Katrina in Your Rearview Mirror

W. F. Marcuson III, Ph.D., P.E. and Lawrence H. Roth, P.E., G.E.

Abstract In August 2005, Hurricane Katrina flooded 80 percent of New Orleans to depths of 10 feet, killed over 1,100 people, and caused more than $30B in damages. More than 400,000 fled the city, 125,000 jobs were lost, and tight-knit communities were destroyed. New Orleans, the bull’s-eye for many hurricanes in the Gulf of Mexico, lies in a bowl at or below sea level between the Mississippi River and Lake Pontchartrain. Following Hurricane Betsy in 1965, the US Army Corps of Engineers planned improved defenses, a Hurricane Protection System (HPS) comprising 350 miles of levees and floodwalls. The HPS, unfinished when Katrina hit, proved inadequate. Despite earlier studies that predicted both a storm of Katrina’s magnitude and its consequences, the risks were not recognized and residents were hesitant, unwilling, or unable to evacuate. Looking back, the lessons from Katrina are straightforward. A political culture that did not understand the potential for catastrophe, compromised by poor land use planning and questionable engineering decisions, gave insufficient consideration to risk and was unwilling to pay the price for adequate protection. Though Katrina overwhelmed the HPS, the HPS did not meet the design intent. Uncontrolled overtopping caused 50 breaches and four floodwalls collapsed before water reached design levels. Other contributing factors include: 1) the HPS was a system in name only; 2) because of many jurisdictions the management of the HPS was dysfunctional, no one was in charge; and 3) pressure to cut costs compromised safety. As we look back on the 10 years since Katrina, were the lessons merely noted, or have we adapted to meet today’s challenges: recognizing and communicating risks posed by flooding; preparing for emergencies; making required investments; reducing vulnerabilities; and improving our flood risk reduction systems?

Introduction In August 2005, Hurricane Katrina hit the gulf coast just east of New Orleans, Louisiana, causing about $30B in damage and the loss of more than 1,100 lives in and around New Orleans. Warren Buffet has been quoted as saying, “The stock market is clearer through your rearview mirror than your windshield.” So are engineering disasters. This paper discusses the setting, what went wrong, the risks, the lessons learned, and finally draws some conclusions.

The Setting Figure 1 is an aerial view of the general New Orleans area. Highlighted is Katrina’s path, the Mississippi River Gulf Outlet (MRGO), Lakes Pontchartrain and Borgne, the Mississippi River, and Plaquemines Parish. A yellow star depicts the location of the French Quarter. Figure 2 is a map of New Orleans, circa 1849. Again, the French Quarter is flagged with a yellow star. Note that the area north of the city and south of Lake Pontchartrain is described as a Cyprus swamp and swamp forest. Figure 3 is a more or less north-south slice through New Orleans and shows that much of the city is below sea level. Actually the banks of the Mississippi River and Lake Pontchartrain are some of the areas of higher elevation in the city.

Figure 1. Aerial view of Southeast Louisiana showing Katrina’s path.

Figure 2. Map of New Orleans area in 1849.

Figure 3. Cross-section through the city of New Orleans.

Figure 4 shows the three main components of the Hurricane Protection System (HPS) around New Orleans. At the top is the earth embankment with a crest nominally at elevation 12 ft. msl and side slopes at 1V to 3H. Below this is a T-wall design, which tops out at el. 12 ft. msl and is supported by piles, two-thirds of which are battered. Below this is an I-wall design showing the sheet pile top at el. 12 ft. msl and tip at el. -20 ft. It is driven in the embankment with a crest at el. 6 ft. msl and side slopes of 1V to 3H. There were about 350 miles of levees, 284 of which were federal levees and 66 miles were non-federal levees. There were 56 miles of I-walls and 2 miles of T-walls. In general, the top of the HPS was supposed to be 12 to 15 ft. msl. Prior to the mid-1960s there had been a levee system in and around the New Orleans area with a crest elevation of about 5 or 6 ft. msl. Following Hurricane Betsy in 1965, Congress authorized raising the crest elevation of the HPS about 6 ft. There are at least three ways to do this. One way is to increase the size of the earth embankment. If the slopes stay at about 1V to 3H and new construction is on the land side, or protected side of the levee, this requires acquisition of real estate, which can be expensive in urban areas. It also increases the overburden pressure on the foundation material making construction difficult when the foundation material is very soft. Remember that much of this area is or was a Cypress swamp. Another way to raise the crest elevation is by driving a sheet pile wall in the levee centerline to form an I-wall. This increases the height of the levee, does not increase the load on the foundation material significantly, and the actual construction is reasonably quick and easy. The third approach is a T-wall. This approach involves building a floodwall that is supported on a pile foundation. Batter piles are used to support hydraulic loads. Probably for ease of construction, the new I-walls were placed on the centerline of the existing levee. This, however, had the unfortunate effect of cutting off vehicular access to areas of concern; for example for flood fighting or levee repair. An enduring image of the Katrina disaster is the helicopters hauling large bags of earth materials to close the levee breaches.

Figure 4. Typical design components of the New Orleans hurricane protection system.

Levee

T-Wall

I-Wall

7

Figure 5 is an aerial view of the Gulf of Mexico during Hurricane Katrina. Note the size of this hurricane. Katrina stretches from the Yucatan Peninsula in Mexico to the Gulf coast areas of Louisiana, Mississippi, Alabama, and Florida – Katrina was a massive storm. The observed peak storm surge was at Waveland, Mississippi, and was about 27 ft. Hurricane wind vectors in the northern hemisphere are counter-clockwise. Looking back at Figure 1 and the path of Katrina shows that counterclockwise winds would tend to push water from the Gulf into Lake Pontchartrain. Figure 6 shows a hindcast of the storm surge caused by Hurricane Katrina; note that the water level in most of Lake Pontchartrain was increased by 6 to 8 ft.

Figure 5. Aerial view of the Gulf of Mexico showing Hurricane Katrina.

Figure 6. Peak storm surge during Hurricane Katrina. Figure 7 shows rainfall data from Hurricane Katrina. It is remarkable that much of New Orleans received 9 or 10 in. of rainfall during Katrina and some areas received as much as 13 in. Since New Orleans lies in a bowl, all of that precipitation had to be pumped out of the city. When Katrina came ashore, it was likely a category 2 hurricane and many people wondered how the weakening storm could wreak such havoc. However, the seeds of disaster were sown when Katrina was in the Gulf and was a category 4 or 5 storm. As Katrina crossed the Gulf, many days of wind from this huge hurricane pushed the storm surge in a counterclockwise and northwesterly direction, right at New Orleans.

What Went Wrong Nothing about Katrina should have surprised us as we saw it coming. In fact, an article entitled “The Creeping Storm” was published in the June 2003 issue of Civil Engineering, the magazine of the American Society of Civil Engineers (ASCE), which states “...if a lingering category 3 storm – or a stronger storm, say category 4 or 5 – were to hit the city, much of New Orleans could find itself under more than 20 ft. (6 m) of water...”. The catastrophe during Katrina was borne out because of a failure to recognize how fragile the levees were and how devastating the consequences would be to the local population. Katrina simply overwhelmed the HPS. While the storm loads exceeded the original design criteria, the constructed project did not meet the design intent. Katrina damaged about 170 miles of levees and caused 50 breaches. These breaches increased the flooding by about 300 percent. Figure 8 shows an analysis of flood water levels in the New Orleans area. The left side of Figure 8 is actual data, which shows the depth of flood water during and after Katrina. The right side of Figure 8 shows the results of a

numerical analysis using rainfall data and assuming there was no breaching of the levees. Much of New Orleans was under as much as 10 ft. of floodwater after Katrina. If the levee system had not failed, however, the analysis shows much of New Orleans would have still experienced water levels to a depth of about 2-3 ft.

Figure 7. Rainfall during Hurricane Katrina. The HPS, begun in 1965, was scheduled for completion in 2015. It comprised an extensive array of earth levees, I-walls, T-walls, gates of many kinds, and pump stations. Numerous levee districts in the New Orleans area operated and maintained various reaches of the HPS, but no one entity was in charge of the whole system. Unfortunately, the HPS was not ready when Katrina came ashore. Surging floodwaters found gates left open, and water poured through incomplete sections of the HPS. Major pump stations, which had not been made hurricane-safe were abandoned, or were otherwise inoperable. As a result, the HPS did not perform as a system – Katrina found the weak, incomplete, or inoperable parts of the HPS and floodwaters surging through these flaws contributed to the disaster. There were two direct causes of levee breaching. First, uncontrolled overtopping and the ensuing erosion led to catastrophic failures of levee and floodwalls. Second, four I-walls collapsed before the water reached the design levels. Figure 9 is a schematic showing how the uncontrolled overtopping of levees constructed of erodible materials can lead to levee failure. Not all levees around New Orleans that were overtopped performed poorly, however. Figure 10 shows a levee being overtopped at the top (Figure 10a); the same levee section after the waters subsided is shown at the bottom (Figure 10b). Note that Figure 10b is taken from a slightly different location than Figure 10a. Clearly this levee section was overtopped and performed remarkably well.

Figure 8. Comparison of flood levels in Orleans East Bank with and without breaching.

Figure 9. Schematic showing uncontrolled overtopping and ensuing erosion that led to levee and I-wall failures

Figure 10. View of levee section during and following Hurricane Katrina.

a. Levee overtopping during Katrina.

b. View of the same levee after Katrina (Courtesy of Peter Nicholson, University of Hawaii at Manoa).

Four I-walls collapsed before the water reached the design levels because the design failed to account for: 1) the variability in soil strength; 2) I-wall deformation, which opened a water-filled gap on the flood side; and 3) critical water pressures beneath the levee. Let’s discuss the variability in soil strength first. The soil borings performed to collect information for I-wall design were made at the levee centerline. Why was this? Because it was easier and more economical to get a drill rig on the crest of a levee than at the toe, and it did not require an easement or access onto property owned by others. Apparently, the designer assumed the soil strength at the levee toe was equal to the soil strength under the centerline. However, the soil strength of normally consolidated clay is a function of overburden pressure. In this case, the weight of the levee increased the overburden pressure on the soil beneath the levee and the levee had been in place long enough to consolidate the underlying soil. Therefore, the strength at the toe was less than the strength under the centerline. Figure 11 is a plot of shear strength in tons/sq. ft. vs elevation at about -20 to -30 ft. msl. This is the critical depth as the sheet piles generally terminated at elevation -20 ft. Also shown on Figure 11 are the shear strength envelope used by the Corps during design and the shear strength envelopes at the centerline and toe that were developed by the Interagency Performance Evaluation Taskforce (IPET) after Katrina. Note that the shear strength at the toe is substantially less than that at the centerline, a critical oversight that contributed to I-wall failure.

Figure 11. Shear strength data showing the Corps’ design envelope and IPET’s interpretation of shear strength at the centerline and levee toe.

Let’s turn our attention to the wall deformation that opened up a water-filled gap on the flood side of the wall. Water pressures acting on I-walls during Katrina caused the walls to deflect, which opened up a water-filled gap as depicted in Figure 12. The water-filled gap had the unfortunate effect of greatly increasing hydrostatic pressures on the wall, and introducing critical water pressures to soil layers below the wall.

a. Schematic showing the water-filled gap.

b. Photograph showing the water-filled

gap following Hurricane Katrina.

Figure 12. Formation of the water-filled gap during Hurricane Katrina.

The water-filled gaps contributed to two I-wall failures on London Avenue – London Avenue South and London Avenue North. The foundation at London Avenue South consisted of a layer of marsh directly above dense sand. The sand had Standard Penetration Test (SPT) blow counts of 50 or greater and an estimated friction angle of 40 to 45 degrees (Duncan et al. 2008). The water-filled gap allowed full hydrostatic pressure to be applied directly to the sand, increasing the uplift pressure, leading to a blow-out and massive erosion of the sand. Figure 13 is a photo showing the results of massive erosion of sand at the I-wall failure at London Avenue South. At London Avenue North the marsh was underlain by a thin layer of lacustrine clay, which was in turn underlain by relatively loose sand. The SPT blow counts in the sand at the north failure site were between 2 to 14 and averaged about 10, and the friction angle of the sand was estimated to be about 30 to 34 degrees (Duncan et al. 2008). At London Avenue North the water filled gap again allowed full hydrostatic head to be applied directly to the sand foundation, which increased the uplift pressure on the lacustrine clay leading to failure. The I-wall failure site on the 17th Street Canal can be described as a levee sitting directly on marsh deposits that are underlain by lacustrine clay, which is in turn underlain by sand. Here the failure was in the lacustrine clay layer. The water-filled gap developed and just cut the levee in half, which shortened

the failure surface and applied full hydrostatic pressure directly to the full height of the I-wall. This hydrostatic pressure is greater than the active earth pressure, which led to failure of the weak lacustrine clay. Driven by hydrostatic pressure, the wall moved the soil like a plow moves snow. Figure 14 shows the side of an excavation at the 17th Street Canal failure and shows the tongue of the clay protruding into the marsh. The reader is referred to Duncan et al. (2008) and Brandon, Wright, and Duncan (2008) for the original work on these failures.

Figure 13. Photograph showing massive erosion at the London Avenue South I-wall failure (Courtesy of J.M. Duncan, Virginia Tech).

Figure 14. Excavation showing the failure surface at the 17th Street I-wall failure (Courtesy of J.M. Duncan, Virginia Tech).

During the mid-1980s, the Corps’ Lower Mississippi Valley Division conducted a field test on an I-wall section at a site near Morgan City, Louisiana, which had soft soil conditions similar to those at New Orleans. Details of this work are summarized in an E 99 report, which says “Although the test wall was not loaded to failure...failure may have been imminent” (Jackson 1988). Figure 15, based on the report, illustrates this near-failure condition by showing the top of the wall deflection becoming asymptotic as the water level increases from 7 to 8 ft. Many people have since criticized the Corps for apparently ignoring its own research. This criticism is not completely valid since the test section near Morgan City was specifically designed so a water-filled gap could not occur. Prior to applying the hydrostatic load, a trench was excavated on the inside of the test section next to the sheet pile and a layer of plastic sheeting was placed in the trench and rolled up and over the sheet pile. The trench was then backfilled so that water was prohibited from entering the gap should it develop. Nevertheless, the mode of failure precipitated by formation of the water-filled gap was unfortunately overlooked by the Corps during its design process. In hindsight, there is nothing wrong with using an I-wall to raise a levee, but a water-filled gap should be assumed in the design.

Figure 15. Wall deflection vs. depth of water from the Corps’ E 99 report.

Risk Eric Holdeman, former director of emergency management for King County, Washington, says there are four stages of denial (Ripley and Kluger 2006):

1. It won’t happen. 2. If it happens, it won’t happen to me. 3. If it happens, and it happens to me, it won’t be that bad. 4. If it happens to me and it’s bad, there is nothing I can do to stop it anyway.

Many of the residents of New Orleans proved Holdeman correct during Katrina as they were unwilling or unable to evacuate before the storm’s arrival. The risk to the citizens of New Orleans was underestimated and misunderstood. Figure 16 is a log-log plot showing the annual probability of failure vs. both fatalities caused and dollars lost. The diagonal lines represent commonly accepted limits of risk

tolerance. Events that plot below and to the left of the solid green line are generally considered acceptable risks by society. On the other hand, events that plot above and to the right of the dashed red line are generally considered unacceptable to society. Events that plot between the two lines are subject to individual judgment. For example, it is a widely accepted societal goal to reduce the health risks posed by cancer and heart disease; that is, we encourage efforts of modern medicine to move the risks posed by cancer and heart disease down and to the left. This figure also shows the risks posed by several common occurrences indicating that, for example, the risks posed by bridge scour and plane crashes lie within a range generally considered acceptable to society. Note that the oval identified as “Hurricane protection system” depicts pre-Katrina risks in New Orleans. There are at least two ways to reduce the risk to the public. One is to reduce the likelihood of death, or move the oval to the left, for example by improving evacuation procedures to move people out of harm’s way. The second is to increase the robustness and reliability of the HPS around New Orleans, or move the oval down, for example by building taller and more robust levees. Increasing the reliability of the HPS will likely take a substantial amount of money and time. Getting people out of harm’s way requires effective public policy that motivates people to evacuate in a timely manner.

Figure 16. Annual probability of failure vs. fatalities and dollars lost (courtesy of Jean-Louis Briaud, Texas A&M University)

The oil industry has a number of offshore oil platforms in the Gulf of Mexico, which suffered similar damage in terms of dollars during Katrina as did the New Orleans area (both suffered about $30B in damages). The oil industry, however, evacuated 100% of its people and lost no lives during Katrina. New Orleans evacuated about 80% of its citizens and suffered more than 1,100 fatalities. These data indicate

that if you reduce risk by getting people out of harm’s way there is much less likelihood that they will be injured or killed. Effective evacuation is the key!

Lessons Learned, or Merely Observed? One of the lessons from Hurricane Katrina is that engineers need to think globally and act locally. For example, it is common knowledge that coastal Louisiana is subsiding. In fact, coastal Louisiana is subsiding at a rate of about 6 in. per decade or more. This subsidence and vertical datum adjustments were never considered or accounted for by the Corps. Going forward, we must be sure to account for the impacts of sea level rise caused by climate change. In addition, the HPS was originally designed for the standard project storm, which was the state of practice in the 1960s. The Saffir-Simpson Hurricane scale was developed in about 1969, yet the standard project storm for the New Orleans HPS was never updated despite new information on hurricane strength. Another lesson is there was a failure to understand, manage, and communicate risk. As depicted in Figure 16, the risks were seriously underestimated and the engineering designs pushed the envelope at each and every stage. The I-walls were not sufficiently conservative to deal with unknowns. A water-filled gap should always be assumed in the design. The Corps failed to build in quality. There was no rigorous internal review process (QA-QC) in place that would have assured that designs met the project goals. A rigorous QA-QC process could have been effective at: 1) embedding an appropriate margin of safety into the culture of the design process, and 2) ensuring that the designs met the appropriate standards of practice. A fair question to ask is: Have these lessons been actually learned, or have they merely been observed so that we are likely to experience them again in future disasters? The lessons from Katrina were considered by many as a call to action. In 2014, ASCE published the results of its examination of our national response to this call in Flood Risk Management: Call for a National Strategy. This examination concludes that in the nearly ten years since Katrina “while some progress has been made, in general the flood challenge continues to receive scant attention, and much remains to be accomplished to safeguard the well-being of the people and property at risk. If the devastating impacts of [Superstorm] Sandy and the losses sustained in floods and hurricanes since Katrina were used as the measure of progress, the nation has failed to heed the call.” In other words, the authors of Flood Risk Management believe Americans have not yet applied the lessons from Katrina. Are we thinking globally and acting locally? ASCE’s Flood Risk Management notes that climate change and population growth are exacerbating the challenge of effectively managing the risk posed by flooding. FEMA reports that because of these changes, the 100-year floodplain could increase by about 45 percent by the end of this century. If we fail to begin managing this risk today, future generations will inherit a potentially insurmountable challenge. ASCE’s Flood Risk Management sounds an optimistic note in that we are beginning to shift from the paradigm of controlling floods to one of managing risk. This new mindset recognizes that absolute protection against flooding is not possible and that we need to focus on first identifying risks, then on implementing approaches to deal with these risks. We are also becoming more effective in communicating risks to the public, especially in light of recent catastrophic flood events. We cannot, however, rest on our laurels. Too many people still believe that absolute protection can be achieved, and this mindset encourages development in flood-prone areas.

Are we building quality in? Perhaps. The National Committee on Levee safety published a draft report in 2009 entitled, Recommendations for National Levee Safety Program. Among its 20 comprehensive recommendations are “Develop and Adopt National Safety Standards” that promulgate implementing best engineering practices, and “Develop a National Levee Safety Training Program” that will increase the level of expertise and knowledge in all aspects of levee safety. In addition, The International Levee Handbook, published in 2013, aims to be a single reference source on good practice for the management and design of levees. Despite this progress, ASCE’s Flood Risk Management notes that much of our flood infrastructure, including levees, remains in marginal condition and there is no plan in place for systematic improvement. Resources at the local, regional, and federal levels are all squeezed leaving us to rely on flood risk reduction systems that are outdated and lack the quality to ensure dependable performance and public safety.

Conclusions It is the author’s opinion that the following conclusions can be drawn from the failure of the New Orleans HPS during Hurricane Katrina. 1. There was a failure to understand and embrace safety. Engineers need to keep safety at the

forefront of public priorities. In fact Canon 1 of ASCE’s Code of Ethics states “Engineers shall hold paramount the safety, health, and welfare of the public...”

2. Engineers need to quantify the risks, communicate the risks to the public, and help decide how much risk is acceptable. For example, when communicating with the public using phrases like—the chances of seeing a 100-year storm is about 1% per year is not as effective as saying if you have a 30-year mortgage on your home and your home is behind a levee and the levee was designed for a 100-year storm, the likelihood of seeing that storm during your mortgage is about one third. That is about the same as playing Russian roulette with two bullets in a 6-shot revolver.

3. Lastly, we engineers need to continuously upgrade our design procedures and make every

attempt to evaluate all possible failure modes. When appropriate, independent experts need to be involved. Public safety must be our first priority.

Finally, the authors also believe that ignoring the lessons is not an option. ASCE’s Flood Risk Management states: “…failing to heed the lessons we should have learned…will have enormous future consequences.” We stand on the shoulders of giants who learned the lessons from past disasters. If we fail to learn from today’s disasters, we will be doomed to repeat them.

Acknowledgements The authors would like to acknowledge the efforts of the US Army Engineer Research and Development Center (ERDC) in the preparation of this paper. We also acknowledge the Corps’ Interagency Performance Evaluation Task Force; specifically the Levee and Wall Performance Group, which was composed of Reed L. Mosher and James M. Duncan. We also thank Professors Robert B. Gilbert, Peter Nicholson, Jean-Louis Briaud, and James M. Duncan for the use of their photos, figures, and data. Lastly, we need to thank Tracey Waddell of ERDC for her help with the figures.

References 1. Brandon, Thomas L., Stephen G. Wright, and J. Michael Duncan. “Analysis of the Stability of I-walls

with Gaps between the I-wall and the Levee Fill.” Journal of Geotechnical and Geoenvironmental Engineering, ASCE, May 2008.

2. Duncan, J. Michael, Thomas L. Brandon, Stephen G. Wright, and Noah Vroman. “Stability of I-walls in New Orleans during Hurricane Katrina.” Journal of Geotechnical and Geoenvironmental Engineering, ASCE, May 2008.

3. “Flood Risk Management: Call for a National Strategy,” ASCE, 2014.

4. IPET, “Performance Evaluation of the New Orleans and Southeast Louisiana Hurricane Protection System,” Interagency Performance Evaluation Task Force, Final Draft, January 2006.

5. Jackson, R. B. “E-99 Sheet Pile Wall, Field Load Test Report.” Technical Report No. 1, U.S. Army Engineer Division, Lower Mississippi Valley, Vicksburg, MS, 1988.

6. Ripley, Amanda, and Jeffery Kluger. Katrina, One Year Later, Time, Aug. 28, 2006.

7. “Recommendations for National Levee Safety Program.” Published by The National Committee on Levee Safety, draft report, 2009

8. “The Creeping Storm,” Civil Engineering, American Society of Civil Engineers, June 2003.

9. “The International Levee Handbook,” CIRIA, London, UK, 2013.