N/A
Posts: 0

# proc netflow constraint

hi,
im using proc netflow to minimize the cost of transferring water from the source to the consumers.
example: i have 1 source passing through 3 connections to 3 consumers. with a supply of 1000 cubic meters and demand of 300 cubic meters for each consumer. Each connection is subject to a water loss with coefficient of 3%.

supply to connection1
supply to connection2
supply to connection3
connection1 to consumer1
connection1 to consumer2
.
.
.
connection3 to consumer2
connection3 to consumer3

how can i code the water loss constraint? like if you will deliver 100 cubic meter of water going through connection1 then you can just supply 97 cubic meter of water..how can you incorporate this in the optimization?

thanks!!hope someone can help me..

regards,
paeng
N/A
Posts: 0

## Re: proc netflow constraint

Posted in reply to deleted_user
Since I am not that familliar with PROC NETFLOW I will just keep my comments to methods to deal with this in the traditional NA method.

If your network is simple enough, which it sounds like it maybe [i.e. every path experiences the same loss rate] then you can "Gross Up" your demand [i.e. if you know each and every path is going to have a 3% loss and your demand is 100 then adjust your demands to [100/0.97]=103.0928

However, a more robust way to do it is to have each path have a dummy sink which is constrained to demand loss from the main connection, this way it doesn't matter if the rates are diffferent. The idea being each real line flows to two sinks, 97% to the real one [consumer] and 3% to the virtual one [loss]

Just some thoughts,

Ike Eisenhauer
N/A
Posts: 0

## Re: proc netflow constraint

Posted in reply to deleted_user
Hi Ike,

thanks for the help..can you please give me a sample code for the inclusion of the dummy sink and its constrain if its possible..im new to optimization..

Thanks!

Regards,
Paeng
N/A
Posts: 0

## Re: proc netflow constraint

Posted in reply to deleted_user
can i use a data set to represent all the sink nodes?
thanks again..
Discussion stats
• 3 replies
• 230 views
• 0 likes
• 1 in conversation