Week Seven – August 4-August 8

Joined the cool BDML summer backpacking — memories for life in the Trinity Alps!

After returning, we started setting up the initial pool experiment and conducted some exploratory tests. I also designed and installed a bridge proof for our whisker sensor.

Week Six – July 28–August 1

📅Monday

1. Met with Mark. Mark proposed some new design ideas including seagrass and beads.

2. Fabricated a new whisker sensor with over 10× sensitivity improvement compared to the original. The improvement was mainly due to extending the bridge length.

3. Investigated sources of different signal frequencies. By adding an external shell to isolate WIV, we confirmed that the 40–60 Hz component stems from the robotic arm. The sensor exhibited high sensitivity—even light contact on the arm induced detectable vibrations in that range. This result is reasonable and supports the system's responsiveness.

📅Tuesday

1. Started constructing and iterating on a wake generator.

2. Collected experimental data using the first-generation wake generator. The whisker successfully detected the wake frequency.

3. Built a new experimental setup, including: a stationary whisker sensor, a new wake generator powered by a different motor, 3D-printed structural components designed for the above

📅Wednesday

1. Ran experiments with the new wake generator, testing 2, 4, 6, and 8 Hz frequencies at a 20deg oscillation angle.

Results: At 2 Hz, the measured dominant frequency closely matched the input. However, for higher frequencies, the detected dominant frequency tended to shift upward and broaden into a frequency band rather than a clear peak.

Analysis: Seals may not detect flow through exact frequency matching but instead extract information from broader spectral features such as amplitude envelopes, phase, or bandwidth. A referenced study also used 2 Hz as its main demonstration frequency, claiming it yielded the best response. Their tests included 1, 1.5, 2, 2.5, and 3 Hz—frequencies we plan to test next.

2. Searched literature and found that typical fish tail beat frequencies are around 1 Hz, further motivating lower-frequency testing.

📅Thursday

1. Started preparations for outdoor water tank experiments.

2. Started to design a new set of clamping fixtures tailored for the open-water setup.

Week Five – July 21–July25

🔬General Progress

1. This week began with a literature-based discussion to analyze possible causes of current experimental results and identify next steps. Two types of experiments are planned next: one with an artificial fish tail and another with a reciprocating cylinder, representing two types of wake generators. However, before moving forward with these designs, the priority remains improving and validating the whisker's sensitivity to flow disturbances.

some relevant ideas are recorded here: https://docs.google.com/document/d/1r2pHJWd1X9XK9VlDJinfKkQ7s5WzrhjP4voA-18InC0/edit?usp=sharing

2. Met with Gianluca on Tuesday and reviewed some promising 3D simulation results. Also finalized a new set of eight models for simulation to study how key geometric parameters (tapering, undulating frequency, undulating intensity, scale size) affect the whisker's ability to suppress VIV.

New model files are uploaded here: https://drive.google.com/drive/folders/15pnSUQYFNWJa3njersTFQlEt06EJh_fN?usp=sharing

More experimental design details: https://docs.google.com/document/d/1z8LfJ_du8KOKEEa-QfvEeRmpgRtO6JUXNjehSBU4bmE/edit?usp=sharing

3. Tested the FBG signals of three types of whisker bridges under robotic arm dragging and manual tapping at the whisker tip. The three configurations were: short–tall, short–short, and long–tall. Comparative analysis showed that the whisker with the longest bridge exhibited significantly higher sensitivity, while bridge height had a less pronounced effect—shorter bridges performed slightly better.

To further validate these observations, two new whisker bases (long–short and extra-long–short) were 3D-printed using the Objet printer and will be tested experimentally soon (first time using the printer, during which I became familiar with its operation and the process of removing support material from printed parts.)

Week Four – July 14–July18

🔬General Progress

This week started with a meeting with Mark. On Monday, we presented the preliminary data analysis of last week's experiments and clarified the experimental plan for the next steps.

Background & Motivation According to the FBG signal results, the whisker did not successfully detect the vortices in the wake of the cylinder ahead. However, we are still uncertain about the cause — whether it lies in the whisker itself, the FBG sensor, or other factors.

Experimental Plan To investigate this, we used slow-motion GoPro recordings to observe the whisker's vibration during the dragging process. (Our reasoning: If the FBG sensor is working properly, sufficiently large vibrations should also be visible to the naked eye.)

We altered two variables:

• Presence or absence of a front cylinder
• Wide face or narrow face of the whisker facing the flow

Under the same flow speed, this resulted in four experimental conditions.

Key Findings

• When the wide face faced the flow, the whisker exhibited 'significant lateral vibration'.
• When the narrow face faced the flow, vibration was minimal but still visible.

These observations align well with basic theory: The projected area facing the flow affects the magnitude of vortex-induced vibration (VIV).A larger projected area (wide face) leads to stronger oscillations. Thus, the whisker can indeed detect its own vortex-induced vibration (VIV).

However, under otherwise identical conditions, the presence or absence of the front cylinder caused almost no noticeable difference, even when the whisker and cylinder were placed very close. This suggests that the whisker's current sensitivity to wake-induced vibration (WIV) may still be insufficient.

Faced with this problem, we revisited some literature for inspiration.

For example, one paper emphasizes the role of the natural resonance frequency in amplifying the oscillation response. The response is maximized when the vortex shedding frequency matches the whisker's own natural frequency. This inspired us to consider: perhaps the vortex frequency generated in our setup is far from the whisker's natural frequency, resulting in weak response. In the coming weeks, I plan to establish a classic vibration model of the whisker to evaluate this possibility.

Additionally, we drew inspiration from a fish-tail experiment in the literature, which showed the application in target motion status monitoring and tracking. This might provide ideas for future experimental setups. link: https://doi.org/10.1016/j.nanoen.2022.107210

1. Designed and 3D-printed the third version of the fixture, which allows flexible adjustment of the distance between the object and the whisker.

2. Conducted robotic arm dragging experiments, and performed data acquisition and processing.

• Each drag generated a csv data file, where the arm moved from one end of the tank to the other end.
• Three key parameters were varied for preliminary comparative studies: dragging velocity, object-whisker distance, and cylinder diameter.
• Data processing of a single csv file included the following key steps: identifying the active segment, conducting frequency domain analysis, and extracting the dominant frequency and its amplitude from the power spectral density (PSD).
• Some comparative plots are also drawed and discussed.

This week began with an Arduino tutorial led by Mark, where I used Tinkercad to simulate basic circuits and gain a clearer understanding of microcontroller programming.

At the same time, I studied the control system and code structure for the Flexiv robotic arm, preparing for upcoming experiments involving object dragging and signal recording.

In parallel, I practiced 3D printing and learned to handle common issues — for example, adding a 2-layer raft or splitting parts to prevent spaghetti-like failures.

1. Design of the initial fixture To conduct the whisker sensor dragging sensitivity tests, we needed a device that could precisely adjust the distance and angle between the whisker and the object. Therefore, I designed such an adjustable fixture using SolidWorks and printed it using a 3D printer, followed by assembly to prepare for the experiments

2. Setup of the robotic arm dragging test Together with Hao, we assembled the initial testing environment using the Flexiv robotic arm and the first version of the fixture. During the first round of dragging tests, noticeable vibrations in the mechanical structure were observed, which introduced noise and could affect sensor accuracy. This result indicated the necessity for optimizing the fixture.

Learned the hard way that “looks good in CAD” doesn’t always mean “a good design.” Tolerance, stability, and tool access all matter—big time.

3. Optimization and retesting In response, I modified the fixture design to improve mechanical stability, then printed and reassembled the updated version. Repeating the dragging tests under identical conditions showed significantly reduced vibration and improved experimental stability. (Additional minor refinements were implemented the following week, most notably, the upper rod was significantly shortened.)

4. Baseplate assembly improvement We also identified an assembly issue with the three-hole whisker baseplate. Excessive glue application during assembly reduced the available space, preventing proper nesting of parts. I redesigned the baseplate to address this problem. However, the updated design still has a minor limitation: the spacing between the posts and the holes is slightly tight, which prevents one of the three coils from fitting properly. This means only two out of three coils can be inserted without difficulty. This small issue will be addressed in future design iterations.

Explored vortex formation around seven different object shapes in two dimensions under identical flow domains, inlet velocities, and hydraulic diameters. The simulations allowed for comparative analysis of wake vortex street patterns and their distinct characteristics.

1. Conduct underwater dragging experiments to test the sensitivity of the whisker sensor, and iteratively optimize the whisker structure based on performance.

2. Perform 2D CFD simulations comparing objects of varying sizes. If notable findings emerge, plan to discuss wake region behavior with Professor Gianluca for further expert insights.

• Completed online, lab-specific, and PRL-specific safety trainings to gain full lab access
• Joined the BDML Slack and Google Drive
• Helped with lab cleanup
• Participated in the 2025 BDML Summer Kickoff at Redwoods State Park — met the team, shared summer goals, and enjoyed a group hike
• Created and updated my personal Wiki page and this summer blog page

• Followed Hao to learn the whisker assembly process and basic signal testing methods. Detailed notes on this topic are available here: Whisker Assembly Protocol.
• Studied two methods for flow field visualization in COMSOL: arrow plots and streamline plots. Comprehensive notes are posted here: COMSOL Arrow & Streamline Plot Notes (Notion)

Had a quick brainstorming session with Hao and Tianyu to finalize next steps and research plans.

• Decided to use simulation to analyze the so-called 'wake region' of the object.
• Developed a new experimental plan for whisker sensitivity testing.

• Used Arrow Plot and Streamline Plot in 2D simulations to visualize flow distribution around the object.
• Experimented with changing object size and shape to better match experimental data.
• Goal: Explore how varying the object's interface shape and dimensions affects whisker placement for effective fluid sensing — identifying the true wake (flow) region.
• Simulations are ongoing and will continue next week.

• Design and 3D print the fixtures needed for the new Whisker sensitivity testing experiment ✅
• Continue the ongoing simulations ✅