Delays in human population: baby, child, teen, adult, marry, old

[Anilda]

Dear vensim experts and users,

I'm a new user in system dynamics field. To learn system dynamics
i've download and installed vensim PLE version. I've still don't understand
enough about delays in a model so far. That's why i need helps from you all.

For example, i still don't understand how to set some delays for human population
problems: baby (0-3), child (4-10), teen (11-18), adult (19-50), and old (51-100).
married ages (19-50), have a baby (19-50), and so on.
how to build stocks & flow diagram for that problem? Any vensim sample file that
close enough to that problem?

Thank for your help.

Regards,

Anilda

[tomfid]

There are some population examples in the User Guide and elsewhere in the help system. See also the SHIFT IF TRUE function and
Fiddaman, Thomas with Larry Yeager Dynamic cohorts: a new approach to managing detail at https://www.systemdynamics.org/assets/c ... index.html

[LAUJJL]

Hi Tom

I downloaded the zip archive file from the 2017 conference. but the vensim model lacks a .vdf file named 'JapanCohortControl' that appears in the graphs.

I could not find either Bob Eberlein's model that makes the continuous calculation mentionned in the paper. When you consult the SDR review there is no supporting model of the paper.

Regards.

JJ

[tomfid]

The earlier version of the Eberlein-Thompson paper can be downloaded at https://www.systemdynamics.org/assets/c ... index.html - that may be the version I used. Unfortunately the cohort control external function seems to have a bug or incompatibility with current Vensim. But the other structures are still useful.
https://www.systemdynamics.org/assets/c ... index.html
Incidentally, the dynamic cohorting approach in Ventity can be done in Vensim as well - you just have to waste some array space.

[LAUJJL]

I attended Bob's presentation at the St Gallen SD conference of his continuous ageing chain model.

I am interested by the possibility to describe things differently with Ventity as I have plannings in my models that I have to shift every day and that are stored in arrays. If you say that it can be represented too in Vensim, this makes Ventity less useful. But it may still be easier to understand and to manipulate, my problem being not a problem of capability to model these ageing chains, but the complexity generated and the difficulty to represent it mentally or to build intelligent policies.

Regards.

JJ

[tomfid]

Hi JJ -

It's definitely simpler to represent such things in Ventity, but it's probably also a lot of work to convert your model.

In a Vensim aging chain, you have a dimension representing time:
age: age 0 to 1, age 1 to 2, ... , age 100+
Then you have to move stuff from one element to the next every year (or continuously).

An alternative, analogous to Ventity, would be to have cohorts:
cohort: born in 1965, born in 1966, born in 1967 ...
Then you need a pointer to the current cohort where births are directed, but you never need to move a person from one to another.

The downside in Vensim is that you simulate a lot of 0s, because the "born in 2099" cohort has to exist when you start the model in 1999, and you can never get rid of the "born in 1720" cohort even though everyone's dead. However, Vensim is so fast that this isn't a huge burden.

Taking a wild guess, in your case, you might change:
plan : 30 days out, 60 days out, 90 days out
to
plan : Jan, Feb, Mar, Apr ...
Then the plans would have to know how much time remains to their implementation, but you wouldn't have to move plans around.

Tom

[LAUJJL]

Hi Tom

Thank you for the explanation.

To avoid moving stuff between levels you kill old levels and create new ones.

I use arrays to store vehicles availability in different towns (around 50) in a 60 days horizon.

I shift the horizon availability every day for each town, with a 1 / 512 per day time step.

It is possible to have a stable horizon, each dimension value representing a stable date, with no shifting. But it will increase the data storage. I already need a savelist with a total time of 180 days with a 60 days shifting horizon. If I do not shift, it will increase the size of the .vdf file.

There is another solution: keeping the trace of every discreet vehicle flow: a flow has a starting town, an end town, a starting date and an end date or a duration (between 1 and 3 days). But I need to know the availability of each town during the next 60 days (the availability can be negative). A negative availability generates a risk of having to transfer a vehicle from another town at horizon 0.

And I am obliged to use discreet values because there can be from one vehicle up to 6 vehicles available per town.

If I store the flows, I am obliged to rebuilt the entire planning for the starting town and end town, because I use it to optimize my prices based on the transfer risk and many other parameters, avoiding the planning being too negative. This rebuilding process will be highly expensive in calculation.

There is maybe a solution in Ventity. I will take the time to study the population example.

Is Ventity compiled or only interpreted?

Regards.

JJ

[tomfid]

Ventity is essentially always compiled, but the big speed differences occur where there are architecture differences between models.