1 August 2005
On the same page
Real-time operations intelligence for plant floor, business.
By Fayez T. Kharbat and George Bauer
Running a refinery, a chemical plant, or a pharmaceutical assembly operation is a complex business. Planning and scheduling, process control, and maintenance require dedicated, multifaceted solutions necessitating highly trained and experienced users. The inevitable system and user specialization can result in the proliferation of disparate data sources, incoherent information, inconsistent decisions, and the failure to realize corporate objectives.
Large organizations are managing their operations using different application classes. Enterprise resource planning (ERP), supply chain management, customer relation management, and others are one class of applications that supports the business and administrative aspects of corporate.
On the other hand, the corporate plant-floor gets support from totally different types of systems, such as distributed control systems (DCS), supervisory control and data acquisition (SCADA), laboratory information management systems (LIMS), and data historians.
Although it is of critical importance for a plant manager to know what is happening throughout his plant or organization, the manager has neither the time nor the inclination to dive into all the systems that manage plant activities. This is why managers are relying on subordinates to provide them with the information required to make decisions. This information is usually composed of business-like data extracted from business applications and process-like data extracted from plant-floor applications. However, the process of extracting valuable information from data repositories and efficiently delivering it to the end-user is complex. This is mainly because of the vast reservoirs of corporate data accumulated over the years and the diversity of storage technologies, database management techniques, and data semantics.
Furthermore, corporate and plant floor systems often can't interact, so users have to jump from one application to another to get the required information. This is why information takes such a long time to prepare and usually is late reaching the manager's office. This information traditionally flows to the plant manager through status reports, verbal updates, turnover meeting notes, phone calls, and e-mails.
As a result, the "latest" information is out-of-date, filtered, and only partially complete. Worse yet, in the event that a problem requires further attention, he has no mechanism to quickly and conveniently find detailed supporting information.
Real-time Operations Intelligence (RtOI) is emerging as a new class of applications that can gather information from a job and present it graphically through a standard Internet browser. On the main screen, summarized information automatically draws the user's attention through highlights of high activity and exceptions. These "actionable items" allow the user to click the mouse and drill down to sequentially unfold more detail in supporting information.
RtOI should deliver a coherent view of the current activity, future plans, and production history to all plant desktops in near real time. RtOI is data source neutral. Data exists in back-end systems, gets collected and buffered in a virtual cache, and is presented to the clients/users in an Internet/Intranet browser or wireless devices, such as PDAs and handheld computers. Data sources accessed include SAP, OSI PI, Oracle databases, various planning systems, LIM systems, and document management systems. Information goes to clients from the RtOI virtual cache with minimized impact on network bandwidth and back-end systems.
RtOI users are anyone in the business who needs tactical information to improve operational performance including management, planners, maintenance, and operations personnel.
Real-time operations intelligence applications gather information pertinent to a particular job and present it graphically through an Internet browser.
Operations intelligence differs dramatically from business intelligence.
Business intelligence provides market information to support strategic decisions. These products slice and dice transactional data from databases and data warehouses where the information is essentially of a single type (transactions held in a database), is static (not real time), and the analysis goes on intermittently and on user demand.
Conversely, operations intelligence helps support tactical decision making by accessing dynamic data of different types from multiple distributed sources in near real time. Information pushes dynamically to a wide audience of subscribing users throughout the enterprise enabling them to make decisions that can improve the bottom line.
Operations intelligence seeks to bridge the gap between the technical plant information systems and the business environments providing a single point of access.
Today, few plant information systems across the process industry have fully exploited technology to deliver real bottom line impact.
In most cases:
Application development often is an end user activity.
There is no real change in the engineer's or manager's business process.
There is very little stewardship of benefits capture; users perceive them as "soft" benefits.
Data is consistent but there is inconsistent analysis by multiple end users.
Inconsistent analysis results in inconsistent information, which leads to inconsistent decision-making.
This is very different from what happens when implementing advanced process control:
The applications are an integral part of the project.
Users define benefits, and the selected applications undergo rigorous justification.
The applications development/implementation plan is clear, and the applications lock in and apply best practices in plant control.
There is a real change in the way technicians work. Could the same approach apply to plant information systems? Rather than advanced process control (APC), this is advanced management control (AMC).
AMC, supported by an operations intelligence solution, would lock in and apply best practices in plant monitoring, asset management, and compliance.
AMC would support a real change in the way plant people work.
A significant difference between APC and AMC is APC is a closed loop solution while AMC relies on the abilities of the users to react to emerging situations and events. In the context of operations intelligence, "Management" refers to anyone in the enterprise including corporate senior management, maintenance engineers, plant operators, planners, and traders. Indeed, an operations intelligence solution provides essential information to support tactical decision making, in [near] real time and confers the greatest benefit when deployed enterprise-wide. On-demand processing only works for a limited number of users owing to the unpredictable loading imposed on the network infrastructure and backend data sources. Deploying a solution enterprise-wide demands a more robust and controlled architecture and platform technology.
Call to action
Ensuring compliance with corporate objectives assumes the user established targets. Setting these corporate objectives frequently involves the determination of key performance metrics (KPI) in order to ensure the business is on target. However, simply the setting and reporting of these key metrics is meaningless and certainly insufficient to ensure success unless the means to drill down (to understand the underlying causes of any deviations) and to manipulate control points is available. Furthermore, these targets have to disseminate across the business to ensure the entire workforce focuses on achieving the corporate objectives so an enterprise-wide capability is essential.
A simple weekly or monthly report is barely adequate for monitoring the progress toward the targets. By the time anyone finds deviations, they are already locked into place, and the report merely serves to record the historical facts.
Any key metrics have to occur in a timeframe that enables the user to identify the causes and take action. To do this you need a real-time enterprise. Inherent in the real-time enterprise should be the ability to present information in [near] real time and to provide the user with intuitive mechanisms for navigating and the capability to dig for more detail.
A real-time enterprise should provide a mechanism for exploring the reasons for any deviations from plan, drilling down to the "control points," and causing a beneficial change to limit, or eliminate, the cause of the deviation. Disparate sources provide this information. A typical process plant will include, at least, the following:
Strategic planning application
Process modeling application
Linear planning/optimizing tool (LP)
Document management system
ERP—handling HR, Finance, Inventory, etc.
Plant information system
Asset/maintenance management application
These systems handle fundamentally different and mutually incompatible data that travels throughout the enterprise making the task of integrating and blending information a difficult technical proposition.
In summary, the problems faced by the process industry include:
Disparate data: Process industry data is in lots of different systems.
How to scale the solution to accommodate lots of users and multiple sites or plants.
The [near] real-time enterprise: How to deliver data in a timely way so users can effect a beneficial change to events.
How to enable the user to explore and discover the cause of problems.
Rendering KPIs in [near] real-time that truly reflects the current, underlying activity in the business and monitoring progress toward these targets.
The solution should deliver these capabilities without having to replace any existing enterprise applications.
Enterprise monitoring (EM) can be an additional element of the information structure. A visibility layer, geared for operations but not an application itself. EM uses live data from other systems to create an environment for users to perform their jobs more effectively while the transactional systems stay unchanged. EM is a Web-centric application. However, by itself, the Web cannot create or distribute the information securely and efficiently within your enterprise. It does not give the tools you need to mange information, keep it current, or keep it organized. Furthermore, Web servers cannot handle processing-intensive business applications on the Internet, and they do not have the required features to support such applications.
EM solutions require a business-processing or application server to handle the processing and interfaces requirements, a processing server that delivers the high performance, and scalability required to meet future growing demands of an enterprise. EM solutions operate in n-tier architectural environments with four major logical components: business servers, application connectors, Web servers, and browser-based clients.
A business user utilizes a Web browser to send HTML requests through HTTP protocol to a Web server. The Web server then forwards these requests to the business server. The business server utilizes the application configured connectors to fetch the required data from the external systems and perform any pre-processing requirements on the data before routing the result back to the Web server, which delivers the result to the user's Web browser. The Web server component acts as a presentation server or conduit for receiving requests from delivering views and data to Web browsers while the business servers which are the integral component of the solution handles the majority of the remaining activities.
This solution content is in components, which are easy to develop and store on the application server. Instances of components assemble into a model that naturally reflect the organizational, physical asset base, process, logical, activity, or workflow structures present in the business. Each component manages its associated data and renders this data in any of many possible appropriate views. Views render data as graphics, forms, tables, or time-series charts. The model appears as a tree structure with branches and nodes. Data from nodes lower in the tree can aggregate and present at any higher node. This feature enables RtOI to easily summarize and present data from multiple data sources. When the data changes in the lower (child) node, the aggregated data in the summarizing node updates to reflect the change.
Data delivers to the solution components via connectors. Unlike technology simply based on Web servers and active server pages, which can generate multiple, asynchronous requests for the same information, RtOI connectors aggregate these multiple requests for data into a single request. You can tune the performance of the resulting architecture to optimize the consumption of network bandwidth and the impact on the data sources. Connectors are highly configurable in the way they acquire data. The update frequency and data acquisition mechanism comes from the needs of the user community and the capabilities of the backend data source.
Unlike conventional browser based applications, RtOI data automatically propagates to subscribing users whenever the model thinks it needs a data change. The updates require no user intervention, and users end up with the freshest information.
Also unlike conventional Web-based applications that present the data from each data source to the user into unique "tiles" of data, the application seamlessly integrates data from disparate data sources and types. This blending of data makes it easier for the user to assimilate information and understand what is occurring.
Harnessing operations information through operations intelligence has significant implications for the business. First, operations intelligence is the primary mechanism for the enterprise to realize operational excellence. Improving adherence to best practices, reducing costs associated with maintenance, optimizing organizational performance, and aligning operations to enterprise wide goals are some practical benefits.
Another benefit of operations intelligence comes from its integration of content from operational systems and business applications. The marriage of these two seldom-bridged worlds engages in the business, the people immersed in the day-to-day operations. For these users, they are able to frame real-time operations content with the meaning for the business as a whole. Through operations intelligence, users have a better understanding of how their actions impact the business and how they can align with the corporation.
Behind the byline
Fayez T. Kharbat is with Saudi Arabian Oil Co. (Saudi Aramco) in the Integrated Solution Services Department in Saudi Arabia. George Bauer is with IndX Software Corp. in Aliso Viejo, Calif.
Return to Previous Page