1 November 2005
A data dream
Far-reaching enterprise data management in machine-to-machine communications
By Matthew R. Miller and Laurence E. Keefe
Improvements in machine-to-machine or mobile-to-mobile (M2M) communication and more efficient operating systems for small devices will expand the collection of operating data to a new class of assets, improving operational visibility throughout the enterprise. A new technology for embedding data storage makes it possible to archive, use, and publish data right at its point of origin, making it simpler to optimize performance and meet regulatory requirements. The embedded component historian object technology promises to overcome limitations created by discontinuous network connectivity.
The maturing of M2M technology will alter the way we see and manage the entire enterprise. Acquisition and retrieval of data throughout the enterprise used to be a specialized activity that clustered around certain activities and assets. This has changed because of a new technology for embedding data collection at the point of use. Data collected will be available where it originates and every place else you need it throughout the enterprise. You can act on this data automatically to optimize processes, document regulatory compliance, and access it from any point in space or time you need it. Factors forcing this process include advances in M2M communications as well as improvements in the performance of embedded 32-bit hardware due to a standardized OS and its attendant software support. The key element in this new advance is the embedded data historian.
Embedded data historian technology
The embedded data historian is a new type of tool M2M developers are only now becoming aware of. It's not a flat file, and it's not a relational database, although it has some of the advantages of both. It's a highly efficient repository for automatically collected data.
This data, also called temporal or time series data, consists of two components: a recorded value of a user determined type, and a time stamp. The input/output (I/O) point identifies organized data in data stream series. This efficient format makes it possible to archive, retrieve, and organize data with minimal demand on system resources. It also makes it optimal for use in embedded devices. The growing acceptance of open standard operating systems such as Windows CE and Windows Mobile makes it even more convenient to incorporate this kind of origin at the point of data origin. The result is a new generation of intelligent devices that can perform local process optimization, communicate among themselves as required, and publish the data on the enterprise network business systems can use.
The future course of process management for the enterprise might require embedding data management in local repositories as well as in centrally managed archives. An embedded component historian object would enable an instrumentation OEM, machine builder, or other user to embed a custom version of a data historian right inside their own application and make it an integral part of their own solution for analysis and optimization.
Although problems could include techniques for compressing large quantities of data without losing any of the original fidelity, ways to create files that work together across multiple time zones, the creation of a 21 CFR Part 11-compliant audit trail, techniques for backfilling data that comes in to the archive out of sequence, and the ability to archive and retrieve huge quantities of data at high speed. At first glance, it might seem to be a straightforward problem to keep a record of data as it comes in from an I/O point. Up to a certain point that's true; even a conventional flat file can work for simple applications. But as the number of data points grows, the problems with managing the data grow.
The technology for embedded data historians could change the entire way M2M devices work together. Consider some of these reasons for embedding an historian in an M2M-enabled device:
- It handles discontinuous network connections and telemetry requirements:
- Lowers cost transmission rates and has a low bandwidth.
- Uses existing or available networks.
- Buffers high fidelity data when needed.
- Offers event based response to events.
- It provides local historical context:
- You can see what happened and when.
- Process optimization offers preventive measures.
- It gives diagnostics and repairs.
- It offers trusted records (compliance or otherwise).
Specific applications for M2M historical data
- Product and market research. Users will be able to look at usage patterns, preferences, and transaction timing as they track flows throughout multiple stores.
- Improved service and maintenance procedures. Knowing an event sequence can help identify complex problems quickly. Fixing design flaws and usability issues. – Faster response to design flaws will permit needed improvements. Manufacturers will be able to embed data historians in their product to track performance and machine response in critical situations over the entire life of the product.
- Better root cause analysis and local troubleshooting. This prevents tampering, fraudulent insurance claims, and disputes about who will pay for service, maybe even prior to dispatch.
- Process optimization and compliance reporting. M2M-based systems can provide the insight and integrity to support a wide range of standards, such as FDA, EPA, and NRA.
- Business transparency. You're able to view from a central location all assets and monitor their real-time status and capacity.
- Predictive metrics. History and real-time information will identify potential problems before they become crises.
As M2M devices proliferate, the cost and ease of local data storage continues to improve; data can be available everywhere you need it. The effects of this are difficult to predict, but you might foresee some of the immediate consequences.
Enterprise real-time performance management
Because of M2M-enabled data historians, enterprise managers can expect to see a cloud of data everywhere as well as waiting on call and data collected automatically about every activity and every operational parameter in the enterprise. Its absence will be the exception rather than the rule.
The enterprise is now a complex environment with rapidly changing requirements for information. Companies need to think and act globally. Situations change minute to minute, so response in real time is important if you want to act in time to take advantage of opportunities. But in order to support an effective response, you need to present real-time data on time and in historical context. Retaining and retrieving this information helps address the demands for accountability and compliance to standards.
From the first days of electronic data acquisition and record-keeping, some attempted to centralize the archiving of information, whether in control rooms as production records (productivity reports, alarm logs, and compliance reports) or in the business end of the enterprise (ERP, MES and similar accounting solutions). However, you can only expand central data retention archives so far before they run into problems of scale. The embedded data historian can help meet these problems of scale.
In the past, we took it for granted we could take selected, predefined measurements on the performance of certain enterprise assets, usually the most capital intensive ones, such as a cracking column, a large turbine, a press, or a wellhead. The proliferation of M2M devices is changing that because it's now economically feasible to collect data on the performance of even relatively inexpensive assets.
Nearly everybody knows about basic M2M devices; in the consumer environment, smart phones are an ideal example. But wireless connectivity and lean resource operating systems and open standards have caused the proliferation of M2M devices into thousands of unexpected and specialized applications. Small, interconnected intelligent devices now reside in or connect to the least expensive, most remote and until lately relatively ignored assets of the enterprise.
Of course, intelligent devices by themselves are nothing new. A hot water heater with rudimentary computing ability has been around for years, and even a cheap robot has appeared to do vacuuming.
What's new is the ability of these devices to collect data automatically, archive it locally, publish it over a network (either to the enterprise or for use by other machines), and do it cheaply, efficiently, and robustly. Even an intermittent connection to the network can enable a device with an embedded data historian to report vast quantities of data, whether it's for production tracking, compliance reporting, optimization, fault analysis, or any combination of these and other uses.
If you want to collect data on heavy equipment that is rented for use on a construction site, and the operator has leased it for a month and agreed to use it under standard terms and conditions, you could collect operational data on all key operating aspects during operation, measuring parameters such as engine temperature, RPM, torque, speed, and load weight. If any of these parameters exceed the agreed terms, you might not be able to notify the operator and leasing company. If there were indications that you needed maintenance, you could call a technician on-site through the maintenance management system, and post operational event data for diagnostics. Once you diagnose the problem, you can order parts to have on hand for the repair. GPS data would signal the location of the equipment for the technician upon arrival. In this case, the cost savings to the lease company and customer could be significant.
If we use these approximate figures:
Lost leasing time = 1 hour x $300/hr = $300
Service time saved = 2 hours x $100/hr = $200
Logistical time saved = 1 hour x $100/hr = $100
Total downtime saved = 4 hours x $600/hr = $2400
Customer satisfaction value = priceless.
Total event savings is $3,000. Assuming an initial system implementation cost of $100,000, the system would have an ROI of 33 service events.
But so far, all these applications are direct operational uses of historical data. The potential impact of data historian technology embedded in M2M devices is greater, however, when we use it for operational collaboration across the enterprise for real-time performance management (RtPM), ensuring timely and secure analysis, visualization and collaboration across the enterprise. With the increased availability of data from the RtPM infrastructure, each group in the enterprise gets access to a common data set (appropriately contextualized). This means all parties can share a common vision of the truth about what is happening, allowing them to focus their attention on how to best address operational and strategic issues.
An RtPM infrastructure provides the data integration discipline that moves information in systematic fashion from taking measurements to making better decisions. These convert into a long-term competitive advantage. The idea behind having an RtPM infrastructure is to give knowledge workers a tool that feeds the continuous improvement process, rather than just doing a series of individual improvement projects that require going out to gather new data. Availability of M2M historians now expands RtPM to other assets in the enterprise, beyond the current network. The increased transparency brings decisions closer to the customer and workers, and the option of low-cost solutions means faster ROI from better decision making. IC
Behind the byline
Matthew R. Miller is OEM relations manager at OSIsoft in Saline, Mich. Laurence E. Keefe is in OEM business development at OSIsoft in Victor, N.Y.