 Good morning and good afternoon. My name is Pedro Arduino. I'm a professor at the University of Washington in Seattle and I am also a Coggy member. I will be this webinar moderator. Coggy is the committee on geological and geotechnical engineering. And is the one of the standing committees of the National Academies of Science, Engineering and Medicine Board of Earth Sciences and Resources. Coggy was established as the focal point within the National Academies for government industry and academia and technical and public policy issues related to earth processes and materials, soil and rock mechanics, responsible human development and mitigation of natural and human hazards. If you have questions about Coggy, please contact Samantha Maxina from the National Academies, she's one of the staff directors there. This webinar is part of a quarterly webinar series produced by Coggy through the support of the National Science Foundation. This webinar will be posted on our YouTube and announcement will be sent out when it is available. First, I would like to thank Samantha Maxina, Sara Hedridge and Mandy Enriquez for helping us to organize and produce this webinar. Dr. Martin McCann, who is the Coggy chair, is also who will help us today fill questions from participants following the presentation. The audience can submit their questions at any time using the Q&A tab on the Zoom panel on their screens. We will pose and respond as many questions as the time permits. First, a small disclaimer. Any opinions, conclusions or recommendations expressed by Mr. Franz or anyone during this webinar are those of the individuals and do not represent conclusions or recommendations of the National Academies of Science, Engineering or Medicine. So without more of this introduction, I would like to introduce you our speaker, John Franz. Mr. Franz, we are really happy to have you here today. He has more than 45 years of consulting engineering experience, most of his technical work for the past 37 years has focused on dams and water retention structures. He has been responsible for the analysis, design and construction of embankment and roller-compacted concrete dams and their apartment structures. He has served on numerous senior technical review boards and panels for dam safety projects for the US Department of the Interior, Bureau of Reclamation, the US Army Corps of Engineering, BC Hydro and Brookfield Renewal Energy. Mr. Franz was the team leader for the independent forensic teams tasked with the investigation of the 2017 automobile dam spillway incident and the 2020 failures of Edinburgh and Sanford dams. With that, we are really happy to have you, John, here and the floor is yours. Pedro, thank you very much. It's a pleasure to be here and we had a large number of subscriptions for this webinar. I'm not sure how many will actually show up for the live version, but that's very heartening to me because I feel the findings we've had at Edenville are important. And I'm really looking forward to sharing with you today and glad to see there's so much interest back early in my career. I did a lot of work in liquefaction with Gonzalo Castro and the late Steve Poulos, including some research funded by NSF. Here, late in my career, I'm finding liquefaction coming back into my life in a significant way again. So with that, I'm going to start going through what I want to present to you today and it's a lot of information so it'll go fairly quickly. We're going to talk about the chronology of these events, talk about the physical mechanisms that we believe explain failure of these two dams in Michigan. Talk a bit about what static liquefaction is, what we believe is our evidence that supports that as the primary cause of primary failure mechanism at Edenville. Briefly tell you where our independent forensic team work goes from here and then close with some, some of my thoughts on what I think are the ramifications of this finding for our profession. I'm going to begin by most of you have probably seen this, but I want to want to go through and try to run this video for you and see if this is the video of the Edenville dam failure and this is real time starting 535pm on May 19, 2020. So we have about a 40 second video, and really is pretty astounding and quite frankly without this video I don't know that we would have been reaching the conclusion we reached it was a very significant piece of evidence in in our work. And I will talk a little bit more about how that video came to be as we go through the chronology of events but just wanted to start with that if you had not previously seen it. The Sam and the Sanford dam or two of four dams owned by organization of a voice hydro at the time. The four dams owned by boys hydro or secord, Smallwood, Edenville and Sanford from upstream to downstream on the Tidabawassee River in Michigan. And Edenville actually spans both the Tidabawassee and one of its tributaries the tobacco river fans across both of those but the four dams are in sequence on that river. They're all power dams built in the 1920s and operated since then. And this is not operating as a hydropower dam at the time of the failure because its license had been revoked back in 2018. There are three other dams in the basin that contributed flows into the the boy stands but our focus is principally on the boy stand. This was a rainfall event on principally May 18 and 19 2020. And this is the rainbow pattern apologized that the image is a little blurry when it, when it is enlarged, but our dams are located here. In this general area of Michigan, and you can see that a bit to the northeast. There was over seven inches of rain but in this basin it really was typically four to six in the basin affected by the dams. In particular, at the four dams themselves, if we tabulate the rainfall that happened there was a little bit on the 17th principally most of the rainfall was the 18th, and then tailed off some into the 19th and the maximum rainfall there is, you know, approaching six inches in the at the core dam bit less more than three to four inch range in the other three dams. So it was a significant rainfall but not really an extreme what we call an extreme rainfall for this particular part of the country. The operations that were going on the morning of the 18th. The water was beginning to rise in the lakes, all of these dams have gated spillways. So the owners started to operate those gates by the afternoon of May 20, all of those gates were open, and the lakes at that time were about their normal water levels. So not very far off with a normal operating water level with all gates open at that point. The rain continued, and the water began to rise and all of these lakes by around midnight, even though which is one that we're focusing mostly on the water was over two feet high above normal pool. But one in the morning it actually reached its previous record pool at about two and a half feet, continued to rise and actually ultimately reached, we estimate about five and a half feet above above the normal pool three feet higher than it had ever experienced before. The Edenville dam itself on just to lay out the dam it's about 6000 feet long consists of four different embankment sections tobacco right tobacco left. The Edenville right and even bill left the failure actually happened over here in this Edenville left embankment section in there was a combination spillway powerhouse structure over here three gates and then the power. Another spillway over here on the tobacco side three gates maximum the height of this dam is somewhere around 50 feet so it's not a terribly large dam over here at the section that failed the embankment height was actually about 30 feet. I want to now go through sort of the sequence of events that happened early in the morning on the 19th. As the water had been coming up the operators of the dam notice that they were getting a lot of erosion on the upstream slope. So they deployed contractor out there to try to help put silt fences and other material and to try to minimize that that erosion on the upstream slope. They did have some increased flow in the tow drains. Mostly back both side and parts of the Edenville side of the embankment over here in this Edenville left. They didn't see anything going on over there for the most part, other than the erosion there was nothing significant noted until about a half an hour before the failure. Some residents on the upstream left bank would that had been watching was going on noticed that a depressed area developed in the crest, relatively suddenly at about five o'clock in the evening on the 19. And when that happened, they walked around the left end of the dam and walked down to a substation consumers energy substation that looked over the dam. And we're just watching with what was going on and at about 531 so this is four minutes before the start of that video. This is what the dam looked like. Other than the subtle part of the crest right here, there's really not much going on there's a little discoloration down here. I can't say for certain that that's not possibly some seepage coming out but we don't think so. If you look at historic photos there's variations in colors of this grass and we're not not convinced if it was water coming out is coming out at a very slow rate it's not flowing, and it's not muddy, it's not anything going on to that degree. So I want to now with that background at 531 that's what it looked like. I'm going to go through several enlarged still frames from the video. So at zero seconds, we see this water coming down the face of the dam. Now we think that happened we know what happened after 531 because the photo at 531 doesn't show it. Now what the gentleman who took this photo this video said is that this water coming down the face is the reason he picked up his camera and started taking the video. So we think it happened really seconds before that video started is when that water started to come down the face. Four seconds later, we see we start to get a little kick here at the downstream toe. Again, six seconds so another two seconds later really starts to bulge at the toe. We do have some some water coming out here at this point, the crest is starting to sag seven seconds it's moving even more. Eight seconds 10 seconds so between like six seconds and 10 seconds is when most of the action is taking place, we see water spouting out of this mass as it's moving. And then at about 15 seconds that massive soil is laying down here at the toe of the dam, but nothing else is going on here at that point there's no water coming out and till about 28 seconds. When we start to see water start to come out so we think this mass fail left the remnant in place that stood in place for about 10 to 13 seconds, and then started to give way and the reservoir started to be carried through through the breach. At about 38 seconds, or 36 seconds rather we have a full breach through the dam, and that breach then just enlarges over the next couple of hours, and eventually releases the entire reservoir downstream. So, Sanford Dam, just before going back to Edenville and talking about its causes I'm going to talk briefly about Sanford because it's a mechanical or mechanism of failure is pretty clear. The Sanford Dam like Edenville was constructed in the 1920s insist of embankment dams, six gated spillway bays and a powerhouse, and then more recently they built the fuse plug spillway to increase the hydrologic capacity of Sanford. This is Sanford after the failure and people were there watching the failure of Sanford and it's really just clear that the amount of water coming through in the flood and then increased by the failure flood from Edenville was just more than that gated spillway and fuse plug spillway could handle that about 745 in the evening, the water went over the top of Sanford Dam and it washed out by over topping. So the mechanism is pretty clear it's a traditional embankment over topping failure just spillways could not handle the amount of incoming water and the structure was over top. Now for Edenville, as we went into this investigation, one of the things we wanted to try to do was to understand the embankments, and particularly the embankment at the location that failed. Again, as probably not surprising as often in the case in these investigations to conflicting information. The original specifications said that this dam was going to be constructed with upstream downstream slopes of 2.5 to one and two to one respectively. The slope at the failure section, according to topographic data we have available was actually probably a little steeper than two to one in places maybe one seven to one and an overall of one nine to one or something like that from crest to toe. The fill according to the specifications was supposed to be placed in layers and compacted. The upstream section was generally to be constructed of lower permeability material, the downstream of higher permeability material, and there was to be a clay tile pipe drains embedded in gravel space. 10 to 15 10 to 20 feet apart along the entire length of the dam extending from centerline to the downstream toe. We're confident there was those were constructed we see them in construction photographs, and quite a number of them were actually inspected back in 2012 so we know they exist and were constructed. As far as that other information about the specifications were very dubious about the placement methods or the donation. The test borings show the embankments can contain both clean to silty sands, some with low blow counts, some as low as five blows per foot or less. And there also are clay sands and sandy clays in the embankment. Construction photographs really don't show any compaction equipment and actually show some evidence of sidecar dumping of Phil in the embankments in Edenville. Breach photos when we look at some of the breach photos from color differences there's some indication of that pot that upstream downstream zoning being in place but it's certainly not a slam dunk and the borings don't fully support it. The borings do, however, indicate that pretty much through underneath all of the embankments and certainly underneath the section that failed. There was a clean sand layer extending from upstream to downstream under the embankment over a low permeability hard pan. In the years berms and buttresses with filters were constructed at the taller sections of the embankment to address seepage and sloughing that had occurred historically at the failure section. The failure section actually remained the tallest section of the embankment did not have a burn or a buttress. What mechanisms did we think about for Edenville four of them that we thought about very seriously and considered very hard embankment overtopping internal erosion, or what's historically been called piping. What I call conventional static instability and by that I mean a conventional drain strength analysis static instability. And lastly static liquefaction flow instability and that static liquefaction flow instability is where we ultimately concluded the, the failure mechanism to be the primary failure mechanism to be but I'll go through the other three first, and give you an idea why we believe those are not viable. And then we'll talk about the flow instability and some detail. So embankment overtop for this and photographic evidence. It really is quite clear that the embankment was not overtopped until up to seconds or perhaps minutes before the failure when we saw that stream of water coming down the downstream slope. The photo down there at the bottom that's four minutes before the failure there's nothing going over the top. And in the eyewitness saying that the water flowing down there was the reason he picked up this camera leaves us to believe it really was seconds before and there wasn't time for that overtopping to significantly drive this failure. And the kinematics. Excuse me, the kinematics of what we see and I don't think are consistent with what we've historically seen. In overtopping failures where the water comes over the crest starts to erode the toe and head cuts its way back through the embankment which takes longer than that 40 seconds that we see in the video and doesn't kinematically look like that. So we're quite confident this was not an overtopping failure. Internal erosion. There really was no seepage of significance exiting the ground surface detected. Really, we don't think we see any water on the surface until just moments before the failure. There certainly was no turbid water discharge seen in that in that photograph at 531. We see opening pipe a single progressive sloughing the kind of things that you see with internal erosion failures, and the kinetics of that 10 seconds of failure duration really don't jive very well with with internal erosion either. In addition, the soils that had been sampled in the embankment over the years, and that we saw in the remnants on either side of this breach were, as I said before they were generally sands and quite quite uniform fine sands, or clay sands or clay sands and when you look at filter compatibility there's really no filter incompatibility between those materials there's no zone of open work gravel and there's and similarly in the foundation there's no open features for seepage to move material and and generate the internal erosion failure mechanism. We do realize there's possibility that there may have been some animal burrows up in the crest up in the embankment near the top. It's also possible there could have been abandoned railroad ties from a railroad system they use for bringing in the fill. Those may have contributed to that settlement we saw in the crest but we don't think they're the primary driver for what we saw as the failure modes. So for all those reasons combined. We don't think internal erosion is a very plausible explanation for what we see in that failure. The conventional static instability. We've done some stability analysis that indicate the fratix surface would have had to be extremely high to reduce the factor of safety below one with drain strength. And we would think if that were to happen we would have seen breakout of seepage much higher on that slope. We would have seen water coming down the face from seepage, we would have seen some evidence of that higher fratix surface. In addition, the kinetics of the failure don't really are not really consistent with a conventional static instability failure. I would expect that kind of failure to be more progressive and slower. So what I would envision happen would be the poor pressures would rise as the reservoir was coming up, the factor of safety with the drain strength might drop below one and you might get a bit of slope movement. But that bit of slope movement would actually be enough to temporarily restore stability with some limited deformation. Then there'd be further rise in the water pressures of the fratix surface. And this looks to the stability of factor of safety below one again, and a bit more movement. And that would progress over time, until there was enough deformation to cause water to to overtop the dam and then we'd have an over topping, or possibly an internal erosion damage to the embankment, but that process that I just can't envision that all happening in 40 in 10 seconds, and it doesn't look like what we see in that in that video. So for those reasons we don't think this was a conventional static instability. So static liquefaction flow instability what is it. The situation where the mobilize shear strength in a saturated loose sand or silky sand decreases rapidly to value significantly less than the applied static shear stresses and I'll talk about how that happens. But what happens is that very quickly the the available mobilize strength is less than the applied static forces that creates a creates a force imbalance and by the old force equals mass times acceleration equation force imbalance creates acceleration and creates velocities and it creates the kind of flow slide behavior that we saw. And that can only happen with a soil that has a brittle stress or brittle strain weakening stress strain behavior. Sometimes it's been called strain softening but curry hug at the NGI pointed out to me that he thought strain weakening in this context is a better term and I agreed with them and I started to adopt that as the as the term for this stress strain behavior. And it's a behavior here and it's a behavior that is typical of undrained shear behavior of loose saturated sand. And this is an example where we have a anisotropically consolidated sample. So we have some initial shear stress. If we load it undrained, it goes up to some peak strength, and then, and that occurs often at very low strain. And then it drops quickly down to a much lower steady state for residual strength, as opposed to a ductal behavior which would be the case in a drained test of this kind of material, where it would go up to a strength and stay close to or at for continued strain in a stress path plot of shear stress versus mean effective stress. What we find if we look carefully at that stress path is that that peak undrained strength over here actually occurs at a level that is less than the friction angle line. And then as those poor pressures generate, and the sample continues to strain down to the residual or steady steady state, it eventually intersects that drained friction angle envelope out here at the steady state. So that's the kind of behavior we're talking about. Interesting. Greg Becker pointed out to me that in an historical note, the original term of use of the term liquefaction that we found in the literature goes back to the 1920s by Hazen. When he referred to the failure of Calaveras Dam, which was a static instability failure. So quite interesting that liquefactions first use was for static failure. There are some historic examples out there. Fort Peck in 1938 during construction. What uses dam in 1907 that was studied by Scott Olson. And then that failed during first filling in the upstream slope failed during first filling. But then liquefaction was later used in reference to earthquake induced failures, particularly beginning with the lower San Fernando Dam failure in the 1970s. And let's say, at this point in time, most geotechnical engineers associate liquefaction only with earthquakes, certainly up until recently that's been the case, and more and more we're starting to talk about static liquefaction in large part because some of our tailings dam colleagues have been focusing on that after a number of failures. Most recently, the Brumadino failure in 2019 in Brazil, the tailings dam in Brazil, and it actually suggested you haven't looked at it, look up the video on the Brumadino and look at it and compare it to to Edenville. And it has a number of similar characteristics in that you see at the beginning of it, you see a little kick at the toe, and then it just progressively begins to move and accelerate faster. The difference being that in in Brumadino, it's the entire width of the tailings dam that fails at once. Here at Edenville we have a limited somewhere between 40 and 80 foot wide section of the embankment that fails while the rest doesn't. And I'll come back to, to why I think one possibility of why it was just that that particular section. I think if you think about it. And what we know about earthquake induced liquefaction now, one could really suggest that all liquefaction flow failures are static liquefaction flow failures, and I'll illustrate that in the next slide I think we have come to the point now that if we if we look at the static liquefaction or an earthquake induced liquefaction flow failure. What really is happening is we have a sample or I have a soil in place that has driving shear stresses from gravity that are less than the drain strength. This sand material, the earthquake comes along, shakes it. And actually, if you could imagine, if you did a that a monotonic loading of this, its stress strain curve would look something like this. And what the earthquake really does is drive it across under that peak generating some core pressures, and then it drops down to its steady state strength. What drives the deformation isn't the earthquake the earthquakes the trigger what drives the deformation is the strength drops suddenly to below the driving forces. That creates acceleration, which creates velocity and causes the flow slide to develop. So in a way it's a static flow slide failure triggered by the earthquake. The interesting test that's out there in the literature interesting references out there in the literature came out in university Alberta with a graduate student the university Alberta working with with Peter Robertson and with Nordy Morgan's turn. And they did a quite interesting test. This is a test where they anisotropically consolidated a sample up to a certain point. They then left the drainage valve open in the triaxial sample, and they just gradually increased the back pressure so they increase the poor water pressure in the sample, while maintaining a constant shear stress. The stress strain space what happened is as that for water pressure increased in this case for about four tenths of a percent axial strain. The sample continued along at that constant stress, and suddenly at a certain point, it collapsed on its own it collapsed and behaved undrained generated high poor water pressures and and failed and actually failed so quickly that they couldn't measure the the proper measure they have the responses, and it was done load controlled and it's reported in their paper that the load frame came down so quickly and hit on the restraints that they could feel it in the in the lab. In the stress path plot, we went up to this point and then here's where they started that increased poor water pressure. And out here at an obliquity corresponding to a mobilized friction angle of about 20 degrees. The collapse happened, even though the soil had a drain steady state friction angle of about 30.6 degrees. They did that particular test they did a second one. Sorry, I'm not sure what happened there. There we go it's better. They did a second one at a different shear stress and got very very similar results. They also compared this to what the stress path is for the monotonic loading, which is shown here in these open circles, and you'd see what happens is the sample actually moves across until it hits that monotonic loading loading curve, and then it collapsed. There's actually a state long or a state plane that when hit triggers this collapse, and it triggers the collapse at less than the drain friction angle. So their conclusions from this is that you could have slopes that fail suddenly undrained if the soil is very loose in the poor pressures rise slowly in a drained manner. So drained increase in poor pressure triggers this failure, and that collapse boundary is is a friction angle significantly lower than the drain strain. The strain during that drain loading of the poor pressure building up is only point four point six percent in the test that they did. And with that behavior, there really would be no clearly observable earth movements prior to sudden collapse. So the conclusion there is even though you have a slope that may have a conventional factor of safety above one based on drain strength the slope could actually be close to an undetectable catastrophic under drain failure. And from that our colleagues in in tailings damn work have sort of hypothesized that if we have a sample with relatively high shear stresses of a loose sand or silky sand with this brittle strain weakening behavior, you could trigger these undrained failures by increases in poor pressure increases in shear stresses, potentially sudden increases in shear stresses, or some combination of the two to follow those kind of stress paths. Now for even though, as is the case for most damn failures, unfortunately, we don't have exact material failed because it's been washed away. We don't know its exact characteristics, but for a demonstration we did is collect some samples from one of the embankment remnants, a clean uniform silty sand sample, prepare it to 30% relative relative density, and tested at three solidation stresses in the laboratory. And you see that what we got out of that was this very abrupt drop in strength, and it that drop in strength began here at an instability line of somewhere around 17 to 18 degrees of obliquity compared to 31 degrees of drain friction angle. So we demonstrated that there are materials in that embankment that, if loose enough, could generate the kind of behavior that's consistent with static flow slide liquefaction. In addition, an ASCE team did a study and they also reached a conclusion that the failure was likely static liquefaction. One of the very interesting things they did is they did some pixel tracing of that video. They picked five different points and traced them over time. This P5 down at the foundation they found did not move so the movement really was isolated up in the embankment. And the other thing that they did is they calculated velocities, and you can see the velocities here approached five meters per second or 16.4 feet per second, and were predominantly horizontal, again all consistent with a flow slide type behavior. They also did a simplified kinetic analysis of Edenville following a procedure that Davis had all developed, and we're still working on this and we're finding a little bit but the initial go at it indicated that we could create velocities as high as 20 feet per second, which is in this concept of what the ASCE folks came up with, and the whole failure would happen over a span of about six or seven seconds, which is not inconsistent with the time we see of the failure. So putting all of that together, we judged static liquefaction flow instability to be the primary mechanism for failure. I believe that's supported by the accelerations and velocities that we see in that video. The strong evidence of loose uniform sand in the form of the very low blow counts from previous investigations. The strength loss behavior in the laboratory on loose samples taken from that breach remnant and the reasonably close match in the kinetic analysis. In my view, it's very difficult to explain the kinetic failure or kinetics of this failure with any mechanism other than static liquefaction flow. A couple of contributing factors. The lake was at a record high level. So that could have indeed elevated the water pressures in the sand to levels high enough to trigger this where that had not happened before. And there's a apparent lack of functioning foundation drains at the location of the failure, which I think helps explain why it happened where it did. In 2012 they did an inspection of these drains actually extending a hose up into them. And in that inspection we note that in this area where the failure likely happened there appear to be drains that either were never installed or they are no longer functional because they couldn't find them and inspect them. So the lack of drains in that area certainly could cause or could contribute to water water pressures in that area than in the area where the drains are present. The triggering of the static liquefaction is a little harder to get our hands around. You know, we think there are plausible explanations, but we probably won't be able to reach a definitive conclusion. Increased pore pressures certainly are a potential trigger. And one possibility is through that foundation layer that I mentioned that is the clean sand foundation layer under the embankment as the reservoir rose, it could have transmitted pore water pressures into the saturated downstream materials and also begun to raise the surface. There could have been flow through the embankment itself as that water rose since we don't know whether the embankment had any zoning or not. It's possible that in the upper part of the embankment that had never been wetted before, there was a permeable layer going through the embankment that allowed water to come through and then rain down through the downstream section and raise the free attic surface. It's also possible that settlement happened on the crest. It opened up cracks that could have allowed more water through to allow those water pressures to rise. I think they're all viable. There's some degree of increased shear stress from the reservoir level, the higher reservoir level, but that's pretty small. However, the problem is with this section of the embankment having these very steep steeper than two to one downstream slopes, the static shear stresses are already quite high. And we could be very, very close to that instability collapse line and not need much to trigger it over that line and it's very sad that it's on off when you reach that line, the sample of the soil on its own, generates these and generates the collapse. It's also possible there could have been some shock load from an initial conventional static instability I think all of those are plausible, and it may be some combination of them. To wrap up here, the future work of the independent team we're continuing to do some further analysis of the stresses and stability analysis to refine our geotechnical analysis. We're evaluating the flood a bit further one of the questions where my answer is why did the lake go to this record level in that particular storm that wasn't that extreme a precipitation. We also are looking at the various decisions actions in actions over the years on this project and how did they contribute were there were there opportunities to intervene with things that might have prevented this, even if you didn't the potential for static liquefaction. For example, if you thought conventional stability was not adequate, you might have built a buttress or a berm, and that in itself might have stopped this failure from happening. We are working towards preparing a final report, including lessons to be learned with a target date for early next year 2022. Lastly, my thoughts on ramifications of this finding. Static liquefaction has not been considered traditionally for water dams according to the existing guidance, geotechnical texts and dam safety guidance. If you look at it it says that you really don't look at high pore water pressures or undrained shear behavior in sands. That's for rapid loadings like earthquakes, instead use drain strength grabbers. As I said our tailings dam colleagues have begun to recognize that may not be true. But to my knowledge that there really hasn't yet been a guidance that's co less direct how do we specifically deal with this issue of having these loose materials in place that could potentially trigger static flow slide when we don't understand the triggering I think this failure demonstrates that static flow liquefaction is possible if strain weakening soils are present. Saturated loose sands and silty sands. It also begs a question of with the video being a primary piece of evidence. Are there other failures in the past that may actually have been static liquefaction that we haven't recognized as such and that perhaps it's more common than we think it is because in general the cases reported are pretty rare. I think the profession is now challenged to develop appropriate protocols and guidance to address static liquefaction. And the IFT's work kind of ends when we put our report out and explain what we think happened and what we think the lessons to be learned are. And now it's up to our profession to see if we can develop some some guidance about how we how we best deal with it. And with that, Pedro I'd be happy to begin to address any questions if we have any. John, thank you. Thank you very much. Great great presentation and with an incredible case with very good information. So we have been getting several several questions here from the audience, and we are trying to organize them a little bit to see what how to answer them. And one that is very, very, very basic. And I know that you mentioned this, but what's the difference between a static liquefaction and piping. So I got several, several questions with that is a very basic one. Can you can you give us. Sure. Sure, piping internal erosion. What's going on with internal erosion is that the water flowing through the embankment or the foundation through a variety of different mechanisms can begin to move the solid soil particles and convey those solid soil particles out of the dam out of the foundation, and it can either create a pipe or avoid that extends back through the foundation or through the embankment, as it progresses and larges and eventually allows the water to either flow through that and and fail the dam or possibly the crest of the dam to settle and and the reservoir water to overtop it, but it's the flowing water moving particles. So the static liquefaction is all related to the, the mobilize strengths and the applied stresses, and is an imbalance in forces in the, in the soil mass rather than from directly from the flowing water. Yeah, it's an imbalance of forces that's the main one of the main things that at the moment of the triggering of the event. And did you look at other sections. So when I saw the video so the video is very good. So there were some remnants of the cross sections that you could also consider or or look at there were over the years at a few sections of this embankment. There were geotechnical investigations done with test foreings with SPT samples and in some cases, even some tube samples and some various analysis done gradation fundamental soil properties out of our limits those sorts of things. There were even some direct shear test done and reconstituted samples, there was all that sort of information available so we looked at that and that's where the, the general understanding that the embankment samples are composed of this mixture of soils and different types. That's where that comes from. And we did go out, I didn't personally but two folks on our team went out and did look at the remnant sections on either side of the failure section and looked at those and map those, and they were markedly different. The failure actually extended over very close to the spillway structure. And on that side the embankment was mostly clay material all the way upstream downstream. On the other side it extended almost to the abutment, and it became a little hard to distinguish what was embankment what was abutment soils. But that section tended to show somewhat a finer grain material on the upstream side sandier materials on the downstream side. And that's actually where we took the sample that we tested in the laboratory. But we looked at those we looked at the borings unfortunately the boring data over time was mostly almost entirely from either the crest or downstream of the dam at the toe. So we didn't have borings in the intermediate slope so it was very hard to determine much about the cross section. The, the taller sections of the embankment that did not fail. In all cases those taller sections had for various reasons of observed seepage or sloughing or different reasons had had the downstream slope flattened, or a buttress constructed with internal drains put in place. At any standpoint their stability had been improved. In this section that failed was actually the tallest section of the embankment, still existing that did not have a buttress or a flattening constructed. So what do you think that the Ed and Bill that was under the stress at multiple locations at the same time, or it was just this one unlucky section. This one was probably the most critical it probably had the, the highest static shear stresses before, in place before the reservoir started to rise. Just because it had the steep slope, and was the highest section that didn't have the buttress. And I suspect that other sections of the embankment of that East embankment where this section was located there were other sections that probably were getting very close to failure, but the lack of drains may have been the thing that the kick this one over. Yeah, I have a couple of questions here more related with internal erosion. One of one says, an inspection report of Ed and Bill dam just a few months before failure indicated several and even areas at the dam crest that could be a sign of internal erosion. What is your opinion in this regard. You know it's possible. However, I think if those things may be settled in uneven crest, the number of different explanations you can have over time, traffic and trafficking on the crest by people by vehicles whatever could lead to some of that. And we get some unevenness like that from some settlements related to animal burrows that collapse from the from the old railroad ties that might have been left in place in the higher sections of the dam in addition to railroad ties. So we had a trussle that they were using to, to bring some of those railroads in and they may have left some of those trussles in place. So you have potential rotting wood within the embankment. So those kind of things could produce the crest unevenness also the, the thing that we felt is after we looked at all the gradation data we had on foundation and dam that it's very hard to conceive of internal erosion, leading all the way into a failure mode without the water breaking the surface on either the downstream face or the toe, and beginning to carry particles out at that location. So that would really be the indication and from the reports that we've seen that wasn't happening and they did have, and this is possible. This certainly was going on to those foundation drains serrated by gravel were not filter compatible with the sands, and they did at times have some reports of sediment in those drains. So they did probably have some ongoing internal erosion through those drains but I think that would lead to a slow settlement not to a rapid failure like we saw. So you show a couple of one slides with you mentioned the, the work by Daniel Pradel so we have Daniel in the audience here. So he's asking some questions too. He said that the AC team obtained the drain strength by testing perform transient analysis and found a low static factor of safety around 108 hands and did not exclude it in their publication as a possibility. So what is your point, what factor of safety did you obtain for a static loop function, and how did you exclude it in your mechanisms. So what was the criteria for excluding it. The criteria for it for excluding it is, and I think we were still working on some of our stability analysis also, but depending on the on the free attic level in the embankment which unfortunately we have no data on, because they did not have any information in place at this section of the embankment so we don't know. We can run some, you know, we're going to run some seepage models to try to get free attic surfaces but the problem is, we don't really know the cross section very well either. So, for a hypothetical cross section we can run free attic surfaces. If we were to do a free attic surface with a uniform sand cross section. I think we get static factors safety very similar to what Daniel mentioned. The biggest reason that we lean very strongly to static liquefaction rather than the convention what I called conventional instability is the kinetics is a vision that if you that calculated factor safety of 108 and if the free attic surface rises a little more you can certainly get less than one, but if the stress strain curve does not have that brittle collapse behavior, I can't get to the kinetics. The only way is the kinetics and the speed and and the match. When we do the kinetics model that was leading us to believe that static liquefaction is the is the most likely explanation rather than a conventional. That's interesting that is the is the kinetics of the problem that is guiding you in the method that in the mechanism that you think it was. I didn't present it but we also did and will present in the final report, we did do a drain triaxial test on that same sand at 30% relative density, and they are not strain weakening. They're not they're not brittle at all they're a very ductile behavior. So if from a drain stress frame behavior of that loose sand. You would expect it to do more like I described in the slide where it would go, maybe a point nine five it slumps a little bit it stops of frantic surface comes up it slumps a little bit more and goes and not this very high velocity movement that we saw in the video. So, so let me let me switch a little bit and then I will be back to the to the to the physics, but I have here a question by Eric helping, and he says in the rapid nature of the failure implies traditional inspection and monitoring would be ineffective in being able to intervene. So some inside analysis indicates that even with brittle failures, there is some small indicators, weeks or months ahead of the failure, if they can be measured, but any comments on these technologies that could be used. Yeah, you know, I, I have heard some of those technologies I've seen some of them and, and yes, I think they do potentially have some, some usefulness for us. To me that sort of part of the challenge to the profession is recognizing. First, I guess we have to decide whether the profession agrees with our conclusion that this is what happened and that this problem now exists. And it's a problem that now exists that we haven't been addressing. And then I think we need to start thinking about how we best address that. Exactly questions like that are things, things to be addressed because if you, if you look at this compared to the way we deal with earthquake liquefaction it's an interesting comparison because our procedures for doing earthquake liquefaction now are to first look at that, if we have low blow count sands the first thing we do is a triggering analysis is do we think the earthquake loading is enough that it will trigger the liquefaction. And the next thing we do then is, okay, if the liquefaction is triggered, does that would that lead to instability. And if it does, then we usually do a mediation for it. This one's a little trickier now because what it really says is if you have these loose sands in place, we're not right now fully understanding the triggers, how comfortable can you be if those loose sands are in place, and they could sit there right on the edge of that collapse line, and then suddenly trigger. So, I, Eric suggestion I think is something we need to think about, but the question will come how confident, can we be that that kind of monitoring would give us sufficient warning so that we would know what's happening. So, follow up, you know, Devon and my blade here from G engineers is asking, was the owner or regulators aware that this section of the embankment was maybe critical for the reasons you discussed before the failure or this was just an apparent hindsight in this respect. Another thing we're going to be addressing a bit more in the final report is when we look at the, the history of the geotechnical evaluations and analyses that were done on the project. It seems that they were, they tended to be reactive. They tended to be a situation where some seepage or wet areas were observed on the slope, maybe a little bit of sloughing on the downstream slope. So, we came in did an investigation of that area. Did some stability analysis on it concluded well we should put a buttress or on it or we should flatten the slope. We should do something about it. But there was not over the history, a systematic let's look at all the sections. And after these buttresses had been put on the other taller sections to the best that we can tell at this point, no one asked the question well. What's the stability over here at this remaining tallest section that doesn't have a buttress. Yeah, so the regulators were not aware, but in that that's one of the questions is, should we have been aware. And, you know, we are we are getting close to the to the time here but I have a personal question here. And it seems to me that we need to understand better static look of action here. And that's one of the outcomes. Do you think that with the type of lab test that we do is possible to get this steady state points and how much the formation you need to reach this steady state points. With these sands like we're talking about here. And I actually did a lot of this with with Gonzalo Castro and Steve pull us back. Long time ago when I was a young buck. And we did a lot of those with the with the cleaner sands and even silty sands. And certainly, it's very typical that that peak stress happens at 1% 2% or less, and by 10% strain you drop down to that residual steady state strain. As it gets silty or the stress being starts to stretch out a little bit, but we still were pretty successful with silty sands, getting down to getting steady state strength within the limits of a conventional track seal test. As it gets very silty, or in particular that starts getting plastic, it gets harder and harder. And the sands we would be worried about in this kind of phenomenon, you can actually get them. So you cannot get or you need to devise something new to try to get to that, I don't know because this could be happening at, I don't know 50% the formation 100% the formation. And so it's very, it's difficult to have a piece of equipment to handle this with those for sure. Yeah. So, and so we are we are exactly at the time and you know I don't know, we have, we have been receiving questions and questions and questions. And, but, and we can continue if, if possible, and for the people that they don't have time this this presentation is being recorded and it's going to be distributed in the YouTube channel. A huge, a YouTube channel, but I have one final question that maybe you can address for everybody and maybe then we can continue more. So, and it says the findings of the automobile forensics teams indicated there are systemic issues with respect to the way in which them safety is managed by owners consultants and regulators in the US. The Adamville failure brings to light a mode mechanism of failure that has not been focused for water retention structures, albeit the static leg function is not new. What are the implications of the finding of these two events to them safety. Okay, I, one of the things I think we're seeing in in Edenville that we also saw an orville and we mentioned there as a lesson to be learned is the need to to not so much depend on physical visual inspections of our dams as a primary method of managing dam safety but to at some point, and then perhaps periodically after that to periodically go back into the history of the structure and do a more comprehensive review of what do we know about how it was built. And how it was performed and whether there are issues that perhaps need to be addressed. And the questions I said, I suggested that that raises is one question is, how does it, does it compare with modern dam safety practice is this was the structure according to sort of modern dam safety safety practice. If it wasn't, how does it deviate and others do those deviations potentially present a risk. And then that should that, if that's true, then you need to look at that further. So I think that lesson still remains and, and the FERC in its new guidelines is trying to address that with modifying its for part 12 procedures to have some of these comprehensive reviews. So that's an important aspect of that I think that that's ongoing, but state dam safety programs and some are doing that already need to start doing that to the aspect of it of this one being a mechanism that we haven't looked at. I think part of the reason of that has been that it is relatively rarely been reported. We've got what you said we got four pack, maybe we've got a few others and that we have some of the tailings dam instances. So that makes us think well it's not that big a risk because it doesn't happen that often. But maybe we need to rethink a little bit well even if it's not that big a risk do we still need to understand, and understand that it's possible, and make that risk decision of whether we can live with that risk or not. So, I think it does open up some of those kind of questions. So, and I think that for the benefit of the audience I think that we should, we should stop and we should stop here. And we have plenty of more questions that we are going to distribute to you, john, in case that maybe you want to look at them and, and maybe answer some of these, these questions, and we can maybe produce a little document with some answers to this question. And then later that to you. I see that we are losing some people in the audience here as we go. And so before we conclude I want to thank everybody everyone to attend the webinar will be posted as I mentioned in the in the YouTube channel. Thank you to to the link, and please also give us your feedback for future for future webinars, what are the topics that the audience would like to, and to see. And again, I have to present a disclaimer that any of the opinions and conclusions or recommendations expressed by anyone during this webinar. All are those of the individuals and do not represent conclusions or recommendations of the National Academy of Science, Engineering, or medicine. So with that, john, this was a, an incredible presentation. It opens a lot of questions about the triggering of static legal facts, if it's static legal faction. What was that race of water a little bit that was enough to generate for pressures to bring everything to the instability or what. So, and thank you very much for, for, for your, for your presentations. My pleasure, thank you.