The rule to avoid splitting an activity just to make it shorter applies to all related activities to the pipe installation deliverable; excavation, bedding, pipe alignment & lamination or bell & spigot, partial backfilling, hydrotest, complete backfill etc. are all separate activities related to the same deliverable. Each of these activities shall not be split if are to be modeled as contiguous. This is understood by even a rookie scheduler, especially by those who resource load their schedules. If the schedule is long then the activities will all have long durations when there is concurrency logic.  It is important to understand how differently a CPM model might schedule a set of activities versus a single contiguous activity.

If you separate each activity under many activities using only FS sequential logic it would be a fatal flaw of logic that will result in inefficient schedules.

http://www.me.utexas.edu/~jensen/network_02/topic_pages/kanegaronkar/opinions.htm

Even the fanatics that insist on very strict rules can accept the use of SS and FF with lag as an acceptable method to model some concurrency, it is not perfect but better than FS only.

Linear diagrams are good at this and usually a few lines are enough to represent all works in a very simple way.  To my knowledge no one ever suggested putting some time limit to the lines.

I do not use P6 but for users of P6 Ron Winter provides software to analyze P6. Ron Winter does not advocate for inflexible rules, he advocates for some guidance to be used wisely.

http://www.ronwinterconsulting.com/DCMA_14-Point_Assessment.pdf

Another reason not to literally interpret guidelines happens when the schedule is used as a tool for payment purposes. The following link provides some reasons why cost loading for payment purpose is not a good idea and what to do if you have no other option because it is on the specifications. 

http://www.nflaace.org/index_files/john_orr_cost_loaded_schedule_updatin...

Another common exception to general guidelines is when the specifications require to keep under separate activities the resources that do some portion of the activity but who report to a separate contractor or subcontractor.  If they are required to do the job at the same time all other crews you shall keep all crews working as a group within the same activity or the resources might be assigned at different time when in reality they must work the activity at the same time. 

In my jobs a common case is when working on elevated slabs where we have electrical, plumbing and civil contractors working on the same slab during the same time period. If you assigned them under separate activities the resource leveling might schedule them working at different periods.

Total Float is a calculated value it is what it is based on your model.

Activities usually represent deliverables or related tasks necessary to make the deliverables to happen.  Activity durations are whatever they need to be.  

We use resources to make them happen, for us not taking into consideration the resources is senseless.  We make use of resource constraining even if all resources are infinite, well this is hard to believe.  Good resource planning requires planning for low idle resources as you cannot hire and fire resources on a daily basis.  For this you use what-if models in your search of a plan that minimizes idle resources. One approach is to artificially limit non critical resources and see if idle resource time is reduced without impacting the schedule total duration or contractual milestones.  

Sorry but there are no fixed rules and some forensic analyst want you to believe scheduling is an exact science, otherwise why experts from both sides cannot agree?

Take for example an equipment or materials delivery activity that takes 6 months.  Why not? What is the purpose of subdividing it into six monthly activities.  What if delayed, to which activity will you apply the delay?  It does not require much brain to figure out remaining duration based on expected delivery date as per most recent available information.

Take for example the installation of a water main 40 km long and a project duration of 36 months.  Why dividing it into 36 activities when a single activity is enough?  Is it that you bill by the activity count?  Is it that more but unnecessary activities make it look better and you sell your schedules by the look? Is it that you need more activities to know how much work have been performed?

By dividing a contiguous activity into several activities the resource leveling might spread apart several segments, especially if some resources such as equipment is shared with other activities. This is not what you want if the activity is to be performed in a continuous way.

To determine progress of an activity only intermediate school math is required.  Most construction activities have a volume of work related to it and percentage complete equals actual volume of work divided by total volume of work, it is very easy, this is not kindergarten. 

We believe scheduling specifications shall always be Performance based specifications that allow the contractors to bring their own expertise to the schedule process without restricting them in any way to predetermined methods.  If the contractor is responsible for the means and methods why restraining him to yours. If you want to keep control of the means and methods why hiring a contractor, better do it yourself. 

No matter what the scale of the project at home we find such rules senseless and and childish. 

The DCMA (Defense Contract Management Agency) uses float as one of it's key indicators when assessing the health of a cpm schedule. The DMCA sets this parameter for testing high float values at 44 days. The DCMA has set the total number of activities with float values at 44 days. The threshold is set at 5% or less. This means that of the total number of activities with float values of 44 days should be less than 5% of the total number of activities in the schedule.

As a side note the DCMA also suggests that individual projects may wish to expand or contract that threshold based on the length of the project and the type of project being scheduled; however, any changes in thresholds should be coordinated with the customer first to confirm the viability of the alternate measurement.

I have revised my threshold to 80 days and I do not consider long lead time times in my analysis. 

I see a lot of answers about the length of the duration when it is really a matter of logic. Duration is also another DCMA test with high durations this is al so set at 44 days and the threshold is also at 5%. This maybe in conflict with tighter scheduling specifications which limit the durations to 20 or 21 working days. Which as previously mentioned is a monthly updating period.