Making an Impact

Chair:  John Hopes, President, The Operational Research Society (UK)

Academics and practitioners have different incentives, and different contractual, personal and professional objectives, but there are considerable areas of overlap and opportunities to learn from each other. This is the third successive EURO conference to bring academics and practitioners together for a question-and-answer panel. A summary of the output from the previous two discussions is here.

Previous panels have considered the obstacles in the way of academic-practitioner collaboration, and ways of overcoming them.  This panel will focus particularly on how we can get the best from each other: what would academics like practitioners to know and do, and vice versa. A brief recap of the conclusions from previous panels will be followed by short introductions from several experienced academics and practitioners, followed by questions and contributions from the audience.

Introduction: Ruth Kaufman, Stream leader, Making an Impact and Chair, EURO Working Group on Practice of OR

Discussants:

Sally Brailsford

+ Sally is Professor of Management Science at the University of Southampton. She has a background in mathematics but also worked for many years in the UK National Health Service as a nurse. Her second career, in academia, began in the late 1980s, and since then she has worked in the area of healthcare operational research (OR), mainly using discrete-event simulation and system dynamics. She has applied both approaches to a wide variety of healthcare-related problems. Her current research interests include hybrid simulation methods and behavioural OR.  Sally is chair of the EURO Working Group on OR Applied to Health Services (ORAHS) and co-chair of the UK OR Society’s Special Interest Group on Behavioural OR.  She is the only person to have won the UK OR Society’s Goodeve Medal three times, for the best paper published in the Journal of the OR Society (JORS) that year: in 2004 for modelling emergency healthcare services in Nottingham, in 2006 for modelling chlamydia infection, and in 2015 for modelling the dental workforce in Sri Lanka.  She is a co-Editor-in-Chief of the OR Society journal Health Systems and is on the Editorial Boards of Health Care Management Science, the Journal of Simulation and Health Care Operations Research.  In 2016 she was made a Companion of OR by the UK OR Society in recognition of her contribution to health OR.

Olli Bräysy

+ biography to follow

John Poppelaars

+ John Poppelaars is Practice leader Advanced Analytics at BearingPoint in the Netherlands. In this role John acts as senior advisor and analytics expert. John is a specialist in the field of Transportation & Logistics, Supply Chain, and Workforce Planning & Scheduling, and a passionate proponent of the use of analytics as a key business enabler. Throughout his career, he has applied the business maxim of ‘improving decision-making quality’ to projects across a myriad of industries and supported his clients to optimise their businesses. John was awarded the 2012 INFORMS Franz Edelman Award for his work at TNT Express. John is advisor to the Erasmus University Rotterdam and regularly lectures at universities and conferences on applied analytics.

Rommert Dekker

+ Rommert Dekker is a full professor of Operations Research and Quantitative Logistics at the Erasmus School of Economics. Prior to this job, he worked for seven years with Shell Research and Shell International. His research interests include reverse, service and green logistics, maintenance optimisation and maritime and port logistics. He has supervised over 150 students doing thesis projects in companies. He has participated in several industry sponsored research programs, amongst others. on service and port logistics, including IBM, Fokker Services, ECT container terminal, Shell, Vopak, KPN, ASML, vanderLande, Thales and DAF Trucks. In many of his papers industrial cases with real data are given.

Chair: Ruth Kaufman

People with OR skills and knowledge are in great demand in industry. There are many possible career paths open to OR people outside academia, whether they are starting with no more than a first degree in a scientific subject, or with a PhD and post-doctoral experience in OR. In this session, panel members with very different career paths will briefly describe their career history and current ambitions, the skills they started with, and the important developments along the way; and then open up to discussion with the audience. The purpose of this session is for people at all stages of an OR career to explore ideas about what they may want to do in the future, and what sort of development and training may be useful.

Panel members are:

John Hopes, President of the Operational Research Society (UK)

Kerry Malone, Senior Scientist, Department of Sustainable Urban Mobility and Safety, TNO

Michele Quattrone, R&D engineer, operational and data science team, Air Liquide

Nicole Havinga, ORTEC

Chair: Ruth Kaufman

At a EURO conference there is an enormous amount of interesting material that any one person can only see a small fraction of. This session will help boost that fraction. Each presenter will have just 5 minutes to present a lightning talk: a maximum of 20 slides at 15 seconds per slide. This will be an opportunity for presenters to get a bigger audience for your key ideas and for the audience to get a wide variety of fast-moving and stimulating talks. Details of presenters within the session will be available in due course. The line-up and order are subject to change.

Lightning talk abstracts

Professional Football Planning
Lukas Bach, Sintef
Football is important to many people, just see the FIFA World Cup. When planning a tournament, this gives rise to many wishes from fans, associations, TV and clubs. Among these we have to find the optimal solution.

Influencing client meetings through conversation framing
Ashley Carreras, Loughborough University
Framing strategies – ie how you focus clients’ attention on key topics – influence the path of client/consultant meetings. Conversational analysis, within a framing context, has been used to analyse client/consultant meetings occurring between a series of facilitated workshops. What can we learn from this?

Expanding your set of objectives: Behavioural Decision Analysis in action
Valentina Ferretti, London School of Economics
Behavioural science has shown that, when confronted with strategic decisions, we tend to generate about half of the relevant objectives. What could be possible solutions? This talk will share lessons learned from a recent project with a Foundation in Hungary that provides educational opportunities to underprivileged children.

Agri-Food Value Chain Decision-Making under high risk and uncertainties; strategies for EU and South America
Jorge Hernandez, University of Liverpool
What are today’s key agri-food challenges? How are they affected by Food Safety, Climate Change, Nutrition and Food Waste considerations? I will be describing the international collaboration between UK, France, Spain, Italy, Poland, Chile and Argentina to design new agriculture production systems policies.

Anticipating the Future Using Problem Structuring Methods
David Lowe, Defence Scence and Technology Lab
Organisational change is typically triggered in reaction to performance issues.  This presentation examines whether routine internal use of Problem Structuring Methods can guide anticipatory change with reference to a case study drawn from the United Kingdom Ministry of Defence.

Connecting research with policy – some examples from TNO
Kerry Malone, TNO
TNO connects university research with business and policy, for governments and industry. The overview provides examples of research questions that span various Technology Readiness Levels – from 2 or 3 in the Early Research Programs, to 5-7 in our knowledge development programs.

Behaviour, behaviour, and behaviour in multi-criteria policy analysis
Gilberto Montibeller, Loughborough University
Every policy decision involves conflicting objectives and policy makers with distinctive perspectives. From my experience in supporting complex policy prioritisations there are three key behavioural factors for high-impact and successful interventions, which I will highlight in my talk.

Waste treatment & disposal supply chain optimisation (in practice)
Matteo Pozzi, Optit
Practical application of a multi-period, multi-commodity flow model to the waste treatment and disposal supply chain of the largest Italian Waste Management Company will be described, as an example of how OR can be integrated in real-life business processes (with great impacts!)

Matching number of trucks to demand
Michele Quattrone, Air Liquide
Air Liquide ships a huge amount of gas cylinders by trucks, serving more than 2 million customers. How can we estimate the cost and feasibility of an entire year of delivery from one truck to a group of customers? Can Montecarlo methods help?

Solving the real problem
Björn Thalén, Jeppesen
I will share experiences from airline crew planning and talk about how to Jeppesen leverages MIP-solvers in practice. Will also talk about the difference between applied OR science and applied OR practice.

Evaluation as a service for solving practical optimization problems
Szymon Wasik, Poznan University of Technology
Evaluation as a Service is a concept of storing all evaluation data and procedures in the cloud. Here, I will present an Optil.io – a unique platform that can be used to solve industrial optimization problems by evaluating algorithms solving them and by organizing crowdsourced programming challenges.

A comparison of different curbside waste collection systems
Sanne Wøhlk, Aarhus University
If countries are to meet the EU goal of 50% recycling, how can local authorities best adapt their systems for curbside collection of household waste? We analyze the driving needed under different scenarios, comparing ‘collect, then sort’ and ‘sort, then collect’ systems using single and multi-compartment vehicles.

Rolling stock maintenance scheduling
Daniel Palhazi Cuervo, Sintef
We discuss the scheduling of the maintenance tasks that need to be carried out by Mantena, the main provider of rolling stock maintenance in Norway. Handling multiple operational constraints and various levels of granularity make solving such problem particularly challenging.

Multiagent Planning in Aviation
Anders Albert, Sintef
From aviation, both airspace and airport management require multiagent planning. We will present a general problem of coordinating planning activities that require multiple resources controlled by different stakeholders in an airport setting.

Chair(s): Björn Thalén

Looking for a career outside the universities? Thinking about recruiting new colleagues?

Leave your details at the Making an Impact stand in the exhibition centre, and we will put you in touch.

Are you grappling with an issue where outside support would be helpful? Or would you like some personalised career or development advice? Sign up for a 20-minute 1-1 mentoring session with an experienced practitioner, to:

• Gain valuable advice – Your mentor’s experiences could shed a whole new light on your problem!
• Develop your knowledge and skills – They could help you identify the skills and expertise you need to succeed. They might advise you on where to go for the information you need.
• See new perspectives – At the conference you could speak to someone you’d never normally meet in your day to day work.
• Build your network – Maybe you’ll keep in touch after your session!

To arrange a meeting with one of our mentors (see below) visit the Making an Impact stand in the Exhibition Area.

To get the most from the session, prepare:

What problem do you want help/advice on; what would you like to know from your mentor?

 

 

Sofiane Oussedijk
Dr Sofiane Oussedik is the Leader of Decision Optimization within the Data Science Technical Sales Group at IBM in Europe. He has a PhD in Computers/Decision Optimization.

As a lead Decision Optimization and Data Scientist, he works with his team to support clients harness the value of Decision Optimization and Data Science within their business processes, to improve their operations and optimize usage of their resources, enabling high ROI from Decision optimization and Data Science.through successful projects in a wide range of industries, including manufacturing, energy and utilities, finance and logistics.

Topics for mentoring:

1. Switching from Academia to Practice
2. Dealing with time pressure and projects delivery deadlines
3. Optimization/Data Science consulting team management

Mentoring languages:  English, French

 

Matteo Pozzi
Matteo Pozzi is CEO of Optit srl, spin-off of the Operations Research team of the University of Bologna, founded in 2007 by Prof. Daniele Vigo.  He has an MSci in Physics, Diploma in International Relations, over 15 years in Management Consulting in Italy and UK.

Since 2010 he has led the growth of Optit, from start-up to the current state, with over 35 professionals developing and deploying solutions and services based on OR and Advanced Analytics.

Topics for mentoring:

1. From academia to practice
2. Integration of OR with Data Science, Machine Learning
3. From models to Decision Support Systems

Mentoring languages:  English, Italian

 

Ruth Kaufman
Ruth Kaufman fell into an Operational Research career by accident, having taken a maths BA in the school of social sciences at Sussex University.  This led to a long career in public sector OR and wider management, at London Transport, London Electricity, Department of Health, and Export Credits Guarantee Department (ECGD).

Ruth is currently Immediate Past President of the OR Society (having been President for two years up until 31/12/17), a visiting Senior Fellow at the London School of Economics, Chair of the EURO Working Group on Practice of OR, and  undertakes a variety of other activities including freelance consulting.

Topics for mentoring:

1. Professional and career development
2. Dealing with difficult customers
3. Managing professional staff

Mentoring language: English

Gerben Groenendijk

Often brilliant optimization results are not viewed as such by various stakeholder during OR implementations. There can be multiple reasons for this. Is the solution of high quality (low cost, high revenue), but are the stakeholders just not used to the solution which is presented to them? Or did we not optimize the actual problem that needed to be solved? Maybe there are conflicts in requirements: the stakeholders may not be aligned on what they find the best solution to their problem.

The challenge in practice is how to tackle the problems mentioned above. This means first of all finding which problem we actually have to solve. It also means that stakeholders have to accept that the solution presented to them could be different than they expect, but could still be of high quality when looking at the goals. Therefore it is essential to this process to define the success criteria. When multiple success criteria are defined, it needs to be identified what the prioritization is, especially when there are conflicting requirements. Quantifying the results and comparing them to a baseline is how we can show the stakeholders that the presented solution solves their problem.

During this session we will open the discussion with some example use cases. We are looking forward to your experience and input.

Nicole Havinga, Bryan Kuiper

Due to a sudden increase in the number of customers, parcel delivery company X realizes the time has come to apply optimization software for automatically creating efficient delivery routes. To prepare the shift from manual to automated planning, a team of OR specialists working for company X get together to compile a complete and unambiguous list of requirements, a clear definition of the problem that needs to be solved, and a desired outcome, supported by some insightful, illustrative examples. A full problem definition is presented to the optimization software consultancy.

If the situation described above was at all realistic, our job at ORTEC would be a lot simpler than it is. In practice, developing supportive optimization software for a wide range of customers operating in a fast-moving, competitive, and unpredictable market is no walk in the park. Most steps leading to the problem definition have gaps and difficulties that ORTEC first needs to resolve. This workshop is aimed at anyone interested in applying OR in practice. In an interactive way we will share our best practices, address challenges faced, and skills required when bringing OR into practice with help of real-life vehicle routing problems.

Ronald Buitenhek, Tim van Luxemburg

The amount of data in the world is constantly growing at an enormous pace. That offers opportunities for consultants to provide factual insights to their customers. The customer can profit from these insights by taking the right (fact-based) decisions. This is in fact what OR-consultants have been doing for a couple of decades already. Recently however, terms like Analytics, Predictive modelling, Machine learning, and Data Science have become much more popular than Operations Research. Does that mean that OR has become redundant? Can quantitative consultants support decision making without using or even knowing typical OR-models and OR-techniques? Are techniques the most important at all?

We think Data Science and Operations Research should go hand in hand. And we believe business understanding and modelling remain indispensable skills in our work.  We illustrate this with some of our typical projects of the last few years. And we invite you to discuss with us. How do you think quantitative consulting should be done?

Matteo Pozzi, Optit

The first conference of the Euro Working Group on Practice of OR was held in Paris-Saclay in February, resulting in an exceptional opportunity for more than 40 top level practitioners from various industries to confront each other on the issue of “measuring the impacts of OR projects”. Starting from the key highlights of the event, this session is aimed at carrying on this discussion leading towards the next event, to be held in Bologna next winter.

Come if:

– you were in Paris and cannot wait to bring this forward;

– you are keen to see how OR is applied in the real world;

– you have been wondering about what are the best KPIs to demonstrate the impacts of your OR models;

– you are curious to know what interests the most in the community of industrial service players who use OR to innovate and improve their processes.

Bring your own views, share them with us, join the community, contribute with your own experience to help OR becoming increasingly relevant in the industry.

Kim Warren

This session is aimed at analysts and consultants, and at teachers/trainers of business modeling. No prior experience is needed. Simulating business challenges and plans has until recently been difficult and time-consuming – but things have changed! We will take you through an “agile” process that makes quantified, working simulations practical for non-experts to build, quickly and reliably. And we need these tools. Spreadsheet-based methods just cannot handle the interdependencies, feedback, thresholds and intangible factors that infect all but the simplest cases. The resulting Living Business Models display with total transparency precisely the factors that management recognises, and exactly the causality that drives the performance outcomes they are interested in. Since everyone sees the same rigorous picture, they get that “joined-up” view that everyone says they want – fully explaining how everything has been changing, and allowing them to explore likely future outcomes under alternative assumptions, decisions and strategies.

Stefan Hug, PTV Group

What is the main goal of our customers in Transport Logistic Sector when it comes to route optimization? Not surprisingly – saving money. In order to achieve this goal it is essential to be able to calculate ‘practically adequate’ shortest paths in huge road networks extremely fast. Practical adequacy requires the consideration of a plethora of additional data such as predicted and live traffic information, vehicle- and time-dependent toll prices as well as physical and legal limitations. Since this data is becoming increasingly accurate and available it is essential to exploit this technological progress in the calculation to get the most realistic result. As vehicle characteristics and start times vary in shortest path web services with every single calculation the use of speed-up techniques often has limitations in real life use-cases.

In this session we present challenges we as PTV Group had to face when incorporating different data in our shortest path algorithm, we show how we dealt with these challenges in our applications and use-cases and invite you to comment and share your own experiences.

Sofiane Oussedik, IBM

The session is aimed at OR practitioners who deploy applications to business users. Deploying applications requires a functional architecture and the corresponding technical architecture to be well defined. Most of the Euro conference is dedicated to OR techniques and applications; this will be a short introductory session to the 50-70% of work that needs to be done to get an OR model to deliver value to a business user. Participation in the discussion from practitioners is very welcome.

Joaquim Gromicho

When it comes to modern and widely accessible programming languages two camps seem to attract comparable amounts of attention and collectively dominate: R and Python.

R seems to be preferred by the statistics and data science communities, while Python seems to be the language of choice of the artificial intelligence and mathematical optimization communities. This classification is by no means limitative, since there are many other good choices such as Julia, Jump, Java, C, C++, C#, Matlab, Octave, etc.

Nevertheless, the perception that R is not for optimization may interfere with the possibility to merge great libraries and tooling for (big) data analysis with optimization.

This tutorial gives data scientists familiar with R a glimpse on how to use it for optimization and other scientists familiar with optimization a glimpse of the power of R. The tutorial includes examples of data driven optimization using R both as a platform to develop heuristic methods and as an interface to mathematical optimization solvers. To get the most from it, please bring your own laptop with an internet connection.

Ruth Kaufman

Ethical practice is important if practitioners are to retain the trust of their clients as well as their own self-respect. There are a number of codes of practice available for anybody who wants guidance on this. But even for the most ethical practitioner, real life can sometimes throw up ethical dilemmas, or tempt us into imperfect behaviour: perhaps fulfilling our contract obligations may require turning a blind eye to unethical conduct from the client; or the client does not want to engage with the caveats around our analysis; or we want to convince the client to employ us even where the work area is new to us, to name just a few situations where a purist approach may result in no OR work happening at all.

This session will start with a short introduction to the ethical dilemmas that can arise in everyday OR, and then open out into discussion inviting participants to share their own experiences and approaches. It is intended to give an opportunity to people engaged in practical OR to reflect on their own practice and to share learning and ideas with others, so that they may be better equipped to make good choices in future.