| Copyright 2014 Simio LLC | All rights reserved. 1 Executing Simulation Experiments in the Cloud C. Dennis Pegden, CEO Simio LLC
| Copyright 2014 Simio LLC | All rights reserved. 2 Cloud computing Avoid infrastructure costs. Focus on value added activities. Shorten development time. Less maintenance, easier to manage. Scale resources to changing demand.
| Copyright 2014 Simio LLC | All rights reserved. 3 Cloud Manufacturing Cloud-based manufacturing paradigm based on Internet of Things (IoT) and virtualized/service oriented technologies. Encompasses the life cycle of a product – design, simulation, production, test, maintenance. Cloud-based simulation for both facility design and production planning/scheduling.
| Copyright 2014 Simio LLC | All rights reserved. 4 Cloud Computing Drivers 1. Mobile and shared information. 2. Scalable demand. 3. Scalable (parallel) computation. Most applications leverage 1 and 2. Simulation applications leverage all three.
| Copyright 2014 Simio LLC | All rights reserved. 5 Facility Design Allow 3D animated simulation models to be built and run in the cloud for improving system design. Model building Data integration Animation Experimentation
| Copyright 2014 Simio LLC | All rights reserved. 6 Example: Facility Design
| Copyright 2014 Simio LLC | All rights reserved. 7 Experimentation Models are used to compare alternative designs, or optimize design parameters. Randomness requires that each scenario is replicated. During experimentation animation is not required. Multiple processes allow scenarios and replications to be run in parallel.
| Copyright 2014 Simio LLC | All rights reserved. 8 Cloud Challenges/Opportunities Cloud-based 3D animation environments don’t exist. Model data is dispersed and not easily accessed. Experiments can be executed in parallel. Data interfacing is the highest priority issue limiting the vision.
| Copyright 2014 Simio LLC | All rights reserved. 9 Working within the existing limitations Models are built using desktop software. Data is first integrated into the model – the project (model + data) is then uploaded to the cloud. Experimentation can leverage the full scalable processing power of the cloud (e.g. 25 replications of 10 scenarios simultaneously executed).
| Copyright 2014 Simio LLC | All rights reserved. 10 Planning and Scheduling Allow simulation-based scheduling systems to be executed in the cloud, and the results deployed across the enterprise. Deterministic model used to generate schedule. Interface to ERP/MES data. Evaluate alternative scenarios (expediting jobs, overtime, etc.). Analyze delivery risks by replicating the schedule with uncertainty and unplanned events. Publish the selected plan to mobile devices for execution.
| Copyright 2014 Simio LLC | All rights reserved. 11 Example: Planning and Scheduling
| Copyright 2014 Simio LLC | All rights reserved. 12 Risk Analysis Deterministic plans assume away uncertainty/unplanned events – they provide optimistic results. By replicating the plan with variation/uncertainty added into the model we can estimate schedule risk. Multiple processes allow replications for risk analysis to be executed in parallel.
| Copyright 2014 Simio LLC | All rights reserved. 13 Cloud Challenges/Opportunities Cloud-based modeling environments don’t exist. Model data is dispersed and not easily accessed. Experiments can be executed in parallel, providing quick comparisons of alternative schedules and risk analysis. Data interfacing is the highest priority issue limiting the vision.
| Copyright 2014 Simio LLC | All rights reserved. 14 Working within the existing limitations Models are built using desktop software. Data is first integrated with model – the project (model + data) is then uploaded to the cloud. Experimentation can leverage the full scalable processing power of the cloud (e.g. 25 replications of 10 scenarios simultaneously executed). Risk analysis can leverage the full scalable processing power of the cloud.
| Copyright 2014 Simio LLC | All rights reserved. 15 Top Priority is Data Integration Cloud manufacturing solutions will initially be hybrid environments (e.g. ERP cloud, MES on premise). Solutions must interface to dispersed data – some on premise – some in the cloud. Data integration between cloud and on premise components (e.g. ERP, MES, IoT, custom data sources) needs to be simple and seamless.