Dear All,
I am currently working on an assignment using Vensim PLE, modelling professional services allocation in a software firm. I am facing a '0/0' problem I was hoping someone could help.
One of the important variables is Current Product Installation Time which is Backlog/Installation Rate. In order to avoid backlog running negative, I am using the suggestion in Business Dynamics (Sterman), ch. 14, whereby a 'Maximum Completion Rate' is defined as Backlog/Minimum Delivery Delay.
The problem is that when Backlog is zero, the max completion rate is zero and, hence, the Installation Rate [MIN(max completion rate, and an endogenously defined rate) is also zero. Therefore, the installation time overfloats.
I thought about using a IF, THEN, ELSE to work around (if backlog is zero, then product installation time is zero, else calculate), but, does not look like a good SD practice.
Any ideas are really welcome! Thank you and congratulations for having this forum.
Regards, Daniel
Calculating Delivery Delay when Backlog is Zero
-
- Senior Member
- Posts: 1107
- Joined: Wed Mar 12, 2003 2:46 pm
One approach -
Product Installation Time = XIDZ( Backlog,Installation Rate,Minimum Installation Time)
(note the terminology change, mixing up delivery delay and installation time is likely to be confusing).
This actually has the problem that if the Backlog is positive, but the installation rate is 0 it will give a small time, when it should give a large time. Using IF THEN ELSE will guard against this, but the formula also becomes more complicated if you want to consider both cases, so the XIDZ is still probably the best choice.
Product Installation Time = XIDZ( Backlog,Installation Rate,Minimum Installation Time)
(note the terminology change, mixing up delivery delay and installation time is likely to be confusing).
This actually has the problem that if the Backlog is positive, but the installation rate is 0 it will give a small time, when it should give a large time. Using IF THEN ELSE will guard against this, but the formula also becomes more complicated if you want to consider both cases, so the XIDZ is still probably the best choice.
I think one question is what you're doing with the result - if a long delivery delay is driving away customers, the perception process might operate differently than if it's an internal management issue. Plant managers presumably know the state of the system regardless of output level, whereas a long delay for few or zero customers is of less consequence, because few people will hear about the situation.
Tom
Tom
/*
Advice to posters (it really helps us to help you)
http://www.ventanasystems.co.uk/forum/v ... f=2&t=4391
Blog: http://blog.metasd.com
Model library: http://models.metasd.com
Bookmarks: http://delicious.com/tomfid/SystemDynamics
*/
Advice to posters (it really helps us to help you)
http://www.ventanasystems.co.uk/forum/v ... f=2&t=4391
Blog: http://blog.metasd.com
Model library: http://models.metasd.com
Bookmarks: http://delicious.com/tomfid/SystemDynamics
*/
Hi Tom,
The result (average product installation time) feeds back to a loop related to product attractiveness. The idea is to capture the effect of the time it gets to install the software on the willingness of the customer base to provide references (sort of a social diffusion reinforcing loop).
Thanks, Daniel
The result (average product installation time) feeds back to a loop related to product attractiveness. The idea is to capture the effect of the time it gets to install the software on the willingness of the customer base to provide references (sort of a social diffusion reinforcing loop).
Thanks, Daniel