Optimal Fleet Mix and
Ambulance Dispatching in
Emergency Medical
Service Systems
Kenneth C. Chong
Shane G. Henderson
Mark E. Lewis
Cornell University
School of Operations Research and Information Engineering
Ithaca, New York 14853
We consider the problem
of selecting the number of Advanced Life Support (ALS) and Basic Life Support
(BLS) ambulances— the
fleet mix— to deploy in an emergency medical service system. ALS units can
treat
a wider range of emergencies, while BLS units are
less expensive to operate. Because the choice of fleet mix
can affect how ambulances are dispatched to
incoming emergency calls, we construct a two-tiered model
incorporating both types of decisions. We model the dispatching
problem as one of control of a two-class
queueing system with two types of servers, which we analyze using Markov
decision process theory. We
show that the optimal policy takes the form of a
monotone switching curve, but demonstrate that the
corresponding value function is, in general, not concave.
From the value function induced by the optimal
policy, we obtain a measure by which system
performance under a given fleet mix can be evaluated. We then
use this measure to solve a single-variate optimization problem to identify the optimal fleet
mix, subject
to a budget constraint. Our numerical experiments
indicate that, within the context of our model, all-ALS
fleets do not significantly outperform well-chosen mixed fleets.