After the Explosion...
A renewable-energy power plant has a chance to improve its technology after an electrical explosion wipes out much of the control network infrastructure.
Imperial Valley Resource Recovery (IVRR) is a biomass-to-electric power plant delivering more than 16 MW of power onto the grid. In the spring of 2004, during a maintenance shutdown, the electrical paralleling switchgear suffered a major electrical explosion that destroyed the connection to the grid and adjacent control room electronics. As a result of the extensive damage, Dynalectric Co. was contracted to provide design, installation, programming, and support to update the prior electrical and process telemetry, and ultimately incorporate Ethernet-based Modbus RTU communications and controls. The new Ethernet-based system has been in place for
approximately one year with the generator again 'on-line' and with significant improvements adapting Ethernet-based I/O Net Concentrator Systems (NCSs) from Moore Industries.
Flash and boom
Imagine that you are in the control room, when suddenly the adjacent ac power panel room lights up brilliantly white, accompanied by a loud boom. A plasma fireball bursts out of the ac power panel main contactor area. It immediately vaporizes the paint off of the nearby power cabinets and melts the aluminum overhead cable trays. All your control room cables carrying signals to and from the facility are incinerated.
You exit your workstation—after all, the melted cables have definitely shut down the facility. The fireball remains for a few seconds, but there is no subsequent flame. The metal walls of the contactor room do their job so nobody is injured, but there is soot everywhere.
The fireball destroyed a few ac cabinet panels, nearby cable trays, and contaminated all of your control room electronics with soot. The rest of the facility is undamaged. You are now faced with the opportunity to upgrade your control architecture, replacing your twisted pair contact closure, thermocouples, analog signals, and relay logic. This involves an in-depth review of your operation and both short- and long-term plans.
Basic plant layout
Design considerations for this facility have many common characteristics with other power plants and with all manufacturing facilities. Wood chips for the furnace that heats the boiler arrive via truck, are unloaded and sent through conveyors that remove contaminants, re-chip the wood, and sort raw material by size. Additional conveyors restack raw materials into holding areas. There are two such independent raw material preprocessing systems to enable redundancy. Each raw material system will be independently controlled both in the field and via the new Ethernet-based communication link with the control room. The furnace area has feed conveyors, combustion air blowers, and thermocouples. The control room provides precise monitoring and control of these functions.
The furnace heats a circulating water boiler that provides high pressure steam to operate the turbine-generator set. Exhaust steam is routed to a large condenser and fed back to the boiler.
This heat-to-electricity conversion requires additional instrumentation and controls (thermocouples, pressure sensors, motor controls, condenser flows, etc.) to assure proper water flow rates and manifold switching. Manifold switching enables switching pumps for maintenance, thus ensuring 100% operational time on the turbine generator. All of these processes are automatically monitored and controlled from the control room. Also, there are multiple manual gauges installed throughout the facility. However, these are not routinely used to control the automated processes.
IVRR is located in the desert near Baja, CA. Temperatures hover near 120 °F (49 °C) in the shade (hotter in the direct sun) during the summer months. This is a critical consideration for facility electronics. Field communications devices, sensors, and actuators need to operate in this environment without any air conditioning. "We are in the power generation business, not the air conditioning maintenance business," says Jim Medland of Dynalectric. The only exception is an air conditioner in the operator control room.
There are also extended seasons of blowing desert dust and flying insects, so electrical panels must be NEMA 4 rated with no external ventilation openings. The electronics packed within the enclosures can only use internal air circulation fans with no outside air access.
Guidelines for rebuilding
IVRR's management team, including Dynalectric and consultants, directed the following guidelines for the rebuild:
Rebuild what is necessary, and do not redesign the entire facility from scratch. Only implement changes that will provide measurable improvements. There are only two damaged areas and the rest of the facility is undamaged, so IVRR wants to get back into normal production promptly.
Use technology that is proven, something that we are already familiar with, and is readily available.
Use technology that meets current and future plans with open and flexible architecture. Do not get locked into a single vendor or a high maintenance technology.
The team decided as follows:
A. Delete multiple long twisted-pair cable runs between sensors, actuators, and controls . Replace them with bidirectional field data concentrators to collect multiple sensor and control pairs near their point of installation throughout the site. Communications for all channels need to be via a simple communications link. This will eliminate hundreds of pairs of previously installed twisted pairs and their maintenance, along with expensive A/D converters and control logic at the DCS location.
The team selected Moore's NCS with a communications module plus up to sixteen plug-in I/O modules. Each I/O module supports multiple discrete channels of inputs, relay outputs, or analog inputs (voltage current loops) or low level sensors such as RTDs, thermocouples, and potentiometers. The NCS is expandable as necessary by adding I/O modules as future growth occurs. It handles temperature extremes of -40 to 185 °F (-40 to 85 °C) to handle all I/O and communications in the desert environment. All of the prior contact I/O pairs plus the analog and temperature pairs can now be routed very short distances from their various facility locations to the nearest NCS.
The team selected the EIM (Ethernet interface module) as the communications module for the NCS, which can use either Ethernet (LAN-based) or RS-485 (twisted pair-based) communication modules. These both support Modbus RTU and OPC communication protocols. However, the EIM supports an internal Web Page Server that is ideal for programming and maintenance. The EIM also provides its own internal data logger, and PLC-like programming using ISaGRAF (IEC611311-3 approved) languages.
B. Use a communication protocol that is already known with a recognized open architecture so that future growth and support will be readily available from any vendor . The team selected Modbus RTU. This is the most popular protocol, and estimates say it is used in more than half of process control applications. It is well suited to legacy equipment, and can transfer the status of coils and analog signals. The communication is all digital, so analog and digital facility data is not degraded no matter where on the facility the information originates or is being directed.
C. Communications link technology needs to be flexible enough to use from anywhere on site to any remote location . Use a modern serial protocol that is readily available and reliable. Eliminate unnecessary pairs of wires where possible to simplify numbers of cables and associated maintenance.
The team selected Ethernet communications with the familiar LAN cable. While Modbus RTU uses twisted pair wire, Ethernet uses Modbus TCP. The Modbus message is identical in both communications protocols. In Modbus TCP the data byte is prefixed by the LAN TCP addressing and suffixed by message error checks. Thus the originating and destination locations merely insert or extract the data bytes from the overall message. The Modbus TCP protocol communicates by using any common LAN cables, switches, fiber optic modems, and goes anywhere in the world. Thus communications within the IVRR facility is easily incorporated.
While installing the new LAN cables throughout the facility, it was convenient to remove the hundreds of now abandoned pairs of communications wires. The speed of the Ethernet approaches 10 MHz and higher, and carries all of the facility data in one cable. The data enters the DCS using one Ethernet card. The prior design required multiple pairs of wires each requiring an expensive interface card at the DCS. Thus the overall investment to go with Ethernet based NCS systems is a considerable cost saving over any thought of replacing piece-for-piece earlier technology.
D. The DCS needs to be familiar and readily easy to use . The team selected a brand of controller with which the programmers were already familiar. Additionally, it has a communication port to accept bidirectional Ethernet LAN communications, and thereby eliminates the need to have multiple scanners located at the DCS workstation to collect sensor/contact signals and originate outgoing signals. Again, this saves wire pairs. Control and display are via readily available desktop computers and monitors. The control room is air conditioned for the host computer workstation, display panels, and operators.
The entire electrical room was demolished, and consequently, all power distribution equipment had to be removed. The damage was so extensive that all walls and roofing were also replaced. The AC power panels were replaced as well. The overhead cable trays were replaced with much smaller designs since they only have to carry a few twisted pair cables plus an Ethernet cable. Relay logic circuits are still used throughout the facility, however many of the communications links to the control room now ride on the Ethernet cable. Portions of the original cabinets make a good home for the new Ethernet-based NCS electronics.
The DCS control room was also cleared, cleaned of soot, repainted, and new computers with monitors were installed. For convenience, five flat screen monitors provide entire graphic coverage of the system. An additional computer in a nearby office connected to the same LAN cable network monitors NCS activity and DCS programming.
Four new NCS units, some containing up to 16 I/O modules of multiple channels each, are installed into these old cabinets or into newly constructed cabinets throughout the facility. All cabinets are sealed to prevent dust and insect migration. Power, I/O pairs, and an Ethernet LAN cable are all that enter the cabinet.
Lengthy sensor and control wire pairs that were cut just outside the ac panel room are now trimmed further back, making even more space in the cable trays. Sensor and control pairs are connected to the nearest NCS panel. Mapping diagrams identify each sensor and control wire pair by Ethernet IP address, I/O module, and channel number. Each NCS I/O channel is completely electrically isolated from every other channel, thereby enabling complete flexibility of thermocouple, RTD, potentiometer, current loop, dry contact closure, and other signals avoiding adverse communication effects on each other. Furthermore, the isolation prevents ground loops in-between channels. This is important because sensors are widely separated with intervening large motors, and there are inherent ground loop concerns for low level sensor signals.
Each of the NCS analog input channels uses a 20 bit A/D converter and communicates this analog value as a digital number all of the way back to the DCS. Thus the DCS now is able to obtain a true reading regardless of sensor distance. This enables the DCS to operate with better precision and higher stability from all sensors.
The Modbus TCP protocol provides simple and successful communication between the remote NCS and DCS. The DCS costs less because it no longer requires multiple manufacturers' unique I/O cards to collect and send information. Communication for the entire facility is via one Ethernet card and one LAN cable. The LAN cable is split using standard industrial grade LAN switches. No particular setup is required on the switches since the entire system is co-located at one facility and no outside LANs are connected. If desired, the LAN switches may be configured to communicate with selected outside virtual workstations so programmers and maintenance staff may securely check performance of the system from anywhere by using any Ethernet port for access. The NCS has a built-in Ethernet Web Server providing an HMI (human machine interface) with security for this purpose, both on- and off-site with connected laptop computers.
Making it work
These new improvements did require some extra efforts:
Unused relay logic and many interposing relays were removed from earlier cabinets to make room for the new NCS electronics. The NCS electronics replaces the cable pair bundles and provides up to 16 sets of I/O modules. Each input and output is isolated, diminishing the need for interposing relays. Input signals arrive on one side of the NCS I/O modules with outputs on the other side. This makes for high density wiring to the terminals since I/O modules typically have up to 16 wire terminals per I/O module base, but this can make for awkward access in tight quarters. However, each I/O module's electronics is quickly detachable from its own terminal base as a maintenance feature for hot swaps. Removing the electronics portion of the I/O module provides easier access to the terminals. Reinstallation of the I/O module electronics is just as fast.
On the sunny side of the enclosures, the temperature of the enclosure can climb several degrees. Since the enclosures are sealed, Dynalectric chose to add two air circulation fans inside certain enclosures exposed directly to the sun. This moves the inside air sufficiently to assure inside temperature does not exceed 185 °F (85 °C). The decision to go with the Moore Industries' NCS and its higher operating temperature is a benefit that has resulted in a "no maintenance concern" according to in-house engineers.
Previous relay logic panels distributed throughout the facility are replaced with computer programs within the DCS. These new programs duplicate the prior relay logic and sometimes complex interlock function algorithms. Programming is a time consuming task, but is now successfully accomplished. Now fine tuning of functions is readily monitored and improved by this upgrade. Many of the facility's manual push-button panels have their contacts duplicated and sent to the DCS via the NCS and Modbus TCP. The DCS can see these actions, display them and can even perform the same actions as the manual controls without the need for personnel to be sent outside to do these same tasks. These are a sample of the new flexibility of a DCS with flexible programming and using the Ethernet to fetch and send sensor data and control information. What was previously a physical rewiring of a task in the field is now a computer subroutine and communication with IP addresses on the Ethernet.
The flue gas exiting the furnace has to pass through a baghouse to capture particulates before going out the stack. The system monitors the dust load in the baghouse by measuring pressure drop from the inlet to outlet. This pressure difference is measured by pressure sensors and reported via Modbus TCP to the DCS. At a given threshold, the DCS runs a subroutine to send a very brief ON/OFF pulse burst in a precise sequence to valves on an air manifold to knock accumulated dust particles down onto a screw conveyor and ash is removed from the system. The valve process requires a lengthy communications distance, but was necessary to fine tune the process for optimization.
As a future alternative to the valve communication link process, the programmer can take this same DCS algorithm, rewrite it using ISaGRAF control engine software and install it within the NCS EIM module located at the baghouse. The NCS has a 'built in' PLC capability using an ISaGRAF run time module. The pressure sensors may be measured directly by the NCS and then trigger the NCS to perform the correct valve sequence without involving the DCS. This same NCS algorithm may also be triggered or inhibited by the DCS, as desired, for more flexibility.
The Ethernet link enables the DCS operator and field technician to connect a laptop anywhere there is a LAN connection at the facility. Via any Ethernet port on the LAN, all of the NCS I/O modules may be monitored using the NCS built-in Web server. Thus the NCS has an HMI (human machine interface) panel, and there is no software to purchase for this ability. Similarly, the configuration of any of the NCS I/O channel setups may be accomplished from the same Ethernet connection. A password prevents unauthorized editing of the system. A data logger on-board each NCS allows the last 64,000 transactions to be recorded on any selected channels. The data logger configuration and log file are also available via Ethernet. Thus the maintenance technician can see and exercise valve positions, actuators, temperatures, and so forth at any point without the need to have the DCS operator or an extra helper for these tasks.
IVRR is up and operating now with its well-planned Ethernet-based system. It is current technology that enables programmers to incorporate capability enhancements or make adjustments to improve peak process performance. These decisions reduce maintenance costs and the need for extra maintenance personnel as compared to the earlier control system. IVRR is confident they are already reaping the benefits of this new and reliable platform to be competitive for the future.
Read this online at SEPTEMBER 2009 at www.controleng.com/archive for links to more about industrial Ethernet.
Jim McConahay is senior field applications engineer for Moore Industries. Reach him at firstname.lastname@example.org .