Save Article Instructions

Using Data from Instruments to Ensure Safety

At its 1,050-MW Boundary project, Seattle City Light collects data from monitoring instruments installed in the dam. To improve analysis of this data, the utility is establishing normal ranges for readings from instruments. Any deviation from these normal readings can be investigated and the problem corrected before it becomes a serious safety concern.

By Walter L. Davis and Robin G. Charlwood

The goal of a surveillance and monitoring program is to ensure continued safe performance of a dam. Data from instruments installed in or on the dam are used to determine whether structures are performing as expected as well as to warn of developing conditions that could lead to safety concerns.

At Boundary Dam, which impounds water for the 1,050-MW Boundary hydro project, Seattle City Light has established normal (threshold) ranges for readings from key instruments, such as plumblines to measure dam and foundation deflections. These ranges help guide the review of this instrumentation data by engineers and technicians and increase the utility’s ability to identify potential anomalies in time to take effective actions. These actions could involve obtaining specialized expertise, implementing repairs, or even lowering the reservoir.

Construction and instrumentation of Boundary Dam

Boundary Dam is a 340-foot-high concrete arch dam in a narrow canyon of the Pend Oreille River, in the northeastern corner of Washington. The dam, which was completed in 1967, has a crest length of 508 feet, not including the spillways on either side of the dam.

Designers of Boundary Dam needed a way to quickly lower the water level if working lead, silver, and gold mines under the reservoir began to flood. For this purpose, designers included seven large sluice gates through the body of the dam, 190 feet below the dam crest.

Initially, the dam was designed as three horizontal structures separated by sliding joints. However, structural analyses performed by the original dam designers indicated that differential deflections would not always provide enough force to actually make the joints slide. As a result, the revised design consisted of one continuous structure with short transitions between the components.

A second challenge in designing the dam was the variable stiffness (modulus of elasticity) of the inter-bedded dolomite and limestone rock that would serve as the foundation. The stiffness depends on the presence and orientation of natural jointing features in the rock. There are softer zones of rock at mid height on the western (or left, when looking downstream) side of the dam and near the top of the eastern side of the dam. These softer zones do not accept as much load from the concrete as the adjacent harder rock. To overcome this problem, designers arranged the dam to partially bridge these softer zones. Thus, even though the valley is symmetrical, the stresses in the dam are not.

Boundary Dam was keyed deeply into the abutment rock to obtain a good transfer of load and resistance to sliding along the contacts. During construction, the civil contractor filled an erosion feature just downstream on the left abutment with concrete. Both curtain and consolidation grouting of the dam foundation were performed. Grout takes were generally low, except for a few isolated zones high in the abutments. No drains were drilled from the dam foundation gallery during construction, which is a common practice for thin arch dams.

Dam monitoring for the first 20 years

Strainmeters, thermometers, and deflection surveys were used to monitor structural performance of the dam during construction and first filling of the reservoir. The behavior of the foundation rock was monitored using flat jacks, deformeters, and joint movement pins. Data from these instruments confirmed that the dam and foundation were performing as expected. Measured deflections (and therefore stresses) matched well with design predictions.

Once the project began operating, Seattle City Light personnel continued to monitor instruments to check for any changes. About eight years after reservoir filling was completed, Seattle City Light personnel noticed cracking on the downstream face near the transition between the arch ring containing the sluice gates and the more flexible upper portion of the dam. An independent consultant hired to analyze the dam in 1977 attributed this cracking to the unique geometry of the dam and did not consider it a safety concern.

In addition, inclined cracks more or less perpendicular to the contact between the dam and foundation rock were visible on the downstream face. This is typical of the type of cracking that occurs in arch dams built in narrow valleys. It is not a safety concern in arch dams because there are no significant transverse shear forces applied along the plane of the cracks.

In the mid-1980s, some 20 years after construction of Boundary Dam was complete, Seattle City Light personnel performing routine surveillance and monitoring noticed slight changes in the cracking pattern. One area of diagonal cracking extended further than usual. This crack extended from the left foundation just above the sluice gates, across four vertical cantilevers, and diagonally upward toward the center of the dam. PRC Engineering and International Engineering Company conducted three-dimensional (3D) finite element model structural analyses, which showed that the dam was safe and stable even for the largest floods and earthquakes – and even if the cracks went through the dam.

In accordance with Federal Energy Regulatory Commission (FERC) dam safety requirements, Boundary Dam undergoes inspection and review by an independent consultant every five years. In 1987, the independent consultant concluded that the dam was safe and stable, even with the observed cracking. However, the consultant recommended that Seattle City Light install multiple point borehole extensometers to better understand what was causing the changes. The extensometers, supplied by Geokon, Inc., of Lebanon N.H., were installed by HWA Geosciences Inc. (formerly Hong West Associates) in 1990. The goal was to answer specific dam safety questions: “Is the cracking in the dam associated with small inelastic creep of the abutment rock? Although there are no kinematically capable rock wedges in the foundation, could cyclic annual temperature or daily reservoir loads be closing joints in the rock over time?”

If it is geometrically possible for a rock wedge to move downstream and daylight into the air without shearing through large sections of intact rock, it is considered “kinematically capable.” Such rock wedges need to be analyzed to make sure there is an adequate safety factor against sliding for usual, flood, and earthquake loading conditions. Because arch dams transmit a large share of the weight of the reservoir to the abutments by compression, it is critical that the foundation always stays put and accepts the load. Comprehensive geologic investigations and analyses performed by dam designer Bechtel Leedshill during the original design of the dam and by MK Engineers in 1993 show no such wedges exist in the dam foundation.

Improving monitoring systems

In November 1990, at FERC’s direction, Seattle City Light retained a board of consultants to review the cracking in the dam, the potential for abutment movement, dam stability, and monitored performance. The four-member board met six times between 1990 and 1998. The board recommended three rehabilitation measures, which Seattle City Light implemented between 1992 and 1994:

– Drill foundation drains to lower water pressure;

– Epoxy grout the cracking in the dam to reduce seepage; and

– Secure a wedge of rock underlying the left spillway (which supports a portion of the spillway apron and provides confinement for the dam foundation) with 13 post-tensioned anchors to resist earthquake loads.

The board also recommended that dam body and foundation deflections be monitored with new technology to obtain a “better, more frequent, more precise, instantaneous, and verifiable overview of the dam’s behavior” than available from ground surveys. Seattle City Light undertook several measures to implement this recommendation. First, the utility discontinued monitoring of instruments that had deteriorated or that were no longer providing useful information. Second, in 1991, Seattle City Light installed additional pneumatic piezometers provided by Slope Indicator (now Durham Geo Slope Indicator) in the left abutment. Third, in 1994, the utility mounted an array of 20 tiltmeters supplied by Applied Geomechanics Inc. in Santa Cruz, Calif., on the dam and connected them to a data logger from Campbell Scientific Inc. Finally, in 1994, Seattle City Light installed two inverted plumblines supplied by Hydro-Québec 270 feet deep into the foundation on either side of the dam.

Boundary Dam is located in a narrow canyon, making it subject to cyclic deflections based on reservoir level and temperature. Numerous instruments installed in the dam measure deflections and behavior under various conditions.
Click here to enlarge image

Because of the timing of the special board of consultants’ review, the next five-year inspection of the dam was not conducted until 1995. During this inspection, the independent consultant expressed concerns about slight trends in the monitored data at the dam. As a result, in 1998, Seattle City Light engaged Acres International (now Hatch Acres), together with Shannon & Wilson Inc. and several other specialist consultants, to carry out a detailed review of all project instrumentation.

The review included project design and construction records, all available monitoring data, data gathering procedures, and procedures to take the raw data and convert it to easily understood graphs and charts for use by engineers in analyzing structural performance. The consultants evaluated the accuracy and precision of all ground surveys, foundation piezometers, drains, foundation extensometers, foundation deformeters, embedded strainmeters, tiltmeter arrays, crack monitors, foundation inverted plumblines, strong motion accelerometers, and thermometers to measure air, reservoir, and concrete temperatures. They also evaluated the ability of these instruments to function as intended.

Next, the consultants reviewed existing project geology and rock mechanics reports and previous structural analyses, to validate the need for each instrument.

Finally, they developed normal (threshold) values for dam deflection measurements from the tiltmeter array.

Over time, the tiltmeters at Boundary Dam experienced drift and required periodic calibration in difficult-to-reach mounting locations. To overcome this problem, in 2003 the team of consultants designed a “real time” deflection monitoring system consisting of seven plumblines supplied by Geokon, Lebanon, N.H. Plumblines are more capable of monitoring for small long-term trends, which is important for this application. Shannon & Wilson installed these plumblines on the dam in 2003. Seattle City Light uses RxTx optoelectronic readout devices available through Roctest Ltd. or RST Instruments Ltd. for remote continuous monitoring of the plumblines. These readout devices have been in service at Boundary Dam since 1994 to monitor the foundation plumblines and have proved very reliable.

In 2003, an automatic data acquisition system (ADAS) supplied by Geomation, Inc. of Golden, Colo., was installed to collect real-time data from the dam and foundation plumblines, foundation extensometers, and thermometers to measure concrete temperature. Shannon & Wilson custom-designed an integrated data management system to manage data, generate automatic engineering plots, apply normal values, and facilitate Seattle City Light’s reporting of instrument data and analysis to FERC. This system uses Grapher by Golden Software to produce time history plots.

In its final reports released in 1994 and 1998, the board concluded that the dam and foundation are safe and stable for all loading conditions, with no evidence of long-term foundation creep. Board members suggested the left abutment cracking might have been caused by differential cooling of the concrete and foundation after construction. They also re-emphasized that vigilant surveillance and monitoring of Boundary Dam should continue, as with all major dams, for the life of the project.

In 2003, Seattle City Light installed seven plumblines on Boundary Dam to monitor small long-term trends in deflection of the dam.
Click here to enlarge image

To establish priorities and improve the focus of the monitoring program, the board grouped the instruments at Boundary Dam into five categories:

– Primary (detector) instruments need to detect any anomaly in the dam behavior, no matter where it takes place and no matter what its causes. Deflections of the dam and foundation, along with foundation water pressures and seepage rates, are the directly measurable physical quantities that can serve this purpose. These instruments need to be read frequently and interpreted immediately.

– Interpretation and control instruments are for calibration of the primary instruments and for understanding loading conditions. The two loading conditions of importance to Boundary Dam are the reservoir level and dam concrete temperature.

– Support instruments can be used to help explain or interpret problems. These instruments are read only as needed to keep staff familiar with reading procedures and instrument maintenance.

– Special instruments installed for a specific purpose, such as the strong motion accelerometers to record earthquake effects.

– Instruments to be retired because they no longer work properly or serve a useful purpose.

Setting normal (threshold) levels

Normal values are essential to help identify the early development of potential failure modes at Boundary Dam, against a background of significant cyclic movements driven by temperatures and reservoir level. The summer’s heat expands the dam and increases its compression against the foundation, causing it to deflect upstream. In the winter, the dam contracts and moves downstream. The maximum cycling between winter and summer is about 1 inch (radial direction) at the top center of the dam.

The Boundary project is operated in a load-following mode that shapes available water to deliver power during peak-load hours. The lake is drawn down by about 10 feet during the day and refills at night. Lowering the lake 10 feet unloads the dam and foundation and results in an upstream (radial) deflection at the center top of the dam of a little more than 0.1 inch. In the winter, when Seattle’s electricity demands are highest, the lake may cycle as much as 20 feet a day. Lowering the lake 20 feet results in movement in the upstream direction of about 0.2 inch.

In 1987, an independent consultant developed normal values for Boundary Dam instrumentation. This work involved preparing time history graphs for the entire period of record for each instrument. Where seasonal cycles were apparent in the data, Seattle City Light personnel took monthly readings from these instruments over a period of a year to better define the seasonal shape. The consultant then plotted trend bands about one-third of the magnitude of the annual cycling above and below the average value. These trend bands were superimposed over the period of record and were used as benchmarks to identify any small long-term trends. In 1995, the independent consultant noted some small long-term trends in the ground survey deflection data and the new foundation plumbline data.

In the 1990s, MK Engineers developed a more analytical method of calculating threshold levels. The company built a finite element computer model of the dam to predict dam deflections for various reservoir levels and temperature schemes. The output was then used to develop influence functions (third order polynomial equations selected to fit the output data) for various points on the dam. By plugging the reservoir elevation and measured concrete temperatures for a specific time into these third order polynomial equations, personnel could calculate an expected deflection for that point on the dam. Extreme Access, Vancouver, British Columbia, installed tiltmeters at these points on the dam to measure actual deflections and compare them to the theoretical values. However, because of difficulties in reducing the tiltmeter data into a useful form, this method did not work consistently.

Figure 1: Daily readings from plumblines installed on the foundation and body of Boundary Dam indicate the dam and foundation are performing as expected as reservoir elevations change during cold weather.
Click here to enlarge image

After the plumblines were installed in 2003, the concept of comparing actual and theoretical deflections was further developed by the team of Shannon & Wilson and Acres. Acres’ structural engineers established specific threshold levels for measurements of dam movement for both normal loading due to seasonal temperature and reservoir changes and for extreme seismic events. The threshold values chosen for normal dam displacements include an allowance for the approximations in the expected displacement predictions, measurement precision and accuracy, and unmeasured influences of uncertainties in the concrete temperature and reservoir level.

Threshold values are being developed in two phases. In 2000, initial threshold values were calculated to aid in design of the plumbline monitoring system and for the initial years of operation of this system. These values are plus or minus 0.3 inch for the top center of the dam, or about the same as the previously estimated trend bands.

With the benefit of several years of data, a new correlation system is being developed to assist in filtering out the secondary effects of reservoir elevation and temperature. This will allow for a finer threshold level, about plus or minus 0.1 inch for the top center of the dam.

In 2005, the independent consultant recommended that thresholds also be established for the foundation monitoring instruments (multiple point borehole extensometers, plumblines, piezometers, and seepage rates) based on the previous five years of data. These thresholds will be similar to the trend bands previously set for dam deflections, as there is no practical way to easily predict foundation responses to changes in loads using the existing modeling.

A measured displacement that is outside the threshold band is interpreted as probably indicating a real change in the behavior of the dam, until and unless it can be shown to be something else. Such a reading might be due to an instrumentation problem or a transient “unusual” or extreme loading condition. But it also could reflect some real change in the structure that requires immediate action. Therefore, while a reading that exceeds a threshold values does not indicate an immediate dam safety problem, it does require that the cause of the reading be quickly investigated and understood. Actions include rereads, review by dam safety engineers, correlating readings with other instruments, and increasing the frequency of readings.

Results to date

Plumbline 4 monitors deflections of the thin upper part of the dam, at the right quarter point of the crest. In general, the dam deflects upstream in the summer and downstream in the winter. Data from this plumbline indicated there are three excursions beyond the threshold bands, all of which occurred during extremely cold weather. The normal average early January temperature at the project is about 25 degrees Fahrenheit, but temperatures can fall below zero. The thin upper sections of the dam and outer few feet of the foundation respond quickly to unusually cold weather. An abrupt downstream deflection as the concrete cools and contracts during cold snaps is about one-third of the seasonal cycling. This is a significant effect that needs to be considered and understood in reviewing the data.

Figure 1 shows the deflections of plumbline 4 on the dam body and plumbline 1 in the foundation during a typical week of reservoir cycling during the extreme cold weather noted above. The amplitudes of measured deflections of the dam are in good agreement with predicted values for a 10-foot lowering of the reservoir. Note that even the plumbline in the abutment senses the cycling load, as the pressure the dam imparts on the foundation rock changes. Even though the extreme cold weather results in out-of-threshold readings, the dam and foundation are behaving as expected. There is no indication of a dam safety problem.

Mr. Davis may be reached at Seattle City Light, P.O. Box 34023, Seattle, WA 98124; (1) 206-684-3657; E-mail: Dr. Charlwood may be reached at Robin Charlwood and Associates, PLC, 1001 Maple Street, Edmonds, WA 98020; (1) 425-712-1750; E-mail: robin@


Kuperman, Selmo C., M. Regina Moretti, Julio C. Pinfari, and Edvaldo F. Carneiro, “Placing ‘Limit Values’ on Instrument Readings,” HRW, Volume 15, No. 3, July 2007, pages 24-31.

Walt Davis, P.E., is dam safety supervisor for Seattle City Light. Robin Charlwood, PhD, P.E., principal of Robin Charlwood and Associates, led the safety review of Boundary Dam.

To access this Article, go to: