Developing a Quantitative Method for Assessing Plant Conditions

The authors have developed a quantitative rating process and standardized methodology for determining the condition of a hydropower plant by assessing parts, components, units and overall plants.

By Qin Fen Zhang and Brennan T. Smith

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.

Many hydropower plants in the U.S., particularly the larger ones, were constructed decades ago. As the physical condition of the generating units in these facilities deteriorates over time, efficiency and capacity may suffer. In addition, the constraints under which these units operate have changed as a result of more stringent protective and mitigation requirements, resulting in decreased electricity production. Concomitantly, advances in turbine designs and other technologies may offer potential for increased efficiency and reliability.

Therefore, there is potential to increase the generation and value from the existing hydropower fleet. In this context, performance improvement involves operational optimization for plant dispatch and scheduling but also improved unit efficiency and reliability through advanced technology and asset upgrades. Condition assessments of equipment, control systems, intake structures and water conveyances are one way to identify opportunities for performance improvements.

This article discusses a standardized condition assessment methodology for quantitative evaluation and comparison of plants from disparate regional and institutional contexts. Discerning the condition of aggregate U.S. hydropower at the national level is one objective of the U.S. Department of Energy’s Water Power Program. Thus, also included is consideration of the representativeness of results from a limited number of plants to which the methodology has been applied. The methodology considers efficiency and reliability improvement opportunities. It is distinguished from previously developed condition assessment methodologies1,2,3,4,5 by the use of two-dimensional rating matrices. The methodology uses a bottom-up scoring process to aggregate the condition scores from parts, components, units as elements of an entire plant.

Condition rating process and framework

The assets of a hydropower fleet are classified hierarchically as plants, units, components and parts. The methodology centers on the civil, mechanical and electrical components, as well as instrumentation and controls of the unit. These typically include the unit intake, water conveyances, turbine, generator and main transformer. The scoring process is aligned with the hierarchy of the facility, with individual part and component scoring aggregated to units and plant scores.

The methodology is documented in a guide book, spreadsheet scoring form, inspection form and checklist for each of 15 major components to be assessed, including trashracks, intakes, water conveyances, turbines and generators. The guidebook provides the rationale and rubric for evaluating the component condition, the Excel-based spreadsheet implements formulas for aggregation of scoring, and the inspection form and checklist facilitate thorough on-site data collection.

These documents were developed by subject matter experts in the hydro industry, with reference and alignment to previously developed guidelines and reports.3,4,5

Assessing parts and components involves five principal metrics:

1. Age, in years, a part or component of equipment has been in service;

2. Physical condition as observed through visual or non-destructive evaluation;

3. Installed technology level, or the extent to which the design of the component lags the state-of-the-art;

4. Operating restrictions placed on the component after commissioning, typically arising from safety, damage or environmental concerns; and

5. Maintenance and repair demands of the component, measured as those demands that exceed the original planned maintenance.

The meaning of physical condition can vary depending on the part or component undergoing assessment. For turbine runners, it means surface roughness, cracks and cavitation damage, while for generator windings it may refer to insulation resistance and polarization index. The physical condition is scored based on visual inspections and data collected from previous tests and measurements. Installed technology level is included because outdated technologies may bring difficulties in terms of parts replacement and a prolonged outage period if failure occurs.

Operating restrictions represent a gap and improvement opportunity between the original design and the desired function of a component. The rough zone of a turbine that limits its dispatch to near peak efficiency and precludes the provision of power system regulation services is one example of such an operating restriction. The maintenance requirement metric reflects historical and current demands for repairs and maintenance, particularly the amount of corrective maintenance required.

For electrical components, the results from some tests and data analyses may be more important than visual inspection as indicators of equipment health and condition. More so than visual inspection, results of tests of insulation resistance, polarization index and winding resistance are crucial to assigning a generator condition score. For instrumentation and controls, a different set of condition parameters are developed to better indicate the health and condition of these components. The details are documented in guides of individual component condition assessment.

Table 1 on page 32 is an example of a rating matrix for a major component, a Francis turbine in this case. The methodology includes similar tables for Kaplan and Pelton turbines. Each unit will have one table for its turbine parts and scoring. The matrix of condition scores are assigned according to the turbine rating criteria included in the aforementioned guide books.

The data quality score, SD(K), indicates the quality of available information and confidence the local facility staff and expert assessment team have in the data used for assessment.3

Assessment of an entire unit requires population of the rating tables for all of its components. Some components (such as the transformer) may be shared by several units in a plant. Such a component is assessed once and its condition affects the rating of each unit it supports. Some parts and components are specific to a unit. For example, an upstream pressurized water conveyance system may be shared by multiple turbine-generator units. In this case, the penstock sections have to be numbered and appropriately mapped to the individual units.

Once the component parts scoring table is established and a matrix of scores SC(J, K) is assigned, the CI of the component can be calculated:

Equation 1


– M is the total number of parts associated with a component;

– K is the identification number of parts (from 1 to M);

– N is the total number of condition parameters;

– J is the identification number of condition parameters (from 1 to N, for the physical condition, age, technology level, etc.);

– F(J) is the weighting factor for a condition parameter, determined based on the relative importance of the parameter to the overall assessment; and

– F(K) is the weighting factor for a part, determined by its relative importance to the overall condition of the component.

The weighting factors are based in the intuition of experienced hydropower engineers and facility staff members consulted during development of the methodology. They will likely require calibration as more facilities are assessed and results are interpreted. Note that the relative values of the weighting factors, rather than their absolute values, set the influence of each component for the overall unit CI.

The DI of a component is the weighted summation of all data quality scores received for its associated parts/items:

Equation 2

DI will be between 0 and 10 to indicate the level of confidence in the quality of data used for condition rating and rating results.

Table 2 illustrates the weighted summation of component CIs into a unit condition indicator (UCI):

Equation 3


– i is the component identification number (from 1 to N);

– CI(i) is the condition score of component i;

– W(i) is the weighting factor of component i; and

– N is the total number of components associated with the unit. The weighting factors for components are based on intuition of experts consulted during the development of the methodology and may require further adjustment.

Similarly to the component DI, the unit data quality indicator (UDI) is calculated as:

Equation 4

These computations yield component and unit CIs between 0 and 10. Poor condition (a low CI score) may be correlated with potential for generation and efficiency improvements through asset upgrade and indicate a higher risk of failure and the resulting economic losses. Operating restrictions or decisions for further evaluation can be made based on the range of CIs. A CI of 7 or greater may support continued operations and maintenance without restriction; a CI of 3 to 7 would be consistent with strategy to continue operation but perform a more detailed evaluation; and a CI of less than 3 would trigger immediate detailed evaluation, operational adjustments and corrective maintenance planning.3

Finally, all the CIs of components and units will be aggregated (see Table 3) to provide an overview of plant’s condition. The plant CI is simply the average of CIs of all assessed units in the plant.

The assessment team is an important aspect of the process. Consistency of approach across disparate facilities, staffing paradigms and operational contexts can be achieved by having a team of experts trained in the standard assessment methodology and executing that methodology for multiple facilities. However, the engineering and operations staffs for facilities often are the best source for design information, history, operating data, anecdotal information and insights into data quality. A combination of these is achieved by having the standard team conduct a scoping and data gathering phase to pre-populate assessment forms and checklists, followed by on-site collaboration with faciilty staff to intepret the assembled information, conduct physical assessments, and assign part and component ratings. Preliminary findings are then shared with the engineering and facility staff for further discussion, intepretation and revision.

Prospects for generalizing to regional and national fleets

Statistically valid estimates of broad opportunities for technology and asset upgrades require survey techniques that support expansion of the results from a limited sample assessment to the entire fleet. Selecting a subset of representative plants for assessment will provide useful insights but may not give a statistically valid basis for inferring the state of the entire fleet.

Sample size determination is important for economic reasons. An undersized study can lead to spurious conclusions and bad investment decisions, while an oversized study uses more resources than needed. The large sample size of assessments may require multiple teams to carry out the condition rating, increasing the challenge of maintaining consistency and comparability across all assessments.

Rationale of sample size

From statistical theory, for a certain confidence level (the level of certainty with which you will estimate the true population value) and confidence interval (the desired level of accuracy of the estimate), the needed number of samples can be calculated for an infinitely large population:

Equation 5


– z value (also called z-score or standard score) is the number of standard deviations a datum is above or below the mean;

– p specifies the expected proportion of the population to have the attribute being assessed; and

– c is confidence interval desired, typically 5%.

The needed number of random samples can be calculated for a finite population:

Equation 6


– NP is the population size.

Sampling techniques

To validate the limited sample assessment results, the concept and techniques of “stratified sampling” should be applied. When subpopulations vary within a population, stratified sampling takes random samples from each subpopulation (stratum) independently to improve the representativeness of the overall sample. This technique can decrease variances of sample estimates and alleviate the requirement for overall sample size. Proportionate allocation uses a sampling fraction in each of the strata that is proportional to that of the total population. For example, more plants should be selected from regions with dense hydro plant populations.

Also, based on the entire plant population, more plants with Francis turbines should be selected to mirror the proportional percentages of turbine types in the U.S. hydro fleet. Similarly, sampled plants should include proportional numbers of plants owned by federal, municipal and other non-federal public, private utility, private non-utility, industrial and cooperative companies. Plants selected also should be distributed in all major river basins.

Final comments

Condition assessment is a critical process to identify the opportunities for efficiency and reliability improvements at hydropower plants. A well-designed quantitative rating process and standardized methodology could provide benchmarking metrics for upgrading opportunity at individual plants, saving facility owners resources. At the same time, the aggregated assessment results can be used to characterize and trend the current hydro asset condition and upgrading opportunities across a broad fleet.

The condition assessment has to be combined with performance analysis to find the potential annual generation increases and values from the recommended upgrades. The condition assessment methodology presented in this article was implemented in 2011 and 2012 at eight plants in the U.S. The resulting CIs were correlated with performance analysis results, which validated its usefulness and promising success for broader implementation.

The methodology could be further improved by:

– Including more components for assessment (circuit breaker, surge arrester, powerhouse crane and station power service);

– Refining weighting factors through more demonstration assessments with variable plant features in terms of the power scale, technology type, ownership type, operating mode and market regionality; and

– Performing analyses to correlate unit and plant condition indicators to the performance improvement potentials, reliability impacts, and operation and maintenance cost impacts.


1Final Report of Hydropower Modernization Initiative – Asset Investment Planning Program, prepared by MWH for U.S. Army Corps of Engineers’ Hydroelectric Design Center, 2010.

2Benson, B., et al, “Aging Plants -Time for a Physical: Conducting a Comprehensive Condition Assessment and the Issues Identified,” Proceedings of HydroVision International 2008, PennWell, Tulsa, Okla., 2008.

3HydroAMP, Hydropower Asset Management-Using Condition Assessments and Risk-Based Economic Analyses, 2006.

4Roose, D., and D. Starks, “The California State Water Project: Asset Management & Condition Assessment,” Proceedings of Operations Management 2006, Environmental and Water Resources Institute of ASCE, 2006.

5Hydro Life Extension Modernization Guides, TR-112350-V1-7, EPRI, Palo Alto, Calif., 1999.

Qin Fen (Katherine) Zhang, PhD, is a hydropower mechanical engineer in the Environmental Sciences Division and Brennan Smith, PhD, is program manager of the Water Power Technology Program at Oak Ridge National Laboratory.

Previous articleUpdate from Ontario: Recalibrating for the Long Term
Next articleTVA seeks candidates for dam safety review board

No posts to display