Page 1

ANNUAL 2014-15 • DEPARTMENT OF AERONAUTICS AND ASTRONAUTICS • MASSACHUSETTS INSTITUTE OF TECHNOLOGY


Department Head Jaime Peraire peraire@mit.edu

Associate Head Eytan Modiano modiano@mit.edu

Editor & Director of Communications William T.G. Litant wlitant@mit.edu

AeroAstro is published annually by the Massachusetts Institute of Technology Department of Aeronautics and Astronautics, 33-240, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139, USA. http://aeroastro.mit.edu AeroAstro No. 12, September 2015 Š2015 The Massachusetts Institute of Technology. All rights reserved. DESIGN Opus Design www.opusdesign.us Cover: On March 4, 2015 at 3:30 a.m. EST, the Microsized Microwave Atmospheric Satellite was successfully deployed from the International Space Station. MicroMAS is a 3U (30cm x 10cm x 10cm) nanosatellite fitted with an atmospheric sensor. The project is a joint effort between MIT AeroAstro students, who built the dual-spinning spacecraft bus, and MIT Lincoln Laboratory, which led development of the passive microwave radiometer payload. MicroMAS was designed to make temperature maps of tropical storms and hurricanes, characterizing structure and improving prediction models. (NASA)

Cert no. XXX-XXX-000


Where we can go What a centennial year it’s been! It seems like only yesterday we were preparing for the department’s Centennial Symposium, fine-tuning the agenda, prepping panel moderators, and confirming travel arrangements for some of the most brilliant people in aerospace. With more than 1200 in attendance, the Symposium was a great success. Whether you were most interested in Elon Musk’s views on Mars colonization or the Apollo astronauts’ memories of Doc Draper and their training in Cambridge, there was something for everyone. We were honored to have not only so many luminaries on stage, but also hundreds of alumni, students, and department friends in the audience. In the pages of this annual, you will find photos of the event. We encourage you to take a look at the videos available on our website http://aeroastro.mit.edu/ video-categories/2014-centennial-symposium-videos; they make excellent and entertaining viewing. Never content to rest on our laurels, the department has embraced its next century eager to explore new opportunities and build on existing strengths. We are delighted to report that renovations to Building 31, home of the Gas Turbine Lab, are underway. This project is critical to the future of our department and a testament to the confidence that the MIT administration and our many friends have in us. Of the 65,000 sq. ft. building, the department has historically occupied ~40,000 sq. ft., and will gain ~6,000 sq. ft. in the revamping and redesign. We are not only expanding our seat count for students, we’re also improving the quality of those seats. What’s more, we’re building a high bay flying robotics area, an outdoor flying area, and a bridge connection to Building 37, bringing together heretofore separate department facilities. The new space will also include an area designated for joint AeroAstro-Lincoln Beaver Works activity to further enhance our partnership with Lincoln Laboratories. For more on Building 31, we refer you to Mark Veligor’s article in this issue of AeroAstro.

Eytan Modiano (left) and Jaime Peraire.

In January 2015, 40 sophomore Course 16 majors were joined by four faculty and three staff members on a field trip to the West Coast. Among our stops were the Boeing Company in Everett and Renton, the Museum of Flight, Blue Origin, JPL, SpaceX, and Northrop Grumman. The students and chaperones were both excited and enthused. One student’s remarks reflect the general sentiments: “It was amazing. Thank you so much for planning this! I thought it was a good mix of aero and astro. Each company was distinctly different and offered a different insight on what AeroAstro can be like. I really learned a lot about how planes and rockets are manufactured.” We anticipate making such trips a regular January Unified feature. We may tweak the itinerary, but rest assured our students will benefit from the experience. As part of our outreach efforts, and consistent with our strategy to leverage online resources for education, in the Spring of 2015 we offered the online subject 16.00x Introduction to Aerospace Engineering: Astronautics and Human Space Flight taught by Professor and former astronaut Jeff Hoffman. With more than 12,000 students registered representing 150 countries, the course was received with tremendous enthusiasm, students commenting: “Amazing course!!” “The best edX experience I have had so far,” “Videos were interesting and energetic,” “Thank you, MIT and Prof. Jeff Hoffman for this mind blowing experience.”

Lab report

i


AeroAstro sophomores, along with several faculty and staff, visit Boeing’s Everett, Wash. factory as part of their January 2015 West Coast aerospace facility tour. (Boeing)

We’re offering our upperclassmen an equally exciting opportunity during Academic Year 2015-16 with the introduction of SuperUROPs (advanced Undergraduate Research OPportunities) in AeroAstro. Already a great success in Electrical Engineering and Computer Science, SuperUROPs offer upperclassmen the opportunity to tackle real-world problems under the direct supervision of MIT faculty members. The students work approximately 10 hours per week in their supervisor’s laboratory, in addition to attending a special weekly lecture. Students receive a named stipend (generously funded by gifts from industry and alumni) as well as course credit. At the time of this writing, we have 21 students signed up and ready to begin their SuperUROPs in September 2015. We’ll keep you posted on the program’s progress. While there is much that is new in the department, we never lose sight of ongoing issues. Our Strategic Plan “Where We Can Go,” developed over two years, is now complete and you can find it on our The 2015 AeroAstro website http://issuu.com/mitaeroastro/ Strategic Plan docs/strategic-plan-2015. The plan, which highlights our core competencies and identifies strategic opportunities in the areas of Air Transportation, Autonomous Systems, Small Satellites and Engineering Education, is our blueprint for the future. We would like to take this opportunity to thank our alumni as well as more than 60 aerospace, government, and academic leader aerospace for their contributions to the development of the plan.

Here in AeroAstro, we sometimes forget how remarkable are the people associated with our department. Whether faculty like Dava Newman, Apollo Professor of Aeronautics and Astronautics, who was recently appointed NASA Deputy Administrator, or alumni like Steve Isakowitz, President of Virgin Galactic, interviewed in this issue, or students like Edward Obropta and Forrest Meyen, members of the MIT 2015 100K competition winning team, the people of AeroAstro are what makes this department so special. Where else would Apollo 11 Command Module pilot Mike Collins visit for a one-on-one conversation? As a Sloan Fellow outside our department commented, “only at MIT would you get 10 Shuttle and ISS astronauts (nine of whom are MIT alumni) and the future Deputy Administrator of NASA on stage and only give them 90 minutes.” We do move quickly here, it’s true. We have our next 100 years to think about! We hope you enjoy the stories and photos. As always, we welcome your feedback.

JAIME PERAIRE Department Head

EYTAN MODIANO Associate Department Head


1

Advanced turbomachinery design for supercritical carbon dioxide applications

35 Building 31 renovations to transform

AeroAstro’s research space

by Zoltán S. Spakovszky

7

By Mark Veligor

Making order of chaos

43 “The reinvention of space”

by Qiqi Wang

A conversation with Virgin Galactic president Steve Isakowitz

13 The challenge of living on Mars

51 Celebrating AeroAstro’s 100 years

by Andrew Owens and Olivier de Weck

1914 – 2014

21 “Autos” racing beneath MIT teach

By William Litant

embedded system basics

58 Lab Report

by Sertac Karaman

A 2014-2015 review of Aeronautics and Astronautics Department laboratories

27 Quantifying uncertainty in the

physical world By Youssef Marzouk

1

7

13

21

27

35

43

51

58


Professor Zoltรกn Spakovszky, director of the MIT Department of Aeronautics and Aeronautics Gas Turbine Laboratory, is developing new compressor technologies for supercritical CO2 applications. (William Litant/MIT)

iv

AEROASTRO 2014-2015


Advanced turbomachinery design for supercritical carbon dioxide applications by Zoltán S. Spakovszky

Carbon dioxide, the so-called “gas of life” and vital molecule for plants, has recently moved in to the ecological, economical, and political spotlight.

With the aide of sunlight plants convert CO2 to sugars and oxygen through photosynthesis, and increased levels of CO2, for example in greenhouses, dramatically improve plant growth. However, the increase in CO2 levels in our atmosphere from both natural and human activities, has had a major impact on global warming. Fossil fuels account for about 84% of the annual total U.S. energy supply and make up roughly 90% of the worldwide primary energy consumption. Nearly one third of CO2 emissions in the U.S. are from coal power plants.

Technologies are being developed to mitigate the CO2 impact on global warming by capturing the carbon dioxide from power plants and industrial processes, transporting it in compressed form to suitable locations, and storing it in deep underground rock formations. This process is called “carbon capture and sequestration.” The carbon sequestration part requires efficient, high-pressure CO2 compressors. To put the economical challenge in perspective, a 500 megawatt power plant emits about 3.5 million tons of CO2 annually, or about 400 tons per hour. Assuming the CO2 can be captured, deep underground storage requires roughly 60 megawatts of power with current compressor technology; this is 12.5% of power plant output, a huge operating cost. Advancing the state of the art of CO2 compressor technology to address this issue, in collaboration with industry, is the topic of this article.

Advanced turbomachinery design for supercritical carbon dioxide applications

1


THE GOOD, THE BAD AND THE UGLY The geological sequestration sites are often miles beneath the earth’s surface, and the gas pressures required to inject the CO2 are above the so-called critical point, where the distinct liquid and gas phases we are familiar with (e.g., water and steam) do At supercritical conditions, surface tension vanishes, not exist. At supercritical condipreventing formation of bubbles and droplets. (NASA) tions the surface tension vanishes, so no bubbles or droplets can form. At this state, the CO2 becomes a fluid with striking properties: the very low viscosity makes fluid friction practically negligible, much to the advantage of the aerodynamicist; the thermodynamic properties change rapidly close to the critical point, which tantalizes the thermodynamicist; and the water-like density greatly increases the machine’s power density, making shaft vibrations, much feared by the rotordynamicist, dangerous. For comparison, the power density of a large jet engine is a few kilowatts per kilogram, but the power density of a supercritical CO2 compressor of about one-tenth the diameter is roughly 100 kilowatts per kilogram.

THE OPPORTUNITY While the features of two-phase (liquid and vapor) substances in power systems are part of the syllabus in the undergraduate thermodynamics course I teach, I had a wonderful opportunity to gain much closer acquaintance with CO2 when on a sabbatical leave at Mitshubishi Heavy Industries (MHI) in Japan, several years ago. As a technology special advisor for MHI, I assisted them in solving research and development and field problems in different business branches, determined areas where new technology capability is needed for current or proposed products, and carried out other special assignments. Reviewing MHI’s CO2 compressor product line and related technology, I realized there was an opportunity for improving compressor efficiency and product cost, which are important drivers in a competitive marketplace. The goal was to achieve a step change in performance of multi-stage compressors for dense gases and supercritical fluids.

2

AEROASTRO 2014-2015


Further investigations indicated that conventional turbomachinery wisdom might not necessarily apply to supercritical working fluids. More specifically, high density, high-pressure supercritical CO2 requires high-speed compressors operating at very low volume flows. As a consequence, so-called “parasitic losses,” such as leakage flows and disk windage loss, become dominant. Further, the narrow blade passage heights (of order millimeters) lead to increased friction losses which can outweigh the benefits of low viscosity supercritical CO2. It thus became clear that new strategies and a paradigm shift in compressor design were required.

UNCHARTED DESIGN SPACE Enhanced Oil Recovery utilizes high density, high-pressure Returning to MIT’s Gas Turbine Laboratory, we initiated a supercritical CO2, which is supplied by super high-pressure research project in collaboration with MHI and explored the compressors (MHI) uncharted design space, carrying out first principles based modeling combined with tailored numerical simulations and laboratory-scaled experiments. Our aim was to characterize the internal flow behavior of supercritical fluids and to establish a foundation for advanced compressor design. Combining insight from pump design (for liquids) with new knowledge of behavior near the critical point and an unconventional compressor architecture (for gas), our study suggested a large increase in efficiency relative to the state of the art. Three years into the project, MHI matured our technology into a full-scale, 700 atmosphere, supercritical CO2 prototype compressor experiment at one of their research centers. Relative to current practice, the tests demonstrated the predicted increase in efficiency and improved the stable operating range, exceeding rotordynamic performance targets with machine size reduced to two-thirds and machine weight cut in half. MHI dubbed the resulting new product line “super high-pressure CO2 compressors.”

DOES CONDENSATION MATTER? Pushing the performance boundaries further, the next question was how close to the two-phase region, where vapor and liquid coexist, and how near the much-feared critical point could one

Advanced turbomachinery design for supercritical carbon dioxide applications

3


operate these compressors. The advantage of such operating points is a cooler compressor inlet temperature requiring less shaft power input. A major unknown, evidenced by a lack of data and useful information in the literature, is the effect of condensation (phase change from gas to liquid) in such compressors for carbon capture and sequestration (CCS) and for enhanced oil recovery applications. Phase change in turbomachinery has been extensively studied for steam turbines, where steam (water vapor) condenses into droplets as it expands through the last turbine stages. The droplet formation can lead to aero-thermodynamic penalties and mechanical damage. This phenomenon is uncommon in The author (front row, fourth from left) and the MHI compressors where the compression process occurs away from Takasago R&D team with which he collaborates on numerous projects including CO2 compressor development. (MHI) the two-phase region, but there is a real concern and a debate in the CO2 compressor community. Our supercritical CO2 gas calculations showed that, while local flow acceleration near the leading edge of the compressor blades can lead to condensation pockets (this is the opposite of cavitation where the liquid boils, as in high-speed ship propellers and rocket engine turbopump inducers), condensation would have little impact on overall performance and stability. To assess real gas effects on internal flow behavior near critical conditions and to validate the established modeling tools, we designed and built a first-of-its-kind supercritical CO2 experiment at the Gas Turbine Laboratory. The experiment together with calculations allowed us to establish a new criterion for condensation onset in compressors. We demonstrated that away from the critical point condensation does not occur because the flow residence time typical in these compressors is much shorter than the time required for extensive condensation. The work not only resolved a longstanding and basic question of condensation occurrence; it allowed MHI to operate their machines closer to the critical point. In addition to the performance improvements from advanced design features, compressor operation at reduced inlet temperatures led to a substantive reduction in shaft power input. In summary, the combination of these technologies reduces the current energy requirement for carbon sequestration by about one quarter, contributing to the economical viability of CCS.

4

AEROASTRO 2014-2015


OPPORTUNITIES AHEAD Encounters with supercritical fluids are not limited to the gas processing industry. The rise in fuel prices and environmental concerns drive the development of transport aircraft propulsion systems. These propulsion systems comprise a gas turbine engine, which converts fuel power into shaft power, and a propulsor, typically a large ducted fan, that converts shaft power into propulsive power to balance the drag power produced by the airframe. The overall metric is the product of the thermal efficiency of the gas turbine and the propulsive efficiency of the propulsor; to reduce fuel burn, both high thermal efficiency and a high propulsive efficiency are needed. The latter can be improved by larger diameter, lower pressure ratio fans; the former depends on the turbomachinery component efficiencies and overall cycle parameters such as pressure ratio and turbine inlet temperature. The increasing trend in cycle pressure ratio indicates that fuel injection can occur at supercritical conditions. As the surface tension vanishes, fuel droplets do not exist and, therefore, the fuel stream does not break up into a spray. This opens new challenges and opportunities for reduced emissions and improved fuel consumption. The Gas Turbine Laboratory is working on a wide range of power and propulsion problems important to the industry partners and government agencies that support our work. These are exciting times in aviation for us to pursue our passions in turbomachinery and propulsion research. We are also establishing a center of excellence for radial turbomachinery, tackling a variety of problems such as cavitation in rocket engine turbopumps, instabilities in advanced turbochargers, real gas effects in supercritical CO2 compressors, and non-linear oscillations in high-pressure power plant pumping systems. ZOLTI SPAKOVSZKY is a Professor of Aeronautics and Astronautics at the Massachusetts Institute of Technology and the director of the Gas Turbine Laboratory. He obtained his Dipl. Ing. degree in Mechanical Engineering from the Swiss Federal Institute of Technology (ETH) Z端rich and his MS and PhD degrees in Aeronautics and Astronautics from MIT. His principal fields of interest include internal flows in turbomachinery, compressor aerodynamics and stability, dynamic system modeling of aircraft gas turbine engines, micro-scale gas bearing dynamics, and aero-acoustics. Spakovszky may be reached at zolti@mit.edu

Advanced turbomachinery design for supercritical carbon dioxide applications

5


NASA accurately predicted the time-averaged vibration of its Space Launch System by simulating the chaotic airflow around the rocket. (NASA)

6

AEROASTRO 2014-2015


Making order of chaos by Qiqi Wang

In 1972, MIT meteorology professor Edward Lorenz (SM ’43, ScD ’48), addressed the 139th meeting of the American Association for the Advancement of Science.

His talk, titled “Does the flap of a butterfly’s wings in Brazil set off a tornado in Texas?” changed scientists’ view of the world. Lorenz had discovered that how weather evolves cannot be predicted well into the future. According to the laws that determine how the atmosphere evolves, a tiny perturbation, such as the flap of a butterfly’s wings, can cause large changes in the weather a few weeks later. Unless all such tiny perturbations on the Earth are observed and modeled, which is, of course, impractical, any attempt to predict the weather more than a few days in advance will be wrong. This phenomenon, now known as the butterfly effect, has since been observed in many natural and engineering systems. These systems evolve according to deterministic and apparently predictable laws, but exhibit seemingly random motions and, due to the butterfly effect, are unpredictable beyond a short period of time. Such systems are called chaotic systems. My research concerns chaotic systems found in many aerospace applications. For example, airflow inside a jet engine forms turbulent boundary layers that have chaotic dynamics. The turbulent flow increases drag, heats the engine components, and makes noise and vibration, so engineers design the engine to delay it and mitigate its effect. Engineers can also use chaotic systems to our benefit. In the same engine, the mixing of jet fuel and air is designed to be chaotic, so that efficient burning can be achieved in a compact and lightweight combustion chamber.

Making order of chaos

7


ORDER FROM CHAOS If the flight of an airplane depends on chaotic systems, could the next flight you take be as unpredictable as the weather? The answer is no. Despite all the chaotic dynamics involved, flying is one of the safest way to travel because the chaotic systems in aircraft components operate at a very fast time scale, typically on the order of a millisecond or less. This is much shorter than the timescale on which we perceive the behavior of an airplane or its jet engine. We are not fast enough to see the flames flickering in the engine combustor or feel the turbulent eddies rushing over the aircraft’s skin. We observe only the time-averaged effect of these phenomena and, fortunately, the safe flight of an airplane depends only on these time-averages. In my research, I predict time-averaged effects by simulating these chaotic systems computationally. The time-averaged heat transfer in a chaotic fluid flow is an example of important quantities to predict. Some jet engine components are exposed to burning flames of temperatures up to 2,000 degrees Celsius. Designing a reliable jet engine depends on predicting the time-averaged heat transfer on the surface of these components. The fluid flow governing the heat transfer can be simulated by numerically solving the Navier-Stokes equations, the governing laws of fluid motion. Due to the butterfly effect and imperfections in modeling the geometry, boundary conditions and in the numerical method, the instantaTime-averaged effect neous heat transfer in the flow field is unpredictable. However, the is predictable, while time-averaged heat transfer can be predicted by the simulation.

instantaneous snapshots are not. This phenomenon can be observed in daily life.

Time-averaged effect is predictable, while instantaneous snapshots are not. This perhaps surprising phenomenon can be observed in daily life, for example in the flow of water over a dam. In a highspeed photographic exposure, the individually visible droplets and splashes of water are apparently positioned randomly and are unpredictable. The time-averaged flow, photographed with a long exposure time, appears to be deterministic, and in principal, can be predicted using the geometry of the underlying terrain and the upstream flow conditions.

8

AEROASTRO 2014-2015


In a high-speed photo of water flowing over a dam (left), the individually visible water droplets and splashes appear randomly positioned and unpredictable. However, when photographed with a long exposure time (right) the flow appears deterministic, and in principal, can be predicted using the geometry of the underlying terrain and the upstream flow conditions. Hence, we have created order from chaos. (William Litant/MIT)

Because time-averaged effect is predictable, NASA succeeded in predicting the time-averaged vibration of the Space Launch System, a heavy launch vehicle designed to replace the Space Shuttle, by simulating the chaotic airflow around the rocket. The simulation, which matches wind tunnel measurements, gave NASA engineers insight into the cause of the vibrations and how to design the rocket to reduce the vibration.

DESIGNING WITH CHAOTIC SIMULATION The primary challenge in my research is performing design using chaotic simulations. With an ability to predict the time-averaged effects, simulation of chaotic systems can help engineers design aerospace vehicles. It enables them to iterate through different designs and configurations, comparing them and finding the one with most desirable time-averaged effects. Performing this design procedure with a chaotic simulation is similar to doing so with a non-chaotic simulation, except for the key component of comparing among different designs. More precisely, analyzing how the time-averaged effects respond to various design changes, namely sensitivity analysis, has a unique difficulty in chaotic simulations.

Making order of chaos

9


This unique difficulty arises because of the sampling error. We stated before that time-averaged effects in a chaotic system are predictable. Now, let’s make this more precise: Only infinitely-long time-averaged quantities in a chaotic system are completely predictable and, therefore, eligible as quantities of interest in engineering design. Any finite time average is less predictable in that it differs from its infinite-long time average counterpart by a small amount. This small amount, called the sampling error, behaves much like an instantaneous quantity in chaotic systems: it is unpredictable, apparently random, and is sensitive to small perturbations. The longer we timeaverage a quantity, the smaller the sampling error we can expect it to have. However, its rate of diminishing is so slow that the sampling error is negligible in few practical simulations. Causing more trouble than its non-negligible size is its sensitivity to perturbations. A small change in the design can change the sampling error unpredictably. When performing sensitivity analysis on a finite time average, which is the sum of the infinite-long time average and the sampling error, the sensitivity of the sampling error often overwhelms the sensitivity of the infinite-long time average. When this happens, it is difficult to deduce the sensitivity of the infinite-long time average, which is the true quantity of interest, from the sensitivity of the finite time average, which is what can practically be computed from a simulation. One way to overcome this problem is by filtering. The sampling error can be modeled as statistical noise that is independent for each design. Statistical filtering through a large number of designs would remove the effect of the noise. However, because it requires many simulations, filtering is expensive, to the point that it may not even be practical when the space of possible designs has many dimensions.

LAST SQUARES SHADOWING I have developed what I call the Least Squares Shadowing method, designed for sensitivity analysis of time-averaged quantities in chaotic simulations. The method relies on the fact that a pair of simulations can “shadow� each other, meaning their respective snapshots look like each other under a time transformation. Such shadowing simulations cannot be obtained from the same governing law with different initial conditions, because the butterfly effect will cause them to have totally different snapshots after a while. Nor can shadowing be achieved by simulating different governing laws, such as different designs of a jet engine turbine blade, but starting from the same

10

AEROASTRO 2014-2015


initial condition. The simulations would again diverge from each other. However, shadowing can be achieved by simultaneously perturbing both the governing law and the initial condition. The existence of such shadowing is guaranteed by the shadowing lemma of dynamical systems under strict conditions. My Least Squares Shadowing method finds such pairs of shadowing simulations, and performs sensitivity analysis with them. My current research in chaotic systems has a theoretical branch and a computational branch. The theoretical branch is establishing a foundation for the Least Squares Shadowing method. I have proven the convergence of this method in ideal cases. We are not only generalizing our proof to more realistic cases; we are improving the method so that it is more accurate for these more realistic cases. The computational branch of my research is accelerating the simulation large scale chaotic fluid flows, their sensitivity analysis and design particularly using the Least Squares Shadowing method. We are designing computational algorithms to achieve these goals, and implementing these algorithms in software widely used by engineers. For example, I am working with NASA, to integrate these algorithms into FUN3D, NASA’s flagship computational fluid dynamics solver, which is being used to design next generation vehicles. With a better understanding of chaotic systems, and ability to model, simulate, and design them, we are contributing to the development of more efficient and safer jet engines, more environmentally friendly aircraft, and more capable launch systems. QIQI WANG is an associate professor in the MIT Department of Aeronautics and Astronautics. His specializations and research interests include engineering design of chaotic dynamical systems, unsteady aerodynamics and turbulence, numerical methods for exascale computation, and design optimization of uncertainty. He maintains a blog at http://engineer-chaos.blogspot.com and may be reached at qiqi@mit.edu.

Making order of chaos

11


In this artist’s concept of a future Mars mission, after driving a short distance from their landing site, two explorers stop to inspect a robotic lander and its small rover. This stop also allows the crew to check out the life support systems of their rover and space suits within walking distance of the base. (NASA/Pat Rawlings)

12

AEROASTRO 2014-2015


The challenge of living on Mars by Andrew Owens and Olivier de Weck

To put humans on Mars, keep them alive when they get there, and bring them home safely, all while maintaining a reasonable budget, we will have to rethink the way we plan and execute human spaceflight missions.

So far, most human spaceflights have been relatively short — from Yuri Gagarin’s brief orbit of the Earth in 1961 to Apollo 17’s nearly two-week trip to the moon and back in 1972. Longer missions, such as those to Mir or the International Space Station, receive regular resupply from home. In the latter cases, the endurance of the system — the maximum time between subsequent resupply opportunities has been relatively short, and, should it be necessary, crews almost always have the option to abort their mission and expeditiously return to Earth. In fact, all human spaceflight to date has been Earth-dependent, relying on proximity to home for logistics support and risk reduction. In contrast, some future missions will take humans farther away from Earth than ever before. Opportunities for resupply will be more limited; mission durations will be longer and require greater endurance from our systems. If something goes wrong, it may take months or even years for space travelers to return home. We are facing never-before-seen logistical challenges, and we will have to adapt and become more Earth-independent.

The challenge of living on Mars

13


Mars is the current goal of human spaceflight, just as the moon was the focus of attention in the 1960s. The giant leap from Earth orbit to the surface of the moon was a formidable challenge, particularly with regard to propulsion and guidance, navigation, and control. The leap to put humans on Mars will be significantly more difficult. At its closest approach, Mars is well over 100 times farther away than the moon — but even more challenging than the long distance and flight time is the fact that the launch window for Mars only opens up every 26 months, when the planets are properly aligned. While some Mars mission plans are shorter than others, there is no such thing as a short trip to Mars. The farther away from Earth we go, the more complicated the logistics of space exploration become. Analysis tools, like SpaceNet, examine the network of possible waypoints on the path to Mars in order to determine optimal logistics strategies for a Mars exploration campaign. (Takuto Ishimatsu/MIT)

MORE COMPLEX MISSION, MORE COMPLEX SUPPLY NETWORK

The ISS is supported by direct resupply flights from a set of launch sites on the ground; the Apollo flights to the moon carried with them all of their supplies for the entire mission. Missions to Mars will require a more complex supply chain network, and more resources will be needed to sustain crews between resupply opportunities. Many different pathways are possible, and there are many questions that must be examined to guide the development of a Mars exploration plan. Should lunar orbit or the moon itself be waypoints on the path to Mars? Can the Martian moons,

14

AEROASTRO 2014-2015


Phobos and Deimos, be used as intermediate destinations? How (and where) can in-situ resource utilization be used most effectively to produce useful materials? How do new technological capabilities, such as 3D printing or automated plant growth systems, affect the logistics costs of these missions? More importantly, how do variations and uncertainties in system parameters — production rate, efficiency, mass, reliability, commonality, crew size, etc. — impact the results, and how can examination of these factors inform technology development efforts? With complex problems such as these, it’s important to maintain a systems view, searching for global optima rather than splicing together a set of locally optimized system elements. In AeroAstro’s Strategic Engineering Research Group, we’re developing tools and techniques to examine these questions holistically, and give mission planners and technology developers the information they need to design the exploration campaigns of the future.

We’re developing tools and Our work in this area began in 2005 as part of the MIT Space Logistics Project in support of NASA’s Constellation program. techniques to give mission Along with partners at the Jet Propulsion Laboratory, United planners and technology Space Alliance, and Payload Systems (now Aurora Flight Sciences), developers the information they our research team examined the problem of supply chain manageneed to design the exploration ment in space. The goal was to develop an integrated logistics and campaigns of the future. supply chain modeling capability to guide the development of an interplanetary supply chain. This involved examining terrestrial analogs, building probabilistic models of supply and demand, developing a flexible modeling and simulation tool to examine the supply chain network and its performance in different mission applications, and conducting trade studies to examine the impact of different architectural decisions on the cost-efficiency and reliability of the resulting supply chain network.

The challenge of living on Mars

15


MODELING AND SIMULATING MISSION LOGISTICS One product of our effort is SpaceNet, a software package that allows decision-makers to model and simulate the logistics of exploration missions and campaigns in order to determine the feasibility of different mission plans and generate quantitative metrics for use in trade studies. These metrics include measures of productivity, such as the crew-days at the destination, and mass of exploration and research equipment delivered, as well as measures of efficiency such as cargo capacity utilization. SpaceNet also calculates the cost and launch mass of a given mission or campaign. These metrics provide a quantitative basis for the comparison of different alternatives. In addition, SpaceNet provides the capability to optimize a scenario in terms of one or more of these objectives, as well as to visualize the flow of elements and resources through the supply chain network over time. As a result, SpaceNet provides mission planners with a powerful and flexible tool to examine supply chain options and explore the impact of logistics requirements throughout the mission SpaceNet provides mission design process.

planners with a powerful and flexible tool to examine supply chain options and explore the impact of logistics requirements

SpaceNet addresses the problem of how best to set up an interplanetary supply chain network for a given mission context. Another side of the logistics problem is demand; how much material does that supply chain need to provide, and when and where does it need to be provided in the Earth-moon-Mars system? For human exploration missions, one of the major drivers of logistics requirements is the mass associated with environmental control and life support, or ECLS. Humans have adapted very well to our home environment on Earth, but when we travel into space we must artificially recreate this environment. ECLS systems handle the major functions required to keep humans alive: controlling pressure and temperature, providing oxygen and removing carbon dioxide, supplying water and food, and removing waste. These functions require careful management of resources such as oxygen, nitrogen, water, food, and waste products. In addition, spare

16

AEROASTRO 2014-2015


parts are required to maintain the systems that control, process, and recycle these consumables. In order to develop a detailed and flexible model of the resource requirements associated with ECLS, as well as enable the concurrent modeling and optimization of habitation and logistics systems, we started the HabNet project in 2012. The goal of this ongoing research is to complement SpaceNet capabiliProposed new technology, ties by developing a similar modeling and simulation tool to such as 3D printers, can be examine the habitation problem: For a given mission profile implemented notionally in and selection of ECLS technologies, what are the resource requirements to sustain the crew? the HabNet framework HabNet simulates the function of different elements within a spacecraft or surface habitat along with crew activities, including sleeping, exercise, and extravehicular activity, while monitoring environmental conditions and resource consumption within the habitat. The results of these simulations indicate whether or not a given ECLS architecture is feasible and, if it is not, it helps designers determine what changes need to be implemented to make it feasible. If ISRU technology is to be used to produce resources on site, the simulation can output production rate requirements. In addition, HabNet utilizes reliability and maintenance data for components within the system to determine how many spares are required to keep it running for the mission duration, and how this spares requirement can be traded with the risk that not enough spares will be supplied. HabNet allows users to conduct trade studies on different habitat options and mission profiles by varying ECLS and ISRU technology selection, crew size, operations profile, and duration. More complex ECLS systems that incorporate technology to recycle waste products into useful consumables can reduce the resupply requirements for these consumables, but they often require more spare parts to maintain. HabNet can examine both of these impacts, as well as how changes in reliability and recycling efficiency affect the results. Proposed new technology, such as 3D printers that can produce spare parts and other useful items, can be implemented notionally in the HabNet framework in

The challenge of living on Mars

17


NASA’s Habitat Demonstration Unit, part of an effort to develop sustainable living quarters, workspaces, and laboratories for next-generation space missions, undergoes desert testing in Arizona in 2010. (NASA)

order to examine how a new capability impacts resource demands. In effect, HabNet allows system designers and mission planners to determine the resource demand associated with a system, and SpaceNet examines the supply chain required to fulfill that demand.

AN HISTORIC CHALLENGE The human exploration of Mars presents one of the greatest logistical challenges in history. Modeling and simulation tools like SpaceNet and HabNet give mission planners a holistic view of the logistics of deep space missions, including both the habitat that sustains the crew and the supply chain that sustains the habitat, in order to facilitate optimization at the system, mission,

18

AEROASTRO 2014-2015


and campaign level. In addition, they allow for evaluation of new technology in the context of the greater mission plan to inform research and development investment. These capabilities support the larger human spaceflight effort, and AeroAstro researchers find it exciting to be a part of this endeavor, helping to inform the development of cost-effective systems and mission plans to accomplish the exploration goals of the future and rise to the challenge of putting humans on Mars. ANDREW OWENS (SM ’14) is a PhD candidate and NASA Space Technology Research Fellow concentrating on risk and supportability analysis for long duration human spaceflight and how new technology can make future missions safer and more efficient. He may be reached at acowens@mit.edu. OLIVIER DE WECK (SM ’99, PhD ’01) is an MIT Professor of Aeronautics and Astronautics and Engineering Systems. His research focuses on systems engineering with a focus on lifecycle properties, integrated modeling and simulation, and multidisciplinary design optimization with applications to space systems, manufacturing and production, infrastructure systems, and city design. He may be reached at deweck@mit.edu.

The challenge of living on Mars

19


Surrounding an autonomous mini race car used in the 2015 Independent Activities Period embedded systems class are instructors (from left) Michael Park, Sertac Karaman, Michael Boulet, and Owen Guldner, and winning race team members Guy Rosman (postdoc), John Alora (G), Valerio Varricchio (G), and Abhishek Agarwal (G). Not pictured is team member Ostin Zarse (sophomore). (William Litant/MIT)

20

AEROASTRO 2014-2015


EDUCATION

“Autos” racing beneath MIT teach embedded system basics by Sertac Karaman

Embedded systems are dedicated computers that monitor and control mechanical/electrical systems in the physical world.

They are an essential component of virtually all aerospace vehicles. On one end of the spectrum, we might find palm-size drones with tiny cameras recording our latest hike. They use embedded systems to monitor their orientation and command their motors every few milliseconds, so that the drones can stay stable in the air. On the other end of the spectrum lie the large satellites with mega telescopes, searching for planets outside this solar system. Here, embedded systems carefully position the satellites towards a target to acquire the best image. From applications that are small and inexpensive, to those that are large and expensive, embedded systems are critical components. Embedded systems differ from more familiar computers, like your laptop computer, in a number of ways. Firstly, embedded systems are built for a specific purpose; hence they are referred to as “dedicated.” They use all their resources for the task at hand, as opposed to your laptop, which is a “multipurpose” and “multitasking” system. Therefore, embedded systems are able to attend to monitoring and control tasks around the clock. In technical terms, they provide “real-time guarantees.” For example, they guarantee that your mini-drone’s motors receive commands every few milliseconds. Most laptop computers provide no such guarantees, and an app can commonly take seconds or minutes to boot up, or may even fail to start altogether. Because embedded systems provide these real-time operation guarantees, they can monitor and control systems that operate in the physical world, including aircraft, spacecraft, cars, and even home appliances.

“Autos” racing beneath MIT teach embedded system basics

21


PREPARING FOR AEROSPACE EMBEDDED SYSTEMS

During three-day hackathon, students worked in AeroAstro’s Neumann Hanger developing algorithms to autonomously navigate their team’s race car through MIT’s tunnel network. (Michael Park)

As vehicles assume more autonomy (independence from human piloting), the embedded systems that govern them have become increasingly complex. As a result, in many current aerospace projects, substantial development effort is devoted to the embedded systems component. This trend is likely to intensify in the future. Given the integral role played by these components, we in the MIT Aeronautics and Astronautics Department are committed to preparing the next generation of aerospace engineers to understand, develop, and apply aerospace vehicle embedded systems.

During MIT’s January 2015 Independent Activities Period (IAP), we offered a three-week short course introducing the basic concepts of programming embedded systems, culminating in an autonomous mini-racecar competition through the tunnel network that interconnect MIT buildings. The course was held under AeroAstro’s auspices, and developed and taught in collaboration with Beaver Works, the joint MIT Lincoln Laboratory and MIT School of Engineering hands-on project-based learning initiative. The instructors were Michael Boulet, Owen Guldner, and Michael Park of Lincoln Laboratory, and Professor Sertac Karaman of MIT AeroAstro. The course included special lectures focusing on the operation of autonomous vehicles. For instance, the students learned the basics of how autonomous vehicles find their location on a map through sensory measurements, and then chart paths to avoid collisions through complex environments. In order to teach the embedded systems concepts, the course championed “learning by doing.” The students learned how to program using the Robot Operating System (ROS) environment. ROS provides an environment for software developers to implement various apps for robotics, and

22

AEROASTRO 2014-2015


an environment for those apps to run in harmony. In this case, the apps include those that accomplish location-finding or path-planning tasks. During the first week, the students were taught how to navigate in ROS and how to use the ROS environment to implement and test software.

RACE CARS AND THE INVERTED LABORATORY

The Beaver Works Initiative.

Starting from the second week of the course, the 20 students were split into four groups. Each group was given a mini race car, equipped with the latest robotics technology. An important part of the hardware was the Jetson, referred to by manufacturer Nvidia as the “first embedded supercomputer.” With 192 computational cores, the Jetson provides roughly 10 times the computational power that is available on standard laptops today, for only 1/10th of the price. The Jetson ships for only $192. We made the Jetson available to our students only six months after it was commercially available.

AeroAstro/Lincoln Lab collaboration through the Beaver Works Initiative

We planned an “inverted laboratory” experience with the race cars. More specifically, the “labs” were done at home, and projects were worked out in class. To enable this experience, the students were allowed to take their race cars home. This inverted lab is in contrast to most laboratory courses at MIT in which students can work with the experimental platform only in the laboratory and only during the lab hours. Our approach allowed the students to work with the platform on their own time. In particular, they could

Like the embedded systems Independent Activities Period course, AeroAstro’s capstone project classes have been planned and executed with input from the Lincoln Laboratory through the Beaver Works initiative. For example, for the 2014-2015 capstone project, the students designed, built, and demonstrated an autonomous seaplane that can land on water, and autonomously dock to a submergible charging station. Collaborations such as these allow the AeroAstro students to work on real projects and interact with Lincoln Laboratory’s experienced staff. For more on Beaver Works, see “AeroAstro students design, build novel vehicles for real-world customers,” by John Hansman in AeroAstro issue 9, 2011-2012, http://issuu.com/mitaeroastro/docs/aeroastro-2011-12/18 and visit http://engineering.mit.edu/programs/beaverworks

The Beaver Works concept, named after MIT’s beaver mascot, originated when Lincoln Laboratory approached AeroAstro about increasing interaction and collaboration among the MIT Main Campus, Lincoln Laboratory, and the U.S. Air Force. The original idea was to commission students in the aircraft systems design classes to design, build, test, and deliver actual flight vehicles, which would be used to support Air Force testing. Beaver Works brings MIT students, faculty, and researchers together with Lincoln Laboratory staff to create innovative design/build projects across a broad range of engineering disciplines. Projects are sized to allow participants to work closely together developing solutions that can be implemented on a large scale by using support and resources provided by the Laboratory. Beaver Works operations will have a home in Building 31 when renovations are completed in 2017.

“Autos” racing beneath MIT teach embedded system basics

23


The mini race car with which the students were provided featured a laser range finder, camera, optical flow sensors, and an inertial measurement unit. Computation was handled by an onboard Nvidia Jetson supercomputer. (Michael Park)

work through the examples and exercises that were explained in documents we made available online. If they students had any questions, they could share them with the instructors and fellow students via email. The questions were answered rapidly, even after-hours, either by the instructors or by fellow students. From a pedagogical point of view, these home exercises replaced the traditional labs. We used traditional lab hours for small projects, allowing students to work together and learn from each other. Roughly one-and-a-half weeks of the course were devoted to the special lectures, home exercises, and in-class projects.

24

AEROASTRO 2014-2015


During the final days of the class, the students applied what they learned to a competition. In a three-day hackathon, each team designed and implemented algorithms that could autonomously navigate its race car through a small section of MIT’s tunnels. The software ran on the Jetson boards, and no external input to the cars was allowed during the race. On the day of the race, the teams ran their cars through the racecourse one by one. The fastest team was declared the winner. For safety reasons, the race cars’ speed was limited to approximately 8 mph, roughly the speed at which team members could run behind them. The winning team’s software allowed its racecar to complete flawlessly the 515-foot course in about 49 seconds at an average speed of 7.1 mph. Videos of the competition are available at http://racecar.mit.edu. We conducted a survey among the participating students. They gave a rating of 4.6 out of a possible 5 for the statement “I enjoyed the class.” While “I knew a lot of applicable material prior to the class” got a 2.4/5 rating, the students rated 4.3/5 for “I learned a lot of new material.” We instructors believe the inclusion of an exciting experimental platform and a competition had much to do with the students’ enjoyment. We also surmise that our approach to the laboratory experience allowed the students to learn a large amount of material in a short time. A second round of the course is planned for January 2016. We are considering introducing palm-size mini-drone platforms as our test vehicles. We hope that, in the future, the experimental platforms, the course material, and the pedagogy will be adopted by undergraduate classes throughout the core professional-area subjects and the capstones, thus preparing MIT AeroAstro students for a world where embedded systems are ever more prevalent. SERTAC KARAMAN is the Charles Stark Draper Assistant Professor of Aeronautics and Astronautics. His research interests lie in the broad areas of robotics and control theory. He studies the applications of probability theory, stochastic processes, stochastic geometry, formal methods, and optimization for the design and analysis of high-performance cyber-physical systems. Sertac Karaman may be reached at sertac@mit.edu.

“Autos” racing beneath MIT teach embedded system basics

25


Computational simulations offer unprecedented means of predicting diverse physical behaviors, ranging from the performance of aircraft systems still in their design stage, to the impact of Antarctic ice sheets on sea level rise. But how confident can we be in these predictions? (NASA)

26

AEROASTRO 2014-2015


Quantifying uncertainty in the physical world By Youssef Marzouk

Alongside theory and experiment, computational simulation has become an essential tool for understanding and predicting the behavior of physical systems.

Many such systems are inaccessible to complete observation or empirical testing. In this setting, computational simulation lets engineers and scientists fill in the gaps, try new scenarios, and discover new behavior. Examples range from future aircraft systems that exist, in their entirety, only on the drawing board, to ice sheets in Antarctica — whose behavior is critical to predictions of future sea level rise, but depends on melting, sliding, and deformation processes that cannot be directly observed. In these settings, the ability to make computational predictions gives engineers the ability to guide the design process (i.e., to perform design optimization) and can even give policymakers the ability to make science-informed decisions about global issues, such as climate change and air pollution. But computational predictions are hardly foolproof, nor do they exist in a vacuum. On one hand, every computational simulation contains many sources of uncertainty: incomplete or unresolved physics, unknown or intrinsically variable parameters, initial conditions or boundary conditions that may not match reality. One would like to assess the impact of all these uncertainties on the computational prediction of any quantity of interest, that is, to quantify uncertainty in predictions. On the other hand, we often can make limited comparisons between computational simulations and experimental observations, or otherwise collect data that are indirectly related to uncertain aspects of a computational model. Thus, data can be used to inform or refine computational models, and to help achieve better predictions than can be produced by simulation or empirical data analysis alone.

Quantifying uncertainty in the physical world

27


The past decade has seen a surge of research activity focused on these issues. MIT AeroAstro researchers are making significant contributions in this regard, advancing the theory and methodology of uncertainty quantification and demonstrating its applicability to vital problems. This endeavor is strongly interdisciplinary, lying at the intersection of modeling (in the relevant engineering and science disciplines), applied mathematics, scientific computing, and statistics. In the following, we describe several highlights from our recent efforts.

INFERENCE WITH COMPUTATIONALLY INTENSIVE MODELS Mathematical models of physical systems often contain parameters that must be estimated from indirect and noisy observations. But such observations can seldom constrain the model parameters precisely; as a result, it is important to produce not only a best estimate of the parameters, but also a quantitative assessment of their uncertainty. This problem can be addressed from the perspective of statistical inference. The Bayesian statistical approach provides a natural route to quantifying uncertainty by characterizing the posterior probability distribution of the parameters. But, extracting information from the posterior distribution can be challenging: the usual approaches, which involve sampling, require many thousands or even millions of evaluations of the underlying mathematical model. When the model represents a complex physical system — for instance, a set of partial differential equations describing fluid flow — direct sampling is completely intractable.

Schematic of the approach for constructing approximations of computationally intensive models, for the purpose of Bayesian parameter inference. Blue dots represent evaluations of the expensive physical model and the green contours mark the probability distribution we would like to explore. Approximations are constructed on-demand using the model evaluations falling within each red circle, and improve in quality as the blue dots become denser and the red circles shrink.

28

AEROASTRO 2014-2015

An important line of research seeks to replace such computationally intensive models with principled approximations, often called “surrogates,” that are much faster to evaluate, but that can guarantee accurate parameter inference. A host of approximation methods have been employed in this regard: polynomial or Gaussian process approximations of the input-output map induced by the model, projectionbased reduced order models, coarse discretizations, simplified physics, and more. Several years ago our group at MIT showed that if a surrogate can be refined so that its error vanishes at a certain rate, then errors in the inference process decay with at least the same rate, and, hence, that many computational surrogates widely used


for uncertainty propagation could immediately be used for parameter inference too. But these approaches still require some potentially difficult a priori decisions: since approximations are constructed before sampling, how accurate an approximation should one make in order to control the bias in subsequent parameter estimates? Also, since the data provide new information about the parameters, constructing a surrogate offline, before this information is available, can waste significant computational effort. For example, why make an accurate surrogate in regions of the parameter space that the data exclude? Our most recent work overcomes these difficulties by developing surrogates on the fly during posterior sampling. These surrogates employ approximations based on local (i.e., at nearby parameter values) evaluations of the expensive physical model. As sampling continues, these evaluations become more densely spaced in the important regions of the parameter space. We have shown that the resulting scheme guarantees convergence to the probability distribution that would have been induced by the exact physical model, but at a fraction of the computational cost. And, it is automatically posterior-focused; in applications where the data are informative, this is particularly important because the posterior probability distribution can concentrate with respect to the prior distribution by many orders of magnitude. We have applied this inference scheme to many problems, including models of ice-ocean interaction at Pine Island Glacier in the western Antarctic. Here, surrogates render an essentially intractable inference problem tractable: we are able to use temperature and salinity measurements to learn about parameters governing melting and mixing processes at the interface between the glacier and the ocean. Compared to a direct sampling approach, many orders of magnitude fewer model evaluations are required to achieve a given level of accuracy. Our methods also extend very naturally to large-scale parallel computing, where multiple “chains� of samples contribute to a common pool of model evaluations and thus borrow strength from one another. Parallel chains do more than simply produce a larger number of samples; each chain converges more quickly than it would on its own.

Quantifying uncertainty in the physical world

29


INVERSE PROBLEMS IN HIGH DIMENSIONS

1000 900

36ʼ

800 700

48ʼ

600 500

o 75 S

400 300

12ʼ

200 100

24ʼ o 102 W

30ʼ

o 101 W

30ʼ

100oW

AEROASTRO 2014-2015

Our approaches involve analyzing the parameter dependence of the mismatch between data and model predictions in linear and nonlinear settings, and developing transformations that exploit any low-rank structure and/or sparsity that may be present. With these tools, we are able to tackle inverse problems of a size that seemed intractable only a few years ago. Examples include atmospheric remote sensing, where we infer gas concentration profiles from star occultation measurements. The parameter dimension in this problem is in the thousands and the data number in the tens of thousands, but the intrinsic dimension is only 20.

30ʼ

A satellite image (top) of Pine Island Glacier in western Antarctica and (bottom) of the computational domain used to model ice-ocean interactions in the Pine Island system. The locations of temperature and salinity observations are indicated by colored circles; from these observations, researchers infer parameters governing transport and mixing at the ice-ocean interface.

30

In many modeling scenarios, the parameters we wish to infer from data are actually spatially distributed quantities; for example, profiles of gas concentration in the atmosphere, or permeability fields governing contaminant transport in the subsurface. Discretization immediately makes these quantities very high dimensional. High-dimensional data arise frequently as well, from satellite observations and other information sources. Even if the data are “big” in this sense, the information they carry is not necessarily big. In fact, when the data are indirect, as is often the case in inverse problems, there many be many aspects of the parameters that the data cannot constrain or inform. The intrinsic dimension of an inverse problem can thus be much smaller than the discretization or data dimensions and, in many cases, independent of both. Many of our recent research efforts have focused on how to discover such low-dimensional structure in inverse problems, and how to ensure that all subsequent calculations have a cost that scales with this dimension.

More broadly, this effort points to the challenge of understanding the conditional independence structure of complex physical models. Conditional independence is central to many models and inference methodologies used in other fields, such as machine learning. But conditional independence is far less obvious in the nonlinear partial differential equation models that we work with — naively, all variables of interest can seem strongly coupled — and transformations of the parameter space are necessary to uncover it.


Complementing our work in sampling methods for high-dimensional inverse problems are new methods that dispense with sampling altogether. In this work, we are establishing links between statistical inference and optimal transportation, a rich area of mathematics that describes procedures for transporting mass from one place to another. Replace “mass” with “probability” to obtain a new perspective on the inference problem: we perform inference by constructing mappings that transform the prior probability distribution on the parameters, which describes our state of knowledge before seeing the data, to the posterior distribution, which is the solution of the inference problem. This approach has a number of advantages, including the ability to generate independent samples at low cost. Building maps in high dimensions depends crucially on understanding the structure of the inference problem, however, and thus makes full use of the dimension reduction ideas discussed above.

OPTIMAL EXPERIMENTAL DESIGN So far, we have discussed how to infer various aspects of physical models from experimental data. This process naturally reduces uncertainty and improves the accuracy of model-based predictions. But not all experimental data are equally useful. At the same time, many experimental observations can be difficult, time-consuming, and expensive to acquire. Maximizing the value of experimental observations (i.e., designing experiments to be optimal by some appropriate measure) is therefore a critical task. Experimental design encompasses questions of where and when to measure, which variables to interrogate, and what experimental conditions to employ.

⇡(✓)

T (✓)

We have been developing model-based approaches to optimal experimental design, which guide the choice of experiments for a particular purpose, such as parameter inference, prediction, or model discrimination. These methods quantify information gain, in a particular quantity of interest, due to any candidate

⇡ ˜ (r)

p(r)

T,T −1

⇡(✓)

T˜(✓) Transport maps can transform simple Gaussian distributions into arbitrarily more complicated probability distributions in order to solve high-dimensional inference problems.

⇡ ˜ (r)

Quantifying uncertainty in the physical world

31


0.09 3

0.08 0.07

2

0.06 2

0.05 0.04

1.5

U (1 , 2 )

2.5

experiment. We must consider expected information gain because the decision of what experiment to perform must be made before the experimental data can be acquired; thus, we consider all possible outcomes of the experiment and find the experiment that is best on average. In this sense, experimental design is an “outer loop� around the inference methodologies described earlier and is a computationally challenging undertaking.

Our approaches to this problem have employed reducedorder or surrogate models that can accelerate the estimation 0.02 of information gain, coupled with efficient optimization 0.5 0.01 methods for stochastic or noisy objectives, since expected 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 1 utilities are typically evaluated with Monte Carlo methods. Recently we have been developing fast sample-based estimators of information gain in targeted subsets of parameters, An example of the optimal experimental design approach, applied to a problem in materials science, the image depicts a map of applicable in general nonlinear and non-Gaussian settings; information gain in the length scale of a spatially heterogeneous these estimators are useful when only one or a few paramsubstrate, based on observations of the spinodal decomposition of a thin film deposited thereon. The optimal combination of experiments eters are of interest to the experimenter, and the other is on the top left, while least informative experiments are in the uncertain parameters are simply nuisances. We have also middle of the map. been developing design techniques that help scientists and engineers choose between competing models for the same physical process. After any experiment, each model is assigned a probability that reflects how well it is supported by the data. One would then like to find an experiment that makes this distribution strongly peaked on one model or another; this is the most decisive experiment, in a quantifiable information theoretic sense. 0.03

1

SOFTWARE TOOLS AND OUTREACH While our research has focused on the development of new computational approaches for uncertainty quantification, applied to a broad range of problems in engineering and science, an essential next step is to make these tools easily accessible to others. We wish to place our tools not only in the hands of other computational scientists, but also engineers and scientists who need to quantify

32

AEROASTRO 2014-2015


uncertainty in their own physical models. Accordingly, we have been developing the MUQ (MIT Uncertainty Quantification) library, a C++ and Python framework for performing uncertainty quantification with complex computational simulations. MUQ has been released under an opensource license via MIT’s Technology Licensing office, and has been finding its way into several new application areas such as ocean modeling, glaciology, subsurface modeling, chemical kinetics, and materials science. We have also been using MUQ to facilitate education and outreach. Last January, the lead developers of MUQ (all MIT Aerospace Computational Design Laboratory postdocs and graduate students) led a week-long MUQ tutorial for other Aero-Astro researchers. This summer, we took MUQ on the road and used it in uncertainty quantification summer schools at the NSF-sponsored Institute for Mathematics and its Applications (IMA, at the University of Minnesota) and Institute for Computational and Experimental Research in Mathematics (ICERM, at Brown University). MUQ has also been adopted by the Department of Energy’s SciDAC (Scientific Discovery for Advanced Computing) Institute on uncertainty quantification. Working with other researchers who use MUQ facilitates a vital interaction between algorithms and applications, which will likely lead our research in many new and unexpected directions. The author thanks his students, postdocs, and many other collaborators at MIT and elsewhere, who have made possible the work described here. YOUSSEF MARZOUK is an Associate Professor of Aeronautics and Astronautics at MIT and director of MIT’s Aerospace Computational Design Laboratory. He is also co-director of educational programs for the MIT Center for Computational Engineering. His research focuses on uncertainty quantification, inverse problems, statistical inference, and Bayesian computation for complex physical systems, and on using these approaches to address modeling challenges in energy conversion and environmental applications. For more information about his work, see http://uqgroup.mit.edu. Youssef Marzouk can be reached at ymarz@mit.edu.

Quantifying uncertainty in the physical world

33


The three-year, $52 million renovation of MIT Building 31 will include a highbay space between the two wings where AeroAstro researchers can test-fly unmanned aerial vehicles. (Imai Keller Moore)

34

AEROASTRO 2014-2015


Building 31 renovations to transform AeroAstro’s research space By Mark Veligor

MIT’s Building 31 is a strange conglomeration of quirky spaces and huge antiquated machinery.

Throughout its labyrinthine interior are immense electrical motors, ancient compressors, vintage machine tools, mazes of pipes, and towering control panels. In fact, several years ago Pixar Animation Studios sent a team of artists to sketch its interiors as inspiration for retro-fantasy video games. Blocks of water-stained wood support dented air conditioners, there’s no handicapped access to upper floors, and a recent visitor was warned not to open a second-story window because, “The last time we did that, it fell out and nearly hit someone on the sidewalk.” This is about to change. With the support of generous alumni and industry partners, and a commitment by MIT to tackle its aging infrastructure, Building 31, officially the “Sloan Laboratories for Aircraft and Automotive Engines,” is undergoing a three-year $47 million transformation into a state-of-the-art home for the Gas Turbine Laboratory, an autonomous systems research center, new offices for faculty and staff, and expansive workspaces for researchers and students in AeroAstro and Mechanical Engineering.

Building 31 renovation to transform AeroAstro’s research space

35


FOR BETTER ENGINES Funded by a gift from General Motors CEO and MIT alumnus Alfred P. Sloan Jr. (1895), Building 31 was opened in late 1928 as a single-story home for MIT’s internal combustion engine research. Shared by the Department of Mechanical Engineering and its Aeronautics Course 16 (Aeronautics did not become a distinct department until 1939), the Sloan Laboratories were outfitted with a Wright E4 aircraft engine, a Chrysler automobile engine, a small wind tunnel, and a photographic darkroom for processing spectrogram and oscillograph recordings.

The Sloan Laboratories under construction in late 1928. In following decades, east and west wings were added to complete the current Building 31. (George Davis/MIT Museum)

With increased demand for work on national defense, a two-story east wing was added in 1940, also funded by Sloan, to relieve congestion on the testing floor and relocate office, lecture, and machine shop space. MIT’s increased contribution to the war effort led to another addition in 1944, a three-story west wing. It became home to the Gas Turbine Laboratory.

Over the decades, many key research advances have been made within the walls of Building 31 in such fields as engine design, advanced propulsion systems, and fluid mechanics. In recent years, the building has supported the activities of approximately 150 AeroAstro and MechE faculty, students, and researchers. Despite the important work done there, Building 31 has been slated for demolition numerous times over the last 50 years as the building and equipment became increasingly outmoded. For example, in the 1960s, when Building 37 was built just a few feet from Building 31, plans called for replacing 31 with a courtyard, but that part of the campus plan was never implemented.

36

AEROASTRO 2014-2015


NEW HOME FOR AUTONOMOUS SYSTEMS One of the most notable features of the renovated Building 31 will be a new high-bay space positioned between the two wings. This 24-foot-high room will allow researchers to test-fly aerial vehicles. The new facility and its corresponding space for students and faculty were made possible by a generous pledge of support from Course 16 alumnus Kent Kresa (‘59, SM ’61, ENG/EAA ’66), former chairman and CEO of Northrop Grumman and General Motors, and current Lincoln Laboratory Advisory Board chair. A second space, funded by Northrop Grumman, will be available for FAA-compliant outdoor flight. This area, which will face the Great Dome and the new MIT.nano facility, will double as a venue for outdoor events. Thanks to a gift from Course 6 (Electrical Engineering and Computer Science) and Course 16 alumnus Paul AeroAstro alumnus and technologist Paul Kaminski (SM’66) and his wife Julia Kaminski (SM ’66) and his wife Julia, the renewal of have contributed to the Building 31 renovation to “enable the development of our future best and brightest in aerospace and defense technology.” Building 31 also will allow Assistant Professor Julie Shah to bring her Interactive Robotics Group together under one roof. Due to space constraints, her team, which works on allowing humans and robots to work collaboratively on manufacturing and disaster relief efforts, has been split between two campus locations. Paul Kaminski has won numerous awards, including the 2006 National Medal of Technology, after a decorated career in the public and private sector. His esteemed resume features a 20-year career in the Air Force, a 1994-1997 stint as the Under Secretary of Defense, and time as the chairman of the RAND Corporation and CEO and president of Technovation Inc.

Building 31 renovation to transform AeroAstro’s research space

37


Some of Building 31’s compressor and electric motor/generator equipment used for many years by the Gas Turbine Lab was salvaged from the World War II submarine USS Halibut (SS-232) following a 1945 attack by Japanese aircraft that left the sub unusable. (William Litant/MIT)

When asked about why they support the project, the Kaminskis replied, “We decided to donate to the building project not as an end, but as a means. This building, the labs, and study areas will enable the development of our future ‘best and brightest’ in aerospace and defense. A key technology — autonomy — that will be pursued in this building will be an important enabler of future capabilities in aeronautics and astronautics.”

FUTURE OF POWER AND PROPULSION The 30 members of the Gas Turbine Lab also stand to benefit from the renovation. GTL research is at the cutting edge of advancing fluid machinery for propulsion systems, energy conversion, and power. (See “Advanced Turbomachinery Design for Supercritical Carbon Dioxide Applications” elsewhere in this issue.) The group is involved in multiple research “This building ... will enable projects such as developing novel aircraft to increase efficiencies and decrease aviation’s impact on the environment. A prime the development of our example is the recent D8 “double bubble” aircraft concept which future ‘best and brightest’ in could reduce fuel consumption compared to current commercial aerospace and defense.” aircraft by as much as 70 percent.

38

AEROASTRO 2014-2015


AeroAstro Professor and Gas Turbine Laboratory Director Zoltán Spakovszky said, “This Building 31 project will provide the GTL with the space we need for the next stage of our work, maintaining and expanding our research capabilities and collaborations, while providing better work conditions for our faculty and students.” AeroAstro will continue to share Building 31 with Mechanical Engineering. The Sloan Automotive Laboratory, led by Professor Wai Cheng works to radically improve transportation systems for the one billion automobiles on the road and the Electrochemical Energy Lab, headed up by Professor Yang Shao-Horn, researches novel solutions to energy storage and battery technology challenges.

The case for the project was bolstered by AeroAstro Visiting Committee recommendations.

TOWARD MIT 2030 In addition to the support of many generous alumni, this $47M project was made possible by a commitment from MIT administration to invest in the “accelerated capital renewal” of spaces critical to advancing MIT’s mission. These investments are part of the MIT 2030 plan (http://mit. edu/mit2030): a flexible framework designed to guide MIT’s senior leadership in the evolution of the campus and the surrounding area over the coming decades. A $750M bond issued by the Institute in 2011, when MIT capitalized on historically low rates, fuels this plan. Due to the hefty deferred maintenance backlog in Building 31 and the mission-critical work done in its laboratories, the project was seen as an appropriate opportunity for these accelerated capital renewal funds. The case for the project was bolstered by AeroAstro Visiting Committee recommendations for physical space upgrades.

Building 31 renovation to transform AeroAstro’s research space

39


Building 31’s west wing was added in the 1940s and became home to the Gas Turbine Lab. (William Litant/MIT)

“It’s very gratifying that Department Head Jaime Peraire, working with the MIT Administration, was able to make the case and put together the funding necessary for the renovation of Building 31,” said Visiting Committee chairman Mark Gorenberg (EECS ’76). “Many of our Visiting Committee members were so moved, that they contributed philanthropically to the project as well. We can’t wait to see the results.” Gorenberg pledged a gift to the renovation, as did fellow visiting committee members Art Samberg (AeroAstro ’62) and Dan Schwinn (EECS ’83).

40

AEROASTRO 2014-2015


A VERY HAPPY 89TH BIRTHDAY The architects on the project are Imai Keller Moore Architects of Watertown, Mass., with Barr & Barr of Framingham, Mass. providing construction management, and WSP Group of Boston, selected as the primary engineering firm. Construction is expected to be completed in 2017, 89 years after the building was first occupied. With these renovations, Building 31 will be fully accessible to all members of the MIT community. Elevators are planned for each wing and a footbridge to Building 37 will connect 31 to the endless internal network of MIT hallways. There will also be student project workshop space, co-developed with MIT Lincoln Labs, to expand Beaver Works, an MIT/Lincoln joint venture focused on projectbased student learning that addresses real-world problems. The renewal of Building 31 stands as a prime example of what can happen when alumni, researchers, industry, and MIT faculty, staff, and administration come together to enable the Institute’s mission-critical work. The result will be a modern space honoring the building’s strong tradition of utility while bringing a new energy ready to inspire the next generation of engineering leaders. MARK VELIGOR is a development officer with the MIT School of Engineering. He raises philanthropic support for projects within the Aeronautics and Astronautics and Biological Engineering departments. He can be reached at mveligor@mit.edu.

Building 31 renovation to transform AeroAstro’s research space

41


Virgin Galactic president Steve Isakowitz (Aeronautics and Astronautics ’83, SM ’84) with Scaled Composites’ White Knight Two, the aircraft used to lift suborbital spaceplane SpaceShipTwo to launch altitude. Virgin Galactic plans to operate spaceplanes in private passenger service. (Courtesy Virgin Galactic)

42

AEROASTRO 2014-2015


AEROASTRO ALUMNUS INTERVIEW

“The reinvention of space” A conversation with Virgin Galactic president Steve Isakowitz Steve Isakowitz (Aeronautics and Astronautics ’83, SM ’84) is the

government sectors including NASA, where he was deputy associate

president of Virgin Galactic, the spaceflight company within the Virgin

administrator for the Exploration Systems Mission Directorate. At

Group that is developing commercial launch vehicles and aims to

NASA, he played a pivotal role in the return to flight from the Space

provide suborbital spaceflights to space tourists, suborbital launches

Shuttle Columbia accident and architecting a roadmap that laid the

for space science missions, and orbital launches of small satellites.

foundation for human exploration plans upon retirement of the Space

Steve has broad leadership responsibilities across a range of areas including the company’s suborbital spaceflight program, SpaceShipTwo,

Shuttle. For his work at NASA, Steve received the agency’s Outstanding Leadership Medal.

as well as development of its satellite launch offering, LauncherOne.

He was the U.S. Department of Energy’s chief financial officer through

Steve joined the company in September 2011 as the executive vice

two presidential administrations. Other experience includes branch

president and chief technology officer. The company has about

chief of science and space programs at the White House Office of

700 spaceflight participants signed up to fly and dozens of satellite

Management and Budget, an aerospace engineer and project manager

customers for orbital launch with 500 employees located mainly in

for space launch vehicles at Lockheed Martin, and senior consultant

Mojave and Long Beach, California.

in the commercial space division of Booz Allen Hamilton.

Prior to joining Virgin Galactic, Steve held a wide variety of senior

Steve is author of the International Reference Guide to Space

engineering, business, and management roles across the private and

Launch Systems.

Alumnus interview: “The reinvention of space” A conversation with Virgin Galactic president Steve Isakowitz

43


AeroAstro: When did you first know you were interested in aerospace engineering? Isakowitz: My first interest in engineering, particularly aerospace, was sparked by the same event that enticed many of my generation — the Apollo moon program. I remember as a second grader watching those blurry black and white images as Neil Armstrong first stepped on the moon. I thought, how cool is that! I was particularly enthralled watching the football field-sized Saturn V rocket majestically liftoff from the launch pad with such immense power, and with only the thinnest of margins to get it right. I marveled at its success. So, I began to look for whatever I could to learn more about physics and how a rocket worked. Needless to say, there was not too much written for an elementary school kid. Nonetheless, I would try to read whatever I could find and even tried to calculate what it would take to build a rocket, powered by gunpowder, to achieve orbit. Fortunately, I never tried to build it as I learned it wouldn’t work. I learned from this childhood experience how important it is to reach kids when they are young, find avenues for them to explore their passion, and if possible, have an adult who cares enough to take the time from their busy schedule to help keep that passion alive.

44

AEROASTRO 2014-2015

AeroAstro: What did you do while in AeroAstro? Isakowitz: I worked hard to receive both my bachelor’s and master’s degrees in Course 16. I also took advantage of those things that the department offered outside the classroom. I attended outside lectures and seminars, joined student groups such as AIAA, and assisted with research, particularly in the Man Vehicle Lab under Professor Larry Young.

AeroAstro: What things stand out about your time at MIT/AeroAstro? Isakowitz: First, the rich history. We truly stand on the shoulders of giants that came before us, such as Doc Draper and his work on inertial guidance, Robert Seamans on Apollo, Tom Young on Mars Viking Program, James Abrahamson on Space Shuttle, and many more. Indeed, I sat in awe at the AeroAstro Centennial celebration last year listening to the many MIT alum who have flown as Apollo and Shuttle astronauts. An impressive group! Second, the great faculty. It was Professor Jack Kerrebrock who enticed me to come to MIT. Professor Manuel Martinez-Sanchez who helped get my first internship at NASA. And, Professor Walter Hollister who inspired me to get my master’s degree and enabled me to write my thesis while interning away.


Steve Isakowitz examines White Knight Two’s landing gear at Scaled Composites. (Courtesy Virgin Galactic)

Lastly, my fellow students. You don’t realize it while sitting in Unified Engineering, but the students around you who probably look sleepy having stayed up doing problem sets, will be the future titans of industry, world experts in design and technology, and visionaries of things to come. These are people that you want to stay in contact with throughout your career.

AeroAstro: What have you done since leaving MIT? Isakowitz: I have pursued a career that, by choice, covered many aspects of aerospace. I have worked in the private sector, from large established aerospace corporations to small startups. I have also served in the public sector, from agencies in civil space to national security, and policy offices, from the White House to

Alumnus interview: “The reinvention of space” A conversation with Virgin Galactic president Steve Isakowitz

45


Briefing local politicians about Virgin Galactic’s plans, Steve Isakowitz discusses a model of White Knight Two. (Courtesy Virgin Galactic)

Congress. This might seem terribly wide-ranging, and perhaps haphazard. But actually, it has been by design. I learned early on that the space industry is moving along across a multitude of fronts and it takes many players to make it happen. Hence, my diverse experiences have been by design. I enjoy trying new things and seeing the industry from many perspectives — and I believe it has made me better for it. So, what ties this journey together? Well, the common linkage is working on activities that have big societal impact, promise breakthroughs, and represent major innovations in science and technology.

46

AEROASTRO 2014-2015

AeroAstro: Tell us more about your current job. Isakowitz: I am the president at Virgin Galactic, a startup company founded by Richard Branson that promises to open up the space frontier for us all. To date, only about 500 people in the history of the world have experienced space. Our plans are to expand that experience to the thousands — with the ultimate vision that anyone who wants to go, can go. Those few who have flown in space have said it has had a profound impact how they view themselves and the world. Just think: if we all had that chance.


So, we are building a vehicle called SpaceShipTwo, which, once flight testing has been completed in California, will begin spaceline operations from New Mexico. In addition to human space travel, we are building a small air-launched vehicle that will support the emerging revolution in small satellites and place them in orbit much more affordably and rapidly than is possible today. The vehicle, called LauncherOne, is undergoing propulsion and structural ground testing now. We expect both SpaceShipTwo and LauncherOne will be in commercial operations within the next two years.

AeroAstro: What are your favorite aspects of your job? Isakowitz: Being a part of something new — the reinvention of space. Today, there are a myriad of startup companies, backed by visionaries and serious investors, who see a different kind of space program. A space program that no longer depends on the government to make it happen. A space program unencumbered by the slower-pace and risk-aversion of bureaucracies but willing to bet private funds on new innovations. Some will succeed. Some will fail. And this is okay. Competition brings out the best of new ideas while being disciplined by market demand, quality products, and profitable delivery. This commercialdriven space program will not replace what NASA is doing, which is focused on the frontiers of our solar system and universe, but it does promise a change in the economics. Think about it. What if, through these efforts, we could create our own Moore’s Law that predicted remarkable improvements in integrated circuitry but in our case, would see regular, dramatic improvements in the cost and capability of human and robotic space travel.

AeroAstro: How did AeroAstro prepare you for your career? Isakowitz: I learned the importance of teamwork. Oddly enough, this is not something you think about when being admitted to MIT. After all, your admittance is driven by your individual achievement in high school and your standardized scores. So, when you start at MIT, you think this will continue. However, I learned that although individual scores still matter, your ultimate success can depend on your living group, who you make friends with, your study group, and how you can help others. In today’s increasingly complex world, these skills continue to be even more important. Aerospace projects are not a one-person show. It takes a team of people, a spirit of collaboration, and a workplace culture of support that enable these projects to succeed.

AeroAstro: What advice would you offer to high school students about considering engineering careers? Isakowitz: I would suggest keeping an open mind where your career will ultimately take you. I never would have predicted the path I have taken when I was in high school. My path has been driven by the people I talked to, the class subjects I learned, the things I read, and the internships I tried. Although I retained my interest in aerospace throughout most of my journey, it took me on unexpected turns, unlocked new doors and opened up wonderful opportunities. Also, please remember that once you get into your college engineering program of choice, it is not the school that will make you, but what you do when you get there. Before you know it, you’ll be graduating from MIT. Take full advantage of those things around you. You won’t be disappointed!

Alumnus interview: “The reinvention of space” A conversation with Virgin Galactic president Steve Isakowitz

47


AeroAstro: What advice would you offer to AeroAstro students to best position themselves for their careers? Isakowitz: Internship, internship, internship. It’s very easy when studying hard at MIT to pay no attention to preparing for a career. It’s easy to think it will be like high school, namely, just get good grades and solid extracurricular activities and the rest will readily follow. Yes, those things are important. Indeed, achieving high grades is impressive and will improve your chances to get into top graduate programs if you so choose. However, it is not enough. Good job experience can be just as, if not more important. It shows that you have the demonstrated experience and maturity to hit the ground running. I do lots of hiring and seeing “MIT” on the resume is an impressive start, but seeing great experience is what most catches my eye — as it does for many others who recruit. Internships have another benefit that is not as obvious — it will help inform you what career path you might choose. Being at an established company, startup, consulting firm, federal agency, or national lab, you will have the chance to try new things and learn what you like and don’t like. This way when you start interviewing for permanent positions, you will know more definitively what you want, your passions, and where to target your efforts. Plus, you will have internship supervisors who can serve as key references and they may even give you your first job offer! So don’t delay. Work hard and early to find your internship opportunity.

48

AEROASTRO 2014-2015

AeroAstro: How do you perceive current career opportunities for those who graduate with an AeroAstro undergrad degree? Isakowitz: I would love to be a Course 16 student again given all the wonderful opportunities out there — highly autonomous robots exploring the solar system, huge new telescopes peering to the beginning of time, advanced observatories discovering earth-like planets around others stars, low cost rockets being built by startup companies or huge new rockets that will take us to Mars, small satellites connecting the entire globe to the internet, sophisticated sensors enhancing our understanding of earth’s climate, new commercial jets flying safer or someday supersonic, unmanned vehicles promising to revolutionize commerce and national defense, and the list goes on. Moreover, improvements in computer design tools and 3D manufacturing will enable us to build things faster, cheaper, lighter, and more reliably. It truly is the best of times to be an aerospace engineer!

AeroAstro: What do you like to do in your spare time? Isakowitz: What spare time? Seriously, I spend much of my time with my wife and our four fantastic kids. They have always been a major part of my career. When they were young children, I often visited their schools to share the wonders of space travel. To see the students eyes brighten up as I talked to them about voyages to strange planets, stunning pictures of distant galaxies, magical stories about dark matter, or raised questions whether there is life


elsewhere. Also, sharing stories about the pioneers who opened up the frontiers of space as astronauts, engineers, and scientists. I have yet to meet a young child who does not have some curiosity about flight and space. The challenge is to help them channel that wonder into possible careers, or even to keep the fire of curiosity lit for many years to come.

AeroAstro: Anything else you’d like to tell us? Isakowitz: Give back. Remember those who made a difference in your life and those who will in your career. Do the same. You all have something to offer no matter where your path should take you after MIT. Mentor students. Give advice. Volunteer time. Give talks at your local schools. Organize reunions. And visit MIT and thank your professors. …oh, and don’t forget to buy your ticket to space!

Alumnus interview: “The reinvention of space” A conversation with Virgin Galactic president Steve Isakowitz

49


Professor Mark Drela explains D-8 aircraft wind tunnel model features to AeroAstro alumnus Carl Deitrich (’99, SM ’07, PhD ’09) during the Centennial Symposium. Created by an MIT-led team, the commercial aircraft is designed to use 70 percent less fuel than current 737-size planes, with reduced noise and emissions. (William Litant/MIT)

50

AEROASTRO 2014-2015


Celebrating AeroAstro’s 100 years 1914 – 2014 By William Litant

“President Maclaurin stated that he believes (MIT) should lead in the study of aërial navigation in the United States.” Technology Review, July 1909

With that statement, MIT President Richard Maclaurin set the stage for what, in 1914, would become class 13.72 Aeronautics. That same year, an aeronautics master’s degree course was approved. In 1926, Aeronautics became a course (Course 16) in the Mechanical Engineering Department. In 1939, Course 16 became a department unto itself. Throughout 2014 and into 2015, MIT AeroAstro celebrated the centennial of the first aeronautics class, starting with a three-day symposium, October 22-24, 2014. The first day featured a history of aviation and space exploration, and a reception hosted by Draper Laboratory. Day two focused on future of aerospace talks and panels, and a reception and banquet at the John F. Kennedy Library and Museum. The last day covered aerospace education, talks by students, posters, exhibits, and lab tours. On April 8, 2015, the department hosted the AeroAstro Open House. Held in conjunction with the Cambridge Science Festival, hundreds of visitors from MIT, the Cambridge community, and across Massachusetts and the country, partook in a wide range of activities: lab tours, flight simulators, vehicle demonstrations, displays, and hands-on activities for kids including a parachute-build/ egg drop, air-launched rockets, and a paper airplane competition.

Celebrating AeroAstro’s 100 years 1914 – 2014

51


Participating in panels about Apollo-era space exploration were (from left) Professor Jeff Hoffman (co-moderator), Charlie Duke (MS ’64), Phil Chapman (MS ’64), Vance Brand, Mike Collins, Karol Bobko, Rusty Schweickart (’56, MS ’63), Professor Larry Young (co-moderator), and Buzz Aldrin (ScD ’63). (Dominick Reuter)

MIT AERONAUTICS AND ASTRONAUTICS DEPARTMENT CENTENNIAL CELEBRATION | OCTOBER 22-24, 2014

Speakers and panelists

52

NORMAN AUGUSTINE

PATRICK BLONIGAN

KERRI CAHOY

Former CEO and President Lockheed Martin

AeroAstro graduate student | MIT

Assistant Professor of Aeronautics and Astronautics | MIT

BUZZ ALDRIN

Col. USAF (Ret.) Apollo-Soyuz Test Project support crew STS-6, STS 51-D, STS 51-J

KENNETH CAMERON

Col. USAF (Ret.) Gemini 12, Apollo 11 DOMINIC ANTONELLI

VANCE BRAND

CHRISTOPHER CASSIDY

Capt. USN (Ret.) STS-119, 132

Apollo-Soyuz Test Project STS-5, STS 41-B, STS-35

Capt. USN STS-127, Expedition 35/36

AEROASTRO 2014-2015

KAROL “BO” BOBKO

Col. USMC (Ret.) STS-37, 56, 74


MARK CAYLOR

CASEY DENHAM

CHARLES DUKE

President of Enterprise Shared Services Northrop Grumman

AeroAstro undergraduate student MIT

Brig. Gen. USAF (Ret.) Apollo 16

FRANKLIN CHANG-DIAZ

MARK DRELA

ANTONIO ELIAS

STS 61-C, 46, 60, 75, 91, 111

Professor of Aeronautics and Astronautics | MIT

Chief Technical Officer Orbital Sciences

PHILIP CHAPMAN

Col. USAF (Ret.) CATHERINE COLEMAN

Col. USAF (Ret.) STS-73, 93, Expedition 26/27 MICHAEL COLLINS

Maj. Gen. USAF (Ret.) Gemini 10, Apollo 11 EDWARD CRAWLEY

Ford Professor of Engineering, Professor of Aeronautics and Astronautics | MIT TOM CROUCH

Senior Curator, Aeronautics Department National Air and Space Museum WALTER CUNNINGHAM

Col. USMCR (Ret.) Apollo 7 DEBORAH DOUGLAS

Curator of Science and Technology MIT Museum

Professor Kerri Cahoy offers comments during the Centennial Symposium panel “Small Satellites and Access to Space.” In the center is Orbital Sciences Executive VP and Chief Technical Officer Antonio Elias (‘72, SM ’75, PhD ’79) while on the right is Virgin Galactic President Steve Isakowitz (’83, SM ’84). (Dominick Reuter)

JAMES DRAPER

son of Instrumentation Lab Head Charles Stark Draper

Celebrating AeroAstro’s 100 years 1914 – 2014

53


KARL GANTNER

AeroAstro graduate student | MIT GWENDOLYN GETTLIFFE

AeroAstro graduate student | MIT HELEN GREINER

CEO | CyPhy Works JOHN GRUNSFELD

STS-67, 81, 103, 109, 125 CHARLIE GUTHRIE

Senior Vice President and CTO | Insitu R. JOHN HANSMAN

T. Wilson Professor in Aeronautics | MIT KOKI HO

AeroAstro graduate student | MIT JEFFREY HOFFMAN

AeroAstro Professor of the Practice | MIT STS 51-D, 35, 46, 61, 75 A young visitor prepares a rocket she made for air-powered launch during the AeroAstro Open House. The build-and-launch activity was sponsored by the Graduate Association of Aeronautics and Astronautics. (William Litant/MIT)

54

ALAN EPSTEIN

JACK FISCHER

Vice President of Technology and Environment Pratt & Whitney

Col. USAF completed training, qualified for assignment

E. MICHAEL FINCKE

KENNETH GABRIEL

Col. USAF (Ret.) Expedition 9, Expedition 18, STS-134

CEO and President Draper Laboratory

AEROASTRO 2014-2015

BRAD HOLSCHUH

AeroAstro graduate student | MIT JON HOW

Richard Cockburn Maclaurin Professor of Aeronautics and Astronautics | MIT XUN HUAN

AeroAstro graduate student | MIT STEVE ISAKOWITZ

President | Virgin Galactic


(Left) AeroAstro senior Connie Liu discusses aspects of rocket propulsion during her five-minute “Lightning Talk.” Liu and eight other students won a competition to make the mini-presentations at the Symposium. (Dominick Reuter) (Right) Professor and AeroAstro Department Head Jaime Peraire (right) chats with Congressman Joseph P. Kennedy III (D, Mass.) at the Centennial Banquet Kennedy Library reception. Kennedy is a member of the House Science, Space, and Technology Committee, serving on the space and energy subcommittees. (WilliamLitant/MIT)

LUKE JENSEN

CONNIE LIU

DAVID MINDELL

AeroAstro graduate student | MIT

AeroAstro undergraduate student | MIT

JEFF KATZ

ANNE MARINAN

Founding CEO and Chairman | Orbitz Former CEO | Swissair

AeroAstro graduate student | MIT

Frances and David Dibner Professor of the History of Engineering and Manufacturing, Professor of Aeronautics and Astronautics MIT

MICHAEL MASSIMINO

KENT KRESA

STS-109, 125

ELON MUSK

Chairman Emeritus Northrop Grumman Corporation

ROBERT MEYERSON

JOHN LANGFORD

President | Blue Origin

Chairman and CEO Aurora Flight Sciences

ROBERT MILLARD

FUK LI

DAVID MILLER

Director | Mars Exploration Directorate Jet Propulsion Laboratory | NASA

Chairman | MIT Corporation

CEO and co-founder | SpaceX and Tesla DAVA NEWMAN

Professor of Aeronautics and Astronautics and Engineering Systems, Director of Technology and Policy Program, MacVicar Faculty Fellow | MIT

Chief Technologist | NASA

Celebrating AeroAstro’s 100 years 1914 – 2014

55


ROBERT NIEWOEHNER

Capt. USAF (Ret.) David F. Rogers Professor of Aeronautics U.S. Naval Academy JAIME PERAIRE

MIT AeroAstro Department Head and H.N. Slater Professor of Aeronautics and Astronautics RAFAEL REIF

President | MIT ROBIE SAMANTA ROY

Vice President, Technology and Innovation Lockheed Martin DARRYL SARGENT

Vice President for National Security and Space Systems | Draper Laboratory SANJAY SARMA

Director of Digital Learning, Professor of Mechanical Engineering | MIT

(Top)Two Mikes: Astronauts Mike Collins (Gemini 10, Apollo 11) and Edward “Mike” Fincke (’89) (ISS Crew 9, 18; STS-134) meet up at the Centennial Banquet held at the John F. Kennedy Library and Museum in Boston. (William Litant/MIT) (Bottom) Tesla chief executive Elon Musk responds to a question from AeroAstro department head Professor Jaime Peraire during the Symposium session “One-onOne With Elon Musk.” Musk has drawn attention from the media and technology communities throughout the world with his Symposium comment, “I think we should be very careful about artificial intelligence. If I were to guess like what our biggest existential threat is, it’s probably that.” (Dominick Reuter)

56

AEROASTRO 2014-2015


Hundreds of visitors experienced brief “tests” in the Wright Brothers Wind Tunnel during the Open House. Each received a certificate bestowing upon them the title of Distinguished Wind Tunnel Model Subject. (William Litant/MIT)

RUSSELL SCHWEICKART

TONY TAO

LAURENCE R. YOUNG

Apollo 9

AeroAstro graduate student | MIT

NORMAN SEARS

IAN A. WAITZ

AeroAstro Apollo Program Professor Emeritus | MIT

former Assistant Director of the Apollo Program MIT Instrumentation Laboratory

Dean of Engineering, Jerome C. Hunsaker Professor of Aeronautics and Astronautics, MacVicar Faculty Fellow | MIT

PAT SHANAHAN

KAREN WILLCOX

Senior Vice President and GM of Airplane Programs Boeing Commercial Airplanes

Professor of Aeronautics and Astronautics MIT

MARIA ZUBER

Vice President of Research and E.A. Griswold Professor of Geophysics | MIT

Celebrating AeroAstro’s 100 years 1914 – 2014

57


LAB REPORT A 2014-2015 review of Aeronautics and Astronautics Department laboratories Reports provided by the research laboratories and centers

AEROSPACE COMPUTATIONAL DESIGN LABORATORY. . . . . . . . . . . . . . . . . 59 AEROSPACE CONTROLS LABORATORY.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 AEROSPACE ROBOTICS AND EMBEDDED SYSTEMS GROUP.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 THE AUTONOMOUS SYSTEMS LABORATORY. . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 COMMUNICATIONS AND NETWORKING RESEARCH GROUP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 GAS TURBINE LABORATORY.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 INTERNATIONAL CENTER FOR AIR TRANSPORTATION. . . . . . . . . . . . . . . . . 66

In the Space Systems Lab cleanroom, AeroAstro graduate students Pronoy Biswas (left) and Mark Chodas prepare the REgolith X-Ray Imaging Spectrometer (REXIS) radiator for cleaning and multi-layer insulation application. REXIS is an MIT/Harvard student-designed component of NASA’s OSIRIS-Rex asteroid-explorer and sampleretrieving spacecraft, scheduled for launch in 2016. (William Litant/MIT)

LABORATORY FOR AVIATION AND THE ENVIRONMENT. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 LABORATORY FOR INFORMATION AND DECISION SYSTEMS.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 THE LEARNING LABORATORY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 MAN VEHICLE LABORATORY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 NECSTLAB. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 SPACE PROPULSION LABORATORY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 SPACE SYSTEMS LABORATORY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 SPACE TELECOMMUNICATIONS, RADIATION, AND ASTRONOMY LABORATORY.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 SYSTEM ENGINEERING RESEARCH LAB. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 TECHNOLOGY LABORATORY FOR ADVANCED MATERIALS AND STRUCTURES. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 WIRELESS COMMUNICATION AND NETWORK SCIENCES GROUP. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 WRIGHT BROTHERS WIND TUNNEL. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

58

AEROASTRO 2014-2015


AEROSPACE COMPUTATIONAL DESIGN LABORATORY The Aerospace Computational Design Laboratory’s mission is the advancement and application of computational engineering for the design, optimization, and control of aerospace and other complex systems. ACDL research addresses a comprehensive range of topics including advanced computational fluid dynamics and mechanics; uncertainty quantification; data assimilation and statistical inference; surrogate and reduced modeling; and simulation-based design techniques. Advanced simulation methods developed by ACDL researchers facilitate the understanding and prediction of physical phenomena in aerospace systems and other applications. The lab has a long-standing interest in advancement of computational fluid dynamics for complex three-dimensional flows, enabling significant reductions in time from geometry to solution. Specific research interests include aerodynamics, aeroacoustics, flow control, fluid structure interactions, hypersonic flows, high-order methods, multi-level solution techniques, large eddy simulation, and scientific visualization. Research interests also extend to chemical kinetics, transport-chemistry interactions, and other reacting flow phenomena important for energy conversion and propulsion. ACDL’s efforts in uncertainty quantification aim to endow computational predictions with quantitative measures of confidence and reliability, while addressing broad underlying challenges of model validation. Complementary efforts in statistical inference and data assimilation are aimed at estimating and improving physical models and predictions by conditioning on observational data. Current research has developed effective methods for error estimation, solution adaptivity, sensitivity analysis, uncertainty propagation and the solution of stochastic differential equations, the solution

of large-scale statistical inverse problems, nonlinear filtering in partial differential equations, and optimal experimental design. Applications range from aerospace vehicle design to large-scale geophysical problems and subsurface modeling. ACDL research in simulation-based design and control is aimed at developing methods to support better decision-making in aerospace and other complex systems, with application to conceptual, preliminary, and detailed design. Recent efforts yielded effective approaches to PDE-constrained optimization, real time simulation and optimization of systems governed by PDEs, multiscale and multi-fidelity optimization, model order reduction, geometry management, and fidelity management. ACDL applies these methodologies to aircraft design and to the development of tools for assessing aviation environmental impact. ACDL faculty are professors Youssef Marzouk, (director), David Darmofal, Mark Drela, Jaime Peraire, Qiqi Wang, and Karen Willcox. Research staff include Steven Allmaras, Robert Haimes, and Cuong Nguyen. Visit the Aerospace Computational Design Laboratory at http://acdl.mit.edu

AEROSPACE CONTROLS LABORATORY The Aerospace Controls Laboratory researches autonomous systems and control design for aircraft, spacecraft, and ground vehicles. Theoretical research is pursued in such areas as decision making under uncertainty; path planning, activity, and task assignment; mission planning for unmanned aerial vehicles, sensor network design, and robust, adaptive, and nonlinear control. A key aspect of ACL is RAVEN (Real-time indoor Autonomous Ve-

Lab report

59


Aerospace Controls Lab undergraduate Wally Wibowo (left) and graduate student Justin Miller in front of AeroAstro’s Newman Hangar with vehicles ACL has outfitted with sensors of the type used in self-driving cars. This project, part of the Ford-MIT Alliance, aims to predict pedestrian behaviors on short time-scales, while providing data to support a mobility-on-demand system for the MIT campus. (William Litant/MIT)

hicle test ENvironment), a unique experimental facility that uses a motion capture system to enable rapid prototyping of aerobatic flight controllers for helicopters and aircraft, and robust coordination algorithms for multiple vehicles; and ground projection system that enables real time animation of the planning environment, beliefs, uncertainties, intentions of the vehicles, predicted behaviors (e.g., trajectories), and confidence intervals of the learning algorithms. Recent research includes the following: Robust Planning in Uncertain Environments: ACL developed consensusbased bundle algorithm (CBBA) as a distributed task-planning algorithm that provides provably good, conflict-free, approximate solutions for heterogeneous multi-agent missions. Aside from extensions to task time-windows, coupled agent constraints, asynchronous communications, and limited network, CBBA has been validated in real-time flight test experiments. ACL has

60

AEROASTRO 2014-2015

also extended its development of chance-constrained rapidlyexploring random trees (CC-RRT), a robust planning algorithm to identify probabilistically feasible trajectories, to new aerospace domains. For instance, ACL recently developed CC-RRT to solve robust pursuit-evasion problems. ACL is also involved in a multiyear Draper URAD on precision landing of guided parafoils, with robustness to large and dynamic wind environments. Finally, ACL is participating in a multi-year MURI focused on enabling decentralized planning algorithms under uncertainty. Ongoing ACL research has demonstrated that the use of flexible nonparametric Bayesian models for learning models of uncertain environment can greatly improve planning performance. UAV Mission Technologies: ACL has recently demonstrated autonomous, closed-loop UAV flight in MIT’s Wright Brothers’ Wind Tunnel. This novel capability allows the ACL to test flight con-


trollers for windy environments in a controlled and systematic manner. ACL has also developed a novel hovering vehicle concept capable of agile, acrobatic maneuvers in cluttered indoor spaces. The vehicle is a quadrotor whose rotor tilt angles can be actuated, enabling upside-down hovering flight with appropriate control algorithms. Additionally, as part of research on long-duration UAV mission planning, ACL has constructed an autonomous recharge platform, capable of autonomous battery replacement and recharging for small UAVs. This capability allows ACL to demonstrate complex, multi-agent missions lasting for several hours. Information-Gathering Networks: Recent ACL research has addressed maximizing information gathering in complex dynamic environments, through quantifying the value of information and the use of mobile sensing agents. The primary challenge in such planning is the computational complexity, due to both the large size of the information space and the cost of propagating sensing data into the future. ACL researchers created adaptive efficient distributed sensing in which each sensor propagates only high value information, reducing the network load and improving scalability. Recently-developed algorithms embed information planning within RRTs to quickly identify safe information-gathering trajectories for teams of sensing agents, subject to arbitrary constraints and sensor models. Task Identification and Decision-Making: Markov Decision Processes (MDP) and Partially Observable MDPs (POMDP) are natural frameworks for formulating many decision-making problems of interest. ACL has identified approximate solution techniques which can utilize this framework while lessening the curse of dimensionality and the curse of history typically encountered for exact solutions. ACL has also developed a Bayesian Nonparametric Inverse Reinforcement Learning algorithm for identifying

tasks from traces of user behavior. This technique allows a user to “teach� a task to a learning agent through natural demonstrations. ACL has also enabled fast, real-time learning in combination with cooperative planning in uncertain and risky environments, while maintaining probabilistic safety guarantees for the overall system behavior. Finally, by efficiently using potentially inaccurate models of physical systems, ACL has developed a method that minimizes samples needed in real-world learning domains such as a car learning to race around a track. Robust State Estimation: Many navigation and robotic mapping systems are subject to sensor failures and sensor noise that do not match the assumed system models. In many cases, this model mismatch can cause divergence of the state estimates and poor navigation system performance. ACL has developed several robust state estimation algorithms that address these issues by learning a model for the sensor noise while simultaneously generating the navigation solution. These algorithms apply hierarchical and nonparametric Bayesian models along with inference techniques such as Expectation-Maximization and variational inference to learn the noise models. In practice, the robust algorithms provide significantly more accurate solutions while requiring little additional computation relative to non-robust state estimation techniques. ACL has also applied this Bayesian framework to the Simultaneous Localization and Mapping (SLAM) problem to develop algorithms for vision-based SLAM that are robust to landmark misidentifications that cause non-robust SLAM algorithms to fail catastrophically. ACL faculty are professors Jonathan How and Steven Hall. Visit the Aerospace Controls Laboratory at http://acl.mit.edu

Lab report

61


AEROSPACE ROBOTICS AND EMBEDDED SYSTEMS GROUP The Aerospace Robotics and Embedded Systems group’s mission is the development of theoretical foundations and practical algorithms for real-time control of large-scale systems of vehicles and mobile robots. Application examples range from UAVs and autonomous cars, to air traffic control, and urban mobility. The group researches advanced algorithmic approaches to control high-dimensional, fast, and uncertain dynamical systems subject to stringent safety requirements in a rapidly changing environment. An emphasis is placed on the development of rigorous analysis, synthesis, and verification tools to ensure the correctness of the design. The research approach combines expertise in control theory, robotics, optimization, queuing theory and stochastic systems, with randomized and distributed algorithms, formal languages, machine learning, and game theory. Current research areas include the following: Autonomy and Future Urban Mobility: Autonomous, self-driving cars are no longer science fiction, but will be ready for commercial deployment soon. The group’s work on self-driving vehicles is very broad, spanning the whole spectrum from technology development to the analysis of socio-economic impact of such technology. Recent work includes: »» Affordable autonomy: can we design safe and reliable selfdriving vehicles at a cost that make them affordable for the general public? Our demo vehicles at the SingaporeMIT Alliance for Research and Technology were developed with less than $30,000 worth of computers and sensors. AeroAstro graduate student Abhizna Butchibabu prepares an unmanned aerial vehicle for her research in the Interactive Robotics Lab. Abhi and Professor Julie Shah are working to better coordinate autonomous vehicles and humans, improving performance and decision making in applications like search and rescue operations and disaster response. (William Litant/MIT)

62

AEROASTRO 2014-2015


»» Provable safety: how do we make sure that the vehicle will behave safely, and respect all the rules of the road? We developed algorithms that provably satisfy all “hard” rules, while minimizing violations of “soft” rules or recommendations. »» Autonomy for Mobility on Demand: How would self-driving vehicles impact urban mobility in the future? We envision fleets of shared self-driving vehicles, develop algorithms for their sizing and operations, and analyze their effects using real data from several cities worldwide. Real-time motion planning and control: The group is developing state-of-the art algorithms for real-time control of highly maneuverable aircraft, spacecraft, and ground vehicles. Focus areas include optimality and robustness, as well as provable safety and correctness with respect to temporal-logic specifications (e.g., rules of the road, rules of engagement). Current projects include high-speed flight in cluttered environments and high-speed offroad driving. Multi-agent systems: Large, heterogeneous groups of mobile vehicles, such as UAVs and UGVs, are increasingly used to address complex missions for many applications, ranging from national security to environmental monitoring. An additional emphasis in this work is scalability; our objective is not only the design of distributed algorithms to ensure provably efficient and safe execution of the assigned tasks, but also to understand exactly how the collective performance and implementation complexity scale as the group’s size and composition change. Transportation networks: Traffic congestion, and extreme sensitivity to, for example, environmental disruptions, is a well-known effect of increasing access to transportation. As infrastructure development saturates, new approaches are necessary to increase the safety, efficiency, and environmental sustainability

of transportation networks. The group’s research in this area concentrates on the exploitation of real-time information availability through wireless communications among vehicles, and with existing infrastructure, to achieve this goal. Emilio Frazzoli directs the Aerospace Robotics and Embedded Systems group. Visit the Aerospace Robotics and Embedded Systems group at http://ares.lids.mit.edu

THE AUTONOMOUS SYSTEMS LABORATORY The Autonomous Systems Laboratory is a virtual lab led by Professors Brian Williams and Nicholas Roy. Williams’ group, the Model-based Embedded and Robotics (MERS) group, and Roy’s Robust Robotics Group are part of the Computer Science and Artificial Intelligence Lab. ASL work is focused on developing autonomous aerospace vehicles and robotic systems. ASL-developed systems are commanded at a high-level in terms of mission goals. The systems execute these missions robustly by constantly estimating their state relative to the world, and by continuously adapting their plan of action, based on engineering and world models. Below are several recent demonstrations. »» Operating autonomous vehicles to maximize utility in an uncertainty environment, while operating within acceptable levels of risk. Autonomous underwater vehicles enable scientists to explore previously uncharted portions of the ocean, by autonomously performing science missions of up to 20 hours long without the need for human intervention. Performing these extended missions can be a risky endeavor. Researchers have developed robust, chance-constraint planning algorithms that automati-

Lab report

63


cally navigate vehicles to achieve user specified science goals, while operating within risk levels specified by the users. (Video at http://www.csail.mit.edu/videoarchive/ research/robo/auv-planning) »» Another demonstration involves human-robot interaction between a robotic air taxi and a passenger. The task is for the autonomous vehicle to help the passenger rethink goals when they no longer can be met. Companies like the MIT spinoff Terrafugia offer vehicles that can fly between local airports and can travel on local roads. To operate these innovative vehicles, one must be trained as a certified pilot, thus limiting the population that can benefit from this innovative concept. In collaboration with Boeing, MERS has demonstrated in simulation the concept of an autonomous personal air vehicle in which the passenger interacts with the vehicle in the same manner that they interact today with a taxi driver. (Video at http://www.csail.mit.edu/videoarchive/research/robo/personal-aerial-transportation.) »» A third demonstration involves human-robot interaction between an astronaut and the Athlete Lunar Rover. MERS has developed methods for controlling walking machines, guided by qualitative “snapshots” of walking gait patterns. These control systems achieve robust walking over difficult terrain by embodying many aspects of a human’s ability to restore balance after stumbling, such as adjusting ankle support, moving free limbs, and adjusting foot

64

AEROASTRO 2014-2015

placement. Members of the MERS group applied generalizations of these control concepts to control the JPL Athlete robot, a six-legged/wheeled lunar rover that performs heavy lifting and manipulation tasks by using its legs as arms. (Video at http://www.csail.mit.edu/ videoarchive/research/robo/athlete-mers.) ASL faculty are Brian Williams and Nicholas Roy. Visit the Model-based Embedded Systems at mers.csail.mit.edu and the Robust Robotics Group at http://groups.csail.mit.edu/rrg

COMMUNICATIONS AND NETWORKING RESEARCH GROUP The Communications and Networking Research Group’s primary goal is design of network architectures that are cost effective, scalable, and meet emerging needs for high data-rate and reliable communications. To meet emerging critical needs for military communications, space exploration, and internet access for remote and mobile users, future aerospace networks will depend upon satellite, wireless, and optical components. Satellite networks are essential for providing access to remote locations lacking in communications infrastructure, wireless networks are needed for communication between untethered nodes, such as autonomous air vehicles, and optical networks are critical to the network backbone and in high performance local area networks. The group is working on a wide range of projects in the area of communication networks and systems, with application to satellite, wireless, and optical systems. The group has been developing


efficient network control algorithms for heterogeneous wireless networks. Existing wireless networks are almost exclusively confined to single hop access, as provided by cellular telephony or wireless LANs. While multi-hop wireless networks can be deployed, current protocols typically result in extremely poor performance for even moderate sized networks. Wireless Mesh Networks have emerged as a solution for providing last-mile Internet access. However, hindering their success is our relative lack of understanding of how to effectively control wireless networks, especially in the context of advanced physical layer models, realistic models for channel interference, distributed operations, and interface with the wired infrastructure (e.g., the internet). CNRG has been developing effective and practical network control algorithms that make efficient use of wireless resources through the joint design of topology adaptation, network layer routing, link layer scheduling, and physical layer power, channel, and rate control. Robust network design is another exciting area of recent pioneering research by the group. In particular, the group has been developing a new paradigm for the design of highly robust networks that can survive a massive disruption that may result from natural disasters or intentional attack. The work examines the impact of large-scale failures on network survivability and design, with a focus on interdependencies between different networked infrastructures, such as telecommunication networks, social networks, and the power grid. The group’s research crosses disciplinary boundaries by combining techniques from network optimization, queueing theory, graph theory, network protocols and algorithms, hardware design, and physical layer communications.

Eytan Modiano directs the Communications and Networking Research Group. Visit the Communications and Networking Research Group at http://web.mit.edu/aeroastro/labs/cnrg

GAS TURBINE LABORATORY The Gas Turbine Laboratory’s mission is to advance the stateof-the-art in fluid machinery for power and propulsion. The research is focused on advanced propulsion systems, energy conversion and power, with activities in computational, theoretical, and experimental study of loss mechanisms and unsteady flows in fluid machinery, dynamic behavior and stability of compression systems, instrumentation and diagnostics, advanced centrifugal compressors and pumps for energy conversion, gas turbine engine and fluid machinery noise reduction and aeroacoustics, novel aircraft and propulsion system concepts for reduced environmental impact. Current research projects include: »» a unified approach for vaned diffuser design in advanced centrifugal compressors »» investigation of real gas effects in supercritical CO2 compression systems »» modeling instabilities in high-pressure pumping systems »» aeromechanic response in a high performance centrifugal compressor stage ported shroud operation in turbochargers »» manifestation of forced response in a high performance centrifugal compressor stage for aerospace applications

Lab report

65


»» multiparameter control for centrifugal compressor performance optimization »» performance improvement of a turbocharger twin scroll type turbine stage »» a two-engine integrated propulsion system »» propulsor design for exploitation of boundary layer ingestion »» aerodynamics and heat transfer in gas turbine tip shroud cavity flows »» secondary air interactions with main flow in axial turbines »» compressor aerodynamics in large industrial gas turbines for power generation »» flow and heat transfer in modern turbine rim seal cavities »» modeling cavitation instabilities in rocket engine turbopumps »» diagnostics and prognostics for gas turbine engine system stability characterization »» investigation of the origins of short-wavelength instability inception in axial compressors »» assessment of thermal effects on compressor transients. »» investigation of surface waviness effects on compressor performance Faculty and research staff include David Darmofal, Fredric Ehrich, Alan Epstein (emeritus), Edward Greitzer, Claudio Lettieri, Zoltán Spakovszky (director), Choon Tan, Neil Titchener, and Alejandra Uranga. Visit the Gas Turbine Lab at http://mit.edu/aeroastro/labs/gtl

66

AEROASTRO 2014-2015

INTERNATIONAL CENTER FOR AIR TRANSPORTATION The International Center for Air Transportation undertakes research and educational programs that discover and disseminate the knowledge and tools underlying a global air transportation industry driven by technologies. Global information systems are central to the future operation of international air transportation. Modern information technology systems of interest to ICAT include global communication and positioning, international air traffic management, scheduling, dispatch, and maintenance support, vehicle management, passenger information and communication, and real-time vehicle diagnostics. Airline operations are also undergoing major transformations. Airline management, airport security, air transportation economics, fleet scheduling, traffic flow management, and airport facilities development, represent areas of great interest to the MIT faculty and are of vital importance to international air transportation. ICAT is a physical and intellectual home for these activities. ICAT, and its predecessors, the Aeronautical Systems Laboratory and Flight Transportation Laboratory, pioneered concepts in air traffic management and flight deck automation and displays that are now in common use. ICAT faculty include R. John Hansman (director), Cynthia Barnhart, Peter Belobaba, and Amedeo Odoni. Visit the International Center for Air Transportation at http://mit.edu/aeroastro/ labs/ICAT/

LABORATORY FOR AVIATION AND THE ENVIRONMENT The Laboratory for Aviation and the Environment was founded in the 1990s as the Aero-Environmental Research Laboratory by Ian A. Waitz, now dean of the MIT School of Engineering.


One of the aviation industry’s defining challenges is addressing aviation’s environmental impact in terms of noise, air quality and climate change. LAE’s goal is to align the trajectory of aerospace technology and policy development with the need to mitigate these impacts. It does so by increasing the understanding the environmental effects of aviation, by developing and assessing fuel-based, operational and technological mitigation approaches and by disseminating knowledge and tools. LAE also contributes to cognate areas of inquiry in aerospace, energy, and the environment LAE researchers are analyzing environmental impacts and developing research tools that provide rigorous guidance to policy-makers who must decide among alternatives when addressing aviation’s environmental impact. The MIT researchers collaborate with international teams in developing aircraft-level and aviation system level tools to assess the costs and benefits of different policies and mitigation options. A current LAE focus is on studying the environmental sustainability of alternative aviation fuels from biomass or natural gas. This research includes both drop-in fuel options, which can be used with existing aircraft engines and fuel infrastructure, as well as non-drop-in options such as liquefied natural gas, which would require modifications to aircrafts and infrastructure. Environmental metrics considered include lifecycle greenhouse gas emissions, land requirements and water consumption. LAE researchers are also estimating trade offs among different metrics and usages to better understand the full consequences of introducing a certain alternative fuel into the aviation system. LAE has developed and publicly released a code that allows for modeling and evaluation emissions and their impacts throughout the troposphere and stratosphere in a unified fashion. LAE has also recently released a new global emissions dataset for civil

aviation emissions, which represents the most current estimate of emissions publicly available. It is widely used by researchers worldwide, in areas including atmospheric modeling and aviation and the environment. Other recent work quantifies air pollution and associated health effects attributable to the different economic sectors in the United States, and the environmental and economic impacts of higher octane gasoline usage for road transportation. LAE faculty include Steven Barrett, director; Robert Malina, associate director; Hamsa Balakrishnan; John Hansman; Ian Waitz; and Karen Willcox. Also associated with LAE are Ray Speth, research scientist; and Brian Yutko, postdoctoral associate. Visit LAE at http://lae.mit.edu

LABORATORY FOR INFORMATION AND DECISION SYSTEMS The Laboratory for Information and Decision Systems is an interdepartmental research center committed to advancing research and education in the analytical information and decision sciences: systems and control, communications and networks, and inference and statistical data processing. Dating to 1939, LIDS has been at the forefront of major methodological developments, relevant to diverse areas of national and worldwide importance, such as telecommunications, information technology, the automotive industry, energy, defense, and human health. Building on past innovation and bolstered by a collaborative atmosphere, LIDS members continue to make breakthroughs that cut across traditional boundaries. Members of the LIDS community share a common approach to solving problems and recognize the fundamental role that mathematics, physics, and computation play in their research. Their

Lab report

67


THE LEARNING LABORATORY The AeroAstro Learning Laboratory, located in Building 33, promotes student learning by providing an environment for hands-on activities that span our conceive-design-implementoperate educational paradigm. The Learning Lab comprises four main areas: Robert C. Seamans Jr. Laboratory. The Seamans Laboratory occupies the first floor. It includes: »» The Concept Forum — a multipurpose room for meetings, presentations, lectures, videoconferences and collaboration, distance learning, and informal social functions. In the Forum, students work together to develop multidisciplinary concepts, and learn about program reviews and management. »» Al Shaw Student Lounge — a large, open space for social interaction and operations. During a December 2014 Course 16.62x Experimental Projects display in AeroAstro’s Gelb Lab student workshop, undergraduate Ana Vazquez explains her team’s creation to Professor Emeritus John Dugundji. The project features a wing surface with integrated bioinspired microelectromechanical sensors. Biologically inspired engineering is a scientific discipline that applies biological principles to engineering. (William Litant/MIT)

pursuits are strengthened by the laboratory’s affiliations with colleagues across MIT and throughout the world, as well as with leading industrial and government organizations. LIDS is based in MIT’s Stata Center, a dynamic space that promotes a high level of interaction within the lab and with the larger MIT community. AeroAstro faculty affiliated with the laboratory are Emilio Frazzoli, Jonathan How, Eytan Modiano, and Moe Win. Visit LIDS at http://lids.mit.edu/

68

AEROASTRO 2014-2015

Arthur and Linda Gelb Laboratory. Located in the building’s lower level, the Gelb Laboratory includes the Gelb Machine Shop, Instrumentation Laboratory, Mechanical Projects Area, Projects Space, and the Composite Fabrication-Design Shop. The Gelb Laboratory provides facilities for students to conduct hands-on experiential learning through diverse engineering projects starting as first-year students and continuing through the last year. The Gelb facilities are designed to foster teamwork with a variety of resources to meet the needs of curricular and extra-curricular projects. Gerhard Neumann Hangar. The Gerhard Neumann Hangar is a high bay space with an arching roof. This space lets students work on large-scale projects that take considerable floor and table space. Typical of these projects are planetary rovers, autonomous vehicles, and re-entry impact experiments. The structure also houses low-speed and supersonic wind tunnels. A balcony-like mezzanine level is used for multi-semester engineering projects, such as the experimental three-term senior capstone course.


In June 2015, the Rocket Team took first place at the Annual Intercollegiate Rocket Engineering Competition, besting 34 other teams, with their entry reaching an altitude of 9,690 feet. The team does most of its construction in the Gelb Lab student workshop. (MIT Rocket Team)

Digital Design Studio. The Digital Design Studio, located on the second floor, is a large room with multiple computer stations arranged around reconfigurable conference tables. Here, students conduct engineering evaluations and design work, and exchange computerized databases as system and subsystem trades are conducted during the development cycle. The room is equipped with information technologies that facilitate teach-

ing and learning in a team-based environment. Adjacent and networked to the main Design Studio are two smaller design rooms: the AA Department Design Room, and the Arthur W. Vogeley Design Room. These rooms are reserved for the use of individual design teams and for record storage. The department’s IT systems administrator is positioned for convenient assistance in an office adjacent to the Design Center.

Lab report

69


Swedish astronaut Christer Fuglesang tests a centrifuge-mounted pedaling device in the Man Vehicle Lab that could be used on the International Space Station to analyze artificial gravity levels and ergometer exercise on musculoskeletal and cardiovascular functions, as well as motion sickness and comfort. Emeritus Professor Larry Young assists. (William Litant/MIT)

MAN VEHICLE LABORATORY The Man Vehicle Laboratory improves the understanding of human physiological and cognitive capabilities as applied to human-vehicle, and human-robotic system safety and efficacy, as well as decision-making augmented by technological aids. MVL develops countermeasures and display designs to aid pilots, astronauts, clinicians, patients, soldiers, and others. Research is interdisciplinary and uses techniques from manual and supervisory control, signal processing, estimation, robotics, sensory-motor physiology, sensory and cognitive psychology, biomechanics, human factors engineering, artificial intelligence, and biostatistics. MVL has

70

AEROASTRO 2014-2015

flown experiments on the Space Shuttle, the Mir Space Station, and on many parabolic flights, and developed experiments for the International Space Station. Space applications include advanced space suit design and dynamics of astronaut motion, adaptation to rotating artificial gravity, mathematical models for human spatial disorientation, accident analysis, artificial intelligence, and space telerobotics training. Ongoing work includes the development of countermeasures using a short radius centrifuge and development of a g-loading suit to maintain muscle and bone strength. New major projects include a collaborative study of adaptation in roll


On May 12, 2015, AeroAstro Professor Dava Newman was sworn in as NASA deputy administrator by video link between Washington and her MIT office. Dava’s partner, Gui Trotti, holds the Bible during the ceremony. Dava is on leave from her Man Vehicle Lab co-director position. (Maia Weinstock/MIT News Office)

Lab report

71


tilt perception and manual control to altered G environments using a centrifuge at the Massachusetts Eye and Ear Infirmary, and a study with UC Davis on customized and just in time space telerobotics refresher training. Non-aerospace projects include GE locomotive cab automation and displays, advanced helmet designs for brain protection in sports and against explosive blasts, the development of wearable sensor systems and data visualizations for augmenting clinical decision-making, and data fusion for improving situation awareness for dismounted soldiers. Research sponsors include NASA, the National Space Biomedical Research Institute, the National Science Foundation, the Office of Naval Research, the FAA, the FRA, Draper Laboratory, the Center for Integration of Medicine and Innovative Technology, the Deshpande Center, and the MIT Portugal Program. The laboratory also collaborates with the Volpe Transportation Research Center, Massachusetts General Hospital, and the Jenks Vestibular Physiology Laboratory of the Massachusetts Eye and Ear Infirmary. MVL faculty include Professor Jeffrey Hoffman, director; Professor Emeritus Laurence Young; Dr. Chuck Oman; Professor Julie Shah; and Professor Leia Stirling. They teach subjects in human factors engineering, space systems engineering, real-time systems and software, space policy, flight simulation, space physiology, aerospace biomedical engineering, the physiology of human spatial orientation, statistical methods in experimental design, and leadership. The MVL also serves as the office of the director for the NSBRI-sponsored HST Graduate Program in Bioastronautics (Young) and the Massachusetts Space Grant Consortium (Hoffman). Visit the Man Vehicle Laboratory at http://mvl.mit.edu

72

AEROASTRO 2014-2015

NECSTLAB The necstlab (pronounced “next lab”) research group explores new concepts in engineered materials and structures, with a focus on nanostructured materials. The group’s mission is to lead the advancement and application of new knowledge at the forefront of materials and structures understanding, with research contributions in both science and engineering. Applications of interest include enhanced aerospace advanced composites, multifunctional attributes of structures such as damage sensing, and microfabricated (MEMS) topics. The necstlab group has interests that span fundamental materials synthesis questions through to structural applications of both hybrid and traditional materials. Much of the group’s work supports the efforts of the NECST Consortium, an aerospace industry-supported research initiative that seeks to develop the underlying understanding to create enhanced-performance advanced composites using nanotechnology. Beyond the NECST Consortium Members, necstlab research is supported directly or through collaboration by industry, AFOSR, ARO, NASA, NIST, NSF, ONR, and others. The necstlab maintains collaborations around the MIT campus, particularly with faculty in the Mechanical Engineering, Materials Science and Engineering, and Chemical Engineering departments; and MIT labs and centers including the Institute for Soldier Nanotechnologies, Materials Processing Center, Center for Materials Science and Engineering, and the Microsystems Technology Laboratory, as well as Harvard’s Center for Nanoscale Systems. Important to the contributions of the necstlab are collaborations with leading research groups from around the world. In the fall of 2014, the group moved into new laboratory space in MIT Building 35.


Example past and current research projects include: »» BioNEMS materials design and implementation in microfluidics »» buckling mechanics »» carbon nanostructure synthesis from non-traditional catalysts »» continuous growth of aligned carbon nanotubes »» electroactive nanoengineered actuator/sensor architectures focusing on ion transport »» nanoengineered (hybrid) composite architectures for laminate-level mechanical performance improvement »» multifunctional nanoengineered bulk materials including damage sensing and detection »» nanomanufacturing »» polymer nanocomposite mechanics and electrical and thermal transport »» silicon MEMS devices including piezoelectric energy harvesters, microfabricated solid oxide fuel cells, stress characterization, and 3D MEMS »» vertically-aligned carbon nanotube characterization and physical properties necstlab faculty include Brian L. Wardle, director; John Dugundji, emeritus; and visitor Antonio Miravete Visit necstlab at http://mit.edu/aeroastro/labs/necstlab

SPACE PROPULSION LABORATORY The Space Propulsion Laboratory studies and develops systems for increasing performance and reducing costs of space propulsion and related technologies. A major area of interest to the lab is electric propulsion in which electrical, rather than chemical, energy propels spacecraft. The benefits are numerous, hence the reason electric propulsion systems are increasingly applied to communication satellites and scientific space missions. These efficient engines allow exploration in more detail of the structure of the universe, increase the lifetime of commercial payloads, and look for signs of life in far away places. Areas of research include plasma engines and plumes, and their interaction with spacecraft and thruster materials, and numerical and experimental models of magnetic cusped thrusters. SPL also has a significant role in designing and building microfabricated electrospray thrusters, including their integration into space missions. In addition to providing efficient propulsion for very small satellites in the 1 kg category (like CubeSats), these engines will enable distributed propulsion for the control of large space structures, such as deformable mirrors and apertures. A recent line of research is focused on the favorable scaling potential of electrospray thrusters for applications in power-intensive missions. SPL has delivered flight hardware to test the first electrospray thrusters in space in CubeSats. The science behind electrosprays is explored as well, mainly on the ionic regime where molecular species are directly evaporated from ionic liquid surfaces. Also, applications beyond propulsion are investigated; for example, the use of highly monoenergetic molecular ion beams in focusing columns for materials structuring and characterization at the nano-scale and also applications in vacuum technology. SPL

Lab report

73


facilities include a computer cluster where plasma and molecular dynamics codes are routinely executed and a state-of-the-art laboratory including five vacuum chambers, clean room environment, electron microscope, materials synthesis capabilities, nanosatellite qualification equipment (vibration/thermal and in-vacuum magnetically-levitated CubeSat simulator), plasma/ ion beam diagnostic tools to support ongoing research efforts and a laser micromachining facility. SPL faculty are professors Paulo Lozano, director, and Manuel Martinez-Sanchez, emeritus. Visit the Space Propulsion Lab at http://mit.edu/aeroastro/labs/spl

SPACE SYSTEMS LABORATORY The Space Systems Laboratory research contributes to the exploration and development of space. SSL’s mission is to explore innovative space systems concepts while training researchers to be conversant in this field. The major programs include systems analysis studies and tool development, AeroAstro student-built instruments and small satellites for exploration and remote sensing, precision optical systems for space telescopes, and microgravity experiments operated aboard the International Space Station and the NASA reduced gravity aircraft. Research topics focus on space systems and include dynamics, guidance, and control; active structural control; space power and propulsion; modular space systems design; micro-satellite design; real-time embedded systems; software development; and systems architecting. SSL has a unique facility for space systems research, the Synchronized Position Hold Engage and Reorient Experimental Satellites (SPHERES). The SPHERES facility is used to develop proximity satellite operations such as inspection, cluster aggre-

74

AEROASTRO 2014-2015

gation, collision avoidance and docking, as well as formation flight. The SPHERES facility consists of three satellites 20 centimeters in diameter that have been inside the International Space Station since May 2006. In 2009, SSL expanded the uses of SPHERES to include STEM outreach through an exciting program called Zero Robotics (http://zerorobotics.mit.edu), which engages high school and middle school students in a competition aboard the ISS using SPHERES. It has expanded to over 100 U.S. and 50 European teams annually. In 2014, finalists were joined at MIT by Dante Lauretta, PI of the Osirix REX Mission and at ESA ESTEC by Matthew Taylor, Project Scientist of the Rosetta mission, as guest speakers. MIT alum and retired NASA Astronaut Catherine “Cady” Coleman hosted the students at MIT while Astronaut Paolo Nespoli, from the Italian Space Agency hosted the students at ESTEC. The competition was run aboard the ISS by three crewmembers: Cosmonaut Elena Serova from Russia, Astronaut Samantha Cristoforetti from Italy, and Barry “Butch” Wilmore from the USA. For the first time teams from Russia and Mexico also participated in this growing competition. There have been several recent and exciting hardware augmentations to SPHERES. In October 2012, the SPHERES facility on the ISS was expanded to include vision-based navigation through the Visual Estimation for Relative Tracking and Inspection of Generic Objects (VERTIGO) program. VERTIGO uses vision-based navigation and mapping algorithms through a stereo camera system and an upgraded processor. In 2013, the University of Maryland Space Power and Propulsion Laboratory, Aurora Flight Sciences, and SSL upgraded the SPHERES facility to include the Resonant Inductive Near-Field Generation System, which have been used to test electromagnetic formation flight and wireless power transfer through a pair of tuned resonant coils that generate a


time varying magnetic field. The SPHERES-Slosh module was also launched in 2013, in collaboration with the Florida Institute of Technology, and has enabled surveys of fluid slosh behavior in zero gravity. Recently, SSL partnered with the Naval Research Laboratory and Aurora Flight Sciences for work on the Defense Advanced Research Projects Agency Phoenix program for satellite servicing and assembly missions. To this end, a cadre of Universal Docking Ports and Halos will be launched and operated on the SPHERES ISS facility in the fall of 2014. The Universal Docking Ports enable active docking and undocking of the satellites creating a rigid assembly; they add fiducial-based vision navigation. The Halo structure enables attachment of up to six electro-mechanical devices around a single SPHERES satellite, allowing researchers to study complex geometrical system reconfiguration. During the fall 2013 and spring 2014 semesters, the MIT 16.83x senior space systems design capstone class developed a prototype for the INSPECT (Integrated Navigation Sensor Platform for EVA Control and Testing) program, which added a thermal imager, optical range finder, and control moment gyros to the ground SPHERES + Halo system. SSL is also active in the area of nanosatellites, particularly CubeSats. In March 2014, the Microsized Microwave Atmospheric Satellite (MicroMAS), a joint effort between MIT SSL students, led by Professor Kerri Cahoy, and MIT Lincoln Laboratory staff, led by Dr. Bill Blackwell, was delivered to NanoRacks for launch to the ISS, from which MicroMAS will be deployed in summer 2014. MicroMAS is a dual-spinning three-unit CubeSat hosting a microwave radiometer payload that captures temperature map images of Earth and is important for characterizing hurricanes and tropical storms. SSL will develop a follow-on to MicroMAS called the Microwave Radiometer Technology Acceleration mission (MiRaTA) which pairs an advanced miniature microwave

radiometer with a GPS radio occultation receiver to help improve radiometer calibration. MiRaTA is sponsored by the NASA Earth Science Technology Office and scheduled for a 2016 launch. SSL students are working with engineers at Aurora Flight Sciences and the AeroAstro Space Propulsion Laboratory on a cluster formation-flight nanosatellite project. SSL students are also engaged with Professor Sara Seager and students in the MIT Earth and Planetary Sciences department, engineers at the Charles Stark Draper Laboratory, and NASA Jet Propulsion Laboratory on the novel ExoplanetSat nanosatellite which uses a two-stage control system (reaction wheels plus piezo stage) to maintain precise pointing at a target star to obtain exoplanet transit measurements using advanced photometry. The Wavefront Control Laboratory, led by Cahoy, is a part of SSL that focuses on precision active optical systems for space applications. WCL students are working on three projects. One is the Deformable Mirror Demonstration Mission (DeMi), which will validate and demonstrate the capabilities of high actuator count MEMS deformable mirrors for high contrast astronomical imaging. The DeMi optical payload will characterize MEMS deformable mirror operation using both a Shack Hartmann wavefront sensor as well as sensorless wavefront control. The second project is a CubeSat Free Space Optical communications downlink that uses a staged control system with MEMS fast steering mirrors. The third project is a bistatic laser system for active characterization of the bidirectional reflectance distribution function of space materials. WCL students also are investigating whether or not we can use standard communications satellite components as space weather sensors, and developing algorithms that can predict the onset of space-weather related anomalies. (http://web.mit.edu/cahoylab).

Lab report

75


SSL is also developing and building the REXIS (REgolith X-ray Imaging Spectrometer) student collaboration instrument, which will be aboard NASA’s next New Frontiers mission: OSIRIS-REx (Origins, Spectral Interpretation, Resource Identification, Security Regolith Explorer). The mission is an asteroid sample return mission, which will launch in 2016 to visit the Near-Earth Asteroid Bennu. REXIS is one of five instruments onboard and uses a 2x2 array of Lincoln Laboratory designed charged-coupled devices to measure the X-ray fluorescence from Bennu to allow characterization of the surface of the asteroid among the major meteorite groups as well as a coded aperture mask to map the spatial distribution of element concentrations in the regolith. Professor Richard Binzel, who maintains a joint MIT EAPS–AeroAstro appointment and Dr. Rebecca Masterson are leading the project in collaboration with EAPS, Kavli Institute, and Harvard College Observatory. REXIS has included the work of more than 50 undergraduate and graduate students and has successfully completed its Critical Design Review. The team is wrapping up the engineering model testing and is on the path to the flight hardware build. The REXIS flight instrument will be delivered to Lockheed Martin for integration onto the OSIRIS-REx spacecraft in the summer of 2015. SSL is directed by Dr. Alvar Saenz Otero while Professor David W. Miller is on leave from MIT as NASA Chief Technologist. Professors Kerri Cahoy, Jeffrey Hoffman, Olivier de Weck, and Richard Binzel participate in the multiple SSL projects. Dr. Rebecca Masterson manages REXIS. Dr. Danilo Roascio leads the SPHERES team. The group is supported by research specialist Paul Bauer, fiscal officers Suxin Hu and Ngan Kim Le, and administrative assistant Marilyn E. Good. Collaborators include AeroAstro professors Manuel Martinez-Sanchez and Paulo Lozano, and EAPS professor Sara Seager. Visit the Space Systems Laboratory at http://ssl.mit.edu

76

AEROASTRO 2014-2015

SPACE TELECOMMUNICATIONS, RADIATION, AND ASTRONOMY LABORATORY The Space Telecommunications, Radiation, and Astronomy Laboratory, or STAR Lab, is affiliated with the Space Systems Lab. It focuses on developing instruments and platforms that enable weather sensing on Earth and other planets, including exoplanets, as well as monitoring “space weather” — the highly energetic flow of radiation, or charged particles, that is constantly streaming towards Earth from the sun. In addition to the flight CubeSat weather sensing projects described in the Space Systems Laboratory section (MicroMAS and MiRaTA) and work to autonomously optimize spacecraft scientific observation return given constrained resources, STAR Lab projects also involve the use of active optical elements, sensing, and tracking systems. Nanosatellite Optical Downlink Experiment: NODE is a miniaturized laser communication module for small satellites that incorporates commercial off-the-shelf components, including a MEMS fast-steering mirror for fine pointing, onto standard threeaxis stabilized spacecraft to achieve free space optical data rates better than 50 Mbps downlinking from low Earth orbit to amateur-astronomy-class 30 cm telescope ground stations. A flight demonstration of a NODE prototype is expected in late 2015 or early 2016. The lab is also developing an atmospheric sensor that has overlap with the NODE configuration. Direct Imaging of Exoplanets: STAR Lab is involved in mission design and technology demonstration efforts toward direct imaging of exoplanets, a method where a space telescope equipped with an occulter is used to image planets around other stars by blocking out (occulting) the parent star and measuring the spectra of the ever so faint exoplanets orbiting it. The spectra are used to tell us about the atmosphere and weather on the exoplanets, as well as give indications about life and habitability. One demonstration mission, called DeMi, the Deformable Mirror Dem-


onstration, involves looking at bright stars and testing a MEMS deformable mirror on a 3U CubeSat using a miniaturized Shack Hartmann wavefront sensor. STAR Lab also collaborates with students in EAPS and at staff at NASA Ames Research Center and Space Telescope Science Institute to do modeling of the retrieved atmospheric spectra. Space Weather Sensing: In collaboration with the MIT Nuclear Science and Engineering Department, STAR Lab is helping to develop miniaturized radiation sensors for satellites that can help us to learn more about the particle types and energies that affect our orbiting assets than a simple dosimeter can. This effort complements work being done to analyze commercial operator satellite telemetry and anomaly databases to understand the sensitivity of spacecraft and components to space weather events and develop new spacecraft and system health monitoring algorithms. The STAR Lab is directed by Professor Kerri Cahoy Visit the STAR Laboratory at http://starlab.mit.edu

humans operating these systems have a much more difficult job than simply following pre-defined procedures. We can no longer effectively separate engineering design from human factors and from the social and organizational system in which our systems are designed and operated. The System Engineering Research Lab’s goal is to create tools and processes that will allow us to engineer a safer world. Engineering safer systems requires multi-disciplinary and collaborative research based on sound system engineering principles, that is, it requires a holistic systems approach. LSSR has participants from multiple engineering disciplines and MIT schools as well as collaborators at other universities and in other countries. Students are working on safety in aviation (aircraft and air transportation systems, unmanned aircraft, air traffic control), spacecraft, medical devices and healthcare, automobiles, nuclear power, defense systems, energy, and large manufacturing/process facilities. Cross-discipline topics include: »» hazard analysis

SYSTEM ENGINEERING RESEARCH LAB

»» accident causality analysis and accident investigation

The increasingly complex systems we are building today enable us to accomplish tasks that were previously difficult or impossible. At the same time, they have changed the nature of accidents and increased the potential to harm not only life today but also future generations. Traditional system safety engineering approaches, which started in the missile defense systems of the 1950s, are being challenged by the introduction of new technology and the increasing complexity of the systems we are attempting to build. Software is changing the causes of accidents and the

»» safety-guided design »» human factors and safety »» integrating safety into the system engineering process »» identifying leading indicators of increasing risk »» certification, regulation, and standards »» the role of culture, social, and legal systems on safety »» managing and operating safety-critical systems

Lab report

77


Recently we have discovered that our safety techniques are also effective for security, and we are now involved in cyber security and physical (nuclear) security in work for the DoD, FAA, and DoE. The System Engineering Research Lab is directed by Professor Nancy Leveson. Dr. John Thomas is an SRL-affiliated research engineer.

»» thermostructural design, manufacture, and testing of composite thin films and associated fundamental mechanical and microstructural characterization

Visit the System Engineering Research Lab at http://sunnyday.mit.edu/safety.html

»» continued efforts on addressing the roles of lengthscale in the failure of composite structures

TECHNOLOGY LABORATORY FOR ADVANCED MATERIALS AND STRUCTURES A dedicated and multidisciplinary group of researchers constitute the Technology Laboratory for Advanced Materials and Structures. They work cooperatively to advance the knowledge base and understanding that will help facilitate and accelerate advanced materials systems development and use in various advanced structural applications and devices.

»» numerical and analytical solid modeling to inform, and be informed by, experiments »» continued engagement in the overall issues of the design of composite structures with a focus on failure and durability, particularly within the context of safety

»» MEMS-scale mechanical energy harvesting modeling, design, and testing

In supporting this work, TELAMS has complete facilities for the fabrication of structural specimens such as coupons, shells, shafts, stiffened panels, and pressurized cylinders made of composites, active, and other materials. TELAMS testing capabilities include a battery of servohydraulic machines for cyclic and static testing, a unit for the catastrophic burst testing of pressure vessels, and an impact testing facility. TELAMS maintains capabilities for environmental conditioning, testing at low and high temperature, and in hostile and other controlled environments. There are facilities for microscopic inspection, nondestructive inspection, high-fidelity characterization of MEMS materials and devices, and a laser vibrometer for dynamic device and structural characterization. This includes ties to ability for computer microtomography.

»» MEMS device modeling and testing, including bioNEMS/MEMS

With its linked and coordinated efforts, both internal and external, the laboratory continues its commitment to leadership in

TELAMS has broadened its interests from a strong historical background in composite materials, and this is reflected in the name change from the former Technology Laboratory for Advanced Composites. Thus, the research interests and ongoing work in the laboratory represent a diverse and growing set of areas and associations. Areas of interest include: »» composite tubular structural and laminate failures

78

»» structural health monitoring system development and durability assessment

AEROASTRO 2014-2015


the advancement of the knowledge and capabilities of the materials and structures community through education of students, original research, and interactions with the community. There has been a broadening of this commitment consistent with the broadening of the interest areas in the laboratory. In all these efforts, the laboratory and its members continue their extensive collaborations with industry, government organizations, other academic institutions, and other groups and faculty within the MIT community. TELAMS faculty include Professor Paul A. LagacĂŠ, Professor John Dugundji (emeritus), and visitor Antonio Miravete. Visit the Technology Laboratory for Advanced Materials and Structures at http://mit.edu/telams

WIRELESS COMMUNICATION AND NETWORK SCIENCES GROUP The Wireless Communication and Network Sciences Group is involved in multidisciplinary research that encompasses developing fundamental theories, designing algorithms, and conducting experiments for a broad range of real-world problems. Its current research topics include location-aware networks, network synchronization, aggregate interference, intrinsically-secure networks, time-varying channels, multiple antenna systems, ultra-wide bandwidth systems, optical transmission systems, and space communications systems. Details of a few specific projects are given below. The group is working on location-aware networks in GPS-denied environments, which provide highly accurate and robust positioning capabilities for military and commercial aerospace

networks. It has developed a foundation for the design and analysis of large-scale location-aware networks from the perspective of theory, algorithms, and experimentation. This includes derivation of performance bounds for cooperative localization, development of a geometric interpretation for these bounds, and the design of practical, near-optimal cooperative localization algorithms. It is currently validating the algorithms in a realistic network environment through experimentation in the lab. The lab has been engaged in the development of a state-of-the-art apparatus that enables automated channel measurements. The apparatus makes use of a vector network analyzer and two vertically polarized, omni-directional wideband antennas to measure wireless channels over a range of 2–18 GHz. It is unique in that extremely wide bandwidth data, more than twice the bandwidth of conventional ultra-wideband systems, can be captured with high-precision positioning capabilities. Data collected with this apparatus facilitates the efficient and accurate experimental validation of proposed theories and enables the development of realistic wideband channel models. Work is underway to analyze the vast amounts of data collected during an extensive measurement campaign that was completed in early 2009. Lab students are also investigating physical-layer security in large-scale wireless networks. Such security schemes will play increasingly important roles in new paradigms for guidance, navigation, and control of unmanned aerial vehicle networks. The framework they have developed introduces the notion of a secure communications graph, which captures the informationtheoretically secure links that can be established in a wireless network. They have characterized the s-graph in terms of local and global connectivity, as well as the secrecy capacity of connec-

Lab report

79


tions. They also proposed various strategies for improving secure connectivity, such as eavesdropper neutralization and sectorized transmission. Lastly, they analyzed the capability for secure communication in the presence of colluding eavesdroppers. To advocate outreach and diversity, the group is committed to attracting undergraduates and underrepresented minorities, giving them exposure to theoretical and experimental research at all levels. For example, the group has a strong track record for hosting students from both the Undergraduate Research Opportunities Program and the MIT Summer Research Program (MSRP). Professor Win maintains dynamic collaborations and partnerships with academia and industry, including the University of Bologna and Ferrara in Italy, University of Lund in Sweden, University of Oulu in Finland, National University of Singapore, Nanyang Technological University in Singapore, Draper Laboratory, the Jet Propulsion Laboratory, and Mitsubishi Electric Research Laboratories. Professor Moe Win directs the Wireless Communication and Network Sciences Group. Visit the Wireless Communication and Network Sciences Group at http://wgroup.lids.mit.edu

WRIGHT BROTHERS WIND TUNNEL Since its opening in September 1938, the Wright Brothers Wind Tunnel has played a major role in the development of aerospace, civil engineering and architectural systems. In recent years, faculty research interests generated long-range studies of unsteady airfoil flow fields, jet engine inlet-vortex behavior, aeroelastic tests of unducted propeller fans, and panel methods for tunnel wall interaction effects. Industrial testing has ranged over auxiliary propulsion burner units, helicopter antenna pods, and in-flight trailing cables, as well as concepts for roofing attachments, a variety of stationary and vehicle mounted ground

80

AEROASTRO 2014-2015

antenna configurations, the aeroelastic dynamics of airport control tower configurations for the Federal Aviation Authority, and the less anticipated live tests in Olympic ski gear, space suits for tare evaluations related to underwater simulations of weightless space activity, racing bicycles, subway station entrances, and Olympic rowing shells for oarlock system drag comparisons. In its more than 75 years of operations, Wright Brothers Wind Tunnel work has been recorded in hundreds of theses and more than 1,000 technical reports. WBWT faculty and staff include Professor Mark Drela and technical instructor David Robertson. Visit the Wright Brothers Wind Tunnel at http://aeroastro.mit.edu/wbwt


Aeroastro magazine 2014-15  

Aeroastro magazine 2014-15

Read more
Read more
Similar to
Popular now
Just for you