CD1011_CovStryButton
CD1011_CovStryButton
CD1011_CovStryButton
CD1011_CovStryButton
CD1011_CovStryButton

The Great Data Link-Up

Nov. 10, 2010
Streams of Factory-Floor Information, Aided by Performance Monitoring Tools, Converge With Business-Level Data to Solve Problems, Enhance Decision-Making and Increase Profitability

Hey, Mr. Machine, how're you feeling today? Mr. COO wants to know. Seriously?

Well, it's about time.

Operators, engineers and managers always want to know how well their machines are running. When machines were simple and shops were small, this was pretty easy to do. However, as machines and their corporations and staffs grew larger and more complex, it became harder to stay aware of how an individual machine's operational health could affect its company's economic health. You know, the old right hand doesn't know what the left hand is doing. 

About the Author

Jim Montague is the executive editor for Control. Email him at [email protected].

Still, many sophisticated devices have been developed and implemented to check on the health of machine and production lines, and as these increasingly software-based tools pushed for more and better operations, they also opened up new levels of proactive and predictive maintenance. Likewise, enterprise resource management (ERP) software and other business-level tools also have grown more sophisticated. However, the persistent problem is that both the production and business sides and their monitoring tools apparently still don't know how to talk to each other.    

New Links to Basic Facts
Despite the historical drawbacks and other new hurdles, some machine builders are finding ways to get machine performance data up to their business-level managers and clients. One of the newest ways to secure and distribute machine information is with the MTConnect standard that puts data from different types of equipment and applications into a standard format. Based on XML and HTTP, MTConnect fosters interoperability between machines, controls, equipment and software by employing a standard vocabulary to gather, publish and distribute machine tool data via Internet Protocol (IP) and Ethernet TCP/IP networking.

"We can network and monitor multiple machines on one server with MTConnect," says Milton Ramirez, product technical specialist for turning centers at Haas Automation Inc. in Oxnard, California. "It even sends text messages and emails to our phones. This allows users to see status, cutting conditions and other performance monitoring information. This means they can check spindle load, idle times and motion time, and monitor overall load on the tool. On the controls side, information from MTConnect can prompt users to do more required maintenance. For example, a prevention page on the interface shows when certain machines need their oil levels checked, filters cleaned, coolant added, chucks greased or seals replaced."

MTConnect's developers say machines use numerous data and communications methods, and so standardizing their language and its structural grammar was crucial for them to talk to and understand each other. Users can learn about MTConnect and even download its source code at MTConnect's website. This software can be used as is, modified for special needs, reverse-engineered or used as a template to create an individual software interface that meets the standard's requirements. Most recently, MTConnect reported that it is partnering with the OPC Foundation to jointly develop MTConnect/OPC-UA, which will be a companion specification set to ensure interoperability and consistency between them.

To better understand the capacity of its plastic medical device packaging machine, Benlan recently employed Plantnode real-time performance management hardware and software from Shoplogix, which secures data from direct digital inputs or via PLCs. Located in Oakville, Ontario, Canada, Benlan needed usage data for its form-fill-seal machine to aid capacity planning and productivity trending, and Plantnode provided it and also email, voice mail or PDA alerts when the machine reached its film replenishment threshold. This reduced many typical 10-15 minute downtime periods, and increased Benlan's production by 25% for this application. 

Self-Made Monitoring
Instead of waiting for others to bring performance monitoring from the factory to the enterprise, Jim Brown, engineering manager at Makino in Mason, Ohio, reports that his firm developed two suites of productivity-improving software, which can connect up to 36 of its mold machining centers, and then store spindle characteristics, alarms histories and other information in a historian database for future analysis. Makino's multi-client Machine Productivity Maximizer (MPmax) will be released in January.

"We found we had a lot of monitoring capabilities that worked well in a machine cell-based control system, such as machine availability and productivity, and that these functions could be communicated via Ethernet, so we decided to build our own solution," Brown says. "This came about because, in the past year, we'd developed our Autonomous Spindle Technology (AST), which can do in-depth monitoring of the spindle, and use embedded software algorithms to automatically adjust parameters on the fly to better maximize the cut, aid material removal, improve finishing and extend tool life (Figure 1). So, we just thought it would be nice to have the same alarming and machine availability functions in our cell system that we now have with AST. That's what led us to develop MPmax."

Savvy Spindle
Figure 1: Autonomous Spindle Technology (AST) in its MAG1 five-axis machining center inspired Makino to develop its upcoming MPmax software, which gives the machine availability, alarming and other performance monitoring capabilities of AST to the mold machining firm's cell-based systems and equipment.   Source: Makino

MPmax can get spindle and other data to the control locations, letting users better monitor availability, alarms, power consumption and other operations, Brown says. "Consequently, if a business manager has an MPmax client, he could display kilowatts per hour used relative to a machine's productivity over a given time period, show what parts and processes are using the most power, and maybe decide to run some more equipment during other shifts to reduce costs."

Makino recently began supporting MTConnect interfaces and their standardized, XML-based data formatting on its CNC machines over Ethernet TCP/IP, and Brown expects many of these functions to be offered by Internet-based, third-party or "cloud computing" services as they become more widely available. 

Got Good Parts?
Another big hurdle in getting users to adopt performance monitoring software is that, even though some solutions have been around for years, the information they provide isn't specific or sophisticated enough to improve decisions on the business side.

"Ten years ago, PlantStar and Maptec software and hardware gathered data on how machines were running, and could send cycle times for numbers of parts produced to the ERP system, but the unanswered questions are still: 'Are these good parts?' and 'Are you sure?'," says Mark Elsass, manager of applications and technical services, Cincinnati Milacron in Batavia, Ohio, which makes injection-molding machines and other equipment.

"For instance, we can get a lot of cycle data on recovery times and cushions, which is the amount of plastic left on the screw after a shot," Elsass explains. "We also can see averages and highs and lows, and set alarms for when a cycle goes outside those limits. However, some good parts can still set off alarms, and some bad parts can get through without setting them off. So, the monitoring question for us is, if the cushion is off, what level indicates a bad part? In my opinion, figuring out the limits of a bad part is very difficult because finding and understanding the limits of a bad part is a very tedious job."

From the Plant Floor to the Enterprise

So, you want to put your detailed machine performance data in a metaphorical taxi, and send it
from the factory up to the business? Well, you can get there from here. Follow this road:

  • Examine your application, evaluate and document how it's performing now, and compare that to what your company wants to accomplish.
  • Ask your business-level managers what types of plant-floor information could help them make better decisions.
  • Select and standardize on a controls platform and architecture that's flexible, and make certain it meets your specific needs.
  • Pick a primary automation vendor and work closely with its engineers.
  • Implement higher-feature I/O modules and components that can be easily networked and serve data into local HMIs and ERP systems without added middleware.
  • Establish communications between machines and the business by using PackML, MTConnect or other standardized communications protocols.
  • Make sure that upper-level engineers and business managers actually turn on their new plant-to-business monitoring tools.
  • Convert plant-based data values, such as scrap percentages and kilowatt hours, into dollar values per item produced to help direct improved and more profitable decision-making.

Elsass reports that one method for beginning to solve this problem is using cavity-pressure monitoring, and he says Milacron is mounting cavity-pressure monitoring devices in some of its machines. "The drawback is that you can't put 128 pressure sensors in a 128-cavity mold, but we can use pressure transducers to measure in enough cavities to give us useful data," he says.

Elsass adds that another way to gather and deliver these and other crucial measures to operators and engineers, and later to business-level managers, is to use more electrically driven machines, which have more readily available information than their hydraulic counterparts, and also can save 50% on energy. These accessible parameters include motor temperature, drive temperature and motor control information for injectors, extruders and clamps.

Similarly, Veski, a consultant in Zagreb, Croatia, recently built its Computerized Diagnostics System (CoDiS) for monitoring, analyzing and diagnosing vibrations in rotating machinery around National Instruments' CompactRIO real-time module, field-programmable gate array (FPGA) module and LabView software, and CoDiS is helping 11 Croatian and Slovenian power plants reduce maintenance and repair costs on 30 generators. CoDiS monitors rotor vibrations, stator and foundations dynamics, air-gap conditions, electrical exploitation processes, power quality and hydraulics on rotating machines.

"This monitoring system consists of a measuring component with all sensors and hardware used to acquire data, a signal conditioning component, CompactRIO's processing unit with a PDA local-control unit for data acquisition and analysis, and a central computer that houses a permanent database and analysis procedures for remote LAN users," says Boris Mesko of Veski. "The system also includes communication with SCADA systems using analog or fieldbus outputs. This means that all relevant data, including waveforms and analysis results, are available as shared variables over a shared network. So far, we've reduced the cost of maintenance and repairs in these power plants by up to 50%."

New Tools and the Cloud  
To secure enough of the right data, analyze it thoroughly, and provide it to the right users, some developers and builders are cooperating to create better software tools that delve deeper into what's affecting machines and what to do about it.

For instance, Mazak in Florence, Kentucky, uses System Insights' vimana software in some of its machining centers to collect real-time data, find previously hidden patterns and trends, identify sources of inefficiency, predict production losses, and devise strategies to eliminate those losses. The software accesses data from IP-enabled PLCs, CNCs and its own machine-based wireless sensors, performs its data processing and analyses on a secure software-as-a-service (SaaS) system—also known as cloud computing—and then reports back to users on their PCs, mobile phones and other handheld devices, and also via MTConnect (Figure 2). In fact, vimana and its dashboard were used to display detailed production data from seven linked machines in Mazak's exhibit at the recent IMTS 2010 show in Chicago. 

"Our focus isn't just on what's happening in a machine, but why is it not producing as well as it could," says Athulan Vijayaraghavan, System Insights' CTO. "We look at data to predict machine disruptions, but then combine that with production line flow data to go from predictive maintenance to predictive manufacturing. Because SaaS, or the cloud, can handle much more complex calculation, our software can convert millions of events per second and is well abstracted enough for us to, for example, ask and get answers about the load on a stuck spindle as it's happening during a downtime event. Having all the data before, during and after a machine goes down can aid machine health, but we also look at all the data around an event to find root causes and help users run their businesses better."

Its reporting abilities enable vimana to communicate via MTConnect to plant-floor users and equipment, but it also can talk to higher-level ERP modules. Vijayaraghavan adds that value stream mapping functions will soon be added to vimana, so it can better identify and differentiate between unavoidable and avoidable sources of waste in it machines.

Find the Why

Figure 2: System Insights reports that its vimana Manufacturing Learning Engine analyzes data from CNCs, PLCs and sensors to find hidden patterns and trends that indicate sources of production losses, and gives users strategies to eliminate them.
Source: System Insights

About the Author

Jim Montague | Executive Editor, Control

Jim Montague is executive editor of Control. He can be contacted at [email protected].