Efficient design and eliminating silly alternatives

[Doug Clover]

Dear All

The purpose of my SC model is to simulate the purchase of electric vehicles in NZ over a period of time taking into account declining costs and improving vehicle performance (e.g. range and speed).
When I set the min and max attribute the levels to take account the effect of changes over time I get combinations that are technically feasible but can only be considered ‘silly’. For example it is feasible to build an electric car now that will travel 400 km, but at current battery prices it will cost around NZ$160,000. Conversely with these values included in the design you could get a combination that has a $160,000 vehicle that only travels 100 km between recharges (a more realistic price is between NZ$50,000 and NZ$65,000).
Am I correct in assuming that if I optimise using B-efficiency in NGENE these types of silly alternatives should be removed from the design? If not how does one go about eliminating these types of alternatives?
Kind regards

Doug

[Michiel Bliemer]

Using B-efficiency, which is balancing alternatives, will not solve your problem.

An option you could consider is to use constraints, look at the chapter about constraints in the Ngene manual. Then you can set certain attribute level combinations that make sense. For example, if the price of the car is high, then the actie radius of the car is also relatively high. This type of constraint is also called 'nesting', the syntax as described in the manual is as follows:

Code: Select all

;cond:
if(alt1.A = 0, alt2.B = 1) ,
if(alt1.A = [1,2], alt2.B = [2,3])


[Doug Clover]

Thanks Michiel.

Doug