Deliverables – SmartWind

The list of deliverables submitted to date is as follows:

TITLE	DESCRIPTION
D1.1 Report on use case definitions and evaluation metrics	This document describes the use case definitions and evaluation metrics that will be carried out in the SmartWind project. It contains the final results obtained in the process of scope definition and initial design of the solution to be developed in the project SmartWind. The contents included in this deliverable will allow the reader to acquire a better understanding of the objectives of the project SmartWind, how the different members of the consortium will cooperate to comply with the objectives of the project, and what to expect to be obtained after the development of validations of the solution. A list of 12 use cases have been developed, which ranges from the detection and prediction of failures in most of the main mechanical and electrical subsystems that compose wind turbines (WT), to the provision of Operations and Maintenance (O&M) recommendations and analytics at a wind farm level. The approach that each of the use cases proposes to achieve the objectives of the project has been defined, as well as the mechanisms which will be used to measure the success of each one of them. Therefore, a list of Key Performance Indicators has been defined that summarises general optimization targets of the project but also of the different use cases considered in SmartWind.
D1.2 Requirements foruse cases’ information architecture	Within the project SmartWind, an integrated platform for cost reduction and revenue optimisation for wind farms operators is developed based on advanced and automated functions for data analysis, performance diagnosis, fault detection, performance diagnosis, root cause analysis and Operations and Maintenance (O&M) recommendations. These functions are designed and integrated in a cloud platform for the important and relevant subsystems of a wind farm and the turbines, which are defined in the use cases investigated in the project. In this deliverable, the optimization objectives of the use cases are summarised including the early detection of component failures and measures to minimize the main deterioration effects that limit the energy yield of the wind farm and that deteriorate the lifetime and availability of the assets. As a central part of the document, the requirements for the information architecture of the integrated platform are defined. Specifically, the required measurement data and the corresponding technological, functional and non-functional requirements are summarised for each use case of the project. Furthermore, business requirements of the WF operator regarding the overall information architecture have to be taken into account to ensure the application-oriented research of SmartWind. To give an overview of the results of D 1.2, conceptual illustrations of the existing and planned information architecture visualize the data flow and the interdependencies of different use cases in the project SmartWind. They include the data transfer from the wind farm to the platform with its data pre-processing and evaluation within the use cases to fulfil tasks of predictive maintenance and operation optimization. The results of the analyses are brought together and evaluated to provide O&M recommendation and action to the wind farm operator. To conclude, D 1.2 forms the basis for the information architecture that will be developed in SmartWind, as the requirements posed by the different use cases at the available measurement signals are merged. Therewith, the feasibility of planned evaluation processes within the use cases is ensured and restrictions can be considered in an early stage of the development process.
D2.1 Existing and desirable sensors to develop predictive maintenance for each use case	This document describes the existing and desirable sensors of the test wind farm to develop predictive maintenance and operation optimization for each use case. Monitoring of signals requires the identification of suitable sensors solutions. Sensor selection, deployment and characterization are therefore fundamental steps for the implementation of condition-based maintenance strategies and technical developments envisaged within the project. Initially in Chapter 3, this document details the signals that will be used to determine the condition of the production equipment components and the sensor systems currently available on the demonstrators. In Chapter 4, this document proposes a set of guidelines for selecting the sensor system solutions best suited to monitor the condition of use cases’ equipment. Such guidelines are intended to be applicable to different technological fields and will be tested in the context of the end users’ demonstrators.
D2.2 Integration and interconnectivity analysis	In this part of the project, various studies have been done on integration and interconnection analysis. With these studies, the integration of the models to be installed into the existing systems has been ensured and all partners in the project have brought a new perspective. In addition, when this document is examined in detail, how the outputs of these integration studies are visualised and how they are used in the sector are emphasised. The purpose of this document is to analyse and integrate appropriate data for predictive and maintenance studies in cyber physical systems. Analyses are performed on the tools used within the scope of the project with the integration and interconnection analysis. The data analysed in these tools are transferred to the SCADA system, and the IOT communication between SCADA and the devices is completed.
D2.3 Network and storage architecture report	In this deliverable, network and storage architecture structures regarding the SmartWind project are explained. SmartWind data warehousing will utilize Netaş’s IoT platform ION which will acquire the data from SCADA tenants, process it and forward it to external parties. The details of this process and of ION are narrated throughout the deliverable. ION is using the ELK system for data storage where Elasticsearch and Kibana are used and data is stored in the NoSQL format. A SCADA ION Adapter service will be integrated to ION to collect SCADA data which will then be reformatted and transferred with telemetry API in MQTT-HTTP format to ION. External parties will be able to query this data in several ways including via Elasticsearch through different databases such as mongoDb, influxDb, elasticsearch, redis etc. and via utilizing the reporting HTTP API of ION or by using the request / response method. In conclusion, this deliverable presents the work done on the adaptation of the IoT platform ION to the SCADA servers used for the wind turbines operated by Zorlu Energy as well as the preparation made to make the collected data accessible to the relevant external parties in the consortium. Furthermore, use cases related to the network and storage architecture designed by Netaş, Isotrol, RUB, and Enforma are presented.
D2.4 Sensors, network and storage architecture configuration and software	This deliverable includes the main results obtained in the development of the WP2 Data characterisation and modelling of the project SmartWind. These results have been obtained from the participation of the members of the consortium in the design, development, and preparation for the implementation of the different subsystems that will conform the SmartWind system as a whole. From data acquisition to data storage and integration with sensors and SCADA data, the architecture presented here allows the seamless integration between the different modules and networks from different actors, to allow the extraction of data from the edge layer of the system to its later storage and preparation for the execution of algorithms. The resulting architecture of this process is presented at the beginning of this document, as an overall visualization of the complete systems and their interactions. Later, the specific processes followed by each of the members of the consortium to successfully achieve the integration of the systems to conform said architecture is presented.
D3.1 Physical Signal and Attribute Analysis related to each use case	The present document describes the main results obtained after the execution of the task T3.1 Physical Signal and Attribute Analysis, one of the tasks that conform to SP3 Data Analytics for Performance Monitoring and Fault Detection. This deliverable is structured following a division between the processes carried out in the different use cases by the different partners in the project for the initial analysis of data, alarms and parameters to be used in the training of models, preparation of algorithms, and their later execution while implemented in the totality of the O&M decision support system to be developed in the project SmartWind. One of the first steps to be carried out in any data analysis project is the initial processing of the input data, to allow the preparation of the initial inputs for the later stages of the process. In the context of the project, data will be captured and treated in an initial process of data cleansing and data quality, implemented as a part of the UC12. After that initial process, the modules responsible for the execution of the different use cases will execute an initial preparation of the data for the proper detection and prediction of problems and failures. In the same way, the different models being developed in the modules for the implementation of the different use cases will require the processing of signals and their preparation (using clustering, outlier removal, among other techniques) to ensure a proper development of the models. As a result of the process of analysis of the available magnitudes and alarms for the later detection of issues and failures in each of the subsystems as well as the design and development of the best tools for the initial analysis of them in the context of each of the use cases, some initial preliminary results were achieved, such as potential underperformances or issues detected in magnitudes related to each subsystem under study. The current dataset is composed of data relative to a total of 5 wind turbines (WT), with a frequency of 10 minutes in a timespan that encompasses a total of 6 months (from January 2020 to June 2020). Wider datasets will allow a more detailed identification of problems to polish the mechanisms and preliminary results presented in this document.
D3.2 Report with the implemented Advanced Data Analytics, Diagnostics & Prognostics	The present document describes the main results obtained after the execution of the task T3.2 Implementation of Advanced Data Analytics: Diagnostics & Prognostics, one of the tasks that conform to WP3 Data Analytics for Performance Monitoring and Fault Detection. This deliverable, in addition with the software included in D3.3 Data analytics software, includes the main advancements achieved by the consortium in the development of tools for the detection of issues and faults in wind turbines, as well as underperformances, along the second year of the project’s execution. In the third and final year, the final tuning of the algorithms according to the results obtained in an initial validation process, as well as taking into consideration the specific requirements from the operation and maintenance (O&M) perspective. The document is structured in a report manner in which the main identification of problems, deviation from normal operation, and overall analysis of the operation of the wind farm under study, which will be analysed in detail in the last year of the project during the validation process, as well as the polishing of the results using feedback from O&M experts from Zorlu.
D3.3 Data analytics software	This document describes the software used to obtain the main results obtained during Task 3.2, namely “Implementation of Advanced Data Analytics: Diagnostics & Prognostics”. This comes as one of the tasks under work package 3, “Data Analytics for Performance Monitoring and Fault Detection”. This deliverable can be considered a companion to Deliverable 3.2, “Report with the implemented Advanced Data Analytics, Diagnostics & Prognostics”. It shows results from the software development work of Isotrol/Tecnalia and Enforma The software will later be used alongside the O&M decision support system in the succeeding work package in order to obtain meaningful recommendations for future actions.
D4.1 Failure modes and criticality analysis for wind farm elements	As basis for the development of an O&M decision support system that provides recommendations for the operation and maintenance planning of wind farms, at first potential failures and causes for underperformances need to be identified and quantified in their criticality. Therefore, methods of the Failure Mode, Effects and Criticality Analysis (FMECA) have been applied to the available measurement data sets of the test wind farm Gökçedağ WPP to analyse the occurring failure modes of the main wind turbine’s components. Lists of criticality ranking have been derived for each component and as an overview for all occurring failure modes. Furthermore, causes for underperformances and their effects have been analysed considering also wake interactions within a wind farm cluster. Mechanical and lubrication failures are present in most of the subsystems under study, such as the generator, the pitch, the yaw and the gearbox.
D4.2 Decision Support System for O&M at wind farms	The second deliverable of the WP4, D4.2 Decision Support System for O&M at wind farms, include the main results obtained in the definition and development of a decision support system for O&M in wind farms. The main results have all been produced in the execution of the task T4.2 Decision support system for O&M, and they have made possible that the milestone MS4.2 Decision support system for O&M was successfully met. The document begins with a high-level exposition of the main findings after an analysis of the state of the art regarding O&M in renewable generation and maintenance in wind farms using big data and machine learning as empowering tools. After that, an exposition of how criticality for different failure modes and problems in wind farms in the context of the project SmartWind is presented. The main metrics, KPIs and magnitudes to consider in the project are exposed, and their utility and viability will be evaluated in the validation process of the project. Then, the recommendation engine for O&M is described in detail, including the integration of the system with algorithms and the rest of elements of the project, as well as the expected coordination with other algorithms and environments according to how the different members of the consortium integrate the developments of the project. Finally, a brief exposition on how feedback is going to be gathered from O&M personnel and how that feedback will be used for the fine-tuning of the recommendations according to their accuracy, usefulness, and how criticality is assigned to different subsystems.
D5.2 Report containing the implementation of machine learning techniques – first prototype	This document includes the machine learning implementations for the SMART-WIND use case. In this document, the processing of data collected from wind turbines and the machine learning modelling phase are explained. Also, this document includes the visualization of all analysis and results. In addition, at the last stage, a prototype was modelled with the data collected from wind turbines and the results of this prototype were presented. As a result, this document includes the modelling of the prototype and evaluation with data collected from wind turbines.
D5.3 Software for the implementation of machine learning techniques – first prototype	This document includes a high-level description of the first prototype for the SmartWind system, as well as the main results obtained in this process. The results here included are simple visualizations and representations of the algorithms and solutions that have been developed in the context of the WP5 of the project in the second year of its development. This first prototype developed in the second year of the project is a preliminary approximation to the solution that will be generated at the end of the project. This implies that the final results will include modifications and improvements in comparison with the product generated this year. The document includes an exposition of the software generated by each of the members of the consortium in this first prototype. The software that this document makes reference to is available by request to the responsible member of the consortium, in case it was necessary for reporting matters by any competent agency or organization.
D6.2 Dissemination Plan	The objective of the deliverable is to outline the strategy for dissemination activities carried out during the project. The planned standardisation efforts will foster a further industrialisation of the SmartWind results and create a solid base for investment and commitment of the SmartWind partners.
D6.3 Exploitation Plan Draft	The objective of exploitation and dissemination in SmartWind is to best utilise the results of the project and also to create awareness about the project’s research and development pipeline: from scientific results over technology advancements to success stories in industrial use cases. This deliverable includes the main results obtained in spreading the project results.