Menu
The MONSOON project - MOdel-based coNtrol framework for Site-wide OptimizatiON of data-intensive processes - aimed to establish data-driven methodology to support identification and exploitation of optimization potentials by applying model-based predictive controls so as to perform plant and site-wide optimization of production process. The ambition of MONSOON project is shared by 2 significant process industries from the sectors of aluminium and plastic.
"Process industries represent a significant share of European industry in terms of employment and turnover, but also in terms of energy, resources consumption and environmental impact. MONSOON vision is to provide such industries with a dependable, replicable and cost-effective methodology that helps them achieving significant improvements in the efficient use (and re-use) of raw resources and energy, easing effective use of cross-sectorial competences."
MONSOON dissemination material:
About the MONSOON project:
Start Date: 1st October 2016
Duration: 36 months
Budget: 5 million €
Coordinator: Claudio Pastrone, LINKS
Innovations features
The MONSOON product provides a unique real time infrastructure to deal with heterogenous systems of data-intensive processes. The Cross sectorial DataLab is a unique platform where data is collected and data scientist tools are available to build tests and improve predictive functions. The semantic modeler allows a formalization of the data analysis process and creates a tight bond between the data scientist and the domain experts, in order to accelerate data scientist's comprehension nd predictuve function relevance. Online life cycle assessment pluging is a powerful tool for energy intensive companies to follow the trend of environmental cririucal parameters.
MONSOON Framework
Data Analytics Services
The aim of the services is to exploit the MONSOON Data platform's capabilities to collect, store and process large volumes of data from several sources, as well as the possibilities to easily deploy the predictive models in the plants. MONSOON Data Analytics Services supports the standard process of the data analysis in multiple phases. Data and problem understanding enables sharing of the domain knowledge between the domain experts and the data scientists. Semantic framework enables the domain experts to create the production processes models, describe related data and predictive functions with associated KPIs to improve data scientists knowledge of the industrial domain.
Life Cycle Assessment Service
Life Cycle Assessment (LCE) is adopted in the MONSOON solution as backgroung methodology for the life-cycle-management plugin. LCA plays the role of the environmental sustainability assessor, ensuring a cross level approach to support a quantitive evaluation of the environmental impact of the investigated system.
Runtime Container
It proposes a common infrastructure for running/exploiting predictive functions on various industrial plants. Experts and operators can access to the predictive functions results via graphical dashboards (Grafana).
System Integration Service
The MONSOON Platform can combine the integration of new components (new predictive functions or new connectors), based on the nned specification tobe used, and deliver solid methodology and tools to support the continuous integration. It follows different types of services in a Docker based infrastucture: the industrial plant data are collected using specific connectors and ingested into MONSOON Platform (DataLab and Runtime Container) by a NiFi data flow and a KairosDB database and the automatic deployment of the platforms (DataLab and Runtime Container) is based on Ansible and Docker Swarm.
DataLab Platform SAS
The DataLab developed in MONSOON project is the data analytical platform for storing and collaboratively analysing the data coming from the factories and serves as development environment for the predictive functions that can be deployed seamlessly into the Runtime Container. It is intented to be used by the data scientists and process managers.
The MONSOON project consortium
Tweets by @MONSOON_EU
The Technical Objectives (TO) of MONSOON project were:
MONSOON have defined an effective data-driven, model based, multi-level distributed control methodology suitable to link with each other robust model based monitoring and control systems and coordinate systems existing at different layers. The increased amount of correlated information and knowledge have provided plant and site managers with increased awareness and control over the effective use of equipment as well as over the use of raw material and energy resources.
In order to achieve seamless, plant- and site-wide distributed control, MONSOON developed a comprehensive integration methodology, suitable to connect all components in a common, dynamic ICT infrastructure supporting the monitoring of data-intensive flows and distributed control. The proposed methodology had been verified in the field by deploying dedicated resource-aware software connectors, suitable to integrate existing process industry systems.
Building upon its real-time and dependable integration infrastructure, MONSOON employed semantic techniques to define, inter-link and share distributed models describing all the key aspects of multi-scale processes of interest. Defined models are now accessible by means of pragmatic open APIs, suitable to help factory systems developers to look-up existing relationships and entities.
MONSOON have integrated a library of scalable analytics functions suitable for application in the domains of interest. Such functions which are now executable either on stored off-line data or in real-time fashion based on data fed in real-time by the field, and deliver new, refined metrics and information which can in turn be seamlessly fed in model based control loops. The novel data-driven techniques that are applied either on historical off-line or in real time data are statistical functions, trend analysis functions, machine and deep learning algorithms.
MONSOON had included a core set of analytics and visualization tools to fuse data coming from disjoint plant levels. Such tools detect complex patterns of manufacturing processes and provide useful information both for supporting short-term analysis (optimization, scheduling, monitoring of KPIs) and refinement of long-term manufacturing strategies (re-design, new processes, ramp-up).
In order to cope with the growing data complexity MONSOON has defined and implemented hybrid simulation infrastructure suitable to facilitate design, development, integration, deployment and testing of predictive control algorithms. The proposed modular framework is enabled to generate data streams produced by virtualized entities as data generators based on models of the represented entities and historical data. The framework employs plant-wide and site-wide structures of virtual entities by the selection, constraining and configuration of available formally described models.
MONSOON has integrated life-cycle (LC) management tools in a symmetric fashion e.g. enabling on the one hand LC tools to access data from enterprise resource planning (ERP) and manufacturing execution system (MES) system and on the other hand to feed LC targets and elaborated metrics back into the control infrastructure e.g. allowing definition and deployment of dedicated LC controls in the infrastructure itself.
The cross-sectorial MONSOON solution is developed and evaluated in two industrial sites from the Aluminium and Plastics sectors to assess its acceptance and usability by its intended end-users and for its potential effectiveness and impact on resource optimization. In the Aluminium production industry, the selected scenario focuses on predictive monitoring of a large potlines, where the increased amount of collected data from low-level control systems is exploited for data-driven predictive control, leading to earlier detection of anomalies and identification of potential optimization gains. In the Plastics domain, the selected uses focuses on enriching existing injection molding equipment with additional data-intensive in-mold sensors, and on the integration of extracted results with information from higher levels of the SCADA pyramid, leading to a faster and more precise identification of potential production problems.
MONSOON proposed methodology for plant-wide monitoring and control
This section outlines the overall high-level conceptual MONSOON methodology for developing model based predictive function suitable to monitor and control data-intensive production in process industries. MONSOON follows a cognitive OODA cycle (Observe, Orient, Decide, Act).
As a pre-requisite, a scalable infrastructure for monitoring process must be in place, spawning one or more production plants or sites (observe phase). Data from the monitored data-intensive processes is continuously collected, stored, annotated with relevant meta-information suitable to keep relevant relationship among different data sets and made available within a development environment designed to support collaborative development of model based predictive functions, namely the MONSOON Data Lab. By using such component, “pain” situations are detected by process experts analysing data (e.g. by benchmarking against standards or similar production processes) or just reported and enriched with detailed information by the production floor manager suspecting inefficiencies (orient phase). Once a “diagnosis” is made, a suitable control solution is developed or selected among the set of controls belonging to the company’s control functions knowledge base. At this stage, multi-scale controls are developed through different iterative cycles of development and evaluation, performed with different approaches depending on the required evaluation needs at hand (e.g. evaluating feasibility, performance, potential impact, etc.) using a hybrid mix of off-line and on-line processing techniques and predictive analytics. At this stage, typically also the business feasibility of the developed solution is evaluated, to verify whether expected gains are sufficient to justify investments e.g. in deploying new sensors or actuators to support the developed controls in the production environment.
Once a feasible and sustainable solution matching the needs is identified and pre-evaluated, a rapid prototyping and deployment stage occurs, resulting in the controls that are made operational within a dedicated production runtime integrated in the plant control infrastructure (act phase). At this stage, the control is able to interact with automation systems managing processes at different levels (i.e. PLCs, SCADA systems, MES, ERP, etc.) to monitor data and perform controls. At this stage, the available monitoring infrastructure can be used to perform evaluation of impact and/or life-cycle Key Performance Indicators (KPIs). Throughout the whole process, a set of dynamic, multi scale models suitable to describe both domain-specific and cross-sectorial phenomena in process industries are employed to ensure interoperability among all systems supporting the process. Such models are chosen among available standard models and/or proposed as standardization output by MONSOON, and allow semantic inter-linking of meta-information from devices, process, management and enterprise information from the production floor, as well as of data extracted by processing from the available collected data sets.
The collaborative companies with MONSOON project are Aluminium Pechiney and GLNPlast. Their plants are used as pilot sites for the implementation of MONSOON project's techniques and methodologies.
Aluminium Pechiney Company is a world leader in aluminium production and the French subsidiary of Rio Tinto Aluminium. Aluminium Pechiney has worked for more than 40 years on the development of electrolysis process equipment, process control, manufacturing execution and advanced data analysis and has proposed its Dunkerque plant as an indicative use case as there is an intensive need for plant-wide monitoring within its aluminium production, carbon and potline process.
Based on the electrochemical reduction of aluminium oxide (alumina), a process invented at the end of the 19th century and radically improved since then, the Aluminium Dunkerque plant is directly employing 550 people and more than 400 indirectly. Most of the contractors, as well as 90% of employees are coming from the neighbouring community, which shows the commitment of the smelter to develop local economy and local skills.
The Dunkerque smelter is in fact the highest-producing primary aluminium smelter in the EU-28 area. Located in northern France, on the North-Sea coast close to the Belgian border, it is benefitting from the very dense transport and energy infrastructure of the Hauts-de-France Region. Its main clients are located in France, Germany and the Netherlands, mainly in the canstock and transportation industries.
Built in 1990, and started up in 1991, the Dunkerque plant is also the first aluminium factory in France with 65% of total national production and Europe’s largest sheet-aluminium producer and an important player in the ingot market as well as one of the most modern smelters, with a state of the art technology and equipment, insuring as well a minimised environmental impact.
The smelter is equipped with 264 AP39 electrolytic pots in one potline operating at 390 kA, for a total yearly production of about 280.000 tons of primary aluminium. As an energy intensive process, its 450 MW nameplate power input result in a power consumption of 3.7 TWh of electricity per year, equivalent to a 1-million people city consumption.
For the past GLN Plast dedicates its activity to the injection thermoplastic components for use in various and di-verse segments, such as the automotive, pharmaceutical, medical, electronic, cosmetic and food packaging.
GLN Plast specializes in the mass production of castings in an environment certified through ISO/TS 16949 and ISO 9001. It has nearly 40 plastic injection machines and is capable of testing molds in machines weighing be-tween 40 to 420 tons and can guarantee the production of pre-series for the approval of new products as well as the production of large, medium and small series. Hence, GLN Plast can produce final parts, ready to be distributed and that can also include small assemblies, packaging and laser engraving. It also offers a wide range of high added value services to deliver market-ready products and develops customized injection solutions to match the productivity and efficiency requirements of optimized productions. As such, the company has an integrated offer, capable of supplying mold and injection, susceptible to providing benefits in terms of knowledge enhancement and cross-selling opportunities.
The company is a member of a larger industrial group, named GLN Group, which incorporates three companies, all tied to the molding and plastic industry. Several decades of experience and constant technological innovation, establish it as a dynamic group of excellence that integrates the phases of Product Development and Engineering, the Development and Manufacture of high precision molds, and the injection of plastic parts, supported by complementary high added value services. This association provides GLN Plast with far reaching knowledge of the sector and production technologies, as well as the necessary financial muscle to provide it with the investment capacity to enable it to easily adjust to market oscillations and adapt to new technological developments. This corporate reality has also provided the opportunity for the creation of a shared R&D Centre – GLN Innov.
For GLN Plast, human resources development is a high priority. In this light, GLN Group has a GLN Academy – its own knowledge center, in order to sustainably assure a highly qualified workforce. It organizes and provides courses focused on capacitating increases in productivity and is based on a wider knowledge sharing and innovation based corporate culture. The GLN Academy is the first and only of its kind in Portugal, within the context of mold and plastic industries.
The MONSOON Consortium consists of 11 complementary partners from 7 different European Countries, namely Italy (Torino, Castellamonte), Germany (Munich, Ludenscheid), Greece (Thessaloniki), Slovakia (Kosice), France (Voreppe, Suresnes, Montbonnot Saint-Martin), Portugal (Maceira) and Spain (Madrid). All partners are combining knowledge to achieve project aims.
Here is a list of partners of the MONSOON project:
External Stakeholders Group (ESG)
The infographic of MONSOON project outcomes is provided below:
The MONSOON platform consists of two major components: the cross sectorial data lab and the real-time operations platform as can be seen in the high-level architecture. Each of them contains several subcomponents which are connected in various ways. For example, data from the real-time operations platform is fed to the big data storage which is contained in the cross sectorial data lab and contents from the cross sectorial data lab are deployed in the runtime container which is part of the real-time operations platform.
The Cross-sectorial Data Lab provides a collaborative environment where big amounts of data are collected, stored, and processed in a scalable way. It enables multidisciplinary collaboration of experts allowing teams to jointly model, develop and evaluate distributed controls in rapid and cost-effective way. The Data Lab eases the definition of predictive control and life cycle management functions, allowing to work in a simulated environment or to exploit co-simulation by mixing stored data with data flowing in real-time from the real systems.
The depicted architecture of the Cross Sectorial Data Lab platform it consists of the following main components:
Big Data Storage and Analytics Platform
The Big Data Storage and Analytics Platform provides resources and functionalities for storage as well as batch and real-time processing of the operational data from multiple site characterized as Big Data. The platform combines and orchestrates existing technologies from the Big Data and Analytic landscape and sets a distributed and scalable run-time infrastructure for the developed data analytics methods.
Development Tools
The Development Tools provide the main collaborative and interactive interface for data engineers, data analysts and data scientists to execute and interact with the data processing workflows running on the Data Lab platform. Using the provided interface, data scientists can organize, execute and share data, and code and visualize results without referring to the internal details of the underlying Data Lab cluster.
Real-time Plant Operations Platform
The Real-time Plant Operations Platformis in charge to (a) communicate with the heterogeneous existing systems already used in process industries, (b) support data collection, storage, and interaction with the process industry systems respecting required real-time / dependability constraints and under the assumed data intensive conditions. The relevant information acquired from the plant is forwarded to the “Cross Sectorial Data Lab”.
The presented architecture of the Plant Operations Platform is composed by the following main components:
The Monitoring Tools support monitoring of plant wide resources in order todetect problems and faults in the systems and provide support of various industry-standard protocols like SNMP and is based on Cactiiand Zabbix – open-source industrial grade monitoring solutions. Notifications in case of errors or alarms was performed through visualization dashboards and through notifications by email or other suitable methods of communications. The main function of the Virtual Process Industries Resources Adapter (VPIRA) is data integration, mediation, and routing.
As shown, different connectors provide data to the VPIRA main component, devoted to the data pre-processing. This module called ‘abstractor’ is in charge to prepare the data according to the specific request coming from the Big Data Storage and Runtime container as well as to send the elaborated data back to them. VPIRAs can integrate devices supporting a wide range of standardized protocols such as OPC-UA, Euromap63as well as business systems like ERP, MesAL or Historian PI.
The Runtime Container is the environment where predictive functions are executed over real-time collected data.
The Data Orchestrator is the central component of the Runtime Container. Its goal is to be the interface between the Runtime Container environment and the outside world as well as to coordinate the data communication inside the Runtime Container. It also filters the raw data according to the needs of the deployed predictive functions. Various Data storage solutions e.g. databases are supported to store/retrieve raw data, predictive function results and visualization data. Through the Visualization Dashboard customized solutions can be developed to present data from the predictive functions. In addition to that management components for the Predictive Functions and the Runtime Container as well have been implemented.
Cross-sectorial domain model
The main tool introduced by the Cross-sectorial Domain model to improve communication between the users is the shared vocabulary collaboratively created by the domain experts and data scientists. The domain concepts are described as the vocabulary entries which unambiguously define their meanings.
The model enables the users to specify the predictive function, input data, output data and evaluation results,andextend the predictive functions description in more detailed fashion including model structure and performance metrics. The structure of the controlled vocabulary is domain independent and the proposed JSKOS format can be used to describe concepts in any domain.
The top concepts can be divided into:
In order to collaboratively design semantic domain models a modeller was developed that is based on open-source components. This web-based graphical user interface allows to create the particular elements of the semantic domain model and different domain model aspects, including the modelling of the processes, process segments, equipment used in those processes, data attributes and predictive functions, and create dependencies between those elements. Along with the web-based interface, Semantic Framework also provides a RESTful API interface, which enables the modelling tool to communicate with external applications.
Development and Testing Environment
Since it is used different technologies and different languages to build components of MONSOON platform, there is no single wayof developing and testing them. Indeed, developing a web application with user interface does not require same testing as developing a Predictive Function which will always run in backend, neither does it requires the same environment.
Each service deployed in MONSOON project were deployed as a Docker container inside a Swarm cluster. As such, every component can be developed and tested locally by developers. Itcan be provided a production-like environment for testing purpose. This environment consists in a Swarm cluster running on several virtual machines with limited resources.
For managing development of the MONSOON PlatformtheGitLabwas used which is a web-based Git-repository manager providing wiki, issue-tracking and CI/CD pipeline features, using an open-source license, developed by GitLab Inc. An instance of GitLab was set up at CERTH. The choice for GitLab is motivated by the fact that CI services are fully integrated into it and is enabled by default in all projects.
The MONSOON project comprises many different software components from simulation and visualization to predictive function modules and more. As such, and according to best practices each specific task will have its own relevant repository, i.e. the repository that contains the code that implements the software component that addresses the problem/Proof-of-Concept (PoC) introduced for the particular task.
Not technical outcomes
The outcome of the project is not only technical, it includes also the methodology development and know-how, as well as the activities and achievements in terms of publications and dissemination. Such outcome can be found in the dedicated pages in the website. A number of lessons learned during process are: the implementation of a data oriented methodology in real industrial plants, the difficulties in implementing R&D initiatives in real plants in operation also addressing aspects like maintenance operations, the internal R&D roadmaps, the interaction with and needs from customers, the interaction with the technology providers e.g. for updating machines.
List of public deliverables of MONSOON project:
D2.1 - MONSOON platform usage scenarios (Public)
Leader: Fraunhofer FIT | 30 November 2016
The purpose of this deliverable is to document and describe a set of plausible usage scenarios in the year 2020 and beyond for the MONSOON platform in the two domains Aluminium and Plastics but should represent scenarios that are also common or at least applicable to other process industry sectors as well.
D2.2 - Initial Process Industry Domain Analysis and Use Cases (Public)
Leader: Aluminium Pechiney AP | 30 December 2016
The purpose of this deliverable is to document and describe the aluminium and plastic domain-specific and cross-sectorial use cases, defining how the MONSOON platform will be used for predictive optimization and scheduling tasks in production plants and sites.
D2.3 - Updated Process Industry Domain Analysis and Use Cases (Public)
Leader: GLNPlast GLN | 30 March 2018
This deliverable is the first update of D2.2 “Initial Process Industry Domain Analysis and Use Cases” where the initial version of the state of the art analysis for the aluminium and plastics domain was documented, as well as the first overview of the domain and business use case descriptions which focus rather on the technological and business aspects.
D2.4 - Final Process Industry Domain Analysis and Use Cases (Public)
Leader: Kunststoff-Institut Luedenscheid KIMW | 30 January 2019
This deliverable is the update of D2.3 “Updated Process Industry Domain Analysis and Use Cases” where the initial version and the first iteration of the state of the art analysis for the aluminium domain and the plastics domain was documented, as well as the first overview of the domain and business use case descriptions which focus rather on the technological and business aspects.
D2.5 - Initial Requirements and Architecture Specifications (Public)
Leader: Fraunhofer FIT | 29 March 2017
This deliverable D2.5 Initial Requirements and Architecture Specifications describes the initial requirements, the first version of the architecture description of the MONSOON platform, including stakeholders which might be relevant for this project and the platform, as well as a scenario description for the use cases in the aluminium and plastics domain. There will be an update of all these topics in D2.6 Final Requirements and Architecture Specifications which is due in month 21.
D2.6 - Final Requirements and Architecture Specifications (Public)
Leader: Fraunhofer FIT | 30 December 2018
This deliverable D2.6 Final Requirements and Architecture Specifications describes the updated requirements, the final version of the architecture description of the MONSOON platform, including stakeholders which might be relevant for this project and the platform, as well as a scenario description for the selected use cases in the aluminium and plastics domain.
D2.7 - Initial Cross-sectorial Domain Model (Public)
Leader: Technicka Univerzita V Kosiciach TUK | 29 September 2017
Main objective of the work presented in this deliverable is to define the initial semantically-enriched model that serves as a basis upon which all other models will be developed. Since the MONSOON platform will integrate various systems and technologies, proper representation of the information and data is required. Semantic model will cover production processes, which will be modelled using functional blocks with specified inputs and outputs. Model will cover the data pre-processing steps as well as predictive functions. The semantic model will leverage existing standards for description of both production processes and predictive functions.
D2.8 - Final Cross-sectorial Domain Model (Public)
Leader: Technicka Univerzita V Kosiciach TUK | 30 December 2018
This deliverable describes the final specification of the Cross-sectorial domain model. The goal of the deliverable is to introduce the extensions of the initial version of Cross-sectorial domain model presented in deliverable D2.6 Initial Cross-sectorial domain model. The initial model specification presented the main concepts required for modelling of production processes, data elements and predictive functions. The idea was to provide the basic concepts to describe the production processes and related elements. Initial model served as a basis for modelling of application domains in MONSOON project. This deliverable presents the extension of the initial model with the new properties and relations, which were identified and incorporated from the experiences gained during the modelling of pilot aluminium and plastic domains.
D3.1 - Initial Real Time Communications Framework (Public)
Leader: Capgemini CAP | 30 December 2016
The document describes:
D3.2 - Updated Real Time Communications Framework (Public)
Leader: Fraunhofer FIT | 30 November 2017
This document is an updated version of D3.1 – Initial Real time Communication Framework. In D3.1, the way that data will be collected on both aluminium and plastic domain as well as a pre-selected list of tools of monitoring tools (Active & Passive mode) have been proposed. After D3.1 was delivered, the project team has been focusing on implementing and deploying solutions proposed in D3.1. As a result, this deliverable is dedicated to document the preliminary results of such activities.
D3.3 - Final Real Time Communications Framework (Public)
Leader: LINKS Foundation LINKS | 30 May 2019
This document is a final version of D3.3 – Final Real time Communication Framework. The D3.2 described the way in which data has been collected for both aluminium and plastic domain, as well as the chosen list of tools for monitoring (Active & Passive mode). After D3.2 was delivered, the project team has been finalized the deployment of the solutions proposed in D3.1 and D3.2. As a result, this deliverable documents the final outcomes.
D3.4 - Initial Virtual Process Industries Resources Adaptation (Public)
Leader: LINKS Foundation LINKS | 31 January 2017
The D3.4 “Initial Virtual Process industries Resources Adaptation” collects the initial specification of the virtual process industries resources connector along with the description of the environment infrastructure, in both pilot sites, and the initial architecture and modules.
D3.5 - Updated Virtual Process Industries Resources Adaptation (Public)
Leader: Capgemini CAP | 30 November 2017
The D3.5 “Updated Virtual Process industries Resources Adaptation” collects the updated specification of the virtual process industries resources connectors along with the description of the environment infrastructure, in both pilot sites, and the updated architecture and modules. It contains an update of the document D3.4 “Initial Virtual Process industries Resources Adaptation”. This document provides the updated description of the fundamentals of the connectors used for the phase 1 of the MONSOON project and an overview of the role that the components will cover in the final framework (phase 2).
D3.6 - Final Virtual Process Industries Resources Adaptation (Public)
Leader: Fraunhofer FIT | 31 May 2019
The D3.6 “Final Virtual Process Industries Resources Adaptation” collects the final specification of the virtual process industries resources connectors along with the description of the environment infrastructure in both pilot sites, as well as the final architecture and modules. It contains an update of the document D3.5 “Updated Virtual Process Industries Resources Adaptation”. The Section1.2 gives all updated/new sections in D3.6 versus D3.5. This document provides the final description of the fundamentals of the connectors used within the MONSOON project and an overview of the role that the components will cover in the final framework (phase 2).
D3.7 - Initial Runtime Container (Public)
The task T3.3 of the MONSOON project aims to implement the Runtime Container, namely the component executing model based predictive control functions at runtime, running within the overall plant infrastructure. The document D3.7 “Initial Runtime Container” collects the initial specification of the Runtime Container along with the description of the environment infrastructure, in both pilot sites, and the initial architecture and modules.This document provides the initial description of the fundamentals of the container components used for the period 1 of the MONSOON project and an overview of the role that the components will cover in the final framework (period 2).
D3.8 - Final Runtime Container (Public)
Leader: Capgemini CAP | 30 May 2018
The document D3.8 “Final Runtime Container” collects the final specifications of the Runtime Container along with the description of the environment infrastructure and the final architecture and modules. The deployment on specific Aluminium and Plastic domains is described in document. This document provides a description of the fundamentals of the components in the Runtime Container used by the MONSOON project.
D4.1 - Initial Semantic Framework for dynamic multy-scale industry modelling (Public)
Leader: Technicka Univerzita V Kosiciach TUK | 30 November 2017
The purpose of the Initial Semantic framework for dynamic multi-scale industry modelling deliverable is to provide an initial overview of the semantic framework tools architecture by providing architectural views and perspectives of the various system design models.The goal of this deliverable is to effectively communicate the architecture of the semantic framework tools to the members of the MONSOON project.
D4.2 - Final Semantic Framework for dynamic multy-scale industry modelling (Public)
Leader: Technicka Univerzita V Kosiciach TUK | 30 July 2019
The purpose of the Final Semantic Framework for dynamic multi-scale industry modelling deliverable is to provide a description of the final update of the Semantic Framework tools and their architecture by providing architectural views and perspectives of the various system design models. The deliverable summarizes the extensions from the initial version of the Semantic Framework and describes the Semantic modeller final version as well as its usage incontext of data analysis methodologies. The goal of this deliverable is to effectively communicate the architecture of the Semantic Framework tools.
D4.3 - Initial Big Data Storage and Analytics Platform (Public)
Leader: Fraunhofer FIT | 31 March 2017
This document describes the distributed platform for Big Data Storage and Analytics that provides resources and functionalities for storage batch, and real-time processing of the big data. The platform combines and orchestrates existing technologies from Big Data and analytic landscape and sets a distributed and scalable run-time infrastructure for the data analytics methods developed in the project. The high-level architecture with its provided interfaces for cross-sectorial collaboration is presented. The solutions and technology options available for each logical component of the architecture are briefly explained.
D4.4 - Updated Big Data Storage and Analytics Platform (Public)
Leader: Probayes PROB | 30 November 2017
D4.5 - Final Big Data Storage and Analytics Platform (Public)
Leader: Capgemini CAP | 30 July 2019
D4.6 - Initial Multi-scale Model based Development Environment (Public)
In the context of MONSOON work package structure, Task 4.3 (Multi-scale Model based Development Environment) deals with the tools and interfaces that will cover the whole life cycle of the planning, implementation and deployment of data analytics functions developed using the algorithms provided by WP5 (Site-wide Scheduling and Optimization Toolkit) into the plant production supporting simulation/co‑simulation features.
D4.7 - Final Multi-scale Model based Development Environment (Public)
Leader: Center for Research and Technology Hellas CERTH | 31 July 2019
This report provides an update of the status of the architecture and tools mentioned before. The final version of the Development environment is based on the combination of Apache Zeppelin and JupyterHub. Both notebook based tools provide similar interfaces and user experience, and both use various programming languages such as Python, Scala, or R. Also, an update on the pipeline of a predictive function inside a Docker container is provided. As for predictive functions, they use various file formats, such as JSON and CSV, as input and output and the output files are further stored in a database, namely, Kairosdb and Mongodb. Kairosdb stores the visualization data, while Mongodb stores also predictive function results.
D5.1 - Initial Trend Analysis Functions (Public)
Leader: Center for Research and Technology Hellas CERTH | 31 December 2017
This document presents the initial trend analysis functions developed until M15 of the MONSOON project. This document is part of the “Task 5.1 – Trend analysis for local control & optimization in data intensive scenarios” mean to design and implement trend analysis techniques for and on significant and key process variables of both aluminium and plastic domains. This deliverable defines the initial approaches for the core set of algorithms, techniques and methodologies for trend analysis. With the implementation of such techniques we aim to provide detection of possible deviations from normal conditions, based on the need of aluminium and plastic production processes.
D5.2 - Final Trend Analysis Functions (Public)
This document presents the final trend analysis functions developed until M34 of the MONSOON project. This document is part of the “Task 5.1 – Trend analysis for local control & optimization in data intensive scenarios” mean to design and implement trend analysis techniques for and on significant and key process variables of both Aluminum and plastic domains. This deliverable defines the final approaches for the core set of algorithms, techniques and methodologies for trend analysis. With the implementation of such techniques we aim to provide detection of possible deviations from normal conditions, based on the need of Aluminum and plastic production processes.
D5.3 - Initial Online and Deep Machine Learning Techniques (Public)
Leader: Probayes PROB | 31 December 2017
This document describes the data science approaches lead during the first iteration of the MONSOON project. The analyses were done by applying machine and deep learning algorithms, in order to answer the different industrial use cases identified in the first iteration. The analyses are related to abnormality detection or process optimization for Aluminium and Plastic domain.
D5.4 - Final Online and Deep Machine Learning Techniques (Public)
Leader: Probayes PROB | 31 July 2019
This document describes the data science approaches led during the second iteration of the MONSOON project. The studies were done by applying machine and deep learning algorithms, as well as statistical analyses, in order to provide answers regarding the different industrial business cases identified during the project. The document is split in two main parts: the first concerns the Aluminium domain, and the second the Plastic domain. For each domain, we briefly describe the business cases, and explain in detail the data science approaches that were conducted, both in terms of methodology and obtained results. The data being used, and the algorithms that were tested, are carefully explained.
D5.5 - Initial Integrated Resource Optimization Toolkit, Decision Support (Public)
Leader: Technicka Univerzita V Kosiciach TUK | 30 August 2018
This deliverable is the initial specification of the Integrated Resource Optimization Toolkit. The main goal of this module is to provide real-time resource (raw materials, anodes, waste, energy, quality control etc.) optimization system. In MONSOON, optimization of the production process is based on the application of the predictive functions. This deliverable specifies general procedure, how the locally applied predictive functions can be combined in order to globally optimize overall production process targeting various global KPIs.
D5.6 - Final Integrated Resource Optimization Toolkit, Decision Support (Public)
Leader: Center for Research and Technology Hellas CERTH | 31 July 2018
This deliverable is the final specification of the Integrated Resource Optimization Toolkit. The main goal of this module is to provide a real-time resource (raw materials, anodes, waste, energy, quality control etc.) optimization system. In MONSOON, optimization of the production process is based on the application of the predictive functions. This deliverable specifies in detail the consolidated procedure, how the locally applied predictive functions can be combined in order to globally optimize overall production process targeting various KPIs. The initial deliverable regarding optimization (D5.5) described main concepts and general architecture of the Integrated Optimization Toolkit module. Also, it extended domain and business use cases for aluminium and plastic domain to overall process, in order to validate that the processes can be globally optimized by combination of the locally applied predictive functions. The current deliverable describes how the predictive functions developed within the scope of the Task 5.1 and Task 5.2 are combined in order to globally optimize the production process and to provide decision support to the end users.
D5.7 - Initial Lifecycle Management Plugin (Public)
Leader: Life Cycle Engineering LCEN | 31 December 2017
This deliverable represents the first step towards the complete definition of LC plugin properties and architecture, expected at the end of the MONSOON project. In this version, focus will be devoted to methodological aspects, to highlight the role of the component in the platform, as well as its similarities and differences with the rest of the toolkit which is the core, innovative potential of MONSOON.
D5.8 - Final Lifecycle Management Plugin (Public)
Leader: Life Cycle Engineering LCEN | 30 April 2020
This deliverable represents the update of D 5.7 – Initial LCA Management Plugin (RD.4). While the functional layout of the plugin has been designed in the first half of the project, second half has been devoted to implementation of the component in the investigated domains for different business cases. This deliverable provides an overview about how the LCA plugin is now working as an active part of the MONSOON platform. Whenever a reminder to architecture is present, RD.4 shall be considered as the main reference.
D6.1 - Test and Integration Plan (Public)
This document describes the comprehensive test, integration and deployment strategy for the software components developed in MONSOON, and how these software modules, sub-systems and systems will be integrated and deployed into a common prototype platform.
D6.2 - Initial Integrated MONSOON Platform (Public)
Leader: Aluminium Pechiney AP | 30 April 2018
This deliverable will present the different components of the MONSOON platform and their integration and deployment on pilot sites in aluminium and plastic domains. One phase of the deployment is the Factory Acceptance Tests, which describe the scenarios to be tested in the integration environment before deploying in the production environment and the objectives expected by each industrial partner. They will be thoroughly developed in this deliverable.
D6.3 - Final Integrated MONSOON Platform (Public)
Leader: GLNPlast GLN | 30 August 2019
This deliverable presents the last updates on the different components of the MONSOON platform and their integration and deployment on pilot sites on the aluminium and plastic domains, according to the first iteration described on RD.2. After a first set of tests (Factory Acceptance Tests - FAT and Site Acceptance Tests - SAT) on the site of each industrial partner, several combinations had to be updated and refined in order to achieve the project goals and the functionalities required in the different production scenarios.
D6.4 - Initial Deployment and Maintenace Report (Public)
Leader: GLNPlast GLN | 30 April 2019
This deliverable “D6.4 - Initial Deployment and Maintenance Report” describes the initial deployment of MONSOON platform into demonstration sites. In this document MONSOON platform refers to the Runtime Container only. The MONSOON platform congregates a set of functionalities oriented to a predictive management of the production area, integrated with the production equipment’s, machines and existing management tools. For the scope of MONSOON project, two different domains were considered – Aluminium and Plastic with two business cases each, in order to amplify the range of the cross-sectorial applicability.
D6.5 - Final Deployment and Maintenace Report (Public)
Leader: Aluminium Pechiney AP | 30 September 2019
This deliverable “D6.5 - Initial Deployment and Maintenance Report” describes the final deployment of MONSOON platform into both Aluminium and Plastic demonstration sites. The MONSOON platform congregates a set of functionalities oriented to a predictive management of the production area, integrated with the production equipment’s, machines and existing management tools. For the scope of MONSOON project, two different domains were considered – Aluminium and Plastic with two business cases each, in order to amplify the range of the cross-sectorial applicability.
D7.1 - Initial Evaluation Framework (Public)
Leader: Life Cycle Engineering LCEN | 30 March 2018
This document aims at providing the initial specifications for the evaluation framework; this is a set of KPIs that allows to quantify the benefits of the application of data-driven optimization methodologies from several perspectives (environmental, industrial, cross-sectorial and circular economy).
D7.2 - Final Evaluation Framework (Public)
This document aims at providing the final specifications for the MONSOON evaluation framework; this is a set of KPIs that allows to quantify the benefits of the application of data-driven optimization methodologies from several perspectives (environmental, industrial, cross-sectorial). This deliverable is the follow-up of D7.1 – Initial Evaluation Framework published at month 18 of the project.
D7.3 - Initial Demonstrators in the Aluminum and Plastic Domains (Public)
Leader: Aluminium Pechiney AP | 30 March 2019
This deliverable documents the process followed to develop the use cases in the Aluminium and Plastics domains. The development of use cases consists of 5 steps:
D7.4 - Final Demonstrators in the Aluminum and Plastic Domains (Public)
Leader: GLNPlast CLN | 30 August 2019
This deliverable D7.4 Final Demonstrators in the Aluminium and Plastics Domains describes the updated process followed to develop the use cases in the Aluminium and Plastics domains, as described in D7.3. The development of the use cases was structured in five steps, namely the Proof of Concept (POC), Proof of Value (POV), Proto or Feasibility per use case, which corresponds to the building of the Proto solution, Demo or Deploying for real time operation on the corresponding pilot site and deploying the model in different environments.
D7.5 - Initial Demonstrators Evaluation and Impact Report (Public)
Leader: GLNPlast CLN | 30 March 2019
This document is a first evaluation of the KPI framework defined in D7.1 from task 7.1. This task is a part of WP7, which is dedicated to stablishing tools to evaluate the effectiveness of the application of data-driven optimization methodologies developed by data scientists along with other tasks of the project and validation of the presented methodologies. The goal of this document is to present an analysis of the hypotheses as the first results basd on the calculation of the KPIs and evaluate if the tool framework is adjusted to what is expected to achieve and deliver feedback in terms of improvements or changes according to the actual needs/requests, based on each domain business cases.
D7.6 - Final Demonstrators Evaluation and Impact Report (Public)
Leader: Aluminium Pechiney AP | 30 July 2019
This document is the final evaluation of the KPI framework defined in D7.1 and D7.2 from task 7.1. This task is a part of WP7, which is dedicated to stablishing tools to evaluate the effectiveness of the application of data-driven optimization methodologies developed by data scientists along with other tasks of the project and validation of the presented methodologies. This deliverable firstly presents a study on data flow simulation, as an accelerator for predictive function development. A literature study is developed, explaining the state of the art of this subject. Then the current and future impact are described. The final analysis and associated calculations of the KPI framework defined and implemented during the project are also presented, according to the actual needs/requests, based on each domain (Aluminium and Plastic) business cases. The KPIs were categorized into different clusters as it was described on deliverable.
D8.1 - Communication and Dissemination Strategy (Public)
Leader: Center for Research and Technology Hellas CERTH | 30 December 2016
This deliverable presents the Communication and Dissemination Strategy for the European Union (EU), under its Horizon 2020 Research and Innovation programme (H2020), Sustainable Process Industry through Resource and Energy Efficiency (SPIRE) “MOdel-based coNtrol framework for Site-wide OptimizatiON of data-intensive processes (MONSOON)”.
D8.2 - Project Website (Public)
This document constitutes the Deliverable D8.2 – MONSOON Project Website of the MONSOON project (Grant Agreement No.:723650), and presents the online dissemination channels of MONSOON as they were created through the first three months of the project.
D8.3 - Initial Project Advertising Materials and Results (Public)
Leader: GLNPlast GLN | 30 October 2018
This document constitutes the Deliverable D8.3 – Initial Project Advertising Materials and Results of the MONSOON project (Grant Agreement No.:723650) and presents the project dissemination Material and the tools that been used to publicise the developments of the project also relying on channels and online platform.
D8.4 - Final Project Advertising Materials and Results (Public)
Leader: GLNPlast GLN | 30 November 2019
This document constitutes the Deliverable D8.4 – Final Project Advertising Materials and Results of the MONSOON project (Grant Agreement No.:723650) and presents the project dissemination Material and the tools that been used to publicise the developments during the 36 months of the project and also relying on channels and online platform.
D8.5 - Report on the Standardization Landscape and Applicable Standards (Public)
Leader: Asociacion Espanola De Normalizacion Y Certificacion AENOR | 31 March 2017
The D8.5 Report on the standardization landscape and applicable standards deliverable contains a detailed study on the relevant existing standards and standards under development currently in the International Standardization System, which are relevant for the Monsoon project.
D8.6 - Report on the Contributions to Standardization (Public)
Leader: Spanish Association for Standardization UNE | 30 September 2019
Standardization has been included in the MONSOON project with the aim of facilitating the acceptance and utilisation by the market of the developed solutions. Other objectives are to provide information about useful existing standards covering the needs of the different working packages, ensure compatibility and interoperability with what already exists in the market through standards, as well as to use the standardization system as a tool for dissemination of the project results and interaction with the market stakeholders. In addition to using existing standards, the standardization system is a way to promote that some innovative result, not yet standardized, of those developed in the project becomes some kind of normative document and can become in the future a reference standard for the sector, increasing the impact in the long term.
D8.9 - Updated Report on Standardization Landscape and Applicable Standards (Public)
Leader: Spanish Association for Standardization UNE | 30 January 2019
The D8.5 Report on the standardization landscape and applicable standards deliverable contains a detailed study on the relevant existing standards and standards under development at the time it was released within the Monsoon project (M6).
Articles:
Conferences:
The MONSOON newsletters are available in the following links:
Newsletter 2017 - Issue 1
In the newsletter we welcome you to the MONSOON Project and provide you an overview of the study cases, as well the presentation of the Pilot partners.
Newsletter 2017 - Issue 2
In this newsletter we share with you the “Thelast project developments”, introducing to the “Big Data Technologies” and the “Project Architecture”. You will also know a bit more about our partners and what we have been promoting.
Newsletter 2018 - Issue 1
New year, more news! In the this issue we achieve the Project mid-term and with it the last developments. You will get familiar with “CEN/CENELECWorkshopAgreement Development” and with the “Life Cycle Management inside”. We share also our “MONSOON Social Communication”, as well the presentation of more project partners.
Newsletter 2018 - Issue 2
… and another six more months have passed! In the newsletter we will update you on the last stages and give you an overview of the “Optimization Toolkit - tools for a continuous efficiency growth” and offering you a flavour… “Predictive culture, what else?”. We will present you two more partners and their involvement in the project, as well our last Dissemination events.
Newsletter 2019 - Issue 1
Welcome aboard on 2019 with MONSOON Project. In this newsletter we present our developments from the last semester. We address you our vision for “Digitizing the Plastics” and a brief explanation of the “Integration, thecore of thePlatform”. You will meet two more partners and check the last “MONSOON Social Communication”.
Newsletter 2019 - Issue 2
Farewell MONSOON!
In this last newsletter we share with you the very last news and developments regarding the MONSOON Project, introducing you to our “MONSOON SOLUTION” and the report of our last event “MONSOON FINAL WORKSHOP”. Finally, few more details about Coordinator Partner are presented along with the last Dissemination activities.
Project coordinator:
Claudio PastroneHead of IoT and Pervasive Technologies Research Area
LINKS Foundation Via Pier Carlo Boggio 61, 10138 Torino, Italy Tel: +39 011 2276612 Email: claudio.pastrone@linksfoundation.com