 Welcome everyone. I'm Marty McCann. I'm the the chairman of the committee on geologic and geotechnical engineering. Welcome to our webinar. This is the third or fourth depending upon how you count webinar that we've hosted on various topics in geotechnical and geologic engineering. I'm just going to say a few words of introduction here, and then I'll turn it over to our moderator and speaker to get started. CAGA, as we refer to it, is a standing committee of the National Academies and it is a committee under the board of Earth Science Resources. And we overlook studies that are that are done in the geotechnical and geologic engineering area. And we've been hosting these webinars as I said now for a few times over this past year. One of the things that we try to do is be a resource for the profession. So Sam Magseno, our director whose email you'll see at the end of the presentation, is your point of contact for getting information about the committee and the committee's work. I did want to make everyone aware, particularly given that you're attending this webinar. The committee has a meeting, October 22nd, part of which is public, and the subject of the the open session will be the same topic that we're talking about today. We'll be having a number of panelists speaking and have open discussion with Q&A and it will be web broadcast and participants online and those in the room are welcome to ask questions and to participate in the dialogue. You'll hear a little bit more about that at the end and you'll be getting an announcement about that for those who are now on the mailing list and registered for this webinar. So we encourage you to participate in that. A little bit of a disclaimer, this of course is a webinar hosted by the academies, but the views and opinions that are expressed are those of the speakers and there's no Academy endorsement of those views or of the methods that are being presented. And lastly, I'll introduce Scott Anderson of BGC Engineering, formerly with Federal Highways. He will be today's moderator and I'll turn it over to Scott to introduce our speaker and the topic for today. Thank you, Marty. Good morning or good afternoon, everyone. It's my pleasure today to introduce Dr. Kanichi Soga, who will be speaking today on the large-scale deformation modeling that you see on the screen and the animation that's being shown there. Hopefully you're all seeing that. Dr. Soga attained his bachelor's and master's degree from Kyoto University in Japan and his PhD from the University of California at Berkeley. He has over 350 journal and conference papers and he's co-author of the third edition of the Fundamentals of Soul Behavior with Professor Jim Mitchell. That's a volume that many of us are familiar with, I think. He's also a fellow at the UK Royal Academy of Engineering and a fellow of the institution of civil engineers and the recipient of many awards including the George Stevenson Medal and the Telfer Gold Medal from the institution of civil engineers and Walter L. Huber Civil Engineering Research Prize from the ASC. So it's really a great pleasure to have him here today and to share his experiences and ideas in this area. I do want to say that before I hand the mic over to him, so to speak, that there is a Q&A button at the bottom of the screen and that's the way that you can ask a question that we will answer at the end. That's where my moderation skills will be evident or not, but that's for you to put the questions in as you're listening to the webinar and we'll be monitoring that and get to that about 30 minutes into this hour. Please don't use the raise hand and the chat kind of thing. We won't catch those messages. All right, was that kind of introduction? Dr. Soghi, it's all yours. Thank you, Scott, for your kind introduction. First of all, I'd like to thank the members of the Committee on Geological and Geotechnical Engineering for giving me this opportunity to present the work on large deformation modeling for geological and geotechnical engineering. Well, you may wonder why I show this particular animation provided by Professor Jory Taran, who is an applied mathematician at UCLA. He and his colleagues use a particular numerical technique called the material point method, NPM, to simulate a snowball rolling along a slope. In fact, this method was adopted for some scenes in the Disney movie Cold Frozen. Before I give my presentation, I'd like to acknowledge the contributors who helped me prepare this presentation. They're from UC Berkeley, Cambridge University, where I used to work for the Anwar 3D NPM research community and colleagues from other universities. The material point method NPM allows us to simulate large deformation behavior of a material. Here are simulations of fluid structure interactions conducted by Professor Zhang of Tsinghua University and Professor Arduino of the University of Washington. Here are some simulations from Professor Ken Cameron of MIT, showing a steel bowl shot into sand from different angles. The top figure shows a steel ball penetrating into soil surface at a large impact angle, whereas the bottom figure shows a ball penetrating at a smaller impact angle. Large-scale runouts resulting from landslides or failures or containment systems have resulted in disasters such as the 2014 also landslide in Washington shown in top right photo. We all know that there are different types of landslides as shown in this slide. The top figure shows a flow slide where failure mass continuously shear inside and becomes a flowing debris. The middle figure shows a slide with a continuous failure surface. Most of the shearing happens along the failure surface and this is where energy dissipation happens. The bottom figure shows a spread slide in which multiple shear bands develop a sequence of blocky mass features. An illustration of the spread failure is shown in the photo at the right bottom. In this presentation I'll show some simulation example of different failure types. Another large deformation modeling application can be failure of levies, which requires soil and water interaction modeling. Again, I'll present some simulation of the levy failure later on. A simple physical model to show a flow slide is to conduct a granular column collapse test. Here a sand column confined in a cylinder is suddenly released to spread radially outward by the gravitational force. The soil mass continuously shears and spreads downwards and sideways. One possible way to simulate the granular column collapse problem is to use the discrete element method in which soil is modeled as particles and the interaction forces between particles are explicitly modeled by spring dashboard and slider. The simulation can be done in dry condition or in liquid. The animation at the bottom shows the result of a simulation in which movement of the water in the pores as well as outside the particles is modeled by the lattice Boltzmann method and the pressure forces from the fluid model is added to inter particle forces in order to simulate the soil-poor-fluid interaction. Unfortunately, this type of simulation is computationally very costly. The other option to model large deformation is to use continuum-based methods in which the conventional finite element method or finite difference model method is replaced by other numerical techniques such as arbitrary Lagrangian or Eulerian, coupled Eulerian Lagrangian, smooth particle hydrodynamics, particle finite element method, and material point method NPM. I do not have time to go through each of the techniques today and for those interested in knowing more about it, please have a look at the paper by Sova Et al published in Geotechnique in 2016. The material point method NPM, which is proposed by Professor Soski of the University of New Mexico in 1994, is one of the methods to conduct large deformation analysis. In the first step, entire mass is divided into groups of massive points. At the same time, the computation spatial domain is divided using a background mesh connected by nodes like a finite element mesh. When a computation starts, the information such as mass, volume, and stress of each material point plays inside a given mesh or map to the nodes where that are associated to the given mesh. The equilibrium equations are solved as the nodes like the finite element method. The acceleration velocity field obtained at the nodes are used to compute the strain and strain rate at the material points and the stress state is updated by a continuum-based constant model like CAMCLEI. The density of material point is recalculated according to the volumetric change of the material point. The mass of the material point is always kept constant to have the mass conservation. The acceleration and velocity field also provide the convection term that is rigid body movement of their points at a given time step. Using this convection term, the material points then move to new positions and then the next step of calculation is performed. This is a simulation of 2D plane strain granular column collapse using conventional more cruel model. The contour shows the deviator strain magnitude. A small cone of non-moving part is observed but most of the soil is sheared extensively. The unique feature of MPM are that moving material points carry the history-related information such as pre-consolidation pressure in CAMCLEI or void ratio in North San. Computation is performed at the fixed nodes and the numerical discretized equations can be solved in explicit, implicit or combined manner depending on the problem. So the computational framework is similar to the finite element method. This animation shows granular column collapse cases with different fission angles. As the friction angle increases, the extent of lateral spreading decreases as expected. We can modify the friction angle or dilation angle of the more Coulomb to be plastic deviator strain dependent. That is friction angle start from maximum one but then with an increase in plastic deviator strain, friction angle can reduce to the critical state friction angle. At the same time, dilation angle can start from the maximum one but it gradually decreases to zero. This is called more Coulomb strength softening model and used in practice to capture the variation in shear mobilization with increasing deviator strain as shown in the right figure. The model is history dependent one in which the history variable is plastic deviator strain in this particular case. It is important to know that in this model, the solar behaves elastically until the shear stress mobilizes to peak friction angle. Another alternative to model the stress softening behavior is to use a model called North Sand proposed by Jeffries. The model utilized the critical state concept. The mathematical structure of this model is that it has a yield surface in which the current stress state lies on, then it has an out of bounding surface determined by the current void ratio or the state parameter. The relative difference of the two gives the plastic hardening processes. By using this concept, it is possible to have a plastic deformation from the start of the shearing rather than after the peak strength in the more Coulomb strength softening model. This means that if we have a strength softening behavior of a dense sand or relatively loose sand, we have two options. One is to model it by more Coulomb strength softening model and the other is to use North Sand model. But the former exhibits elastic deformation during initial shear mobilization wherein the latter, the entire process from the beginning of the shearing is modeled as elasto-plastic behavior. Both models contain history variables and hence the use of MPM is attractive to simulate the large deformation behavior of glanear collapse. Here we see two simulations using two models. The model stress strangulation ship is from the loose one shown before. The top is more Coulomb strength softening model whereas the bottom is the North Sand model. The failure mechanism looks similar but note that the roundout distance is smaller for the North Sand model. This is because plastic dissipation North Sand occurs from the start of shearing. For the upper one plasticity only happens when the stress state reaches the more Coulomb failure line. The results illustrate the importance of considering plastic deformation and energy dissipation mechanism for this particular problem. Here are simulations of dense sand case using more Coulomb softening model and the North Sand at the bottom. We can notice the difference in the failure mechanism for this particular case. The difference in constituted model result in different failure patterns and failure shapes. So sometimes the choice of constituted model influence the roundout behavior. As discussed earlier, strength softening behavior so is one of the ingredients to initiate localization and bifurcation in shear band formation. The progressive development of localized shear band potentially leads to catastrophic failure as we see in many landslide failures. The strength softening drain shearing of dense sand was illustrated earlier. In undrained condition, the large positive excess pore pressure develops during shearing in very loose soil or sensitive clay due to contractive nature of the soil structure. Seepage also leads to reduction in effective stress and softening of the soil. In dense sand or heavily overconsoiled clay, the shear band often exceeds negative excess pore pressure. However, since the shear band thickness is often thin, water can locally migrate from the surrounding and the shear band itself becomes partially drained or drained. That is the localized drain shear band in undrained soil gives softening. The energy dissipated inside shear band has to be converted to heat and there will be local thermal expansion of the soil. Since the thermal expansion of the soil skeleton is smaller than that of the fluid, excess pore pressure can develop in undrained conditions, which in turn softens the shear band. This was discussed in the recent ranking lecture by Professor Eduardo Alonso. The ray effects of soil include creep and stress relaxation will lead to structure softening and the forced entry of water will also soften the soil. In this presentation, I'll show some simulations of the ones highlighted in red. The most well-known example of strain softening is landslide in quick clay or sensitive clays in which the large positive excess pore pressure develops during shearing by the soil structure collapse as shown by the stress relationship and the stress path of such clays in undrained direct simple shear test presented by Professor Lokot in 2015. Chen Wang and Professor Bipo Haolada of Memorial University of Newfoundland implemented strain softening model in abacus coupled orelarian Lagrangian CEL large deformation code and simulated the 2010 Saint-Jew landslide in Canada. The progressive development of shear bands can be observed from their simulations. This is the final geometry of the landslide. The horse and grabbing feature observed in the field are captured. The simulated failure surface matches qualitatively well with a sliding location determined by comb penetration testing. The important point here is that the large softening in undrained conditions and the initiation of failure on this particular geometry led to this failure pattern. Further details can be found in several papers published by Professor Haolada. We all know that seepage force reduces the effective stress leading to softening of the soil. This happens in the soil that starts from unsaturated state in which capillary forces at particle contacts provide apparent cohesion. Which can which then reduces by saturation and then further decrease its strength by decrease in effective stress. This is illustrated well in the stress pass presented by Dr. Scott Anderson who is moderating the stock in their paper published in 1995. There are two possibilities in NPM to include the seepage force. One approach is to model the soil material point as the mixture of soil skeleton, water and sometimes gas and each material point is partitioned in different phases. The movement of soil and pressure of the fully phases are computed at the nose. The other option is to create two layers of material points. One for soil skeleton and the other for water. We call this two point formulation. This simulation led by Dr. Jero Virginia Tech shows a failure of initially unsaturated slope using the first approach. The suction is positive in this particular contour plot. Initially the slope is under high suction. Heavy rain is simulated by increasing the pressure at the slope surface. The pore pressure propagates from the slope surface and a shallow surface failure mechanism can be observed. In 2015 Dr. Bandera and I published a paper describing a theoretical formulation of multiple point mixed phase system. The model contains both soil material points and water material points. This simulation saw seepage induced soil failure. The movement of soil material points are tracked in the top figure and liquid material points in the bottom figure. The liquid material points seep in the soil pores medium. The seepage four reduces the effective stress in the soil material points then at some point the seepage induced failure starts as shown in the red contours. I would like to present a video showing a sand levy model leading to seepage induced failure which was conducted by Dr. Mori and his colleagues at the Public Works Research Institute in Japan. The half model levy has a three meter wide crest and the height is three meters. The water table at the upstream boundary is two point three meters causing seepage into the levy model. The soil above the water table is unsaturated by having a small amount of simulated rain. The person standing here is holding a memorabella. The video shows the progressive nature of the levy failure. There are blocky features with tensile fractures which is characteristics of capillary induced cohesion of unsaturated soil. The failure is catastrophic. In this two point MPM model the soil material point above the water table has a suction to keep the material effect the effective stress relatively high. If the soil material point crosses the water table mixes with water material points then the effective stress is reduced to zero as it loses its capillary force to bond the soil particles. This results in softening behavior. The top figure shows the shear strain development. The middle is the pore pressures and the bottom is the vertical effective stress. As the seepage force starts to reduce the effective stress especially at the toe the slope starts to move. Shear band develops due to brittle nature of the unsaturated soil and some blocky features are observed. The soil model is a simple conventional more cooler model with constant critical state friction angle with zero dilation angle. If we artificially make the dilation angle to be a negative value for example minus five degrees in this case the soil becomes very contractive and positive excess pore pressure developed due to undrained shearing. This leads to catastrophic failure of the slope with multiple shear bands as shown in this animation. It is to note that this is just a simulation to show the capability of the MPM. It is very unlikely a soil has a constant negative dilation angle. The two-point formulation allows us to investigate interesting soil fluid and adhesion problems. For example one problem may be soil fluidization problem in which the water volume fraction the soil water mixture becomes more than a certain level that soil particles do not touch each other to transmit its effective stress. Then the soil water material becomes liquid like material. The cutoff criteria can be a certain value of prosity as shown in this slide but other possibilities can be considered. An overtopping an erosion or model using MPM as shown in this particular movie. Professor Turan and his colleagues created a very nicely rendered animation of seepage induced dam model failure with some cohesion applied in the soil particles. MPM has a unique feature in which the material points moving points can be used for rendering. This is why it is attractive to be used for animation with stunning images shown in this particular slide. My final example is simulation of submarine landslide as shown in the slide. A movement of an ancient landslide was simulated. The source areas identified from the geomorphological feature from the basimetry data. The aim of the simulation exercise was to investigate the applicability of large deformation MPM code to simulate such landslide. There are several key features of submarine landslide and they can be observed from this movie of sliding clay model underwater. The interaction between the sliding mass and the surrounding water can be observed especially the fluid drag along the top surface and the hydro plane like feature at the bottom. This figure is a schematic diagram of submarine land run out and its possible boundary conditions. First of all because of the ambient water there is a large drag force of skim friction on the surface of run out. Due to this suspension flow may be generated but this is less important in terms of impact because its density is much smaller. We are more interested in water entrainment which is an intrusion of water beneath the front of the run out. If there is mixing of water and soil the base of resistance is reduced which could explain the larger travel distance of typical submarine run out. The clay at this site had a very large water content and laboratory tests show that it is a sensitive clay. Hence cam clay model was modified to capture the soil structure degradation due to shearing. By doing so strange softening due to positive excess pore pressure is modeled. Since the movement happens very quickly the clay remains undrained as shown by this state path on the right figure. To model the mixing of the water at the sliding base it was assumed that the void ratio state of the bottom part of the slide mass increases along the critical state line in the specific volume of void ratio mean effective stress plot. With that increasing water entrapment the void ratio increases and this is associated with decrease in mean effective stress. This in turn reduces the shear resistance as shown in the stress path plot on the right bottom figure. This model was assigned to the base of the sliding mass as shown in the top figure. These figures show various scenarios with different water entrainment conditions. As the figure moves from left to right the amount of entrainment increases hence the run out distance increases as expected. Again the feature of NPM to incorporate a history dependent constituted model in this case cam clay model with water entrainment history is highlighted in this simulation exercise. This is a summary of today's presentation. Simulation tools to conduct large deformation modeling of geotechnical and geological problems are becoming available. These tools can be computationally demanding especially for 3D problems. However fast computers such as high performance computing GPUs, multicore processors are also becoming available so they are within our reach to conduct 3D simulations. On the geomechanics side of large deformation modeling I believe that we need better appreciation of failure mechanisms and energy dissipation processes. Although this presentation shows some example to large deformation results I believe that more theoretical developments are needed such as improvements in localization modeling such as shear band fracture, free surface interface and multi-phase porous needed material. We also start to see the use of these methods for other problems. You'll find papers that describe simulations of pile driving, tonal excavation collapse, comb penetration testing, soil construction equipment interaction and geological modeling of land formation. Finally I would like to propose that we need to have a better discussion, more discussion on the value of these large deformation modeling approaches to geotechnical and geological engineering and beyond. For example what is the value in extending our conventional stability and deformation analysis business to include after failure stress testing analysis that is understanding the consequence of failures using these new tools? Can we make better engineering judgments and to aid stakeholders decision-making processes using some stunning vigils that we saw in this presentation or for better communication of the results? If we do embrace these tools and approaches I think we geotechnical and geotechnical and geological engineers have an opportunity to become the leader in this area. If you are interested in this topic the University of Cambridge team is hosting the second international conference on the material point method this coming January. The conference will have research paper presentation, training courses of NPM code called ANR3D and publication of a new NPM book for geotechnical engineering. Thank you very much for listening. Thank you very much Kenichi, it's very interesting and some of you have already started putting questions into the question and answer icon I guess at the bottom of your screen and I encourage you to keep doing so and we have plenty of time left now to address some of those questions and I'll start with some of the early ones that came in and you know one of them is interesting to me is has the NPM been used to model liquefaction due to cyclic loading? The answer is yes I've seen some but then I think we start to see more at UC Berkeley we have a PhD student working on that particular problem. Thank you. How important is it to get precise values of the sheer parameters of the soil? How sensitive is the modeling to that? Would you mind saying that again? Sorry. How important is it to get precise values for the sheer parameters of the soil in your constitutive models? Yes I think it depends on the problem and if you see a sort of the granular collapse problem which the runout distance seems to affect by the initial geometry of the mass then the precise modeling of the stress strain blade behavior may be very important. However if you have a lot of shearing going on and most of the material become critical state or some sort of a large deformation shear straight maybe the initial stress state doesn't matter and therefore the application of critical state friction angle may be applicable so I think it really boils down to the models that you're looking at and how the landslides are behaving as it moves along. Okay so related to that question and maybe a follow-up would be is there any different kind of soils testing that you would recommend or are we doing the right things currently? It's a very good question and for example a large deformation I understand there a lot of work done in the ring shear testing which will give some idea of the how the material behaves in the shear band if it moves very long but then I think we can do more in terms of how shear band develops from continue to localize behavior and I've seen a lot of work done in the past in finite element method but how can we use that in this sort of MPM or new techniques we need to improve upon. We already see that something like in hydro fracturing when FEM adoptive X FEM which is used to simulate fractures I think we probably need to do similar things in the other techniques like in MPM. Okay a couple of questions are related to the computational cost of this method versus some others and in particular I guess I had the same question about the two-point MPM over the one point what the computationally complicated is this get? The two-point formulation is complicated because we have to solve both two equations simultaneously and then you have more material points in general compared to one material point and this is why in we use one material point for three phase problems like soil skeleton water and gas whereas two material point becomes an interesting one to pick although it's computationally expensive in certain cases the water can like we see in the erosion problem you see an interaction of the water particles where water is flowing as a material point and then interacting with porous material which is a soil skeleton now you can argue that yes we can put some material point for air or gas but then that becomes computational expensive and I don't think we have seen anybody done that yet maybe we can do that now the code has to be run more efficiently but as I said in my summary slide high-performance computing and GPUs and other techniques are becoming available from the computer computational side and for example we are working on to develop much faster codes using GPUs and multiple processing units so could you give us an idea Kanachi in terms of the you know current processing capabilities that a practitioner might have how long would it take or how would they go about running one of these models yes so so for example most of the npm code use explicit sort of our integration scheme so so sometime it may take hours to run the simulations there are some implicit like what we see in implicit fine element code that you can solve with a much bigger time steps and then there are situations where you want to use a combined one where you have a sort of a combinational explicit and implicit for pressure calculation for example but again I have to say that this is still in the academic stage and as the codes that available called annual 3d can be used for some engineering applications and we start to see a lot of people using this code for their problems and again it will take a few hours to run of 2d like problems okay here's another question you know the existing apparent fluid models are computationally efficient like Dan 3d and rams and titan 2d and and they have been used in practice to solve engineering problems and help and design what advantages do you think mpm brings in modeling large you know flow type landslides like debris avalanches over this apparent fluid approach yes as I said in my presentation I think one of the attractiveness of the npm or either these techniques is that this material points will carry its history as it move along and a soil can typically has its history variables like pre consolidation pressure which may change as it gets sheared and changes volume and therefore I think the techniques like npm I showed is attractive in that particular case I think that's the difference we see and I'm not sure I'm not really familiar with the other codes that you mentioned so I can't really comment too much in terms of advantages and disadvantages so it sounds to me and I'm not an expert in this area but it sounds to me like the advantage would be the you know no constitutive models are applicable to the npm method where they really aren't to the others so maybe it's a it's a comfortable step for the practicing geotechnical engineer in using constitutive models they're familiar with yes so I agree that we can use the models that you're familiar with and implement it let's say in 2d or 3d like you can do a 2d plane strain analysis to see how the material is shearing and also sliding in a large deformation if you use a depth average technique for example which is often used and which is very useful but only looks at more or less the interface between the sliding mass and the basal resistance so so I think different methods have different capabilities and I think that will be a good it's a it'll be a good discussion that we can have within the community which programs are used for what problems okay I guess after that comment on familiar constitutive models as the model been used for snow avalanches yes so I see my colleague from UCLA professor Joey Taran has a paper in nature recently which shows his avalanche simulations and all the real ones whereas you saw at the first slide a more of a one that you see in a frozen movie okay I'm having a little technical difficulties here with a frozen screen speaking movie frozen so bear with me for a moment while I try to get this thing unlocked if uh yes Scott this is Marty I have a question so maybe I can that's perfect I can give you some give you some buffer here uh Kenichi um I'm curious of your thoughts with regard to uh the potential of these methods to model uh the the initiation and and progression of uh internally erosion and embankments and I ask it from the perspective a little bit different than your your simulations that you showed in the sense that um internally erosion may may initiate and develop over very long time scales um and on the one hand um you know simulating longer periods of time may be difficult um computationally but on the other hand it may it may be a unique tool to to sort of model the potential that these events could take place have you looked into that or have others and and what promise do you see in uh in that particular area yes that's a very good question and actually uh Dr Alba Iero from Virginia Tech when she was at Cambridge we had a project to trying to model the internal erosion process using npm which is to point npm and the idea behind that was that if you have or you can do it with single layer um npm but the really looking at the mass transfer from one phase to the other as a representation of modeling the internal erosion that is that some of the particles in the soil will move to water and water becomes a muddy water and a two material point method will be interesting to use we haven't finalized that yet but the idea is that the the mass in some of the solid material points will be passed to the liquid material point and that muddy liquid material point having a little bit density change or when they potentially change with viscosity will seep in to the outside of the porous material the challenge at least at then in in any case in the erosion process or is what are the criterias in terms of starting the initial erosion internal erosion and then what is the rate of mass transfer from sort of a sort of a soil within the porous material becoming mobilized and flow into outside so I think the the framework was there I think the question really becomes how do we model these processes did that answer your question yeah it it did it sounds like it's very much a work in progress but I think it's also fair to say that the as you pointed out simply the initiation of the internal erosion is is a difficult thing to define and under what circumstances that would occur but yeah thank you yes the other point I wanted to add was that the the the old code that we used was more of a explicit code which requires a lot of small time steps and if you simulate water system in explicit code you have to use very small time steps whereas recently we've been sort of a going more implicit pressure calculation which allows us to go much larger time step and hopefully we can simulate large much larger time frame that we we we couldn't do before so that is also in progress thank you Marty I need to ask you to continue on for a second here I'm still having some technical difficulties so perhaps you can solicit some other comment while I try to resolve this okay I do have a general question Kenichi that that you sort of alluded to in your in your reference to the paper you showed earlier and that is that there are other methods out there and your paper discusses those can you give us a general understanding of benefits of different methods to address different problems you know are certain methods seeing a bit more advancement in terms of the type of I'll say pragmatic geotechnical geologic engineering problems that are being applied or are all the methods pretty much on the same plane in terms of their application at this point in time that's a very good question and it's really my personal opinion is that I think npm has been embraced by many of our colleagues more than the other methods perhaps because it's a little bit easier for us to understand because we understood the finite element method for example and it's very similar to how finite element is done but how do you update your material points using convection term is something that it's easier to interpret and therefore I see many researchers embracing npm having said that as I showed in my presentation like the one that I showed with abacca ceo that professor how louder at memorable university in newfoundland use that technique to really simulate very nice feature of failures that we see in sensitive clays so I think it's really trying out these different methods to see what is the best approach for certain problems it's a good discussion we can have having sort of conferences like what we're trying to propose what's going to happen in january at Cambridge also brings the community together to discuss what are the possibilities that we can do for example with npm but I know that other disciplines have their own techniques and so like particle fem is using other disciplines and they have conferences sph smooth particle hydrodynamics sph is being used in different areas and I saw some of the geotechnical academics use sph in very effective way for landslides uh the sph I had we also tried sph in the past and we had difficulty how to treat the boundaries and that sort of thing but then that's always a problem for any other methods and so what I'm trying to say is that I think there are communities developing I think there's a flavor of npm but I'm just I want to see the discussion and then we really see the sort of advantage and disadvantage of npm and I think that's what we're trying to do so uh thanks thanks you can you cheat I have a question that's maybe a little bit different we've got some other technical ones that are coming in too but when I when I think back and when I first started seeing deformation modeling fem and finding different stuff going in slope stability analyses I I thought it wasn't going to be long before the community evolved to doing that kind of work on a routine basis yet yet I still see a lot of work that is limited equilibrium base um yeah the models you showed today were fascinating and and be easy to get yourself in a place where where you didn't even believe that there was really physics behind them they were just interesting images uh but do you have any thoughts on on like how we could uh as a profession you know bring this type of modeling into a wider use maybe uh more rapidly it's a good question as you say um we are moving from liquid equilibrium method but we always use the slope stability slip circles and all these things for certain problems and then uh we go for finite element method or finite different methods for looking at more of a deformation issues and failure patterns and um I think these large deformation problems extends from that to go to much larger deformation to see what kind of mechanisms that we'll see as the soil start to move and go further so in some cases you may not need that for certain reasons but then sometimes you need it because the community that is exposed to this risk needs that to understand what the consequence of potential failure can be so so so I think it goes both ways we need to start developing our capabilities to show and build confidence in using these techniques and they took say 20 30 years for us to become like that with the finite element method it may take like that for large deformation analysis I hope it won't be but we have to build our confidence in using these techniques and at the same time we want to ensure that there's a value for stakeholders on this particular approach and I think maybe in the October session that you'll have you should it'll be good to discuss with stakeholders and clients what will be the value of having this kind of tools unless we embrace that I think it'll be very difficult to move forward yeah thanks for those thoughts yeah one question that's come in and maybe it's a way to answer that last one too is is um you know will the MPM method have application in geologic geotechnical disaster emergency response in other words predicting secondary failures or hazard identification uh what are your thoughts there that's a very interesting question and I do hope that may happen uh to be one of the possibilities that if you if this happens what's going to happen next but then you want to simulate in a very short time to ensure that you get sort of a confident results so so I think I like the idea of that particular one and using these tools to sort of see the potential secondary hazards again that may be also related to monitoring of the slopes or monitoring of the infrastructure to ensure that where should we monitor to ensure that there's no catastrophic failure in the secondary one because if it's just moving slowly maybe there's a possibility to do something within a given time but the always the danger is that some failures happens with a little bit movement but sudden collapse and I think that's what we want to avoid and I I'm hoping modeling a sheer band more carefully and then doing that kind of simulation it will allow us to understand the catastrophic failure which may be uh something that you don't want to have as a secondary failures that that that brings up a question I had to it you did show the um a photograph certainly of the osso or the sr530 landslide in washington um and I've you know maybe this was certainly an issue there was you know what is the safety of the first responders after an event like that and being able to do some analyses you know would have been help of this type would have been helpful there and I just sort of related to that you know have has the npm been used to model what happened at the osso site yes I didn't have time to present today but we did work on osso landslide to these analysis and depth average analysis using npm I know that there are other colleagues in the community using different techniques to simulate the osso landslides we did a simulation for a certain purpose and we we get the results of course there are a lot of assumptions and the problem is very complex and therefore I would like to see the uses of these techniques but also like to see a lot of discussion of why it happened what are the so these site conditions what is the geology what are the properties I think more discussion because I think by working on it I felt that it's a very complex problem so some of the questions that I have available to me here can each of you know is the you know questions can particle shape effect be somehow considered in this approach and yes but why don't you explain that yes so particle shapes becomes important when you use discrete element method which was the first option I showed when you see the granular collapse where you meet individual particles and therefore having particle shapes do make a difference in terms of runout now in the npm model or other models that continuum method particle shape will represent as a stress strain relationship and it's the stress strain relationship that we're modeling as an inconstituted model and that particular constituted model with certain particle shape will have a unique some sort of a stress strain relationship that we're curve fitting which will be used in the npm simulation therefore particle shape can be incorporated by changing the parameters of the continuum model if it allows to do and then trying to simulate in that way very good I think we're going to get Sam Maxino here who's on the call to ask a couple questions so hold on one moment we have about five minutes left if you if you have burning questions still we'll get to the ones we can what one thing I can say while Sam is getting ready is that we will post these slides and the recording of this webinar it takes about a week's time afterwards for us to do that but they will be available at that point thank you Scott this is Sam Maxino sorry for a little bit of technical confusion but I think we can probably get through one or two more questions one question that came in normally we have limited soil properties information to to landslides or dam or a dam is npm a good tool for debris flow occurrences prediction or dam failure prediction any numerical model requires good material point model and good material property parameters input so if you can get good thing for a final element model yes you need to have a good input parameters for npm so I can't say that npm will be better than the other methods for that particular one you have to have a good material properties and material models to simulate that okay and another question is this a coupled seepage and deformation model if not how do you combine seepage forces into the deformation model it is coupled so the seepage part of the presentation I showed two different methods one is use material point but then each material point do the coupled analysis the second option I showed you was that material point for water and material point for soil and they are interacting as they work together and that's also coupled so the seepage part of the presentation is fully coupled analysis and let's see we have about three minutes left let's see if I can pick a question that we'd be able to answer in that time how much solid fluid interaction modeling is robust in the npm method the question of robustness is also a tricky question to answer because it depends on who you talk to for example we recently published a paper in terms of how to model the fluid more sort of accurately using what we call implicit method and because the book modulus of water is high and therefore running in explicit sometimes requires very small time step and you'll see that in a lot of pressure oscillations when you use explicit calculation so so so my answer to that is that yes we're trying to improve and there is a progress in that particular area and I think one more question can npm be applied to modeling a brittle solid material like rock yes you can uh the question then becomes how do we measure model fracture and uh shear fracture tensile fracture and uh the as we see in x f e m using in fine element the npm can simulate opening of the crack by material points moving apart but then the mathematical treatment needs to be improved to may have a better accuracy in that particular one so so answer is yes uh framework allows you to do that and um we have about we have about 20 more questions but unfortunately only one more minute so i'm going to hand this back over uh on behalf of uh scott anderson and the rest of the community on geological and geotechnical engineering um we'd like to thank you kenny chief for participating in this webinar and i'll hand it back over to our committee chair marty mccann for final words yeah thank you sam and kenny chief thank you very much for for bringing your your expertise today uh to our to our webinar and introducing uh what was a pretty large audience uh to these methods and the and the potential they offer and and obviously as as i said earlier um we're pretty excited about it and so our next meeting is going to devote an awful lot of time to further discussion of of some of the topics that you brought up so again thank you very much for agreeing to do this and um your presentation and answering all the questions uh and thank you to all the participants for joining um we will be having more of these um roughly about every quarter and if you have suggestions for future webinar topics feel free to send them in to sam and the committee will will consider those and we invite you to our meeting um on october 22nd when we discuss this further so thank you all very much