Blog 19 – Sarkout: The Data Acquisition System

A solar powered flying boat has a lot of sensors. All these sensors measure a lot of data. All this data has to be stored somewhere and somehow it has to be analysed and displayed. Fortunately you learn a lot as a computer science student. First of all you have to know what you have to make, a field called requirements engineering. This is the most important thing because making something no one will use is not a very good investment of the limited time available. This year I have learned for example it is not fun to make certain features no one will use. After knowing what to make you have to design the overall layout of the system. In our case we iterated the previous layout of the design and improved some parts of it.

For my fellow nerds: We used an embedded module with 4g to transfer the data from the boat to a powerful server provided by our dear friends at Hewlett Packard Enterprise. On our powerful server we parsed the data and stored it in a database. At the same time it was send to a Play framework application which sends the data to the end client by using Websockets. This means that we don’t need to query databases which can be time consuming. At the same time the data is stored for later analyzation. A meteor application on the user-side receives the data and visualizes it.

Naturally there is still a lot to do , for example determining the optimal speed for the boat depending on the weather and battery conditions is still mainly a manual task. This could be an optimized algorithm which takes all these factors in account and determines the most efficient speed for the boat. Also analysing the data could be a huge factor in determining if the strategies used in the past were indeed as effective as we thought they were. These kind of analysations need a lot power and a team who knows how to handle these amounts of data. Luckily there are new courses at the Delft University of Technology focused on Big Data and cloud powered applications. This means that we will be able automate the previous mentioned tasks as soon as possible. Which in turn could give us the final push for victory.

Blog 18 – Luc: Energy Box Mini Series Part V

In the previous part of this blog mini-series I have discussed on the fuel gauging system used in the energy box. In this part the cooling system will be the main topic.

Before I spent all of my time building solar powered hydrofoil boats, I’d like to use my free time for gaming. I even built my own ‘gaming rig’ so I could play all of my games in 1080p at 60fps. At some point in time I even decided to compete in a casemodding competition, surprisingly I actually managed to make in into the finals of the contest. One might wonder why I am telling you this in a blog about the solar boat. Well, as you may have noticed in my previous blogs, I get my inspiration for designs from seemingly unrelated topics, and with casemodding this is also the case (pun intended).

The energy box is designed to be extremely efficient in terms of power loss and weight. However, at maximum power we can discharge the battery at about 200 amperes of current. An important thing to note is that power loss due to resistance is actually the current squared times the resistance. This means that at a discharge of 200A, 1mohm of resistance dissipates 40W of power. Even though it is designed for high currents, our energy box would dissipate around 500W of power, equal to a powerful desktop computer. A desktop computer uses fans to blow away the hot air and pull cold air in. However, regulations for the Dutch Solar Challenge state that the battery may not come into contact with water, even while the boat has practically sunk. This is where it gets fun.

Using fans is of course a no-go since it requires us to make huge cut-outs through which air can flow (or water for that matter). Other possibilities include the use of heat sinks, but dissipating 500W passively through conduction would require very heavy heatsinks. Now to come back to my story of casemodding. In casemodding there is actually only one real way to go and that is water cooling. Usually water is pumped through heatsinks and led through radiators which are cooled by fans and cold air. In our case, it is pretty much the other way around. We heat up cold water in a radiator by blowing hot air in the box through the radiator. The water is then pumped through the rear strut which acts as a heatsink and is cooled by the flow  of the water we fly over. Based on my knowledge of casemodding I knew what components to find. We contacted Aquatuning, a supplier and manufacturer of casemodding components and luckily our plans were received with great enthusiasm. Before too long we got our hands on three Alphacool 360mm radiators and a large supply of liquid coolant, including red UV ink which is convenient for discovering leaks and spills (and it looks really cool under a blacklight). We also partnered up with EBM Papst for the fans that we use since we are big fans of their fans. We even selected the fans on their noise production, nobody wants a loud boat.

The nice thing about the custom watercooling loop is that we can incorporate all other hot components as well. These are the motor and the motor controller. Luckily, these components were already made to be water cooled so we did not have to make any custom parts. A practical problem is that connecting the energy box to the motor, and inside of the energy box the radiator to the motor controller would mean loss of modularity, one of our main design focuses. If we had to sever and refill the cooling loop every time we took out only one of the components, many valuable hours would be lost and the chance for errors is much greater. After scouring the casemodders fora we found yet another solution in the form of quick-release water couplings built by EPC. Normally these are used in medical centers and clean rooms, but we saw a chance for us to keep our modularity. By using the couplings we can just pop the energy box in and out of the boat without having to vent the cooling loop every time.

This is the end of part five of our mini-series on the energy box.

Blog 17 – Luc: Energy Box Mini Series Part IV

In the previous part of this blog mini-series I have discussed on the safety issues tackled by the energy box. In this part fuel gauging will be the main topic.

One key part that is needed for our strategy is something that might seem obvious, but in practice is quite difficult to determine: how full is the battery? Batteries are often said to be full at a specific voltage and empty at another voltage, e.g. a rechargeable NiMH battery is full at 1.25V and empty at 1.0V. However, a battery’s voltage drops as a current is drawn from it and so it might seem that a loaded battery is empty while in fact it still has some juice left. The way we determine how “charged” the battery is, is by actually measuring the charge stored in the battery (charge is a physical quantity with coulomb as its unit). Since current equals coulomb per second, we can determine how much charge is stored in the battery by integrating the current over time.

A major problem with this method is that even a small offset in the current measurement leads to large errors over time. If this were the case, then after a while the calculated charge will be either much lower or higher than it actually is. To tackle this problem, we have partnered with Isabellenhütte to use their IVT-MOD line of high precision current and voltage sense systems. Using this system we can very accurately tell how much charge is stored in the battery.

The charge stored in the battery gives us a fairly good indication of how much energy is stored in the battery (but not perfectly due to losses in the cell chemistry). However as I said before, the voltage of a battery cell drops with increasing current drawn from it. I have also said in a previous part of this series that we limit the cell voltage of the battery to a certain safe bottom limit. Now picture this: The boat has almost finished a stage with the finish line in sight. Of course the battery will be nearly empty at this point, but we are only a few seconds in the lead of our competitors. If they didn’t now any better, the strategy team will now tell the pilot to give all the throttle he can to make a final sprint. If this were the case, the battery might be stressed so far that the EMS triggers a shut-down of the boat to keep the minimum cell voltage above the set limit and we would lose the stage. This shows that it is very important how the voltage drop corresponds to the current.

This sounds much easier than it actually is, in reality the relationship between voltage drop and current depends on a huge number of parameters especially temperature and state-of-charge. We have given our best shot at determining the relationship anyhow using a test set-up using some very nice instruments from Keysight. The results are shown below. The characteristic shows that the voltage drop per amp (internal resistance) actually increases when the battery is near-empty. Using this graph, the strategy team can give a reasonable estimate of how much throttle the pilot can give resulting in a somewhat slower sprint but in this case we will cross the finish line.

This is the end of part four of our mini-series on the energy box. In the next episode we will talk about one of the coolest things of the energy box: the cooling system (get it?).

Blog 16 – Luc: Energy Box Mini Series Part III

In the previous part of this blog mini-series I have discussed the battery pack housed in the energy box. In this part safety will be the main topic.

Lithium-based batteries are quite nice. They can contain (fairly) large amounts of energy at a low weight will also being able to deliver lots of power. They do have their downsides however, for example they can’t be turned off. This is true for all chemical batteries but since they are so energetic it is a larger problem. Below you can see the energy that is delivered by one of our battery modules as a function of its voltage. Normally we would say that it is ’empty’ at 3.0 volts. However, this is not entirely true or rather it is entirely not true. As can be seen, the battery can be discharged even further (down at 0V when it is truly electrically empty). However, once a Lithium-ion battery is discharged below 2.4 volts it is irreversibly damaged and going down to 0 volts it is deemed dead. In most cases this very unwanted, but not very dangerous. However, if someone were to try and recharge the battery it could potentially catch fire or explode due to the chemical processes in the battery cell.

So there are two problems here. The first is that we can’t power off the battery and so the boat will continue to consume energy slowly discharging the battery. The second problem is that we need to shut down the battery in case of an emergency or when the voltage drops too low. We evade these problems by not actually shutting down the battery, but disconnecting it from the rest of the systems by means of two relays. One big Gigavac relay which can handle hundreds of currents for discharging and one smaller relay for charging. The relays are controlled by our energy management system and are wired through the emergency stop and deadman switches, interrupting the current whenever the pilot or the system detects possible danger.

Even though the relays can handle hundreds of amps and can interrupt even more, they are not meant for emergency in-operation switching. In the case something does awfully go wrong we also employ a Carling Technologies circuit breaker. This specific circuit breaker is designed to interrupt over a thousand amps. A circuit breaker acts like a fuse in the sense that it will stop the current flow when it senses the current is too high, but it has the nice ability that it is resettable and therefore does not need replacement.  Unfortunately, if the pilot is too enthusiastic the circuit breaker might trip, setting the boat to a standstill. Normally it would be necessary for the pilot to open the hatch to the energy box and manually reset the breaker. This would cost precious racing minutes, so we fitted the circuit breaker with a remotely operated motor so it can be reset with just a simple press of a button.

So now we have a system that can interrupt the current flow out of the battery in any given situation, sounds safe right? Well technically yes, but very rarely there may still be a problem. Battery technology has rapidly advanced since the development of the first lithium based battery back in the 90s allowing us to build the energy box in a very compact manner. However there is still a very slim chance that a battery cell might be faulty coming fresh out of the factory, think of the recent problems with “hoverboards”. If this is the case, our relays and circuit breakers are useless because a battery cell might catch fire just out of nothing. It would be a waste of the energy box not to speak of the danger it might be to the boat and the pilot.

A battery based fire is not easy to control. Some people may remember the “triangle of fire”, the three necessities for a fire to break out. These are heat, fuel and oxygen. Unfortunately, all of these three factors are fed to the fire by the battery cell itself and therefore a fire cannot be controlled in the traditional manner. Fortunately, the “triangle of fire” is actually the “tetrahedron of fire”, which sounds way cooler but also shows us another factor which must be present for a fire to exist: A chain reaction. A fire is nothing more than the combining of oxygen to other molecules in a chemical reaction. If you could in some way block the reaction from happening, the fire will die out. This is exactly what our firepro fire extinguisher does. It uses an aerosol mixture that interrupts the chemical reaction and thus it is able to stop a battery fire. It even leaves all of the electronics intact, unlike powder extinguishers.

This is the end of part three of our mini-series on the energy box. Want to read more about our energy box and his friends energy box 2 and energy box 3? continue reading part four which is on the topic of fuel gauging.

Blog 15 – Luc: Energy Box Mini Series Part II

In the previous part of this blog mini-series I have discussed the general layout of the energy box. In this part the battery will be the main topic.

Ten months ago, I wrote a blog about the design of the battery. At the moment of writing, I had only just selected the battery cells we would be using but I had spent almost no time in the actual design of the battery itself. I thought that choosing the right cell for the job would be the hardest part in building the battery pack. I was right but nevertheless it would take another three months and about four iterations before the first battery pack was finally finished.

As depicted in the photo, the battery pack consists of twelve battery modules placed in series to form a ‘U’ shape, having the positive and negative poles of the battery at the same end. In this way, every cell is exposed to free airflow for cooling and the pack would be just a bit smaller than a solar panel meaning we would only have to manufacture a single hatch for the electronics.

Each battery module in its place consists of fourteen cells placed in parallel. The battery tabs (plus and minus terminals) of the batteries are soldered to a copper busbar, which is very uncommon in the world of batteries. Normally, batteries are spot-welded to nickel. The drawback of nickel is that it has a higher resistivity than copper and thus it incurs more losses at high currents. Spot welding copper to the batteries to a copper strip is unfortunately impossible however. Why? You might wonder, well it is because the welder itself is also made of copper and if you were to try and weld the battery you would only end up in welding your welding equipment to the battery.
Being the naive designer I am, I said: ‘I am going to weld it anyway!’. I knew spot welding would be difficult so I had replaced spot welding with /laser/-welding, how cool is that!? Even better, the pulsed laser-welder we were using is located in the Reactor Institute Delft, an actual nuclear. The early test samples we made of busbars laser-welded to batteries came out fine and so we decided to stick with the plan I had. Then Murphy struck and we found out that one in about every hundred laser pulses was far more intense than the other pulses and could potentially penetrate the battery cell housing and damage the cell.

At this point I did not know how to solve the problem. In my desperation I turned to the previous chief electronics Gijs Bruining for advice. He had the very simple, but in my head preposterous idea of directly soldering the battery tabs to the copper. In every single paper I read on connecting batteries it was always stated that soldering batteries was a no-go since the heat of the soldering iron would damage the cell and send it to battery hell. Since I had nothing to lose I tested this method and to my astonishment, the battery was still fine and the contact resistance between the busbar and the battery was even lower than when welded. After some more samples I reluctantly thanked Gijs and mass production of battery modules started. This went slowly at first, but after some practice we got the hang of it. Fun fact, our two racing packs were built in halve a day.

After testing the battery pack thoroughly I was pleased to find it had a capacity of 1508Wh, within a percent error of the maximum and thus it would not result in a penalty.

And thus I conclude the this part on the energy box. Tune in next time for a story about safety.

Blog 14 – Luc: Energy Box Mini Series Part I

The energy box might just be the most complex component of the boat, judging by the number of its components in SolidWorks (not to speak the many thousands of lines of code written for its functionality). Due to this complexity and the number of functions it fulfils, we will dedicate a sort of mini-series of blogs to the energy box. In this first part we will have a friendly introduction to the functions, components and design philosophy of the energy box.

The energy box is the beating electronic heart of our solar boat. All of the high-power electronics is housed in this box (with the exception of the MPPTs). This includes the batteries, the motorcontroller, our battery-management-system as well as our energy-management-system and all of the safety features such as relays, circuit breakers and even a fire extinguisher. In previous years, the same functionality was split over two separate boxes: the battery box and the power box. However, you could only use any one box if you had its partner with you as well. Therefore, we decided to make them into a single box. Not only does this make testing a lot easier (if you have a problem with energy box 1 just pop in energy box 2 or 3), but it also decreases weight and losses in the connections needed to couple the two boxes.

The energy box is divided into two compartments, one for the battery and the other for everything else. The most notable part is the motorcontroller, a solid aluminium block with potted electronics inside which can drive our motor. Of course we want to be able to physically shut-off the battery from other electronics systems when the battery is depleted or in case of a malfunction. To do this we use relays, electronic switches comparable to a light switch but remotely controllable. Between the battery and the main relay we have placed a current sensor so we can determine how much power the boat is consuming or producing and what the state-of-charge of our battery is.

All of the power connections in the energy box are made using busbars, which is a fancy name for solid copper strips. We have chosen busbars over regular cables since they can be bent in much more extreme forms, allowing us to make the box very compact. Busbars also are capable of handling more current compared to a cable of the same thickness (if you ignore the skin-effect). Eventually, the busbars end in the power connections on the outside of the energy box which are waterproof connectors which I find exceptionally sexy myself.

This is the first inside of our energy box, check back for the second part about the energy box in which I will discuss the battery we have built.

Blog 13 – Bart: From sensors to wing actuation: the control system

To win the Dutch Solar Challenge, it is key that our boat can perform a stable flight on its two wings. To prevent our boat from becoming completely airborne on one hand, but still lifting the boat out of the water on the other, a control mechanism is necessary. The new TU Delft Solar Boat will therefore feature active control to fulfil this job.

The new boat features double wing control. The controller attached to the front wing is used to compensate for disturbances such as waves and gusts. The rear wing controller, which is the wing that produces most of the lift, aims to trim the boat. This means that the deck is completely horizontal. If the boat is completely level the aerodynamic drag created by the hull and deck will be minimum and that helps us to fly more efficient and win the Dutch Solar Challenge!

In a nutshell a control system does the following: First it senses the relevant parameters of the boat, like for instance the height above the water and accelerations. Then it uses this knowledge to send a feedback signal to the actuators. The actuators then start moving the wings of the boat to change the amount of lift they produce. In this way the flying height of the boat can be controlled.

There is more to the control system than meets the eye, so let’s go into a bit more technical detail. The actual control loop itself as you would design it in Simulink is just one link in the chain. I will walk you through the controller from sensor to actuation.

To achieve high run-time performance on the controller and synchronization of all the data inputs and outputs, a real-time operating system (RTOS) is implemented. The advantage of an RTOS is that it allows you to execute different tasks at the same time by using threads. Every thread can be considered a stand-alone program. A scheduler allocates processor time to every thread so that it seems like they are executed simultaneously. You can compare this to a university quarter where you are taking 4 courses at the same time. You have 4 courses to do, but your lecture schedule allocates time slots in which you can work a single course.

The state of the boat is measured by sensors. In our case, the height, acceleration, rotation and velocity are being measured. Ultrasonic height sensors and an inertial measurement unit with GPS are used to collect all these state parameters. Every sensor has its own thread in which data is collected. This allows sensor data to be collected at specific sampling frequencies, depending on the sensor.

In-house designed filters are then applied to process the sensor signals to obtain a robust and accurate estimation of the dynamic state the boat is in at that moment. This is where your Simulink model comes into play. A feedback signal is computed using the measured states. For instance if the boat is losing height and is flying to low, the feedback signal will tell the actuators to increase the angle of attack of the wings to create more lift. In this way the control system will keep the boat at the desired flight height and attitude.

To test if all the sensors are working together before the boat is actually finished, all the electronics and sensor were mounted on a simple wooden board. All the sensors are constraint in the same manner as they would be in the real boat. This allows us to simulate a flying boat by simply moving the board. I am really looking forward to start testing the control system on our new boat in a couple of weeks!

Blog 12 – Floor: The evolution of solar energy

How can a boat with a tiny battery of only 1.5 kWh sail a distance of 60 km within two hours? Simply, with the use of solar energy! Answering this question was not as obvious as 30 years ago. The solar cell industry has taken huge steps the past years. Around 1980 the efficiency of a research Silicon solar cells was around 16 percent, meaning only 1/6th of the received light is transmitted into electrical energy. Furthermore this energy was relatively expensive, one Watt of energy cost around 20 US dollars.

The efficiency of research Silicon solar cells is currently 25.6 percent, which means more than a quarter of the energy is transmitted into electrical energy at only a fraction of the price of the 1980’s. The price of one Watt of solar energy is currently only 0.30 US dollar.

This is why solar energy becomes attractive, on houses but also on vehicles. Especially when lightweight panels are used, encapsulated with polymers to protect the cells to the environment. Our boat currently has 8.2 square meters of 23.9 percent efficient Silicon solar cells (the most efficient cells which are currently on the market), together delivering 1750 Watt during a sunny day.

These panels are composed of 527 cells, encapsulated into 34 panels which are again divided into 11 arrays. These arrays are connected to our MPPT’s (which we made ourselves) which will find the maximum power point on the voltage/current diagram and make sure we will have the highest output possible. With this energy the boat is able to sail over 50 kilometres per hour at its top speed.

The potential of solar energy is not only seen by racing boats like ours but also on pleasure yachts. Only they use the power transmitted from solar energy for their on-board systems (fridge, lights etc.). However, given the current developments of solar cells and electrical engines, sailing 100 percent climate neutral might be a possibility in the near future!

Blog 11 – Sophie: Production: where every detail is crucial

What is the point of designing a perfect, fast and reliable boat, if it is impossible to produce?

The step between the design-part and the produce-part of our solar boat isn’t a small one. It is actually a major transition, which is easily underestimated. When it comes to the hull & body, there are a lot of aspects to take in consideration.

As you probably know by now, it takes quite some time and a lot of focus to design a solar boat. The building-part is just as complex. Every little thing needs to be perfect. This means that laying even a bit of carbon fiber, can take an entire day! Even worse; if you find out you didn’t produce or design a part optimally, you have to start all over again. All these risks should be included in the planning of the production. Apart from margins that are already taken, there are always non-predictable influences. An example: at our production location, we were never sure when we could use the oven to cure our product.

Time is not our only enemy when it comes to producing the hull and body. The solar boat isn’t an object that could just be built in your garage. Of course it is not that small, and there are facilities and tools necessary. Where do you normally find a 6-meter oven? We found one, and the rest of what we needed, at Rondal all the way in Vollenhove (North of Flevoland), which is a three hour drive from Delft. Not something you want to travel twice a day for seven weeks. That’s why we stayed in a very tiny mobile home in the middle of nowhere. This was a challenge itself.

Taking the logistic part into consideration,  there were more aspects you wouldn’t think about while designing a solar boat. For example: the hull was first too wide to fit through the door and leave the building. The choice was to molest the door and building, or to make sure the hull would become some millimeters smaller. We choose the last one, surprisingly.

Let’s go back to answering my first question: there is absolutely no point in designing a perfect, fast and reliable boat, if it is impossible to produce. It will either take forever, or the entire hull will get stuck somewhere (if you even find a place to produce it).