Picture source:


Making the right preparations for natural disasters

When a disaster occurs, delivering relief supplies such as food, water and shelter to people in the affected areas as early as possible is of critical importance for effective relief operations. We could strategically position the resources in advance based on the most likely disaster scenarios.

There is significant uncertainty in factors such as location and severity of disaster, demand quantities and transportation infrastructure that affects how effective a response is. It is crucial to develop models incorporating the inherent uncertainty to make sound decisions.

In this paper, former doctoral student Xing Hong, professor Miguel Lejeune from George Washington University and professor Nilay Noyan from Sabanci University determine the size and location of the response facilities and inventory levels of relief supplies at each facility while guaranteeing a certain level of network reliability.

Their optimization models feature a chance constraint to ensure that demand for relief supplies across the network is satisfied with a high probability.

Calculating the cost of accepting new orders

Scheduling can be a nightmare for companies that offer dozens, hundreds or even thousands of different products made from the same process.

The auto film company’s scheduling problems are investigated in “Understanding and Managing Product Line Complexity: Applying Sensitivity to Analysis to a Large-Scale MILP Model to Price and Schedule New Customer Orders.” In this paper, Zhili Tian from Florida International University; Panos Kouvelis, a professor at Washington University in St. Louis and director of The Boeing Center on Technology, Information and Manufacturing; and professor Charles Munson from Washington State University use a form of global sensitivity analysis that applies novel regression techniques to a mixed-integer, linear-programming, scheduling model in order to provide decision support tools managers.

Production costs of products vary from one production run to another. By using data from historical customer orders, the authors developed relatively simple equations that managers can use in real time, without needing to rerun the full scheduling model, in order to estimate the capacity usage and material waste that a new order would impose on the system. Managers can use these estimates to make accept/rejection decisions on the phone with customers and to price new orders properly.

Mixed-integer programming for asset replacement

The management of capital assets is vital to the efficiency and profitability of operations in any industry. A decision often faced by managers is the replacement of assets at the minimum possible cost while continuing operations to meet customer demand.

There are number of factors motivating the replacement of capital assets. These factors include increased operating and maintenance costs and reduced capacity caused by deterioration of the currently owned asset or technological advances that make newer assets more efficient to operate with lower operating and maintenance costs.

The ability to solve problems of parallel asset replacement with multiple challengers is critical in industry today, as managers often must trade off the value between suppliers.

Esra Buyuktahtakin of Wichita State University, J. Cole Smith of Clemson University of Massachusetts Lowell, and Shangyuan Luo from UBS AG address this problem in their paper “Parallel Asset Replacement Problem under Economies of Scale with Multiple Challengers.”