Tuesday, February 15, 2011

Self-tacker design

To work out the radius of the self-tacker track i started with working out what i wanted from it as follows:

  1. The leech tension to remain the same on all positions of the track: This was so that i could adjust the car position without having to adjust the jib sheet as well, so that simple tweaks could be made.
  2. The foot remains that same tension on all positions of the track: This was to keep the foot tension on the jib an independent variable from the outward position of the car. I adjust the foot tension by the position of the jib sheet on the clew board.
Some people like to set the boat up so that the leech tension decreases as the car is eased or have the foot tension increase as the car goes out. I'm personally not a fan of this as you are effecting two parameters with one variable and i feel you could end up in a position that you need to constantly tweak two controls just to setup the one parameter you want to vary.

If order for these criteria to be meet the jib car will need to follow the jib clew as it moves exactly. To do this it means the distance from the head to the jib car must remain the same as it travels. Secondly the distance from the tack to the car must also remain the same. This gives us the simple problem where we have a triangle with constant length sides that pivots about the forestay.

Therefore it can be found quite easily. Either by the geometry of the jib and then find the line length that runs through the clew and is perpendicular to the forestay. Or once the boat is rigged up you could just use a tape measure and builders square to find the distance perpendicular to the forestay.

The only other things you need to consider is to account for the fact you self tacker track will sit something like 30mm above the deck at the centreline and to allow for pulley's etc.

Also if the track has a slightly smaller radius than needed the sheet tension will increase slightly as it passes the centreline during a tack. If the track is on a slighlty larger radius then the sheet will loosen slightly during the tack. I personally favour the track to be on a slightly larger radius as i don't want the sheet to tighten up and jam during a tack.

From memory my current track is a 1580mm radius

Monday, February 14, 2011

Self-tacker track construction

In order to fit a self-tacking jib to the boat i needed a method of curving a track for the jib car to run on. On other boats i have seen this done in a couple of ways:

  1. A ronstan I-beam (series 19) track is ordered from ronstan to be bent to a given radius, which can then just be mounted to a support at each end and a support in the middle to prevent the track from twisting or warping off its supports. (e.g. 3 mounting points that are not co-linear are the minimum required to make the mounting stable). When i investigated this the track was roughly $300, with a lead time of 3 weeks and the car cost around $170.
  2. A thinner track like a ronstan series 14 BB track or a Harken Micro CB track is flexed to a rigid backing or a series of brackets attached to the deck.
  3. The track as above is directly attached to a concave deck that matches the curvature of the track.
I chose to use a rigid backing support that i could then force the bend in the track as it was bolted to the support.

To make the support i screwed two pieces of 18mm form ply together giving a piece of timber 36mm thick. Next the required radius was traced using a pencil and piece of string. This was then cut out carefully with a jigsaw and neatened up with a belt sander. To give a radius on the edge a quarter round radius bit was used in a router (with a ball bearing).

This mould was then covered in packaging tape as a mould release and two strips of 10mm foam where cut to cover the mould.

The laminate is described in the image below. First the two layers of glass are laminated onto the mould surface with a +-45 degree orientation to the length of the mould. Next two layers of 300gsm UD carbon that is only wide enough to cover the flat surface are laminated. If the strips are wider they will not stick to the sides of the mould.

There are several reasons for laminating the glass on the -+45 deg (or carbon if you wish, i would still use two layers ontop and atleast 1 on the bottom).
  1. When the fabric is on the -+45 deg it wraps very easily around the corners of the mould and also traces the curved radius of the track well. Otherwise you need to place cuts all the way along the edge if the fabric is on the 0/90 deg
  2. When this beam is in bending the one outside skin will be in tension while the other is in compression. If there were no skins on the side the foam core would need to carry a shear load between the two outside skins. Since foam has a low shear modulus (stiffness) the will deflect significantly. Since woven fabrics have there highest shear modulus and strength when the fibres are on a -+45 deg it is a good idea to place these skins on the side as it will give you a stiffer and stronger beam.
  3. Unidirectional composites are generally stronger in tension than in compression because when the fibres are under compression the individual fibres tend to buckle. In order to improve this compressive strength a laminate on the -+45 deg to the unidirectional fibres should be used. This is because the cloth layer on top allows the load to be passed from one fibre the neighbouring fibre through a shear stress which is taken by the woven fabric. I'm fairly sure that i read somewhere this can improve the compressive strength by 10-15%.

Next spread a runny bog of Q-cells and epoxy resin onto all surfaces of the foam and place them on top of the laminate. Then continue to laminate the rest. Once the peel ply is placed over the top carefully place strips of packaging tape over the laminate starting in the centre and working towards the ends.

Once cured it can be easily trimmed to length and the track is bolted to the support. Care needs to be taken when clamping the track to the support. If you place a clamp in the centre and tighten it up you will buckle the track and place a bump in the track, which can stop the car from moving.

Bowsprite idea

Also thought i would mention this,

Trev Fay pointed it out to me at the nationals in Adelaide mentioning it was one of the best solutions for it he has seen. I can't remember which boat was using it, though i think it is a very elegant solution to the problem.

When setting up the tip of the bowsprite you want it to do a couple of things;

1. Be able to get the spinnaker tack as far forward as possible, as the bowsprite is limited in length.
2. Be a low friction setup for the strop to move through
3. Lightweight, since it is at the extremity of the boat hull e.g. effects the rotational inertia of the hull to pitching and turning.

I normally use two sheave boxes, which cost about $20 each and you need to reinforce the tip of bowsprite as you cut most of the material away to fit the sheave boxes.

The idea used on this boat i saw used a turned plastic plug in the pole tip with a hole straight through the tip. The ends of the hole where nicely rounded to avoid chaffing the rope and the plug was made from a nylon or low friction plastic.

I image this could be made to be very light weight if the inside of the plug was hollowed out.

Vang Design

I first saw the lever style vang on a forum after a 12 ft skiff had trialled the setup. Shortly after i saw it on some international 14's. The advantages that i saw immediately were the increase in cockpit space for the crew over a conventional vang system and (for my boat particularly) the clearance for the centreboard to be raised downwind.

Image 1: 12ft skiff design, i think it was built around 2006
Image 2: i14 design, possibly built around 2003
Images from (http://forums.sailinganarchy.com/index.php?showtopic=31443)

I began to look at this design concept for my older boat "Heart Shaped Box" and began a basic force analysis to compare a conventional setup to a lever style and the loads involved. This was fairly basic and going over it now i can see that there are several errors but i believe it represents a reasonable comparison between the two systems. The assumptions that i made were a 50kg load along the leech of the main sail and i only considered the lever vang forces parallel to the boom (not accounting for misalignment, which I'll explain later)

The results of this analysis was that an axial load on the purchase system needed for each system was as follows at 50kg of leech load.

Conventional = 1805 N (184 kg)
Lever = 1345 N (137 kg)

This gives a 25% reduction in the load at the rope. Meaning that where a 16:1 purchase was used before a 12:1 system could now be used for the same power. Having less purchase means less excess rope in the boat when the vang is sheeted on hard giving less rope to tangle in the boat.

While i was looking at this system a cherub at Lane Cove 12ft skiff club "Ajax" trialed the lever vang systems as they had played with the system on a 12ft skiff. After discussing with them to mentioned that you need to be wary of the sheeting angle at the tip of the lever as it can place a vertical load on the goose neck and this needs to be understood to make sure it isn't neglected.

To understand this lifting force you need to see that the boom/lever member is pivoting about the goose neck and the point which you are applying to load is the end of the lever. When this system is pivoting the tip of the lever will trace an arc with the centre point being the goose neck. If the load at the tip of the lever isn't applied tangent to this arc then the force can be broken into a vertical and a horizontal force. As this misalignment gets larger the vertical component starts to grow, but is still fairly small.

The next this i looked at the was the clearance on the centreboard as on my older boat the centre case was fairly close to the mast meaning that there was not a lot of room between the vang and centreboard. It showed i could get a significant increase in the clearance.

When it came to fitting out my new boat i looked at how i would actually set up the system. In order for the vang to work properly the leech tension should remain the same as the boom is eased through its travel. Since the boom is pivoted about the goose neck if the ropes on the end of the lever is pivoted about a similar position on the mast then the leech tension should remain the same. I was first concerned about the bending this would place on the lower mast and how this would effect the mast bend further up the rig. So i looked at options for having the pulley systems attached to the hull rather than the mast. The image below shows different loci that i had traced from different ways of wrapping the ropes around the mast to approximate an arc from the mast. The blue lines show that the system followed an arc about the centre of the mast until one of the ropes stopped touching the side of the mast, at which point it started tracing an ellipse from the pulley on the hull and the mast. I felt that this would approximate the arc enough to be a practical system.

These images below show the setup that i used. The lever was made from carbon over a polystyrene foam core with a section of windsurfer mast as the main structure inside. This was don't to make it as stiff as possible as flex in the boom or lever results in a loss of leech tension when the boom moves from dead centre to slightly eased as the pulley's at the boom tip stop taking the vertical load of the leech.

The purchase i went with was a 3:1 followed by a 2:1 around the mast, giving a 6:1 all up on the lever tip. Although this was lower than first calculated it was decided to try it as the old vang system was rather easy.

After sailing with this setup for a little while now there are several changes I plan to make. First of all, we are struggling to get the height out of the boat that others have and I'm starting to feel that this is due to the leech on our mainsail. The vang is slightly underpowered meaning that in the stronger breezes it is hard to get a good trim on the leech.

Secondly the way the vang wraps around the mast gives some deviation from an arc so i will move the pickup point to the back of the mast. As a small deviation at the lever tip is multiplied by four at the tip of boom (ratio of lengths) so 2 mm at the lever = 8 mm at boom tip, which will reduce the leech tension quite a bit.

1. Power: I will increase the purchase of the system to 10:1 0r 12:1 from the 6:1 i originally had.

2. Pick up points and locus: i will move the pick up to the back of tha mast and if mast bend is excessive i will place a strap/strop to the hull forward of the mast.

Other than this the vang shows promise, giving good clearance to the centreboard and more room for the crew. The lever i added to my boom is on the heavy side (about 1 kg), as it was placed on the boom two days before the nationals and i didn't want it to fail. So improvements could definitely be made there.

just a few more photos

Tuesday, February 1, 2011

More Bulkhead Repairs

After returning from the nationals we identified another small crack in the bulkhead/deck joint but this time it was much closer to the centre line of the boat. Upon closer inspection the crack occurred at the ply drop of a unidirectional layer from two layers down to one layer, yet still with a -+45 deg carbon cloth patch covering the ply drop.

The crack is clearly a compression failure and as can be seen even propagates past the carbon fibre material and into the bog covered fibreglass cloth each side.

To me this suggests that only 1 layer of carbon uni at 50mm wide is not sufficient to carry the loads running through this part of the hull. While considering that i suspected the load in this bulkhead/deck joint to diminish as the distance from the side stay increased (spreading the load into the adjacent bulkhead and deck material). This decrease is either non-existent or smaller than i anticipated. Also it is clear the failure occurred at the ply drop, which is a known stress concentration point. In future i would recommend that ply drops in a similar situation should be tapered (e.g. cut the end of the strip to become a point) as this should spread the stress concentration over a longer distance and prevent a straight line that is perpendicular to the load path being formed.

Also noted was that when i peeled up the damaged area that the epoxy did not stick particularly well the to the bog/filler that was next to the laminate e.g. from the old paint and filler. In future i think that needs to be sanded back further and more attention paid to make sure that all "stop putty" is removed with acetone before laminating.

I suspect that the reason this is happening is for two reasons, both involving the stiffness of the structure.

1. As mentioned earlier, stiffer structures generally experience higher peak shock loads as the structure deflects less when loaded (due to being stiffer) the time over which the load is applied and reacted to is reduced e.g. a shorter impact time. This gives a higher peak acceleration to the overall structure, which for a given mass gives a higher force (F=ma). Therefore a stiffer boat with stiffer rigging is going to suffer higher accelerations when slamming on waves. The next question is whether my new boat is actually any stiffer than other boats out there at the moment. I feel that i could be but its rather hard to prove or test and therefore hard to judge with this has a significant effect or not.

2. The second reason i feel this has occurred is due to the relative stiffness of the laminates used in and around the bulkhead/deck joint. Woven fibreglass cloth laminates have an elastic modulus (stiffness) which is drastically lower than carbon fibre fabric or unidirectional tapes as per below.

Young's or Elastic Modulus:

Woven E-glass (0/90 deg) = 25 GPa
Woven E-glass (-+45 deg) = 12.3 GPa
Woven Carbon (0/90 deg) = 70 GPa
Woven Carbon (-+45 deg) = 17 GPa
Uni Carbon (0 deg) = 135 GPa

Source: (http://www.performance-composites.com/carbonfibre/mechanicalproperties_2.asp) 01/02/2011

Now assuming that the E-glass fabric each side of carbon fibre strips is at 0/90 degrees then the unidirectional carbon is approx 5.5 times stiffer then the glass next to it. I feel that it is this reason that very little load is being transferred into the deck or bulkhead but instead travelling along the much stiffer carbon. This is because as the carbon compresses and strains the glass next to it strains an equal amount but because of the low stiffness it only corresponds to a low stress level and therefore contributes to very little load carry ability of the structure.

If this fibreglass is in fact not on an 0/90 degree angle the situation only gets worse until the angle reaches -+45 degrees when the stiffness difference becomes approximately 11 times...

With this now in mind i need to consider how i can make this carbon fibre laminate strong enough to solely carry the rigging loads and allow it to transfer to the fore stay without just moving the problem further along the boat.

Since the failure occurred in a single ply of unidirectional carbon yet not in the section with a double ply it would be okay to assume that this would be sufficient. Though considering this appears to be quite highly stressed and it is unlikely that the boat has experienced its highest loads that it ever will. With this in mind and also the effects of fatigue to be considered (although UD carbon has an excellent fatigue life) it would be a good idea to increase the factor of safety. Therefore i will try to locate the previous ply drops and rejoin them to give a 3 ply laminate to carry from side stay to centre line on the deck on both sides.

From here i will need to try to "dissipate" this load to the foredeck and the "V" beam underneath the deck. To do this i plan to arrange a patch or two of carbon cloth at different fibre orientations to try and spread it over a large enough area in the E-glass deck to carry the load.

Otherwise the next step would be to run carbon all the way along the deck to the fore stay mount. Though this would be rather undesirable as quite a large area would then need to be faired in and re-painted, which would be quite a major repair.

I will try to document the repair process as much as possible when i undertake it later on in the week.


The preparation for the nationals was very much a rush job considering the late completion of the boat and the fact that uni studies where pulling me a away from tuning and finishing the boat.

The rig was tested in the front yard for the bend characteristics of the mast under different leech loads using wire, digital scales and a block and tackle. A tight string was then run from tip the goose neck and measurements every 1/8th of the luff length taken. This gives useful information to the sail maker for determining the luff curve.

The data in the graph above shows the curves produced for different leech loads and lower stay positions e.g. (tight or loose). From the graph you can see that the difference between tight and loose lowers is quite substantial giving approximately 20-30mm difference in the point of max deflection and the position at which this point occurs. This data alone for me suggests that easily adjustable lowers on the water would give quite an advantage on days where the weather is quite unstable. This idea will be explored later on. Note: Data is for the 40mm ID series High modulus "Stiff" rig from CST made in late 2010.

After this it turned out that the rig performed rather closely to the older rig off Heart Shaped Box and it was therefore decided that due to the lack of time left and the close fitting of the main we would leave it as it was and not build a new one. This is certainly an area that can be improved on this boat though.

Overall the boats performance at the nationals was reasonable and certainly developed significantly at the end of the regatta, but our major downfall was lack of training/tuning and sailing in the new boat. Our sailing skills where not up to scratch, which cost us a lot of places and prevented us from getting the most from the boat. In terms of the performance of the hull, it appeared to nose dive no more or worse than the current Matthews hull design and held good pace downwind. Upwind pace was good but to combine this speed with height proved to be difficult. By the end of the regatta the boat had good upwind height and pace but it wasn't exceedingly quick, so i feel there are still plenty of improvements to be found.

Pushing the boat hard to the wind mark in a large chop and 25 knot winds

approaching the mark

Cruising along it the lighter breezes

Bulkhead repair

After the sail at lane cove sailing club I noticed that there was an uneven surface that had appeared around the port side stay mount. After further investigation it turns out that strands of unidirectional fabric buckled and were poking up through the filler and paint layers. Also the plain weave fabric closer the stay mount had also buckled.

This damage was more evident on the port than the starboard side, yet since both sides were pretty much symmetrical i first thought it may have just been voids from manufacture since this area was not originally vacuum bagged.

Upon further investigation the unidirectional fabric had buckled where it did not have a covering layer of fibreglass or carbon cloth. Yet in areas the carbon fabric had also buckled in a few places. This suggested that we had underestimated the loads on the bulkhead in compression along that top surface/corner.

Yet speaking to other class builders they had mentioned that they usually just put a simple layer of carbon cloth over the corner. My laminate was two layers of glass cloth and essentially two tapes 50mm wide of uni carbon running along the corner. With carbon cloth closer to the side stay mount where the load was expected to be higher.

After further thought i concluded that if the failure was not from voids and from the boat experiencing higher loads than expected. Then it would need to be strengthened significantly since the boat had only sailed in moderate conditions. This higher load could potentially be from shock loading, as a stiffer structure will generally experience higher peak loads during shock loading when compared to a "softer" one. So at least on the positive side it my suggest that the boat is quite stiff :)

To bulk up the laminate I used 3 layers of 300 gsm uni-directional carbon 50mm wide running from the chain plate towards the centre of the boat, and tapered each layer to avoid a large laminate drop or stress concentration. Then uni strips from the ratchet blocks and possible side stay positions the meet up with the main laminate.

Then to improve the compressive strength of the unidirectional laminate a covering layer of 200gsm carbon fabric was applied at the -+ 45 deg to the uni-directional laminate. This improves the strength by allowing compressive stresses to transfer from fibre to the fibre next to each other via shear stresses through the 45 deg laminate, or to simply support and stabilise the unidirectional fibres .

The laminate was applied using the "Poor mans Pre-preg" technique and then vacuum bagged to the hull to improve consolidation and reduce void content.