Posted 11 March 2005 - 15:05
Upon popular (?) demand, here is the nowhere-near exhaustive, but nonetheless probably very exhausting, stressed skin structure discourse I threatened with in my prior post, so please bear with me…
The critical parameters for what static load (fatigue is another issue which would require yet another post!) a given panel of a stressed skin structure is able to carry are the Young’s modulus of the material (its stiffness), the yield strength of the material, its thickness, its area and the relationship between width and height of the panel, the boundary conditions (edge restraint) and panel curvature (if any). The intention, of course, is to select the optimum combination of these parameters that enables carrying of the load with the lightest mass of structure – so material density is thus also added to the equations.
Let’s have a brief look at each of these parameters to see how we can ‘play around’ with them:
Young’s modulus: Stiff means better – but by some interesting quirk of nature, the specific stiffness (i.e. modulus relative to density) of the primary structural metals/alloys, i.e. steel, aluminium, magnesium and titanium, is fairly constant. As increasing thickness helps exponentially, the optimum panel size is larger for the less dense materials, which is one of the reasons we see a lot more aircraft structure made from aluminium alloys rather than from steel or titanium, even though steel and titanium alloys can be produced with much higher specific yield strength than aluminium alloys. In applications where the geometry of the structure does not influence its stability, which basically means in tension members or short, ‘fat’ compression members, the higher specific modulus materials win hands down – but obviously not so on stressed skins unless we move to way stiffer materials such as carbon fibre.
Yield strength (and ultimate strength) : Obviously, the higher, the better – but again, more, weaker material of lesser density helps where the geometry of the structure is important, as in stressed skin structure, and thus again it is actually the specific strength that becomes important. If it is a stiffness-critical rather than a strength-critical structure we’re looking at, ultimate strength may actually never need to be approached. This often means that if the stiffness of the structure is adequate and high, then the strength will usually be sufficient – very few racing car chassis have actually failed globally from normal use, even though they may have been highly flexible, whereas whole wings of aircraft may collapse globally, but not before having deflected in the extreme. As an aside, most racing car chassis structural failures are local and relate to poor detail design at points of load introduction and transfer.
Thickness: This is one that really does wonders: As functions of thickness, and with all else being equal, weight increases linearly, but resistance to bending, and thus to buckling, increases exponentially. Thicker means better – but panel thickness may be traded off by reducing panel size and locating the material in stiffening members at the panel edges, so there is a limit to the optimum thickness that we bump into rather quickly. In this department, sandwich structures really show their worth: Obtain density by filling in the area around the ‘non-productive’ neutral axis with low density material with adequate specific strength and stiffness. Think of ‘Mallite’ and other sandwich structures here. Sandwich panels can thus be very large relative to monolithic panels for a given load.
Area: The smaller the area of each skin panel, the more stable, but the more stiffening members are required in a complete structure of a given area. As mentioned above, sandwich construction enables increasing panel area, and thus minimizing the number of stiffening members.
Height/width relationship: This is typically defined by the geometry of the overall structure and the direction of load. There are two basic ‘schools’ used in aircraft fuselages (which have more similarity with a racing car chassis than the wings do) : Wide frame spacing with short stringer spacing, versus short frame spacing and wide stringer (longeron) spacing – obtaining the optimum compromise also involves manufacturing cost etc. What one rarely sees is square panels – typical panel ‘format’ is approximately 1.4: to 3:1.
Boundary conditions: The more restraint against panel edge rotation, the better. Deep-section edging members (stiffeners or other substructure) with rigid attachment helps the panel immensely – but they add weight which contributes less from the load carrying capability of the skin as it moves away from the surface.
Curvature: Single curvature skins with the rules parallel with axial load helps panel stability immensely – rules perpendicular to the load are detrimental. Happily, the section sizes of stressed skins structures are typically such that a very small portion of the structure at its extremities will carry all the axial load, the majority of the panel seeing shear, where any curvature will always aid panel stability. Compound curvature panels are difficult to make, but due to their geometrical stability, fewer additional stiffening members are needed.
But how does all this lead us into a discussion about the merits of stressed skin construction relative to aircraft and relative to racing cars? Not sure after I read the above again, but please bear with me and I’ll try below:
A typical aerospace structures design exercise is to find the optimum panel/stiffener configuration. As aircraft have to be certified, produced and maintained at the lowest possible cost, additional important parameters are added to that task! For the racing car designer, certification used not to be an issue, but with mandatory crash testing it is now, and cost and maintainability were and are way less important to the racing car designer, even for ‘production’ racing cars, relative to the aircraft designer. Aircraft have to fly reliably with minimal inspection and maintenance for thousands of hours. Not so with racing cars – the chassis typically become obsolete after a few hundred running hours, they receive almost what amounts to a ‘C-check’ every few hours of running, and ‘lifeing’ of components is much more acceptable; in the days of riveted sheet aluminium chassis it was not unusual to replace chassis in the course of the season. Also imagine if airlines or air forces had the same staff-to-vehicle ratio as a current F1 team...not even the richest customers or governments could afford that. Another important difference between aircraft structure and racing car chassis structure is that the flexure of aircraft structures can be accommodated in a way that does not adversely impact flight performance at all (it can even be made beneficial by so-called aeroelastic tailoring), whereas racing cars need to be very stiff for the suspension to work (although active suspension could be made to accommodate a flexible chassis – within reason).
A very important feature of composite structure design is that you actually design the material and its properties concurrently and integrally with the structure. Using isotropic, metallic materials with well-known and well-defined properties makes calculation and certification relatively easy, whereas with composites it gets rather more complicated. This entails that, due to the stringent safety requirements needed by society to enable transportation of humans and goods in a mode where failure means gravity wins every time, the certifying authorities for aircraft are very conservative and that certification of a new material and/or structural concept is an expensive and drawn-out processes. Racing car designers have, or rather had, much more freedom in this respect. This, and other cost issues, had the implication that although adhesively bonded structures in general and carbon fibre composites in particular originated in the aerospace industry, in practical use and degree of structural refinement, the racing car industry has actually surpassed the aviation industry. Kilogram for kilogram, the modern carbon composite racing car chassis is structurally more efficient than practically any aircraft flying or being built today – but also more expensive to procure and maintain.
Although somewhat off-topic, another problem with aircraft, especially combat aircraft, is that they have a very high density of internal equipment that requires easy access for inspection and maintenance. This means removable structural and non structural openings, covers and doors abound, and such are anathema to structure in general and composite structures in particular. Once you have designed a cockpit opening into a racing car chassis you don’t really need to be able to access anything other than the fuel bag and the pedals. In order to overcome this problem in aircraft structural design, there is a trend to attempt to make internal equipment super-reliable and self correcting so that it can be built into the structure with no need for subsequent access.
There are indeed similarities between aircraft structural design and racing car chassis design – but as we have seen there are also some very important, fundamental differences: Both types of structure are dependent on light weight, or rather high specific strength and high specific stiffness, in order to achieve the desired dynamic performance. Aircraft are more strength-critical and durability-critical, and structure buy-and-maintain cost constitutes a significant proportion of overall vehicle and operating cost, whereas racing car chassis were and are more stiffness-critical, although, happily, increased focus on crash safety has put strength at the top of the agenda, and the cost of the basic chassis structure and its maintenance constitutes a much smaller fraction of the overall initial and operational cost of the vehicle.
I think the differences in requirements outlined above, especially when cost is involved in the calculations, is one of the reasons that aircraft stressed skin construction methods took so long to catch on, and are still not universal. The increased affluence of society spills over into racing, but more so at the top level than at the grass roots. As related many times elsewhere, welded tubular steel space frame chassis with separate non-structural body work are so much easier, quicker and cheaper to repair than either aluminium or carbon composite stressed skin chassis, but at the ‘cost’ of higher weight they can actually be made both adequately stiff and quite ‘crashworthy’ relative to the modern carbon composite chassis.
Specifically regarding adhesively bonded structure: Again this was born in aviation (think de Havilland and Ciba “Redux” epoxy, before then the casein and phenolic glued wood and plywood structures). The big advantage of bonding is the elimination of all the stress raisers that fasteners and their holes constitute, which primarily improves durability (fatigue resistance), but also enables some weight saving by saving fasteners and local beef-ups around fastener holes. Bonding allows a smoother distribution of load transfer in a joint. And as GeorgeTheCar pointed out, sandwich and laminated structures need adhesives for their very existence.
The difficulty with bonded joints is again in the details: The design needs to be right, parts fit needs to be excellent and everything must have the proper surface preparation and be absolutely clean. The bond doesn't like out-of-plane loading in general and 'peel' in particular - hence the odd 'chicken' rivets still often seen (soemtimes also used simply to clamp the bond line ends, and/or to clamp during cure). Add to this the need for autoclaving for the really high-performance adhesives, and the fact that the integrity of the joint is difficult to verify without expensive methods and equipment – none of this is really within the realm of the ‘backyard special’ manufacturers of yore. Proper rivet installation can be verified visually, and a bit of grime doesn’t matter too much. Actually a riveted sheet metal chassis need not be more difficult or expensive to make than a welded tube space frame – but damage always affects a larger section of the stressed skin chassis making ‘scrap’ more likely than ‘repair’. A very popular compromise that (almost!) combines the advantages of the space frame with those of the stressed skin structure used in some lesser formulae and sports racing classes was, and is, the ‘stress panelled space frame’ – i.e. a welded tubular steel ‘skeleton’ (bulkheads and longerons) with blind-riveted sheet metal skins providing the shear resistance normally provided by the diagonals of a truss structure. This type of construction invariably provides some structurally redundant material, so it will be heavier than a ‘pure’ stressed skin design, because more of the material is not located right at the surface where it does most good, but it often provides a more crashworthy design, and is also typically less prone to fatigue failure than ‘pure’ aluminium skin designs –perhaps this is why this method is still quite popular. ‘Tis also often dubbed ‘Ferrari aero style’ chassis, BTW.
Oh stop me! – I must finish this long rant…
I personally think that many of the structural performance and durability problems with sheet aluminium chassis of the 1960’s and 1970’s was due to using too weak alloys and tempers, too large unstiffened panels and long torsion boxes, inadequate rivet types, sizes and spacing, poor hole preparation, poor fit of parts and inadequate surface preparation when bonding was used, poor load distribution from load introduction points, lack of de-burring of holes, chips and swarf between mating faces, inadequate reaction of sudden changes in load path geometry – in short due to inadequate detail design and inadequate manufacturing quality. But one way of obtaining lightness is to stress the structures very highly and live with fact they don’t last long. F1 teams could live with this, few else could and aerospace certainly could not. Conventional riveted aluminium alloy sheet metal structure can be made to last for decades and thousands of hours of use. There are a lot of very old aircraft out there to prove that point. If the racer’s had the incentive they could do it also – but always at the cost of a few kilograms of added weight. That may be acceptable in aerospace, but racing is all about winning a competition, so racers just don’t want to sacrifice lightness for reliability to the same extent the aerospace guys have to – to them competition is about cost as much as anything else. That, to me, is the biggest difference between aerospace and racing.
Apologies for yet another long post…
George,
Re the benefits of liquid engineering, aka Loctite, the late, great Carroll Smith addressed this in some detail in one of his excellent books.
Allen,
Re Indy Car chassis I'll have to do some research and think about it for a while...