The Delay Converter allows you include builtin functions that have implicit stocks in them without drawing out the actual stock and flow structure they represent. This allows you to simplify the diagrammatic representation, typically in a portion of a larger model.
Consider what you would need to do when managing a bakery. Every day you need to bake goods hoping to have full shelves throughout the day, and empty shelves at the end of the day. In order to do this, you need to pay close attention to what you usually manage to sell and adjust that based upon things like the day of the week and the time of the year. A simple representation of this would be:
Here historical_sales is a delay converter and has the equation:
SMTH1(sales, time_to_average_sales, 100)
This model can be downloaded from here.
Without a delay converter, it would be necessary to treat historical_sales as a stock:
This model is exactly the same, and will give the same results. The representation is simply not as compact. The stock version of the model is available here.
Simulate this model and see what it does. Production buffer is set to 0 (produce exactly as much as you expect to sell). The results may seem surprising at first, but on reflection it should be clear that this is what would happen with a demand that goes up and down.