Problem in Reconstructing System Dynamics Models

[anjomshoae]

Hi,
I have a problem in rebuilding a system dynamics model that I found in a journal article. Unfortunately, many of the equations in the paper are not supporting all the variables in the model. So in an effort to rebuild the model, I end up with a handicapped model that equations are partially completed. I write this post to seek for helpful comments regarding this issue. I have attached a paper and a Vensim model of this paper to this post as an example. In this particular model, the author did not include all the model equations in the paper.(Equation 2 and 5)

Best,

[Administrator]

I think it's probably best to contact the author and ask them for a copy of the model.

[anjomshoae]

I think it's probably best to contact the author and ask them for a copy of the model.

Unfortunately, most of the time they are not willing to share their models. May I ask another question ? I hope I don't bother you !
In this particular model, what could be the equation for " Information delay generator" ?
the author mentions in his article something about "negative exponent". What is exactly a negative exponent in Vensim ?

I cite a part of this article :

"So first we use a negative exponent in generator to depict the main tendency of delay, its index points out how quick the level drops. Then we add a uniform random on it to depict the fluctuation. These two generators have the same structure, one for the Transport Delay, the other for the Information Delay."



I attached a picture of this structure.

[Administrator]

My guess is that "Information delay generator" is a random number.

http://www.vensim.com/documentation/ind ... random.htm

[anjomshoae]

My guess is that "Information delay generator" is a random number.

http://www.vensim.com/documentation/ind ... random.htm

SO I managed to find out what is this information delay generator. It is a negative exponential function, and it is uses the EXP + LN + LOG.mdl in FunctionExamples. I attached the model to this post.

another issue is that, in this model, two variables (Average Lead Time, std Lead Time) has input of (INITIAL TIME, TIME STEP, Time, Feedback Lead Time ). these variables are colored in green in the model. What could a possible equation for calculating average lead time and std lead time ? any idead?

Thank you,
Best


P.S; Model is attached to this post.

[Administrator]

I cannot help on this, I don't know.

Where did you get the model? The author hasn't bothered with units, or any documentation. And a very quick look shows up obvious errors (the equation for "supplies consumed" is incorrect).

[anjomshoae]

I cannot help on this, I don't know.

Where did you get the model? The author hasn't bothered with units, or any documentation. And a very quick look shows up obvious errors (the equation for "supplies consumed" is incorrect).

The model is from a published paper (attached to this post (File number 1) ). The author has based his other published papers on this model. So its looks like a complex model!

I have corrected the information delay generator and transportation delay generator, and they are working fine. However, other equations are not correct, that's why it's not performing correctly. I attached the last version of this model to this post. The behavior of information delay generator and transportation delay generator is in accordance with the results in the paper. But I am trying to figure out what is the right equation for variables in the model, since the author has not revealed the equations fully, it is difficult to rebuild it.

P.s. the model together with papers attached here.

[Administrator]

Why do you need this? And have you tried contacting the authors?

There are errors in this model, it doesn't even pass the most basic of tests, and it is undocumented. The model is also very sensitive to time step changes (change the time step within reason and the results should be very similar, the results from this model are very different). If you take a quick look at the model, the smallest delay I can find is "1" (I think it's a day, but no units so not sure). The time step should be at least 1/3 of this (see http://www.vensim.com/documentation/ind ... e_step.htm). Running the model with TS=0.25 gives very different results to the baseline. So what are the correct results?

[anjomshoae]

I want to run this model because it probably helps me in my own research.Yes, I've contacted the author and I didn't receive any response. So do you think the model is wrong?!

[Administrator]

Yes, I think the model is wrong (I've edited my last post to give you an example using time step).

Maybe I am just not using it correctly.

[anjomshoae]

Yes, I think the model is wrong (I've edited my last post to give you an example using time step).

Maybe I am just not using it correctly.

Thank you for your comments. They are really helpful. Let me read them carefully and I try to find out what's going on in this model.

Best,