Untitled Document

Hydro Review

Asset Management: Using ADCP-Based Flow Monitoring to Improve Operations at Niagara

Installation of a flow measurement system based on acoustic Doppler current profiler technology allowed the New York Power Authority to improve operational control of its Robert Moses and Lewiston hydroelectric plants.

By Patrick A. March, Paul J. Wolff, Jiankang Zhu, Richard Fremming and Thomas Key

The New York Power Authority (NYPA) has implemented an acoustic Doppler current profiler (ADCP) flow measurement system for near real-time flow measurements in the large-scale, diverging transition at its 2,755-MW Niagara Project. These flow measurements are improving operational control of the closely coupled 2,515-MW Robert Moses Niagara Power Plant and 240-MW Lewiston Pump Generating Plant, which together form the Niagara Project. The ADCP system is providing insights into control system enhancements with the potential to increase overall operational efficiencies by up to 1.5% at Robert Moses and up to 3.4% at Lewiston.


The Robert Moses plant contains 13 Francis turbine-generator units, and Lewiston has 12 reversible pump-turbine units. The two intakes for the Niagara Project are on the Niagara River, upstream from Niagara Falls. Water flows by gravity through two underground conduits into a "pantlegs" transition, then into the forebay of Robert Moses (see Figure 1). The Lewiston facility pumps water from the Robert Moses forebay into the Lewiston reservoir or generates power, returning water to the Robert Moses forebay. The Lewiston facility is operated to balance diversion flows from the Niagara River and to provide on-peak generation and storage for future generation.

The Niagara Project operates in a highly complex manner, due to:

- Operation of the Robert Moses plant in automatic generation control (AGC);

- Dynamic interactions between the plants;

- Robert Moses forebay hydraulics;

- International treaty obligations between the U.S. and Canada for diversion flows;

- Transmission system requirements; and

- Market constraints.

Accurate, timely knowledge of the diversion flow rate is vital for proper operational control.

Results from five methods for determining the diversion flow rate for the Niagara Project have been compared. The methods are:

- Acoustic Doppler current profiler;

- Waterways Coefficient method using the forebay level measurement at Robert Moses, river level measurement at the intakes, and conduit head loss relationship developed by NYPA;

- Waterways Coefficient method using forebay level from the ADCP, river level at the intakes, and a modified version of the NYPA conduit head loss relationship;

- Volumetric method using the Robert Moses unit characteristics for Robert Moses flows, Lewiston upper reservoir volumetric method for Lewiston flows, and Robert Moses forebay level measurement for forebay volumes; and

- Volumetric method using the Robert Moses unit characteristics for Robert Moses flows, derived Lewiston unit characteristics1,2,3 for Lewiston inlet (i.e., generating) and exit (i.e., pumping) flows, and Robert Moses forebay level measurement for forebay volumes.

Comparisons were conducted using data from November 2009, April to May 2010, and June to July 2010. These comparisons provide insight into current operations and potential improvements.

Diversion flow measurement with ADCP

An ADCP-based flow monitoring system was installed in the large-scale diversion channel in 2008 to measure the diversion flows. At the measurement site, the diversion channel is 250 feet wide, and the water depth varies from 40 to 60 feet. The ADCP is a SonTek SL unit operating at 500 KHz. There are three acoustic transducers, two larger ones for the Doppler current profiling function and a smaller one for level measurement. The SonTek SL unit also incorporates a pressure transducer to ensure the proper range for the acoustic level measurement. Figure 2 shows the location of the access platform installed for the ADCP.

Figure 3 provides a simplified view of the ADCP. The flow monitoring system was calibrated by SonTek using a towed, vertically oriented ADCP downstream from the access platform. The manufacturer's documentation provides details on the ADCP and its calibration.

Description of the Waterways Coefficient method

Details about the development of the Lewiston unit characteristics for determining Lewiston flows are available.1,2,3 A limited discussion of the Waterways Coefficient method is provided below.

The diversion flow rate can be computed using the Waterways Coefficient method by assuming constant hydraulic characteristics for the conduits and by using the energy equation to relate the flow rate through the conduits to the head change between the conduits' entrance and exit. The hydraulic characteristics developed by NYPA are presented as:

Equation 1

QHL1 = 1000 x (K x (h2 - h1 - ~1))0.5


- QHL1 is diversion flow rate in cfs;

- K is the Waterways Coefficient;

- h1 is the Robert Moses forebay level in feet measured at Robert Moses; and

- h2 is the Grass Island Pool water level in feet measured at the intake to the diversion conduits.

Plant and diversion flows

Figure 4 shows the water levels measured at the Robert Moses forebay and by the ADCP. The ADCP water level is typically about 0.5 feet lower than the Robert Moses forebay level because of the higher velocities in the transition section where the ADCP is located. Occasional dropout in the level signals for the ADCP and the corresponding effect on the ADCP flow measurements is apparent. Figure 4 also shows the diversion flows measured using the Waterways Coefficient method and the ADCP. With the exception of the ADCP flows adversely affected by the signal dropout, the two diversion flows agree closely. However, when the diversion flow increases or decreases rapidly, the ADCP flows respond more slowly than the Waterways Coefficient flows because of the ADCP averaging time.

ADCP flow and level data with 5-minute averaging times were compared with Niagara Project data from April 29, 2010, to May 19, 2010. Results were similar to the November 2009 results.4 ADCP flow and level data with 2-minute averaging times were also compared with Niagara Project data from June 21 to June 28, 2010, and from July 15 to July 20, 2010. Much of the time delay observed in the ADCP diversion flows4 was eliminated with the 2-minute averaging time.

ADCP flow and level data with 1-minute averaging times were also compared with Niagara Project data for a 50-minute period on July 15, 2010. With the 1-minute averaging time, the ADCP flows scatter by about +/- 10% around the Waterways Coefficient flows. Oscillations with a 5-minute period, which is characteristic of the Robert Moses forebay oscillations, were observed in the Robert Moses forebay level measurement and ADCP level measurement. The ADCP level measurements are out of phase with the Robert Moses forebay measurements, which indicates the oscillations result from a forebay "slosh" mode.

The ADCP level measurements are closer to the exit of the diversion conduits and levels show an average difference of 0.45 foot compared to the Robert Moses forebay level measurement used for the original Waterways Coefficient flow, which assumes an average difference of 1 foot. Subsequently, radar-type level transducers were installed at Lewiston, and a period of comparison data was used to determine an improved Waterways Coefficient and level correction factor for Equation 1.5

Oscillations in power and flow

Insights can be gained by plotting the Robert Moses plant flows (based on unit characteristics from efficiency testing), Lewiston plant flows (based on derived unit characteristics), diversion flows (based on the ADCP measurements and improved Waterways Coefficient method), and diversion flow targets (see Figure 5). The Lewiston and Robert Moses flows show significant oscillations that typically appear to be initiated on the hour. Corresponding oscillations were also observed in the Robert Moses plant power, Robert Moses unit powers, Lewiston plant power and Lewiston unit powers. Similar oscillations were observed in analysis results from November 2009 but not in analysis results from August 2007, before the supervisory control and data acquisition (SCADA) system upgrade.

The corresponding Lewiston efficiency losses during generating were analyzed for the 8-day period from June 21 to June 28, 2010. Based on these analyses, the potential efficiency improvement in generating mode for Lewiston is 3.4%. The Robert Moses efficiency losses also were analyzed for the 8-day period from June 21 to June 28, 2010. The potential efficiency improvement for Robert Moses is 1.5%. In addition to the efficiency losses, the flow and power oscillations correspond to increased wicket gate oscillations and the potential for increased wear and tear on the Lewiston and Robert Moses units.

Niagara simulator-controller

A review of the flow comparisons and data analyses identified significant opportunities to manage diversion flows with reduced flow and power oscillations. A conceptual control mode, including a control simulator, was developed and used with actual Niagara Project data to predict project operations.

The Niagara simulator-controller:

1) Reads the ISO setpoint and sets the Robert Moses plant power so that it and Lewiston together meet that setpoint;

2) Computes the Robert Moses unit commitment, unit flow and total plant flow for the plant power computed in the previous step;

3) Computes the Lewiston flow (positive or negative) that balances the Robert Moses outflow and target diversion inflow;

4) Computes the Robert Moses forebay elevation that would produce the target diversion flow;

5) Computes the difference between the target and actual forebay elevation;

6) Uses the forebay elevation difference in a PID control algorithm to compute a correction to the Lewiston flow;

7) Computes the Lewiston unit flow corresponding to the plant flow;

8) Computes the Lewiston unit and plant powers;

9) Recalculates the Robert Moses power (i.e., the current ISO setpoint minus the Lewiston power); and

10) Returns to step 2 unless the new value for Robert Moses power equals the previous value for Robert Moses power.

The Niagara simulator-controller also computes the Robert Moses forebay level versus time given the simulated plant operation. A maximum unit ramp rate of 10 MW/min is assumed for Lewiston, and a maximum unit ramp rate of 30 MW/min is assumed for Robert Moses, based on a review of the operating data.

Figure 6 provides typical flow and water level results from the predictive controls simulation, using June 2010 data. The diversion flows computed follow the diversion target closely compared to the flow excursions in the actual diversion flow (i.e., the Waterways Coefficient flow). Similarly, the Lewiston flows computed vary smoothly, while the actual Lewiston flows oscillate considerably. The Robert Moses forebay water levels computed vary smoothly, while the actual Robert Moses forebay water levels vary considerably.

Results were also computed for Niagara Project power deviations (i.e., ISO setpoint minus total Niagara Project power) and diversion flow deviations (i.e., diversion flow target minus Waterways Coefficient flow), using June 2010 data and the predictive controls simulator. The diversion flow deviations are significantly lower for the predictive controls simulation compared to the actual diversion flows. The Niagara Project power deviations are slightly lower for the power values from the predictive controls simulation compared to the actual power values.

Conclusions and recommendations

This article presents results from multiple methods for determining diversion flow rate for the Niagara Project. The flow comparisons provide insight into current operations and potential improvements. Significant conclusions and recommendations are:

- For the range of flows in the comparison data, the Waterways Coefficient is an accurate method for determining diversion flows through the conduits. By using the ADCP for periodic calibration checks, a more accurate Waterways Coefficient can be obtained, improving control of the diversion flows.

- Further improvements for tracking the Lewiston unit flows can be obtained through improved forebay and reservoir level measurements and Lewiston unit characteristics, similar to the computation method for Robert Moses unit flows.

- Subsequent to the SCADA system upgrade, flow and power oscillations occur in both Robert Moses and Lewiston, especially in automatic mode during times of change in power levels and diversion flows. These oscillations could shorten the useful life of the units and increase unit downtime and maintenance costs. In addition, the oscillations cause the units to operate less efficiently, resulting in wasted water and lost revenue opportunity.

- Significant value can be obtained by improvements to the control system that allow simultaneous, integrated control of power levels and diversion flows. The Niagara simulator-controller provides a useful tool for testing control methods and refining control algorithms. Based on results, improvements to the Niagara Project control system are under way.


1Wolff, P.J., and P.A. March, Extraction of Lewiston Unit Characteristics from Operational Data, New York Power Authority's Niagara Project, Hydro Performance Processes, Nashville, December 2008.

2Wolff, P.J., and P.A. March, Improved Lewiston Unit Characteristics from Operational Data, New York Power Authority's Niagara Project, Hydro Performance Processes, Nashville, March 2010.

3Wolff, P.J., P.A. March, T. Key, and J. Zhu, "Using Operational Data to Determine Pumped-Storage Unit Characteristics in Generating Mode and Pumping Mode," Proceedings of HydroVision 2010, PennWell Corporation, Tulsa, Okla., 2010.

4March, P.A., and P.J. Wolff, Diversion Flow Comparisons, Hydro Performance Processes, Nashville, March 2010.

5March, P.A., Level and Diversion Flow Comparisons, New York Power Authority's Niagara Project, Hydro Performance Processes, Nashville, December 2010.

Patrick March is president and principal consultant with consulting firm Hydro Performance Processes Inc. Paul Wolff, PhD, is president of WolffWare Ltd., which creates software applications for automating data analysis at hydro plants. Jiankang Zhu, PhD, is a research and technology development engineer and Richard Fremming is senior operations engineer with the New York Power Authority. Thomas Key is technical executive for power delivery and utilization with the Electric Power Research Institute.

This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.

Hydro Review Past Issues

Issue 5


Volume 34
Issue 5
Issue 4


Volume 34
Issue 4
Issue 3


Volume 34
Issue 3
Issue 2


Volume 34
Issue 2
Issue 1


Volume 34
Issue 1

Related Articles

Kim Ogren: Graduating Researcher of the Hydro Research Foundation

Kim Ogren is a PhD candidate in Geography at Oregon State University in Corvallis, Ore., and...

Manitoba Hydro CEO Scott Thomson stepping down

Manitoba Hydro's Scott A. Thomson has announced he will step down as president and CEO of th...

Manitoba Hydro again extends call for turbine-generators at Canada's 78-MW Pointe du Bois hydro project

Canadian utility Manitoba Hydro again extended the deadline for proposals to design, supply ...

BC Hydro seeks wildlife mitigation, monitoring for Canada's 1,100-MW Peace River Site C hydro project

Canadian utility BC Hydro seeks bids to provide wildlife mitigation and monitoring services ...

Recent Comments

Hydro Slideshows

more slideshows >>  

Hydro Whitepapers

Industry News

Editor's Picks

Volume 34, Issue 5
Volume 23, Issue 3

Buyers Guide Products