A parallel evolutionary strategy based simulation–optimization approach for solving groundwater source identification problems

Groundwater characterization involves the resolution of unknown system characteristics from observation data, and is often classified as an inverse problem. Inverse problems are difficult to solve due to natural ill-posedness and computational intractability. Here we adopt the use of a simulation–optimization approach that couples a numerical pollutant-transport simulation model with evolutionary search algorithms for solution of the inverse problem. In this approach, the numerical transport model is solved iteratively during the evolutionary search. This process can be computationally intensive since several hundreds to thousands of forward model evaluations are typically required for solution. Given the potential computational intractability of such a simulation–optimization approach, parallel computation is employed to ease and enable the solution of such problems. In this paper, several variations of a groundwater source identification problem is examined in terms of solution quality and computational performance. The computational experiments were performed on the TeraGrid cluster available at the National Center for Supercomputing Applications. The results demonstrate the performance of the parallel simulation–optimization approach in terms of solution quality and computational performance.     

Quantitative measurement of channel-block hydraulic interactions by experimental saturation of a large, natural, fissured rock mass

The hydrodynamic behavior of fissured media relies on the relationships between a few very conductive fractures (channels) and the remaining low-conductivity fractures and matrix (blocks). We made a quantitative measurement of those relationships and their effect on water drainage and storage in a 19,000 m3 natural reservoir consisting of karstified limestones. This reservoir was artificially saturated with water by closing a water gate on the main outlet during a varying time (delta t) fixed by the operator. The water gate was completely or partly closed until the water pressure reached a particular specified value. If the water gate was left completely closed long enough, the water pressure was fixed by the elevation of temporary outlets at the site boundaries. The water elevation within the reservoir was monitored by means of pressure cells located on single fractures representative of the bedding plane and the two families of fractures of the massif network. The comparison of pressure variations with the network geometry allows us to identify a typical double permeability characterized by a few very conductive channels along 10 vertical faults. These channels limit blocks consisting of low-conductivity bedding planes and a rather impervious matrix. Depending on the closure interval, delta t, of the water gate, the total volume of water stored in the reservoir can vary from 0.8 m3 (delta t = 5 minutes) to 18.6 m3 (delta t = 2 days). Such a variance of storage versus closure time is explained by the reservoir's double permeability that is characterized by blocks that saturate much more slowly than channels. If plotted versus time, this injected volume fits a power relationship, according to the saturation state of the blocks. In less than 34 minutes, storage is close to zero in the blocks and to 1.6 to 2 m3 in the channels. For closing times shorter than 1 hour, only 1% of the volume that flows in the channels is stored into the blocks. Depending on the water pressure and for a given delta t = 3000 minutes, the volume of water stored is controlled by the geometry of the joint network and of the aquifer boundaries. Such an experiment shows that the flow is concentrated in about 10% of the fractured network (channels). The water storage that takes place in the 90% remaining fractures (blocks) depends mainly on time during which pressure remains high into the 10% channels. The accurate modeling of such typical double-permeability media needs a careful study of the geometry of the channels whose narrowings modify the flow and induce dam effects that maintain a high pressure gradient with surrounding blocks.     

Site effect on vulnerability of high-rise shear wall buildings under near and far field earthquakes

Worldwide experience repeatedly shows that damages in structures caused by earthquakes are highly dependent on site condition and epicentral distance. In this paper, a 21-storey shear wall-structure built in the 1960s in Hong Kong is selected as an example to investigate these two effects. Under various design earthquake intensities and for various site conditions, the fragility curves or damage probability matrix of such building is quantified in terms of the ductility factor, which is estimated from the ratio of storey yield shear to the inter-storey seismic shear. For high-rise buildings, a higher probability of damage is obtained for a softer site condition, and damage is more severe for far field earthquakes than for near field earthquakes. For earthquake intensity of VIII, the probability of complete collapse (P) increases from 1 to 24% for near field earthquakes and from 1 to 41% for far field earthquakes if the building is moved form a rock site to a site consisting a 80 m thick soft clay. For intensity IX, P increases from 6 to 69% for near field earthquake and from 14 to 79% for far field earthquake if the building is again moved form rock site to soft soil site. Therefore, site effect is very important and not to be neglected. Similar site and epicentral effects should also be expected for other types of high-rise structures.     

Effective stress analyses of liquefaction-induced deformation in river dikes

A series of effective stress analyses is carried out on the seismic performance of river dikes based on the case histories during the 1993 Hokkaido-Nansei-oki and 1995 Hyogoken-Nambu earthquakes in Japan. Seven case histories selected for the analyses involve a crest settlement ranging from none to 2.7 m in the dikes 3–6 m high with evidence of liquefaction at foundation soil. The effective stress model used is based on a multiple shear mechanism and was developed by one of the authors. The soil parameters are evaluated based on the site investigation and laboratory test results. The results of the analyses are basically consistent with the observed performance of the river dikes. In particular, the effective stress model shows a reasonable capability to reproduce the varying degree of settlements depending on the geotechnical conditions of foundation soils beneath the dikes. The analyses also indicate that the effect of a cohesive soil layer mixed with the liquefiable sand layers beneath the dikes can be a primary factor for reducing the liquefaction-induced deformation of dikes.     

Species sensitivity distributions: data and model choice

Species sensitivity distributions (SSDs) are increasingly incorporated into ecological risk assessment procedures. Although these new techniques offer a more transparent approach to risk assessment they demand more and superior quality data. Issues of data quantity and quality are especially important for marine datasets that tend to be smaller (and have fewer standard test methods) when compared with freshwater data. An additional source of uncertainty when using SSDs is appropriate selection from the range of methods used in their construction. We show through examples the influence of data quantity, data quality, and choice of model. We then show how regulatory decisions may be affected by these factors.     

MATHEMATICAL MODELS FOR LIQUID-SOLID TWO-PHASE FLOW

I INTRODUCTIONTurbulence models for single-phase fluid flows have been developed and widely applied in mechanical,aeronautical, environmental and hydraulic engineering, and other fields. The closure techniques for theReynolds-averaged Navier-Stokes equations for various levels of models including the simple turbulencemodel, one-equation turbulence model, k-s turbulence model and turbulence stress/flux model have beenverified to be physically reasonable and have acceptable accuracy in app…     

“Site & region-specific” response spectra: a probabilistic approach

Design seismic forces depend on Peak Ground Acceleration (PGA) values and on the shape of Response Spectrum (RS) curves dictated by Building Codes or which need to be evaluated in every particular case. The PGA values and RS curves strictly depend on earthquake magnitude and distance, as well as on the regional and local geological conditions. At present, there is no doubt that it is necessary to construct so-called “Site & Region-specific” Building Code provisions reflecting the influence of different magnitude events at different distances that may occur during the life time of the construction, as well as the variety of local ground conditions. A scheme of Uniform Hazard Response Spectra and PGA estimation considering local site response is described in this paper. The assessments of these design parameters are obtained on the basis of Uniform Hazard Fourier spectra using the conception of “dominant earthquakes”. The effect of local geology is included by means of the soil/reference site spectral ratios.     

Straightforward methods to detect non-linear response of the soil. Application to the recordings of the Kobe earthquake (Japan, 1995)

Simple straightforward methods are applied to testtheir ability to detect the non-linear response of thesoil. Recordings of the main shock and aftershocks ofthe 1995, Hyogo-ken Nanbu (Kobe) earthquake are used.Non-linear effects are investigated using twodifferent techniques, on a collection of data for 12sites situated on different geological structures inthe Kobe and Osaka areas. The first method used is theso-called receiver functions technique (Langston,1979), which consists of computing the spectral ratiobetween horizontal and vertical components of motion.This ratio has been shown to reveal the fundamentalfrequency of a site (Lachet and Bard, 1994; Lachet etal., 1996; Theodulidis et al., 1995, 1996). For eachsite, recordings of the main shock and a set ofaftershocks are considered. The variation of thisspectral ratio for different values of the maximumacceleration recorded at a site is investigated. Bothvariations of the amplitude of the H/V ratio (due tonon-linear behavior, on the horizontal components inparticular) and of the frequency position of theamplified band-width are observed. The secondtechnique used in this study is related to thevariation of the high frequency content of therecordings during the main-shock and its aftershocks.The high frequency spectral decay of the motion,characterized by parameter, is assumed to berelated mainly to the near-surface attenuation. Itshould then increase with increasing peak velocity, incase of non-linearity. The value of kappa iscalculated for the 12 sites in the Kobe area, fordifferent types of soil conditions, and againdifferent values of peak ground acceleration.Variations of kappa are then related to non-linearbehavior of the soil during the Kobe earthquake.     

Tsunamis Generated by Subaquatic Volcanic Explosions: Unique Data from 1996 Eruption in Karymskoye Lake, Kamchatka, Russia

—The 1996 subaquatic explosive eruption near the northern shore of Karymskoye Lake in Kamchatka, Russia, generated multiple tsunamis. We document the explosive process that produced the tsunamis, and describe the tsunami effects and runup around the 4-km diameter lake. These data enable the determination of an attenuation relation of runup (wave) height for these “explosive” tsunamis, which is compared with theoretical models of wave height distributions. For the proximal zone, involving radial distances (r) up to 1.3 km from the source, the runup height (R) shows rapid attenuation (from > 30 m to 8 m) with distance as log R = ?1.98 log[r] + 2.6. For the distal zone, r > 1.3 km, involving mainly wave travel southeastwards along the body of the lake away from the explosion source, R decays more slowly (from 8 m to 3 m) as log R = ?0.56 log[r] + 1.9. Rapid decay in the proximal zone suggests that near the source of the explosion, the tsunami propagated radially as a collapsing wave (bore) with discontinuous change in height. The break-in-slope of the runup plot at 1.3 km suggests that beyond this distance the tsunami propagated approximately as a decaying one-dimensional wave in a channel of approximately constant width.