Even from the discussion between Colin Ricks and I, it is apparent that the method and/or approach to modeling can and does differ slightly in the professional world.

Regarding WaterCAD/GEMS’s parent-child dataset (i.e., alternative) relationship, I personally do not feel it causes more problems. Yes it is a little more complex, but the water modelers in my company generally prefer this feature over the current setup in IWP. We find it is easier to update records in the parent dataset as this allows the changes to cascade through to the child datasets (i.e. alternatives).

Regarding implementation to IWP, this would not have to change the entire interface of the datasets in IWP. The current setup could be default, but with a setting to manually change the datasets relationships. To Colin Ricks’ point, if younger or inexperienced modelers are unfamiliar with the program or how a parent-child relationship works, this may precipitate error. However, I believe this feature would not detract, but rather add value and enhance the capabilities of IWP. This would be appreciated by modelers who choose a more complex form of modeling or prefer to work with child-parent relationships.

I recognize that there are lots of use cases and everyone likes to approach modeling a differently. I use Bentley WaterGEMS about as much as I use InfoWater, so I feel like I can provide some insight on my experience with the way this implmented in WG.

WaterCAD/GEMS has a parent-child relationship for datasets*, and can have records in a dataset that are inherited from the parent, or can be "local" (manual overide of parent data). Because you can have local data, it's hard to know what impact a change you make in one scenario will have on other scenarios that reference a parent dataset.

For example, if I have a parent and child pipe dataset, and I make a change to a diameter in a scenario which uses the child dataset, will that change also be made in the parent dataset (and consequently any scenarios that reference the parent)? If I update the parent, it doesn't update the local data, but how do I know what is local data, or if there is any local data at all? I can figure this out, but it takes time and effort, making it easy to introduce errors by making a change without checking the dataset relationships and local settings.

This is compounded when you also have parent-child relationships with scenarios. Not to mention, if I come back to a model after a few months, it's hard to remember how all those relationships will work and easy to introduce errors by making a change without diving into the dataset relationships.

So I find it's easier and faster to just have separate independent datasets and do the comparison/updates in Excel if I really need to, which is rare. It's also easier for less experienced modelers to understand this approach and harder for them to mess up the model.

Innovyze could certainly come up with a better implementation than Bentley that's easier to use and understand, but I'm just not sure how. Maybe the answer is to add this feature to IW, and some people will make use of it for specific use cases like Joe Lawrence described, and others will ignore it? However, I fear that the added complexity will make it easier to introduce errors to a model.

*Bentley refers to these as "alternatives", but I'll use the term "dataset" here.

Thanks Colin Ricks for your comment. It is good to hear from other water modelers. To respond, I would argue that it depends on the type of modeling you do, or the way you approach modeling. There are likely different modeling approaches even between the users of InfoWater Pro. Maybe this is an option that the modeler could choose in the project preferences for a model to allow or not allow cascading changes through the datasets.

The situation I run into the most with this issue is when we receive an update from the client on their existing system parameters or data. I have already created multiple proposed scenarios based off the existing system (it can be upwards of 10 proposed scenarios). If I must update the existing pipe diameters, adjust valve settings, turn off or on initial settings...I must go into each scenario that has a unique dataset and manually make those updates. It would be simpler and more efficient if I made the change to the "parent" existing scenario dataset(s) and then let the program update all child datasets (i.e., proposed datasets) unless there is a manual override. The current situation (no parent-child relationship) inherently consumes time and money. It also introduces potential for errors as someone must physically input the values into each scenario with a unique dataset. I’ve found from experience the more manual entries performed the higher percentage of error(s) will be present in the model.

Hope the sample situation above helps explain how the current setup is not always "simple" and "efficient" on my end.

The parent-child relationships in the Bentley "datasets" can easily get really complicated and confusing. As an experienced modeler, I prefer the simpler approach InfoWater currently uses, which I think strikes the right balance.