I am sure that a number of people have found the forum useful (and will continue to do so).
If you look back, you will see that most questions are answered quickly. If you haven't had a response to a question, it just means that those of us who do answer are a little busy. I am sure you understand that.
All the best,
Tony.
Point of this forum
point of this forum
Hi Nik
I always try to help people, as I think that if you want to be helped sometimes, one has to help in return.
But there are four cases where I do not reply.
First when I do not feel competent.
Second if I feel that the question is somewhat absurd.
Third if I notice that the user never answers to the persons that kindly tried to help him.
It seems to me that the least that one can do is to reply by a short thank's, showing that everything is now all right!
It is sadly not always the case!
Fourth when I have not the time.
In your case, I think that a number of time steps of more than 8000 is absurd in the context of SD.
Maybe it is not SD that you must use for your problem, but I cannot imagine a case where such an amount of time steps can be useful to something.
Probably the administrators know better than me!
Regards.
JJ.
I always try to help people, as I think that if you want to be helped sometimes, one has to help in return.
But there are four cases where I do not reply.
First when I do not feel competent.
Second if I feel that the question is somewhat absurd.
Third if I notice that the user never answers to the persons that kindly tried to help him.
It seems to me that the least that one can do is to reply by a short thank's, showing that everything is now all right!
It is sadly not always the case!
Fourth when I have not the time.
In your case, I think that a number of time steps of more than 8000 is absurd in the context of SD.
Maybe it is not SD that you must use for your problem, but I cannot imagine a case where such an amount of time steps can be useful to something.
Probably the administrators know better than me!
Regards.
JJ.
Morning Laujjl
Thanks for your response. However, I don't think that 8000 time steps is absurd, because it merely equates to 8000 hours (one year). As Vensim has the capability to have time steps of seconds, it would seem reasonable that Vensim expects its users to work in these short intervals so, as a consequence, I would expect a far higher number of time steps than 8000.
Thanks
Thanks for your response. However, I don't think that 8000 time steps is absurd, because it merely equates to 8000 hours (one year). As Vensim has the capability to have time steps of seconds, it would seem reasonable that Vensim expects its users to work in these short intervals so, as a consequence, I would expect a far higher number of time steps than 8000.
Thanks
number of time steps
Hi Nick
I prefer to talk about concrete problems than about generality, because problems are always different.
So to explore your problem it would be better to expose it in a more concrete manner, for instance your model or something that is like it.
The 8000 time steps is just an upper limit that is probably never used. I still maintain that the need of such a high number of time steps may be the consequence of a badly defined problem. But to prove it, I must have an idea of your problem!
Regards.
JJ
I prefer to talk about concrete problems than about generality, because problems are always different.
So to explore your problem it would be better to expose it in a more concrete manner, for instance your model or something that is like it.
The 8000 time steps is just an upper limit that is probably never used. I still maintain that the need of such a high number of time steps may be the consequence of a badly defined problem. But to prove it, I must have an idea of your problem!
Regards.
JJ
Laujjl - apologies for the late response - I've been on a course. I don't understand your point about 'concrete' problems as opposed to 'generalities' because the principles are the important matter - not specifics. However, I will explain my problem a bit more.
We have a training device that has a specific training pattern, throughout the year (the training programme HAS to be in hours). Linked to this device, we have other training devices that have there own specific patterns. Whenever a device fails catastrophically (when linked), it will stop the operation of all the linked devices. Now, the failures will be of Poisson distributions types and when they fail, they will be repaired (by a variable number of hours). What I need to know, is the exact number of 'downtime' hours. I am therefore trying to run the simulation over one year (8760 hours) and looking at the result at 8760 hours - but off course I can't because of the limitation.
Regards
We have a training device that has a specific training pattern, throughout the year (the training programme HAS to be in hours). Linked to this device, we have other training devices that have there own specific patterns. Whenever a device fails catastrophically (when linked), it will stop the operation of all the linked devices. Now, the failures will be of Poisson distributions types and when they fail, they will be repaired (by a variable number of hours). What I need to know, is the exact number of 'downtime' hours. I am therefore trying to run the simulation over one year (8760 hours) and looking at the result at 8760 hours - but off course I can't because of the limitation.
Regards
Point of this Forum
Hi Nick
About the specificities, I mean the specificities of the problem.
If only generalities mattered, you would find on the market plenty of SD model already made with parameters for the specificities. This is not the case and I think this is one of the reasons of the difficulty of SD to be more used. Bob Eberlein once told me that most of the time the clients of the Ventana consultation division had different problems and that for this case they had to start a new model for them.
There are some archetypes but only at the simplest level like in molecules.
About your specific problem, it seems more to be a discrete problem that is studied in the mathematical queuing theory: stochastic processes.
You need to study your problem software where the time is continuous and the events discrete.
The events will be represented with a poison distribution and the repair eventually with an exponential distribution and you will not need to slice the time in a multitude of hours.
Your problem may eventually too be solved without simulation, with formulas that you find in the mathematical books.
But I think that pure discrete software whose main capability is to deal with queues, may be more adapted for this situation than continuous simulation.
You can of course study the problem with Vensim, but you will have to adapt it to the characteristics of the software that is mainly continuous.
Regards.
JJ
About the specificities, I mean the specificities of the problem.
If only generalities mattered, you would find on the market plenty of SD model already made with parameters for the specificities. This is not the case and I think this is one of the reasons of the difficulty of SD to be more used. Bob Eberlein once told me that most of the time the clients of the Ventana consultation division had different problems and that for this case they had to start a new model for them.
There are some archetypes but only at the simplest level like in molecules.
About your specific problem, it seems more to be a discrete problem that is studied in the mathematical queuing theory: stochastic processes.
You need to study your problem software where the time is continuous and the events discrete.
The events will be represented with a poison distribution and the repair eventually with an exponential distribution and you will not need to slice the time in a multitude of hours.
Your problem may eventually too be solved without simulation, with formulas that you find in the mathematical books.
But I think that pure discrete software whose main capability is to deal with queues, may be more adapted for this situation than continuous simulation.
You can of course study the problem with Vensim, but you will have to adapt it to the characteristics of the software that is mainly continuous.
Regards.
JJ