Conventional mechanisms for planning supply networks are largely manual and target stable network configurations. For example, rapid delivery times are achieved by co-locating suppliers in "supplier parks" next to the client's manufacturing facilities, but this locks smaller suppliers to the client. More flexibility is provided by the Fourth-Generation Logistics providers, but they also work from the perspective of one manufacturer and not for the benefit of the whole network. Supply networks evolution is also conducted in a haphazard manner, and any run-time disturbances to transport arrangements are dealt with manually. Researchers (e.g. TeleTruck [1], MaBE [3]) have approached this problem using run-time bottom-up aggregation of orders and service bundling. This achieves flexibility at run-time, but sacrifices optimality of solutions due to the short time for seeking partnerships, and increased transaction costs.