Possiblity excel to ED

[ditmar]

Hello,

I am trying to make a simulation of a weighing bridge for a company.
I have all raw data for this simulation which contains the following aspects.

All arrival times for the whole year ( incoming and outgoing ) because both have to pass the same weighing bridge.
The arrival times are the times that a truck driver receives his badge to continue to weighing bridge.
Weighing in or outgoing takes average 1min15sec. it is neg. exponentiel distributed.

However I want to know if its possible to put all raw data in ED without calculating the inter-arrival times myself because when you take this data over a whole year it is very hard to get the right distribution.

I have included my excel doc for your information, any help would be very appreciated.

Best regards
Ditmar

[Nick]

Hi Ditmar,

Enterprise Dynamics is indeed capable of reading data from excel. To do so, you can use the ExcelActiveX atom. (Available under 7. DATA). This is explained in the example model ExcelActiveX.mod (Examples > English > Active X Samples > ExcelActiveX.mod)

Instead of using the Excel Active X atom you could also:
1. copy all values to a table and model the interarrival times exactly as they occured in the real system.
2. create an empirical distribution.
3. find a statistical distribution. To do so, use the autofit (available under tools).

Regards, Nick

[ditmar]

Inkomend data !.xlsx

(948.87 KiB) Downloaded 289 times

Hi Nick,

Tried and tried over and over again, but ED can't put my excel file in a proper way in her own tabel. Also tried the second option. But my data is over 365 days and all have there arrival times so you can put them in classes but it gives the wrong result. Hard to explain via this way.

I attach my data file maybe it is better for you to understand.

Inkomend data !.xlsx

(948.87 KiB) Downloaded 289 times

Best regards

Ditmar

[ditmar]

Hello,

Does anybody else knows how to convert this in to ED, cause still don't find the way to do this!

Regards,

Ditmar

[Nick]

Hi Ditmar,

Based on your latest posts I think you can use the ExcelActiveX atom.
Just mind that your input data is in hours/minutes/days and that ED calculates in seconds.

A small example of my thoughts:
If you choose 1/2/2013 0:00:00 to be time 0 (excel value 41276),
then 10/24/2013 14:11:00 (41571.59097) is 41571.59097 - 41276 = 295.5909722

As ED calculates in second you have to convert the number of days to seconds

Code: Select all

hr(295.5909722 * 24)

Regards, Nick

[ditmar]

ok thankx will try that

regards

[ditmar]

Hi Nick,

I worked with your last replie. But now I have the following. What is the best way to integrate a 30% growth rate if you are moddeling inter arrival times. I don't see any options to do this when your basically moddeling your exactly as in the excel file.

So I calculated the Interarrival times and have seen they follow a Gamma Distribution.

So my question is can I do the 30% growth via your earlier explained way or do I need to just simply set up the Gamma distribution ? Also what are two parameters which need to be filled in after Gamma. Is it alpha and beta ? I cant understand how it can model if you only have alpha and beta and no actual vallues.

Thanks in advance

Regards
Ditmar

[ditmar]

anyone ?

regards

Ditmar

[marlies]

Hi Ditmar,

I had a look at your excel data and suggest the following approach:

- Define the distribution for the number of orders arriving per day. This distribution can easily be increased with a grow factor or 30%, just multiply the number by 1.3.
- Generate all orders for the day at time is 0.
- Then let the products enter your system in an appropriate order. It is very likely that the distribution for the inter arrival times is negative exponentional (which is a specific type of the gamma distribution), negexp is typical for indepent arrivals.
- You'll get a negexp interarrival time by distributing the arrivals uniform over a certain time period. E.g. let them arrive at time is uniform(0, hr(24)) will distribute the products randomly over the day, and the interarrival times will be negexp.
- In your excel data I saw that there is a typical pattern: during night hours only a small amount of orders arrive, there is a peak in the morning and then it's more or less uniform during day hours. I translated that into an empirical distribution that can be used to define the hour of arrival for each product. Each product gets an arrival hour in the exittrigger of the arrivallist.
- If a product gets arrival hour 6 for example, it will leave the multiserver atom between hr(5) and hr(6).
- So in this case, after the multiservice the products will go into your model.

Find attached my example to model this problem. If found the distribution for orders per day (in the arriallist) by using the autofit tool. See the table for the growth factor.

Just post againg if anything is not clear!

Regards, Marlies

[ditmar]

Hai Marlies,

This is very very helpfull.

I have some relating questions:

1. Is it possible to switch the grow factor on/off. Like when i adjust the table from 1.3 to 1 is it than back to normal or do I have to adjust the whole formula, and in what kind of formula ?

2. In the multiservice atom: Is it there you want me to let them arrive in the following formula: uniform(0, hr(label([24], i)-1), 0(label([24], i) ) ) ? Or do I need to adjust that ? don't understand this one.

3. Could you explain how to use to autofit option, because when I understood it well. My model would look like
arrivallist > multiservice > queue (With 15 cars capacity) > server(wheigingbridge) > Sink
And I have done some random tests of the server times which also are distributed most probably in a normal distribution, so how can I use the autofit option to model this?

4. When I run my model with some random figures in the server etc. I get a popup which states: fucntion 0 is no longer a valid function ? Which after I cannot close the popup.

Thanks in advance

Best regards

Ditmar