Prakash Kripakaran

Multimodel Structural Performance Monitoring

Measurements from load tests may lead to numerical models that better reflect structural behavior. This kind of system identification is not straightforward due to important uncertainties in measurement and models. Moreover, since system identification is an inverse engineering task, many models may fit measured behavior. Traditional model updating methods may not provide the correct behavioral model due to uncertainty and parameter compensation. In this paper, a multimodel approach that explicitly incorporates uncertainties and modeling assumptions is described. The approach samples thousands of models starting from a general parametrized finite-element model. The population of selected candidate models may be used to understand and predict behavior, thereby improving structural management decision making. This approach is applied to measurements from structural performance monitoring of the Langensand Bridge in Lucerne, Switzerland. Predictions from the set of candidate models are homogenous and show an average discrepancy of 4-7% from the displacement measurements. The tests demonstrate the applicability of the multimodel approach for the structural identification and performance monitoring of real structures. The multimodel approach reveals that the Langensand Bridge has a reserve capacity of 30% with respect to serviceability requirements.

Configuring and enhancing measurement systems for damage identification

Engineers often decide to measure structures upon signs of damage to determine its extent and its location. Measurement locations, sensor types and numbers of sensors are selected based on judgment and experience. Rational and systematic methods for evaluating structural performance can help make better decisions. This paper proposes strategies for supporting two measurement tasks related to structural health monitoring - (1) installing an initial measurement system and (2) enhancing measurement systems for subsequent measurements once data interpretation has occurred. The strategies are based on previous research into system identification using multiple models. A global optimization approach is used to design the initial measurement system. Then a greedy strategy is used to select measurement locations with maximum entropy among candidate model predictions. Two bridges are used to illustrate the proposed methodology. First, a railway truss bridge in Zangenberg, Germany, is examined. For illustration purposes, the model space is reduced by assuming only a few types of possible damage in the truss bridge. The approach is then applied to the Schwandbach bridge in Switzerland, where a broad set of damage scenarios is evaluated. For the truss bridge, the approach correctly identifies the damage that represents the behaviour of the structure. For the Schwandbach bridge, the approach is able to significantly reduce the number of candidate models. Values of candidate model parameters are also useful for planning inspection and eventual repair.

Design of tensegrity structures using parametric analysis and stochastic search

Tensegrity structures are lightweight structures composed of cables in tension and struts in compression. Since tensegrity systems exhibit geometrically nonlinear behavior, finding optimal structural designs is difficult. This paper focuses on the use of stochastic search for the design of tensegrity systems. A pedestrian bridge made of square hollow-rope tensegrity ring modules is studied. Two design methods are compared in this paper. Both methods aim to find the minimal cost solution. The first method approximates current practice in design offices. More specifically, parametric analysis that is similar to a gradient-based optimization is used to identify good designs. Parametric studies are executed for each system parameter in order to identify its influence on response. The second method uses a stochastic search strategy called probabilistic global search Lausanne. Both methods provide feasible configurations that meet civil engineering criteria of safety and serviceability. Parametric studies also help in defining search parameters such as appropriate penalty costs to enforce constraints while optimizing using stochastic search. Traditional design methods are useful to gain an understanding of structural behavior. However, due to the many local minima in the solution space, stochastic search strategies find better solutions than parametric studies.

Support vector regression for anomaly detection from measurement histories

This research focuses on the analysis of measurements from distributed sensing of structures. The premise is that ambient temperature variations, and hence the temperature distribution across the structure, have a strong correlation with structural response and that this relationship could be exploited for anomaly detection. Specifically, this research first investigates whether support vector regression (SVR) models could be trained to capture the relationship between distributed temperature and response measurements and subsequently, if these models could be employed in an approach for anomaly detection. The study develops a methodology to generate SVR models that predict the thermal response of bridges from distributed temperature measurements, and evaluates its performance on measurement histories simulated using numerical models of a bridge girder. The potential use of these SVR models for damage detection is then studied by comparing their strain predictions with measurements collected from simulations of the bridge girder in damaged condition. Results show that SVR models that predict structural response from distributed temperature measurements could form the basis for a reliable anomaly detection methodology.

Feature selection using stochastic search: An application to system identification

System identification using multiple-model strategies may involve thousands of models with several parameters. However, only a few models are close to the correct model. A key task involves finding which parameters are important for explaining candidate models. The application of feature selection to system identification is studied in this paper. A new feature selection algorithm is proposed. It is based on the wrapper approach and combines two algorithms. The search is performed using stochastic sampling and the classification uses a support vector machine strategy. This approach is found to be better than genetic algorithm-based strategies for feature selection on several benchmark data sets. Applied to system identification, the algorithm supports subsequent decision making.

Optimal Sensor Placement for Damage Detection: Role of Global Search

Optimal sensor placement is one that maximizes the likelihood of identifying future damage models. Based on assumptions from engineers, damage models of a structure are simulated and their predictions are computed. Computational approaches are used to place sensors at locations that maximize the chances of identifying damage. This paper studies the application of global search for optimal sensor placement. The global search methodology uses stochastic sampling to find optimal locations for sensors. In a previous study, Robert-Nicoud et al. proposed a greedy strategy that places sensors sequentially at locations where model predictions have maximum entropy. Performance of the two strategies are compared for the Schwandbach bridge in Switzerland. The results show that global search is better for designing measurement systems on a previously unmonitored structure while the greedy algorithm is better for incremental measurement- interpretation strategies.

MGA – a mathematical approach to generate design alternatives

Optimization methods are typically proposed to find a single solution that is optimal with respect to the modeled objectives and costs. In practice, however, this solution is not the best suited for design as mathematical models seldom include all the costs and objectives. This paper presents a technique – Modeling to Generate Alternatives (MGA), that instead uses optimization to generate good design alternatives, which the designer may explore with respect to the unmodeled factors. The generated alternatives are close to the optimal solution in objective space but are distant from it in decision space. An application of this technique to design of moment-resisting steel frames is illustrated.

Long-term structural health monitoring of the Cleddau bridge: evaluation of quasi-static temperature effects on bearing movements

Abstract This paper illustrates how long-term measurements can be analysed to understand bridge behaviour under changing environmental conditions and how the developed understanding can help explain the performance of its critical components. Measurements from the Cleddau bridge, a structure that has been continuously monitored for more than two years, are used to investigate thermal effects in steel box-girder bridges and, in particular, their bearings. Observed temperature distributions are very different to the recommended distributions in design codes (BS EN 1991-5: 2003). These temperature distributions create plan bending of the box girder, which in turn impose forces at the bearings that have contributed to its wear. This paper investigates bearing movements of the bridge using numerical models, and estimates the resulting forces at the supports. A physics-based model of the bridge is created to which temperature distributions inferred from in situ measurements are supplied as input. Model predictions are validated against measured deformations at the bearings. Subsequently the model is used to predict forces at the bearings due to plan bending and bearing locking. Results quantify the impact that thermal effects have on the performance of the bearings. They also highlight the significance of considering a range of temperature distribution scenarios that go beyond those given in the design codes in order to reliably evaluate thermal effects at the design stage.

Considering sensor characteristics during measurement-system design for structural system identification

This paper presents a method for measurement-system design through criteria related to model based structural identification. Using a multi-model approach and results from previous research carried out at EPFL, an improved algorithm is proposed. The algorithm accounts for various types of sensors having different accuracies and taking different kinds of measurements. The algorithm selects sensor types and locations that minimise the number of non-identified candidate models. The results show that the approach provides an alternative to selecting and placing sensors using engineering experience alone, and that a scientific approach based on sensor characteristics and modelling error is feasible. A single span composite bridge is used to illustrate the algorithm. It is shown that adding more than 9 sensors, from a possible set of 34, will not provide further useful information for structural identification.

Structural Identification to Improve Bridge Management

This paper presents results from static loads tests performed on the new Langensand Bridge built in Switzerland. A systematic study of over 1000 models subjected to three load cases identifies a set of 11 candidate models through static measurements. Predictions using the set of candidate models are homogenous and show an averaged discrepancy ranging of 4 to 7% from the displacement measurements. All candidate models have values for material proprieties that are close to expected values. This finding confirms that the behaviour of the structure conforms to the design expectations. Comparing the candidate model set to a design model that takes into account only main structural elements shows that the structure has approximately 30% reserve capacity with respect to a typical deflection risk scenario according to Swiss codes. The population of candidate models may be used to understand and predict the behaviour of the full bridge prior to its completion.