OT WWII Memoirs (was SAFE Winch Launching ...)

Back to metrics - if you are measuring the wrong thing, the best you can
hope for is that it is irrelevant...

So - for a "good" winch, there are a couple of things you may want to
measure.

How much does it cost?
- Capital to purchase (and how long it lasts)
- Operating cost to service and run
- Repair cost when something breaks (including damage to the gliding fleet)
- How fast can it launch relative to requirement

Is it safe?
- How easy is it to control the acceleration?
- How easy is it to achieve the correct speed in all phases of the launch?
- How well does the set up accommodate for unexpected events (big
thermal, windshear, pilot error)

So far my experience and inclination tend to support the theory that the
key requirements to satisfy the above favour simplicity and human
control. I am a computer / automation professional, but there are some
things I would prefer a human doing. Driving and controlling the winch
is one of those. There are too many variables to concern yourself with.

As an exercise - Let's test a theory. (in the empirical sciences the
difference between a theory and an idea being that a theory can be
disproved by observation and deduction).

Lets say that my theory is that relying on a single metric for control
of a winch - specifically cable tension as measured anywhere on the
cable or drum is safe.


OK - can we disprove that? -

Let's say for argument that the pilot fails to rotate appropriately. For
any number of reasons. Constant tension must be applied in this model,
so energy transfer continues to be done and the glider speed increases
as there is no increase in potential energy.
After a very short period the glider is past Vw and the pilot dare not
rotate. If the pilot is incapable of, or is slow at releasing (for
whatever reason) the speed will continue to increase, as the available
landing area is consumed at an ever increasing rate. Clearly this is
not a safe situation.

Let's argue a different approach - Now let's say for argument that the
belief that the control over tension in the cable is so good that we
have disregarded the weak link specification and are relying on the
ability of the system to regulate tension. Now we follow a launch where
the cable predictably hooks on some grass causing a bow in the run. At
the all out the grass resists the considerable lateral pressure
momentarily, the winch achieves designed tension. Then the rope/wire
breaks free of the grass and the bow becomes slack cable. In attempting
to maintain tension the winch accelerates, removing the slack. At some
point in time the cable becomes straight, all slack has been absorbed
and the cable tension rapidly increases past the design tension.
Consider the reaction time of the system, given the rotational inertia
of the drum and transmission, to say nothing of the whip action on the
cable as the wave travels toward the aircraft. Once again I fail to see
how this is going to be safe.

In this case the judgement available in a manual control system
increases safety - you can't rely on automation to magically create
safety.

So what can we deduce?

Basically you are trying to measure how well a complex and potentially
dangerous operation is progressing. In a constantly changing medium,
with multiple sources of error and variance (not least caused by the
multiple human inputs) I know it is very tempting to believe that there
is one thing you have to worry about and it will all work out perfectly.
Regrettably this is a little like perfect landings. Rumour has it there
are only three things you have to get right. Problem is no-one can agree
on which three things...

In a winch launch one needs to decide on what to do based on aircraft
speed, height, position relative to the available runways, terrain and
traffic. Always considering the capabilities of the aircraft, pilot,
winch driver and not least the prevailing weather.

Clearly - Relying on one metric is dangerous oversimplification.
Now a tension limited winch may be a good thing, and a winch that can
smooth the tension in the cable to avoid spikes would be desirable.
And certainly I have seen some scary old manual winches. Funny thing is
- as long as you have people who communicate about what the performance
and experience is, and who learn and adapt to the vagaries of whatever
winch they are using. It tends to work. Usually surprisingly well.

So - in reality the technically complex winches use a lot of technology,
and then rely on the same judgement that a lot simpler and cheaper winch
uses.

To quote Hydrowinch
"The Hydrostart winch still runs on the original automatic force and
speed control software. A manual override is just one of the regulatory
safety features that have been there since the winch is in operation.
Because when it comes to safety, we can only rely on the winch driver's
judgement. The automatic control system is not running the winch. It's
there to assist the winch driver maintaining a constant quality and
launch performance."

Is this system better and safer - From an engineering perspective ,
definitely. Is is better value for money and safer than the person using
it? Unfortunately not.

Now - if people put as much effort into building and using the things as
is getting put into this debate - we would be doing well.

So far - lots of attempts to substitute technology have been
unsuccessful at displacing simple , robust semi manual systems with
proven reliability and controlability. There may be a prefect winch
design out there, and so far the evidence indicates that this is
something less than over engineered.

Again I submit - the metrics that matter are safety and economy.
For safety you need reliability, and controlability.

For economy you need simplicity, robustness and maximum use of "off the
shelf" components.

The only place I could possibly see justifying a technological super
winch, would be in a very high launch intensity 7 days a week operation.

If anyone can, and desires to make a better winch than any of the
commercial alternatives out there, profitably. Please do so - there are
customers who would love to make you commercially successful.

If you want to , and can afford to use/build something as far into the
overkill region as a Hydrostart, be grateful and enjoy the toys. Just
don't expect great economy/commercial success. As a technocrat - I would
love one. As a club member it would be a financial, maintenance and
operational disaster. Our best winch driver loves spending hours a day
getting the best out of our ancient single drum - but he can't use a
computer... Personally, I would love to be able to put a new "Skylaunch"
style winch at his disposal. It's the same reason the much maligned
Windows OS dominates desktops. It is not the "best" but it works for
most...

Bruce

Damn - I'm feeding trolls....

Don Johnstone wrote:
As I see it the "constant tension" theory relies on being able to measure
the tension being exerted on the glider release at the winch drum. Quite
how this might be achieved is very puzzling and has no relevance to the
information required to give a safe and effective launch.
If you were to say that measuring the tension at the glider release and
using telemetry to pass this information to the winch then that might
indeed work, however if you were going to the trouble of doing that you
might as well send useful information, like the airspeed of the glider, so
the winch driver could maintain a constant speed.
Cable tension during a winch launch has sod all to do with anything except
as an indicator to the winch driver of possible over or underspeed. It is
the speed which is of relevance and importance.

At 13:15 27 July 2009, Del C wrote:
The 'constant tension' theory of winch launching was dreamed up by
someone in the US who has no practical experience of winch launching
whatsoever!

So far nobody has managed to built a true tension winch (which would
measure actual cable tension), so we don't know if the theory would
work
or not. The concept seems to have become a bit of a Holy Grail in the
US,
which is probably inhibiting the design and building of more
conventional
winches that would work just fine.

On the Yahoo Winch Design site I have suggested carrying out some
autotowing experiments, where it would fairly easy to mount an in-line
load cell to find out if CT would work, but this suggestion was rejected
by the above person and his followers as not being relevant due to the
mass and inertia of the towcar. Such an experiment would work in calm
conditions.

There are a number of constraints in real life winch launching.

1) The minimum airspeed has to be at least 1.3 x the normal stalling
speed
(Vs), to avoid the risk of stalling or spinning at the increased wing
loading due to the cable pull. At the high levels of pull suggested this
might increase to 1.4 Vs.

2) The optimum climbing airspeed for best gain of height seems from
practical experience to be in the range 1.5-1.6Vs.

3) Most gliders have a fairly low maximum winch launching speed (Vw),
which is set for structural reasons. There should also be a weak link
(fuse) included in the cable line which will break before the glider
does.

4) Many gliders, particularly older ones such as the K13, only have a
very
limited speed range in which they will climb safely and well without
exceeding Vw. The stalling speed of a K13 can increase to over 50knots
near the top of the launch, its optimum climb speed is about 56knots and
its Vw is 58knots. Some more modern types such as the K21 are a bit more
speed tolerant.

5) You have to fly the glider in such a manner that you can always
recover from a cable break or winch power failure, and not risk a stall
or
flick spin. This entails a fairly shallow initial climb followed by a
controlled rotation rate of not more than 10 degrees per second. I
believe
the Germans once managed to kill 12 pilots in one year (1995) by
carrying
out what are known as 'kavalier starts' where the glider climbs very
steeply straight off the ground to maximise height. We have also had a
few
such accidents in the UK, always on very powerful winches so rapid
acceleration doesn't make them safe.

The theory behind constant tension is you provide a pull or tension that
is close to the breaking strain of the weak link. Thus you maximise the
pull and the height gain in accordance with the Goulthorpe formula:
h = P/W/(1+P/W) x l
where h = height, P= Pull, W = glider weight and l = notional cable run
from the point of rotation.
Thus for a Pull equal to the weight of the glider you would expect to
get
a height of 50% of the effective cable length.

However, the above equation is idealised and assumes zero cable weight
and
zero drag, and is based on 100% transfer of energy.

For many years I launched on very powerful manually driven Tost winches.
Many of the launches were way over Vw until you signalled too fast, but
it
was quite rare to break a weak link in the early part of the climb. I
therefore suspect that the constant tension as a large fraction of the
weak link strength idea would just vastly overspeed the launches. In
order
to contain the speed according to the theory, you would have to climb at
an
achieved climb angle of about 60 degrees. Most gliders run out of up
elevator well below this angle. Such an angle would also represent more
than a 'kavalier start' as described above!

The other idea in the 'constant tension' theory is that the glider
pilot
would control the speed by pulling back harder to slow the launch down
and
easing forward to speed up. However I worry that a pilot trying to
control
the speed at the same time as the winch is trying to sense and control
the
tension would just lead to an oscillating or hunting situation. As a
winch
driver myself, I always try to avoid 'chasing the glider pilot' as
this
generally makes things worse. If I have to make a speed adjustment I
just
move the throttle to a slightly different setting and then hold it still
again. The technique for controlling the airspeed from the glider end
does
work on a Skylaunch winch where you are giving a constant power setting
and
also works on constant torque Supacat (diesel + fluid flywheel))
winches.
With either type of winch you have to start backing off the throttle
setting near the top of the launch to avoid overspeeding the glider.

We don't know if constant tension would give a constant and appropriate
airspeed, or whether it would need to be varied for different stages of
the launch to achieve this.

Derek Copeland



At 19:45 26 July 2009, Don Johnstone wrote:
As interesting as it is the discussion about who did what in the last
war
has about as much relevance to gliding and safe winching as a tesion
controlled winch.

The differences are too numerous to mention except that a Spitfire,
Hurricane and Mustang all worked and did a useful job, unlike the
mythical
tension controlled winch.