I did this buoy design over a couple days as part of a demonstration of capabilities in 2021. It is reproduced here with small adjustments for this format, but with the errors left in.
You can also check out the buoy CAD model on Onshape here:
Problem Overview
The design challenge was to develop a rugged buoy capable of operating in extreme marine environments, including arctic conditions, for a client. The enclosure needed to protect sensitive internal components while incorporating sealed optical windows and maintaining wireless communication capabilities. Key considerations included waterproofing methods, material selection for durability and signal transparency, and efficient manufacturing processes that could meet both timeline and budget constraints.
Requirements
From the above problem overview, the product must:
- Float
- Have an anchor line mounting point
- Protect the electronics from seawater ingress
- Resist corrosion or degradation when immersed in seawater
- Resist degradation due to sunlight
- Operate in low temperatures expected in arctic waters
- Be strong enough to withstand potential collisions with marine debris
- Contain a power source to power the sensors and RF antenna
- Contain sufficient stored energy to last for a reasonable duration without service
- Have UV-transparent windows to allow fluorescence sensors to measure seawater
- Have an RF-transparent hull to enable satellite communication
As with any project, there are also unstated requirements, where it would be nice if the product:
- Was easily serviceable with standard tools in a fishing boat
- Used standard off-the-shelf parts wherever possible
- Used easily replaceable subassemblies to enable swapping damaged parts
- Required a minimum number of tools for disassembly
There are also prototyping and manufacturability constraints due to the short project duration and (inferred) scale of around 100 buoys. These constraints mean that it would be ideal if the manufacturing method:
- Was cheap
- Had a short leadtime
- Was easy to change design for fast prototyping
- Minimized the number of unique parts
Given that the buoy will be deployed in the Arctic, it needs to contain sufficient energy storage to power itself for a reasonable amount of time. A ballpark figure of 6 months is estimated for the following report, but the desired lifetime should be discussed with the client to get an idea of how often it is practical to service and recharge a buoy.
The brief states that the buoys are deployed in arctic seas, but does not give a more detailed description of the environment and latitudes in which the buoys are expected to operate. As a conservative first assumption, the buoy can be expected to operate in a saline slush, which reaches a minimum temperature of around -20 C. Further discussion with the client needs to occur to determine if the buoy needs to be able to survive being frozen in the ice.
Hull Design
Material Selection
The need for the hull to be tough and UV resistant while remaining RF transparent rules out metal enclosures and limits the material choices to a number of engineering plastics such as ABS, polycarbonate, and acrylic.
ABS is relatively cheap and appears to be available in marine grades. However, it is weaker and has worse chemical resistance than polycarbonate.
Acrylic is very UV resistant, but is prone to chipping and cracking under stress. It may not be tough enough to survive collisions with marine debris.
Polycarbonate is tough, but more expensive. Polycarbonate seems to be used in marine applications and available in UV-stabilized and marine-grade formulations. Polycarbonate can have issues with stress-corrosion cracking in salt water, which will need to be investigated further to ensure polycarbonate’s suitability in a submerged application.
Given polycarbonate’s toughness, lightness, and ease of manufacturing, it is the primary choice for the hull material.
Comparison of Potential Manufacturing Methods
Milling from Solid Billet
One approach is to mill out the desired hull shape from a solid plastic billet. However, given that hulls are typically thin, contoured shapes, milling from a solid billet will be tremendously wasteful and time-consuming. In addition, fixturing such a large hull to prevent distortion or vibration during machining will be challenging, which will impact the surface finish of the sealing surfaces.
3D Printing
Another approach would be to 3D print the outer hull with a fused deposition modelling (FDM) printer. 3D printing gives excellent flexibility for designing the hull shape and allows the mounting points for the inner components to be easily integrated into the hull.
However, because FDM parts are composed of strands of plastic laid side-by-side into layers and then stacked, the parts will be inherently porous and difficult to seal. In addition, the rough layered surface of the part will be challenging to seal against and may require post-machining or smoothing to improve the surface roughness.
The size of the buoy also presents a challenge for using FDM, as printing large parts with the fine layer thicknesses required for good surface finish is very time consuming, potentially taking many days to print a single part. This long manufacturing time will slow the design process by increasing the time to produce prototypes and will make scaling up to hundreds of buoys cost- and time-prohibitive.
Selective laser sintering (SLS) 3D printing will give a better surface finish and may prove easier to seal, but reduces the range of available materials and also presents the same scaling and print time concerns. SLA printing will have excellent surface finish and sealing, but it is likely infeasible to make such large parts as required for the hull.
Injection Moulding
The hulls could be injection moulded, which would produce water-tight parts with good surface finishes. However, the tooling cost for large injection moulds would be in the many tens of thousands of dollars, potentially over $100k given the size. Injection moulds also take 8-12 weeks to manufacture and are difficult to change after manufacture without incurring significant tooling costs. Given the probable scale of hundreds of buoys, the per-part mould amortization cost will likely be prohibitively expensive.
Vacuum Forming and Post-Processing
Another approach is to vacuum-form the hulls to produce the desired shape, then laser-cut, waterjet, or mill the desired mounting features into the hull. The single-sided male die that defines the hull shape can be cheaply CNC machined from wood and easily modified or replaced to enable quick design changes. Plastic sheets can be purchased in standard sizes and UV-stabilized colours, allowing for easy sourcing of material. The heating, forming, and cooling process is relatively quick (though dependent on part thickness), so forming a hull in under an hour should be feasible. Given that the part begins as a solid plastic sheet, the vacuum-formed hulls will be water-tight.
The surface finish of the die will set the surface finish of the inside of the part, so the mould must be sanded to ensure good surface finish on critical sealing surfaces.
However, vacuum forming is limited in the maximum sheet thickness that can be formed, with cursory research showing a maximum polycarbonate sheet thickness of ~6mm. It may be possible to form thicker sheets by heating from both sides and applying a gentle vacuum above the sheet to counteract gravitational forces, but there is technical risk in attempting to develop the process of vacuum-moulding thicker sheets. Depending on the size of the buoy, additional reinforcements may be necessary to strengthen the hull.
Vacuum forming also limits the shape of features that can be created in the hull. Features must be drafted to allow the part to separate from the mould, and bends must be radiused to allow the material to deform. Mounting features for internal components must be either cut or post-machined into the hull after forming.
The dimensional control will also be worse than machined, 3D printed, or injection moulded parts as the sheet can contract and distort during cooling. Thick, compliant gaskets are recommended to reduce the effect of undulations in the sealing surfaces.
Registering the post-processing features to the moulded part may be difficult due to distortion. Pre-drilling registration holes in the clamping zones of the plastic sheet before vacuum forming will help to register the part for post-processing.
Laser cutting and waterjetting are both fast processes that can be easily adjusted by changing the toolpath, which makes both excellent choices for both prototyping and production. Both methods are limited to making perpendicular cuts completely through the hull, which limits the possible shapes of features cut into the hull. If necessary, the hull could be loaded into a mill for post-machining of key features.
Selected Method
Given the cheap cost, easily changeable mould, fast production time, and good sealing, vacuum forming of polycarbonate will be selected. However, polycarbonate cannot be laser cut, so waterjet cutting will be used to cut the mounting holes. Care must be taken to protect the sealing surfaces from abrasion damage from the waterjet spray.
UV Fluorescence Sensor Design
The project brief states that the client has a working sensor design, but does not elaborate on the sensor size or shape. A rough estimation of the expected sensor design based on first principles is necessary to design the sensor mounting.
Fluorescence Sensor Physics and Design
Fluorescence measurement consists of exciting a fluorescent material with a light source of known frequency, then measuring the light emitted at a lower characteristic fluorescence frequency.
The excitation source must have a tight frequency distribution, so an LED or laser is often used. To detect the emitted fluorescence, a photodiode or other light receptor sensitive in the range of the characteristic frequency is often used. However, the receptor will often also be sensitive to the light frequency from the excitation source, so an optical bandpass filter must be placed in front of the receptor to isolate light at the fluorescence frequency.
A rough block diagram of the sensor is shown below.
If the excitation frequency is relatively far from the characteristic fluorescence frequency, an LED may be usable; if the frequencies are close together, the tighter frequency range of a laser may be necessary to limit the overlap in excitation and fluorescence signals.
It is assumed that it is possible to package these components into a small ~1”x3” plug-shaped sensor geometry that is common in industrial applications.
Sight Glass Material
The sight glass must be transparent to the UV excitation and fluorescence frequencies. It must also be able to withstand sustained UV radiation without degradation, and should be scratch-resistant enough to prevent scratches from obscuring the sensor’s view of the water.
Acrylic appears to have good UV tolerance and transmission, with up to around 90% transmission of UV light. Its transmission properties in the excitation and fluorescence frequencies would need to be confirmed with the optics team, but as acrylic is cheap, relatively tough, and easy to manufacture, it is a good option for the sight glass material.
Cursory research into UV transmitting glasses show that fused quartz and fused silica glasses are available for UV transmission. While the glasses would be more scratch-resistant, they would also be more fragile and would present a more difficult part to seal reliably. The strength and available thicknesses has not been explored in depth, so the difficulty in obtaining suitable glass parts is unknown.
Some candidate materials have been identified, but further discussion with the optics team is required to ensure that the window is suitably UV transparent.
Electronics & Control Considerations
Given the potentially small fluorescence signal intensity and high RF noise from the satellite antenna, it seems pragmatic to locate the sensor and signal conditioning electronics in the sensor body in a shielded enclosure. The sensor can then digitize the signal and send data to the control board over a noise-tolerant protocol such as CAN bus. It may also be necessary to use shielded cables to prevent RF energy from being picked up in the sensor cables.
There is also a possibility that excitation light from adjacent UV sensors could distort measurement results, so it may be necessary to take measurements in series to prevent optical cross-talk. Given that the pollutant concentration likely does not vary significantly from second to second and that the bandwidth and power budget for the satellite uplink is likely low, the reduced sampling rate from measuring serially will likely be acceptable.
Control Board
The project brief states that the electronics have already been designed, so this section is an attempt to estimate the size and raise any mechanical issues that may be relevant to the electronics.
To reduce the possibility of damage due to small water ingress, it is recommended that all boards be thoroughly conformally coated and that the control board contain a moisture sensor to detect a slow hull leak.
The control board needs to:
- Read data from the UV sensors
- Read battery/power source health data
- Read in-buoy moisture sensor
- Communicate with satellite over antenna
Given the likely slow data rate and low power consumption, a small single-board computer may be sufficient to handle the control requirements. A power source monitoring system such as a battery management system will be required to detect power source issues. The digital telemetry signal will need to be packaged, modulated, and amplified for transmission to the satellite, which may be possible with an all-in-one RF antenna/transceiver module.
The control electronics module is modelled as a 100mm x 100mm x 20mm box mounted under the mounting plate. There is currently space for larger or multiple control modules if necessary. Given that the buoy will need to contain a sufficient amount of air to float, space constraints are not anticipated to be an issue.
Energy Budget & Power Sources
As the buoy must remain at sea for long periods, the energy source is expected to occupy significant mass and volume inside the buoy. The type of power source chosen will affect the design of the hull and sealing, so a consideration of the possible power sources is necessary to guide the buoy design.
Potential Power Sources
The buoys are expected to be deployed in the arctic seas, so solar may not be a good option given low sunlight in the arctic winter. If the buoy area is large enough and the power budget is small enough, it may be possible to use an array of solar panels, but this is considered likely to be impractical.
If batteries are used, the batteries need to be light and function in low temperatures. The energy density of the battery is critical as increasing battery weight increases required hull size, further increasing cost. Cursory research into marine batteries indicates that LiFePO4 batteries are often used in marine environments, are ~50% lighter than lead-acid, and function down to -40 C, so a LiFePO4 battery bank may be suitable for powering the buoy. These batteries are also available in standard sizes from a variety of manufacturers, making replacement of a failed battery relatively cheap and easy. However, the energy density of the batteries is relatively low, which may require a prohibitively large and heavy battery bank.
It may be possible to use a fuel cell to power the buoy, which would give a better energy density than a battery bank. However, fuel cells need to vent exhaust gasses, which would complicate the seal design by requiring an exhaust valve. Fuel cells also tend to be expensive and are not as easily replaceable as a battery bank.
Scavenging power from the wave energy with a mechanical device such as an off-center weight and generator similar to a mechanical watch may be possible, but presents risks of mechanical failure during storms and would further unbalance the buoy, potentially impacting the RF antenna effectiveness. Scavenging wave power may be feasible but presents a larger technical risk.
A longshot idea would be to mount a stack of dissimilar metallic plates on the bottom of the buoy and use seawater as the electrolyte to form a galvanic cell, reducing the weight required for electrolyte and increasing the air volume inside the hull. However, the energy density of such a system is unknown, the technical risks are not well understood, and exhaust of metal ions to the environment may be unacceptable.
Given the maturity of the technology, low cost, robustness, and lack of moving parts, LiFePO4 batteries are selected as a first choice for the power source. Whether the energy density is sufficient to be practical depends on the buoy energy budget discussed below.
Energy Budget
As a ballpark estimate, assume that the buoy consumes power at the same rate as an iPhone 11. The lack of display and reduced processing workload may be compensated by the higher power required to transmit data to the satellite.
A battery life test conducted by Tom’s Guide showed that the 3.7V ~3000 mAh lithium battery in the iPhone lasts around 11 hours under constant use, which corresponds to around 1W average power consumption.
At 1W constant power and using a 12V LiFePO4 battery, the buoy would need ~360 Ah of energy storage to last six months, which is far more than is practical.
The power consumption and energy storage budget of the buoy will need to be discussed with the client to determine the size of the battery bank and hull necessary, or to prompt the switch to a more energy-dense energy storage mechanism.
RF Considerations for Satellite Communication
Buoy will likely be bobbing and rolling significantly in the waves, so a mechanically pointed antenna may be impractical given the cost and scale of the buoy. It may be possible to gimbal the antenna to compensate for bobbing, but the gimbal design is likely to be expensive. I am not sufficiently knowledgeable in RF antenna design to know if a phased array antenna could compensate for the bobbing quickly enough. Further discussion with the satellite link team is required to understand how to design the buoy to enable consistent antenna functionality.
Using a directional antenna such as a dish, Yagi-Uda, log-dipole, or phased array antenna will reduce the power required to communicate with the satellite by directing the energy at the satellite receiver. However, if accurate pointing cannot be achieved, a more omnidirectional antenna such as an eggbeater antenna could be used at the expense of increased transmission power.
Another consideration is the antenna proximity to the water’s surface. If the antenna is sitting at or below the water level, the water may shield the signal and prevent it from reaching the satellite. It would be advantageous to mount the antenna as high as possible, which could include using an off-the-shelf panel-mount marine antenna passing through the top hull.
The client has stated that the satellite uplink design has been completed, however it would be worth going over the requirements of the uplink design to ensure that the buoy mechanical design does not compromise its effectiveness.
Proposed Design Overview
The proposed design is around 0.75 meters in diameter and uses a three-part polycarbonate body. The upper and lower hull elements are vacuum-formed on the same die and share the same waterjetted perimeter holes, while the bottom hull has additional holes to expose the UV sensors, anchor point, and waterproof charge/debug port. To reduce the number of hull seals, a mounting plate made from the same polycarbonate separates the two hulls and has mounting features for the control electronics, batteries, and RF antenna.
The UV sensor subassembly is based on a lathed 316 stainless steel housing that holds the UV transparent sight glass and UV sensor. The housing provides the sealing surfaces for gasket seals against the hull and around the sight glass.
Hulls & Mounting Plate
The hull will be composed of a top and bottom hull both vacuum formed on the same die. Both hulls will have perimeter holes waterjetted, while the bottom hull will have additional holes to fit the UV sensors, anchor point, and waterproof charge/debug port. The waterjetting should be done from the outside surface of the hull so that the perimeter sealing surface is undamaged. Pre-drilling the piercing holes for the sensor, anchor loop, and charge/debug port is recommended to prevent surface damage from piercing.
To reduce the number of seals required, a middle mounting plate will be waterjetted from the same polycarbonate sheet stock as the top and bottom hull. The mounting plate will share the same perimeter hole pattern as the hulls to complete the perimeter seal and give a good mechanical connection to the hulls. Cable pass-throughs holes will be waterjetted into the mounting plate, while mounting holes for internal components will be waterjetted undersize, then drilled and helicoiled. If helicoils are not sufficient to prevent thread tear-out, heat-weldable threaded inserts can be installed to improve strength.
The housings will be separated by thick, conforming gaskets that are made from marine-grade material such as neoprene or EPDM and are either laser-cut or waterjet from a sheet. Using thick, conforming gaskets will reduce the need to maintain flatness on the sealing surfaces.
The perimeter seals will be compressed by the 316 stainless steel M8 fasteners around the perimeter. The spacing of the bolts will need to be adjusted based on the polycarbonate stiffness to ensure that bowing of the hull between bolts does not allow water leakage.
In the top hull, an RF-transparent foam float is installed to prevent the buoy from sinking in the event of a water leak. The symmetrical hull design leaves plenty of room for the satellite antenna, the shape and size of which is currently unknown. A phased array antenna is modelled in the renderings, but could be easily changed.
The battery, sensors, and majority of electronics will be slung under the mounting plate to keep the centre of mass low and keep the buoy upright. A replaceable desiccant packet is included to remove any moisture introduced from servicing the buoy while on a boat. All components on the mounting plate are mounted with M6 screws to minimize the number of tools required for assembly.
The UV sensors, anchor loop, and charge/debug port are installed in and sealed against the lower hull. The anchor loop is an off-the-shelf 316 stainless steel eye bolt. The charge/debug port is an off-the-shelf marine-grade IP68 waterproof port which should survive being permanently submerged.
Visibility & Markings
Cursory reading of Canadian regulations around buoy markings suggest that a private buoy can be yellow and that flashing lights or retroreflectors may be acceptable as visual aids. The proposed hull will be made with opaque yellow polycarbonate for high-visibility and will have yellow marine-grade retroreflective tape applied to improve visibility while not requiring increased power consumption for a flashing light.
Buoys are also required to be marked with the owner and phone number, so the top of the buoy will need to be painted with the contact information and buoy serial number. This painting could likely be screen or pad printed with a marine-grade paint.
The applicable marking and visibility regulations for the seas in which the buoys will be deployed will need to be reviewed in detail to ensure regulatory compliance.
UV Sensor Housing and Gasketing Design
The UV sensor housing is lathed from 316 stainless steel with internal threads for holding the gaskets, lens, and gasket compression locknuts. 316 stainless is selected for its resistance to salt water, machinability, and to maintain a consistent metal type on components exposed to the salt water to prevent forming a galvanic cell.
To ensure consistent compression, a mechanical stop for the locknuts is cut into the housing. The placement of the stop can be changed in the design to vary the gasket compression.
The sight glass is cushioned between two face gaskets and with a perimeter gasket to ensure that it experiences gentle, well-distributed loads. This enables the use of both acrylic or more fragile glasses without risk of shattering the glass.
If acrylic’s UV transmission properties are sufficient, the sight glass gasketing assembly can likely be redesigned to place the acrylic on the outside of the housing and transmit the gasket compression forces through the acrylic, which will simplify the design and prevent bubbles from collecting in the pocket.
The upper locknut is threaded to fit the UV sensor body, while the lower locknut is drilled larger to clear the threads. The internal locknuts can have the external and internal threads cut on a lathe, then be parted off for slotting on the mill. A custom tool for tightening the locknuts using the inside slot can be made by spot-welding a strip of steep edge-on to the front of a deep-drive socket.
The sight glass seal is locked by applying a theadlocking compound to the internal threads of the housing, tightening the first locknut against the mechanical stop in the housing, then tightening the second locknut against the back of the first to prevent loosening. Use of threadlocking compound in sealed optical assemblies can sometimes cause issues with the threadlocker outgassing and fogging the optical elements, so a low-outgassing threadlocker may be needed in this application. In addition, an activator may be necessary to allow the threadlocker to cure against the 316 stainless steel housing.
A slippery UHMW washer sits atop the upper face gasket to distribute the compression force and allow the locknut to turn without tugging on the gasket. It may be possible to source the an off-the-shelf washer; if not, the part can be easily lathed in-house.
The housing features a flange for sealing against the hull. The housing body is externally threaded and either an off-the-shelf nut or custom nut is used to tighten the housing gasket against the hull. A UHMW washer may need to be added against the inside of the hull to prevent the nut from damaging the plastic hull surface.
The top of the housing has two flats for gripping with a wrench while tightening the nut. The handle clearance of the wrenches contacting other sensor bodies during installation may be an issue, so custom short-handled wrenches may need to be waterjetted to make assembly easier.
To reduce the part count, the gasket material specified for both the sight glass and hull seals is the same gasket as the hull perimeter seals. These smaller gaskets can be cut from the gasket sheet at the same time as the outside perimeter gaskets to save cost and time.
The UV sensor can be pre-assembled and tested to catch issues like failed sensor, glass contamination, etc. early in the assembly process. Building the UV sensor as a subassembly also allows easy replacement in the field in the event of a sensor failure.
Assembly Process
The buoy is composed of the following four major subassemblies assembled in the following order:
- 6x UV sensor
- Lower hull
- Mounting plate
- Upper hull
Given more time, the assembly process would be documented in work instructions that outline the assembly steps, parts required including part numbers, tools required, and annotated diagrams or pictures of each step.
Note: I did later build a more detailed work instruction here: Sight Glass WI Public
UV Sensor
The UV sensor can be assembled top-down on the assembly bench with the following steps:
Assembly:
- Put on gloves for handling optical elements
- Place housing on bench with flange down
- Place sight glass face gasket into bottom of housing tube. Ensure gasket sits flat against bottom of tube.
- Clean inside face of sight glass with optical wipe
- Using sight glass positioning pliers (modified long-nose pliers with soft and compliant gripping faces) to place sight glass on top of face gasket and centred in tube
- Place perimeter gasket around sight glass and gently press into shape with popsicle stick or other soft poking tool. Ensure that gasket is well seated and surrounding sight glass
- Place second face gasket atop the sight glass and perimeter
- Place UHMW washer atop gasket
- Apply threadlocker and activator to internal threads of housing
- Thread in through-hole locknut with no internal threads. Tighten using locknut tightening tool until locknut contacts mechanical stop in housing
- Re-apply threadlocker to internal threads for second locknut
- Thread in locknut with internal threads and tighten against first locknut
- Place hull gasket around sensor housing and slide down until it contacts the front flange
- Apply threadlocker to external threads on sensor body
- Thread sensor body into internal threads of upper locknut
Test:
- Install sensor subassembly into calibration jig, then plug sensor wire into calibration computer
- Power up sensor and confirm that it can be read by test computer
- Confirm basic functionality such as reading signal, turning exciter on and off
- Calibrate sensor readings against known fluorescence samples
- Depending on sensor design, potentially write sensor calibration curve back into sensor EEPROM
Lower Hull
- Place lower hull inside-up on bench
- Lift hull and pass UV sensor subassembly through sensor mounting holes
- Thread on backing nut and tighten sensor body to hull
- Repeat steps 1-3 for 6 UV sensors
- Flip hull over and check compressed sensor gasket thickness with go/no-go shim gauges
- Install waterproof charge/debug port through charge/debug port hole and tighten as per manufacturer’s recommendations
- Place anchor loop gasket onto anchoring loop
- Apply threadlocker and activator to anchor loop threads
- Lift lower hill and place anchor loop through anchor hole in bottom hull
- Place M12 washer and nut onto the anchor loop threaded post, then tighten nut with socket and ratchet. Use a metal tube through the anchor loop to provide resisting force.
- Note: this step will be easier if the assembler has a pocket in the table for the anchor loop to sit so that the hull does not rest on the anchor loop
- Check anchor compressed gasket thickness with go/no-go shim gauges
Mounting Plate
- Place mounting plate on bench with bottom of plate facing upwards
- Install battery upside down with terminals going into terminal cutouts of plate
- Place battery bracket over battery
- Apply threadlocker to 4x M6 x 8mm screws, then install screws into battery bracket mounting holes in the mounting plate. Torque to known safe torque
- Place desiccant packet onto desiccant packet mounting area on plate
- Place desiccant bracket overtop of desiccant packet
- Apply threadlocker to 2x M6 x 8mm screws and install into desiccant bracket mounting holes. Torque to known safe torque
- Install control electronics to mounting plate with long M6 screws
- Connect battery and antenna cables to control electronics
- Thread battery and antenna cables through cable pass-through in plate
- Flip mounting plate over
- Connect battery cables to batteries
- Install antenna with long M6 screws as per antenna mounting instructions
- Connect antenna cable to antenna
Upper Hull
- Clean upper hull with appropriate solvent, then paint logo, contact info, and serial number onto outside of hull and allow to dry.
- Note: paint is difficult to do well in a prototype shop. May be worth outsourcing this to local painting or screen printing shop
- Cut 6x retroreflective tape strips and apply to cleaned hull surface
- Flip hull over apply adhesive spray to flat surface on inside of hull
- Stick foam float to hull and allow to adhesive to dry
Final Assembly
- Place lower hull on bench
- Note: anchor point cutout in bench or raised blocks will make this step easier
- Wipe lower hull sealing surface to remove debris and dust, then place perimeter sealing gasket onto lower hull
- Place mounting plate on-edge beside the lower hull
- Note: may be easier with a jig that holds mounting plate in place, as batteries will likely be heavy and unbalance plate
- Connect UV sensor cables and charge/debug port cables to control electronics module
- Rotate and lower mounting plate onto lower hull, aligning the perimeter bolt holes by eye
- Note: in proposed design, sufficient space exists for rotational alignment of hull to mounting plate to not be of concern. A registration feature could be waterjetted into the hulls to ensure alignment if necessary
- Wipe mounting plate perimeter to remove debris and dust, then place perimeter gasket onto top mounting plate and align holes by eye
- Place top hull onto perimeter gasket
- Install 32x M8 bolts and washers through top of top hull
- Apply thread locker to M8 bolt threads
- Install 32x M8 washers and nuts to bottom of threads
- Tighten around perimeter, then check gasket compression with go/no-go shim gauges
Test:
- Flip buoy over
- Connect debug cable to charge/debug port
- Run control module diagnostics to ensure control module can read all UV sensors, battery health, moisture sensor, antenna controller, etc
Burn-in:
Note: burn-in would be conducted in a salt-water pool with a pulley and fluorescent plastic target at the bottom.
- Attach rope from bottom of pool to buoy anchor point with carabiner
- Place buoy into pool
- With winch, pull buoy until it is underwater to desired test depth (say 1m)
- Confirm that control board is still communicating via debug port
- Enable debug sensing mode where telemetry is sent via debug port instead of antenna
- Leave underwater and sensing for 24h
- Confirm using moisture sensor that housing is not leaking
- Review test data and ensure no anomalies exist in sensing
- Unwind winch to remove buoy, then remove from pool
Post-Test:
- Hose down buoy with fresh water to remove salt, then dry with blower
- Install sealing cap onto charge/debug port as per manufacturer’s instructions
- Package into custom wooden crate
- Insert any relevant product/shipping documentation into pocket attached to crate
Discussion of Technical Risks and Improvements
Maximum Thickness of Vacuum Formed Polycarbonate
Cursory review of vacuum forming literature suggests that the maximum thickness of polycarbonate sheet for vacuum forming is ~6mm, which may not be thick enough to provide sufficient strength to survive collisions with marine debris. It may be possible to develop a process to form thicker polycarbonate, but there is technical risk that the process may not be feasible or may have low yield.
Polycarbonate Stiffness in Perimeter Seal
In the hull perimeter seal, the gasket will resist compression and cause the polycarbonate to bow outwards between the perimeter bolts. With thin polycarbonate and wide bolt spacing, this may cause leaks to occur at peak of the bowing. The bolt spacing will likely need to be adjusted to ensure sufficient stiffness in the perimeter seal. If necessary, a set of polycarbonate rings with the same shape as the perimeter gasket could be cut and placed on the outside of the hull to provide additional stiffness in the perimeter seal.
Power Source and Energy Budget
The stored energy required to power the electronics and satellite connection dramatically impacts the choice of power source and size of the buoy, both of which are tightly connected to the total cost. The ballpark calculations based on the power consumption of an iPhone 11 show significant energy reserves being necessary, but an iPhone 11 may not be representative in terms of power consumption. The stored energy requirements will have a dramatic effect on the design of the buoy and must be better understood to complete the design. Until such understanding is reached, there remains significant technical risk that the proposed design is infeasible due to size or battery life concerns.
Satellite Uplink on Rolling Buoy
The shape of the buoy will cause it to roll in the waves, which may make maintaining a satellite connection with a directional antenna challenging. In addition, the antenna may need to be mounted on the outside of the upper hull to reduce attenuation from the water. Further discussion with the satellite uplink design team is needed to fully understand the mechanical requirements of the satellite uplink system.
UV Sensor Housing Material
The UV sensor housing is proposed to be lathed from 316 stainless steel, but it may be possible to cost-reduce the part by switching to a machinable engineering plastic such as Delrin at the expense of strength and size. Quotes should be obtained for stainless and plastic designs to determine if switching to plastic is cost-effective.
UV Sight Glass Material
Further discussion with the optics team is required to evaluate the suitability of different sight glass materials. The current design has the sight glass tucked inside the housing to ensure good sealing and protection from outside debris. However, this approach may trap bubbles that interfere with the UV fluorescence measurement. If changing to an acrylic sight is feasible, it will be advantageous to redesign the housing and sight design to mount the sight on the bottom outside surface of the housing to allow bubbles to roll off the glass unobstructed.