STUDY OF LARGE HYDROGEN LIQUEFACTION PROCESS

H. Matsuda and M. Nagami
Nippon Sanso Corp., Kawasaki-ku, Kawasaki-city, Kanagawa 210


ABSTRACT

In WE-NET (World Energy NETwork) project, liquid hydrogen is the most expected energy carrier for the clean energy system which doesn't discharge carbon dioxide. For this system it is necessary to develop a large scale hydrogen liquefaction plant with high process efficiency for this system. The liquefaction capacity of 300t/day for one plant is estimated to be suitable and the target process efficiency is set to be more than 40 % Carnot. Up to now Hydrogen Claude cycle, Helium Brayton cycle, Neon Brayton cycle have been studied. Based on the results of these process calculation a trade-off evaluation has been conducted. Though the Neon Brayton cycle gives the best efficiency, Hydrogen Claude cycle and Helium Brayton cycle has been chosen for further detailed study.


1.INTRODUCTION

WE-NET project to study clean energy system started in 1994. In this project several energy carriers are investigated and hydrogen is considered the most expected one. Hydrogen gas is generated by electrolysis utilizing the electric power produced by renewable energy sources such as hydroelectric power, wind power, or solar power. Then hydrogen gas is liquefied, transported to energy consumption area, stored there, and utilized as fuel with oxygen gas for electric power generation by a combustion turbine. For this purpose a large scale hydrogen liquefaction plant with high process efficiency has to be constructed. Nippon Sanso Corp. together with Mitsubishi Heavy Industries and Teisan is responsible for the study to develop this large hydrogen liquefaction plant in this project.
When a capacity of one power generation plant is 500MW, the required amount of liquid hydrogen is about 600t/day. Taking this size into account, we have decided the capacity of one plant as 300t/day, which is still 10 times larger than the largest existing plant in the world. Allowable energy consumption for liquefaction of hydrogen has been fixed to be less than 1 kWh/Nm3 in this project, which is corresponding to about 36% Carnot efficiency. Thermodynamically the hydrogen liquefaction of 300 t/day is equivalent to the refrigeration power of about 700 kW at 4.2 K, of which Carnot efficiency will be around 38% on "the Strobridge Curve (process efficiency vs. refrigeration capacity at 4.2K)[1]". In order to study liquefaction processes, it would be convenient to ignore several factors such as pressure drop and heat loss which have to be considered in detail design stage. Therefore we have decided to set the target process efficiency to be more than 40% Carnot, which is a little higher than required.
Several liquefaction processes have been studied and the results are described in this paper, together with the result of trade-off study among the processes.


2.PROCESS CYCLES

The followings are liquefaction processes that we have studied. They have been selected not only to conduct as much wide comparison as possible but also to find the maximum process efficiency.

(1)HYDROGEN CLAUDE CYCLE
This is the most adopted cycle for large hydrogen liquefaction plants in the world. Hydrogen gas is used not only in the feed line to be liquefied but also in the recycling line which generates necessary cold for the liquefaction. The feed gas is compressed to approximately 5MPa, and then cooled down to 80K and converted to 47% para hydrogen by a ortho-para hydrogen converter (O-P converter) at the same time. From 80K to lower temperature region, continuous O-P conversion method, where the cool down and the O-P conversion to its equilibrium concentration of the feed hydrogen is done simultaneously in heat exchangers, is adopted. Finally the feed hydrogen gas is liquefied at 0.1MPa, 20.4K by expansion at the J-T valve. In order to improve the process efficiency, a process with a super critical expansion turbine installed in the feed line has been developed. Fig.1 shows the comparison between the process with and without the super critical turbine. The efficiency of the process is normalized by the efficiency of the process at the feed pressure of 5.07MPa without the super critical turbine. As shown, for the process without the super critical turbine lower feed pressure gives better efficiecy. This is because the exergy loss at the J-T valve is increased when the feed pressure is increase. On the other hand, the efficiency of the process with the super citical turbine is increased when the feed pressure is increased. In the whole pressure range it is shown that the process with the super critical turbine has better efficiency than that of the process without the super critical turbine. Because the outlet pressure of the super critical turbine is limited to above the critical pressure of hydrogen(1.315MPa) to avoid two phase condition at the outlet of the turbine, high pressure of the feed hydrogen is preferable to have a large expansion ratio. Fig.2 shows the typical process which we have studied this time. One stage O-P converter is installed after the outlet of the super critical turbine in order to avoid a large gap in para concentration at the inlet of the following heat exchanger.
Not only the room temperature compression process as Fig.2 has been studied, but also a cryogenic temperature compression process is examined. In this process recycled hydrogen is compressed by a cold compressor at 80 K and the compression heat generated is rejected to liquid nitrogen. Though required input power to the cold compressor is much reduced, required amount of liquid nitrogen is too large and therefore we can not get good process efficiency[2].

(2)HELIUM BRAYTON CYCLE
This cycle is adopted for relatively small hydrogen liquefaction plants. A typical flow of this process is shown in Fig.3. Helium gas is used in the recycling line. The pressure conditions of feed gas and the installation of the super critical turbine and O-P conversion of the feed hydrogen are almost the same as Hydrogen Claude Cycle.

(3) NEON BRAYTON CYCLE
It is difficult to compress a gas of small molecular weight to high pressure. Especially if we consider the size of the plant, the preferable type will be centrifugal compression. Hydrogen and helium used in the recycling line are not suitable in this point due to their small molecular weight, because so many compression stages are required in the centrifugal compressor. In order to reduce the number of compression stages, we have to utilize a gas of larger molecular weight as recycling media. One possibility is a Neon Brayton Cycle as shown in Fig.4. The boiling temperature of neon is about 27K, which can be used as the final cooling of hydrogen of which boiling temperature is 20.5K. The saturated temperature of liquid neon can be lowered by the reduction of the saturated pressure, which will more suitable refrigeration due to the smaller temperature difference. Though it will require lots of energy to reduce the saturated pressure by a room temperature compressor, a process with a cold pump has been developed to do the same work with smaller power input, as shown in Fig.5. The subatmospheric pressure of saturated liquid neon is set to be 0.05MPa, of which saturated temperature is 25K. This is considered to be suitable minimum because the triple point of neon is 0.043MPa.


LIQEFACTION PROCESS CONDITIONS

The followings are the conditions on which our process calculations are based.
Hydrogen liquefaction capacity: 300t/day
Pressure of feed hydrogen: 0.106MPa
Liquid hydrogen tank: Capacity 50,000kL
Pressure 0.106MPa
Evaporation rate 0.1%/day
Para hydrogen concentration>95%
Boil off gas is recovered by a room temperature compressor.
Efficiency of expansion turbine: 85% (power recovery efficiency: 90%)
Efficiency of compressor: 80%
Unit power consumption of saturated liquid nitrogen: 0.5kWh/Nm3
Unit power consumption of saturated gaseous nitrogen : 0.14kWh/Nm3
Heat loss and Pressure drops: Ignored
Target process efficiency: 40%

The process efficiency is defined by the equation below;


where h[%]: process efficiency
    Wmin[kW]: minimum work
    Wcomp[kW]: required power input to all the compressors
    WN2[kW]: power consumption corresponding to nitrogen used
    WET[kW]: recovered power from all the turbines
The minimum work is based on the theoretical required energy to liquefy hydrogen at 0.101 MPa and 300K at the same pressure by continuous O-P conversion.


PROCESS CALCULATION RESULTS

The process calculation results about the cycles in Fig.2`Fig.5 are indicated in Table 1. Each process efficiency is better than the target efficiency as shown in the table.

Table 1. Process calculation results

Item
Process
Hydrogen
Claude
Helium
Brayton
Basic Neon
Brayton
Neon with
Cold Pump
Compressor Total Power[MW]74.9585.2082.9779.10
Input power by Nitrogen[MW]35.0728.0528.3629.82
Recovered Generated Power[MW]|3.40|4.63|3.93|3.70
Total Power[MW]106.6108.6107.2105.2
Minimum Work[MW]49.549.549.549.5
Process Efficiency[%]46.445.646.247.1


TRADE-OFF STUDY

As a preceding step to further process study, we have decided to select two promising processes by a trade-off study. Each process is evaluated relatively among the processes because the absolute evaluation is difficult.

(1) EVALUATION ITEMS
‡@Process Efficiency
The higher efficiency, the higher point.
‡APlant Cost
E Equipment Cost
The evaluation is done by the basis that the complex process will be expensive.
E Refrigerant Cost
Because of the plant size, required amount of gas in the recycling line will be so large that we can not ignore its cost, or availability.
E Maintenance Cost
At maintenance it will be necessary to purge equipment after the inspection. The similar evaluation to the gas cost above.
‡BOperation and Management of Refrigerant
Evaluation is done whether refrigerant management is necessary and leakage countermeasures is easy.
‡CCompressor
The required number of the compression stages will be increased when the molecular weight is smaller.
‡DExpansion Turbine
As same as "Compressor" above.
‡EHeat Exchanger
In order to make the design and the manufacturing of heat exchangers easy, it is desirable that the volumetric flow of each fluid in a heat exchanger is nearly equal in design and manufacturing. Volumetric flow of each fluid is evaluated.
‡FSafety
It is evaluated as safer when the gas in the recycling line is not combustible.

(2)TRADE-OFF RESULTS
Each processes is classified as A, B and C from higher point for each evaluation item. For numerical evaluation it is set that class A gets 10 points, B gets 6 points, and C gets 2 points. Concerning "Process Efficiency" and "Plant Cost (equipment cost, refrigerant cost)", the points are doubled because they are the most important items. Table 2 shows the results of the evaluation of Hydrogen Claude Cycle, Helium Brayton Cycle and Neon Brayton Cycle with Cold Pump.

Table 2. Trade-off results

Hydrogen
Claude
Helium
Brayton
Neon with
Cold Pump
Process Efficiency202020
Plant Cost - equipment cost202012
refrigerant cost20124
maintenance cost10102
Operation & Management
of Refrigerant
6610
Compressor2610
Expansion Turbine6610
Heat Exchanger1026
Safety266
Evaluation Point Total96/12088/12080/120
Selection ResultsSelectedSelectedNot Selected

From the results in Table 2, Hydrogen Claude Cycle and Helium Brayton Cycle have been selected as the target process which will be examined in more detail. The process conditions such as process pressure and temperature difference in the heat exchangers at the different temperature levels have to be set to give the same effect to the process efficiency of each process as much as possible. These are required for the fair and final evaluation of the process. In this stage of the study ignored conditions such as pressure drop and heat loss will be considered. In this fiscal year the final trade-off will be conducted and along with the process study the important technology which has to be developed for construction of this large plant will be clarified and the plan for this development will be proposed.


REFERENCES

1. T. R. Strobridge, NBS TECHNICAL NOTE 655, U.S. DEPARTMENT OF COMMERCE, 1974.
2. K. Iwamoto, IHCE '95 Proceedings, 215-218, 1995.


ACKNOWLEDGMENT

The authors thank Mitsubishi Heavy Industries and Teisan for the discussion. The authors also thank ENAA and NEDO for the opportunity and the permission of this presentation.