Hi Mike,
Thank you for the clarification.
Spider Project does not abort resource constrained scheduling if Finish No Later Than restriction is impossible to meet. It just moves activity finish for the least possible time and informs the scheduler.
Lower gang size is not always possible. Sometimes you need to move some activities to avoid overallocation. Avoiding overallocation of one resource you may cause overallocation of some other resource. So the process is iterative and may be very time consuming if done manually.
When your task is to define the number of resources of each type necessary for construction project at some remote place, you deal with the situation when almost all resources are restricted. And adjusting required number of rersources is multidimensional task. Resources that will be selected will be moved to the site and it costs a lot. It does not make sense to move the resource that will be used only a small amount of time, so you will avoid peak requirements. This task is too complicated for manual levelling. We had done this for Caspian Pipeline Construction program simulating the work of more than 1000 resource types. Even with automated levelling this task is rather complex with a lot of what if scenarios.
If overloads are few then it is much easier but still you will need to adjust the schedule manually each time when changes (like entering actual data) occur.
Why you do not use automated levelling? You will see the results, which activities were moved and where, which resources were overallocated at any moment. If the results are unsatisfactory you can make all necessary adjustments or assign priorities making the software to take into account your preferences. In most cases resource levelling change Critical Path. New critical activities with zero resource constrained floats create Resource Critical Path or Critical Chain. If it was calculated by the software you will not need to struggle with the resource overallocation all the time. The software will do it for you considering your preferences.
Best Regards,
Vladimir Liberzon

Hi Mike,
I did not try PowerProject but expected that it has good levelling capabilities.
I use Spider Project and its automated levelling is reliable and the schedules are at least close to optimal (shortest).
The software can try much more options than you can do manually. Levelling 3000 activities with 20 limited resources is not easy and finding the best solution manually is just impossible when resources are really constrained. And doing it again and again is a nightmare.
Best Regards,
Vladimir

Mike,
what size have your models? What is the number of limited resources? I am not sure that it will take 30 minutes in programs consisting of many thousand activities with several hundred resources. In this case you shall play not piano but organ.
Besides you will repeat this play each time after entering actual data. It will not be funny very soon.
In any case I am not sure that you will find optimal solution.
I think that you have an experience of using poor PM software - that is why you avoid to use automated resource levelling and I understand you.

Best Regards in the New Year,
Vladimir

Exactly!

Andrew,
when planning we try to optimize necessary resources - to calculate the amount necessary for achieving the result on time. If one day there is a need in 60 resource units, next day 40, then 70, etc. we can find that 53 is sufficient.
This is the usual task for project planning.
Besides some resources can be limited by the reasons that we cannot change. Like no space for installing the third crane on the construction site.
Regards,
Vladimir

Andrew,
I think the same - that Critical Chain theory did not suggest anything new but it proposed to pay more attention on resource constrained scheduling and using contingency reserves (buffers) for management by collecting all contingencies in one place. We had done it long before 1998 when Goldratt printed his book. In my presentation I described the approach that we use for many years, we call it Success Driven Project Management. Approaches are somewhat similar.

I don’t understand your point on resources for short term planning. We are involved in many large scale projects with limited resources. Limitations may have many reasons - supply limitations, lack of space, shortage of skilled workers, financing constraints, etc.
Besides peaks in resource requirements shall be avoided if possible. This is also one of the reasons for resource constrained scheduling in large programs.

Andrew,
resources are always limited. Not all of them but some. For an example you can be able to install not more than two cranes on the construction site, or you may have financial restrictions, or there are some peaks of manpower requirements that may be unreasonable to follow. So resource constrained scheduling is useful for optimizing resource requirements for completing the project on time and within available budget restrictions.
Resource levelling is necessary for proper resource planning. It is useful to understand which resources are critical in your project and what activities belong to Resource Critical Path (or Critical Chain - this is the name given to RCP by Goldratt). Activities that belong to RCP have zero total float calculated in the levelled schedule. Activities that belong to RCP may be unlinked and create a sequence basing on resource availability.
By the way most packages (including MS Project) calculate wrong total floats (slacks) when resources are constrained.

We use the approach that is similar to Critical Chain for many years (long before Critical Chain was invented by Dr. Goldratt). It includes RCP calculation and creating amd managing project buffers basing on risk simulation. Unlike Critical Chain approach for buffer management we use buffers not only for time but also for cost and materials.

I suggest to look at http://www.spiderproject.ru/library/tt.ppt and some other papers and presentations at http://www.spiderproject.ru/publ_e.php#eng

Best Regards,
Vladimir