Research and development of an initial model of hydrogen turbine
combustor for WE-NET project in Japan

M. Murayama, H. Toh and S. Yasu
Aero-Engine & Space Operations, Ishikawajima-Harima Heavy Industries
229 Tonogaya Mizuho-machi Tokyo 190-12 Japan

T. Saito
Research Institute, Ishikawajima-Harima Heavy Industries
3-1-15 Toyosu Koto-ku Tokyo 135 Japan


The WE-NET project aims to establish technologies for constructing a hydrogen energy network on a world-wide scale based on non-carbonaceous renewable energy[1]. In the project we are considering a hydrogen-combustion turbine system in the future for clean power generation with high efficiency. Mitsubishi Heavy Industries Co., Hitachi Co.[2], and we, IHI (Ishikawajima-Harima Heavy Industries Co.,) are respectively conducting research about a hydrogen combustor for the turbine system.

A typical cycle for this research is based on the cycle studied at the Graz University of Technology[3]. Figure 1 shows a diagram of a cycle which is a Brayton-Rankine combined cycle and contains a condenser. The design conditions for the combustor are 350°C, 2.5 MPa steam at the combustor inlet, hydrogen/oxygen stoichiometric combustion, and 1700°C at the outlet. Therefore, it is important to keep the metal temperature of the combustor liner within a tolerable level and the amount of residual reactants at the physical minimum.

To realize the combustor, we considered the following points:

  1. Annular type combustor has been selected to minimize the liner cooling surface and to easily control the outlet temperature distribution.
  2. Oxygen is premixed with steam to prevent metal damage by pure oxygen and to decrease the flame temperature. Hydrogen is multi-injected in a swirling flow of oxygen/steam mixture to enhance hydrogen mixing.
  3. Combustor liner has double-wall structure to enhance reliability.

In addition, a continuous gas analysis system which analyzes hydrogen and oxygen concentrations in steam was developed for combustion control to minimize residual reactants.


Figure 2 shows an initial sector model of the combustor and the test apparatus. The combustion chamber is a box-shaped, 220 mm in width, 120 mm in height and 242 mm in length. The combustor outlet is 40 mm in height and 140 mm in width. It has two gas injectors which inject hydrogen and oxygen. Oxygen is injected before a swirler and premixed with steam. The concentration of oxygen in a oxygen/steam mixture is about 30%, and flame temperature of the primary region is maintained under 2300°C to decrease non uniformity of the outlet temperature distribution. The combustor liner has double wall structure. Inner walls are heat shields, and outer walls form the structural strength. Impingement cooling was used with film cooling for the combustor liner durability.

Figure 3 shows the flow chart of the continuous gas analysis system. Sampled gas mainly contains water vapor, so after adding nitrogen, vapor is removed at the condenser. The hydrogen detector is a catalytic combustion type (TM-9098, Shin- cosmos Denki Co.). The oxygen detector is a paramagnetic analyzer (model 755, Fisher-Rosemount Japan Co.).


The test conditions were as follows:
inlet steam temp. : 350°C,
inlet pressure : 0.10-0.12 MPa,
outlet temp. : 1000-1700°C, near stoichiometric combustion,
steam flow rate : constant,
outlet flow velocity : 140 m/s at the 1700°C condition in outlet temp.

Tested Nozzles and Ignition Properties
Each hydrogen nozzle has 3 concentric sets of 8 holes. Two types of hydrogen nozzles, nozzle A, B were tested. Diameters of the nozzle B are 20% less than that of the nozzle A. Ignition occurred smoothly for each nozzle even in a condition that the concentration of oxygen in the oxygen/steam mixture is 8%, 1/4 of the design condition for ignition.

Gas Temperature Distribution at Combustor Outlet
Table 1 shows the test results. Each test condition is represented by the adiabatic flame temperature respectively. In the case of nozzle A, the pattern factor increased with the increasing flame temperature, and exceeded 0.2. On the other hand, in the case of nozzle B, the pattern factor is around 0.14. The nozzle B was used for the tests below.

Combustor Liner Temperatures
Figure 4 shows inner wall temperatures of the combustor liner. Test conditions are represented in the legend by the adiabatic flame temperatures at the combustor outlet. For reference, outer wall temperatures are shown at location 4' represented by circles with internal symbols indicating the test conditions. In the case near the design condition, inner wall temperatures were around 600°C. But temperatures of 2 points were over 700°C level. Therefore, liner cooling should be improved. On the other hand, the outer wall temperature was near the inlet steam temperature, and scarcely increased with the increasing flame temperature.

Residual Reactant Concentrations
Figure 5 shows residual hydrogen and oxygen concentrations. Test conditions are represented in the legend by the adiabatic flame temperatures at the combustor outlet. Temperatures of tested conditions were lower than that of the design condition. The error range of equivalence ratio was about 4% calculated from preciseness of the gas supply system. Therefore, residual reactant concentrations were evaluated by intersections of residual hydrogen and oxygen curves as indicated by circles in the figure. Residual reactants are mainly due to the decomposition of water vapor and poor mixing of the reactants. Tested conditions retarded thorough mixing of the reactants because the ratio of steam was larger than that of the design condition, nevertheless concentrations at intersection were suppressed below 1% and did not increase with the increasing flame temperature. Therefore, the residual reactant concentrations of the designed flame temperature are most likely to be below 1%, and the same order of the equilibrium concentration.

Response Time of the Gas Analysis System
Figure 6 shows a example of time chart of hydrogen and oxygen detector outputs. In this case the total response time of the system was about 30 seconds. Residence time from the probe to the detectors was about 10 seconds, and response times of the detectors were about 15 seconds. Since the tested conditions were near stoichiometric, the residual reactant concentrations were sensitive to perturbations in the gas supply system. So the apparent response time contained the influence of gas supply system perturbations.

A conventional analyzer, that is a gas chromatographer, is a batch process, and takes about 10 minutes to detect gas concentration for single case. In comparison, this system is effective for control of reactant flow rates to minimize residual reactants, and could monitor the combustor operation continuously.


In this experiment, we confirmed the following:
  1. Ignition occurred smoothly.
  2. The pattern factor of the outlet temperature was suppressed to the 0.14 level.
  3. The metal temperature of the combustor liner was within the tolerable level.
  4. Residual hydrogen and oxygen were suppressed to the same order of the equilibrium concentration of the designed flame temperature.
  5. The gas analysis system developed here is effective to continuously detect residual reactants, and could monitor the combustor operation.

We are planning a high-pressure test in 1998 to evaluate the performance of the combustor at the design condition.


This work has been conducted under a contract with the NEDO (New Energy and Industrial Technology Development Organization) and the Central Research Institute of Electric Power Industry as a part of the New Sunshine Program of MITI (Ministry of Industrial Trade and Industry). We appreciate their advice and support.


  1. M. Murase, "R&D Plans for WE-NET", Proceedings of IHCE'96.
  2. T. Hashimoto et al., "Hydrogen Combustion Characteristics in a Model Burner with a Coaxial Injector", Proceedings of the 11th World Energy Conference, Vol.2, pp. 1493-1502, 1996.
  3. H. Jericha et al., "Towards a Solar-Hydrogen System", 1991 ASME COGEN-TURBO IGTI-Vol.6.