Automated Skeletonization of models to increase simulation speed

Thank you Joel, Tom and Andy for the useful comments. I have set the status to Gathering Support so we can engage in consultations with a wider array of stakeholders to gauge their interest and gather additional perspectives on the proposed feature.

Should we observe a substantial level of enthusiasm and demand, we will be inclined to advance the status of this suggestion to "Under Consideration."

This stage will involve a more detailed analysis, including a feasibility study, resource allocation assessment, and preliminary planning to understand how this feature could be seamlessly integrated into InfoWorks ICM. At this stage, I will be reaching out to validate few wireframe feature designs with you to ensure that the feature does address your needs.

Would be good to have an option to run / simulate a selection only of the model for a selected area of interest using a selection list that you can drag in into the sim set up. If the software could do a pre-validation of selection and say setting the downstream node to an outfall as part of the sim. The user would then have to decide on the limitation of downstream boundary conditions, but this could be assessed when creating the selection list for the area of interest.

Agree with Andy's comments below. The automisation is the key for me. As to use the prune tool, users would be required to make small selections of areas, prune those and move onto another branch. Something similar on a model wide basis as Andy suggests and as shown in the WS video linked would be really useful.

Samer,

The big drawback with the pruning (and merging) facility (in ICM) is that it is a very manual process and is generally only done once, when first building a network. Pruning is used to trim off private drainage systems - the 'twigs' around the periphery of the trunk and branches. In the UK, models would very rarely have included private sewers but this changed when private sewers were transferred to the water and sewerage companies in Oct 2011. The more recent a model, the more likely it is that a download from a sewer record system will include (formerly) private sewers. (Some private sewer networks were surprisingly extensive and were far more than just the small pipes).

Going back to early days of modelling, it was routine to simplify networks - how many manholes had been simplified out was a data field for each modelled pipe! In current model networks I think would struggle to find any examples of the use of the merging tool.

What is needed is something to automate the process of simplifying a network - pruning and merging. The example link given in the original post is for WS Pro and it seems to have more automation in it than we have in ICM.

Key points are:

• we don't want to have to maintain multiple networks with varying amounts of detail but to have one network from which easy-to-produce simplified versions can be generated.

• there is currently too much time and effort required to produce and maintain simplified models. This has resulted in a decline in their use, even though there are useful benefits to be realised.

I would envisage a two part process - an automated tool working with a set of rules and a set of user-defined exceptions (local specific rules) to undo or override the automated part where necessary. If the base is updated the pair of tools would be used to recreate the simplified model quickly (perhaps with some tweaks to the user-defined exceptions). Thee would need to be a way to save the user-defined rules.

Thank you for the suggestion and thanks to those who voted. I am assuming that this is an ICM feature suggestion, not a CS one.

This feature already exists in InfoWorks ICM as users are able to select a group of links and go to Model >> Simplify >> Prune Selection - when this is applied to a selection, the network will be skeletonised and the subcatchment will be diverted to the new compartment.

Please give that a try and let me know how you get on

I have been thinking about this recently as well. This suggestion is aligned with the UK past practice of having a macro model (Type 1), a DAP Model (Type 2) and detail model (Type 3). There has been tendency to move away from having different Types (and therefore away from having to maintain 2 or 3 models for the same catchment) to have one model for everything. The facility to convert part of the model to Type 1 but keep the areas of interest at Type 2 or 3 would indeed be useful.

To skeletonize a model I suggest that we would want to have this facility within a scenario and would use selection lists to pick which areas to retain at a detail level - this might just be for overflows and pass forward flows but in an area of flooding it might be everything in an area of flood risk.

I have previously suggested that we should be able to run a model but only save full results for a selection picked using a lasso and to be able replay that part of the model. This is also about getting results for the part you are interested not the whole model. At present we cannot replay a model unless we generate results for the whole model. So, for the example of flood risk area, to be able to replay results would be useful to understand how the worst flooding in the (say) 10 year rainfall set rose and fell. Some strange things can happen in 10 year rainfall sets so being able to replay the detail provides useful insight.