DISRUPT SOLUTION: THE ARCHITECTURE
DISRUPT aims to provide a smart decision support and analytics platform that will address the lack of intelligence in the management of huge data volumes, which are being collected from automated plant floors, system-enabled manufacturing environments, collaborative smart supply chain networks, and connected IoT infrastructures. More specifically, the proposed solution is an integrated platform for supporting decision making, which harness event-based observations and data coming from physical and digital entities in the factory plant floor, the manufacturing systems and the connected supply chain networks. This platform orchestrates the integration of services that facilitate modelling, multi-level simulation, optimization and analytics for the implementation of business cases that aim to improve production quality, efficiency and throughput. DISRUPT primarily addresses the manufacturing sector, giving emphasis on the automotive and consumer electronics industries. However, the solution also relates to the needs of further manufacturing, like pharmaceutical companies, food and beverage production departments, etc., which require modern ICT to transit to the new digital transformation era.
This market is still growing, as individual vendors of existing enterprise information systems, like ERP and MRP, may offer similar capabilities, but to a limited extend, while the integration of relevant mechanisms is not yet mature. The respective market surveys foresee a 9.70% increase in CAGR for the global smart factory market by 2022 (reaching an estimated expenditure rate of 74,5b$ globally), which give an undoubtable opportunity for the DISRUPT offering to reach a significant market share in the next years. Through DISRUPT, we aim to advance the ability of the manufacturing sector in improving their operational efficiency and productivity and managing the decision processes that would support intelligence at the business level. The platform offers a data driven implementation of user scenarios that enhance awareness on what is happening within and across a factory environment, through the integration of big data. It employs state-of-the-art software solutions for observing the evolution of key performance metrics and estimating their impact on the production and manufacturing capabilities. Further to it, it incorporates a set of decision support services that assist key players in handling disruptions in the production processes and improving their capacity planning by exploiting innovative modelling, simulation and optimisation technologies
Complex Event Processing (CEP) Module
The core architecture in DISRUPT follows best-practice from industrial solutions based on IoT architectures, while considering core concepts of the lambda-architecture (with its Speed- and Batch-Layer paradigm), a unified messaging layer that facilitates connectivity between the “outer world” and the DISRUPT platform, and a consistent alignment with principles of an Event-Driven Architecture (EDA) approach. The use of Complex Event Processing (CEP) technology is a relatively new field which gained more and more popularity with the last decade. Forrester Research describes it as a “Software that can filter, aggregate, enrich, and analyse a high throughput of data from multiple, disparate live data sources and in any data format to identify simple and complex patterns to provide applications with context to detect opportune situations, automate immediate actions, and dynamically adapt. CEP technology is sometimes referred as a typical database model turned upside down: Where databases typically stored data and runs queries against the data the CEP stores queries and runs data against the queries. Streaming analytics is neither a scheduling technology nor a decision tool but a re-active solution which always needs an external trigger. Respective rules are hereby defined at design time in EPL (c.f. subsection on Apama’s EPL), an SQL-like dialect. It doesn’t allow for a perpetual analysis and executes observations in a defined and well-prescribed time-window (e.g. last five seconds, last 10,000 observations, etc.).
![]()
FCA Applications: Supply Chain Monitoring and Truck Location
The Supply Chain Monitoring scenario from CRF monitors truck movements from different suppliers to a central warehouse where containers are reordered according to the actual production plan and eventually shipped to the plant. Truck movements are based on real, recorded data along the checkpoints of the respective routes. Once at the warehouse, the truck monitoring is continued until the plant is reached. For the reloading at the warehouse, the minimum time taken from historic data is taken. The important key figure of interest in this scenario - which is continuously updated - is the estimated time of arrival (ETA) which is continuously checked against the required time of arrival (RTA) directly taken from the production schedule. Whenever ETA is later than RTA an alarm is issued by the CEP to inform the KPI dashboard and CloudBoard about this incident. In addition to the historic data, information from actual and most recent traffic is taken into the equation to provoke delays as well as current weather conditions in terms of weather warnings. The CEP thereby produces 2 kinds of external events on the messaging bus: an continuously updated stream of information which is consumed by the KPI Dashboard as well as Traffic Alarms events only generated in case of serious issues which are consumed by the Cloud Board (c.f. Figure 3 below). The CEP makes use of external services for traffic information and weather conditions (c.f. respective subsections below) to ensure real-time operation on most recent data.
![]()
Traffic Information Adapter
To check the current traffic situation the HERE routing API is used which returns the normal duration of the route and also the calculated time assuming current traffic conditions. Out of the difference of these two values we get the expected delay. This delay gets added on the mean duration of the past Truck movements on the route. The traffic and rerouting information is taken from a real-time traffic service. It provides precise instructions to a destination using various transport modes (e.g., car, truck, public transit, bicycle) and leveraging different algorithms (e.g., matrix, isoline routing). The following code sample shows a REST request on the HERE API for the fastest truck routing from GPS coordinate 52.5,13.4 to 52.5,13.45 while considering actual traffic conditions.
Cyber-Physical System (CPS) Module
In the last years, there is a rapid advance in multidisciplinary systems which have been emerged from the development of the ICT sector such as cloud computing, internet of things (IoT), sensor networks (SN), social networks computing, and Cyber Physical systems (CPS). For this reason, CPS will particularly present a new generation of systems that will integrate the communication models and different computing capabilities with the physical and engineering systems. The CPS paradigm is composed of physical and cyber objects (software, hardware, sensing devices, computational applications) for monitoring, interaction, manipulation, control of the physical world. CPS represents a new generation of digital system, which consists of two main functional components: (1) the advanced connectivity that ensures real-time data acquisition from the physical world and information feedback from the cyber space; (2) intelligent data management and computational capability that constructs the cyber space.
The core of the CPS concept and module developed for DISRUPT embraces this latter aspect, focusing on the monitoring (and control at later extent) of physical processes and tries to extend it by providing computational cyber mechanisms to predict future effects of the (mutual interaction of) ongoing physical processes, based on existing and past process evidences. CPS module, continuously fed by data coming from diverse information sources, typically real time data from shopfloor, manufacturing cells and machines, identifies, rebuilds and update the digital image of the complex ongoing process, which is complex, very articulated and to some extent not explicit, and use it to detect and predict incoming disrupting evidences from shopfloor. Outputs of such mechanisms are the events that are raised to DISRUPT platform, which will result – as utmost consequences of DISRUPT tools’ decisions – into CPS control actions.
Test line machine and performance monitoring-Arcelik Case
The Test line monitoring scenario from Arcelik addresses the monitoring of different test lines in Arcelik production plant, aimed at performing several functional and operational tests on incoming Printed Circuit Boards (named cards). Each line is composed of a sequence of different automatic test machines, each performing a specific test type and endowed with variable features and fixtures to the purpose, so as to process a diversified set of products. The product portfolio as input of all the test lines is very wide and the consequent incoming batches, their distribution over the various test lines and machines, always changing. As mentioned, in performing the specific functional/operational family of tests, each machine can specifically respond either by positively or negatively certifying the performance of each card (test passed or failed), or can end up in an inconsistent state (terminated test), depending on circuital/contingent situation. The important key aspects of interest in this scenario - which are continuously updated by CPS module – are the dynamic evolution and related monitoring of the card specific estimated test time (ETT) and the machine specific testing performance (MTP). Both of these parameters are continuously processed by CPS module based on models which consider and harmonize several raw information coming live from shopfloor testing machines, so as to compute the card specific ETT evolution along time and across machines, as well as the card-independent MTP evolution along time (and therefore across cards).
![]()
Different families of events are the result of the complex identification and runtime computation: CPS module raises event at card level, evidencing - based on parameterized warning and alarm thresholds – whether the ETT is dynamically evolving toward potentially critic (in the case of warning) or highly critic conditions and estimating when this will happen. On the other hand, it evidences if a MTP is drifting and when this will be impacting, based on the same logic in terms of events. In both cases, the consequences of the conditions detected by various events would strongly affect the performance of the test lines, impacting on the overall balance and expected through put of each line. Events are therefore raised to the DISRUPT platform, where they are further processed and simulation/ optimization based assessment of potential consequences of these happenings take place.
Digital Platforms
The output of DISRUPT will be a digital cloud platform which is able to provide different kind of services to users to create organisation's ecosystem, enable collaboration and rapid application of the best decision to face the incoming disruptions. Nowadays, considering the impressive increase of the digital solutions and the implementation of the digital technologies, it is interesting to understand the value of the digital economy and the future changes in value creation. It is estimated (Accenture, 2016) that the digital economy will account for 25% of the world’s entire economy by 2020, up from 15 percent in 2005. Moreover, the top 15 public ‘platform’ companies already represent $2.6 trillion in market capitalization worldwide, and they’re attracting more and more capital investments through the value creating power of their platform ecosystems and digital assets. This is changing also the characteristics of value chain of the business model: from the traditional and linear value chain we will pass to a circular, two-way and continuous value creation where the platforms substitute the products and growth is driven by the network effect with the aim to optimize the ecosystem. The network effect exists when two user groups generate network value for each other, resulting in mutual benefits that drive demand-side economies of scale.
![]()
A typical example of this kind of platform driven business model is the iOS App Store. It includes an ecosystem of nearly 380,000 developers that have created 1.5 million applications have generated $33 billion in sales by the end of Apple’s fiscal year 2015. (Accenture, “Platform Economy: Technology-driven business model innovation from the outside in”, 2016)
News & Events
![]()
DISRUPT will participate to the 9th IFAC Conference MIM (Manufacturing Modelling, Management and Control) presenting two papers. The first one “On Event-Aware Manufacturing systems” is written by AUEB, the second paper is presented by AUEB and SIMPLAN: “Rescheduling and co-simulation of a multi-period multi-model assembly line with material availability restrictions”. The mission for the 9th MIM is to investigate Connect Control, Industrial Engineering, and Computer Science for better management decision-support systems in digital, resilient and sustainable manufacturing and supply chains in the era of Industry 4.0. The conference will be held in Berlin on 28-30 August 2019 (https://blog.hwr-berlin.de/mim2019/).
Watch the DISRUPT video
It is now possible to watch the video presentation of the project with all the general information and the main characteristics of each architecture module and industrial cases.