28. ELEMENTS of STRUCTURAL RELIABILITY

Traditionally, evaluations of structural adequacy have been expressed by safety factors SF = C/D, where C is the capacity (i.e. strength) and D is the demand (i.e. load). Whereas this evaluation is quite simple to understand, it suffers from many limitations: it 1) treats all loads equally; 2) does not differentiate between capacity and demands respective uncertainties; 3) is restricted to service loads; and last but not least 4) does not allow comparison of relative reliabilities among different structures for different performance modes. Another major deficiency is that all parameters are assigned a single value in an analysis which is then deterministic. Another approach, a probabilistic one, extends the factor of safety concept to explicitly incorporate uncertainties in the parameters. The uncertainties are quantified through statistical analysis of existing data or judgmentally assigned. This chapter will thus develop a procedure which will enable the Engineer to perform a reliability based analysis of a structure, which will ultimately yield a reliability index. This is a 'universal' indicator on the adequacy of a structure, and can be used as a metric to assess the health of a structure, and compare different structures targeted for possible remediation.

In this introduction, we learn about how structural reliability is traditionally assessed using safety factors. A safety factor is a ratio of the structure's capacity to its demand, expressed mathematically as SF = C/D. However, this method has several limitations. It does not understand the different types of loads acting on a structure, does not account for uncertainties in measuring capacity and demand, and cannot compare reliability between different structures effectively. Therefore, engineers have developed a probabilistic approach that incorporates these uncertainties through statistical data analysis, allowing for a more reliable assessment. This chapter aims to teach you how to perform such an analysis and obtain a reliability index to evaluate and compare different structures.

Imagine a bridge builder assessing whether their new bridge is safe. Using the safety factor method is like checking if a car can fit under a bridge by just measuring the lowest part of the bridge and the height of the car without considering potential fluctuations like the bridge sagging slightly or the load of passengers in the car. However, if they used a probabilistic approach, they’d consider different scenarios, weather challenges, and the bridge’s wear and tear, leading to a much more dependable assessment of safety.

Elementary statistics formulae will be reviewed, as they are needed to properly understand structural reliability. When a set of N values x is clustered around a particular one, then it may be useful to characterize the set by a few numbers that are related to its moments (the sums of integer powers of the values): Mean: estimates the value around which the data clusters. (Mean = Σxᵢ/N)

This chunk discusses the importance of basic statistics in structural reliability. The mean, a fundamental statistical concept, is emphasized as it helps to determine the central tendency of a data set. The mean is calculated by summing all values (xᵢ) and dividing by the number of values (N). Understanding the mean is crucial for engineers as it allows them to recognize typical load conditions on a structure, which aids in evaluating performance reliability.

Think about how teachers calculate the average score of a class on a test. If each student had a different score, the teacher adds all the scores together and divides by the number of students to understand how well the class performed overall. Similarly, when engineers assess loads on a bridge, they calculate the average load to predict how the structure will behave under realistic conditions.

Skewness: characterizes the degree of asymmetry of a distribution around its mean. A positive skewness indicates that the distribution has a tail extending toward more positive values. Kurtosis: is a nondimensional quantity which measures the 'flatness' or 'peakedness' of a distribution. A negative value suggests a distribution that is flat, while a positive value indicates a sharp peak.

In this chunk, we explore two key statistical concepts: skewness and kurtosis. Skewness describes how symmetrical a data distribution is. If it skews to the right, it means there are more high values. Kurtosis, on the other hand, reveals the shape of the distribution—whether it is flat (indicating more average values) or has a sharp peak (indicating more extreme values). Understanding these characteristics is vital for engineers when assessing how loads behave under various conditions, as it offers insights into the variability and risk associated with structural reliability.

Imagine a set of data points representing the ages of attendees at a concert. If most ages are young with a few older attendees, the data is skewed to the right. If you think about the concert's atmosphere, a crowd dominated by teenagers creates more energy, which is different from an older crowd where age is more evenly spread. The shape of the crowd's age distribution impacts the overall concert experience—the same applies to understanding load distributions on a structure.

--