This topic covers all elements of Deterioration Modelling in JunoViewer Web
The Benefit-Cost Analysis Modeling Process
|
Fritz Jooste Administrator Posts: 81
5/19/2020
|
Fritz JoosteAdministrator Posts: 81
This post provides an overview of the key aspects of Benefit-Cost Analysis (BCA) models in JunoViewer. Key stages of the model are explained and links are provided to further help post that delve into detailed aspects and modelling options.
The BCA models in JunoViewer run over two main stages, each of which require several sub-processes. These stages are (a) a Strategy Generation stage; and (b) a Global Optimization stage. In a model run, these two stages are automatically run one after the other and thus they are distinguished here mainly as a means of explaining how the BCA models work. An overview of these two stages is provided below:
Stage 1: Strategy Generation Stage (also see figure below)
- For each modelling segment, the model uses your treatment trigger settings and specified reset values to generate all viable treatment strategies over the full modelling period. Different approaches to strategy generation can be used and these may affect the running time of your model. For more details, see this post.
- For each candidate strategy, the discounted Benefit-and-Cost of Treatment is calculated. Benefits are calculated by comparing a specified Cost of Use parameter under each strategy to a Do Nothing scenario. For more details on how Benefits are calculated, please see this help post.
- A subset of Pareto Optimal strategies (i.e.g a Pareto Front) is selected from the total set of candidates on the modelling segment. For more details on how and why the Pareto Front is used, please see this post.
- The Pareto Front strategies are added to a Global Candidate Set. This set holds the Pareto Optimal strategies over all model segments;
Stage 2: Global Optimization Stage
- The first step in this stage is to handle the Committed Treatments assigned to your model. For each committed treatment, viable follow-up strategies are checked and the optimal Committed strategy is applied. The available budget in each treatment category is reduced accordingly for each applied Committed strategy and the committed treatments are removed from the Global Strategy Set.
- Strategies are now selected from the Global Strategy Set on the basis of the specified Optimization Objective, which could be either (a) Net Present Value; or (b) Maximum Benefit Cost ratio.
- A Heuristic Search method is used to select a combination of strategies such that (a) the user defined budget constraints are satisfied over all modelling years; and (b) the specified Optimization Objective is maximized.
In a typical modelling run, the Strategy Generation stage typically takes the longest time to complete. The run time for this stage can vary from several minutes to several hours, depending on the modelling period and how tightly your treatment triggers are specified. JunoViewer allows a user to specify the Strategy Search method, and this setting may significantly decrease running time for the strategy generation phase. For more details on the available Strategy Search methods, please see this post.
The Global Optimization stage typically finishes within a few minutes. However, the Global Optimization stage is an area of active research within Lonrix. Future improvements may involve algorithms such as Genetic Algorithms or Simulated Annealing that may take somewhat longer to complete.
It is important that you explore the outcomes of using different settings for your BCA models. Also note that, in some scenarios, you may get better results by using a Ranking model than a BCA model. This will depend on your specific modelling objectives. For more information about modelling objectives and deciding "which model is best?", please see this help post.
edited by admin on 5/26/2020
|
|
|
0
link
|
