Zhongdong Duan

Finite element model updating of structures using a hybrid optimization technique

Finite element model updating of structures usually results in a nonlinear optimization problem. Finding an optimization technique with high efficiency is one of the key issues for model updating. A hybrid optimization technique is proposed in this paper, which draws together the global searching capability of the chaos-based optimization technique and high searching efficiency of the trust-region optimization method. The hybrid approach is demonstrated to be more efficient and prone to obtain a global minimum as compared to conventional methods using a two dimensional test function. Then this hybrid method is employed to update a 14-bay frame model. An optimization problem for model updating using modal frequencies and modal shapes is formulated. Studies using numerically simulated data and experimental data show that the proposed hybrid optimization technique is very promising for structural model updating.

Structural Seismic Performance Evaluation in Consideration of Earthquake Ground Motion Uncertainties Using Convex Set Model

When data is insufficient to support a probabilistic assumption, the structural seismic performance evaluation results with probabilistic approach are suspicious. In this paper, the convex set theory, which requires much less information, is employed to model the uncertainties of peak acceleration and frequency characteristics of earthquake ground motions. The bounds of shear forces are first derived with Chinese seismic code, and then a lateral bound load method for Push-over analysis is conducted. Then the convex analysis is integrated into Push-over analysis to study the bounds of structural capacity. Furthermore, the bounds of earthquake demand are deduced with the bounded convex set model and Chinese seismic code. The bounds of target displacement are obtained using capacity spectrum method. The seismic performance of structure is then evaluated with an interval. Finally, seismic performance evaluation of a three-storey RC building is given to demonstrate the benefits of the proposed method.

Bound Direct Displacement-Based Seismic Design of Buildings

Limitations of probabilistic model appear when information is insufficient. Convex set model, which much less information is needed, is employed to model uncertainties of target displacement, peak acceleration and predominant period of ground motion. A new bound direct displacement-based seismic design method is proposed. The interval of seismic design results is obtained by considering the uncertainties of design variables and the relative risk of any design scheme within the interval is also simultaneously provided. The results of a four-storey RC building show that the base shear and beam sectional area by Xues design scheme are close to the lower bound of the bound design method, and the column sectional area and the weight of steel bar are just over the middle value to the upper bound. The sensitivity analysis indicates that the base shear is more sensitive to target displacement.

Convex Model for Earthquake Damage and Loss Estimation of Individual Buildings

Probabilistic results drawn upon inadequate information are suspicious. The convex set theory, which requires much less information, is employed to model the uncertainties of the spectral displacement and damage state medians. Furthermore, a convex model of fragility function is established based on the Envelope Bound Convex Models of the spectral displacement and damage state medians. A bound loss estimation method is derived by integrating HAZUS-AEBM module with the convex set theory. The loss bounds of a hotel in southern China are obtained and compared to the loss calculated by HAZUS-AEBM method, which locates in the lower half interval of convex analysis results. The uncertainty propagation is analyzed and damage state medians are found to be most critical factor to the loss. Finally, the PEERs probabilistic loss estimation methodology is also applied to this example to deduce the probability of loss exceeding the bound values of convex analysis results.

Structural damage detection by wavelet transform and probabilistic neural network

Artificial Neural Networks (ANNs) have been applied in structural damage detection as a classifier, but generally a capable ANNs has to be trained with a certain amount of samples. When both damage locations and damage extents are to be identified, the amount of training samples is tremendous because of the combinations of damage locations and extents. By wavelet transform of the structure free motion equations, the Residual Wavelet Coefficient Vector (RWCV) is deduced. A damage feature parameter is defined as the ratio between RWCVs in two different frequency bands. This parameter has a unique property that its sensitive only to damage locations, and is independent of damage extents. The damage feature parameters are then fed to the neural network for damage localization. After the damage sites are detected, the damage extent is further identified by another neural network with RWCVs as inputs. This two-phase approach for damage localization and extent identification can simply the neural network and reduce the training samples tremendously. Finally a numerical example is given for damage detection of a 10 DOFs system using the proposed approach.

Suction-based active aerodynamic control of wind loads on super high-rise buildings

This article introduces a new method, active flow field control through steady suction, to reduce wind loads on buildings. To test its effect, suction equipment was developed and installed on a Commonwealth Advisory Aeronautical Research Council super high-rise model. A series of experiments were then conducted in a wind tunnel. The effectiveness of this method for reducing wind loads is discussed through comparison of the suction bottom moment ratio under different wind incidence angles and different positions of steady suction. The results of the experiment demonstrate that steady suction reduces wind loads effectively.

Convex Set and Probabilistic Model-Based Hybrid Earthquake Loss Assessment Method

In order to make the most of the advantages of probabilistic model and non-probabilistic model, a hybrid earthquake loss assessment method based on convex model and probabilistic model was proposed. The exceeding probability of earthquake loss bounds within convex model was derived based on the probabilistic earthquake loss assessment model proposed by PEER. Moreover, the uncertainties of repair cost and collapse probability in the complete damage state were modeled by the envelope bound convex model, and the bounds of exceeding probability of earthquake loss were obtained by the hybrid earthquake loss model. The loss results of a hotel indicate that non-collapse parts mostly contribute to economic loss, and earthquake loss is more sensitive to repair cost than collapse probability. The presentation form of earthquake loss results is enhanced by the hybrid loss method and is also easier for owners and decision-makers to understand and accept.

Convex model for gross domestic product-based dynamic earthquake loss assessment method

Ground motion intensity was presented as Modified Mercalli Intensity in gross domestic product (GDP)-based earthquake loss assessment method, and the variable uncertainties and the variation of GDP with time were neglected. In this study, the hazard-loss/GDP relation was determined based on peak ground acceleration (PGA), and the uncertainties of PGA and the hazard-loss/GDP relation were modeled with the Envelope Bound Convex Model and the curves of GDP with time were also derived. Finally, the earthquake loss of Ningbo city was estimated and disaggregated,and sensitivity of uncertain variables was investigated. The results show that the loss with Chen method locates in the lower half intervals of convex analysis results, and the most loss is from frequent earthquake. Hazard-Loss/GDP relation gives the most contribution to loss uncertainty, and the reminder sequence of sensitivity is the annual occurrence rate v, the b value and the upper bound magnitude Mu.