Luca Talarico

MISTRAL : a game-theoretical model to allocate security measures in a multi-modal chemical transportation network with adaptive adversaries

In this paper we present a multi-modal security-transportation model to allocate security resources within a chemical supply chain which is characterized by the use of different transport modes, each having their own security features. We consider security-related risks so as to take measures against terrorist acts which could target critical transportation systems. The idea of addressing security-related issues, by supporting decisions for preventing or mitigating intentional acts on transportation infrastructure, has gained attention in academic research only recently. The decision model presented in this paper is based on game theory and it can be employed to organize intelligence capabilities aimed at securing chemical supply chains. It enables detection and warning against impending attacks on transportation infrastructures and the subsequent adoption of security countermeasures. This is of extreme importance for preventing terrorist attacks and for avoiding (possibly huge) human and economic losses. In our work we also provide data sources and numerical simulations by applying the proposed model to a illustrative multi-modal chemical supply chain.

A decision model to allocate protective safety barriers and mitigate domino effects

Abstract In this paper, we present a model to support decision-makers about where to locate safety barriers and mitigate the consequences of an accident triggering domino effects. Based on the features of an industrial area that may be affected by domino accidents, and knowing the characteristics of the safety barriers that can be installed to stall the fire propagation between installations, the decision model can help practitioners in their decision-making. The model can be effectively used to decide how to allocate a limited budget in terms of safety barriers. The goal is to maximize the time-to-failure of a chemical installation ensuring a worst case scenario approach. The model is mathematically stated and a flexible and effective solution approach, based on metaheuristics, is developed and tested on an illustrative case study representing a tank storage area of a chemical company. We show that a myopic optimization approach, which does not take into account knock-on effects possibly triggered by an accident, can lead to a distribution of safety barriers that are not effective in mitigating the consequences of a domino accident. Moreover, the optimal allocation of safety barriers, when domino effects are considered, may depend on the so-called cardinality of the domino effects.

Ambulance routing for disaster response with patient groups

We consider a routing problem for ambulances in a disaster response scenario, in which a large number of injured people require medical aid at the same time. The ambulances are used to carry medical personnel and patients. We distinguish two groups of patients: slightly injured people who can be assisted directly in the field, and seriously injured people who have to be brought to hospitals. Since ambulances represent a scarce resource in disaster situations, their efficient usage is of the utmost importance. Two mathematical formulations are proposed to obtain route plans that minimize the latest service completion time among the people waiting for help. Since disaster response calls for high-quality solutions within seconds, we also propose a large neighborhood search metaheuristic. This solution approach can be applied at high frequency to cope with the dynamics and uncertainties in a disaster situation. Our experiments show that the metaheuristic produces high quality solutions for a large number of test instances within very short response time. Hence, it fulfills the criteria for applicability in a disaster situation. Within the experiments, we also analyzed the effect of various structural parameters of a problem, like the number of ambulances, hospitals, and the type of patients, on both running time of the heuristic and quality of the solutions. This information can additionally be used to determine the required fleet size and hospital capacities in a disaster situation.

Metaheuristics for the risk-constrained cash-in-transit vehicle routing problem

This paper proposes a variant of the well-known capacitated vehicle routing problem that models the routing of vehicles in the cash-in-transit industry by introducing a risk constraint. In the Risk-constrained Cash-in-Transit Vehicle Routing Problem (RCTVRP), the risk of being robbed, which is assumed to be proportional both to the amount of cash being carried and the time or the distance covered by the vehicle carrying the cash, is limited by a risk threshold. A library containing two sets of instances for the RCTVRP, some with known optimal solution, is generated. A mathematical formulation is developed and small instances of the problem are solved by using IBM CPLEX. Four constructive heuristics as well as a local search block composed of six local search operators are developed and combined using two different metaheuristic structures: a multistart heuristic and a perturb-and-improve structure. In a statistical experiment, the best parameter settings for each component are determined, and the resulting heuristic configurations are compared in their best possible setting. The resulting metaheuristics are able to obtain solutions of excellent quality in very limited computing times.

The k-dissimilar vehicle routing problem

In this paper we define a new problem, the aim of which is to find a set of k dissimilar solutions for a vehicle routing problem (VRP) on a single instance. This problem has several practical applications in the cash-in-transit sector and in the transportation of hazardous materials. A min–max mathematical formulation is proposed which requires a maximum similarity threshold between VRP solutions, and the number k of dissimilar VRP solutions that need to be generated. An index to measure similarities between VRP solutions is defined based on the edges shared between pairs of alternative solutions. An iterative metaheuristic to generate k dissimilar alternative solutions is also presented. The solution approach is tested using large and medium size benchmark instances for the capacitated vehicle routing problem.

A large neighbourhood metaheuristic for the risk-constrained cash-in-transit vehicle routing problem

In this paper, we propose a new metaheuristic to solve the Risk constrained Cash-in-Transit Vehicle Routing Problem (Rctvrp). The Rctvrp is a variant of the well-known capacitated vehicle routing problem and models the problem of routing vehicles in the cash-in-transit sector. In the Rctvrp, the risk associated with a robbery represents a critical aspect that is treated as a limiting factor subject to a maximum risk threshold.A new metaheuristic, called aco-lns is developed. It combines the ant colony heuristic for the travelling salesman problem and a variable neighbourhood descent within an large neighbourhood search framework.A new library of Rctvrp instances with known optimal solutions is proposed. The aco-lns is extensively tested on small, medium and large benchmark instances and compared with all existing solution approaches for the Rctvrp. HighlightsA new metaheuristic named ACO-LNS is developed to increase security in the CIT sector.The ACO-LNS combines the ant colony optimization with a VND heuristic.The ACO-LNS uses an effective large neighbourhood search heuristic.A set of benchmark instances for the RCTVRP is proposed and made publicly available.The ACO-LNS outperforms the existing algorithms for the RCTVRP.

An optimization algorithm for the workforce management in a retail chain

A model to generate optimized work shifts in a chain of supermarkets in Italy is proposed.A mathematical formulation of the workforce management problem is defined.An exact approach to solve the problem is described.An efficient hybrid heuristic is presented and statistically tested.27 real problem instances faced by the firm within its supermarkets are solved. This work proposes a scheduling problem for the workforce management in a chain of supermarkets operating in Italy. We focus on determining the ideal mix of full-time and part-time workers which are needed every week to guarantee a satisfactory service level during the check-out operations. The generation of working shifts, to be assigned to retail workers, is subject to several constraints imposed by both labour laws and enterprise bargaining agreements.We present a mathematical formulation of the problem followed by an exact solution approach which relies on the definition of feasible daily working shifts. The number of feasible daily shifts, that are combined to determine feasible weekly shifts, could drastically increase, depending on the selected planning interval. In addition, there may exist additional constraints, that are difficult to incorporate into the mathematical model. For these reasons, a hybrid heuristic, which does not require the generation of all feasible weekly shifts, is proposed in this paper.Using appropriate statistical techniques, a sensitivity analysis is performed to test the design of the hybrid heuristic. Computational tests are carried out by solving several real instances provided by the retail firm. The results obtained by the heuristic are compared both with an exact approach and with the solutions adopted by the retail company, which have been determined by using a naif approach. Our hybrid heuristic exhibits excellent performance finding optimal or near optimal solutions in a very limited CPU time.

A hybridised variable neighbourhood tabu search heuristic to increase security in a utility network

We propose a decision model aimed at increasing security in a utility network (e.g., electricity, gas, water or communication network). The network is modelled as a graph, the edges of which are unreliable. We assume that all edges (e.g., pipes, cables) have a certain, not necessarily equal, probability of failure, which can be reduced by selecting edge-specific security strategies. We develop a mathematical programming model and a metaheuristic approach that uses a greedy random adaptive search procedure to find an initial solution and uses tabu search hybridised with iterated local search and a variable neighbourhood descend heuristic to improve this solution. The main goal is to reduce the risk of service failure between an origin and a destination node by selecting the right combination of security measures for each network edge given a limited security budget.

Intelligent Systems in Managerial Decision Making

This chapter aims at giving a general overview of the existing intelligent systems that can be used to support decision making in a variety of domains. Intelligent systems comprise both the methods, processes and technologies able to gather, analyse and interpret data with the goal of helping decision makers improve the performance of any organization. Such intelligent systems can be based upon several techniques borrowed from different fields such as operations research , decision theory , data mining and game theory . For each category the ideas behind these methods are explained and the operating principles are summarized. Practical applications and tools used for managerial purposes are also provided.

A bi-objective model to increase security and reduce travel costs in the cash-in-transit sector

In this paper, we present a variant of the vehicle routing problem (VRP) to increase security in the cash-in-transit sector. A specific index is used to quantify the exposure of a vehicle to the risk of being robbed along its route. In addition, the problem is subjected to a traditional capacity constraint, according to which a maximum amount of valuables can be transported inside the vehicle. This constraint might be imposed, for example, by insurance companies. A bi-objective formulation, aimed at reducing both the risk and the travel cost, is proposed. The objectives are conflicting since higher risk exposures allow a reduction of the travel cost needed to visit and collect valuables from all customers. A mathematical model of the problem is proposed and solved by using a progressive multi-objective metaheuristic. Realistic instances are also generated considering the geographical coordinates of several customers (e.g., stores, banks, shopping centres) located in Belgium. The proposed solution approach is tuned and tested both on these realistic instances and on standard benchmark instances for the capacitated vehicle routing problem.