Training New Hires 25

[HSThompson]

OK, This isn't a technical question but one that I am sure that we are struggling with. We, like everyone, are hiring new stressers everyday. I've been a stress engineer since '91. I was trained in the old Boeing way, find the grizzliest old stress guy you could, then shut-up and listen. In addition, at Boeing we had the Tech Excellence classes.

My problem is this, the OEM I’m at now does not have very good training courses, and there aren’t enough grizzlies to go around. We’re getting worn out explaining the same things over and over to a different person; i.e. IDT, Mc/I, Cripple, Buckle, Repairs, DADT, FEM, Load Paths, FAR’s, what references to buy etc. etc., etc.

My question is this, for all you old timers and newbies out there, what are some training methods you have used or seen that can get them up to speed and quickly? I can think of a few but a fresh perspective would be nice.

 

[crackman]

To answer a few questions, the liasion structures course that I spoke of is just that, a collection by chapter of progressively more difficult structures examples and problems. There are several of these courses that the OEMs used to put on about 20 or 30 years ago but alas no more. Lockheed used to teach a very good series of courses included a very famous one on Thin Sheet theory.
The course notes basically do not have an actual document number or title or any restrictions, it is more a compilation of course work and so has not proprietary stamps or references on it since it is all derived from textbook equations. Its the type of course companies and the government used to send engineers to in order to get trained on "real" world problems right after graduating. The information in it is basically what one would get from several structures textbooks but only in a simpler and more focused manner as applicable to aircraft liasion work.
Hope this clarifies some questions.

James

 

[Sparweb]

Crackman,

I'll also contact you for a copy of that training course. The test sounds... humbling.


Steven Fahey, CET

 

[Asanga]

Thank you for the notes Crackman. Even if its not as detailed as Bruhn, it gives quite an insight into the background of the engineering fundementals knowledge expected of a liaison engineer, or for that matter any engineer involved with aircraft structures. The course outline for the correspondance course that I want to take (Based on Flabel's book) follows a similar pattern to Macair's.
Cheers
Asanga

 

[aerodog]

Crackman,

I would be curious to hear your opinion on the use of CATIA/FEA vs. the old hand crank method. In particular, which is the more accurate predictor of the stress levels that get recorded during static test.

 

[crackman]

aerodog

Depending the type of test being performed (small component to full scale) both can make good and/or bad predictions based on how they are applied. Typically though, the most direct and best way in my own opinion for large scale tests is to use some sort of FEA BUT this means a normal industry standard (or that which used to be standard in our industry) "coarse" grid airframe model built by a well experienced stress analyst who knows load path very well following standard modelling practices. NOT, some ultra fine mesh or autogenerated model (dont even think about solid elements) with every minutia modeled as seems to be the practice these days. Another good alternate method for wing boxes and even fuselage structure (as long as you account for cutout redistributions) is to use a Unit Beam method (ie Cozzone method) which was used by most OEMS for decades. In fact, most OEMs designed and built all of the 1940s to late 1960s transports with this method in various forms. Boeing used it on the 707, 727, 737, 747 (and I have heard still does for prelim design purposes). The Boeing fuselage code was named TES057 while the wing was TES170 as I recall. From the old ultimate correlation reports I have seen they did a pretty good job for the most part in predicting the values.

One thing to keep in mind is that whenever you are correlating always pick very clean areas to instrument. That is to say, I have very rarely ever seen test correlations to strain gages in stress concentration areas provide good (within 10%) correlation. For instance, if you are trying to validate a wing coarse grid FEM, place your gages on basic general spar cap, stringer, or skin locations as nearly close to the NA as possible and away from any cutouts, splices or discontinuities. The best way to correlate areas with stress concentrations is to perform very localized tests and even then it is difficult to obtain very good correlation. The main point in any aircraft FEM is to correlate internal loads not necessarily detail stresses. This is typically what regulation agencies are looking for in order to gain confidence in model results.

James Burd
Avenger Aircraft and Services

 

[SuperStress]

Crackman makes an excellent point about strain gage correlations.

Stress engineers tend to want to locate strain gages for the static test right on the "hot spots": areas of very high strain and very high strain gradient. Gages in these location do a lousy job of providing data for a correlation effort.

The best strain gage layout for a correlation is to put gages in areas of high strain but LOW strain gradient. This way the gages are less sensitive to mislocation.

The best way to validate a FEM distribution on a wing, for instance, might be to put axial gages on every stringer and spar cap on both surfaces at a single wing station, with a good number of rosettes scattered on the skins in the bays between stringer gages.

Too often the engineer is asked to do a correlation without the right tools.

Another sign of untrained management... asking for FEM results that are "conservative". For a given set of externally applied loads, there is exactly one solution for any given structural system. Inasmuch as the model overpredicts the load in one location, it must necessarily be underpredicting somewhere else.

Maybe the engineers need to be training the management!

</soapbox mode off>

SuperStress

 

[aerodog]

crackman/Superstress,

Thanks for the good answers to my off topic question?

The basis for my question is to see if the reliance on FEA is the reason why manufacturers (especially start-up business jet mfg) miss their empty weight projections so often.

This might make a good topic by itself since there is great effort being expended trying to make lighter and stronger materials, avionics are certainly lighter but everyone keeps missing empty weights.

aerodog

just because a process can be automated
does not mean it should.

 

[SuperStress]

I think the reason manufacturers miss their empty weight projections is because they are made by the Marketing department, not Engineering.

SuperStress

 

[GregLocock]

More cynically (or even more cynically) - in order for the project to proceed it has to offer better payload etc than the competition. Therefore, in order for everyone to keep their jobs, at the start of the program it is in EVERY team meber's interests to emphasise the superior performance.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[aerodog]

Greg/Super,

There is inconsistency here. On one hand a lament, industry is losing the ability to do good stress work yet blaming marketing/management for not hitting empty weight targets.

An outsider might assume bad stress work leads to an unsafe airplane. Bad stress also leads to a heavy airplane.

Take the Eclipe airplane for instance. It sailed through static test with no failures and management is thrilled. Their press release proudly proclaimed "no detrimental permenant distortion at limit or ultimate load".

Is that good stress work or extreme conservatism? If an individual sheet metal part is .063 and it could have been made from .050, then it is 25% heavier than it should be.

If .050 called out where .040 would have worked out, again the part is 25% heavier.

If .040 instead of .032, again a 25% weight penalty.

Keep this up and you have an airframe 25% heavier than it should be and it will sail through static test. Is this good stress work?


aerodog

don't just rock the boat...jump up and down on the gunnels.

 

[prost]

Would it even be possible to design the kinds of really advanced, high performance airplanes (such as the 787 will be) without the sophisticated design and analytical tools we have now? You can cite the SR-71 as one of those really advanced airplanes that was designed and built without many of the fancy tools that we have now--I doubt the SR-71 could be built in the current budget and legal environment we have now; there'd be way too much risk for the GAO and other bean counters.

The positive correlations between the test data and the analytical tools are what is allowing the 'pushing of the envelope' we are seeing now, IMO. While it may trouble me to see phrases such as "Insights from FEA" (Machine Design just had such a front cover), there is no doubt in my mind that the current tools for analyses such as CFD have enabled signficant evolution of design practices.

 

[prost]

Trying to get back close to the original topic--Anybody out there get as disturbed as I do when some person not under your supervision (say at another company) starts asking you how to use engineering software that takes advanced knowledge and years to training to understand the results (such as FEA or fracture mechanics analysis tools) and you know from the questions that the person doesn't have a mentor and/or supervisor that is helping that person to understand? Would you feel the obligation to help the person? Forget about the cost of your time; how can a company expect a just graduated engineer understand how to use the software and interpret the results when such understanding would take years of experience to obtain?

 

[Sparweb]

What ever happened to the practice of designing the airplane 15% under the required ultimate strength, and then correcting the failures once the static tests are performed? Sorry I don't have a "been there, done that" story to add to my point, it's just what I heard thru the grapevine.

Back to the topic at hand: it was asked earlier, what training is required for new hires. I work in a mod/repair environment. Being expected to give on-the-spot answers about whether somebody's proposed solution is acceptable or not is common. Other times, the responsibility is upon us to create the solution. These situations require a lot of resources in your head. I find the "socratic method" works well for me: keep asking stupid questions. And, never be afraid to ask the same question again, later! Once you're more familiar with the problem at hand, you may find that you will understand the same answer better the second time around. Or, better yet, the person explaining it to you will include more info, because the process may have clarified their understanding, too.


Steven Fahey, CET

 

[SuperStress]

aerodog,

While I don't have any special insight to the Eclipse program, I would agree that if the press release is correct, having an airplane with no detrimental permanent deformation at ultimate load is indeed overkill.

Boeing would call this "growth potential", but I doubt that's what the boys in Albuquerque are after with a VLJ.

Is this bad stress work? It's impossible to say without more information. Parts of the design might be driven by damage tolerance requirements more than static strength.

Are heavier planes "poorly designed"? Redundant structure and alternate load paths provide added safety for occupants. Think Aloha 737.

Having airplanes that are too light are "poorly designed" as well. Nobody wants to own a hanger queen that can only fly 100 hours between scheduled inspections.

Airplane structure needs to be "right-sized", having just enough strength to meet certification and performance goals, while not carrying unnecessary weight in non-value-added locations. This is where good stress work comes in.

I think the lament about losing the ability to do good stress work comes from the new breed of "FEM analysts" that aren't well-schooled in classic theory and analysis techniques. Allow me to quote from an AV-8 stress memo from 1990.

"There has been a tendency in the last decade for inexperienced people to make finite element models or for the models to be made by a 'model group'. These can be very dangerous approaches. Modelers should be experienced load path persons and must work closely with the Stress and Design people that are sizing the layouts and the parts. Finite element models cannot be a substitute for free body diagrams and load path layout work. Some hard experiences have demonstrated this."

In my career I've come across these types of "FEM analysts" that couldn't free body a simple box; analysts that don't understand what shear center is; analysts that are completely ignorant as to how fasteners transfer load in a joint. Some of these same analysts are fluent only in linear static analysis, and they use it incorrectly! And these were "experienced" engineers and lead engineers! I feel sorry for the new graduates who find themselves being "mentored" by such engineers!

I've witnessed failures in test due to bad modeling assumptions; neglecting cutouts, kick loads, eccentricities, etc. The test I'm thinking about failed catastrophically at 75% of LIMIT load, when we were supposed to go to ultimate load without permanent deformation. This part was deliberately "overdesigned" because schedule didn't allow for a second "test-only" unit to be built - what we were testing was supposed to be flight hardware! The responsible analyst just sat there with his mouth open... he couldn't believe what he was seeing. Worse yet, HE DIDN'T UNDERSTAND WHY IT HAPPENED.

These are the engineers who scare me......

SuperStress

 

[crackman]

Superstress

Right on! Couldnt agree with you more. I worked for a company who recently went thru the same exact experience. They argued and argued with the certification authorities that due to their state of the art finite element modeling capabilities including linear and non-linear analysis, that no full scale ultimate test of their wing was required. Well......, the wing failed at about 137% of limit! So much for state of the art. This ended up costing them millions.

In the past 20 years, I have developed several airframe FEMs and IMO here is what much of the industry has forgetten about analytical methods and particularly FEM from its original inception and use in the aircraft industry. Back in the olden days, EVERYTHING was tested and the results evaluated (that is what all of the OEM stress manuals are based on - the famous Lockheed bathtub fitting method was based on P38 and B17 wing attach fitting tests). Analysis was used only for preliminary estimates until the test could be completed. THEN, and ONLY THEN, would methods be reviewed, modified, altered in order to predict the test failure modes. Having done this and established these methods, THEN the methods could be used to help evaluate new designs without repeating the expensive testing. LESSON NUMBER ONE - ALL methods must be correlated and validated. The FAA does not accept FEM without test validation, period. IMO the industry has become far too arrogant.

Now keeping this in mind, the experienced engineer knows that he must validate his method/FEM throughly at least ONCE and capture all of significant failure modes. Then, the method/FEM can be used subsequently for any new designs, mods, etc. This is the proper way to approach the issue. Most OEMs have spent millions of dollars validating their methods and have a large volume of substantiation data (even if there arent any engineers left at the OEM who know where the find the data). In fact, Boeing has a wonderful manual on exactly how to idealize airframe FEM which is based on decades of validation.

Unfortunatly, the people running companies today have no clue about this. Worse yet, many management teams of upstart companies think ANY testing and/or methods development to be a complete waste of money since CATIA with FEA is the solve all solution. Worste yet, young impressionable engineers start to believe this as well because "senior management must know what they are talking about", NOT. Unfortunately for them, they end up paying thru the nose when the day they submit their mod for an STC, the FAA says the following "...thats all fine and dandy but where's your validation to test data?".....The engineers are left bugeyed and the management team disappers or better yet blames the engineers. Shortly thereafter, the company folds up and lays everyone off. No one benefits from this.

One last comment then I'll get off my soap box. Many an old stress engineer always told me: "If half the design team and management arent pissed off at you, then, youre worthless to me as a stress engineer and you should call yourself one". I believe this to this day. Stress engineers must have a backbone and stand up when need be. Stress is the last defense line for a safe airframe.

Good luck all.

James Burd

 

[GregLocock]

"Would it even be possible to design the kinds of really advanced, high performance airplanes (such as the 787 will be) without the sophisticated design and analytical tools we have now? ...

The positive correlations between the test data and the analytical tools are what is allowing the 'pushing of the envelope' we are seeing now, IMO. While it may trouble me to see phrases such as "Insights from FEA" (Machine Design just had such a front cover), there is no doubt in my mind that the current tools for analyses such as CFD have enabled signficant evolution of design practices. "


I agree. The amount of what-if and optimisation that can be done in FEA is an important extension of the engineer's toolkit. We can also run analytical DOEs and end up with a more robust design.

SparWeb (Aerospace) 5 May 06 11:28
"What ever happened to the practice of designing the airplane 15% under the required ultimate strength, and then correcting the failures once the static tests are performed? "

Ah, the Colin Chapman design method! It's a bit bogus - if you truly designed to 85% then you'd have to improve EVERYTHING by 18%, so to some extent it only worked because the design process was generally conservative.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Sparweb]

The Chapman that hooked up with Burt Rutan years ago? Then I stand corrected. That may go over well with ultralights, but clearly everyone here prefers to talk about transport category and large passenger aircraft.


Steven Fahey, CET

 

[GregLocock]

No, Colin Chapman from Lotus, two apocryphal stories:

A designer was working on a suspension bracket. They tested the car, and it broke. So the designer redesigned it, made it heavier, they tested it , it broke. So, he redesigns it again, made it heavier, and they test it. It doesn't break. Chapman looks over the designer's shoulder, and says " Oh, I think we'd better go back to the previous design".

The ideal racing car should fall apart as it crosses the finishing line.







Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[SoonerMitch]

Systems guy here. My standard method is to give a new engineer a brand new bound engineering notebook day one. Step 1: A reasonable response given to a reasonable newbie question. Generally, guidance and directions on where to go look it up, but not answer question. Then expect them to come back with answer.
Step 2: Occasional repeat offenders who are making progress are directed to look in their notebook. Repeat Step 1.
Step 3: Habitual offenders are shown to the door. Small business + at will state.

Generally, a 6 month window to usefulness is expected.

 

[StressMan2506]

FAO Greg Locock:

Did the suspension component that Colin Chapman wanted to revert to undergo static or fatigue failure? Presumably the "ideal" racing car fails in fatigue as it crosses the line...


Louis