astex fyt2009 final report v2 - european space...

Education

Fly Your Thesis! 2009 – Final Report 1

Team AstEx – Simulating Asteroidal

Regoliths: Implications for Geology and

Sample Return

Fly your Thesis! 2009

Final Report

Student contact:

Ben Rozitis ([email protected]) Naomi Murdoch ([email protected])

Tomi de Lophem ([email protected])

Endorsing professor contact:

Dr. Simon Green ([email protected]) Dr. Patrick Michel ([email protected])

1. Executive Summary The AstEx experiment investigates the dynamics of regolith on asteroid surfaces. Despite their very low surface gravities, asteroids exhibit a number of different geological processes involving granular matter. Understanding the mechanical response of this granular material subject to external forces in microgravity conditions is vital to the design of a successful asteroid sub-surface sampling mechanism, and in the interpretation of the fascinating geology on an asteroid. The AstEx experiment used a microgravity modified Taylor-Couette shear cell to investigate granular flow caused by shear forces under the conditions of parabolic flight microgravity. The aim of the experiment is to characterise the response of granular material to rotational shear forces in a microgravity environment. A particular emphasis has been put on investigating the timescales to reach steady state flow and the memory effects of sheared glass beads in microgravity. During the period of microgravity in a single parabola a granular flow was initiated by applying rotational shear forces to the granular material. High speed cameras imaged the flow of the top and bottom layers of glass beads. Once a steady state flow was achieved particles continued to be imaged for the duration of microgravity. After the flight the individual particles on the surface layers of the granular material were

Education


tracked using a particle tracking algorithm so that their displacements and velocities could be determined. By calculating the particle velocities, it is then possible to determine the timescales involved in initiating a steady state flow in a granular material. These timescales can then be compared to Earth based results. Another investigation considered the effect of reversing the direction of shear on the steady state flow already started in microgravity. These three investigations (time to start a steady state flow, monitoring the flow, and effect of shear reversal on the flow) were repeated with granular materials of different particle, and with different shear rates to determine the effect of these variables. In some parabolas experiments were also conducted in the 1.8 g regimes giving data for three gravity conditions. The AstEx team were presented with several challenges related to both hardware and software during the parabolic flight campaign. However, despite this and the fact that neither the experiment nor any of the experimenters had ever flown before the team were still successful in taking useful data in 96% of the parabolas. Due to the complexity of the particle tracking and the extremely large volume of data collected there was not enough time to perform a full analysis of the outcomes at the time of preparation of this report. Analysing the images for just one experimental run is a very lengthy and computationally demanding process. As a result we are still very much involved in the data analysis phase. All the results presented are therefore preliminary results.

Figure 1.1: The AstEx team working in microgravity!

Education


2. Student Team Description

2.1 Team Photo

Figure 2.1: The AstEx team photo. (Top) Dr. Simon Green and Dr. Patrick Michel

(Bottom) Ben Rozitis, Naomi Murdoch, and Tomi de Lophem.

2.2 Core Team Members

2.2.1 Ben Rozitis The AstEx team leader and a postgraduate research student affiliated with The Open University’s Planetary and Space Sciences Research Institute (PSSRI) in the UK. His PhD thesis focuses on the characterisation of Near-Earth asteroids for a sample return mission. He graduated with a first class degree in MPhys Physics with Space Science and Technology from the University of Leicester in summer 2007 before beginning his PhD at The Open University in autumn 2007. As team leader he had many responsibilities including keeping the AstEx project on track, developing initial experiment design ideas, obtaining building materials and equipment for the experiment, sourcing additional funding for the project, developing the experiment data acquisition software, testing the experiment hardware, keeping the experiment ESDP up to date, performing the microgravity experiments during the actual parabolic flights, and finally helping in the initial data extraction phase just after the campaign.

2.2.2 Naomi Murdoch The granular material expert of the team and a postgraduate research student affiliated with The Open University’s PSSRI in the UK, and The Côte d’Azur

Education


Observatory (University of Nice Sophie-Antipolis) in Nice, France. Her thesis focuses on modelling the behaviour of granular material on the surfaces of planetary bodies (including asteroids). She graduated with a first class MPhys Astrophysics degree from The University of Edinburgh in summer 2007 and then went to work as a Young Graduate Trainee with the Advanced Concepts Team at ESA for a year. Naomi began her PhD at the start of 2009 and began immediately working on the AstEx project since the parabolic flight experiment will provide experimental validation of the numerical models she will later develop in her PhD. As the granular materials expert her many responsibilities included ensuring the design of the experiment produced scientifically viable results, discussing and presenting the experiment design to over seas granular material experts, obtaining suitable beads to act as the granular material in the experiment, designing and producing the AstEx website, helping Ben with some of his responsibilities, adding the finishing touches to the experiment hardware, transporting the experiment to the south of France, and performing the microgravity experiments. Her main role after the parabolic flight campaign is to analyse the ~1.5 million images taken during the three flights.

2.2.3 Tomi de Lophem The team’s expert mechanical engineer and is now currently studying for a second degree in Physics at The Open University. He originally worked for the Education Department at ESA and was responsible for setting up the ‘Fly Your Thesis!’ program. He also played a key role in the 2009 program organisation and team selection process. After leaving ESA he came and worked with the AstEx team providing invaluable expertise on a number of engineering issues with the experiment hardware. His responsibilities included producing detailed engineering and CAD designs for the experiment hardware, running simulations to ensure the experiment hardware could withstand the necessary 9g crash loads without damage, adding finishing touches to the experiment hardware, and helping Ben and Naomi to perform the microgravity experiments during the parabolic flight campaign.

2.2.4 Dr. Simon Green The main endorsing professor behind the AstEx project and a Senior Lecturer at PSSRI in The Open University. His research focuses on physical studies of planetary surfaces and small solar system bodies (asteroids, comets, interplanetary dust and space debris) through analysis of spacecraft data, ground and space-based observations and computer modelling. Simon is Ben’s main PhD supervisor and one of Naomi’s two supervisors. His responsibilities included ensuring Ben was doing his job of leading the AstEx project properly, sourcing additional funding for the project, ensuring appropriate insurance covering the experiments was in place, and assisting the rest of the team during the parabolic flight campaign. He is now also providing support and advice to Naomi as she continues with the data analysis phase.

2.2.5 Dr. Patrick Michel The second endorsing professor behind the AstEx project and is the head of the Planetology Group of the Cassiopee Laboratory of the Côte d’Azur Observatory. His research is devoted to the understanding of the collisional processes between small bodies and to their dynamical evolution in our Solar System and other planetary systems. Patrick is Naomi’s other main PhD supervisor. He came up with the original concept of the AstEx experiment following a scientific discussion with Wolfgang

Education


Losert. He then proposed the project to the key AstEx team members back in summer 2008 when the team first all met together at the Asteroids, Comets, and Meteors conference in Baltimore, USA. Patrick has provided valuable contacts to other granular material experts during the development stage of the AstEx project, and is currently assisting Naomi in analysing the parabolic flight data.

2.3 Additional People Involved In The Project In addition to the main five team members the AstEx project received valuable support and contributions from the following people.

2.3.1 Kevin Dewar Kevin is an engineer working in the workshop that constructed the AstEx experiment hardware at The Open University. He helped to design and build the experiment whilst putting up with the constant design changes requested by Novespace and ESA (which usually involved redoing a lot of work he had already done)!

2.3.2 Damian Flack Damian also works in the workshop at The Open University. He too worked hard along side with Kevin throughout the design and construction phase of the AstEx experiment.

2.3.3 Chris Hall Chris is the manager of the workshop at The Open University. During the design and construction phase of the AstEx experiment he remained incredibly patient when the project took up almost all of the time of his workshop staff. At the beginning of the project he helped to find the appropriate indestructible materials to construct the experiment out of, as well finding a suitable powerful motor for the experiment that rotated at an incredibly slow speed.

2.3.4 Nick Aldermann Nick is PSSRI’s electrician and installed most of the wiring for the experiment. He was able to install interlocks that automatically stopped the motor when the experiment’s double containment was opened in response to a last minute request from the ESA safety engineers.

2.3.5 Wolfgang Losert Wolfgang is one of Naomi’s and Patrick’s contacts who is an expert on non-linear granular dynamics. He is based at the Losert Lab at the University of Maryland, USA, and provided valuable technical and scientific advice regarding the experiment. The AstEx hardware that was designed and built is a microgravity version of one of his ground based granular material experiments. He is now collaborating strongly with Naomi as she analyses the data – providing routines to help in the particle tracking and subsequent analysis as well as important advice.

Education


3. Project Description

3.1 Scientific Objectives

3.1.1 Motivation Asteroids are leftover building blocks from the formation of the Solar System 4.5 billion years ago and offer a wealth of information regarding the early Solar System and the mechanisms responsible for its formation and evolution. They range in size from small (~10 m) boulders to bodies 1000 km across with consequently a larger range in mass, from a few thousand tonnes to 1021 kg. However, even the most massive of asteroids, Ceres, has a mass which is only a fraction of a percent of the mass of the Earth. As a result, asteroid surface gravities are many orders of magnitude smaller than that of the Earth.

Figure 3.1: The size range of asteroids (JAXA/NASA). (Left) Itokawa (target of the

Hayabusa sample return mission) vs. The International Space Station (Right) Ceres vs. our Moon.

The geology and geophysics of asteroids is a fascinating and constantly surprising field. Regolith, granular material covering the uppermost layer of solid planetary bodies, plays an important role in the surface geology of asteroids. Regolith dynamics on asteroids has been both observed by in-situ spacecraft and modelled. Space agencies are planning sample return missions, other than the current Hayabusa mission, to near Earth asteroids to bring back to Earth a pristine sample of an asteroid’s surface. These missions aim to investigate early Solar System processes by applying the vast array of laboratory analytical tools to these samples, to link meteorite classes to asteroid classes, and study components (such as interstellar grains, organics and volatiles) that do not survive the atmospheric entry or terrestrial contamination of meteorites. To ensure that the returned samples are unaltered there are requirements that the sample shall contain grains of sufficient size so that their interior will have been protected from cosmic radiation and weathering effects. Ideally, the sample should be retrieved from a sufficient depth to ensure large surface temperature variations did not affect it. However, physical turnover of the surface through impacts (called “gardening”) makes this impractical. The expectation is that by collecting a large number of samples from as deep as possible, some will have been buried or located in cooler regions of the surface when these heating episodes occurred. An additional requirement is that sufficient sample

Education


mass must be returned to satisfy some of the ground based laboratory analyses of rare components.

Figure 3.2: The surface of asteroid 433 Eros taken by the NEAR Shoemaker

spacecraft (NASA). These two images show that the surface is covered with a substantial regolith and numerous boulders. They also display evidence of regolith movement where the crater floors are flat where finer regolith has moved into them

and filled them out. To meet the sample requirements a suitable sampling mechanism must be chosen and developed. However, the surface properties of an asteroid can vary and are not easily determined from ground based astronomical observations. Hayabusa used a generic impact sampling device to ensure sample collection from any type of target [1]. However, this mechanism returns a limited mass consisting of only small grains from the surface. To ensure larger grains and greater mass are collected a sticky pad has also been developed and tested on parabolic flights [2]. However, the sticky pad does not allow retrieval from depth or uncontaminated retrieval. A multitude of different sampling mechanisms for retrieving samples at depth can be thought of including simple scoops and rotating corers. Designing a suitable sub-surface sampling mechanism requires knowledge of how the regolith granular matter responds to various different external forces/pressures in microgravity conditions.

3.1.2 Granular Materials The granular materials which form regolith consist of a very large number of discrete particles that interact with each other only through dissipative contact forces [3]. The particles are massive enough (with sizes from 100 to 3000 µm [4]) so that their potential energy is orders of magnitude larger than their thermal energy [5]. Without an external drive their kinetic energy is rapidly lost and the system is referred to as being non-thermal. Despite this deceptively simple description, granular matter exhibits many complex behaviours that are difficult to predict. Although individual particles are solid, granular materials exhibit both solid-like and liquid-like behaviour. In a solid-like state, such as a heap or pile, the material is said to jam [6]. When the material jams, the individual particles are in a stable mechanical equilibrium with their local neighbours. In a liquid-like state the material is said to "flow". Granular flows can be divided into two types: rapid dilute flows, and dense flows [7]. Rapid dilute flows can be described with some success by kinetic theories for hard spheres that interact through uncorrelated and instantaneous binary collisions [8]. Dense flows are dominated by many-body interactions and occur when

Education


particles have long-lived contacts with many neighbours. In turn, dense flows can be divided into two regimes: fast dense flows dominated by the inertia of particles, and slow dense flows where inertia is negligible. Inertia dominates when the kinetic energy of a particle is much greater than the energy needed to move a particle past its neighbour. In the low gravity regime of an asteroid the potential energy of an individual grain is lower and thus the energy required to move past its neighbours is reduced. Assuming an individual grain has the same kinetic energy as on Earth, the flow which on Earth would have been in the slow dense regime, may move to the fast dense regime on the surface of the asteroid due to the inertia no longer being negligible. This implies that bodies with low surface gravity may be very sensitive to processes that appear irrelevant in the case of larger planetary bodies. For example, seismic vibrations, induced by small impacts can occur throughout a small body. These vibrations can be at the origin of motion of its granular surface. The regolith motion resulting from such seismic vibrations has been proposed to explain the lack of very small craters both on Eros [9] and Itokawa [10]. Due to their dissipative nature, granular materials will only flow under the application of a continuous input of energy. This energy can be provided by applying shear stresses to the material. For instance impact phenomena [11] [12], tidal forces from planetary encounters [13], and YORP spin up [14] could apply shear forces to the surfaces of asteroids.

3.1.3 Taylor-Couette Shear Cell In order to investigate the effect of shear stresses on granular materials experimentally, a Taylor-Couette shear cell can be used [15]. The Taylor-Couette geometry is shown in Figure 3.3.

Figure 3.3: The Taylor-Couette Geometry [16]. (a = Inner Cylinder Radius, b = Outer Cylinder Radius, w = Width of Shear Region, r = Radial Distance,

θ = Angular Distance, ω = Inner Cylinder Rotation Rate)

Education


There are two concentric cylinders. The outer cylinder is fixed and its inside surface is rough with a layer of glued on particles, the outer surface of the inner cylinder is also rough but it is free to rotate, and the floor between the two cylinders is smooth and fixed in place. The gap between the two cylinders is filled with granular material on which the rotating inner cylinder applies shear stresses. Large velocity gradients are then produced near the inner cylinder as the energy input in to the granular system by the rotating inner cylinder is dissipated by friction in a narrow band. This localised region of shearing is known as a shear band. The size of the gap between the two cylinders is made to be ~50d where d is the average diameter of particles filling the gap.

3.1.4 Previous Experimental Results It has been shown that the flow of granular matter is strongly influenced by the network of direct contacts with neighbouring particles. This contact network, in turn, is shaped by how the material evolved with time. When uniform shear or compression is applied a stronger contact network in the direction of forcing develops. When the shear direction is reversed, or the direction of compression is changed, the material rearranges until it forms a new contact network that can best support the new direction of compression or shear [17]. This is illustrated for shear in Figure 3.4 where it can be seen that force chains aligned to cause jamming in one direction are not suited to jam under the reverse driving. Therefore, time is required to re-form a force chain network subsequent to reversal.

Figure 3.4: The breaking and re-forming of a contact network through shear reversal in granular matter in Taylor-Couette geometry is illustrated schematically [17]. The

arrows indicate the relative movement of the cylinder walls and the red lines through the particles in the upper diagrams indicate the stress transmission. The positive and negative signs show the principal direction of the stress transmission. When shear is started opposite to the prior shear direction, transiently the material compacts and is

easy to shear. In Figure 3.5 below the six lines show the angular velocity measured in six concentric rings during the experiment described above. When sheared in the initial direction the system reaches steady state during which three of the six rings do not experience any appreciable flow. If the driving is discontinued and then reapplied the system reaches the same steady state immediately. However if the driving is stopped and then applied in the reverse direction, a transient sets in during which flow is evident both in the regions that were flowing and in the previously jammed regions. Average

Education


flowing velocities in all regions are faster initially and drop off with roughly the same time scale in all regions. The implication is that regions that normally do not move under steady shear, move significantly during reversal of the shear direction. Studying the reversal of shear in a granular material in microgravity has the potential to shed light on different modes of deformation that are evident when granular material is sheared in different directions. Additionally, the flow fields shown in Figure 3.5 during shear reversal are accompanied by compaction due to gravity. It is not clear how the force chains would break and reform in the absence of a preferred guiding direction such as gravity.

Figure 3.5: Angular velocity measured in six concentric rings during granular flow in a Taylor-Couette cell [17]. Driving stops restarts in the same direction, and the steady

state recovers instantaneously. Then the shear is reversed, and a transient flow regime precedes the establishment of steady state shear in the opposite direction. Each concentric ring is represented by a line on the plots of angular velocity above.

3.1.5 AstEx Experimental Objectives The AstEx experiment will investigate how a steady state (constant) flow is achieved in a granular material in microgravity conditions. A flow will be started by applying rotational shear forces to the granular material contained within a microgravity modified Taylor-Couette shear cell. Individual particles of the granular material will be tracked so that their displacements and velocities can be determined. By monitoring the particle velocities, the time needed for a steady state flow to start can be determined. Once a steady state flow is achieved particles can be tracked for the length of microgravity time available during a single parabola. This will tell us how a steady state flow in microgravity differs from a steady state flow on Earth. Another investigation will determine what effect reversing the direction of shear has on the steady state flow already started in microgravity. Again this will be compared with ground based results. These three investigations (time to start a steady state flow, monitoring the flow, and effect of shear reversal on the flow) will be repeated with

Education


granular materials of different particle sizes, and with different shear rates to determine the effect of these variables. The experimental results will also be used to validate an N-body model based on the parallel code pkdgrav [18] which will be developed to simulate the behaviour and response of granular materials to various external pressures and forces. This pkdgrav code has previously been used to successfully compute the gravitational re-accumulation phase during catastrophic disruption of pre-shattered parent bodies [19], perform simulations of the collisional and gravitational dynamics of aggregates, with or without cohesion [20] in addition to many other complex N-body numerical simulations. Once this code has been validated for a number of different gravitational conditions it can be applied to any solid planetary body to determine the dynamical response of any granular material found on its surface.

3.2 Experimental Set-Up

3.2.1 The Experimental Rack Figures 3.6 and 3.7 show how a Taylor-Couette shear cell is mounted inside the A300 Zero-G aircraft for testing in microgravity conditions. The experiment rack consists of two parts: a test compartment, and a laptop work station. The test compartment is where the intended experiments will take place and it will contain one removable shear cell that is to be tested. Built into the test compartment is a motor powered by an inverter, an enclosed toothed belt pulley system, four illumination lights, and two high speed cameras, which are required to conduct the experimental tests. Situated next to the test compartment is the laptop work station where two laptops are mounted to allow two people to control the various components of the experimental hardware and to perform the experimental tests. The entire experiment rack is 1006 by 1250 by 750 mm in size, and has a total mass of ~170 kg.

3.2.2 The Microgravity-Modified Shear Cell The shear flow and shear reversal is studied using Taylor Couette geometry as described above. Figure 3.8 shows a single shear cell. They are mounted on a polycarbonate base 650 mm long and 450 mm wide, the outer cylinder has an outer diameter of 400 mm and a height of 200 mm, and the whole unit has a total mass of ~27.5 kg when filled with beads. The polycarbonate base features two carrying/shaking handles used to carry and shake the shear cell. The outer cylinder is made from a cast Acrylic tube with a 400 mm outside diameter and a 5 mm wall thickness, and the inner cylinder is made from the same material but with a 200 mm outside diameter and a 3 mm wall thickness. This gives a gap size of 95 mm between the inner and outer cylinders, or ~32 bead diameters if 3 mm size beads are used. The inner cylinder has a steel shaft running through the centre and is attached to the outer cylinder by two bearings contained inside a housing unit. This steel shaft attaches to the driving shaft and transmits the torque produced by the motor to the inner cylinder. Figure 3.9 shows the internal workings of a shear cell. In a normal Taylor-Couette shear cell the granular material does not have a top constraint [15]. This allows the granular material to change its depth and density during shear. However, without a top constraint in microgravity the granular material will just float away.

Education


Figure 3.6: The AstEx experimental rack (front).

Figure 3.7: The AstEx experimental rack (back).

Education


A movable and transparent pressure disk is used to keep the granular material contained. It is loaded by 3 weak springs to apply a very small force e.g. by sprung loaded roller balls (see Figure 3.10). The transparent pressure disk can also be fixed in place so that experiments can be performed at constant volume. To maintain a granular seal the spaces between moving parts are made to be very small at 0.5 mm to prevent a single glass bead from escaping. The glass bead granular material fills the shear cells to a height of 100 mm. Since the shear cells are self-contained units each shear cell contains glass beads of a certain size that cannot be easily exchanged. In total three shear cells were built: one with 3 mm glass beads (~0.37 million beads), one with 4 mm glass beads (~0.16 million glass beads), and one constant volume shear cell containing 3 mm glass beads (~0.32 million beads). The shear cells were designed to be easily exchanged between flights when the plane is on the ground to allow testing of different shear cells during the parabolic flight campaign.

Figure 3.8: A photo of the microgravity modified Taylor-Couettte shear cell.

3.2.3 Shear Cell Mounting Figure 3.11 indicates how a shear cell is mounted inside the experiment rack. The shear cell itself is attached to the polycarbonate base via two aluminium securing bars. These can be easily undone to replace the current shear cell with a different one without the need for making a separate mounting for each shear cell. The polycarbonate base, with a test shear cell attached, is secured to the rack by four guide rods running through four holes in the polycarbonate base. Two sliding locking bars connect the two guide rods on each side of the shear cell. The positions in which these are set determine whether the polycarbonate base is free to move up and down the guide rods, or whether it is fixed in place. During a parabola the polycarbonate base is locked in place to allow experiments to be performed. It is unlocked during level flight to allow the experimenters to shake the shear cell between parabolas to reset the granular material contained inside the shear cell for the next experimental test. This is required to prevent the shear memory effect of granular materials affecting the outcome of consecutive independent experimental tests. During a shear experiment the grains rearrange themselves to form a contact network that suits the last direction of shear imposed upon it. This contact network is not broken when the shear force is removed, and shaking is one of the easiest options available to break this contact network and create a random arrangement of grains in the granular material.

Education


Figure 3.9: Internal workings of a microgravity shear cell. (1) Housing unit with

bearing (2) Movable pressure disk (3) Pressure springs (4) Steel shaft (5) Fixed outer cylinder (6) Fixed bottom plate (7) Rotating inner cylinder

(8) Camera viewport.

Figure 3.10: The ball transfer unit.

3.2.4 Counteracting Imperfect Microgravity The microgravity environment on a parabolic flight is not perfect; there are small fluctuations about 0 g of magnitude ± 0.05 g. This means at times the experiment will experience small amounts of negative g. It is uncertain whether this negative g will reset the grain contact network. To ensure that all gravity fluctuations are positive, very low positive gravity must be simulated on the experiment during the constant pressure experiments. This is achieved by using the sprung loaded movable pressure disk to provide a small force in a preferred direction to simulate the effect of very low positive gravity (see Figure 3.10). This small force is equal to the maximum amplitude of the gravity fluctuations i.e. 0.05 g. When applied to the grains in the experiment they experience an effective gravity fluctuating between 0 and 0.1 g.

Education


Figure 3.11: Shear cell mounting. (1) Shear cell securing bars (2) Guide rods

(3) Sliding locking bars (4) Silent blocks (5) Support structure.

Vibrations may also reset the contact networks set up in the granular material. To get statistically valid results the experiment ideally needs to be free of aircraft vibrations and free of gravity jitters during the microgravity phases. To minimise these effects the shear cell is isolated from vibrations. This is done by mounting silent blocks between the two strut profiles on which the shear cell is resting, and the rest of the support structure frame (see Figure 3.11).

3.2.5 The Mechanical System The experiment requires rotating the inner cylinder of the shear cell at a very slow speed i.e. <1 RPM. To provide the driving force a Watt Drive HU50C 64K4 inline helical geared motor is used. It is a three phase motor and is controlled by a Moeller DF51-322-025 three phase inverter. The motor operates at 120 W and runs at 1330 RPM. However, the motor has an in built gear system with a ratio of 123.14:1, giving an output speed of 10.8 RPM. To achieve the desired rotation speed for the experiment a toothed belt pulley system is used to achieve the final 10:1 reduction in rotation rate. The driving shaft of the pulley system connects to the shaft of the inner cylinder via a pin and groove slotting principle. The inner cylinder shaft then rotates the inner cylinder at the desired very slow speed of <1 RPM. A specific rotation speed can be set by the inverter, and a load trip is programmed to protect the shear cell if the torque transmitted to the inner cylinder becomes too great.

3.2.6 Data Collection The two high speed cameras (Matrix Vision Blue Fox 120aG) image the top and bottom glass bead layers of the shear cells. The imaging speed was designed to be ~100 frames/sec so that the particles do not move more than 1/10 d between consecutive frames. The cameras are mounted to the experiment rack in the test compartment and image the glass beads through camera viewports built into the shear cells. See Figures 3.6, 3.7 and 3.9 above. Four GU10 energy saving reflector lamps are mounted next to the cameras (two for each camera) to illuminate the glass

Education


beads. These lamps were chosen for their low power consumption and low operation temperature. The cameras also measure the height of the pressure disk as well as tracking the glass bead motions. Suitable beads are also glued in place onto the top surface of the pressure disk so that they can be imaged by the cameras. The separation distance between these bead markers as they appear in the camera images will allow a calculation of the height of the pressure disk. In order to monitor the torque exerted on the inner cylinder the current consumed by the motor is measured by the inverter powering the motor. The two laptops (one per camera) mounted to the experiment rack are used to collect and store data from the experiment. It was anticipated that ~150 GB of data per flight and ~450 GB of data in total will be generated.

3.2.7 Health And Safety Requirements Since the glass beads used in this experiment are small (3-4 mm) they are given the same precautions as a fluid onboard a parabolic flight and therefore require double containment. The shear cells themselves are self-contained units and act as a single layer of containment. The double containment is provided by covering the experiment rack with 5 mm thick polycarbonate panels. Half of the top panel is able to slide open to give access to the shear cell in order to shake it and exchange it. However, shear cells are only exchanged between flights when the plane is on the ground. The complete electrical system of the experiment is protected by a single, easily accessible emergency stop button. In the event of an emergency, actuating this button will cut off all 220 V AC power to the equipment. It is also protected by a differential circuit-breaker rated at 30 mA, and a fast fuse rated at 6 A. Individual components of the experimental hardware are also protected by their own fuse but of lower values specific for their needs.

3.3 Experimental Procedure Upon entry into the A300 Zero-G aircraft the experiment rack is assembled with the appropriate shear cell to be tested. During the experimental phases (described below) the high speed cameras imaged the granular material by imaging reflections off the coloured particles. This allows for subsequent tracking of particle motions. The motor inverter measured the amount of current being consumed by the motor in order to monitor the amount of torque exerted on the inner cylinder. Two experimenters were needed throughout the experiment: one to supervise the Taylor-Couette shear cell, and one to control the two laptops and monitor the data acquisition. Slightly different experiments are performed in each of the three parabolic flights in order to investigate three different effects. Each experiment type is performed with the same granular material for 5 experiments i.e. 5 parabolas. Described in Tables 3.1 to 3.3 below are the experiments planned for each of the three parabolic flights. There are two different modes of experiment:

• Mode 1 - The motor rotates the inner cylinder in the same direction for the full 20 seconds of microgravity.

Education


• Mode 2 - The motor rotates the inner cylinder in one direction for the first 10 seconds of microgravity and then stops and reverses the direction for the remaining 10 seconds.

There are three phases during a single parabola of a parabolic flight: a ~20 second 1.8 g injection phase, a ~20 second microgravity phase, and a ~20 second 1.8 g recovery phase. There is a 2 minute 1 g rest between parabolas and so parabolas are repeated every 3 minutes. After 5 parabolas it is standard for the A300 Zero-G to take a longer 1 g rest of 4-8 minutes. The experimental procedures are timed with these different phases. Approximately 5 seconds before the start of the parabola the experimenters start the high speed cameras recordings. During the 1.8 g injection phase the experimenters do nothing except monitor the experimental equipment for any malfunctions. As soon as the microgravity phase starts the experimenters blink start the motor, and then halfway through the microgravity phase the direction of the motor rotation can be reversed if needed. During the second 1.8 g recovery phase the motor is normally left running in the direction it was at the end of the microgravity phase. In the last flight the AstEx team also made use of this second 1.8 g period to perform shear reversal experiments. Finally, when the 1 g rest phase starts the motor and high speed cameras are stopped. During the 2 minute rest period the experimenters must ensure all data from the high speed cameras are saved to the laptop hard drives, and prepare the shear cell for the next experiment. The shear cells are prepared for the next experiment by inspecting them for any damage or leaks, and shaking them by hand to reset the glass beads back to a reasonably consistent initial arrangement. This procedure is repeated for each parabola; however, during the longer 4-8 minute rests the motor rotation speed was also adjusted. Parabola Number

1-4 5 6-9 10 11-14

15 16-19

20 21-24

25 26-29

30

Bead / mm

4 4 4 4 4 4 4 4 4 4 4 4

Frequency / mHz

4 4 4 4 8 8 8 8 16 16 16 16

Mode 1 1 2 2 1 1 2 2 1 1 2 2

Shake? Y N Y N

Adjust Motor Speed

Y N Y N

Adjust Motor Speed

Y N Y N

Table 3.1: Microgravity Flight 1 – Constant Pressure 4 mm Beads Parabola Number

1-4 5 6-9 10 11-14

15 16-19

20 21-24

25 26-29

30

Bead / mm

3 3 3 3 3 3 3 3 3 3 3 3

Frequency / mHz

4 4 4 4 8 8 8 8 16 16 16 16

Mode 1 1 2 2 1 1 2 2 1 1 2 2

Shake? Y N Y N

Adjust Motor Speed

Y N Y N

Adjust Motor Speed

Y N Y N

Table 3.2: Microgravity Flight 2 – Constant Pressure 3 mm Beads

Parabola Number

1-4 5 6-9 10 11-14

15 16-19

20 21-24

25 26-29

30

Bead / mm

3 3 3 3 3 3 3 3 3 3 3 3

Frequency / mHz

4 4 4 4 8 8 8 8 16 16 16 16

Mode 1 1 2 2 1 1 2 2 1 1 2 2

Shake? Y N Y N

Adjust Motor Speed

Y N Y N

Adjust Motor Speed

Y N Y N

Table 3.3: Microgravity Flight 3 – Constant Volume 3 mm Beads

Education


4. Parabolic Flight Campaign

4.1 Before The Flights During the preparation week the AstEx team had a busy time implementing changes requested by the parabolic flight safety team whilst simultaneously resolving issues with the experimental hardware and practising the in-flight operating procedures. The evolution of the experiment during this time is described in the relevant subsections below.

4.1.1 Changes Requested By The Parabolic Flight Safety Team In the few weeks leading up to the flight campaign Novespace, ESA, and CEV conducted pre-campaign safety reviews. A potential health and safety concern was raised regarding the required shaking of the shear cells between parabolas by the experimenters. Although the weight to be shaken (~27.5 kg) was within the limits of safe lifting by a single person (30 kg) it was thought the bending required to pick up the shear cell from within the experiment rack would put excessive stress and pressure on the lifting person’s back. To make the lifting and shaking easier it was requested that the shaking handle bars should be extended. This option was not preferred by the AstEx team as such handles, being less strong, could wobble around and potentially crack the supporting polycarbonate base when shaking the shear cells. Thus a set of extended handles were made before the campaign but not installed. They were to act as a replacement for the non-extended handle bars in case they were needed. Upon arrival at the Novespace premises and after several demonstrations of the shear cell shaking process it was decided that the extended handle bars should be installed. The extended handle bars were easily installed and luckily they did not wobble around or crack the polycarbonate base as we feared they might do (see Figure 4.1).

Figure 4.1: Installation of extended shaking handle bars. (Left) Original handle bar (Right) Extended handle bar.

Education


In addition to extending the shaking handle bars it was requested that the experimenters who would shake the shear cell wear protective gloves (see Figure 4.2). This was to protect the experimenter’s hands from being accidently cut from sharp edges, and to protect them from being burnt from the light bulbs required for high speed imaging. Although the light bulbs were safe to touch (even when having remained turned on for a long a time) in open spaces they did become a little too hot to touch when inside the enclosed space of the experiment rack. No other suitable bulbs that provide the continuous brightness required (most bulbs appear to flicker in the images taken at high speed) but operated at a lower temperature could be found.

Figure 4.2: Ben wearing protective gloves.

The ESA safety engineer, after conducting a safety review of the experimental hardware, requested additional redundancy to be integrated into the AstEx electrical wiring. The first addition being to add an extra grounding wire to the motor even though tests conducted both at The Open University and at the Novespace premises showed that it was grounded sufficiently already. The additional grounding wire was easily added (see Figure 4.3). The second addition was to add protective plastic covers to the light switches since they operated at 240 V (see Figure 4.4). The technical data sheets of the light switches showed that they were rated at 240 V for several amps. However, the ESA engineer still felt that the protective plastic covers should be added as they were directly switching 240 V.

Figure 4.3: Installation of additional motor grounding wire.

(Left) Original motor grounding (Right) Two motor grounding wires.

Education


Figure 4.4: Addition of protective plastic covers to the switches.

(Left) Unprotected switches (Right) Protected switches. During the final safety review onboard the aircraft a CEV safety engineer felt that there were an insufficient number of attachment points connecting the shear cell mounting to the rest of the experiment rack i.e. we had 5 attachment points instead of 6. Although this is more than sufficient for the normal operation of the plane during parabolic flight, it was felt the rack didn’t quite have the sufficient safety margin during a crash landing i.e. when the peak lateral forces are ~9 g. A simple solution to this was to add a removable strap around the shear cell mounting and the experiment rack to increase the number of attachment points to 6. This strap would be added before takeoff and landing but could be removed once the aircraft had reached level flight to allow the experiments to be conducted (see Figure 4.5).

Figure 4.5: Removable attachment strap.

(Left) Strap removed (Right) Strap in place.

4.1.2 Problematic Data Connections From Electromagnetic Interference During the final month before the flight campaign several last minute design changes to the experiment were requested following the safety reviews conducted by Novespace, ESA, and CEV. As a result the experimental hardware was fully finished just a few days before the campaign started, leaving insufficient time for the AstEx team to properly ‘test and debug’ the experiment in its fully integrated form. This ‘testing and debugging’ phase therefore occurred during the preparation week of the parabolic flight campaign.

Education


During testing it was quickly discovered that when the motor is running it would cause an image transfer error in the data connection between the high speed cameras and the laptops. This was occurring for roughly a third of all images taken and this would have been an unacceptable loss of data for the intended parabolic flight experiments. By showing that the problem still occurred when the laptops and high speed cameras were run from the laptop batteries only i.e. by separating the two power supplies, we were able to identify the source of the problem to be electromagnetic interference from the motor and its inverter. Similarly, the motor inverter data connection to one of the laptops was unstable and the communication repeatedly timed out. These problems did not manifest when the components of the experiment were tested individually but only when tested as a whole unit. Unfortunately the motor inverter was placed right next to the laptops and the USB data cables and could not be moved. We managed to mostly solve the problem by making three changes. Covering the USB data cables with aluminium foil to act as a shield (see Figure 4.6), moving the USB data cables as far away from the motor and inverter as possible, and by experimenting with which laptop/high speed camera/motor inverter connection combination worked best. Once we had made these changes the problems occurred far less frequently but they did not disappear completely. If the experiment is to fly again we would place the motor inverter on the opposite side of the experiment rack, install a special electromagnetic interference dampening device for the motor inverter, and coat the USB data cables in proper shielding tape.

Figure 4.6: Naomi wrapping the USB data cables in aluminium foil.

4.1.3 Slow Laptop Hard Drive After Defrag Our initial intention was for our high speed cameras to record ~10,000 images into the laptops random access memory during each parabola, which would then be written to the laptop hard drives during the 1 g phases between parabolas. This requires ~6 GB of data to be written to the laptop hard drives in less than two minutes! Before the campaign special data acquisition software was written that would allow this amount to be saved that quickly. This was tested several times before the campaign and did work very well as long as the laptop hard drives were running fast and efficiently. To ensure they were fast and efficient it became routine

Education


to defrag the laptop hard drives on a regular basis between testing. However, on one occasion the defrag process had the opposite effect on one of the laptop hard drives causing the speed at which the data was saved to reduce drastically. At this slower speed there would not be enough time to save all the data during the two minute interval between parabolas. The problem was resolved by partitioning the slow laptop hard drive into two separate disk drives and performing a complete format on one of them. When the data acquisition software was run from the freshly formatted partition it was running at full speed again. It still remains unknown as to why the defrag process had such a counterintuitive effect and even has computer scientists baffled! If that solution had not worked we had an alternative solution of running both high speed cameras from one laptop but at a slightly slower speed.

4.2 During The Flights By the end of the flight campaign preparation week the experiment and its operating procedures were well tested and rehearsed on the ground. However, since we couldn’t test the experiment in actual flight conditions before the parabolic flights started we encountered some minor problems as described below. Despite these problems we collected most of the data we wanted and the experiments were a great success!

4.2.1 Slow Laptop Hard Drives Due To Aircraft Vibrations The first problem we immediately encountered during the first parabolic flight was that both laptops were saving the high speed camera images to their hard drives much slower than intended. After just a couple of high speed imaging runs they had become out of sync with the timing of the parabolas. To reduce the time spent saving the data and to get back into sync with the parabola timing we reduced the amount of data the laptops had to save to their hard drives by reducing the camera frame rate from ~100 fps to ~60 fps. In addition, we shortened the length of time the cameras were taking images thus we reduced the number of images taken per run from 10,000 to ~2,500. We didn’t record any data for one parabola whilst we adjusted the camera settings, and it took us another 5 parabolas to finally decide on when to stop the camera recording during the parabola sequence (see Figure 4.7). Thereafter, we were routinely taking data during the remaining 20 parabolas of flight one. Despite these problems we collected most of the data we needed during the first flight. The reduction in frame rate and number of taken images has no consequence for our data analysis and scientific results. The beads must move less that one tenth of a particle diameter between frames for the tracking and the original experimental set-up (frame rate, images taken etc) was over designed for the size of bead we were testing. After landing of the first flight it was determined that aircraft vibrations were interfering with the hard drive write processes and thus making them much slower. Unfortunately there was no immediate fix for that problem so we decided to work around it for the remaining two flights. We kept the camera frame rate at ~60 fps and chose to record between 2,500 and 5,000 images for each parabola to ensure that the data could be saved in time. This worked very well for the remaining two flights and we collected most of the necessary data from the cameras we needed. If the experiment is to fly again a permanent solution to allow higher frame rates of ~100 fps could be to mount the laptops on top of foam pads/mats to absorb and

Education


dampen the aircraft vibrations. In the configuration in which the experiment flew the laptops were mounted directly onto the experiment base plate. An alternative solution is to replace the laptop hard drives with expensive solid state hard drives as they have no mechanical moving parts and thus would be completely immune to the aircraft vibrations.

Figure 4.7: Ben and Naomi resolving the laptop issues mid-flight.

4.2.2 Motor Inverter Connection Timeouts During the first flight, after we had solved the issues with the slow laptop image recordings, we noticed that the motor inverter data connection to one of the laptops had timed out multiple times. To reinitialise the connection would have taken too long so we decided to stop the motor inverter data recording. This meant we collected very little data from the motor for the first flight. The possible causes for this are again electromagnetic interference, or due to the laptop hard drive slow down. For the second and third flights the data connection worked almost perfectly with the connection timing out only once, meaning that we collected all the motor data from those flights. The fact that we collected very little data from the motor on the first flight doesn’t affect our science greatly since the motor data is secondary to the data from the high speed cameras. To prevent the problem from reoccurring in the future would be to minimise the electromagnetic interference and laptop aircraft vibration problems as already described above.

4.2.3 Stuck Shear Cell Between each parabola it was standard procedure for one of the experimenters to shake the shear cells to reset the granular material contained within for the start of each experiment. However, on the third flight it became increasingly difficult to shake the shear cell between parabolas 11 and 17. And finally, on parabola 18 it became permanently stuck and no matter how much force was applied it would not move (see Figure 4.8)! For parabolas 18 and 19 the motor was not started whilst attempts we made to free the shear cell. However, the high speed cameras were still started providing useful data on how the beads behaved in imperfect microgravity conditions without any shearing force applied. On parabola 20 the motor was started and the experiment still worked even though the shear cell was stuck in the vertical direction. For the remaining parabolas of the third flight experiments were performed without the shaking of the shear cell between experiments. This should not impact the

Education


science greatly as similar experiments of the same bead size were performed on the first flight with shaking. The shear cell also only became stuck on the last flight so it didn’t affect the experiments of the previous two flights. After the campaign had finished the experiment was transported back to the workshop at The Open University to free the stuck shear cell. It took a lot of heat (from a blow torch) and brute force to free it! Inspection of the shear cell drive shaft afterwards indicated that corrosion had bonded the two drive shafts together. The corrosion occurred because the drive shafts were made out of mild steel which was a cheaper alternative to stainless steal. It was a surprise that the mild steel had corroded so quickly! If the experiment is to fly again then the shear cell drive shafts should be remade out of stainless steal to prevent this problem from reoccurring.

Figure 4.8: Tomi trying to shift the stuck shear cell.

4.3 Between The Flights After the first flight everything happened quite smoothly in setting the experimental hardware up for the second flight. The shear cells were easily exchanged, the data downloaded from the laptops, and the high speed camera settings easily adjusted to suit the slower laptop hard drive write speeds during flight. However, we encountered one problem, as described below, when setting the experimental hardware up for the third flight after the second flight had landed.

4.3.1 Jammed Constant Volume Shear Cell To prepare the experimental hardware for the next parabolic flight required exchanging shear cells between flights as each shear cell performs a slightly different experiment. On the first flight we flew the 4 mm shear cell, and on the second flight

Education


we flew the 3 mm shear cell both with the constant pressure plate lid. On the afternoon of the second flight we were supposed to remove the constant pressure lid of the 3 mm shear cell and replace it with the 3 mm shear cell lid for constant volume experiments. However, when we came to use the constant volume lid it had become attached to another shear cell and we could not remove it. The metal of the two parts had bonded and even with lubricant, hot water, brute force, mallets, tools and the help of half the staff of Novespace we could not remove it at all (see Figure 4.9)! Unfortunately, we were forced to abandon the constant volume experiment and we then had to decide between re-flying a shear cell or making a new shear cell by mixing 3 mm and 4 mm beads. Finally, we chose to re-fly the 4 mm cell (the one from the first flight). We hoped to get some more data from the cameras during the different phases and also from the inverter if it behaved correctly. Our procedures were also changed slightly. The 0 g phases were kept identical to the first flight for reproducibility and increased data points but we used the 1.8 g phases this time too. In the first two flights we allowed the motor to continue to run in 1.8 g but did nothing special. On the third flight, in every parabola where shear reversal was not performed in 0 g, we performed the shear reversal in the second 1.8 g phase of the parabola. This will not affect the 0 g results in any way but will give us some new science. Using these new procedures means we have shear reversal data in three different gravity regimes (0 g, 1 g, and 1.8 g). This will be very important for determining how the dynamics vary with the level of gravity and also for more detailed comparisons with the simulations. The 1.8 g is not perfect but we will get the flight data with records of the gravity levels which we can input into the simulations just as we will do with the 0 g experiments. After transporting the experiment back to the workshop at The Open University it was found that the drive shaft, which was also made out of mild steel, was again to blame. This design fault can easily be rectified, if the experiment is to fly again, by remaking it out of stainless steel.

Figure 4.9: Simon and Novespace staff trying to free the jammed constant volume

shear cell.

Education


5. Scientific Results

5.1 Collected data 5.1.1 The Experiments Performed Three investigations were carried out during the parabolic flight campaign:

1. Time to start a steady state flow 2. Monitoring the flow 3. Effect of shear reversal on the flow.

These three experiments were repeated with granular materials of two different particle sizes (3 mm and 4 mm diameter), and with three different shear rates (4, 8 and 16 mHz) to determine the effect of these variables. In some parabolas experiments were also conducted in the 1.8 g regimes giving data for three gravity regimes (1 g, 0 g and 1.8 g). We additionally conducted a few experiments in which we did not shake the granular material prior to the parabola in order to determine if the shaking did or did not influence the results. A final test that was performed was to image the beads during two parabolas without the motor rotating. This should give us an indication of the influence of the aircraft vibrations and fluctuating gravity levels on the movements of the beads. On average we were taking images at a rate of ~60 fps on both cameras simultaneously and we were able to collect on average ~68 seconds of data per parabola. This giving 731,123 images (209 GB) of data in total from the three flights plus 796,984 images (228 GB) of data taken in experimental runs on the plane whilst on the ground.

5.1.2 Summary Of Data Collected A summary of the experiments performed and data collected during the parabolic flight campaign is given in Table 5.1. In some parabolas we did have a few issues with the data collection as outlined in Section 4.2. In Table 5.1 the parabolas during which we encountered problems have only been included if we got enough useful data from either or both cameras.

5.2 Data Processing And Analysis

5.2.1 Data Extraction And Conversion In order to speed up the data saving process on board the aircraft the camera software which we wrote did not save individual files but instead created an avi file from the individual images of each experimental run and saved this one output file per experimental run/parabola. Therefore, the first task of the data analysis was to extract the individual image files from the avi files so they could be processed individually.

Education


No. of parabolas

during which data collected

Particle size (mm)

Rotation rate (mHz)

0 g phase (Uniform

shear/ Shear Reversal)

Second 1.8 g phase

(Uniform Shear / Shear

reversal) 5 3 4 Uniform Uniform 4 3 4 Reversal Uniform 5 3 8 Uniform Uniform 5 3 8 Reversal Uniform 5 3 16 Uniform Uniform 5 3 16 Reversal Uniform 5 4 4 Uniform Uniform 10 4 4 Reversal Uniform 5 4 4 Uniform Reversal 6 4 8 Uniform Uniform 6 4 8 Reversal Uniform 6 4 8 Uniform Reversal 6 4 16 Uniform Uniform 8 4 16 Reversal Uniform 5 4 16 Uniform Reversal 2 4 0 - - 88 Parabolas used in total out of a possible 92*

TABLE 5.1: A summary of the experiments performed during the 3 parabolic flights. Contained in column one is the number of parabolas during which we performed

each experiment and were able to record useful data over the course of the ESA 51st Parabolic Flight Campaign (*one parabola was aborted during the 3rd flight). The

other columns show the details of the different experiments performed throughout the campaign.

The camera software we had written provided an output file containing the exact computer time at the moment of each image taken per experimental run/parabola. Using this output file a new data file was created containing the exact time in seconds that each image was taken assuming the first image per parabola/experimental run was taken at time 0 seconds. We then had an accurate record, in seconds, of the frame rate for each parabola and ground based run.

5.2.2 Particle Tracking In each image approximately 600 particles can be seen (see Figure 5.1). Almost every single one of these particles is identified and subsequently tracked between consecutive images using a particle tracking algorithm [21]. The particle tracking process is completed in several stages. Firstly the individual particles must be identified. This is done using software that specifically searches for bright particles on a dark background. To aid particle detection, particle size (in pixels) must be specified and a band pass filter applied to smooth the image and subtract the background. Secondly the coordinates of the particles in the image must be identified. The band pass filter applied previously will have resulted in several bright spots being visible in the images. The locations of the true particles are found by specifying threshold

Education


brightness. Detected bright spots with brightness above this threshold level are normally the real particles and these are filtered out. Next, in order to improve the accuracy of the positions, the centroid of each detected particle is then calculated. At this stage a precise list of particle coordinates in one image can be produced. These positions can then be plotted over the original image to verify the accuracy of the detection (see Figure 5.2). The following step is to detect the particles in all of the images taken during an experimental run and finally construct two-dimensional particle trajectories from the scrambled list of particle coordinates determined for each frame (i.e. at discrete times). The algorithm which performs this task outputs the original data sorted into a series of trajectories assuming that no particle moves more than one tenth of a particle diameter in between frames. Each of the identified particle trajectories is assigned a unique ID number and a list containing the coordinates of the particle trajectories at each discrete time is produced. Finally, a test must be performed to determine if there is any pixel biasing affecting the particle tracking. The test involves plotting a histogram of the decimals of each of the x and y coordinates for each particle trajectory at each time i.e. if the position is 18.3 then 0.3 is plotted. In an ideal case these histograms should be flat showing there is no bias in the results (see Figure 5.3). As is often the case, the histograms do not appear flat (see an example of a biased tracking run in Figure 5.4) and the parameters must be adjusted and the particle tracking process run from the beginning again.

Figure 5.1: An example image obtained of the top layer of beads during the

experiment.

Education


Figure 5.2: Example of particle detection in one image (before any filtering).

Figure 5.3: Histogram demonstrating good tracking with no pixel-biasing. The

decimal of the particle position is plotted against number of particles for all position coordinates recorded during parabola 13 of flight 1. Blue shows the decimals of the

x-direction coordinates and red shows the y coordinates.

Education


Figure 5.4: Histogram demonstrating pixel-biasing. The decimal of the particle

position is plotted against number of particles for all position coordinates recorded during parabola 13 of flight 1. Blue shows the decimals of the x-direction coordinates

and red shows the y coordinates.

5.2.3 Calculating The Pixel Scale The pixel scale was calculated for the check beads, which are glued to the top surface of the confining pressure plate. The real separations were measured, centre to centre, with vernier callipers. The measured distances, shown in Figure 5.5, are for the 4 mm shear cell and have an estimated uncertainty of ± 0.2 mm. The check beads are then identified in the images and their centres are located. See Figure 5.6.

Figure 5.5: The measured separation distances for the check beads on the 4 mm

shear cell. The distances have an uncertainty of ± 0.2 mm.

Education


Figure 5.6: The identified check beads in an image and the distances between them

in pixels. The distances are then compared to give the pixel scale at the height of the check beads (i.e. on the surface of the pressure plate). See Table 5.2. The pixel scale, at the height of the check beads is found to be 5.43 ± 0.02 pixels/mm.

Measurement (mm) Measurement (pixels) Pixel Scale (pixels/mm) 75.2 408.71 5.43

54.5 295.24 5.42

74.7 407.28 5.45

56.2 304.20 5.41

AVERAGE 5.43 +/- 0.02

TABLE 5.2: Comparison of distances between check beads in mm and pixels in order to determine the pixel scale at the height of the check beads.

The viewing geometry of the check beads in comparison to the top layer of beads within the shear cell is shown in Figure 5.7. The pixel scale that has been calculated for the check beads on the surface of the pressure plate will not be the same as the pixel scale for the layer of beads below the pressure plate. The distance from the camera to the top of the pressure plate is 141 mm. The check beads on the surface are 4 mm in diameter and they sit at a depth of approximately 2 mm into the pressure plate (a small hole was made to accommodate them). The pressure pate has a thickness of 5 mm. Therefore, assuming the pressure plate is resting on the top layer of beads contained below, the difference in height between the top of the check beads and the top of the beads below will be 7 mm. The distance to the camera from the check beads, d1, is 139 mm and the distance from the top layer of beads below the pressure plate, d2, is 146 mm. The pixel scale (PS) will decrease with distance from the camera according to 1/d, where d is the distance to the camera. Therefore PS2 = PS1*(d1/d2).

Education


Assuming the pixel scale at the check beads (PS2) is 5.43 pixels/mm we can then determine that the pixel scale at the top layer of beads in the shear cell (PS1) is 5.17 pixels/mm.

Figure 5.7: The viewing geometry of the check beads which are glued to the top of the confining pressure plate and the top layer of beads contained within the shear

cell.

5.2.4 Calculating The Centre Of The Shear Cells Before any radial, angular displacements or velocities can be calculated the centre of the two dimensional disk must be known (i.e. the centre of the two concentric cylinders). The first step was to take one image and identify, by hand, several points distributed around the walls of the two cylinders. Points were chosen on the inner wall of the outer cylinder and on the outer wall of the inner cylinder. See Figure 5.8. This then gives a series of points for the large cylinder inner edge as (XL1, YL1), (XL2, YL2)… (XLn, YLn), and for the small cylinder outer edge as (XS1, YS1), (XS2, YS2)… (XSn, YSn).

Education


Figure 5.8: This shows the 38 points selected by hand on the inner wall of the outer

cylinder and the 32 points selected on the outer wall of the inner cylinder. To calculate the centre points accurately it was necessary to minimise the difference between the known radii and the calculated radii for a given set of centre points. To do this a Matlab routine was written to minimise the summed functions of A and B shown below. The values of the two centre coordinates that correspond to the minimum values of A and B are thus the centre coordinates for the two cylinders.

( ) ( )2

1

22∑=

−+−−=

n

i

iiYCYLXCXLRLA

( ) ( )2

1

22∑=

−+−−=

m

i

iiYCYSXCXSRSB

Where RL = 195 mm x pixel scale = 1008.15 pixels, RS = 100 mm x pixel scale = 517 pixels. (XC, YC) are the coordinates of the centre position. As the circle centre coordinates and the circle radii are now known the full circle solutions can be plotted and compared to the original image to check for accuracy. See Figure 5.9. The coordinates of the two centre points, in pixels, are:

• Outer cylinder (1103.41, 253.63)

• Inner cylinder (1103.61, 255.73)

Education


This disagreement is 0.04 mm in the x direction and 0.41 mm in the y direction corresponding to within 1/8 of a particle diameter or better in both cases. From this point onwards the mean of this centre position (1103.51, 254.68) is assumed for both cylinders.

Figure 5.9: The circle solutions found via minimisation of the difference between the

known radii and the calculated radii plotted over the original image.

5.2.5 Removing Spurious Particles Several procedures must then be carried out to ensure there are no spurious particles in the data. These can be caused, for example, by reflections of the lamps on the pressure plate or reflections of particles in the walls of the shear cell. Identifying the spurious particles within the bulk of the sampled data is done via determining the intensities and sizes of the tracked particles. Normally the spurious particles lie outside the normal range. Figure 5.10 shows the particles in one image plotted as a function of their brightness. Figure 5.11 shows the same particles plotted as a function of particle size. The average brightness and particle size per trajectory is calculated and then using the information obtained from the plots a filter is applied to the data to remove the abnormal (spurious) data. By this method the majority of the spurious particles within the bulk of the material are removed. Those that are not detected via this method must be identified by hand. The ‘particles’ outside the shear cell, i.e. with radii greater than that of the outer cylinder or less than that of the inner cylinder are filtered out using the calculated radial coordinates of the shear cells as boundary conditions.

Education


See Figures 5.12 and 5.13 to demonstrate a fully filtered image with no spurious particles. All spurious particles have been removed by a combination of particle brightness filtering, particle size filtering and radial boundary filtering.

Figure 5.10: Particles detected in one frame plotted as a function of their brightness.

Figure 5.11: Particles detected in one frame plotted as a function of their detected

size.

Education


Figure 5.12: Fully filtered plot of all particle positions with no spurious particles.

Particles have been filtered as a function of their radial position, size and brightness.

Figure 5.13: The particle positions shown in Figure 5.12 plotted over the original image to demonstrate which particles have been tracked and which have not.

Education


5.2.6 Calculating Angular Displacements Using the particle positions in each frame the average displacement and velocities of particles can be computed. Firstly, the displacement vectors between every set of consecutive frames are calculated for each particle in Cartesian coordinates. This coordinate system is then transformed into polar coordinates using the coordinates at the centre of the cylinders, as calculated in Section 5.2.3, giving the angular displacements. The angular displacements, between two consecutive frames, are determined for each particle. The particles are then separated according to their position, at every time interval, into five equal cylindrical rings (radial bins) reaching from the inner cylinder wall to the outer cylinder wall. See Fig 5.14. Once the particles have been divided into radial bins the mean particle angular displacement per bin can then calculated.

Figure 5.14: One image with the boundaries of the five radial bins plotted over the

top. The particles are split into these radial bins based on their radial position.

5.2.7 Conversion Into Real Units In order to convert the displacements into velocities the exact times in between consecutive frames must be known. Normally, the frame rate should be a known and constant value, however during the parabolas our frame rate varied slightly due to interference with the data transfer. The camera software, as described in section 5.2.1, provided an output file containing the exact computer time at the moment of each image. By converting this time to seconds for every single frame it will then be possible to calculate the angular velocity in radians per seconds.

Education


When a detailed analysis is performed of the radial velocities the camera coordinates will also have to be converted into real coordinates. However at this stage we have focussed on the angular velocities so the coordinates have been left in camera units.

5.3 Results So Far

5.3.1 Angular Displacement During Uniform Shear An example plot showing the mean angular displacement, in radians, per interval, per radial section is shown in Figure 5.15. These are the results from parabola A13 - a uni-directional shear experiment performed in microgravity with 4 mm beads and a rotational rate of 8 mHz in a counter-clockwise direction. The shear cell was shaken before the experiment. It can be seen that the angular displacement per frame is greatest closest to the inner cylinder wall as expected.

Figure 5.15. This plot shows the results of one uni-directional shear experiment

performed in microgravity (parabola A13). The mean angular displacement in each section, per interval, is plotted with the standard error. The particles are divided into 5

radial bins with 1 being closest to the shearing cylinder wall.

5.4 Work Still To Be Done Analysing the images from one camera for just one experimental run is a very lengthy and computationally demanding process. As a result at the time of preparation of this report we are still very much involved in the data analysis phase. The key analysis that remains to be done in order to achieve our objectives is the following:

• Complete the particle tracking process for all parabolas and all ground based results for both cameras.

Education


• Investigate how the particle angular displacements and velocities vary with time throughout the experiment.

• Compare all of the experiments that should be identical to check that we have the same results.

• Investigate the effect of all experimental parameters (particle size, shear velocity, gravity level) on uni-directional shear.

• Investigate the effect of all experimental parameters (particle size, shear velocity, gravity level) on shear reversal

• Use the check beads glued to the upper surface of the pressure plate to determine if the height and thus the bulk density changes during the experiments. If this is the case the particle tracking will also have to be adjusted at the positions of the beads relative to the camera will alter, hence altering the pixel scale.

• Examine the data recorded by the motor inverter to determine if we can see a difference in shear strength of the material during shear reversal.

• Compare the results with the exact plots of the gravity levels to determine any effects of the varying gravity.

• Determine the effect (if any) of shaking the shear cell before the parabola. Once all of the above points have been addressed this will allow the main AstEx objectives; time to reach a steady state, monitoring the flow and the effect of shear reversal on the flow, to be achieved. Eventually, we will have direct measurements of granular flow caused by rotational shear forces, direct measurements of the timescales required to reach a steady state flow in three different gravity regimes and we will also be able to determine the effect shear reversal has on the dynamics of a granular flow in 0 g and 1.8 g. Finally, once the data analysis is complete we aim to compare the results directly with numerical simulations. In these numerical simulations we will use the exact levels of gravity recorded on the parabolic flights for the different gravity regimes.

5.5 Expected Outcome In previous experiments the history dependence of the shear flow was found in a situation where gravity is confining the system. In this situation when the shear stress is reversed the initial contact network breaks but then gravity quickly compacts the system. This is seen at the start of the shear flow. In microgravity, if the shear rate is large compared to the gravitational timescales then we expect the system may fall apart dramatically rather than compact. If pushed too hard along an axis that does not have force chains, the ensemble of grains might offer essentially no shear resistance. Since the resistance is already low during shear reversal even though the system compacts, it should be even lower in microgravity! So we expect a very dramatic weakening of the material in microgravity.

Education


6. Conclusions

6.1 Design Changes For Future Flights The AstEx experiment worked very well but it could still be improved. Based on our experiences during the parabolic flight campaign a list has been made of the design changes the AstEx team would like to implement before any future flights. The AstEx team would like to:

• Remake the mild steel shaft and mild steel fittings using stainless steel to avoid all problems with corrosion.

• Rearrange the components inside the experimental rack to try to minimise electromagnetic interference from the motor and inverter. We would also add a special electromagnetic interference dampening device for the motor inverter, and coat the USB data cables in proper shielding tape.

• Use solid-state hard drives to ensure no problems are caused by the aircraft vibrations when trying to save the data.

• Implement a more accurate and reliable method of measuring the torque, and thus the shear strength, during the experimental runs.

• Adjust the camera illumination lights so that their reflections off the pressure disks do not appear in the images.

• Consider implementing a different technique for shaking the shear cell in between parabolas if this is found to be an essential part of the experiment.

6.2 Scientific Conclusions At this stage, it is too early to make any firm scientific conclusions. The data analysis is a very complex process and we recorded a very large quantity of data during the parabolic flight campaign. However, we are very happy with the way in which our experiment performed. Throughout the parabolic flight campaign the AstEx team were able to overcome many challenges and were successful in taking useful data in 96% of the parabolas! We are very confident that all of the microgravity experiments were a great success and that much valuable science will be produced as a result.

6.3 Science To Be Investigated The ESA 51st parabolic flight campaign provided the AstEx team with a fantastic opportunity to obtain a large quantity of valuable scientific data. However, with the AstEx hardware that has now been built there are many important but simple experiments we would still like to have the opportunity to perform. Some of the investigations, possible with a few minor changes to the existing hardware, are listed below. They are not given in order of priority because it is likely that the results of this flight campaign will influence them. Future microgravity experiments the AstEx team would like to perform include:

• Shear and shear reversal at constant volume.

• Bimodal size distribution of particles undergoing shear and shear reversal.

• Investigations of larger ranges in rotation speed and thus shear velocities acting on the particles.

Education


7. References [1] H. Yano et al., Proceedings of Asteroids, Comets, Meteors, 103-106 (2002). [2] M. A. Franzen et al., Lunar and Planetary Science, XXXV, 1716 (2004). [3] P. Richard et al., Nature Materials, Vol. 4, 121-128 (2005). [4] R. L. Brown et al., Principles of Powder Mechanics, Pergamon Press (1970). [5] M. Schröter et al., Phys. Rev. E, Vol. 71, No. 3, 030301 (2005). [6] S. Slotterback et al., arXiv.org:0802.0485 (2008). [7] E. I. Corwin, Physical Review E (Statistical, Nonlinear, and Soft Matter Physics), Vol. 77, No. 3, 031308 (2008). [8] R. Delannay et al., Nature Materials, Vol. 6, No. 2, 99-108 (2007). [9] D. C. Richardson et al., Science, 306, 1526 (2004). [10] P. Michel et al., Icarus, 200, 503 (2009). [11] P. Paolicchi et al., Asteroids III, 517-526 (2002). [12] K. Holsapple et al., Asteroids III, 443-462 (2002). [13] W. F. Bottke et al., Nature, Vol. 381, 51-53 (1996). [14] K. A. Holsapple, Icarus, 205, 430-442 (2010). [15] M. Toiya et al., Phys. Rev. Lett., Vol. 93, No. 8, 088001 (2004). [16] M. Toiya, http://hdl.handle.net/1903/3886 (2006). [17] M. L. Falk et al., arXiv:0802.1752 (2008). [18] D. C. Richardson et al., Icarus, 143, 45-49 (2000). [19] P. Michel et al., Icarus, 168, 2, 420-432 (2004). [20] D. C. Richardson et al., DPS Meeting, 40, 55.02 (2008). [21] D. Blair and E. Dufresne, http://physics.georgetown.edu/matlab/ (2009).

Education


Figure 6.1: The AstEx team are incredibly happy with the outcome of the experiment.

astex fyt2009 final report v2 - european space...

Documents