This page is under construction. Some links may not work properly.

Use of Existing Powerplant Instrumentation and Data Logger Makes Index Testing Simple and Economical
DamThree unitsCross section
Aerial view of powerplantOrientation of 3 turbines in powerplantCross-section view of one turbine in the powerplant

Introduction: Benefits of Index Testing

Potential benefits of index testing and optimization of hydroelectric turbines are huge; increased generated energy, extended equipment life and reduced environmental impact provide a full trifecta of benefits from this basic maintenance practice. In this time of concern for the environment, the multi-faceted “green” nature of index testing and optimization as a means to increase energy production must not be overlooked.

Index testing refers to the use of relative flow measurements to characterize the efficiency profile for hydroelectric turbines. When the highest efficiency operating envelopes of the turbines are known, more intelligent unit allocation decisions can be made by Joint Load control schemes. This relative efficiency measurement is most beneficial for double-regulated Kaplan turbines, which have more complex setup procedures than single-regulated impulse and Francis turbines.

Kaplan turbine optimization is the practice of using this relative flow data to define the optimum blade angle under given head and gate setting. If every individual Kaplan turbines is not tuned-up to correct for variations in the turbine’s manufacturing & setting and the powerplant’s instrumentation, peak efficiency operation will not be enjoyed. The new data collection and manipulation methods presented herein can be used for any turbine type, but are the most beneficial for Kaplan turbines. The tools useful for all turbines are:

1. The SteadyState algorithm plucks steady-state data points out of the continuous data stream, thus allowing unattended
    data collection during normal hydroelectric turbine operation, and,
2. A new data graphing method was developed for the Constant Power method data that shows all of the data on a single
   graph. This graphing method could also be used for Francis and impulse turbines.

 Tools specifically useful for Kaplan turbines are:

1.  The Constant Power testing method for Kaplans that can perform a transparent* index test of these machines, and
2.  A topographical data mapping method that aids visualization of the efficiency contour profile and determination of the optimum “on-cam” line at the
     existing net-head.

*(A “transparent” index test refers to the Constant Power method’s ability to sweep the gates and blades back and forth at a single power level, completing about 1/5th of the index test points without disturbing the operators, dispatchers and water management folks. As far as they are all concerned, its all-smooth running while the Constant Power test is running.)

Steps to Optimization

Index testing is only a first step for optimization of hydroelectric power generation. It is strongly recommended that as the first part of any index test that the turbine control system condition and performance gets evaluated in accordance with published industry standards  (ASME PTC-29 and IEEE Std-125 & 1207).

The second step to optimizing a powerplant is to employ Joint Load control strategies that can produce even greater improvements. Joint Load optimization is the practice of using individual turbine efficiency profile data (such as that produced by index testing) in the decision making process for unit allocation and dispatch. Although relative data can provide a fair approximation, for the best results absolute data must be used in Joint Load schemes for a group of turbines; but absolute data is more expensive and difficult to acquire, so what typically happens is a few points are measured concurrently with the relative and absolute methods, and the relative method is corrected to agree with the absolute values then extrapolated across the entire measured spans.

A companion draft article* submitted recently to Hydro Review magazine by Lee Sheldon details a procedure that will synthesize suitable absolute data by merging model test data and relative index test data; this article is currently undergoing technical peer-review. That paper explains how to massage the relative field test data derived by relatively inexpensive index testing methods so it can be incorporated as if were absolute data into a Joint Load scheme.

*(A New Method To Convert Relative Index Test Data Into Absolute Flow And Absolute Efficiency Performance Data by Lee H. Sheldon, P.E., Consulting Engineer.)

These pages will focus on the first step of optimization by presenting new, improved methods for index testing individual hydroelectric units, and provide a few case studies to show the potential payback from these efforts.

Real-World Example

Test setting and conditions

Use of Existing Powerplant Instrumentation and Data Logger Makes Index Testing Simple and Economical

This is the story of optimization efforts in a powerplant with a trio of 18MW Kaplan bulb turbines that are situated in parallel.  Maximum power output of all three units is 54 MW at a design-head 5.7 meters gross head with a total flow of 1,056 cubic meters/second. Water flow in this "run of the river" setting is normally limited; average Gross Head in the most recent data is 6.2 meters with a total flow of 560 cubic meters/second producing a total 31 MW of electrical generation.

These units run mostly during peak periods during the day when bulk energy rates are high. To reduce repetitive thermal stressing, the units are not shut down overnight; they are operated at lower power levels to keep generator temperatures consistent.  As is the norm in most hydropower installations, limited water availability increases the need to do everything possible to maximize generation and revenue.

Prior Index Testing

One of these turbines was index tested when the powerplant was commissioned in 1985, but the new cam surface was not installed in the machine at that time.  Kaplan blade cams were two-dimensional metal cams, so this cam surface upgrade would have required new metal gate-to-blade cams (at least 4 cams for each of three turbines) to be cut and installed on the governor’s gate restoring shafts. This final step (installing the new nethead & gate to blade surface data) of the index testing and optimization was not completed, so the benefits of this index test were delayed for 20 years.

After 20 years of running time, a governor upgrade was installed in 2005, at which time the original index test report was located and the new 3-Dimensional cam surface was installed in the Kaplan turbine 3-D cam as part of the upgrade.  The new digital 3-D Cam and blade control system made this cam surface update easy and quick. Before-and-after spot-checks of unit performance found an 8% increase in power at the same head and flow, which justified upgrading the two other turbine’s control systems right away.

In preparation for these upgrades, another index test was conducted in 2006; the test report stated that the machine was operating close to the optimum, but some small improvements could still be achieved.

When the control systems for the other two machines were upgraded in 2007 and the same 3-D cam profile was installed into them, a comparable power increase was not seen - much to the consternation of the powerplant personnel. Field-testing of hydroelectric turbines using classical methods is expensive, tedious and time-consuming, so they were reluctant to purchase more index tests on these machines.

In 2008, we started working on a new method to instrument and index test these three machines as an integral function of the governor and control system in a cooperative effort between the powerplant, the governor supplier and Actuation Test Equipment Company.

New Tools Facilitate New Methods

The existing instrumentation at this powerplant is very comprehensive and of high quality, well suited for efficiency measurements and optimization work. Modern computer based turbine control systems, instrumentation and data recorders make data collection in the field simple and economical, the remaining problem is getting the “steady-state” data needed for efficiency performance analysis; the most vexing and labor-intensive part of index testing is establishing and holding the hydroelectric unit at steady-state operating conditions while these measurements are taken. To implement this new method, a few signals were added to the station data logger and the scan rate was increased from every 5-seconds to every second. The station datalogger already had most of the signals needed. SetPoint signals for gates and blades were added to the data logger so we could checkout the control system. Trashrack transducers were added so we could watch for accumulated debris and a blade-offset signal was added to the 3-D cam and blade controller.

The governor supplier added a means to offset the blades from the “on-cam” position using the governor HMI (Human Machine Interface) or from the control room.

When the control systems for the other two machines were upgraded in 2007 and the same 3-D cam profile was installed into them, a comparable power increase was not seen - much to the consternation of the powerplant personnel. Field-testing of hydroelectric turbines using classical methods is expensive, tedious and time-consuming, so they were reluctant to purchase more index tests on these machines. 

In 2008, we started working on a new method to instrument and index test these three machines as an integral function of the governor and control system in a cooperative effort between the powerplant, the governor supplier and Actuation Test Equipment Company.

New Tools Facilitate New Methods

The existing instrumentation at this powerplant is very comprehensive and of high quality, well suited for efficiency measurements and optimization work. Modern computer based turbine control systems, instrumentation and data recorders make data collection in the field simple and economical, the remaining problem is getting the “steady-state” data needed for efficiency performance analysis; the most vexing and labor-intensive part of index testing is establishing and holding the hydroelectric unit at steady-state operating conditions while these measurements are taken. To implement this new method, a few signals were added to the station data logger and the scan rate was increased from every 5-seconds to every second. 

The governor supplier added a means to offset the blades from the “on-cam” position using the governor HMI or from the control room.

The advent of OPC communications for SCADA system control in powerplants creates a means to access unit operating data easily. By using the OPC method to gather the data - it is just as easy to record operating values from all three units at the same time as it is to record data for just one. This leads to the penultimate improvement – the ability to simultaneously recorded data for all three units, allowing us to index-test all three units at the same time. All three units are being efficiency tested at the same time because every single “steady-state data point” is a valid efficiency point for all three machines.

Using a new SteadyState software algorithm and PostProcessor data analysis tools, steady state data points are gleaned from the continuous data stream emanating directly from the turbine control systems and manipulated to sort out the new optimum cam surface. SteadyState points can also be plucked from the pre-recorded “canned’ field data captured by the powerplant’s data recording equipment. We’re getting better data now than has ever been acquired from these machines before.

The SteadyState software has a two-pass signal processing technique that applies limits to data sampling at both high (0-100 Hz, which is high frequency for hydro) and super-low frequency. The high frequency sampling is looking for acoustic signatures of vibrations or cavitation, the low frequency looks for very slow signals, such as wave action in long approach canals and flumes. An example of a powerplant with a mile long approach canal is included below in this text.

Another plus was that powerplant personnel were able to add the channels and speed up the scan rate of the data recorder themselves. This saves money on the index test setup and gets the plant personnel more involved in the index testing process.

Because the plant personnel did all of the instrumentation setup and calibration, they have a higher confidence level in the results. Bonus.

Test Procedures

The only recurring tasks for the powerplant personnel is to maintain NIST traceability on all transducers and sensors, to periodically (once or twice a day) flush the pressure transducer sense lines to remove any accumulation of silt and/or air bubbles, and to mail the thumb-drives to me for analysis every week. To get the off-cam data, powerplant personnel use the blade-offset control feature added to the turbine control system to move the blades off-cam a few percent once day. The units run for 24 hours and remove a thumb drive once a week and mail it to me so that the data can be analyzed off-line, here in my shop.

Testing “On the Governor,” Articulation Methods for Kaplans

Kaplan turbine variable-pitch blades provide an added degree of freedom that must be setup and tuned correctly for maximum efficiency. If these blades are not continuously positioned at the optimum angle, Kaplan turbines are less efficient and create greater harm to the environment. Index testing and optimization of each and every Kaplan turbine and accompanying rigorous checkouts of the governor & blade control systems in accordance with published industry standards is essential to satisfactory performance.

 During many “classical” method index tests the blades are decoupled from the gates and positioned manually; the gates are also positioned manually with the Gate Limit control on the governor cabinet, even for the “before” On-Cam test run. This practice allows for more precise positioning of the gates and blades for the index tests, but overlooks potential problems that may exist with the governor, 3-D cam and blade controller. Any problems that exist within the control system would remain undiagnosed and uncorrected, problems, which may well defeat the entire purpose of index testing and optimization.

By collecting the unit performance data with the system data logger as a routine part of normal powerplant operation (as described above) the data is representative of what the machine is actually doing, instead of how the test personnel arbitrarily positioned the gates and blades as dictated by the index-test plan. To be fair, some index test reports we have seen have the on-cam line prescribed while operating “on the governor,” and on some tests large & small loop hysteresis plots have been taken to check for slop and wear in the gate and blade servos and linkages, but many index testing procedures overlook these critical items.