Simulation of the internal stress in a rotor disk during the fast cold start of a gas turbine. The transition from blue to red shows the increase in stress levels.The simulation shows that the stress is highest at the center of the rotor disk.
September 15, 2020
6 min read

How computers are helping turbines with stress

Frank Krull

Starting power plants fast from cold may be the norm in a low-carbon world, but it puts gas turbines under a lot of stress. So, what’s your turbine’s limit? Here’s how mass material simulations on high-performance computers can tell you.

Kai Kadau works with his team at Siemens Energy in Charlotte, N.C., to develop simulation tools for designing large generators and turbines, and he has good reason to be enthusiastic. Mrugesh Gajjar is one of the leading experts in graphics processor based computing and software performance optimization at Siemens Corporate Technology in Bangalore, and he’s found a way to increase the calculation speed of his tools by between ten and 100 times. That’s especially good news for Kadau’s flagship, a simulation tool used in turbine development at Siemens Energy to calculate the mechanical fracture strength of forged-steel components. Calculations now take just minutes or seconds rather than hours, even though the tool, which uses probabilistic simulation, has to run through several million calculation sequences to cover all the material and load combinations that determine fracture strength.

Assembly of a gas turbine rotor made of stacked rotor disks.


Highly reliable details on fracture strength are needed for the forged components that Siemens Energy uses in its large turbines, especially the rotor disks used in the turbine rotors. Weighing up to seven metric tons and up to two meters in diameter, the disks – with forty or more blades installed around the outer diameter – are the turbine components that have to withstand some of the toughest loads. During fast cold starts, for example, points on the disks that heat up more slowly than others are subjected to extremely high stress. “If these starts happen regularly, the result can be dangerous, because it can lead to disk fractures over the longer term,” Kadau notes. “When it’s in full swing, a complete large turbine rotor with 20 disks has the kinetic energy of 200 trucks driving at full speed. The consequences of a disk fracture aren’t something you want to think about.”


The fracture strength calculations at Siemens Energy using Kadau’s simulation tool have played a critical part in ruling out the risk of disk fractures for a number of years. In addition to determining the stress a rotor can withstand during operation – before they have to be inspected and replaced as necessary – the tool is also used in turbine development to ensure that the disks can cope with new demands. This is especially important right now, because more and more power is being generated using solar cells and wind turbines whose capacity fluctuates strongly and frequently, depending on the weather. If there’s no wind, or if cloud blocks the sun, traditional power stations have to step in at short notice to ensure a stable supply of electricity. Gas-fired power stations are preferred for this task, because they’re capable of starting up very quickly, and they produce a very small amount of harmful emissions compared with other fossil-fuel power stations. This means that their disks will need to withstand many more quick starts in the future than were previously required.

The rotor inside the SGT6 9000HL gas turbine is a stack of rotor disks with turbine blades attached to the outer ring.


What the tool can achieve in terms of building up starting strength is shown by a new high-performance natural gas turbine that was fired up for the first time at Duke Energy’s Lincoln Combustion Turbine Station near Denver, N.C., in April 2020. “At 85 MW per minute, the SGT6 9000HL powers up extremely quickly, enabling it to reach its full capacity of 400 MW in just a few minutes,” affirms Hans Maghon, who has been monitoring the development of the new gas turbine as program director at Siemens Energy. “The 9000HL also guarantees 1,250 starts between inspections, the level of flexibility that’s a precondition for commercial viability in the energy market of the future.” For Maghon, this is more confirmation of why probabilistic simulation of fracture strength has been an established part of his projects for some time now. He’s looking forward to the faster computing speeds every bit as much as Kadau.


Kadau has already done a lot to increase the calculation speed his tool offers, whether using performance-optimized software codes or distributed computing systems. With each improvement, the simulations ran a little faster. “But we never achieved the level of acceleration provided by Gajjar’s new graphics processor-based high-performance computers,” Kadau says, acknowledging the importance of this new step. This extraordinary leap in speed isn’t just because even more computing operations can be processed simultaneously using graphics processors compared with traditional high-performance computers. “Even more important was the fact that Gajjar and his team, with their expert knowledge, skillfully reorganized and reprogrammed key parts of the software to get the most out of the benefits offered by graphics processor-based high-performance computing,” Kadau explains.


It’s already conceivable that Gajjar’s acceleration coup will provide a new boost to innovation in more areas than just rotor disk development. “The simulations now run so fast that we can also apply probabilistic simulation to entire turbines,” Kadau observes. “The jump in speed also enables services that we were previously unable to offer in this form because of insufficient computing capacity and the costs involved,” adds Uwe Gruschka, who’s responsible for technology and innovation strategy orientation at Siemens Energy. “More and more customers are asking us what more frequent quick starts will mean for their older power stations. For us to be able to answer that question, we first need to thoroughly simulate appropriate load scenarios using a digital twin of their plant. Until now, that would have been simply too time-consuming. But that all changes with simulation times coming down to just a few minutes.” 

September 15, 2020

Frank Krull is a physicist and journalist and works in the Communications department at Siemens Energy.

Combined picture credits: Siemens Energy