Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
Dam Failures in Derna, Libya 25
- Thread starter SSCon
- Start date
The numbers: Estimation of the surface runoff depth of Wadi Derna Basin (sebhau.edu.ly)
100 mm over 575 km[sup]2[/sup] (Wikipedia, Wadi Derna) is 57.5 Mm[sup]3[/sup]. That approximates the 1945 flood volume. This rain event was 100 - 300 mm (reliefweb.int) over that same area which could be double that volume of water or more. The dam capacities are in the order of 20 and 1.5 Mm[sup]3[/sup] (various sources). Interestingly, the study included a single event of 145 mm but the abstract does not mention capacity issues for the dams. I'm curious about the rainfall rate and runoff/discharge rates used. I suspect a 100 Mm[sup]3[/sup] event could easily overwhelm the structures, even if they were in good condition.
I also suspect that a single surge from a dam failure would have an outsized impact relative to a steady heavy flow. The war time event would have been spared that peak. I still question whether a quadruple or quintuple capacity system is ever in the cards. I suppose an infrastructure comparison would be the lone Japanese town of Fudai (Youtube) that built an outsized Tsunami break wall.
[b said:
100 mm over 575 km[sup]2[/sup] (Wikipedia, Wadi Derna) is 57.5 Mm[sup]3[/sup]. That approximates the 1945 flood volume. This rain event was 100 - 300 mm (reliefweb.int) over that same area which could be double that volume of water or more. The dam capacities are in the order of 20 and 1.5 Mm[sup]3[/sup] (various sources). Interestingly, the study included a single event of 145 mm but the abstract does not mention capacity issues for the dams. I'm curious about the rainfall rate and runoff/discharge rates used. I suspect a 100 Mm[sup]3[/sup] event could easily overwhelm the structures, even if they were in good condition.
I also suspect that a single surge from a dam failure would have an outsized impact relative to a steady heavy flow. The war time event would have been spared that peak. I still question whether a quadruple or quintuple capacity system is ever in the cards. I suppose an infrastructure comparison would be the lone Japanese town of Fudai (Youtube) that built an outsized Tsunami break wall.
-
2
- #42
SymP, it would indeed be interesting to see fainfall rates, duration, storm travel speed and basin coverage. Unfortunately we will probably be left guessing.
Have you been able to download that first link? I can't seem to do it. Site keeps jumping back to the original icon. Can't even keep it's login page open.
Medicane
It's a new term to describe known phenomenon (reaching hurricane like size, similar high water temp cause and effects) occurring in regions where they haven't been seen at the same magnitude before.
Brian, no hydrotest is required. These effects can be predicted by a simple Excel open channel flow calculation, or more elaborate CFD study for a few thousand bucks. We don't have to fill dams with water and blow them up. The potential dangers these dams presented are easily recognisable and, no doubt, would have been recognised assuming that even a minimal amount of study was conducted. Such extreme risk would have been identified and the project halted. But it was built. Why? Could we assume that at the time these were built, the risk was classified as minimal. With hindsight of this disaster, how it could have presented a minimal risk might seem unfathomable today, but that could easily have been the case.
Was the design of these dams perfectly adequate for the design conditions identified back in the day? Why would we assume otherwise, noting that at least so far, there is no evidence to the contrary and we have no reason to believe that they were poorly designed. We expect engineers, even way back then, to be able to build dams. So let's consider what might have happened since then that made these dams such a high risk project that eventually killed 11,000 people.
We are immediately inclined towards thinking, well there must have been a problem with the physical design or construction of the dam, or, think, Of course it happened, it wasn't maintained for years. Or, they let people move into that high risk area. While it is true that all those factors could have contributed to increasing risk, there is at least one more important factor we should consider. A factor that is probably more easily recognised in projects designed for a specific production capacity, such as chemical plants, refineries, power generators and pipelines, among others. When you run at more than design capacity, things start going wrong fast. For dams and reservoirs, we usually think of maximum capacity in terms of water level, but not necessarily about how that was determined. And when we do think about maximum rainfall and drainage area, we tend to think that those numbers are sufficiently high, relatively permanent constants that will give us safe results, rather than considering them as highly dynamic variables. Even if we did, would we consider them to have a potential variation range of 2 or 3 times what is assumed as "a basically maximum constant design value". Probably not. We would have to build a refinery with twice our initially planned capacity. We would never do that. It would never be an economically viable plan.
So, what this disaster suggests is that rainfall rates, i.e. climate data was once probably adequate for the safe design of these dams, and they survived for 50 some years. People got killed by living in the wrong place, but that perspective may be based on hindsight alone. Before today's rainfall data, the dams could have been safe, but today we know they are not. Sure, more people are living in high risk areas up the consequences, but those people had nothing to do with the high rainfall rates. High casualties are a symptom, not the root cause. What this disaster apparently proves is that the design conditions for these dams have changed. Nobody would build them as they were based on this weeks rainfalls.
What can we do to solve these specific types of problems (dams, reservoirs, flood controls)? Each and every dam and potential flood zone (maybe in the whole world, you be the judge) needs to be revaluated using the new data. If I lived downstream of one, you can bet I surely would have done it already. The full monty, CFD and all. In some cases it may be feasible to build bigger and better, but where high variability is present, tear down may be the only solution. I predict a great future for hydrologists.
We can be skeptical of reasoning behind this, but it seems obvious to me that changing design conditions are forcing us to consider higher winds, higher tides, more fires, more and higher flood controls, desertification, invasive species, glaciers and disappearing water resources , Great Salt Lake and Colorado River diminishing flow and higher offshore platform deck levels. All those are facts and we are already taking those measures. Being skeptical does not prevent disaster. Even the most ardent former skeptics are being forced to admit to it and are now saying, Oh just let it happen. Its too expensive to do anything about it, but what that fails to recognize is that it may also be too expensive not to do anything about it. You can be skeptical if you want, but [spoiler alert] the skeptic frog in a warming pot of water does not survive. Doing nothing also has its costs.
Is it true that medicanes will become prevalent in the Med, with their very few hour rainfall rates totalling 77% of the previous yearly accumulations for these cities. Only time will tell, but it looks like that's where we're going.
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
Have you been able to download that first link? I can't seem to do it. Site keeps jumping back to the original icon. Can't even keep it's login page open.
Medicane
It's a new term to describe known phenomenon (reaching hurricane like size, similar high water temp cause and effects) occurring in regions where they haven't been seen at the same magnitude before.
Brian, no hydrotest is required. These effects can be predicted by a simple Excel open channel flow calculation, or more elaborate CFD study for a few thousand bucks. We don't have to fill dams with water and blow them up. The potential dangers these dams presented are easily recognisable and, no doubt, would have been recognised assuming that even a minimal amount of study was conducted. Such extreme risk would have been identified and the project halted. But it was built. Why? Could we assume that at the time these were built, the risk was classified as minimal. With hindsight of this disaster, how it could have presented a minimal risk might seem unfathomable today, but that could easily have been the case.
Was the design of these dams perfectly adequate for the design conditions identified back in the day? Why would we assume otherwise, noting that at least so far, there is no evidence to the contrary and we have no reason to believe that they were poorly designed. We expect engineers, even way back then, to be able to build dams. So let's consider what might have happened since then that made these dams such a high risk project that eventually killed 11,000 people.
We are immediately inclined towards thinking, well there must have been a problem with the physical design or construction of the dam, or, think, Of course it happened, it wasn't maintained for years. Or, they let people move into that high risk area. While it is true that all those factors could have contributed to increasing risk, there is at least one more important factor we should consider. A factor that is probably more easily recognised in projects designed for a specific production capacity, such as chemical plants, refineries, power generators and pipelines, among others. When you run at more than design capacity, things start going wrong fast. For dams and reservoirs, we usually think of maximum capacity in terms of water level, but not necessarily about how that was determined. And when we do think about maximum rainfall and drainage area, we tend to think that those numbers are sufficiently high, relatively permanent constants that will give us safe results, rather than considering them as highly dynamic variables. Even if we did, would we consider them to have a potential variation range of 2 or 3 times what is assumed as "a basically maximum constant design value". Probably not. We would have to build a refinery with twice our initially planned capacity. We would never do that. It would never be an economically viable plan.
So, what this disaster suggests is that rainfall rates, i.e. climate data was once probably adequate for the safe design of these dams, and they survived for 50 some years. People got killed by living in the wrong place, but that perspective may be based on hindsight alone. Before today's rainfall data, the dams could have been safe, but today we know they are not. Sure, more people are living in high risk areas up the consequences, but those people had nothing to do with the high rainfall rates. High casualties are a symptom, not the root cause. What this disaster apparently proves is that the design conditions for these dams have changed. Nobody would build them as they were based on this weeks rainfalls.
What can we do to solve these specific types of problems (dams, reservoirs, flood controls)? Each and every dam and potential flood zone (maybe in the whole world, you be the judge) needs to be revaluated using the new data. If I lived downstream of one, you can bet I surely would have done it already. The full monty, CFD and all. In some cases it may be feasible to build bigger and better, but where high variability is present, tear down may be the only solution. I predict a great future for hydrologists.
We can be skeptical of reasoning behind this, but it seems obvious to me that changing design conditions are forcing us to consider higher winds, higher tides, more fires, more and higher flood controls, desertification, invasive species, glaciers and disappearing water resources , Great Salt Lake and Colorado River diminishing flow and higher offshore platform deck levels. All those are facts and we are already taking those measures. Being skeptical does not prevent disaster. Even the most ardent former skeptics are being forced to admit to it and are now saying, Oh just let it happen. Its too expensive to do anything about it, but what that fails to recognize is that it may also be too expensive not to do anything about it. You can be skeptical if you want, but [spoiler alert] the skeptic frog in a warming pot of water does not survive. Doing nothing also has its costs.
Is it true that medicanes will become prevalent in the Med, with their very few hour rainfall rates totalling 77% of the previous yearly accumulations for these cities. Only time will tell, but it looks like that's where we're going.
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
-
1
- #43
Brian Malone
Industrial
1503-44 your analysis is very complete and knowledgeable and I certainly recognize humans are affecting the very world we live, yet does not change my belief the impact of weather events to humans is not necessarily because the events are more intense - it is because either humans have moved into risk prone locations or our population has increased in such locations - often because humans use technology for mitigation or financial methods such as insurance to encourage utilization of areas with potential for loss. The parable of a frog in a warming pot is analogous to human behavior not that of a frog in a natural setting because a frog floating in warming liquid would leave before the temperature is dangerous. If given no path of leaving the frog would exhibit distressed behavior. Frogs do not live in pots - only humans do.
Estimation of the surface runoff depth of Wadi Derna Basin (sebhau.edu.ly)
The PDF is attached below
The PDF is attached below
-
4
- #45
Reader's response to an article in today's Times (UK, London version):
Its not just the locals, the media need a hard navel gaze too. We used to shrug and blame "gods" - now media shrug and blame climate change and "worst floods on record". Few have asked why the Wadi is what it is, why (and by who) the dams were built on it, how these dams ended the city's flood problem, which was causing huge losses of life and property. These include the 1941 flood, which caused great losses to the German army, the 1956 flood, the catastrophic flood of 1959, the 1968 flood, and the 1986 flood, which, although large, the dams prevented damage to the city. So the city grew, oblivious to the old threat. The dams have not been maintained for over 20 years since Libya is a failed state (wonder why), and then another storm arrived, as science predicted. Not man's fault - bend the knee to the climate change god and carry on BAU.
Politicians like to panic, they need activity. It is their substitute for achievement.
Its not just the locals, the media need a hard navel gaze too. We used to shrug and blame "gods" - now media shrug and blame climate change and "worst floods on record". Few have asked why the Wadi is what it is, why (and by who) the dams were built on it, how these dams ended the city's flood problem, which was causing huge losses of life and property. These include the 1941 flood, which caused great losses to the German army, the 1956 flood, the catastrophic flood of 1959, the 1968 flood, and the 1986 flood, which, although large, the dams prevented damage to the city. So the city grew, oblivious to the old threat. The dams have not been maintained for over 20 years since Libya is a failed state (wonder why), and then another storm arrived, as science predicted. Not man's fault - bend the knee to the climate change god and carry on BAU.
Politicians like to panic, they need activity. It is their substitute for achievement.
Brian, thanks. I try to make some sense of these things.
Hpwever I do think your reasoning confuses mortality data with rain gages. Saying that mortality data depends solely on how many people live in risk prone areas is certainly true, but it ignores why the area is classed as high risk. Take earthquakes. No association with climate change that I know of, so its pretty safe to assume your theory is 100% correct there. If I did a regression of earthquake mortality and climate data, I.e. rainfall rates and durations, size of storms etc., there is no way I would expect to see any connection whatsoever to any of those data points. But if I did the same analysis on drunk driving and traffic deaths, that connection would be very obvious. Now, why don't you do a regression analysis of flood deaths and rainfall intensity in a flood risk region and tell us what that regression study shows. Sure its important to realize that regression analysis is not proof of cause and effect, but it also does not prove that it isn't a cause and effect. What it does is add strongly to the evidence when a possible connection is seen.
In the end, as I say, we are increasing the design conditions anyway. The insurance companies demand that to minimize their payout risk. Doesn't matter if we see a connection or not. They see it and costs of increased design robustness are already on us one way or another. Nobody will insure an offshore platform with a 20ft lower deck height. Have you tried to get flood insurance lately?
Multiple Variable regression analysis
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
Hpwever I do think your reasoning confuses mortality data with rain gages. Saying that mortality data depends solely on how many people live in risk prone areas is certainly true, but it ignores why the area is classed as high risk. Take earthquakes. No association with climate change that I know of, so its pretty safe to assume your theory is 100% correct there. If I did a regression of earthquake mortality and climate data, I.e. rainfall rates and durations, size of storms etc., there is no way I would expect to see any connection whatsoever to any of those data points. But if I did the same analysis on drunk driving and traffic deaths, that connection would be very obvious. Now, why don't you do a regression analysis of flood deaths and rainfall intensity in a flood risk region and tell us what that regression study shows. Sure its important to realize that regression analysis is not proof of cause and effect, but it also does not prove that it isn't a cause and effect. What it does is add strongly to the evidence when a possible connection is seen.
In the end, as I say, we are increasing the design conditions anyway. The insurance companies demand that to minimize their payout risk. Doesn't matter if we see a connection or not. They see it and costs of increased design robustness are already on us one way or another. Nobody will insure an offshore platform with a 20ft lower deck height. Have you tried to get flood insurance lately?
Multiple Variable regression analysis
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
Debunked. But you get the picture, right?
Puds post, Reader comment is the reason behind the high number of deaths, but it misses the probable root cause entirely. It most likely hits the action plan square on.
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
Puds post, Reader comment is the reason behind the high number of deaths, but it misses the probable root cause entirely. It most likely hits the action plan square on.
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
TugboatEng
Marine/Ocean
Lack of maintenance wasn't the probable root cause?
TugboatEng
Marine/Ocean
The dams were built for the rainfall! The valley has a long history of deadly floods and the dams stopped that for some time. Nobody has indicated that the dams spilled over. As others have noted, the basins are dry most of the year. It seems the dams broke while filling and not due to being overfull.
There is no precise information on exactly when or why the dams did not hold.
What has been reported is
30Mm3 of water passed through the city in an apparently very short time.
Water depth reached 7m in the city.
LittleInch has estimated the dam retention volume as roughly 2Mm3.
The dams are now gone.
If the dams held, that would mean that there was a 32Mm2 of rain water shed from the basin, 2M of which would still be behind the dam. They did not hold, so we assume shed rainfall was 30M.
A water shed 15 X dam capacity tells me that, if the dams broke or held isn't important, as in either case the city would be whipped out by either 30M, or 32Mm3 of water. That's an inadequate capacity issue. If these dams were designed properly 50yrs ago, we can suppose that a 2Mm3 rainfall runoff was the expected runoff, including whatever safety factor was used in determining runoff volume and adequate dam retention design capacity. Today we see runoff is now 30M, 15X what apparently was considered adequate 50yrs ago. Why is that? 50yrs ago they knew that it would be dangerous to build dams there if they got the capacity wrong; they knew the wadi had severely flooded several times already. They must have wanted to get it right used the best available rainfall data at the time.
Today we have only two possible answers as to why dam retention capacity is inadequate.
1) Engineers 50 years ago had no idea what they were doing and underestimated runoff, using a value of only 2/30 = 7% of what they should have. They were complete idiots? Or was the data totally wrong and they made no attempt to verify it by looking at trace levels of previous floods.
2) Engineers 50yrs ago did a proper job with their rainfall/runoff calculations, but for some reason runoff expectations today need to be 15X higher than the 2Mm3 they calculated 50yrs ago. "The answers have changed."
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
What has been reported is
30Mm3 of water passed through the city in an apparently very short time.
Water depth reached 7m in the city.
LittleInch has estimated the dam retention volume as roughly 2Mm3.
The dams are now gone.
If the dams held, that would mean that there was a 32Mm2 of rain water shed from the basin, 2M of which would still be behind the dam. They did not hold, so we assume shed rainfall was 30M.
A water shed 15 X dam capacity tells me that, if the dams broke or held isn't important, as in either case the city would be whipped out by either 30M, or 32Mm3 of water. That's an inadequate capacity issue. If these dams were designed properly 50yrs ago, we can suppose that a 2Mm3 rainfall runoff was the expected runoff, including whatever safety factor was used in determining runoff volume and adequate dam retention design capacity. Today we see runoff is now 30M, 15X what apparently was considered adequate 50yrs ago. Why is that? 50yrs ago they knew that it would be dangerous to build dams there if they got the capacity wrong; they knew the wadi had severely flooded several times already. They must have wanted to get it right used the best available rainfall data at the time.
Today we have only two possible answers as to why dam retention capacity is inadequate.
1) Engineers 50 years ago had no idea what they were doing and underestimated runoff, using a value of only 2/30 = 7% of what they should have. They were complete idiots? Or was the data totally wrong and they made no attempt to verify it by looking at trace levels of previous floods.
2) Engineers 50yrs ago did a proper job with their rainfall/runoff calculations, but for some reason runoff expectations today need to be 15X higher than the 2Mm3 they calculated 50yrs ago. "The answers have changed."
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
A Turkish firm was contracted in 2007 to carry out maintainence on the two dams and build another dam in between. The firm, Arsel Construction Company Ltd., said on its website that it completed its work in November 2012. It didn’t respond to an email seeking further comment.
la belle vie
Structural
Another tragedy is the lack of failure modes analysis, surveillance and monitoring, and emergency planning. There are plenty of dams around the world that have vulnerabilities. But the potential failure modes need to be recognized, continuously monitored, and planned for. Every dam whose failure could kill people (high hazard) needs to have an emergency action plan kept up-to-date and regularly exercised. This way when conditions indicate that a PFM could be triggered, people can be clearly warned to get out of harms way. For all of this to happen there needs to be a stable government to regulate and enforce a dam safety program. Here in the US, dams are regulated by the states and the federal government, but we are not perfect. Alabama started regulating their dams only this year, and with some dams there can be political pressure to keep unsafe dams operating to maintain waterfront property values.
Regardless of the magnitude of the rainfall event that was used to design the dams, they should have been designed to handle that magnitude of event when they were already full. Even if the dams were designed solely to manage "rare" flood events and the dams are normally empty, there is a very real possibility of two or more rain events in short succession, such that the dams are still full when a subsequent major rain event strikes.
The only way to do this without overtopping the dam crest is some sort of emergency spillway, which I cannot see on the aerial images of either dam.
The only way to do this without overtopping the dam crest is some sort of emergency spillway, which I cannot see on the aerial images of either dam.
Per LittleInch and the Google Earth image, both dams had overflow systems in place. As was demonstrated at the Oroville dam, a spillway can be insufficient even given a relatively minor amount of spill. It is very difficult to provide a spillway on earthen structures as even a slight over-capacity of the spillway will immediately begin eroding the dirt from around and under the spillway.
concur about the difficulty... the Oroville dam failed partially because the concrete spillway was poor... concrete spillways are often used with earthfill dams and generally function as they should.
-----*****-----
So strange to see the singularity approaching while the entire planet is rapidly turning into a hellscape. -John Coates
-Dik
-----*****-----
So strange to see the singularity approaching while the entire planet is rapidly turning into a hellscape. -John Coates
-Dik
If they had an overtopping spillway, it most certainly would not have been designed for 15X and we could add that material to whatever went through the city. They certainly would have failed.
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."
Brian Malone
Industrial
Without accurate data on the rainfall from 1941, 1956, 1958, or 1968 I am not convinced the 'medicane' event is a new weather event. I don't think it is known what input/output criteria was used for the dam designs and if they were designed with sufficient flow and storage capacity for an event such as those reported in 1941, 1956 . . . I have looked at the Google Earth images of the Derna wadi dams and they are retention flow dams with bellmouth overflow structures, so they would be damaged/ destroyed with overtopping. Maybe the dams were underdesigned and for the last 50 years the rainfall has not reached levels such as the level that occurred in the flood years - is enough information available to decide?
- Status
Similar threads
- Locked
- News
- Locked
- Question
