CAPS testing (long)

On Tuesday (3/26) and Wednesday, I was at the Cirrus plant in Duluth, where I was invited as a representative of COPA to observe and participate in a series of CAPS tests.
By now many of you have read this Letter from Alan Klapmeier (link available to members); so this report is primarily to fill you in on my perspective on the testing, and other topics not covered in the letter. This is a somewhat abbreviated version of the report I’ve posted on the Members Forum.
The first two tests, last night, were exactly as described in the letter. Prior to the testing, I was invited to talk to anyone about anything, and examine anything at all, related to the test. It was abundantly clear that every effort had been made to make this as “real” a test as possible – N109CD was a fully functioning demonstrator airplane, chosen in part precisely because it was an early SR20, with the same two CAPS SBs as we have all had done. The only changes made to the airplane were…

  • One door was removed, so that the cameras could see the action inside;
  • Fuel was drained;
  • The airplane was tied down in an extremely secure manner – this to prevent any possible dragging by the parachute should it catch the wind;
  • Air gaps in the aft bulkhead (between the baggage compartment and the CAPS compartment) were sealed with tape. This was done to minimize the amount of smoke in the airplane when the rocket fires. When the CAPS system is used in flight, this is not a concern – the smoke is blown away by the slipstream – but it’s pretty noxious stuff to have to breath on the ground.
    When it was time for the test, the ramp in front of the building was cleared of personnel except for Cindy Broman and Paul Johnston (Cirrus Chief Engineer), who were both in the airplane; the video crew; and a handful of other people essential to the operation. Alan and Dale Klapmeier, representatives from BRS, Cirrus sales and engineering, and I, watched from an upstairs window.
    The first two tests failed; the first because of a failed component (found only in the first 14 airplanes; expect an SB on this). The second because Cindy, who weighs 85 lbs, could not exert enough pull on the handle. Alan’s comment was that it wasn’t good enough.
    The failing light dictated that testing for the day must end.
    Many of the staff worked through the night (although probably not solely on this testing). When I arrived at the plant at 6:30 a.m. on Wednesday, the parking lot was full. By now, N109CD’s handle had been fixed using the current bracket. It did not take long before we were ready to test again. This time, when Cindy pulled the handle, the rocket fired correctly. The “cap” covering the CAPS compartment blew off cleanly, and settled a few yards away. I was surprised at how far the forward risers pulled out of the skin - almost to the wing root on the left, and halfway across the wing on the right.
    Ian Bentley said that the next test would be on an airplane from the production line. While that airplane was moved into position, Alan asked if I would like to do the next test - an offer which I immediately (and gratefully) accepted. I went down to Flight Ops, where I met Cindy, and asked her about the test. She said it was “very easy indeed”. While I donned the Nomex flight suit and crash helmet (obvious precautions), Paul Johnston briefed me on the logistics - how we would get signaled that the airport was shut down for the test, that the wind was OK, etc.
    I went out to the airplane. It was an SR22 with tail number N46X, which I later discovered means not much - it’s a temporary tail number used for airplanes as they shuttle to and from the paint shop in Hibbing. Only one other modification was done to this airplane – the forward risers were disconnected, to save the extra damage to the fuselage.
    I got into the pilot seat, and Paul Johnston occupied the right seat.
    When Paul told me to pull, I removed the safety pin, and “popped” the handle out of it’s channel. This was the first (for me) surprise. This is definitely a two-step process.
    Step one is getting the handle to dangle on 4-6 inches of slack cable. This is done by an easy pull of the handle forwards, to release it from its tube - it’s held in there by a rubber O-ring.
    Step two is getting the “T” cross bar aligned fore-and-aft instead of left-to-right, clasping it with both hands, and pulling down. It felt as though it moved 2-4 inches. It was as though the other end of the cable was connected to a corkscrew, and I pulled the cork out of a bottle; except that after the cork was out, the cable stopped moving. Engineers told me that it was probably about a 60 lb effort, but I have to say that it felt like a LOT less, and that was without the benefit of Adrenalin.
    It took a fraction of a second (1/4? 1/2?) for the rocket to fire. It sounded like fairly loud white noise – like water rushing out of a fire hose. I’m pretty sure that it would not be heard above the sound of a running engine, and probably not even above the sound of a “quiet” (failed or shut down) engine, if the airplane were in flight. The wind was calm, and there was no jerk on the airplane, nor any other sensation; although the cockpit did fill with smoke, which poured through the open door from the outside. I breathed through the mask which had been provided, and Jeff opened the passenger door. Within a few seconds, the smoke cleared.
    Aside from now having a TON more confidence that the system works (after the literal and figurative kinks are gone), I feel really good knowing what to do if I ever really need to pull that handle. Actually doing it is a HUGE advantage that every Cirrus owner should try to experience.
    After the tests, Alan showed me the CAPS demonstrator that is nearing completion, and which will be at Sun-'n-Fun. It’s clear that significant resource - and time - has gone into its development. It consists of a fuselage from just forward of the empenage to the firewall, with a pretty complete cockpit. A special version of the CAPS release mechanism has been developed - the production version is unsuitable, because it’s essentially a “one-time” mechanism that requires rebuilding after activation. The demonstrator version is designed to automatically reset, and to faithfully reproduce the expected worst-case loads and “feel” of the real thing. If you were thinking of going to Sun-'n-Fun, but weren’t quite decided… I feel that the two minutes you’d spend actually pulling that handle are worth the trip. Those who will attend the CPPP (Cirrus Pilot Proficiency Program) in Dublin, GA, might get the opportunity to use the simulator there – we are working with Cirrus to bring it there, provided that we can clear a few logistical hurdles.
    I’d like to take this opportunity to thank the company somewhat publicly for their prompt, decisive, effective actions in their dogged pursuit of whatever issues there are with CAPS, and in fact any/all safety aspects of the airplanes. Alan’s comments, time and again, reflect that Cirrus is no more satisfied with an imperfect CAPS system than any customer is. At one point he said that it doesn’t matter WHY someone chooses to use the system - it could be that they simply feel that they no longer have sufficient confidence to continue a flight; whatever their reason, it’s an emergency to them, and if they pull that handle, CAPS must work. Those comments were repeated in spirit, if not in exact wording, by Dale, Ian Bentley, and many others. Thank you, Cirrus.

Finally…

I have spent most of today at Cirrus, then in transit from Duluth back home to New Jersey; I have a ton of emails and voicemails that I have yet to look at and listen to. If yours is among them, and you are asking questions about the CAPS testing, please contact me again (preferably by email) if you need more information.

One of the many photos I have is attached. (A different one is on the Members Forum)

Mike.

4 Likes

Thankyou, Mike! Excellent report, and really usefull.

Timm Preusser N747TG

Thanks Mike! Ignore my question via e-mail of a few days ago. What a great Company!

I share the point of view that the people at CD have acted in a way that took the high road, and have clearly earned the continued confidence of their constituents.

One further observation - reading Alan’s letter and Mike’s. If anything, Alan was more “conservative” or “reserved” than Mike. Point being that with all that is going on, Alan refrained from commenting on the positive, and merely put forth the sobering facts. Mike, of course, was in a better position to put forth his interpretations.

Proud to be associated with Cirrus; proud to be a member of COPA. Alan and Mike should be congratulated for doing their jobs in an exemplary fashion!

Andy

I fail to see the enthusiasm. Based on this test you only have one chance in 3 that the chute will work. I am sure that eventually somebody will figure out how to fix the problem but in the meantime we are flying around in a plane with a high incidence of conponent failure, a higher than normal accident rate, and the most promenant safety feature of the plane is more likely to fail than to work.

I fully agree with the actions and taken by Cirrus so far and would like to thank Cirrus and Mike for the information on this subject.
This does however not imply that it is acceptable that handles break / bend or rockets fail to ignite when fired with over 100lbs force.
After 300+ airplane have been delivered to customers, who trust that there is an additional last safety option when all other options are exhausted in an in flight emergency, the CAPS should work the first attempt.
Engineers know that every modification to a system requires retesting of the essential functionality to predetermined criteria and this seems not be the case for the CAPS.
Why was the handle changed after 14 planes? Was it considered too weak?
It is my opinion, that the CAPS system should be fully retested, in-flight, to ensure that the chute does not only fire but that it will deploy and that the plane will not drop from the cables because of some “seemingly unrelated” design or production modification to the cable routing or structure.

PS
I also would like to take this opportunity to express my appreciation for all the COPA work and the valuable information I receive from all of the participants by reading all the interesting subjects. Keep up the good work. Waiting for #527Â……(long)

This group (cirrus) has INTEGRITY.
Judging from their exemplary customer service, exemplary corporate behaviour, and up-front “has nothing to hide” attitude I will have no second thought about buying and flying the next upgraded product that they may come up with.
I TRUST them.

Moses Grad

In reply to:


Based on this test you only have one chance in 3 that the chute will work.


I’m just curious about the math – how did you arrive at this?

Mike.

Regarding…

In reply to:


I fail to see the enthusiasm. Based on this test you only have one chance in 3 that the chute will work. I am sure that eventually somebody will figure out how to fix the problem but in the meantime we are flying around in a plane with a high incidence of conponent failure, a higher than normal accident rate, and the most promenant safety feature of the plane is more likely to fail than to work.


I must take issue with this analysis and opinion.

(1) Cirrus has already identified and will soon fix the bracket problem on the 14 early-production aircraft effected.

(2) After the bracket was replaced on N109CDR with the newer version, the system worked (with reasonable pull force).

(3) Cirrus has committed to resolving the pull-force problem encountered in N174CD

(4) The current production-line aircraft systems worked as expected.

Taking the most worst-case outlook, the success-rate (with the new bracket) was 2/3 - not 1/3. Also, the pull-force problem in N174CD might well have been overcome in a real emergency by adrenaline, making it 3/3 :slight_smile:

The bottom line is that Cirrus took a few days to line up some engineering issues and then did serious testing. Further, they have been forthright with their test results, have clearly committed to making the systems better, and are arranging for pilot education with non-trivial simulators. This is pretty damn good for a small start up company.

Conpil,

You wrote that “the most promenant safety feature of the plane is more likely to fail than to work”. I was always taught that the most important (don’t know what “promenant” means) safety feature on any airplane is the PILOT.

Now that I’ve had an opportunity to read a lot of your posts, I have little doubt that the most IMPORTANT safety feature on your airplane is, in fact, unreliable!

:slight_smile:

S.F.

1st test—failed
2nd test—failed
4th test—passed

3rd test doesn’t count. You don’t get a chance to do repairs in the air in a real life situation.

If you add the apparent experience in Lexington then the odds are one in four.

I am not trying to be negative, just analyzing the results as I interpret them.

There were 3 attempts, 2 failed on the first attempt. 1 in 3. If you throw in the last accident that is really 1 in 4. If you include the one last year that becomes 1 in 5.

In reply to:


1st test—failed2nd test—failed4th test—passed3rd test doesn’t count…just analyzing the results as I interpret them.


But the third test absolutely does count for all but the first 14 production models. The first test only applies to production models 1-14. Thus, at worst, fail/pass/pass. The Lexington problem’s cause is not yet determined (system vs. use?)

In reply to:


3rd test doesn’t count. You don’t get a chance to do repairs in the air in a real life situation.


Michael,

Agree somewhat. I was editing my previous post when you posted this one, and so I added this:

As I edit this, I just caught your post that the 3rd test doesn’t count. Perhaps you’re right - my take was that it was very important to confirm that the repair worked. Also, it is possible that the 'chute could have been deployed even in that test. The bracket that failed holds the entire handle assembly to the “ceiling”; it seemed that it could still be fired holding the assembly with one hand and pulling the T-handle with the other. They elected to stop that test at that point - the attitude seemed to be “not working properly is not good enough”.

Everyone was disappointed by the failure, but was happy to have found - and have the opportunity to correct - a flaw.

I said I “agree somewhat”. Basically, while I’m certainly no statistician, I don’t really see the difference between testing a DIFFERENT airplane after a failed test, and testing one where a known failure point was corrected. I guess in real life, you wouldn’t get a chance to do that, either.

On the other hand, testing the fix also provides some data that tells me that everything ELSE in the chain was working. Otherwise, I might wonder if there’s yet another link that might fail.

My .02…

Mike.

I agree. Test three certainly counts. The work that was done allowed the test to be performed with the same bracket as is installed in all production models after #14. That work also confirmed the need to modify the first 14.

The one last year? are you serious? CFIT accident. The only thing that would have saved those poor souls is to not have taken off in the first place. As for the Lex. accident, we still don’t know if the forces applied to the handle were adequate and or in the right direction. MIke

In reply to:


There were 3 attempts, 2 failed on the first attempt. 1 in 3.


Art,

Strictly to keep the facts straight: There were four attempts. Two failed, two succeeded.

Mike.

The one last year? are you serious? CFIT accident.

A CFIT after a failed deployment, that had it performed as the pilot had every reason to suspect it would, might have saved them. According to BRS it won’t deploy in an accident unless it is armed by the pilot and it was found deployed. Some say it must have deployed as a result of the fire. I say that the fire result when the chute rocket ignited the fuel after the crash. Neither position can be proven, but neither can be discounted. You may not like the idea of fatalties being caused by the chute but it is a possibility that must be considered.

As for the Lex. accident, we still don’t know if the forces applied to the handle were adequate and or in the right direction.<<

The accident investigators stated that I had pulled in the proper direction and with forces estimated in excess of 100 lbs. They based their conclusion on the direction and distance that the mounting bracket was bent and several simulated pull tests on my airplane both before and after the cable rerouting.

…I agree. Test three certainly counts. The work that was done allowed the test to be performed with the same bracket as is installed in all production models after #14. That work also confirmed the need to modify the first 14…

I am trying not to be argumentative. I now see how people get sucked into these tiresome long “he said/she said” threads.

The first plane had a significant force applied to the cable. Enough force to break the bracket. The bracket was then replaced and the handle and cable re-stowed. On a subsequent test the CAPS deployed correctly.

On the second test excessive force was required and a working hypothesis from CD has to do with cable routing, memory, etc.

Having attempted to deploy by exerting significant tension on the cable of plane #1 then pulling the cable a day later demonstrates that the rocket does indeed fire when the igniter is activated but does not tell you that it will fire unless you pull your cable with some unknown tensile force less than 85 lbs then restow it and try again tomorrow. In other words the attempt to pull it the first time MAY have set it up to work the second time 12 hours later. Had the bracket not broken, it is reasonable to assume the chute might not have deployed, since there had already been enough force applied to break the bracket.