Summary:
Phasor Measurement Units (PMU) are monitoring devices currently dedicated to the transmission grid, where high voltage and large coverage makes it necessary to have some devices capable of wide-area oscillation detection, in order to prevent massive blackout. Over the years, the costs of PMUs have decreased and its functionalities have been improved with the advances in telecommunication technologies. New constraints resulting from the energy transition also suggest future requirements for better visibility in the distribution grid, which currently remains largely unmonitored.
This thesis explores the possibility of integrating PMUs in the distribution grid and evaluates its cost/benefit trade-off. We start by introducing the reader to the electricity industry, particularly the power grid sector and its shift towards a digitalized and “smart” grid. We then introduce PMUs and the various applications they can provide, with a focus on two topics: integration of Distributed Energy Resources (DER) and outage management. Using case studies and literature reviews, we prove that PMUs have a positive Net Present Value (NPV) for distribution grid fault location, while literatures suggest an important contribution of PMUs to future DER integration.
Given the regulated nature of the power grid industry, it is impossible to make a proper cost/benefit analysis without consulting regulatory frameworks. The last section of the thesis deals with various type of regulations in Europe. We show that policies have a large influence on the economic viability of PMU deployment
Results:
Considering the ability of PMUs to locate faults quickly and increase the hosting capacity for DERs, these monitoring devices have high potential for application in the distribution grid. The results indicate a positive NPV based on the Value of Lost Load (VoLL). VoLL is a societal cost of power interruption, and therefore proper regulations are required to share these benefits with Distribution System Operators (DSO). DERs have yet to reach a critical penetration which would justify the need to consider alternatives to grid reinforcement, but studies have shown that smart technologies such as PMUs are promising in reducing the cost of massively integrating solar and wind while maintaining grid reliability.
Summary:
In the context of a large scope and long-term duration project, asset management inputs and scheduling are essential to provide. Multiple constraints such as numerous work packages and limited duration to complete the project prevents to apply usual methodologies from the industry. The purpose of this thesis is to provide an analysis of the current tools and workflows to propose meaningful asset management procedures and propose a planning tool. The key objectives are the tracking of information about the equipment and activities performed on a virtual platform and to enhance quality management in the project while meeting the tight planning given by the coordination team.
Results:
Through the analysis conducted during this thesis, a meaningful data structure of connected assets has been built to enable to a smooth workflow of activities to complete on the assets. Relevant templates allow the user to store in an efficient and simple way the essential data to the operation of the project. The creation of an autonomous workflow from the asset creation guides the operators in the task to complete. To organize the overall activities in this large project, dynamic tools for scheduling with flexibility and planning have been designed considering the internal and external dependencies to optimize the duration of the project
Summary:
The main goal of this thesis was to answer to a business interrogation that might happen in every industry and company. We wanted to assess what are the factors that could impact a dashboard’s adoption, the value added it brings and its appreciation from the end-users. To assess this, two dashboards were created, one with tables visualizations and the other with graphical ones. Multiple A/B tests were done on the whole sales department of the company. In particular, we checked the impact of the specific tasks of a particular job, of the geographical region where the employee is working, of the level in data manipulation and of the experience in the company.
Results:
Overall, the introduction of the dashboards led to an increase in efficiency of 54% for the sales department of the company. Concerning the results of the different A/B tests, we discovered that:
– In terms of appreciation, tables seem to be preferred for the main tasks of an employee, whereas graphical visualizations shall be used for topics on which he/she is less familiar with
– In terms of usefulness, graphical visualizations increase the cleavage between known data and new data. Tables seem to encourage people in discovering some of the data they were previously not using
– In terms of usage, employees that are already using data on their daily routine take more value from a visualization tool than their colleagues and even more when being provided with tables.
Summary:
Due to the urgency of acting on sustainability to save the planet, many big companies have set-up sustainability targets and are actively working towards achieving them. Logitech is one of those companies, and since plastic is widely used across its products, it has recognized the need to explore sustainable rigid plastic materials (non ABS) for the use across its PC peripheral product range. Prior to this project, scouting and preliminary moulding trials of a range of different sustainable materials had been undertaken by experts at Logitech and four categories of materials were downselected for further investigation: recycled commodity plastics, natural fibre polymer composites, bio-based plastics, and a cellulose-based material with some recycled and bio-based plastic content. In this study, the consumers’ point of view and preferences were integrated in the evaluation of the materials in order to constitute a better understanding of the consumers’ sentiments with regards to the use of sustainable materials in PC peripherals. This will help in both the selection of suitable materials and in the marketing and communication related to future products made with these materials. The methods adopted in this study were a survey and a user experience (UX) testing. The sample size for the survey was 264 Logitech employees distributed across four office locations: Lausanne, Cork, Hsinchu, and Suzhou. As for the UX study, it was targeted on Generation Z and was performed with 8 users of which 6 are interns at Logitech based in Lausanne. These methods helped in gauging the consumers’ knowledge about plastic materials, assessing their attitudes towards each material based on their sustainability stories, determining the factors that influence their choice for the material, and characterizing the materials based on their look, tactility, and usability.
Results:
The results of the survey suggested that in general people do not know much about plastic materials. In addition, in order to model the influencing factors for choosing a sustainable material, a multinomial regression was built. The model suggested that the consumers’ choice of material can not be explained much by their socio-demographic characteristics. However, it was found that the green values of the consumer have a significant positive effect on their preference for bio-based plastic and natural fibre polymer composites. Moreover, based on the UX study, it was found that the performance and aesthetics of the material used in the mouse are more influencing for the choice of mouse material than the story of the material, unless the individual really cares about sustainability. Therefore, two strategies, depending on the individual’s level of concern towards sustainability were discussed and presented. The first strategy is to develop specific products with eco-design aesthetics to satisfy those who really care about sustainability. The second strategy is to use the most performing material for the rest of Logitech’s product portfolio for the company to meet its sustainability goals in the long term.
Summary:
This thesis concerns the supply and internal flows of the after-sale workshop of a luxury watch manufacture. The first goal is to adapt the supply flows of the workshop to a transformation in the distribution system of the manufacture due to the arrival of a new storage facility. This storage will cut direct access of the workshop to 30% of its supply. The second goal is to analyse the internal flows of the workshop in order to evaluate the opportunity of implementing an automatized storage. Researches were conducted by analyzing relevant datas in the manufacture’s ERP and by brainstorming with the workshop’s employees to understand their needs.
Results:
We redesigned the incoming flows of the workshop, considering the temporal and handling contraints. The study of the supply flows spotlighted a non-value-added activity that we removed and gained 5% of the human resources.
To understand if adding an automated storage in the workshop is valuable we studied the internal flows of the workshop. It allowed us to understand that the major area for improvement was the service level; and it could be increased by decreasing the lead times during the reparation flows. However, a new automatized storage would not bring any improvement of the workshop’s service level. The original idea was reavaluated and the project redefined.
Summary:
Along with governments and businesses, citizens as individuals represent a strong lever for action against global warming. As part of the Goodtrack project, this master thesis aimed at building relevant solutions to foster a citizen-driven sustainable transition at the Swiss level.
The first section both identifies major hinders to citizen engagement as well as highlights possible avenues to fight against them. The second one dives into the Goodtrack project, explaining its scope, objectives, and main functionalities. The third and last section discusses to what extent these features offer meaningful solutions to promote green transition from
individuals.
The Goodtrack project consists in creating a web platform which both collects and proposes practical daily Actions in the various themes of sustainable transition. It aims at bringing a quantitative qualification of those Actions’ impacts, bringing its members ways of estimating the positive effects they have.
Results:
Using both theoretical knowledge from environmental science and brainstorming around existing web functionalities, the work led to the design of core functionalities for Goodtrack: Calculators that will estimate personalized impacts of Actions, features to allow for the catalog of Actions to be improved by the application’s community, Transition Accounts showing each members their aggregated impacts and early reflections on customized content and recommendations.
Summary:
Faced with a 400% increase in mobile users between 2011 and 2019 and an exponential growth in data consumption, Swisscom is hard pressed to scale its network. To assist its cell tuning experts, a parameter suggestion tool is developed for 4G macro outdoor cells. Using physical and LTE parameters as well as past performance data, a performance prediction model made up of one ridge regression model per Performance Metric Indicator is trained and evaluated using cross-validation and the pseudo-Huber loss. It is then fed to a Bayesian optimization algorithm to suggest optimal parameters.
Results:
The results are conclusive. They demonstrate a clear improvement and consistent performance. Processing efficiency for expensive operations has also been increased by up to 80\%, cutting down processing time from days to hours. The project is hence ready to be tested with production data.
Summary:
This thesis presents an improvement in Inventory Management of an FMCG company through digitization of inventory data visualization and reorder point calculation. The thesis explores two main research questions: “how does digitization improve performance in Inventory Management?” and “how to digitize Inventory Management practices and which challenges are faced?”. Two practical problems are identified to serve as a case study to answer the previous questions: “how to ensure spare-parts inventory classification and visibility about it for all stakeholders?” and “how to reduce repetitive manual labor in Inventory Management through process automation?”. Two main digital tools are designed and implemented to obtain empirical results: (i) a descriptive analytics dashboard and (ii) a software robot. Tool (i) aims at providing granular visibility on the spare-parts inventory in the partner company. Tool (ii) aims at automating the Inventory Management process from data extraction to order input into the company systems.
Results:
The implementation of tool (i) had both direct and indirect results in the spare-parts Inventory Management of PMI. The two main direct results obtained are the following:
- Increased visibility over the spare-parts inventory in terms of items classification and location
- Increased capability to convert data into insights
These results drove an increase in the level of adoption of the tool among the spare-parts community. Decision-makers started leveraging the dashboard to manage the spare-parts inventory leading to the following indirect results:
- Adoption of a more data-driven decision-making approach
- Achievement of an 8.2% inventory value level decrease in 3 months
The implementation of tool (ii) has the following expected results:
Reduction of process duration from 8 to 1 working day
- Reduction of manual labor costs and consequent release of 10 people to focus on value-added activities
- Elimination of manual input errors leading to inventory ordering and carrying costs decrease
Summary:
Through the implementation of an experimental study testing the effects of dream journaling and bedtime dream reading on creativity, this thesis tested the adequacy of a dream journaling app for conducting research studies. Previous literature exposed a correlation between the frequency of dream recall and creativity, and it was found that daily dream logging positively affects dream recall and creativity. This study tested the hypothesis that in addition to dream journaling, reading previous dream reports at night would increase the beneficial effect on creativity. The study was conducted in Capture, a well-established dream journaling mobile application, and served as a pilot for evaluating the adequacy of the app as a tool for conducting research studies. The main actors of the dream market were identified, and several development strategies were discussed.
Results:
The experimental study found no significant effect of dream journaling and bedtime dream reading on dream recall and creativity. The most plausible explanation for these results comes from the fact that participants of the study were already familiar with dream journaling, which may indicate that the beneficial effects of this process on creativity and dream recall tend to manifest early and mostly for low dream recallers at baseline. Overall, it was shown that Capture is a good fit for conducting large-scale dream studies, and the opportunity to develop the product towards research should be pursued.
Summary:
With the goal of introducing new scientific techniques into the business, the thesis took place in the retail cosmetics environment of the L’Occitane group. Among the different departments of the group, this thesis took place within the retail IT team which ensures the stability of the checkout software and the launch of new projects affecting the checkout process in stores. Nevertheless, checkout processes are very complex with many interdependencies between systems and especially a great variety of implementations between stores in the park. These variations can occur from one country to another or within the same country. Thus, in order to support the work of the team, this thesis will attempt to identify the different causal structures of the checkout process for the French park stores, and then group the stores with similar causal characteristics. The goal of this study is to provide the retail IT team with a working support when investigating IT problems in the retail store. In addition, this tool could be used to estimate the impact on other systems when launching a new project. To address this problem, this thesis will use causal inference techniques and in particular causal discovery techniques with the use of the PC algorithm. To do so, the first step of the analysis will be to define in more detail the field in which the study takes place in order to understand the issues and the particularities behind this work. Then, a theoretical part will be presented in which the principles of causal inference and the PC algorithm will be detailed. Furthermore, the PC algorithm will be applied on a store to illustrate how this algorithm works and how it converges to the solution. Following this theoretical part, the use of the PC algorithm will be generalized to the whole park of stores in France. Nevertheless, with a large number of stores in France (nearly 100), it is difficult to manually compare the causality networks between all these stores. Thus, it was decided to perform a clustering of the causality networks of the stores. The goal of this approach is to identify groups of stores with similar causal structures. However, to perform this clustering it was necessary to identify an adequate distance measure that could be applied between two graphs. To do this, the analysis was based on the use of the Structural Hamming Distance (SHD). Following the clustering, 4 groups of stores with similar causal structures were identified within the French store network. Finally, following the clustering, the most representative graph of each cluster will be presented and will serve as a basis for the team’s future work in detecting anomalies.
Results:
Following the optimization of the hyper parameters of the model, it appears that there are 4 groups of stores with similar causal structures in France. Among these groups, one is composed of only one store, which is the Champs Elysées. This observation is interesting because this store is very special because of its collaboration with the pastry chef Pierre Hermé. In addition, it offers restaurant services, which makes its IT network structure very different from other stores. Then another group is the one composed of the Erborian and Mervita boutiques, this observation also makes sense because these boutiques are very different from the L’Occitane en Provence boutiques. Indeed, at the moment, because of their place in the group, the development of projects is more numerous and more important within the L’Occitane en Provence brand. Finally, there are two groups for which it is difficult to attribute a title even by consulting managers in the France area. These two groups present technical interpretations but not business ones.
Summary:
The thesis showcases options for improving the management of supply-chain risk and resilience within the Global Healthcare Operations to support the achievement of profitable growth objective while delivering high-quality, safe, and continuous supply of life-saving products to patients. The thesis presents alternatives to the predominantly classical (qualitative) and decentralized risk management approach with illustrative business applications. Specifically, the thesis addresses three main questions to improve operational, tactical, and strategic decisions. First, how can risk assessment be improved to provide visibility on single-risk exposures? Second, how can risk assessment and mitigation be improved by accounting for risk interdependencies? Third, how can supply chain resilience be measured and monitored? To address these questions, the thesis develops bespoke tools and frameworks based on emerging risk management approaches, namely loss distributions, risk networks, and a supply chain resilience index.
Results:
The quantitative risk analysis using Monte-Carlo simulation provides several improvements over the current (likelihood x impact) risk score. For example, the developed model makes explicit the uncertainty through failure rate and disruption-time distributions, and the associated costs given acceptable disruption thresholds. The value-at-risk and loss exceedance probability can help management decide on whether the residual risk is acceptable or not, and therefore on resource allocation for mitigation and/or the selection of optimal mitigation strategy. The risk network methodology provides a semi-automated way to collect expert knowledge of risk interactions and their respective strengths and to develop decision-centric visualization and analytics allowing better risk ranking and prioritization. The supply chain resilience framework offers a candidate structure for a holistic assessment of a product’s end-to-end supply chain resilience. The contributions include (i) translation of the multifaceted resilience concept into a 9 X 6 matrix structure with dimensions of resilience and SC process, including a set of plausible metrics and (ii) best-practice recommendation—based on Analytical Hierarchical Process—for composite score calculation.
Summary:
Le marché suisse de l’énergie est actuellement en pleine mutation. Un projet de libéralisation du marché de l’électricité est en effet en cours. La production d’énergie, aujourd’hui, centralisée se meut peu à peu vers une production décentralisée, particulièrement marquée pour l’énergie photovoltaïque. Les consommateurs actuels se transforment ainsi en prosommateurs (consommateurs équipés d’une installation de production d’électricité), plus actifs et moteurs de cette évolution.
Dans ce contexte apparaissent les plateformes peer-to-peer (P2P) « énergétiques » qui offrent la possibilité d’effectuer des transactions d’électricité directement entre les prosommateurs et les consommateurs. Ce phénomène émergeant modifie les modèles d’affaires traditionnels en établissant des marchés décentralisés basés sur des offres et des demandes locales.
Dans le cadre de ce travail est analysé la faisabilité économique d’une plateforme P2P dans le marché suisse de l’électricité. Cette étude aborde, en complément, les contraintes et facteurs clés de succès au développement d’un projet de ce type tout en considérant l’intérêt que peut représenter une telle plateforme pour des consommateurs et prosommateurs résidentiels.
Results:
Sur la base d’une modélisation de la plateforme, il a pu être montré qu’il n’y aurait probablement pas d’intérêt économique à développer une telle plateforme, au moins en tant que produit unique. Une analyse qualitative a cependant permis de montrer qu’un intérêt stratégique pourrait subsister notamment en termes de facteur différenciant et de complémentarité avec une gamme de produits plus large. Le développement d’une telle plateforme pourrait ainsi être justifié pour une entreprise d’approvisionnement en électricité de grande taille.
Par ailleurs, les participants à la plateforme, autant les consommateurs que les prosommateurs, pourraient eux y trouver un intérêt individuel, que celui-ci soit d’ordre économique ou écologique, ce qui pourrait générer une potentielle demande pour ce type de plateformes.
Summary:
The healthcare industry is currently facing increasing demands with limited resources in terms of funding, workforce, and workspace leading to increase of turnaround time, cost reduction, space limitation and to drop the level of quality. Medical laboratories have then evident interest in Lean management. The specificity of Medical Device Testing (MDT) laboratories lies on diversified daily manual experiments production instead of processing samples. The Lean implementation in the MDT laboratory aims to reduce report’s delivery time, and customer communication through formal studies follow-up. Focusing on four types of wastes (defects, motion, transportation, and excessive inventory), Lean principles help to introduce effective formalism of study management, implement relevant KPI, restore strong information flow, optimise consumables and technicians flows to, in fact, improve internal quality.
Comparison between MDT and medical laboratories on Lean adoption and adaptation strengthen the Lean solutions that are already implemented and bring opportunities to nonmassive-production companies in healthcare industry.
Results:
The standardised study management guarantees an accurate prediction of the report delivery date, same margin for quotes, a scheduling tool, an accurate progress rate for each project. Implemented KPI such as measuring the rate of “Right the First-Time” update delivery prediction. Internal communication has been improved using a daily Kanban meeting. Digitalisation of traceability restores qualitative information required for accreditations. The average turnaround time of the studies decreases with flows and layout optimisations. In first intention, unlike production laboratories, the global improvements are the only ones considered. Indeed, some Lean concepts are difficult to apply due to the wide variety of experiments. While production laboratories focus on waiting time waste between processes, MDT laboratory margin and turnaround time are strongly affected by defects.
Summary:
This report is composed of the main part, the first one aims to explain in detail what is venture capital and gives the majors insight from literature about the performance of venture capital fund. The second part display in detail the investment process used by DAA Capital Partners and discuss the usage of lean operation and continuous improvement to improve the process efficiency. Finally, the third chapter shows a concrete example of how Venture capitalist build their investment thesis by displaying all the different part of a real DAA deal’s investment memorandum under the form of a case study.
The first chapter highlight, using evidence from literature but also example on concrete individual case, that the Venture Capital industry is an important factor of economic growth. The different methods to assess the Venture Capital performance elaborated in the literature are presented. Even if the asset class as a whole tend to earn zero Alpha, literature shows that top VCs have persistence in performance. The different VC firm parameters and investment criteria that affect VC performance are then presented.
In the second part, the DAA capital investment process is presented. A measure of efficiency is proposed for the process as being the ratio between the added value time and the overall time spent on the process. Lean operation principles are used to identify inefficiencies. Concrete methods and tools are proposed to improve the process efficiency and reduce the “knowledge work deficit”.
Summary:
A key measure of success for any digital business is to engage with its customers with relevant and time-critical communications. To better understand their customers, the Customer Relationship Management team uses segmentation to engage with relevant and targeted communications. The goal of this project was to provide recommendations for engaging inactive customers. For this, we deep-dived into the movements between different customer segments, identified what factors and communications made them active, and based on these provided recommendations for future communications.
Results:
– We noticed that majority of customer movements were from active to inactive segments.
– In our deep-dive, we found that were three key factors explaining these movements.
Based on this, we provided communication recommendations in order to engage these customers.
Summary:
Alpana Ventures SA is a Geneva-based venture capital advisory company. Its first fund, created in 2016, is entering the last phase of its life, after the two initial steps of choosing start-ups and reinvesting in the best performing ones. Today, some portfolio companies reached important milestones and are mature enough to prepare their exit strategy. Such strategy relies on important human decisions. Even though existing literature can help, most of the work is dependant on factors specific to each company, that require unique consideration.
The purpose of this thesis is to bring a useful analysis regarding how company-related factors, and how the competition influence the exit outcome for a Swiss VC fund ; while each industry environment helps to define the most valuable exit route. I wish it will also benefit Alpana Ventures when preparing the exits for their second early stage fund (2019).
Results:
The literature brings to light the various mechanism in place in the venture capital industry. One can understand the juridical aspect behind every VC contract and how do Venture capitalists manage to keep sufficient control of the company’s strategy and decisions.
The comparable analysis allows understanding the competition scope of each startup correctly. Alpana Ventures now knows how competitive their portfolio companies are compared to the industry standards.
The model based on public companies’ revenue multiples has been a precious estimation of the valuation for each startup. Alpana ventures can now back up their negotiations with an approximation of each of their portfolio companies’ value. Such a model will likely be advantageous to Alpana when entering negotiations for future M&A.
Summary:
The goal of this project was to determine how many customers can be prioritized at a certain service level using the company’s future ERP’s allocation framework. The ERP’s allocation framework requires the user to segment customers into priority classes. To determine how many customers can be placed in each class while reaching target service levels it was necessary to first determine which customer are to be placed in each priority class. Indeed, the customer’s ordering habits influence how many customers can be served at a certain service level. Then a data set was built to backtest the ERP’s allocation policy. A strategy is described to find feasible numbers of customers in each priority class by backtesting and iterative variations of the customer segmentation. Due to difficulties implementing the ERP’s allocation policy in a tractable way, the backtesting was left for a future project. The last part of the project was to research improvements over the ERP’s allocation policy that anticipatedly backlog low priority orders to ration inventory. A rationing model using holding costs and penalties for late allocations was solved using dynamic programming.
Results:
The performance of 4 different customer segmentation were compared using partially simulated data. A reduction of 7.8 M$ of the present value of lost profit was achieved using a heuristic optimization technique.
A first come first serve (FCFS) allocation policy was implemented and backtest over 6 months considering inventory replenishment, expiration and demand. The simulated allocation policy achieved an 87.1% rate of on time and in full allocations (OTA).
The rationing model was tested on different examples, and it was shown that in situation of scarce inventory, the model enables savings compared to a myopic allocation policy such as the one used in the ERP system.
Summary:
This paper aims to find out whether data governance is a necessity in a financial services company nowadays. It is structured into three main sections. Firstly, we analyse the evolution and the current state of the data governance in an asset management company. We discover lack of data governance policies and control which lead to data quality issues and misalignment in methodology of performance and risk calculation across business teams. Secondly, based on the analysis we present a centralising tool. We seek to uniquely define the methodology of financial measures and implement them in the performance and risk calculation tool. The analysis shows that the tool improves the company’s hybrid data governance model. The effects are, namely, reduced duplicate processes between business units, increased collaboration and centralised methodology and calculation of performance and risk measures. Lastly, we define differences in the methodology across business units and provide a basis for discussion in their standardisation in the financial industry.
Results:
First, we discover lack of data governance policies and control which lead to data quality issues and misalignment in methodology of performance and risk calculation across business teams. Second, the analysis shows that the tool improves the company’s hybrid data governance model. The effects are, namely, reduced duplicate processes between business units, increased collaboration and centralised methodology and calculation of performance and risk measures. Third, we define differences in the methodology across business units and provide a basis for discussion in their standardisation in the financial industry.