1 00:00:00,833 --> 00:00:06,833 [music playing] 2 00:00:16,733 --> 00:00:22,633 - Welcome to the 2016 NASA Ames Summer Series. 3 00:00:22,633 --> 00:00:26,733 When Robert C. Hook, in 1653, 4 00:00:26,733 --> 00:00:29,333 took a thin slice of cork 5 00:00:29,333 --> 00:00:31,900 and observed it under a microscope, 6 00:00:31,900 --> 00:00:38,366 he discovered modular chambers he termed "cells." 7 00:00:38,366 --> 00:00:41,400 Upon further studies, 8 00:00:41,400 --> 00:00:44,066 we found that living cells 9 00:00:44,066 --> 00:00:46,500 have the capability to react 10 00:00:46,500 --> 00:00:50,633 and adapt to the environment. 11 00:00:50,633 --> 00:00:54,966 Imagine a world where vehicles start off 12 00:00:54,966 --> 00:00:56,766 with a particular form 13 00:00:56,766 --> 00:01:01,633 and have the capability to morph, during travel, 14 00:01:01,633 --> 00:01:06,666 to the environment that they experience. 15 00:01:06,666 --> 00:01:08,733 Today's seminar, entitled 16 00:01:08,733 --> 00:01:11,233 "Building Blocks for Aerostructures," 17 00:01:11,233 --> 00:01:16,666 will be given by Dr. Kenny Cheung. 18 00:01:16,666 --> 00:01:18,833 Dr. Cheung received a Bachelor of Arts 19 00:01:18,833 --> 00:01:21,466 from Cornell University, 20 00:01:21,466 --> 00:01:24,733 then came over to this side 21 00:01:24,733 --> 00:01:28,500 and received a Masters in Computer Science-- 22 00:01:28,500 --> 00:01:30,300 actually didn't. Sorry, I apologize. 23 00:01:30,300 --> 00:01:32,666 Stayed at MIT. 24 00:01:32,666 --> 00:01:35,533 Did a Masters in computer science at MIT, 25 00:01:35,533 --> 00:01:41,600 followed by a PhD in applied physics. 26 00:01:41,600 --> 00:01:44,166 He joined Ames and is currently 27 00:01:44,166 --> 00:01:45,666 a research scientist with the Office 28 00:01:45,666 --> 00:01:47,466 of Chief Technologist, 29 00:01:47,466 --> 00:01:49,700 and he conducts research on the application 30 00:01:49,700 --> 00:01:51,800 of building blocks, base materials, 31 00:01:51,800 --> 00:01:57,633 and algorithms for aeronautical and space systems. 32 00:01:57,633 --> 00:02:01,400 Please join me in welcoming Dr. Cheung. 33 00:02:01,400 --> 00:02:05,766 [applause] 34 00:02:05,766 --> 00:02:12,200 - Thank you, and thank you all for coming today. 35 00:02:12,200 --> 00:02:16,133 I'd like to begin by acknowledging the various 36 00:02:16,133 --> 00:02:17,433 contributing organizations 37 00:02:17,433 --> 00:02:19,800 to the work that you'll see in this talk. 38 00:02:19,800 --> 00:02:23,866 In particular, the Aero Research Mission 39 00:02:23,866 --> 00:02:27,166 Directorate's Convergent Aeronautics Solutions program, 40 00:02:27,166 --> 00:02:30,400 which is right now our biggest sponsor, 41 00:02:30,400 --> 00:02:32,100 if you will, under the Transformative 42 00:02:32,100 --> 00:02:34,233 Aeronautics Concepts program. 43 00:02:34,233 --> 00:02:37,300 That work extends from NASA 44 00:02:37,300 --> 00:02:40,800 Aeronautics Research Institute, Team Seedling. 45 00:02:40,800 --> 00:02:42,766 That was very enabling. 46 00:02:42,766 --> 00:02:45,266 And we also have some work that you'll see today 47 00:02:45,266 --> 00:02:48,266 that's been funded under the Space Tech Mission 48 00:02:48,266 --> 00:02:51,466 Directorate's game-changing development program. 49 00:02:51,466 --> 00:02:55,100 Also involved are-- 50 00:02:55,100 --> 00:02:57,266 with in-kind support 51 00:02:57,266 --> 00:03:01,966 as well as just collaborators, UCSC, 52 00:03:01,966 --> 00:03:04,000 the Massachusetts Institute of Technology Center 53 00:03:04,000 --> 00:03:09,066 for Bits and Atoms, and MOOG's base systems. 54 00:03:09,066 --> 00:03:15,366 I also want to acknowledge the two 55 00:03:15,366 --> 00:03:17,700 most significant mentors I've had here at Ames, 56 00:03:17,700 --> 00:03:20,200 in the short period of time I've been here so far, 57 00:03:20,200 --> 00:03:23,066 which is Harry Partridge and Sean Swei. 58 00:03:23,066 --> 00:03:28,033 And Sean Swei has also spent time sitting in the lab 59 00:03:28,033 --> 00:03:30,766 as well as all the faces that you see here, 60 00:03:30,766 --> 00:03:32,966 which are the minds 61 00:03:32,966 --> 00:03:34,600 and hands behind the work 62 00:03:34,600 --> 00:03:38,166 that you'll see in this presentation. 63 00:03:38,166 --> 00:03:42,433 And being here at NASA Ames Research Center 64 00:03:42,433 --> 00:03:45,133 has been really wonderful. 65 00:03:45,133 --> 00:03:48,733 You can see this is just some of all of the people here 66 00:03:48,733 --> 00:03:50,600 who have been involved in this work, 67 00:03:50,600 --> 00:03:52,833 and it hasn't been going on that long here. 68 00:03:52,833 --> 00:03:55,200 So you get an idea of the kind of support 69 00:03:55,200 --> 00:03:57,533 we have here at NASA Ames. 70 00:03:57,533 --> 00:03:59,200 So, thank you all. 71 00:03:59,200 --> 00:04:03,000 So I'm here today to talk about 72 00:04:03,000 --> 00:04:05,433 building blocks for aerostructures. 73 00:04:05,433 --> 00:04:08,866 And this comes from an intersection 74 00:04:08,866 --> 00:04:12,333 of digital materials, cellular solids, 75 00:04:12,333 --> 00:04:14,433 and fiber composites. 76 00:04:14,433 --> 00:04:16,700 And so what I aim to show is that it's possible 77 00:04:16,700 --> 00:04:18,333 to make things lighter, 78 00:04:18,333 --> 00:04:19,666 in a way, stronger, 79 00:04:19,666 --> 00:04:22,533 as well as possibly faster and cheaper 80 00:04:22,533 --> 00:04:24,600 by assembling composite materials 81 00:04:24,600 --> 00:04:26,900 from small, discrete parts. 82 00:04:26,900 --> 00:04:29,533 Short-term applications include infrastructure 83 00:04:29,533 --> 00:04:31,900 like buildings and bridges, 84 00:04:31,900 --> 00:04:33,900 or vehicles like airplanes. 85 00:04:33,900 --> 00:04:36,266 In all cases, the most significant benefit 86 00:04:36,266 --> 00:04:38,966 might be immaterial life cycles. 87 00:04:38,966 --> 00:04:42,800 So, the most obvious short-term benefit 88 00:04:42,800 --> 00:04:44,633 comes from the reduced weight. For example, 89 00:04:44,633 --> 00:04:46,700 less structural weight in an aircraft 90 00:04:46,700 --> 00:04:49,400 means less required fuel. 91 00:04:49,400 --> 00:04:53,900 Oops. Sorry--which itself means less weight, 92 00:04:53,900 --> 00:04:56,166 which means lower structural requirements 93 00:04:56,166 --> 00:04:57,666 and even less required fuel. 94 00:04:57,666 --> 00:05:00,500 So I'll say a bit about each of these areas, 95 00:05:00,500 --> 00:05:02,100 and then explain how it all fits together 96 00:05:02,100 --> 00:05:03,666 and show a couple of examples 97 00:05:03,666 --> 00:05:06,833 that we're pursuing as applications. 98 00:05:06,833 --> 00:05:09,366 So what are digital materials? 99 00:05:09,366 --> 00:05:11,566 Digital materials are made from a small 100 00:05:11,566 --> 00:05:13,700 and discrete set of parts, 101 00:05:13,700 --> 00:05:16,500 with discrete locations and discrete connections. 102 00:05:16,500 --> 00:05:18,300 They form as low-cost and reversible ways 103 00:05:18,300 --> 00:05:19,533 of making things, 104 00:05:19,533 --> 00:05:21,866 with low precision assembly requirements. 105 00:05:21,866 --> 00:05:24,200 Think about the precision of a 3-year-old, 106 00:05:24,200 --> 00:05:27,400 relative to the precision of the assemblies 107 00:05:27,400 --> 00:05:31,500 that he or she can make from, say, Lego bricks. 108 00:05:31,500 --> 00:05:33,933 So this is accomplished with aero reduction, 109 00:05:33,933 --> 00:05:35,966 tolerance, and correction mechanisms, 110 00:05:35,966 --> 00:05:37,800 which is just like what we see in digital 111 00:05:37,800 --> 00:05:39,966 communication and computation systems. 112 00:05:39,966 --> 00:05:42,166 So digital systems in general, 113 00:05:42,166 --> 00:05:44,466 as we understand them, employ many-- 114 00:05:44,466 --> 00:05:46,033 and as we've engineered them-- 115 00:05:46,033 --> 00:05:48,600 employ many fungible units to form assemblies 116 00:05:48,600 --> 00:05:51,033 with a wide range of size and function. 117 00:05:51,033 --> 00:05:52,666 All from low and finite sets of types 118 00:05:52,666 --> 00:05:54,333 of discrete building blocks. 119 00:05:54,333 --> 00:05:57,300 If the application of this to materials sounds 120 00:05:57,300 --> 00:05:58,800 to you a bit like biology, 121 00:05:58,800 --> 00:06:03,533 and Jacob helped to introduce this analogy, 122 00:06:03,533 --> 00:06:06,400 that's because it is indeed similar to our basic models 123 00:06:06,400 --> 00:06:09,033 for how materials are used in living organisms. 124 00:06:09,033 --> 00:06:10,266 So in biology, 125 00:06:10,266 --> 00:06:12,966 we see discrete assembly resulting in structures 126 00:06:12,966 --> 00:06:15,766 with a wide range of size and function in organisms 127 00:06:15,766 --> 00:06:17,200 as a whole, 128 00:06:17,200 --> 00:06:19,566 or specialized organs and tissues, 129 00:06:19,566 --> 00:06:21,300 all from a low and finite set of types 130 00:06:21,300 --> 00:06:22,866 of discrete building blocks. 131 00:06:22,866 --> 00:06:27,600 So the most, the simplest way 132 00:06:27,600 --> 00:06:29,800 you can look this is amino acids 133 00:06:29,800 --> 00:06:31,633 as building blocks. 134 00:06:31,633 --> 00:06:33,033 And so the diversity of things 135 00:06:33,033 --> 00:06:35,666 that can be made with this sort of combinatorial system, 136 00:06:35,666 --> 00:06:37,500 at various levels of hierarchy, 137 00:06:37,500 --> 00:06:38,800 is plainly evident 138 00:06:38,800 --> 00:06:40,933 and the small number of required building blocks, 139 00:06:40,933 --> 00:06:43,033 along with system scalability, 140 00:06:43,033 --> 00:06:45,433 has allowed synthetic biologists 141 00:06:45,433 --> 00:06:46,666 to make new systems 142 00:06:46,666 --> 00:06:48,866 from the same or similar material. 143 00:06:48,866 --> 00:06:51,466 It's amazing work with incredible rates of progress, 144 00:06:51,466 --> 00:06:53,066 some of which is going on here. 145 00:06:53,066 --> 00:06:58,166 For many of our current manufacturing applications, 146 00:06:58,166 --> 00:07:00,333 however, we may be able to take advantage 147 00:07:00,333 --> 00:07:02,900 of the ability to work at a much larger scale, 148 00:07:02,900 --> 00:07:05,466 and with many fewer building block types, 149 00:07:05,466 --> 00:07:09,500 or diversity of parameters and objective functions. 150 00:07:09,500 --> 00:07:13,700 So short of setting out to try and re-create life, 151 00:07:13,700 --> 00:07:16,533 which is being done 152 00:07:16,533 --> 00:07:20,700 in the synthetic biology world right now, 153 00:07:20,700 --> 00:07:25,300 we're looking at modular systems that, on the surface, 154 00:07:25,300 --> 00:07:26,800 appear not that different from those 155 00:07:26,800 --> 00:07:30,100 that have been studied for ages, 156 00:07:30,100 --> 00:07:32,400 but historically, 157 00:07:32,400 --> 00:07:34,600 these kinds of physical systems are often considered 158 00:07:34,600 --> 00:07:38,200 to be a mechanical compromise relative to conventional 159 00:07:38,200 --> 00:07:39,966 manufacturing methods. 160 00:07:39,966 --> 00:07:41,233 So there's a couple things 161 00:07:41,233 --> 00:07:43,566 I want to point out on this slide. 162 00:07:43,566 --> 00:07:45,133 The first is that, 163 00:07:45,133 --> 00:07:49,166 while it's sad that the Lego house got torn down, 164 00:07:49,166 --> 00:07:51,033 it's not sad that most of the parts 165 00:07:51,033 --> 00:07:53,033 got distributed amongst many children 166 00:07:53,033 --> 00:07:56,266 and therefore maintained most of their utility. 167 00:07:56,266 --> 00:07:59,766 It's harder to do that with your typical demolition job. 168 00:07:59,766 --> 00:08:01,900 So can I have a show of hands 169 00:08:01,900 --> 00:08:05,666 for how many of you played 170 00:08:05,666 --> 00:08:08,866 with toy sets like that? 171 00:08:14,133 --> 00:08:16,133 And so can I now have a show of hands 172 00:08:16,133 --> 00:08:17,900 for how many of you, 173 00:08:17,900 --> 00:08:21,166 given a set like the one in the middle, 174 00:08:21,166 --> 00:08:23,166 would only use the provided bricks 175 00:08:23,166 --> 00:08:27,233 to make the things shown on the cover of the box? 176 00:08:27,233 --> 00:08:33,333 So in a way, that set, 177 00:08:33,333 --> 00:08:36,633 for a wind turbine set, is very realistic 178 00:08:36,633 --> 00:08:38,000 because it comes with a truck 179 00:08:38,000 --> 00:08:40,033 that is almost as big as the turbine. 180 00:08:40,033 --> 00:08:43,166 And today's terrestrial wind turbines 181 00:08:43,166 --> 00:08:44,566 are indeed size-constrained by 182 00:08:44,566 --> 00:08:46,466 that truck, which is, in turn, 183 00:08:46,466 --> 00:08:48,600 constrained by the geometry of the roadway system. 184 00:08:48,600 --> 00:08:54,700 So we have wind turbines that are set by the roadways. 185 00:08:54,700 --> 00:08:56,966 So, the funny thing is that the Lego version 186 00:08:56,966 --> 00:08:58,933 doesn't actually need the enormous truck. 187 00:08:58,933 --> 00:09:01,200 And you can bet that there are some kids 188 00:09:01,200 --> 00:09:03,700 that have made some curiously big wind turbines, 189 00:09:03,700 --> 00:09:05,933 together with mysteriously small trucks, 190 00:09:05,933 --> 00:09:08,433 just like how the parts are meant to be exchanged 191 00:09:08,433 --> 00:09:11,533 among the hundred or so explicit designs 192 00:09:11,533 --> 00:09:14,200 in the reconfigurable plane sets. 193 00:09:14,200 --> 00:09:16,200 Okay? So, one of the things 194 00:09:16,200 --> 00:09:17,866 that make toys like this work so well 195 00:09:17,866 --> 00:09:21,566 is that the parts fit together into a regular lattice 196 00:09:21,566 --> 00:09:23,000 and when it comes to studying 197 00:09:23,000 --> 00:09:24,833 the mechanics of lattice structures, 198 00:09:24,833 --> 00:09:27,000 we have the science of cellular solids. 199 00:09:27,000 --> 00:09:29,033 And so this was forged in the '80s 200 00:09:29,033 --> 00:09:31,400 by Lorna Gibson and Michael Ashby, 201 00:09:31,400 --> 00:09:33,700 and it shows that we can treat these kind of materials 202 00:09:33,700 --> 00:09:36,166 as continuous materials with bulk properties. 203 00:09:36,166 --> 00:09:38,333 That really changes how you can design 204 00:09:38,333 --> 00:09:40,900 with these kinds of materials. 205 00:09:40,900 --> 00:09:42,666 And so the result has been the modern 206 00:09:42,666 --> 00:09:45,733 engineering use of materials 207 00:09:45,733 --> 00:09:48,133 with useful characteristics 208 00:09:48,133 --> 00:09:51,000 like lower density, greater elasticity, 209 00:09:51,000 --> 00:09:52,833 and better energy absorption 210 00:09:52,833 --> 00:09:54,400 than previously commonly used. 211 00:09:54,400 --> 00:09:57,366 So these include the foam in bicycle helmets, 212 00:09:57,366 --> 00:09:58,666 optimize uses for wood, 213 00:09:58,666 --> 00:10:00,600 and ultralight materials like aerogel. 214 00:10:00,600 --> 00:10:03,000 So the properties of any cellular solid 215 00:10:03,000 --> 00:10:05,066 are defined by the properties of the solid material 216 00:10:05,066 --> 00:10:06,433 that it's made from, 217 00:10:06,433 --> 00:10:10,300 and the geometric spatial configuration of that material. 218 00:10:10,300 --> 00:10:13,033 So if we're going to assemble a cellular solid 219 00:10:13,033 --> 00:10:15,100 from small parts, 220 00:10:15,100 --> 00:10:18,433 we could do whatever we wanted with a spatial configuration. 221 00:10:18,433 --> 00:10:22,600 So, what should we aim for? 222 00:10:22,600 --> 00:10:24,066 To get an idea of what we could try, 223 00:10:24,066 --> 00:10:26,033 we can look at a material property chart 224 00:10:26,033 --> 00:10:27,933 like those pioneered by Michael Ashby, 225 00:10:27,933 --> 00:10:30,366 and this shows most of the broad categories 226 00:10:30,366 --> 00:10:32,866 of materials that we use in engineering today. 227 00:10:32,866 --> 00:10:34,633 It has density along the X-axis 228 00:10:34,633 --> 00:10:37,000 and modular stiffness along the Y-axis. 229 00:10:37,000 --> 00:10:40,233 So metals and ceramics are the densest and stiffest 230 00:10:40,233 --> 00:10:43,100 and foams are the least dense and least stiff. 231 00:10:43,100 --> 00:10:45,100 If we were to make anything out of these materials, 232 00:10:45,100 --> 00:10:46,633 it's simple to see what the optimal 233 00:10:46,633 --> 00:10:48,066 performance could be 234 00:10:48,066 --> 00:10:50,033 by following the proportional modulus 235 00:10:50,033 --> 00:10:51,833 to density lines down the chart. 236 00:10:51,833 --> 00:10:53,200 So if you build something sparse 237 00:10:53,200 --> 00:10:55,466 that optimally uses any of these materials 238 00:10:55,466 --> 00:10:58,166 that you see in the bubbles 239 00:10:58,166 --> 00:11:00,100 to support the loads, 240 00:11:00,100 --> 00:11:03,266 with behavior that's dominated by stretching of the material, 241 00:11:03,266 --> 00:11:04,466 then the relative stiffness 242 00:11:04,466 --> 00:11:06,466 will stay proportional to the relative density, 243 00:11:06,466 --> 00:11:09,600 so it's gonna follow down along those lines. 244 00:11:09,600 --> 00:11:12,233 So this is sort of the absolute best case design goal, 245 00:11:12,233 --> 00:11:14,433 the physical boundary, if you will. 246 00:11:14,433 --> 00:11:15,966 So we therefore know 247 00:11:15,966 --> 00:11:17,933 that we should be able to access the area 248 00:11:17,933 --> 00:11:19,400 in yellow on this chart, 249 00:11:19,400 --> 00:11:21,366 by properly making a cellular solid 250 00:11:21,366 --> 00:11:24,700 out of a known constituent solid material. 251 00:11:24,700 --> 00:11:25,966 In practice, however, 252 00:11:25,966 --> 00:11:27,266 we've only followed these lines 253 00:11:27,266 --> 00:11:29,666 down about in order of magnitude and relative density. 254 00:11:29,666 --> 00:11:31,900 So we've had reasonable success 255 00:11:31,900 --> 00:11:33,133 getting down to around 1/10 256 00:11:33,133 --> 00:11:35,233 the volume of the materials 257 00:11:35,233 --> 00:11:38,233 that you see here-- most of them-- 258 00:11:38,233 --> 00:11:40,866 before reaching the limits of conventional 259 00:11:40,866 --> 00:11:42,266 manufacturing methods. 260 00:11:42,266 --> 00:11:46,366 So if the designs that are truly volumetric periodic structures, 261 00:11:46,366 --> 00:11:49,000 a pioneer of lattice-based metallic structures, 262 00:11:49,000 --> 00:11:51,600 Hayden Wiley, has said that it is not trivial 263 00:11:51,600 --> 00:11:54,333 to ensure that the increased cost of periodic cellular 264 00:11:54,333 --> 00:11:56,466 manufacturing processes is compensated 265 00:11:56,466 --> 00:11:57,933 for by the reduction in weight. 266 00:11:57,933 --> 00:12:01,766 So whatever cost gains you get in your application 267 00:12:01,766 --> 00:12:03,166 through the reduction and weight. 268 00:12:03,166 --> 00:12:05,333 So this presumably refers to the technologies 269 00:12:05,333 --> 00:12:06,733 like the cast aluminum structure 270 00:12:06,733 --> 00:12:08,033 that you see on the lower right. 271 00:12:08,033 --> 00:12:10,633 So architecture and civil engineering 272 00:12:10,633 --> 00:12:12,500 have employed space frame truss structures 273 00:12:12,500 --> 00:12:15,966 for a very long time in buildings and aircraft 274 00:12:15,966 --> 00:12:17,800 but have yet to implement the principal 275 00:12:17,800 --> 00:12:19,533 in a way that takes advantage of the ability 276 00:12:19,533 --> 00:12:23,133 to treat such structures as a continuum material. 277 00:12:23,133 --> 00:12:25,466 In any case, the appreciation for designs 278 00:12:25,466 --> 00:12:26,900 that have lots of similar elements, 279 00:12:26,900 --> 00:12:28,633 sharing structural duties is evident 280 00:12:28,633 --> 00:12:31,233 throughout our history of high performing structures. 281 00:12:31,233 --> 00:12:32,666 You think of the Eiffel Tower 282 00:12:32,666 --> 00:12:34,933 or you could think of the geodetic aircraft, 283 00:12:34,933 --> 00:12:38,666 the Vickers aircraft, from the '30s. 284 00:12:38,666 --> 00:12:41,300 So we're still left with the red area 285 00:12:41,300 --> 00:12:42,966 on this materials property graph, 286 00:12:42,966 --> 00:12:44,233 which is physically possible, 287 00:12:44,233 --> 00:12:45,866 but for which we still have yet 288 00:12:45,866 --> 00:12:49,666 to show suitable methods of fabrication. 289 00:12:49,666 --> 00:12:52,766 So we do have methods of creating cellular materials 290 00:12:52,766 --> 00:12:54,966 with greater reductions in mass density 291 00:12:54,966 --> 00:12:58,100 than those 1 to 10 that I just mentioned. 292 00:12:58,100 --> 00:12:59,633 So these, however, have the property 293 00:12:59,633 --> 00:13:01,433 that instead of material stretching, 294 00:13:01,433 --> 00:13:03,200 the phenomenon that governs their behavior 295 00:13:03,200 --> 00:13:04,566 under load is bending. 296 00:13:04,566 --> 00:13:07,000 This results in the bulk 297 00:13:07,000 --> 00:13:10,166 following reduced mechanical property laws, 298 00:13:10,166 --> 00:13:12,066 which is shown here. 299 00:13:12,066 --> 00:13:13,400 And many of these materials 300 00:13:13,400 --> 00:13:15,366 can be produced very efficiently. 301 00:13:15,366 --> 00:13:16,600 For instance, by bubbling 302 00:13:16,600 --> 00:13:19,766 or mixing a liquid state with air before hardening. 303 00:13:19,766 --> 00:13:21,300 It is these kinds of the materials 304 00:13:21,300 --> 00:13:23,866 that have achieved ultralight status 305 00:13:23,866 --> 00:13:26,433 at the extreme left-hand side of this chart. 306 00:13:26,433 --> 00:13:28,466 But this still leaves this entire 307 00:13:28,466 --> 00:13:29,966 red region of the chart behind, 308 00:13:29,966 --> 00:13:31,766 due to the effect of the beam-bending 309 00:13:31,766 --> 00:13:33,566 dominated behavior 310 00:13:33,566 --> 00:13:35,433 and the order of magnitude practical limit 311 00:13:35,433 --> 00:13:37,266 for the processes that produce materials 312 00:13:37,266 --> 00:13:38,600 with proportional scaling. 313 00:13:38,600 --> 00:13:41,266 So it's this region of material property space 314 00:13:41,266 --> 00:13:44,900 that we focused on with some success. 315 00:13:44,900 --> 00:13:47,100 So it's interesting because for many applications, 316 00:13:47,100 --> 00:13:49,400 it's not a strength or stiffness that matters 317 00:13:49,400 --> 00:13:50,866 or weight that matters, 318 00:13:50,866 --> 00:13:53,333 it's strength or stiffness per weight. 319 00:13:53,333 --> 00:13:55,400 So you can make something 320 00:13:55,400 --> 00:13:57,000 with the same stiffness per weight, 321 00:13:57,000 --> 00:13:58,633 but at a lower weight. 322 00:13:58,633 --> 00:14:01,200 Then it costs less to do things like move it around, 323 00:14:01,200 --> 00:14:03,866 keep it steady, heat it up or cool it down. 324 00:14:03,866 --> 00:14:06,133 So one current industry strategy 325 00:14:06,133 --> 00:14:09,133 for trying to get total systems into this regime 326 00:14:09,133 --> 00:14:11,300 is with fiber composites, 327 00:14:11,300 --> 00:14:13,066 which is a technology that's undergoing a bit 328 00:14:13,066 --> 00:14:15,366 of a Renaissance in prototyping capabilities 329 00:14:15,366 --> 00:14:18,733 and extensibility of manufacturing equipment. 330 00:14:18,733 --> 00:14:20,466 So, here we traditionally have a process 331 00:14:20,466 --> 00:14:22,233 where the manufacturing systems are larger 332 00:14:22,233 --> 00:14:24,300 and more complex than the product, 333 00:14:24,300 --> 00:14:25,966 which can be a problem when your product 334 00:14:25,966 --> 00:14:29,100 is as large and complex as a passenger aircraft. 335 00:14:29,100 --> 00:14:32,300 So this diagram is from a guide that does an admirable job 336 00:14:32,300 --> 00:14:35,466 of distilling the best practices for composites material design 337 00:14:35,466 --> 00:14:37,266 analysis and manufacturing, 338 00:14:37,266 --> 00:14:38,666 as well as testing, 339 00:14:38,666 --> 00:14:44,733 into a concise, 693-page handbook. 340 00:14:44,733 --> 00:14:48,800 Commercial aircraft are used as examples in this guide 341 00:14:48,800 --> 00:14:50,966 and new composite aircraft architecture can be partly 342 00:14:50,966 --> 00:14:53,033 considered as attempts to reduce the enormity 343 00:14:53,033 --> 00:14:56,400 of this pyramid of design analysis and testing. 344 00:14:56,400 --> 00:14:59,233 For example, we now have the cockpit 345 00:14:59,233 --> 00:15:00,933 and forward fuselage 346 00:15:00,933 --> 00:15:05,433 down behind the ninth passenger window. 347 00:15:05,433 --> 00:15:07,700 All the way from basically in the nose 348 00:15:07,700 --> 00:15:11,333 to behind the ninth passenger window, 349 00:15:11,333 --> 00:15:14,233 as a single piece manufactured as a single, 350 00:15:14,233 --> 00:15:16,200 monolithic carbon fiber winding. 351 00:15:16,200 --> 00:15:17,800 So this one part is about 1/6 352 00:15:17,800 --> 00:15:19,466 the size of the entire plane, 353 00:15:19,466 --> 00:15:21,533 and needs a tool or framework 354 00:15:21,533 --> 00:15:23,866 that's about 1/3 the size of the entire plane, 355 00:15:23,866 --> 00:15:25,266 which needs to go into an autoclave 356 00:15:25,266 --> 00:15:26,500 to cure the composite 357 00:15:26,500 --> 00:15:28,600 that's almost the size of the plane, 358 00:15:28,600 --> 00:15:30,300 and then get processed on a CNC mill 359 00:15:30,300 --> 00:15:33,066 that is also almost the size of the plane. 360 00:15:33,066 --> 00:15:34,566 When it comes to shipping the part 361 00:15:34,566 --> 00:15:36,133 to the final assembly plant, 362 00:15:36,133 --> 00:15:37,700 it has to go on a plane 363 00:15:37,700 --> 00:15:41,000 that's twice the size of a plane. 364 00:15:41,000 --> 00:15:42,566 This plane was custom-designed 365 00:15:42,566 --> 00:15:45,433 and built as part of the manufacturing process 366 00:15:45,433 --> 00:15:46,700 for another plane. 367 00:15:46,700 --> 00:15:49,300 So this isn't particularly scalable. 368 00:15:49,300 --> 00:15:54,766 By the way, not only one manufacturer has done this. 369 00:15:54,766 --> 00:15:57,933 This isn't the first time in history this has been done. 370 00:15:57,933 --> 00:16:00,100 But it isn't particularly scalable, 371 00:16:00,100 --> 00:16:02,566 though they've done an admirable job up to this point. 372 00:16:02,566 --> 00:16:05,666 But it's largely in the name of lighter weight. 373 00:16:05,666 --> 00:16:08,566 So what digital composites propose 374 00:16:08,566 --> 00:16:11,533 is to assemble structures from a lot of linked parts. 375 00:16:11,533 --> 00:16:13,666 The core concept is that 376 00:16:13,666 --> 00:16:16,600 if we can effectively couple loops of fibers, 377 00:16:16,600 --> 00:16:18,900 then a chain of discrete fiber composite parts 378 00:16:18,900 --> 00:16:21,400 can be close to the strength of a monolithic part, 379 00:16:21,400 --> 00:16:23,233 and would have advantages with manufacturing 380 00:16:23,233 --> 00:16:25,166 processes, serviceability, 381 00:16:25,166 --> 00:16:26,933 and reusability, 382 00:16:26,933 --> 00:16:29,233 in addition to the tunability and extensibility 383 00:16:29,233 --> 00:16:32,833 that are general goals of digital materials. 384 00:16:32,833 --> 00:16:34,433 And the tunability in particular 385 00:16:34,433 --> 00:16:38,800 is also a very specific goal of composites. 386 00:16:38,800 --> 00:16:42,300 And so, at the sort of larger scale than amino acids, 387 00:16:42,300 --> 00:16:44,433 of course, recent work has shown 388 00:16:44,433 --> 00:16:46,266 that a building block approach 389 00:16:46,266 --> 00:16:49,066 can be used to achieve novel material properties 390 00:16:49,066 --> 00:16:50,566 and cellular materials 391 00:16:50,566 --> 00:16:52,566 with centimeter scale lattice pitch. 392 00:16:52,566 --> 00:16:54,666 So these are essentially three-dimensional, 393 00:16:54,666 --> 00:16:56,600 optimized fiber composites. 394 00:16:56,600 --> 00:17:01,866 You get the unidirectional material performance scaling, 395 00:17:01,866 --> 00:17:05,633 but now in a quasi-isotropic cellular solid. 396 00:17:05,633 --> 00:17:09,700 So we've shown better specific modulus scaling 397 00:17:09,700 --> 00:17:11,933 than previously known ultralight materials, 398 00:17:11,933 --> 00:17:14,300 and this is possible because assembly allows us 399 00:17:14,300 --> 00:17:16,933 to produce geometries that have not been achieved 400 00:17:16,933 --> 00:17:19,266 with conventional ultralight processes. 401 00:17:19,266 --> 00:17:21,600 So, the strength of the material performed similarly 402 00:17:21,600 --> 00:17:23,866 to the expectations of other high-performance 403 00:17:23,866 --> 00:17:25,233 ultralight materials, 404 00:17:25,233 --> 00:17:28,033 and the stiffness is much better. 405 00:17:28,033 --> 00:17:30,133 So despite how light it is, 406 00:17:30,133 --> 00:17:32,800 you don't need much of this stuff to, for instance, 407 00:17:32,800 --> 00:17:35,300 support the weight of a person. 408 00:17:35,300 --> 00:17:37,600 And you'll see some application slides 409 00:17:37,600 --> 00:17:38,900 that demonstrate that. 410 00:17:38,900 --> 00:17:40,866 But we're not that interested in quasi-static 411 00:17:40,866 --> 00:17:42,966 applications here. 412 00:17:42,966 --> 00:17:44,666 There are many metrics that we can use 413 00:17:44,666 --> 00:17:46,366 to compare the usefulness of this material 414 00:17:46,366 --> 00:17:48,700 to what we use today and a simple example 415 00:17:48,700 --> 00:17:50,066 is a beam performance index, 416 00:17:50,066 --> 00:17:52,800 first described by Ashby and Cebon, 417 00:17:52,800 --> 00:17:55,066 which simply quantifies beam-bending stiffness. 418 00:17:55,066 --> 00:17:58,433 So if we're given description of a beam, 419 00:17:58,433 --> 00:18:00,800 a given cross-sectional mass distribution, 420 00:18:00,800 --> 00:18:02,533 or a second moment of inertia, 421 00:18:02,533 --> 00:18:04,533 this is governed by the square root of the material 422 00:18:04,533 --> 00:18:07,266 modulus divided by the mass density of the material. 423 00:18:07,266 --> 00:18:10,733 So, we can immediately recognize that exceeding 424 00:18:10,733 --> 00:18:12,733 the quadratic relative density 425 00:18:12,733 --> 00:18:15,366 to relative modulus of cellular solid scaling 426 00:18:15,366 --> 00:18:16,833 will have a very positive effect 427 00:18:16,833 --> 00:18:18,166 on beam performance. 428 00:18:18,166 --> 00:18:20,533 So by reducing the weight 429 00:18:20,533 --> 00:18:25,433 for these very specific performance metrics, 430 00:18:25,433 --> 00:18:27,733 we don't have to maintain ideal performance, 431 00:18:27,733 --> 00:18:30,733 as long as we don't go below the quadratic scaling 432 00:18:30,733 --> 00:18:33,266 to actually increase the functional performance 433 00:18:33,266 --> 00:18:34,766 in an application. 434 00:18:34,766 --> 00:18:38,500 Even for one as simple as a beam. 435 00:18:38,500 --> 00:18:42,633 So, here we see 436 00:18:42,633 --> 00:18:45,666 that we've achieved 437 00:18:45,666 --> 00:18:49,400 an experiment of better performance 438 00:18:49,400 --> 00:18:51,500 than conventional aerostructure materials. 439 00:18:51,500 --> 00:18:53,933 So this is based on the linear elastic regime, 440 00:18:53,933 --> 00:18:56,466 which is generally considered to be the useful bit. 441 00:18:56,466 --> 00:18:59,433 And when you exceed the linear elastic regime, 442 00:18:59,433 --> 00:19:04,733 we do see a non-dissipative super elastic mode. 443 00:19:04,733 --> 00:19:07,633 And so this-- 444 00:19:07,633 --> 00:19:10,666 I won't go too much into detail here, 445 00:19:10,666 --> 00:19:16,766 but what you see is 446 00:19:16,766 --> 00:19:18,466 there's this effective twisting, 447 00:19:18,466 --> 00:19:19,833 the free end node is transferring 448 00:19:19,833 --> 00:19:22,666 a substantial bending moment through these nodes, 449 00:19:22,666 --> 00:19:27,400 and you'll see from the strut spacing 450 00:19:27,400 --> 00:19:31,000 and the buckling of the struts 451 00:19:31,000 --> 00:19:34,700 that you have this coordinating twisting motion. 452 00:19:34,700 --> 00:19:36,166 And so throughout a large structure, 453 00:19:36,166 --> 00:19:39,100 this appears to develop into a coordinated buckling behavior 454 00:19:39,100 --> 00:19:40,900 that is something like a three-dimensional 455 00:19:40,900 --> 00:19:42,600 folding or pleating. 456 00:19:42,600 --> 00:19:44,933 This is also readily apparent in simulation work 457 00:19:44,933 --> 00:19:48,200 being done by Khan and Joseph. 458 00:19:48,200 --> 00:19:50,066 And the rotational displacements 459 00:19:50,066 --> 00:19:53,266 that develop for each node are anti-symmetric 460 00:19:53,266 --> 00:19:56,000 relative to most of its neighbors. 461 00:19:56,000 --> 00:19:58,000 A kind of unconstrained coordinated twisting 462 00:19:58,000 --> 00:19:59,933 stress response has been reported 463 00:19:59,933 --> 00:20:02,466 in two-dimensional surfaces, 464 00:20:02,466 --> 00:20:04,400 particularly some natural ones. 465 00:20:04,400 --> 00:20:06,233 And we're currently working to describe this 466 00:20:06,233 --> 00:20:08,766 for such a partially constrained condition 467 00:20:08,766 --> 00:20:10,566 in three-dimensional lattices. 468 00:20:10,566 --> 00:20:11,766 So what happens if we keep 469 00:20:11,766 --> 00:20:15,400 loading the material past this point? 470 00:20:15,400 --> 00:20:17,433 A significant perceived pain point 471 00:20:17,433 --> 00:20:18,966 with fiber composites 472 00:20:18,966 --> 00:20:21,566 is with failure modes of typical engineered parts. 473 00:20:21,566 --> 00:20:23,733 A caricature of the stress strain curve 474 00:20:23,733 --> 00:20:25,500 for a typical fiber composite part 475 00:20:25,500 --> 00:20:27,900 is a straight linear elastic line 476 00:20:27,900 --> 00:20:29,766 that abruptly ends in breakage. 477 00:20:29,766 --> 00:20:31,433 And what this means, in a practical sense, 478 00:20:31,433 --> 00:20:33,566 is that breakages can be accompanied by elastic 479 00:20:33,566 --> 00:20:36,500 unloading of a great deal of stored energy. 480 00:20:36,500 --> 00:20:38,600 And this can be quite dangerous. 481 00:20:38,600 --> 00:20:40,333 So now we still have a great deal of work 482 00:20:40,333 --> 00:20:41,500 to do on this topic, 483 00:20:41,500 --> 00:20:43,866 of failure mechanisms for digital composites, 484 00:20:43,866 --> 00:20:45,533 but early tests suggest the possibility 485 00:20:45,533 --> 00:20:47,000 of being able to tune materials 486 00:20:47,000 --> 00:20:48,966 to behave in new ways. 487 00:20:48,966 --> 00:20:53,133 So this is an open frontier in materials science. 488 00:20:53,133 --> 00:20:55,233 So why does this work? 489 00:20:55,233 --> 00:20:57,166 So it's possible that we have mechanics at work 490 00:20:57,166 --> 00:20:59,966 that can be described as stretch-bend coupling. 491 00:20:59,966 --> 00:21:02,800 And this has been described, 492 00:21:02,800 --> 00:21:09,400 not so much in the bulk material solids literature... 493 00:21:09,400 --> 00:21:13,566 but... 494 00:21:13,566 --> 00:21:15,100 once there's any tendency 495 00:21:15,100 --> 00:21:17,933 towards microstructural bending under load, 496 00:21:17,933 --> 00:21:20,366 for the kinds of materials 497 00:21:20,366 --> 00:21:22,366 that have been considered 498 00:21:22,366 --> 00:21:24,166 in the class of stochastic materials 499 00:21:24,166 --> 00:21:27,066 in the cellular solids literature, 500 00:21:27,066 --> 00:21:29,900 it seems that the geometries that result 501 00:21:29,900 --> 00:21:33,100 from the typical ways of making engineered foams, 502 00:21:33,100 --> 00:21:36,033 stochastic engineered foams, predispose the structure 503 00:21:36,033 --> 00:21:38,066 to fall into a bending-dominated behavior. 504 00:21:38,066 --> 00:21:40,933 So we don't really see this middle regime 505 00:21:40,933 --> 00:21:43,300 in most of engineered cellular solids. 506 00:21:43,300 --> 00:21:46,900 There just doesn't appear to be much observable middle ground. 507 00:21:46,900 --> 00:21:48,866 In contrast, studies of biomaterials 508 00:21:48,866 --> 00:21:52,233 such as intracellular meshworks of actin, 509 00:21:52,233 --> 00:21:54,366 microtubules and extracellular matrices 510 00:21:54,366 --> 00:21:57,166 of fibrin and collagen have shown mechanical performance 511 00:21:57,166 --> 00:22:00,166 that exceeds predictions based on framework rigidity theory. 512 00:22:00,166 --> 00:22:02,166 So these are the same theories 513 00:22:02,166 --> 00:22:06,033 that underpin the cellular solids work. 514 00:22:06,033 --> 00:22:08,433 And they use these rigidity criteria as bases 515 00:22:08,433 --> 00:22:10,700 for estimating the balance of degrees of freedom 516 00:22:10,700 --> 00:22:12,133 and constraints in a structure. 517 00:22:12,133 --> 00:22:14,133 So recent work suggests 518 00:22:14,133 --> 00:22:15,700 that departing from these principles 519 00:22:15,700 --> 00:22:17,566 by assuming that significant bending moments 520 00:22:17,566 --> 00:22:20,000 can be transmitted through the nodes in the structure, 521 00:22:20,000 --> 00:22:25,233 can result in effective models of these biomaterials. 522 00:22:25,233 --> 00:22:26,966 So we think, 523 00:22:26,966 --> 00:22:29,033 while many support this model 524 00:22:29,033 --> 00:22:30,700 for its simplicity with an understanding 525 00:22:30,700 --> 00:22:32,733 of the relative magnitude of the forces concerned 526 00:22:32,733 --> 00:22:38,633 when you're talking about biomaterials, 527 00:22:38,633 --> 00:22:41,766 and if in principle this kind of framework exists, 528 00:22:41,766 --> 00:22:44,233 then we should be able to create one with a macroscale 529 00:22:44,233 --> 00:22:46,833 and it looks like we have. 530 00:22:46,833 --> 00:22:48,866 The level of description that's being used 531 00:22:48,866 --> 00:22:52,700 for the biomaterials actually suits 532 00:22:52,700 --> 00:22:55,800 what we're doing perhaps even better. 533 00:22:55,800 --> 00:22:57,500 So if we apply a simple adjustment 534 00:22:57,500 --> 00:22:59,300 to the dimensional scaling arguments 535 00:22:59,300 --> 00:23:01,933 that so efficiently describe classical cellular solids, 536 00:23:01,933 --> 00:23:03,400 we produce a scaling rule 537 00:23:03,400 --> 00:23:05,000 for stretch-bend coupled lattices 538 00:23:05,000 --> 00:23:06,600 that's very close to the behavior 539 00:23:06,600 --> 00:23:10,633 that we actually see. 540 00:23:10,633 --> 00:23:14,233 And so we have these different lattice types 541 00:23:14,233 --> 00:23:16,966 that we've essentially been playing with, 542 00:23:16,966 --> 00:23:20,766 with building blocks that we produce. 543 00:23:20,766 --> 00:23:26,666 We snapped them together, initially by hand. 544 00:23:26,666 --> 00:23:29,266 And we have these different types of geometries 545 00:23:29,266 --> 00:23:30,533 that we're working with. 546 00:23:30,533 --> 00:23:32,033 There's nothing magical about any one 547 00:23:32,033 --> 00:23:33,600 of these particular geometries, 548 00:23:33,600 --> 00:23:35,033 there's just ones that work and ones 549 00:23:35,033 --> 00:23:37,600 that we can get to work together. 550 00:23:37,600 --> 00:23:39,466 And so, we can ask the question of what happens 551 00:23:39,466 --> 00:23:42,800 if we start mixing them? 552 00:23:42,800 --> 00:23:44,866 And so, to illustrate 553 00:23:44,866 --> 00:23:47,966 the sort of exponential fan-out and variability 554 00:23:47,966 --> 00:23:49,500 that you can get by doing this, 555 00:23:49,500 --> 00:23:50,966 this is an example of designer 556 00:23:50,966 --> 00:23:52,833 material properties from this approach. 557 00:23:52,833 --> 00:23:55,333 So each of these towers are built 558 00:23:55,333 --> 00:23:57,466 with the same set of components 559 00:23:57,466 --> 00:23:59,266 but compress or buckle in different ways, 560 00:23:59,266 --> 00:24:00,900 under identical boundary conditions, 561 00:24:00,900 --> 00:24:02,533 all as a result of differences 562 00:24:02,533 --> 00:24:06,266 in the ordering or spatial arrangement of the parts. 563 00:24:06,266 --> 00:24:12,300 So this becomes a jumping-off point for us, 564 00:24:12,300 --> 00:24:17,200 as an example of bringing it down to an application 565 00:24:17,200 --> 00:24:18,966 that utilizes the variability 566 00:24:18,966 --> 00:24:20,600 that you get as you mix these parts. 567 00:24:20,600 --> 00:24:22,266 So the key thing to understand here 568 00:24:22,266 --> 00:24:24,500 is that there's this global variability 569 00:24:24,500 --> 00:24:26,000 that we can produce by 570 00:24:26,000 --> 00:24:28,866 mixing a very small set of different types of parts, 571 00:24:28,866 --> 00:24:30,400 as long as we have a lot of them. 572 00:24:30,400 --> 00:24:32,133 And so it's exponential, the behaviors 573 00:24:32,133 --> 00:24:35,033 that we get are exponential 574 00:24:35,033 --> 00:24:36,233 with a number of parts 575 00:24:36,233 --> 00:24:39,133 that compose our system. 576 00:24:39,133 --> 00:24:45,433 So the first example is wings. 577 00:24:45,433 --> 00:24:53,933 If I can get this to play. 578 00:24:53,933 --> 00:24:56,433 So this shows, 579 00:24:56,433 --> 00:25:00,533 with the same models of the large lattices 580 00:25:00,533 --> 00:25:03,533 that I just showed, 581 00:25:03,533 --> 00:25:05,300 what kinds of deformations can be produced 582 00:25:05,300 --> 00:25:07,433 by simple actuators 583 00:25:07,433 --> 00:25:11,700 with just specified arrangement of stiffer 584 00:25:11,700 --> 00:25:14,433 and less stiff parts in a structure. 585 00:25:20,600 --> 00:25:22,333 And so we did some initial testing 586 00:25:22,333 --> 00:25:23,900 of different kinds of shapes 587 00:25:23,900 --> 00:25:27,066 going between different kinds of shapes, 588 00:25:27,066 --> 00:25:29,333 and seeing if in principle, 589 00:25:29,333 --> 00:25:30,933 they could hold up for instance, 590 00:25:30,933 --> 00:25:33,533 in a wind tunnel. 591 00:25:33,533 --> 00:25:35,333 And the results look promising enough 592 00:25:35,333 --> 00:25:39,400 that we went forward with the NARI 593 00:25:39,400 --> 00:25:43,833 Team Seedling project, 594 00:25:43,833 --> 00:25:48,933 which basically takes the pyramid of tests 595 00:25:48,933 --> 00:25:52,233 and takes out the middle. 596 00:25:52,233 --> 00:25:55,800 So the top of the pyramid of tests 597 00:25:55,800 --> 00:25:57,566 are the small coupon tests, 598 00:25:57,566 --> 00:26:00,866 and our parts look like those small coupons. 599 00:26:00,866 --> 00:26:03,833 And so we can go very quickly, 600 00:26:03,833 --> 00:26:07,266 as far as the entire process from design 601 00:26:07,266 --> 00:26:09,000 and manufacturing, 602 00:26:09,000 --> 00:26:11,033 from those small coupon tests, 603 00:26:11,033 --> 00:26:14,066 to a complete structure 604 00:26:14,066 --> 00:26:19,666 with apparently complex behavior. 605 00:26:19,666 --> 00:26:22,733 And so this shows the process... 606 00:26:32,633 --> 00:26:38,366 Including some testing of our skin structure. 607 00:26:42,300 --> 00:26:45,933 That was on campus, by the way. 608 00:26:45,933 --> 00:26:47,200 And so we can go through-- 609 00:26:47,200 --> 00:26:50,166 and the skins are also discrete units. 610 00:26:50,166 --> 00:26:53,433 And so this obeys the same sort of property 611 00:26:53,433 --> 00:26:56,500 of what we can do, in terms of the design and testing, 612 00:26:56,500 --> 00:26:58,666 and very quickly get to the wind tunnel 613 00:26:58,666 --> 00:27:00,133 model that you see here. 614 00:27:00,133 --> 00:27:02,933 And in fact, we had two. 615 00:27:02,933 --> 00:27:06,800 And we have the parts for another one. 616 00:27:06,800 --> 00:27:10,933 And so you see the bench testing that we've done 617 00:27:10,933 --> 00:27:12,733 to validate 618 00:27:12,733 --> 00:27:17,400 the behavior and the stiffness of the structure. 619 00:27:17,400 --> 00:27:20,933 And we've been able to bring 620 00:27:20,933 --> 00:27:25,533 the modeling down in the same way. 621 00:27:25,533 --> 00:27:29,466 So, here we're looking at vibrational modes, 622 00:27:29,466 --> 00:27:30,766 using a modeling strategy 623 00:27:30,766 --> 00:27:34,200 that actually obeys the same kind of principles 624 00:27:34,200 --> 00:27:37,233 as our fabrication strategy. 625 00:27:37,233 --> 00:27:38,633 And so what I mean by this is 626 00:27:38,633 --> 00:27:43,400 that a traditional finite element model 627 00:27:43,400 --> 00:27:44,900 will break down the geometry 628 00:27:44,900 --> 00:27:47,166 into a number of finite elements 629 00:27:47,166 --> 00:27:49,133 that compose your mesh. 630 00:27:49,133 --> 00:27:51,566 And what we've been able to show 631 00:27:51,566 --> 00:27:56,100 is that we can have a number of different mesh element 632 00:27:56,100 --> 00:28:00,200 types that equals the number of different parts of we have. 633 00:28:00,200 --> 00:28:02,600 And we can model and exhaustively test 634 00:28:02,600 --> 00:28:05,900 to tune the model of those individual parts, 635 00:28:05,900 --> 00:28:09,100 and then each one of those is a single simplex 636 00:28:09,100 --> 00:28:11,400 in the mesh for our simulation. 637 00:28:11,400 --> 00:28:15,033 And that allows us to do things, 638 00:28:15,033 --> 00:28:18,433 in terms of the simulation and design, 639 00:28:18,433 --> 00:28:21,666 essentially in real-time and with computational loads 640 00:28:21,666 --> 00:28:27,566 that vastly undercut traditional processes. 641 00:28:27,566 --> 00:28:31,466 And so all this work in the wind tunnel tests 642 00:28:31,466 --> 00:28:33,366 that you just saw videos of 643 00:28:33,366 --> 00:28:36,066 led up to a direct comparison 644 00:28:36,066 --> 00:28:38,900 between the white 3-D printed model 645 00:28:38,900 --> 00:28:40,333 you see on the bottom. 646 00:28:40,333 --> 00:28:42,400 It's not a biplane. 647 00:28:42,400 --> 00:28:44,266 They're just stacked on top of each other. 648 00:28:44,266 --> 00:28:47,633 And the twist model on top. 649 00:28:47,633 --> 00:28:51,166 And you can see on the bottom model 650 00:28:51,166 --> 00:28:53,766 that we have conventional ailerons 651 00:28:53,766 --> 00:28:56,400 that are replaceable on the bottom model. 652 00:28:56,400 --> 00:28:58,600 The bottom model, by the way, was 3-D printed 653 00:28:58,600 --> 00:29:04,766 by Gary Wainwright at Langley. 654 00:29:04,766 --> 00:29:06,900 And so we were able to do a direct comparison 655 00:29:06,900 --> 00:29:10,966 between these two 656 00:29:10,966 --> 00:29:12,233 and show the viability 657 00:29:12,233 --> 00:29:15,100 of the digital composite structure. 658 00:29:15,100 --> 00:29:20,033 So as opposed to a traditional wind tunnel model 659 00:29:20,033 --> 00:29:24,500 that is often built extremely stiff, 660 00:29:24,500 --> 00:29:29,800 sometimes out of a solid block of metal, 661 00:29:29,800 --> 00:29:34,433 we were able to use a structure that, 662 00:29:34,433 --> 00:29:37,066 in this case, weighs 100 grams on each side. 663 00:29:43,733 --> 00:29:47,466 And we had some fun with the wind tunnel tests. 664 00:29:47,466 --> 00:29:53,166 Here, we're driving it with a forced oscillation 665 00:29:53,166 --> 00:29:57,400 and also driving the wings 666 00:29:57,400 --> 00:29:59,566 to actually counter the motion 667 00:29:59,566 --> 00:30:03,300 of the forced oscillation of the wind tunnel rig. 668 00:30:03,300 --> 00:30:06,300 So this is the team, 669 00:30:06,300 --> 00:30:08,700 including the flight dynamics group at Langley, 670 00:30:08,700 --> 00:30:13,033 that was wonderful to work with. 671 00:30:13,033 --> 00:30:16,966 And so we moved forward with 100 grams per wing, 672 00:30:16,966 --> 00:30:19,800 and surviving to full pressure of the wind tunnel 673 00:30:19,800 --> 00:30:21,633 that we used, 674 00:30:21,633 --> 00:30:25,733 we thought, "Why not fly it?" 675 00:30:25,733 --> 00:30:33,266 And so we took the same wings and flew them. 676 00:30:37,366 --> 00:30:41,733 There was sound, but you can imagine the sound that it makes. 677 00:30:41,733 --> 00:30:43,066 Vroom! 678 00:30:49,566 --> 00:30:55,133 And that's a barrel roll, 679 00:30:55,133 --> 00:30:58,233 thanks to a very skilled pilot. 680 00:31:07,033 --> 00:31:10,333 And so what we're going forward with for this project, 681 00:31:10,333 --> 00:31:12,600 which is the Mission Adaptive Digital Composite 682 00:31:12,600 --> 00:31:14,200 Aerostructures Technologies project, 683 00:31:14,200 --> 00:31:17,033 or MADCAT for short, 684 00:31:17,033 --> 00:31:21,466 are what we're calling X to the N planes. 685 00:31:21,466 --> 00:31:23,900 And so you can see that, 686 00:31:23,900 --> 00:31:26,366 if we have the set of building blocks, 687 00:31:26,366 --> 00:31:28,933 that we can show can be rapidly integrated 688 00:31:28,933 --> 00:31:31,533 into a flying aircraft, 689 00:31:31,533 --> 00:31:34,833 and we can take that same set of building blocks 690 00:31:34,833 --> 00:31:37,100 and just by snapping them apart 691 00:31:37,100 --> 00:31:38,666 and snapping them back together again 692 00:31:38,666 --> 00:31:44,066 in a different configuration, 693 00:31:44,066 --> 00:31:46,200 meet very rapidly, 694 00:31:46,200 --> 00:31:49,300 through design and analysis, 695 00:31:49,300 --> 00:31:52,933 all of the goals of aircraft design 696 00:31:52,933 --> 00:31:55,766 analysis and manufacturing. 697 00:31:55,766 --> 00:31:58,566 That means that we can very quickly 698 00:31:58,566 --> 00:32:01,200 get into the air lots of different, 699 00:32:01,200 --> 00:32:04,633 maybe novel, maybe theorized in the literature, 700 00:32:04,633 --> 00:32:09,300 but difficult to fabricate aircraft types including, 701 00:32:09,300 --> 00:32:12,833 in particular, the morphing aircraft types. 702 00:32:12,833 --> 00:32:15,166 Because what we've been able to show 703 00:32:15,166 --> 00:32:18,766 is the ability to change 704 00:32:18,766 --> 00:32:20,966 where a stiffer part and a less stiff part 705 00:32:20,966 --> 00:32:22,600 is throughout the structure, 706 00:32:22,600 --> 00:32:24,800 in order to define global deformation modes 707 00:32:24,800 --> 00:32:27,000 with very simple actuation. 708 00:32:27,000 --> 00:32:34,233 There are only two servos on the twist wing 709 00:32:34,233 --> 00:32:35,733 that I just showed. 710 00:32:35,733 --> 00:32:38,766 And we think we can keep 711 00:32:38,766 --> 00:32:40,800 the number of actuators extremely low 712 00:32:40,800 --> 00:32:44,633 for essentially arbitrarily complex deformation modes, 713 00:32:44,633 --> 00:32:46,733 where the number of states relates 714 00:32:46,733 --> 00:32:48,333 to the number of actuators, 715 00:32:48,333 --> 00:32:53,433 not the complexity of the deformation mode. 716 00:32:53,433 --> 00:32:55,433 And the manufacturing techniques 717 00:32:55,433 --> 00:32:57,166 that we look to employ with these methods 718 00:32:57,166 --> 00:32:58,566 and that we're going forward with, 719 00:32:58,566 --> 00:33:01,533 are ones that suit mass manufacturing. 720 00:33:01,533 --> 00:33:04,700 Which means that the cost of all 721 00:33:04,700 --> 00:33:07,933 of the components can come very close 722 00:33:07,933 --> 00:33:11,900 to the theoretical energy cost of those components. 723 00:33:16,833 --> 00:33:18,600 And the long-term vision, 724 00:33:18,600 --> 00:33:22,633 if you have a material ecosystem 725 00:33:22,633 --> 00:33:25,900 that works this way, 726 00:33:25,900 --> 00:33:28,933 is the ability to take that fundamental energy 727 00:33:28,933 --> 00:33:33,833 cost and amortize it over many different applications, 728 00:33:33,833 --> 00:33:37,500 because the actual cost of assembling these things 729 00:33:37,500 --> 00:33:41,166 by snapping them together is extremely, extremely low. 730 00:33:41,166 --> 00:33:42,833 And so you can start to envision 731 00:33:42,833 --> 00:33:49,166 completely new ways of managing material. 732 00:33:49,166 --> 00:33:56,166 So this brings us to this question of-- 733 00:33:56,166 --> 00:33:59,933 I've brought this to this question of price-performance. 734 00:33:59,933 --> 00:34:01,466 And so to put this in perspective, 735 00:34:01,466 --> 00:34:03,666 we can briefly look at how this proposal relates 736 00:34:03,666 --> 00:34:07,066 to aerostructures fabrication as we know it. 737 00:34:07,066 --> 00:34:09,800 And so the closest example 738 00:34:09,800 --> 00:34:12,400 of an aerostructure built with the principle 739 00:34:12,400 --> 00:34:13,833 of distributing load requirements 740 00:34:13,833 --> 00:34:15,500 over many smaller parts is probably 741 00:34:15,500 --> 00:34:17,433 the Vickers geodetic aircraft. 742 00:34:17,433 --> 00:34:19,166 So one of them is a plane 743 00:34:19,166 --> 00:34:22,100 that still holds the record for distance flown nonstop 744 00:34:22,100 --> 00:34:23,633 by a single engine aircraft. 745 00:34:23,633 --> 00:34:26,133 Over 7,000 miles. 746 00:34:26,133 --> 00:34:29,500 So we can assume that this was not a heavy aircraft, obviously. 747 00:34:29,500 --> 00:34:32,533 So it also had the reputation of it 748 00:34:32,533 --> 00:34:34,733 being extraordinarily damage tolerant, 749 00:34:34,733 --> 00:34:37,400 and the unusual robustness came at a perceived 750 00:34:37,400 --> 00:34:40,133 cost of manufacturer ability. 751 00:34:40,133 --> 00:34:42,000 And the factory countered this perception 752 00:34:42,000 --> 00:34:43,600 was a much-publicized exercise 753 00:34:43,600 --> 00:34:45,500 where they built an entire aircraft from start 754 00:34:45,500 --> 00:34:47,366 to finish in 24 hours, 755 00:34:47,366 --> 00:34:51,033 actually taking off after that much time. 756 00:34:51,033 --> 00:34:52,733 There's not a lot of information out there 757 00:34:52,733 --> 00:34:54,800 on how many parts are in those aircraft. 758 00:34:54,800 --> 00:34:57,066 But the time of production is similar to today's 759 00:34:57,066 --> 00:34:58,666 commercial production capability 760 00:34:58,666 --> 00:35:03,533 for, say, a Boeing 737. 761 00:35:03,533 --> 00:35:05,633 So the current 737 fuselage 762 00:35:05,633 --> 00:35:08,066 is comprised of a couple hundred of thousand parts, 763 00:35:08,066 --> 00:35:11,300 an additional couple hundred of thousand fasteners, rivets, 764 00:35:11,300 --> 00:35:13,466 are used to attach these parts to each other. 765 00:35:13,466 --> 00:35:16,133 With the factory system of about 700 people, 766 00:35:16,133 --> 00:35:19,166 about 1 fuselage per day is completed. 767 00:35:19,166 --> 00:35:21,533 And therefore, we can estimate that on average, 768 00:35:21,533 --> 00:35:24,066 7,000 parts are added to the assembly per hour, 769 00:35:24,066 --> 00:35:25,933 or 10 per person, per hour, 770 00:35:25,933 --> 00:35:27,166 or about 2 per second. 771 00:35:27,166 --> 00:35:28,800 And estimating the total structural 772 00:35:28,800 --> 00:35:31,100 volume of the fuselage as a cylinder, 773 00:35:31,100 --> 00:35:35,000 4 meters in diameter, 40 meters long, 774 00:35:35,000 --> 00:35:36,333 25 centimeters thick, 775 00:35:36,333 --> 00:35:41,000 gives us about 120 cubic meters. 776 00:35:41,000 --> 00:35:43,966 And this would require about 900,000 parts, 777 00:35:43,966 --> 00:35:45,766 if we simply use the cuboct truss 778 00:35:45,766 --> 00:35:48,500 digital composite design presented earlier. 779 00:35:48,500 --> 00:35:50,733 So I know that trained students 780 00:35:50,733 --> 00:35:53,733 can place one of the cuboct truss parts 781 00:35:53,733 --> 00:35:56,566 every 5 minutes or faster. 782 00:35:56,566 --> 00:35:57,900 So with the same workforce, 783 00:35:57,900 --> 00:36:00,233 the job would get done in about five days. 784 00:36:00,233 --> 00:36:01,533 So we're not even talking about 785 00:36:01,533 --> 00:36:03,600 a necessarily great increase in parts 786 00:36:03,600 --> 00:36:05,166 to realize some of the advantages 787 00:36:05,166 --> 00:36:06,833 of digital construction. 788 00:36:06,833 --> 00:36:10,466 And the difference here is that for an entire plane, 789 00:36:10,466 --> 00:36:12,433 we would now have a few million of the same part, 790 00:36:12,433 --> 00:36:15,133 instead of around a million of custom parts, 791 00:36:15,133 --> 00:36:18,600 with around a million of the same rivet. 792 00:36:18,600 --> 00:36:21,200 So the life cycle implications are not trivial, 793 00:36:21,200 --> 00:36:23,266 from design through fabrication, 794 00:36:23,266 --> 00:36:25,166 to repair and reuse. 795 00:36:25,166 --> 00:36:28,766 It's easy to see the potential advantages. 796 00:36:28,766 --> 00:36:33,866 So this points out that your typical 797 00:36:33,866 --> 00:36:37,566 toy building block price per kilogram 798 00:36:37,566 --> 00:36:42,200 comes very close to what you would expect 799 00:36:42,200 --> 00:36:44,966 for the cost to produce the raw material 800 00:36:44,966 --> 00:36:49,600 that it's composed of. 801 00:36:49,600 --> 00:36:54,566 The next vertical line over are the materials 802 00:36:54,566 --> 00:37:01,033 that we're currently using for our experiments. 803 00:37:01,033 --> 00:37:04,533 And we can expect to come reasonably close 804 00:37:04,533 --> 00:37:08,666 to that cost, at scale, 805 00:37:08,666 --> 00:37:13,733 for producing arbitrarily large structures 806 00:37:13,733 --> 00:37:15,466 per kilogram. 807 00:37:15,466 --> 00:37:17,833 And so that's compared against things 808 00:37:17,833 --> 00:37:20,133 like tennis rackets and bicycles, 809 00:37:20,133 --> 00:37:22,100 consumer drones, 810 00:37:22,100 --> 00:37:25,533 which, I'm not sure why they're that expensive except 811 00:37:25,533 --> 00:37:30,200 that maybe that's what you all will pay for them. 812 00:37:30,200 --> 00:37:36,500 And then we have the newest composite aircraft 813 00:37:36,500 --> 00:37:39,066 and your phones, 814 00:37:39,066 --> 00:37:42,300 which actually cost more per unit kilogram. 815 00:37:42,300 --> 00:37:46,400 But, at least looks less than an orbital launch, 816 00:37:46,400 --> 00:37:49,133 except when you consider 817 00:37:49,133 --> 00:37:55,166 what SpaceX seems to be able to do, 818 00:37:55,166 --> 00:37:59,266 which is actually just under what your phone costs 819 00:37:59,266 --> 00:38:02,866 per kilogram. 820 00:38:02,866 --> 00:38:04,800 And so we can look 821 00:38:04,800 --> 00:38:09,933 at what this means for space applications, 822 00:38:09,933 --> 00:38:16,233 where this cannot only change how you get to space, 823 00:38:16,233 --> 00:38:19,433 but what you do once you get there. 824 00:38:19,433 --> 00:38:22,833 And so, what does it mean to be building things in space 825 00:38:22,833 --> 00:38:24,400 and what does it do for you? 826 00:38:24,400 --> 00:38:27,733 And so the launch environment, many of you know, 827 00:38:27,733 --> 00:38:29,866 almost always presents the largest 828 00:38:29,866 --> 00:38:34,033 and governing structural requirement for a spacecraft. 829 00:38:34,033 --> 00:38:35,800 And so using dimensional scaling arguments 830 00:38:35,800 --> 00:38:37,800 in the difference between the launch environment 831 00:38:37,800 --> 00:38:40,533 and, say, space station deployment, 832 00:38:40,533 --> 00:38:42,966 for an assembled cube set, 833 00:38:42,966 --> 00:38:45,466 you could see a factor of 10 difference 834 00:38:45,466 --> 00:38:48,300 in the low frequency accelerations, 835 00:38:48,300 --> 00:38:51,933 which can theoretically result in about a factor of 3 reduction 836 00:38:51,933 --> 00:38:54,366 in the required structural mass. 837 00:38:54,366 --> 00:38:55,766 And so this is even more pronounced 838 00:38:55,766 --> 00:38:58,466 when you're considering launch vibration, 839 00:38:58,466 --> 00:39:01,066 given the typical strategy for avoiding vibration-induced 840 00:39:01,066 --> 00:39:03,400 damage by keeping natural 841 00:39:03,400 --> 00:39:05,166 resin frequencies higher 842 00:39:05,166 --> 00:39:07,833 than the significant driving frequencies during launch. 843 00:39:07,833 --> 00:39:09,700 And so for these same cube sets, 844 00:39:09,700 --> 00:39:12,400 this could theoretically mean a hundredfold or more decrease 845 00:39:12,400 --> 00:39:13,866 in the required structural mass, 846 00:39:13,866 --> 00:39:17,633 depending on what you need to carry. 847 00:39:17,633 --> 00:39:20,666 So that's a pretty big deal. 848 00:39:20,666 --> 00:39:26,300 And the state-of-the-art energy 849 00:39:26,300 --> 00:39:30,100 to low Earth orbit right now, 850 00:39:30,100 --> 00:39:34,366 and that is based on the Falcon 9. 851 00:39:34,366 --> 00:39:37,000 The gap between that and what we know 852 00:39:37,000 --> 00:39:38,666 is the minimum theoretical energy 853 00:39:38,666 --> 00:39:40,100 if you were, say, 854 00:39:40,100 --> 00:39:43,400 reusing all of your launch mass-- 855 00:39:43,400 --> 00:39:47,300 not including the fuel-- 856 00:39:47,300 --> 00:39:48,866 is quite large, 857 00:39:48,866 --> 00:39:54,433 and actually spans the space 858 00:39:54,433 --> 00:39:58,600 of the production energy per unit kilogram 859 00:39:58,600 --> 00:40:01,466 of a lot of the materials that we use in space, 860 00:40:01,466 --> 00:40:04,700 or in engineered materials in general. 861 00:40:04,700 --> 00:40:08,066 Which means as we come down this chart 862 00:40:08,066 --> 00:40:09,566 with the right-hand line, 863 00:40:09,566 --> 00:40:11,833 it starts become much more sensible 864 00:40:11,833 --> 00:40:15,866 to be not only building 865 00:40:15,866 --> 00:40:17,633 and doing manufacturing 866 00:40:17,633 --> 00:40:19,700 in space in general, 867 00:40:19,700 --> 00:40:22,433 but to be looking for the materials 868 00:40:22,433 --> 00:40:26,166 to do so as well. 869 00:40:26,166 --> 00:40:30,066 Complete raw manufacturing in space. 870 00:40:30,066 --> 00:40:33,066 And so this is something 871 00:40:33,066 --> 00:40:37,366 that's a lot of progress has been made on, 872 00:40:37,366 --> 00:40:39,000 since the '80s, 873 00:40:39,000 --> 00:40:41,233 and we can leverage the progress-- 874 00:40:41,233 --> 00:40:46,700 the amazing progress done by NASA, 875 00:40:46,700 --> 00:40:50,466 to look forward to what this means 876 00:40:50,466 --> 00:40:53,833 when we know that we can now create these bulk materials 877 00:40:53,833 --> 00:40:57,266 and cellular solids out of composite materials. 878 00:40:57,266 --> 00:41:01,766 And now using potentially a much smaller 879 00:41:01,766 --> 00:41:03,266 building block, 880 00:41:03,266 --> 00:41:05,366 and taking advantage of all of the advances 881 00:41:05,366 --> 00:41:09,866 in robotics since then. 882 00:41:09,866 --> 00:41:12,000 And so here is the building block 883 00:41:12,000 --> 00:41:13,466 as what you see on the left, 884 00:41:13,466 --> 00:41:15,833 which itself is composed of building blocks, 885 00:41:15,833 --> 00:41:18,633 smaller building blocks. 886 00:41:18,633 --> 00:41:21,500 And they're injection molded fiber composites, 887 00:41:21,500 --> 00:41:25,866 and the tubes are pultruded carbon fiber. 888 00:41:25,866 --> 00:41:29,333 And these can be assembled into these structures, 889 00:41:29,333 --> 00:41:32,100 which we can evaluate on a load testing machine 890 00:41:32,100 --> 00:41:35,700 or by standing on them. 891 00:41:35,700 --> 00:41:39,633 And so, the bridge that you see below weighs 18 kilograms. 892 00:41:39,633 --> 00:41:43,933 And we fit it into two backpacks. 893 00:41:43,933 --> 00:41:48,500 And assembled it in a day. 894 00:41:48,500 --> 00:41:53,533 And then we disassembled it and made into a boat. 895 00:41:53,533 --> 00:41:59,600 And disassembled it again and made it into a structure. 896 00:41:59,600 --> 00:42:03,466 And so, some proof that it's not that heavy. 897 00:42:03,466 --> 00:42:13,400 And this is the structure, 898 00:42:13,400 --> 00:42:19,000 the small habitat structure that we built. 899 00:42:19,000 --> 00:42:23,133 And so the fungibility of this unit, these units, 900 00:42:23,133 --> 00:42:27,233 actually extends beyond mechanical systems. 901 00:42:27,233 --> 00:42:31,533 And so we've also been exploring 902 00:42:31,533 --> 00:42:36,233 what it means to combine the electrical systems, 903 00:42:36,233 --> 00:42:39,066 the power systems, 904 00:42:39,066 --> 00:42:41,700 and the communication systems, 905 00:42:41,700 --> 00:42:43,166 whatever protocols for actually 906 00:42:43,166 --> 00:42:48,633 running a mission on board a spacecraft. 907 00:42:48,633 --> 00:42:51,933 This is largely Daniel and Greenfield's work. 908 00:42:51,933 --> 00:42:58,066 Sitting in front here. Swell. 909 00:42:58,066 --> 00:43:01,533 This has a name for the larger project, 910 00:43:01,533 --> 00:43:05,500 which is OuroburoSat. 911 00:43:05,500 --> 00:43:09,900 And Daniel would be happy to discuss that name with you. 912 00:43:09,900 --> 00:43:13,733 And the OuroburoSat board 913 00:43:13,733 --> 00:43:15,266 is one of these white boards. 914 00:43:15,266 --> 00:43:18,166 And each one can carry a payload, 915 00:43:18,166 --> 00:43:22,033 and what you see here 916 00:43:22,033 --> 00:43:24,266 includes a coms payload on the one on the right 917 00:43:24,266 --> 00:43:25,900 and actually an experimental payload 918 00:43:25,900 --> 00:43:28,166 on the one on the left. 919 00:43:28,166 --> 00:43:31,666 And so these boards can all communicate with each other, 920 00:43:31,666 --> 00:43:35,233 not only share information, but share power. 921 00:43:35,233 --> 00:43:38,166 And so they can each handle power production 922 00:43:38,166 --> 00:43:41,500 with solar panels and power storage. 923 00:43:41,500 --> 00:43:46,033 And they're reconfigurable into different layouts. 924 00:43:46,033 --> 00:43:48,833 So we were lucky enough 925 00:43:48,833 --> 00:43:52,366 to be able to work with Marcus Murbach 926 00:43:52,366 --> 00:43:55,300 to get a flight. 927 00:43:55,300 --> 00:43:59,400 And the space that we were given was the shape of this, 928 00:43:59,400 --> 00:44:01,200 instead of the shape of this. 929 00:44:01,200 --> 00:44:04,866 So we just reconfigured it 930 00:44:04,866 --> 00:44:11,000 and flew that on board a sounding rocket. 931 00:44:13,266 --> 00:44:15,000 So there's obviously no underestimating 932 00:44:15,000 --> 00:44:18,466 what things you can learn 933 00:44:18,466 --> 00:44:20,500 from an actual flight experiment. 934 00:44:20,500 --> 00:44:23,033 And so going forward with that, 935 00:44:23,033 --> 00:44:27,500 this version shows the inner connectivity extending out, 936 00:44:27,500 --> 00:44:30,300 not just to a cube, but in three dimensions 937 00:44:30,300 --> 00:44:33,700 to fill a volume. 938 00:44:33,700 --> 00:44:37,033 And so where do we go from here? 939 00:44:37,033 --> 00:44:43,200 And so this is the direction that we're trying to head in. 940 00:44:43,200 --> 00:44:51,733 And what we think is that this is critical to getting here. 941 00:44:51,733 --> 00:44:55,000 So these are much larger structures 942 00:44:55,000 --> 00:44:57,766 than anything we've built in space, 943 00:44:57,766 --> 00:45:00,500 in fact, they're many orders of magnitude, 944 00:45:00,500 --> 00:45:02,966 more mass, 945 00:45:02,966 --> 00:45:09,100 than everything we've put in space combined. 946 00:45:09,100 --> 00:45:11,000 But there's no physical principle 947 00:45:11,000 --> 00:45:13,966 saying that these aren't possible. 948 00:45:13,966 --> 00:45:15,466 It's just that our current methods 949 00:45:15,466 --> 00:45:21,133 for producing space hardware don't suit. 950 00:45:21,133 --> 00:45:24,900 And so what does it mean to have something, 951 00:45:24,900 --> 00:45:28,966 to have a system that doesn't require your factory 952 00:45:28,966 --> 00:45:31,833 to be larger than the thing you're making? 953 00:45:31,833 --> 00:45:34,433 And so we think that we may be able 954 00:45:34,433 --> 00:45:39,233 to make progress in this direction. 955 00:45:39,233 --> 00:45:42,666 And if this seems far-fetched, 956 00:45:42,666 --> 00:45:46,100 I like the example of "The Brick Moon," 957 00:45:46,100 --> 00:45:49,066 which is a short story by Edward Everett Hale, 958 00:45:49,066 --> 00:45:51,500 published in 1869, 959 00:45:57,333 --> 00:46:02,033 which is said to be the first historical proposal 960 00:46:02,033 --> 00:46:04,333 for a satellite. 961 00:46:04,333 --> 00:46:08,966 And the story goes that it was an accident 962 00:46:08,966 --> 00:46:12,233 that people were still on the brick moon 963 00:46:12,233 --> 00:46:14,066 before it got sent up, 964 00:46:14,066 --> 00:46:15,900 and they started a colony up there. 965 00:46:15,900 --> 00:46:19,633 And this was completely fanciful in 1869. 966 00:46:19,633 --> 00:46:25,166 That wasn't that long ago. 967 00:46:25,166 --> 00:46:31,066 So I want to leave plenty of time for questions. 968 00:46:31,066 --> 00:46:35,133 So I'm going to leave it at that. 969 00:46:35,133 --> 00:46:37,333 And open it up to questions. 970 00:46:37,333 --> 00:46:40,333 [applause] 971 00:46:45,833 --> 00:46:47,900 - If you have a question, please raise your hand 972 00:46:47,900 --> 00:46:49,500 and wait for the microphone 973 00:46:49,500 --> 00:46:51,533 and ask a single question at a time. 974 00:46:51,533 --> 00:46:52,533 Thanks. 975 00:47:02,566 --> 00:47:06,866 - Okay, so, imagine you're on a space mission 976 00:47:06,866 --> 00:47:10,466 and you need to have the ability 977 00:47:10,466 --> 00:47:12,533 to make repairs and do maintenance. 978 00:47:12,533 --> 00:47:16,300 And so there's the philosophy-- 979 00:47:16,300 --> 00:47:21,700 or an approach of carrying a lot of the modular parts, 980 00:47:21,700 --> 00:47:25,433 which are nearly all identical like you describe, 981 00:47:25,433 --> 00:47:27,200 or you could be even more fundamental 982 00:47:27,200 --> 00:47:29,966 and carry the completely raw materials 983 00:47:29,966 --> 00:47:32,700 and 3-D print parts as you need them. 984 00:47:32,700 --> 00:47:34,566 And so have you given some thought 985 00:47:34,566 --> 00:47:36,133 about how you allocate those resources 986 00:47:36,133 --> 00:47:40,966 in those directions? 987 00:47:40,966 --> 00:47:45,766 - It's a very good question. 988 00:47:45,766 --> 00:47:48,366 And in particular because we are doing 989 00:47:48,366 --> 00:47:51,866 so much work on 3-D printing, 990 00:47:51,866 --> 00:48:01,166 NASA as a whole and it's state-of-the-art in general. 991 00:48:01,166 --> 00:48:02,700 I think there's some question 992 00:48:02,700 --> 00:48:04,566 as to how you define 3-D printing 993 00:48:04,566 --> 00:48:07,866 and what counts is that additive manufacturing. 994 00:48:07,866 --> 00:48:10,300 In a way, what I'm proposing was snapping together-- 995 00:48:10,300 --> 00:48:12,333 what we're proposing was snapping together 996 00:48:12,333 --> 00:48:16,500 building blocks here, is additive manufacturing. 997 00:48:16,500 --> 00:48:19,833 So in that sense, it's one in the same. 998 00:48:19,833 --> 00:48:24,000 In terms of what we think of as 3-D printing, 999 00:48:24,000 --> 00:48:26,666 in terms of the 3-D printing devices 1000 00:48:26,666 --> 00:48:32,500 that we see for instance, in the Space Shop. 1001 00:48:32,500 --> 00:48:36,500 One thing you can look at is the energy cost. 1002 00:48:36,500 --> 00:48:39,133 And so, it becomes a question of 1003 00:48:39,133 --> 00:48:44,433 what your energy sources are. 1004 00:48:44,433 --> 00:48:48,300 By and large, when you look at the total energy cost 1005 00:48:48,300 --> 00:48:52,600 to arrive at a piece of hardware, 1006 00:48:52,600 --> 00:48:55,466 it appears that the majority of that is worked up 1007 00:48:55,466 --> 00:48:58,566 in the raw materials production. 1008 00:48:58,566 --> 00:49:01,900 And a large proportion 1009 00:49:01,900 --> 00:49:05,666 of that you can arrive 1010 00:49:05,666 --> 00:49:08,800 at or estimate just by, 1011 00:49:08,800 --> 00:49:10,200 for instance, 1012 00:49:10,200 --> 00:49:13,800 needing to overcome latent heating fusion. 1013 00:49:13,800 --> 00:49:21,833 So needing to liquefy any given material. 1014 00:49:21,833 --> 00:49:26,500 And since that dominates the production cost, 1015 00:49:26,500 --> 00:49:32,233 processes that have to re-liquefy the material 1016 00:49:32,233 --> 00:49:35,566 are sort of gonna double your base 1017 00:49:35,566 --> 00:49:36,866 material production cost, 1018 00:49:36,866 --> 00:49:38,966 or significantly add to it. 1019 00:49:38,966 --> 00:49:43,566 And so, the versatility is undeniable 1020 00:49:43,566 --> 00:49:46,733 and it's moving towards-- 1021 00:49:46,733 --> 00:49:50,966 of what we now think of as conventional 3-D printing, 1022 00:49:50,966 --> 00:49:55,633 it's moving towards being able to produce 1023 00:49:55,633 --> 00:49:59,000 well-characterized parts. 1024 00:49:59,000 --> 00:50:01,200 But as it is, in a situation 1025 00:50:01,200 --> 00:50:08,500 where you have a very strict energy limit, 1026 00:50:08,500 --> 00:50:11,800 it can be difficult to justify some of the processes 1027 00:50:11,800 --> 00:50:15,666 that are currently used for 3-D printing. 1028 00:50:15,666 --> 00:50:18,500 And so, some of this comes down 1029 00:50:18,500 --> 00:50:20,666 to a question of whether 1030 00:50:20,666 --> 00:50:23,100 or not you can even tell the difference. 1031 00:50:23,100 --> 00:50:27,766 And so this is a question of pixelating a picture, 1032 00:50:27,766 --> 00:50:29,366 whether or not a digital picture 1033 00:50:29,366 --> 00:50:31,966 is sufficient versus 1034 00:50:31,966 --> 00:50:34,200 what you would term an analog picture. 1035 00:50:34,200 --> 00:50:37,000 But some point, when you get down to the molecules, 1036 00:50:37,000 --> 00:50:38,800 your color molecules in the picture, 1037 00:50:38,800 --> 00:50:43,200 you could argue that you have a digital picture anyway. 1038 00:50:43,200 --> 00:50:46,000 And so the only fundamental difference 1039 00:50:46,000 --> 00:50:49,700 is your conventionally- considered digital picture 1040 00:50:49,700 --> 00:50:54,200 has the pixels on a regular lattice. 1041 00:50:54,200 --> 00:51:00,700 And so what we've found with digital photography 1042 00:51:00,700 --> 00:51:05,133 and digital images that there are all of these advantages. 1043 00:51:05,133 --> 00:51:08,033 You can do compression, 1044 00:51:08,033 --> 00:51:13,466 you can operate over the actual functional image 1045 00:51:13,466 --> 00:51:15,166 in terms of the shapes that are on the image, 1046 00:51:15,166 --> 00:51:18,366 with algorithms, extremely efficiently. 1047 00:51:19,766 --> 00:51:24,933 And we think that the same thing applies 1048 00:51:24,933 --> 00:51:27,266 to three-dimensional systems. 1049 00:51:27,266 --> 00:51:30,033 Where if you're defining three-dimensional systems 1050 00:51:30,033 --> 00:51:31,833 in terms of voxels, 1051 00:51:31,833 --> 00:51:34,633 for given applications there's a minimum resolution 1052 00:51:34,633 --> 00:51:37,833 that you need, just like with pictures. 1053 00:51:37,833 --> 00:51:39,200 And as long as you meet that, 1054 00:51:39,200 --> 00:51:41,900 then there could be all of these advantages 1055 00:51:41,900 --> 00:51:45,700 to being able to address things 1056 00:51:45,700 --> 00:51:48,900 in terms of the discrete states. 1057 00:51:48,900 --> 00:51:52,200 And so I think there are these advantages 1058 00:51:52,200 --> 00:51:53,500 in terms of scope, 1059 00:51:53,500 --> 00:51:56,400 of the kinds of things you can do, 1060 00:51:56,400 --> 00:51:58,400 if your systems are built in this 1061 00:51:58,400 --> 00:52:02,533 fundamentally reconfigurable way versus 1062 00:52:02,533 --> 00:52:07,066 what's right now considered as 3-D printing, 1063 00:52:07,066 --> 00:52:10,200 which sort of compresses a lot 1064 00:52:10,200 --> 00:52:12,966 of the industrial infrastructure 1065 00:52:12,966 --> 00:52:14,933 producing hardware 1066 00:52:14,933 --> 00:52:19,566 as we know it into a small box. 1067 00:52:19,566 --> 00:52:21,200 Which is amazing. 1068 00:52:21,200 --> 00:52:23,333 And really useful. 1069 00:52:23,333 --> 00:52:30,066 But the extensibility of the digital approach 1070 00:52:30,066 --> 00:52:35,000 I think is very promising and interesting. 1071 00:52:35,000 --> 00:52:36,500 Thanks. 1072 00:52:36,500 --> 00:52:37,700 Yes. 1073 00:52:43,000 --> 00:52:45,366 - For the snap together structures that you've shown, 1074 00:52:45,366 --> 00:52:48,166 would it be reasonable to eventually have automated 1075 00:52:48,166 --> 00:52:51,500 assembly and reassembly? 1076 00:52:51,500 --> 00:52:53,400 - That's an easy one. 1077 00:52:53,400 --> 00:52:56,233 Yes. 1078 00:52:56,233 --> 00:52:58,366 And that's what we're working on right now, 1079 00:52:58,366 --> 00:53:04,400 that's a major focus of our work in the lab right now. 1080 00:53:04,400 --> 00:53:08,766 And we're talking about robots that are actually quite simple. 1081 00:53:08,766 --> 00:53:11,933 You don't need many, many degrees of freedom, 1082 00:53:11,933 --> 00:53:17,366 but you could have many robots. 1083 00:53:17,366 --> 00:53:19,700 - Hi, for the structures you were showing, 1084 00:53:19,700 --> 00:53:21,233 that were composite materials, 1085 00:53:21,233 --> 00:53:24,900 they weren't separated in discrete layers 1086 00:53:24,900 --> 00:53:27,300 but more combined in a marble structure. 1087 00:53:27,300 --> 00:53:29,366 I was wondering how is that determined 1088 00:53:29,366 --> 00:53:31,400 if you were measuring the force loads 1089 00:53:31,400 --> 00:53:34,566 or just have a random algorithm 1090 00:53:34,566 --> 00:53:38,500 that determines how that is made? 1091 00:53:38,500 --> 00:53:41,833 - So, the one where I showed 1092 00:53:41,833 --> 00:53:45,466 the blue low connectivity structure 1093 00:53:45,466 --> 00:53:50,366 and the red denser structure 1094 00:53:50,366 --> 00:53:54,300 and then the sort of marbled one, that one was random. 1095 00:53:54,300 --> 00:53:56,800 But it's a very good question 1096 00:53:56,800 --> 00:53:58,566 and it points to the direction 1097 00:53:58,566 --> 00:54:04,566 that we're headed in at that point, 1098 00:54:04,566 --> 00:54:08,933 which is to look at how to optimize the distribution. 1099 00:54:08,933 --> 00:54:12,100 And Holly, I'm not sure she's here right now, 1100 00:54:12,100 --> 00:54:17,000 Holly Jackson has done some work on exactly that question. 1101 00:54:17,000 --> 00:54:21,500 Of not only the question between the red 1102 00:54:21,500 --> 00:54:23,566 and the blue ones was sort of, 1103 00:54:23,566 --> 00:54:25,866 if you have a stiffer one and a less stiff one. 1104 00:54:25,866 --> 00:54:28,400 And the most basic version of that question is, 1105 00:54:28,400 --> 00:54:31,966 do you have a part or not in that location? 1106 00:54:31,966 --> 00:54:34,666 And so, it re-addresses 1107 00:54:34,666 --> 00:54:38,400 the topological optimization problem 1108 00:54:38,400 --> 00:54:42,166 within the scope of the finite element 1109 00:54:42,166 --> 00:54:44,333 tools in a way 1110 00:54:44,333 --> 00:54:46,200 that takes advantage of our ability 1111 00:54:46,200 --> 00:54:49,066 to modularize the finite element process as well. 1112 00:54:49,066 --> 00:54:50,933 So we can look at some of these incredibly 1113 00:54:50,933 --> 00:54:57,166 computational-intensive topological optimization methods 1114 00:54:57,166 --> 00:55:01,700 and apply them to something that would have been also 1115 00:55:01,700 --> 00:55:04,333 an incredibly computationally-intensive 1116 00:55:04,333 --> 00:55:07,533 finite element analysis problem. 1117 00:55:07,533 --> 00:55:10,666 But we can cut most of the computational intensiveness 1118 00:55:10,666 --> 00:55:12,433 out of the finite element analysis problem 1119 00:55:12,433 --> 00:55:16,066 and just focus on the optimization problem. 1120 00:55:16,066 --> 00:55:21,266 And so that's the crux of the approach 1121 00:55:21,266 --> 00:55:24,000 to the X to the N planes, in fact. 1122 00:55:24,000 --> 00:55:25,466 Because what we'd like to be able to do 1123 00:55:25,466 --> 00:55:28,433 is define the overall boundary shape 1124 00:55:28,433 --> 00:55:31,166 and the deformation modes 1125 00:55:31,166 --> 00:55:32,433 and then really optimize 1126 00:55:32,433 --> 00:55:34,066 how the parts are placed within that, 1127 00:55:34,066 --> 00:55:39,666 in order to meet that objective. 1128 00:55:39,666 --> 00:55:42,466 - So, please join me in thanking