Work team building and planning problem : models and experiments in the service-to-business context

PhD student: 
Co-supervisor(s): 
External supervisors: 

José Francisco Ferreira Ribeiro, Professeur associé

Faculdade de Economia
Administração e Contabilidade de Ribeirão Preto
Universidade de São Paulo, BRASIL
Email: jffr@fearp.usp.br

Starting date: 
October 2016
Host institution: 
Other institution: 
University of Sao Paulo, Sté RESOLV, Brésil

Subcontracting companies organize their agents into work teams and create their work plans to fulfill clients’ demands. Multiple constraints have to be met, and several economic and social performance criteria have to be attained. Making a decision, satisfying all these conditions, becomes increasingly difficult, especially in a context of variable demand. A generic problem characterization in the form of a class diagram, containing all the characteristics of the clients, the demands, the agents, the travel routes, and the vehicles, allows us to identify a large number of variants of the problem in the service-to-business as well as service-to-individual contexts. Each variant corresponds to a combination of decision-making problems, demand variability, agents’ flexibility, and economic and social performance indicators. We study more deeply two variants, inspired by a real problem found in a Brazilian company in the service-to-business sector, with stable and variable demands. Through a literature review, we identify several potential organizational levers to increase the flexibility of agents, and appropriate modeling and resolution approaches. We use the multi-objective mixed integer linear programming method for the two variants. In the context of stable demand, we consider a compromise between the company’s travel costs and agents’ work trip duration in a two-level approach: cyclic weekly planning for new customers ensuring assignment stability for a long term and, at a given frequency, re-planning for all active clients, allowing global optimization. When the demand is variable, we consider the reconciliation between the travel costs, the workload balance between agents, and their preference satisfaction for work periods. As planning is made for each short horizon and without repetitiveness, the dependence of agents’ activities between consecutive horizons can be relaxed by adding buffer zones between them; in this variant, organizational levers, such as flexible work contracts and multi-skilled agents, are considered.

Keywords: team building, workforce planning, mixed integer linear programming, multi-criteria decision support, variability of demand, flexibility of resources.