Best practices when running SLACA (Simplify By Straight Lines and Circular Arcs)

[JasonBessert]

Need assistance with determining the safest parameters to enter into the dialog for the Simplify By Straight Lines and Circular Arcs.

We ran SLACA over a large parcel dataset, upon fabric loading, and have not built up the fabric yet to lines yet.  We only have polygons in this dataset as we are seeking out gaps and overlaps with the highlight tool.  As we scrub the data further, we are noticing several parcels have had their geometries corrupted, to the extent that false curves were created (see screenshots)

In most instances, we've seen the SLACA tool work as advertised in recent meetups from ESRI, but from the screenshot examples, we're seeing an unfortunate influx of parcels that had curves created where there shouldn't be, and curves that fail to be created when they should be, and we just want to know if there is a best-fit array of parameters to have entered to keep this data's geometry as intact as possible while eliminating the densification of curves as the tool is intended to work.

[JasonBessert]

I just don't think the SLACA tool should ever be taking a polygon like this:

And turning it into this:

We just need to know how we can have it run without these kind of geometry corruptions.

[JasonBessert]

Here is an update, below are the parameters used when I re-loaded these parcel polygons from the source and ran the SLACA in a manner that best output to match the original, while creating arc segments, even for the small radius ones.  It also prevented those straight segments from inadvertently 'eating' into the rectangular unit polygons.  It's the solution I will apply to other areas similar to this, laden with stacked condo parcels and complex road parcels having curves with various radial sizes.  Looks like the key was to have "Fit To Segments" selected over fit to vertices, while having a minimum radius set to 1  (Not sure what the default is if no minimum is entered).  We would have to be careful with this, as it may not be the solution in other areas in the event that it creates small curves where it shouldn't.  Perhaps there is NO one-size fits all set of parameters to apply across the board on a large dataset of all kinds of parcel geometries, ranging from detailed condos to large parcels in remote areas.      

[JasonBessert]

Here's an update - I re-ran the entire dataset with the "solution" parameters I listed above from my post last week, and there was some unfortunate results.  My original post was a gripe that the tool, under certain parameters was 'over-creating' curves in instances where they data should be straight lines.  In this latest attempt to run SLACA, I was left with the opposite problem, where the tool failed to create curves in geometries that should obviously be true curves (two straight segments were the unfortunate outcome)

So I'm back to where I started, not having a set of input parameters to run over the entire dataset as a one-size-fits all means of running SLACA to remove the densification in curve segments without creating false curves and corrupting a parcel geometry over the course of a conversion from shapefile to parcel fabric.