Modeling decision making with inventory

[Hamilton]

Hi all.

I have been trying to model a laboratory game that simulates business processes (producing, selling and managing inventory). Typically, the decisions are taken after each period. My model is a very simple one (see attached - "model test2"). It includes a condition for market turbulence.

The problem is: I cannot simulate correctly the input from inventory into the decision of how much to produce ("effective production"). I tried to include a first order control that only sells what is available in inventory but it did not work. I tried several alternatives of first order controls (MIN function, for instance) but each one produced bizarre behaviors (such as having a huge inventory but only selling half of what the market demands). I suspect the problem is related to the nature of high turnover stocks (inventory, in the model, is filled and depleted simultaneously after each month).

My last model (see attached) does not control for negative inventory. I do want to include excess inventory in the decision of producing (i.e. only producing the net amount necessary to satisfy the market demand). I have also attached the previous version ("model test"). See that inventories skyrocket. In a real laboratory setting, players would stop producing after a while to let the inventory diminish. Does anyone have an idea on how to solve this problem?

Thank you.

Hamilton.

[Administrator]

I'd start by putting units into your equations. Once it passes the units check, It will be much easier to debug and figure out what is wrong.

For example, I'd assume "Invetory" is measured in something like "widgets". It looks like "changing" is also measured in widgets (it should be widgets/month as it's a flow) so this is an error and will give incorrect calculations. Units really help others to understand a model, and also show up the most common errors very quickly.

[Hamilton]

I tried to put all the units but failed regarding "actual demand" and "market turbulence".

Anyway, I am attaching an updated model. Why can't I meet the actual demand (or the estimated demand)? I have been losing my sleep over this problem.

Thank you.

Hamilton.

[LAUJJL]

Hi

Two advices:

check Change in model>settings>use stricte testing. It will be easier to understand your errors using stricte testing.

Transform all your data like 1000 (that is understood by Vensim as being DMNL and in fact is probably product/month) etc.. in your equations into constants and give them a dimension. It will make too your equations easier to understand.

Regards.

JJ

[Administrator]

It might be a good idea if you take a look at the beer game, there is a Vensim version at
https://forio.com/simulate/kishoremumma ... ame/model/

You are producing and selling high volumes, and the capacity is also changing rapidly. In most real world situations, this would not happen so quickly. Estimated demand lags behind demand as well, this will need looking at to see if you can produce a better measure. Maybe smooth the demand

[tomfid]

Your Effective Production decision is the problem. Your version takes the min of capacity or expected demand, then subtracts existing inventory (implicitly, this is a pressure to adjust toward zero).

I think you want a decision that does the following:
- replace expected demand
- adjust inventory toward a desired level, over some adjustment time
- not exceed capacity

One option might be:

Code: Select all

Effective production=
	min(Capacity,Estimated demand+(desired inventory-Available products)/Time to make decision
)
Units: products/Month

**************************************************************
desired inventory=
	Estimated demand*desired inventory coverage
Units: products

**************************************************************
desired inventory coverage=
	2
Units: Month

**************************************************************

That already works a lot better. However, given your test input, you might want to add a term that follows a trend or expects seasonality.