeks of reading XCSoar's glide solver code. Maybe safety MC is
a setting to account for long distances of sink lacking any lift, and
the setting should be the expected sink value (and not the expected
lift value in the next thermal!).

Max

On Feb 28, 9:22*pm, Max Kellermann wrote:
Maybe safety MC is
a setting to account for long distances of sink lacking any lift, and
the setting should be the expected sink value (and not the expected
lift value in the next thermal!).

Max

That's a really insightful thought.

-Evan Ludeman / T8

On Feb 28, 6:22*pm, Max Kellermann wrote:
Ramy wrote:
Please explain why safety MC doesn't make a lot of sense.

Safety MC was implemented to degrade the polar, nothing else. *It has
no effect on speed suggestions or anything else related to MacCready's
theory.

So what you probably want is a third polar degradation setting:

1. "polar degradation" adjusts the configured polar to match your
* *real-world glider (persistent, new XCSoar 6.3 feature)

2. "bugs" adjusts the polar to match the dirty glider (not persistent,
* *has always been there)

3. "safety degradation" to replace "safety MC"; basically the same as
* *"safety MC" just with a name and definition that makes more sense.

I'm not 100% sure about this; these are just my unfinished thoughts
after weeks of reading XCSoar's glide solver code. *Maybe safety MC is
a setting to account for long distances of sink lacking any lift, and
the setting should be the expected sink value (and not the expected
lift value in the next thermal!).

Max

Max, I understand your points from pure theoretical perspective, but I
believe in practical most pilots use STF varios for speed, not PDA
flight computers, since STF varios are already doing this very well
and have audio signal as well. This is why the decoupling was
important. I also believe most pilots prefer to use MC for safety
degradation instead of polar degradation, or a combination of both.
But I guess one can just use the MC in XCSoar instead of safety MC.
Not sure about expected sink value, since this should be average sink,
and as such hard to predict or estimate (sink always seem stronger
than it really is, since we fly faster and see the combined result of
our increased sink rate and the neto sink).
Anyway, I think the most important thing here is the ability to
decouple the vario MC setting from the flight computer, or
alternatively use polar degradation. One thing is clear, using the
same MC for STF and glide calculation is a sure way to arrive short
without additional polar degradation.

Ramy

On Tuesday, February 28, 2012 2:45:58 PM UTC-8, Ramy wrote:
On Feb 27, 7:15*pm, John Cochrane
wrote:
I wrote a new article on how to use computers to help judge glides.

http://faculty.chicagobooth.edu/john...ocs/safety_gli....

or the first item here

http://faculty.chicagobooth.edu/john...ndex.htm#maccr....

This will probably end up in Soaring sooner or later, but I always get
a lot of help from early readers. If it's not clear or you see
problems etc. let me know. (john dot cochrane at chicagobooth dot
edu0

John Cochrane

Finally! I've been arguing this with flight computer developers
(Winpilot, XCSoar) for years. STF MC, and safety MC are 2 different
things that needs to be decoupled. Use your STF Vario MC setting for
speed to fly, and keep a constant MC for safety (I use 4 which seem to
work for almost any situation) in your glide computer (i.e. PDA).
Problem is, that if you connet your 302 to winpilot/xcsoar (and
probably others) you could not decouple the two. The good news, is
that XCSoar 6.3 will give the option to decouple the two different MC
settings.

There is another solution though. Instead of setting a high safety MC
in your glide computer, you can degrade the polar using the bug factor
to achieve the same results. (typically 33% -50% degradation, depend
how aggressive you want to be). Probelm is, that some flight
computers, such as XCSoar, did not store this value, which means you
had to remember to set it before every flight. The good news is that
this is also addressed in 6.3, which will have persistent polar
degradation.

And last, this excellent article also demonstrate why just keeping a
safety altitude does not work, as it will be appropriate only for one
distance. The further you are the highest it would need to be. This is
why Safety MC, or polar degradation are better solutions, since they
are not depending on distance.

Ramy

So, assuming an undegraded polar, if the last thermal is 4 knots and your Safety MC is 4, at what speed do you go home and where is the safety?

David

Thanks John.

I tend to think about it slightly differently. On the one hand I think about worst-case, inescapable sink for the conditions as measured by minutes times sink rate (result - a fixed number of feet that I am at risk of losing - I don't generally consider a string of sink occurrences - I assume one low probability sink patch is worse than multiple, higher probability sink patches and that multiple low probability events aren't likely enough to worry about). On the other hand I think about the probability of being able to find lift to recover after hitting a sink patch - which is a function of altitude above ground.

On very short final glides the constraint is the rate of sink (not much glide time left), on longer final glides the constraint is the probability of finding decent lift to get back up. The pinch point for me tends to be around 25 miles out - chances of finding lift are declining, odds of a long of stretch of sink still decent. This seems roughly consistent with your square root rule, though the math is different.

Your square root rule breaks down for me on very long final glides where I tend to optimize more around trying to transition away from climb and glide to cruise-climb in an effort to avoid thermal centering losses. This translates to an altitude buffer of maybe -1000 feet on a 50 mile "final glide" that you are trying to "bump-up" to +1000 feet by the time you get to 25 miles out. Not sure if/how that figures in your analysis.

9B

This article of John's ranks in my top ten for competition pilots.
Written with enough technical mumbo jumbo to be believable yet enough
plain talk to be understandable. I don't have a speed to fly Vario.
I am one if the cheepies that uses the PNA for speed to fly. This
article makes me want a speed to fly vario so I can decouple the PNA
from elevator.

Thanks John,
Lane
XF

Thought provoking article but I feel that Nine Bravo's comments are most
interesting too. It is my belief that final gliding is far more of an art
than a science. So many variables that theory is only useful to inform some
rules of thumb. Most of the decision making has to based upon circumstance
and experience.

Jim


At 15:36 01 March 2012, Nine Bravo wrote:
Thanks John.

I tend to think about it slightly differently. On the one hand I think
abou=
t worst-case, inescapable sink for the conditions as measured by minutes
ti=
mes sink rate (result - a fixed number of feet that I am at risk of
losing
=
- I don't generally consider a string of sink occurrences - I assume one
lo=
w probability sink patch is worse than multiple, higher probability sink
pa=
tches and that multiple low probability events aren't likely enough to
worr=
y about). On the other hand I think about the probability of being able
to
=
find lift to recover after hitting a sink patch - which is a function of
al=
titude above ground.

On very short final glides the constraint is the rate of sink (not much
gli=
de time left), on longer final glides the constraint is the probability
of
=
finding decent lift to get back up. The pinch point for me tends to be
arou=
nd 25 miles out - chances of finding lift are declining, odds of a long
of
=
stretch of sink still decent. This seems roughly consistent with your
squar=
e root rule, though the math is different.

Your square root rule breaks down for me on very long final glides where
I
=
tend to optimize more around trying to transition away from climb and
glide=
to cruise-climb in an effort to avoid thermal centering losses. This
trans=
lates to an altitude buffer of maybe -1000 feet on a 50 mile "final
glide"
=
that you are trying to "bump-up" to +1000 feet by the time you get to 25
mi=
les out. Not sure if/how that figures in your analysis.

9B

On Mar 1, 9:36*am, Nine Bravo wrote:
Thanks John.

I tend to think about it slightly differently. On the one hand I think about worst-case, inescapable sink for the conditions as measured by minutes times sink rate (result - a fixed number of feet that I am at risk of losing - I don't generally consider a string of sink occurrences - I assume one low probability sink patch is worse than multiple, higher probability sink patches and that multiple low probability events aren't likely enough to worry about). On the other hand I think about the probability of being able to find lift to recover after hitting a sink patch - which is a function of altitude above ground.

On very short final glides the constraint is the rate of sink (not much glide time left), on longer final glides the constraint is the probability of finding decent lift to get back up. The pinch point for me tends to be around 25 miles out - chances of finding lift are declining, odds of a long of stretch of sink still decent. This seems roughly consistent with your square root rule, though the math is different.

Your square root rule breaks down for me on very long final glides where I tend to optimize more around trying to transition away from climb and glide to cruise-climb in an effort to avoid thermal centering losses. This translates to an altitude buffer of maybe -1000 feet on a 50 mile "final glide" that you are trying to "bump-up" to +1000 feet by the time you get to 25 miles out. Not sure if/how that figures in your analysis.

9B

Good thoughts. But just to clarify, the article is not about final
glides -- how to do it efficiently. The article is about safety
margins -- how to do it with x percent chance of landing out.

John Cochrane