Part 1: identify use cases for modeling information to identify requirements for Model Fidelity.
Today modern data science and culture is pushing the progression of regression models. This is not always the right choice for a process. The first step in any modeling initiative is to identify the need. The customer or process needs should be clearly identified. When embarking upon the modeling process one must be able to state this in a simple sentence. The first example shows how a model may only form a single insight for the facility (statement 1), while another may form multiple insights for the facility (statement 2). inform multiple insights for a facility, while another
- I am modeling the volume of the diffuser; so I can calculate the draft correctly.
- I am modeling the rate of solids deposition on a pressure leaf filter media; so I can understand these things: filter cycle time, filter solids capacity/cycle, filter capacity of facility.
As a consultant/engineer we need to be able to explain this to anyone quickly. Their question may next be: why do I need to know the draft. That can become a harder one to answer, because operations does not always see that knowing things like the current draft, are extremely valuable. This may not be used to identify a failure mode in real time but instead develop insights for decision making on a future capital project.