Energy efficiency is an important operational element for power plant operators. A recent project has benchmarked the energy efficiency of the entire power generation fleet of the Netherlands, comparing it against the ‘world’s best’.
Jan van der Marel and Ingrid Bins, Jacobs Consultancy, Netherlands
In 1999 the Dutch government and energy intensive industries agreed on a covenant on energy efficiency as part of the country’s climate change policy. In this covenant the industry committed to attain ‘world class performance on energy efficiency’ by 2012. At the same time, the government agreed not to expose the benchmarking parties to additional energy taxes.
In the benchmarking process the industry is obliged to hire an independent third party consultant every four years to define ‘world class performance on energy efficiency’, also referred to as ‘World’s Top’ (WT), and the distance from the world’s top for their installations.
The combination of KPMG Sustainable B.V. (KPMG) and Jacobs Consultancy acted as an independent consultant for the Dutch power production companies to define WT for power production and the distance from WT for their installations.
Figure 1. Difference between design and operational efficiency over time
In 2000, the first round of benchmarking was carried out based on design efficiencies, while for the 2004 round, Jacobs and KPMG were able to develop a benchmarking methodology based on operational efficiencies. The use of operational efficiencies provides a more accurate result as it takes into account the actual state of the installation under consideration of external factors such as part load, ambient conditions, and cooling water.
Jacobs and KPMG chose to use the decile method, where WT is defined as the top ten per cent of international installations. The distance to WT, expressed in TJ primary energy per year, is the difference in specific energy consumption of the WT and the installation and the production volume of the installation.
The results of the first benchmarking round of 2000 were used in the allocation process of CO2 emission allowances under the EU emission trading scheme, where installations with a high energy efficiency were allocated relatively more allowances as a reward for early action.
Design vs efficiency
In practice the operational net electric efficiency of a power plant is not equal to the design efficiency. The difference in design efficiency and operational efficiency is caused by the following effects:
- Natural degradation of components
- The state of maintenance
- Operational effects like part load operation
- External effects like ambient conditions.
In spite of the difference between design and operational efficiency, the design efficiency is still the first indicator of a power plant’s energy efficiency. Operational efficiencies are considered confidential and data from international installations cannot be obtained, let alone all the information required to perform the necessary corrections. Therefore the first round of benchmarking, in 2000, was executed using design efficiencies only.
The obvious disadvantage of using design efficiencies is that the actual plant performance does not influence the benchmarking result. The second disadvantage is that using design efficiencies indirectly implies a correction for plant age. Using design efficiencies is a disadvantage for installations newer than the average of the world’s best ten per cent and an advantage for most installations older than the average.
Thirdly, improvement measures that do not influence design efficiency have no effect on the benchmarking result. Finally there is the disadvantage that design efficiencies are higher than operational efficiencies giving the impression that the design efficiency approach is beneficial for the power sector.
The advantage of using design efficiencies is the unambiguous and transparent process of determination of WT and distance to WT, while the use of operational efficiencies requires a number of corrections before a WT and distance to WT can be determined.
For the 2000 benchmarking project, KPMG and Jacobs researched extensively the availability of data on power plants worldwide. The result of the study was that operational data on fuel consumption and electricity production from a large range of plants was not available. However, basic data determining design efficiencies is available in commercial databases.
After the evaluation of a number of public and commercial databases the UDI Platts World Electric Power Plant Database was selected as the basis for the benchmarking. This database was found to be the most up-to-date, extensive and complete.
From the database the gas and coal fired plants larger than 50 MWe were selected and their design information collected. With a simple model in the heat and mass balance software GateCycle, the design efficiencies were calculated. The model was tuned to match closely the known design efficiency of the Dutch power plants then the design efficiency of all 6000 selected installations in the coal and gas databases was calculated.
With the design efficiency, at reference conditions, the installations can be ranked. The WT2000 for gas and coal follow from that ranking. In this document the results from 2000 are used as a reference and the WT2000 for both gas and coal is set to 100. Other results are presented relative to this.
The distance to the WT follows by subtracting WT fuel consumption (from plant power production and WT efficiency) from the plant design fuel consumption (from plant power production and plant design efficiency). See equation below with distance to WT in GJ and production in MWh. A negative result implies better than WT performance.
The challenge for the KPMG/Jacobs team was to develop a method using operational efficiencies in the benchmarking without the availability of international operational data. It is important that the method is simple, accurate and transparent.
In theory a brand new installation could operate at its design efficiency for a certain period if the external conditions are equal to those defined at the design. In practice external conditions are not equal to the design conditions and degradation of heat transfer surfaces, compressors and turbines begins the minute they start operating. Deterioration is caused by three effects: ageing, O&M practices and external factors, the latter of which must be dealt with on a plant-by-plant basis.
Given the three causes of performance degradation a method for bench-marking using operational efficiencies can be developed. This method has a two-sided approach:
- The design efficiencies of the international installations will be determined as in the first round of benchmarking. After that, the design efficiencies are adjusted to incorporate the generic effects of ageing and O&M-status. In this way the design efficiency of the internal installations is ‘translated’ into operational efficiency. This adjustment is done based on plant age only. The result is a ‘corrected operational efficiency’ (COE). ‘Corrected’ because using the design efficiency as a starting point implies an implicit correction for external factors.
- For the Dutch installations the operational efficiencies will be taken as a starting point. The rough operational efficiency (electricity production divided by fuel consumption) will be corrected for external factors like part load, ambient conditions, etc. Like in step 1 the result is a corrected operational efficiency (COE).
The result of step 1 and step 2 is that the operational efficiency of the international and Dutch installations can be compared. The WT on operational efficiencies (WTCOE) can be determined through ranking the international installations after step 1. The distance to the WTCOE for Dutch plants can be determined with the plant COE that is available after step 2.
Implementation of this methodology requires the determination of the generic performance degradation as a function of plant age covering ageing and O&M-status as well as the development of a model using plant specific input and plant operational input to determine the correction for external factors. The result is a generic correction for international installations and an individual correction for Dutch installations.
Generic correction: From a screening of public data like the Internet, literature, suppliers, (commercial) databases, design codes, etc. it was concluded that only limited data on plant degradation through ageing and O&M-status was available. On a component level, however, sufficient well-founded data on ageing could be found, although data on O&M status remained limited. Further analysis found that an absolute figure for the O&M status-driven performance degradation cannot be determined. However, the effect of the O&M status is relatively small compared to the effect of ageing.
Individual correction: For the Dutch installations, all the information required to carry out the plant specific individual corrections is available. A spreadsheet-based model has been developed to correct the rough operational efficiency (electricity production divided by fuel consumption) to find the COE.
For the individual correction for the Dutch installations, a number of external factors were included. Corrections are calculated based on monthly average data, summing up to a total annual correction. The monthly corrections were found to be frequent enough to account for the relative slow changing variables like cooling water temperature and ambient conditions but not frequent enough to account for daily load following patterns. Therefore in the part load correction extra provisions had to be taken.
Part load: Generally, when an installation is operated in part load the efficiency decreases compared to full load. With the given efficiency curve of an installation, the plant efficiency at monthly average load can be determined. The correction for part load results from the difference between the theoretical fuel consumption at design efficiency and at ‘curve efficiency’.
Because the efficiency decrease at part load is not linear to plant load the impact of plant load behaviour is significant. For the part load correction it makes a large difference if a plant is continuously operated at 80 per cent load for a month or if the plant is operated at 100 per cent load for half a month and at 60 per cent load for the other half. This information cannot be obtained from monthly data. Because most plants follow a daily load pattern it does not make sense to increase the calculation interval from monthly to daily either. Therefore, as a test, the corrections for a number of plants were calculated on an hourly basis.
The calculations on hourly data showed that the part load correction for load following and peak load installations was significantly higher. The part load correction for base load installations and the other corrections were much less sensitive to the calculation interval.
Because supplying hourly data requires a large effort for the production companies an alternative was found in so called load profiles. Three typical load profiles were defined – base load, medium load and peak load. From the monthly data a load profile can be determined. The part load correction is calculated depending on the selected profile for the applicable month.
As a part of the test with hourly data the effect of ambient conditions and cooling water temperature relative to the most representative efficiency curve was analyzed. From the analysis it could be concluded that the effect was relatively small.
Start/stop: The correction for additional fuel consumption due to start/stop operation of the plant can be calculated by the number of hot and cold starts per month and the typical additional fuel consumption for a hot and cold start.
Grid frequency control: In the Netherlands most large-scale power production units participate in ‘grid frequency control’. This means that they have to respond to frequency fluctuations in the grid to keep the grid frequency within fixed limits. Because of the continuously changing set points the optimum set point from an energy efficiency point of view cannot be fixed. Therefore a general, installation independent, correction on fuel consumption of 0.5 per cent for participation in the grid frequency control is used.
Biomass co-firing: Because of the relatively low calorific value of the biomass fuels used, plant efficiency decreases with biomass co-firing. With a general, installation-independent efficiency curve, the co-firing efficiency as a function of calorific value can be defined. With the co-firing efficiency and the percentage of biomass (on energy) the additional fuel consumption due to biomass co-firing can be calculated.
Results of benchmarking
With the COE from the design efficiency and the generic correction based on age the installations can be ranked and the WTCOE can be determined. The WTCOE for gas and coal fired plants relative to 2000 are:
- WTCOE gas: 104
- WTCOE coal: 100
Figures 2 and 3 give the ranking of the top 400 to 500 installations and WTCOE for gas and coal fired installations.
In Figure 2 the black dots represent the design efficiency of the best gas fired installations in 1999. In 1999, the population of gas fired plants consists of 1990 installations, resulting in a decile of 199 and a reference WT of 100. The dark blue dots represent the design efficiency of the best installations in 2003. The large difference with 1999 is caused by the large scale construction of new combined cycle plants.
Figure 2. Ranking and WTCOE of gas fired installations
In Figure 3 the black dots represent the design efficiency of the best coal fired installations in 1999. In 1999, the population of coal fired plants consists of 2770 installations, resulting in a decile of 277 and a reference WT of 100. The dark blue dots represent the design efficiency of the best installations in 2003. The difference with 1999 is mainly the increased number of supercritical units.
Figure 3. Ranking and WTCOE of coal fired installations
With the WT based on COE, the distance to the WT of the individual installations can be determined. The distance to the WT follows by subtracting the WT fuel consumption (from plant power production and WTCOE) from the plant fuel consumption minus corrections.
UDI Platts ‘World Electric Power Plant Database’, December 1999 Covenant benchmarking energy efficiency 1999