Dataset of rise velocities for gas bubbles with and without gas hydrate shells

Flow loop apparatus: between September 1, 2018 and February 7, 2022, the flow loop at the Woods Hole Coastal and Marine Science Center was modified to optimize the bubble rise velocity data collection. This process step outlines the apparatus configuration and environmental controls. The flow loop device (FLD) configuration is imaged in the file Flow_Loop_Apparatus.png. The closed loop can be pressurized to 1 MPa (~150 psi) using a syringe pump (not pictured). Temperature control is via the cooling water bath, which pumps chilled water through a tube wrapped around the FLD circulation pump and into a heat exchanger above the circulation pump (Flow_Loop_Apparatus.png, image A). FLD water temperatures can be held between room temperature and ~11 degrees Celsius.
Bubbles are introduced into the FLD at the base of the observation chamber via a gas-filled syringe connected to a vertically oriented needle (Flow_Loop_Apparatus.png, image C). A computer-controlled solenoid valve can be used to allow only a single bubble's worth of gas into the chamber at one time. Bubbles introduced through the needle rise through a 0.91 meter (36") tall acrylic observation chamber (Flow_Loop_Apparatus.png, image A). The chamber's inner diameter (101.6 millimeter, 4") is designed to be large enough for centralized bubbles to rise without contacting the walls.
Bubbles are imaged via a pair of computer-controlled cameras mounted so they face into the cylinder from perpendicular directions (Flow_Loop_Apparatus.png, image B and Browse Graphic). The cameras are set to image only the central portion of the cylinder, near the cylinder's top. This location allows the bubble to reach a terminal rise velocity prior to entering the camera's field of view. The camera collects data from ~90 millimeters of the bubble's rise, with a frame rate of 200 frames per second.
To enhance the image contrast and allow bubble edges to be easily identified, the bubbles are backlit by LED panels, and that backlit light is diffused with a layer of diffuser film attached to the observation cylinder (Flow_Loop_Apparatus.png, image B). Example images collected with this system are given in the image file Flow_Loop_Bubble_Example_Images.png.

Image collection: between February 7, 2022 and March 11, 2022, data contained in this dataset were collected on bubbles observed in the flow loop at the Woods Hole Coastal and Marine Science Center. This process step outlines how the bubble images were collected and how the bubbles themselves were identified within a given image.
Bubble images were collected with a pair of Edmund Optics IDS uEye cameras controlled by a Streampix 8 software package. To maximize the frame rate and the viewable bubble-rise interval, the field of view was set to a relatively narrow 440 pixels wide by 1024 pixels tall. With this field of view, each camera could record images taken at a rate of 200 frames per second.
Images were stored as .avi video files for analysis with Matlab. These images were used to obtain frame-by-frame estimates of the position, size, shape and orientation of each bubble. A Matlab script was written to acquire these parameters with the following strategy: 1) Calculate the difference between frame of interest and the initial frame (the initial image was always recorded prior to the bubble arriving in the field of view). 2) Convert the difference image into a binary image using a threshold of 0.17. 3) Identify any objects in the frame using the regionprops MATLAB function. 4) If the area of a detected object is greater than 50 pixels, then properties such as the centroid, major/minor pixel axis, and tilt angle were extracted using the regionprops MATLAB function, as well as the time in seconds of the frame (used to calculate bubble velocity).
Because the entire perimeter had to be captured with high contrast for this image processing sequence to be effective, images were excluded if the bubble extended outside the field of view of the camera. Additionally, bubble perimeter contrast was low due to low lighting at the upper and lower edges of the images, so bubbles within the ranges 0-100 pixels (top of the image) and 925-1024 pixels (bottom of the image) were discarded. Converting from pixel-based information to bubble position and size in 3-dimensional space is discussed in the next process step.

Image calibration: between February 7, 2022 and March 11, 2022, data contained in this dataset were collected on bubbles observed in the flow loop at the Woods Hole Coastal and Marine Science Center. This process step outlines how the images were calibrated for each run to attain bubble location, size and shape information.
To establish a conversion between the position of a bubble in an image to the position of a bubble within the chamber, we attached a printed grid of points to outer surface of the cylinder such that the two cameras could simultaneously image the grid. One such grid was affixed to what the cameras viewed as the front of the cylinder, and another set of images was taken with the grid affixed to the cameras would see as the far side of the cylinder. The grid had points printed at equal spacings of 2.54 millimeters (0.1"). Because the grids themselves were directly attached to the observation cylinder a known distance below the top steel cap (see Browse Graphic), and the cameras were permanently mounted at a known distance from the cylinder (Flow_Loop_Apparatus.png, image B) the grid images provided a means of converting a camera's image pixel location to a pair of points on the near and far side of the cylinder. These two points define a ray, which can be parameterized by the two points on the cylinder, and a distance between them. When a bubble is imaged by both cameras, each camera provides a pixel correlated with the bubble's center, and hence, each camera provides a ray along which the center of the bubble must exist within the cylinder. The intersection of the two "bubble center" rays (one from each camera) defines the position of the center of the bubble in x, y and z coordinates (see Browse Graphic). Once the location of the center of the bubble is known, the size of the bubble can be calculated from the pixel locations defining the bubble edge.
Once the cameras were calibrated using the grid approach, the grids were removed and a second piece of grid paper was affixed to the cylinder and matched to exactly overly the diffuser panels, which we never removed from the chamber. With the cameras on, the corners of each camera's field of view was drawn onto the grid in pen. Prior to each set of bubbles being collected at a given pressure, temperature and gas type, this field of view "calibration check" paper was replaced atop the diffuser panels, and the cylinder position was adjusted until the corners of each camera's field of view matched those indicated on the paper. Once a match was simultaneously attained for both cameras, the calibration check paper was removed, and the experiment could begin. In this way, the camera calibration could be held constant across each of the bubble conditions tested.

Parameter calculation: between February 7, 2022 and March 11, 2022, data contained in this dataset were collected on bubbles observed in the flow loop at the Woods Hole Coastal and Marine Science Center. This process step outlines how the calibrated bubble images were used to calculate the reported bubble geometries, rise rates and rise paths.
For each bubble image calibrated and processed in Matlab, several parameters were obtained: 1) The timestamp at which the image was taken. 2) The position of the bubble centroid in millimeters for the x, y and z coordinates (see Browse Graphic). 3) The lengths (in millimeters) of the major and minor axes of the bubble. 4) The tilt angle, meaning the angle away from horizontal at which the major axis measurement was made (in degrees).
From these values, the following data were calculated for each bubble image:
1) The bubble volume, recorded as an equivalent diameter, de, in millimeters. The equivalent diameter is the diameter of a sphere with the same volume as the original (generally ellipsoidal) bubble. The equivalent diameter was calculated from the major and minor axes values as the cubed root of the minor axis length, b, times the square of the major axis length, a: de = (ba^2)^(1/3). 2) The aspect ratio, calculated as the ratio of the major to minor axis lengths: a/b. A sphere would have an aspect ratio of 1, and the ellipsoidal bubbles would have aspect ratios exceeding 1. 3) The instantaneous bubble rise velocity, calculated as the total distance traveled by the bubble from one frame to the next, divided by the time between frames. 4) The instantaneous vertical bubble rise velocity, calculated as the vertical distance traveled by the bubble from one frame to the next, divided by the time between frames. Unless the bubble rose only vertically, the total velocity would exceed the vertical velocity.
Once all of the images for a particular bubble rise were processed, the average vertical rise velocity was taken as the slope of the straight line fit to the vertical position of the bubble as a function of time.
Two numerical data file styles are provided in this dataset, and their column descriptions are provided in the attributes sections below. To summarize: the summary spreadsheet, AllBubbles_RiseVelocity_Size_Shape, contains the rise velocity, size and shape information for all bubbles reported in this work. The remaining four spreadsheets contains point-by-point information for four individual bubbles that exemplify endmember rise paths: (1) Small_Clean_Sphere has data for a small, hydrate-free bubble with a nearly vertical rise path, (2) Large_Clean_Ellipsoid has data for a large, hydrate-free bubble with a helical rise path, (3) Spherical_Xenon_Hydrate has data for a nearly spherical, xenon hydrate-coated bubble with a zigzag rise path (see Browse Graphic overlay for this bubble's rise path), and (4) Distorted_Xenon_Hydrate has data for a non-spherical, xenon hydrate-coated bubble that moves laterally while rising rather than oscillating around a single vertical line. The image file Flow_Loop_Bubble_Example_Images.png has the four bubble shapes.

Rise velocity measurements: between February 7, 2022 and March 11, 2022, data contained in this dataset were collected on bubbles observed in the flow loop at the Woods Hole Coastal and Marine Science Center. This process step outlines which gasses and test conditions were used, and how the bubbles were released.
To measure rise rates for a specific gas, pressure and temperature, the chosen gas was drawn into a syringe pump. The laboratory air (room air) was simply air drawn in from the laboratory, with no further chemical analysis. Methane and xenon were purchased from the manufacturer with the manufacturer’s purity ratings given below in this process step. The syringe pump pressure was then balanced with the flow loop water pressure, which was controlled by a separate syringe pump. Two strategies were used to obtain individual bubbles:
1) The burst strategy for obtaining a single gas bubble was to elevate the gas pressure approximately 1 psi (pound per square inch) above the water pressure. The solenoid valve in the gas line was triggered to open for between 25 and 45 milliseconds. While this would often result in a single bubble, this process could also generate multiple bubbles, in which case no images were taken. Only individual bubbles, rising independently, were recorded for this dataset.
2) The continuous strategy for obtaining an independent bubble was to hold the solenoid valve open. The syringe pump was then set to continuous flow mode rather than continuous pressure mode. The flow rate was then reduced until the bubbles appeared at intervals of 20 seconds or more. At this interval, their rise rates were indistinguishable from single bubble rise rates obtained using the burst strategy.
For both strategies, the exact parameters required to obtain single bubbles varied from gas to gas and pressure to pressure, and had to be tuned for each experiment.
Prior to each test, the flow loop water was replaced with fresh tap water, and the flow loop was circulated to establish the desired water temperature. Because the flow loop water is initially fresh tap water, the xenon concentration is essentially zero to start with. Hydrate formation on a bubble is facilitated by elevated concentrations of the hydrate-forming gas in the surrounding water, so for the xenon experiments at hydrate-forming conditions (Flow loop water set to: pressure = 1 MPa, temperature = 13.1 Celsius), the initial bubble injections will be hydrate free until the flow loop water concentration is raised enough for hydrate to form. Consequently, for those flow loop conditions, the list of tests run below includes both hydrate-free and hydrate-coated xenon bubbles. Establishing what this critical xenon concentration in the flow loop water is for triggering hydrate formation will be included in a separate data release looking at gas dissolution rates from bubbles with and without gas hydrate shells.
For the hydrate-free bubbles, the following flow loop water conditions were tested: Laboratory air (atmospheric pressure): Pressure: .101 Megapascals, Temperature: 26.1 Celsius Methane (99.99% purity) (atmospheric pressure): Pressure: .101 Megapascals, Temperature: 24.3 Celsius Methane (99.99% purity) (elevated pressure): Pressure: 1 Megapascals, Temperature: 26.0 Celsius Xenon (99.999% purity) (atmospheric pressure): Pressure: .101 Megapascals, Temperature: 23.1 Celsius Xenon (99.999% purity) (elevated pressure): Pressure: 1 Megapascals, Temperature: 23.3 Celsius Xenon (99.999% purity) (elevated pressure): Pressure: 1 Megapascals, Temperature: 13.1 Celsius For the hydrate-coated bubbles, the following flow loop water conditions were tested: Xenon (99.999% purity) (elevated pressure): Pressure: 1 Megapascals, Temperature: 13.1 Celsius

Data archiving: Microsoft Excel version 16.63.1 (22071301) was used to consolidate all data in a series of spreadsheets. Results were then exported to a comma-separated values (csv) file format.