Hmmm… I think I see what you are saying but let’s extend your analogy a little. I beg your indulgence just a little longer!!!
Suppose my communications system has three types of user and, according to my customer there are 100 users in total.
1. A fixed land-line user
2. A mobile cell phone user
3. A UHF Radio Transmitter user
Let’s suppose that the land-line users need to be able to talk to other land line users and also mobile users. Similarly, the mobile users need to talk to other mobile users or fixed land-lines. However the only person that needs to talk to a UHF radio transmitter user is another UHF radio transmitter user.
Now, there are a finite number of calls that my system can sustain at any given time. Suppose I have 2 design options (A) and (B) both of which can provide 200 “capacity units”.
Let’s assume for option A the “capacity cost” of placing a fixed land-line call is “1”, a mobile user “2” and a UHF user “5” – meaning that it is more costly to have large numbers of UHF users than it is fixed land-line users, and for option B the cost is “1” for each user type, but option B is 10 times as expensive as option A
Then suppose my customer provides an expected user population that is split 40/40/20 between fixed, mobile and UHF. Since UHF and mobile traffic is more “expensive”, this hurts my overall capacity and I find that for option A: 40 x 1 + 40 x 2 + 20 x 5 = 220 (i.e. more than the total overall capacity), so I am forced to choose option B (40 x 1 + 40 x 1 + 20 x 1 = 100)
What I want to be able to do is create the scenario as defined by my customer and show them why my system suddenly has increased in cost by a factor of 10.
If my customer then goes away and decides, actually they only needs 10 % of their users to use UHF radios then the user population demographic becomes 45/45/10 and the capacity cost for option A becomes 45 x 1 + 45 x 2 + 10 x 5 = 185, so I can use option A which is cheaper (and probably simpler).
Potentially I may need to assess many hundreds of different user demographics and possibly many more design options, each with its own “capacity cost” for a particular type of communication link. My customer may in time begin to specify the proportion of user traffic (say 50 fixed to fixed, 20% fixed to mobile, 10% mobile to mobile and 20% UHF to UHF). I will also need to add this sort of information to my model.
I cannot manually create each allowable link for each scenario that the customer dreams up because there are too many variations and the risk for error is too great. Instead what I want is a way to define a set of “abstract” rules that govern the types of connection and the “cost” of each and then drop “instances” of land-line, mobile and UHF users onto a diagram to define a specific scenario and determine the overall effect on the system in a repeatable way.
The able example is a (simplified) version of the problem I am facing and you may very well ask why I need to do this using EA at all. Well as you have probably guessed, the real problem is more complicated with many things to consider aside from the issue of communications “cost”.
