The Diablo Canyon installation consists of two 1,100-megawatt pressurized water reactors. This type of reactor pumps highly pressurized water within a primary loop through the reactor core, heating it to temperatures that can exceed 600 °F. High pressure keeps the water from boiling. The water is then sent to the steam generator, where it flows through thousands of tubes, where its heat converts lower-pressure water, in a secondary loop, into steam.

As in power plants that run on conventional fuels, the steam spins the generator turbines, producing electricity serving Pacific Gas and Electric (PG&E) customers. Steam then cools in a condenser loop that draws cooling water from the Pacific Ocean. A system of turbine exhaust condensers, air ejectors, heaters, and pumps recovers the condensed steam and returns it to the steam generator. An automatic control system ensures that the operation returns the exact amount of water necessary to balance the water volume equivalent of the exiting steam, maintaining a constant water level in the steam generator. No water in any of the loops ever comes in contact with water in the other loops.

Accurately controlling feedwater is critical to this process. Should water volume fall significantly, the reactor will not cool sufficiently. Too much water would cause water to pass through the steam header, which can cause turbine damage. The control system that maintains this balance must run reliably—without failing—and control the process when the plant is operating and while it undergoes routine and non-routine maintenance.

Scott Patterson, project manager for instrumentation and control systems at PG&E, says that as recently as three years ago the company lacked any kind of coherent strategy for upgrading the plant.

The plant went on-line in the mid-'80s, but some of its analog equipment was from as much as 20 years earlier. The Westinghouse-designed facility included vintage Westinghouse controls on critical systems, all of which faced imminent obsolescence.

PG&E wanted a digital-control system from one vendor that could provide a common platform across all instrumentation. The goal was to increase the plant's reliability, reduce maintenance and associated costs, and extend the plant's useful life.

The company's engineers spent more than 6 months evaluating available digital platforms before choosing Triconex hardware from Invensys—primarily because of its triple-modular redundant architecture, which would provide an extra margin of safety for plant operators. The design features three main processors and internally triplicated I/O cards that can be changed out on-line. The turbine control system receives signals from the several places, including the steam supply system and the turbine peripherals, which are sent to the processors for analysis. This approach ensures that all of the processors receive the same information and can "vote" on the appropriate action. It also permits the plant to continue operating normally despite a single or even double fault.

Any fault in one of the I/O cards, for example, sends an indicator or alarm to the control room where the display shows which assembly generated the fault. Diagnostics are run automatically by the Triconex hardware. Following the diagnostic results, a maintenance person goes to the site of the fault, installs an operable card and takes the faulty card out of service. In this way, the reactor can run without interruptions. The bad card is repaired (if possible) off-line and returned to the spares inventory.

The Triconex control system chosen by PG&E interfaces with four governor valves with dual linear variable displacement transformers (LVDTs) for position, four stop valves, reheat and intercept valves, a speed sensor, and dual servo-position controllers. The hardware also offers overspeed protection and can ramp down the unit to prevent a turbine trip if support systems like a feedwater or circulating water pump trips off line. This is to prevent a reactor or turbine trip. The control system software comes with a set of pre-configured functional blocks within process libraries optimized for turbine control. In PG&E's case, however, engineers also created proprietary custom blocks.

Typically, the hardware/software vendor or one of its system integrator affiliates will install and integrate its systems in the field. PG&E, however, performed its own integration. The reason? Patterson contends that installing a control system teaches the end-user a great deal about it.

On the other hand, Patterson admits that self-guided installation inevitably strains already limited supply of people resources. Although the installation required plant shutdown, Patterson reported a smooth transition. People involved were highly familiar with the old system, so they knew the required steps to get the new one up and running.

After completing installation on one of the two reactors at Diablo Canyon, PG&E was sufficiently pleased with the system's performance to replace the other reactor's turbine control system.