Marco Schutten

Distributed Decision Making in Combined Vehicle Routing and Break Scheduling

The problem of combined vehicle routing and break scheduling comprises three subproblems: clustering of customer requests, routing of vehicles, and break scheduling. In practice, these subproblems are usually solved in the interaction between planners and drivers. We consider the case that the planner performs the clustering and the drivers perform the routing and break scheduling. To analyze this problem, we embed it into the framework of distributed decision making proposed by Schneeweiss (Eur J Oper Res 150(2):237–252, 2003). We investigate two different degrees of anticipation of the drivers’ planning behaviour using computational experiments. The results indicate that in this application a more precise anticipation function results in better objective values for both the planner and the drivers.

Inventory routing for dynamic waste collection

We consider the problem of collecting waste from sensor equipped underground containers. These sensors enable the use of a dynamic collection policy. The problem, which is known as a reverse inventory routing problem, involves decisions regarding routing and container selection. In more dense networks, the latter becomes more important. To cope with uncertainty in deposit volumes and with fluctuations due to daily and seasonal effects, we need an anticipatory policy that balances the workload over time. We propose a relatively simple heuristic consisting of several tunable parameters depending on the day of the week. We tune the parameters of this policy using optimal learning techniques combined with simulation. We illustrate our approach using a real life problem instance of a waste collection company, located in The Netherlands, and perform experiments on several other instances. For our case study, we show that costs savings up to 40% are possible by optimizing the parameters.

An approximate dynamic programming approach to urban freight distribution with batch arrivals

We study an extension of the delivery dispatching problem (DDP) with time windows, applied on LTL orders arriving at an urban consolidation center. Order properties (e.g., destination, size, dispatch window) may be highly varying, and directly distributing an incoming order batch may yield high costs. Instead, the hub operator may wait to consolidate with future arrivals. A consolidation policy is required to decide which orders to ship and which orders to hold. We model the dispatching problem as a Markov decision problem. Dynamic Programming (DP) is applied to solve toy-sized instances to optimality. For larger instances, we propose an Approximate Dynamic Programming (ADP) approach. Through numerical experiments, we show that ADP closely approximates the optimal values for small instances, and outperforms two myopic benchmark policies for larger instances. We contribute to literature by (i) formulating a DDP with dispatch windows and (ii) proposing an approach to solve this DDP.

An agent-based simulation framework to evaluate urban logistics schemes

Inefficient urban freight transport has a negative impact on both livability in cities and profit margins in the supply chain. Urban logistics schemes, consisting of governmental policies and company initiatives, attempt to address these problems. However, successful schemes are difficult to realize due to the divergent objectives of the agents involved in urban logistics. Traditional optimization techniques fall short when evaluating schemes, as they do not capture the required change in behavior of autonomous agents. To properly evaluate schemes, we develop an agent-based simulation framework that assesses the interaction between five types of autonomous agents. Compared to existing studies in this field, we contribute by (i) explicitly including company-driven initiatives, and (ii) adopting a supply chain-wide perspective. We illustrate the working of our framework by testing a number of schemes on a virtual network.

A unified race algorithm for offline parameter tuning

This paper proposes uRace, a unified race algorithm for efficient offline parameter tuning of deterministic algorithms. We build on the similarity between a stochastic simulation environment and offline tuning of deterministic algorithms, where the stochastic element in the latter is the unknown problem instance given to the algorithm. Inspired by techniques from the simulation optimization literature, uRace enforces fair comparisons among parameter configurations by evaluating their performance on the same training instances. It relies on rapid statistical elimination of inferior parameter configurations and an increasingly localized search of the parameter space to quickly identify good parameter settings. We empirically evaluate uRace by applying it to a parameterized algorithmic framework for loading problems at ORTEC, a global provider of software solutions for complex decision-making problems, and obtain competitive results on a set of practical problem instances from one of the worlds largest multinationals in consumer packaged goods.

Advanced Production Planning and Scheduling Systems

In this chapter, we present algorithms for a number of functions of the production planning framework presented in Chap. 5. We focus on models for integrated Capacity and Master Production Planning, Job Planning and Resource Group Loading, and Shop Floor Scheduling and Control. At the Master Production Planning level, we exploit a simple Linear Programming formulation to set appropriate capacity levels and in particular to decide whether a temporary expansion of capacity is needed (e.g., through overtime work). With the same formulation, we decide what end-items are to be produced in which period. By applying the lead time offset procedure that is the heart of Materials Requirements Planning, and using the Bill of Materials information, the same is done on the level of part manufacturing (basic level). Essential in the above procedure are two parameters, the effective overall capacity of each manufacturing shop and the final assembly department, often indicated as the maximum throughput, and the lead times needed to complete a part or product in each department. A significant portion of these lead times may in fact be waiting times in front of individual workstations that are busy. To minimize these waiting times, workload control norms are often used which in turn may influence the effective capacity. An essential question then is what these workloads should be in order to match a desired throughput and production lead time. That question is answered by exploiting a Closed Queueing Network approach that explicitly determines the relation between a preset work-in-process level, throughput and the resulting lead times (advanced level). Finally, we exploit a detailed shop floor scheduling procedure, called the Shifting Bottleneck approach, that basically serves to ascertain that internal due-dates, following from the above defined internal manufacturing lead times are indeed met (state-of-the-art).

Scheduling a single machine with release dates, due dates and family setup times

In machining environments the combined goal of efficient and effective production may lead to complex control problems. Efficient production in such an environment is translated in minimizing the loss of capacity due to setups and therefore combining jobs with similar setup characteristics. Effective production in an order-driven environment is translated in completing jobs in time, or at least minimizing tardiness. Clearly, these two objectives may be conflicting: clustering jobs with similar setup characteristics may lead to a severe delay of other “badly fitting” jobs, thereby causing a substantial job tardiness. On the other hand, an “earliest due date” schedule may lead to too many setups, hence a capacity overload and more and more late orders, contrary to its objective. Any solution to this problem should be based on a combination of batching and sequencing procedures.