Using Project to Model?

[new2vista]

I'm working on a plan which has a rather long list of tasks. Right now, these
are all sequential, i.e., they are serially ordered in FS relationships. I'd
like to be able to model how the timeline changes if I factor in some
parallelization in the tasks. I know I can just create a separate plan file,
but what I'd like to do is to have multiple views of the task list in the
same file, with each view depending on whether the tasks are executed
serially or in parallel, and be able to flip from one view to the other
easily. Is this possible? I'm using MSP 2003 SP2.

Thanks.

 

[John]

I'm working on a plan which has a rather long list of tasks. Right now, these
are all sequential, i.e., they are serially ordered in FS relationships. I'd
like to be able to model how the timeline changes if I factor in some
parallelization in the tasks. I know I can just create a separate plan file,
but what I'd like to do is to have multiple views of the task list in the
same file, with each view depending on whether the tasks are executed
serially or in parallel, and be able to flip from one view to the other
easily. Is this possible? I'm using MSP 2003 SP2.

Thanks.

new2vista,
Generally speaking, no. What you will need to do is to change links
between tasks and within a given project file, only one set of link
relationships can exist. Creating a new view will not change the project
data, it simply allows you to take a different look at the same set of
data.

Your best bet is to indeed create a new separate file for each set of
task relationships. I suggest you do a Save As to your existing file.
Change the links and do another Save As. Repeat that process until you
have as many separate files as you want to analyze. You can then do some
comparison of those files by creating a master file (Project/Insert) so
you can display all scenarios in one view. Filtering in the master can
help to isolate areas of interest.

By the way, this process applies to all versions of Project. Even
Project 2007 will not allow you to have multiple link relationships
between given tasks in a single project file.

John
Project MVP

 

[new2vista]

Thanks.... what I was thinking of was fiddling with lag time to do the
modeling. For instance, I'd like to define a variable (custom field?) that I
can use to define lag for a set of tasks. Perhaps, even, a number of custom
fields that I can use for different sets of tasks. I don't see a way to set
lag in the task definition using such a custom field. Does that exist?

If it did, then, in one view, I could set a negative lag for one task set to
create some parallelism. It wouldn't be ideal since the lag would have to be
the same across all the tasks in the same set, but I guess I could play with
the ganularity of these task sets to approximate what I want. Of course, it
would also be nice to have the ability to define a profile for each task set,
such that each task in the set would have it's own unique lag defined, and
then to have multiple profiles you could play around with.

See what I'm after?

 

[Dave]

I'm working on a plan which has a rather long list of tasks. Right now, these
are all sequential, i.e., they are serially ordered in FS relationships. I'd
like to be able to model how the timeline changes if I factor in some
parallelization in the tasks. I know I can just create a separate plan file,
but what I'd like to do is to have multiple views of the task list in the
same file, with each view depending on whether the tasks are executed
serially or in parallel, and be able to flip from one view to the other
easily. Is this possible? I'm using MSP 2003 SP2.

Thanks.

I suppose that if you created a copy of the tasks in the same plan you
could open two views and apply filters to each of those views
(Window/New Window). Use flag values for example to exclude the
unwanted tasks from each of the views.

The plan wouldn't really be a plan then but you would be able to
visualise the effect of the changes you were making which is what you
asked about.

If you allowed this situation to persist, it would play havoc with your
resource allocations if you have resources applied to your plan. You
could elaborate the idea by creating dummy resources, but that might be
a lot of work.

Strictly speaking, if your tasks are serially ordered it shouldn't be
possible to execute them in parallel as the links are meant to indicate
what order the tasks can be done in rather than what order you have
decided to do them in.

I would probably make the links respect the order that the tasks can be
executed in. Then you can use the levelling function and priority
parameter to introduce degrees of parallelism if you have a resource
loaded plan.

If you aren't using a resource pool you could also make separate copies
of your plan (or even if you are, detach from it) as a new file and then
open both simultaneously to flip between the plans.

 

[John]

Thanks.... what I was thinking of was fiddling with lag time to do the
modeling. For instance, I'd like to define a variable (custom field?) that I
can use to define lag for a set of tasks. Perhaps, even, a number of custom
fields that I can use for different sets of tasks. I don't see a way to set
lag in the task definition using such a custom field. Does that exist?

If it did, then, in one view, I could set a negative lag for one task set to
create some parallelism. It wouldn't be ideal since the lag would have to be
the same across all the tasks in the same set, but I guess I could play with
the ganularity of these task sets to approximate what I want. Of course, it
would also be nice to have the ability to define a profile for each task set,
such that each task in the set would have it's own unique lag defined, and
then to have multiple profiles you could play around with.

See what I'm after?

new2vista,
Oh, I know what you are after, you just can't get it with Project.
Predecessors, of which lag is a part, cannot be customized with custom
fields - at least not in the normal operation of Project. You could set
up some custom fields for various predecessor modifiers (e.g. lag) and
then run a macro to insert those values into the actual predecessor
field, but again, you would only be able to see that particular project
data set by itself.

You could set up some schedule parameters (e.g. total slack, start
variance, etc.) that could be dumped into spare fields for each data set
and then do a comparison that way, but it wouldn't be the same as
actually seeing different scenarios of the same plan in one view.

I think the best you can do is to follow my original suggestion -
separate scenarios in separate files and then consolidate in a master
for direct comparison.

John
Project MVP