Changing scale bar interval and other settings on individual Data Driven Pages?

I don't think you can change the properties of the scale bar using the out-of-the-box Data Driven Pages functionality.

However, that is not to say that what you're looking to do is impossible.

It wouldn't be terribly difficult to do this using ArcPy and is roughly described in this [ESRI Help Page][1]. Using their example, you could set up multiple scale bars, one for each setting you prefer, and then move them on and off the page based on scale of a given data-driven page.

To get it to work you would have to bypass the normal data-driven pages export and loop through the pages using Python and the above example.

Something along these lines (I HAVE NOT TESTED THIS, YOU WILL HAVE TO ADAPT IT TO YOUR DOCUMENT):

#You're going to need to reference your current map document and the data frame, assuming you just have one 
mxd = arcpy.mapping.MapDocument("CURRENT")
df = arcpy.mapping.ListDataFrames(mxd)[0]

#Associate scale bars in your layout with variables that arcpy can manipulate
m_scale = arcpy.mapping.ListLayoutElements(mxd, "MAPSURROUND_ELEMENT", "m scale bar")[0]
km_scale = arcpy.mapping.ListLayoutElements(mxd, "MAPSURROUND_ELEMENT", "km scale bar")[0]

#Iterate through all the pages, for each page, look at the scale, then adjust the scale bars to be on, or off, the page.
for page in range(1, mxd.dataDrivenPages.pageCount +1):

    #Check the scale and move the elements around to get the right scale bar on the page
    if df.scale < 25000:
        m_scale.elementPositionX = 5 #on the page
        km_scale.elementPostitionX = 15 #off the page

        #Export the current page to a pdf, using a specified path
        arcpy.mapping.ExportToPDF(mxd, r"C:\Project\Output\Project1.pdf", df)
    else:
        m_scale.elementPositionX = 15 #off the page
        km_scale.elementPostitionX = 5 #on the page

        #Export the current page to a pdf, using a specified path
        arcpy.mapping.ExportToPDF(mxd, r"C:\Project\Output\Project1.pdf", df)
    mxd.dataDrivenPages.refresh()