USPA BOD... Nothing more than a mouth piece for manufacturers

[mark 102]

billvon

> How sure would you like to be that your test drops would detect a problem?

Relatively sure.

I'm looking for a statistical confidence level. 95%? 99%?

Mark

[billvon 2,400]

>I'm looking for a statistical confidence level. 95%?

Probably 95%.

[mark 102]

billvon

>I'm looking for a statistical confidence level. 95%?

Probably 95%.

Ok. Check my math, please: 30 tests have a 95% chance of detecting a problem if the underlying failure rate is at least 1 in 500.

Now let's assume that we get a failure in that test series, duplicating the problem that led to the testing. What should be the course of action? Should we:

-- continue to jump rigs with the same configuration until we can isolate the problem and identify a solution?

-- ground other containers of the same make, model, and size and with similar-sized reserves until that combination has been adequately (whatever that is) tested?

-- ground that particular reserve make, model, and size in other containers, until it has been adequately tested in other containers?

-- something else?

Mark

[billvon 2,400]

>Now let's assume that we get a failure in that test series, duplicating the problem
>that led to the testing. What should be the course of action?

-Identify the problem
-If the problem is solvable via a simple fix (change to packing directions, a simple field-installable mod) then solve the problem that way
-If the problem requires a more systematic change then ground* the risky combination of gear (say a Mirage MT with an Optimum 143) until the change is made
-If no change can be made then ground* that combination of gear


(* - of course all any independent effort could do was recommend a grounding; it would have no force of law)

[mark 102]

billvon

Identify the problem

That's the rub, isn't it.

The 30-drop test series may be enough to identify that there is probably a problem, but may not be enough to determine exactly what the problem is nor how widespread it might be.

Let's imagine I have a low reserve opening after a total malfunction on my Curv/Optimum-143/Katana-120, and that I get a similar opening in the retest series using exactly that rig. Should I test with bigger or smaller mains?

It might be a problem with that particular rig. Should I test with the same canopies in another production model rig the same size?

It might be a problem with that particular reserve make/model/size. Should I test with a low-bulk Icarus reserve of the same pack volume?

I think my Curv reserve packs about as tightly as my Mirage. In case the problem is with the reserve canopy and not the container, should I test with the Mirage as well?

Mark

[Southern_Man 0]

mark

***>I'm looking for a statistical confidence level. 95%?

Probably 95%.

Ok. Check my math, please: 30 tests have a 95% chance of detecting a problem if the underlying failure rate is at least 1 in 500.

Mark

It's been a while since I've done statistics, so I could be remembering something wrong, but unless the expected n>= 5 then you are not going to be able to generate a confidence interval with any accuracy (I think I remember that right).

You have to quadruple your sample to halve the confidence interval.

"What if there were no hypothetical questions?"

[Southern_Man 0]

masterrigger1

Quote

In that case we are almost certainly in compliance now. Very few jumpers have experienced longer delays in reserve openings. (We, of course, only hear about the ones who DO experience longer delays, especially when that delay results in a fatality.)

Not quite!
You really need to look at some recent cutaway videos. I think you might just get a wakeup call. There are several that are in the 4-6 second range and some even a little longer.

MEL

Hi MEL,

can you post a few of those videos?

Thanks,

Ken

"What if there were no hypothetical questions?"

[billvon 2,400]

>The 30-drop test series may be enough to identify that there is probably a problem,
>but may not be enough to determine exactly what the problem is nor how widespread
>it might be.

Right. But if your instrumentation is good (primarily good video recording) you have a very good shot at understanding the problem as soon as you see it.

Is the closing loop not releasing? Is the reserve container opening but retaining the D-bag? Is a hesitator loop doing too good a job? Most of that can be determined by visual inspection of a (good) video.

>It might be a problem with that particular reserve make/model/size.

It might. If, for example, the problem is due to tightness, then it will likely be seen only in reserves that are very tight (as in the Mirage/Optimum example from previously.) Thus "ban" that combination and you have likely solved the problem.

[masterrigger1 2]

Ken,

Here's is one that falls in the 4-6 second category.

http://www.deepseed.com/d-spot/blog/liam/inside-no-pull-cypres-save-0

Look for the shudder of when the P/C is released and also when his feet rotates which is about 1-2 seconds into it.

You can find them on YouTube. You can also find some of them on here.

Simply type in parachute malfunction and sort through them with a stop watch.

MEL

Skyworks Parachute Service, LLC
www.Skyworksparachuteservice.com

[nigel99 143]

mark

***>I'm looking for a statistical confidence level. 95%?

Probably 95%.

Ok. Check my math, please: 30 tests have a 95% chance of detecting a problem if the underlying failure rate is at least 1 in 500.

Now let's assume that we get a failure in that test series, duplicating the problem that led to the testing. What should be the course of action? Should we:

-- continue to jump rigs with the same configuration until we can isolate the problem and identify a solution?

-- ground other containers of the same make, model, and size and with similar-sized reserves until that combination has been adequately (whatever that is) tested?

-- ground that particular reserve make, model, and size in other containers, until it has been adequately tested in other containers?

-- something else?

Mark

I think your maths is right. As Bill said in test conditions it should be relatively easy to identify the issue.

I don't think you'd need to ground the rigs unless it was a very high failure rate. But considering people's reactions to AAD failures in the past, it would be unlikely the product combination would be accepted.

Experienced jumper - someone who has made mistakes more often than I have and lived.

[mark 102]

billvon

if your instrumentation is good (primarily good video recording) you have a very good shot at understanding the problem as soon as you see it.

If it were up to you, would you do another 30 test drops to verify you'd identified and corrected the problem?

Mark

[billvon 2,400]

>Look for the shudder of when the P/C is released and also when his feet rotates which
>is about 1-2 seconds into it.

I didn't see any shudder although I did see a lot of motion (which would be understandable for someone who cut away at 800 feet.)

He said he cut away about 800 feet. If it was his AAD that saved him it's going to take an AAD 3-4 seconds to reach activation speed from a cutaway at that altitude; he was lucky it fired at all. That means he's getting an opening at about 400 feet, and getting canopy opening right about at ground level.

[billvon 2,400]

>If it were up to you, would you do another 30 test drops to verify you'd identified and
>corrected the problem?

If it were up to me and time/money were no object - yes.

[masterrigger1 2]

Quote

I didn't see any shudder although I did see a lot of motion (which would be understandable for someone who cut away at 800 feet.)

You probably need to look at several times.

He is probably near terminal again if not at terminal when the AAD fired.

Just going from the rotation of his feet is 4.3 seconds to impact.

The early video on the same page has about a 4.0 flat second deployment.


MEL

Skyworks Parachute Service, LLC
www.Skyworksparachuteservice.com

[billvon 2,400]

>The early video on the same page has about a 4.0 flat second deployment.

We must have different stopwatches then. I got about 3 seconds.

[masterrigger1 2]

I guess it depends on when you start timing. I am going from handle pull.

MEL

Skyworks Parachute Service, LLC
www.Skyworksparachuteservice.com

[Southern_Man 0]

masterrigger1

I guess it depends on when you start timing. I am going from handle pull.

MEL

From cutaway handle pull? He did not pull his reserve. (oh, sorry you are looking at the other video on this one)

Anyway, I agree with what Jerry said earlier in the thread--time of deployment from an AAD fire is not a TSO standard.

"What if there were no hypothetical questions?"

[nigel99 143]

Southern_Man

***I guess it depends on when you start timing. I am going from handle pull.

MEL

From cutaway handle pull? He did not pull his reserve.

Anyway, I agree with what Jerry said earlier in the thread--time of deployment from an AAD fire is not a TSO standard.

While what Jerry is said is technically, it sure would raise a few issues if manufacturers started blaming the AAD for slow deployments. For a start I believe AAD's are required to not affect the TSO of a rig.

So for example we say that pulling your reserve handle will get you a reserve in 300ft, but the AAD cutting the loop takes 1000ft. That means that effectively the AAD HAS affected the TSO of the rig. Very dangerous 'excuses' to get into and I'm sure it is not the case anyway.

Regardless we have all got sidetracked on detail from the core of Ron's original post. That is the fundamental issue that the USPA should be representing the members needs and views. A major benefit of being part of an organisation is that the collective voice is more powerful than an individuals. So the USPA should be there to represent us to the FAA and the manufacturers.

All the evidence points to them doing a very poor job and that for whatever political or financial reasons the BOD is giving far too much weight to manufacturers wishes. Not necessarily in this issue but in a number of issues.

Experienced jumper - someone who has made mistakes more often than I have and lived.

[yoink 321]

Right on.

I was just going to ask how we proceed from here... is it to lobby the USPA and what specifically for?


Lots of good thinking in this thread, and it'd be a shame to see it just dissapear.

[Ron 7]

Quote

How would they do that? Ban rigs that are not retested? (I can just see Ron's reaction to _that_!)

What exactly would be my reaction to that?

"No free man shall ever be debarred the use of arms." -- Thomas Jefferson, Thomas Jefferson Papers, 334

[JerryBaumchen 1,048]

Hi Ken,

I've been away for a few days but this discussion is a very interesting one. And I hope that it has brought some knowledge to a lot of us on this matter.

Quote

You have to quadruple your sample to halve the confidence interval.

I was on the TSO committee when it was suggested that the Functional Tests be increased from 12 ( TSO C23b ) to 48 ( all subsequent versions ).

The suggestion also had ( IMO ) some mumbo-jumbo math to support the increase in Functional Testing. It was back in the 70's and I no longer have the data.

Now, it has been over 40 yrs since I took 'Statistics for Engineers' and I have been retired for over 10 yrs, so my statistics could be better.

IMO to get a decent increase in probability of opening within the 3 second period ( remember, this is the issue ), you would have to go from 12 tests to about 1200 tests. Now you are getting some significant statistical data. By going from 12 tests to 48 tests, we have only increased the costs of obtaining a TSO certification. The increase in probablility is insignificant.

Just a thought or two,

JerryBaumchen

PS) If there is a sharp, current statistician out there, please provide us with your thoughts. I am more than OK with being corrected.

[JerryBaumchen 1,048]

Hi Peter,

Quote

Good stuff, delving into the details of what our parachutes really are supposed to be able to handle.

Absolutely. By having a good, civil discussion on issues ( and numerous sides bringing their thoughts/experience to the table ) we all gain knowledge.

And that has to be a good thing.

JerryBaumchen

[Ron 7]

Quote

I was just going to ask how we proceed from here... is it to lobby the USPA and what specifically for?

Call the members on the BOD. Tell them that you are upset that they seem to be caving to the demands of an equipment manufacturer.

I have called 2, PM'ed one, facebooked two more, and talked in person to one of them. Further, I have an appointment to talk to one more next mth at the nationals.

Long story short.... Most think this was the only way to save lives. But the fact they seem so eager to change the BSR's but have close to ZERO inclination to demand better performance from the manufacturers is the point of this whole thread.

The only way they are going to start listening to the members is if the members TELL them what they want. In this case the only voice they seem to hear (or care about if you want) is the manufacturers. Change that.

Then if after you have talked to them and if you don't like their answer.... Vote them out.

Maybe, just maybe, we can get the USPA to support jumpers again.

"No free man shall ever be debarred the use of arms." -- Thomas Jefferson, Thomas Jefferson Papers, 334

[ChrisD 0]

You guys are talking about the mean, median , and mode and the standard deviation from the mean.

But what's the context?

If you are just measuring something fixed like opening time then if the std dev is within an alpha value that you set, what are you going to set it at and why.

More tests are not necessary if the std dev is close.

If the std dev is large then more tests might make some sense. Your just approaching an average at that point.

So much of numbers is how you define the parameters of what you are looking at, be wary of approaching something like opening time with words like confidence levels and such because you start to get confused with experiments where we manipulate some independant variable, The N depending upon effect size and the alpha level due to the possibility of chance then have some meaning.

Never forget things like effect size. If I was to manipulate something like some experiment where I drop a hammer, with considerable force on your foot. The level would approach 1. No need to run the trials much more than once. On the other hand If I was to drop a feather on your foot, large numbers N might be nessasary to tease out the effects of chance, the results then would depend upon what a level we set and the results may or not be signifigant. Signifigance at which level you set is the term I think some of you are getting confused with...

If your just measuring some value then the std dev is what is signifigant. Not much math really... the std determines the N in this case. The real question is how are people measuring the time? With a stop watch? This is the real problem and is the reason behind the scenes that no official body really wants to get into this....

It is not possible to get any actionable information from a film and some one with a stop watch watching said film.

Unless of course the numbers started varying by large amounts, say ten seconds one trial and 15 seconds the next. But when every test opening only varies by hundriths not much is too be gained.

You guys are also forgetting the margin for error with these devices. Think about the speeds at 800 feet, the time left is the time needed to open! Which is of course what is driving this whole thread....

cause there ant much time left to steer or decelerate!!!!

Depends upon the canopy!!!! Not the opening altitude.
C

But what do I know, "I only have one tandem jump."

[excaza 1]

ChrisD

The real question is how are people measuring the time? With a stop watch? This is the real problem and is the reason behind the scenes that no official body really wants to get into this....

Opening times can be determined using a pressure altimeter and video to corroborate. This is done routinely and is not a difficult problem. The difficult problem, as yourself and others have already pointed out, is defining a reasonable scope for the study.