In general, fragility curve is plotted probability of certain damage state of an or a group of Element of Risk Vs hazard intensity. In case of flood as you want to know, assume a town near the flood plain which has a probability of damage (buildings) is 50% for a 2 meter depth flood (you can also consider velocity as a flood hazard intensity, or depth+velocity+duration of flood). You can easily plot fragility curve. You can also consider absolute damage cost as a probability of damage state.
visit following web pages or simply google for further reading. I hope this will help you.
http://www.nfrmp.us/IFRMA/docs/pre/summary/SimmPaper.pdf
http://web.sbe.hw.ac.uk/frmrc/downloads/asset_mangement_wallingford/4-Simm-fragility-for-FRMRC-workshop-cut-down.pdf
As of to how to create them, there are 4 possibilities:
1. A detailed damage assesment on flooded areas, where hydraulic variables, damage rate are surveyes along with any nummber of discriminating variables, that allow you to generate curves for specific asset types.
2. A conceptual development for specific assets types. You make a sketch of your asset, for example a two story house and you place a typical household inside. Then you proceed with a healthy rationality, answering the question: What percentage of the household is damaged when the water reaches a number of depths? Of course you may use simple asumptions, for example lets say you have a television, you may assume that as soon as the water reaches it, the TV is totally lost. I have given you an example for a house, you may also use it on other assets.
In both of the above cases, water depth is usually used because it is both simple to grasp and observe, but in many cases it would be better to use also water velocity and in the case of agricultural areas, the duration of the flood conditions.
3. At least ins theory you may do physically based numerical modelling, but this is in many cases not practical (time consuming, expensive). Maybe just for a very important asset (nuclear plant for example).
4. You may transfer functions obtained by any of these methods to your study area. This would be some sort of regionalization approach to vulnerability.
If you do not have detailed damage data available I would advice you to proceed with methods No. 2. Methods No.4 is tempting (because it is easy), but it is also a good way to spread an infection. I mean this were developed under given conditions, that may not apply to your study area and a very big problem is that you rarely find a full description on how they were created.
Hope it helps.
what i have to do is to assess the flood hazard using hydro logical modeling techniques and then assess the vulnerability using various spatial and temporal analysis after that i will have to create the fragility curves to assess the capacity of the buildings in the surroundings of floodplain to with stand the hazard . now i m confuse about the assessment of probability of failure of buildings ( in the fragility curves creation) as the inundation depths increase. should i assess it using recent hazard data or should i assess it using expert knowledge that this building will react like this in case of any event occurrence.
The fragility curves cannot be determined on the basis of hazard alone, you need some way to relate the hazard variable, lets say water depth, with the "reaction" of the building. Either you use damage data from recent events for which you also know the water depth or you do expert knowledge assesment.
The second option is in deed not the best choice, because you may ask how do I know how the building will react (for the household it is more straight foreward), but it is usually the only practicable approach when you do not have damage data.
Yes, but it will depend on how detailed is your information. For any given class of buildings you have to proceed along this way. But be carefull I am assuming the data that you have is total damage expressed as percentage of the value of the building.
- Normalize your damage data, for example according to the area of the buildigs.
- Select a given class of buildings according for example to material and use. These criteria represent a proxy for describing vulnerability, but you can choose others.
- For each class arrange them according to water level and plot your data:
Water Level vs. Normalized Damage.
Ideally you would become a clear positive relation between the variables with little dispersion, and fit any curve visually or math function to your data and this would be your first version of the vulnerability curve.
But it is very likely that you will get a fairly ammount of dispersion on your data, and even maybe no positive relation between the two variables. Then it is your job to assess what other vulnerability descriptors would be better, or if maybe a multivariate approach would work better (then you will get a family of curves for each building class) and explore your possibilities until you get some satisfying result. Be aware though that usually these sort of data presents a very large variability. It is therefore very important that you also include confidence intervals.
Hope it helps.
Dear Rafael,
Do you have any published guideline for development of fragility curve?