Development

Progress Log

Week 1 | 13 – 19 May 2019
We arrived at this idea after going through several failed ideas. We wanted to do a project on biomimicry, and shark skin denticles presented an interesting problem, as it had been extensively studied in water, but was less well studied in air.

We initially hoped to scale up the size of the denticles such that they could be manufactured more easily and possibly on a high speed vehicle such as a bullet train.

Week 2 | 20 – 26 May 2019
We decided to focus on riblets inspired by denticles instead of the denticles themselves, as we quickly realised from literature that the performance and effect of denticles were complex and results varied quite widely between papers. Furthermore, the drag reduction properties of shark skin are greatly enhanced by skin deformation due to its flexibility. Riblets on the other hand are a simplified structure, which are simpler to manufacture and simulate.

At this point, our plan was to simulate the riblets on a CFD software such as COMSOL or ANSYS Fluent. We decided to split into teams, each focusing on one CFD software to conduct simulations on the various riblet geometries to compare and corroborate the data obtained. After determining an optimal riblet geometry, we intend to manufacture the textured surface on a model vehicle for wind tunnel testing purposes, allowing us to measure and observe the drag effects.

Week 3 | 27 May – 2 June 2019
We met Assoc. Prof. Eddie Ng from MAE to consult him on the general direction of our project, and to discuss our simulations. Upon reviewing our plans going forth, Prof. Ng highlighted the numerous difficulties we might face when working with CFD simulations, particularly in terms of getting usable results. We were cautioned against

He also advised against using the wind tunnel, as a lot of work needed to be done to calibrate the instruments used. Furthermore, there was a long queue for booking the wind tunnel, and the MAE building was to undergo renovation soon. As such, we abandoned the wind tunnel idea, although we still aimed to manufacture a set of riblets, possibly for further testing.

On both Ansys and COMSOL, we decided to get used to wind tunnel simulations by running them on a flat surface in a wind tunnel. Once we were used to the simulations, we imported the riblet geometries from Fusion 360 and ran the simulation on them. Throughout all these simulations, we set the wind flow velocity to 88.9m/s, roughly the speed of a bullet train.

Week 4 | 3 – 9 June 2019
Two of our members were away on a trip to CERN this week. As such, progress was slower.

We realised our simulations had been erroneous due to a lack of sufficient understanding of the literature. It turns out that the airflow is supposed to go along the riblets as opposed to against the riblets. This allows the streamwise vortices that form in turbulent conditions to be lifted from the surface, should the riblets be sufficiently tightly spaced.

Progress Update 1 |5 June 2019
Not much feedback received. Our findings thus far were quite rudimentary, due to insufficient understanding of the theory, as well as the steep learning curve of the CFD software.

We also started sourcing for external printing vendors who may be able to produce our riblets at the desired scale.

Looking into alternative applications of the riblets as a contingency plan, we investigated using the riblets on blades (such as in turbines or drones).

Week 5 | 10 – 16 June 2019
Having revised our simulations since the previous week, we continued to seek out the optimal geometry for riblets in air.

After a discussion, we felt that a viable application of our riblets would be on a wind turbine, as they turn at relatively high speeds and are likely to experience quite some drag on their blades.

We also further refined our models based on other papers on riblets, changing our model to LES (more detail on Simulations & Analysis).

Week 6 | 17 – 23 June 2019
After meeting with Mr. Kanesh on 18 June, we revised our plan based on some of his suggestions.

Firstly, the idea of a turbine was discarded, as he mentioned that turbine blade designs are highly-confidential within the industry, of which optimised geometries are unlikely to be easily found online. As per his suggestion, we decided to look towards vehicles for testing our drag, potentially a model car.

On the simulations front, we decided to run simulations progressing from a laminar flow model until a LES-based turbulent model. By running this against a standard geometry, such as a sphere, we could collect values such as the drag coefficient and plot it over a series of Reynolds numbers, comparing it with existing literature and theory. This allowed us to know whether our simulations were producing reliable results, a particularly important consideration at small length scales.

Progress Update 2 | 19 June 2019
After receiving feedback from the second progress meeting, we were advised to incorporate more hardware into our project by the end of the module, and thus decided to shift the focus of our project away from simulations.

One of the suggestions given was to to apply our ribleted surface on a flexible ‘skin’ and attach it to a Tamiya car for a ramp test. However, not only were we were unable to get the desired feature size on such flexible materials, but the geometry of Tamiya cars were also rather complex in nature.

As such, we looked into the possibility of using a more simplified geometry such as an Ahmed body, which is a benchmark model widely used in the automotive industry for validating CFD simulations. We planned to conduct ramp tests using  four-wheeled Ahmed bodies of varying surfaces – one smooth and the other ribleted. A standard Ahmed body is depicted below.

 

Apart from the dimensions of the Ahmed body, there were several other design considerations to be made. These include whether the body should be hollow/solid, whether a motor should be installed to drive the system, and whether the ramp surface should be lubricated (which might potentially cause slippage in addition to rolling).

With these, we started sourcing for the various components needed.

In order to attain the desired feature size on the surface of the Ahmed body, we decided to look into external precision laser solutions such as micro-cutting and engraving. The main idea was to print flat ribleted plates that adhered to the top and side faces of the Ahmed body, and these textured plates would then be attached to a 3D-printed Ahmed body. However, this idea came to a standstill due to cost limitations (most quotes were above a thousand dollars 😱 for the pieces needed).

Several ideas were proposed for the ramp. As the surface of the ramp had to be as smooth and sturdy as possible, we considered using a full-body mirror.

By tilting the above mirror from its original vertical position to the orientation as shown, we are able to slide the Ahmed bodies down the inclined surface to conduct ramp tests. However, we were apprehensive towards this approach as the length of the mirror was relatively short, and this meant that the time taken for the Ahmed bodies to travel down the ramp would be rather short, increasing the inaccuracies in our data obtained due to human error.

In view of this, an alternative method to use two inclined measuring tapes placed side by side at as the ramp, with one ramp each for the left and right sets of wheels of the Ahmed body. Yet, despite this approach offering a long enough surface for the path of the Ahmed bodies, the tapes are easily bendable and are not very sturdy. Hence, it was not advisable for us to risk damaging such an expensive prototype via this approach.

Whilst we were thinking of ways to proceed with this new project, we decided to keep in mind several other ideas as well. This was because we felt that ramp tests alone might be too insignificant as such experiments were quite common in existing literature, notwithstanding the novel surface tested.

Week 7 | 24 – 30 June 2019
Through some discussions and brainstorming, we came up with the idea of building a measuring instrument for drag! The main idea of this prototype involves rotating textured cylinders at a high RPM, then measuring the deceleration time taken for the cylinder to come to a complete stop. Attached is the very first rough draft of our preliminary design drawn by Hannah, and more details for this prototype can be found under Design Specifications and Working Principle. 

With this new plan in place, we started deciding on the design specifications of our measuring instrument. The first half of the week was spent making a list of all the materials and parts we would need to build the prototype, and started sourcing for them.

A vital part of our prototype were the textured cylinders, as one of the recurring problems brought up in earlier project ideas were the constraints in fabricating a textured surface of our desired feature size. With this, we met with Mr. Rahman from the Mechanical Workshop for the manufacturing of our cylinder. It was prudent to settle this before moving on to building the remaining framework and circuitry as the cylinders would take more than a week to be manufactured.

We also met with Mr. Kanesh on 26 June, and he advised us on several issues with our current idea that we ought to address before going forth. This issues include accounting for the difference in mass due to the varying surfaces, possible back emf generated, and overall stability of the set-up. We decided to look into these pointers in greater detail in the following week.

One concern with the designs we had so far was with the interactions between the motor shaft and rotating axle. As such, we decided to include an additional component between the two. This led to yet another draft that uses an actuator and linear guide to disengage the motor from the rotating axle.

Nevertheless, the prototype design was revised several times over the course of the week after receiving more suggestions and feedback –  the final design can be found at ‘Working Principle’!

Along with editing our prototype design, we split into pairs and set out to source for components locally. Not only were we able to obtain several desired parts immediately, we were also able to ask around for better alternatives and possible modifications to the components we had in mind.

Week 8| 1 – 7 July 2019
In preparation for the upcoming third progress update, we came up with a sketch for the inner workings of our main framework as of then. Here, we decided to employ a z-axis platform mount to engage and disengage the friction clutch connecting the motor to the cylinder for ease of operation.

In addition to this, we considered several other revisions to our design over the weekend due to certain issues that arose over the past week.

The initial framework for our prototype was intended to be built using 20×20 aluminium profiles. However, after constructing a skeletal draft and testing its sturdiness, we decided that the profiles were too fragile and may break apart under the vibration from the motor’s high RPM. As seen from the picture below, this frame has since become a glorified storage shelf for our other miscellaneous items 😂 We have since opted for 30×30 aluminium profiles instead.

Apart from this, the motors that we bought a few weeks ago turned out to be unsuitable for our intended purposes as they were unable to produce the desired torque to spin the cylinders at high RPM. Upon a closer look at the specifications of the motor, it seemed as though they were more suited for powering the wheels of toy cars. In view of this, we consulted Tony for alternative motor options in order to prevent making the same mistake 😅.

Lastly, there was the issue of back emf in the motor causing a magnetic breaking force that would decelerate our cylinder faster than desired. This was a major concern for the feasibility of our prototype as it could potentially damage the motor and power supply.

As such, we further revised our design in the later part of the week, and decided to invert the order of components for the motor to be mounted on top of the frame instead of the bottom. Attached is a picture of the updated framework!

Apart from this, the complete set-up would also comprise additional circuitry including a potentiometer controlled by a stepper motor, an Arduino driver board and a Raspberry Pi for data computation and storage purposes.

Progress Update 3 | 2 July 2019
Since incorporating more hardware elements into our project, the feedback received seemed to be much more favourable. We were greatly encouraged by this and strived to improve our prototype 😊

After the progress meeting, we focused on settling the framework and components of our prototype, along with the design blueprint. As luck would have it, our cylinders were ready for collection! Pictures and specifications of all test cylinders can be found at ‘Design Specifications’.

We spent some time testing the swiveling of components and discovered that the current bearing we had at that time was suited for heavy duty purposes, and hence incurred high amounts of friction. Furthermore, we found that our axle – made of regular aluminium – was too easily bent when spun at high RPM, thus interfering with the alignment of our components. We went with a reinforced aluminium rod instead. As such, the re-ordering of new components was in order.

An alternative to our current bearing proposed was an Igus closed linear plain bearing – its interior was made of a special patented ‘iglidur’ material which allowed an axle to be rotated with minimal friction inside the bearing.

However, one downside of this bearing was that it did not come with housing, and we would have to print an additional mounting bracket for it. Also, such bearings are built to slide along the axle with ease, and are thus not able load-bearing.

Another alternative was the bearing that was amongst the components of an handheld electric drill. Even though we had to test the smoothness of the bearing before determining its suitability for our prototype, the housing of this bearing was ideal as it could be easily mounted onto the 30×30 profiles.

As this bearing came with several other internal drill components as a set, we experimented with these other parts to see if they could be repurposed in our set-up.

Over the course of the week, we went through several draft prints of shafts and caps. Dimension adjustments were made with each new draft to better accommodate our test cylinders. We also included indented grooves (not visible in the pictures) in several of the cap geometries to account for the weight differences between different test cylinders.

To further enhance the overall stability of our prototype, we opted to mount the entire framework on a 1x1m piece of plywood of 2.5cm thickness. This addition was done in hopes of weighing down the entire set-up, as well as for the plywood to absorb some of the vibrations from the high RPM rotations.

With this, a new consideration arose – how were we going to dismantle the set-up and swap cylinders out for another during testing? We settled on mounting our frame on the plywood at the very end after confirming our prototype design, right before conducting our experiments. Several other changes were made to the order of components within the frame to accommodate the removal of cylinders.

Week 9| 8 – 14 July 2019
In preparation for the last progress update on the following week, we decided to settle the circuitry components and the rotational alignment of our prototype. This would allow us to showcase the set-up in its entirety along with the smooth rotation of the central axle. 

After some thought and experimentation, we decided to use the drill chuck to connect the shaft of our motor to the friction clutch. This mechanism is explored in greater detail at ‘Design Specification’. On this note, some time was spent sourcing for materials to make the friction clutch. Possible options were rubber mats, rubber rollers used in printing, and anti-static mats. We eventually went with the anti-static mat as it was already available in the lab, and worked pretty well in our set-up upon testing.

We decided to use a potentiometer controlled by a stepper motor to restrict the current flowing through the main spindle motor. This would be operated with an Arduino driver board which controls the movement of the z-axis as well. More details on our circuitry is elaborated on at ‘Design Specifications’.

To facilitate the organisation of the various components in our prototype, we printed some additional parts to properly contain the circuitry.

 

As the smoothness of the cylinder’s rotation at high RPM largely depends on the alignment of the axle and bearings with respect to one another, a fair bit of time was spent on making precision adjustments to our set-up. 

Lastly, we made the decision to remove the cap from the set-up altogether, and instead fixate the cylinder on the shaft via two crossbeams through the axle – one approximately at the middle of the shaft to prevent the shaft from slipping around the axle, and another at the top of the shaft to prevent the cylinder from sliding up the shaft during rotation. As such, we included grooves around the circumference of the shaft’s base to enable us to account for any weight differences between the various textured cylinders. The finalised sketch of our prototype can be found at ‘Working Principle’!

In light of all these major changes, we kept Assoc. Prof. Eddie Ng updated on our progress.

Week 10 | 15 -21 July 2019
Finally, all the various parts and components had arrived and we were able to prepare for the final assembly of our prototype 🙂 As the final presentation looms closer with each passing day, our goal for this week was finalise and assemble our prototype so as to conduct some preliminary experiments!

We started off the week by heading down to the Mechanical Workshop to get several holes drilled into our axle for the fitting the crossbeams – attached is a diagram.

Once this was done, we spent the rest of the day mounting the cylinders onto the shaft and assembling the prototype. We pasted reflective tape on a strip of black paper before attaching it around the shaft’s base to create a greater contrast and allow the tachometer to sense the rotations more accurately.

Progress Update 4 | 16 July 2019
We showcased a skeletal version of the prototype as shown above, and shared the working principle behind our prototype in brief. Going forth, all that was left was refinements to our set-up, as well as incorporating the circuitry and safety precautions before conducting experiments. 

In preparation for conducting our experiments, we made further precision adjustments to the set-up. This routine consists of tightening all screws and re-aligning the bearings to ensure that the shaft is able to spin smoothly. Also, we inverted the z-axis to prevent it from sagging and causing the motor to be misaligned.

Apart from this, we were also sourcing for possible safety measures to assuage concerns highlighted by Dr. Ho, Mr. Kanesh and Tony. We mounted the prototype on the plywood using self-tapping screws, and this alleviated the issue of stability as the plywood could absorb some of the vibrations during the high RPM rotations.

We also designed and printed a lid for the circuitry box to prevent accidental contact with circuitry components when the power supply is turned on during experiments.

Due to the risk of small objects flying out of the set-up whilst the central axle is rotating, we wanted to build a mesh cage to surround the prototype for shielding purposes. This cage would be held up using bamboo poles and cable ties, and a sketch is shown below.

After building a draft cage, we tested its structural integrity and realised that the cage was too wobbly, failing to hold its shape. As such, we abandoned this idea and decided to cable tie the mesh layers to each other and the bamboo poles directly.

With our prototype fully-completed, we started to run some preliminary tests to gauge how much data we could collect per round. 

We also began working on the slides for the final presentation. 

Week 11 | 22 – 28 July 2019
As the CNYSP Freshmen Orientation Programme took place on 22-24 July, all our group members took a short break from MnT to partake in camp activities!   

For the remaining part of the week, we spent most of our time conducting experiments with the various cylinders. As we were moving on to high RPM rotations, we conducted all experiments with the cage placed over the prototype. To ensure that we were not in the line of fire of any loose objects flying out during the experiments, we ensured that our prototype could be operated remotely and made sure to stay at a safe distance throughout.

We also tried to understand the data collected by tallying the numbers against previous data sets and corroborating the results obtained with the simulations we have already run.

Unfortunately, our progress came to a temporary halt when our motor clutch was damaged during rotation. We noticed this when the top half component of the friction clutch started to jut out of its original position at an angle during one of the runs. After detaching the component from the prototype, we found that we were unable to salvage the part as the screw head had dug into the solid.

As such, we started re-printing an identical part as shown on the right. Since a new part would also require applying a layer of epoxy to attach the anti-static mat cutout, we decided to continue with the rest of our experiments the following week.

Finally, the end is in sight! 💪

Week 12 | 29 July – 4 August 2019
It’s finally here – Presentation week! All that’s left is data collection and preparation for the final presentation itself 😊

Monday was spent conducting one experiment after another, with routine screw re-tightening checks done in between. We were met with a minor hiccup when the bottom portion of our friction clutch was ‘beheaded’ in the middle of an experiment. Luckily, our cage prevented it from flying out and hitting us 😅 What a relief! After replacing it with a spare part, we resumed experiments once again.

Selected data can be found at ‘Working Principle’.

Tuesday was a lab-free day as we spent some time rehearsing for the final presentation. D-1 to freedom 🙌