A new 3.0 mgd water treatment plant on North Carolina's Cape
Hatteras is believed to be the first in the United States to apply Foundation
fieldbus for device-level control communications. The plant, built by the Dare
County Water Department near the famous Cape Hatteras Lighthouse, has been
operating without a problem since startup more than two years ago. Fieldbus has
been widely applied in all process industries including electric power
generation, an industry also cautious to adapt new technology.
The new facility, located in the town of Frisco, is unique
in another way. A single, non-PLC
automation platform runs all of the plant's equipment (i.e., reverse osmosis trains
and pumps, anion exchange units and pressure filters, blending and distribution
storage tanks and pumps, and remote well and tank telemetry). In most water
treatment plants, these functions are served by several control panels or
platforms that either do not talk or poorly talk to one another.
To best understand the control system, a short discussion on
the makeup of the Frisco plant helps. The facility treats and blends water from
[if !supportEmptyParas] [endif]
long-used shallow well field (40- to 80-feet deep) of highly colored, variable
iron content water whose output cannot be expanded (for environmental reasons)
beyond the field's present 1.6 mgd capacity, and
new deep wells (200 to 300 feet) that take brackish water from high yield
[if !supportEmptyParas] [endif]
To prolong the life of the existing shallow well field,
production is being limited to 0.9 mgd. Supplementing this water are the
brackish deep wells, whose water is treated by two 0.7 mgd trains (of three
eventual trains totaling 2.1 mgd) of standard-pressure, high-rejection,
single-stage reverse osmosis (RO) equipment. The resulting 2.33:1 RO
permeate-to-shallow well production ratio will provide a well-buffered blend
even at maximum plant output.
The shallow well water is being treated for color/TOC
reduction by anion exchange plus oxidation and filtration for iron removal. By
significantly reducing THMFP and virtually eliminating iron, the natural
calcium hardness and alkalinity of this water blends advantageously with the RO
permeate that contains virtually no hardness or alkalinity. Most of the vessels
used in the former shallow well treatment plant were reconditioned and
installed in the new plant.
Each RO train is composed of 18 vessels of seven elements
each, with each element providing 400 ft2 of membrane. Because solute modeling
predicts a steady increase in deep well total dissolved solids (TDS) over time
to a limiting value of approximately 15,000 mg/L, the RO system has been
designed to operate eventually at 50 percent recovery with an initial design
recovery of 70 percent. The inclusion of energy recovery devices coupled with
the planned reduction in recovery will maintain a relatively constant feed
water pressure as recovery is reduced. This allowed the feedwater pumps to be
selected for the full range of operating conditions. Space and foundations have
been provided for a third RO train, to be installed when warranted by water
To help minimize power costs as RO pump pressures eventually
rise to 400 to 425 psi to overcome the rising TDS, each RO train utilizes high
efficiency vertical turbine pumps driven by high efficiency 300 hp VFDs. The
trains also incorporate energy recovery turbines that are expected to pay for
themselves in 41?2 years at the projected rate of TDS increase.
Other features of the new plant include permeate flushing of
the RO systems plus, on the shallow well side, permeate backwashing of anion
beds and permeate makeup for anion brining. Additionally, the RO system has an
ultraviolet sterilizer to protect membranes should stored permeate become
biologically active. The two water sources are blended prior to the addition of
post-treatment chemicals (i.e., caustic soda, hydrofluorosilicic acid, sodium
hypochlorite and a corrosion inhibitor). CO2 stripping is not required.
Operating the facility is the PlantWeb field-based process
automation architecture from Emerson Process Management, Austin, Texas. The
centerpiece of this architecture at the Frisco plant is a DeltaV digital
process management system (ee "Looks Like a PLC" on page 12)
consisting of the following.
[if !supportEmptyParas] [endif]
single compact controller fitted with hot-standby redundant central processor
unit (CPU) and power supply cards, two 2-segment Foundation fieldbus cards,
numerous 4?20mA and discrete I/O cards, and one 2-segment serial
communications card. (See photo above.)
operator, application and engineering workstations. (See photo below.)
Ethernet network tying the controller and workstations together.
fieldbus as well as conventional hard-wired field devices.
configuration from a library of preengineered control modules using IEC
Function Block Diagram and Sequential Function Chart languages.
(OSI Software, Inc., San Leandro, Calif.) PI historian for data capture, trend
graphs and reports.
[if !supportEmptyParas] [endif]
The automation was of particular interest to Dare County
because a troublesome or failed component can be taken off line and the rest of
the system will continue to run. Plant operations need not be interrupted to
replace a component or recalibrate.
DeltaV is a powerful process control system that can be
scaled from as few as 25 tags for running an OEM equipment skid to as many as
30,000 to operate a refinery. Because of this scalability, the cost of
automating the Frisco plant was competitive with comparable PLC automation.
The fieldbus cards serve 33 Rosemount pressure, temperature
and dp pressure transmitters connected to venturi flow tubes and magnetic flow
transmitters. The 4?20mA cards connect 12 turbidity, chlorine, pH and
conductivity instruments plus 10 servomotor-operated butterfly valves (MOVs)
for non-regulatory flow rate setting of lines serving two anion and three
pressure-filter blending vessels. Fieldbus MOVs were not yet available. The
serial card takes care of communications for the radio telemetry, the RO VFD
pumps, three VFD distribution and backwash pumps and two VFD booster pumps
feeding shallow well water to the anion system.
Although the cost of fieldbus devices was somewhat higher
than conventional 4?20mA versions, the difference was more than overcome by
the need for fewer conventional I/O cards, reduced engineering time when
preparing drawings, faster configuration, reduced wiring material and labor
costs and quicker installation and commissioning. Instead of 33 separate loop
diagrams, two fieldbus one-lines were sufficient. When connected, the fieldbus
devices were instantly auto-recognized, appearing on the engineering
workstation with license plate data and settings available without looking in
registers. No wire tracing and labeling, or instrument field calibration, was
required. Startup proceeded quickly and smoothly.
All process logic is run within the controller, even though
fieldbus devices can perform logic. There was no need to distribute control to
enhance process reliability or heighten response. Plant operators are finding
that troubleshooting fieldbus devices is simpler. Although the systems
integrator, who is located in Florida, has the ability to access the automation
and fieldbus devices remotely using pcAnywhere software, a need to assist
operators in this manner has arisen only two times.
The systems integrator largely configured the Frisco plant
process control system on site while supervising installation of both the RO
equipment and the plant-wide control equipment. By bringing all plant areas
into the single automation platform (RO system, anion and filter system,
distribution and telemetry) configuration was much easier, faster, cheaper and
better coordinated than trying to program and cobble together separate
standalone control platforms. Further, operator monitoring/control of all plant
functions is concentrated in the display screens of one interface.
The controllers, graphics, fieldbus instruments and
communications and serial communications were developed using one configuration
package. The system's global database also eliminated the need to map between
separate controller and HMI databases. If a tag name was changed, the change
automatically was reflected everywhere the tag was used.
The systems integrator divided the configuration into
numerous small sections to ease troubleshooting by operators and maintenance
technicians and to make it simple for an unfamiliar engineer to follow at some
later date. All function blocks were drawn from a provided library with the
exception of a derived totalizer block and faceplate prepared for the daily
Sequential function charts (SFCs) served as the step-by-step
and interlock backbones of the control system with function block diagrams
added to create PID loops and other control details. For instance, function
blocks were simply popped into the filtration system's SFC to tell the five
valves on each vessel to perform their routines without having to write
separate lines of ladder logic. Also, to shorten configuration time and
minimize checking, the logic for one vessel was simply copied and pasted to
create the other vessels. To make changes easier in the future, the logic for
all vessels is located in one place in the configuration.
Graphic configuration simplified cascading the caustic and
chlorine injection PID loops for pH control. Cascading is necessary because
caustic and chlorine both affect pH, and water properties vary greatly
depending on whether one or two RO units are in use or whether water is being
blended. The cascaded loops react quickly to sudden changes in water quality
and flow and slowly when near their setpoints, all the while precisely
coordinating the ratio of caustic and chlorine injected.
The automation's historian generates three performance and
efficiency reports per RO unit, a plant-wide alarm report and numerous
plant-wide trend graphs for operators and management to assess operations and
backtrack for finding causes should problems arise.
The scalability of the control system means that the
addition of the third RO unit at the Frisco plant as well as other possible
expansions can be handled at minimal incremental expense.
It also means that the advanced automation is useful for any
new or retrofitted water or wastewater treatment plants of any size or
complexity. In fact, the larger and more complex the project, the more the
strengths and resources of the process automation can be utilized. A 50-mgd or
larger integrated membrane system for brackish surface water treatment plus
conventional treatment for non-brackish water is well within the power of the