South Big Data Hub Materials and Advanced Manufacturing Workshop, 13 Feb 2017

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: CALPHAD as a Critical Component of Data Infrastructure for ICME/MGI Part 2

Description

ThermoCalc, Kaisheng Wu

A

Hello, so I'm going to introduce our first speaker. Have him come up, so it's dr. Kevin Wu, who is the principal material scientist at thermo-calc, so we're gonna hear about Cal fed technique, which is a phase based approach to modeling and underlying thermodynamics and phase equilibrium for equilibria from multi component alloy systems, so I'll? Let him take it from here and.

A

Okay, thanks acrimony everybody, my name is Kevin whoo from a thermo-calc software inc located in Pittsburgh I, come here and on behalf of Paul Mason, who is the president of our mocha Software Inc here? He cannot make rate before because he's on vacation before he also notified for this wonderful chance for us to come here. So I came here on behalf of him. So I also like to thank my colleague back, in my hope. Our headquarter in Sweden, dr.

A

Chinchin, Joann Verburg and her Sandstrom from the macaques offer a be located in Stockholm, Sweden and endosteum is actually a prison of our software company. So my talk today is about comfort method as a critical component of editing for structure for a CME and MGI.

A

So some of you may know about our software companies thermo-calc, and we know that our company is actually that develop the software tools and the database is based on what we call the comfort method. So this method is being involved for about 30 years. If you look at package but more than 40 years for from the starting point, so buddy so over 30 years to working as alloy development psycho so to design, and you want materials. So all the software and the database have been war for quite a long time.

A

So today, I looked first, we talking about what cavities and how you fit into the current struggling society of ICMP and MGI.

A

So and then we give the profile and talk about history, history of our cafes, software tools and what the future should look like and as I talk about thing about our databases of course, and what we are one emphasizes some challenges we are facing, especially for database developments throughout these years and know for ongoing projects and also, we hope that we can get a benefit. Some other data data infrastructures.

A

So, first many people, honest, can be more familiar with canna triangle, microstructure, property and processing triangle. Actually, we want to add another kind of corner to pick the prison as a chemical, a chemistry, that's what really what the cafe is working on. So we try to understand what every chemistry can be can affect the microstructure and probably and processing conditions.

A

So Carrodus previous. This originates come from the acronym of CAF calculation or phase diagrams. So it's a society and it started in the 1970s I've seen actual East Studies from the developed, a method by Eric Hoffman from MIT and gradually we grew up as a society to generate the face recruitment and phase diagram theatres for industrial use. Since 1980s, we have a bunch of software like thermo-calc and lucas program in later have a prism campus room, so we divestment softwares and the databases, so industrial can uses this to develop their alloys.

A

So prom with a bunch of information like a fish information. Thermodynamic information connect information for users to use. So the basic idea is to capture the conversation, temperature and pressure dependence. So we based its kind of faith based approach, so we develop different, the cubes French, for example, for different phases, and we build out the other basis and from there then we use some theoretical background that gives free energy to keep theorem or a phase transformation theory to provide information like conversation, temperature temperature, dependence of different phases and also faced relations nature.

A

So we have a regular, you know Koch Alfred, special record, and so actually we have active participants for throughout the world.

A

So for ICM e is the integrate computational materials engineering. This initial published is released in a 2008 graduate becomes, as me, it's developed very fast in recent years to become a very active Society. So the basic world here is a integrate and engineering.

A

So it's try to integrate multiple in computational math models from town from the sub atomic scale to up to a production scale, to try to integrate it as a holistic system that we can use it in regular Android development to try to shift this kind of concept from a trial. Error can be experiment technique to some more computational orientated with technology, so the key was an integrated and in engineering, and also the therapist is very, is the key components in the system. So.

A

So many people know the MD is a materials genome initiative that we are lost by were housed in the Trinity 2011.

A

Actually, this the purpose of this one's tried to top of the speed of alloy development materials, innovation cycle. So the core of this amici is kind of three kind of integrated parts. The efficient has report computational force and high throughput experimental techniques and also digital data informatics, so the three the combine together to build up the materials innovation infrastructure.

A

So if you look at that, actually MacArthur is kind of a nexus of this kind of a min-gi, so in his unique way, because cafard has a lot of a very close relation with the database with the experiment data. So then the database that build up is considered as the digital data, although we don't have really enough knowledge about the data informatics from a computational perspective, but we we based our hotel a base based on the kind of material science theory like thermodynamics and a phase transformation.

A

So based on that, we do some a computational tours that can be very efficiently. I could provide the data for the industrial use. Okay,.

A

So give some brief idea, the comfort, so first I didn't think that we clever Lauretta either from the theory. There are quantum at Hanks statistics and also connive DFT calculations to generate a master Muhammad data and also we care from experiments like a tt8, TFT and x-ray diffraction at improvisation, etc. To cut all the some lab information like entropy entropy gives range and aphasic ruined data. So we create all this data together to pure up can a physical based models?

A

The wonderful thermodynamics just like thermodynamic you Spanish for mobility or connecting that's kind of autonomy, ability, databases for diffusion, so we build up with tailors and with with some parameters that can be fit to the experiment data. So then we store this order: data as a Gibbs, free energy and autonomy mobility into the databases. So then we use the advanced computational software tool with very advanced algorithm, to provide all this data to industrial use, either to track click at a phase.

A

Information like a phase fraction, very Caribbean and matrix composition, etc, or they can build up a appeared on your own bottle with our with our kind of sdk tools. Okay,.

A

So if you look at this in the IC IC, any framework Kevlar can't be as an important bridge to multicam room prediction, so we we can allow the can micro scale. So we can collect data from a festival, for example, for smoke equations, then we can provide, can be credible. Data like gives free energy data SOP data interface generator, won't bother an elastic housing data, etc.

A

The pure data base databases that we provided software to provide to provide a stator to the interrupts, kill, cannon calculation like a face, face, VM method, diffusion transformation equation or language type in field approach to persuasion, kinetics and yesterday, and we can move up to another scale for the like fvn or fluid dynamics, calculations, nature.

A

So this is trial that provide by doctor professor gray oxen with Northwestern University, who is also the founder of the quad tech innovation company, a quiz which is the lead. Actually, the leader right now in the ACM II Society for using the computational can integrate tools to design alloys.

A

You know in a way that, in this kind of a so-called hierarchical kind of a structure to design materials- and it successfully I think it's the Society in mind as a series of iron and steel Airways. So if look here is a critical, hierarchical structure, they started from the north girl like a materials genome and tools. ah / I see SMD, like I, said optimization kind of throws.

A

So if on a computational tours in the left side and the databases right side, so Donna scratch, you can see that actually in a meteor I can pools in the bottom of this diagram.

A

So, actually they probably first don't you know the proto data is the data for prototype system, so the chrono experiment, data, DFT data and the user tools to generate data, and this data be precise and that storage database using the caustic Odom.

A

That's what we called a cada method to develop a database using a software like the thermal cracking teacher to travel database, so I think in the sense of their kind of development confidence actually semi GI for them, I mean in the sense for that, so the developer database from service in the material regime they move up to next next level to a CME region. Then they develop Dominic as models to onions, understand, process, structure and started, probably models.

A

So in our case, there's also use dive software for, like a thermal crack and a teacher so photo McCarty, provide phase transformation, information-, re-enforce and a face, accrue information and for teacher for, like the future for diffusion control transformation. So, if not I can see cadet atomic abilities and using this information I can build up the models like a transformation model. Snackable one inside is starting temperature and also for Zener painting or preserving precipitation.

A

Modeling is for the person information, so you can see like the cathodic, actually play a very quick role in the our development of the of the arrays.

A

So the cover-based tools that divide me now come: please have a shriek and four branches, first of all, as a flagship software, of course, Sequoia thermo-calc, which is the basically the thermodynamic calculations, and we have another software to auto pictures. The diffusion control the transformation. So this kind of diffusion based at all so reasoning, with lapse quality, see Prisma, is developed for presentation, connects.

A

So for copper based software, the Summa of its mission is flagship of the software, so it's back in the night is develop, add back in 92 80s. So this the purpose of this software is for some anomic and face equipping decorations, so they can calculate stable and metastable, hitting yourself facie Caribbean for multi-component, multiple modifiers systems and information are provided about the conversation faces. So that's really important information to design the processing conditions and its other transformation temperatures.

A

The liquidus starters temperatures also can predict the trying forces for first generation either diffusion or the diffusion is faced faster machines, the provide phase diagrams panel, eternally ISOs, my suppressor and different type of our phase, diagrams informations and also recently. We also introduced other kind of information like more volume density and also expansions and there's a lot of module core. We can do some certification simulation and shell equations dr. summer, chemic chemical data, such as enthalpy episode, hit capacity and activities or information before the important for certification process.

A

So we have some similar great progress for chemical reactions. Okay, so all these imaging can be very helpful and very efficient to provide for users to design optimize the ROI and also for the processes.

A

So for teachers, diffusion culture, the motor component efficient. So in addition to some endeavors for this problem, we introduced more character correctly issues. So we build up a lot of database called atomic mobility ways to teach some kind of similar the diffusion process to the processing conditions. So there's a what what application for his software like a carburization, antique appraiser, aging process, micro circulation tool in certification and the homogenization process after that and Priscilla kuro-san data solution to simulate.

A

For example, karate solution to resolution determines principally costing, and also for holding substrate kind of an inter diffusion process to help design the coating materials and how how to understand how the interaction between the coating, the substrate, for example, for aerospace and the co-pays of our publications and as a of course, as all other, while applications for here to show you the example of the colorization process.

A

So another software quality series, my squad, young software tools to simulate the Prisma, connects so yeah. Another can model model the christian model, so we can consider concurrent the creation process ruching and a cross in process. It's better than language was theory, so it's method approach, but we still can use up to islamic history and the multiple insights to simulate representation.

A

Multiple phases are the different endocrine size, so either the grain boundary, for you know, painting effect or we think that with the park to simulate kind of strengthening christmas transiting effect so information, I provided face: rajan our accommodation in the precipitant matrix and a particle size distribution of operative precipitates. So all the information that be very helpful for if you build up our strengthening models to simulate the mechanic, progress and optimize the Richmond history to achieve the goal of different mechanical progress.

A

So that's what we'll think about it! We're gonna we're gonna do in the future. So in the left side, it's what we already have right now we have different modules like diffusion module and a pretty model. So we try to build out more modules in the future like introduce more effects like a stress effect and others, and we also provide the data. Optimization kind of module that'll be help for users to develop their own databases using kind of a subscrib organization algorithm, and we also provide sdk tools for users.

A

So if you want to build up your own model based on our data information, you can do that. Okay.

A

So of course we still approach still giving developer different databases and academic connect, and in future we want to build a more, like other, probably database, neck, a viscosity and elastic constants nature. So, on the other hand, we want to provide more kind of model. Libraries like computational method.

A

Traditional maturities are in process structure, probably model libraries for different Airways, like steel nickel base, for you know, you know, is and a titanium alloys, and we also want you- can ever develop the more modules for application already. Four different Processing's like steelmaking and casting. Let's just assemble all these modules. With the embedded into thermo-calc, then we can can assemble our different modules to similar eight application oriented it first processes.

A

So for ongoing work, we can more get more involved in the property, visualization, optimization applications. So for property perspective, we provide more property that user can use to to the calculations like phase transition train force. Then you develop the model, slack interfacial energy and a causing rate that we can use this information. He was talking to the calculation to carry this information. Okay and in the future, users can also can be used to verbal. There are models that based on that, for example, pattern script.

A

So if you write a pencil script, you can use our software to get all the information from the three o'clock and a butyl model.

A

So visualization we provide more kind of visualizing tools like, for example, heat for heat map in the left side or a control map in the right side. So from this information you can just design the kind of a design window based on this diagram to get the information.

A

And also we try to get some uncertainty Kanak analysis. So if you wanted to do some master said sensitivity, analysis, you can use this for to get its information. Okay, so of course databases are very important. So that's what we strive to to a more database development for that from the co-pays Tamara steel and also go to oxides and a slack suture okay, so tell about if M is a very critical and it's a very difficult.

A

Actually, if you look at the little it's a very, very efficient Nikolay the base, we have a twenty elements for alloy and three elements for gas. So if you look at binary ternary system, it's very big number of that. So, although usually we try to focus on some nickel, pure nickel, Turnagain region to the other database. But if you move on to that, for example, high entropy Airways, you cannot do that. You have to size to the whole conversation range.

A

It does a big task so right now we provide high entropy, no errors for fitting elements. So in that case, we assess all the anyway system for for composition range, so some challenges for the other database, for example, come first why's complete, complete server database so for I just mention that that this database developer is a very difficult task. So if you look at ICU diagram, you have a a bunch of the binary and the ternary system, but you cannot collect all them. We can only collect some just part of them.

A

So how do you choose this one? So it's very critical. So other challenge is that conversation temperature dependent data. So we read good quality data for that, unfortunately, we usually right now we still don't have that much information about that, so industrial air, that is the multi component system. Well, we cannot directly use that we based on binary tree and then we use that to predict the molecular system. So we cannot actually directly use that either directly to pull out a de base as or rather will use it as a validation.

A

So no haha intelligent temperature data is so very difficult to get low temperatures that efficient, very slow, so we will not. Allow is really chic Ruben high temperature is sometimes a contaminating mother or a teaching a researcher to cause a problem. Okay, so another thing is that it's not a big issue, so sometimes we will know some. We have some issue as I'm afeeling about it, asserting if I know attorney, but we don't know how to propagate is asserted propagate to a multi-camera system. So hopefully you guys have some information, but based on studies.

A

It can help us to understand this one so because the extended to higher component system is highly nonlinear behavior. So we really not only how to address this issue in the medical system, so good quality data from community common system. So if you collect all the data formally in industrial public domain, so we don't really know what quality of this. So we hope that in the future, who have society to develop kind of standard of benchmark can arrow a system that we can follow to validate our databases.

A

So, of course, the explainer how how we define is good or bad. So that's something we just list here. So hopefully we have some comedy or something to have some standard to set some standard for us to see. Okay, what candidates we need. So let's hear what is some some of the specifics about what we really need for our database like you should really Cuban or not. It can reduce you to persevere it or not. It's accurate or not so Nick.

A

Terrific examples we support on if ever born in the system actually reckon just very small mantra- can really reduce the other increase, the licorice quite a lot.

A

So what accurate is that there is and for multi-camera system so accuracy and that contaminations of a key issue, especially high temperature experiments, so open and the coil system, so especially for some of our other system and the incongruity paralyzation, so hopefully in data infrastructure we can get a early simulation okay, so we hope that can get a benefit if another data infrastructure so like, for example, of a chart. It's a center for hierarchical maturity, sign it's in to partner with the materials he see. A seam image.

A

I took him to ok the program, so that's what we will emphasize that it can generate scan a prototype system data. So that's what we really need it and we will actively participate in this program to provide our software to the users. Then they can generate some prototype there for us to use to develop a basis. The other data sources that we are interesting is that data repositories like, for example, at least they have experimental data repositories, especially for diffusion thermodynamics, and you can probably Stella base.

A

That's what we really need- and hopefully here can help us for all this data. Okay. Okay, to summarize, confer is a faith-based approach, tradition. It is for stamina, Cafe secret information, but now is a beyond as seminaries. We provide all other humans, so we keep working on software and a databases, and we face a lot of challenges also, of course, so, hopefully, the other data infrastructures can help us about on this. To address this issue. Okay, thank you for attention.