Photo: Paul Daly (AP)

Apologies to those visiting my blog searching for updates on Evolution Cedrus, my next generation carbon fiber hydrofoil mast. But in the wake of the loss of Titan, I am feeling a need to address the false information that is being shared by the media and self-proclaimed “experts” regarding the assumed role of composite materials in the loss of 5 people. These assumptions, about a very complex and unknown failure, are actually causing me to question my entire consumption of the media as I’m realizing it’s not just social media, but major outlets like CNN and NYT capable spreading lies whether intentional or not. 

I’m not trying to call out James Cameron, or make the case for composite materials to be used in any and all applications. My goal here is to bring attention to the challenges of multi-disciplinary optimization as it pertains to product design and engineering. When designing something like Titan, engineers must balance a combination of performance criteria from structural requirements, hydrodynamic characteristics, weight, cost, manufacturing feasibility, and schedule. The same is true when building airplanes, cars, and hydrofoil masts, although the constraints on each are different. I do hope to further educate readers on the benefits and drawbacks of composite materials, and give evidence as to why they may have been used in this particular application. We may never know if failure was due to a delaminated composite vessel, cracked titanium end caps, a sheared stainless steel bolt, or the acrylic viewport. Until we do, it is both foolish and disrespectful to assume that failure was simply due to composite materials alone.

Before I jump in, I feel it’s important to share a bit more on my background. I have worked closely with composite materials for 20 years. I began my professional career as a structural analyst/engineer on the 787 program, the first all-composite commercial aircraft. There I was responsible for ensuring various components, both metallic and composite, were capable of meeting structural requirements in support of FAA certification. I then spent time in Silicon Valley, as a lead structural engineer on the earliest flying car company. But when I realized that flying cars are decades away (10 years ago) I left for Apple where I used the same tools (Finite Element Analysis) to improve the design and reliability of various products from iPads to watch bands. I have founded or co-founded 3 companies, one of which developed a low-pressure (vacuum) chamber out of composite materials to store highly perishable produce. I have 12 issued patents and others pending, most of which are related to composite materials. Through Adherend Innovations, I help support companies and clients solve challenging problems related to product design, materials selection, structural optimization, testing, and more. To be clear, I’m not just a composites engineer or fanboy; I enjoy finding the right material for the application. For example, I was the first to introduce a carbon hydrofoil mast with an aluminum mount and end fittings, a design that many foil brands have since followed. I know the benefits and drawbacks of various materials, from carbon fiber to titanium. This is perhaps why so many of the Titan-related headlines and baseless comments on social media are so offensive. I have devoted 20 years of my life to develop a level of expertise that allows me to influence products that impact our world, yet a 10 minute YouTube video with over 2.6 million views can have a bigger influence by spreading completely false information.

Myth: Carbon has “no strength in compression” – James Cameron, NYT

It is true that composites are well suited for structural applications where stresses are primarily tensile. This is why composites are used so heavily in high pressure vessels like hydrogen storage tanks. However, to say they have no compression strength is a gross overstatement and simply not true. Composites are typically anisotropic materials, meaning their properties are different in all directions. As a result, a material technical data sheet (TDS) will offer critical mechanical properties so that engineers can properly design and analyze structures. Here is a TDS for Toray T700S, a standard modulus carbon fiber used with epoxy to form a composite material. While the compressive strength (1,450MPa) is  only half the tensile strength (2,860MPa), it is still 50% higher than even the strongest grades of Titanium (980MPa). If you are designing a weight critical structure, which OceanGate claims to have been doing, you can further improve strength by using twice the amount of carbon material (it’s less than half the density of titanium!). This significantly reduces compressive stress levels (by over 50%), meaning you now have effectively 3-4x the safety factor vs. using Titanium. Now I’m not saying based on this alone they should have used carbon fiber as the pressure vessel, I am simply offering evidence as to why one would consider it and refute false statements that are being made by “experts.” Carbon fiber composites are regularly used in applications where compressive states of stress are prevalent, not the least of which are airplane wings from both Boeing and Airbus which see massive levels of compressive stress in upper skins and spar caps, potentially to the point of buckling under critical loads without failing. We have also shot composite rockets into space for decades, subject to major axial compressive forces from liftoff to orbit. And let’s not forget the F-1 Halo, which has been attributed to saving the lives of drivers during catastrophic crashes where extreme compressive loads are reacted. Whether the load is tension, compression, or shear dominated, it is critical that the designers/engineers size accordingly. If the Titan failure was attributed to the use of composites, I would not consider it a material failure, but an engineering failure.

The carbon wings of the 787 Dreamliner undergo ultimate load testing, putting upper surfaces and spar caps into a state of significant compressive stress.


Myth: Composites have worse fatigue characteristics than metallics. “Over the course of repeated pressurizations, they tend to weaken.” (CNN) Or, they suffer from “degradation failure.”

Now we’re getting into concepts that are much more complicated than simple tension/compression loading. However, once again this statement has caveats. First, I assume both of these experts are referring to fatigue, which is a technical term for reduced strength of materials following cyclic loading well below yield strength. Now an argument can be made that Titan was hardly fatigue-critical. Even low-cycle fatigue can require 10,000 load cycles. Those commercial aircraft wings I mentioned above will cycle 30-40,000 times throughout their life. Some structures see millions of cycles. So even before we get into the fatigue performance of various materials, it can be argued that with so few dives (unknown amount, but less than 200), fatigue is less of a concern for Titan than the static strength of materials. Setting aside the argument that this couldn’t be fatigue failure, because there simply weren’t enough dives to qualify as even low cycle fatigue loading, let’s address the fatigue characteristics of composite materials. Believe it or not, composite materials are actually known within the aerospace industry to have better fatigue characteristics than metals. Boeing selected carbon fiber for the fuselage of the 787 largely due to fatigue performance, the benefits of which resulted in lower aircraft weight, higher cabin pressures and humidity (more comfort for passengers), and longer intervals between major service checks. When loaded in tension, unidirectional carbon fiber can be shown to have less than 5% reduction in strength after 10,000,000 cycles vs. a 50% reduction in strength for Titanium. Once again factoring in the initial strength and density of carbon, you can see 6X fatigue strength by weight.

Composites Fatigue

Reduction in strength due to fatigue of various materials, up to 10 million cycles. Source

But carbon fibers are not always 0 degrees, and not always loaded in tension, and in some cases not even continuous. As a result, fatigue characteristics are much more complicated, and I highly recommend reading the above sourced paper if you want to better understand the topic. In short, the key with composites, fatigue, and “progressive damage” is keeping the load in the fibers (often referred to as in-plane stress), which can absolutely be done in a vessel subject to external pressures. It is important to note that design affects load path and level/state of stress, not the material. Good composite design will keep load out of the matrix, and avoid states of stress like inter-laminar tension, which I’ve discussed extensively as it pertains to hydrofoil mast design.  Once again, this boils down to design and engineering, and is not so simple as material selection alone. A good engineer can make any material work in any application, with tradeoffs of course. The vessel I designed for RipeLocker, which is subject to nearly 20,000lbs of load under vacuum, is made of recycled milk carton material (PP/HDPE) to enable higher-rate production and very low costs.

Myth: You cannot simulate composite materials with a computer, like you can metals or ceramics – James Cameron.

A lot has changed in the world of computing since James designed his submersible 30 years ago. He’s referring to Finite Element Analysis (FEA) which is a tool used by engineers to simulate the application of structural, acoustic, aerodynamic, or other loads to predict behavior before actually building or testing anything. FEA is an extremely powerful and valuable tool, especially for composite materials due the previously mentioned anisotropy. It’s so powerful, I spent $30,000 on the software to do it, which is nothing compared to what companies like Boeing, BMW, or Apple are spending on annual software licenses. I regularly simulate entire aircraft made entirely of composite materials subject to emergency landing conditions or high G maneuvers on my MacBook Pro. A simulation that used to take weeks or even months running on a Unix server can now be run while I sleep on a personal computer. The software has advanced as much as the hardware, allowing the input of much more complex material data (anisotropic properties and damage criteria) and individual ply layups to be incorporated into computer models. I can look at various states of stress throughout a structure, and determine where more material is needed, or ply angles should change, in order to meet structural performance criteria like strength and stiffness. I have literally spent my entire career simulating composite materials using finite element methods, savings months or even years from a design cycle by reducing the amount of testing and physical DOEs that would otherwise need to be executed. So for someone to say it can’t be done, is not only false but doing a huge disservice to me and other engineers trying o make a living convincing companies of the value of our services. It’s a hard task to be honest, because so few people understand the complexity of it, and can justify the associated costs, even though payback is massive. It is the anisotropic behavior of composite materials that make them so powerful, allowing engineers like me to truly optimize for multidisciplinary problems that have constraints from aerodynamic to acoustic.


Cedrus Torsion FEA
Cedrus Bending FEA

Finite Element Analysis was used extensively to optimize the stiffness and strength of this hydrofoil mast, eliminating the need to design, build, and test many prototypes.

Myth: Titanium is a wonder material and should be used everywhere, especially submersibles.

The use of Titanium, like all materials, offers benefits and drawbacks. When it comes to corrosion resistance and performance at elevated temperatures, Grade 5 Titanium can offer benefits over steel, aluminum, and CFRP (carbon fiber reinforced plastic). This being said, Titanium is both heavier than aluminum with less strength and ductility than steel. As shown above, it has half the strength of CFRP and almost 3X the density. I won’t get into cost, or the challenge with Titanium supply chains (hint: Russia), but these are the other main reasons I have had very few good applications for Titanium throughout my career. Jet engine exhausts, turbo fan roots or compressor blades, and other high-temperature, high-cycle applications seem to make the best candidates. Titanium has half the Modulus of Elasticity of Steel (112GPa vs. 190GPa), which means it will strain twice as much under a given load. Not necessarily a bad thing, but an important consideration in stiffness critical applications. For some reason, there is a common misconception that Titanium is a super material and should be used wherever possible. I’ve had many hydrofoil clients lose wings or suffer from loose hardware after swapping their stainless steel machine screws for Titanium bolts of an unknown quality. They are very surprised when I have to tell them that cheaper grades of Titanium (Grade 2) has an ultimate strength of ~350MPa, while an A4-80 (stainless) screws offers at least 800MPa tensile strength, which is only slightly below that of very expensive Grade 5 Ti. Factoring in weight, you get a property called “specific strength,” which is valuable for assessing material strength per pound. This is where Titanium can shine compared to steel, but when we’re talking about a 2oz machine screw, the weight benefits do not justify the costs.

Myth: You can’t join composites with other materials

Adhesively bonding dissimilar materials is hard. But again, to say it can’t be done, is simply false. Joints with composite materials are much different from metallics. Metal parts use nuts and bolts or screws, something human beings have done since 400BC. As a result, we are more comfortable with the use of screws to join materials together. Fasteners aren’t the best option for composites, mainly because you break the fibers when drilling holes and put more load in the weaker matrix (epoxy). Fortunately, thanks to major advances in chemistry throughout the last 20-30 years, modern adhesives can be used to join composite materials, and can lead to a joint that is actually stronger than the base material! It’s true, a well designed adhesive joint will fail the substrate, or adherend. A poorly designed joint will fail cohesively (adhesive is not strong enough) or adhesively (surface prep or adhesion is not good enough). There mare many new primers and processes available now strictly to improve the strength of hybrid joints between carbon and metals. Modern jet engines use hybrid joints on the leading edges of the fan blades with great success. But again, it comes down to engineering, not just material. Joining a stiff, low CTE carbon fiber vessel to a relatively soft, higher CTE Titanium dome certainly poses challenges. In this case, I may be interested in urethane-based adhesives which offer more elasticity at the joint than stronger, more brittle epoxy-based adhesives. Or maybe there’s a case for not using adhesives at all, and relying on traditional fasteners through a super thick composite laminate. Either way, joint design is incredibly complex and requires significant testing and validation, but a common practice in high performance structural engineering.

Myth: They should have just followed the regulations and certified or “classed” the sub.

Without getting political, there are understandable concerns around the design decisions made by OceanGate, especially with respect to safety. This being said, I want to clarify that it’s not so simple as many have made it sound. I can only speak to this topic through my experience in certifying aircraft and designing to the rigorous standards of the automotive industry. Commercial fixed wing aircraft are certified to FAR 14 CFR Part 25 airworthiness standards. These are laws, written and enforced by our government to keep us safe. They do cover materials, redundancy requirements, and general design guidance. They took decades to complete, and are constantly being revised. We are paying for them, and they cost taxpayers and the flying public billions of dollars. Obviously they are worth every penny, no question there. In the automotive world, there are FMVSS (Federal Motor Vehicle Safety Standards) and DOT (Department of Transportation) regulations that govern the design and manufacturing of our vehicles. There are no federal rules related to the design of submersible vehicles designed to travel to the depths of the Titanic, just as there are no rules for home-built aircraft which are simply tagged “experimental” and flown at risk all the time. There are no formally agreed upon and accepted regulations in place yet for electrically propelled aircraft or self-driving vehicles, either… which is part of the reason I left Silicon Valley 7 years ago. It will be decades before these technologies see significant real world adoption not because of the technology per se, but because of the challenges and costs associated with even setting up the rules of the game. How “safe” should a self-driving car be? How much extra range should a battery powered VTOL aircraft have? Should it fly over populated areas? How statistically significant should the tensile strength of the materials used in assembly be? There are so many questions and defining the answers is going to be expensive and time consuming. Nevertheless, there are companies and governments attempting to do just this, for the betterment of society. In the case of deep-sea submersibles used by a tiny fraction of the population, there is simply no payoff in developing a reliable path to certification. How many cycles? How deep? How do you test? It is up to the engineers to determine safety factors, design depths, the types of materials and manufacturing processes used. The same is true for Elon Musk launching and landing rockets-he’s making the rules, not our government. So the blanket statement that they should have “classed” or “certified” the sub simply shows a clear misunderstanding of the certification process in general. Again I am not saying questionable decisions weren’t made, or justifying their said decisions, but only trying to shine light on the challenges of engineering new, world-changing products.

There are some other benefits to composites that are important to mention. First, the manufacturing processes associated with composites lend themselves very well to these types of vessels/structures. Forms of automated fiber placement allow 1-piece structures like barrel sections of aircraft fuselages, rocket fuel tanks, and boat hulls to be produced with little to no waste using a layup optimized for the load path. Further, single piece structures eliminate joints and connections (screws, welds) which are common forms of stress concentrations. To be honest I’m not sure how you’d make a 5-person metallic submersible from a single piece of Titanium, as many suggest. You’d probably first need to design and build a CNC mill big enough to machine down a chunk of raw material larger than you could even transport. This would be orders of magnitude more expensive and time consuming, which is why it’s not done in practice. Secondly, if low cycle fatigue or abuse loads from transport of the submersible are of concern, there are many forms of non-destructive inspection (NDI) which can be used to identify potentially damaged (delaminated) areas within the laminate before catastrophic failure occurs. NDI can be conducted throughout the manufacturing process as a form of quality control, and/or used at various intervals of the vessel’s service life to ensure continued safe operation. 

In closing, I want to reiterate that I am in no way attempting to justify the various engineering and business decisions made by OceanGate. I know only what is in the media, as I didn’t design the sub, or know anyone involved in the design or business. I am simply offering supporting evidence as to why composite materials could actually a good choice for this application, in an attempt to refute many false statements and assumptions that are being made by “experts.” I am also trying to share some of the challenges involved with engineering and product design, specifically with respect to materials selection and manufacturing. I do not know what caused Titan to implode. No one does, yet. It very well could have been delamination within the composite hull due to LCF, but again it’s important to note that this would not be a material failure, but an engineering failure. I am amazed they’ve managed to recover so much debris, and look forward to more conclusive evidence of the failure modes. Until then, I think it’s important to not jump to conclusions but instead use this experience to motivate our next generation of scientists and engineers as we work to solve some incredible engineering challenges related to our life on planet earth, above and below the ocean surface. 


On a lighter note, I’d like to close with this classic clip; enjoy!

Thanks for reading, Kyle