 In the United States, there are more than 74,000 dams serving immensely valuable purposes. Flood control, water supply, hydropower, navigation, erosion control, recreation, and environmental management. Of these dams, about 2100 are federally owned and operated. The rest, about 72,000, belong to other governmental and private parties. The challenge for dam owners is to maintain an active awareness, both of the operating condition of these structures and of the risks they pose to the public. It is easy for owners to concentrate on the cost management concerns of their business and to lose the self-discipline necessary to monitor and maintain structures so that they are safe. Catching and storing water presents a potential hazard. The stored water possesses tremendous energy and can cause extensive damage and loss of life if the reservoir is released suddenly uncontrollably from dam failure. We've all seen the destructive power of large volumes of rushing water. The most frequent cause of dam failure is water overtopping the structure because the reservoir is inadequate to hold a flood. The second most frequent cause of failure is seepage and piping. Ever since dams have been built, failures have been caused in this way, often without apparent warning and usually catastrophically. Water seeps and erodes through the dam or its foundation or abutments. It pushes through paths of weakness or joints in rock, creating tunnels by erosion. These channels may be enlarged by continued erosion until the water quickly rushes under, around, or through a dam and the entire reservoir is suddenly empty. Piping is a condition difficult to detect. It requires an aggressive, continuous monitoring program. It is such a critical problem that the Federal Interagency Committee on Dam Safety has asked an internationally renowned engineer and educator to develop a presentation based on his extensive experience in dam and foundation engineering. Dr. Ralph Peck is graciously consented to share his experience and knowledge through these two videotapes. The first video is about failures caused by piping and how they developed. The topic of the second video is surveillance and preventive measures. Dr. Peck's involvement in saw mechanics and foundation engineering coincides with the development of saw mechanics as an engineering discipline, starting when he worked on the Chicago Subway Project in 1939 as an assistant to Dr. Carl Tersage, the father of saw mechanics. He has since served as a consultant for many governmental organizations, including the Bureau of Reclamation and Corps of Engineers. The San Francisco Bay Area Rapid Transit Project and the James Bay Hydro Project in Canada are two of many significant projects that he has been associated with, as well as a large number of major dams around the world. Dr. Peck was awarded the National Medal of Science by President Ford for his contributions in the United States, currently as Professor Emeritus at the University of Illinois. Dr. Peck. Almost a century ago, there were three notable dam failures in the United States. In 1907, Hauser Lake Dam was built across the Missouri River near Helena, Montana. It was a steel structure about 70 feet high and it rested on alluvium in the river bottom. There was a cutoff wall near the upstream toe that extended only about halfway through the alluvial stratum. The first time the dam was filled, the water broke through beneath the cutoff wall, washed the dam out, and it failed completely. In 1909, Ashley Dam near Pittsfield, Massachusetts, was constructed. It was about 50 feet high. It was a slab and buttress concrete structure, quite long, with abutments on rock, but a center section over alluvium. The first time it was filled, the dam failed completely, leaving the dam itself spanning over the hole where the alluvium had been. This failure was reported in the press at the time as being a remarkable test of the concrete superstructure, but of course, it wasn't much of a dam. In 1912, Port Angeles Dam was constructed near Bremerton over the River. It was about 100 feet high and it was a solid masonry dam. Initially, it didn't have a cutoff wall, but workers were beginning to construct a concrete case on under compressed air beneath the structure to the bottom of the alluvial valley in order to make a cutoff. To reduce the flow a little bit, they decided to construct a sheet pile cutoff wall downstream of the toe. The river came up while this work was in progress and washed out the dam again, causing a complete failure. These three failures, although they were of entirely different kinds of structures, a steel dam, a slab and buttress dam, and a solid masonry dam, had one thing in common. They all failed on first filling and, it's fair to say, they failed before the advent of modern soil mechanics. If we come up to 1951, Ontario Hydro built Sir Adam Beck number two power plant. This is at Niagara Falls and the scheme of the layout was to construct tunnels from above the falls through the dolomite bedrock to the escarpment above the gorge downstream from the falls. And a pump storage reservoir was constructed above the escarpment. The power plant was down below, of course, and the idea was that water could be stored at times when the falls were not a scenic attraction, that is late at night when there weren't people to look at it, for example. And then the water would be released to make power at times when the scenic attraction of the falls wouldn't allow people to reduce the flow. The reservoir was contained by a dyke, which had a conventional sloping clay core resting on a dumped rock fill with a couple filters in between. The dam rested on the Niagara dolomite, the bedrock of the region, and the rock fill itself was separated from the underlying rock by a cushion layer, two layers actually of finer grain material, so that the rock fill could be embedded on it instead of resting its sharp points on the underlying bedrock. One day, sometime after the dam was constructed, seven years as a matter of fact, one of the watchmen was walking the ditch, as he was supposed to do every day, and noticed that there was muddy water, and he set in motion the procedure for lowering the reservoir. And when the reservoir was emptied, it was discovered that there was a sinkhole in the dam and a sinkhole through the core. It was also discovered that the limestone dolomite beneath the structure was jointed. There were vertical joints, there were horizontal joints, and some of these joints near the upper part were filled with glacial till. And some of this fill had washed out so that there was an open channel from upstream above the dam to apart beneath the structure, and literally the force of the water, as the filling had washed out of this joint, was like a fire hose on the underside of the fill, it washed the cushion up into the rock fill, it let the rock fill subside, and that in turn let the core subside and cause the sinkhole in the core. And it also turned out that this particular joint that washed out had upstream from the dam been the location of a light pole that was used during construction. And to establish the pole, a hole had been blasted into the limestone bedrock. And that hole gave access to the joint for the water in the reservoir. It took seven years for that joint to wash out, but when it finally did the dam failed. The failure wasn't complete, it was localized simply because the warning was given soon enough by the alert inspector. And quite incidentally it was a holiday, a Canadian holiday, on which the dam failed and the inspector certainly gets credit for doing his job on that day. In 1965, the US Bureau of Reclamation constructed Fontanelle Dam. This was a rather long earth-filled dam going between sandstone and shale cliffs. It had its spillway on the right-hand side against rather steep abutments. The abutments were not treated by dental concrete. The treatment that was chosen for preventing leakage through the foundation and abutment materials was grouting. But the rock against which the core was placed itself was not treated with any sort of dental concrete, so they're occasional open joints. We see some here several inches wide, and against these open spaces the dam fill material, including core material, was placed. In the spring of 1968, when the dam was filled, seepage appeared near the toe of the dam, near the junction between the embankment and the spillway. And rather rapidly over a period of two or three days, the seepage made an enlarged hole, began to work its way back on the abutment. We see here in the fill a stage just before the roadway collapsed that goes to the spillway. When the roadway collapsed, the dam by that time had been pretty well drawn down so that it was not an actual failure in that the reservoir was dumped, but it was very close to being a failure. There were a series of violent surges just before the final collapse that took the spillway bridge with it, but the dam as a body was saved. The dam was repaired, it continued to exhibit some seepage downstream, and eventually many years later the cutoff wall was put through the dam to reduce the seepage and make it safe to fill completely. On the Far Flung Churchill Falls project in Labrador, a very large reservoir was constructed made by a series of dikes around the edges and the lowlands. Most of these dikes were established on rocks, some of them on glacial till. The bedrock up here is the Canadian Shield Granitoid rocks, nices, very strong component materials that have been polished off by the glaciers, and in most instances the foundations were entirely successful. One structure, however, developed some trouble at the time that the reservoir was being filled, a dike known as dike GJ-11A. That dike is an abutment of a control structure as you can see here. It was an earth fill glacial till core rock fill shoulders with several transition zones. Here we're looking down on it from above after a washout occurred the first time the reservoir was filled alongside the end of the dam. You can see that the wing wall for the control structure sits on rock and we can see that the rock had many joints in it. This is ancient pre-Cambrian granites and nices. There are joints primarily because under the weight of the glacier the rock was compressed and when the glacier melted away the rock rebounded and produced a number of flat-lying joints as well as right-angle joints vertical in several directions. It turned out that one set of these joints terminated on a rock surface, as you can see in this slide, where there was no opportunity for water to escape, but there was a wide open joint going upstream terminating in the reservoir that delivered water directly against this flat rock surface and shoved it up through a four-inch wide joint and as soon as the material washed out of that joint again the water behaved like a fire hose tore out the material from the core trench material above it and caused sinkholes at the surface and led to reconstruction of the dike. In 1975 we're getting more modern now. Walter Bolden Dam was constructed in Alabama. This is an air view of the completed works. We see a long wing dike on either side of the powerhouse which is located here in the middle. The curved structure upstream of the dam is the intake and down below we see the powerhouse itself and the tail race. Of course when the project was operating the tail race was full of water so that you didn't see the submerged part of the structure. The main part of the dam where the powerhouse is located was carried down to rock and into the underlying shift that formed the bedrock of the region. Above the shift was a layer of cretaceous clays and sands, a cohesive quite strong formation and above that terrace deposits that cover the entire territory and on which most of the dikes were founded. The project operated successfully for a number of years. This is a picture of the construction in which we can see the powerhouse embedded in the rock. We can see the base of the terrace up at this upper level and we can see that there was a berm in the dam itself at the level of the roof of the powerhouse. You can also see from the outlets how much of the tail race would be submerged when the powerhouse was in actual operation. Even though the facility operated successfully for a number of years it was being carefully inspected by a watchman whose job it was to tour the project every few hours. And one night, one dark foggy night, the watchman had quite a story to tell. He had made his rounds early in the evening, come back about midnight, went into his office and sat down to read the Sunday papers. Notice that water was coming through the door, the base of the door which he had closed in his office. You realize that was unusual, looked out, saw there was water on the deck of the powerhouse, more than there should have been just from the rain. So at one ten in the morning he called his supervisor and said something was wrong. The supervisor told him to go take a look around. He opened the door and looked out, saw that the water was getting deeper, decided it would be a good idea for him to leave, and came out to where his truck was parked and started to leave the property. This is a view of the path of his inspection you might say as he beat his retreat. You can see at this point labeled A the watchman's shanty. You can see that he went around to the edge of the structure where his truck was parked. By the time he got to the truck the water was about four or five inches deep and was very muddy and it seemed to be coming, he thought, from across the far corner of the powerhouse. He could see things not too well because there was a lot of switch gear and the like on top of the powerhouse, but he did notice that a lamp on top of the dike was still illuminated and he could see by the light of that lamp. He got in his car, drove around to the gate of the property, had to get out to unlock the gate so that he could leave, and he turned around and looked back and discovered that the water seemed to be going down. It wasn't as deep as it had been when he drove out. This got his curiosity so he walked back a ways to a bend in the road where he could see across the top of the powerhouse and as he stood there for a few minutes looking he heard the sound of large rocks falling. The water came back again, there was a big flash and all the lights went out. The lights went out at about 20 minutes after he called his supervisor. When the lights went out he could see very little but he had thought he had seen mist rising from behind the dam somewhere in the neighborhood of the old lamp post. He couldn't really be sure. At this stage of course he realized something serious was happening so he left waited outside the property for his supervisor to show up. By the time they got there within an hour of his first call the dam had failed completely and washed out. There was a lot of discussion as to the cause of the failure but the description of the watchman who was not a technical person who had no reason to think he understood what was going on he could simply describe things as he saw them makes a classic picture of a failure of piping that is an erosion tunnel. Most likely what happened that seems to agree with all of the information and with the stratigraphy is that somewhere underneath the level of the tailrace water and the tailrace was always full of muddy water because the streams are muddy in this part of Alabama. Somewhere there was erosion going on and a tunnel being formed working backward removal of the material grain by grain as it worked back toward the head water level. This is the way erosion tunnels work. It probably started either at the base of the powerhouse along one of the old excavations it hadn't been completely backfilled perhaps or tightly back filled or at the contact between the cretaceous and the underlying schist and the tunnel evidently worked its way back until it got up to the level of the roof of the powerhouse where it was then able to discharge water over the roof and it was discharging a considerable amount of water. As time went by the tunnel got bigger and when the big collapse occurred when the lights went out when he heard the watchman heard the rocks falling on the roof and the like is most likely the final breakdown of the arch over the tunnel and the dam failed by breaking through. Here we see a cross section of one edge of the structure up here is the lamp post and the top of the dam down here is the roof of the powerhouse the construction slopes that were built and even though people thought the cretaceous was presumably pretty impervious and not erodable we noticed that there were three lines of well points that were used to dewater the cretaceous while it was while construction was going on so most likely down here somewhere the erosion tunnels started as I've indicated and worked its way up and finally got up to a place where the lamp post fell down into the hole there had always been the consideration that the dam would be subjected to some seepage and piping at its base particularly the upper part of the dam but the material that was suspect was the terrace material and there had been over the years relief wells other drainage facilities to handle the seepage through that terrace material but the cretaceous material was felt for some reason to be relatively unerodable but after the failure occurred you could see all sorts of erosion channels and gullies in the cretaceous here's one example where you can see a number of crevices that had been eroded out another hole through which the water obviously had come the cretaceous was certainly not an unerodable material and was in all likelihood the seat of the failure very shortly after water bolden dam failed which was a delay of some years after it was first filled teton dam failed teton failed on first filling and I suppose it's the most famous of the dam failures that we have heard about in recent years the cause of the failure is a subject of still a certain amount of controversy there because there were a number of things that might have contributed to it here we see a plan of the dam looking down from the top there was first seepage noted down near the bottom of the groin it worked its way up the groin there was a tunnel that one could see at the top an erosion tunnel so obviously water was coming through and before long that tunnel approached the crest of the dam people tried to fill the hole with bill dozers they lost the bulldozers in the hole and before very long within a few hours the dam broke through and before the day was over the dam had not only failed but the reservoir was empty at teton although there were a number of factors that undoubtedly contributed to the failure one of them most likely was like at fontanel the compaction of core material against bedrock that had not been treated by some sort of dental concrete so that in effect core was compacted against open holes in the rock such as this one we see in the key trench this was a trench at the end of the core excavated in the rock to tie the core into the rock itself as you can see the rock was jointed and there were spaces like those that you see in the picture against which fill was placed with no resistance really against the materials being able to move into such an open space well we have seen in these illustrations three different kinds of failure all classified these days as failures by piping the first we call today piping by heave that's the kind of piping that occurred at the dams at the turn of the century for example they always occur on first filling of the reservoir that is if they're going to occur at all they occur when the water first acts on the base of the dam and we'll see more about that type shortly there is a second type of which walter bolden is probably a very good example that is backward erosion beneath some kind of a cohesive roof that permits an erosion tunnel to stand open until it gets to be a large span in which time it collapses that backward erosion between such a roof is necessarily then a delayed failure doesn't necessarily occur on first filling of the reservoir it hardly could occur at first filling although under some particular conditions it might and then there is a third type which is sort of intermediate in a way it's the removal of filling in joints and altered rock and other weak materials which we saw exemplified in serratum beck for example and that could happen very quickly after the reservoir is filled or it could happen as it did a serratum beck a number of years later it could be a first filling or it could be a delayed failure for piping by heave there is a good theoretical basis on which we can analyze the type of failure we may not be able to get numbers for final design but we get the concept and understand what's going on and can put a quantitative numbers on it this was first developed by Tertzage in about 1921 and illustrated by this flow net of the water going beneath a simple concrete block resting on the surface of an alluvial deposit retaining a body of water this flow net is just a graphical representation of the mathematical solution of the flow of water through a permeable material in this case one that has constant uniform properties in horizontal and vertical directions it's a homogeneous material and the mathematics lets us draw a diagram like this in which the flow can be divided into flow channels as we see here through each channel the same quantity flow passes in the same length of time and as the water goes past and goes through the flow channels it loses head and these dashed lines represent lines of equal potential or equal water pressure that represents how the head is lost as the water goes to the downstream toe since the same quantity of water flows through each of these flow channels at the same time it's pretty obvious that where the flow channels are very constricted and very small the velocity is greater than it is elsewhere and the corresponding seepage pressures are greater the uplift pressures at the downstream toe for example are very large in this area right next to the edge of the structure that's the smallest area the largest velocity so here's where the failure could occur and Tertzage demonstrated by model tests that this phenomenon agreed with the theory at the same time he came up with a method of taking care of the problem this use of filters or weighted filters because he investigated this problem for a client that was bidding on a particular job he patented the procedure and these photographs are from his original patent papers they are very simple but you can see a block here that represents a dam you can see a filter downstream and a weight on top of the filter and the whole idea was that where the flow lines would be concentrated and tend to come up just downstream of the dam the water could escape into the filter which is coarser than the underlying material and which is held in place by the weight on top he even went so far as you see in the second sketch as to put in some instances the filter underneath a part of the dam structure itself so that the dam provided some of the weight that held the material in place and by means of the flow nets for various conditions it's possible to see what kinds of remedial measures or preventive measures would be helpful there are several flow nets sketched in this in the following slides which illustrate these points in this flow net we see a cutoff wall placed near the downstream toe of a dam ordinarily if you had your choice you would put a cutoff wall all the way to the bottom so you could cut off the flow entirely but sometimes that isn't possible and this flow net shows a cutoff wall that only goes part way to rock down on the downstream toe and you can see that the flow nets have the the flow lines spread out on account of the presence of the sheet piles near the discharge point and that is a favorable element in reducing piping at the downstream toe the next flow net has a short cutoff wall on the upstream side which reduces the flow but of course doesn't cut it off completely but the flow channels are collected downstream in a filter like that that Tertzage had in his patent papers the next one shows a drain near the downstream toe and no cutoff at all there's more flow of course through this foundation there are fewer equipotential lines as you can see but the flow is harmless because it comes into a filter which is weighted down by the structure and this fourth one shows the effect of a long upstream blanket which makes the water flow a long way before it can get to the toe of the dam it reduces head all the time as it goes and that reduces the uplift pressures near the toe in this particular flow net there is also a filter at the downstream toe but the effect of the blanket is the thing that this primarily illustrates all these are tricks that can be used to reduce the likelihood of piping when a dam is being constructed in in the first place all these apply to a mathematically homogeneous material not stratified not with lenses and peculiarities of that sort in the material and all real materials have these defects all these that I have shown you so far are for materials that have the same permeability horizontally as vertically but we know that in nature in sands the permeability in a horizontal direction is usually very much greater than that in a vertical direction so these only give us ideas as to protective measures that we can use but they aren't really a sound basis for final design real soils are stratified with horizontal permeabilities greater than vertical permeabilities we can draw flow nets theoretically for those conditions just as well but the flow lines and the equipotential lines no longer make squares they make distorted figures and the amount of distortion as you see here it depends on the ratio of horizontal to vertical permeability but again the problem can be solved mathematically could be solved very nicely if we really knew what the permeability of the deposit was all these things are very helpful to us in understanding the problem and in conceiving a design but they're not things that we can rely on to give us numerically correct precise designs in actual practice the use of a blanket for example doesn't always work as a flow net might suggest here is a blanket on one of the dikes up on the james bay project you can see the dike here not very high you see a blanket going upstream a considerable distance but unfortunately this particular blanket didn't do the job because it turned out that at depth at great depth alongside the dike there was a buried channel with big boulders in it and when the water came up the first time a string of sink holes took off and you can see several of them here took off along around the end of the dam went on downstream and seepage occurred downstream and landslides began to develop as much as a quarter of a mile downstream from the toe of the dam obviously the water was getting into the foundation materials far upstream of the blanket and in this case it was necessary to draw the reservoir down while it was being filled for the first time put in deep cutoff walls some of these cutoff walls went through boulder beds at depths of a hundred or more feet that hadn't been discovered because they were not along the lines of the dikes they were between dikes the foundation materials between dikes the abutments were the leaky members not the dikes themselves fortunately there is a way to get around this problem of not knowing exactly what the stratification is we don't have to depend on the flow nets to determine what these uplift gradients are at the toe the gradients that cause the piping failures we can make simple measurements as you see here we can put in two piezometers one at depth and one near the surface and we can measure the difference in head between the piezometers and if the lower one rises higher than the one that's installed at a higher level then we know there's an uplift gradient having the knowledge of that gradient we can calculate the safety of the dam against blowing up at the toe so you might say that the piezometric installations can be used to verify the theory that tells us what the mechanics of piping are but more importantly these direct observations tell us what the real gradients are and we can base a design on that in 1913 the city of Dayton Ohio was submerged almost by a very great flood this led to the creation of the miami conservancy district to control flooding on the greater miami river in Ohio in the vicinity of Dayton and the surrounding communities you can see here how serious the flood was it devastated the downtown region and the miami you can see miami conservancy district was a very progressive organization arthur morgan was the chief engineer the conservancy district developed many of the features that were later adopted by most engineering organizations in the design of dams they built five large hydraulic fill dams here is one in process of construction the dams themselves are still in excellent shape the valley of the river however is filled with various fluvio glacial deposits and at the time the dams were built it wasn't realized that piping by heave could occur and so there's no particular effort spent in the construction of the foundations to cut off the water they made studies and decided that the foundations would be adequate later piezometric observations of the kind that I just described were made and indicated that there were uplift pressures these dams have never yet been filled the dams all have openings through them the idea is that they simply restrain the flood waters pond them for a while and let them out easily so that the city doesn't get flooded and at lower stages of filling the piezometric observations indicated small uplift pressures and if you extrapolate that to the full pool level you find that there would be a danger of the dams failing when the reservoirs were full so remedial measures of various kinds have been taken the principal kind being drainage so downstream of the dams today there are a number of relief wells and other measures have been taken to guard against this particular type of failure so piping by heave can be foreseen by flow net studies and a careful study of the stratification of the soil it can be checked up on by the installation of piezometers and careful observations at partial pool and the conditions under full pool can be determined by extrapolation with considerable accuracy actually long before piping by heave was understood and the theory demonstrated by tritzage people had tried to quantify the piping problem try to decide when there would be danger of piping when they wouldn't and the oldest efforts go back to a gentleman named bligh in 1912 who developed a line of creep theory he developed this by studying the failure of various dams mostly low dams associated with the irrigation system of india and he got the idea that the water flowed from upstream to downstream under a uniform hydraulic gradient and the longer the seepage path the less dangerous the condition might be and so he related the factor of safety against failure to the line of creep so called which on this protector sketch you can see we're represented by a particle of water that enters the subsoil just at the upstream end of the dam goes down the first sheet pile comes back up goes along the base of the dam down around the second sheet pile that whole length of path is the so called line of creep and bligh found that the ratio of the line of creep to the head at which the dam failed a quantity that he called the creep ratio could be related to the grain size of the material in the foundation the finer the material the more like of the dam was to fail somewhat later quite a bit later ew lane of the u.s. bureau reclamation made a much more detailed study he concluded that the vertical sections of the line of creep those that went up and down alongside the sheet piles for example were much more effective in reducing erosion produced much more head loss than the horizontal sections so he came up with a weighted creep ratio in which the horizontal sections were three times as ineffective as the vertical sections of the line of creep and his expression for the line of creep we see here and he too found that the line of creep was related that the that the critical creep ratio at which failure occurred was related to the grain size and those grain sizes for lanes theory are shown in this next table we can see that the weighted creep ratio has to be much higher for a very fine sand or silt than for a gravel neither lane nor bligh realized that they were dealing with two kinds of piping failures that some of the failures in the statistical studies that both of them made were failures due to heave today we realize that some of those examples should be thrown out of the study you might say but nevertheless what has turned out is that failures associated with backward erosion the second most important kind of piping are indeed related statistically to the grain size of the material and we can use these creep ratios of lane to get some guidance on the point the trouble is that any statistical study is just that it's based on experience and the experience may not cover all the possible situations so that there is always a possibility that the next dam that is designed on this basis to prevent backward erosion may be the exception and there is not a theory that one can completely rely upon however it was an important step in determining the causes and the control of piping by backward erosion so we have now looked at the causes of two kinds of failure by piping and the third kind the washing out of joints which is sort of intermediate between the two and we see that this is a very dangerous form of failure and we have gotten a glimpse as to the mechanisms that are associated with it and now we are faced with the question how do we prevent failures by piping if we've already got a dam how do we prevent them from occurring in the future how do we detect that a piping failure may be developing and if we see that there is a problem developing that might lead to piping how do we remedy the situation those will be the subjects of the second part of this discussion