Eli Angela Vitor Toso

Erratum to: Production setup-sequencing and lot-sizing at an animal nutrition plant through atsp subtour elimination and patching

This paper considers the usefulness of a production lot sizing and scheduling model at an animal nutrition plant with sequence-dependent setup times. The model covers multiple periods and is based on the asymmetric travelling salesman problem (atsp). It is applied initially to the case where the setup state is zeroed between periods, and then revised to model the carryover of the setup state from one period to the next. An iterative solution procedure based on subtour elimination is applied, and then enhanced by the inclusion of a subtour patching procedure. Case-based tests with actual plant data show that the subtour elimination is practicably fast where the setup state is zeroed between periods, but needs the patching procedure when the setup state is preserved, as is the situation at the plant. In this latter case, the subtour elimination and patching can be very fast, showing the method’s viability for operational lot sizing and sequencing in animal nutrition plants of the kind studied. Tests on perturbed plant data show that further algorithmic development is needed to tackle certain challenging variants found in other plants.

Effective location models for sorting recyclables in public management

The recycling of urban solid wastes is a critical point for the “closing supply chains” of many products, mainly when their value cannot be completely recovered after use. In addition to environmental aspects, the process of recycling involves technical, economic, social and political challenges for public management. For most of the urban solid waste, the management of the end-of-life depends on selective collection to start the recycling process. For this reason, an efficient selective collection has become a mainstream tool in the Brazilian National Solid Waste Policy. In this paper, we study effective models that might support the location planning of sorting centers in a medium-sized Brazilian city that has been discussing waste management policies over the past few years. The main goal of this work is to provide an optimal location planning design for recycling urban solid wastes that fall within the financial budget agreed between the municipal government and the National Bank for Economic and Social Development. Moreover, facility planning involves deciding on the best sites for locating sorting centers along the four-year period as well as finding ways to meet the demand for collecting recyclable materials, given that economic factors, consumer behavior and environmental awareness are inherently uncertain future outcomes. To deal with these issues, we propose a deterministic version of the classical capacity facility location problem, and both a two-stage recourse formulation and risk-averse models to reduce the variability of the second-stage costs. Numerical results suggest that it is possible to improve the current selective collection, as well as hedge against data uncertainty by using stochastic and risk-averse optimization models.

Otimização no dimensionamento e seqüenciamento de lotes de produção: estudo de caso numa fábrica de rações

Neste trabalho, e proposta uma abordagem para otimizar o problema integrado de dimensionamento e sequenciamento de lotes de producao em uma empresa do setor de nutricao animal. Tal problema consiste em decidir quanto produzir de cada produto em cada periodo, considerando a sequencia de producao dos lotes, de maneira a satisfazer a demanda e minimizar os custos de producao e estoques. Uma das grandes dificuldades para a programacao da producao na empresa e integrar estas decisoes, uma vez que os tempos de preparacao (setup) sao bem dependentes da sequencia produtiva. O problema e modelado por programacao linear inteira mista e resolvido por meio da linguagem de modelagem GAMS/CPLEX com alguns procedimentos para reduzir os tempos computacionais. Experimentos realizados com dados reais mostram que esta abordagem e capaz de gerar resultados melhores do que os utilizados pela empresa.

Combinação de abordagens GLSP e ATSP para o problema de dimensionamento e sequenciamento de lotes de produção de suplementos para nutrição animal

In this paper we study the combination of GLSP (General Lot Sizing and Scheduling Problem) and ATSP (Asymmetric Travelling Salesman Problem) approaches with sub-tour elimination and patching to a lot sizing and sequencing problem in the animal nutrition industry. This problem consists of deciding the lots size for each product as well the production sequence of the lots, while meeting demand without backlogs and minimizing production and inventory costs. The coordination of these decisions is a challenge for production scheduling in this industry as the setup times are sequence dependent. The ATSP approaches are compared with relax-and-fix approaches applied to the GLSP (General Lot-sizing and Scheduling Problem) formulated in previous research, using real data from an animal nutrition plant in Sao Paulo state. Portuguese: Neste artigo estudamos a combinacao de abordagens GLSP (General Lot Sizing and Scheduling Problem) e ATSP (Asymmetric Travelling Salesman Problem) para o problema de dimensionamento e sequenciamento de lotes na industria de nutricao animal. Este problema consiste em determinar o tamanho de cada lote de producao para cada produto, assim como a sequencia de producao destes lotes, de forma a satisfazer a demanda sem atrasos e minimizar os custos de producao e estoques. Uma dificuldade para a programacao da producao nesta industria e integrar estas decisoes, pois os tempos de preparacao da linha de producao sao dependentes da sequencia produtiva e nao obedecem a desigualdade triangular. A abordagem proposta e comparada com abordagens relax-and-fix para o modelo GLSP (General Lot-sizing and Scheduling Problem) estudadas em trabalhos anteriores, utilizando dados reais de um estudo de caso de uma fabrica de nutricao animal localizada no interior de Sao Paulo.

Production planning in the molded pulp packaging industry

We study the production planning in molded pulp packaging companies.A key decision is to define which molding patterns should be used.Other decisions are how long the patterns should be used and their sequence.We propose a novel optimization approach to deal with this problem.Tests show that the obtained production plans are suitable in real settings. Production planning in molded pulp packaging companies involves decisions about setting up appropriately the production process to meet the demands of different products. This production process comprises stages such as blending, molding and drying, and the major challenges for the production planning occur in the molding stage, where the mix of products manufactured simultaneously depends on the combination of molds attached to the molding machine, called molding pattern. Thus, the problem lies in deciding which molding patterns should be used, how long each pattern should be used and the production sequence that they should be scheduled. In this study we propose a novel optimization approach to deal with this problem by considering some possible settings for the molding machine based on specific technological constraints. These preset settings are included into a mixed integer programming model to define the molding patterns and to deal with the production lot sizing and scheduling decisions. Since the molding patterns are defined by the approach, three types of production setup times and costs (some of then are sequence-dependent) must be also calculated and taken into account in the planning decisions. Computational tests with real data from a Brazilian molded pulp plant show that the production plans generated by this approach are suitable in practical settings, as well as define molding patterns different from those commonly used and simultaneously reduce setup times and costs, when compared to the production plans of the company.

BIOMASS SUPPLY CHAIN ANALYSIS WITH DENSIFIED WASTE

The possibility to vary the energy matrix, thus reducing the dependency on fossil fuels, has amplified the acceptance of biomass as an alternative fuel. Despite being a cheap and renewable option and the fact that Brazil is a major producer of waste from agriculture and forestry activities, the use of these materials has barriers due to its low density and low energetic efficiency, which can raise the costs of its utilization. Biomass densification has drawn attention due to its advantage in comparison to in natura biomass due to its better physical and combustion characteristics. The objective of this paper is to evaluate the impact of biomass densification in distribution and transport costs. To reach this objective, a mathematical model was used to represent decisions at a supply chain that coordinates the purchase and sale of forestry and wood waste. The model can evaluate the options to deliver biomass through the supply chain combining demand meeting and low cost. Results point to the possibility of an economy of 60% in transport cost and a reduction of 63% in the required quantity of trucks when densified waste is used. However, costs related to the densifying process lead to an increase of total supply costs of at least 37,8% in comparison to in natura waste. Summing up, the viability of biomass briquettes industry requires a cheaper densification process.

Planejamento agregado na indústria de nutrição animal sob incertezas

One of the greatest challenges of production planning in the animal nutrition industry is determining the amount of each product that should be produced during each period, given the perishability of the products, the manual execution of the setups and the need to adjust the production capacity in a stochastic demand environment that is characterized by the seasonality of the products and raw materials. This paper investigates an aggregate production planning problem in a plant that produces supplements for horses, cattle, pigs and poultry. To address this problem, we proposed an extension of the classical capacitated lot-sizing problem to incorporate decisions about lost sales and inherent uncertainties in production planning, such as demands, setup times and perishability. To generate solutions that are less sensitive to changes in scenarios, we also developed a risk-averse stochastic model with an absolute semi-deviation-based risk measure. An analysis of the expected value of perfect information and the value of the stochastic solution confirmed that the stochastic approach outperformed the deterministic approximations in handling uncertainty. Furthermore, the results indicated that it is possible to significantly reduce the variability of the second-stage costs without sacrificing the expected total cost.