This topic covers all elements of Deterioration Modelling in JunoViewer Web
Strategy Generation in Benefit-Cost-Analysis Model
|
Fritz Jooste Administrator Posts: 81
5/19/2020
|
Fritz JoosteAdministrator Posts: 81
As explained in this post, the first stage of a Benefit Cost Analysis model (BCA) model run involves the generation of a set of viable treatment strategies on each model segment. A set of treatment strategies will include all treatments that can be triggered in sequence over the length of the modelling period. The figure below shows a schematic representation of a set of strategies generated on a single model segment:
As indicated, for each strategy a discounted Cost and Benefit is calculated. Benefits are calculated by comparing the total of the specified Cost of Use parameter under each strategy the the Cost of Use under a Do Nothing scenario. [The Cost of Use parameter is specified on the General Sheet of your DMS file]. The calculated Benefits and Costs are then used to identify Pareto Optimal strategies which are then put forward to the Global Optimization search. For more details on how Benefits are calculated, see this post.
The generation of viable strategies over all segments is the most time consuming element of a BCA modelling run. For long modelling periods (20 years or greater), and with relatively open triggers, hundreds or even thousands of strategies can be generated on each segment. By default, JunoViewer limits the number of strategies allowed on each segment to 300. However, this setting can be overridden on request. Contact Lonrix for assistance if your model requires a higher maximum strategy setting.
JunoViewer currently allows three approaches to Strategy Generation, as outlined below:
Exhaustive Tree Search: This is the default method. It is the most thorough but also slowest method of strategy generation. In this method, all possible viable strategies are considered. With relatively open triggers this method may easily result in excessive strategies being generated (most of which will not be on the Pareto Frontier).
Exhaustive Tree Search with Trigger Horizon: This method is somewhat faster than a full exhaustive tree search. Essentially, the Trigger Horizon limits the number of years that a treatment can be postponed after it is first triggered. Using a trigger horizon of 3 or 4 years can substantially reduce the number of strategies generated without the user having to use complicated trigger settings. Please refer to this help post for more details on strategy generation with a Trigger Horizon.
Phased Strategy Generation: This is the fastest method for strategy generation and is a more sophisticated application of the trigger horizon concept. It typically results in a reduced strategy set that includes the most optimal treatments on the Pareto frontier. In this approach, strategies are generated in stages. Users can specify the treatment stage (typically 12 years), and the length of follow-up periods (typically 8 years each). By limiting the number of strategies generated in the first treatment stage, the decision tree is significantly trimmed which in turn speeds up the strategy generation process significantly.
To specify which of the above methods are to be used, you need to correctly specify the following parameters on the General Sheet of your DMS file:
|
|
|
0
link
|
