Situated just north of Monterey Bay in Davenport, Calif., RMC Lonestar's 875,000-tpy Santa Cruz facility has been producing cement continuously since 1906. The plant was instrumental in the rebuilding of San Francisco after the great earthquake and in supplying cement for the Golden Gate Bridge, the Panama Canal, and Pearl Harbor. Although output today is the highest ever, the plant and associated mines require only 150 employees compared with 260 before the 1979 modernization. Automation has played a role is this reduction.

When plant personnel recently upgraded the controls from a distributed control system (DCS), installed in 1979, to the latest programmable logic controller (PLC) system, the Santa Cruz plant solved a looming Y2K problem while boosting plant reliability enough to pay for the upgrade in only 36 months.

Costly failures "The aging DCS, which was coupled with 12 stand-alone PLCs for discrete tasks, had become prone to failure and very difficult to diagnose," explained Karl Willis, electrical superintendent at the Santa Cruz facility. "Typical of DCS equipment of that era, discrete inputs had to pass through four I/O cards to get into the DCS, and discrete outputs had to pass through another four I/O cards to get out.

"When the process 'winked' because of an intermittent fault in one of the eight I/O cards, it was nearly impossible to tell where the fault originated," Willis said. "The only solution was to replace the most likely card and hope that doing that solved the problem. Sometimes it took several tries before the right card was found."

According to Willis, the cost of these failures was growing exponentially. Each failure meant the process shut down because of interlock logic. Plant downtime costs an estimated $25,000 per hour at Santa Cruz, with no way to make up the loss since the plant normally operates 24 hours per day, 365 days per year at 100% capacity. Bringing the plant up to full capacity after a shut down could take anywhere from four to 48 hours, the longer time needed to clear vessels plugged by 2,700 degrees F feedstock acting like molten glass. Added to this cost was an $80,000 per year DCS service contract.

Y2K spurs action What drove RMC Lonestar to act quickly to replace the controls was the DCS's Y2K problem. In October 1997, the vendor notified the cement plant that its DCS was too old for a Y2K fix. RMC Lonestar immediately instituted a crash program to upgrade automation during the plant's scheduled February 1998 shut down.

Investigation suggested that the plant could either replace the 6,000-point DCS with another DCS, or develop an entirely PLC-based system. The objective in either case would be to upgrade control without necessarily improving on it. The idea was a like-for-like functional replacement with the latest hardware and software.

"Outside of reliability questions, the aging DCS still performed admirably in providing continuous full production with minimum manpower," Willis said. "The choice of a simple functional replacement also would make it easier to develop a new control system in five months, and to install and cut it over in only three weeks. The only new production equipment was to be several VFDs on fans."

Open environment To assure competitive initial and long-term costs, an open PLC system not dependent on an OEM vendor's configuration, service, and parts channel seemed to be the best route for Willis. PLC technology has advanced to where it can duplicate most traditional DCS functions, including those typical of a cement plant. The points in the plant are approximately one-third analog, two-thirds discrete.

Because of limited staff, RMC Lonestar retained FLS Automation Group, Hunt Valley, Md., to serve as system integrator. FLS Automation has developed a number of custom-loadable PLC function blocks for cement plant operations, such as a flop gate and a motor-reversing starter, which helped speed program development.

"Even then, RMC Lonestar's control room supervisor-with the instrument loop book in hand and plant operations in his head-spent most of his time between October 1997 and February 1998 at FLS Automation," Willis said. "It was impossible to print out the old DCS's loops. The supervisor also worked to assure that the new workstation screens followed the old screens as closely as possible, so as not to overload operators at startup with unfamiliar graphics and routines."

At the same time, RMC Lonestar's lead electrician, Stephen Faree, organized the installation effort using Microsoft Project software. His job was to make sure that all construction work proceeded in the correct order and on schedule.

Six PLCs in place Six Schneider Automation Quantum 486 Automation Controllers provide all analog and discrete intelligence and control, replacing the DCS and the 12 older Modicon 984 PLCs. Most of the new controllers are located in MCC equipment rooms in the same cabinets that housed the old DCS and PLC equipment.

"FLS Automation assembled and wired new backplanes for the existing cabinets, which sped the retrofit," Willis said. "Field wiring was simply disconnected from the old terminal blocks and laid back, the old backplanes removed, the new backplanes installed, and the field wiring brought back and reconnected to new blocks. No new field wiring was run."

Because of the small size of the new controls, the elimination of separate analog and discrete panels, and the fact that a series of I/O cards for DCS-to-discrete communications was no longer required, cabinets on average are now about one-third full. Although the new Quantum I/O is remote architecturally, it is generally located in the same cabinet as its Quantum CPU, or in an adjacent cabinet.

"A reduction from 12 to six controllers was possible because of the greater memory and speed of the new 486 microprocessor-based units," the electrical superintendent added. "The former PLCs, many driving 20-year-old Modicon 200 Series I/O, are being recycled to replace relay logic on many of the site's 311/42 miles of conveyors."

Two types of networks A new control room console features two redundant Digital Equipment minicomputers and three Dell workstations as operator interfaces. All run proprietary FLS Automation software. Operator input is by keyboard and mouse, versus keyboard alone used previously.

HMI communications between all workstations and controllers is on a fiber optic IEEE 802.3 10BaseFL Ethernet local area network (LAN) with ST connectors. The Ethernet modules in the Quantum controllers were preloaded with TCP/IP stacks and Modbus protocol application support in user-upgradeable flash memory.

"In addition, the controllers alone are networked on a redundant Modbus Plus high-performance token-ring LAN" Willis said. "Considerable data is interchanged between controllers in the burning and finishing areas of the plant.

"The independent controller network was established to isolate control from HMI functions, thereby maximizing control reliability and speed. The entire control system's electric power source is backed by a UPS and an emergency generator to assure controlled process shut downs," Willis stated.

Fiber optic cables were specified for both LANs to reduce EMI and to make it less likely that new cables would need to be pulled if communications protocols ever change.

All points relabeled "Prior to construction, we verified the ID numbers of all field wiring to make sure that every point thought to exist actually did exist," Willis explained. "The instrument shop, process department, and electrical department each had developed their own record books over the past 19 years. Scott Renfrew, process engineer, brought these three databases together, found the discrepancies, and then ensured the questionable circuits were traced.

"Once tracing was completed, all 6,000 points were relabeled and validated," he added. "We had to be very careful during construction not to detach the labels when replacing the cabinet backplanes."

A local contractor installed the equipment during the three-week plant shut down, with assistance from the RMC Lonestar's 10 electricians. Because the new control program was written from the old one, startup was relatively clean and the plant went back on line on schedule.

"In the intervening year since startup, we have had no shut downs resulting from control equipment failures," Willis said. "Based on output losses over the previous year incurred from failures in the old equipment, we were able to establish the36-month payback by assuming that the former failure rate would have continued."

This article was adapted from materials provided by RMC Lonestar and Schneider Automation, North Andover, Mass, (+1) 978-975-9696.