Massimiliano Giorgio

An age- and state-dependent Markov model for degradation processes

Many technological units are subjected during their operating life to a gradual deterioration process that progressively degrades their characteristics until a failure occurs. Statisticians and engineers have almost always modeled degradation phenomena using independent increments processes, which imply that the degradation growth depends, at most, on the unit age. Only a few models have been proposed in which the degradation growth is assumed to depend on the current unit state. In many cases, however, both the current age and the current state of a unit can affect the degradation process. As such, this article proposes a degradation model in which the transition probabilities between unit states depend on both the current age and the current degradation level. Two applications based on real data sets are analyzed and discussed.

A competing risk model for the reliability of cylinder liners in marine Diesel engines

In this paper, a competing risk model is proposed to describe the reliability of the cylinder liners of a marine Diesel engine. Cylinder liners presents two dominant failure modes: wear degradation and thermal cracking. The wear process is described through a stochastic process, whereas the failure time due to the thermal cracking is described by the Weibull distribution. The use of the proposed model allows performing goodness-of-fit test and parameters estimation on the basis of both wear and failure data. Moreover, it enables reliability estimates of the state of the liners to be obtained and the hierarchy of the failure mechanisms to be determined for any given age and wear level of the liner. The model has been applied to a real data set: 33 cylinder liners of Sulzer RTA 58 engines, which equip twin ships of the Grimaldi Group. Estimates of the liner reliability and of other quantities of interest under the competing risk model are obtained, as well as the conditional failure probability and mean residual lifetime, given the survival age and the accumulated wear. Furthermore, the model has been used to estimate the probability that a liner fails due to one of the failure modes when both of these modes act.

Very short-term probabilistic wind power forecasting based on Markov chain models

Wind power forecasting methods generally provide estimates of future wind power as point forecasts, but most of the decision making processes in electrical power systems management require more information than a single value. For this purpose, additional methods - complex or based on strong assumptions - have been developed for estimating so-called interval forecasts associated to point forecasts. The method proposed by the authors is based on the use of discrete time Markov chain models of a proper order, developed starting from wind power time series analysis. It allows to directly obtain in an easy way an estimate of the wind power distributions on a very short-term horizon, without requiring restrictive assumptions on wind power probability distribution. With reference to an application, results obtained via a First and Second Order Markov Chain Model, respectively, are compared to those of Persistent Model evaluating the related prediction errors.

A State-Dependent Wear Model With an Application to Marine Engine Cylinder Liners

In this paper a new wear model is proposed in which the transition probabilities between process states, unlike models with independent increments, depend on the current system state. The model is used to describe the wear process of the cylinder liners of some identical heavy-duty diesel engines for marine propulsion. The application is developed on the basis of a real dataset of wear measures obtained via staggered inspections. A time and state space discretization is introduced to obtain the likelihood function of the observed data. The model parameters and reliability characteristics of the liners are then estimated and the wear growth during future inspection intervals is predicted. The homogeneity of wear data and the goodness of fit of the proposed model are tested. A simplified maintenance scenario is also considered to show the need for accurate modeling of the wear process for planning condition-based maintenance activities. Finally, inferential, predictive, and decision-making results derived within the proposed model are compared to those obtained within one of the most widely used age-dependent wear models. Fortran codes and executable programs, as well as the cylinder liner data, are available online as supplemental material.

Real-time risk analysis for hybrid earthquake early warning systems

Earthquake Early Warning Systems (EEWS), based on real-time prediction of ground motion or structural response measures, may play a role in reducing vulnerability and/or exposition of buildings and lifelines. In fact, recently seismologists developed efficient methods for rapid estimation of event features by means of limited information of the P-waves. Then, when an event is occurring, probabilistic distributions of magnitude and source-to-site distance are available and the prediction of the ground motion at the site, conditioned to the seismic network measures, may be performed in analogy with the Probabilistic Seismic Hazard Analysis (PSHA). Consequently the structural performance may be obtained by the Probabilistic Seismic Demand Analysis (PSDA), and used for real-time risk management purposes. However, such prediction is performed in very uncertain conditions which have to be taken into proper account to limit false and missed alarms. In the present study, real-time risk analysis for early warning purposes is discussed. The magnitude estimation is performed via the Bayesian approach, while the earthquake localization is based on the Voronoi cells. To test the procedure it was applied, by simulation, to the EEWS under development in the Campanian region (southern Italy). The results lead to the conclusion that the PSHA, conditioned to the EEWS, correctly predicts the hazard at the site and that the false/missed alarm probabilities may be controlled by set up of an appropriate decisional rule and alarm threshold.

Uncertainty in early warning predictions of engineering ground motion parameters: What really matters?

[1] From the engineering perspective, the effectiveness of earthquake early warning systems (EEWS) depends only on the possibility of immediately detecting the earthquake and estimating the expected loss at a location of interest, in order to undertake actions to manage/mitigate the risk before the strike. The simplest proxy for the earthquake’s destructive potential is the peak ground acceleration (PGA), which is predicted through probabilistic seismic hazard analysis in the framework of EEW. In this paper, the effects of different sources of uncertainty on the prediction of PGA are assessed with reference to the ISNet (Irpinia Seismic Network) EEWS. First the analyses show how the uncertaintyof the ground motion prediction equation (GMPE) dominates those of magnitude and distance, almost independently of the information available for the event. Secondly, based on these findings, information-dependent lead-time maps are provided for the Campania (southern Italy) region. Finally, different real-time magnitude estimators are compared in terms of errors in the prediction of PGA, as a more efficient estimator may give additional lead-timeforriskreduction. Citation: Iervolino, I., M. Giorgio, C. Galasso, and G. Manfredi (2009), Uncertainty in early warning predictions of engineering ground motion parameters: What really matters?, Geophys. Res. Lett., 36, L00B06, doi:10.1029/ 2008GL036644.

Conditional Hazard Maps for Secondary Intensity Measures

Abstract Vector-valued ground-motion intensity measures (IMs) have been the focus of a significant deal of research recently. Proposed measures are mainly functions of spectral ordinates, which have been shown to be useful in the assessment of structural response. This is especially appropriate in the case of structures following modern earthquake-resistant design principles, in which structural damage is mainly caused by peak displacements experienced during nonlinear dynamics. On the other hand, there may be cases in which the cumulative damage potential of the earthquake is also of concern, even if it is generally believed that integral ground-motion IMs, associated with duration, are less important with respect to peak parameters of the record. For these IMs, it seems appropriate to develop conditional hazard maps, that is, maps of percentiles of a secondary IM (e.g., duration-related) given the occurrence or exceedance of a primary parameter (e.g., peak acceleration), for which a design hazard map is often already available. In this paper, this concept is illustrated, and conditional hazard is developed for a parameter, which may account for the cumulative damage potential of ground motion, the so-called Cosenza and Manfredi index ( I D ), given peak ground acceleration (PGA). To this aim, a ground-motion prediction relationship was derived for I D first. Subsequently, the residuals of PGA and I D were tested for correlation and for joint normality. Finally, the study obtained analytical distributions of I D conditional on PGA and on the corresponding design earthquake in terms of magnitude and distance from hazard disaggregation. As shown by the application to the Campania region (southern Italy), I D maps conditional on the code design values of PGA may be useful, for example, for a more refined ground-motion record selection as an input for nonlinear dynamic analysis of structures.

Reliability analysis of accelerated life-test data from a repairable system

This paper analyzes accelerating testing of a repairable item modeled by a nonhomogeneous Poisson process with covariates. We extensively analyze a single accelerating variable with two stress levels, and derive closed-form maximum likelihood (ML) solutions. These closed-form solutions provide: (1) an easier way to obtain point estimates of the unknown parameters under usual operating conditions, and (2) a way to obtain confidence intervals on the process parameters and function thereof which are more accurate than those based on asymptotic normality of ML estimates. We analyze the circumstances in which a drawback to closed-form estimation arises, and guides the extent that our procedures may equally apply. An example application drawn from a real situation of accelerated testing is presented, and numerical estimates based on our procedures are derived and discussed. Theoretical and simulation results show that estimation procedures based on the power-law process and regression methods can be a flexible, useful tool for reliability analysis of a repairable item. >

A Wear Model for Assessing the Reliability of Cylinder Liners in Marine Diesel Engines

Stringent in-service and operating requirements oblige marine diesel engines to possess high levels of reliability and availability. To achieve the desired reliability and availability levels, it is necessary to carry out costly maintenance activity on the cylinder liners, the wear on which is a major factor in causing diesel engine failure. This paper presents a s-based methodology that can be used to carry out a condition-based estimate of liner reliability with respect to failure through excessive wear, and to plan condition-based maintenance activity. The wear process is described through an ad hoc cumulative damage model, and the reliability of the liners is estimated on the basis of wear measures, and predictions of wear increase. The approach proposed allows reducing maintenance costs without noticeably affecting liner reliability. In fact, it gives the probability of inspecting or replacing liners only when there is a high likelihood that their wear level will exceed the wear limit before the next inspection date. The proposed model, and planning procedure have been applied to a data set consisting of wear measures of the cylinder liners of two SULZER RTA 58 engines equipping twin ships of the Grimaldi Group

Operational (Short-Term) Earthquake Loss Forecasting in Italy

Abstract The seismological community is currently developing operational earthquake forecasting (OEF) systems that aim to estimate the seismicity in an area of interest, based on continuous ground‐motion recording by seismic networks; the seismicity may be expressed, for example, in terms of rates of events exceeding a certain magnitude threshold in a short period of time (days to weeks). OEF possibly may be used for short‐term seismic risk management in regions affected by seismic swarms only if its results may be the input to compute, in a probabilistically sound manner, consequence‐based risk metrics. The present article reports on the feasibility of short‐term risk assessment, or operational earthquake loss forecasting (OELF), in Italy. The approach is that of performance‐based earthquake engineering, in which the loss rates are computed by means of hazard, vulnerability, and exposure. The risk is expressed in terms of individual and regional measures, which are based on short‐term macroseismic intensity (or ground‐motion intensity) hazard. The vulnerability of the built environment relies on damage probability matrices empirically calibrated for Italian structural classes; the exposure is represented in terms of buildings per vulnerability class and occupants per building typology. All vulnerability and exposure data are at the municipality scale. The developed procedure, which is virtually independent of the seismological model used, is implemented in an experimental OELF system that continuously processes OEF information to produce nationwide risk maps applying to the week after the OEF data release. This is illustrated by a retrospective application to the 2012 Pollino (southern Italy) seismic sequence, which provides insights on the capabilities of the system and on the impact of the methodology currently used for OEF in Italy on short‐term risk assessment.