Elevator Dispatch Problem

Optimization Problem:
To define a dispatch pattern of elevators group so that global waiting time and/or PassKm/PasH are minimized.

The problem is usually described in off-line terms as an optimization problem solved with operational research methods.

I don't know yet if this problem is considered in terms of stochastic process with demand model defined as a random event described with distributions obtained from demand data.

Therefore my idea of a solution to elevators dispatch problem is:
 to find an optimal waiting-for-orders-position of elevator based on historic demand data. Objective function would be a function of vector [P]=[P1,P2,...Pn] of positions of elevators. Value of this function would be either scalar or vector cost of operations in terms of waiting time/passKm/Energy etc. 
Objective function value would be calculated based on historic demand database defined stochastically. Value would be obtained from a modeled stochastic process, (series of processes) and value would be result of random stochastic process.
Database would become a Knowledgebase with growing data sample size (growing history), and random process would become more and more described and defined though still random.

References:
the initial idea was mine and independent, though those guys were first and should be referenced:
M. Ruokokoski, H. Ehtamo, S. Sorsa, M.-L. Siikonen, 2008:
Passenger Allocation to Capacitated Elevators. Working
paper.
http://www.sal.tkk.fi/publications/pdf-files/cruo08.pdf