Just in Time levelling

[stoods]

Hi,

Project is doing a great job of levelling teh resources, but I cannot figure
out how to get project to just in time various tasks. It is currently moveing
tasks with a start as soon as possible, but it I want to delay these tasks
until they are closer to the last task in a group of tasks. Ie, task 1 is
followed by task2 which is followed on by task 4 and so on. How can I level
so that the start of task 1 minimises time between task 1 and task 4?

Thanks,

Mike

 

[Rod Gill]

Are you sure you want to do that? If all tasks start as late as possible,
then they are all critical. Delaying any task delays the project finish
date.

 

[Jan De Messemaeker]

Hi,

Project is very poor at that. As you may see, many claim it is not necessary
(whereas it is a generally accepted practice in production environments);
maybe that is why the requests to develop it aren't honored.

Let's be positive: if you can fix the date at the end of the chain you can
use As Late As Possible constraints; if not you'll be condemned to make
manual adjustments, sorry:-(

 

[stoods]

Thanks,

I do appreciate the replies.

Yes, we do need to do this.

The situation is a woodworking factory. The processes are typically cut the
board, edge the board, drill the board, paint the board and then put it
together. The bottleneck is in assembly, and teh saw can produce more parts
than can be assembled. So, if all the sawing happens as soon as possible, you
have to find somewhere to put all the parts that have been cut while they are
waiting to be assembled. So, it does make sense to do this. Also, it is
acceptable that all tasks become on teh critical path, that is the trade off
needed in this style of application.

I guess also on another level, you would also need to purchase all the raw
materials well in advance of needing them, cuasing more storage issues, as
well as incurring costs needlessly.

 

[davegb]

Thanks,

I do appreciate the replies.

Yes, we do need to do this.

The situation is a woodworking factory. The processes are typically cut the
board, edge the board, drill the board, paint the board and then put it
together. The bottleneck is in assembly, and teh saw can produce more parts
than can be assembled. So, if all the sawing happens as soon as possible, you
have to find somewhere to put all the parts that have been cut while they are
waiting to be assembled. So, it does make sense to do this. Also, it is
acceptable that all tasks become on teh critical path, that is the trade off
needed in this style of application.

I guess also on another level, you would also need to purchase all the raw
materials well in advance of needing them, cuasing more storage issues, as
well as incurring costs needlessly.

You might consider looking at Production Scheduling software for your
situation. Project is primarily designed to do CP scheduling, something
you're not really doing. I would think, having no direct experience
with it, that production scheduling software would allow for JIT
processes, which Project doesn't do well. I Googled for production
scheduling software a week or so ago, and got lots of hits. I imagine
one of those products would serve you better.
Hope this helps in your world.

 

[John Sitka]

davegb tells us they exist,
Google tells us they exist,
the software vendors tell us they exist.
Guess what, they don't exist.

 

[Jan De Messemaeker]

Hi John,

Just for once I disagree with you.
AXAPTA (from Microsoft now, used to be Navision) has an elaborate Production
management module.

BUT! Project specialists recommending the use of such a monster don't
realize the scale difference there is between Project and these tools.
They're not just about job scheduling, they're about Bills of Material and
versions thereof, Work in Process accounting, job costing, invoicing, links
to HR data bases, to general ledger...

I humbly admit that I very optimistically tried to get into this part of
AXAPTA and that I have given up.

So I do understand people who try to use Project for job scheduling. It's
very simple (as compared to the elusive production packages) and but for a
few weaknesses it can produce very sensible results. Just in time is one of
these weaknesses, an other one is the impossibility to manage material
resources and having their availability influence the schedule.

Greetings,

 

[John Sitka]

They're not just about job scheduling, they're about Bills of Material and

versions thereof, Work in Process accounting, job costing, invoicing, links
to HR data bases, to general ledger...

I think that is the real problem. The act of scheduling dosen't have to be so close
coupled to those things. The product to produce is just a box of tasks.
These should be allowed to float freely within the resource pool without being
accountable to the traditional measures of progress,or tied to other things
like worker resident time.

The magic factory, raw materials in, finished goods out. Sit back and just enjoy the results.
Why do the accountants feel it is their responsiblity to peak inside the factory
and help decide how and when it is best to paint something?
(It will take me 2 hours to do this, 4 if you help)
Why are work cell managers held accountable for their efficiencies in isolation of what
goes out the back door?
(We are not a polishing company we are a shiny widget builder, to evaluate us based
on our polishing skill is rediculous. The gleam on our widgets is second to none. And we can have
your widgets ready for you fast and high quality, how many can I put you down for.)

The software dosen't exist becasue it wouldn't sell, because the scheduling idea for a vast
majority of the world is cost accounting based, not finished product efficiency based.
Operational measurement is cost accounting based not bottleneck erasure based.

 

[davegb]

I think that is the real problem. The act of scheduling dosen't have to be so close
coupled to those things. The product to produce is just a box of tasks.
These should be allowed to float freely within the resource pool without being
accountable to the traditional measures of progress,or tied to other things
like worker resident time.

The magic factory, raw materials in, finished goods out. Sit back and just enjoy the results.
Why do the accountants feel it is their responsiblity to peak inside the factory
and help decide how and when it is best to paint something?
(It will take me 2 hours to do this, 4 if you help)
Why are work cell managers held accountable for their efficiencies in isolation of what
goes out the back door?
(We are not a polishing company we are a shiny widget builder, to evaluate us based
on our polishing skill is rediculous. The gleam on our widgets is second to none. And we can have
your widgets ready for you fast and high quality, how many can I put you down for.)

The software dosen't exist becasue it wouldn't sell, because the scheduling idea for a vast
majority of the world is cost accounting based, not finished product efficiency based.
Operational measurement is cost accounting based not bottleneck erasure based.

I'm not following you at all, John, perhaps because I know next to
nothing about production. Are you saying that these sites Google is
bringing up are bogus? The software they sell doesn't work? Or just
that it doesn't apply in this case? Please elaborate. Dummy it down for
those of us who don't know the jargon.

 

[stoods]

Hi All,

Thanks for the responses.

I am perservering, and have had a measure of success. If I turn the
predeccesors around, so that the last task is the first in the chain of
predecessors, and then make the links start to finish, I ma getting close.
The only issue with this is that when I set the tasks to start as soon as
possible, it is making the start of the chain, ie now the last task, start on
the project start date. This then has the effect of making all the tasks
after it (in realty before, becuase we have start to finish) start BEFORE the
project start data. So, project is putting the tasks in the correct order,
but the dates now don't work. Any ideas?

Thanks,

Mike

 

[stoods]

Hi All,

This is becoming an interesting thread.

I have had a small exposure to AXAPTA. The guys who were running it were
accountants, it must admit that their perspective of how a factory makes
items lost me. Instead of moving from one work centre to the next, they were
moved in and out of ledgers, in and out of various accounts, and I am afraid
that the whole description of the process lost me.

So, I really got the impression that these type of systems did not really
model what was happening on the floor.

Yes, it is true that there are various external factors that affect a
project as it moves through a factory. In our case we have our own system
that models these, we just need to put the model in to project so that it can
resource level for us. I am not inclined to reinvent the wheel. Project
certainly has the inputs we can work with, and the data model is broad enough
for us to work out various inputs, it just needs to be able to level
resources more flexibly. The limitation of project in this model is that if
you set the tasks to start as late as possible, it just does not resource
level, it only resource levels if you set the tasks to start as soon as
posisble. This is really ignoring a complete sector of project managers who
need to level to for just in time outcomes.

I have done a search in google on "just in time". Yes, I agree that their a
lot of systems out their, but I suspect that thay all work along the
accountants view of a factory, not the production managers view of the
factory. It is intersting to see that these systems proudly list clients such
as Boeing, Toyota etc etc. Their price tags show it. Project is around $1500,
add a 0 to this and you don't even get close to the starting point of these
systems. Also, I suspect that their is a fair bit of effort required in just
making these systems run.

So, if their is any system out their that does not require all the overhead
of BOM, etc etc, and performs as a resource leveller better than project, I
would welcome advice as to what they are.

Thanks,

Mike

 

[John Sitka]

These guys used to have an addin that would invert the dates
http://www.prochain.com

start BEFORE the project start data (sp? date)

yes this happens really frequently, it shows that what
you are trying to accomplish is impossible but only as you have modelled
and estimated it though.

 

[Jan De Messemaeker]

Hi Stoods,

I feel this is genius.
Can't you put a formula in a custom start, finish and/or date field that
shows the correct date?

 

[John Sitka]

Yet more disease,

http://www.microsoft.com/dynamics/sl/product/pmademo.mspx