Modeling an In Transit effect

Jack Harich · Post by **Jack Harich** » Wed Sep 15, 2004 9:23 pm

I must say I'm very impressed with Vensim. Great product. I've used it for a month now.

In trying to model an in transit effect concisely, I've come up with no good way. The purpose is of course to avoid overshoot and oscillation, and avoid the gimmicky approaches that merely come close. The preferred method seems to be as shown in Sterman's classic, Business Dynamics, page 415. The sample equation is:

Material in Transit = INTEGRAL(Inflow(t) - Outflow(t), Material in Transit(0))

In Vensim we certainly do have INTEG( ) but we don't seem to have the ability to refer to the value of a variable using Time, such as by using something like INTEG(variable(Time) - variable(Time - delay), 0), which would be the effect in transit if I'm thinking correctly.

I thought the doc on DELAYP might help, but no luck there. No luck on searching the web either, but did get a nice education.

Is there any documentation or literature out there that would address this?

Thanks,

Jack

[Edited on 9-15-2004 by Jack Harich]

Pruyn · Post by **Pruyn** » Thu Sep 16, 2004 7:52 am

If I understand you correctly you want to keep track of the in transit amounts. If there is a fixed delay between entering and exiting you could use the following:

In = Variable
Transit = INT(In - out, 0)
Out = delay fixed ( In, Delay,0)

As in attached model

Jack Harich · Post by **Jack Harich** » Fri Sep 17, 2004 11:55 am

Hi there,

Thanks for taking the time to think about this. What I've done is about what you've suggested, without having to setup variables for a stock and its in and out flows, which would greatly clutter the model. The in transit variable now has this equation:

in transit = intermediate increase - DELAY3I ( intermediate increase , delay , 0 )

An intermediate variable is necessary to avoid a circular reference.

With this problem resolved, I can get back to the real issue, a generic technology growth structure that can be reused in many subsystems. It handles investment in transit and a falling cost per tech unit, resulting in the classic exponential growth curve of technology without overshoot.

Thanks. I guess a model a day keeps the doctor away.

Jack

bob@vensim.com · Post by **bob@vensim.com** » Fri Sep 17, 2004 8:50 pm

Be sure to include and check your units. The definition of in transit you are using does not match any definition I know for that concept (which seems like it should be a level as in the previous post), so you may mean something different. Dimensional analysis should let you know if you have what you want.

LAUJJL · Post by **LAUJJL** » Sat Sep 18, 2004 3:52 pm

Hi

Did you try the delay conveyor for your in transit problem. It has many advantages.
It is a pipeline delay, and you can vary the transit time evey time step and specify the position of the stuff on the conveyor at the start of the simulation as well as inputing every time step a different amount of stuff in the transit level.
Regards.
JJ Laublé