Springe direkt zu Inhalt

Business Track

The conference includes a dedicated stream with speakers from practice. In these sessions, industry experts present OR & Analytics case studies and success stories.

The talks should be no marketing or product shows, nor do they need to focus on the scientific background, but they should mainly discuss business cases for OR & Analytics.

The main purpose is to demonstrate the value of OR, its impact and its usefulness in general. Presentations may but need not be technical and should be accessible to participants from other industries. The presentations may serve as showcases for the classroom. Talks are focused and should last no longer than 20-25 minutes incl. discussion

We invite industry experts to contribute to this stream of presentations, which will take place on Wednesday (September 06). Pease ask for invitation code under or2017@fu-berlin.de.

Presentations in this year's Business Track:

Einführung von Optimierungssystemen in Verkehrsbetrieben

In diesem Vortrag sollen die Probleme bei der Einführung von Systemen zur Fahr-, Umlauf- und Dienstplan-Optimierung betrachtet werden. Mittlerweile sind universell konfigurierbare Systeme zur Lösung der Problemstellungen in Verkehrsbetrieben verfügbar, dennoch ist die Einführung solcher Systeme aufwändig und zeitintensiv. Bei den bisher eingesetzten manuellen oder teilautomatischen Verfahren beruht vieles auf dem Expertenwissen der Bearbeiter, das nur unzureichend dokumentiert ist und nur in iterativen Arbeitsschritten in die Konfiguration der Optimierungsverfahren einfließen kann, was einerseits zu längeren Einführungszeiten andererseits häufig zur Unzufriedenheit seitens der Kunden führt. Weiterhin werden die Aspekte der Datenqualität und der Vergleichbarkeit von Lösungen betrachtet. Dies ist insbesondere bei der Dienstplanoptimierung relevant, da dabei nicht nur finanzielle Kosten in die Optimierung eingehen, sondern schwer quantifizierbare Aspekte wie z.B. Sozialverträglichkeit.

Strategic network analytics for the postal industry

Advanced analytics are permeating almost all aspects of network operations of postal and logistics companies and result in significant improvements in planning accuracy and agility. A vast number of both off-the-shelf and proprietary tools exist to support most aspects of day-to-day operations from depot-level volume forecasting to sort center yard management or dynamic delivery route planning. Interestingly, however, there seems to remain a last bastion within most companies that hasn’t been taken by analytics by storm yet – that of long-term strategy making. Many times, decisions on significant real-estate investments, long-term contracts or regulatory issues are taken with little consideration of complex resulting effects on network operating cost, service quality implications or agility to support an evolving product portfolio (moving, e.g., into time-window or same day delivery). Is this because existing tools aren’t suited to fully support strategic decision processes? What would be needed in terms of tools, capabilities and processes to bring truly analytics driven decision making into postal board rooms? In our talk, we will briefly outline strategic challenges most postal networks will face (or are already facing) in a world of ever-growing and diversifying eCommerce, globalizing trade flows and increasing competition. We will then argue that a new and different kind of analytics will be needed to address these challenges and allow decision makers to understand their networks as core strategic assets rather than a complexity to be dealt with. Finally, we’ll provide a brief overview of CEP2NET powered by OPTANO, McKinsey’s proprietary and tailored solution for postal network analytics.

Impact of efficient algorithm implementation on the performance of heuristic optimization

The inola Advanced Optimization Core is a reliable, self-learning, heuristic optimization technology which is particularly useful for solving practical complex constrained operations research problems. Generic heuristic optimization methods have the disadvantage of not being able to provide evidence for global optimal solutions, so the computed solution should be delivered fast and of high quality.In the presentation, the importance of efficient algorithm implementation will be demonstrated by means of practical routing tasks, which has been implemented with the inola AOC. Using different algorithm designs, the impact on performance will be analyzed by pre-post comparison. Small changes in just parts of selected algorithms can lead to a significant improvement in the application’s performance. In addition to the importance of efficient algorithm implementation another aspect, why heuristic optimization methods can fail, is the lack of a proper solution space exploration. The optimization method shall only cut-off areas of the solution space, in which the optimum will never be found, which is by default a challenge of optimization.

Research in operations - the aircraft maintenance, repair and overhaul market

For maintenance, repair and overhaul (MRO) of aircrafts, used parts from other airplanes are used. The sell, loan and exchange of used aircraft parts is a big market. MRO providers offer flat rate deals to airlines that cover all expenses of MRO operations. These deals require a lot of operations research: Demand for spare parts need to be forecasted, values for spare parts in various conditions must be determined and most efficient economical use of existing stock calculated. The presenter will give a brief insight in a solution for a big MRO provider. Applied methods include forecasts based on stochastic processes for demand prediction. Additionally, a combination of various regression models is used for value determination. Optimization models support buy/sell decisions and the most efficient use of stock parts. The models itself are of interest not only due to their complexity but also due to the enhancements to handle all kind of missing data cases. The solution for this MRO provider is tailor made. The presentation will show examples for other custom-made OR solutions. That should support the thesis that successful OR implementations are mostly tailor made.

User-centered Smart Data - Decision Analytics for Digital Engineering

Currently, the strategic use of Big Data for decision analytics is a challenge of utmost relevance for numerous leading enterprises. Following a clear management commitment, most companies have prepared themselves with a considerable IT infrastructure to meet the challenges of data-driven decisions. However, there is often a serious discrepancy between the technical possibilities for intelligent use of Big Data and its actual application for immediately available use cases. This gap can very often be contributed to the lack of organizational development. In order to tap the full potential in the long term, it is not enough to rely on the know-how of data scientists in the short term. In the future, engineers will have to act as data scientists themselves. Hence, a mindset change and corresponding qualification measures are essential. In this talk, we present the iterative and incremental User-centered Smart Data approach for effective decision analytics. This innovative Design Thinking based methodology involves users and their needs right from the start and rapidly develops highly scalable data analyses in an agile manner and conforming to the respective IT policies. The resulting speedboats in form of prototypes enable sustainable solutions accepted by central IT departments. Customized training concepts provide the required backwind for the necessary mindset change of the engineers. Concrete case studies show how the existing gap between IT and specialist departments is closed and data based and algorithmically sound decision-making processes can be established efficiently. For example, for a premium automobile OEM, the application of User-centered Smart Data facilitated the optimization of the operating strategy of hybrid drives to maximize recuperation performance.

Designing and optimizing an LNG supply chain using LocalSolver

This talk deals with the optimization of the sizing and configuration of a Liquefied Natural Gas (LNG) supply chain. This problem is encountered at ENGIE, a French multinational electric utility company which operates in the fields of electricity generation and distribution, natural gas and renewable energy. Having described the industrial problem and its stakes, we show how to model and solve it efficiently using set-based modeling features of LocalSolver, a new-generation hybrid mathematical programming solver. In the process, we combined ENGIE Lab CRIGEN business knowledge and mathematical modelling skills with LocalSolver agile design thinking and powerful solver components to tackle advanced routing & scheduling problems. The resulting software OptiRetail is now used by ENGIE to carry out design studies of LNG supply chains.Several onshore customers need to be supplied with natural gas from LNG sources. The demand of each client is known for every time step. Different transportation means such as vessels or trucks are available to supply LNG from sources to customers, possibly using intermediate hubs. Each carrier is characterized by its storage capacity and its costs as well as the list of sites that it can visit. A tour is a distribution travel starting from a source with full capacity and visiting a certain number of sites, unloading a fraction of the capacity at each site, and finally getting back to the starting source. A planning is a set of tours over the horizon. The cost of the planning is composed of fixed costs and operating costs. The objective is to minimize this cost over a long-term horizon, typically 20 years.

Delivering on delivery: Optimisation, expectations and the future of vehicle routing.

As the demand for home delivery continues to grow, so too does the need for organisations to implement more efficient delivery operations. For our clients, the constraints imposed on home delivery are numerous - and it is a challenge that causes many organisations to struggle. Our clients have any number of vehicles, each with limited capacity. Vehicles are limited to certain routes, and routes are often limited by traffic, accidents or road closures. Add driver shifts, multiple destinations and the rising desire for nominated time-window delivery (slots) and you get a scheduling problem no human operator can solve with any degree of accuracy. Our data-driven, operations research approach has, and will continue to improve the efficiency of home delivery; reducing costs for our clients, and vastly improving the customer experience.
In this talk, as well as discussing the above in more depth, we will describe how optimisation and machine learning can be leveraged to build highly impactful last-mile delivery solutions. We will illustrate, through case studies from our own consulting experiences, how some of the UK’s largest retailers are using optimisation to schedule their fleets of vehicles, and disclose just how impactful it has been for their costs, and their customers. We will provide an insight into the business challenges associated with applying the latest optimisation techniques into organisations of all sizes, and describe how the rising expectations of customers is forcing the optimisation community to react.