You are here
Repair /Replace Decision Models
DIAGNOSTIC TESTING & DECISION MODELS
A Decision Analysis framework that includes value of information is fundamental to the electric utility asset management problem. For every class of assets the following three elements are present:
- A decision cycle – inspections and repair decisions occur on a periodic basis.
- Diagnostics testing – potentially provides useful information and is an integral part of the asset management process.
- Maintenance and replacement options – the key is to find a combination of diagnostics / inspection options and repair / replace options that minimize the lifecycle costs of the asset class.
The diagnostic testing protocol associated with any class of assets provides information about the state of specific assets in the asset class. However, diagnostic testing results alone do not directly prescribe which repair/replace decisions should be made:
- The best decisions are only identified when the information provided by testing are embedded in a decision model.
- The value of specific testing can be directly computed when the protocols are imbedded in arepair/replace decision model.
- Diagnostic testing is only valuable if the information provided by the testing change the optimal repair/replace strategy. These points can be demonstrated using a simple one-period decision model.
A ONE-PERIOD DECISION MODEL
The Situation: Suppose there is a 50/50 chance that the asset is in poor health. If in poor health and left unrepaired it will fail and cost $1000. However the asset can be repaired for $100 and after the repair the asset will not fail. Also suppose that determining the health state requires a diagnostic test (an inspection).
This situation is a lottery. The decision tree here illustrates the possible choices and outcomes. Without inspection there is a 50/50 chance you will pay $1000. It costs $100 to do the repair and avoid the $1000 risk.
Now consider the inspection option, assuming that the inspection is 95% accurate:
- If the report indicates good there is a 5 percent chance that condition is bad and the cost will be $1000
- If the report indicates bad there is a 95 percent chance that condition is bad and the cost will be $1000
The Analysis: To analyze these kinds of problems we have developed a one period decision model that can be run from this web page. With this model you can determine optimal testing and repair / replace strategies given various test costs and accuracies, repair / replace costs, and failure likelihoods.
Click the Run Model button and you will be able to apply the framework (the figure contains the input parameters for this example):
A WORD OF CAUTION
The model in the example above is for a one period evaluation. In most real world asset management problems, because the objective is to minimize lifecycle costs and because asset condition evolves over time, correctly solving the problems requires multi-period evaluations – what is done at a specific point in time will affect the best course of future actions.
Multi-period models that capture long-term dynamics are required. In real world asset management decision models there are several essential elements, including the following:
- Decisions must be based on the estimated health state of the asset.
- Hazard functions (probability of failure as a function of age) are a starting point for estimating the health state.
- Improving the prediction of the heath state of the asset will require using inspections and diagnostics.
- The decisions in each period are (1) to test or not, and (2) decide what to do about the asset.
Multi-period models are required for another reason. They provide an answer to the questions: (1) is diagnostic testing justified given the costs of the tests and the accuracy of the information provided by the tests, (2) at what age should testing be initiated, and (3) once started how frequently should the tests be applied?
An IEEE paper “The Optimal Replacement of Underground Cables” provides a detailed description of the application of a multi-period decision model that integrates diagnostic testing decisions and repair/replace decisions. This paper can be downloaded from the following link: Link to IEEE Paper - Optimal Replacement of Underground Cables