CIM systems from new sources.

NPE '94 in Chicago last June saw three additional suppliers join the ranks of vendors of computer-integrated manufacturing systems for plastics. Two of them - RJG Technologies and T.G. Branden Corp. - have expanded on their previous process-monitoring products to build complete real-time process and production monitoring systems. The third additional supplier - Nicollet Process Engineering - is new to plastics and launched a full system designed entirely from scratch and specifically for this market. Here is a review of the key features of each system.

FAST-SCANNING PLUS

WINDOWS

Minneapolis-based Nicollet has supplied systems for die casting since 1984. Its new injection molding system consists of three levels for single-machine, work-cell (four-machine), and plantwide implementation. High rates of data acquisition are said to be a notable feature of this system, which is based on 486 PCs and a Novell network. The single-machine (PMM240) and work-cell (WCM240) modules profile both analog and digital signals simultaneously, scanning up to 100,000 data samples per second, and generating a trace less than 1 sec after the shot.

Such high-speed scanning reportedly makes it possible to capture more data points and provide a more complete picture of what happens during a shot. High-resolution scanning may be an advantage on small or fast-cycling presses; it may highlight transient pressure spikes or vibrations during a cycle. According to a company official, these capabilities have attracted multiple orders from the research departments of at least two major resin companies.

In addition, full-featured SPC capabilities are offered for up to 24 parameters, and the job and machine database, holds up to 4000 entries. The PV1000 host-level system collects data from up to 70 machines. Single-screen plant-status display shows machine status, maintenance, and operator help conditions. For traceability, the PMM system audits and logs every occurance in an event record.

Another reported advantage of the Nicollet system is that its Windows operating system is used for all levels of monitoring, from the single-machine level to the plantwide level. Besides user-friendliness, Windows enables data to be easily exported to other software, such as spreadsheets or word processing.

In addition, says director of marketing and sales Dan Whealon, the system runs an Ethernet network all the way to the presses, which is said to provide plenty of flexibility in collecting and viewing information. Another claimed benefit of Ethernet is the ability to carry more data faster (10 megabits/sec) than competitive networking configurations, making it possible to monitor more parameters continuously, obtain better response, and generate reports faster.

In order to provide a rich SPC package, Nicollet teamed up with Quality America of Tucson, Ariz., which specializes in SPC software, to integrate the latter firm's software seamlessly into the Nicollet CIM product. The system offers 11 SPC charts, which Whealon says is a broader range than is typically available. The user can select three levels of control limits. No limits are imposed on SPC charts, so users can do SPC on all 24 parameters monitored if desired. In addition, a design-of-experiments (DOE) module is available for process monitoring.

On the production-management side, Nicollet's system includes job scheduling, as well as downtime, scrap, and machine efficiency reporting. Tying into the job scheduler is a "predictive maintenance module" that builds a history of maintenance on a machine and determines its future maintenance requirements.

According to Whealon, a "high-end" version of the Nicollet system is typically priced about 15% higher than similarly configured systems sold by competitors. A 10-machine system from Nicollet costs anywhere from $50,000 to $100,000, depending on options. A portable machine-analyzer unit is offered for $22,000. While Nicollet's first system was designed for injection molding, the company is readying an extrusion CIM product for release in the first quarter of next year. Also, it is currently beta-site testing a version for injection blow molding at 3M Co., St. Paul, Minn. (CIRCLE 27)

INTEGRATING PLANTWIDE

MANAGEMENT

Over the past year, T.G. Branden Corp. of Wilsonville, Ore., has gradually expanded its Shotscope injection molding process-monitoring system into a plantwide production- and process-management system and is starting to adapt its system to extrusion and blow molding as well. Branden has recently added production-management features, including an intelligent job scheduler that preferentially assigns a mold to the machine where it ran fastest in the past.

Key to the company's move into CIM is the Shotscope's new ability to share information with other software such as MRP-II, preventive maintenance, or accounting packages, according to T.G. Branden president Ken Panck. In the interest of importing and exporting data, the company has been building a utilities library that allows Shotscope to accept or deliver pertinent information to or from its database with minimal custom programming.

Shotscope collects both process management and production scheduling information in a database that can be exchanged between different systems. As an example of manufacturing control, Shotscope's job scheduler predicts completion times, calculates full-run job data, and passes on that information to the MRP system. Conversely, the MRP system can do the scheduling and pass information back to the Shotscope system. A recent addition to the software is a custom-report design option based on Microsoft Access, which allows users to tailor reports to specific format and data requirements.

Panck says the real strength of the Shotscope system remains its process-analysis capabilities, which give molders numerous tools to help achieve continuous process improvement. Shotscope archives parameters of every shot and stores the information on an optical disk for access at a later time. For example, Shotscope can provide a dynamic display of the screw velocity and position during every part of the plastication cycle. Such curves can be stored and recalled for troubleshooting or setting up the same job later.

Among the more recent additions are parts-gauging input and SQC screens for user-defined product attributes such as weight, dimensions, or color. By tagging a gauge reading with a shot number, product-attribute data curves can be overlaid with process-variable curves. Also, scatter diagrams can be created, pairing any product attribute with any selected process variable - the computer will automatically draw a "least-squares" curve through those data points.

Shotscope is an open-loop system, which keeps it from being overly complex, claims Panck. Ifs priced competitively: about $24,000 for the central workstation and file server with software, plus $4000 for each on-machine data monitor. The DOS-based system is being converted to work in Windows. A portable Shotscope machine process analyzer costs $7500. (CIRCLE 28)

A FLEXIBLE APPROACH

The DARTnet system from RJG Technologies, Traverse City, Mich., is comprised of separate "DART" single-channel (i.e., single-variable) data-acquisition modules, each of which can monitor and store individual process variables at 100 samples per second. Up to 255 modules can be networked so that injection presses can be monitored from one PC at a remote location. Critical process parameters that can be viewed include fill time, injection pressure, cavity pressure, and end of fill.

The DARTnet system is based on an approach that allows a molder to view the smallest number of selected variables that the user deems necessary, providing an uncomplicated view of the process, according to RJG v.p. Brad Watkins. It allows the molder to selectively monitor those variables that correlate best to production of good or bad parts. Because it is modular, a system can be configured and proved out on one machine before being implemented on machines in the rest of the plant.

Watkins adds that RJG has the expertise to help molders narrow their selection of variables to monitor. Just two variables are often enough to predict whether a part will be good or bad, Watkins says. For example, monitoring cavity pressure and injection pressure at the cylinder also gives an indication of fill time and generally correlates wen with consistency of part weight, dimensions, or strength.

The graphic waveforms of key parameters provides the molder with a "snapshot" in time of that part, resulting in a sort of blueprint for the part that can be duplicated. The number calculated from the areas under the curve relates to key parameters such as part dimensions, weight, and strength. "Primarily, everything about the process is represented by one number, which is the area under the curve," explains Watkins. If the area changes, then I can automatically detect that the part may be bad and sort it before the mold opens."

An automatic alarm-logging function saves the waveforms whenever an alarm condition occurs. The area is divided into sections to correspond to the process stages of filling, packing, and cooling. This provides detailed troubleshooting information, which can be used to interpret the problem at a later time.

Another capability allows charts of different variables to be overlaid on each other or on a master template to help pinpoint the cause of a process going awry. Eventually, RJG plans an expert system to aid in interpreting the process.

Because DARTnet is a modular system, price varies with its configuration; however, a system that is set up to monitor hydraulic and cavity pressure, including networking and software, would probably run about $4500-5000 per press, estimates Watkins.

DARTnet is a DOS-based system and can run off of an office network such as Novell. In addition to process monitoring, it can also perform production monitoring functions such as parts counting, job status, runtime, and downtime. RJG also offers a series of portable machine monitoring boxes with up to six channels with an approximate base price of $8000. The box has only LED displays and must be linked to a PC for data analysis. (CIRCLE 29)