Subcategories

  • pparrend

    Author: Pierre Collet

    Presentation: Complex systems Prezi - Pierre Collet
    Video: Nature-inspired systems (and algorithms) for the Industry - Video

    Abstract

    Before electronic computers existed, all computation was done by hand. For instance, all the strength and structure resistance computations necessary to design the Eiffel tower were computed by hand. When computers were invented, in the second half of the XXth century, they were thought of machines that considerably accelerated the computations previously done by hand, but not more.

    On the other hand, natural and biological processes have evolved as an answer to the second law of thermodynamics, as interestingly enough, order can be more dissipative than disorder. Therefore, Nature kept popping up with new innovative dissipating structures, emerged from this irrepressible need to better exchange energy. Now, if we can instill this natural quest into computers, they will become as creative as Nature, and will come up with the best possible solutions to solve problems we give them to solve.

    Before 2000, the task of finding better solutions to industrial problems was falling upon engineers. Now that natural emergent processes are better understood thanks to the new science of Complex Systems, it is possible to insert them into computers and have them find out the best ways to solve the industrial problem.

    Since 2000, computers are capable to routinely find better solutions to industrial problems than human engineers. The new trend in industry should therefore be to insert more nature-inspired systems (and algorithms) that can be more efficient in finding efficient solutions than human engineers.

  • pparrend

    Authors: José Gramdi, Jean Vieille
    Keywords: performance, systemic approach, costs, value added, ongoing improvement, real-time management

    Abstract:

    A new world is emerging and in this new world, our old good management and administration
    rules don't work anymore. Agility is now the survival condition: we need to be able to deliver the right product, at the right time, with the right price and with the exigency of maximizing profit. The golden industry period running from 1945 to 1975 and designed as “les Trente Glorieuses” by Jean Fourastier has left a persistent legacy to industry managers: quest for full occupation of resources, mass production, organizations splitting up with obsession of local performances, producing, purchasing and delivering according to economic batch size… All those methods that were pretty good when we were assured that every part produced will be sold in a short delay, are fully obsolete now that we have to produce only what has been sold. The statement is cruel: our production systems have never been so advanced from both hardware and software point of view, we are nevertheless swamped over by inventories, work-in-progress and… data! Aware of this problem, companies are now moving toward approaches like TOC, Lean Management or Six Sigma with the hope to find a solution. This movement may seem promising, yet we unfortunately note many blunders and misinterpretation regarding the management of these approaches, the deployment of their attached tools, the setting up of their goals and the measurement of their benefits. The root cause to this situation can be attributed to the incredible persistence of well-established beliefs known as local optima's. It is no doubt that this behaviour is the most remaining legacy of the "Trente Glorieuses".

    But the world has changed and new management rules are to be imagined. From a systemic approach and pattern of industrial company, we suggest a generic expression of Overall Interactionnal Performance including operating expenses, raw value added, lead-time and customers satisfaction. We then recommend an ongoing improvement loop focused on this KPI and based on TOC, Lean Management and Six Sigma tools. This movement embraces the five value-added processes of the company : sell, design, purchase, produce, deliver. If one and only one of these processes is down no more value added is generated by the company. These five processes are simply described by their four main characteristics: actual lead-time, nominal capacity, quality level and involved operating expenses. Then we try to establish the existing relation between these characteristics and our global KPI. In this way of proceeding we can highlight the bottleneck of the company. This bottleneck may reside in any of the five value-added processes and can be a quality bottleneck that we will eliminate using Six Sigma, a speed bottleneck that we will eliminate using Lean Management, or a throughput bottleneck that we will eliminate using TOC. The new bottleneck of the company now resides in another characteristic of another (or the same) process that we are going to locate in the same way. The virtuous ongoing improvement loop is launched, no more local indicator is needed anymore, and every employee can simply understand the movement and get involved in it : more, better, faster with the same people !

  • pparrend

    Authors: Philippe Pitolli, Pierre David
    Slides: 9.PPittoli_casan_light.pdf
    Video: CASAN : A New Communication Architecture for the Factory of the Future - Video

    Abstract

    Sensor networks have long been an active research field. However, deployments are still often specialized for one particular task while emerging commercial products use proprietary low-end technologies.
    Factory builders are thus faced with multiple and incompatible solutions from vendors.

    The new CoAP protocol, designed for resource constrained devices and standardized by the IETF may bring a partial answer to this divergence. However, complexity of needed protocol stacks may limit real deployments.

    We propose a new, open and pragmatic distributed architecture for sensor and actuator networks, based on CoAP, whose goal is to further reduce constraints on devices. By shifting intelligence to a master node and keeping only what is really needed on sensors, it is possible to reduce software complexity on sensors, while keeping a secure, reliable and scalable system providing sensor resources to the global Internet.

    This architecture is not tied to a particular layer 2 technology and thus can be used on wireless networks as well as classical wired networks for noisy environments such as the factory.



4P-Factory Workshop 2016

Internal error.

Oops! Looks like something went wrong!