Hi Mr Ahmad,

What do you mean by fail in this context? We did validation of thin piping systems attached to the main larger, and the calculation programs overestimated the respons at small Earthquake given by respons spectra, so it was not possible to use. Instead other schemes for construction gave a more accurate model. Larger Earthquakes may need an entirely different approach. 

Hi Lena,

I am happy for your interest to answer my question.

Seismic hazard maps often fail to predict the location and magnitude of future earthquake. These maps provide for Design Engineers with ground motion parameters, if the hazard map output is incorrect, then our calculation is completely wrong. All of our infrastructures should stand against small earthquakes. Our main goal, how to make our buildings resist high seismic load.  

For example:

Japanese seismic hazard map shows that M 9 earthquakes would not occur on the Japan Trench off Tohoku. The largest future earthquakes along different segments of the trench there were expected to have magnitude between 7 and 8 only . The model that used in seismic hazard map assumed that different segments of the trench would not break simultaneously. However, the March 2011 earthquake broke five segments, yielding a magnitude 9 earthquake. This giant an M 9 earthquake

involves a larger average slip over a larger fault area, resulting in a larger tsunami because the maximum tsunami run-up height is typically about twice the fault slip.

The scientists never expect this to happen. As you may know that higher earthquake magnitude leads to higher shear load to be considered in desingination plan.

To answer this question in more details, one should look on the various techniques being used to produce seismic hazard maps all over the world and test these maps after each strong earthquake.

Thank you again you again for your concern.

I heard a lecturer from Latvia after the Tsunami, and he explained the wave dynamics with that framework. It was complex, and I don't remember how/if he related to the

shear at the Earthquake. There must be probability, because the waves are more rare? But a probability could be provided, notifying about the risk, at certain places, and I think there are, together with escape routes.

For small Earthquakes, it is of course also important with design to endure, especially if they are frequent. Normal operation are validated for some other dynamical loads (water hammer, flow oscillation, walve openings), and in many systems, the calculations are the same as for small Earthquakes, but the input is different. 

A bit deviating from the subject: Thermal expansion is the most severe in piping, an therefore it can not be tightly attached, but flexibility is required, and therefore the dynamics give too large, and rapid deflections.

Reliability of seismic hazard results depends on the quality of the used input parameters. That is why seismic hazard map of a given area is just reflecting the state of knowledge at this area.   

Dear Bouhadad,

The quality of input parameters and their accuracy is very important but it is not enough to create trustful seismic hazard map. Note that Japan is one of the most active area seismically, and the collected input parameters are quite accurate but in the example i have given above, the Japanese seismic hazard fail to predict March 2011 earthquake. This results show that our used models should be improved.

I recommend the my recently published article for readers:

http://www.sciencedirect.com/science/article/pii/S2352938516300520

Dear Ahmad,

The irreducible random uncertainties  often didn't affect the reliability of seismic hazard results. What is affecting the reliability is the scientific uncertainties (state of earth sciences knowledge which it was the case even in Japan (Tohoko) 2011.

In general, such modeling  would be of interest for other material as well, where a few cycle cause 'resonance' of a rate dependent material, but the uncertainty is of course always large at such predictions

The main reasons for poor quality of the "hazard maps" are:

1. There is no universal mathematical model describing formation of earthquakes

2. Available data is rather limited

3. There is no common agreement concerning which quantities could be used as predictors

Significant improvements of the "prediction of earthquakes" could be obtained after more extensive application of the recent technologies. Personally I'm inclined to believe that more attention should be paid to the measurements of a magnetic field

http://www.sciencemag.org/news/2014/06/could-pulses-earths-magnetic-field-forecast-earthquakes

The additional monitoring of ozone in the areas of seismic activity is also very advisable

http://aip.scitation.org/doi/abs/10.1063/1.3660763?journalCode=apl

The new promising source of data is monitoring of the time dependent geoid (using atomic clocks)

https://arxiv.org/pdf/1506.02853.pdf

In summary with the constantly increasing number of tools I expect some improvements of the quality of the "hazard maps".

Dear Janusz,

Thank you for your answer, actually earthquake prediction is totally another issue to discuss. my initial question was related to preparation of seismic hazard map. I know that concentration of carbon dioxide Co2 in the atmosphere can be monitored by satellite data and its anomalies was used as indicator for prediction of large earthquake. This is done 10 days before occurring of Gujrat 2001 earthquake in India. Related to magnetic field anomalies, i belief electromagnetic field will show real anomalies before occurring of an earthquake. Actually, seismologists focus on seismicity and earthquake records analysis. Scientists should have different backgrounds to be able to create a new prediction earthquake model. Seismic hazard map shows the seismic intensity level variations and corresponded peak ground acceleration values. These values are much more important for seismic design Engineers.

Thank you again for your sharing us your valuable knowledge. 

I must say that I agree with Y. Bouhadad and may add, the subject of seismic hazard is not an exact science and it will never be. I’m almost convinced that earthquakes generally behave as if everyone of them have its own “personality”. Keeping that in mind I can say now that you have all kinds of relative qualities in this kind of studies, However the best of them will remain speculative at best, Deep knowledge of geology and tectonic regime, stat of stress, physical behavior of the seismic waves and their propagation, attenuation, reflection, refraction, diffraction, isotropy or anisotropy and to what degree, in the variable conditions on their paths. This in only a part of the highly volatile and almost beyond control conditions. The various effects of an earthquake on a given site is in fact a solution to a huge inverse problem admitting an infinity of correct mathematical solutions the true physical solution is only one particular solution from that infinity ! Computers, programs and programmers are fairly good in providing mathematical solutions, they cannot provide the correct choice of the physical unique solution (the true effect). This is why the seismic hazard maps “often” fail. So a seismic hazard map is a mathematical solution of a problem in which most parameters are assumed and the rest are generally poorly constrained or used.

Of course earthquake models do fail because of several factors. First factor could be the identification of seismogenic sources and it may be possible that a fault is hidden often called as blind fault. Second, could  be estimation of maximum potential earthquake (already an example has been given regarding Fukusima Tsunami in Japan). Thrd could be epistemic uncertainties, and 4th could be the return period of probabilistic hazard maps. Most often, probabilistic hazard maps  provide mean/ median values and not the peak value of ground motion for a specific return period design earthquake. For example, earthquake hazard maps of 10% and 2% probability  of exceedance. 

Dear Gupta,

What about our attenuation models? Are they good predictor?

As all other kinds of input data used for seismic hazard analysis, the attenuation models affect the results of seismic hazard analysis. Attenuation models are empirical relationships developed on the basis of strong motion data recorded during earthquakes and are continually improved and updated. So  the available attenuation models represent only the state of knowledge in this domain.  

Dear Ahmad,

To add further to my answer, yes, attenuation models have almost matured and if you take their uncertainties into account, I believe they are good predictor. Remember, we are estimating mean or median values or M plus sigma or 2 sigma values and not the peak values from these attenuation relations. So, there is always a difference in peak parameter (acceleration, velocity, or displacement) and the mean or median values that one has predicted from attenuation models. 

Dear Gupta,

I don't agree with you! Attenuation model often fail. Intensity  distribution is nonlinear behavior and can't be equal on both sides of seismic source or earthquake epicenter. Nepal 2015 earthquake is one example.

Dear Ahmad,

Intensity is a measure of felt effect and involves many more parameters related to structural and non-structural and their design and construction quality. So, please do not get confused with Intensity and ground motion estimates. It is simple intensity attenuation regression model that often fail because of their simplistic low number of parameters and NOT the GMPEs.

I hope, it is clear to you. 

Very interesting and highly relevant question. Happy to note that such a high degree of intellectual discussion has taken place on the subject. I was just comparing this problem with that of landslides. The hazard zonation map is prepared based on many parameters which initiate & accelerate the natural hazard and also on the past records, as all of us know that "Past is the key to the present & future". However any failure according to me is due to lack of complete geologic knowledge of the terrain and proper representation of the weightage of different driving forces in the modelling.