 Welcome everyone to our next installment of our committee on geotechnical and geologic engineering webinar series. Happy to have you here. My name is Marty McCann. I'm the current chair of the Kaga committee, which is a standing committee of the National Research Council at the National Academies. Before we get started, just wanted to pass along a few, a few messages. One is I wanted to thank our staff. Sam Maxino, who's the director of Kaga. She's a senior staffer at the National Research Council, and Remy Shapetta, who is a staff assistant for the committee. I'm the chair of all of the organization and logistics that makes this happen. So Sam and Remy, thank you for your for your time and efforts in putting this together today. I'd like to now turn the the webinar over to one of our Kaga members, Professor Pedro Arduino from the University of Washington. He will moderate the session and go through a few logistics about asking questions, and then introduce our speaker. So, Pedro, with that, I'll turn it over to you. Thank you, Marty. Good morning and good afternoon everybody and welcome. Thanks for a lot for joining us today. As Marty mentioned, my name is Pedro Arduino, and I am a member of the Committee on Geological and Geotechnical Engineering, also known as Kaga. I have the pleasure of serving as your moderator for today's webinar on my daily stamps, together with John Stamatacos, who will help me collecting and organizing some of the questions. We are certainly delighted to have Dr. Alan Mark can join us. Before turning the microphone over to Alan, I want to review a few things on how to provide questions. To ask questions, you can use the question and answer box, which you can locate by hovering your mouse near the bottom of your screen. You should see a bottom label Q&A. Please type in and send your questions there and not in the chat feature. You can send those at any point during the webinar. We are pleased to have so many of you interested in this webinar, but I will ask you please to realize that we won't be able to answer all the questions if there are many of them, but we will try to do our best. Should you have any technical issues, please use the question and answer feature. Additionally, any conclusions or recommendations provided by Dr. Mar are his own and should not be thought of as recommendations from the National Academies or the Congress. So this is just a disclaimer. With that said, let me introduce Dr. Mar. Dr. Mar, I am very pleased to do this very well. By the way, Dr. Mar founded and leads Geocom, one of the foremost providers in the United States of America of real time web-based performance monitoring of civil engineering structures, including dams, deep excavations and tunnels, among others. He also has extensive experience in testing and measurement of mechanical properties of earthen materials, designing earth structures, determining the cause of poor performance of geotechnical structures, developing cost-effective remedial measures for trouble projects and risk management. He's an elected member of the U.S. National Academy of Engineering and of the Malls. He has published widely and has given invited keynote lectures around the world. Dr. Mar has extensive experience evaluating the stability of tailing dams and other waste retention facilities and monitoring their performance. I am very pleased to introduce you, Dr. Mar. So Alan, please, the floor is yours. Thank you, Pedro. I appreciate those kind words. Just double check here. You can see my screen. Hopefully that's okay. The topic today is focusing on why do what I'm calling here mine waste impoundments also known by many as tailings dams experience stability failures. This is a particular geotechnical focus in what I'll talk about today. I will try to deal at a somewhat high level to set the scene and then get into some specific geotechnical issues toward the end. I'll try to go for about 35 minutes to leave as much time for questions as we can. And we've extended the time a bit to try to get in as many questions as possible. Let's see for some reason. Sorry, I just had a little trouble going forward. I will cover a brief review of some of the significant failures that have occurred to tailings dams around the world over my lifetime. And I've chosen ones that had a specific particular impact that I'll describe. I'll then draw from that some characteristics of tailings dam failures that seem to be common and that should really influence how we approach tailings dam safety. I'll review some of the key aspects of geotechnical failures in tailings dams. And a lot of this gets into sheer strength behavior of the materials comprising the dam and of the retained tailings. So I'll review some of the sheer strength concepts for granite materials. I'll list some design goals that ought to be on one's to do list for if they're responsible for the design of a tailings dam or for the review of its safety certification of its safety. And I'll summarize with a slide. Again, these are my views and not those in any way related to the Academy. Just a scope, what do we mean by tailings dam might refer to them more as barriers that are made of earthen or waste materials that retain particulate waste upstream of the barrier and that are placed by where those waste are placed by sluicing at high water content. So I'm dealing mostly with hydraulically placed waste materials behind some form of barrier. Barriers are referred to in the industry by different parts or different groups as dams or impoundments or ponds. All of these share some of the geotechnical characteristics that we're going to talk about. The particulate materials themselves vary in how they're called or referred to as well as in their properties. They're referred to as tailings or washings or ashes or slags chemical byproducts like gypsum produced in the phosphate mining industry. The placement varies by industry and the mechanical and physical properties can vary widely even within one tailings dam, they may vary vertically and horizontally very significantly, which makes it a much more complicated problem to try to deal with than many other typical engineering situations. So let's review some of the failures that have occurred. I chose this one because it was the first one that I had the opportunity to work on when I was still a young engineer in 1971. It was common in Florida in the phosphate mining industry to build tailings dams of sand that were produced as a byproduct of the phosphate mining. So you you mined it or the phosphate you got sand and you got slimes clay slimes and so we would use the sand to build a dam and we put the clay slimes behind the dam. This this particular failure occurred suddenly no warning 45 meter 45 feet high the clay slimes flow downstream and had a lot of bad consequences. The failure was caused by a geotechnical stability failure of the homogeneous dam due to the fact that the flow of water through the dam was not controlled in any way. So we had a condition of higher poor pressures that destabilize the dam. The clay tailings went 120 kilometers down the river with a very large fish kill no life loss. But the significance was in Florida became the first use of the courts of a strict liability doctrine for hazardous use of land. And so this widely greatly broadened the potential liability of an operator of a tailings dam first in Florida and then expanded out into the US. It also led to the implementation of design regulations for tailings dams in Florida that were developed by the engineering profession became a part of the laws. And one key part of that in fact was I think the first use of a regulatory requirement for minimum factor of safety on a tailings dam. So it really changed the rule of the day and how we do these things in the United States kind of in a separate industry but also very important was a failure of another type of impoundment in West Virginia in 1972. This was just basically a an impoundment placed across the creek to create a sedimentation basin on which to catch the washings off of mind cold so they didn't go downstream and pollute the river. It rained heavily. The water got up and was flowing through this homogeneous dam caused it to burst suddenly the water and the tailings the washings all go downstream. The town of Sanders is only about a mile or so downstream and everything was wiped out there as well as many people killed. The same time you talk about catastrophe this the pile of coal ash just down or waste just downstream of that was on fire. And when I was at the site three days later looking at this failed mass and destruction down the valley. There was also this fire still burning despite there having been a massive flood things it's amazing what happens to us sometimes this occurred suddenly no warning. It was a collapse caused by slumping and sliding of the downstream portion of the dam after a heavy rain that the impoundment itself was sitting on cold tailings. So just overall not an engineered impoundment facility at all 125 people died within 30 minutes or so of this because the water went downstream in a very narrow valley and big wall of water waste debris and everything just just washed everything out of the valley. This was significant not only for the terrible damage but it led to the US Dam Safety Act which gave the Corps of Engineers responsibility to conduct a nationwide program for safety inspections of dams. And this was the first time we really discovered how many dams we had in the United States and how many of them were in such poor condition. First really step up in recognition of the importance of dam safety and the big job we had ahead of us in the US. Another one I worked on in 1980 then getting into some of the or iron or the metal business was a 66 meter high dam constructed by the upstream method large area used for copper tailings in a copper mine. It occurred suddenly again no warning was we determined that it was a stability failure through the dam and the tailings that resulted because these tailings are low permeability materials and when you add new weight on top of them. They they develop they don't have time for the poor pressures to adjust to that new weight and so they are they can't gain enough strength so you're adding more weight than you're gaining strength. And that just caused an undrained stability failure classic geotechnical failure but once that containment let go the contained tailings basically turned to mud and flowed down the valley. No one killed in this but a very significant environmental consequence. It took years to try to clean up. Another one in Italy it's not all the United States this was 1985 I'm following the eastern time sequence in which that tragic tragic failure of a tailings dam. Unindated the village of Stavia. This again was sudden no warning tailings dam made by the upstream method which I'll describe in a minute. In this case though the failure occurred through the Silt Foundation. Again a geotechnical stability failure and then that let loose the tailings that developed into a mud flow of liquefied tailings floated a very rapid rate downstream killed 268 people and destroyed a bunch of property. Back to the US in 2008 there was a major failure at Kingston Tennessee on a TV a facility. This is coal ash in the blue shown in my slide here is hydraulically placed. This is the residue from burning coal to produce power. So we start with a starter dyke down here on the right side of compacted earth materials and then we start using some of the ashes as compacted ash as construction material and then pound the hydraulically placed tailings behind this very barrier structure and so you keep building. They got to a certain point this was an active facility still adding weight to it and it experienced a sudden slide. Mass movement that released the ash that then liquefied and flowed miles actually upriver in some instances. Fortunately no one was killed here but it has been a major billions of dollars cost to TV a to deal with the consequences of this. It's also led in this case to you EPA regulations much more strict on building constructing operating these types of it and found it. A expert assessment of this found that there was a very thin layer of mud like the plastic debris left in the bottom on which this dam was actually constructed. And it was failure. The failure was triggered in an undrained stability of that very thin six inch layer of clay slime like material. This occurred suddenly with no warning even though it had been inspected the day before. It was an undrained failure through the thin week foundation layer over four million cubic meters of liquefied ash flowed quite a long distance. A lot of property damage no life lost billions of dollars in cost to TV a. Just two more a couple of three more I guess your Mount Polly up in British Columbia was a major tailings dam that let go in 2014. You can see here on the upper right the breach that developed the tailings are up in the upper right corner of that diagram. The dam itself was an engineered and construct designing constructed facility. This shows a cross section of Mount Polly where the red material is compact Rockville. Then we see transition zones and a core. This is an engineered dam section. This is as the pieces the features we like to see in a well engineered dam and yet it's still failed. It failed due to a weak play layer here in the foundation that got overstressed for the weight that was being added to it. A case where more weight was being added faster than the foundation soils could adjust. So we had an undrained stability fire through the foundation that consisted of our play layer. This led to liquid fashion of the stored tailings followed by mud flow that went into a pristine lake and down the Frasier River for as much as 600 kilometers. It's one of the biggest environmental disasters in modern Canadian history. There weren't life loss. I don't believe in this and you know the monies are less than say TV a experience, but it was kind of interesting to me here that this is a case. Now we're another consequence of these peers and that is the design engineering firms being found responsible for negligence or unprofessional conduct in this and that had a big hit to them and they're they're standing. Couple more from Dale in Brazil in 2015, the largest environmental disaster up at that until that time in Brazil. This is an iron ore tailings facility. Let go and mud and water wiped out several villages below. This is a look by the expert team post mortem. What caused the failure and they were able to determine that it was a stability failure on one section of the dam where using appropriate geotechnical concepts and analyses. They back figured to factor safety less than one for the conditions that existed at the time that the dam failed. This was attributed to high pore pressures being generated by adding tailings to the top of the dam at a rate faster than the tailings could dissipate filling too fast. There was no warning sudden failure. It had been inspected shortly before the failure. Again, it was a case of an undrained failure of the containment dam followed by liquefaction of the stored tailings. There were 19 people killed here and a lot of property damage and billions of dollars paid out by the owners. The last one is very recent in Brazil, the Brimadinho tailings dam. There's a video of this online. I didn't I'm not trying to show it here take too long, but it's for those of you in this involved in this area strongly urge you to go look Brimadinho tailings dam failure and watch this video. It's a real live shot several frames a second of the development of the failure. I've picked out something here that I think is about six seconds after the failure started. The tailings are up here behind. This is the dam portion you can see where my cursor is going. This dark area across here is the exposed scarf from the slip that is developing. At this point it's dropped maybe some 15 meters or so the dam itself is 86 meters high. The front part of this is coming out at us out of the screen. So it's the back parts going down and the front parts coming out. So that looks to me like a geotechnical stability failure. And then the stored tailings which is the black part in the background here. Let go and liquefied and went downstream. 248 people were killed. There's still 22 missing damages that I know I can trace right now or at least $5 billion and no one knows yet where that's going to go. This was a sudden failure. No warning. The dam had been inspected the day before. 12 million cubic meters left this site and again I urge you to watch the video. So what are some of the characteristics that we're seeing here? I think you've seen me try to emphasize that failure can occur quite suddenly with little to no warning. This just seems to be a theme over and over again and something that ought to be at the back of our minds if we're in responsible charge for these facilities. In my view, generally from my review and from my knowledge of these things generally something triggers a failure within the barrier itself. That is that outer part that's holding the creating the containment or within the foundation. And then this triggers a loss of containment of the loose tailings. That then results in the tail the store tailings liquefying and when they liquefy they have no shear strength they flow like molasses. So the key here is to keep that outer barrier portion stable and safe. And so we contain contain the tailings. I like to think of it when I'm designing is think of the tailings as though we're storing liquid just like an earth retention dam. I mean sorry water storage dam and and then make sure we can hold that in place. Then we can avoid these liquefaction type static liquefaction type problems liquefied tailings can flow very fast for very long distances and they can present great risk to downstream people in the environment. There's little time to warn and evacuate people if they're located with directly below the dam. This was a case of remedial or even within a few kilometers of the dam. Visual inspections may not reveal the threat of an imminent failure. You noticed in the cases I reviewed I mentioned several of those had been inspected visually the day before or recent just prior to the failure itself. Most monitoring systems as they've been put in the place in the past will not give us adequate warning and I'll explain that a little bit more toward the end here. These above characteristics should be strongly considered in the design and operation of any tailing stand. They should govern our decision making and our efforts to do the right thing and try to keep these facilities safe. Where are we in this overall waste storage industry and I think it's helpful to realize that these are generally waste materials. So anything we do with them costs money and so a part of any prudent management system is trying to minimize your costs. So waste storage is always driven a lot by how can we do it for the least cost. But that has a consequence. Here is a diagram that Professor Becker and I put together many years ago for a client in the petroleum industry where we're trying to figure out how much risk is acceptable if there's no government standards or guidelines. So we went out and looked at all different types of facilities and tried to assess what was the accepted average rate of failure. And if they failed what was a typical consequence of that in dollars dollars is across the bottom axis annual probability failures on the on the vertical axis. So you see a case like mine pit slopes the consequences there could be managed. And so there's not if we get a slope failure there's not necessarily on average a lot of loss there so we typically mine was steep side slopes and accept failures. On the other hand something like a large earthen dam could potentially take out hundreds of lives. And so we designed those with a lot more a lot lower probability of failure. Typically the data would suggest that our well modern modern engineered dam earthen dam storing water has a statistical average rate of failure about one in 10,000 per year. And you'll see that shown here and kind of the yellow on this slide. We put all this together and Greg and I kind of drew a couple of lines through this stuff saying you know here's kind of an envelope of accepted risk because risk is really probability times consequence. So and then we had marginally accepted risks and our particular client in Japan at the time had a risk right up here. So this was a very useful chart to help say it's too high we need to get it down. So recently I'm thinking about these tailings dams failures I got interested in well where do they sit in this this chart which is kind of a concept idea chart to just give relative comparisons. You could argue with the specific numbers as not being more than maybe an order 100% or so off rough numbers. The data suggests that the rate of tailings dams filters is about one in 1000 we have we have on average about two significant tailing dam filters a year. And there have been studies by folks kind of looking at frequency of losses and one in 1000 is a reasonably representative number and I've sketched that on here with the blue dashed lines. That's average a particular tailings dam might be anywhere here it could it could have a higher annual probability of failure. You could have a higher consequence of failure. Every one of these you almost have to think about it in terms of what is the potential consequence. And then if that's high to shoot to try to reduce the chance of failure by having a more stable or a stronger facility. I think in some ways one pot shot at this as to where are we as an industry today is I just kind of threw out here for concept. You know we're probably up in that pink zone with a lot of our tailings dams that the consequences can be pretty significant in the billions of dollars and the rates of failure are in some cases for some specific cases potentially higher than what the average is. So I think that raises the cause of concern you know are we too high with the probabilities of these things failing and do we need to do something about it. To review a little bit about the methodologies here and why does this happen. Let's look at a typical concept of an upstream method of constructing a tailings dam. It starts with a starter dam which is usually a compacted earthen embankment of some type or other built with some good control and then we start putting tailings behind that. They will try to attempt to put the coarser granular materials out in this so called structural zone and then the fine grain materials out here. And if we can get the coarser materials here and if we can keep the water out of them then they're strong enough that they can help make up this barrier. As we raise the dam we might take material out of the structural zone and pile it up here to make these smaller outer dikes that are used to perform to provide the containment for raises. The problem with this methodology is there's not good control over what is going on in the structural zone. This dash red line we don't know exactly where it is and in fact hydraulically placed it's not a line like that at all but it's a zigzag pattern of fine grain materials in or woven with some of the coarser ones. So we wind up not really having much control over what materials get placed in the structural zone and yet what goes in that zone is critical to the stability and safety of this overall dam barrier. There are other ways of doing this. I just described the upstream method. There's something called the downstream method in which you start with a starter dike. Then you take competent as you raise it you take competent materials and you raise that which means you also add a barrier or you increase your barrier or your burn downstream. Your dam gets to be bigger and bigger and looks a lot like what we would use as a conventional earthen dam to store water. Clearly this takes a lot more earth work and therefore costs more to achieve than what one would do in the upstream method. And then there's something that's somewhat halfway in between in which we the center line method in which we build a downstream half of what would look like an engineered dam and then the upstream portion is more like uncompacted tailings. For those of you who are in I used to teach dam design years ago and one of the fundamental things I taught at the very beginning is you have to have positive seepage control in every dam. These simplified sections don't show any seepage control and I realized the diagram was put together by my colleague Steve Vicks. Many years ago to just demonstrate just to illustrate the different methods of construction but this diagram unfortunately has been reproduced over and over again and has led I think to some erroneous beliefs that we can build these things without positive seepage control. What do I mean there a simple thing like a drainage blanket that is constructed at the beginning that then pulls the water down into that drainage blanket so we can keep the downstream portion of that dam unsaturated. Without positive pour water pressures and working as a structural barrier to help hold the stored materials upstream. This is one critical shortcoming in many of our facilities we do not have these positive seepage control measures. Second part of it is that the waste materials that are used to construct the barrier portion and that we're also trying to contain their complicated materials and some of the things that make them complicated are summarized here. They're placed in a loose state without compaction and without very few controls on their placement. Many of them are hydraulically sluiced in the place and left there they have a high water content. A large fraction is fully saturated which makes them incompressible that means when if they're saturated and incompressible if we have anything like a shock or a sure stress applied to them they will typically try to decrease in volume. And that causes them to turn to a fluid like material and I'll come back to that in a minute. They may be highly variable in composition with alternating layers of different gradations and plasticity you may have thin layers of clay like material and over that a course like material. But from a stability standpoint that thin clay layer may be dominant in and whether the section is going to remain stable or not. Many may be chemically altered or weathered or aged which can alter their strength characteristics considerably. Some of these have some true cohesion to them but then that may be may disappear under certain conditions. Once something triggers shear strain in many of these materials they switch from the so-called drained mode which can be safe to an undrained shear behavior in which they lose strength. And that's where they're turning to that liquid like material. In this situation almost no strain occurs before they switch to undrained behavior lose strength and liquefy. These are described as brittle like materials. My colleague Andrew Forre down at University of Western Australia has done a lot of work on this and talks about when you have a brittle material and you don't have redundancy in the design you're inherently dealing with an unsafe situation. So I've talked a little bit about this stability being the important thing. What are the key geotechnical factors that do affect stability of tailings stands? Geometry is key. How steep is that outer slope? The subsurface profile that is what's in the foundation? What is the layering of the tailings? Do we have weak layers that are going to dominate? The stability is always dominated by the weakest material. Have we found all of those? Poor pressures within the dam and the foundation and the stored tailings. This is one of the more complicated parts of dam stability. What are these water pressures because they directly affect the strength of the materials? Getting those poor pressures is complicated. We have computer programs that can calculate them, but how good are they because the materials themselves are very complex, heterogeneous. They're anisotropic, non-homogeneous. We get poor pressures if we're still building and filling things. We get changes. We get changes from weather. So poor pressures really to know what they are, you have to have a reliable measuring system in the field to measure what's actually developing in the field. So you have a better understanding of what your true stability is. And then the strength of the materials. And there's key words that are played here, whether the materials are drained or undrained, contracted or dilated, cohesive or non-cohesive. I'm going to take this up on the next slide a little bit more. Drained means that the soil, the materials are loaded slow enough so that as they try to change in volume, any water that needs to flow in or out to adjust to that volume change can occur and no excess poor pressures can occur. This drained case is something we understand. We can predict drains or measure drain strength pretty well. Our students learn this and know how to do these calculations pretty straightforwardly. Most people think of sands as being drained because they have a high permeability, water can flow out easily. Contrast that to undrained materials where the rate of loading is fast so that water cannot flow in or out of the element as we're shearing it. And this causes excess poor pressures to develop. It also causes them to fail in a mode we call undrained, which the strength of which is quite different than the drain strength. In some cases it may be more, in some cases it may be less, but it's a very difficult concept for many people to understand. And many people who do undrained stability analyses fail at assigning a proper undrained shear strength for those cases. That undrained shear strength is affected a great deal by whether the soil is so-called contractive or dilated. Contractive means that when we take an element and we try to strain it in some way or other, it wants to decrease in volume. Which if the element is undrained, this causes positive excess poor pressures and a decrease in strength. So this key word in tailings is is the materials, are the materials contractive? Because they can become brittle. If they're contractive, they can exhibit this brittle behavior, lose strength and statically liquefy. On the other hand, the opposite of that is dilative, which where if I have an element, I strain it, it wants to increase in volume. It wants to spread out. And in this case, if it was undrained, that will create negative poor pressures, which increase the strength of the material. So if I could build my dam, my barrier of all dilated materials, which means mostly things that have been compacted so that they're at higher relative densities, then I don't have to worry because it's just going to get strong. If it's drained, I know what its strength is. If it's undrained, it's just going to get stronger on shear. So the real challenge for us is dealing with contractive materials that are going to strain in an undrained mode. We have to find a way to avoid that. And I would add a big strong caution here. Most textbook classifications of sands as always being a case of drain stability is terribly misleading. Because in many of these tailings dams, we analyze them as though they're drained because they're sand, they're draining our materials. But if they're saturated and if they develop this tendency to want to start to strain and develop positive poor pressures that can't drain off fast enough, you actually have a sand that's now failing in undrained mode. And that can be catastrophic. This slide shows some sources of geotechnical problems where failures have occurred. I don't have time to go through all of this. I just tried to pull out an example from some of the many cases I've looked at. There's no one simple triggering action in these things. We have to look at foundations. We have to look at the construction. We have to look at how they're operated. We have to look at environmental conditions. And so if you get a chance that the video will be posted and if you want you can come back and go through these a little more care. Factor of safety is a standard number that we're using. Many of you have worked in this, say we need to get a factor of safety of 1.5. What is the factor of safety? What does that mean? It's the chance that a massive soil is going to let go or waste is going to let go and slide away in an uncontrolled way. It's defined as the ratio of strength, sheer strength of the soil to the sheer stress that is created by gravity. So it only exists where we have sloping ground. So what's going to alter factor of safety? You can see if we decrease the numerator here, if we decrease the sheer strength factor of safety goes down or if we increase the sheer stress factor of safety goes down. So how would we decrease sheer strength? In the tailings dam, this mostly comes due to increases in pore pressure from water flowing through the dam. We raise the reservoir level, we raise the height of water in the reservoir. We get an adjustment in pore pressure throughout the cross section that generally is an increase in pore pressure. That's resulting in a slope that was stable up to now. We in fact could be decreasing factor of safety. An increase in sheer stress. This occurs if we're adding load that is adding tailings to the top. We're taking away from the tow area in effect trying to make the slope a little steeper. Increasing pore pressure also will increase your stress because the water flowing through the dam creates a seepage force that adds to the destabilizing forces. We could have external forces such as earthquakes, blasting or loads collapse from piping or desolution of contained materials. All can result in increases in sheer stress. So these are things. And if we have a contracted material that's subject to liquid static liquefaction, these would be so-called triggering could be triggering actions that could be just enough to set off an instability failure. Getting close here to the end, a couple more key concepts. Going back to drained and undrained strength, I just want to emphasize the importance here of getting this part right. Drain strength we refer to as the friction angle of typically say 30 to 40 degrees. And coefficient of friction would be the tangent of that. So the tangent, the coefficient of friction for drain strength would be about 0.6 to 0.8 as I show on the slide here. That exact same material, if it was shared in undrained mode would have an undrained strength ratio of about 0.2 to 0.3. So that says if I have a condition where I've designed with drain strength or I think I have drain strength, but something could switch the material in a way that it would now behave undrained, it would only have one-third to one-half the sheer strength that I used in that drained calculation. I hope you can see that this is a critical piece for the safety of a dam and for its design. Again, if the materials are contractive, then they have the potential to shear in undrained mode and their shear strength is going to be for that undrained condition much, much less than it would have been had I assumed that the failures would be drained. I think as a guide, a comprehensive stability assessment should consider both drained and undrained loading. Don't take shortcuts. You know, if the material is dilative, the drain condition will usually be the more critical. If the material is contractive, undrained condition will usually be more critical. But why in today's modern world computer programs and everything, why take the risk? Do like our structural engineering friends do, consider all load cases and prove that each is in an acceptable condition. Coming kind of a stepping back out from that GeoSpeak kind of slide or two and looking at what are the most important design requirements for all dams. And that is we start, do not lose containment of the contents of the dam. You pick out a book on design of water retention reservoirs. You know, this is the primary rule that we say and that means we don't have, we don't allow stability failures and no piping failures, no over topping, no uncontrolled erosion and no washouts around hard structures. This is just as important in tailings dams as it is to water retention structures. So I kind of put that in a simple slide. What does that really mean? Make sure our foundation has enough strength for all possible load cases. Remember two of those cases I showed you were three actually were failures through foundations. Have all the materials that comprise the barrier portion of the dam, the structural part of it be dilative materials. And then we don't have this risk of static liquefaction contributing to a stability failure. We control the internal water pressures with internal drains to keep the factor of safety for drains and unconstrained conditions greater than 1.5 for all possible load cases. That's a safe way to go about this work. We prevent over topping. We control internal external erosion. And then there are other requirements for seismically active regions, which we don't touch on in today's presentation. Many existing tailings impoundments do not meet these goals and that raises the question. What do we do about that as a society and as an industry? Just a slide, quick slide on monitoring. There's been a lot of thought that that we could improve this situation a great deal with real time monitoring. There's actually some countries have gone to mandated required real time monitoring. I think we have to be careful and I'll give this caution at the end. What can we monitor that works in tailings dams particularly? We can detect lateral movements in the foundation beneath the outer slope of the dam. So there was three cases I showed you that failed through the foundation. A good monitoring system could have detected those. We can detect changes in flow rates from internal seepage that might be precursors of piping failures or failures along hard structures. We could have sufficient locations with poor pressure measurements to establish the flow pattern through the dam at several sections so that when we do a calculation of factor of safety, we're doing it with confidence that our poor pressures, which are key to stability, are meaningful. But that means that you've got to have measured poor pressures along the critical failure surfaces that you're analyzing. And I see an awful lot of dam sections where the poor pressure measurements are made in the wrong place and they're not made with sufficient frequency to be meaningful. Instrumentation should be reliable and redundant. I prefer these days to see automated readings. The technology is there. It can be done without a lot of added cost and we do that several times a day. Dams can change. Surprisingly, they can behave and look like they're the same thing, but undergo significant changes quickly. Any instrumentation monitoring program should be complemented by visual inspections using trained people who know what to look for. And then have a team available to help evaluate that measured data and interpret it. What does it mean? Have an action plan that gets triggered when certain measurements exceed pre-selected action levels. So just a warning for tailings dams made of contracted materials. Monitoring deformations most likely will not provide warning because these materials are brittle. They can lose strength suddenly with little to no strain occurring before that event occurs. You can show this in the laboratory testing. You can see this from some of the failures that have occurred. So summing up, we know how to design and build tailings dams that are safe. The geotechnical profession knows how to do this. We're not challenged by reshort shortages there. We could use some better tools to help us more closely define, separate, contractive and dilated behavior. There are always some improvements we can make, but the fundamentals necessary to do safe design we know. The problem enters when the designers don't use what we know and the builders don't build what the designers design. And this is particularly a problem in the mining industries where there's a separation many times between those who design the facilities and those who actually construct them. I'm not harping on anybody. I'm just trying to make some observations of where are potential areas for improvement. The problem we have is compounded by many existing tailings dams that were not designed or built using what we know is required for the dam to be safe. Tailings dams can fail after they are retired from service. Rumidinho had been out of service for three years. Water was being taken off the top of the dam slowly over time. Its safety should have been going up. Its factor of safety should have been going up by simplistic assessments. So the idea that once we take it out of service it's no longer a threat is up for serious argument now. The average failure rate of tailings dams is too high in my opinion. I showed you about one in a thousand compared to water retention dams. And half the dams are even higher than that rate. I think that raises a question for us as to is this acceptable and if not what do we do to cost effectively bring that rate down. Risks from favor of tailings dams are higher than that from many other industries as I showed in that one plot of different industry risks. And so that's something we need to be aware of. It seems logical that steps are needed to reduce these risks by at least an order of magnitude below the present state. So the challenge I serve on Cognate we're looking at ways we can improve things for society. And so one big picture view maybe can we can we help identify ways that this can be done efficiently and cost effectively. And that is by reducing probability of failure and reducing potential consequences of failures. So I thank you very much. It's really strange to me to be sitting here speaking to 576 people and I don't see any faces. But I hope I was able to keep you all the way. OK. Thank you. Thank you for a great presentation. And you have created a problem for me because now I have a ton of questions. That we could try to answer. So remember if you have not already sent a question you can still use the question and answer a box which you can locate at the bottom of the screen. Please type and send the questions and we will try to answer them as they arrive. So we have a can you listen me Alan here. Yes I can Pedro. I'm clear. Can you see also the question and answer just in case the box I will I will be asking the questions but maybe you can also look at them. No they we purposely kill that so I don't get confused. So you don't get perfect. That's your job. So I will be even mentioning the name of the person's just in case. And one is by Edgar Sanjan is one of the first one is a simple one. He asked are you considering slimes as a granular or particular material in particular because of the case at the beginning that you showed. Yeah that's a good question Ed and I would say no most of my comments are more dealing with granular materials. Slimes typically have plasticity to them. And I think most people involved in most slimes are actually behind an engineered embankment. So so sure strength of them is less of a concern but I'm not considering slimes in my presentation today. OK so there is another one here by Stephen Emberman and he said you seem and actually I kind of know this one but maybe you can answer. You seem to be talking about liquefaction as a phenomenon that occurs after failure of the dam rather than as the cost of the failure. Could you please say a little bit more about that. Is liquefaction induced because of the failure or liquefaction was the reason for this. So this is a really good interesting question and I think probably one open for further debate and study but my I spent a fair amount of time looking at this. And going back in some of these cases I tried to review this morning plus others. And I think if you if you look at this really carefully it appears that that we get a stability failure first which then releases containment of the tailings and then then they they're subjected to shear strains. They collect they liquefy and then flow out. I know if you some papers out there they taught they describe this whole process is just failure by liquefaction. But I think there's there's a first step that where where there is a stability failure that occurs and then let's let's the tailings forces the tailings to start to shear and liquefy. The exception of that would be if we have seismic shaking where potentially there you shake you liquefy the tailings and if the barrier hasn't been adequately designed that could increase the load on the barrier and force a stability failure. But that's the only case that I can identify that where liquefaction occurs first and then the stability failure follows. Yeah there is a question here by Haimet Castro that is a little bit more technical but he's asking it seems that there has been some studies on the effect on shear strength by principle stress rotation and it seems that there is some work that has been done related to the stability of dams. Do you have any comments on the effect of principle stress rotations on the reduction of strength. No I'm not I'm not an expert in that. I worry a lot more about the more fundamental principle causes of brain versus unbrained strength. You know and then I guess the other thing I worry about is you're making sure the tests that we are doing are representative of the conditions in the field. So maybe indirectly I take care of this problem in that a lot of the shear strength testing that I do and want to see done for these kinds of problems is more like direct simple shear test conditions which more closely capture the stress pass in the field. Then what we might get in a triaxial compression test so I kind of avoid this stress rotation question he's he's bringing up I think. Yeah. So Yeah I don't know that I avoided I just I address by running a test that tries to try to mimic what happens in the field. Yeah I think the question was some some people have been asking also about possible stress path that are followed by the material. Before the failure occurs and there may be due to the load it may be a stress path that puts you in in a point where failure is more imminent. Just going closer to a P equal to zero which is the effect of for what the pressure that you are mentioned. Yes. Here there is a question by Jorge Macedo. What is approximately the number of existing upstream dams being operated in the US and what percentage represents them in terms of the total numbers of dams that exist. That's an approximate numbers. Yeah I don't know that I. I'm sure that's in the literature somewhere I just don't have it. Probably the moment. Let me see some other questions. So here it says by Lindsay Newland Boker and she says it's a long question in his mellow lecture this year Dr. August and estimated that at least 50% of all catastrophic failures in history were due to static look faction and state that this year was to the level of this was to the level of expertise and mastery industry in stability analysis. The case records of these dams support his his observation and witnesses misuse of parameters in the to be as you have a Bernadino case. So it seems that what she's implying is that there were a lot of signs in some of these tailing dams that even though that the failure was imminent and very fast, there was a lot of evidence that this was going to happen. And she's asking if this is common in others that even though that they occur very quickly, but there are evidences that people don't consider. So she has several questions related to this. Yeah, let me try to see if I can take that apart. Partly, you know, Dr. Morgenstern making a statement that 50% of catastrophic failures in history are due to static liquefaction. I don't recall it being stated that way. I'll go back and check again. You know, he did a great job in the two or three of these failures he looked at it from Dale, for example, and that Matt Pauley. And, you know, I think in both of those cases, static liquefaction occurred, but it did not cause the failure something else triggered it. And then static liquefaction was a consequence of that initial failure. So I would read very carefully what he said and I will. It's a great lecture. All of you should go read it. Dr. Morgenstern is one of the experts in this field and he writes very clearly. So I strongly recommend that it's available on the web. I think the question knows that there are, the second part of the question was, you know, there's lots of evidences out there that we miss. I think what I was trying to say is when we do visual inspections of dams, what are we looking for? We're looking for cracks, we're looking for slumps, we're looking for water that's exiting in an uncontrolled way. And so when I say people go do inspections, that's what they're looking for. And you usually don't see in these tailings dams failures changes in those visual kinds of things. So what is she referring to? I think it's if you're a very knowledgeable person in this field, you can go back and look at postmortems of failures. You say, oh, obviously this material was contractive and obviously this stability analysis was inactive. Obviously there were high pore pressures in this dam and that's what caused it to fail. But those are things you look at in backward. It's hard to walk the surface of the dam and come to the conclusion that the material is contractive or that the pore pressures are too high. So when I say, you know, visual observations, visual inspections that we do can't tell us a lot about these inward things that really drive the safety of the dam. Here I have a kind of a continuing question by Bruce Catter and he says, rightly, you emphasize the importance of the dam stability, but is there a reliable way to predict the run out distance of the tailings given that failures will happen? Do you agree that entrainment and mixing of water with tailings flow continually, softens the stuff to slurry or a suspension that allow the material to travel several kilometers? Hence to predict run out, we need to account for entrainment. Oh, absolutely. Professor Catter, nice to hear from you. I totally agree and I often wondered if we didn't have water stored on top of these things, would they still liquefy and flow? And I think they will but not anywhere as severely as what happens when we have a fair amount of water stored and then it all gets entrain. I even suspect that we get failures in which we have a lot of solid like material riding on a film of air and water that can go quite long distances. I've looked a little bit at the literature about trying to predict run out distances. You know, this is a really complicated problem and my hats off to those people trying to model this. As a design engineer, I guess I give up and say, I just want to make sure my dam never gets to that stage. So I focus on how can I prevent this tragedy from ever occurring. And I have another question here that goes back to the material properties by Murali Tehran. How do you determine whether a material is contractive or dilating in a highly variable material? What kind of lab or in-city testing do you recommend, CPT? There were several questions related to that. Yeah, no, that's a great question. Alan, before you, Mr. Smarty, before you answer that question, I just wanted to let everybody know that we have a number of questions and we'll stay live for a number of more minutes so we can squeeze in as many as possible. And then at some point, of course, we'll have to pull the plug. But thanks everyone for all of your questions. We'll get to as many as we can here. Yeah, I see that we got up to something like 750 at one point and it's dropped off to 570, so there's still a lot of interest in this. I'll go as long as we can. I said that we know how to do these things, but we have certain challenges. And one of the challenges is exactly this question. How do we determine whether the material is contractive or dilated with any degree of reliability? And so, you know, one technique that people are using and it's growing in its use is the CPT, comb penetration test. And there are some great work that's been done out there where it's pretty simple. You take the cone resistance and you plug it into a equation and you draw a nice chart that says, you know, this part of the material appears to be contractive, this part appears to be dilative. In fact, it's a nice simple line that's plotted with depth to tell you at what depths materials are dilative and contractive. And I see a lot of people using that and they seem to be using it with a high degree of confidence. I caution you in concept that can work, but it's not that simple of a break of a solid line drawn with depth between contractive and dilative. That really should have a zone there of uncertainty where you can't be really sure. And if you read the papers that talk about that methodology, there's always a sentence in there somewhere that says, you know, this is an approximate method and for site specific conditions, you really need to calibrate. And if it's important, you really need to calibrate this methodology to your own situation. That's kind of tough to do because how do you do your own CPT calibration to determine whether materials dilative or contractive? That's a big research project. So what do I do? You know, I typically want to see CPT work. I want to try my best to get samples. I know people say you can't sample granular materials, but we do a reasonably good successful job at that. I will try to do sureway velocity measurements in situ. I'm looking for two or three different sources of information. I'll get samples so that I can classify the materials. I'm looking for multiple pointers to tell me how is this material likely to behave as a contractive or dilated material? And then the other part of her question or his question is kind of where you have high variability. You know, how do you deal with that variability? Do you kind of go for, since it's a stability problem, the lowest values? And this is a challenge. You as a designer really kind of have to figure out what do you think is the best characteristic representation of this field conditions? And it certainly isn't the average, which I see some people using in some published methodology suggest the average. I use the word characteristic, which kind of falls back on some of the work by the Corps of Engineers many years ago. Take the average minus one standard deviation or so to kind of say, well, maybe that's a safer representative strength. But don't try to be overly conservative because you're going to cause your client money. So there were a couple of questions here, but if you think that the regulatory agencies and methods are appropriate. If they should go more into probabilistic methods or should continue to use safety factors, what are your comments? I am trying to bring together several questions here. Several on the, what are the recommendations that should be followed? Are there regulatory agencies doing a good work? Should be pushing for probabilistic methods. What are your comments on that? Oh, yeah, this is kind of a general question across all different types of facilities, isn't it? We're trying to get away from kind of deterministic methodologies to more risk-based stuff. I do a lot of risk work. Of course, as a decision kind of tool, putting the problem in terms of probabilities and consequences to me is very satisfying. It's kind of hard to carry that out in the regulatory framework. Regulators have to keep things pretty simple and for a number of reasons. And I think if they just simply wrote some good rules or guidelines that were really clear about what we should be doing for deterministic methods, we would probably be better served right now. You look at some of the, around the world, you look at some of the written requirements for the stability of a dam. And it says, you know, thou shalt have a minimum factor of safety of 1.5. But then there are variations. So you can get into all kinds of situations where it's not really clear when you can deviate from that or whether it's possible to deviate from that. If I've got a temporary condition and I'm working to improve it, what's acceptable in that circumstance? So, and I think really more of the emphasis right now ought to be on making sure that we do the deterministic methods right, that we do the site characterization right, that we understand the concepts of fluid flow right without unnecessarily complicating it by adding in the probabilistic approaches. That's not to say they aren't useful, but I think from just a design standpoint, we don't want to overly complicate our work. And in that constant, do you believe that the federal guidelines for dam safety should be applied or can be applied to tailing dams? I think so. I don't know why not. The consequences, increasingly now the consequences, well, let me modify that. If the tailings dam is one in which a failure will get off site, then I think, yes, it's not going to impact people and I don't really see the difference in a tailings dam failing or a water retention dam failing. If it's a situation where the consequences can be contained within the site, that to me is the, you know, that's up to the owner and an operator of the facility. So with that, I think that we are already 10 minutes over the hour and I am getting some comments here from the bosses in Coggy that maybe we should start wrapping up this event. Do you have any additional comment that you want or something they want to say, Alan, before we finish here? Well, I think, you know, thank you for your attention and for this opportunity to talk about this subject. I'm hoping that we can somehow find a way that we get enough focused positive energy towards how do we improve safety of tailings dams. They don't need to have a black eye. They don't need to be viewed as a bad thing. They're necessary for our society to produce metals and other materials. Can we figure out a way that we can go positive on this and find cost effective ways to do a better job? Okay, so I think that we are reaching the end of this meeting. We have plenty of more questions that we can really ask to Alan. I want to thank Alan for his presentation and the great audience for the engagement in this event. I want to say that if you have questions about Coggy, including ideas for topics you would like to seek over in our ongoing webinar series, please, you can reach out some Maxino with, there is an email that you can follow in the screen. I will also note that the presentation and the audio recording from today's webinar will be posted within seven to 10 days in our website. Please watch your email for announcements about future webinars and events. And again, thank you very much, Alan. Great presentation and goodbye to all of you. Thank you very much.