Hi Jáchym,
As David Turner is now longer at ecoinvent, I'll do my best to reply to your questions.
First, I just want to clarify the different products related to steel scrap that are available in the ecoinvent database.
There is the product "scrap steel", which is classified as a waste (not recyclable). It gets treated, and whatever burdens are associated with the transport and subsequent treatment of scrap steel are attributed to the activity that produces it.
These treatment activities include some incineration activities. In the underlying Undefined datasets, these have as byproduct "iron scrap, sorted, pressed". This product is classified as recyclable, therefore, the burdens associated with its transportation and recycling are cut off, so none of the associated burdens are attributed to the treatment activity. You'll see in the cut-off system model an input of "iron scrap, sorted, pressed" (with negative amount) supplied by the dataset "iron scrap, sorted, pressed, Recycled Content cut-off", for example in the activity "treatment of scrap steel, municipal incineration".
There is also the product named "iron scrap, unsorted", which classified as recyclable, and is the one that is commonly used for modelling an output of iron-containing waste that is sent to recycling, such as steel scrap. Activities that produce it have a negative input from technosphere that is supplied by the dataset "iron scrap, unsorted, Recycled Content cut-off", which is burden-free. This means that none of the burdens of the collection, transportation and recycling of "iron scrap, unsorted" are attributed to the activities that produce it.
So the answer to your first question is: no, it's not quite correct. The burdens of transportation of the "iron scrap, unsorted" and its sorting and pressing are attributed to "iron scrap, sorted, pressed". These burdens are then attributed to whatever activity turns this sorted scrap into iron and steel products. This is because "iron scrap, unsorted" is classified as recyclable. Your statement is however correct for products classified as waste, such as the product "scrap steel".
Regarding your second question: to model the impacts of steel recycling, in the ecoinvent cut-off system model, you should use the impacts associated with the dataset "sorting and pressing of iron scrap", which produces "iron scrap, sorted, pressed". Then, you would need to add to your model an activity that produces steel from "iron scrap, sorted, pressed". In ecoinvent, for example the activities "cast iron production" and "steel production, converter, low-alloyed" get the iron in part from "iron scrap, sorted, pressed" through the "market for iron scrap, sorted, pressed". This way, consumers of "iron scrap, sorted, pressed" get the burdens of the transportation of "iron scrap, unsorted" to the sorting facility, the burdens of the sorting, and also the burdens of transportation after the sorted iron scrap leaves the sorting and pressing facility.
You would thus have 2 steps that cover the recycling and production of steel, which could be compared to the production of steel purely produced from virgin iron (although as far as I know, most steel produced nowadays always includes a significant fraction of recycled iron).
The description above is also valid for plastics and paper, but you would need to first look for the products that represent plastic or paper that goes to recycling and their corresponding recycling activities.
Please keep in mind that all of the above is only valid for the system model "Allocation, cut-off". It is just one of the possible ways to allocate burdens among multiple products along a supply chain. The other two system models would give different results, due to the different philosophies that are applied.
Best regards,
David FitzGerald
Junior Project Manager
ecoinvent