Followers

Sunday, June 22, 2008

SimCity for Big Boys: Preparing for the Katrina of earthquakes

Feature -Have you ever wondered what could have been, if New Orleans and the federal government had been prepared for a devastating Hurricane such as Katrina? How many lives, structures and businesses could have been saved? Where do you put your money to build up a region’s defense against the next natural disaster? These are the questions scientists from the Mid-America Earthquake Center (MAE) are trying to answer in order to prepare local, regional and federal governments for a destructive earthquake that will hit the Midwest sooner or later.

Chasing news from the desk sometimes can distract you from what is going on in the real world and while I am personally not the guy who likes waiting at airports to jump on a crowded plane and fly coast-to-coast to visit press events (luckily we have other editors here at TG Daily who like to do that), an opportunity to leave the office for half a day to see something new is a welcome change.

Such an opportunity recently came along when Trish Barker, PR representative for the National Center of Supercomputing Applications (NCSA) at the University of Illinois at Urbana-Champaign, invited me to visit the MAEviz team at the Mid-America Earthquake Center (MAE). MAEviz sounded about as familiar to me as it might to you right now (I had no idea what it was), but some research revealed that these guys are involved in predicting earthquake damage and that indication alone was interesting enough to justify the three hour drive down to the NCSA and learn how advanced earthquake research and damage prediction is these days.

The researchers were nice enough to walk me through their projects over several hours and patiently answer my questions and the more information I absorbed and wrote down, the more I wondered how this fascinating and complex work could be combined into an article. Close to the end of the day, I felt convinced that the prediction of earthquake damage is, in its very basics, very similar to playing SimCity. If you are in your late 20s or somewhere in your 30s and had access to a PC early on, I am sure you are familiar with this title (and if not, check Abandonware sites to download the game or play an online version here): You build a city, strategically place city services, balance a tight budget, come up with a structure of residential, business and industry areas, fine-tune the infrastructure over time in order to build a strong economy and prepare for disasters such as floods, fires or earthquakes (not to mention SimCity’s monsters).

MAEviz could be considered a very sophisticated and scientific version of SimCity. However, here you are not building a virtual city, but you are working with real cities, real streets, real businesses, real resources and also real lives. And it may reveal vulnerabilities of U.S. cities you may not expect, not like and not hear about from your government – for understandable reasons: At the very least, such data may give terrorists an easy map where U.S. cities are most vulnerable.


Construction and destruction

Much of the earthquake damage prediction is done through computer simulation, but simulation needs real-world data in order to be credible mature over time. The MAE, led by Amr Elnashai, is headquartered at the University of Illinois and is spread out over the University of Michigan, Washington University, University of Memphis, Georgia Tech, Texas A&M, University of Texas at Austin and the University of Puerto Rico, pulls data from a variety of sources, including its own engineering efforts.

One of the numerous NCSA buildings in Urbana hides a huge construction hall that is actually used to construct or reconstruct actual concrete and steel structures to test their seismic vulnerability. An enormous, three story high concrete/steel wall (six feet of the structure are anchored into the ground) is used as the front end of engineering tests. It holds a movable steel structure weighing “several hundreds of tons” that can put structures such as reconstructed bridge peers under enormous horizontal and vertical stress. While it was unclear just how powerful this apparatus is, I was told that the setup can hit structures with about a million pounds.
It is hard to comprehend how much force that is, but it is definitely enough to reveal the limits of any concrete structure.

Image

The process in itself is provided by the NCSA as a service to researchers who received the funds to run such a test (typically $50,000 for an average test run that includes construction (of one peer for about $17,000) as well as destruction and will yield several gigabytes of test data). The data set resulting from such a test belongs to a researcher for a time frame of about six months. After that, the Network for Earthquake Engineering Simulation (NEES) will request the data for archival purposes.

I was told that the facility also accepts requests from outside parties such as governments or companies. In all sincerity of the matter, walking through the construction hall gives the inexperienced visitor a sense of a huge playground for earthquake researchers – a dream come true for anyone who likes to build elaborate sandcastles and later level them to the ground. NCSA representatives said that there is currently a testing backlog of a few months for the facility. Once a test is approved and scheduled, it takes about four weeks to set up a test.

Image
Simulated impact on a peer

There is an enormous complexity behind those tests with numerous variables impacting an actual damage to a structure during an earthquake. Depending on the soil, the angle and the strength an earthquake hits a structure, the outcome may be very different. The MAE team says that there simply isn’t enough data to reliably simulate an impact of an earthquake on a bridge through a computer, which is why there are actual physical tests. However, researchers do simulate bridge spans since they do not “travel” during an earthquake. If a bridge collapses during earthquake, it will not be because of the span, but because of the peers, I was told.

There are many fascinating tests going on and I have to admit that I was surprised to learn about the dangers of a possible earthquake in the Midwest, which is apparently just a matter of time and could be much more devastating than an earthquake caused by the St. Andreas fault. The New Madrid fault was the origin in an 1812 quake, which was felt all the way to Washington DC. It is expected that the fault will cause an earthquake very 200 to 300 years, which means that the risk for the Midwest is already substantial. It is worrisome worrying is the fact that there have been very few earthquake preparations in the Midwest and the soft soil along the Mississippi river, whose water has penetrated the ground of nearby to a depth of about 1000 m (3000 ft), provides a perfect scenario for a very destructive quake. According to the MAE scientists, a strong New Madrid quake could travel far, causing skyscrapers to shake and causing windows to fall out of buildings as far as Chicago.

Image
Simulated impact on a wall
Closer areas, such as Memphis or St. Louis could be affected in a much more serious way, we heard, and facing what is now commonly described as the Katrina of earthquakes. Most of us are very aware of Katrina, which has become a synonym for natural disasters that should be taken serious. At the very least, Katrina has woken up people and government agencies that there are real natural threats we should be prepared for.

Apparently, the danger of such a quake is known among scientist and even government agencies, but what exactly is done about? Well, Memphis simply should not have been built where it has been built in the first place, we were told. But correcting that mistake and leveling the city to the ground, or relocating it a few hundred miles away, is not a realistic option. So scientists are trying to find ways to limit the potential impact of an earthquake through simulation: Governments can use the findings and data to reorganize a city structure and limit damage to structures and the economy as well as loss of life.

Simulating the worst case scenario

Hazus has been the standard tool for predicting and estimating earthquake damage for the past 10 years. MAEviz is coming in as a new and more modern solution, but MAE scientists stressed that it is not a replacement for Hazus, but a solution that is intended to be used in parallel. It is based on more recent findings and more importantly, it is open source, which means that scientists around the world can contribute and extend the code. The MAEviz team, which is led by Jim Myers at NCSA, will celebrate the software’s second birthday this August.

Image
MAEviz interface

MAEviz is a data-centric application and since it is open source, anyone can download and use it. However, substantial scientific knowledge, and lots of input data is required to come up with useful results and at least I did not find my way through it. However, if you can get the software running and convince it to process simulations, it can deliver stunning results – and you don’t even need a supercomputer: A decent desktop PC is enough to get a basic result within 15 to 20 minutes and more sophisticated data within 60 to 90 minutes. In the example of Memphis, the behavior and impact to 240,000 buildings can be simulated. These simulations were already done and presented to FEMA, which we heard is very interested in the findings, but hasn’t applied them to potential preparations so far. Detailed results have not been released for obvious reasons - such as exposure of vulnerabilities to terrorists – but there was little doubt that, at least in the case of this city, an earthquake poses a substantial risk.

Broken down to its very basics, MAEviz truly is a complex version of SimCity, with the difference that the city already exists, a natural force is simulated and improvements are suggested by scientists to the local, regional or national government as a result. It is basically a risk assessment and analysis tool - the most sophisticated we have seen so far – that enables scientists to go through What-if scenarios: Data input includes hazards, inventory (buildings and other structures, population, etc.) and fragility models. The output consists of damage prediction and actual decision support.

Image
MAEviz damage forecast
The analysis features cover impacts on buildings, bridges, potentially developing hazards, likely casualties and socioeconomic effects. In fact there is a network of more than 40 analysis tools that allows scientists far beyond the direct damage of an earthquake, but estimate secondary impacts such as likely repair costs, revenues lost by businesses or street damage. As a result, MAEviz supplies information that can highlight recommended changes to city structures, relocation of buildings and a strategic placement of first responder services, possible evacuation routes or even possible safe locations of tent cities. These results are always tied to the estimated cost and answer questions like: What if I had $10 million? Where should I invest these funds to achieve the best possible result in damage protection in the case of an earthquake? Is retrofitting of building the answers or should I build new? It is generally assumed that $1 spent on possible improvements can save $5 in response cost, so MAEviz is a tool that can save governments millions and, in extreme cases, billions of dollars in the case of a devastating natural disaster.


Data preservation

I’d like to also touch a related topic Jim Myers and his team appeared to be very concerned of. Tools like MAEviz rely on data – lots of data that is accumulated over a long time frame. Storing and preserving data as well as making sure that data remains accessible for 50 or 100 years is a key problem for the scientists.
NCSA has data available that is about 20 years old and made sure that the formats the data is stored in are well documented. Microsoft’s approach to keep a tight leash on its formats and keep them proprietary is considered by researchers as a “time bomb”. Even if it is unlikely that Microsoft is going away anytime soon, the question is – will they still be around in 50 years and will there be enough documentation to read data from today’s files? And if not, can the technology be reverse engineered?

Especially projects like MAEviz highlight that there is a need for a common, open data format. The value could tremendous and the possible impact of data loss just as devastating as natural disasters.


Conclusion

I have said it before and I will say it again. We are often just too consumed with our own little world around us to notice how technology is changing and improving our life. In my case that has been Fermilab and the NCSA supercomputer systems before and MAE is just another example. I highly recommend to check out the projects that may be going on in your area and find out whether you can learn more about these projects on site. In many cases, you will find visitor centers and scientists who are willing to share details about their work.

MAEviz, of course, is just one project at NCSA (which, by the way where Marc Andreesen invented the web browser) that will make you think for more than 15 minutes after you have left the building.

I personally was impressed with the sophistication of earthquake damage prediction and the advances scientists have made in this area. However, you can always look at it from different sides and if you are an emotional person, you may be disturbed about the fact that a software simulates the environment you live in and local governments makes “efficient” decisions as a result. Whether we like to hear it or not, the questions asked and answers provided by MAEviz also include human lives: Applying recommendations made by the software will impact human life and may decide which earthquake improvements will be made in which regions.

For the virtual world of SimCity, MAEviz would be the perfect tool to build the perfect city, protect human life and limit earthquake damage to a minimum.

In reality, MAEviz confronts governments with choices. Choices that always appear to represent compromises between limited funds and the best possible protection for a certain geographic area.
Original here

No comments: