Discussion on Safety Factors for Composites Design and Testing

Discussion in 'Composites' started by Othman, Oct 20, 2011.

Help Support HomeBuiltAirplanes Forum by donating:

  1. Oct 20, 2011 #1

    Othman

    Othman

    Othman

    Well-Known Member

    Joined:
    Aug 23, 2004
    Messages:
    355
    Likes Received:
    4
    Location:
    Ottawa, ON, Canada
    Hello everyone, it’s been a while since I’ve posted anything on the forum, although I have been popping in once in a while for a quick read of the interesting things going on. I’ve been meaning to write this post for some time now as it’s quite an important subject and I haven’t seen it discussed too much here.

    In the last few years, I’ve been working more heavily with the design of composite structures, primarily for certified aircraft (development of STC modifications). As such, a full substantiation of structural strength of the new parts/structures was required to the satisfaction of the aircraft certification authorities. Design was done using analysis and testing was required for the official substantiation since, as most of you know, the material properties of composites are highly dependent on the combination of the materials, manufacturing processes and environmental conditions.

    Testing of composite structures is not as straight forward as testing traditional metal structures. Metals have well known properties over a wide range of operating temperatures. The published properties databases (such as the FAA MMPDS document or its predecessor MIL-HDBK-5) have statistical backup such that the values can be applied to say for example any 2024-T3 sheet or 6061-T6 extrusion etc. Therefore, the metal structure can be tested in the shop under typical environmental conditions (unless the design calls for high/really low temperature applications), subject to the limit or ultimate loads. Ultimate loads are simply the limit loads multiplied by a Safety Factor of 1.5 as specified by FAR 23.303, and any other Special Factors (1.15 for fittings etc) as specified by FAR 23.618. Easy.

    With composites, it is required that the effects of variability in manufacturing are taken into account as well as environmental conditions such as hot/cold and dry/wet. So how do you do that? Some use a catch-all Safety Factor of 2 as opposed to 1.5 between limit and ultimate loads. I had come across it in the past in an FAA Advisory Circular (AC107?), but have not been able to find it again because I believe it has since been removed and is not accepted by the aircraft certification authorities. After doing some research on this matter, which I discuss below, I proved to myself why.

    (Note: this an example only)

    Material properties databases available to the public are quite limited, and mainly consist of materials such as pre-pregs, and process involving autoclaves or elevated temperature cures. None-the-less, a typical data set will provide values of strength, such as tensile strength, at varying temperatures and moisture saturation levels (wet/dry). From that data, a material performance envelope can be graphed. The one in the picture below is made from data I was fortunate to find in a military report for wet-layup room temperature cure 7781 fiberglass/Hysol EA9396 epoxy laminate. The points of the blue shape represent the ultimate tensile strength under Cold/Dry, Cold/Wet, Room Temp (RT)/Dry, RT/Wet, Hot/Dry and Hot/Wet conditions. So it’s clear that the tensile strength of the laminate can vary significantly (cold/dry strength is 5 times greater than hot/wet).

    [​IMG]

    It might be tempting for someone to look at the tabulated data and pick out the RT/Dry tensile strength value and apply it to the analysis of their design, but as you see, that’s not representative of the conditions an aircraft will see in service over its life. To establish proper design values, the design operating temperature range must be defined as well as an anticipated worst case saturation level. Fully saturated is overly conservative, and according to Michael Nui (Airframe Structural Design) a typical value of 67% saturation should be used. This is shown by the red line in the material performance envelope below. The temperature range to use is up to the designer. Remember the aircraft can be sitting and baking in the sun for a while. For example, if the maximum allowable temperature is set to be 60 deg C (maybe a bit high), then the design tensile strength at 67% saturation would be approximately 26 ksi… that’s far from the RT/Dry value of 53 ksi (knock-down factor of 53/26 = 2.04).

    For the case above, the ultimate tensile of strength of 26 ksi should be used for analysis purposes. If a load test is to be performed at room temperature conditions, with an unconditioned “dry” test specimen, the applied load(s) would have to be scaled by an overload factor of 2.04 (as calculated above) to account for the degraded strength at the worst case design temperature/saturation level. Therefore the ultimate test load would have a total Safety Factor of 1.5 x 2 = 3.0 over the limit load. This is just an example, as not all composite layups perform as “badly” at worst case conditions. From the data I have studied, it was clear that wet-layup, room temperature cured laminates typically have lower maximum strength and the most degradation in adverse environmental conditions (with more advanced materials/methods, the previously mentioned total Safety Factor of 2 is more applicable). What’s important is that these “low-tech” methods traditionally used to build many composite homebuilts. So if you are designing your own, careful consideration of the design safety factor you use is warranted, and should be representative of the materials and processes employed.

    In lieu of test data, some aircraft certification authorities have accepted Safety Factors of 2.25 for non-structural components and 2.5 for structural components. There is no reference for these values, but these may only be acceptable for one-off type modifications, not production products.

    Some of you may disagree with the method I’ve shown here to compensate the safety factor for composite materials, but the “moral of the story” is that I want to make amateur designers aware of the necessity to use higher Safety Factors when designing for composites, and to generate discussion on this matter. Maybe others can share their methods and how they approach this problem.

    If you made it this far, thanks for bearing the long post.
     

    Attached Files:

    Last edited: Oct 21, 2011
  2. Oct 21, 2011 #2

    wsimpso1

    wsimpso1

    wsimpso1

    Super Moderator Staff Member

    Joined:
    Oct 19, 2003
    Messages:
    6,164
    Likes Received:
    3,466
    Location:
    Saline Michigan
    Welcome back Othman. Sorry, no particular advice...
     
  3. Oct 21, 2011 #3

    Othman

    Othman

    Othman

    Well-Known Member

    Joined:
    Aug 23, 2004
    Messages:
    355
    Likes Received:
    4
    Location:
    Ottawa, ON, Canada
    Thanks wsimpso1. Your craft looks like it's coming along nicely by the way.
     
  4. Oct 24, 2011 #4

    D Hillberg

    D Hillberg

    D Hillberg

    Well-Known Member

    Joined:
    Nov 24, 2010
    Messages:
    1,189
    Likes Received:
    434
    Location:
    very low low low earth orbit
    Try adding engine & exhaust heat on cowled & free stream areas on the fuslage, It gets real ugly and weak in some places.
     
  5. Oct 25, 2011 #5

    autoreply

    autoreply

    autoreply

    Moderator

    Joined:
    Jul 8, 2009
    Messages:
    10,732
    Likes Received:
    2,553
    Location:
    Rotterdam, Netherlands
    Thanks for a very interesting post!

    I wasn't aware of that huge contribution of the moisture content of the air. Would you expect similar results (3 times weaker) for a typical airframe?
     
  6. Oct 25, 2011 #6

    Othman

    Othman

    Othman

    Well-Known Member

    Joined:
    Aug 23, 2004
    Messages:
    355
    Likes Received:
    4
    Location:
    Ottawa, ON, Canada
    Thanks autoreply,

    As is with most things, the answer is "it depends".

    The data shown in the original chart is only one sample for a specific laminate. Different materials and processes will result in laminates with different performance (strength loss due to moisture ingress).

    It is up to the designer to make the judgement on what is a suitable design condition (how hot and how wet) and apply the necessary knock-down factor to the strength values. Looking at it from a homebuilt airplane perspective, the designer should take into consideration if it is a one-off airplane or if it is intended to be a kit for anyone to build. If it is a one-off, and you know the airplane will always be in a dry climate etc, selecting the design conditions will not be as critical as if the aircraft was a kit that could live in Papua New Guinea, Sahara Desert or Iqaluit.

    Also, often as amateur designers we have limited access to good data sets for composite laminates, and the data we do have is not exactly for the laminate we intend on using. This is another area where the designer will have to apply good judgement. For example, if you look at data for some pre-preg, elevated temperature cured laminate, you might find that the strength at your design point is not too far off from the room temperature/dry strength (say a factor of safety of 2 would be sufficient). You cannot go an apply this value to a laminate that is wet-layup room temperature cure. This is an extreme example, more judgement is needed when the laminates are very similar but maybe the epoxy is different.

    What's interesting in all of this is that we can't just simply test a composite structure under the normal design load, as it may not exactly reflect the strength of the structure after many years of being exposed to the environment. That's why you need to justify an appropriate overload factor to apply to scale the design load.

    This is my take on the matter, which has come from experience in certifying aircraft components. It would be interesting to hear the experiences of someone who has possibly worked for any of the OEM's certifying composite aircraft, or even for kit aircraft.
     
  7. Oct 25, 2011 #7

    Himat

    Himat

    Himat

    Well-Known Member

    Joined:
    May 5, 2011
    Messages:
    2,858
    Likes Received:
    668
    Location:
    Norway
    Thanks for posting, sure it is interesting.
    Different materials do have different properties.
    Some "plastic" materials is stronger with a certain moisture ingress than without.
    One example is glass filled nylon, wich become brittle if completely dry.

    In my work I find that "plastic's" often have sought after properties.
    But it's very difficult to get propper data on how they behave at elevated temperatures, low temperatures, high ambient pressure and other "special" cases.
    And then there is the batch to batch differences and changes done through the years of manufacture.
    The data sheet might be the same, but quite often I'm not sure if all the properties are the same five or ten year's after the product was introduced. The manufacturer might have tweeked the prosess, substituted some component or just changed the operator.
     
  8. Oct 25, 2011 #8

    BBerson

    BBerson

    BBerson

    Well-Known Member HBA Supporter

    Joined:
    Dec 16, 2007
    Messages:
    12,664
    Likes Received:
    2,590
    Location:
    Port Townsend WA
    I own the first Grob G109 that was imported into the U.S. and it was used for FAA certification. I was talking to the person who did the certification and he said all composite aircraft gain weight from moisture over time. I would be interested to know how much is gained and why, I didn't think to ask him.
    Anybody have any thought about water gain?
    BB
     
  9. Oct 25, 2011 #9

    craig saxon

    craig saxon

    craig saxon

    Well-Known Member

    Joined:
    Sep 26, 2010
    Messages:
    549
    Likes Received:
    32
    Location:
    Canberra, ACT, Australia
    You should be able to figure that out for your own aircraft by looking at the original certification weight and comparing that with the current weight. Any equipment changes or repairs that might affect weight should be in the log book. You could then work out a percentage. It would be an interesting exercise.
     
  10. Oct 25, 2011 #10

    Himat

    Himat

    Himat

    Well-Known Member

    Joined:
    May 5, 2011
    Messages:
    2,858
    Likes Received:
    668
    Location:
    Norway
    I have seen test results on some "plastic" materials (intended for offshore use).
    From memmory, weight gain after a long time in water range from less than 1% to several %.
    Epoxies formulated for the purpouse perform well, if not their properties could be either good or bad.
     
  11. Oct 25, 2011 #11

    BBerson

    BBerson

    BBerson

    Well-Known Member HBA Supporter

    Joined:
    Dec 16, 2007
    Messages:
    12,664
    Likes Received:
    2,590
    Location:
    Port Townsend WA
    A boat in the water would absorb more, I suppose. An airplane would absorb mostly from humid air and rain. So it might change over time and with the season. The epoxy absorbs the moisture, I think. What about the glass?
    BB
     
  12. Oct 25, 2011 #12

    autoreply

    autoreply

    autoreply

    Moderator

    Joined:
    Jul 8, 2009
    Messages:
    10,732
    Likes Received:
    2,553
    Location:
    Rotterdam, Netherlands
    Yeah, it usually does, doesn't it? ;)

    My astonishment is mostly due to the shear magnitude of the effect moisture apparently has. It's hard to get hard date or knowledgeable comments in GA composites indeed, but what I've heard and seen over the years, moisture was often totally ignored. Temperature often is mentioned and at least EASA CS22 specifies a test temperature of (I think) 55 centigrade.

    I've always assumed (hearsay mostly) that water adsorption is major issue in typical ship building (polyester) but not so much in aircraft with high-quality epoxy and vinylesters and this got further reinforced by several tests on used airframes (the nimbus 4DM which crashed in the states and fatigue testing on other crashed gliders for example)
    The NTSB report has a lot of real-life results. The glider was only 3 years old (200 hours), but I've seen similar results in testing of much older airframes.
    http://www.ntsb.gov/doclib/reports/2002/AAB0206.pdf
    Page 34:
    And, despite fairly low margins (1.55 FoS on many parts):
    Mostly rumors. Ships get up to a few perfect is what I was told, cleaned wings that I've measured were below 0.5 percent weight increase after well over a thousand hours flown in a relatively moist climate.
     
  13. Oct 25, 2011 #13

    D Hillberg

    D Hillberg

    D Hillberg

    Well-Known Member

    Joined:
    Nov 24, 2010
    Messages:
    1,189
    Likes Received:
    434
    Location:
    very low low low earth orbit
    Water ingress will flow along the voids in the strands and migrate with time, Honeycomb will collect pockets of water all through small pin holes. freezing water in rotor blade pockets will throw ballance off and damage bond lines.
     
  14. Oct 25, 2011 #14

    BBerson

    BBerson

    BBerson

    Well-Known Member HBA Supporter

    Joined:
    Dec 16, 2007
    Messages:
    12,664
    Likes Received:
    2,590
    Location:
    Port Townsend WA
    Autoreply,
    That Nimbus 4 NTSB report has quite a bit of detail, thanks for that link.
    I think the connection between the airbrakes and the flaps is a bad idea. It would increase the g-loading when deploying the flaps at high speed.
    BB

    p.s. doesn't seem like moisture is a big problem.
     
  15. Oct 26, 2011 #15

    Othman

    Othman

    Othman

    Well-Known Member

    Joined:
    Aug 23, 2004
    Messages:
    355
    Likes Received:
    4
    Location:
    Ottawa, ON, Canada
    Here is a link to the NIAR AGATE site that has a good collection of data (this link was posted on the forum before):

    http://www.niar.twsu.edu/agate/

    In these test reports, full saturation of the test coupons is defined as the following:

    "Effective moisture equilibrium was achieved when the average moisture content of the
    traveler specimen changed by less than 0.05% for two consecutive readings within a
    span of 7 ± 0.5 days"

    The actual % weight increase data for the specimens is contained in the later sections of the reports, and is typically less than 1% at equilibrium which occurs after several weeks (in the order of 50 days) of exposure at high temperature/high humidity conditions. So we're not talking a whole lot of weight increase.

    The reports also discuss the material performance envelope and interpolation of data within the envelope, as I discussed in the original post.
     
  16. Oct 26, 2011 #16

    autoreply

    autoreply

    autoreply

    Moderator

    Joined:
    Jul 8, 2009
    Messages:
    10,732
    Likes Received:
    2,553
    Location:
    Rotterdam, Netherlands
    In fact, the breaking wing was started by less lift at the inner wing, since popping out the drag brakes on the inner wing works as a spoiler. Lift shifts outward, wing breaks.

    Thanks Othman. I'm familiar with their reports and I've read several of them (not the test ones) in the past, but apparently missed the large influence of moisture. (here, close to 50% on page 30)

    Time for some serious recalculating on my design, though I'm not that afraid that it will change much. Once again, thanks a lot for this contribution.
     
  17. Oct 26, 2011 #17

    BBerson

    BBerson

    BBerson

    Well-Known Member HBA Supporter

    Joined:
    Dec 16, 2007
    Messages:
    12,664
    Likes Received:
    2,590
    Location:
    Port Townsend WA
    Page 30 shows wet numbers at 180 degrees F.
    What about 75-100 F (moist)?
    BB
     
  18. Oct 27, 2011 #18

    Othman

    Othman

    Othman

    Well-Known Member

    Joined:
    Aug 23, 2004
    Messages:
    355
    Likes Received:
    4
    Location:
    Ottawa, ON, Canada
    autoreply,

    I'm glad it got you thinking and looking at your design. That is what I intended by bringing this up in a sort of design example. When I was doing the research for myself I though it would make an interesting post as it isn't discussed very clearly in much of the available documents. I know a lot of people who visit this site are working on some interesting new designs using composites and could benefit from this information.

    BBerson,

    You can linearly interpolate within the material performance envelope to get design values for the conditions you set (say 80 deg F and 67% saturation).
     
  19. Oct 29, 2011 #19

    orion

    orion

    orion

    Well-Known Member

    Joined:
    Mar 3, 2003
    Messages:
    5,800
    Likes Received:
    135
    Location:
    Western Washington
    This is one of the more difficult areas of airplane design, especially in the homebuilt industry or for the case of one-off projects where you are often unable to guarantee a factory-like setting of processes and quality control. The subject could be a fairly long dissertation of design practices and property databases used (and their modification) for the derivation of serviceable design values and safe products. Going into this at any level of detail would be a long discussion and quite frankly, beyond my typing tolerance. But I can put forth some basic thoughts, in no particular order, for folks to consider as they go down this path.

    1) We commonly acknowledge that property values of any particular fiber and weave will change due to the resin used for the particular product. But we often fail to discuss a couple of basic corollaries which are just as true and are beneficial to the process. First off, while properties do of course change with resin, the change across several of my databases (wet layup, epoxy with 7781 fabric) tends to be relatively small. As such, it would not take a sizable property modification factor to account for an unknown (quality) resin. Also, the highest strength resins do not necessarily produce the highest strength laminates. For the case of our applications, it is generally best to use what is termed a "toughened" resin, which means that it has a rubber component as part of its chemical makeup. This additive makes the cured resin less brittle (and less subject to issues of fatigue and localized crack initiation) and tends to provide the laminate with higher peel strengths and at times, even higher inter-laminar shear strengths.

    2) In the design of composite airframes, the term "limit condition" is meaningless since the materials do not have a non-linear portion of their stress-strain curves. The properties are such that the curve is for the most part linear to the point of failure (which is defined as a strain in a particular orientation). As such, it makes sense to only design to ultimate condition. The interesting part though comes when you combine metal structures with the composite ones and you have to decide what component or property will define the failure point (personally, I find that changes depending on application).

    3) We've talked about this one here before: Never use one layer of anything. It is better to use two layers of a thinner material than one layer of a thicker one. There are several reasons for this but most center around the way a single layer may behave under high stress. The goal of composite design is to not only taylor the structure to behave in a particular way when loaded, it is also to work at a laminate that has the same type of stability as the more conventional isotropic metals have. When you have a single layer, some fraction of the load may become more concentrated in a localized region of fibers. Because you only have one layer, the stress field may not distribute itself into the surrounding material thus causing higher localized stresses than what analysis may reveal. The failure will be first realized in those affected strands, which may fail long before the ultimate design condition is reached. With multiple layers you have a much better chance of a more benign stress distribution since the load is carried by overlapping strands. Other issues we've discussed here before but simply said, experience shows that multiple layers serve better than a single layer.

    I know there are aircraft out there that do use single layer laminates in their structure and the argument that gets put forth is that they seem to fly just fine. And I agree - as long as these light aircraft fly the typical flight profile, the potential problems I describe will most likely not make themselves evident. But all it takes is one abrupt maneuver or shock and the damage may start. Airframes rarely fail at one or two Gs. But it's the case of higher loading, such as may be needed for an emergency (high G avoidance maneuver, gust loading due to unpredictable weather, etc.), where the occupants are expecting a certain load capability, where any induced damage may make its presence known. And I think it's this scenario that is equally as important in the design of an airplane as is getting the proper material data.

    4) To the greatest extent possible, try to design your structure around a balanced laminate. This primarily applies to sandwich skin structures where an unbalanced laminate may result in distorted parts (different shrinkages during cure). This is not so much a strength issue but more so one that may affect part quality and build ease, especially for parts that are hand laid (wet layup) and room temp cured. Keep in mind that a room temp cure resin may continue curing for months after you remove it from a mold - during that time it continues to shrink and when there is an unbalanced laminate, that shrinkage will be uneven, potentially leading to part distortion.

    4) Don't be so weight conscious that you compromise safety. There is a vast difference between material optimization and weight paranoia. I've seen several light aircraft programs where a layer of material was arbitrarily removed based on the theory that the design is "obviously" over-designed and performance goals dictated that weight be removed. This in my opinion is the ultimate in dumb management decisions but it does happen. I think most of us would gladly pay the few pounds incurred in a layer of structural material, with a nearly imperceptible difference in performance, in order to assure ourselves of an airframe capable of the full design load.

    5) As far as structural design is concerned, I've seen no better publicly available resources than those provided on the AGATE page referenced earlier. I do have one or two better documents (excerpts from the Boeing Material Specification manual) but for the average designer, this is a great start. To be conservative use the A-values. The B-values can be used however you must really understand your application, your building skills and environment, in order to do so safely. As previously mentioned, design to ultimate conditions. These are as defined by yourself (after all this is an experimental, which puts you in the position of making decisions such as airframe limits) - if you wish to stay conservative adopt the FAR imposed factors and you should be fine (7.6 Gs for Normal Category; 12 Gs for Aerobatic; don't forget gust loading).

    6) Decisions regarding moisture and heat are a bit more problematic and would require a much longer discussion than what we can do here. In general though (for GA products only), most folks consider heating only for aesthetic reasons (don't want your skin to droop or print through). One is unlikely to go up and do unlimited aerobatics or high G combat maneuvers when it's a hundred outside. Most likely the airplane under these conditions is hidden away or is flying to a cooler part of the country. Either way though, heat soaking is rarely an issue mainly because composites (even graphite based) are extremely poor heat conductors and heat soaking through to critical laminates has not been a significant issue for any manufacturer thus far. However, that is not to say that some consideration should not be given - in general it is a great idea to design component assembly bonds to the case of elevated temperatures, just to provide yourself with a bit of insurance.

    Property modifications due to moisture are generally minimal - historical practice suggests that using A-value based material properties is sufficiently conservative to account for the minimal degradation that an enclosed and painted aviation product may see over a lifetime of humidity exposure. I've only seen one program where humidity and moisture effects were considered and that was in the design of Unlimited Hydroplanes - I've never seen it in experimental GA - I am not familiar with Cirrus or Lancair (Columbia) design practices to expand that statement into the certified lines. I'm pretty sure the original Lancair aircraft did not modify the properties for moisture but do not know if the later higher end products did or not.

    Anyway, I know this is very brief and does not go into sufficient detail to make any set decisions. It's also based more on design practices in the experimental and/or light aircraft sectors and no so much on practices of the mainstream industries. But maybe it's enough to allow folks to ask more in depth questions.
     

Share This Page

Group Builder
arrow_white