Around three months after commissioning in September/October 2007 cracks and leaks were encountered in heat recovery steam generator HP evaporator tubes at Trianel’s 2 x 400 MWe Hamm-Uentrop combined cycle plant in Germany. Similar damage was also found at Mark E’s new 1 x 400 MWe Herdecke combined cycle plant (Figure 1), which employs identical technology.

The triple pressure HRSGs, manufactured by Ansaldo under licence to Siemens, are of an innovative design (see MPS, February 2006), employing Benson once through evaporators in their HP sections. The exhaust gas flow is horizontal and the tubes are vertical. They employ the low mass flux concept, as developed by Siemens (licensor of the Benson technology) and first used at Cottam in the UK.

After Cottam, Hamm-Uentrop and Herdecke are indeed only the second and third combined cycle plants in the world with this HRSG technology to enter service, although there are a number of other combined cycle plants currently under construction and planned which incorporate the Benson low mass flux vertical tube design.

In Hamm-Uentrop and Herdecke the HRSGs are installed downstream of Siemens SGT5- 4000F (aka V94.3A) gas turbines (as at Cottam).

The problem

The HP evaporator part of the HRSG, in which the problem was encountered, consists of two sections connected in series, known as HP Evaporator 1 and HP Evaporator 2. The leaks and cracks were found to be located in the inlet section of HP Evaporator 2, in the welds connecting the heat exchange tubes to the inlet header, where the stubs are joined to the inlet headers (see Figure 2).

A metallurgical investigation was carried out by TÜV Rheinland. This concluded that the cracking was caused by fatigue due to low frequency oscillations in the tensions experienced by the tubes.

In addition to the metallurgical analysis, strain gauges were installed and showed that, at full power plant load, the tension in the tubes oscillated with a period of about 240 seconds.

Further investigations were carried out by Siemens. Thermocouples were installed at the upper ends of heat exchanger tubes in HP Evaporator 2, on the 32 first row (ie hottest) tubes of the central module (the tubes being arranged in three modules across the width of the HRSG connected in parallel, each module having 32 tubes width-wise (ie 96 tubes across the full width of the HRSG).

The thermocouples showed there were oscillations in individual tube wall temperatures, with an amplitude of up to ±35 K and a period of ~300 seconds (Figure 3).

Also, as Figure 4 shows, there was a temperature spread of 90 K between the tubes with the minimum temperatures and those with the maximum temperatures.

These variations in temperature among the tubes leads to unacceptably high stresses at the position where the cracks were found.

In turn, the cause of the temperature oscillations was found to be dynamic instability of HP Evaporator 1. These dynamic instabilities (mass flow fluctuations) were found to have the same frequency as the temperature oscillations.

Simulations clearly demonstrated the dynamic instability of HP Evaporator 1 at full load operation, with oscillating steam quality and mass flow (Figure 5).

Figure 5 viewed in combination with Figure 6 demonstrates the relation between the instability of HP Evaporator 1 and the resulting temperature oscillations in the tubes of HP Evaporator 2.

The spread between tubes is caused by non-uniform distribution of the water steam mixture within the HP Evaporator 2 inlet header. Figure 6 shows the relation between mass flow, steam content and wall temperature (average and at the tube outlet) for a single tube. The average steam quality before HP Evaporator 2 is approximately 0.3. But as mentioned there will be tubes in the HP Evaporator 2 seeing lower and higher steam qualities because of the non-uniform distribution. At full load conditions the minimum tube wall temperature was 345 °C and the maximum 435 °C. According to Figure 6 the wall temperature of 435 °C corresponds to an inlet steam quality of ~0.1 and 345 °C to 0.6.

The solution

It is well established that throttling at the evaporator inlet improves dynamic stability and can be achieved by inserting orifices. The orifices are installed at the inlets of HP Evaporator 1 (see Figure 7) and designed to achieve an additional pressure drop of 250 mbar at full load.

In addition, to reduce the spread of temperatures by improving the water/steam distribution at the inlet to HP Evaporator 2, T-pieces (Figure 8) have been installed to double the number of feed points into the inlet headers of HP Evaporator 2. These T-pieces reduce the temperature spread by roughly 20% compared with the original design.

Finally, to better accommodate the stresses arising from remaining temperature differences between tubes, expansion loops have been installed at the inlet end of the HP Evaporator 2 tubes (Figure 9). The expansion loops in fact have the effect of increasing the temperature spread but are designed to deal with a range of ± 50 K (averaged over tube height), providing a good margin when the other countermeasures are taken into account.

The stress analyses and the fatigue calculations were performed by the boiler manufacturer and double checked by Areva.

Temperature measuring devices have been installed at the outlet of all 96 tubes of the hottest tube row in HP Evaporator 1 and the measurements are used to ensure that temperatures are within the allowable limits.

The modifications were completed at Herdecks on 7 June 2008 and at Hamm-Uentrop on 8 August 2008.

The end result

The orifices installed at the HP Evaporator 1 inlets render HP Evaporator 1 stable and successfully eliminate large temperature oscillations, of the kind shown in Figure 3, over the whole load range, from 40% to 100% gas turbine power.

The installation of T-Pieces at HP Evaporator 2 inlets has improved the water/steam distribution, reducing the temperature spread among the tubes.

Figures 10 and 11 show examples of typical measurements taken (at 100 % load) following implementation of the countermeasures.

Figure 10 shows individual tube wall temperatures over time. The large fluctuations have been eliminated – compare with Figure 3.

Figure 11 shows typical time averaged tube wall temperatures (at the tube outlets) for 100% gas turbine power. The temperature spread recorded at the tube outlets is less than ± 47 K over all power levels. But the relevant value for the design of the expansion loops is the tube wall temperatures averaged over tube height, which have a smaller spread than that of the tube outlet temperatures. Figure 6 gives a qualitative indication of typical differences between tube wall temperatures at the inlet and tube wall temperatures averaged over the tube height. Post modifications, the spread of tube wall temperatures (averaged over tube height) is well within the ± 50 K spread that the expansion loops are designed to accommodate. So the remaining stresses can be tolerated by the new expansion loops.

Overall, the countermeasures have achieved the desired result, allowing the HRSGs to be operated as originally intended.