21 June 2011
Service Oriented Architecture (SOA) has made great strides in the world of IT. It has been mostly applied to the design of integrated MES solutions where the individual MES applications were integrated together using middleware software. While the principles of SOA can be made good use of in such integration scenarios, it is not only applicable to Enterprise Application Integration (EAI). Individual applications are now starting to be designed or redesigned, based on the same principles of SOA which will ultimately make the new generation of MES applications more flexible, easier to deploy and richer in user experience. To explain this in more detail:* Applications designed according to SOA principles are more flexible. They will be designed to be extendable by project teams without losing the levels of support often associated with customised applications. The developers can determine how open the system is and which parts will be free to enhance in terms of site specific business rules.* Applications designed according to SOA principles are also more flexible in terms of scaling and redundancy as each service is autonomous and stateless. Applications can be hosted on-site or in data centers.* The user experience becomes richer on thin clients and web based platforms through asynchronous processing. This is mostly a side effect of the ability to effectively construct asynchronous services.Business process managementBusiness Process Management (BPM) is about three things – business (organisation, goals and objectives), processes (methods, practices and procedures) and management (how processes and the business are directed). BPM is a process-centric approach and builds on older methodologies such as Lean and Six Sigma. Middleware defines the technology to model and improve the business processes, whereas SOA allows the processes to be implemented as decomposed processes that follow a defined set of principles. Traditionally, individual applications were integrated through Enterprise Service Bus (ESB) technology. Today applications themselves are being designed based on SOA principles. If applications can already talk to each other using standard interface mechanisms, a dedicated interface may no longer be required.A typical BPM project would follow the following phases in order to establish an optimized MES workflow:* Discovery phase: interviews are conducted with the various stakeholders to establish an AS-IS baseline. The process maps are identified in this phase. * Storyboard phase: the TO-BE model is construed – this does not have to be an ideal optimal state, but a next logical step based on various implementation constraints. * Design phase: the workflows are developed describing activities, events and exceptions and each interface is described in detail. * Development phase: the development of the data model and in particular the master data model is done in this phase.* Simulation: in a MES approach this phase consists of the integrated FAT (IFAT) and integration SAT (ISAT).* Deployment & execution: the actual deployment on a scalable infrastructure is managed.* Monitoring and optimisation: now the business is in control to monitor the deployed solution through dashboards using KPI’s and various metrics. The solution should be flexible enough to further optimise on a continuous basis.Data modelingData modeling is performed at different levels of detail from the storyboard phase up to the development phase. In this activity the relevant entities used in the MES are described together with their relationships to each other. A planning system might know about sales contracts, nominations, production, shutdown and inventory, all of which will be described as entities in a data model. Object Relational Mapping (ORM) techniques will be applied, which describes the entities in terms of classes as is the case in Object Oriented Programming (OOP). Master data, The reference data shared by several applications – Master Data – always requires a careful approach as the MES layer can consist of applications from different vendors that do not easily share a common data model. Although several approaches exist to assign a system of entry and a system of record, it is quite common that the ERP is chosen as the master for resources like materials, contracts and equipment and the Historian forms the basis for tag names. A good data model should anticipate aggregations that the end user is likely to perform, such as aggregations across material, across the asset model and across time. Framework based extensionsSoftware Frameworks are an abstraction of standard programming. They use templates, which leads to a higher focus on business rules and enforces the programmers to follow certain design patterns. Several Frameworks are available to implement SOA based applications and when combined with built-in MES knowledge, a powerful framework allows for the development of new MES applications. Examples of MES knowledge that can be added on top of a typical software framework are:* Data modeling using database entities defined in terms of an asset structure, material movements and a temporal dimension to allow the framework to more easily perform aggregations that are used in MES applications. Often the layer under the MES is defined by the temporal dimension, whereas the layers on top of the MES are defined by resources and material.* Implementation of business logic through services that provide Representational State Transfer (REST). When REST is used, all information about the current application state is transferred between the client and the server, allowing a scale-out of multiple servers, or running applications on central data centers or even in the Cloud on the Internet. The Business Logic layer framework extensions will allow use of standard MES security framework and has access to federated data sources, such as OPC.* The MES framework should also offer a rich user experience for new and redesigned applications while offering a common user interface using web based technology. This will allow the user experience to be similar on both a standard PC as well as any other device such as smart phones with a similar look-and-feel as the entire MES solution.Semantic dataFor most applications in the MES relational data models work perfectly. However for some data the more natural structure to store data is in a graph database, in particular in RDF (Resource Description Framework) format. Here we can describe attributes about nodes using triplets. Take, for example, an asset in the plant. As a RDF triplet this asset is described as an item with a name, attribute and value – such as Pump-101, Pressure, 01PI01.PV. In a similar way we can define several statements about a pump, such as its physical location (Pump-101, Location, Area-A) and location in the plant hierarchy (Pump-101, Unit, U101) in the form of RDF triplets. RDF query languages such as SPARQL enable the application to quickly traverse through the graph database and finding relationships which are not always straightforward to map in a relational structured database. Obviously not all information is best described in a RDF database. However for large amounts of data from diverse systems, a RDF can be a good approach to describe the various information we posses about an object for which a matrix representation would not be suitable. MES in the CloudCloud is a style of computing where scalable and elastic IT-related capabilities are provided as a service to external customers using Internet technologies. For Cloud computing to really take off the technologies described above are needed, including Rich Internet Applications and SOA based services. However, there will always be concern around security, in particular when the cloud is external to the organisation. It would be unimaginable that companies would allow the master data to reside outside their organisation. However, what is imaginable is that certain applications are integrated in the MES, but hosted outside the company. Examples of these include non-critical applications being generated based on a feeder application, or a scenario where an application is tried and tested in a pilot hosted by an external vendor before the customer decides to host the application in a full-scale application in its data center.ConclusionRecent trends have been described which allow for a new generation of integrated MES solutions. For MES vendors these are certainly exciting times and require a shift away from the traditional implementation of single applications. By utilizing SOA down to the level of the applications in the form of service component architecture, graphical configuration tools and more intuitive data models, the end user will have a more fit-for-purpose MES solution that at the same time is flexible for future adaption of business processes or further integration.
Print this page | E-mail this page
This isn't a paywall. It's a Freewall. We don't want to get in the way of what you came here for, so this will only take a few seconds.
Register Now