Abstract

A promising candidate for CO2 neutral power production is semiclosed oxyfuel combustion combined cycles (SCOC-CC). Two alternative SCOC-CCs have been investigated both with recirculation of the working fluid (WF) (CO2 and H2O) but with different H2O content due to different conditions for condensation of water from the working fluid. The alternative with low moisture content in the recirculated working fluid has shown the highest thermodynamic potential and has been selected for further study. The necessity to use recirculated exhaust gas as the working fluid will make the design of the gas turbine quite different from a conventional gas turbine. For a combined cycle using a steam Rankine cycle as a bottoming cycle, it is vital that the temperature of the exhaust gas from the Brayton cycle is well-suited for steam generation that fits steam turbine live steam conditions. For oxyfuel gas turbines with a combustor outlet temperature of the same magnitude as conventional gas turbines, a much higher pressure ratio is required (close to twice the ratio as for a conventional gas turbine) in order to achieve a turbine outlet temperature suitable for combined cycle. Based on input from the optimized cycle calculations, a conceptual combustion system has been developed, where three different combustor feed streams can be controlled independently: the natural gas fuel, the oxidizer consisting mainly of oxygen plus some impurities, and the recirculated working fluid. This gives more flexibility compared to air-based gas turbines, but also introduces some design challenges. A key issue is how to maintain high combustion efficiency over the entire load range using as little oxidizer as possible and with emissions (NOx, CO, unburnt hydrocarbons (UHC)) within given constraints. Other important challenges are related to combustion stability, heat transfer and cooling, and material integrity, all of which are much affected when going from air-based to oxygen-based gas turbine combustion. Matching with existing air-based burner and combustor designs has been done in order to use as much as possible of what is proven technology today. The selected stabilization concept, heat transfer evaluation, burner, and combustion chamber layout will be described. As a next step, the pilot burner will be tested both at atmospheric and high pressure conditions.

Introduction

It is envisaged in the EU 2050 energy roadmap (published by the European Commission in Dec. 2011) that energy supply from electricity is assumed to grow significantly until 2050. It is also stated that 97–99% of electric power production from fossil fuels must be equipped with CO2 capture by 2050.

The opportunities for near-term implementation of low and zero-emission fossil fuel power generation using carbon capture and storage (CCS) is emerging in niche markets. This is primarily motivated by regulations following a growing awareness regarding the potential impact of climate-change and partly the opportunities for use of carbon dioxide (CO2) with enhanced oil recovery (EOR). However, there remain significant technology, engineering, investment, and political barriers that need to be overcome before CCS can be accepted as commercially mature for the power generation industry and the finance community.

One of the three main routes for CCS is the oxyfuel route. The literature regarding gas fired oxyfuel power plants is mainly referring to semiclosed oxyfuel combustion cycles and a modified Graz cycle variant [1]. SCOC-CC are promising candidates for CCS power plants due to their relative simplicity compared to other carbon capture cycles and their independence of chemicals.

These cycles are well suited for all types of pure hydrocarbon fuels such as natural gas and for gasified biomass or coal (syngas). The concept mainly includes turbomachinery equipment well known to utility companies and excludes chemical industry equipment.

A drawback of the oxyfuel concept is the need for an air separation unit (ASU) for the generation of oxygen. The ASU generates both high investment cost and a large foot print, and it also contributes significantly to a reduction in plant net efficiency due to its high energy consumption. However similar types of drawbacks appear in close to all other CO2 capture concepts, e.g., post combustion capture concepts with monoethanolamine (MEA) scrubbing.

This paper presents results from a feasibility project called OXYGT (oxyfuel gas turbine combined cycle), the objective of which is to investigate the performance of a proposed commercial SCOC-CC in the 125 MW range. The objectives for the OXYGT phase 1 are (1) to assess the market potential for the oxyfuel technology including the produced CO2, (2) to assess the technical and economic performance of the oxyfuel engine, and (3) to prepare for a demonstration of a natural gas fired oxyfueled gas turbine in a pilot plant. The overall objective for the whole oxyfuel project (phase 1–3) is to develop a natural gas fired oxyfuel combined cycle power plant concept (CCPP) for demonstration in a commercial scale power plant.

The development work in the project phase 1 is based on optimization of two slightly different oxyfuel power plant concepts with the aim to select one of them as a starting point for developing a concept for a combustion system in the selected SCOC-CC. The project is carried through in a partnership between a turbine manufacturer (Siemens), an engineering company (Nebb Engineering), two partners representing research institutes and academia (SINTEF, Lund University), and the funding organization Gassnova.

Oxyfuel Plant Concept

The main idea behind oxyfuel cycles is to avoid mixing carbon dioxide with nitrogen (N2), which complicates the carbon capture process. By using pure oxygen (O2) instead of air in the combustion process, the combustion products can be reduced to mainly steam (H2O) and CO2. CO2 can then be separated by simply condensing the steam downstream of the heat recovering unit.

The oxygen is generated in an air separation unit. There are three main techniques used in air separation units: cryogenic, membrane, or adsorption. For large scale O2 generation, a cryogenic process is presently the only feasible choice, even if ionic transport membrane (ITM) based pilot plants are in operation today. It should be noted that with a cryogenic process, argon cannot be separated. A cryogenic process means that the gas components are separated by condensation at low temperatures obtained in a refrigeration process. In a cycle perspective, there might be possibilities to integrate the ASU process with the compression of CO2 after the power plant. To reach the high pressures of the CO2 (100–200 bar depending on storage conditions) and low temperatures that are required for final storage, the CO2 will first be compressed to condensation pressure and then pumped to the final pressure.

It should be emphasized that oxyfuel power cycles are only suitable for carbon capture power plants and will never be able to compete with conventional power plant cycles in terms of efficiency and cost of electricity if CO2 is not taxed or utilized as a commodity.

The SCOC-CC consists of a topping Brayton cycle and a bottoming Rankine cycle and has, therefore, obviously many similarities with a regular combined cycle power plant. This makes the SCOC-CC a promising candidate for carbon capture and storage due to its relative simplicity compared to other CCS concepts. It is mainly the topping gas turbine and the working fluid that differ from a conventional CCPP plant. First of all, the WF in the cycle is mainly CO2 and H2O (diluted with small amounts of O2, Ar, and N2). Secondly, the combustion chamber is operated close to stoichiometric conditions with pure oxygen as the oxidizer. The combustion product that consists of CO2 and H2O expands through the turbine. The turbine cooling is provided with a CO2/H2O mixture as coolant extracted from the compressor.

The pressure ratio of the oxyfuel gas turbine is much higher (about 40) compared to a conventional combined cycle gas turbine (about 18) due to the relatively low specific heat ratio for CO2. The heat recovery steam generator (HRSG) does not differ from the one in a regular CCPP in other ways than that the WF is CO2/H2O instead of flue gas, nor does the steam cycle.

The Oxyfuel Combustion Combined Cycle

The first of the two investigated and optimized semiclosed oxyfuel combustion combined cycles is very similar to a recirculated conventional combined cycle. It is named the oxyfuel combined cycle and has got the acronym OCC (see Fig. 1). The main difference, except from the working medium, is that the compression is mechanically divided into two compressors with different rotational speeds to manage to keep up efficiency for all compressor stages and to prove good working condition for the driving turbines. This solution is triggered by the high pressure ratios needed and the low speed of sound in the working medium.

Fig. 1
Layout of the OCC

After the HRSG, the exhaust gas is cooled below the dew point of H2O generating a liquid stream of water. The main part of the exhaust gas is recirculated to the compressor inlet, and the rest is bled off for treatment to export conditions. The bleed stream is compressed, intercooled, and separated into CO2 and water, and the CO2 is pumped to final storage pressure. Since the working medium is cooled down below the dew point of H2O before the recirculation, the compressor has to be protected from erosion of the inlet stages by reduction of the relative humidity in the stream.

The recycled gas will have a moisture content of about 15 vol. % before the condensation, and 7 vol. % after the condensation. This cycle has been simulated for two turbine inlet temperature (TIT) levels, 1340 °C and 1420 °C, but the following is mainly describing the 1340 °C version.

The High Moisture Oxyfuel Combustion Cycle

The second investigated and optimized semiclosed oxyfuel combustion cycle is named the high moisture oxyfuel cycle and has got the acronym HMOC (see Fig. 2). This is a slightly more advanced cycle compared to the OCC but still similar to a recirculated conventional combined cycle plant. There are two main differences compared to the OCC:

Fig. 2
Layout of the HMOC
Fig. 2
Layout of the HMOC
Close modal
  1. (1)

    There is no cooling/condensation after the HRSG, but instead an intercooler is placed between the two compressors. This has three effects: first it will keep a high efficiency for all compressor stages, and second it will reduce the compressor work. Further, water is condensed out in the intercooler at elevated pressure, which enables utilization of the condensation energy in the bottoming Brayton cycle.

  2. (2)

    The bleed is located between the compressors, generating a higher inlet pressure for the CO2 compressor train. This saves at least one stage in the CO2-compressor train.

The recycled gas will have a moisture content of about 40 vol. % before the condensation and 36 vol. % after the condensation.

Method

The method used for the investigation and comparison of the two SCOC-CC cycles was to simulate them at a detailed level using process simulation tools (Siemens in-house tool KRAWAL that is used for tender calculations of combined cycle power plants, and the commercial tool HYSYS). First the plant boundary conditions and the performance characteristics of the individual components were specified for a fair comparison. Then the two cycles were simulated and optimized individually for maximum efficiency and manufacturability and compared to decide which concept would be the preferred choice for design regarding turbomachinery and combustor.

Results

Compressor Data.

Compressor data are shown in Table 1. For the OCC cycle, results from two slightly different design concepts are provided to show the tradeoffs in performance for an enhanced turbomachinery design of the gas turbine. The difference between these cycles is the gas turbine exhaust temperature, which is either 590 °C or 630 °C, referred to as OCC 590 and OCC 630.

Table 1

Compressor data

HMOCOCC 590OCC 630
GT compressor power (MW)108.45104.1887.73
Compressor outlet pressure (bar)10 (after compr 1)46.9934.69
40 (after compr 2)
Compressor outlet temperature (°C)358 (after compr 1)460.5418.8
311 (after compr 2)
Compressor inlet temperature (°C)98 (compr 1)43–4443–44
136 (compr 2)
Mass flow rate (kg/s)230.1 (compr 1)242.0228.5
198.8 (compr 2)
HMOCOCC 590OCC 630
GT compressor power (MW)108.45104.1887.73
Compressor outlet pressure (bar)10 (after compr 1)46.9934.69
40 (after compr 2)
Compressor outlet temperature (°C)358 (after compr 1)460.5418.8
311 (after compr 2)
Compressor inlet temperature (°C)98 (compr 1)43–4443–44
136 (compr 2)
Mass flow rate (kg/s)230.1 (compr 1)242.0228.5
198.8 (compr 2)

It is seen that the HMOC compressor power requirement for generating an outlet pressure around 40 bar is higher than in the OCC 590 case, despite the latter having a higher outlet pressure and that the HMOC compressor is intercooled. The main reasons for this high power consumption are probably that the compressor inlet temperature in the HMOC cycle is much higher, since no external cooling has been applied to the gas stream prior to the low pressure (LP) compressor inlet, and that the moisture content is relatively high due to the high media pressure and temperature in the cooling/extraction part of the cycle. Since the HMOC compressor is intercooled, one would expect that the required compressor power would be lower than for the OCC compressor. However, this could perhaps be partly offset by the fact that the HMOC compressor also performs part of the CO2 compression work that in the OCC is performed by the CO2 compression train.

In the HMOC cycle, the bleed stream is extracted after the intercooler between the compressors, where the moisture content in the working fluid also is controlled. The water mole fraction in the LP compressor is 0.403 whereas in the high pressure (HP) compressor it is reduced to 0.359. The mass flow difference in the simulated case is 31.3 kg/s, which means that 31.3 kg/s excess water in the gaseous phase are compressed from atmospheric pressure to 10 bar in the LP compressor. This generates additional compressor work to be compensated for by the utilization of the heat of vaporization in the Rankine cycle.

The temperature in the cooling/extraction part of the OCC cycle should be >40 °C. The heater in front of the LP compressor raises the temperature to 43–44 °C, i.e., significantly lower than for the HMOC cycle. The slight preheating of 3–4 °C is included in order to reduce the relative humidity of the inlet stream down to 80% to avoid condensation in the compressor inlet.

Power Generation Data.

Power generation data are shown in Table 2. As seen, the steam turbine power is more than 10 MW higher in the HMOC cycle than in any of the two OCC cycles due to utilization of the heat of vaporization in the Brayton cycle. Another effect of this is that the LP steam mass flow becomes very high in the HMOC cycle generating problems to use a one casing steam turbine train.

Table 2

Power generation

HMOCOCC 590OCC 630
Power turbine power (MW)98.54106.6101.0
Power turbine inlet temperature (°C)879912.3948.8
Steam turbine power (MW)70.4655.2858.57
HP steam pressure (bar)6695.092.0
HP steam mass flow (kg/s)26.839.843.8
LP steam pressure (bar)2.74.598.0
LP steam mass flow (kg/s)74.07.234.42
Plant Net efficiency (%)41.948.147.9
HMOCOCC 590OCC 630
Power turbine power (MW)98.54106.6101.0
Power turbine inlet temperature (°C)879912.3948.8
Steam turbine power (MW)70.4655.2858.57
HP steam pressure (bar)6695.092.0
HP steam mass flow (kg/s)26.839.843.8
LP steam pressure (bar)2.74.598.0
LP steam mass flow (kg/s)74.07.234.42
Plant Net efficiency (%)41.948.147.9

Oxygen Production Data.

The ASU power is slightly higher and the O2 compression power is slightly lower for the HMOC than for the OCCs, but there is no significant difference in the auxiliary power consumption for the different cycles (cf. Table 3).

Table 3

Oxygen production data

POWER (MW)HMOCOCC 590OCC 630
ASU power17.5716.6516.72
O2 compression power11.8214.0512.34
Sum29.3930.7029.06
POWER (MW)HMOCOCC 590OCC 630
ASU power17.5716.6516.72
O2 compression power11.8214.0512.34
Sum29.3930.7029.06

Oxyfuel Plant Concepts: Discussion and Conclusions

The preferred design of the SCOC-CC is the OCC cycle working with a flue gas temperature entering the HRSG of 630 °C. The OCC has clearly higher efficiency than the HMOC without any clear disadvantages in either design complexity or cost. The reduction in efficiency by changing the HRSG entry temperature from 590 °C to 630 °C is more than well-compensated by the enhancement in the turbomachinery design. The main reason for the lower efficiency of the HMOC ought to be found in how the cycle concept affects the gas turbine compressor work in addition to the lower temperature of the WF entering the combustion chamber. The choice of the design parameters for these cycles is triggered mainly by the following reasons:

  • High pressure ratios are needed for this type of cycle but the experience of high pressure ratios is limited.

  • The efficiency versus pressure ratio gradient in the OCC cycle is not that steep, and a high oxyfuel turbine exhaust gas temperature can in the future be, to some extent, compensated for by high steam turbine inlet temperatures, around 600 °C. In this study, the live steam temperature was restricted to 565 °C due to present limitations of live steam temperature for this size of steam turbines.

  • High pressure ratio also generates higher working media mass flow, which indicates that the specific work output is going down as pressure is increased.

A pressure ratio of about 35 is chosen as the most favorable for the future development of this concept. This generates a plant efficiency slightly below 48%. This is, according to the reasoning above, a suitable design point for the cycle, which limits the leaps in technology for an implementation of these ideas into hardware.

The increase in TIT from 1340 °C to 1420 °C emphasizes the need for really high pressure ratios, and still the Brayton cycle exhaust temperature tends to become too high to be efficiently utilized in the bottoming cycle. The increase in efficiency is very limited compared to the increased cost of the equipment and the increased maintenance of the Brayton cycle gas path parts. Designing the cycle for very high pressure ratios (about 50) for TIT 1420 °C is not justified by the gain in efficiency. Higher Brayton cycle TIT must include more innovative cycle solutions to be justified.

Some of the arguments from the OCC results are also valid for the HMOC. A pressure ratio of about 40 is chosen as the most favorable for the future development of the HMOC concept. This generates a plant efficiency slightly below 42%.

The opening statement that oxyfuel power cycles are only suitable for carbon capture plants and will never be able to compete with conventional power plant cycles in terms of efficiency is still valid. The calculated efficiency for the OCC close to 48% still keeps the question open if there is a market for this type of plant if CO2 is a commodity. To generate such relatively high plant efficiencies from an oxyfuel power plant, the Brayton cycle equipment ought to be designed for a pressure ratio in the range of 35–40 combined with a TIT of 1340 °C. Design parameters in that range generate favorable plant performance if future developments of power plant equipment for this type of concepts are pursued. This statement is of course linked to the validity of the prerequisites. If breakthroughs in technology affecting the prerequisites are performed, these assumptions may change. To generate a better established view on the possibilities regarding this type of plant, the next step ought to be a more thorough investigation of the conditions in the rotating equipment regarding gas path design.

Combustion Concept

The aim of the present work on the OXYGT combustion concept is to develop a design basis on how to arrange the different streams into the combustor matching both the requirements of flame stability, temperature level, and heat transfer, as well as complying with the relevant emission requirements. This is done using thermodynamic and chemical kinetic analysis both on a global basis and by reactor network representation of the combustor.

The combustor of the OXYGT gas turbine will have three inlet streams; the fuel which in the present work is natural gas, the working fluid from the compressor consisting mainly of CO2 and H2O recycled from the exhaust, and the separate oxidizer stream from the ASU consisting mainly of oxygen. This means that a new burner and combustor concept have to be developed. The three-stream configuration seems to be more complicated than in the air-case. On the other hand, it also provides more freedom with respect to injection, mixing, and combustor operation over the gas turbine load range.

Combustion behavior studies relevant to the OXYGT gas turbine have been done theoretically using thermodynamic and chemical kinetic numerical tools (e.g., Refs. [2,3]), as well as experimentally at small scale (e.g., Refs. [4,5]). From these studies, some main challenges can be highlighted.

First, the combustor should preferably operate close to stoichiometric conditions in order to save costly oxygen. This may result in improper fuel burn-out and excessive CO emissions. This near stoichiometric combustion requirement is a large difference compared to traditional air-based gas turbines, which operate at large O2 excess.

Next, the reactivity and flame stability in the primary zone are affected and will be highly dependent on the local ratio of oxygen to CO2. The reactivity of a combustible mixture can be evaluated by the laminar flame speed SL. As shown in Ref. [4], the laminar flame speed for stoichiometric combustion of methane in an O2/CO2 mixture is just around 10 cm/s when the volumetric O2 concentration is as for air. This is only one-fourth compared to methane-air and implies higher possibility of poor flame stability. In order to achieve the same flame speed as for the air-case, the oxygen concentration in the O2/CO2 mixture has to be raised to about 38 vol. %. However, this also increases the adiabatic flame temperature to about 2110 K, being 160 K above the methane-air value. The challenge appears to be the trade-off between excessive local temperature and flame stabilization. Optimal design will need different oxygen/working fluid ratios in the different combustor zones in order to match the different temperature and stabilization requirements.

Further, in order to save energy in the ASU, the oxidizer stream will contain small amounts of impurities (N2 and Ar). Nitrogen, combined with the possibility of local high temperatures in the flame zone, may generate thermal NOx.

Finally, the heat transfer will be altered. The emissive power of the combustion products in oxyfuel combustion will be about five times higher compared to the air case. Heat transfer analysis of the gas turbine hot gas path [6] shows that for a SCOC-CC case at pressure ratio 40 and with recirculation after condensation, thermal radiation increases with 30% compared to a natural gas fired CCPP case at pressure ratio 17. The convective heat transfer may also be affected, depending on the final combustor design with respect to flow conditions, velocities, and cooling schemes.

Combustor Representation and Simulation Tools

A schematic diagram of the model combustor is shown in Fig. 3. In the present combustion analysis model, the WF from the compressor can be split in three streams. One stream a1 is going through the burner together with fuel and oxidizer. The stream a2 is also fed to the primary zone while stream a3 is the main dilution stream fed downstream the primary zone. One main target of the analysis is to evaluate the ratio between these WF streams.

Fig. 3
Schematic model of the combustor
Fig. 3
Schematic model of the combustor
Close modal

The arrows a1, a2, and a3 show the positions of the WF streams into the combustor. In addition, they are distributed evenly around the combustor circumferentially. The composition of the WF streams used in the combustion calculations is as given in Table 4. It contains mostly CO2 plus some small amount of water vapor, argon, nitrogen, and a minor amount of oxygen. The arrow into the center of the burner shows the inlet position of the fuel stream. Its composition, as used in the combustion calculations, is given in Table 4. The fuel stream contains mainly methane with smaller amounts of ethane, propane, and nitrogen. The oxidizer stream inlet is indicated with an arrow, and its composition is given in Table 4 as well. In addition to oxygen, it contains some smaller amounts of nitrogen and argon.

Table 4

Inlet streams to the combustor section

ItemUnitOCCBase cases
Fuel stream
CH4vol. %92.692.6
C2H6vol. %4.54.5
C3H8vol. %2.22.2
N2vol. %0.70.7
Temperature °C1010
Oxidizer stream
O2vol. %95.095.0
N2vol. %2.02.0
Arvol. %3.03.0
Temperature °C33340
WF stream
CO2vol. %90.3100
H2Ovol. %1.80
Arvol. %5.10
N2vol. %2.70
O2vol. %0.060
Temperature °C42225/400
Combustion chamber
Pressurebar(a)37 (34a)1/5/2010/30/40
TIT °C1400 (1340a)
Equivalence ratio0.9951.0
O2 excessvol. %0.50
ItemUnitOCCBase cases
Fuel stream
CH4vol. %92.692.6
C2H6vol. %4.54.5
C3H8vol. %2.22.2
N2vol. %0.70.7
Temperature °C1010
Oxidizer stream
O2vol. %95.095.0
N2vol. %2.02.0
Arvol. %3.03.0
Temperature °C33340
WF stream
CO2vol. %90.3100
H2Ovol. %1.80
Arvol. %5.10
N2vol. %2.70
O2vol. %0.060
Temperature °C42225/400
Combustion chamber
Pressurebar(a)37 (34a)1/5/2010/30/40
TIT °C1400 (1340a)
Equivalence ratio0.9951.0
O2 excessvol. %0.50
a

Optimized cycle value.

The oxyfuel combined cycle was the preferred alternative from the cycle analysis, and data only for this cycle will be presented. In addition a base case has been analyzed to check the influence of the WF temperature and the effect of combustor pressure. For simplicity, this base case is run with pure CO2 as WF as shown in Table 4.

The pressure and TIT deviate somewhat from the final optimized values in the cycle analysis. The combustion study has used preliminary data from the thermodynamic cycle analysis (37 bar/1400 °C) instead of the data from the optimized cycle analysis (34 bar/1340 °C) as the optimized cycle data were not available when the combustion calculations were started. It has been shown by a parametric variation on pressure and WF dilution that these deviations have no significant influence on the conclusions of the combustion analysis.

In the present report, the CHEMKIN-PRO combustion calculation package [7] has been used to calculate combustion properties. The software uses idealized homogeneous reactors as the basis. Inhomogeneity, flow effects, and incomplete mixing will not be reflected in the results. However, the results will give a good representation of the chemical kinetics and flame temperatures. By arranging several reactors in a reactor network, it is possible to mimic the combustor in a better way.

As a chemical kinetic reaction mechanism, the GRI-Mech 3.0 [8] has been used. It consists of 325 reactions and 53 species and is one of the most widely used mechanisms for gas combustion.

Basic Combustion Properties

The adiabatic flame temperature Tad and the laminar flame speed SL are seen as the two most important combustion parameters with respect to evaluating the trade-off between flame temperature and flame stability. In the air-case, these parameters are commonly presented as a function of the mass of air to the mass of fuel ratio (AFR), which is a measure of the flame dilution.

In the present work, an equivalent term for oxyfuel combustion has been defined as the mass of the WF plus oxidant to the mass of fuel (mWF + mOX) / mF. This term is named WOF ratio (WF plus oxidant to fuel) and analogous to the AFR. As long as the oxygen is supplied at least in stoichiometric proportion, the additional oxygen and WF act as a diluent to the flame in terms of thermodynamics, whether it is CO2 and H2O in the oxyfuel case or N2 in the air case. The stoichiometric amount of oxygen for the fuel in Table 4 is 3.90 kgO2/kgfuel. The difference between the AFR and WOF ratio in order to achieve equal value of some combustion property is just a consequence of the different properties of the diluting components, e.g., such as CO2 having higher heat capacity than N2.

The laminar flame speed SL and adiabatic flame temperature Tad for the base case are shown in Figs. 4 and 5 as function of combustor pressure and WF inlet temperature versus dilution represented by the WOF ratio on the abscissa axis. In these calculations, the combustor is treated as a perfectly stirred reactor (PSR) without any flow splits.

Fig. 4
Laminar flame speed for the base case as a function of dilution with working
                        fluid
Fig. 4
Laminar flame speed for the base case as a function of dilution with working
                        fluid
Close modal
Fig. 5
Adiabatic flame temperature for the base case as a function of dilution with
                        working fluid
Fig. 5
Adiabatic flame temperature for the base case as a function of dilution with
                        working fluid
Close modal

For equal dilution and WF temperature, the laminar flame speed decreases with increasing pressure. Therefore, the oxyfuel case follows the general trend known from air-based systems. The difference in flame speed between the high and low pressure decreases when the WOF is increased. The flame speed difference between the 30 bar and the 40 bar cases is generally very small and in principle negligible at the higher WOF ratios.

The adiabatic flame temperature shows an increase with pressure for equal dilution and WF temperature. Also, here the difference between the 30 bar and 40 bar cases is very small and vanishes more or less completely at the higher WOF ratios.

Combustor Flow Estimates From the Basic Properties

The laminar flame speed and the adiabatic flame temperature of the OCC case are shown in Fig. 6 as a function of dilution (WOF ratio). In this case, the pressure is 37 bar and both the WF and the oxidizer streams are at elevated temperatures as given in Table 4. The equivalence ratio is kept constant at 0.995 so the variation in WOF ratio is only caused by variation in the amount of WF.

Fig. 6
Adiabatic flame temperature (above) and laminar flame speed (below) for the
                        OCC case as a function of dilution with working fluid and comparison with
                        relevant data from an air case
Fig. 6
Adiabatic flame temperature (above) and laminar flame speed (below) for the
                        OCC case as a function of dilution with working fluid and comparison with
                        relevant data from an air case
Close modal

A representative gas turbine air case is also shown in the same figure as a function of AFR. The lower AFR point of 33.5 represents a lean primary flame zone, and the high AFR point of 41.8 represents the combustor outlet conditions. The fuel composition is the same as for the OCC case.

A first estimate on the WF flow split for the OCC combustor has been made based on a comparison between the adiabatic flame temperatures and the laminar flame speeds of the OCC case and the air case. First, the amount of dilution to the OCC combustor primary zone (a1 + a2 in Fig. 3) is found from the requirement of having the same laminar flame speed as the primary zone of the air case as indicated by the arrows in the lower part of Fig. 6. This will, however, create a much higher temperature compared to the air case but is nevertheless chosen as a starting point. The additional WF to be fed to the dilution zone (a3 in Fig. 3) is decided from the requirement of having the same outlet temperature as the air case as indicated with arrows in the upper part of Fig. 6.

The results of this estimation are summarized in Table 5. Some important design issues can be highlighted: First, the WF split to the dilution zone (a3 in Fig. 3) should be about half (0.52) of the total WF to the combustor.

Table 5

Combustor flow estimates

Air caseOxy case
Power input (fired heat)MW260260
Fuel heat content (LHV)MJ/kg48.348.3
Fuel mass flowkg/s5.45.4
Combustor pressurebar a22.7537.0
Air mass flow to PZkg/s179.9
Working fluid mass flow to PZkg/s86.4
Oxidizer mass flow to PZkg/s22.3
Burnt mixture leaving PZkg/s185.2114.1
Burnt mixture leaving PZm3/s43.915.3
Dilution stream (a3, air or WF)kg/s45.094.8
Air for dilution/air total0.20
WF for dilution/WF total0.52
Burnt mixture leaving combustorkg/s230.2208.8
Burnt mixture leaving combustorm3/s48.919.3
Burnt mixture densitykg/m34.710.8
Burnt mixture sound speedm/s789633
Burnt mixture Prandtl number0.810.86
Air caseOxy case
Power input (fired heat)MW260260
Fuel heat content (LHV)MJ/kg48.348.3
Fuel mass flowkg/s5.45.4
Combustor pressurebar a22.7537.0
Air mass flow to PZkg/s179.9
Working fluid mass flow to PZkg/s86.4
Oxidizer mass flow to PZkg/s22.3
Burnt mixture leaving PZkg/s185.2114.1
Burnt mixture leaving PZm3/s43.915.3
Dilution stream (a3, air or WF)kg/s45.094.8
Air for dilution/air total0.20
WF for dilution/WF total0.52
Burnt mixture leaving combustorkg/s230.2208.8
Burnt mixture leaving combustorm3/s48.919.3
Burnt mixture densitykg/m34.710.8
Burnt mixture sound speedm/s789633
Burnt mixture Prandtl number0.810.86

The volumetric flow into the primary zone is much smaller for the OCC case than the air case. This may affect aerodynamics such as sizing of swirl and recirculation. The volumetric flow of burnt mixture out of the primary zone is just above one-third of the air case. The total volumetric flow out of the combustor is less than half of the air case.

It should be noted that the results from these estimates are highly dependent on the prerequisites made, the most important one being the laminar flame velocity in the primary zone (PZ). However, this calculation gives several insights into the complexity of converting a gas turbine from air to oxyfuel mode as exemplified below.

Even though the volumetric flow for the OCC case is much smaller than the air case, it does not necessarily imply a considerable reduction in size. One interesting exercise is to look at the Reynolds number at the combustor outlet for the air and OCC case in Table 5. The dynamic viscosity is nearly the same; about 5.9 × 10−5 for both streams. If, as an example, the Reynolds number were to be kept constant for the two cases it would imply that (V * D)OCC = 0.43 * (V * D)air, where V is the flow velocity and D is the related flow diameter. The constraint giving equal Reynolds number for these two cases will then be DOCC = (Dair/0.43) * (VFOCC/VFair), where VFOCC and VFair are the volumetric flows in the oxyfuel combined cycle case and the air case, respectively. For the volumetric flows given at the combustor outlet (i.e., “burnt mixture leaving combustor” in Table 5) this means that DOCC = 0.92 * Dair. Therefore, the flow diameter for the OCC case should be only marginally smaller than for the air case if the Reynolds numbers were to be equal. In this case, the velocity of the OCC case must be about half of the air case meaning that the residence time will increase, something which may be beneficial for fuel and CO burn-out.

Under the constraint of equal Reynolds number, the Mach number for the OCC case would be just 58% of the air case at the combustor outlet. In order for the Mach numbers to be equal, the OCC diameter should be 0.7 times the air value. Because of the lower sound speed in the OCC case, the velocity will of course still be lower than the air case, but the Reynolds number will be about 31% higher.

The convective heat transfer coefficient increases with both the Reynolds and Prandtl number. An increase in Reynolds number would, thus, mean that the convective heat transfer would increase since the Prandtl numbers are fairly equal. This increase will add to the possible increase in thermal radiation.

Reactor Network Calculations

A reactor network setup of the model combustor configuration in Fig. 3 was made in the CHEMKIN-PRO package. This gives a better representation of the combustor than one single PSR since the flow and chemistry interaction will be more realistic. The reactor network representation is shown in Fig. 7. The fuel, a share a1 of the WF and most of the oxidizer stream (75%) are fed to a near burner mixing zone with short residence time. This zone is modeled as a PSR. A small portion (30%) of recirculated hot gas from the flame zone is also fed back to the mixing zone. Next follows the main flame zone, which is also modeled as a PSR. Here, the rest of the oxidizer is fed together with the share a2 of the WF and 70% of the recirculated hot gas. 20% of the gas leaving the flame zone reactor is sent back as hot gas recirculation. Thereafter the dilution zone follows where the share a3 of the WF is injected. No streams enter after the dilution zone. The next zone is a post dilution zone modeled as a plug flow reactor (PFR). Finally, the flow in the turbine stages is included in the model in order to take into account any slow chemistry. Since CO burn out is expected to be challenging in oxyfuel combustion due to the high concentrations of CO2, the additional residence time in the turbine can affect the final composition at the gas turbine outlet. The turbine PFR has a gradual pressure and temperature decrease down to the turbine outlet conditions of about 630 °C and nearly atmospheric pressure.

Fig. 7
Reactor network representation of the model combustor
Fig. 7
Reactor network representation of the model combustor
Close modal

The network calculations were done for a large range of WF splits with the purpose to find optimum values from an emission point of view and to verify whether these values match the estimated value of the a3 share of about 0.5 found in the Combustor Flow Estimates From the Basic Properties section. The different values of a3 calculated are 0.15, 0.25, 0.40, 0.60, and 0.80. For each of these a3 values, the share between a1 and a2 were varied so that a1/(a1 + a2) went from 0.05 to 0.95.

For some cases, with low a1 there is a very rapid ignition already in the mixing zone and the temperature immediately becomes very high. Except from this the results clearly indicate that the share between a1 and a2 is of minor importance compared to the variation in a3. The rest of the results are, therefore, only for the cases where a1 and a2 are equal so that a1/(a1 + a2) is 0.5. For the species concentrations only the results downstream the dilution zone are shown. Then all the feed streams have entered the combustor and the concentrations are directly comparable. Concentrations are given as ppmv (parts per millions volumetric) without standard corrections as is normally done for air-based cases.

Figures 8 and 9 show the temperatures and CO concentrations. A high amount of WF in the primary zone (low a3) gives low temperatures in the mixing and flame zone. On the contrary, a low amount of WF to the primary zone (high a3) causes excessively high temperature in the flame zone. The CO in the dilution zone becomes high and so does the O2 concentration (not shown here). As discussed in Ref. [1] this is probably due to the increased importance of the dissociation reactions of H2O and CO2 at high temperatures, which results in higher emissions of CO and O2. In addition this high flame zone temperature generates a considerable amount of NOx, the level being around 800 ppm at the outlet.

Fig. 8
Temperature through combustor and turbine as a function of the working fluid
                        share a3 to the dilution
Fig. 8
Temperature through combustor and turbine as a function of the working fluid
                        share a3 to the dilution
Close modal
Fig. 9
CO, O2, and NOx concentration as a function of the
                        working fluid share a3 to the dilution zone
Fig. 9
CO, O2, and NOx concentration as a function of the
                        working fluid share a3 to the dilution zone
Close modal

The case that seems to give the best results is the one with a3 equal to 0.4. It shows the lowest CO values, the temperature in the mixing zone indicates more rapid ignition than the other cases, and the flame zone temperature is below 1800 °C. The NOx concentration at the outlet (not shown here) is only about 2 ppm. Even though this case provides the lowest CO emission among the cases calculated here, the actual value of about 500 ppm is still rather high. In addition, the CO value is not at a stationary level at the outlet of the combustor and some CO oxidation continues through the turbine. This is not desirable.

Better CO burn-out can be achieved with higher O2 excess. The case of a3 = 0.4 has been run again but now with different oxygen excess, ranging from 0.5 vol. % to 5.2 vol. % (equivalence ratio from 0.995 to 0.95). In addition, the total amount of WF has been slightly increased in order to reduce the TIT from about 1400 °C as in the former case down to about 1340 °C, which is the value from the more optimized cycle.

The results are shown in Fig. 10. First, the increase in total WF contributes to a slight decrease in both temperature and CO as seen from the curves that are comparable with the results of Fig. 8 and Fig. 9 (excess O2 equal to 0.5 vol. %). Next, the increase in O2 excess reduces the CO significantly and less CO burn-out is happening in the turbine. However, the reduced CO comes at the cost of increased O2 production in the ASU and O2 emissions.

Fig. 10
CO and O2 concentration as a function of oxygen excess for the
                        case with a3 equal to 0.4
Fig. 10
CO and O2 concentration as a function of oxygen excess for the
                        case with a3 equal to 0.4
Close modal

It should be noted again that the present calculations use idealized homogeneous reactors and the effects of inhomogeneity and incomplete mixing will not be captured. In the real case, it is likely that such effects will be present to some degree and most probably contribute to increased CO emissions. The combustion calculation package used in the present work (CHEMKIN-PRO) does have the possibility to use nonhomogeneous reactors where incomplete mixing can be taken into account through mixing models and choice of mixing parameters. This is a sensitivity study on its own and has not been included in the present paper but is relevant for further work.

Overall Combustor Geometry

The concept of the test combustor incorporates features that have been gained from rules used in Siemens for design of gas turbine combustors. Similar design principles will be applied for both testing in the HP test rig at SINTEF and in an anticipated continuation in a project Phase 2 in a small scale gas turbine pilot plant at Lund University. Overall, the design scheme of the test combustor is shown in Fig. 11, which also shows the flow path of the working fluid.

Fig. 11
Overall design features of the OXYGT test combustor
Fig. 11
Overall design features of the OXYGT test combustor
Close modal

The basic features of the OXYGT combustor are as follows:

  • burner with radial swirler: swirl number close to the critical, i.e., the flame will be stabilized by the central recirculation zone

  • fuel is injected through fuel pegs in the inlet to the radial swirl generator

  • secondary fuel is injected axially in the center of the burner

  • oxidizer (O2) is mixed with the WF in a dome upstream the swirl generator with the help of a flow conditioner for a homogeneous mixture

  • cylindrical combustion chamber (can or silo) with a smooth expansion (quarl) between the burner outlet and the flame tube

  • combustor liner is cooled with the dilution portion of WF (a3) by convection

  • the dilution WF enters the hot gas duct in the transition part, which is also a mechanical connection to the outlet duct in the test rig and to the turbine nozzle guide vane in the pilot turbine

Combustion Concept: Discussion and Conclusions

The OXYGT combustion concept has first been evaluated by a simplified approach matching the laminar flame speed and the adiabatic flame temperature of the oxyfuel case with a relevant reference air case. The results show that the WF split should be 0.48/0.52 through the burner and for dilution, respectively. In this case, the volumetric flow of burnt mixture out of the primary zone is just above one-third of the air case whereas the total volumetric flow out of the combustor is less than half of the air case.

Reactor network calculations have been done to evaluate temperatures and emissions. In this analysis, it is found that the WF split should be around 0.6 through the burner and 0.4 for dilution. The flame zone temperature is below 1800 °C, and the NOx concentration at the combustor outlet is only about 2 ppm. A higher WF share through the burner would reduce the flame zone temperature further, but the laminar flame speed will then be low and flame stability will be affected.

CO burn-out is very dependent on the O2 excess. At an O2 excess of 0.5% compared to stoichiometric value, CO is 600 ppmv at the combustor outlet and 400 ppmv at the turbine outlet. At an O2 excess of 5.2% the CO values are 43 and 20 ppmv at combustor and turbine outlet, respectively. In this case, a lot of extra O2 has to be produced in the ASU, being costly in terms of efficiency loss.

It should be noted that in the oxyfuel case, the exhaust is not liberated to air but will go to storage or be used for EOR. There will be other requirements on the composition and values have been provided both from the Dynamis project [9] and from NETL [10]. The values from the reactor network analysis fall generally well within the given limits except for the CO limit given by NETL and for the O2 concentration if the stream is to be used for EOR. The CO limit given by NETL is 35 ppmv and is just the time weighted average concentration. This seems to be a very stringent limit for the concentration within a pipeline since any leakage will be diluted, which is used as methodology by Dynamis. The Dynamis requirement of 2000 ppmv CO will be of no problem according to the reactor network analysis.

A schematic design of the OXYGT combustor has been provided. The burner will be swirl stabilized by radial swirlers, and there will be a cylindrical combustion chamber (can or silo) with a smooth expansion between burner outlet and the flame tube. The combustor liner is to be cooled with the dilution portion of the WF by convection. The next step will be to manufacture a burner–combustor arrangement that will be tested both at atmospheric and pressurized conditions in a dedicated oxyfuel test facility [11].

Summary and Conclusions

The results from the cycle comparison show that the more straight forward OCC cycle with low steam content has a clearly higher efficiency (∼48%) without any clear disadvantages in either design complexity or cost than the more advanced HMOC cycle with high steam content in the WF that generates a lower efficiency (∼42%). The preferred design for the SCOC-CC is the OCC cycle working with a flue gas temperature entering the HRSG of 630 °C. There is a slight reduction in efficiency (0.2%) in the OCC by changing the HRSG entry temperature from 590 °C to 630 °C, but this is more than well compensated for by the enhancement in the turbomachinery design.

The main reason for the lower efficiency of the HMOC ought to be found in how the cycle concept affects the gas turbine compressor work, in addition to the lower temperature of the WF entering the combustion chamber. The LP compressor work is increased due to the higher temperature level at the compressor inlet and the ballast steam that is condensed and extracted after the LP compressor.

These drawbacks are not fully compensated by the reduced power consumption in the CO2 compression train, the HP compressor, and the power generated from the heat of condensation in the LP steam cycle.

The general conclusion from this oxyfuel cycle comparison is that the OCC seems more promising from a number of reasons and that the future combustion development will be concentrated on this cycle.

The OXYGT combustor concept has been evaluated both by a simplified approach using laminar flame speed and flame temperature of an air case as basis and by more refined reactor network analysis using the CHEMKIN software package. The results indicate that the WF share going to the burner should be in the range 0.5–0.6, while the rest is used for dilution downstream of the primary flame zone. This WF fluid split gives the best compromise between flame zone temperature, flame zone laminar flame speed, and flame stability as well as emission levels. NOx emissions are low, about 2 ppm, while the CO emissions are very much dependent on the O2 excess. An O2 excess of 0.5% compared to stoichiometric value gives CO emissions of 400 ppm while an excess of 5.2% gives 20 ppm. This additional O2 will imply higher duty on the ASU, which is costly in terms of efficiency.

The exhaust will be subject to storage or EOR. Compared to requirements and guidelines given by the EU project Dynamis and NETL of the US, the present numerical reactor network analysis falls generally well within the given limits except for the CO limit given by NETL and for the O2 concentration if the stream is to be used for EOR.

A schematic design of the OXYGT combustor has been provided. The burner will be swirl stabilized and there will be a cylindrical combustion chamber. The combustor liner cooling will be provided by the dilution portion of the WF by convection. The next step will be to manufacture a burner and combustor arrangement that will be tested both at atmospheric and pressurized conditions in a dedicated oxyfuel test facility.

Acknowledgment

The research leading to these results has received funding from the CLIMIT program in Norway through Project No. 212784 OXYGT, in addition to contribution from the partners. The authors acknowledge the partners SINTEF Energi AS, Siemens Industrial Turbomachinery AB, Siemens AS, Nebb Engineering AS, and Lund University.

Nomenclature

    Nomenclature
     
  • AFR =

    air/fuel ratio

  •  
  • ASU =

    air separation unit

  •  
  • a1, a2, a3 =

    stream split of WF

  •  
  • CCPP =

    combined cycle power plant

  •  
  • CCS =

    carbon capture and storage

  •  
  • D =

    diameter (m)

  •  
  • EOR =

    enhanced oil recovery

  •  
  • GT =

    gas turbine

  •  
  • HMOC =

    high moisture oxyfuel cycle

  •  
  • HP =

    high pressure

  •  
  • HRSG =

    heat recovery steam generator

  •  
  • ITM =

    ionic transport membrane

  •  
  • LHV =

    lower heating value (kJ/kg)

  •  
  • LP =

    low pressure

  •  
  • m =

    mass (kg)

  •  
  • MEA =

    monoethanolamine

  •  
  • NOx =

    nitrogen oxide

  •  
  • OCC =

    oxyfuel combined cycle

  •  
  • OX =

    oxidant

  •  
  • OXYGT =

    oxyfuel gas turbine cycle feasibility project

  •  
  • PFR =

    plug flow reactor

  •  
  • ppmv =

    parts per millions volumetric

  •  
  • PSR =

    perfectly stirred reactor

  •  
  • PZ =

    primary zone

  •  
  • Q =

    specific heat flow (kW/kg)

  •  
  • SL =

    laminar flame speed

  •  
  • SCOC-CC =

    semiclosed oxyfuel combustion combined cycle

  •  
  • T =

    temperature (°C)

  •  
  • TIT =

    turbine inlet temperature (°C)

  •  
  • UHC =

    unburnt hydrocarbons

  •  
  • V =

    velocity (m/s)

  •  
  • VF =

    volumetric flow (m3/s)

  •  
  • WF =

    working fluid

  •  
  • WOF =

    ratio working fluid plus oxidant to fuel ratio

Subscripts

    Subscripts
     
  • ad =

    adiabatic

  •  
  • air =

    air stream

  •  
  • L =

    laminar

  •  
  • OCC =

    oxyfuel combined cycle

  •  
  • OX =

    oxidant

References

1.
Jericha
,
H.
,
Sanz
,
W.
, and
Göttlich
,
E.
,
2006
, “
Design Concept for Large Output Graz Cycle Gas Turbines
,”
ASME
Paper No. GT2006–90032.10.1115/GT2006-90032
2.
Liu
,
C. Y.
,
Chen
,
G.
,
Sipöcz
,
N.
,
Assadi
,
M.
, and
Bai
,
X. S.
,
2012
, “
Characteristics of Oxy-Fuel Combustion in Gas Turbines
,”
Appl. Energy
,
89
(1), pp.
387
394
.10.1016/j.apenergy.2011.08.004
3.
Hasegawa
,
T
.,
2012
, “
Combustion Performance in a Semiclosed Cycle Gas Turbine for IGCC Fired With CO-Rich Syngas and Oxy-Recirculated Exhaust Streams
,”
ASME J. Eng. Gas Turbines Power
,
134
(9), p.
091401
.10.1115/1.4006985
4.
Ditaranto
,
M.
, and
Hals
,
J.
,
2006
, “
Combustion Instabilities in Sudden Expansion Oxy-Fuel Flames
,”
Combust. Flame
,
146
(3), pp.
493
512
.10.1016/j.combustflame.2006.04.015
5.
Kutne
,
P.
,
Kapadia
,
B. K.
,
Meier
,
W.
, and
Aigner
,
M.
,
2011
, “
Experimental Analysis of the Combustion Behavior of Oxyfuel Flames in a Gas Turbine Model Combustor
,”
Proc. Combust. Inst.
,
33
(2), pp.
3383
3390
.10.1016/j.proci.2010.07.008
6.
Hammer
,
Th.
,
Keyser
,
J.
, and
Bolland
,
O.
,
2009
, “
Natural Gas Oxy-Fuel Cycles—Part 2: Heat Transfer Analysis of a Gas Turbine
,”
Energy Procedia
,
1
(1), pp.
557
564
.10.1016/j.egypro.2009.01.074
7.
Reaction Design,
2011
, “
CHEMKIN-Pro Release 15112 2011
,” http://www.reactiondesign.com/lobby/open/index.html
8.
GRI-Mech, 1999, “
GRI-Mech Version 3.0 7/30/99 CHEMKIN Format
,” http://www.me.berkeley.edu/gri_mech/
9.
EU FP6 Project Dynamis,
2007
, “
Dynamis CO2 Quality Recommendations
,” Public Report No. 019672/D3.1.3.
10.
NETL,
2012
, “
CO2 Impurity Design Parameters
,” National Energy Technology Laboratory, Public Report No. DOE/NETL-341/011212.
11.
Ditaranto
,
M.
,
2011
, “
Description of a High Pressure Oxy-Fuel Combustion Facility HIPROX
,”
2nd Oxyfuel Combustion Conference
, Yeepoon, Australia, September 12–16.