MRP and Production Planning

This forum contains all archives from the SD Mailing list (go to http://www.systemdynamics.org/forum/ for more information). This is here as a read-only resource, please post any SD related questions to the SD Discussion forum.
Locked
Francisco Perez francisco77pp ya
Junior Member
Posts: 18
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Francisco Perez francisco77pp ya »

Posted by Francisco Perez <francisco77pp@yahoo.com>


Hi People!. I would like to know whether any of you have been involved
in modelling Planning processes in the manufacturing or service systems.
I would like to know what simulation strategy and approach can be given
to construct a simulation containing the Demand Planning, Production
Planning and Distribution planning together and to a not-so-aggregated
level. Any comment is valuable. Any comments on how to desaggregate to
deeper levels without loosing sense of SD Sim is highly valuable.

Thanks in advance.!

Francisco
Posted by Francisco Perez <francisco77pp@yahoo.com>
posting date Mon, 2 May 2005 11:27:19 -0700 (PDT)
Jean-Jacques Laublé jean-jacques
Senior Member
Posts: 68
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Jean-Jacques Laublé jean-jacques »

Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
Hi Francisco


I have never seen any model or paper dealing with planning and utilization
of resources.

I have worked for about three years with planning the utilization of our car
fleet on a day to basis trying to 'optimize' our price setting. The problem
is the level of aggregation.

In my case the demand has a lot of characteristics that have all a big
influence on the level of the demand and on the price elasticity. So the
more you disaggregate, the smaller the samples and the less valid the
statistics. On the contrary if you do not disaggregate enough the price
elasticity has no more meaning. The demand can be better disaggregated than
the price elasticity. The model was finished, but whatever the level of
aggregation of the demand, it was somewhat bugged principally by price
elasticities.

I will probably go back to the causal loop diagram, and work at this level
first, keeping the model to a pure qualitative level with loop polarity and
if the node is a level, a rate or an auxiliary, and trying to get the most
of that qualitative model first and see how I can move step by step to an
eventually partially quantitative model.

I probably made a mistake common to many beginners, jumping to quickly to
the more powerful and sophisticated tool before trying first to explore the
problem qualitatively first.

This is the first advice I could give although it might look evident.

It has to do too with the discussion on optimization, where Alan Graham
writes

that one must take optimization as not definitive. I think it is the same
with quantification.

One should first work on qualitative models and once one has found
everything possible from that step, eventually go further on.

On top of that qualitative models cost only a fraction of a quantitative
model (for my case 5% for the qualitative and 95% for the quantitative).

And you are not restricted by a problem of size. You can extend the boundary
of the model to a level where it would be not manageable quantitatively.

In short the more precise you are (the more quantitative), the more you have
to make hypothesis and the more you take the risk to err from reality by
narrowing your model and being obliged to value the influences even if you
do not know how to do it.

Hope this helps a bit.

Regards.

J.J. Laublé. Allocar

Strasbourg. France
Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
posting date Tue, 3 May 2005 15:03:22 +0200
Carlos Alberto Brandão brandao
Newbie
Posts: 1
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Carlos Alberto Brandão brandao »

Posted by =?iso-8859-1?Q?Carlos_Alberto_Brand=E3o?= <brandao@correios.com.br>
Francisco,

Regarding the desaggregation issue, you could use Agent-Based Computational
Economics (http://www.econ.iastate.edu/tesfatsi/ace.htm).

It is still uncharted territory for me. I have just started studying this
subject, but - from what I've already read - it seems one way to suit the
necessity for a deeper desaggregation level. At the same time, the ""without
loosing sense of SD Sim..."" might be answered by a software like RePast
(http://repast.sourceforge.net/) which is claimed "" to include built-in
system dynamics modeling"".

Posted by =?iso-8859-1?Q?Carlos_Alberto_Brand=E3o?= <brandao@correios.com.br>
posting date Tue, 3 May 2005 16:35:32 -0300
Justin Lyon justin1028 yahoo.com
Junior Member
Posts: 17
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Justin Lyon justin1028 yahoo.com »

Posted by Justin Lyon <justin1028@yahoo.com>
Jean-Jacques,

You make a good point about the power of qualitative
modelling. I think that developing the CLD and then
turning that into a architecture might provide you
with a nice intermediate step between the purley
qualitative CLD model of your situation and the
quantitative SD model. I use to do CLD development and
now use resource mapping ideas to develop
architectures that are more easily translated into
simulation models than CLDs. Plus, the architectures
seem to make sense to non-analytical managers and
executives who hate diving into detailed mathematical
models. Once you teach them how to read the
architecture (which usually only takes a few days of
work), then it seems like the light bulb goes off over
their head and when you start doing quantitative
simulation, they can see the connections from the CLD
to the architecture to the computer SD simulation more
easily. An added benefit is that I've seen amazing
insights come out of the resource mapping exercises
without ever going to time and expense of making a
computer simulation.

Best,
Justin
Posted by Justin Lyon <justin1028@yahoo.com>
posting date Wed, 4 May 2005 08:54:04 -0700 (PDT)
Alan Graham Alan.Graham paconsul
Junior Member
Posts: 11
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Alan Graham Alan.Graham paconsul »

Posted by ""Alan Graham"" <Alan.Graham@paconsulting.com>
In answer to Fransisco Perez's question on studies of MRP and
production planning:

The System Dynamics literature has a major preoccupation with
production planning and stability. Doubtless it starts with Jay
Forrester's Industrial Dynamics book itself, with the many chapters
devoted to a production-distribution model, and many appendices
devoted to various aspects, e.g. advertising and forecasting.
Geoff Coyle's System Dynamics Modeling book likewise devotes several
chapters to a production distribution model. John Sterman's
Business Dynamics book has pretty thorough treatment and numerous
references. (By the way, that's very generally true--SD list
subscribers should always check Sterman before asking ""does anybody
know about X?"". Also, I'd expect the System Dynamics Review to have
published numerous articles on such topics over the decades, but I
haven't personally collected the references.)

For a look at somewhat more detailed issues, check out John
Morecroft's PhD dissertation at MIT's Sloan School of Management on
production planning and the impact of MRP systems on stability.
Such theses are typically available online. Google MIT.

Cheers,

Alan

Alan K. Graham, Ph.D.
Decision Science Practice
PA Consulting Group
Alan.Graham@PAConsulting.com
One Memorial Drive, Cambridge, Mass. 02142 USA
Posted by ""Alan Graham"" <Alan.Graham@paconsulting.com>
posting date Wed, 4 May 2005 15:22:14 -0400
Bill Braun bbraun hlthsys.com
Member
Posts: 29
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Bill Braun bbraun hlthsys.com »

Posted by ""Bill Braun"" <bbraun@hlthsys.com>
>From the pen of: Justin Lyon justin1028 yahoo.com

>> You make a good point about the power of qualitative
>> modelling. I think that developing the CLD and then
>> turning that into a architecture might provide you
>> with a nice intermediate step between the purley
>> qualitative CLD model of your situation and the
>> quantitative SD model.


I've moved to starting in between, with a mostly CLD approach but
explicitly identifying stocks. It is a slightly longer learning curve for
those who have never thought in causal loop influnce terms. Two simple
examples seem to effectively get them over initial puzzlement: first is
Income Statement as Revenue [Inflow], Expenses [Outflow] and Net Income
{Stock/Accumulation]; second is the well known number of people in a room
being the net accumulation of everyone who has ever entered net of
everyone who has ever left. With these examples in hand, most catch on
pretty quickly.

Bill Braun
Posted by ""Bill Braun"" <bbraun@hlthsys.com>
posting date Thu, 5 May 2005 10:38:35 -0400 (EDT)
Jean-Jacques Laublé jean-jacques
Senior Member
Posts: 68
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Jean-Jacques Laublé jean-jacques »

Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
Hi Justin.



Thanks for answering my e-mail.

Can you give some references about your architectures based
on resource mapping ideas, like articles, books, software, web site?

My principal difficulty but I think it must be common to most SD
modelling, is to formulate a question that will if answered properly
generate a sufficient pay back, will take a reasonable time and effort
to solve it and verify that the pay back will exceed without any doubt the
cost of the effort involved.

This sounds rather evident, but I think that it is right at the beginning of
the modelling effort (the definition of the problem and its formulation)
that lies the potentiality of a profitable or not profitable model.
And it is at this time of the modelling process that imagination is the most
needed.

For instance I can formulate my problem of price setting as:
What is the best price I can charge a customer taking into account
all the other factors: seasonality, resource available, type of customer
(with potential regular needs or not), type of car needed, propositions of
the competition etc. (lots of etc.)

When faced with a too complex question, I can simplify and narrow it like:
What will a price discrimination between regular or not regular customers
have on the level of activity and on profit?

Seen that way I can formulate hundreds of other questions like this.
They may have the particularity of being easier to model and still have a
sufficient
pay back.

So my first concern is to ask questions whom the payback and cost are
relatively easy to evaluate, because I manage my own personal planning
(with a very simple SD model) strictly on the basis of the profit / cost
ratio.

And it is at this time of the modelling process that imagination is the most
needed.

Regards.

J.J. Laublé Allocar
Strasbourg France.
Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
posting date Thu, 5 May 2005 16:11:24 +0200
Thompson James. P (Jim) A142 Jim
Junior Member
Posts: 8
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Thompson James. P (Jim) A142 Jim »

Posted by ""Thompson, James. P (Jim) A142"" <Jim.Thompson@CIGNA.COM>
Alan Graham wrote: ""For a look at somewhat more detailed issues, check
out John Morecroft's PhD dissertation at MIT's Sloan School of
Management on production planning and the impact of MRP systems on
stability. Such theses are typically available online.""

Better yet, for less than the price of one printed thesis, those
interested can purchase 200 theses on system dynamics in The MIT System
Dynamics Group Literature Collection on DVD. Forty-one are doctoral
works including John Morecroft's prescient ""Influences from Information
Technology on Industry Cycles"". The others are mostly master's level
texts.

The Literature Collection is a remarkable resource and a remarkable
bargain. The DVD contains over 3,600 research memoranda (D-memos)
produced between 1956 and 2003, the entire content of the Road Maps
self-learning program, and a complete Guided Study program in PDF.
http://www.systemdynamics.org/MITCollectionDVDinfo.htm

Joel Rahn, George Richardson and I review the collection in an up-coming
System Dynamics Review.
Jim Thompson
Email:
james.thompson@strath.ac.uk
jim.thompson@cigna.com
Posted by ""Thompson, James. P (Jim) A142"" <Jim.Thompson@CIGNA.COM>
posting date Thu, 5 May 2005 11:18:18 -0400
Francisco Perez francisco77pp ya
Junior Member
Posts: 18
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Francisco Perez francisco77pp ya »

Posted by Francisco Perez <francisco77pp@yahoo.com>
Thanks Alan... I checked out all the references you've mentioned before
posting the message. Perhaps I should be more precise with my question.
I'm wondering if someone had any experience in modelling and simulating
with SD production planning, demand planning and distribution planning
to a not so high level as illustrated in the references previously
mentioned. It happens that when one wants to modell jus a bit deepr
complexity and model confidence is a real subject. If you or any of you
has been involved in real life projects regarding these questions, then
I would like to know what your approach was regarding the aggregation
issue, for example all SKU's (stock keeping units) cannot be aggregated
to 1 but perhaps to 4 or 5 (because dynamics are not equivalent and
smoothing or averaging is not particularly significant), the same happens
to distirbution phenomena and demand planning processes.

Thanks anyway.... anyother comments are welcome!

Francisco Perez
Posted by Francisco Perez <francisco77pp@yahoo.com>
posting date Thu, 5 May 2005 07:34:26 -0700 (PDT)
Jean-Jacques Laublé jean-jacques
Senior Member
Posts: 68
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Jean-Jacques Laublé jean-jacques »

Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
Hi Alan

You write that there are plenty of models dealing with
planning production processes. But unfortunately
I never saw any model dealing with planning services where there are no
inventories.

The main difference between service and production, is that as the time
is flowing, the service that is not sold is lost, while the product is still
there in
the inventory.

This makes the modelling of service systems completely different then
production.

In an economy that is more and more changing into an economy of service,
which is too a good way to protect oneself again the foreign competition,
the SD community should take the effort to investigate other domains then
the traditional backlog, production inventory, sales inventory that one sees
in all SD text books.

I wonder if it is not one of the reason why I never saw any use of SD in the
Journal of Revenue and Pricing management since its beginning about three
years ago, where is only used classical maths, probability and statistics. I
recently sent an e-mail to two of the authors of articles: Sheryl E. Kimes
from Cornell University, School of Hotel Administration and Matthias Fuchs
University of Innsbruck, Austria, asking them if they had considered using
SD in their studies. I did not even got an answer!

Regards.

J.J. Laublé Allocar
Strasbourg, France
Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
posting date Thu, 5 May 2005 16:57:32 +0200
Jean-Jacques Laublé jean-jacques
Senior Member
Posts: 68
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Jean-Jacques Laublé jean-jacques »

Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
Francisco

I asked a question called 'Aggregation techniques' the 16-1-2005 which
summarizes approximately your question. It generated many answers that you
can consult on the web site archives of the sd mailing list:

http://www.ventanasystems.co.uk/forum/

So not wanting to go back to a subject already discussed, I can write down
the different strategies that I Think can be used, to address partially the
problem, because there was no real solution to my question, although the
answers gave me new ideas.

The level of aggregation becomes problematic when you are facing units in
your stocks that behave very differently relatively to the problem you are
trying to explore.

In some cases working with averages has no meaning, because the average of a
function is not necessarily the function of the average. Average F(x) ~
F(average (x)).

When faced with this I devised different strategies.

1.. The first one is to evaluate the bias generated by the use of an
average, and eventually live with it.
2.. The second is to cluster the level studied into pieces that behave
more homogenously relative to your problem. This solution is not always
easy. In my case, I did not found a solution to disaggregate price
elasticities.
3.. Reformulate your question so that the behaviour of the elements in
the stock behave more homogenously relatively to that problem (Study the
behaviour of something else).
4.. Restrict your model to a limited subset of your stock that will behave
more homogenously. (Make different models for each subset)
5.. Eventually restrict the boundaries of your model. Making endogenous
variables exogenous may lessen the inhomogeneous characteristics of the
elements of the stock. Of course you will loose insights by narrowing the
boundaries.
6.. Instead of trying to find results, just work with a model with no real
data, but with plausible data, to learn about the general behaviour of
reality even if it is not strictly the reality. I think this should be the f
irst thing to do, to get a feel of the problem. (Explore possible
realities). It is always the fundamental question, are you looking for
results or understanding?
7.. Do a happy mix of the above solutions.
Any other ideas are welcome.

Regards. J.J. Laublé

Allocar Strasbourg
Posted by =?iso-8859-1?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
posting date Fri, 6 May 2005 16:37:00 +0200
Nijland lukkenaer planet.nl
Junior Member
Posts: 4
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Nijland lukkenaer planet.nl »

Posted by Nijland <lukkenaer@planet.nl>
Existing models of ""production of services""

Jean-Jacques Laublé Allocar wrote:

""The main difference between service and production,
is that as the time is flowing, the service that is not sold is lost,
while the product is stil there in the inventory.

This makes the modelling of service systems completely different then production"".


We remember that this was the famous thesis of Karl Marx,
who explained the loosing position of the labour class
against the class of the owners of the production sector, by this principle.

It is a real mechanism, indeed.



However with somewhat more abstraction, formally, there is not so much difference

between the production of services and the production of goods.

The model structure may be the same, only the parameters should be different.



Inventories of services may be inconceivable,

but inventories of people delivering services are conceivable.



Especially some high tech goods if not sold are also lost, just as services.

The ""specific relative loss fraction"" of unsold computers is very high.

The specific relative loss fraction of services is 1 (the maximum value).



I think that there are more SD models of service systems than suggested by Jean-Jacques Laublé Allocar.

20 years ago I worked on a model of ""dental health care services"" (Nijland et al., 1985)

Literature: Nijland, G. Klabbers, J., Plasschaert, A. Truin, G., 1985. Dynamics of Dental Care in the Netherlands: a systems dynamics approach. Proc. of the1985 International Conference of the Systems Dynamics Society, Keystone, Col.

I think that in the mean time several other models have been developed. I might be wrong.

May be the thesis of Vennix (and some later publications about the health care sector of him)

give information about the modelling of services with SD.

Publication: J.A.M. Vennix (Nijmegen University, 1990)
Mental models and computer models: design and evaluation of a computer-based learning environment for policy making.

Another interesting website is that of Jan Klabbers,

who wrote different publications about the integration of gaming and simulation,

and how to tackle decisionmaking problems in the flield of planning of services.

Geert Nijland lukkenaer@planet.nl
http://www.ifi.uib.no/staff/jan/publications-list.html
Posted by Nijland <lukkenaer@planet.nl>
posting date Sat, 07 May 2005 21:03:15 +0200
Jean-Jacques Laublé jean-jacques
Senior Member
Posts: 68
Joined: Fri Mar 29, 2002 3:39 am

MRP and Production Planning

Post by Jean-Jacques Laublé jean-jacques »

Posted by =?Windows-1252?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
Hi Geert



Thanks for your references.

Two remarks.

The particularity of non storability of services works too for some kind of
productions:

example to order productions. Generally people that work on order, have to
wait until they receive an order, and in the meanwhile they may have nothing
else to do unless they organize to produce something more standard that can
be stored.

Another example is people who sell or produce perishable goods. They face
exactly the same problem.

One can find other cases probably.

A second remark. When I wrote that I had not seen any model on services, I
meant, models that are well known, by text books principally or papers or
subjects discussed in forum.

I have unfortunately not much time to read academic or not academic papers.
I need to have pre-digested tools and examples to work with. Otherwise I
have to look for other solutions. My current work is to run a business. One
of the problem of SD too is that there are practically no already made
models available commercially as it exists in other fields and that can be
easily integrated in the current business work.

Your reference to Karl Marx is interesting and is still topical.

Regards.

J.J. Laublé Allocar

Strasbourg France
Posted by =?Windows-1252?Q?Jean-Jacques_Laubl=E9?= <jean-jacques.lauble@wanadoo.fr>
posting date Sun, 8 May 2005 15:11:29 +0200
Locked