Forrest M. Holly Jr., Allen Bradley, Benedetta Rocco
Iowa Institute of Hydroscience and Engineering
The University of Iowa, Iowa City, IA 52242
USA
John B. Parrish Iii
Parrish Engineering, Beaverton, Oregon, USA
Corresponding Address:
Forrest M. Holly Jr. (forrest-holly@uiowa.edu)
IIHR
300 S. Riverside Drive
The University of Iowa
Iowa City, IA 52242, USA
Tel: (319) 335-5229; Fax: (319) 335-5238
Abstract: This paper describes the development and application of an unsteady thermo hydrodynamic model of a system of nuclear powerplants and their associated riverine and lacustrine cooling resources. The purpose of the model is to predict critical surface-water temperatures over a several day forecast horizon taking into account time-dependent meteorological, hydrologic, and plant-operation data. The temperature forecasts provide decision support for plant personnel who must respect certain river and cooling-lake temperatures in complying with environmental and safety requirements.
A unique feature of the one-dimensional unsteady model is its automatic, real-time acquisition of actual and forecast forcing data from internet resources. These actual and forecast data include hydrologic inflows, meteorological parameters, and in situ water-temperature measurements at selected locations.
The thermo-hydrodynamic forecast system requires procedures for internal imposition of in situ data where available; detection, reconstruction and repair of missing and anomalous internet data; and procedures for coping with uncertainty in forecasts.
Keywords: Powerplant，cooling，real-time simulation，computational hydraulics，hydroinformatics;
thermo-hydrodynamic modeling; unsteady flow; contaminant transport
Following a period of extreme air temperatures and humidity in late July 1999, the Nuclear Generating Group of the Chicago power utility identified the need to acquire and maintain tools for real-time analysis and planning of generating-station operations to comply with intake and outlet water temperature standards. The objective was to be able to forecast intake, outlet, and cooling pond temperatures as accurately as possible over a several-day period; and to test alternative generation and water-management scenarios as a basis for real-time decisions. Simulation tools to meet these objectives had to be capable of acquiring meteorological, thermal, and hydrologic data as automatically as possible, and require minimum expertise and data entry on the part of the user. The resulting modeling system, CS2, is a Java-based Graphical User Interface and data manager wrapping the CHARIMA Fortran-based computational engine (Holly et al, 2000).
Figure 1 shows the approximate location of the four nuclear powerplants included in this project. This paper is focussed on the Braidwood, Dresden, and LaSalle powerplants, all of which are treated in a single modeling system. A separate modeling system developed for the Quad Cities station is described elsewhere (Holly et al, 2000).
For all three
powerplants, condenser circulating water excess heat is dissipated to the
atmosphere during circulation through a cooling lake. In the case of Braidwood
and LaSalle plants, this is essentially a closed-cycle operation with only
minimal riverine discharge (“blowdown”) and corresponding makeup from the
rivers. However the Dresden plant has a relatively small cooling lake, and
therefore is permitted to operate in closed cycle during the winter, but
indirect open cycle in the summer (heated water is routed through the cooling
lake before being discharged to the Illinois River). Dresden is also endowed with both hot-side and cold-side
mechanical draft cooling towers to augment the cooling capacity of the lake. All
three plants are subject to environmental restrictions on the temperature of
blowdown discharges to the adjacent rivers; and to operational and safety
restrictions on the condenser inlet temperatures. The cooling lake performance,
as well as the riverine intake temperatures, are strongly influenced by
transient meterorological conditions as well as by upstream thermal discharges
from other industrial and power-generation facilities.
The purpose of the modeling system is to provide the capability to obtain, on a regular basis, forecast intake and blowdown temperatures over a several-day time horizon, thus enabling plant operators to anticipate the possible need to “derate” (i.e. limit the maximum allowable power level) the units to ensure compliance with appropriate temperature standards. It is very important that the system enable the user to override observed and forecast data (operational, hydrological, and meteorological) to determine forecast sensitivity to uncertainty in driving parameters, on a real-time basis.
Figure 2 shows a topological layout of the one-dimensional
thermo-hydrodynamic model. The cooling lakes are known to be reasonably well
mixed. This fact, along with the need to be able to make month-long simulations
within just a few minutes of cpu time, justifies the one-dimensional
approximation. The model topology includes not only fluvial links to represent
the rivers, canals, and cooling lakes; but also special links for condensers,
weirs, pumps, cooling towers, etc. In one time step, a classic Preissmann-method
solution of the de St. Venant hydrodynamic equations is immediately followed by
a Holly-Preissmann solution of the advection-diffusion-source equations for
transport of heat; in this case the source term is the surface heat exchange
with the atmosphere (including shortwave and longwave radiation, evaporation,
and conduction). Condenser links include empirical relations giving the
temperature rise as a function of flowrate and generation level. Cooling tower
links include semi-empirical relations giving the temperature drop as a function
of tower configuration and atmospheric conditions.
The computational
engine for the CS2 system is the CHARIMA program for one-dimensioanl
simulation of unsteady flow, sediment, and contaminant transport in multiply
connected systems of mobile-bed channels (Holly et al, 1990). The overall model
runs at a time step of 7.5 minutes. A four-week simulation (see below) requires
1-3 minutes of cpu time on an Intel-based PC.
The primary purpose of a CS2 simulation, performed at time “now”, is to forecast water temperatures from “now” forward in time for up to seven days. These future temperatures depend partly on forecast meteorological and hydrological conditions over the forecast period; but they also depend on the state of the system “now. If one thinks of the system from upstream to downstream, the upstream-most locations are immediately affected by forecast conditions starting from “now”. But the more downstream the location, the longer it will take before the current conditions wash out and then cease to influence the forecast temperatures.
The consequence of the above is that the thermal and hydraulic state of the entire system (i.e. temperatures, depths, velocities, discharges, etc. in all channels and lakes) must be initialized for “now”. Since it would be impossible to install enough sensors to define this initial state in the necessary detail, the CS2 program must generate the initial state through its own simulation procedures. CS2 does this by beginning its simulation a full three weeks prior to “now”. During the first few days of the simulation, transient conditions provoked by the arbitrary initial condition gradually propagate out of the system, and the actual observed meteorological conditions, river inflows and temperatures, station operating conditions, etc. progressively establish a coherent thermal-hydrodynamic regime in the system.
This process continues through the entire three weeks preceding “now. During this period, the model is forced to honor actual in situ water temperatures reported by eleven probes at strategic locations in the system through a special procedure (see below). The overall objective is to produce dynamic model conditions “now” that are as close as possible to the actual conditions that cannot be measured, using CHARIMA’s own simulation as a surrogate for a comprehensive but unfeasible detailed measurement network.
The final week of a simulation, i.e. for seven days beginning “now”, comprises a continuing CHARIMA simulation driven by the simulated initial condition “now” as well as by forecast meteorological, hydrological and thermal inflows, and station operating conditions.
The forecast
strategy described above implies that complete meteorological, hydrological, and
plant operational time-series data streams, both for observed and forecast
periods, be immediately available any time CS2 is executed from any
one of numerous locations within and outside the utility. This is accomplished
through automatic acquisition of all relevant data from internet resources.
At regular
intervals throughout each day, a data-management program running at the CS2
host organization (IIHR, the Iowa Institute of Hydroscience and Engineering)
automatically contacts multiple internet sites (hosted by the U.S. Geological
Survey, U.S. Army Corps of Engineers, etc.) to download hydrologic time series
for the observed period. At different intervals, the utility company’s data
operations facility automatically uploads observed and forecast meteorological
data time series, as well as powerplant operational time series, to the host
site. In parallel, the host computer regularly dials up the cell phones mounted
on the in situ data loggers and
downloads the newest temperature data streams.
Whenever the host
data-management program detects that any observed or forecast data streams have
been updated, it executes a series of programs designed to replace missing
and/or anomalous data with generated series, perform hydrologic forecasts of
stream inflow and associated water temperatures (see Bradley et al, 1998), and
add newly acquired in situ data series
to existing series. This procedure, performed several times daily, results in an
automatic archiving of expired data, and uploading of coherent data series to a
dedicated ftp site at IIHR. Whenever CS2 is executed from any
computer with a live internet connection, it downloads all required data from
the IIHR ftp site and performs a forecast simulation as described above.
A particular feature of the CS2 capability is the direct integration of in situ temperature measurements during the observed period of an automatic simulation. Three of the probes measure the station hot-water outlet temperatures at Braidwood, Dresden, and LaSalle Stations, i.e. the temperature of the blend of circulating and service flows entering the cooling lakes at their upstream (hot) sides. During the observed period of an automatic run, the CHARIMA simulation forces the correct heat load into the lakes by honoring these probe temperatures. This is accomplished through the use of so-called virtual powerplants, identifed in Figure 2 as links 652, 3069, and 622, respectively. In CHARIMA, if probe data is available at any of the three particular sites, the program diverts all of the flow through the respective virtual powerplant and computes the effective temperature change across the link such that the probe data will be respected on the downstream side. If the probe data is not available (e.g. during the forecast period), then CHARIMA routes all the flow through the normal channel link, i.e. the virtual powerplant becomes invisible to the computation.
One of the YSI in situ probes is located in the Kankakee River intake of Dresden Station. This probe is used to determine the apparent blend of Des Plaines and Kankakee water, and perhaps Dresden blowdown water, implied by the measured in situ data in each computational time step. When probe data is no longer available (e.g. at the end of the observed period of an automatic simulation), the latest computed blend ratios are presumed to persist for the duration of the remainder of the simulation (normally the forecast period).
Figures 3, 4, 5,
and 6 are from a CS2 forecast executed at 1100 hours on 5 September
2000. This was a particularly challenging period since a period of extremely
warm weather was abruptly terminated by the passage of a cool, dry front on 3
September. This is seen in the observed and forecast air temperatures and
dewpoint temperatures in Figures 3 and 4.
Figures 5 and 6
show selected results from just one of the powerplants; these are typical of
those available for all the plants after a forecast run. Figure 5 shows the time
series of water temperature in the condenser outflow. The temperature prior to
“now” (i.e. 1100 hours on 5 September) is from the in situ water temperature probe, which
the computation is forced to respect as described earlier. The temperature after
“now” is the computed one which the program would have simulated without the
in situ probe forcing; the smoothness
of the transition suggests that the model was doing quite a good job for this
particular period (the transition across “now” is not always so smooth). Figure 6 shows the computed temperature
time series at the cool side of the cooling lake, from which the blowdown to the
Illinois River is taken. This water temperature is well below the critical
values for either the condenser intake or the riverine blowdown.
The first season’s
experience with the CS2 system has given the developers and utility
users a much better sense of the role of changing meteorological conditions in
the performance of the cooling lakes. This has led to a more realistic
appreciation of the limitations of the cooling lakes as a cooling resource; but
also has led to an appreciation of how plant operations can be synchronized with
diurnal weather patterns to optimize the use of the cooling resource. Future
developments will include use of more local (rather than regional) meterological
data, accounting for the effects of wind direction, and systematic studies to
quantify the uncertainty in forecast water temperatures provoked by the
uncertainty in meterological data forecasts.
Holly, F.M. Jr., Bradley, A., Wilson, M., Rocco, B., and Parrish, J. B.
III, (2000), “Thermal Environment Forecast Models for Commonwealth Edison Nuclear Stations”, Limited Distribution Report No. 287, Iowa Institute of Hydraulic Research, September.
Holly, F.M. Jr., Yang, J.C., Schwarz, P., Schaefer, J., Hsu, S.H., and Einhellig, R., (1990),
“CHARIMA - Numerical Simulation of Unsteady Water and Sediment Movement in Multiply
Connected Networks of Mobile-Bed Channels”, IIHR Report No. 343, July.
Bradley, A. A., F.
M. Holly, Jr., W. K. Walker, and S. A. Wright, (1998),
“Estimation of
Water Temperature Exceedance Probabilities Using
Thermo-Hydrodynamic
Modeling”, Journal of the American Water ResourcesAssociation, 34(3), 467-480.
Fig. 1 Nuclear powerplant locations
Fig. 2 Topological layout of thermo-hydrodynamic model
Fig.3 Dry-Bulb air temperature
Fig. 4 Dewpoint temperature
Fig. 5 Condenser outflow temperature
Fig. 6 Condenser intake (and blowdown) temperature