Skip to main content
Business LibreTexts

4.2: Simple Decision Making and Forecasting from Condition Assessment

  • Page ID
    21125
  • In many cases, asset management decisions are made without complicated deterioration models at all. The simplest approaches simply use the existing component condition, a linear projection of the component condition over time, or a projection based upon the past history of similar components (using a graph such as those illustrated in Figure 4.1.3). These approaches are discussed in this section.

    Using the existing condition can be augmented with simple decision rules. For example, ‘if the condition is x or lower, then rehabilitation is desirable.’ Components with only two condition states defined are particularly amenable to this approach. For example, an incandescent light either works or is burnt out. The maintenance rule might be to only replace lights when they burn out.

    A subset of this simplified method includes a “run to failure” approach. Unfortunately, this “fix it when it breaks approach” is a widely applied and expensive approach to infrastructure management. There are scenarios, like window air conditioners, when it simply doesn’t make sense to replace before failure. As we will explore, it is nearly always more cost-effective to replace before failure when considering major infrastructure systems.

    Using existing conditions has the advantage of eliminating any costs associated with deterioration modelling. However, the amount of effort may fluctuate considerably as many components cross over the trigger condition for action and this may not be compatible with budget constraints. Also, if deterioration has significant costs, waiting until deterioration occurs may not be the best approach.

    Simple linear extrapolation is another approach that is inexpensive to employ for deterioration models. In this approach, c where ct is the condition at time t, and Δt is some desired time period in the future.

    \[c_{t+Δt} = c_t + (c_t - c_{t-1})\]

    As an example, suppose the component condition is now 3, last year the condition was 4, then the forecast for next year is \(3 + (3-4)*1 = 3 – 1 = 2\) and the forecast for the following year (two years from now) would be \(3 + (3-4)*2 = 1\). More complicated forms of extrapolation could also be used, but linear extrapolation is the most common for infrastructure deterioration.

    A single year maybe two short of a period to effectively capture deterioration, so a moving average of multiple years might be used instead. In this case, \(c_t\) would be the average condition for the current period of years (which might be the past three years). This approach would be useful for very slowly deteriorating infrastructure components. Moving averages of this type are common for smoothing fluctuating time series histories such as stock prices.

    Finally, forecasts of component deterioration might be based upon simple historical records. For example, Figure 4.1.3 shows the average deterioration trajectory of different bridge decks under specific conditions. An infrastructure manager might assume that a particular bridge deck with a particular condition would simply follow this trajectory in the future. Even without a formal database, infrastructure managers might have their mental model of expected deterioration and make subjective forecasts based upon their experience.