Lecture by Tanka Dhamala and Hari Nath

on 24.06.25 
at 14:00 
in 31-302 

Tanka Dhamala will give a lecture on Evacuation Planning: Multi-Objective Approach. Interested guests are welcome.

Abstract:
One of the most prominent mathematical methods for addressing evacuation planning prob-
lems is the use of dynamic network flow models, which aim either to maximize the flow from
danger zones to safe areas or to minimize the total clearance time required to evacuate a given
population. In this presentation, we adopt a macroscopic modeling approach to represent and
analyze these issues. Although network reconfiguration is a computationally challenging task,
it can significantly improve evacuation performance by increasing the possible capacity through
lane reversals directed toward safe destinations. However, implementing a complete contraflow
reconfiguration may block most lanes leading toward the source areas, unfortunately blocking
some paths for specific applications. We propose strategies to avoid reversing unnecessary arcs
during network reconfiguration.
However, since real-life problems are highly motivated by modeling with multi-criteria models,
we combine the above problems with additional characteristics. For example, policymakers
may wish to keep some paths upturned from a given junction or a set of intermediate nodes.
Regardless of flow maximization or time minimization, this requirement implies that the flow
is decreased and the evacuation time is increased. We present the existing algorithms using bi-
objective programming approaches. On the other hand, we model the multi-terminal problems
where Pareto optimal solutions for the maximization of sinks are obtained, contrary to the
priority-based lexicographically maximum flow solutions at sinks.

Hari Nath will give a lecture on Network Flow Location Problems with Maximum Utility of the Facilities: A Bicriteria Approach. 

Abstract:
This study addresses FlowLoc problems on directed networks, which involve optimally placing facilities of given sizes on arcs to minimize disruption to network flow and maximize the utility of the facilities placed. Each facility is associated with a nonnegative utility, and the problem is formulated as a bicriteria optimization model aiming to maximize both flow (static/dynamic) and total utility of the facilities placed. We propose an epsilon-constraint-based algorithm to obtain exact Pareto optimal solutions and develop an NSGA-II-based approach to generate nondominated solutions that progressively approximate the true Pareto front over generations. Through computational experiments on the road network of Bhaktapur city, Nepal, we find that the NSGA-II algorithm is more efficient for larger problem instances, where the epsilon-constraint method either becomes too timeconsuming or the MILP-solver used in its subroutines fails to deliver a solution.