Offshore Optimising
August/September 1998
The recent spate of carbon mast failures has left concerns amongst the generaf yachting public about the reliability of the concept and ultimately the material itself. This concern is justfied in most peoples' minds by the simple evidence on the water backed up by the high profile nature of the boats that have lost their masts. This perception is hard to ignore from the industry point of view. This article is an attempt to bring some rationality back in the debate and is based totally on real world failures and designs using a Windows based finite element model (FEM) so our findings have a real world application. This research was targeted primarily towards the most popular trend in carbon mast development that being the runneriess fractional rig. However, much of what is said here is applicable to carbon masts in general. First some background.
Carbon fibre's main attraction is its strength and low density. The density is around 55% that of aluminium regardless of the fibre orientation but the strength can be greater or lesser, being highly dependent on fibre orientation. For the comparative stiffness of a material, engineers rely on a value called the Young's Modulus or "E". Using an isotropic material (i.e., one that has the same properties in all directions) like the mast builder's favorite 6061 T6 aluminium the E value is a constant 69 Gpa. However for a typical carbon mast laminate it can vary from a typical high of 120 Gpa using high strength carbon in the vertical plane to 55 Gpa in the transverse plane. These values could reduce by as much as 30% if lower strength carbon is used. If you could achieve an E of 69 Gpa with carbon in all the relevant load directions then you could expect a carbon mast to have the same size and wall thickness' as an aluminium mast but to be lighter in section weight by around 45%.
In aluminium mast design different sections are compared by using a value known as moment of inertia or "I". This is a property of the section area not the material. This means that a cardboard object and a steel object of the same pattern can have an identical I value. For a valid comparison you need to add something that denotes the physical properties of the materials in question and this is the E value. If the E value is the same then 1 alone can be used to compare sections. If the E value changes or you wish to compare the capabilities of masts of different materials you then multiply the E value by the I value to create a new value known as "El". Using this concept one can approximately reengineer a known aluminium mast into carbon - calculate the E value for the carbon laminate in the direction of the applied load and then generate an I (by adding or subtracting material in the mold) to achieve the desired El. At least this is theory.
The main design load in the tube is compression. Compression in the mast section is developed from the vessel's righting moment acting through the rigging and this load achieves its maximum in the lower panel. It is also a load that is evenly distributed throughout the cross section of the tube. The mast also bends in the fore and aft plane but is restrained at the partners. This bend occurs upwind (for sail shaping) and downwind in the process of getting the top of the mast further forward to improve spinnaker efficiency. In the case of the swept spreader mast the forward movement of the lower panel is restrained by the aft running diagonals which are fixed at the deck while, with the inline spreader mast, the checkstays provide a similar restraint. We can all understand intuitively that if the mast is allowed to keep going forward it would, at some point, break through bending around the partners. In the swept spreader rig there is no visual evidence of the mast bending forward since the diagonals will not allow significant deflections.
However, if the load keeps increasing (from mainsail, vang and gooseneck) as the wind increases what appears to be a stable situation visually (i.e., the mast "looks" right) can become fatal as the loads are still accumulating, potentially heading well beyond the maximum assumed upwind load case. in fact, these off wind loads can be as much as 100% greater than the upwind load, in arts of the mast when the boat is being driven by mainsail alone off the wind.
These would normally be alleviated substantially in the in-line mast as visual examination of the mast would lead to more load being taken up on the checkstays to keep the system in equilibrium. In a swept spreader, runneriess rig the lower diagonals are doing this job but at a substantially less effective angle. Suddenly their relatively light upwind load case is overwhelmed and they prove inadequate for the task. The solution is basic - size these members on a downwind load case and you will find that they virtually need to be doubled in strength. This will reduce stretch (which helps the lower panel bending) as well as taking them away from permanent yielding from overload. The overloading of the diagonals also means that the maximum upwind compression around which the tube is designed is being exceeded and a substantial additional allowance in the section must be made for the extra loading.
When a bending overload occurs the El equivalence between our carbon mast and the aluminium mast goes out the window since a stress concentration occurs at the partners. This means that our primary loads are no longer shared throughout the cross section of the mast but rather are being concentrated in parts of the section at the partners. If these loads exceed the capability of that small area then that area will fail locally. Local failure usually leads to global and down comes the mast.
Why has failure at the partners not been an issue in aluminium masts? Go back to our El equivalence and you have your answer. Since the aluminium mast has a lower E the I needs to be higher which translates to a thicker wall by around two times. The local stresses generated by the partners are thus distributed over a greater area reducing the opportunity for failure. However, when you produce the carbon equivalent the El requirement generated by the proven aluminium design leads to a reduction in wall thickness of two to three times. Suddenly the loads from the partners are being applied into a much thinner wall as well as being in the weaker load axis of the carbon laminate.
Our conclusion is that a well designed swept spreader carbon mast for unlimited offshore use needs to look something like a traditional aluminium mast in its panel development even if the El values suggest otherwise. This means a more robust section from the gooseneck down to below the partners (not dissimilar in thickness to an equivalent aluminium mast) as well as a second panel designed for the extra off wind compression. The diagonals require independent sizing using an offwind load case. For the in-line spreader Carbon mast the bending stress at the partners is equally relevant and, while the tools exist on board to control it, we would strongly advise caution in this area as it may prove to be too critical for even a good crew in a longer race. Finally, appreciate that this article is for general consumption and the more technically minded reader may wish to go deeper or challenge our findings. We welcome this. Carbon masts are undoubtedly the way of the future but their development needs to be opened up if we are to avoid more failures and the ensuing risk to life, limb, and pocketbook.