Creating Efficient and Beautiful Structures

Xinghuo Yu:My name is Xinghuo Yu, I'm the chair of RMIT Professorial Academy. I'm chairing this distinguished lecture as well. This is just the one, the Academy was established in 2019, consisting of 20 RMIT distinguished professors who are the prominent researchers, and academic in the university. The role of us is certainly there's three roles, we act like a think tank advising university on the strategies, we promote universities and also we represent universities, so today is the ... Lastly, we have two distinguished lectures and this is the first in 2018, so we're very glad to have distinguished professor Mike Xie, to give us a talk. So, this is going to be a very interesting, when Mike sent email to me say, “Hey, here's the talk I'm going to say.” And I looked at I say, “Can you make it as simple as possible? Because we have a very different, wide range of ... People have different knowledge.” So, he made it, so I think the title is great. Make something beautiful, we note when something is beautiful, is expensive as well, but he's also trying to make it efficient.

Xinghuo Yu:So, I just have a very quick introduction of Mike. Mike is Distinguished Professor in RMIT, he's very well known around the world. If you look at his CV, he received a number of very significant awards, for example, the 2017 Clunies Ross Innovation Award from Australia Academy of Technological Science and Engineering. He also was recipient of 2017 Michelle medal by Engineer Australian for his work in mechanical engineering, which surprising is ... I was surprised you're civil engineer, right? But now he got a mechanical engineer award.

Xinghuo Yu:And he's done many work more than, I think one of the things I found Mike is actually a great example, also translation from research to industry application, and vice versa, which he has done very well. So, I think without any delay, let's welcome Mike to deliver his lecture.

Mike Xie:Thank you Xinghou for your introduction, and thank you all for coming all the way to this lecture. It's a great honor to present the first distinguished lecture of RMIT University. I grew up in a small town in Jianzhu Tanzhou. This is what it looks like where I grew up. This is a grand canal connecting all the way from Beijing to Hangzhou. At that time, I wasn't appreciating how lucky I was to live in such a beautiful environment. My parents always ask us to study hard, to go to big cities, to look for opportunities, which I did. The first city I went was Shanghai. This is a photo of Shanghai, and you'll see this kind of buildings, Myers and Myers. And then, I went to London in Britain to study, come to Sydney. My first job was in University Sydney. On the left hand side you'll see ... Some of you may recognize this building. To me, it's the ugliest building in Sydney, this is the UTS building on Broadway when I walked home every night, I would see this building.

Mike Xie:And then, I came to Melbourne. I was living in Footscray, and I'd drive to work. I would see these two buildings in [race coast view 00:04:08] I think some of you know these buildings. So, these are the kind of buildings we see in big cities, and after 40, 50 years ... Nowadays when I go home, I really appreciate how lucky I was. And on the other side, when China become rich and we are doing something different so, because of the wealth generated and the exuberance or this rational exuberance, we are creating [monsters buildings 00:04:45] like this, and in the newspaper we were actually imposing how much steel we used to construct this world record stadium, and there's other buildings created. So, on the left is a CCTV Tower of China, on the right hand side it's the headquarter of People's Daily, and so, it resembles some human organs rather than [inaudible 00:05:19] the buildings.

Mike Xie:So, what I'm going to share with you today is how we can ... Because as engineers we always ask us questions like how we can create some beautiful structures, and also it's efficient. So, not only can we save the materials, we can also create something interesting and beautiful.

Mike Xie:And one of the contributions I made to this field, I'm an Engineer, but I have many friends in architecture. So, we come up with a very simple idea, say, if you want to design a better structure, you can actually start from any silly design, and look at what you have got currently and remove those unneeded parts. Those most inefficient parts you can gradually remove from your design, and then gradually you will end up with a structure, that structure is not only efficient, it looks very beautiful just like what you see in nature, I'll show you some examples soon. The advantages of our simple idea, it's so easy to understand. I'm sure when I show you examples you will see why we are getting this kind of shapes. And also it can be implemented very easily. When people read our papers they can spend a few hours then they can do exactly what we do on the computers.

Mike Xie:So, I'll show you the first example, say, if we want to design a structure for these given conditions, there's one point fixed, I apply gravity, I want to know what would be the best shape to satisfy these given conditions. If we don't know the answer, it can not matter, we can just start from this square shape, and then we can do a simple analysis of the structure. We would find the stress distribution of the structure, fix here apply gravity, and these purple areas, they have no stress. It indicates these corners, they're not useful from the structure point of view. If you remove these parts, it would not affect your structure performance. If they are not useful, why would you keep them.

Mike Xie:So, you can delete all these inefficient materials step by step, and you come up with something more efficient, and it's interesting, looks like apple. It can continue to evolve, and you get a small tiny cherry. It's just a coincidence, and when I published this work, a professor in Cambridge, he said, “Ah, I can do better.” Because I only use one material in my model, so he started to put a different material in the core and by changing the relative stiffness, he got a series of pairs and many other things.

Mike Xie:We are not just getting some beautiful pictures, but if we compare the two designs on the left, you can see different colors throughout the whole structure. On the right hand side, we see the same color on the surface. So, we got a design, which has uniform surface stress. Because this is the criteria we use, if you have low stress we delete it. And if you use different rules to satisfy different design objective, you will get different shapes. Because the idea is so simple, it has been utilized by many engineers, and architects, material scientists for many different applications. So, you can design a building by controlling the drift of a building, the top of the building or it can change the vibration frequency for buildings and the earthquake or wind We also did a project with Boeing several years ago for maximizing the buckling load of composite aircraft wing. And we are lucky to have another Boeing student in the audience. James is working with us, extending this work to the aircraft design.

Mike Xie:So, I'm going to show you some of the examples we have done. So, when you design a structure, it really depends on what material you use, you will get totally different shape. Some of you might have seen the stair cable bridges. These cables that are good for attention, the cables cannot sustain any compression. So, if your material is suitable for attention then, your final structure should be predominantly intention. I show you one example, you'll understand how it goes, because your final structure needs to be intentioned. You can start again from any silly design, and then every time, every step you remove the areas with the highest compressive stress, and you will end up with a material only intention.

Mike Xie:So, this is a very simple example, so, I fix two points and I apply the gravity, and I want to find a shape which would be suitable for a stay cable or some membrane structures. So, every time I remove the areas with highest compressive stress, then I will get a catenary shape. So, when you hold a metal chain, that shape is a catenary and you would only have tension throughout whole structure. Not only have I got a shape for the structure, it also indicates where you need more materials so, on the two sides, because it's supporting half of the weight of the structure, you have more materials here then at the bottom, here the forces are very small so, the relative size is also small.

Mike Xie:If you see around us, most of our structures are made of concrete, the concrete is good for compression. So, if you want to design a structure for bricks, masonry stones, and you want your structure to be in compression, and again you can start from any design and then, every time you remove the areas with the highest tension stress so remove all the areas in tension, you will left with a structure, which will be in compression. And when I was doing this work about 10 or 12 years ago, I was very lucky that I have a colleague, Professor Mark Burry, some of us know him. He's a very well known architect, he's a Chief Executive Architect for this famous building in Barcelona. So, at that time Mark was working on this Passion Façade, and these columns on this Facade.

Mike Xie:There were very few joints left behind by the original architect Gaudi. Gaudi started designing this structure about a hundred years ago. At that time he was using this physical models with chains and weights. If you look at this structure, every point on the chain is in tension, and what Gaudi did, he put this model upside down then everywhere would be in compression. Because he was building this church using stone, which is similar to concrete. It needs to be in compression to be efficient. So, if you turn this model upside down, you will get a shape, which would be suitable for a masonry structure or stone structure. And if you go inside this church, you will see a lot of interesting models. This is one of the models inside, and these chains were being tensioned and there's a mirror underneath, I took a photo from the mirror. And this would be the structure, Gaudi would use to build these structures. You can compare the two, you can see the architectural elements, they are very similar. And if you walk around Barcelona, you will see many of Gaudi's creations similar to this shape.

Mike Xie:Once we got our method, we can actually replicate what Gaudi did a hundred years ago very quickly, because when he created this models, it needs six or nine months in order to get a model working, and also he could only consider gravity. But, with our method you can put all the forces in, and to see, which areas are in tension you remove them. You will leave behind a structure, which will be suitable for the stone material. So, this will be what we get for the Passion Facade, which is very similar to what Gaudi's original joints, and what Mark Burry finally created using parametric studies.

Mike Xie:And it also depends on what forces you use. So, if we build this church in Shanghai, you would get a totally different shape, because in Barcelona the earth quake loading is small, but in Shanghai, Tokyo, the issue is, you have to have very high resistance to the lateral load. And we actually try that, if you put down very strong lateral ... The earth quake load, you would have additional bracings in order to resist forces in the horizontal direction. These are a series of studies I did with Mark Burry. So, if we want to create three columns on a slope, and starting from this by deleting the areas, which has highest tensile stress, within a few seconds, we can get a shape like this. If you go to Barcelona next time, have a close look at these columns on Passion Facade. And these columns have exact the shape as what we can get from our simple rule of Evolutionary Structure [inaudible 00:16:12]

Mike Xie:All we did is we start from any design like this, there's no bias. It only says there's three columns. If I remove areas in tension, you will get a shape, a beautiful shape, and just like we would see at the real church in Barcelona. In a 3D version, say, if we want to build a two columns on a flat surface, so you start from this, every time we remove the areas with the highest tensile stress, you get some strange looking column like this. And people can see that Gaudi as a genius, he actually got this design about a hundred years ago using his physical models. But if we go back to the basic mechanics and apply the simple rules of removing the tensile materials, we would get exactly the same shape as Gaudi did for his famous church in Barcelona.

Mike Xie:And later on we improved our method, in previously we can only remove the inefficient parts from the current design. We can actually do the reverse, say like the apple example, he can plant a seed and let the see to grow into apple so he can ... In structure terms, you can strengthen the most critical part in your structure by adding material to where it is most needed. Or we can do both ways, so you can start from any design then, while you are removing the most inefficient material, you can add those material to where it is critical. So we caught it Bi-Directional Evolutionary Structure Optimization method. So, it would be much faster and it's more efficient. So, I show you one simple example, say I want to design a bridge type of structure. So, I have a deck with uniformly distributed force, four points fixed at the corners, and I leave a gap in the middle so that vehicles can go through. All the rest will be decided by the computer where to put the material. And your initial design can be also very simple.

Mike Xie:So I have applied four columns, and this can also be a bridge, but no one will pay you consulting phase for designing something like this. It can break very easily, there's a stress concentration, but it doesn't matter, we can start from this initial design, and let the computer to decide where to add the material, and where to delete. So you are fine. Some of the earlier material added is removed later on. So, we are not doing some local modifications of a design. That's what people do in consulting office. They have a design and then do some assessments, “Oh, this part needs to be strengthened a little bit.” So, they do a simple change. We actually change a structure concept from this to that. It's not a simple local modification. The structural concepts of the two designs are completely different. We got an arch bridge with additional members like the tree branches we have repeated seeing in Gaudi's designs so that the whole structure has very uniform stress distribution and uniform structure performance.

Mike Xie:We have done many other examples, say this one is to design [kitchen 00:20:03] in a hotel, say you have four columns, you have very few column, you want to transfer the loads from a few columns to upper levels, and our process would enable you to find the best load transfer system, again, it's looks like what we see in nature, those tree branches. We can also do the dynamic optimization, say if we have vibration frequency through our process, we can find a design with a highest frequency, or you can increase the first three or five frequencies of the structure. We can have multiple constraints, for this building, we can minimize a deflection at the top of the building and also control the frequency.

Mike Xie:So, by redistributing this uniform shear walls, we get a bracing and the structure performance will be increased significantly not for both static and dynamic criteria. We have also tried to apply our technique to periodical structures. So, I show you one simple example, you will understand. So, this is a bridge say a few hundred meters long, but it's made of identical units. So, if we apply our rule by first say, divide the structure into a certain number of units. And when you change your one unit, you change the corresponding locations at all units, you will get a periodical optimal design. This example would be easier to understand say, this wheel has eight identical modules. When we apply our method, say, if want to do need any material here or the corresponding location, what have the material removed and you apply all the loads you were experiencing when you're driving the car.

Mike Xie:In our case, when you're accelerating or turning or breaking all the forces on the wheel will be different. You need to consider all the major load cases in order to decide which part needs to be removed or remain. Depending on how many cells you want, for a wheel you can get to do defend designs. And this was done by one of my PhD student. He got this pretty pictures, and then, I asked him whether these are correct or not. He walked around Melbourne street and took two photos and he found this. So, this is what he got using our simple Bi-Directional Evolution Structure of Optimization. These are Cadillac, BMW 17, They're almost identical.

Mike Xie:So, again, this is not a coincidence. This is because our design objectives, and BMW and Cadillac's, their objective are the same. We wanted design the lightest possible wheels, which has certain structure performance. So, they have spent maybe 40 or 50 years to improve their designs. But using our technique, we can develop the next generation of motor vehicles very quickly. Around the world these days, the electric vehicles is very popular, but the loading on the wheels are quite different because of the heavy batteries, the loading on these wheels will be completely different from what you see on BMW or Cadillac. But when we design a new wheel, we can apply this new conditions, and get a very close to the optimum design very quickly. And I started doing this work when I was in the Aeronautical Department in the University Sydney. And because weight reduction is of critical importance for aircraft, and that's why we have been working with Boeing and also Comac in China, tried to apply this technology to aircraft. We had several collaborative project with these companies.

Mike Xie:But strangely, I expected the people who become very excited about our technique was a group of architects. And so about 15 years ago, famous architects around the world started using our technique. I think it's mainly because of the organic shape we were able to create. It look so beautiful, it looks like what you see in nature. So, they started to make buildings, so this was the first building designed using extended Evolutionary Structure Optimization, which was based on my work. This was built in 2004 near Nagoya. There many other buildings, I give you a few.

Mike Xie:And some of you may heard that the Pritzker Architecture Prize, this is the highest price in architecture. It was award last week to a guy called Arata Isozaki. He actually was one of the first architect using our technique to design buildings. This one was in Qatar. It's actually very similar to the bridge example I showed you. If you're using our software, you fix this two points, put uniformly distribute load on the top, within a cup of tea, you will see this beautiful image on your screen. It's as simple as that. It's not just a beautiful picture, it has been built. This was a photo taken in 2012. I once passed Qatar airport, so I took a taxi and went to take this photo, so I took this one. So it has been built.

Mike Xie:And in China, Shanghai, there's another famous building, Himalayas Center, these strange shapes. Next time when you see it, it's not strange anymore. It's based on our simple rule of deleting inefficient material, and to get a different design. So, these are all based on the extended evolutionary structure optimization [master 00:27:01] I originally redeveloped. I have some other architecture friends who had been using my technique to build these different structures. This one has been done in Shanghai as well. I also tried to ... She mentioned that I was good at translating research into real project, but I tried and I failed terribly in Australia. I give you one failed example, it's a lesson I learned to move to the next step.

Mike Xie:So about 12 years ago, the VicRoads was doing a revamp of the Monash Freeway. There are several foot bridges, which they were calling for proposals. And I was very excited because we thought we could use our technique to design a better foot bridge for Malvern. I think we did a week ... We did a very good job. So, we were given the information, how wide the Monash Freeway was, 45 meters, and the boundary conditions. Depending on where you build a bridge, so if you're building against two solid rock, you would have this, if you are supporting here with a simple support, a roller, it will be a different structure.

Mike Xie:So, we did a series of studies. This is what we proposed to VicRoads. This is near Malvern station. So, next time you could have this bridge there in 2007, we presented it to VicRoads because this is really too dramatic, people haven't seen this kind of thing. This is before Isozaki built his structures in Qatar and Shanghai. So, VicRoads said, “Oh, look, it's impossible to build a bridge like this.” It's actually the [tech 00:29:06] is here so, if you walk through this bridge, it will be a really interesting visual experience. So, they said, “No way, this can't be built.” So, in their great wisdom, they built this. This is the Malvern station footbridge. So, they chose the safer way to construct this.

Mike Xie:I was very disappointed because actually it's not possible to build. I built one in the lab. It's building 10 in the old civil lab. The circle has six identical pieces. So, I made a reusable mode, this is timber mode, this foam will become the house. So, I put the steering reinforcement where the connections, putting the concrete six identical pieces I put on the boats, I can construct this very easily one to four scale section of that bridge. So, I wasn't successful to convince VicRoads. But in this check, retrospect, I can understand because this is something too ahead of the market, the market wasn't ready 10 years ago. But if I do this again now, I think I have a much better chance to convince people in their authority to take a risk and do something different in Malvern.

Mike Xie:I tried my hands in China as well. So, this is a proposal in [foreign language 00:30:49]. Again, this beautiful design is from a very simple rule of removing inefficient materials, redundant material from the structure. The byproduct of that simple process is this beautiful design. So, this is the proposal, we are still trying to get a build in China. And not that, we actually did a serious work. Last year I worked with a famous architecture firm in Shanghai called [Atilya Aqui 00:31:20] mixing. So, we proposed a series of foot bridges in Huangpu River, it's a famous river in Shanghai. We can also apply our technique to small scale structures. And so, this is a hot area of meta materials. So, by applying our technique we can, we can improve the material properties of this repeated periodical microstructures. In this case we have been maximize the back modules. So, under any pressure, this structure would have the least amount of volume change.

Mike Xie:The next one is a micro structure which can resist the shear stress. If you put on shear stress on this material because the material has been arranged on the diagonal direction, it can have very high shear resistance. Or we can design a micro microstructure with specified stiffness in different directions, like a bone, when you want to create a replacement for a piece of bone, you want that material to be having similar material properly like the bone being replaced. And normally you would have different stiffness in different directions. We can get exactly the same stiffness ratios by designing the microstructure. So, this is a functional grade it can have changes along the length and have a gradual change in the material property. And this is one of the material we have designed.

Mike Xie:So, normally, normal material when you have compression it will expand laterally. So, in our case, when you compress it, it actually shrink literally, so it does the opposite. So, this is [inaudible 00:33:37] material. And there've been a lot of papers written, but we were very early to be able to produce something which we can show to the people how it behaves. Another one, this is metal version of the [inaudible 00:33:59] material. So, the shapes are a little bit different from the rubber one. You can see a rugby ball shapes, so a horizontal one, a vertical one, this is a 3D, it's not just 2D, it's a Q. So, when you compress it, it will move in laterally. This kind of material has a lot of application in defense industry. We have some contract with the DSG, used to be caught DSTO trying to incorporate this kind of material to resist the strapping of penetration to armored vehicles.

Mike Xie:We have developed another material which behaves different from normal situation. So, normal material, when you put it on compression, it will shrink. So, in this case, I put my material in a freezer bag, and put a tube to suck the air out, which is the same as putting uniform pressure and the material. If I do that, if I suck the air out put the pressure, it actually expands on the pressure. And this is purely because we have designed our microstructure so it will have maximum deflection or deformation in a certain direction. So, I can change the shape of those cavities to make it grow in two directions. We have designed these kind of materials as well. So, you can make it grow in one or two direction. It's not possible to grow in three directions under uniform pressure. We can also make different holes, so that under any pressure, the length of the material will not change or the cross section area will not change. So, by simply changing the shape of the cavity through our optimization algorithms. So, this material has applications in biomedical field. We had to contract with [Wun 00:36:13] management CRC. It's because our special material like this.

Mike Xie:I'm going to talk about something. So, the connections in civil structures, they are very complicated and normally you need to say this are cast, these are weld. And one of the project we have been working with Arab is try to use 3D pending technology to realize these complex connections. So, traditionally we would design the columns and beams and in our project we would like to use standard steel columns, steel beams, but we designed every connection because you can find these forces in every member easily, and then, using this forces as input, you can design the connection. So your initial design could be a solid balL, and then let the computer decide, which area needs to be removed to make it lighter. So, this is one of the examples. So, I fix here, put a [banner 00:37:35] moment here, another banner moment there, then I can get rid of all the unnecessary material to produce the most efficient connection for this given loading condition. We have done many different notes. In this case, I fixed one point, put a shear force here, put a shear force, after you removed being efficient material, you got this beautiful connection. So, if we come to a future building, we are going to construct, when you come to a room, the connections are not like this boring squares. Every connection is a piece of artwork just like this.

Mike Xie:And these are some of the 3D printed metal notes. I have some at the back, when we finish, we can have a look. So, at RMIT, we are very proud of the facility we have at the AMP, Advanced Manufacturing Printing, we can print many off our notes. So, this was done at AMP, this was done somewhere else. But, there's one critical thing about the metal printing, although you can say, oh, it can be done, but there's two ... When I brought this note back to Mike, my friends who are doing the real project in the field, they asked me tWo questions, A; How much does it cost? B; Can you certify your note? The first question, I did some calculation, I give him the prize. It's about a thousand or more times of the normal price they were to use in the construction industry. The second one is the possible defect in this metal printed parts. So, eventually I stopped doing the direct metal printing started to do the casting, because you can print the same model using wax very cheaply, and you can print very large wax models.

Mike Xie:So, in Huangzhou University of Science Technology, they can print those wax models 1.2 meters big. Or if it's three meters you can print two pieces and stick to wax model together. Then with the wax model, you can create a shell outside the wax model, and that Shell can be used to cast metal. So, the next one is the shell. So, this is a ceramic shell you can get after you dipping your wax model into the ceramic slurry and let it dry. You do that several time, you get this very solid shell, and you can pull the molten steel into this.

Mike Xie:So, this is a very good image, it was taken in Malvern three years ago. It looks like a medieval 300 years old picture. The process of casting has been there for hundreds of years, but people are still doing this. But this is a perfect combination of traditional casting technology with modern techniques here. So, here we actually used the lightest topology optimization technique to create the shape. We used a 3D printing to print the wax. And so the advanced manufacturing technology is here, but the last step is casting, and this, we have several examples on display at the back. So, when you see this note, this one was cast. And also two questions I was asked, one the price, this would cost a small fraction of the direct metal printing node. The other thing is because the last step is done through traditional casting technology. It can be used tomorrow because there's standards people have been using casting for real projects for many, many years.

Mike Xie:So, I have 10 more minutes. I will go quickly about some of other [fans 00:42:08] we have been doing. So, the topology optimization technique can be used for many things. We actually made some jewelries, but I deleted that picture. I show you some of the furniture's we have made. This is a stool we made so by applying the load, then we can get this structure, we can 3D print. If you come to my office, you can see some of the tables, my student made for me.

Mike Xie:So, this is jamming and one of the student there. He made this table for me using our technique, using our software to remove the inefficient materials. I want to show you the slight difference. Initially I want to make a table using 3D printing. So, we generate this shape, I send it to print, the bill came back, this leg costs me 700 Australian dollars. So, if I want to have a table with four legs, it's almost $3,000. It's too much, or it didn't get the budget, Xing didn't give me any additional support for that. So, I changed the strategy, so instead of doing the full 3D printing, this is a 2D laser cutting so, it's made of four plates. So, it's a 2D plate, you do laser cutting and it can be flat packed. It takes very small volume and the whole thing cost about $100, that one will be $3,000. And we bought a table top, the glass $99. So, this is $200 table, it looks very nice.

Speaker 3:[inaudible 00:44:00]

Mike Xie:Sorry.

Speaker 3:[inaudible 00:44:03]

Mike Xie:The table top is from Ikea, but that advanced technologies underneath the glass, we want the glass so that we can see what's there. And the other issue is this four plates, we need to have the glue to glue them together, which is actually very tedious because you need to have two people hold it and wait for a lot of time. So, we improved it. So this one has three plates, we use a tough tower system. So the three plate, you can just slot into each other, they can assemble them. So, he can put it in a suitcase because it's so small volume, and when you get there, you can assemble it.

Mike Xie:We also tried our method to [June. 00:44:57] So, this is the initial design and you apply the load when you fly, you're turning the force is on the four propellers are quite different, so you feel when taking off. So, some of the loads in one propeller will be much higher than the rest. So, I think we consider maybe around 15 load cases and removed the inefficient parts, we can get a very light weight and organic June, just like this. It flew perfectly, we tried this in our campus. It's also at the back. We didn't bring the battery, otherwise we could fly in the room here, I could show you.

Mike Xie:I actually tried very hard to put some of my technology into real structures, I'm making some progress. So, this is one of the project we have design. So, in this one we designed 48 tree top column, just like what Gaudi did in Barcelona. We are having this to 200 meter buildings to be completed in the next few months with 48 trees just like this. And the wanting, why? It's me. I was so proud to inspect my team's work when this mock up, this one to one model was built about 18 months ago.

Mike Xie:This is one of our current job. So, this is Observation Tower overlooking a famous lake in China. And this was designed by architect and the owner wasn't quite satisfied, he said, "Oh, we want to create a landmark building for this structure. Can you find someone who can do better?" So, people refer to us, they know we like doing some crazy things. So, we started redesign this Observation Tower. So, this is about 150 meters or 200 meters? About 200 meters. So, there's two lifts going up, there is artificial wetland here. And so, first we need to set up all the loads. So, you have the gravity on the tower, you have the forces on each floor, and the lifts and the artificial wetland winds from four directions. Once we put all the forces in, we can start to evolve, to generate different designs. So, again, by deleting those inefficient materials we can come up with a total different design. So, this is one of the proposals. We can actually control the member sizes to make a sticker. So, this is one design, in another one, it's a similar process, but with thicker members. So, these are some of our proposals so I just go quickly. We are hoping to get this built in the next few years.

Mike Xie:And there's one thing we realized through our work is we need to have much closer collaborative relationship with the architect. Because traditionally in China, in Australia, in many countries, the architect would produce a design, and then they give it to the structural engineer to say, 'Look, you design the thing to make sure it doesn't fall." But the structural engineer wouldn't be able to make much changes to the shapes or forms of the design. But if you're able ... I'm mainly a structure engineer, but if we are able to get involved in the project very early, so before they decided that the design, you need to start to talk with the architect in order to co-design that project, then that would make a big difference.

Mike Xie:In this case, we actually have been talking to the architects very constantly, like when we produced this design, the architects said, "No." Because this is Observation Tower, I can't a solid wall here. You need to have transparency, you need to able to see through. We can achieve that by simply putting a thicker walls at the bottom. If you use a thicker plates, you can get additional holes in these phases. So it's only a matter of proper dialogue rather than just one way flow. It's a two way dialogue in order to come up with a design, satisfying both the architecture intention and a structure performance.

Mike Xie:One of our current researches, when we deal with a mathematicians and when we do optimization, they always say, "Is your solution the absolute best or the unique solution?" That's what the mathematicians want to ask. But these days, actually I want to do exactly the opposite. I don't want a single solution, I want many solutions. That's because I've been working with architects, they hate to give a solution, I've got the best solution if that's a case, what's their job? They want choice, they want input. So, what we want to do is to create many solutions, all the solutions that look dramatically different, but in terms of structure performance, they're very close. So, this is something, and we have made some good progress. So, this Qatar Convention Center, if we only consider the structure performance, we will get something like this. But because this is the existing design, we can't build this in Melbourne or Shanghai. It's has been there, people have been winning awards.

Mike Xie:But the situation is quite common. You have to support uniformly distributed load. For given boundary and loading conditions, we want to create a different looking design. This is what we have achieved because this is a existing solution. Any material which has already appeared in the previous designs, I would penalize it. So, by doing that I get it quite different results. So, this is the additional trust and the structure performance of this is actually very close to that. Another, example say, I have uniformly distributed load on this tech, and if I only consider structure performance, you get two arches, if I penalize the material which has already appeared in this design, I get the second design where the two arches are coming together.

Mike Xie:If I penalize both existing design, I get another one. The third one, the two arches become one and the three looks quite different, but in terms structure performance they're within 3% if you consider their stiffness. So although they look different, from structural engineering point of view, they are very similar. So, that's why we call it diverse and competitive designs. And I'll go very quickly, we have developed some software called Amoeba. It's been running on the Cloud. It can create all these structures very quickly. So, this is how we can design the stool, and this is a 2D version of a chair and there are some of other things. So, I'm very lucky to have so many talented students, and probably 15 of them are here so they have contributed to the work, and we are making some new progress in our team. So, thank you very much.

Xinghuo Yu:[inaudible 00:53:49] So all different kind of discipline areas and they create some of the fascinating structures. So, now we just open for some questions, any questions and comments? There.

Speaker 4:Does your method always converge to a solution that's stationary or do you just have to decide on an arbitrary stopping point?

Mike Xie:We actually have to get them to converge. So, we look at the objective function. If the objective function do not change much for 10 steps, we stop. So, we have a criteria to decide when to stop. It's always we have a very flat line to make sure it converges.

Speaker 4:Thank you.

Xinghuo Yu:Thank you. Any more questions? Yes. Can we have the ... Sorry, can we have the speak [inaudible 00:55:00]

Speaker 5:Okay. Thank you. In designer category or structure, if the minimal size constraint and specify the volume fraction, can we get solution which matches [seratical 00:55:19] solution very well?

Mike Xie:To prove our algorithm is working, we have actually compared a lot of benchmark examples to show that our simple rules can produce those analytical designs. We have done that in both static and dynamic examples.

Xinghuo Yu:Alright. Okay. Can we pass on that to speaker.

Speaker 6:Thank you for the example. I'm just wondering [inaudible 00:55:57] Well, your design is dependent on electro materials that are used, and if you can use a combination of different materials and how many materials you can incorporate into one design, if it's a constraint or?

Mike Xie:It heavily depends on the material you use, like what I showed, whether it should be suitable with tension or compression. You can have multiple materials, because our technique is a simple post processing [inaudible 00:56:25] It's actually whatever you can analyze using finite element software, we can process the output and make changes. So, your problem could be nonlinear, material could large deformation, as long as you have considered this in your analysis, you can do modifications to get a new design and put it back for the analysis to assess its performance.

Speaker 6:But, if I can add to this, the practicality of things like their boundaries between materials, they would be susceptible to deformation or stress, how this is accounted for?

Mike Xie:We haven't considered these details as I keep it mind.

Speaker 6:Thank you.

Xinghuo Yu:Okay. Thank you very much. Yes, there's some other questions from there.

Speaker 7:Thanks Mike. Just a quick question. When you analyze the structure and then you change the shape, do you restart from initial condition or how does it work or do you just keep going the analysis, especially if it's nonlinear. Do you start from scratch when you change the shape?

Mike Xie:We actually continue. Yeah, if it's nonlinear it's far more complicated, you really need to go back again to do it. We haven't done many nonlinear examples, but we did some nonlinear materials. Both are possible, but the result will be different because it's past dependent then. We can do both ways to see what difference we will get.

Speaker 7:Thank you.

Xinghuo Yu:Okay. Oh, well there's another one questions.

Speaker 8:Thank you. It's a very interesting presentation. My question is regarding construction prefabrication, because that's getting more and more popular in the construction industry. Have you considered to inform the lean manufacturing stage for that, ratio design and the optimal solution for the structure, whether that can be considered also to inform the lean manufacturing stage?

Mike Xie:One of the things we can contribute is the aspect I mentioned about the periodical structures, because the prefabrication, you have many identical components, and we can make this identical component. If we can save a small amount for that component because it's producing set such a large number we can achieve a lot of savings. So, we have thinking about along this line.

Speaker 8:Maybe bathroom poles so it's kind of things.

Mike Xie:Could be.

Xinghuo Yu:Okay. Thank you. Yes, there's another question.

Speaker 9:I've a non engineering question. So, because obviously, so you've got the architect that is using the software to produce a very unique looking design, but the design uniqueness is actually sort of inherent to the actual software. So, the software in itself is geared towards this very specific design aesthetic. Does it mean that as a design professional, there is a possibility of coming across IP problems by using the software from multiple design firms because you will inevitably get, although you say they look different, they may not look sufficiently different from plagiarism point of view.

Mike Xie:I guess, it's self regulating, if they are aware that some similar firm being built, they wouldn't produce something similar [crosstalk 01:00:55] That's why the diverse solution becomes so important. We can actually have some random process every time you write because we have put some random modifications to the same material properties, every time you get totally different random design, but we make sure that the structure performance is not sacrificed, although you get random shapes, but every design has very high structure performance.

Speaker 9:But, I'm thinking more from the actual aesthetic point of view. Is there a way to modify, so let's say as an architect, you can have a very specific, We call it design, thumbprint, so design some print?

Mike Xie:Yeah.

Speaker 9:So, it's literally like a thumbprint that recognizes you as a designer because the Japanese designer seems to be using that very natural shape and it's become his signature look. So, is there a way to modify the algorithms so you can have your own, the algorithm takes your signature look into or it will develop some sort of a signature look.

Mike Xie:You can actually specify some regions, some examples, you can say, in your design you can actually put RMIT in the design and that RMIT is always remaining there, and you can say this area has to be a hole or that area has to be solid. And all these things, the designers or architect can control on the screen. So, it's not just a black box, we want to avoid a black box to, we want to empower the architects to use our tools.

Xinghuo Yu:Given the time, we might just stop here so you can still talk to Mike after we have some refresh after those. So please join me to thank Mike for the fascinating talk. All right.