5 Common Composite Design Oversights - and How to Avoid Them!
Wondering how to effectively take advantage of composites in your components to reduce design iterations? Join us in a discussion on the most commonly overlooked areas in designing composite structures and how to effectively avoid them early in your process, allowing you to reduce your design iterations and leverage the full potential of advanced materials.
Director of Analysis Services
Teton Simulation Software
Rick received a MSME from the University of Wyoming with a thesis on cryogenic thermal fatigue of unidirectional composites. Upon graduation he joined Firehole Composites and immediately began work as an FEA-based composite structural analyst. Firehole was the developer of a software product called Helius MCT which is an add-on to commercial FEA codes Abaqus, ANSYS, and Nastran that enables high-fidelity progressive failure analysis of composite structures. While at Firehole, Rick was the lead analyst on several challenging composite projects and worked with companies in a variety of industries such as commercial aerospace, Formula 1, renewable energy, defense, racing yachts, and sporting goods.
Autodesk acquired Firehole in 2013 and at Autodesk Rick continued work as a structural analyst as part of Autodesk Advanced Consulting and was part of a multi-company ICME effort for composite materials.
In late 2018, he joined Teton Composites where he is currently the Director of Analysis Services.
Rick is continuously engaged in the composites community. He is a member of the NAFEMS Composites Working Group and regularly attends industry trade shows such as CAMX, SAMPE, and JEC.
Waterfront Composite Solutions
Originally from New Zealand, Mark Bishop has a degree in Mechanical Engineering from the University of Auckland and has been designing full time in the field of advanced composite structures since 1994 with a background in general acoustics from original work with Marshall Day Associates in Auckland, NZ. Working for High Modulus in NZ, he has overseen the design and construction of composite structures throughout the world.
In 2001 he moved to the US to be the onsite structural engineer for the advanced composite high speed motor yacht Adler that he had previously completed the structural engineering for. He began his own engineering consultancy business before being recruited by Farr Yacht Design in 2003 where he was responsible for the composite engineering of Farr Yacht Designs race boats from 2003 through March 2011 including the Volvo Open 70’s Telefonica Azul and Abu Dhabi Ocean Racing’s Azzam amongst multiple race winners and world championships. Mark has also engineered all of the high speed composite boats from Mystic Powerboats since 2002 including the structurally revolutionary and world’s fastest C5000-R offshore power cats as well as other high speed offshore racing powerboats including the 2005 P1 World Championship winning Dragon 39 ARPRO. In 2008, Mark was engaged independently to provide the composite structural engineering for the 141m Giga Yacht SWIFT 141 including full DnV classification. This project lead to reforming as an independent engineering consultancy with the creation of Waterfront Composite Solutions in 2009 to provide clients with more than composite engineering but rather a complete holistic design service.
Based in Annapolis Maryland, Mark has been a regular guest lecturer at the Massachusetts Institute of Technology (MIT) in Boston, the US Naval Academy in Annapolis MD as well as presenting at the SAMPE and IBEX conferences.
Q & A
Q: Do you have general guidelines on manufacturing and materials selection in different geo's?
A: (MB) I’m assuming this refers to different geographical areas. Generally, any western european country tends to have relatively high labour rates with a rough cost component being, say ⅔ labour ⅓ materials. So it’s usually preferable to spend more on materials to decrease labour content all other factors aside. Examples could be the use of thicker cores and higher modulus carbon skins for structures to reduce framing elements. In other countries with significantly lower hourly wage costs, the reverse is usually true - so rather than a lot of sandwich cored structures you might see more single skin, multi-frame type solutions in some areas, and more use of e-glass rather than carbon. Obviously this is a highly generalised statement but is reflective of the options you do have to adjust methodology with composites, and one is not inherently better than the other. It just depends on what the best “fit” is. Compare this to a steel structure built anywhere and it will always look exactly the same in terms of its form.
Q: Are there communities where I can acquire more information from?
A: (MB) Outside of places like SAMPE I’m not really sure as I’ve never inherently made use of forums etc.
A: (RD) There are several communities that focus on supporting the composites community in various ways (trade shows, blog posts, educational publications, training, and more):
NAFEMS - The NAFEMS Composites Working Group is where you can find links to educational publications, training, etc.
Composites World - This well-known organization is dedicated to composites.
JEC - Another organization that is dedicated to the composites community.
SAMPE - An organization with a long history of serving the composites industry and is associated with the relatively new trade show CAMX.
Q: What techniques do you use to determine where to hybridize? Are there manufacturing considerations?
A: (MB) I think one of the biggest issues is establishing what the fundamental requirements are and/or what characteristics you may want to see. If, for example, you find yourself making a definitive statement that a certain type of laminate or a composite “can’t” work, this might be a hint towards hybridisation... For example, do you have a higher tech structure where you are fundamentally constrained on a bolted joint edge distance in composites but can’t seem to generate suitable shearout margins? Considering a hybrid titanium leaf and carbon composite hybrid can offer a solution that is aligned with the generalised composite structure and avoiding a whole sale and potentially problematic shift to a full metal solution. Another example may be where you have extremely disparate requirements across a structure - one area may be specifically stiffness critical in an otherwise non stiffness critical structure. If that’s the case, it could or should suggest a switch to a different material within the composite (e.g. from glass to carbon). Then, in turn, overlaid onto the macro structure the wholesale choice of carbon is not reasonable, the next and probably obvious solution is to consider a local hybrid approach. Disparate stiffness requirements based on geometric form can also hint at looking at a hybrid (i.e. a carbon reinforcement in one direction and a e-glass in the less critical directions).
Q: In your experience, what is the typical number of design iterations for a composite structure?
A: (MB) That really depends on budget, but as a rule I would say you should always expect to do a minimum of three. Your baseline work - your first iteration to review behaviour - the second to assess the changes you made were doing what you expected - then a final third run to get a reasonable level of developmental optimisation. Obviously doing 2-3 more runs can lead to greater optimisation, particularly if it’s in an area less familiar to you. But practically after that I think you are getting into the area of rapidly diminishing returns. This, however, is all predicated on the assumption you can reasonably identify your correct concept and initial structure baseline as opposed to simply taking a wild guess. If you are simply guessing, then I’d suggest you need to be really cautious about the direction you are steering your analysis, as you might simply end up spending a lot of time doing really good analysis and optimisation on a fundamentally flawed concept... and that’s not good design.
Q: Can you discuss the impacts and considerations of assembly techniques for composites?
A: (MB) Often these go hand in hand with material selections and manufacturing processes. While, for example, it may be beneficial to consider as low as a ply count as reasonably possible for the requirements of the structure to aid manufacturing by reducing time to lay it up, sometimes depending on the geometry of the part and the form or architecture of the ply the opposite could be true. Trying to fit a heavy prepreg ply into an intricate geometry can be more time consuming than multiple lighter plies This may not always be obvious, and subtle changes in ply mass can offer benefits to construction that is difficult to ascertain from the engineering office. That’s why making sure there is a positive feedback process between build and specification is, in my view, the key. Similarly, a one size corner radius spec doesn’t work either unless you want to go toward a larger size. Generally, a tighter radius for a female corner is possible if you are vacuum consolidating, but you will need a larger radius if hand consolidating. Part thickness does also play a role if you are considering bonded in type construction. That’s an assembly at the ply level. For components, if say you want to be able to bond two parts together onto a filled adhesive, and you are not going to have fixturing to positively clamp the mating surfaces, then you will need to ensure the part itself has sufficient stiffness to provide a positive clamping force and not simply bend away. That, in turn, depends on the material being used - carbon being stiffer can be thinner than E-glass, but the rate of stiffness increase in a monolithic section increases rapidly with thickness... so it’s the net EI of the section, not just the E of the reinforcement. Practically this can mean the idea of a “free” bonded joint might end up being heavier, and then you also have to consider the choice of adhesive based on how much or how little bondline thickness control you have. More aggressive weight control or performance of bonds will require more extensive fixturing and direct clamping force so the part is not being required to generate the clamping force itself. The list can go on and on, but again loops back to the notion of understanding the relationship and relative importance placed on manufacturing vs. cost vs. performance, etc. before you start.
Q: Does the manufacturing process influence the structural behavior of composites?
A: (MB) This is perhaps unfortunately, a yes and no answer. I’d start out by saying that I think it’s useful to recognise that the manner in which we are used to describing strength and stiffness for metals which is a cross sectional area type e.g. MPa, GPa, psi, ksi etc is primarily useful only for homogeonous isotropic materials and can be highly misleading in composites where there is a highly disparate mechanical range between fiber and matrix. By this I mean than in principal load carrying capacity of a composite is primarily a function of the fiber (yes, generalised w.r.t. the effect of very low resin contents etc). So basically what we should be thinking about is the load carrying capacity being a number that’s related to the load per unit width per areal mass of fiber. This sounds all complicated but in reality it’s pretty simple as we are effectively only removing the thickness aspect of the composite and in turn indicates that the resin content doesn’t really affect the fundamentals of the structure... if we ignore for now local defects such as voids etc. For example I’ve often heard people say vacuum bagging makes a laminate stronger than hand layup, which if we consider load carrying capacity of the fiber is not true and here’s why - let’s say i have a carbon unidirectional laid up at 35% resin content and we have a thickness for 1 kg/m2 of unidirectional of roughly 1 mm. If we test the 0 deg properties (i.e. along the fiber), then roughly speaking for a generalised standard modulus fiber we would expect to get a tensile stiffness of 130 GPa (yes higher and lower is possible but this is a generalised ballpark number), and we might then test a tensile strain to failure of say 1% for a tensile strength of 1300 N which for 1 mm thick and 1 mm wide is 1300 MPa putting it into “normal” units. If we take the same fiber and lay it up with a resin content of say 45% then we would expect to generate a tensile stiffness of about 104 GPa and a laminate thickness of about 1.25mm and expect to also find about 1% failure in tension all things being equal for a tensile strength of 1040 MPa for a 1mm wide strip so to speak. This is the issue i see with people looking at raw strength or stiffness data in composites and saying “see the vacuum bagged sample is stiffer and stronger”. It’s not it’s exactly the same as we need to normalise our data to the same thickness. If you multiply 104 GPA by 1.25 mm you get low and behold 130 Gpa, which is the same as the 1 mm thick sample acquired by vacuum bagging. So what the vacuum bagging has done is give you the same load carrying performance for less weight, not miraculously some inherently stronger laminae. Then you can start to work into how different consolidation and manufacturing techniques reduce your defects which does intrinsically have an effect on your properties but as long as you are aware that it’s the effect on defects and not inherent load carrying capacity variable manufacturing on a consistent fiber and matrix has then this may help. You can then look at how changing matrix behaviour can change your properties and generally moving to a prepreg type matrix that is not required to be liquid at room temperature can improve properties by improving shear stiffness of the matrix and in turn improving fiber buckling resistance coupled with the prepreg fiber being inherently straighter etc.
Now you need to understand your failure modes, and if they are even going to be influenced by improving mechanical performance and so in turn by the manufacturing method. For example, if you have a sandwich panel where the skin wrinkling limit is say... 0.47% in compression, then changing your manufacturing method to improved the skin outright mechanicals from say 0.65% in compression to 0.75% is going to have absolutely no impact on the critical failure mode of the panel Conversely, if you have an outright compression critical monolithic structure constrained by geometry then moving toward a prepreg intermediate modulus unidirectional offering you an increase in compression strength of 1300 MPa vs. 845 MPa (normalised to the same thickness!) for an infused standard modulus carbon unidirectional is clearly a worthwhile jump predicated by different manufacturing AND associated material selections. You may also find the improvement between the autoclaved prepreg and an out of autoclave prepreg MIGHT also give you a worth while step again, but a lot of the time it may not. So above all it again comes down to understanding what you need to do conceptually and how that is going to inherently influence failure modes before looking at data that says one manufacturing method or material combination is “better”. It can certainly influence it you just want to avoid being influenced by things that don’t really exist in your own reality.