siriuscon 2017 - sirius-powered risk modelling and simulation in industry 4.0 supply chains

25
Centre d’Excellence en Technologies de l’ Information et de la Communication www.cetic.be Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains Christophe Ponsard SiriusCON, Paris 2017

Upload: obeo

Post on 22-Jan-2018

103 views

Category:

Technology


2 download

TRANSCRIPT

Page 1: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be Centre d’Excellence en Technologiesdel’Informationet de la Communication

www.cetic.be

Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

Christophe PonsardSiriusCON, Paris 2017

Page 2: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Context – Risks Management in Supply Chains

Page 3: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Example of Procurement Risks

Supplier Ain Seattle Supplier B

in Osaka

Customerin Hamburg

Riskarea

Legend

LA,R1

T1,B,R1

T2,B,R1T2,B,R2

T2,B,R3

LB,R1

T1,A,R1

T1,A,R2

T2,A,R1

T2,B

T1,BT1,A T2,A

Politics

Technology

Eco-nomy

Ecology

Social

Risk of multinational enterprisesaccording to Dunning

Risktypes

Page 4: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

End-to-end Risk Management in Supply Chains

Warehouse risk(Delay)Quality

Production riskDelayQuality

Warehouse risk(Delay)Quality

[Production Risks (internal)]

[Procurement Risks (external)]

[External Risks (external]

Page 5: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

State of practice in SMEs (survey performed in 2015)

https://www.cetic.be/Management-of-Procurement-Risks-on-Manufacturing-Processes

Page 6: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

SimQRi Project Context

• Towards Quantitative Risk Management in Supply Chains

• Main goal:

• Develop a practical tool-supported methodology to help SMEs assessing the risks and mitigating their impact in the production process

• Approach: modelling and simulation toolbox

• Modelling supply chains

• Identifying delay, quantity, bad quality,…

• Expressing them with model queries

• Efficiently simulate alternative designs to reduce risk

http://simqri.cetic.be and http://simqri.com

6

Page 7: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Modelling Supply Chains and Risks

• Suppliers + supply policies

• Storages

• Processes

Page 8: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Queries

• Probes on model element: e.g. relativeContent(storage) totalWaitDuration(process), …

• Composed probes

• Percentage of working time for a processworktime(p) := 100 * (time – totalWaitDuration(p)) / time

• Carbon footprint:CF(order) := upplier(order).dist*CO2_FACTOR_TRUCK_PER_KM

Page 9: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

SimQRi - EMF Meta-model

Page 10: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

SimQRi - Tool Architecture

• Actually a dual tool with a common simulation engine• Web-based • Eclipse-based

• Technology mix:

10

BIRT(reporting)

Page 11: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Web-based Version (« beer game » model)

11

Page 12: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Eclipse Version (Green Supply Chain)

12

Page 13: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

SimQRi - Web-based UINot a Good Experience

Joint.JS selected as JavaScript among some candidatesGlobally did the job, however:• Not configurator-based as Sirius ⇒ JavaScript programming • Need to buy full (non-open source) version or spend time reinventing the

wheel if you want features like UNDO/REDO• Integrating more stuff, e.g. dashboard, was painful• Proved hard to maintain (kept as “simplified tool” in the end)

Comparing pure UI development effort:

With joint-JS With Sirius

First usable prototype 3 months 3 weeks (1 week training)

Full scope Not achieved after 18 months 6 months

Page 14: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Another feedback on learning curve

Page 15: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

SimQRi - Eclipse VersionComplexity of Project Set-up

Page 16: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

SimQRi - Ease of Evolution in Eclipse Green Supply Chains

• Garbage production• special type of output of a process through a "trash" port • connected to a dedicated overflow storage named "Garbage“• total quantity of garbage= Garbage.totalPut()

• Energy Usage:• “energy counter” updated on processed/storages per processed item• can depend on load, e.g. EnergyCounter+=MaxEnergyPerBatch*(1-Load)

• Carbon footprint:• Transportation: OP_A.quantity*Supplier_A.dist*CO2_FACTOR_TRUCK• Production: using specific attributes

success.CF=part_A.CF+part_B.CF+part_C.CF EnergyPerBatch*(1-Load)*CO2_FACTOR

Page 17: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

SimQRi - Ease of Evolution in Eclipse: Simulation of Cloud Computing Elasticity

• total energy (on whole system) = SaaS level energy modelsum(Process.allInstances()->select(p:Process|p.energy*p.processedRequests))

• missed request ratio (on the FRONT process)FRONT.processedRequests/FRONT.incommingRequests

• average response time (on a chain):avg(sum(ProcessChain.allInstances()->select(p:Process | p.runtime)))

Page 18: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

SimQRI - Community Support

• Web-based: • Join.js: good tutorials, forums• Need to buy RAPPID commercial version to get all features

e.g. undo/redo

• Eclipse-based: • Full editing features • Good tutorials, forums• Strong support from Obeo community • Obeo training useful for a quick learning curve• Collaborative work

Page 19: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Integration with the OscaR engine

A library for operational research encompassing:

• State of the art optimization engines: CP, CBLS, MIP, DFO

• Discrete simulation engine

• Visualization

Scala 2.11, Open source LGPL

Release 4.0.0 in September 2016

• Simulation using OscaR.DES (Open Source)• Built on top of a task-resource model• Incremental evaluation throughout the simulation run

i.e. no trace stored, no post-processing• Minimal updates of only the relevant fragment of queries• Bottom-up updates to allow the sharing of sub-queries

• Monte Carlo techniques for aggregating results

• Availability of specific statistical operators (mean, avg, std dev,…)

• Computation of distributions

oscarlib.org

Page 20: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

OscaR integration

• AQL query to extract queries

• Model transformation to translate into OscaR.DES

• Running MCS over DES engine

• Injecting result back into model

• GeneratingBIRT reports

Page 21: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Integration with BIRT for Reporting

• First: tried to setup ecore as data source

• Finally used intermediate XML format

Page 22: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

On-going work: Optimised Risk Mitigation

• Tuning the model in order to minimise costs induced by risks

• Now: simple “explorer” functionality to change the values of a single parameter to find out its optimal value (the other being unchanged) –e.g. optimal ordering threshold

• Current work (PRIMa-q) : use optimisation (constantly) minimizing risks

• Risk robust strategies (scheduling)

• On-line Stochastic Optimisation

• Based on the Oscar.CBLS engine (efficient, scalable, online) oscarlib.org

Page 23: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

Conclusions and Ideas

• Sirius approach• Proved evolvable to other domains• Some learning curve but quickly recovered• Easier to maintain/evolve• Related technologies are quite widely used and well known• Our work also released as open source to SME (and everyone)

GitHub: https://github.com/cetic/SimQRI

• Web version• Too specific and hard to maintain• Code not released (possible licence issues)• But nice showcase, interesting SaaS mode what about a “web-based” Sirius in the future ?

• Text integration• Xtext integration great• Looking into Office suite integration like Open/Libre Office & Microsoft Office

see EclipseCON talk and more on-going work

Page 24: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

QUESTIONS ?

24

Tool homepagehttps://simqri.cetic.be

Open Source repohttps://github.com/cetic/SimQRI

Page 25: SiriusCon 2017 - Sirius-powered Risk Modelling and Simulation in Industry 4.0 Supply Chains

www.cetic.be

linkedin.com/company/cetic

[email protected]

+32 71 159 362

twitter.com/@CETIC twitter.com/@CETIC_be

www.cetic.be

Aéropole de Charleroi-Gosselies Avenue Jean Mermoz 286041 Charleroi - Belgique

Questions?

25