Impact Magazine Autumn 2022

EDITORIAL

This issue has a European flavour – three articles come from continental Europe. You can read about a German University helping to plan beer production in Switzerland, a French software company helping a client’s customers to optimally move soil at construction sites and an Italian company helping their clients to improve their energy consumption, waste management and logistics operations. A common theme to these examples of O.R. in action is the powerful use of optimisation tools. I’m only too aware of the use of optimisation, as the authors have put up with me changing their optimization to optimisation.

Optimisation is also the basis of US data analytics company FICO’s work for car dealerships, to allow them to make financial offers to potential customers. It is clear that mathematical optimisation continues to make a significant difference to many organisation’s operations.

But analytical work is not solely concerned with optimisation. The report from British Red Cross describes the work of a team of researchers and analysts to help people receive the care and support they need in emergencies by producing evidence at short notice to give decision-makers insight.

I hope you enjoy reading this issue, which shows how O.R. and analytics have made, and can make, an impact. All issues are available at https://issuu.com/orsimpact Please subscribe to this free magazine at https://www.theorsociety.com/impact/

This is my last issue as editor. It is ironic that my last cover features a beverage I don’t drink. But cheers everyone! It has been a privilege over the last eight years to have brought to the attention of readers some of the best examples of the impact of Operational Research. Careful observers of the image at the top of this page in each issue will have noticed that my editorial efforts have miraculously stemmed the ageing process, such a pleasure it has been.

Graham Rand

The OR Society is the trading name of the Operational Research Society, which is a registered charity and a company limited by guarantee.

Seymour House, 12 Edward Street, Birmingham, B1 2RX, UK

Tel: + 44 (0)121 233 9300, Fax: + 44 (0)121 233 0321 Email: email@theorsociety.com

Secretary and General Manager: Gavin Blackett

President: Edmund Burke Editor: Graham Rand g.rand@lancaster.ac.uk

Associate Editor: James Bleach

Print ISSN: 2058-802X Online ISSN: 2058-8038 www.tandfonline.com/timp

Published by Taylor & Francis, an Informa business

All Taylor and Francis Group journals are printed on paper from renewable sources by accredited partners.

OPERATIONAL RESEARCH AND DECISION ANALYTICS

Operational Research (O.R.) is the discipline of applying appropriate analytical methods to help those who run organisations make better decisions. It’s a ‘real world’ discipline with a focus on improving the complex systems and processes that underpin everyone’s daily life – O.R. is an improvement science. For over 70 years, O.R. has focussed on supporting decision making in a wide range of organisations. It is a major contributor to the development of decision analytics, which has come to prominence because of the availability of big data. Work under the O.R. label continues, though some prefer names such as business analysis, decision analysis, analytics or management science. Whatever the name, O.R. analysts seek to work in partnership with managers and decision makers to achieve desirable outcomes that are informed and evidence-based. As the world has become more complex, problems tougher to solve using gut-feel alone, and computers become increasingly powerful, O.R. continues to develop new techniques to guide decision-making. The methods used are typically quantitative, tempered with problem structuring methods to resolve problems that have multiple stakeholders and conflicting objectives.

Impact aims to encourage further use of O.R. by demonstrating the value of these techniques in every kind of organisation –large and small, private and public, for-profit and not-for-profit. To find out more about how decision analytics could help your organisation make more informed decisions see https://www.theorsociety.com/about-or/or-in-business/ O.R. is the home to the science + art of problem solving.

7 OPTIMISATION KEEPS THE BEER FLOWING

Markus Mickein, Matthes Koch and Knut Haase report how a production planning system, designed for the brewing industry, has been implemented for the Swiss brewery Feldschlösschen

16 MODERN CAR FINANCE: OPTIMISING ALTERNATIVE DEAL STRUCTURES

Richard Cowley explains how FICO developed an optimisation tool to enable financial counteroffers to customers to be made in car dealerships

20 RAPID INSIGHTS FOR HUMANITARIAN AID IN THE UK

Matthew Gwynfryn Thomas explores how British Red Cross provided insight in three recent emergencies: Covid, destructive winds and the invasion of Ukraine

25 OPTIT - FROM ANALYTICS AND OPTIMISATION TO OPERATIONAL EXCELLENCE

Michela Mattei reports on Optit’s work with clients to improve their energy optimisation, waste management, logistics and supply chain management

32 IMPROVING THE ENVIRONMENTAL IMPACT OF CONSTRUCTION SITES

Nicolas Blandamour shows how software company LocalSolver enabled the reduction of transportation costs and carbon footprint of truck journeys at construction sites by over 10%

4 Seen Elsewhere Analytics making an impact

11 Further suggestions to support your efforts to understand the potential impact of climate change

Nicola Morrill discusses two further approaches, simulation and forecasting, that can help organisations reduce their carbon footprint

29 Universities making an impact Brief report of a postgraduate student project

35 Measure for Measure

Geoff Royston considers issues of measurement informed by Beyond Measure by James Vincent and The Tyranny of Metrics by Jerry Muller

DISCLAIMER

The Operational Research Society and our publisher Informa UK Limited, trading as Taylor & Francis Group, make every effort to ensure the accuracy of all the information (the “Content”) contained in our publications. However, the Operational Research Society and our publisher Informa UK Limited, trading as Taylor & Francis Group, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by the Operational Research Society or our publisher Informa UK Limited, trading as Taylor & Francis Group. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. The Operational Research Society and our publisher Informa UK Limited, trading as Taylor & Francis Group, shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Reusing Articles in this Magazine

All content is published under a Creative Commons Attribution-NonCommercial-NoDerivatives License which permits noncommercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.

SEEN ELSEWHERE

STATS AND BATS

SAS has created ‘the Batting Lab’, an interactive experience that uses AI, computer vision and the IoT with analytics to help children better use stats and bats. Its primary aim is to improve one’s swing in games such as baseball but could also be applied to any game involving the use of a bat, racket or even golf club. It secondary aim is to make statistics and analytics more accessible, relevant and fun. It can boost data literacy and help young people thrive in a world increasingly driven by data and analytics.

More at https://bit.ly/StatsandBats

FALLACIES OF AI

Dr Harvey Lewis, partner at Ernst & Young, opened the OR Society’s recent Analytics Summit with a talk entitled “Tackling the fallacies of AI in business”. He argued that it was a fallacy to believe that AI could be superior in every aspect to human level performance. It is currently believed that solving quantitative reasoning problems using machine learning will require significant advancements in model architecture and training techniques. He identified seven common fallacies: uniqueness, value, completeness, accuracy, sophistication, automation and intelligence. One of the big issues associated with AI that had to be considered, he said, was that of trust, “how can we trust the machines and how can we trust the people who build the machines?”. He answered the question “How can AI help us realise transformation in the future?” by suggesting what was needed was:

1. Innovation at scale – AI to be integrated with wider strategic transformation initiatives.

2. Humans at centre – AI systems to be designed with people in mind.

3. Technology at speed – AI embraced to enhance creativity to broaden and deeper skills.

during significant moments in people’s lives such as approaching retirement or facing financial shock.

BEYOND BIG DATA

MAX MOULLIN GOES TO WASHINGTON (VIRTUALLY)

In 2015, work of OR Society member Max Moullin was featured in Impact (see https://bit.ly/PSSMaxMoullin). His Public Sector Scorecard (PSS) is an integrated problem structuring and performance management framework for the public and third sectors. It is workshop based, involving staff and service users and its outcome focus is ideal for working across organisational boundaries. In 2022 Max was the only non-US speaker at the US Government Performance Summit which was attended by over 100 senior federal and state government officials. He presented a talk entitled “Performance Management without the Blame: Improving and Evaluating Services across Organisational Boundaries with the Public Sector Scorecard”. This proved very relevant to the US President’s Management Agenda which includes developing interagency teams to improve the public’s experience

Ocient, a leading hyperscale data analytics solutions company serving organisations that derive value from analysing trillions of data records in interactive time, has released a report, (https://bit.ly/BBDOcient), “Beyond Big Data: The Rise of Hyperscale.” Its purpose is to uncover key trends around how organisations are managing the shift from big-data volumes toward ingesting, storing and analysing hyperscale data sets, which include trillions of data records, and the expected technical requirements and business results from that shift. Key findings of the survey include:

• Data analysis is tied to financial success. More than 85% of C-level respondents indicated there is a strong relationship between implementing faster data analytics and growing the company’s bottom line.

• Data workloads are getting bigger, faster and more complex. 97% of respondents indicated the volume of data managed by their organisa tion will grow fast to very fast over the next one to five years.

• Legacy systems aren’t built to handle hyperscale data analysis. More than 59% of respondents plan to switch data warehousing solutions, and 46% of respondents indicated a legacy system is motivating them to switch.

• Security and compliance are among top concerns. 63% of respondents

said maintaining security and compliance is a challenge in scaling data volumes and analytics to hyperscale workloads.

• Staffing is a challenge. 49% of respondents wrote there is a lack of talent to analyse their data.

“Data analysis is no longer a ‘nice-tohave’ for organisations. Hyperscale data intelligence has become a mission-critical component for modern enterprises and government agencies looking to drive more impact and grow their bottom line. With the rapid pace of growth, it’s imperative for enterprises and government agencies to enhance their ability to ingest, store, and analyse fast-growing data sets in a way that is secure and cost effective,” said Chris Gladwin, co-founder and CEO, Ocient. “The ability to migrate from legacy systems and buy or build new data analysis capabilities for rapidly growing workloads will enable enterprises and government organisations to drive new levels of agility and growth that were previously only imaginable.”

NEW DIGITAL WATCHDOG

The Digital Regulation Cooperation Forum (DRCF) – a collaboration of the Competition and Markets Authority (CMA), the Information Commissioner’s Office (ICO) and the Office of Communications (Ofcom) –has been joined by the Financial Conduct Authority to address the concerns of digital technologies within the UK. The DRCF says that algorithms, and particularly those utilising modern AI and ML practices pose significant risks (such as harmful biases leading to discrimination or reinforcement of inequalities) if not adequately checked, regulation

is therefore necessary to protect companies and citizens.

DRCF approach their role with a focus on three overarching goals:

• to promote greater coherence, so that, where regulatory regimes intersect, the DRCF can resolve tensions and clarify regulatory positions;

• to work collaboratively on areas of common interest and jointly address complex problems; and

• to work together to build the necessary capabilities, developing from the learnings of each regulator.

It is hoped that the increased collaboration of regulators within the UK, and by focusing on the three primary goals of the DRCF, that the UK will serve as a coherent and responsive location for the development of digital markets.

See https://bit.ly/DigitalWatchdog

Learning “is most appropriate for technologists in high-stakes domains who care about the broader impact of their work, have the patience to think about what they’re doing before they jump in, and do not shy away from a little math.”

In writing the book, Kush says, “I have taken advantage of the dual nature of my job as an applied data scientist part of the time and a machine learning researcher the other part of the time. Each chapter focuses on a different use case that project managers, data scientists, and other practitioners tend to face when developing algorithms for financial services, healthcare, workforce management, social change, and other areas. These use cases are fictionalised versions of real engagements I’ve worked on. The contents bring in the latest research from trustworthy machine learning, including some that I’ve personally conducted as a machine learning researcher”.

FASTER OPTIMISATION

TRUST IN MACHINE LEARNING

Kush Varshney of IBM has authored a free book (https://bit.ly/KVarshney). He says that Trustworthy Machine

Combinatorial optimisation problems are difficult to solve at high speed and at a reasonable computational cost. Toshiba has developed a ‘fast’ Simulated Quantum Bifurcation Machine+ (SQBM+) on Azure Quantum, based on its Simulated Bifurcation Machine (SBM), using an Ising model solver that can solve complex and large-scale combinatorial optimisation problems with up to 100,000 variables at high speed. There are two algorithms available: Ballistic Simulated Bifurcation algorithm (bSB) designed to find a good solution in a short time and; Discrete Simulated Bifurcation algorithm (dSB) which finds more accurate solutions quickly. See https://bit.ly/FasterOptimisation

© Kush R. Varshney

OPTIMISATION KEEPS THE BEER FLOWING

MARKUS MICKEIN, MATTHES KOCH AND KNUT HAASE

THE INSTITUTE FOR LOGISTICS, TRANSPORTATION, AND PRODUCTION OF THE UNIVERSITY OF HAMBURG implemented a model-based production planning system at the Swiss brewery Feldschlösschen. Feldschlösschen is the leading brewery in Switzerland, with a market share of 40% and revenue of nearly CHF 1 billion (approx. £776 million). The company was founded in 1876 and acquired by the Carlsberg Group in 2000. The largest production site in Rheinfelden produces 1.8 million hectolitres per year.

Beer manufacturing contains brewing, fermentation, maturation, filtration, and filling. Feldschlösschen cooperates with the project team of the University of Hamburg because there is no suitable industrial production planning solution that covers brewery needs, such as unique storage tank operations with processing times. Feldschlösschen defined three requirements for the planning system: considering all operations and resources, modelling the multilevel production system and process restrictions, and optimising production and inventory schedules. Therefore, we

designed a planning system customised for breweries. The mathematical program optimises each resource of the entire product system to support holistic planning.

Figure 1 illustrates the brewery production process. At first, the brew house brews the base beer. The base beer stays in the storage tanks for fermentation and maturation. Next, the filtration filters undesirable particles to produce semifinished beer. Buffer tanks keep semifinished beer until further processing. Finally, the filling lines fill the beer in bottles, cans, and kegs. The warehouse stores the finished beer until delivery to the customer. Various production and storage tank resources are available at each production stage. In addition to the general brewing process, particular beer types such as speciality beer, alcohol-free beer, and mixed beer require additional processing with special equipment.

Challenges regarding changing consumer preferences, complex production processes, and transparent planning necessitate the application of an advanced analytics tool

CHALLENGES AND APPLICATION

The brewing industry is confronted with new challenges as a result of changing consumer preferences. The demand for speciality and alcohol-free beer has increased in recent years. As mentioned before, such beer types require special equipment and hence adaptations to the production system. Feldschlösschen’s product portfolio includes 220 finished goods out of 100 semifinished goods. The production

system contains 13 production resources and 8 storage groups. The growing number of products and resources increase the planning complexity.

Since most supply chain planning issues have strong dependencies, the decision making is centralised in the supply chain planning department. For this reason, the planning department strives at a transparent planning procedure to simplify the decision communication and improve solution acceptance. Challenges regarding changing consumer preferences, complex production processes, and transparent planning necessitate the application of an advanced analytics tool. Therefore, the project team developed a model-based production planning system.

The planning department requires a comprehensive scenario analysis tool for tactical tasks and strategic issues as well as an operative planning tool. Tactical planning tasks include selecting shift schedules or identifying temporarily required capacity extensions. Strategic planning issues contain evaluating the impact on the production system regarding changes in product portfolio, production volume, and production equipment. The detailed process mapping for the

tactical and strategic planning activities guarantees operational feasibility. The operational planning tool provides detailed schedules that includes production quantities, inventory levels, required overcapacities, production sequences, and storage tank allocations.

PLANNING SYSTEM

The planning system consists of a user interface for data collection and validation as well as a dashboard for visualising the optimisation results embedded in a cloud-based optimisation framework (see Figure 2). The user-friendly interface is designed for production managers and planners without a mathematical optimisation background. The server stores the data and hosts the optimisation engine.

The employed brewery-specific production planning problem considers relevant process restrictions, such as the available number of storage tanks, different storage tank groups, and processing times. The planning system implements the optimisation model and algorithm in GAMS/CPLEX. The computationally intensive optimisation runs on scalable cloud services to reduce computing times and hardware costs.

The user interface provides the input data collection from various sources, i.e., master data from the ERP system and additional data from spreadsheets. The basis data query includes the product, resource, and demand data for operational planning. Collecting and adjusting data sets enables the creation of different production scenarios for tactical and strategic analysis. The user interface visualises the collected data for quick checks and provides a comprehensive data validation to ensure structurally correct data sets for optimisation.

The integrated visualisation tool displays the optimisation results in dashboards. Derived key performance indicators predict the future performance in operational planning, e.g., utilisation and costs. In addition, the indicators quantify the impact of different production scenarios for comparison. The management report contains an overview and a detailed dashboard. The overview dashboard enables a simple identification of bottleneck resources. The detailed dashboard provides production and storage tank schedules.

The user-friendly interface is designed for production managers and planners without a mathematical optimisation background

ADDED VALUE

The customised planning system applies to operational, tactical, and strategic planning activities. Considering all relevant process stages and restrictions guarantees practicable production schedules. In addition, the detailed plans reduce the manual planning effort and the need for reactive capacities due to improved

planning quality. The scenario analyses improve cross-divisional decision making by quantifying the impact on the entire production system of planning decisions by other departments in various production scenarios.

Besides nonmonetary benefits, the new planning system reduces investment costs by avoiding reactive capacities and inefficient equipment as well as realising operational cost savings by optimised production and inventory schedules.

SUCCESS FACTORS

The involved departments were integrated in the software development process from the beginning to ensure the acceptance of model-based planning solutions. Besides higher trust in the optimisation results, this procedure gets detailed process knowledge from expired planners and customise the tool for the enduser's need. Thus, the project team implemented a supportive data validation and calculation report to support users in data collection and result interpretation. The data validation ensures correct data sets for optimisation. Additionally, the calculation report displays unexpected relaxations of constraints during the optimisation, for example, the used capacity exceeds the given capacity. This information supports planners in identifying bottlenecks and adjusting production scenarios.

CONCLUSION

The project team developed a modelbased production planning system designed for the brewing industry and the implementation at the Swiss

brewery Feldschlösschen. The main advantage of the proposed system is the consideration of the relevant process stages and restrictions as well as the application of different planning activities in one planning system. Considering the process-specific constraints improves the planning quality. Integrating all planning activities in one planning system guarantees operational feasibility for tactical and strategic decisions. Applying the new planning system supports Feldschlösschen in reducing planning effort, reactive capacities, inefficient investments, and operational costs.

Feldschlösschen’s Head of Supply Chain Planning & Product Change Management said: “The software supports strategic and tactical planning decisions in standardised reports. The most significant value added is the analysis of the interactions between production stages as the consequence of decisions. This enables us to better quantify investment requirements and evaluate future strategies. It reduces investment costs by identifying actual need and realises operational cost savings by analyzing strategic scenarios in a holistic manner. Furthermore, it improves decision communication to the relevant departments.”

The most significant value added is the analysis of the interactions between production stages as the consequence of decisions

OUTLOOK

The project team currently works on higher modularity, integrating advanced optimisation methods, and considering related planning problems to extend the planning system. Higher modularity

simplifies adaptations of the planning system to other production sites and industries. Furthermore, integrated advanced optimisation methods are able to cope with further challenges,

for example, stochastic programming to manage demand uncertainty and multicriteria optimisation to balance different company targets. Moreover, considering additional related planning problems improve holistic decision making in supply chains, e.g., production network design.

Markus Mickein is a PhD candidate in operations research and management science at the Hamburg Business School, Universität Hamburg. His research focuses on advanced optimisation approaches for lot-sizing in the process industry.

FOR FURTHER READING

Matthes Koch is a specialist in operations research and managing partner at the consultancy DESIOR. His research work through a PhD program at Universität Hamburg focused on management systems for large-scale problems in crowd management.

Knut Haase is the director of the Institute for Logistics, Transportation, and Production at Universität Hamburg. His research topics are focused on optimisation approaches for solving large-scale problems with applications in logistics, public transport, and crowd management.

Mickein, Markus, Matthes Koch and Knut Haase (2022). A Decision Support System for Brewery Production Planning at Feldschlösschen. INFORMS Journal on Applied Analytics 52:158–172.

© Pavel

FURTHER SUGGESTIONS TO SUPPORT YOUR EFFORTS TO UNDERSTAND THE POTENTIAL IMPACT

OF CLIMATE

Nicola Morrill

CHANGE

mess to enable a focused exploration of an issue related to the system to be explored.

MOVING INTO IDENTIFYING THE ISSUE TO BE CONSIDERED

If I had an hour to solve a problem and my life depended on it, I would use the first 55 minutes determining the proper question to ask.

Albert Einstein

In my last column, I explored ways that O.R. and Analytics can help with understanding the potential impact of climate change. Here I will explore where more traditional areas of O.R. and Analytics are able to help: covering discrete event simulation and forecasting.

EXPRESSING CLIMATE CHANGE AS A PROBLEM

As a quick re-cap, climate change, expressed as a problem, is an example of a Megatrend and a ‘wicked problem.’ Megatrends are typically slow to form; persist for a long time (circa. 10-15 years); occur at a global or large scale; and are visible and well known to everyone. A ‘wicked problem’ is generally, a social or cultural problem that is difficult or impossible to solve—normally because of its complex and interconnected nature.

My previous article shared some of the approaches and examples that are useful in exploring such situations with a principle focus on increasing understanding. Here, I focus on where an O.R. practitioner extracts a part of the complex

Considering the questions to be explored is a key aspect in increasing understanding of a problem and preparing to address it. This is not a trivial thing to do and is something that warrants a decent amount of time spent on it; sadly, this is too often not the case. Exploring and shaping the question is an area that O.R. as a discipline has much to offer – look to Problem Structuring Methods for more details. The nature of the question will be shaped by whether it is a puzzle, problem or mess that is to be explored. Figure 1 , taken from Mike Pidd’s 1996 book Tools for Thinking, provides an insight into where these vary along the axes of problem formulation and the ‘solution’. The difference for each is related to how much agreement there is on each axis. Puzzles tend to be tightly bound, have lots of certainty and a clear answer, where messes have lots of different interpretations and there is always debate. This is in contrast to a problem where it is possible to have agreement around its focus but as there is unlikely to be one definitive answer, the solution tends to be less agreed. This links into the notion that ‘problems are social constructs,’ where people see issues in diverse ways; re-enforcing the importance of diversity of thought in addressing problems.

Welham (2021) (https://bit.ly/theaccidentalmarketer) provides a clear and succinct overview of the differences between puzzles, problems and messes and discusses the importance of problem formulation; so, it is not just O.R. people that see how tremendously important it is!

In the context of considering how O.R. can help with questions organisations may wish to address related to climate change, McKinsey (see A framework for leaders to solve the net-zero equation | McKinsey ) have identified a range of ‘requirements for ‘solving’ the net-zero equation’ (see Figure 2 ). They have articulated a series of questions related to these requirements that some businesses may

FIGURE 1 PUZZLES, PROBLEMS AND MESSES

© Mike Pidd

FIGURE 2 NINE REQUIREMENTS FOR SOLVING THE NET-ZERO EQUATION

wish to consider in supporting their efforts towards netzero. Clearly this is not an exhaustive list or suitable for all organisations but provides an insight into the micro level questions to consider.

LOGISTICS, SUPPLY CHAINS AND DISCRETE EVENT SIMULATION

Discrete Event Simulation (DES) is a modelling approach used to model real world systems where they can be separated into a set of logical processes that autonomously progress through time. If it is possible to create a process map of the system being considered, it is likely that DES can be applied. In the McKinsey questions, the one related to bottlenecks is well suited to DES.

A good, accessible overview of DES is provided in an Impact article (see https://bit.ly/Elder2015) by Mark Elder. Figure 3 is from a 2018 Impact article on modelling logistics at the Port of Dover: (https://bit.ly/Dover2018). One area the work explored was understanding how the Port could best handle future volumes of traffic and the team decided to use DES as ‘the Port consists of a series of process steps, each preceded by orderly queues.’ The layout and processes of the Port are shown.

This was then used to create a simulation model where the team modelled the current day system as well as modelling potential future scenarios (see Figure 4).

Here are a few examples of where DES has been applied to issues related to climate change.

• Estimating CO2 emissions that result from equipment usage during a construction project with the goal of allocating resources to make projects more eco-friendly. (See https://bit.ly/EstimatingCO2)

• Simulation modelling and analysis for sustainable supply chains. This work identified the benefits of consolidation of transport orders, which would result in a fewer number of trucks on the road. (See https://bit.ly/SCSustainable)

FIGURE 3 SCHEMATIC OF THE PORT OF DOVER EASTERN DOCKS (OUTBOUND TRAFFIC)

• Simulating vehicle movements on the road to explore the impact of traffic congestion on CO2 emissions. The work identified that factors such as synchronization or desynchronization of traffic lights, mode of dispatch rates, and route configurations were important factors to consider. (See https://bit.ly/DESGreenSC)

• An analysis of carbon friendly supply chains, based on a 3-tier system, created a simulation model that ran a range of scenarios. It was considering the CO2 emissions along supply chains from freight energy use to inventories storage and identified that the location of the first-tier supplier is particularly important.

(See https://bit.ly/CarbonFriendlySC)

DES is an O.R. approach that has wide applicability in supporting efforts on looking at climate change with a focus on specific types of questions. Some of the software packages, such as Simul8, have carbon emission features built into them.

CONFIDENCE IN ACHIEVING TARGETS AND FORECASTING

Forecasting is another area of O.R. that has use when exploring issues related to climate change. Forecasting is a

commonly used word and in the context of this article I use the term to refer to the host of formal analytical approaches to undertaking a forecast. An article in the Harvard Business Review (https://bit.ly/forecastingtechniques) provides an accessible insight into the different ways to formally undertake forecasting and a paper by Fildes et al (2008) presents a review of forecasting and also provides a useful insight into different techniques (see https://doi.org/10.1057/palgrave.jors.2602597).

In April 2022, EUROCONTROL issued a new long-term air traffic forecast, focussing out to 2050, (see https://bit.ly/EuroControl2050), which provides insights into how aviation may be able to achieve net-zero emissions over that period. This forecasted a range of scenarios and included consideration of growth in demand and the changing nature of the aircraft fleet. Figure 5, from the report, provides an insight into one aspect of the forecasting work.

Other examples, include:

• Scottish Government Zero Emission Energy for Transport Report, which provides forecasts out to 2045 across a range of scenarios for national demand for electricity and hydrogen. The underlying question for the work

FIGURE 4 BASELINE SIMUL8 MODEL OF THE PORT OF DOVER EASTERN DOCKS

undertaken related to decarbonising the transport sector (see https://bit.ly/TransportScotand2022).

• The Deloitte carbon forecasting model was used by the organisation to improve the sustainability of its own business practises, resulting in changes to its air travel and lease car policies. The question sitting behind the modelling was the goal of carbon neutral business operations within Deloitte Netherland by 2025 (see https://bit.ly/DeloitteNL)

• In the investment world there are a few carbon forecasting models with a focus on supporting sustainable equity decisions.

In some examples related to organisations improving their climate risk forecasting there were discussions around what to do when there is no data available. This is more common a challenge than people may think. It is an important consideration and something that O.R. is well versed in addressing.

Finally, it would be remiss of me not to mention the importance of validation, verification, and governance over what will be complex (simulation or forecasting) models.

SHAPING MY NEXT PIECE

If there is something, related to O.R., that you would like me to consider for future columns in Impact then please get in touch. The goal is to share the discipline with users / potential users of O.R. by highlighting how it could support ‘business’ challenges they may be facing.

WANT TO LEARN MORE?

The OR Society runs training courses on much of the above if you want to bolster your in-house team. We are an active community and there are various events running through the year that may be of interest.

Nicola Morrill is a Systems Thinking Consultant at Dstl, a certified coach and mentor and the current Diversity Champion of the OR Society. She writes in a private capacity – all views expressed are her own and all examples are available in the open domain. You can contact her on Nicola.Morrill@googlemail.com

MODERN CAR FINANCE: OPTIMISING ALTERNATIVE DEAL STRUCTURES

RICHARD COWLEY

THE CRIPPLING

CHALLENGES

COVID AND THE CURRENT COST OF LIVING CRISIS have presented to the motor finance sector will shape the industry’s choices for years to come. But a major problem that was already evident well before the pandemic can be resolved with advanced analytic science. It is an issue facing all those in the industry, not unique to any single organisation.

Car dealers want to get more customers through their doors, virtual

or physical, and quickly back out again with a signed contract and a set of new keys. Of course, for many, part of the process is setting up the finance and this is where many lenders struggle. Automated systems will reject a finance request without making a compelling counteroffer that could win them the business.

Most auto lenders do not have an automated counteroffer process, which means many lenders handle this process manually, based on simple rules. This

©

Torsten Dettlaff

is a time-consuming and inefficient process. The evolution of the online car buying experience is creating a whole new consumer expectation of speed and convenience. Auto finance providers that want to support traditional motor retailers rising to this challenge need to respond with the right tools for the job.

IMPROVING THE COUNTEROFFER

Automating the counteroffer process is much more complex than it first appears. It needs to be comprehensive and consistent in line with FCA (or other regional) regulations. Nearly every possible counteroffer needs to be evaluated for acceptance, and then the bundle needs to be rapidly reduced to those that will interest the buyer.

The process should take into account that some buyers care most about the interest rate, some care most about the deposit, while others care most about the monthly payment. It must also be fast. While many consumers grudgingly accept long waits in the car purchase process as the cost of doing business, this patience will not be permanent. In the UK, motor finance applications can take up to two working days to complete, according to MoneyExpert.com (see https://bit.ly/MoneyExpertdotcom).

Some of the delay is caused by the ‘rehash procedure’ that the dealer has to conduct with their lenders to arrive at an acceptable offer.

The difference between the ‘before experience’ and the ‘after experience’ is the speed with which the finance and insurance provider in the back office was able to get a differentiated deal and approved counteroffer back into the hands of the salesman. Not just a long list of options to sift through, but

a small set of meaningful options that the borrower will want to consider that also work for the lender and dealer. Plus, of course, any approved offer needs to conform to the lender’s credit risk criteria, which can be arduous to determine when the offer terms are constantly changing.

This problem cannot be comprehensively solved within a lender’s existing rules engine. What’s needed instead is mathematical optimisation – the analytic technology that can search through thousands of potential combinations to deliver the best possible counteroffers within the lender’s and dealer’s constraints and return those offers to the dealer in seconds.

What’s needed instead is mathematical optimisation – the analytic technology that can search through thousands of potential combinations to deliver the best possible counteroffers within the lender’s and dealer’s constraints and return those offers to the dealer in seconds

These are the challenges that FICO set out to solve with a solution we call alternative deal structure optimisation (ADS). Optimisation solvers factor in credit and pricing logic plus behavioural analytics, such as the likelihood of response. This kind of optimisation can help car dealers improve the customer experience and underwriting levels. ADS allows dozens, hundreds, or even thousands of options to be evaluated, helping to ensure that the best alternative offers are discovered. It uses a dynamic search path controlled by business objectives

and policy constraints, eliminating the need for lists of all possible options to be considered.

In the indirect auto space, ADS helps balance the competing needs of three parties: consumers, lenders and dealers. And, as long as car purchases are fulfilled by dealers, even online-only purchase/financing will involve some type of dealer commission. Some of the richness of ADS optimisation involves finding that balancing act: how do lenders keep dealers and customers happy while meeting internal profit/ volume/risk objectives?

The most common use of ADS is in generating counteroffers, whether prepopulating for underwriters (manual review) or returning automatically to dealers (conditional approvals). However, ADS has a variety of applications:

• Approval with options - Generate additional options when the initial deal structure is system-ap proved.

• Inventory decisioning - Request for approved financing options on all available inventory at a dealer.

• Pre-approval - Request for approv al limits not linked to any specific vehicle.

• Manual review - Generate alter natives for the underwriter before queuing for manual review.

• Buyer self-service - Guide dealer/ customer24 on approved struc tures via the self-service portal.

REAL-TIME OPTIMISATION

In developing ADS, we needed to create customer-level, real-time optimisation that enables personalised offer generation, rather than an offline, portfolio-level optimisation formulation, which is more common. To do this, we enabled the creation of a multi-objective business strategy through usage of two or more optimisation slots, where each optimisation slot is a unique mathematical formulation, and efficiently solves all optimisation slots within a strategy simultaneously.

Our ADS solution supports objective/constraint definitions for multiple metrics related to consumer needs (e.g., monthly payment amount, loan amount, interest rate, etc.), dealer needs (commission, backend allowance, etc.) and lender needs (e.g., customer-level risk level, customer-level profitability). The solution includes a simulation tool for a strategy manager to evaluate candidate strategies on large volumes of transactions (whether historical or manufactured).

SUCCESS STORY

A large auto finance company in the US looked to ADS when it was forced to reject too many prospective buyers: buyers it believed could have been profitably financed if the deals had been structured properly.

The company wanted to ensure that the maximum number of customers walked out the doors of its dealerships with deals in hand. However, its

credit analysts generated too many credit exceptions, which reduced profitability. That practice also created inconsistencies among individual dealerships, reducing risk manager monitoring effectiveness and putting the company at risk of predatory lending practice accusations.

In addition, manual deal financing processes were time-consuming: credit analysts had to pour over data caseby-case without necessarily arriving at the optimal deal. The company wanted to transfer the majority of deal structuring from its individual credit analysts to its centralised risk managers, who could structure deals to accommodate shifting parameters without sacrificing profitability. It also sought to offer a range of best options to prospective customers so that it could close more deals at a higher loan-to-value rate.

The auto finance company's new chief information officer proposed a fresh vision for structuring deals. By implementing FICO® Optimization Solution for Alternative Deal Structure, it could use the power and flexibility of the most advanced mathematical modelling environment with userfriendly visualisation and control to meet complex pricing challenges.

The company wanted to ensure that the maximum number of customers walked out the doors of its dealerships with deals in hand

AN ALTERNATIVE DEAL STRUCTURE SYSTEM WITH REAL-TIME FLEXIBILITY

When customers are ready to purchase vehicles, the FICO system quickly

generates up to ten different deal structures, all of which are profitable for the company and conform to its lending policies. The credit analyst selects three of those ten dealswhichever fit the particular customer’s situation best - and negotiates with the customer within the framework of those deals.

Using the ADS system, the auto dealer has revamped its deal structuring processes and is experiencing multiple benefits. These include:

• Increased loan approval rates for better dealer and customer satisfaction

Because credit analysts focus on deal execution instead of creation, the company now can ensure that its deals all fall within acceptable ranges. This alternative deal structuring also means that dealerships can speed up their negotiations and make their pricing more attractive to customers.

• Reduced annual losses by up to $12 million

Alternative deal structures are helping convert missed opportunities into deals that maximise profitability. Having alternative structures means that the company can reduce loss ratios, based on improved loan-to-value rates, multiple payment lengths and specific terms. The annual losses were cut by as much as a further $12 million.

The annual losses were cut by as much as a further $12 million

•

Faster negotiations, lower labour costs

The company estimates that using the solution will trim two or three minutes of each financing application, which equates to a saving of up to 150 hours every day. Because credit analysts will save time on every customer application, they can handle more deals, which means that the company can increase sales without adding credit analysts or paying overtime, an additional savings of up to an estimated $3 million every year.

• Greater transparency, reduced risk

Now that its risk managers make more uniform decisions about

deal structures and monitor activities more effectively, the company faces less risk of audits and penalties. Risk managers also have the flexibility to alter policies based on market changes, current promotions, and the competitive landscape.

Those changes went into effect immediately, so the company can maintain a more consistent experience throughout its dealerships.

ADS provides self-service capabilities to both the dealer and customer, who are both looking for more insight and input over dealstructuring decisions.

ADS has widespread and immediate benefits. Crucially, it also improves overall time-to-decision and reduces the need for manual decisioning. This all boils down to an increase in approval

and booking rates, helping motor dealers respond to the challenging market conditions. We believe this is a ground-breaking application for optimisation that has potential applications in other lending areas, including mortgage lending, personal loans and credit cards – anywhere where competition is fierce and a fast, profitable counteroffer can win the business.

Richard Cowley is a Principal Consultant for Analytics at FICO. He has worked within financial services, both on the banking side and as an analyst and consultant for more than 25 years. During his time at FICO he has led many analytics projects across the customer lifecycle, across multiple industries, and has developed and implemented successful optimisation solutions across EMEA.

RAPID INSIGHTS FOR HUMANITARIAN AID IN THE UK

MATTHEW GWYNFRYN THOMAS

HUMANITARIANS HAVE A

LOT TO DO THESE DAYS. The last five years have seen a torrent of emergencies, while the future looks ever less predictable.

The Red Cross—or the International Red Cross and Red Crescent Movement, to give the global humanitarian network its full name—is best known for helping people during

emergencies. The British Red Cross, one of the world’s many National Societies, responds to emergencies across the UK – including floods, heatwaves, and domestic fires.

But that’s not all we do. The British Red Cross also helps people receive the care and support they need without falling through gaps in the health and care system, and we support people to

© David Severn/British Red Cross

feel safe, live with dignity, and have choices and opportunities if they are experiencing displacement. These are complex areas of work, requiring many kinds of action backed by evidence. My team of researchers and analysts takes a mixed-methods approach to producing the evidence required, including literature reviews, semistructured interviews, thematic analysis, PESTLE analysis, scenario planning, data visualisation, and statistical analysis. A lot of the time we need to produce this evidence at short notice, and we often work closely with operational, policy, and strategy colleagues to do so. This article explores three recent (and mostly ongoing) emergencies that required rapid insights.

We designed a set of food aid services, including a partnership with FareShare, that aimed to support people at the sharp end of food insecurity, which was targeted based on the Vulnerability Index

THE COVID-19 PANDEMIC

We aim to support people living in the most vulnerable situations, where no other means of support is available. But when facing a rapidly spreading new disease and support is limited because your nation is locked down, how do you prioritise where and who to help?

A few days before the Government announced a national lockdown for the whole UK, we began to model vulnerability to Covid-19 based on what was known at the time. People with specific health conditions or suppressed immune systems were classed as ‘clinically extremely vulnerable’ to the disease – so we gathered as many datasets as we could find that had information about these conditions.

It was also clear early on that there would be a socioeconomic angle to the crisis, with many people unable to work, financially insecure, unable to access food, digitally excluded, living in higher-risk environments, or working in occupations that increased their risk of catching the disease. Again, we pulled together a range of indicators that spoke to these vulnerabilities.

As mutual aid groups popped up across the country, a cavalcade of friendly nerds also sprang to action. Several people helped us by providing data or offering their analytical expertise, including Obi Thompson Sargoni at University College London, Brian Johnston at Queens University Belfast, and Tom Russell at the University of Oxford. OCSI, the Oxford Consultants for Social Inclusion, were particularly

FIGURE 1 BRITISH RED CROSS COVID-19 VULNERABILITY INDEX

© British Red Cross

generous with their time and data and deserve special thanks for supporting us throughout the pandemic (and beyond).

With all the data in place, we produced a neighbourhood-level composite index for the whole UK that modelled clinical vulnerability, socioeconomic vulnerability, and indicators of broader health and wellbeing. Figure 1 displays the index, showing neighbourhoods that were clinically and socioeconomically vulnerable during the pandemic. The areas lightly shaded were the least vulnerable and those with the darker shading were the most vulnerable. From the beginning, we made our Covid-19 Vulnerability Index (VI) open source and freely available so others could use it –which many did: from Local Authorities and Community Foundations, to the Scottish Public Health Observatory and even Arts Council England.

The British Red Cross used the VI for multiple avenues of support. We set up a ‘hardship fund’, disbursing £5m as shortterm financial support to 13,000 people who were facing (or already experiencing) destitution – and the socioeconomic domain of the VI helped in targeting our cash assistance. We also designed a set of

food aid services, including a partnership with FareShare, that aimed to support people at the sharp end of food insecurity, which was targeted based on the VI. That these services operated alongside our ‘business as usual’ support to people experiencing a multitude of crises is testament to the amazing humanitarian spirit of our volunteers and staff.

ARWEN, BARRA, CORRIE, DUDLEY, EUNICE, FRANKLIN

As if a pandemic wasn’t enough, the UK also got battered by multiple storms in quick succession. Beginning with Storm Arwen towards the end of November 2021, through to Franklin in February 2022, hundreds of thousands of people across our isles experienced power cuts, damage to their homes, injury and, in a few tragic cases, death.

The day before Barra, the second storm of the season, was due to hit, we rushed to provide insights to guide emergency planning and response in areas with a red or amber warning from the Met Office. This resulted in a set of profiles for these areas that included information about atrisk groups, such as older people, people living alone, and people in financially precarious situations.

Although this was an especially rapid analysis, produced within a single day, some local infrastructure organisations (who support other local charity organisations) said they were incredibly useful and shared them widely across their network

These profiles built upon several of the same data sources used in the Covid-19 Vulnerability Index, and we repurposed code from that and other projects—all shared on our organisational GitHub account—to make the profiles in such a short time. (The code for this piece of work is not currently publicly available, although we would be happy to share it with anyone who might be interested.)

We shared the vulnerability profiles with local British Red Cross teams as well as with partner organisations at the Voluntary and Community Sector Emergencies Partnership (VCSEP). The VCSEP is a hub for local and national charities, helping them plan for and respond to emergencies in a coordinated way. Although this was an especially rapid analysis, produced within a single day, some local infrastructure organisations (who support other local charity organisations) said they were incredibly useful and shared them widely across their network.On the other hand, this rapidly conducted experiment revealed some interesting challenges for our ambition to become more anticipatory and proactive in how we plan for emergencies. One of our goals as a team is to support the British Red Cross to become more future-oriented, especially in anticipatory planning for emergencies. However, there can be gaps between strategic intent and

©

Sam Atkins/British Red Cross

operational reality – and these gaps will not necessarily be reduced by throwing more evidence and insight into the mix: they require deeper consideration around culture and even differing worldviews about what it means to respond to an emergency.

UKRAINE

Towards the end of February 2022, the ongoing crisis in Ukraine spilled over, leading to casualties, displacement, and a worsening of the global cost of living crisis. Alongside the International Red Cross and Red Crescent Movement’s support in Ukraine and across the region, my team attempted to shed some light on the potential humanitarian impacts in the UK.

Given the massive uncertainty surrounding the situation, we needed to produce rapid analyses of multiple plausible scenarios to inform how the British Red Cross might support people in the UK, especially those fleeing Ukraine and their loved ones who were already living here.

This was far more qualitative than the previous two examples. Our scenario analysis explored a range of potential humanitarian consequences for the UK and took a deeper look at the potential vulnerabilities and unmet needs for people seeking protection here.

We categorised the multitude of potential impacts by whether they were ‘critical uncertainties’ (highly uncertain, high-impact phenomena), ‘predetermined potential impacts’ (medium-to-high-impact phenomena that are better understood), and ‘ripple effects’ that could have large impacts further into the future.

The predetermined potential impacts included further cost of living increases, the health needs of refugees in an already-pressured health system,

people entering the UK through unsafe routes such as via small boats crossing the Channel, and other factors associated with the Nationality and Borders Act.

Critical uncertainties encompassed a range of worst-case scenarios such as: potential destitution for people fleeing Ukraine after their support in the UK ends; increased trafficking to or in the UK; cyberattacks targeting critical infrastructure such as hospitals, banks, the National Grid, or transport systems; and even nuclear attacks or radiation borne on winds.

Finally, the ripple effects looked at the potential for future refugee crises following global food shortages that may lead to conflicts in other nations; the implications for climate change under short-term decisions around energy supply; the risk of a recession; and increased social and material inequalities.

This depressing smorgasbord of possibilities shaped how we have planned different aspects of our aid in the UK, which so far has focussed on providing emotional support, information and signposting, cash assistance, SIM cards, hygiene products, and mobility equipment.

Critical uncertainties encompassed a range of worst-case scenarios such as: potential destitution for people fleeing Ukraine after their support in the UK ends; increased trafficking to or in the UK; cyberattacks targeting critical infrastructure

We also shared these scenarios with other organisations through a VCSEP network call. They especially piqued the interest of Business in the

Community—a charity that helps businesses improve their impacts on communities and the environment— which used the scenarios in regional meetings and ‘Business Response Forums’ with representatives from various FTSE 350 companies.

A CASE FOR SLOWER INSIGHTS

These three examples showcase how we have increased our agility as humanitarian responders, supported by the rapid production of insights. But slowing down can be beneficial too. By building what you might call ‘insight infrastructure’, we can offer tools that support teams with their decision-making and understanding of situations related to emergencies, health inequalities, and displacement.

We are working on two main tools in this regard, both of which will be publicly accessible web apps. The first is an Emergency Planning Tool (which we are developing in collaboration with the Emergencies Partnership) that will help charities to better prepare by highlighting risks and vulnerabilities to specific emergencies in their areas, signposting users to other organisations who work in a place, and offering guidance on how to plan for emergencies. Our other tool will be a Health Inequalities Explorer, showing who and where are likely to experience inequalities in health outcomes, in the risks of poor health, and in access to high-quality support.

We cannot stop hazardous events from occurring, but we can take steps to become better-prepared so that they do not necessarily become disasters.

Matthew Gwynfryn Thomas is Head of Strategic Insight & Foresight at British Red Cross

OPTIT - FROM ANALYTICS AND OPTIMISATION TO OPERATIONAL EXCELLENCE

MICHELA MATTEI

OPTIT HAS BEEN DEVELOPING IT SOLUTIONS to support business innovation based on Data Science, Analytics and Optimisation for over 15 years, following its mission to unlock the potential of Operational Research (O.R.) and Advanced Analytics in practical contexts to achieve operational excellence. We leverage on proprietary analytics platforms to support our

customers and partners in their Digital Innovation roadmap.

The company operates through two offices in Bologna and Cesena, in Emilia-Romagna, the beating heart of the Italian industrial avant-garde. With consolidated agrifood, packaging, automotive and biotech industries, and the recent opening of the Big Data Technopole where the ECMWF

© Photo by Sam LaRussa on Unsplash

(European Centre for Medium-Range Weather Forecasts) and the Leonardo supercomputer are going to be located, this area is now considered the centre of the so-called Italian Data Valley.

We leverage on proprietary analytics platforms to support our customers and partners in their Digital Innovation roadmap

This thriving ecosystem is the result of the deep influence of the University of Bologna, the oldest in the world. Indeed Optit itself was founded in 2007 as a spin-off of the O.R. group within the DEIS department, under the initiative of Prof. Daniele Vigo. Daniele was joined in 2010 by Matteo Pozzi and Claudio Caremi, both experienced management consultants, and Fabio Lombardi, a talented software engineer. Since then, Optit has seen constant and rapid growth in its size and customer base, and now employs about fifty people, almost all with MS degrees or PhD in engineering and science disciplines, consolidating a leading role at national and international level.

The main characteristic of Optit, which is also one of the reasons for

its growth and success, is the flexible business model, which allows spanning from model-assisted consultancy to the development of complex enterprise applications to better suit the specific needs of the customer. Such flexibility is achieved through the focus on three pillars: Science, ICT and Business. Data Scientists and O.R. Specialists interact with a full-stack Software Factory, while interaction with the end user is facilitated and supported by a team of consultants that keep the focus on business impact. This interdisciplinary approach (which is coherent with the origins and development of O.R.) enables the seamless translation of objectives and constraints into mathematical and algorithmic terms, and eventually ensures that the decision support system’s implementation effectively delivers the intended results to the customers.

Co-founder of the EURO Forum on Practice of Operations Research, Optit keeps a close connection with Universities and Research Centres, participating in various research projects and financing PhD programmes, as well as attracting the enthusiasm and energy of young

researchers across the EU to advance at the tipping point of digital innovation through Advanced Analytics.

PROCESS

OPTIMISATION AS A KEY TO A MORE SUSTAINABLE AND EFFICIENT FUTURE

“We have a very privileged point of view: we share the strategic and economic driven attitude of companies, know the theoretical possibilities expressed by O.R. and have developed the software skills to transform consultancy into actionable processes. The possibilities for companies and public bodies to improve profitability, resilience, environmental impact and transform data into value is really large” says Matteo Pozzi, CEO of Optit. “Not all managers and decisionmakers are always aware of this potential, yet there is a growing consciousness of the value embedded in data and all we do is to unlock this by designing and implementing flexible solutions that support better decisions. The good news is that the return on investment is tangible, and we have defined a methodology to approach this process gradually, which helps developing the culture to resort to digital innovation with a very low risk profile.”

Optit has participated in PlaMES, which provides a Decision Support System to assist energy decision makers in the long-term planning of multi-energy systems at regional or national level designed to achieve decarbonisation targets at the lowest overall cost

Energy optimisation

“For example, in the energy sector” continues Pozzi, “the international and environmental situation requires us to

radically change how we plan and build energy systems. O.R. and advanced analytics can be a very powerful tool in this compelling transformation. There is a wide consensus among research and specialists, but this needs to be implemented both at the industrial, regulatory and policy level. We were lucky to initiate a collaboration, around 12 years ago, with the leading Italian utilities in the sector of urban district heating, eager to improve both the design of the network and the generation side, rapidly evolving towards more complex energy mixes with strong interconnection between heat and electricity coupling. Today, I devote a share of my time as Vice-Chair of DHC+, the Innovation and Knowledge hub of Euro Heat & Power, the Brusselbased organisation that supports the District Energy Industry unleash its potential to support the decarbonisation challenge, and we are working with the company that supplies energy to New York City re-think how digital innovation can support their transition into the new millennium. I hope not to be immodest if I say I feel very proud!”

One of the most recent EU-funded projects Optit has participated in is

PlaMES, which provides a Decision Support System to assist energy decision makers in the long-term planning of multi-energy systems at regional or national level designed to achieve decarbonisation targets at the lowest overall cost. The very high complexity of the underlying models, designed by the RWTH University of Aachen, Fraunhofer Institut and the University of Bologna, is combined with an easy-to-use web platform that allows a smooth navigation of data and scenarios. “We hope that high usability and good user experience will be the key for adoption by public decision makers and transmission/distribution

System Operators Planners, as we witness the growing need for audacious and forward-looking energy programmes that can lead us into a CO2 Neutral future!” remarks Pozzi.

Waste logistics optimisation

The waste sector is another area that has a huge environmental impact in our time. O.R. and process optimisation can improve efficiency for operators while generating great benefits to the environment. (See a previous article in Impact: https://bit.

ly/OptitWaste). “The seminal project at the origins of Optit was a tool to design efficient waste collection services. Today, we help the largest Italian utility leverage on its 100+ treatment plants managing over 5 million tons of waste every year. Our solution supports all planning processes, from the 4-year industrial plan to the weekly collection programme, providing a platform where each single transport is seamlessly allocated, implemented and reported involving all actors in the value chain, from the customer to the driver of the truck” remarks Fabio Lombardi “and we are now investing to leverage on our logistics platform to increase the capacity to implement innovative collection services that use fresh data coming from the field.”

Logistics and S upply C hain What Lombardi is referring to is the platform at the heart of the fastest growing market in Optit: logistics and supply chain. Rooted in the deep expertise of the founding partner Daniele Vigo, one of the leading experts worldwide in Vehicle Routing, a digital platform was designed to support planning and delivery of logistics services, ranging from routing, 3-d bin packing, zoning and flow optimisation.

“Our platform is used every day by a leading Food Processing Company to manage several hundred trips per day, fully integrated with the enterprise resource planning system, enabling the digital evolution of the existing transportation

© Andrey Metelev/Unsplash

© Photo by Pawel Czerwinski on Unsplash

planning process” says Claudio Caremi, “and the nice thing is that other customers use the same infrastructure with models that ensure the achievement of their specific and unique goals. Being full owners of every algorithm in a digital environment designed for superior flexibility and scalability is surely paying its dividends.”

A world leader in home appliance production, with factories all across Europe, reduced its transportation costs by increasing the load factor significantly. “Our solution creates optimised “same-day” solutions in batch form, providing insights into the management of successive days (such as possible mismatches between the volume of goods to be transported and the capacity of the vehicles scheduled), considering the possible unavailability of merchandise, changes in order priority, and every last-minute change. The integration of bin packing models (already developed in vertical projects in road as well as air cargo projects) in the planning process is adding yet another edge to the potential of our multifaceted offering” concludes Caremi.

A world leader in home appliance production, with factories all across Europe, reduced its transportation costs by increasing the load factor significantly

AN INCREASINGLY INTERNATIONAL AND MULTI-FACETED PERSPECTIVE

Science and mathematics know no boundaries, as does our research and attention for the latest development in

our fields. Just like academic research benefits and grows from exchange and communication with different groups, Optit participates actively in its reference academic and business communities, playing its role as members of several international associations, together with selected strategic partnerships with leading industry players and an increasing role into EU-funded projects, like UpgradeDH, Retrofeed, the already mentioned PlaMES, and, most recently, TUPLES.

TUPLES (TrUstworthy Planning and scheduling with Learning and ExplanationS) aims at building trusted planning and scheduling systems that are safe, robust, explainable and efficient. Optit is part of a Consortium financed by the Horizon Europe framework programme that involves four Universities (Communauté d'université et d'établissements université fédérale de Toulouse Midi-Pyrénées (ANITI), Katholieke universiteit Leuven (KU LEUVEN), Universitat des Saarlanders (USAAR), Alma mater studiorum - Università di Bologna (UNIBO) and three industrial partners (Airbus, Optit and Scisports).

The objective of the TUPLES project is to develop hybrid planning and scheduling methods that combine the efficiency, flexibility and adaptability of data-driven learning approaches with the robustness, reliability, and understandability of model-based reasoning methods, designing the verification and explanation methods to provide the user with a high degree of visibility of the properties of the solutions produced by the systems, testing the approach across case studies ranging from airplane pilot assistance and manufacturing, to soccer team recruitment, energy production and waste collection management.

The project, which kicks off in October 2022 and will produce a Digital Lab and a self-assessment tool on trustworthiness of algorithms, will represent yet another landmark towards the integration of data driven and model-based approaches, fostering the integration of O.R. into the wider Artificial Intelligence domain.

PEOPLE REMAINS THE KEY SUCCESS FACTOR

“Although born from the union of mathematics and software development, so called “hard sciences”, the key for success remains the human factor” Pozzi remarks. “In the end my main duty as entrepreneur is to create a good work environment, collaboration between the teams, and a good work-life balance. Employing passionate and talented people, the results emerge with little effort. It is thanks to this that we went through the pandemic with very little impact, and in fact we learned how to be even more flexible in our daily practice. The achievement of the last years set the basis for an even larger growth at international level and I am sure the best is yet to come!”

Michela Mattei has been Communication Manager of Optit since 2021. She has twenty years of experience as a communi cation and marketing specialist. Michela witnessed the web portals gold rush and has known a world without social media. She remains very curious about tech novel ties but is convinced that communication always works more or less in the same way, a challenging way. Daughter of an early Cobol engineer and granddaughter of an entrepreneur she has always been passion ate about science and industry, and how the two fumble together.

UNIVERSITIES MAKING AN IMPACT

EACH YEAR STUDENTS on MSc programmes in analytical subjects at several UK universities spend their last few months undertaking a project, often for an organisation. These projects can make a significant impact. This issue features reports of projects recently carried out at one of our universities: University of Edinburgh. If you are interested in availing yourself of such an opportunity, please contact the Operational Research Society at email@ theorsociety.com

In light of Scotland’s net zero targets until 2045 and the omnipresent effects of human induced climate change through the emission of harmful greenhouse gases, renewable energy technologies and efficient approaches to harnessing the energy produced for consumers are moving to the centre of public attention.

Nikolas’s project explored how to cost-effectively unlock the abundance of Scottish renewable wind energy resources available by coupling it to green hydrogen production for the purposes of industrial decarbonisation – in this case focussed on the highly energy intensive whisky industry clustered on the Hebridean island of Islay. The work was commissioned by EMRC, an independent consultancy providing energy market, regulatory, techno-economic and financing advice for electricity, renewables and gas sector clients globally, with the overarching objective of helping to deliver reliable access to affordable and clean energy.

His work covered many technoeconomic questions including:

• How to optimally design and size the electrolyser and the hydrogen delivery system, given the intermittency of offshore wind, the constant industrial heat requirement from the distilleries and the transportation constraints.

• How best to deploy storage mechanisms – through input battery storage systems or through hydrogen storage at the output of the electrolyser.

• How financially viable the project is in terms of the surcharge required on retail price of whisky or the financial support required for hydrogen production in order to make to make it competitive with the financial costs of current fossil fuel systems.

Based on a thorough discussion of various engineering aspects, different cutting edge technologies, as well as the specific case in the Island of Islay, a selection of apt system components was made and their characteristics/ behaviours subsequently modelled mathematically. To meaningfully

address significant uncertainties involved, such as the highly volatile electricity generation from wind, associated electricity prices or demand uptake, a complex two stage stochastic optimisation problem was set up.

The results of Nikolas’s optimisation, and comparisons made with alternative carbon efficient technologies show that hydrogen has the potential to play an important role in deep decarbonisation efforts of the distillery sector and support the net zero transition on the island. This is thanks to its versatility as an energy carrier allowing for efficient use of wind generation, the little effort necessary for technology conversions and its economic feasibility, as demonstrated by a thorough economic breakeven analysis. Furthermore, valuable insights surrounding the necessary plant components, their technologies, sizing and operation could be gained.

By virtue of the exclusive electricity supply from renewable sources, the system is carbon neutral from day one and could, even for the most

GREENING ISLAY’S WHISKY PRODUCTION (Nikolas Heinloth, University of Edinburgh, MSc Operational Research with Data Science)

conservative demand scenario, help to save 35,598t of carbon emissions making a valuable contribution towards a greener and more sustainable future.

If the positive effects were to materialise as expected, the Islay hydrogen project can serve as an important case study and model for further applications. Lessons learned from its implementation would undoubtedly lead to technology improvements and system enhancements, thus increasing

the rollout potential across other municipalities.

Douglas Caskie, Project Supervisor and Chief Executive Officer, EMRC: “The Islay Green Hydrogen Project, which we designed for Nikolas, covers some of the ‘hot topics’ in energy provision – how to cost-effectively deploy ‘green hydrogen’ for the purposes of industrial decarbonisation –in this case focussed on the whisky industry clustered on the Hebridean island of Islay. Nikolas adeptly covered many techno-economic

questions and provided the energy stakeholders on Islay an extremely useful modelling framework with which to base further analysis. Based on this work, there is a very strong case to be made for the decarbonisation of the island.”

This was a highly technical project that required a diverse set of skills across several disciplines such as physics, engineering and operational research. The work demonstrated an excellent practical application of O.R. in such a challenging domain.

IMPROVING THE ENVIRONMENTAL IMPACT OF CONSTRUCTION SITES

NICOLAS BLANDAMOUR LOCALSOLVER IS A FASTGROWING SOFTWARE COMPANY in optimisation and decision support. The team is developing LocalSolver Optimizer, an innovative optimisation solver. In addition to this flagship software, LocalSolver offers tailor-made optimisation solutions with ergonomic interfaces for various businesses. The company counts among its customers the most prominent companies worldwide, like Starbucks, Macy’s, Microsoft, Chewy, Roadie, Procter &

Gamble, Air Liquide, Bosch, Publicis, JCDecaux, Airbus, Renault, Repsol, Fujitsu, Softbank, and Sony.

LOCALSOLVER’S INNOVATIVE APPROACH

LocalSolver’s team develops a generalpurpose mathematical optimisation solver for operational researchers, data scientists, and software developers to quickly solve optimisation problems and, beyond, build and deploy optimisation applications effortlessly. One of LocalSolver

© Hesus

Optimizer’s strengths is its innovative modelling approach that makes the mathematical modelling of Vehicle Routing, Production Scheduling, Workforce Scheduling, and many more types of problems much simpler than traditional solvers. The resulting models are compact and natural while providing much better results, particularly for large instances in limited running times. LocalSolver’s team also offers reactive and dedicated support to the users of the optimiser. The team works closely with its users and helps them model and efficiently solve their optimisation problems until they are fully satisfied.

HESUS

Recently, the French start-up, Hesus, was able to leverage the power of LocalSolver Optimizer. As a European leader in sustainable solutions for the management of construction site soil, the group is positioned in all strategic, operational, and logistical areas of expertise with proven expertise in polluted sites and soils. They assist all players in the construction and

public works sector during the design, construction, and demolition of infrastructures and buildings. Present in six countries, they manage nearly 2 million tons of soil each year in France and Europe and recover an average of 82% of the spoil on its sites.

They manage nearly 2 million tons of soil each year in France and Europe and recover an average of 82% of the spoil on its sites

A CHALLENGING TRANSPORTATION PROBLEM

Dozens of construction sites delegate their transportation logistics for excavated earth and construction materials daily to Hesus. Transport infrastructure or building sites require the movement of large quantities of excavated soils and building materials. The ground extracted from these construction sites can present different pollution levels, requiring specialised centre treatments. Once the treatment has been completed, the inert soil obtained can be reused as backfill for

earthworks, while the extracted materials, such as gravel, can be used as raw material for other construction sites. The different trips needed to move these materials and the operational constraints lead to a difficult Vehicle Routing Problem (VRP). In addition to the traditional constraints of a VRP, it is necessary to take into account the road traffic, the opening hours of construction sites and treatment centres, and specific regulatory constraints such as the maximum driving time and break times to have the most realistic routes possible.

Mathematically speaking, this problem is related to a VRP with full truckloads. The main objective is to reduce as much as possible empty trips for dump trucks, or in other words, unproductive minutes and kilometres for the whole fleet of trucks.

More than 300 tours are requested per day for evacuations/deliveries. This represents 10,000 tons of material to be moved between construction sites and treatment centres.

HESUS DISPATCH SOLUTION

The resulting optimisation solution based on LocalSolver as the core optimisation engine, called Hesus Dispatch, optimally matches possible drop-off locations with potential loading sites along the truck routes while respecting all their operational constraints. Thus, the application allows Hesus to reduce its transportation costs and carbon footprint of truck journeys by 10% to 15%.

The application allows Hesus to reduce its transportation costs and carbon footprint of truck journeys by 10% to 15%

The optimisation engine can identify how to efficiently mutualise a truck across several sites and treatment centres to reduce the length of empty transition

rides between valuable trips. For instance, it can lead a truck to evacuate excavated soils from a site to a treatment centre further than the nearest ones but close to another site with transportation needs. The extra distance travelled to reach this treatment centre will be more than compensated by the short deadhead ride to move to another site where the truck will be reloaded to continue its journey. Identification of efficient repositioning among the numerous possible combinations resulting from the size of the problem is one of the levers used by the optimisation to improve operational efficiency.

Another improvement comes from a better filling of trucks timetable, which reduces the number of trucks needed to perform the transportation. The engine will exploit differences in sites and centres’ opening hours to increase the used time range of the truck with respect to regulatory constraints. The effective time range of a truck can also be increased by tweaking the end of the truck roadmap to avoid returning too early to the truck depot. For example, at the end of a day, a truck may evacuate soils from a site to a nearer centre than the one used during the rest of the day. This nearby centre can be less attractive in terms of treatment costs than the others but allow an extra tour to be planned at the end of the day,

a tour which is not possible for other treatment centres due to daily travel time limitation.

Finally, the optimisation engine chooses the most suitable contractors for truck providers and treatment centres regarding geographic implantation and prices. The tradeoff is indeed challenging to find for a planner at hand. Choosing the truck depots and processing centres closest to the sites reduces the distance travelled, but it increases the overall cost as those contractors close to sites are located in urban areas and thus generally expensive. On the contrary, choosing the cheapest contractors will increase the distance as they are located far from the sites, resulting in a higher number of trucks and an important transportation cost despite a reasonable unit price.

It is really exciting to discover each day how the routes can be optimised and to contribute to bringing sustainable solutions for construction site soils

To ease the change management and the adoption of the decision-support tool by the logistic planners, Hesus Dispatch allows fine control of the complexity of the routes generated by the engine. Kevin

Jahier, Logistics Director at Hesus: “The optimisation engine helps me every day to dispatch our customer orders. On top of saving me a lot of time that I can dedicate to other high-value-added tasks, it is really exciting to discover each day how the routes can be optimised and to contribute to bringing sustainable solutions for construction site soils.”

AN ENVIRONMENTAL IMPACT

Because the excavated soil from one site can become a resource for another, Hesus, Greentech’s partner for construction sites, puts the circular economy, seen in Figure 1, at the heart of its solutions. They favour re-using soil and materials while ensuring soil compatibility and the absence of contamination risks. Hesus prioritise the re-use of soils, which allows controlling costs while acting to preserve resources and reduce CO2 emissions. Hesus partners with more than 900 sectors, including treatment and decontamination centres, quarries, and storage facilities. Thus, each site benefits from a service adapted to the type of pollution of its soils and waste. In 2020, Hesus achieved an average recovery rate on its sites of over 84%.

O.R. has proven to play a significant role in solving Hesus’ challenging transportation problem both from an operational and environmental point of view. That’s why the LocalSolver team is glad to help Hesus’ team achieve even more ambitious goals in the years to come.

Nicolas Blandamour is a Vehicle Routing Optimisation Expert at LocalSolver. He joined the company right after his gradua tion and has worked since as a member of the R&D team to improve the performance of LocalSolver Optimizer. Nicolas also developed several tailor-made vehicle routing applications in various business fields, among which Hesus Dispatch was selected as a finalist for the Indus’RO price given by the French O.R. society rewarding best O.R. applications deployed in the industry.

MEASURE FOR MEASURE

Geoff Royston

Look around your house. Is there a clock or other timepiece? A ruler or tape measure? What about some weighing scales? Three out of three? You may not be a metrologist, but you are clearly familiar with their subject matter and well equipped with some of their basic tools.

My piece for this issue of Impact has been prompted and informed by the recently published book Beyond Measure by James Vincent and also by The Tyranny of Metrics by Jerry Muller.

Beyond Measure takes the reader through a lively history of measurement from the ancient Egyptians monitoring the height of the floodwaters of the Nile, through mediaeval standards for weights and measures, on to the birth of the metric system during the French Revolution - and then to the present day where it seems that almost everything is being measured by everyone – or, more insidiously, surreptitiously, by someone.

THE SPREAD OF STANDARDISED MEASUREMENT

The ability – and need – to measure time, distance, and mass underpins physical sciences and, arguably, civilisation. For instance, the societal importance of regulating such measurement was recognised in the Magna Carta of 1215, clause 35 of which reads: ‘There is to be one measure of wine throughout our kingdom, and one measure of ale, and one measure of corn, namely the quarter of London, and one breadth of dyed, russet and haberget cloths, that is, two ells* within the borders; and let weights be dealt with as with measures’. (* an ell was the length of King Henry I’s forearm!). The societal importance of weights and measures regulations was further demonstrated by the intertwining of the French and the Metric revolutions, and continues to this day: Beyond Measure notes for example how arguments over use of metric rather than imperial measures - featuring slogans like ‘Rule Britannia - in Inches not Metres’ fostered the growth of UKIP and the Brexit movement.

Historically, searching for a reliable standard has involved striving for ever greater accuracy and precision. So standards have developed from the King’s forearm’s ell of the Magna Carta, to the platinum cylinder kilogram of the French metric revolution, to the modern standard for the second based on the frequency of electromagnetic waves from caesium-133 in an atomic clock.

Reference standards now exist for an astonishing range of things, far beyond fundamental physical units: amongst those mentioned in Beyond Measure are peanut butter, powdered radioactive human lung - and a standardised process for making tea. Similarly, the range of what is routinely measured has expanded hugely. Measurement of inputs, outputs and performance has become ubiquitous in the workplace. Lately, with credit ratings, body mass indices, daily step counts and so on, we have seen the emergence of what has been termed the ‘quantified self’.

MEASUREMENT: GOOD OR BAD?

Positions on the merits of this spread of measurement and quantification vary. The eminent Victorian physicist William Thomson, Lord Kelvin, was in no doubt: ‘When you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind’.

A more nuanced view is given by the management writer Charles Handy In his book The Empty Raincoat: ‘The first step is to measure whatever can easily be measured. This is OK as far as it goes. The second step is to disregard that which can't be easily measured or to give it an arbitrary quantitative value. This is artificial and misleading. The third step is to presume that what can't be measured easily really isn't important. This is blindness. The fourth step is to say that what can't be easily measured really doesn't exist. This is suicide’.

Such criticisms relate to a modern unease that measurement may have gone too far, with a view that excessive quantification discounts the more ‘human’, qualitative, elements of life. In fact, such unease is far from new, as shown in the words of the Greek philosopher Plutarch: ‘The measure of a man's life is the well spending of it, and not the length’, or of Oscar Wilde ‘s criticism about a person who ‘knows the price of everything and the value of nothing’. In Beyond Measure James Vincent comments that measurement can conflict with ‘something deeper and unquantifiable in ourselves’, echoing Keats’ ‘unweaving the rainbow’ – seeing the rationalisation of nature as removing its magic and meaning. (For him, clearly, but for others, understanding the workings of the natural

world enhances rather than diminishes their appreciation; see for example Richard Dawkins’ book Unweaving the Rainbow: Science, Delusion and the Appetite for Wonder.)

This is beginning to take us beyond the scope of this article so, without ignoring criticisms of measurement, let’s focus mainly on measurement in management.

MEASURING THE THING RIGHT AND MEASURING THE RIGHT THING

For measurement in ‘softer’ areas – including management – issues beyond reliability, accuracy and precision become more challenging – not least validity. Does the measurement correspond to what it is meant to be measuring? And how far is it measuring the right thing anyway? So, for example, how well (if at all) does a particular measure for organisational performance correspond to the success of an enterprise? As one of the key figures in the post-war growth of Operational Research, Russ Ackoff, said: ‘Managers who don't know how to measure what they want settle for wanting what they can measure’.

Beyond Measure and The Tyranny of Metrics present numerous instances of what they describe as ‘excessive and inappropriate measurement’, ranging from over-extended applications of Frederick Taylor’s ‘time and motion’ studies in the early 1900s to the infamous example of the Vietnam War’s reliance on ‘body count’.

These books note how problems grow, with unintended consequences such as gaming the system, when measurement is used to penalise poor performance. They relate that when the NHS penalized hospitals with A&E wait times longer than four hours, some hospitals responded by delaying the handover of patients from arriving ambulances, thus keeping the wait within the hospital down - but seriously reducing the ability of the ambulance service to meet its response time targets. Or when surgeons ‘cream’ the easier cases, hesitating to take on the riskier procedures for which the success rate, and so their apparent performance, will inevitably be poor. Such behaviour exemplifies Goodhart’s Law: ‘Any measure used for control is unreliable’.

Such hazards need to be recognised, but not overemphasised. As Muller, in The Tyranny of Metrics, notes: ‘The problem is not measurement, but excessive measurement and inappropriate measurement—not metrics, but metric fixation’.

MEASURED JUDGEMENT

One way of avoiding ‘metric fixation’ is to favour ‘subjective judgment’ over ‘objective measurement’. But although there

will be some situations in which relying on a manager’s personal experience and instincts is appropriate, and others where this approach will be inferior to one grounded on data and analysis, decisions often benefit from deploying both experience-based judgement and data-based measurement. A good example can be found in The Signal and The Noise by Nate Silver in his revisiting of the story of Moneyball. That book – and resulting film – was about how analytics had transformed the process of recruiting baseball players from one relying on the observations and intuitions of traditional scouts to one where analysis of the huge database of baseball performance statistics was successfully used to identify promising players that scouts were missing. Nate Silver points out that at first this seemed to threaten the jobs of baseball scouts but in the years that followed teams that that had leant heavily on statistical analysis added in more scouting to the mix, while teams that had relied strongly on scouts added in more analytics. The best results were produced by a judicious blend of the two approaches.

Which leads me to end this piece with the (somewhat surprising given the book’s title) concluding words of The Tyranny of Metrics, ‘Ultimately, the issue is not one of metrics versus judgment, but metrics as informing judgment, which includes knowing how much weight to give to metrics, recognizing their characteristic distortions, and appreciating what can’t be measured’.

Dr Geoff Royston is a former president of the OR Society and a former chair of the UK Government Operational Research Service. He was head of strategic analysis and operational research in the Department of Health for England, where for almost two decades he was the professional lead for a large group of health analysts.

© Princeton University Press

© Faber