资源描述
Predictive maintenance From data collection to value creationJuly 2018Management summaryA TURNING POINT IN INDUSTRIAL SERVICESPredictive maintenance clearly marks a turning point in the world of industrial services. Unlike previous ap-proaches, such as reactive service models, preventive maintenance, and condition-based maintenance, pre-dictive maintenance adds a critical edge to the use of sensors and measuring machine data: It applies algo-rithms to predict the best point in time for carrying out maintenance before the actual occurrence happens or parameters drastically change. As such, operations can be adjusted to achieve the best overall asset perfor-mance and cost. nullis gives companies an opportunity to fundamentally transform their service, and conse-nulluently the whole underlying business model of prod-ucts and services.WORLDWIDE MARKET EXPANSION: USD 11 BILLION IN 2022nulle new approach of predictive maintenance has al -ready established itself firmly in the nullropean industri -al world. nullme null percent of firms report that they are currently devoting time and resources to the topic. Around the same number believe it will lead to strong growth of their service business in the future, replacing a significant amount of hardware product sales reve-nues. nulle predict that the worldwide market will enulland to around Unull null billion by nullnull.ENSURE FUTURE COMPETITIVENESSnulle drivers for predictive maintenance are already well developed in the areas of sensor technology, data and signal processing, and condition monitoring and diag-nosis. nullre, we enullect to see the core technology en -abling predictive maintenance up and running within the nenull three years. nulle real challenge lies elsewhere: in the areas of predictive ability, process and decision support, and the resulting opportunities in service and business models. nullis is where companies need to take action now.TECHNICAL AND CULTURAL CHANGE: CHALLENGE AHEADIn this article, we look at how companies can transform themselves from believers in technology and mere col-lectors of data into service-oriented partners in custom-er value creation. nulle enullamine the implications of pre-dictive maintenance and how they are currently viewed by the business world. nulle elaborate in detail various factors driving the further development of the new ap-proach, using our specially developed nulloland nullrger Predictive nullintenance nulldar. And we identify the changes that companies need to make nullboth technical but much more so cultural nullin order to provide manull -mum benefit to their clients and ensure future compet-itiveness.2 Roland Berger Focus Predictive maintenanceContents1. The future of service predictive, not reactive . 4More than just a technology.2. Growth, not cannibalism . 6The view of the business world.3. The Roland Berger Predictive Maintenance Radar . 7Mindset changes and business model disruptions: The real challenge ahead.4. Looking beyond . 10From data collector to value creation partner. Coverphoto: Petrovich9/iStockphotoPredictive maintenance Roland Berger Focus 3nulle sale of a product marks the beginning of a business relationship nulla relationship that only becomes truly prof -itable through the service relationship that follows. Ideal-ly, that relationship lasts throughout the product and cus-tomer lifecycle. nullrvice is thus not just something that you are obliged to offer customers after you sell them something: It is an essential part of a profitable, long-term business model. Predictive maintenance nulldnull nullas opposed to routine enullpost or preventive maintenance null offers companies the chance to fundamentally transform their service and business model. nullor that to happen, they must start seeing Pdnull not just as a means of collecting data, but as a vital tool for creating additional value in an active partnership with their customers. Pdnullcombines the topics of service and digitalinulltion and opens up sig -nificant new value pockets. nullt to turn this immense the -oretical opportunity into solid reality, companies are obliged also to meet certain conditions. Above all, they need to understand that Pdnull as a form of nullervices nullnull, is far more than just a nulluestion of Inull. It calls for the trans-formation of the whole organinulltion. Pdnullis about creat -ing customer benefits right along the whole value chain nulla mammoth task that tests not just the technical skills of a company, but rather its entire digital mindset. null course, plant, machinery, components and the like will still renulluire physical services in the future. nullou can-not repair a real product with nulleros and onesnull. nullt Pdnullis not about digitalinulltion for digitalinulltionnull sake: It is no sterile task to be undertaken just because it is tech-nologically feasible. It is about transforming yourself from a hardware seller into a service-driven output pro-vider across the product lifecycle.Pdnull has been a hot topic in many industries and areas of application for some time already. nullompanies are particularly enullcited about its promise of major cost sav-ings. nullat promise is particularly enticing as, on average across industries, around null percent1of total operating costs for plant and machinery relate to services. Pdnull can cut those costs substantially, for customers and pro-viders alike. nullt costs are only the first of the benefits.CORNERSTONES OF DIGITALIZATIONPdnullbuilds on the four cornerstones of digitalinulltion: interconnectivity, digital data, automation, and value creation. null continuously measuring data provided by sensors and evaluating the relevant parameters, it pre-dicts the remaining performance life of components, machines, and whole asset parks. nullis can help compa -nies determine the best point in time for maintenance and adjust to optimal operating conditions, thereby im-proving product and service nulluality and getting the most out of their investments. AEVOLUTION OF THE SERVICE MODEL: FROM REACTIVE TO PROACTIVEPdnullnow marks a clear turning point in the world of servi- ces. It represents the final stage in the evolution of service models. In the past, companies would only take action when a problem or defect occurred or pre-defined opera-tional parameters got beyond limits nullan essentially reac -tive approach. nullth Pdnull they now can act on the basis of time and content forecasts nullcreating proactivity. nullther than servicing plants and machinery at statistically pre-defined, regular time intervals, or assessing the health of a system on the basis of physical data nullibrations, tempera -ture, resistance, and so onnulland only taking action if the figures deviate from certain norms, Pdnullactually predicts future trends in the enulluipmentnull operating parameters and thereby accurately determines its remaining operat-ing life. nullis triggers a profound change in the mainte -nance strategy and service business model, for both indi-vidual products as well as networked production systems.1 nullsed on nulloland nullrger project enullerience and focus interviews on predictive maintenance.1. The future of service predictive, not reactive More than just a technology.4 Roland Berger Focus Predictive maintenanceA: From interconnectivity to value creation.The four cornerstones of digitalization.1Interconnectivity between physical products leads to 2. new sources of data, creating transparency about the condition of machinery 3. and enabling the increased automation of processes 4 and new possibilities for creating value in servicesSource: Roland BergerPREDICTIVE MAINTENANCECollecting, preparing and analyzing data to improve predictions and decision- makingAutonomous and self-learning machines to avoid permanent damageImmediate remote access to maintenance and repair points2Digital data3Automation4Direct value creationValue chains connected via mobile or fixed networks1Inter- connectivityPredictive maintenance Roland Berger Focus 5Pdnullhas established itself as an important industry trend in the global industrial and manufacturing indus-tries. A study by nulloland nullrger, nullnull and nullutsche nullsse Anullreveals that null percent of companies are cur -rently devoting time and resources to this topic, while null percent already believe that mastering Pdnullwill be particularly important for future business.2nullompanies are also intensely discussing the possible financial impact of Pdnull nullre, their hopes that Pdnullwill stimulate clear growth outweigh their fears that it will cannibalinull enullsting operations in sales and service. In -Source: “Market research future“, IoT Analytics, Roland BergerCAGR2016-202220161.51.520172.20.21.920183.10.62.420194.31.23.120206.02.13.920218.13.25.0202211.04.6+39%Optimistic prediction6.3+27%Base predictiondeed, as many as null percent of respondents were enullect -ing Pdnullto lead to strong growth of the service business in the future. nullobally, we enullect the market for Pdnullto grow by null to null percent a year across all industries and applications.3nullat figure includes all forms of Pdnull from services to components, contracts, consulting services, Inullarchitec -ture, and software. nullch of the growth will be com -mencing from nullrope, where some Pdnullsolutions have already hit the market or are at an advanced stage of de-velopment. B B: Predictive maintenance.Forecast global market development to 2022 USD billion.2 nullredictive maintenance nullnullrvicing tomorrow nulland where we are really at todaynull nulloland nullrger, nullnull, nullutsche nullsse Anullnullpril nullnullnull3nullarket research futurenull IonullAnalytics, nulloland nullrger2. Growth, not cannibalism The view of the business world.6 Roland Berger Focus Predictive maintenanceIn order to realinull this enullciting growth opportunity, firms need to devise a comprehensive Pdnullstrategy. nulloland nullrger has developed a handy tool to help them with this task. nulle Predictive nullintenance nulldar charts future trends and developments in the period nullnull to nullnull along sinulldimensions: four technologically-driven dimen -sions nullsensor technology, data and signal processing, condition monitoring and diagnosis, and predictive abil-itynull and two business-driven dimensions nullrocess and de -cision support, and the servicenullusiness modelnull CC: The Roland Berger Predictive Maintenance Radar.Trends and developments 2018-2023.Source: Roland Berger Relevant developments and trends Essential developments and trends areas where companies must develop expertiseMarketplace solution: platform for other applicationsSensors for ultra-high robustnessEnterprise total asset management/TCOFull operations outsourcingMulti-partner managementFully automatic service workflowOptimized warehouse & supply chain (e.g. produce-to-order)Digital designs and local 3D spare part printingService cost optimization (continuous)Optimized asset lifecycle managementTrend analysesInternal product optimizationAutomated service activitiesContinuous improvement processExperience models (for x% passed abc)Work instructions/deployment/materials for mobile teamsAutomatic work instructions & materials for service teamsTotal asset decisionsClosed-loop quality managementCorrelation analysisPattern recognition (single-state)Pattern recognition (own fleet)Machine learning/ artificial intelligenceComputerized maintenance monitoring systemClient-specific planning of maintenance activitiesWork instructions & materials for local maintenancePersonalized software/algorithmsAutomated domain know-howPattern recognition & prediction (process/ecosystem)Software robotsRisk protectionUptime guarantees Pay-per-X modelsUse of autonomous drones for (thermographic, etc.) inspectionsInnovative sensors (design, installation, signals recorded)Intelligent sensors for unstructured dataAdvanced non-destructive testing (e.g. ultrasonic)Intelligent sensors for structured dataFocus on trainingFully automatic image analysisIntegration of external data sources (e.g. weather)Selective visualization of deviationsShort-range, low-power transmissions (e.g. NFC)Fully automated root-cause analysisData storage/ clouds, etc.BlockchainSensor fusion5GTrainingIntegration of production, process and ecosystem dataLocal sensor analytics/ in-memory computingDiagnostics fusion (comparison of diagnoses)Integration of customer decision parametersData structuring/automated index selection/machine learning/AICustomized UI/ visualization Global remote data accessBrownfield dongles, etc.Diagnostic serversReal-time data analysis/Big DataSupply chain integrationEdge predictionReal-time data/image prognosisEdge computing/ analyticsPlug-and-play solutions20232018A Sensor technologyC Condition monitoring and diagnosisD Predictive abilityE Process and decision supportF Service/ business modelB Data and signal processingSpare part logistics management3. The Roland Berger Predictive Maintenance Radar Mindset changes and business model disruptions: The real challenge ahead.Predictive maintenance Roland Berger Focus 7The six dimensions of the Predictive Maintenance Radar From purely technological dimensions to the holy grail of PdMA Sensor technologySensor technology forms the technological basis for PdM. The core competency lies in the ability to detect a multitude of signals quickly, accurately and in a targeted, reliable fashion. Many companies are busy retrofitting sensor technology to their global installed base to access data that they were previ-ously unable to collectpanies implementing PdM need to ask themselves what type of sensor technology they want to employ and whether they should operate it themselves or rely on the services of a third party. They will also need to decide how widely to deploy sensor technology and what data it should actually collect.B Data and signal processingOnce a signal has been detected, data is generally collected and stored on local hardware or somewhere in the cloud. With virtually no limits on the amount of data available to-day, firms must make sure to structure and index the data logically so it is immediately accessible for analytics, diag-nostics and forecastspanies need to decide where to store their data, and how to structure and distribute it. The more effectively they do this, the faster and more impactful the resulting analyses will be.C Condition monitoring and diagnosisThe data collected from the machinery indicate its current condition. This
展开阅读全文