The Kaveri Saga

Kaveri, an afterburning turbofan engine, was sanctioned on 30th March 1989 for ₹382.21 crore and was made to power the then-under development LCA. The venture was placed under GTRE, who already had quite an experience making engines. GTX37-14U was an augmented turbojet engine in 1977 and demonstrated in 1981, or a centrifugal type 10 kN thrust engine (i.e., F107-WR-105), which they made around 1959.

Another low-bypass engine was made called GTX37-14UB with GTX 37-14U as its core, which was an 89 kN Class engine; it was proposed for LCA before Kaveri but was overruled due to its large frontal section. The GTX 37-14U was a flat-rated twin-spool turbojet having a throttle ratio of 1.13 and a bypass ratio of 0.2:1.

Again, GTX 37-14U was also suffering from its own fair share of problems. Let's talk about Kaveri or GTX-35VS: It's an afterburning, twin-spool, low-bypass turbofan engine. It borrows its core, called Kabini, from it. 

In simple terms, twin spool means. A two-spool engine features two concentric shafts that rotate at different speeds: one links the high-pressure turbine stages to the high-pressure compressor, while the other connects the low-pressure turbine stages to the low-pressure compressor and fan. 

A Twin-Spool Engine Diagram

A low bypass means an engine where the amount of air bypassing the engine core (Bypass Flow) is relatively small compared to the amount of air that passes through the core (Core Flow).

The design of Kaveri includes:

• Compressor: 3 Low-Pressure Compressor Stages | 6 High-Pressure Compressor Stages with variable inlet stator vanes on the first 2 stages.
• Turbine: 1-stage LPT and 1-stage High-Pressure Turbine and Low-Pressure turbine using DS both blades at the start. 
• Convergent-divergent short annular system with step diffuser combustion: A nozzle with converging and diverging sections, a compact ring-shaped combustion chamber, and a diffuser with steps to manage airflow and pressure

Kaveri engine cross section model

At the start, the Kaveri had 5 stages of high-pressure compressor, which were increased to 6 based on suggestions of a retired chief designer from GE, Peter Chipporus.

One day in March 1988, Arunachalam told me that I should come to DRDO Head Quarters to take up the position of Chief Controller R&D. I resisted initially as this was a pure desk job, but I yielded. At one stage, I was assisting more than half of the technical laboratories. I was also in charge of the human resources development. But, I got bored doing this desk job as CCR&D and requested Arunachalam to send me back to a laboratory. He gave me a couple of options and 1 chose to be posted as Director of the Gas Turbine Research Establishment at Bangalore with effect from 1 January 1990. I was not a gas turbine specialist, but materials problems of the engine contributed to more than 50% in the engine development. Kaveri design incorporated the use of a variety of nickel base superalloys and titanium alloys. Thus, I was sure of making my contributions in the development of the engine.

Unlike NCML, GTRE, a major systems laboratory had more than 1500 employees and a strong employees union. It had built the after burner for the Orpheus 703 engine, but that was not accepted because it was over-weight. It had built two technology demonstrator engines designated GTX 37-14U and GTX 37-14UB, the former being a straight jet, while the later a bypass version. This had given confidence to GTRE to bid for building the flat rated gas turbine engine (Kaveri) for the Light Combat Aircraft. The design of the Kaveri engine was based on a mixture of eastern and western philosophies. Obviously, several problems did surface. Prior to my joining GTRE, reputed jet engine manufacturers like Rolls Royce, Snecma and General Electric had shown interest in participating in our engine development programme, but that was turned down due to some reason or other. After my joining GTRE, Mr. Peter Chipporus, a retired chief designer from GE came for discussions and based on his suggestion, the number of stages in the high pressure compressor was increased from 5 to 6. Also, it was brought out that the engine design dimensions specified correspond to the operating conditions and one needs to take into account thermal as well as centrifugal expansion to arrive at the dimensions for manufacturing. Another problem pertained to dimensional distortion due to residual stresses arising out of welding/machining. I assisted in developing a step wise annealing schedule to reduce the distortion to the minimum.

Most of the superalloys required for the manufacture of various discs and blades were initially imported, but later MIDHANI developed all the alloys which were certified by CEMILAC, the agency for certification of materials for air-worthiness. From the designers' point of view, it was necessary to generate adequate data of the materials in use and hence a separate facility, Aeronautical Materials Testing Laboratory, was established near Midhani.

While most of the engine testing facilities was available, we had to go overseas to get some specialized testing done. Russia offered these tests at a lower cost, but procedural delays were considerable. GTRE had 5 engine test beds and it used to test the prototype only once a day. I suggested that it has to be done more number of times a day, as otherwise we may not be able to meet the project schedule. And it was done. My moment of happiness was when the first prototype of Kaveri was test run in 1994. 

Kaveri was an overambitious project from the start with a very low budget compared to its equivalents, which is why, with a total of 2,839 crore spent, it was delinked from Tejas in 2008. Everyone was more or less fine with this decision. IAF said they will only order LCA in substantial numbers only if it meets their ASRs.  It was supposed to use even lower compressor stages than RD33 and achieve overambition targets. In 2008 the Kaveri was able to achieve ~73 kN of thrust with AB and 49 kN of dry thrust. The Kaveri was able to achieve its dry thrust more or less; the problems start with the afterburner, where it has a shortfall.  Below are the design specs of Kaveri:

Specification	Value	Specification	Value
Power Plant For	TEJAS (LCA)	Engine Type	Twin Spool, Low Bypass
Air Mass Flow	78 kg/sec	Bypass Ratio	0.16 - 0.18
Overall Pressure Ratio	21.5	TET (Turbine Entry Temp.)	1487 - 1700 K
Max Thrust (Dry)	52 kN	Max Thrust (Afterburner)	81 kN
SFC (Dry)	0.78 kg/hr/kg	Max SFC with Afterburner	2.03 kg/hr/kg
Thrust-to-Weight Ratio	7.8

Kaveri also has a shortfall of around 12%, and it has produced 70-75 kN throughout its development. There are a total of 10 versions of Kaveri: from K1 to K9, K9+, or K9*. Kaveri was started with an ambitious weight target of 1000 kg. Later revised to 1100 kg for LCA and again revised to 1050 kg. The first Kaveri engine (K1) weighed 1423.78 kg. GTRE undertook weight reduction exercise starting in 1993, and the weight of K9 was 1235 kg and finally the K9+, which is 1180 kg. To further reduce the weight, the use of blisks (bladed disks), PMC ducts, which weigh 26 kg compared to metallic 32 kg, and powder metallurgy disks, which will require a high-pressure isothermal press that India does not possess as of now, we do have a 2000 MT ton isothermal forge at MIDHANI, which was used to make disks for all five stages of the high-pressure compressor of andour engines, which power the Jaguar aircrafts of the IAF. The aim of Kaveri has achieved a thrust-to-weight ratio of 6.5 against the target of 8.

The thrust and weight shortfalls are just the tip of the iceberg; Kaveri also had a fair share of other problems:

• U Certain critical and crucial activities for successful development of Kaveri, viz. development of Compressor, Turbine and Engine Control System, have been lagging behind despite increase in cost by Rs 186.61 crore - CAG Report No. 16 of 2010
• GTRE has been unable to freeze the design of the turbine blades, the compressor has witnessed mechanical failure in performance and the engine control system is not flight-worthy. - CAG Report No. 16 of 2010
• The Kaveri engine which is designed to meet the conditions of operating fighters in the Indian environment is perhaps technologically more challenging than the airframe.

The Core Engine demonstration, which was planned for 1990, happened in 1995; Full Engine was demonstrated in 1995 instead of the planned 1992. The historical circumstances and geopolitics didn't help the program either. The LCA was to use Kaveri in the production variant while using Electric F404-GE-F2J3. In 1995, the US approved the sale of the 404 engine to India, and eleven of them were purchased to be fitted in early demonstrators of the LCA. 

In 1998, after the Pokhran Tests, the US sanctioned India, all support was withdrawn, and the LCA's last hope was the Kaveri engine. In 2001, LSP-1 made its first flight at Mach 2.1 powered with an F404-F2J3. The US sanctions were lifted, and Tejas went into production with F404-IN20 engines.

The main issues with Kaveri engines were blade flutter, screeching noise, and afterburner oscillations. There were also problems with the low-pressure compressor, which is why the entire frontal section was redesigned for KDE. The problem with the fan was that of efficiency surge margin and flutter; for HPC, it was blade high-cycle fatigue failure and shortfall of performance compared to what was expected. For combustors it was pressure loss, pattern factor, and structural integrity; for afterburners it was screech, thrust boost, and buckling. At the time of writing this article, all the problems are mostly solved. The issue with afterburner instability was with the fuel spray nozzles, and mostly the afterburner was stretched.

There was also an issue of flickering, which was solved back in 2010. Another issue is the very low bypass ratio, which leaves less air for cooling. The in-house blades used by GTRE initially failed; this led GTRE to procure disks, blades, and control systems from Snecma. The major issues with Kaveri as of today are resolved. Safran did an audit on the engine a couple of times; they noticed some problems with the afterburner, and they were resolved.

13 engines are made till 2021, including 9 full prototypes (K1 to K9) and 4 core engines, all upgraded to the K9+ standard. 

Two of the six Kaveri engines made in 1998 went for testing in Russia at CIMA, and a series of tests over two years were to examine its ability to withstand low pressure and temperature at high altitudes. Later in 2004, Kaveri failed high altitude testing in Russia. By 2008 Kaveri underwent 1,700 hours of ground testing in India and was sent to Russia twice. In 2011-12, Kaveri was tested in Russia by replacing one of the four engines of the Il-76 at a maximum altitude of 12 km and a forward speed of 0.7 Mach. 

After conducting thorough engine ground runs, the scientists successfully completed the taxi trials and the maiden flight test of the Kaveri engine with the IL-76 aircraft on November 3rd, 2011, followed by three additional flight tests. The engine was again successfully tested in 2012 at an altitude of 6000 ft at Mach 0.6 speed for 55 hours in Russia. During this testing, Kaveri was able to achieve 49.2 kN dry thrust against a 51 kN target and 70.4 kN wet thrust against 81 kN. 

In 2012, it was confirmed that a version of Kaveri (K9+) with its afterburner removed, called the Kaveri Derivative Engine (KDE), will be used to power the Indian Unmanned Strike Air Vehicle, or AURA, what is now called the Ghatak UCAV. 

During this time there was also a proposal to invite a foreign engine house to solve the problems with the Kaveri engine called K10. This was in discussion since 2008 but was never approached. In 2014 France proposed investing 1 billion euros as part of the Dassault Rafale offsets deal and suggested a joint venture with DRDO to swiftly revive the Kaveri engine program and make it airworthy by 2018. Everyone was fine with it; it required them to do nothing. Later, France offered the M88 core to be used, and this venture fell apart.

Kaveri was using DS blades in the LPT and HPT section, but the KDE is using CMSX4 for HPT blades, as revealed in an interview by Dr. S.V. Ramana Murthy (GTRE Director), who also recommended India to set up its own strategic materials reserve to store or bank critical materials that India will require for making these engines. He also talked about CMSX4 blades being made in India but raw materials still being imported.

We have also developed a TBC coating unit with ARCI, Hyderabad, and DMRL for yttria-stabilized zirconia coating. The project is more or less in research mode and has not been used for production yet. The SX blades in KDE does use TBC coating, but it is applied by some other machine. ARCI has also developed 150 kW Axial Suspension Plasma Spray, an alternative method for TBC. GTRE is also using EDM machines from Makino for cooling holes.

Talking about single-crystal blades, the HAL makes single-crystal blades for AL31FP engines, which power the Su-30MKI of the IAF, using the Bridgman–Stockbarger technique. The AL31 engine is 53% by cost indigenous and 87% of the components. One can imagine the disks and raw materials are still being imported. A contract was signed with HAL to make 240 AL31FP engines to overhaul Su-30s. The indigenous content will climb to 63% and average 54%.

DMRL has also developed their own single-crystal blades, like DMS4 with intricate cooling channels. The DMS4 is officially a 3rd Gen SX alloy, although there exists an argument that DMS4 is a 4th gen alloy; we will stick to the official definition given in papers published by DMRL. DMS4 is in the same league as CMSX10, Rene N6, and TMS75. DMS4 offers 1140°C TET compared to 1104°C of Rene N6. DMS4 is also patented. It also has a metal temperate capability of 1140°C against 1135°C of CMSX10, 1110°C of TMS 75 and 1150°C of TMS 196 (5th gen alloy) used in XF5(49 kN Japanese engine). Also, one can imagine the TET you can achieve with proper use of DMS4 from its heat treatment. The details about DMS4 show that it lies in the same league as some of the alloys.

Info about DMS4 from:

In the fully heat-treated condition, DMS4 offers more than 80°C metal temperature advantage over the first generation single crystal superalloy CMSX2 and about 8°C advantage over modern third generation alloys such as CMSX10

Fully solutionized between 1315°C - 1360°C over 24 hours with a heat treatment window of 20°C.

Shows superior 1% creep strength and creep rupture life comparable to CMSX10.

 

Alloy: DMS4M

• Solution Heat Treatment:

  • 1340°C for 5 hours
  • 1350°C for 5 hours
  • 1355°C for 10 hours
  • 1360°C for 15 hours, followed by air cooling

• First Aging Heat Treatment:

  • 1160°C for 6 hours, followed by air cooling 

• Second Aging Heat Treatment:

  • 870°C for 20 hours, followed by air cooling 

• Third Aging Heat Treatment:

  • 760°C for 30 hours, followed by air cooling 

There is also DMD4, a directionally solidified alloy derived from DMS4. Developed as a columnar grain superalloy for cost-effective turbine airfoil parts. Solutionized between 1300°C and 1330°C over 30 hours.

Now the question is why aren’t we using DMS4 if we have it? I think it was because the goal was never the performance; Kaveri already had a lot of trouble back in 2010, and they were dealing with it, and on top of it, introducing an untested, new material in the engine would have complicated their already very complicated problems. Since they have chosen tried and tested CMSX4, a second-generation alloy to be used in HPT of KDE. The Kabini was using Supercast 247A; maybe that's been replaced with the superior DMD4 alloy. Kaveri was also using Superni 718A for HPC. MIDHANI has also developed the Superni-115 LPT blade blank, although it can't be confirmed if Kaveri is using it or not.

DMRL has also made low-pressure turbine blisks for STFE engines. DMRL has also worked on serpentine air cooling for SX blades. 

The first KDE was delivered to GTRE in late 2024, soon followed by a second prototype, with both undergoing high-altitude testing in Russia. Where they have more or less achieved their goals. According to a report by ET, on Dec 24, 2024, after completing its high-altitude testing, the engine is ready for real-world evaluation on a flying test bed. 

There is another simultaneous project going on where a redesigned afterburner is being made by BrahMos Aerospace, who won the tender in 2020 from GTRE. The AB will then be integrated with KDE and will probably be used on a dual-engine aircraft or LSP Tejas for certification and demonstration purposes.

Today, India has all the building blocks to make its own 4th-gen engine with many PSUs and private players being the first-tier suppliers for global OEMs. A lot of work is being done on composite materials to sustain temperatures as high as 2000°C. Other than GTRE, there is also the Aero Engine Research and Design Center (AERDC) of HAL's engine R&D wing, which has developed the HTFE-25 and HTSE-1200 engines, although the technology is a generation older than what is used in Kaveri. Then there is also MIDHANI, which has developed a wide range of materials since the program began, including materials like Inconel 718. 

The majority of problems can be solved, and India can have its own engine with adequate manufacturing and testing infrastructure. But to be optimistic, HAL is setting up a national facility with a 50,000-ton die forge press and a 20,000-ton isothermal press. For many uses, the facility will also make titanium bulkheads for AMCA. The 20,000-ton iso press will be used to make powder metallurgy disks. Then India will need to use blisk for weight reduction of the engine. and use of composite materials in the AB.  

Then there is the joint venture deal with foreign OEMs to make a 110 kN engine, which will replace GE F414-IN6 in the future, the front runners of this deal being Rolls-Royce and Safran. Both are offering everything India lacks more or less, and the deal basically includes GTRE getting know-how and know-why, a flying test bed, and other manufacturing and testing infrastructure required to make this engine. The development of this engine will take 10-15 years and will require funding of $4-6 billion USD.