Inspection validation model for life-cycle analysis

Design decisions lead to future maintenance needs. In the past, these decisions were often taken implicitly. During the last decade there has been a development to make more explicit choices, based on the expected condition in time sometimes combined with life-cycle cost analysis. The inspection validation (IV) model is currently under development. This decision-support model can be used to evaluate the a priori design assumptions on the condition in time versus the actually measured deterioration. The aim of the IV model is to bridge the gab between design and inspection, and to provide maintenance engineers with better founded maintenance advises. The variability of the deterioration in time is modeled by a gamma process. Its parameters have been estimated based on inspection data by combining the methods of least squares and maximum likelihood. Some practical examples of the inspection validation model are discussed. structure that Rijkswaterstaat owns is provided with a maintenance plan (Bakker and Volwerk, 2003). These plans forecast the future maintenance costs on element level, based on nationally or locally defined maintenance strategies, derived from the expected condition in time. These strategies can be justified with life-cycle cost analysis. 3 PREDICTED MAINTENANCE COSTS BASED ON EXPECTED CONDITION 3.1 Implicit forecasts During the design phase of a structure many implicit forecasts are made for the expected future maintenance need. For instance: a concrete bridge is constructed using the national building codes with an expected lifetime of 80 years. The codes will provide boundary conditions on concrete cover and concrete quality. These boundary conditions include implicit durability assumptions ensuring the structure to last for 80 years. 3.2 Explicit forecasts Sometimes maintenance is forecasted explicitly. For instance: • A steel retaining wall is constructed with a surplus of steel thickness of 2 mm. These 2 mm should be sufficient to make the retaining wall last for 50 years. • Based on the Reference Documents (Klatter et al., 2002), the forecasted average maintenance intervals for the asphalt pavement on the bridge are: Replacement of the porous asphalt top layer every 10 year Replacement of the total asphalt every 22 year • Sometimes the required lifetime of concrete structures is defined in construction contracts. Contractors have to prove, for instance by DuraCrete calculations (DuraCrete, 2000ab), that the desired lifetime is reached with a predefined certainty. This implies that the concrete structure is theoretically free of maintenance for the desired lifetime. 3.3 Forecasts based on lifecycle optimization One of the tools used at Rijkswaterstaat is the LEM model (Bakker et al., 1999; van Noortwijk and Frangopol, 2004). This model can optimize intervals of replacement and lifetime-extending maintenance (LEM) based on: • expected condition in time, • uncertainty in condition in time, • costs of preventive replacement, • costs of corrective replacement, • costs of lifetime-extending maintenance, • parameters for the effect of lifetime extension on the condition. Based on the lowest present value of the life-cycle costs, the model determines optimal intervals of lifetime-extending maintenance and replacement. The LEM model has been applied to optimize the maintenance of the coatings on the steel gates of a Dutch storm-surge barrier (Heutink et al., 2004). In this application, lifetime extension is defined as local spot repair of the coating and replacement as removing the old coating and placing a new one. The condition parameter is the percentage of the corroded surface. 4 EVALUATION OF DETERIORATION PREDICTIONS WITH INSPECTIONS Predictions of the future condition of civil structures are uncertain. Often, there is not much statistical data available and simulation programs don’t take into account all the (mutually dependent) parameters involved. Therefore, inspections are needed to validate the current predictions of the condition over time and to update them (Figure 1). This is not an easy task, since there is not always a clear relation between the predictions and the relevant inspection parameters. The inspection validation model can be a tool to support inspection validation in a uniform manner. Condition parameters Maintenance forecast Inspection Inspection validation results Condition prediction Figure 1. Inspection validation cycle. 5 INSPECTION VALIDATION MODEL The Inspection Validation (IV) model is a decision-support model. It presents the a priori predicted condition as a function of time versus the updated condition based on inspection results. Based on the updated prediction, the maintenance engineer can decide to: • Do nothing: wait until the next inspection; • Change the inspection strategy; • Change the maintenance strategy; • Make a life-cycle cost analysis based on the predicted condition. Figure 2 shows an example of a graphical output. The solid lines represent the a priori predicted time-dependent condition with the 95% confidence interval and the dotted lines represent the updated time-dependent condition based on the inspection results. Expected condition according to c = r_0 a·t^b 0 10 20 30 40 50 60 70 80 90 100