Hydrogen admixture in the Dutch gas grid

(1)

Leeghwaterstraat 44 2628 CA Delft P.O. Box 6012 2600 JA Delft The Netherlands www.tno.nl T +31 88 866 22 00 F +31 88 866 06 30 TNO report in collaboration with the University of Amsterdam

Hydrogen admixture in the Dutch gas grid

Author Student number

Floris Taminiau 10661301

Supervisor(s) Néstor González Díez Rene Peters

Second assessor Rudolf Sprik

Institute TNO

University Universiteit of Amsterdam

Faculty Faculteit der Natuurwetenschappen, Wiskunde en Informatica (FNWI) Date of submission 10-07-2017

Number of pages 43 Number of

appendices

4

No part of this publication may be reproduced and/or published by print, photoprint, microfilm or any other means without the previous written consent of TNO.

In case this report was drafted on instructions, the rights and obligations of contracting parties are subject to either the General Terms and Conditions for commissions to TNO, or the relevant agreement concluded between the contracting parties. Submitting the report for inspection to parties who have a direct interest is permitted.

(2)

Abstract

The Paris climate agreement states that global CO2 emission must be lowered. To

achieve this, energy production needs to shift to low-carbon sources, such as wind and solar. However, these sources produce a fluctuating energy supply that causes times of deficit and surplus. At times of surplus this electricity can be used later by producing hydrogen. Hydrogen is an energy carrier that can be stored and transported and be transformed back into energy. Since hydrogen is often produced away from demand areas, the possibility of transporting hydrogen through the existing natural gas network is explored. This report will determine how much hydrogen can be admixed into the Dutch gas grid and how to further improve this percentage. Up to 20% hydrogen admixed Groningen-gas is shown to provide only slight behavioural deviations from pure natural gas. Engineering and dispatching issues such as decrease in energy density, increased flow speed and pressure losses and the subsequent need for compression power are all within operating envelopes of the hardware involved. Safety issues such as hydrogen embrittlement, leakage and combustion are shown to be elevated but do not pose a significant threat.

(3)

1 Module 1: Introduction

1.1 The hydrogen economy

The Paris 2015 convention states that global warming must be kept below a 2 °C increase from pre-industrial times. To achieve this, annual carbon emissions on a global scale need to be reduced by 85% by 2050 compared to today’s levels(Körner, Tam, Bennett, & Gagné, 2015). This means that a decarbonization of the energy system is necessary. A transition needs to happen away from carbon-rich fossil fuels (the existing system) and towards carbon-free renewable energy sources (the new system). A transition is always gradual, which means that two systems must operate at the same time and work together, with the existing system is slowly phased out. Thus, a need arises for system integration.

In the new system wind and solar energy are the main renewable energy sources, considering that natural hydro has been tapped already where potential exists. They are advantageous in comparison to conventional energy sources in that they have low life-cycle carbon emissions and that they do not deplete. In the year 2100 the wind will still be blowing and the sun will still be shining. However, at one particular location, they do so intermittently. At night, the sun does not shine and sometimes the wind may not blow. Our demand for energy unfortunately does not follow this intermittent supply, generally it is quite the opposite. Besides the day-cycle, in winter, we also demand more energy to heat our homes, but there is less sunshine. This mismatch causes periods of seasonal supply surplus and deficit.

To solve this mismatch there is a need for energy conversion, storage and transport, which calls for an energy carrier. A good candidate is hydrogen. Hydrogen is a flexible and clean energy carrier that can be used for a wide range of applications. Flexible because it can be stored for a long time and be transported over long distances with minimal efficiency costs, and clean because it contains no carbon, and thus emits no greenhouse gasses in its combustion.

Hydrogen can be produced from water by a process called electrolysis. In this process water is split into hydrogen and oxygen by exploiting their opposite charge.

2 𝐻2O(𝑙) → 2 H2(𝑔) + O2(𝑔)

Electricity is needed for this reaction, which can be provided in multiple ways. When the hydrogen is produced from renewable energy sources such as wind or solar, it is called green hydrogen. The hydrogen generated at times of surplus electricity can be stored in large quantities over long periods and be re-transformed to electricity (power-to-power), at an efficiency of about 30% of the original energy input(Körner, Tam, Bennett, & Gagné, 2015). This provides the new energy system, whose supply is dependent on location and time of day, the flexibility to adjust to our temporal change in demand. Hydrogen can also be converted to synthetic methane (power-to-methane), be used directly as fuel for transport (power-to-fuel) or as building block for other highly valued chemicals, such as ammonia or methanol (power-to-liquids). All these uses make hydrogen a key player in integrating the existing and the new energy systems.

(6)

This integration of hydrogen into our energy supply is called the hydrogen economy, and can only be realised when all sectors work together. The production, market, infrastructure and consumers need to be reorganized simultaneously.

1.2 Outline and motivation

This report will be focused on the infrastructure reorganization and improvement. In the Netherlands, there is a big opportunity besides a pressing need for hydrogen. On the North-Sea there are many wind parks producing electricity for onshore usage. These parks transport their electricity through cables towards land where it is connected to the electricity grid. On the North-Sea there are also still many gas producing and processing platforms that are used to collect natural gas, which is transported to shore using a large network of pipelines. Simultaneously, production from offshore assets is rapidly declining. Article 44 of the Dutch mining law states that an unused mining facility must be removed. The removal of a small platform costs about EUR 10 million, a big one can cost up to 10 times more (Bremmer, 2016). This situation creates a perfect opportunity for finding smart combinations and system integration: green hydrogen can be produced on these platforms from wind power at times of surplus electricity, and be transported through the existing pipelines to shore. This way hydrogen acts as an integrator between the old and new energy system, without the need to build an entirely new network.

However, admixing hydrogen into the natural gas pipelines brings some challenges. Hydrogen and methane are both gases at atmospheric conditions, but they differ in various aspects. They react differently to changes in pressure and temperature, have different burning speeds, and since hydrogen is a smaller molecule than methane, it is more prone to leak from pipes and valves. It is also very reactive with certain metals and alloys, causing corrosion and embrittlement.

1.2.1 Research questions

Quantitative research will be done to determine how much hydrogen can be admixed into the current grid and how it can be optimized for hydrogen transport. A schematic overview of the scope is shown in Figure 1. The research questions answered will be:

1. How much H2 can be mixed into the existing Dutch offshore grid and why?

2. How much H2 can be mixed into the existing Dutch onshore grid and why?

3. What changes need to be made to increase this percentage?

(7)

1.3 Hydrogen production

As stated before, hydrogen can be produced from the electrolysis of water, and there are several ways to do this. The most common is a polymer electrolyte membrane (PEM) electrolyser. In PEM electrolysis, water is split into its components O2-_{and H}+

by exploiting their charge difference. By combining two H+_{and two electrons}

hydrogen (H2) is produced. A cathode (charged -), and an anode (charged +), are

separated by a solid polymer electrolyte that is only permeable to H+_{, and thus}

separates the product gases and acts as an insulation between the two electrodes. The anode splits two H2O molecules into O2-, which combines to O2, and 4H+.

2𝐻2𝑂(𝑙) → 4𝐻+(𝑎𝑞) + 𝑂2(𝑔) + 4𝑒−

The H+_{then travels through the membrane towards the cathode where it will be}

combined with electrons to form hydrogen. 4𝐻+_{(𝑎𝑞) + 4𝑒}−_{→ 2𝐻}

2(𝑔)

In Figure 2 a simplified representation is given. To provide the cathode with electrons an external energy source is needed. Thus, the energy from the electrons from the external energy source is being embedded into the chemical bonds of the hydrogen.

Figure 2: Simplified representation of a PEM electrolyser.

Most hydrogen in the world market is however not produced through electrolysis. About 95% of worldwide commercial hydrogen is produced from fossil fuels, either natural gas or coal. The process to produce hydrogen from natural gas is called steam-methane reforming (Press, et al., 2009). This process has a typical efficiency of 70-85%(Onda, Kyakuno, Hattori, & Ito, 2004). Here, the efficiency is the usable energy stored in the chemical bonds of the hydrogen compared to the total energy input in the form of heat, work, or chemical potential (Wang, Lee, & Molburg). However, this process emits large amounts of CO2. Hydrogen produced through

electrolysis only makes up about 4% of the market share. This is due to the lower efficiency (65-78% for a PEM electrolyser) and to higher costs, about 4-6 times more expensive per kW than steam-methane reforming(Onda, Kyakuno, Hattori, & Ito, 2004).

Although hydrogen has a high energy density by weight, about 142MJ/kg, at atmospheric pressure its energy density by volume is low. At 20 °C and 1 atm the

(8)

specific volume of hydrogen is 11.9m3_{/kg. So, the energy density by volume is}

11.9MJ/m3_{. Compared to methane, 36.4MJ/m}3_{, it is quite low. This means that to}

efficiently transport energy while using hydrogen as a carrier it needs to be compressed. This is done using a compressor1_.

1.4 Existing infrastructure

After the hydrogen is compressed it needs to be transported. The goal is to do this with minimum investment in the infrastructure by using the current gas grid. One of the main gas transport lines offshore is currently operated by NOGAT. It consists of about 264 kilometres of pipes and carries about 6 billion Nm3_{of gas per year. It}

connects many offshore platforms to the Dutch mainland grid through a gas treatment plant in Den Helder. The Dutch onshore gas transport network is operated by Gasunie. It consists of 12000 km of transport pipelines with the necessary connection points, compressors and mixing stations. The transport network is divided into the main transport system (HTL) and the regional transport system (RTL). Each system has its own uses, and therefore design codes and maximum allowances. The HTL is connected to gas producers, import points, large domestic end users (such as power plants and industries), transmission operators in other countries, storage facilities and of course the RTL, which it feeds into. The RTL is connected to regional distribution system operators (DSOs), smaller power plants and industries. The distribution grid is divided into the high pressure distribution grid (HDD) and the low pressure distribution grid (LDD). It consists of about 130000 km of pipes, of which 60% are made from PVC (Weller, Hermkens, & van der Stok, 2016). It connects the RTL to consumers’ homes, and provides gas for heating and cooking.

1.5 Structure of report

To determine the maximum hydrogen admixture possible in the entire grid, each section of each of these transport systems must be considered independently. To set an exact limit for those is beyond the scope of this report, so it will focus on representative cases: the offshore trunk line (NOGAT) and the main transport system (HTL), since their allowances and compositions are best known and they are the least variable in their construction specifications. Of these two systems, first their mechanical properties and chemical compositions will be given in Module 2. Next, in Module 3, the fluid dynamics of pure natural gas will be compared to hydrogen admixed gas to show differences in behaviour. These differences will be derived and calculated to ensure a steady energy flow to the end user. Then, by considering the process of hydrogen embrittlement and leakage, the safety issues accompanied with admixing hydrogen will be described in Module 4.

1_{The compression power needed to compress hydrogen gas to a typical output pressure of 120 bar} is greater than the power needed to compress water to similar pressures. Another type of electrolysis, aptly called high pressure electrolysis, exploits this fact by first pressurizing the water, and then electrolyzing it to immediately form high pressure hydrogen. This type of electrolysis saves about 5% original power input to create the same high pressure hydrogen(Onda, Kyakuno, Hattori, & Ito, 2004).

(9)

1.6 Literature survey

Since hydrogen admixture and the hydrogen economy could be good alternative to our current system, extensive research on the subject has already been performed. Some of the important reports and papers on hydrogen admixture are listed together with their general conclusions.

The International Energy Agency (IEA) Greenhouse Gas R&D programme has published a report in 2003 titled Reduction of CO2 emissions by adding hydrogen to natural gas. They conclude that safety will not be compromised as a result of admixing up to 25% hydrogen into the Dutch natural gas grid. They also conclude that admixture up to 3% would lead to no additional cost in improving the grid. Above this threshold significant investment would be needed for checks and changes to end-user appliances. They advise an admixture up to 12% to minimize investment costs but maximize CO2 emission reduction (IEA Green House Gas R&D programme,

2003).

The National Renewable Energy Laboratory (NREL) of the U.S. Department of Energy published a report in 2013 called Blending hydrogen into natural gas pipeline networks: A review of key issues. They conclude that admixture up to 20% would give no added safety concern in the distribution grid, and that even higher admixtures may be acceptable in the transmission grid. They conclude that admixture of concentrations from 5% – 15% into the U.S. gas grid appear to be feasible with minor modifications to the existing pipeline systems and end-use appliances (Melaina, Antonia, & Penev, 2013).

Klaus Altfelt and Dave Pinchbeck published a paper in 2013 called Admissible hydrogen concentrations in natural gas systems in the journal Gas for Energy. They reported a significantly lower advisable maximum admixture of 10% in most parts of the gas grid, with some parts, such as gas turbines or gas engines only being allowed up to 5%. Their concern is mainly due to the difficulty in detecting hydrogen leaks with current gas chromatographs and other detecting methods such as DIAL (Altfelt & Pinchbeck, 2013).

(10)

2 Module 2: Infrastructure specifications

2.1 Offshore transport: NOGAT

The NOGAT pipelines connect many offshore platforms on the Dutch continental shelf to an onshore gas plant in Den Helder. The basic materials specification for the pipes is DIN 17172 STE 415.7TM. Meaning the steel has been produced under DIN 17172 standards, and has a steel grade STE 415.7TM. The chemical composition of this steel grade is listed in Table 1. It is almost completely made from iron, only the other elements that are present have been listed in mass percentage. Three different sizes are used: 36, 24, and 16 inch. The main trunk line has a 36 inch outer diameter and a wall thickness of 17.8 mm, giving it an inner diameter of 878.8 mm.

2.2 Onshore transport: Gasunie 2.2.1 HTL

The HTL network is the main transmission grid in the Netherlands. It connects supply points such as the Groningen gas field and the NOGAT pipeline to industries, power plants, and transmission operators in different countries. In Table 2 the allowances for pressure (p), temperature (T), flow speed (u), and roughness (є) are listed.

p [bar] T [°C] u [m/s] є [m] min 43.5 -20 - 12 max 80.9 50 20 -

Table 2: Minimum and maximum allowances for the HTL network (Gasunie, Ontwerp

uitgangspunten transportsysteem, 2014).

The types of metal used in the HTL have been determined by looking at a typical pipe used in the system. Pipe and flange specifications have been found for the transport pipes to and from a typical Dutch compressor station. These specifications are regarded to be representative for the majority of the HTL system. The pipe is a

STE 415.7TM [%] (T)STE 355 [%] L415 [%] Mn 1.63 1.55 1.30 C 0.007 0.18 0.28 P 0.01 0.03 0.03 S 0.005 0.03 0.03 Nb + Ti + V 0.134 0.12 0.15 Si 0.22 0.50 - Cr 0.11 0.30 - Cu 0.013 0.20 - Al 0.008 0.02 - Mo 0.228 - - Ni 0.011 - - N 0.008 - -

(11)

Figure 3: Overview Dutch onshore gas grid.

DN1200, meaning an exact outer diameter of 1219.2 mm, and has a wall thickness of 21.7 mm. This gives an inner diameter of 1175.8 mm. Another often used pipe in the HTL is a DN900, with and inner diameter of 876.3mm. The pipe is made from steel grade L415, which has a yield stress of 415MPa. Flange connections are made from (T)Ste 355 grade steel. Both steel compositions are given in Table 1.

2.2.2 RTL

The RTL is the regional transmission grid in the Netherlands. It connects to the HTL and supplies to smaller power plants and industries, and feeds into the distribution grid. While the RTL will not be discussed further, its allowances are given in Table 3 for reference and completeness. Missing values are not stated by Gasunie, but can most certainly not be more extreme than the HTL. Pipe diameters are generally smaller, with the DN600 being prevalent.

p [bar] T [°C] u [m/s] є [m] min 16 - - 18 max 41 - 20 -

Table 3: Minimum and maximum allowances for the RTL network (Gasunie, Ontwerp

uitgangspunten transportsysteem, 2014). 2.3 Distribution grid

The distribution grid connects to residential consumers and small businesses. Because the distribution grid is not nationally maintained but rather by several DSO’s, such as Liander, Enexis and Stedin, it is not possible to set exact limits on the allowances. However, the general pressures for the HDD and the LDD are known and listed in . p [bar] HDD LDD min 4 0.03 max 8 0.10

Table 4: Minimum and maximum pressure allowances for the HDD and LDD 2.4 Overview

In Figure 3 schematic overview is given of the Dutch onshore gas grid. Note that the pressure values are not the minimum and maximum, but rather typical values taken from DNV GL Energy (2015).

(12)

3 Module 3: Flow assurance

3.1 Fluid properties

Natural gas flowing through a pipe over long distances drops in pressure. This is mainly due to viscous forces that slow the gas down. To determine this pressure drop a range of different fluid properties need to be known. For example: density, calorific value, flow speed, viscosity and Reynolds number. For natural gases such as Groningen-gas (G-gas), these properties and thus the behaviour is known. For hydrogen admixed G-gas these can be derived and calculated. It is important to know the behaviour of a gas to ensure a steady energy flow.

3.1.1 Compressibility factor

Ideal gases adhere to the ideal gas law: 𝑝𝑉 = 𝑛𝑅𝑇

where 𝑝 is pressure [N/m2_],_{𝑉 is volume [m}3_],_{𝑛 is number of moles, 𝑅 is the gas}

constant = 8.3145 J/Kmol and 𝑇 is temperature [K]. However, natural gasses do not behave as ideal gasses. A generic way to determine the behaviour of a non-ideal gas is to add an extra factor 𝑧 to the ideal gas law:

𝑝𝑉 = 𝑧𝑛𝑅𝑇

𝑧 is called the compressibility factor and is unitless. The way to determine this compressibility factor depends on the Equation of State (EoS) being used to model the gas. The EoS that will be used in this report is the Soave modification of the Redlich-Kwong EoS (Soave, 1972). It is written as:

𝑝 = 𝑅𝑇 𝑣 − 𝑏− 𝛼(𝑇) 𝑣(𝑣 + 𝑏) Where: 𝑣 = 𝑧𝑅𝑇 𝑝 𝛼(𝑇, 𝜔) = (1 + (0.480 + 1.574𝜔 − 0.176𝜔2_{)(1 − 𝑇} 𝑟0.5))2

Where 𝜔 is the acentric factor for the species, and 𝑇𝑟= 𝑇 𝑇 𝑐

⁄ with 𝑇𝑐 being the

temperature of the species at the critical point. The compressibility factor 𝑧 can be determined by setting an arbitrary 𝐴 and 𝐵:

𝐴 = 𝑎𝑝 𝑅2_𝑇2

𝐵 =𝑎𝑝 𝑅𝑇

(13)

And rewriting to:

𝑧3_{− 𝑧}2_{+ 𝑧(𝐴 − 𝐵 − 𝐵}2_{) − 𝐴𝐵 = 0}

The compressibility factor is numerically calculated using PVTsim 2. Values are computed for hydrogen admixed (0-20%) G-gas, at a wide range of pressures (1-100 bar) and for winter (5 °C) and summer (20 °C) conditions. The composition of pure G-gas is found in Table 5. The corresponding compressibility factors for the range of pressures are found in Appendix 7.1. A plot is given in Figure 4.

component chemical formula molar percentage

[%] methane CH4 81.29 ethane C2H6 2.87 propane C3H8 0.38 n-butane n-C4H10 0.15 n-pentane n-C5H12 0.04 n-hexane n-C6H14 0.05 nitrogen N2 14.32 oxygen O2 0.01 carbon dioxide CO2 0.89

Table 5: Composition Groningen-gas(Gasunie, Basisgegevens aardgassen, 1980).

Figure 4: Compressibility factor values for hydrogen admixed G-gas at different pressures

0.8 0.85 0.9 0.95 1 1.05 0 20 40 60 80 100 120 comp re ss ib ili ty f act o r pressure [bar] 10 % 5 % 0 % 10 % 15 % 20 % 15 % 20 % 5 % 0 % zomer 20 °C winter 5 °C

(14)

Mixtures with more hydrogen admixed have a higher compressibility factor for all pressures, meaning they behave more like ideal gasses than pure G-gas. In the extreme case for 5 °C and 100 bar 𝑧 20

𝑧0 = 1.093, so almost a 10% difference in

behaviour. The subscript 20 means 20% hydrogen admixed G-gas, and the subscript 0 means pure G-gas. The exact consequences of this difference will be discussed next.

3.1.1.1 Density

To determine the density of a non-ideal gas the following equation must be used: 𝜌 = 𝑝𝑀𝑤

𝑧𝑅𝑇 Where 𝜌 is the density [kg/m3_{] and 𝑀}

𝑤 is the molar mass [kg/mol]. The density of the

mixture dependent on the hydrogen molar proportion x is then:

𝜌𝑥=

𝑝[(1 − 𝑥)𝑀𝑤,0+ 𝑥𝑀𝑤,𝐻₂]

𝑧𝑥𝑅𝑇

Dividing 𝜌𝑥 by 𝜌0 and rewriting gives:

𝜌𝑥= 𝑧0 𝑧𝑥 [(1 − 𝑥) + 𝑥𝑀𝑤,𝐻2 𝑀𝑤,0 ] 𝜌0

Comparing values will be done at typical transport pipeline values, 60 bar and 20 °C, and between pure G-gas and 10% hydrogen admixed. At these conditions, 𝜌0 = 50.6

kg/m3_{and 𝜌}₁₀_{= 45.0 kg/m}3_{: 𝜌}₁₀_{is about 10% less dense than 𝜌}₀_{. This stems from the}

fact that hydrogen is very light in comparison to all normal components of G-gas. 3.1.1.2 Calorific value

The combustion of hydrocarbons always respects the following chemical reaction: 𝐶𝑛𝐻2𝑛+2+ 𝑎𝑂2→ 𝑏𝐶𝑂2+ 𝑐𝐻2𝑂 + ∆𝐻

Thus, 𝑎 =1

2+ 3𝑛

2, 𝑏 = 𝑛 and 𝑐 = 𝑛 + 1. ∆𝐻 is the enthalpy change for the reaction

assuming an equal temperature before and after combustion, this is also called the higher heating value (HHV). Every molecule undergoing a combustion process has a specific HHV. Calculating the HHV of a mixture is done by taking the weighted average of the individual components of the mixture. The HHV of the individual components can be found in Table 6.

(15)

component chemical formula molar percentage [%] HHV [MJ/kmol] methane CH4 81.29 890.30 ethane C2H6 2.87 1559.88 propane C3H8 0.38 2220.03 n-butane n-C4H10 0.15 2877.09 n-pentane n-C5H12 0.04 3536.15 n-hexane n-C6H14 0.05 4194.92 nitrogen N2 14.32 - oxygen O2 0.01 - carbon dioxide CO2 0.89 - hydrogen H2 0 285.84 100 784.76

Table 6: Calculation of HHV for Groningen-gas (Gasunie, Basisgegevens aardgassen, 1980).

G-gas contains a fair amount of N2, which does not combust, though it can oxidize at

high temperatures, producing NOx in an endothermic reaction. Because of these high levels of N2, the HHV is lower than that of pure methane. To determine the HHV of a

hydrogen admixed gas the weighted average is taken. Since hydrogen is so light, the HHV when measured in moles is lower than that of methane. In contrast, the HHV measured per kilogram gives 𝐻𝐻𝑉𝐻2 = 141.8 MJ/kg and 𝐻𝐻𝑉𝐶𝐻4 = 55.5 MJ/kg. Thus,

hydrogen has a higher energy density per mass, but not per volume. In Table 7 the HHV of several mixture ratios is given.

H2 [%] HHV [MJ/kmol] 0 784.76 5 759.81 10 734.87 15 709.92 20 684.98

Table 7: HHV of hydrogen admixed Groningen-gas.

3.1.1.3 Volume flow rate and flow speed

End users of the gas are interested in getting the same amount of energy per time interval, but the amount of energy stored per mole of gas decreases with increasing hydrogen admixture. A mole of ideal gas, regardless of its species, occupies a set volume depending on its temperature and pressure. For a non-ideal gas the compressibility factor is once again needed. The volume a non-ideal gas occupies per mole can be written as:

𝑉𝑚=

𝑧𝑅𝑇 𝑝

Where 𝑉𝑚 is the molar volume [m3/mol]. The energy output per second can then be

written as:

𝑃 = 𝑄 𝑉𝑚

(16)

Where 𝑄 is the volume flow rate [m3_{/s] and 𝑃 is the power [kW]. In order to satisfy the}

gas users, the energy output with or without hydrogen admixture must be equal. Hence, 𝑃0= 𝑃𝑥, and thus:

𝑃0= 𝑄0 𝑉0 ∙ 𝐻𝐻𝑉0= 𝑃𝑥= 𝑄𝑥 𝑉𝑥 ∙ 𝐻𝐻𝑉𝑥 Solving for 𝑄𝑥: 𝑄𝑥= 𝑉𝑥 𝑉0 ∙𝐻𝐻𝑉0 𝐻𝐻𝑉𝑥 ∙ 𝑄0 (1) In this equation, 𝑉𝑥

𝑉₀> 1, since 𝑧𝑥 > 𝑧0 (Figure 4), and 𝐻𝐻𝑉0

𝐻𝐻𝑉_𝑥> 1 (Table 7). This means

that the flow rate and thus the flow speed of hydrogen admixed G-gas must be significantly higher than normal to get the same energy output. Since the allowances in construction specifications are given in flow speed, this will be fixed at 10 m/s for pure G-gas. The volume flow rate is then calculated by:

𝑄 = 𝑢𝐴 Thus, equation (1) becomes:

𝑢𝑥= 𝑉𝑥 𝑉0 ∙𝐻𝐻𝑉0 𝐻𝐻𝑉𝑥 ∙ 𝑢0

Where 𝑢0 will be set at 10 m/s. Values are shown in Table 8 for the example of a

DN900 pipe, which has an outer diameter of 914.4mm, and with schedule 40 has a wall thickness of 19.05 mm, giving an inner diameter of D = 876.3 mm(ASTM, 2007).

H2 [%] u [m/s] Q [m3_/s] 0 10.00 6.03 5 10.49 6.33 winter 5 °C 10 11.00 6.63 15 11.54 6.96 20 12.10 7.30 0 10.00 6.03 5 10.46 6.31 summer 20 °C 10 10.94 6.60 15 11.44 6.90 20 11.98 7.22

Table 8: Volume flow rates and flow speeds needed to match the energy supply of pure G-gas.

3.1.1.4 Dynamic pressure

This increased flow speed also causes an increase in dynamic pressure. The dynamic pressure gives an idea of how much kinetic energy the gas flow contains. The dynamic pressure is an important element of the pressure losses in a pipe system, and therefore the power needed to boost the flow from supply point to demand site. It is also related to the potential of the flow to excite vibrations in pipe systems. Calculating the dynamic pressure, symbol 𝑞, is done by:

(17)

𝑞 =1

2𝜌𝑢 2

As seen before, with hydrogen admixture 𝜌 gets smaller, but 𝑢 increases. The result is that the dynamic pressure increases with increased hydrogen admixture, but not by too much. For a flow speed set at 10 m/s and a pressure of 60 bar, the dynamic pressure is calculated and shown in Table 9.

H2 [%] q [kg/ms2_] qx/q0 0 2736 1 5 2824 1.03 winter 5 °C 10 2910 1.06 15 2994 1.09 20 3076 1.12 0 2528 1 5 2601 1.03 summer 20 °C 10 2673 1.06 15 2743 1.09 20 2813 1.11

Table 9: Dynamic pressure for hydrogen admixed G-gas at 60 bar and a flow speed of 10 m/s.

3.1.2 Wobbe index

The Wobbe index is a measurement of the amount of energy a nozzle transports to a burner and therefore it refers to the ability to use the gas with the same domestic or industrial appliances. It is given by:

𝐼𝑊= 𝐻𝐻𝑉 √𝐺𝑠 · (𝜌𝑁 𝑚𝑥 )

Where 𝐺𝑠 is the specific gravity 𝜌𝑥

𝜌_𝑎𝑖𝑟, 𝜌𝑁 is the density at normal conditions (0 °C and

1.013 bar), and 𝑚𝑥 is the molecular weight. This multiplication of 𝜌𝑁

𝑚_𝑥 is done to

transform the units of the Wobbe index from [MJ/kmol] to [MJ/Nm3_{], which is the}

industry standard. The Wobbe index is an intrinsic quantity and does not depend on temperature or pressure, it is given in Table 10.

H2 [%] IW [MJ/Nm3_] 0 43.40 5 42.98 10 42.57 15 42.18 20 41.78

(18)

3.1.3 Dynamic viscosity

The dynamic viscosity, symbol µ, unit Poise (P), or in SI [kg/(ms)], further just called viscosity, determines the resistance a fluid has against shearing. Shear appears whenever there is a different of speed between adjacent streamlines within the fluid flow, induced by the geometry or otherwise. For example, wind blowing over a field of grass. The air in between the grass will move slower than the air moving over the grass, causing a shear stress between them. This also applies to gasses moving through a pipe. Gas flow near the wall of the pipe will be blocked by small imperfections and bumps on the inside. Thus, streamlines will flow faster in the middle of the pipe. This difference in flow speeds creates a velocity profile in the pipe. One of the simpler models for fluid flow in a pipe is called Poiseuille flow. It is applicable for an incompressible laminar and Newtonian fluid that is driven by a constant pressure gradient. Poiseuille flow gives a parabolic velocity distribution. A visual representation is given in Figure 5. To get an idea of viscosity you could see it as the ‘stickiness’ of a fluid, high viscosity means high stickiness and thus a lot of shear stress between the flows. This means that in Figure 5 the velocity profile would become even pointier for high viscosity, and would flatten for low viscosity.

On a molecular level, viscosity in gases can be explained by looking at the momentum transport between layers. When a slower moving particle in a layer close to the wall of the pipe moves inwards through random motion, it is moving its momentum in the 𝑥 direction to a faster flowing layer. Here it will collide with other, faster particles, slowing those down. These particles will move up or down again and adjust the velocities in those layers. The result is a velocity profile where the fastest moving particles are the furthest away from the boundaries.

Since momentum transfer, and thus viscosity, is dependent on the random motion and mean free path of particles, it is dependent on temperature. In liquids, higher temperature generally means lower viscosity. For example, honey and syrup get ‘runnier’ when heated up. The viscosity in gasses, however, rises with temperature increase.

(19)

To understand this, it is once again necessary to look at this from a more precise physics perspective, specifically the kinetic theory of gases. The shear stress, symbol τ, that a particle in a flow layer at 𝑦 = 1 experiences depends on the speed in the flow direction of the particles that enter that layer. In the example of Figure 6 the flow direction is 𝑥. The speed of a particle in the flow direction at a particular layer is 𝑢(𝑦). In this example 𝑢(𝑦) is bigger for bigger 𝑦, on average. So, a particle travelling from layer 𝑦 = 3 to layer 𝑦 = 1 will generally transfer more momentum than a particle originating from 𝑦 = 2 traveling to 𝑦 = 1.

In formula form, and taking an arbitrary flow layer surface 𝐴: 𝜏 =𝑚̇〈𝑢(𝑦)〉

𝐴

Where 𝜏 is the shear stress and 𝑚̇ is mass transfer over time. Since the amount of speed, and thus momentum, being transferred depends on the original layer of the particle, it depends on the amount of distance a particle can travel in the fluid. This is also called the mean free path 𝜆. When the mean free path in a fluid is bigger, it means that particles from layers ‘further away’ can influence each other, resulting in a higher shear stress and thus viscosity.

𝜇 ∝ 𝜆

Next, the temperature and pressure dependencies can be determined. Increasing the temperature of a gas means increases the amount of energy in the translational modes of freedom of the particles, and thus the average speed of the particle. In the

Figure 7: Illustration of temperature dependency on mean free path. Figure 6: Schematic representation of a velocity profile.

(20)

example of Figure 7 the black particle wants to pass the white particles. Because all the particles have a certain temperature they will be vibrating. This means that the black particle has a smaller chance of collision when it moves through the barrier quickly.

So, with a higher temperature, particles can move a bigger distance, on average, meaning a bigger mean free path, and consequently bigger viscosity.

𝜇 ∝ 𝜆 ∝ 𝑇

With increasing pressure in a fluid there will be more particles per unit volume. This results in more particles transferring their 𝑢𝑥(𝑦) over the same area 𝐴. This would

suggest that higher pressure gives higher viscosity. However, as determined before, viscosity is also dependent on the mean free path. It is simple to imagine that a higher pressure, meaning more particles per volume, decreases the mean free path.

𝜇 ∝ 𝑝 & 𝜇 ∝ 𝜆−1_{∝ 𝑝}−1_.

The result is that viscosity is independent on pressure. Do note that this is only applicable in the ideal case. The viscosity in real gasses is very slightly dependent on pressure, because of the change in compressibility factor, which is also observed in the values from PVTsim 2 and from the NIST chemistry WebBook (~+3.5 · 10-8

(kg/ms)/bar). The viscosity of hydrogen admixed G-gas is numerically calculated using PVTsim 2, in the same way the compressibility factors were determined. From this viscosity, which can be found in Appendix 7.2 and plotted in Figure 8, several other important fluid properties will be calculated.

Figure 8: Dynamic viscosity of hydrogen admixed G-gas for a range of pressures. Note the inversion of the pure G-gas and mixtures.

1.00E-05 1.10E-05 1.20E-05 1.30E-05 1.40E-05 1.50E-05 1.60E-05 -10 0 10 20 30 40 50 60 70 80 90 100 110 vis cos ity [ kg/(m s] pressure [bar] 20 % 20 % 0 % 0 % 0 % 0 % 20 % 20 %

(21)

3.1.3.1 Reynolds number

The Reynolds number gives the ratio between the inertial and viscous forces of the fluid. It is defined as:

𝑅𝑒 =𝜌𝑢𝐷 𝜇

Where 𝐷 is the characteristic length. When looking at the flow through pipes, this characteristic length is usually the inner diameter of the pipe. For a DN900 pipe, D = 0.8763 m. The values of the Reynolds number can be found in Table 11.

H2 [%] µ [kg/(ms) · 10-5_] u [m/s] 𝜌 [kg/m3_] Re [· 107_] 0 1.30 10.00 54.72 3.69 5 1.30 10.49 51.33 3.63 winter 5 °C 10 1.30 11.00 48.10 3.57 15 1.29 11.54 44.99 3.53 20 1.29 12.10 41.00 3.45 0 1.34 10.00 50.56 3.31 5 1.34 10.46 47.58 3.25 summer 20 °C 10 1.34 10.94 44.70 3.20 15 1.34 11.44 41.91 3.14 20 1.33 11.98 39.21 3.09

Table 11: Reynolds numbers for hydrogen admixed G-gas for a range of pressures.

3.1.3.2 Turbulent and laminar flows

The value of the Reynolds number determines whether a fluid has a turbulent or laminar behaviour. In a laminar flow all particles move orderly in lines parallel to the pipe walls. Turbulent flow is, as the name suggests, more chaotic, with particles moving completely irregularly. Generally, when Re < 2000 the flow is laminar, and when Re > 4000 the flow is turbulent. Between 2000 and 4000 is called the transition phase.

The Poiseuille velocity profile shown before is one of the simpler velocity profiles, but it only applicable to an incompressible laminar Newtonian fluid. Natural gas through a pipe at typical transport conditions behaves Newtonian and can be regarded as incompressible, because the Mach number of gas transport is very limited (typically 5 - 15 m/s, or M = 0.015 - 0.044, even lower when hydrogen is admixed, since the speed of sound is higher), but it is most certainly not laminar. The Reynolds numbers found are of the order 107_{, which is well within the turbulent regime. This means that}

the Poiseuille model is not applicable.

A turbulent flow in a pipe can be divided into three regions, characterized by their distance from the pipe wall(Munson, Young, & Okiishi, 1994). The layer closest to the wall is called the viscous sublayer. Here, the viscous shear stress is dominating the turbulent stress, since the flow speed sharply increases with distance from the wall. The velocity profile in this region can be approximated by the so-called law of the wall: 𝑢̅ 𝑢∗= 𝜇𝑦𝑢∗ 𝜌 (2)

(22)

Where 𝑦 is the distance measured from the wall, 𝑢̅ is the time-average flow speed and 𝑢∗_{= (}𝜏𝑤

𝜌

⁄ )2, the friction velocity. In the overlap region the following equation has been proposed to fit experimental data:

𝑢̅

𝑢∗= 2.5 ln (

𝜇𝑦𝑢∗

𝜌 ) + 5

(3)

In the fully turbulent region, where 𝜇𝑦𝑢∗

𝜌 ≳ 60, the following equation is often used:

𝑢̅ 𝑉𝑐 = (1 −𝑟 𝑅) 1 𝑛 ⁄

Where 𝑉𝑐 is the centerline velocity, and R is the inner radius of the pipe. 𝑛 is a function

of the Reynolds number, and is set at seven for most practical uses. The equation is then called the venerable 1/7th_{power law. Note that in the example of Figure 9 the}

horizontal scale is logarithmic and may make the viscous layer seem bigger than it is. In reality it is very thin, so the 1/7th_{power law is generally used in the fluid dynamics}

industry to approximate the entire velocity profile.

(23)

3.1.4 Pressure loss

Now that most gas properties are known and described, it is possible to calculate the pressure loss through a pipe due to viscous effects. This is done using the Darcy-Weisbach equation:

∆𝑝 𝐿 = 𝑓𝐷

𝜌𝑢2

2𝐷

Where 𝑓𝐷 is the Darcy friction factor, which can be approximated in many ways for

different cases. The approximation posed by Chen in 1979 will be used since it is valid for very high Reynolds numbers (> 107_{) and it is deemed one of the most}

accurate(Brkic, 2011): 1 √𝑓𝐷 = −2 log [ 𝜖 𝐷 ⁄ 3.7065− 5.0452 𝑅𝑒 log ( 1 2.8257( 𝜖 𝐷) 1.1098 + 5.8506 𝑅𝑒0.8981)]

Now everything needed to calculate pressure loss is determined. But, since the pressure loss is dependent on the density, which is dependent on the pressure, it needs to be evaluated iteratively. Initial pressure will be set at 60 bar and pure G-gas flow speed at 10 m/s. Further D = 876,3 mm and є = 15 m, so 𝜖⁄ = 1.71 · 10_𝐷 -4_{. The}

pressure losses can be found in Table 12. H2 [%] u [m3_/s] Re [· 107_] fD Δp/L [bar/km] 0 10.00 3.69 0.0133364 0.416 5 10.49 3.63 0.0133370 0.430 winter 5 °C 10 11.00 3.57 0.0133376 0.443 15 11.54 3.53 0.0133381 0.456 20 12.10 3.45 0.0133388 0.468 0 10.00 3.31 0.0133406 0.385 5 10.46 3.25 0.0133412 0.396 summer 20 °C 10 10.94 3.20 0.0133419 0.407 15 11.44 3.14 0.0133427 0.418 20 11.98 3.09 0.0133433 0.428

Table 12: Darcy friction factor and subsequent pressure loss due to viscous forces.

As can be seen, the pressure loss increases with more hydrogen admixture. This is mainly due to the increased flow speed that is needed for delivering the same amount of energy to the user. Although, it should be noted that the difference is not very big. After a certain amount of kilometres transported the new pressure should be used to achieve accurate results when transporting over a large distance.

3.1.5 Joule-Thomson effects

As shown, gas grid pressure is not the same everywhere. Generally, pipes nearer to the gas source have a higher pressure than pipes close to the users. In the case of hydrogen, typical output pressure from high pressure electrolysis is 100 - 120 bar, while pressure inside the HTL is 66 - 80 bar, with the RTL and distribution grid being even lower, as shown in Figure 3. Aside from the slow pressure loss from transporting gas over long distances which was calculated above, pressure can drop suddenly due to choked flow. When a real gas is subject to sudden pressure change while kept

(24)

insulated it undergoes a temperature change. This is called the throttling process or Joule-Thomson process and is an isenthalpic process, meaning that no enthalpy is gained. Whether a gas will heat up or cool down when expanded depends on the sign of Joule-Thomson coefficient (JTC) of the gas at the initial temperature and pressure. If the JTC is less than zero, the gas will heat up, and vice versa. The units of the JTC are K/bar, so the magnitude of the JTC gives the exact temperature change due to a pressure change. The JTC for hydrogen admixed G-gas have been determined using PVTsim 2. Values can be found in Appendix 7.3, and a plot is given in Figure 10. As can be seen, the JTC decreases for increasing hydrogen admixture. Meaning that the mixture will cool less when undergoing the same pressure drop, or heat less with increased pressure.

By looking at the JTC for pure species it is possible to determine where this change in behaviour is coming from. In Figure 11 the JTC values are plotted for pure species at two isobars (60 and 120 bar) over a large temperature range (38 – 800K). The fluid data are taken from the NIST Chemistry WebBook. As can be seen in Figure 11 the JTC for hydrogen does not vary much over a wide range of temperatures and pressures, meaning that hydrogen is very unresponsive to isenthalpic change. At almost every temperature, its JTC is lower than that of methane, which is the main component in G-gas. Thus, admixing hydrogen to G-gas lowers the JTC.

0 0.1 0.2 0.3 0.4 0.5 0.6 0 20 40 60 80 100 120 JT C [ K /ba r] pressure [bar] 0 % 20 % 20 % 0 %

(25)

Next, it is interesting to take a quick look at the kinetic theory of gases to see where this unresponsiveness of hydrogen comes from. As stated earlier, real gasses do not adhere exactly to the ideal gas law, but to some EoS. A rather famous EoS is the Van der Waals equation, which is less accurate than the RSK EoS in the case of gas transport, but useful for this explanation. It is written as:

[𝑃 + 𝑎 (𝑛 𝑉)

2

] (𝑉

𝑛− 𝑏) = 𝑅𝑇

Where 𝑎 and 𝑏 are the Van der Waals coefficients. The Van der Waals equation approaches the ideal gas law as the values of these constants approach zero. Constant 𝑎 is a correction for the average attraction between particles. Constant 𝑏 is the volume excluded by a mole of particles. The values of 𝑎 and 𝑏 can be found in Table 13 for some often-used species. One thing that stands out is that the value of constant 𝑎 is 5-20 times smaller for hydrogen than others.

Figure 11: Joule-Thomson coefficient of methane, water, hydrogen and nitrogen at isobars for a wide range of temperatures.

(26)

species 𝑎 [m6_Pa/mol2_] 𝑏 [m3_{/mol · 10}-6_] Cp at 60 bar and 300K [J/Kmol] methane 0.228 42.78 43.041 water 0.554 30.49 75.026 hydrogen 0.024 26.61 29.159 nitrogen 0.137 38.70 32.125

Table 13: Values of the Van der Waals coefficients and specific heat capacity (Fishbane,

Gasiorowicz, & Thornton, 2005).

By considering the famous porous-plug experiment as performed by Joule-Thomson in 1852, where a gas is pushed by a piston through a porous plug and exerts force on a piston on the other side, the temperature change, or JTC, is found to be(Loeb, 2004): ∆𝑇 ∆𝑝= 1 𝐽𝐶𝑝 (2𝑎 𝑅𝑇− 𝑏)

Where 𝐽 is the mechanical equivalent of heat, 𝐶𝑝 is the specific heat at constant

pressure, and 𝑎 and 𝑏 are the Van der Waals constants. From the above equation, it is possible to determine the temperature where the JTC changes sign. When 2𝑎

𝑅𝑇< 𝑏

the temperature change will be negative, and when 2𝑎

𝑅𝑇> 𝑏 it will be positive. The

inversion temperature 𝑇𝑖 can be defined as 𝑇𝑖= 2𝑎

𝑅𝑏. For hydrogen, 𝑎 = 0.024

m6_Pa/mol2 _{and 𝑏 = 26.61 · 10}-6 _m3_{/mol. This gives us an inversion temperature 𝑇}

𝑖 =

198 K, which agrees well with Figure 11. This inversion temperature is crucial in understanding the unresponsiveness of hydrogen to pressure change over such a large range of temperatures. As can be seen in Figure 11 all substances fall of exponentially after they have reached their maximum and thus flatten for higher temperatures. Hydrogen has the lowest 𝑇𝑖 and thus flattens out at the lowest

temperatures. If the graph were to be extended to even higher temperatures all the lines would flatten out and thus have their own region of unresponsiveness to pressure change. It is just the case that hydrogen is already unresponsive at temperatures being handled in pipeage and compression systems.

To draw conclusions, it is a good idea to look at some typical throttling processes. First, when pure hydrogen flows from the high-pressure electrolyser to the national transmission grid, the upstream pressure and the downstream pressure will typically be matched at the injection point. The fluid will experience a rapid depressurization. Taking initial values of 120 bar and 70 °C, the JTC for hydrogen is -0.044 K/bar (NIST). The pressure then shifts isenthalpically to 70 bar. Calculating the new temperature is done by:

𝑇𝑓𝑖𝑛𝑎𝑙 = 𝐽𝑇𝐶(𝑝𝑓𝑖𝑛𝑎𝑙− 𝑝𝑖𝑛𝑖𝑡𝑖𝑎𝑙) + 𝑇𝑖𝑛𝑖𝑡𝑖𝑎𝑙

Thus, the hydrogen gas will heat up to 72.2 °C. This amount of change in temperature is still within the uncertainty of the temperature determination and thus not significant. Secondly, smaller chokes happen during transport. Here, the hydrogen is already mixed with the G-gas so the mixture JTC values need to be used. A typical choke would be from 60 bar to 56 bar. For this type of choke, the temperature changes can be found in Table 14.

(27)

H2 [%] Tfinal [°C] ΔT [°C] 0 3.23 1.77 5 3.37 1.63 winter 5 °C 10 3.50 1.50 15 3.63 1.37 20 3.75 1.25 0 18.43 1.57 5 18.55 1.45 summer 20 °C 10 18.66 1.34 15 18.78 1.22 20 18.89 1.11

Table 14: Temperature change due to an isenthalpic pressure change from 60 to 56 bar.

As can be seen, the temperature change decreases with increasing hydrogen admixture, which could make some parts of the transport system easier to maintain. 3.1.6 Compression power

Pressure lost due to viscous forces in transport and due to chokes must be compensated. To do this compressor stations are needed. Since the pressure drops depend on the amount of hydrogen admixture, so does the compressor power needed to pump it back up to a desired pressure. Compressor power is calculated using:

𝑃 = 𝑚̇𝑐𝑝(𝑇𝑠− 𝑇𝑑)

Where 𝑇𝑠 is the suction temperature, 𝑇𝑑 is the discharge temperature, and 𝑐𝑝 is the

heat capacity [J/kgK] at suction conditions. To determine 𝑇𝑑 the temperature ratio is

needed, which can be expressed in terms of the pressure ratio and the polytropic efficiency as: 𝑝𝑑 𝑝𝑠 = (𝑇𝑑 𝑇𝑠 ) 𝜅𝑐_𝑒 𝜅−1 Where 𝜅 =𝑐𝑝

𝑐𝑣, the heat capacity ratio for a non-ideal gas, and 𝑐𝑒 is the polytropic

efficiency. Isolating 𝑇𝑑 and plugging into the power equation gives:

𝑃 = 𝑚̇𝑐𝑝𝑇𝑠[( 𝑝𝑑 𝑝𝑠 ) 𝜅−1 𝜅𝑐𝑒 − 1]

A schematic representation of the pressure between two compressors is given in Figure 12. By setting the distance between two compressor stations to 50 km, a pressure loss over that distance can be calculated. This pressure loss is bigger for more hydrogen admixture. By setting the suction pressure for each compressor to 60 bar, the discharge pressure of compressor 1 can be calculated to provide compressor 2 with 60 bar again. In Table 15 the discharge pressure needed is given, as well as the compression power to generate this discharge pressure. The values for both heat capacities have been numerically determined using PVTsim 2 and can be found in

(28)

Appendix 7.4. All others values have been calculated using information from previous paragraphs. The polytropic efficiency is taken to be 0.85.

H2 [%] pd [bar] cp [J/kgK] cv [J/kgK] ṁ [kg/s] P [MW] Px/P0 0 88.7 2325 1460 330 39.8 1 5 89.8 2363 1507 325 40.0 1.004 5 °C 10 90.9 2411 1559 319 40.3 1.012 15 91.9 2470 1616 313 40.7 1.022 20 92.9 2541 1681 307 41.2 1.034 0 86.3 2285 1496 305 32.5 1 5 87.2 2332 1543 300 32.9 1.011 20 °C 10 88.0 2387 1595 295 33.3 1.024 15 88.9 2453 1653 289 33.8 1.040 20 89.7 2529 1718 283 34.4 1.058

Table 15: Compression power needed to supply a compression station 50 km further with a suction pressure of 60 bar.

(29)

3.2 Summary

In Table 16, a summary is given of the previously calculated values. All values are computed for a DN900 pipe, with a nominal pressure of 60 bar and an energy flow equivalent to standard G-gas flowing with a speed set at 10 m/s.

H2 [%] HHV1 [MJ/kmol] 𝜌 [kg/m3_] u [m/s] q [kg/ms2_] IW1 [MJ/Nm3_] Re [· 107_] Δp/L [bar/km] Tfinal2 [°C] P3 [MW] 0 784.76 54.72 10 2736 43.40 3.69 0.416 3.23 39.8 5 759.81 51.33 10.49 2824 42.98 3.63 0.430 3.37 40.0 5 °C 10 734.87 48.10 11.00 2910 42.57 3.57 0.443 3.50 40.3 15 709.92 44.99 11.54 2994 42.18 3.53 0.456 3.63 40.7 20 684.98 41.00 12.10 3076 41.78 3.45 0.468 3.75 41.2 0 - 50.56 10 2528 - 3.31 0.385 18.43 32.5 5 - 47.58 10.46 2601 - 3.25 0.396 18.55 32.9 20 °C 10 - 44.70 10.94 2673 - 3.20 0.407 18.66 33.3 15 - 41.91 11.44 2743 - 3.14 0.418 18.78 33.8 20 - 39.21 11.98 2813 - 3.09 0.428 18.89 34.4

Table 16: Summary of fluid properties.

1 _{Independent of temperature.}

2_{For an isenthalpic expansion from 60 to 56 bar.}

3_{Compression power needed to supply a compression station 50 km away with a suction pressure}

of 60 bar.

3.3 Discussion of figures

It can be said that hydrogen admixture into natural Groningen-gas has sizeable implications for its behaviour. This was first observed in the change of the compressibility factor, where 𝑧 20

𝑧0 = 1.093, meaning a 10% difference. The

compressibility factor influences many properties, but most importantly the density. The density is inversely proportional to the compressibility factor, so it too will change by 10%, on top of the density change when mixing species. It is also seen that admixing hydrogen has a large influence on the calorific value of the gas with 𝐻𝐻𝑉 20

𝐻𝐻𝑉0 =

0.873, meaning that the same volume of 20% hydrogen admixed G-gas produces about 13% less energy than pure G-gas when burned. This decrease in energy density means that the flow speed must go up in order to supply the user with a steady amount of energy. Comparing the two opposites again gives 𝑢 20

𝑢0 = 1.21,

meaning a 21% increase in flow speed is needed. This could have serious implications since the entire system needs to run at a higher speed. However, the maximum allowance for flow speed in the HTL is 20 m/s, so only when the system was originally running at 16.5 m/s would it pose a potential hazard. This original flow speed is a lot higher than normally used in gas transport (5 - 10 m/s). The increased flow speed gives rise to an increase in dynamic pressure, or kinetic energy, the gas has: 𝑞 20

𝑞0 = 1.109. This dictates certain construction requirements since the pipes,

valves and elbows may have to be sturdier. Next, the difference in Wobbe-index was shown 𝑊𝐼,20

(30)

run at lower efficiencies. Also, the pressure loss was determined, ∆𝑝 20

∆𝑝0 = 1.124. So,

at a flow speed of 10 m/s, initial pressure of 60 bar and over 10 kilometres, the pressure would drop by 4.68 bar instead of 4.16 bar. The one aspect where hydrogen admixture may have beneficial effects is the Joule-Thomson coefficient. Comparing opposites for a typical choke (60 to 56 bar) gives 𝐽𝑇𝐶 20

𝐽𝑇𝐶0 = 0.707 meaning that with an

increasing hydrogen fraction the mixture would cool down less when undergoing sudden pressure drops. Lastly, the compression power difference has been calculated by setting the suction pressure to 60 bar and placing two compressor stations 50 km apart. Because of the increased pressure loss when admixing hydrogen an increased discharge pressure is needed, which requires more energy. It has been shown that this difference is smaller than expected, 𝑃 20

𝑃0 = 1.034, this is

mainly due to the mass flow rate decrease for hydrogen admixture. Although this increase seems small, the total amount of extra energy needed can become large, since every single compressor needs 3.4% more energy.

3.4 Conclusion

In conclusion, admixing hydrogen into natural gas comes with some engineering consequences if end-users are to be supplied with the same amount of energy. A higher flow speed is needed which causes larger pressure losses. This necessitates increased compression power. Hydrogen admixture also lowers the Wobbe index of the natural gas, meaning that appliances that have been optimized for a certain Wobbe range might operate at lower efficiencies. The only mechanical benefit from admixing hydrogen would be the decrease in Joule-Thomson coefficient, which makes the gas less responsive to sudden isenthalpic change. Nonetheless, it has been shown that most differences are relatively small and in many cases, can still be well within operating envelopes of the hardware involved. Only when a system is already running near its maximum capacity would it be necessary to study the effects of admixing hydrogen in more detail.

(31)

4 Module 4: Integrity management and safety

4.1 Embrittlement

Hydrogen deterioration is probably the biggest concern for safety when admixing hydrogen into the natural gas grid. It has been extensively shown that metals, steels and other alloys suffer severe consequences from prolonged exposure to hydrogen concentrations (Louthan, 2008). The ASM Materials Handbook (1998) lists five types of hydrogen deterioration:

1. hydrogen-induced blistering

2. cracking from precipitation of internal hydrogen 3. hydrogen attack

4. cracking from hydride formation 5. hydrogen embrittlement

These all cause a significant deterioration in the mechanical properties of metallic components. Except for hydrogen embrittlement, the listed types of hydrogen deterioration do not give any extra safety concern in the case of gaseous hydrogen transport through a pipe. To understand why, first the basics of hydrogen induced damage will be explained and then in more detail all five types.

All hydrogen induced damage starts with hydrogen absorption. Hydrogen dissolves into metals as an atom rather than as a hydrogen molecule. Thus, hydrogen must first dissociate before it can be absorbed by the metal layer. A schematic representation of this process is given in Figure 13.

Figure 13: Schematic representation of hydrogen diffusion into a metal.

Thus, the rate of hydrogen absorption into the metal depends on the amount of dissociation at the surface, which depends on conditions at the surface. For example, a thin oxide layer on the metal surface will greatly reduce the amount of dissociation and thus absorption. Lowering the hydrogen density will also lower absorption. There are several types of hydrogen exposure, examples are: gaseous exposure, acid cleaning, hydrogen sulphide exposure and chemical charging. In this range of exposures, gaseous is by far the least violent type and will lead to less absorbed

(32)

hydrogen than others. Absorbed hydrogen can be present in either interstitial sites within the metal lattice, or in extraordinary sites, such as grain boundaries, vacancies and other lattice imperfections. The hydrogen is generally too big to fit comfortably in interstitial sites. In extraordinary sites the interstitial sites have already been dilated, causing the absorbed hydrogen to prefer these sites. Together with the fact that hydrogen is mobile within the lattice, this will cause the hydrogen to shift locations. When hydrogen concentration is higher in a region it will bring a higher chance of deterioration, simply because of the pressure the hydrogen exerts on the lattice. With this basic explanation of hydrogen absorption and migration it is now possible to explain all five types of hydrogen deterioration and which are important when dealing with gas flow through a pipe.

4.1.1 Hydrogen-induced blistering

Hydrogen-induced blistering occurs when a large concentration of absorbed hydrogen atoms accumulates in an extraordinary site and then recombines to hydrogen molecules. This produces a pressure that pushes on the lattice and dilates the site even more. This process, however, only occurs for very high absorption rates. These are found when metals have been exposed to high concentrations of hydrogen by for example chemical charging or acid pickling. Gaseous hydrogen does not produce concentrations needed for this amount of absorption, not even at pressures of up to 700 bar(Louthan, 2008).

4.1.2 Cracking from precipitation of internal hydrogen

Cracking from precipitation of internal hydrogen occurs when hydrogen has been introduced to the metal during forging, welding or casting. During these events, high temperatures are used and hydrogen is produced from moisture in the air. When the metal cools down again the hydrogen precipitates and causes cracking or so-called fisheyes. One way to prevent this is to have all welding aperture and the welding environment as dry as possible. This way of introducing hydrogen into a metal is however not caused by usage of the pipe, but by production. Thus, hydrogen admixture into the gas grid does not increase the chance of cracking from precipitated hydrogen.

4.1.3 Hydrogen attack

Hydrogen attack occurs when hydrogen interacts with certain elements in the alloy to form other gaseous products. The most occurring examples are reactions with copper oxide,

𝐶𝑢2𝑂 + 2𝐻 → 2𝐶𝑢 + 𝐻2𝑂

where steam is formed, and with certain carbons, 2𝐻2+ 𝐶 → 𝐶𝐻4

where methane is formed. The gaseous end product once again pushes on the lattice from the inside, dilating it and causing cracks. However, both reactions only happen around higher temperatures than are generally being used in gas transport (100 °C and 200 °C respectively).

(33)

4.1.4 Cracking from hydride formation

Cracking from hydride formation occurs when hydrogen reacts with certain rare earth, alkaline earth, or transition metals to form hydrides. This most commonly occurs with titanium, zirconium and tantalum. The reaction for forming a titanium hydride is as follows:

𝑇𝑖 + 𝐻2→ 𝑇𝑖𝐻2

Hydrides are typically low density, brittle compounds that will consequently lower the ductility and stress resistance of a metal. Luckily most species that can form hydrides are not generally used in the construction of transport pipes, with the exception being traces of titanium. The above reaction, however, prefers temperatures between 300 °C and 500 °C. It can be concluded that cracking from hydride formation does not add safety issues to hydrogen admixture into the gas grid.

4.1.5 Hydrogen embrittlement

Hydrogen embrittlement is the only type of hydrogen induced damage where no phase transition is associated, meaning that no chemical reaction takes place and thus no products are formed. This makes hydrogen embrittlement the hardest to observe and understand. Hydrogen embrittlement is a delayed failure process. Once hydrogen has been introduced into a metal it will slowly degrade it from inside out without leaving easily identifiable characteristics. Delayed failure caused by hydrogen embrittlement will roughly follow these steps (Louthan, 2008):

1. Hydrogen atoms get introduced to the metal either during production, welding or usage.

2. The absorbed hydrogen prefers extraordinary sites where the lattice has been dilated and will migrate there.

3. In a brand new component extraordinary sites will be randomly distributed. This means that absorbed hydrogen will also be randomly distributed over the component.

4. When putting the component into service, it undergoes stresses. These stresses are focussed on certain locations in the components, causing macroscopic regions of lattice dilation. Such regions of focussed stress are generally the stiffest connections, such as flanges, sharp edges or sharp connecting lines.

5. Hydrogen will now prefer these regions where the lattice has been dilated and will accumulate. When the hydrogen in these regions reaches a critical concentration, a small crack is formed by the pressure the hydrogen exerts on the lattice.

6. This crack causes even more lattice dilation and thus an even higher preference for hydrogen. This means that the crack propagates itself. 7. Hydrogen will keep relocating to these cracks and repeat the process until it

reaches a critical size and the component fails. Hydrogen embrittlement can stop prior to component failure when the hydrogen concentration is not high enough to further dilate the lattice.

This process and its consequences have been observed in metals, steels and other alloys of all kinds, high and low strength (Louthan, 2008). A common way to try to determine whether a component could be embrittled is fractography, to observe intergranular fracture. However, intergranular fracture alone does not mean that a component has been embrittled, as does the absence of intergranular fracture state

(34)

the opposite. It is merely a pointer, as hydrogen embrittlement has no distinct fracture pattern. The fracture pattern depends on the condition of the metal, exposure temperatures, stress patterns and of course hydrogen concentration.

In regular gas pipelines, only the hydrogen introduced into the metal during production or welding would be of concern. If hydrogen were to be admixed to the gas grid, these pipes would be under constant hydrogen exposure, albeit a low concentration. Because hydrogen embrittlement is such a delayed process, even this low concentration could have an effect.

The degree of hydrogen embrittlement is temperature dependent in two ways. First, with an increase in temperature hydrogen atoms gain more mobility, and thus increase chances for hydrogen accumulation. On the other hand, at higher temperatures, hydrogen atoms are less inclined to move to regions with lattice dilation, resulting in less concentrated preferential sites, lowering chances for hydrogen accumulation. These two effects together bring the optimal temperature for hydrogen embrittlement to about 260 K(Louthan, 2008). This is very close to regular operating temperatures, making embrittlement an even bigger concern. (Melaina, Antonia, & Penev, 2013) state that hydrogen embrittlement is most severe at 294 K, which is even closer to regular operating conditions.

Steels, nickel-based alloys, metastable stainless steels and titanium alloys are most prone to hydrogen embrittlement (ASM Handbook, 1998). Both the pipe and flange used in a compressor station in the HTL contained titanium (< 0.15% and < 0.12%, respectively), with fittings such as flanges also containing a fair bit of nickel (0.30%). Aluminium and aluminium alloys showed to be one of the most resistant metals to embrittlement (Louthan, 2008).

High-strength steel, with a yield strength of > 700 MPa, is more susceptible to hydrogen induced cracking, while low-strength steel is only subject to ductility loss (Melaina, Antonia, & Penev, 2013). The yield stress of the DN1200 pipe used in the HTL showed to be 415MPa, so it would only be subject to ductility loss, and not so much hydrogen induced cracking. This means that the pipe would fail in a ductile mode instead of catastrophic brittle fracture.

4.2 Critical condition

However, for actual mechanical failure due to hydrogen absorption either the diffused concentration or the pressure needs to be relatively high. In a research conducted by (Maoqiu, Akiyama, & Tsuzaki, 2006) a steel bar with a diameter of 22mm was rolled. The bar has a steel grade of AISI 4135, which is similar to the L415 grade steel used in the HTL. The yield strength, however, was significantly higher at 1320MPa. In the sides of the bar small notches were made, and subsequently hydrogen was induced by electrochemical pre-charging. After 48 hours of severe hydrogen exposure due to the charging, a cadmium coating was applied to prevent hydrogen release. Multiple samples were made with ranging hydrogen absorption levels from 0.1 to 0.5 ppm. Then, a stress was applied at the notch of up to 90% of the yield strength and sustained up to 6000 minutes if fracture did not occur. They concluded that fracture did not occur when the hydrogen content was lower than 0.2 ppm. Since the stress applied in this research is about four times larger than is being observed in transport

Hydrogen admixture in the Dutch gas grid

Hydrogen admixture in the Dutch gas grid

Abstract

Contents

1

Module 1: Introduction

2

Module 2: Infrastructure specifications

3

Module 3: Flow assurance

4

Module 4: Integrity management and safety