Information, not just automation
Software architectures move plant automation vendors into performance-partner role
By Roberto Michel, senior contributing editor -- Manufacturing Business Technology, 7/1/2006 12:00:00 AM
Just when did the big plant automation vendors evolve from hardware-centric suppliers to suppliers of manufacturing information management software? On the one hand, the transformation has been under way for several years—as evidenced by the vendors' software acquisitions and their growing share of key plant operations software markets.
On the other hand, the evolution of these vendors around software remains incomplete, with much buzz about architectures, but sales still largely driven by control hardware. As Boston-based AMR Research recently noted in a report on the "big software dilemma" for automation vendors, revenue from applications software for most of these multibillion-dollar companies is akin to a "rounding error" on the corporate balance sheet. Even so, AMR figures plant automation vendors control 46 percent of the manufacturing execution system (MES) software market.
But even if automation vendors have only partially transformed their wares and appeal, there is no doubt the evolution is under way. The vendors have modernized their control platforms over the last few years to allow easier migration and open server platforms, and have acquired MES and plant intelligence software vendors. Underpinning these efforts is architecture.
Generally, the architectures facilitate integration of plant-floor data with the enterprise in accordance with the ISA 95 standard, and support reuse of industrial automation engineering work. These frameworks serve as a common software layer for configuring systems, and for commonly used plant-level functions such as alarming and historization. Most contain a unified plant data model that abstracts and organizes factors such as the equipment and assets in each factory. Some also may have workflow and business-process management capability. Sure, the architectures simplify integration of data to higher-level systems, but their overlooked purpose is as a foundation for modular MES and plant analytics applications.
Unified plant models
Users are in need of new plant analytics and MES-level applications, says Colin Masson, a research director with AMR. In a recent survey, AMR found significant gaps in areas such as finite plant scheduling, managing in-process quality, and visibility into current production costs. "These areas fit into the sweet spot for the plant automation and MES vendors because they all require a high-fidelity manufacturing model that ERP vendors don't have at this point," Masson says.
Unified plant data models can be correlated with other data sources such as current energy use and costs so as to then assess manufacturing costs for a product or batch. "If you want to build a manufacturing costing model, you need to start with a high-fidelity manufacturing model—not keep it at the 50,000-foot financial level," says Masson.
Jane Biddle, a VP with Boston-based analyst firm AberdeenGroup, says architectures offered by automation vendors complement the desire among users to standardize their approaches to MES and plant-information systems. In a recent survey on MES needs, AberdeenGroup found that 46 percent of best-in-class performers are standardizing plant applications within the four walls of facilities, and 36 percent are integrating systems into common factory-floor information platforms.
"There is a top-down approach to normalizing the approach to these systems," says Biddle. "It's often driven by corporate IT in partnership with the VP of manufacturing."
Plant-information architectures, says Biddle, are becoming more important as a means of serving up data to enterprise systems in a consistent way. Vendor architectures, she adds, serve as a common software layer for both integration to the ERP level, and for applications at the MES level.
"Before, with plant-level systems, you had to understand the data model for each one, and bring together the data when you needed to, through a custom effort," Biddle says. "Now you can begin to establish what is the overarching data model that includes all the data elements for all the plant systems you have."
Kevin Tock, a VP with Invensys's Wonderware unit, believes users will leverage architectures for both integration, and as a foundation modular applications. Invensys's architecture is called ArchestrA, and includes a plant data model and a software framework for common services such as security, alarming, and historization.
Better architectures may help vendors sell more applications, but the best approach, says Tock, is a framework that allows users to deploy just the applications they need, when they need them.
"The real test is who can modularize the functionality in the MES space," says Tock. "The vendors that are going to be the most successful are the ones that can bring to a customer the most appropriate application pieces at a reasonable cost. The customer may only need one piece—but because all the vendors are talking about frameworks and data models, users must ensure there isn't a lot of overhead in supporting that very first function that is needed. So the vendor architecture that will win is the one that can deliver the most appropriate pieces of that MES puzzle at a reasonable cost."
Performance goals
Kevin Roach, VP of Rockwell Software, agrees that vendor software architectures support user interest in adopting modular applications that share the same information model. Rockwell's architecture is called FactoryTalk. It includes data-sharing services with process-level systems, a Unified Plant Model, and other common services such as security, historization, and audit-trail generation. Rockwell's FactoryTalk suite of applications leverage these services.
But FactoryTalk primarily is about serving user needs, says Roach, rather than paving the way for software sales. "Because all the applications share those services, the integration time is much less, and because the engineering time goes down, your costs go down, and you also speed up peak time-to-market [when introducing changes to manufacturing]. In the end, you have systems that perform at a higher rate, and are easier to maintain."
AberdeenGroup's Biddle agrees thatcompanies are after a more enterprisewide approach to plant applications. The reason, she contends, is corporate pressure to improve performance. "Before, you probably could have gotten by without [a consistent approach] because there wasn't this ratcheting down on manufacturing performance you see today," Biddle says. "But if you look at what has happened in manufacturing, there has been a rationalization of resources, which means the assets that are left have to be at peak performance, and the only way you can do that is to get a consistent handle on the plant environment."
This need to increase performance is expected to fuel the sales of MES and enterprise manufacturing intelligence (EMI) software. For the big automation vendors, these software niches represent higher growth markets than controls.
For instance, Dedham, Mass.-based analyst firm ARC Advisory Group pegs the $12-billion distributed control system market as growing by just 6 percent through 2009, with only 4-percent growth in hardware revenues. By contrast, ARC predicts a 15.2-percent compound annual growth rate over the next five years for collaborative production management (CPM) software and services in discrete sectors.
ARC's CPM term is roughly analogous to AMR's MES category. Last June, AMR estimated the size of the MES market at just over $1 billion for 2004, noting that it had grown by 50 percent since 2001. In its report on the MES market, AMR noted, "Large automation vendors that have invested heavily in unified platforms for their manufacturing software offerings are poised to effect a sea change in this market. AMR Research expects to see market share shift to these players within the next 12 to 18 months."
Buyer beware
Clearly, the major automation vendors are well positioned to meet several needs, including the need for a new generation of production-management applications. However, says AMR's Masson, they don't exactly have the market sewn up, for a couple of reasons.
For one, a pure-play MES vendor still can capture large chunks of the market if it addresses gaps such as better plant scheduling. Second, automation vendor architectures aren't all the same. Some may offer more flexibility, such as being able to configure multiple applications within one framework. In other cases, he says, users must first implement a vendor's MES or plant historian to use its plant-intelligence functionality.
"There is a lot of marketing out there that starts to sound very similar," says Masson. "Buyers need to look closely under the covers to understand how truly flexible a framework from an automation vendor or an MES vendor is," says Masson.
Bob Mick, a VP with ARC, agrees that software architectures will be a key battleground for the automation vendors, but not, in the end, their only way to differentiate. "They will continue to compete on a number of factors: who has the best functional fit for a vertical industry, who is best at support, and who excels at integration," says Mick.
Finally, when SAP acquired EMI vendor Lighthammer last year, and coupled that functionality with its xApps and NetWeaver technology, the plant automation vendors suddenly faced a hefty competitor for higher-level manufacturing visibility solutions. AMR recently posited that SAP remains more of a potential partner than a competitor to the automation giants for problems like energy and cost management.
The bottom line: in this era of composite applications, where the answers on manufacturing performance are less likely to be found in one traditional system, even the automation vendors with their improved architectures and software acquisitions can't lay sole claim to plant intelligence.
