Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
Checking of calculations in the aircraft industry 14
- Thread starter Andries
- Start date
the Brit's used to have a really good system where someone would do a complete reanalysis (a "check stress") independent of the original calcs. obvious there's cross-referencing when conclusions differ significantly. maybe the yanks did this too, i encountered it when i was working in the UK. not sure if they did it these days, i think it might be something that's considered "too expensive" or "not value added".
but then the certification rests with the static test. the stress calcs are pretty much supporting data.
and of course in any company there is usually lots of checking, and usually (hopfully) enough experienced guys to shoo away those pesky project "engineer"s.
but then the certification rests with the static test. the stress calcs are pretty much supporting data.
and of course in any company there is usually lots of checking, and usually (hopfully) enough experienced guys to shoo away those pesky project "engineer"s.
- Thread starter
- #3
Check stressing is also the case in my neck of the woods as are the "project engineers".
A very worrisome thing, at least to me, is checking spreadsheet calculations. People tend to panel-beat data to satisfy the requirements of the spreadsheets instead of first looking at the fundamentals of the analysis.
Andries
A very worrisome thing, at least to me, is checking spreadsheet calculations. People tend to panel-beat data to satisfy the requirements of the spreadsheets instead of first looking at the fundamentals of the analysis.
Andries
ESPcomposites
Aerospace
A lead engineer should be amongst the first to check the results. Philosophies tend to vary though. Some concentrate on the details (i.e. get the adjustment factors right, follow the methods verbatim, while still using the canned tools). I usually do sanity checks with a piece of paper, a book, and calculator...coupled with some good assumptions and previous knowledge of what to expect. This tends to expose problems effectively and is more "independent" than just making sure the I/O of the tools is correct.
Brian
Brian
most companies prefer to use canned/known/"approved" s/sheets for calcs, so they don't have to validate them all the time. there is some "trust" that the analyst isn't going into the s/sheet code and tweeking it (wihtout telling anyone); many places will lock the code so the obvious routes are blocked, but a determined "cheater" can always find a way around.
if you develop your own analysis, this usually needs to be valiadated before someone will believe it ... this'll include philosophy and math ... usually a hand calc of one loadcase.
with all the checks and balances we still get satellites crashing into Mars 'cause guys make fundamental mistakes in their calcs. fortunately for us, most of our strcutural problems are revealed in test ... a la 787 side of fuselage "issue".
if you develop your own analysis, this usually needs to be valiadated before someone will believe it ... this'll include philosophy and math ... usually a hand calc of one loadcase.
with all the checks and balances we still get satellites crashing into Mars 'cause guys make fundamental mistakes in their calcs. fortunately for us, most of our strcutural problems are revealed in test ... a la 787 side of fuselage "issue".
MiketheEngineer
Structural
Had a friend that spent over 6 months checking on a forging for the Hawker that had developed minor cracks.
Another fried was project manager on an F-15 special fuel tank for 20 years or so.
Bet both got analyzed to death!!!
Another fried was project manager on an F-15 special fuel tank for 20 years or so.
Bet both got analyzed to death!!!
MiketheEngineer
Structural
Good point... Friends still alive and both planes still fly!!
Officially we had the system RB1957 alludes too. IE the design engineer was meant to have done his calcs as required as he went along, and the stress guy was meant to be an independent check.
For critical applications we stuck to this, however for a lot of other stuff the Stress guys calcs were the only detailed analysis.
For critical applications we stuck to this, however for a lot of other stuff the Stress guys calcs were the only detailed analysis.
Posting guidelines faq731-376 (probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484
What is Engineering anyway: faq1088-1484
mfgenggear
Aerospace
RB1957
yes you are cranky :>)
would canned software make less mistakes.
than hand calcs?
mfgenggear
yes you are cranky :>)
would canned software make less mistakes.
than hand calcs?
mfgenggear
I said "Design Engineer" A$$hole!;-)
That said, the design engineers in question were mostly HNC/HND/Apprenticeship so in the US probably wouldn't be allowed to be called engineers. Then again the govt approved signatory Stress Engineer was HNC/HND/Apprenticeship not degreed or Chartered - go figure.
That said, the design engineers in question were mostly HNC/HND/Apprenticeship so in the US probably wouldn't be allowed to be called engineers. Then again the govt approved signatory Stress Engineer was HNC/HND/Apprenticeship not degreed or Chartered - go figure.
Posting guidelines faq731-376 (probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484
What is Engineering anyway: faq1088-1484
you've worked in aerospace .. you know "Designers" design (whether they're called engineers or not).
i was voicing my opinion that they have their own body of knowledge to absorb and usually don't have the time to become proficient stress analysts. particularly where FE is concerned; in fact, if i come across a designer stress-wantabe who is good with FEA (not someone who thinks he is), then the question would be "why are you designing ?" and "what have you given up to become proficient in stress analysis ?"
no disrespect was intended or implied.
as for canned FEA ... yes, i am very suspicious of "instant stress analysis". the models are usually limited in how they can apply constraints, how they can load it, what elements you can use, the users are generally not well schooled in subtleties ... but they do produce pretty pictures.
i was voicing my opinion that they have their own body of knowledge to absorb and usually don't have the time to become proficient stress analysts. particularly where FE is concerned; in fact, if i come across a designer stress-wantabe who is good with FEA (not someone who thinks he is), then the question would be "why are you designing ?" and "what have you given up to become proficient in stress analysis ?"
no disrespect was intended or implied.
as for canned FEA ... yes, i am very suspicious of "instant stress analysis". the models are usually limited in how they can apply constraints, how they can load it, what elements you can use, the users are generally not well schooled in subtleties ... but they do produce pretty pictures.
-
1
- #14
SWComposites
Aerospace
would canned software make less mistakes than hand calcs?
> yes! it will probably have worse (and undocumented) assumptions than hand calcs, and is much more likely to be misused - the GIGO effect - too easy to think that if the input produces a "pretty picture" that it must be correct.
> the worst thing ever invented is one button FEA inside CAD programs
> yes! it will probably have worse (and undocumented) assumptions than hand calcs, and is much more likely to be misused - the GIGO effect - too easy to think that if the input produces a "pretty picture" that it must be correct.
> the worst thing ever invented is one button FEA inside CAD programs
-
2
- #15
GregLocock
Automotive
One big difference between those who put up buildings and bridges and other targets, and the manufacturing industries, is that we build and test protos to validate our often wildly imprecise calculations. That is a luxury that the architect's sidekicks can't often justify.
The saving grace with designing to code is that there are a lot of hidden safety factors, whereas if you are designing for function there is nothing to stop you designing any old thing, and (hopefully) relying on a good testing program to catch any errors in the design.
Approximately 25% of my work is the final safety related signoff for a car. My results and process and correlation are reviewed, but I've never had anybody actually check the runs themselves. For all they know I'm giving them the graphs of a completely different car (the input decks are stored, so down the track it is possible to replicate the work pretty easily).
Cheers
Greg Locock
New here? Try reading these, they might help FAQ731-376
The saving grace with designing to code is that there are a lot of hidden safety factors, whereas if you are designing for function there is nothing to stop you designing any old thing, and (hopefully) relying on a good testing program to catch any errors in the design.
Approximately 25% of my work is the final safety related signoff for a car. My results and process and correlation are reviewed, but I've never had anybody actually check the runs themselves. For all they know I'm giving them the graphs of a completely different car (the input decks are stored, so down the track it is possible to replicate the work pretty easily).
Cheers
Greg Locock
New here? Try reading these, they might help FAQ731-376
ESPcomposites
Aerospace
> the worst thing ever invented is one button FEA inside CAD programs
And the second worse thing is the mass use (abuse) of general FEA programs.
Brian
And the second worse thing is the mass use (abuse) of general FEA programs.
Brian
-
2
- #17
Nice to see this subject on the agenda.
As an Aussie, I deal with both civil and military regulators who are both quite anal about their requirement for independent verification of the adequacy of designs.
I personally like to see my analyses verified. It saves grief later.
I have found that the regulators seem to use language which aligns with ISO9000 type design philosophy, where the independent verification of the adequacy of a design is a basic requirement.
I know some individuals resent review for items which are non-structural and passed as acceptable by comparison. In such cases it means that another suitable individual is required to review the design and also find that it is acceptable by comparison (at least), FROM HIS/HER PERSPECTIVE.
From experience, I have found that items reviewed, certified and approved in such a manner will be more likely to be suitable for end-use than items where a review has not been performed.
In fact, an experience that I had with an aircraft OEM in the mid-80's (now defunct) cemented my views on this subject. They designed and certified a HF installation without a design review, only to find that the Tx/Rx was failing in-use. The chief avionics engineer responsible for the design swore black and blue that the design was good and the fault was with the box. After frying a couple more boxes (and with national stock levels in jeopardy), he finally agreed to a design review, whereupon a design fault was found.
Pride cometh before the fall.
I recommend that the review and validation process be built into the design process and not tacked-on after. It has positive effects on the quality of the design for function and cost as well as the engineering of the strength and durability for certification. If done properly it saves money and time and improves reputations.
I'm about to change employment modes shortly.
Watch out for Melbourne Aeronautical Design, operating in Melbourne (Australia), and anywhere else someone has a problem that needs a solution.
By the way, if anyone needs access to ANC-5, MIL-HDBK-5, MMPDS-01 for designs on aircraft which have referred to older issues (and they almost all have), I have a personal copy all of them except MIL-HDBK-5E Vol 1. If anyone has it I would love to obtain a copy.
I have personal copies of a lot of texts and other material. I will share if anyone wishes to examine my copy of anything.
I am cognisant of the need to avoid breaching copyright, so I cannot copy, loan or re-sell where the document is copyrighted (MIL-HDBK-5 is approved for public distribution by the way) or but I like to help as I have been helped myself and numerous good people have given me access to their libraries when I have been searching for info.
"Let me bring order to your chaos"
As an Aussie, I deal with both civil and military regulators who are both quite anal about their requirement for independent verification of the adequacy of designs.
I personally like to see my analyses verified. It saves grief later.
I have found that the regulators seem to use language which aligns with ISO9000 type design philosophy, where the independent verification of the adequacy of a design is a basic requirement.
I know some individuals resent review for items which are non-structural and passed as acceptable by comparison. In such cases it means that another suitable individual is required to review the design and also find that it is acceptable by comparison (at least), FROM HIS/HER PERSPECTIVE.
From experience, I have found that items reviewed, certified and approved in such a manner will be more likely to be suitable for end-use than items where a review has not been performed.
In fact, an experience that I had with an aircraft OEM in the mid-80's (now defunct) cemented my views on this subject. They designed and certified a HF installation without a design review, only to find that the Tx/Rx was failing in-use. The chief avionics engineer responsible for the design swore black and blue that the design was good and the fault was with the box. After frying a couple more boxes (and with national stock levels in jeopardy), he finally agreed to a design review, whereupon a design fault was found.
Pride cometh before the fall.
I recommend that the review and validation process be built into the design process and not tacked-on after. It has positive effects on the quality of the design for function and cost as well as the engineering of the strength and durability for certification. If done properly it saves money and time and improves reputations.
I'm about to change employment modes shortly.
Watch out for Melbourne Aeronautical Design, operating in Melbourne (Australia), and anywhere else someone has a problem that needs a solution.
By the way, if anyone needs access to ANC-5, MIL-HDBK-5, MMPDS-01 for designs on aircraft which have referred to older issues (and they almost all have), I have a personal copy all of them except MIL-HDBK-5E Vol 1. If anyone has it I would love to obtain a copy.
I have personal copies of a lot of texts and other material. I will share if anyone wishes to examine my copy of anything.
I am cognisant of the need to avoid breaching copyright, so I cannot copy, loan or re-sell where the document is copyrighted (MIL-HDBK-5 is approved for public distribution by the way) or but I like to help as I have been helped myself and numerous good people have given me access to their libraries when I have been searching for info.
"Let me bring order to your chaos"
rb1957, ours was a small design team on the peripheries of hardcore aerospace but we were expected to be able to be all rounders. Truth was we were all a bit week on the analysis side and we'd occasionally hire consultants to make sure we got a true 'second set of eyes'.
Certainly over reliance on FEA without validation (which in some cases is classical calcs, some cases testing, some cases all 3) or understanding is problematic.
We didn't do this, in fact we didn't have access to FEA most of the time.
Certainly over reliance on FEA without validation (which in some cases is classical calcs, some cases testing, some cases all 3) or understanding is problematic.
We didn't do this, in fact we didn't have access to FEA most of the time.
Posting guidelines faq731-376 (probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484
What is Engineering anyway: faq1088-1484
PHovnanian
Electrical
In systems engineering, calculations checking is more of an informal process we did within a design group. Prior to releasing our work product, the calculations of one engineer are checked by another.
Formal design verification was done by an outside, independent* group. Rather than checking the calculations of the design group (which supposes that they calculated the right stuff and didn't overlook anything), the V&V people started with the system requirements and developed their own analysis, wrote their own s/w and did their own work.
*The whole question of independence came up from time to time when I was at Boeing. Where the design and certification groups work for the same company and eventually reported to the same first level supervisor.
Formal design verification was done by an outside, independent* group. Rather than checking the calculations of the design group (which supposes that they calculated the right stuff and didn't overlook anything), the V&V people started with the system requirements and developed their own analysis, wrote their own s/w and did their own work.
*The whole question of independence came up from time to time when I was at Boeing. Where the design and certification groups work for the same company and eventually reported to the same first level supervisor.
- Status
Similar threads
