11th Bass Strait Voyage - Software Bugs and Good Luck

Wednesday 17/6/2026 6:30pm. Launch of Voyager 2.9 on a quick 24 mile / 24 hour voyage across the edge of Bass Strait from Torquay to Rye Ocean Beach. 

The vessel was fairly much the same configuration as the previous Bass Strait voyage, in late 2025.

Final Checks Before Departure to Launch Site

Ready for launch at Torquay 6:30pm 17/6/2026

The vessel made it through the destination, but it took 59 hours, rather 24, and took some large diversion on the way.

This voyage revealed new software issues. 

The image below shows the vessel heading eastward, in a generally northerly wind, slowly drifting out of the corridor, around a third of the way across.

The wind was fading, and the vessel was unable to make way in the light conditions, possibly with tidal influence.

The Voyager OS software does change it behavior to cope with course deviation outside the corridor.

This is defined as Cross Track Error (CTE) exceeding Maximum CTE (MaxCTE), the boundary.

The vessel continued to drift out of the corridor, beyond 150% MaxCTE, where a further behaviour threshold commences.

When the wind returned, it failed to return toward the rhumb line, but continued further off course.

Analysis of the logs revealed that the vessel  had adequate wind power to sail, but was running on port tack and refusing to head toward the rhumb line, despite the fac that it could have, if it tried.

I called this the "stuck running" bug, where it was in running mode, and refused to leave this mode.

Good Luck #1

Eventually, the wind backed to north west, which allowed the vessel to successfully make the centre waypoint, while "stuck running".

Once it returns within 25% of MaxCTE, all flags related to exceeding boundaries are cleared and the sailing algorithm returns to normal.

Voyage from Torquay to Rye Ocean Beach, including excursions

As the vessel continued eastward approaching the destination, the scenario repeated.

The vessel drifted out of the corridor beyond MaxCTE in light conditions, and continued to drift south in the light north westerly wind.

Eventually, the vessel drifted several miles south, when the wind returned. 

but "stuck running" meant it wouldn't return to course.

Good Luck #2

The wind did come in from the SSW at around 30knots. 

This allowed the vessel to run back to the final waypoints with good speed and good control.

The image below shows the vessel approaching from the south and then heading eastward to the shore.

Landing at Rye Ocean Beach 5:30am 20/6/2026

Software Bugs

"Stuck Running" Bug

Analysis of the code highlighted several issues related recovering from values of CTE greatly exceeding MaxCTE.

The primary bug, the "stuck running" bug is caused by some specific maneuvering cases not being covered.

When the vessel is running down a course, it is forced to gybe back and forth to avoid sailing closer to dead-downwind than around 30 degrees. 

This is for reasons of speed and more importantly to prevent downwind rolling (which is a problem that afflicts monohulls with self-trimming wing sails).

The downwind tacking behaviour generally consists of tacking downwind from one boundary to other. And in each case when a boundary is reached, the mode changes to running on the other tack.

The "stuck running" bug was complex to unravel, but it in simple terms it was related to a problem where the vessel is outside of the corridor, running on a divergent tack, away from the course, and never meeting the boundary.

This is now believed to be resolved, by ensuring that the gaps in the logic are covered to ensure the vessel tries to return to the course as one of its highest priorities.

"Waypoint Reached" Bug

This has not been a problem, but this latest voyage highlighted a potential problem.

A waypoint is considered passed, when the vessel passes across a line perpendicular to the course, that bisects the waypoint circle. The vessel must be within the circle as it crosses the perpendicular line to be considered past the waypoint.

This means the vessel must be within MaxCTE to do that.

A situation may arise where the vessel is outside of MaxCTE as it passes the waypoint.

It may be pointless, fruitless and counterproductive to force the vessel to pass within a distance of MaxCTE, when the vessel is already outside MaxCTE.

Instead, a change has now been made to remove the requirement to be within the MaxCTE distance of a waypoint when crossing the perpendicular line, for it to be considered passed.

Of course, the vessel will always try to pass precisely through each waypoint, but if it is push past a waypoint outside of MaxCTE, there is little point trying to go back.

 Sat Comms Antenna Update

Testing of the satellite communications has proceeded over the past couple of months in different circumstances. 

Australia is in the Band 255 region for the Skylo NTN service. This band is close to the GPS L-Band.

I've had reasonable success using a passive GPS Patch antenna with the Murata 1SC NTN module

But reliability for telemetry was only adequate. In practice, around 1 message in 3 may be lost.

I purchased a low-cost PCB Antenna intended for use with Inmarsat through AliExpress.

This antenna is intended for the Band 255 L-Band.

This yielded significantly better performance than the passive GPS Patch antenna.

It is larger, but being flat it does not occupy much volume.

Inmarsat PCB Antenna sold through AliExpress for under $10

The SMA Connector is mounted on the top of the board. I moved it to the underside of the board to allow the PCB Antenna to be placed against the top of the interior of the equipment housing, occupying the minimum amount of space.

Inmarsat PCB Antenna installed within the Equipment Housing

On water tests have been limited. But so far results are good, with zero lost messages with reporting every few minutes over several hours.

On water testing - good results so far.

 Satellite Comms - another try

I have previously integrated 2-way satellite IOT communications into the Voyager Sailing Drone design.

Swarm Space 

I started with a modem for Swarm Space. But before I commenced integration they announced to imminent shutdown of the constellation. I did nothing more with this modem.

Astrocast

I purchased the Astronode S+ device from Astronode through Mouser.
The Astronode S+ is a small-footprint, low power and very effective IOT modem. Their constellation consists of a few LEO satellites, so they are visible only a few times per day.

Once integration was completed, Astronode advised me of the "commercial service contract" fee of US$500 per month.
For some reason, the only fees documented on their website for connectivity were the data usage fees ranging from US$1.30 to US $6 per month.  
A fee of $500 per month changes from being a reasonable hobby project to being only a serious commercial enterprise.
As at Feb 2026, it is unclear if Astronode is still a trading entity.

I have been continually monitoring the Satellite IOT market looking for low power, small footprint devices with low cost connectivity and data usage fees.

Monogoto with Murata and Skylo

Then I happened upon a YouTube video by Laurens Slats from Monogoto.
They provide a service marrying together the Murata 1SC NTN modem, with access to geo-stationary satellites via Skylo. The costs are under $10 per month for my usage, so it fine for hobby projects.
Using geo-stationary satellites means that the satellite is always accessible (assuming it is not obscured).

The Murata 1SC is available within an evaluation board.

This costs around US $100 through Monogoto. It is small and low power.

Murata 1SC Development Board

The Murata 1SC Dev Kit does have some design problems which compromise its use when deployed in an operational system:

• It lacks any mounting holes, and so must be clamped in place.
• It has level-shifting interface circuity on board, which consume just over 20mA while the board is powered up. The Murata 1SC does go to sleep and consumes very little power, by the persistent 20mA minimum is the largest current sink in the Voyager Sailing drone, with the exception of the steering servo.
• The Murata 1SC supports 3.3Vdc, but the Dev Kit only support 5Vdc.
• The main connector looks like a standard 16-pin IDC ribbon cable connector.
  But its not. It has a 2mm pitch, rather than the more standard 2.54mm or 0.1" pitch.

In future, I will update the Voyager Controller board to provided switched 5Vdc to allow for powering off the whole Dev Kit board when idle, to significantly reduce power consumption.

The first step was addressing the main issues of installation on a sailing drone, being the antenna.

The antenna supplied with the kit is a multiband antenna, designed to address the different bands used by the Skylo service around the world.

The multiband antenna is pictured below. It doesn't easily fit with the existing equipment bay, and would need considerable effort to deploy and also handle immersion in sea water.

Multiband Antenna included in the Dev Kit

Australia is in the Band 255 region for the Skylo NTN service. This band is close to the GPS L-Band.

I wanted to see if a standard GPS patch antenna could be used with the Murata 1SC Dev Kit in Australia.

Of course, a standard GPS patch antenna is an active device.  I stripped out the active components and attached an SMA pigtail directly to the antenna element to allow for testing with the Dev Kit.

I tried this process on two different sizes of GPS patch antenna.
The 25mm passive patch antenna works well.
The smaller patch antenna could not establish a connection, and was of no use.

Testing with Passive Patch Antennae.

The Murata 1SC Dev Kit mounted above the main board within the equipment housing, including the 25mm passive patch antenna.

The Murata 1SC Dev Kit lifted up on the hinge to reveal the main board below.

I found the documentation of the AT commands provided by Murata was only a subset of the commands available.
The additional AT commands to aid the development were obtained from here:
we-online.com/components/media/o691492v410 Manual-um-acm-adrastea-i-2615011136000 %28rev1.2%29.pdf

 Testing in the back yard is generally good with mostly reliable communications.

I still need to perform on-water testing to ensure good communications while under sail on the local lake.

 

I have set up dedicated UDP listener software to receive and decode the telemetry packets.
At this stage I’m sending around 60 bytes of data, at regular intervals.

It then checks to see if there are outbound commands queued up, and waiting to be sent.
This allows for commands to be sent to change the mission waypoints or other parameters.
They can only be sent in response to the vessel sending a periodic telemetry update.

On-water testing results to follow...

Measuring Angle of Attack of a Self-Trimming Wing Sail

An important performance characteristic of a self-trimming wing-sail assembly is the relationship between trim-tab deflection and the resulting angle of attack (AOA).
Within the normal operating range, this relationship is approximately linear.

A practical method for quantifying this relationship uses the existing control and sensing capabilities of the sailing drone.
The onboard system can both set the trim tab to a defined angle and measure the corresponding equilibrium orientation of the wing sail relative to the wind.

Calibration Method

A dedicated calibration procedure was implemented within the control software.
During calibration:

1. The vessel is held stationary in a steady, uniform breeze.
2. The software sets the trim tab to a known deflection and measures the resulting equilibrium AOA, averaged over several seconds to reduce noise.
3. The trim tab is then set to the same angle in the opposite direction, and the new equilibrium AOA is recorded.

The effective change in AOA produced by the trim tab is calculated as half the difference between these two measurements.
This process is repeated across a series of trim-tab deflections to construct a calibration curve of trim-tab angle vs. resulting AOA.

The illustration below shows a sample measured response of the wing sail to a reversal of trim-tab deflection:

• On the left, the wing has rotated to an equilibrium position 52.5° from the reference line.
• On the right, with the trim tab reversed, the wing equilibrates at 22.6°.
• The total change in wing angle is 29.9°, giving an effective AOA of 15° (half the difference).

Illustation of the trim tab on opposite sides with the change in wing angle

The following image shows a sample run of the Trim Tab Authority test on Voyager 2.9.

Example Run of the Trim Tab Authority test, covering 10 to 17 degrees for the Trim Tab.

A series of four runs were performed and the results recorded in the scatter plot below.

Scatter plot of results with lines of best fit.

The Eppler 169 foil should have its angle of attack kept under 10 degrees, probably around 8 degrees maximum.

For Voyager 2.9, the Trim Tab angle approximately equals the Wing Sail Angle of Attack.

 

Wing Sail Foil Testing - Downwind Settings

The current algorithm used within Voyager to set the Wing Sail Trim Tab is simple:

• The trim tab is rotated clock-wise on starboard tack.
• The trim tab is rotated counter-clock-wise on port tack.
• The angle may be changed, but has typically been set to 15 degrees or majority of sailing.

The following three illustrations (figures 1,2 & 3) show the typical operation of self-trimming wing sails.

In each case, the wing sail effectively doesn't change while the vessel changes heading from beating, reaching to running, on starboard tack.

Figure 1. Starboard Tack Beating, Trim Tab rotated CW 15°

Figure 2. Starboard Tack Reaching, Trim Tab rotated CW 15°

Figure 3. Starboard Tack Running, Trim Tab rotated CW 15°

One issue with self-trimming wing sail operation that has not previously been clear is:

What is the best position of a trim tab whilst running.

Should the position of the trim tab be reversed when deep running ?

The following figure 4, illustrates the vessel running on starboard tack, with the trim tab reversed, and swapped to CCW position.

Figure 4. Starboard Tack Running, Trim Tab rotated CCW 15°

The Test rig was set  up for different headings, including deep running, and the load cells were used to provide a measure of relative forward thrust and force to leeward.

The aim was to measure thrust generated by the two strategies for trim tab position while deep running.

Once tests were performed to assess the forward thrust it became clear that reversing the trim tab is not a good idea when deep running.

Figure 5. Test Rig set for Starboard Tack Running- CCW Trim Tab

Figure 6. Test Rig set for Starboard Tack Running- CW Trim Tab

Experimental test results are presented in the following table.

In all cases the thrust was reduced when the trim table angle was reversed.

The last results with 157.5° off the wind and 20° trim tab showed a reduced difference.

At 135° off the wind the thrust appeared to be reversed.

Of course, at reduced angles to the wind such as 90°, the effect of reversing the trim tab will definitely cause the vessel to travel in reverse.

Figure 7. Test Results

Conclusion

The is no benefit in reversing the trim-tab position when deep running.

It is best to retain the behaviour where trim-tab is rotated CW on starboard tack, and CCW on port tack.

Wing Sail Foil Testing - Trim Tab Size and Control Authority

One objective of testing was to clarify the relationship between trim-tab size and control authority. This was investigated by splitting the trim tab into two equal halves and comparing the aerodynamic response when operating one half versus both halves together.

Test foil with split trim tab allowing testing of a full size and half size trim tab

As expected, the relationship between trim-tab area and effectiveness proved to be approximately linear. Tests conducted with the wing pivot at 18% chord showed that halving the trim-tab area produced roughly half the change in angle of attack. The results confirm that, within this range of operation, trim-tab authority scales proportionally with its surface area.

Conclusion

Trim-tab effectiveness varies linearly with area.

Wing Sail Foil Testing - Optimum Axis Position

What is the optimum position of the axis of a self-trimming wing sail expressed as a percentage of the chord ?

A self-trimming wing sail must naturally weathervane into the wind. 

It is well documented that the aerodynamic centre of a simple thin foil is at 25% of the chord.

NASA - Aerodynamic Center

This means that 25% is the absolute maximum position along the chord, because beyond that the wing sail won't weathervane into the wind.

As the position of the pivot is moved toward the 25% point, the wing sail becomes more balanced, requiring less effort to increase the angle of attack.

This also means that manufacturing tolerances will have an increasing effect as the pivot point moves toward 25%.

 

As the pivot point moves closer to the leading edge, the effort to increase the angle of attack of the foil increases.

Also manufacturing and assembly issues encourage the pivot point to be a reasonable distance back from the leading edge.

The image below shows the sail from Voyager 2.9. A practical issue is the need to have pivot far enough away from the leading edge to allow for mounting the magnetic encoder ring used for measuring the wing sail angle.

Voyager 2.9 wing sail with mast pivot at 16% of chord

The wing sail with 16% pivot point has been used successfully in lake trials and on the recent ocean passages.

Testing Pivot Position

A series of tests were performed with the test rig by setting the position of pivot at the following percentages of the chord:

15%, 16%, 18%, 20%, 25% and 27%.

Then the trim tab moved in 5° steps from -40° to +40°.

The resultant angle of attack was recorded at each step.

The results are plotted below with the trim tab angle on the horizontal access, and the resulting angle of attack (AOA) on the vertical access.

The Eppler 169 foil has an optimum AOA for lift vs drag, of around 5° to 6°.

The overall trends observed in testing were consistent with expectations.
The sensitivity of the wing’s angle of attack (AOA) to trim-tab deflection increases as the pivot axis moves aft toward the 25% chord point.
At 25% chord the system became too sensitive for practical use: we aim to operate the wing at an AOA below about 10°, yet a trim-tab change of only 1–2° was enough to drive the wing into stall.

It is important that the wing’s response to trim-tab input is not excessively sensitive.
The trim tab must move by a practical amount during normal operation—large enough to be repeatable and to overcome any mechanical backlash or stiction in the linkage—while still giving fine control over the AOA.

The tests revealed an apparent offset of approximately –3° in the trim-tab calibration.
This could be due to small manufacturing or alignment errors, but is more likely caused by flow-field asymmetries in the test setup.

A useful benchmark is the trim-tab deflection needed to achieve a 10° AOA:

• about 7° of tab deflection when the pivot was at 15% chord,

• about 5° when the pivot was at 18% chord.

For completeness, we also explored operating with deliberately large trim-tab deflections that drove the wing well past the stall angle, even though such conditions are outside the intended flight envelope.

Conclusion

Overall, a pivot-axis position in the range of about 15–20% chord appears to be well-suited.
Within this range the wing shows stable weathervaning, and the tab response is strong enough to overcome backlash yet not so strong that small tab motions cause abrupt stalling.

I expect to continue using the 16% chord position for the pivot of self-trimming wing sails.

 Wing Sail Foil Testing - The Setup

The designs of the wing sails used with the Voyager sailing drones have been established by studying other designs, by using a lot of intuition and by judging whether it looks right. 

It is time to perform some more rigorous testing to determine the optimum values of some key parameters of a self-trimming wing sail.

Design Questions

These are some questions to be answered to assist in designing and operating the next wing sails.

1. What is the optimum pivot point for a self-trimming wing sail ?
2. What is the relationship between Trim Tab angle and Angle of Attack, and hence what size should a trim tab be ?
3. What is the optimum trim tab angle when running ? Should the trim tab be reversed when running ?

Wing Sail Test Rig

 I developed a test rig to allow a series of relative measurements to be performed indoors.

The airflow is provided by a large domestic fan.

But tests quickly showed that the airflow from a fan is too turbulent to be useful for performing measurements. So a columnator or flow straightener was developed to improve the quality of the airflow.

This was constructed primarily from rolled up sheets of A4 paper, contained within a wooden frame. It wasn't great, but it was good enough to get useful results.

 

Fan, flow straightener and the test article.

Fan and flow straightener

The airflow in the vicinity of the test article was around 2.8m/s.
This was measured using an Air Velocity Sensor Module, the Renesas FS3000-1015.

Test Article

The test foil is an Eppler 169 (E169) with a 400mm chord. This chord size represents the approximate size to suit a wing sail for Voyager 3.

The 400mm chord yields a Reynold's number of around 76,000 for the 2.8m/s airflow.

The test foil has been designed to support testing with the following characteristics and adjustments:

• The position of the axis may be varied from 15% of the chord to well over 35%.
• The trim tab is adjustable with a scale to easily set a desired angle.
• The trim tab is split in two, to allow for measuring authority versus size.
• The test rig includes a scale to measure angle of attack.
• The wing sail mast bearings are supported by load cells to provide a relative measure of the load in 2 dimensions. The load cells are rated at 1kg max, and include digital readouts in tenths of grams as a relative measure of force.

Test wing section, Eppler 169 with 400mm chord

Adjustable pivot point shown at 18%

View of Trim Tab and scale showing +10 degrees.

Angle of Attack scale showing 

View of mast mount and load cells providing independent support in the X and Y axes.

Digital readouts of relative force expressed in grams

Tenth time lucky - Finally !! - a Successful Ocean Passage

Sunday 7th September 2025 around 6pm Voyager 2.9 was launched from Torquay.

She was bound for Flinders in Western Port, around 50 miles or 80kms to the east.

Winds were generally 10 to 15knots, peaking at 20knots from the northwest and backing around to southwest over the next two days.

Launch from Torquay Fisherman's Beach 6pm Sunday.

Course from Torquay, 80km eastward to Western Port

The intended destination was Flinders beach, just inside Western Port. This beach is calm, and surrounded by hills. The hills tend to block the southwesterly wind. As the boat cleared the last waypoints to head into Flinders, there was not enough wind power to overcome the incoming tide.

This meant that Voyager was swept out of its corridor and deeper in Western Port.

I could see from the satellite tracking that it was not going to be able to reach Flinders. I was able to move to high ground and gain line-of-site to establish a telemetry link and redirect Voyager further north to Point Leo, where a landing would be easier, with more favourable winds.  

The telemetry uses 433MHz LoRa. It is reasonably reliable over 2 to 3km provided there is line-of-site. This requires being elevated above the beach, on the hills behind.

Approaching the beach at Point Leo, Tuesday midday, 100m to go.

Happy skipper.

I swam out 50m to bring Voyager to shore.

Safely ashore at Point Leo after 44 hours

Battery Life

The main battery consists of 12 x 18650 cells arranged as 2S x 6P.

When the nominated minimum voltage is set at 6.5Vdc, the estimated battery life is around 10 days.

The Wing Sail battery consists of a pair of 18650 cells arranged as 2P.

With a nominal minimum voltage of 3.5Vdc, the estimated battery life is around 60 days.

Wrap up

The boat was in good order on arrival, responding well to manual override via the LoRa telemetry channel. It appeared that it could have continued at sea for several more days.

Summary of failures on previous ocean voyages:

1. Fatigue failure of aluminium mast, and loss of equipment housing due to inadequate strength and fastenings. Changed to carbon fibre mast, and greatly improved the strength of hull fastenings.
2. Water ingress into the Wing Angle Sensor, which is 3D-printed part. Filled the sensor housing with epoxy resin so that we don't rely on the 3D-print being waterproof.
3. Software errors related to weather changes and wind direction transitions in particular cases that were not correctly handled. Some software errors are not revealed with lake testing on short courses. Some are only revealed on multi-day multi-mile courses with wind direction transition that were not anticipated.
4. Poor compass calibration, combined with software errors lead to failure. This was addressed by focusing on improving compass performance and calibration, and also software improvements to make better decisions and be more resilient in the cases where compass accuracy is critical.
5. Failure of standard servo with a brushed motor. Changed to brushless servo motors for steering. There may still be an issue with the life-span of the mechanical potentiometer used for positional feedback. On low-cost servos, the feedback potentiometer may wear out. So this is still an issue of concern.
6. Failure of the Wing Sail controller due to water damage. This is now potted in epoxy and has performed well over multiple missions at sea. This includes the failed 6th voyage, lying on a beach for 100 days. It worked perfectly after that, once powered up, so the one controller has been reused on all voyages since.
7. Likely failure of the Wing Sail structure. It is difficult to prove, but it is believed that one or two failed missions may have been caused by the Wing Sail suffering a structural failure with the tail.
   A new design has now been established for the Wing Sail that eliminates the separate tail with trim tab and integrates the trim tab into the main foil. This design appears to be inherently stronger because of the elimination of the separate tail.

Ninth Voyage in Bass Strait - Argh! Software Bug!

Conditions were good for a 2-day passage commencing from Torquay Fisherman's Beach on the evening of Monday 28/7/2025.

The weather on the course was initially running, and was predicted to back toward a reach, and continue backing to a beat with moderate winds near the end of the main leg to Western Port.

But it didn't get there, and this time the reason was a software bug.

Ready for Launch at Torquay Fisherman's Beach, 28/7/2025.

The software bug was introduced in 2021, but had not shown itself in ocean conditions before.

When beating to windward the software limits the upwind course with a minimum upwind angle, forcing the vessel to tack to reach a windward waypoint.
In early 2021, a similar constraint was added for downwind sailing, forcing the vessel to tack downwind to reach a waypoint within the minimal downwind angle. This was to improve downwind performance by avoiding sailing dead-downwind.

This all appeared to be fine, but there was a latent error, that was not recognised until now.

The error was revealed when sailing a running course that requires tacking downwind, and then the wind changes so that the course requires tacking upwind, without a period in between where the course is directly sailable.

Given this specific scenario, the software had an error where it did not allow the vessel to leave the running course. I've named it the "Stuck Running" bug.

This lead the vessel off the course and it did not recover.

The software has now been updated to handle the direct transitions between upwind and downwind courses that are not directly sailable. 

On a positive note, everything else with the vessel appeared flawless.

 New Self Trimming Wing Sail Design - Tailless

A symmetrical foil generally has the Centre of Pressure (CoP) at a position about 25% of the chord from leading edge.

Hence a reliable configuration for a self-trimming wing sail is to place the axis of the wing sail at the 25% point, and then add a separate tail to ensure the CoP is well behind the pivot point.

Influential Sailing Drones

These are images of significant Sailing Drones equipped with self-trimming wing sails that have strongly influenced the design of the Voyager Drones.

Saildrone Explorer

Saildrone Surveyor

Maribot

Voyager Sails

The following images show a series of different Self-Trimming Wing Sails on Voyagers.

Voyager 3

Voyager 2.0 with first version of wing sail

Voyager 2.8 with evolved wing sail with tail

As at April 2025, Voyager has made eight voyages into Bass Strait with different problems occurring in each voyage. Before commencing each new voyage the issues encountered on the previous voyage would be addressed and the design evolved. Issues have been related to waterproofing, software design errors, robustness, servo failure during to wearing out.

One problem appeared to repeat on two different voyages.

This was a loss of sailing performance, probably due to the wing sail tail being damaged.

It was determined that the sail assembly might be more robust if the tail is removed from the design.

This can be done if the pivot point is moved forward of the 25% neutral point.

It is unclear what optimum point is, perhaps between 14% and 20%.

Tailless Self-Trimming Wing Sail

Voyager 2.9 has been rebuilt with a tailless self-trimming wing sail with a pivot point at 16% of the chord. 

Voyager 2.9 with Tailless Self-Trimming Wingsail

The Tailless Self-Trimming Wing Sail with an axis at 16% of the chord appears to perform as well as previous sail with tails. It appears to be just as stable, showing no signs of unexpected behaviour.

This sail should be more robust with the elimination of the tail.

A future refinement may be to construct the sail from fibre glass coved foam core for increased strength to resist to damage in breaking waves.

A further design change was the choice of foil. The recent Voyager Wing Sails have all been NACA 0018 (a symmetrical foil with 18% thickness),
The new sail employees an Eppler 169 (E169) foil (~15% thickness). The Eppler foils are intended for use with low Reynold's numbers, typical of those found on small vessels such as Voyager.

Voyager 2.9 with Tailless Self-Trimming Wingsail on the water

Saildrone's Hurricane Sail

I later realised that Saildrone had already developed a tailless version of the self-trimming wing sail.

It is designed with a smaller highly robust self-trimming wing sail for deployment during hurricane season.

Measurement taken from photographs suggest that it's pivot point at around 19% or slightly more.

Saildrone fitted with the Hurricane Sail

Future Study

Future work will involve preparing a test foil which allows for an adjustable pivot point. The aim is to allow the pivot point to be varied between 15% and 30% to allow for testing in a wind tunnel environment.

The likely aim will be to find the position representing the largest percentage away from the leading edge for the axis, where the sail still exhibits stability and "weather vane" behaviour.