Introduction

It is well-known that the usage of non-renewable fossil fuels leads to greenhouse gas (GHG) emissions. GHG emissions contribute to global warming and climate change issues (Nunez 2019). Thus, it is important to reduce usage of fossil fuels and bring global energy-related carbon dioxide (CO2) emissions to net-zero by 2050. This is to limit the global temperature rise to 1.5 °C to ensure a more sustainable future, as proposed by Intergovernmental Panel on Climate Change (IPCC) in 2018. To achieve the abovementioned goals, sectoral changes must be driven. For instance, the energy sector is a prime contributor of GHG emissions in most countries, due to its large scale usage of fossil fuels to produce energy. New forms of cleaner energy must be identified to reduce the dependency on fossil fuels.

Hydrogen (H2) has been identified as a potential alternative as it does not produce carbon emissions when used. H2 burns with oxygen and producing water (H2O) as the only product, hence eliminating any generation of emissions. It is considered a potential resource suitable for replacing fossil fuels as an energy carrier (Marchant 2021). However, the implementation of a clean H2 production process is not straightforward. H2 production can generally be categorised into (but not limited to) grey, blue, green and turquoise H2. Grey H2 refers to H2 which is produced from fossil fuels. Unfortunately, the process that produces the grey H2 also produces CO2 emissions. If the emissions are captured and stored during grey H2 production, the produced H2 is known as blue H2.

Both grey and blue H2 routes possess environmental and economic challenges (Hague 2021). Meanwhile, green and turquoise H2 are produced via systems that are powered by renewable energy such as solar energy and biomass gasification. For instance, green H2 is typically produced using electrolysis of water (H2O). The electrolysis system is powered by renewable energy and it produces green H2 along with oxygen (O2). On the other hand, turquoise H2 uses methane (CH4) pyrolysis to crack natural gas into H2 and solid char, where the latter can be used for agricultural purposes. Interestingly, this pathway can reduce emissions and produce valuable by-products at the same time (Marchant 2021). Nevertheless, green and turquoise processes face challenges in H2 production and storage. Firstly, additional cost will be required to store H2 before being transported to the end-users. Next, the H2 production process is energy intensive. The low efficiency of H2 production processes via renewable energy also limits the amount of H2 produced. This results in poor feedstock utilisation efficiency and energy losses when producing H2. Apart from that, the amount of H2 energy stored is highly dependent on the storage conditions and efficiency of the storage method. Thus, urgent studies are required to evaluate the technoeconomic feasibility of green and turquoise H2 production.

Mathematical optimisation can be a possible way to evaluate the technoeconomic feasibility of green and turquoise H2 production. Mathematical optimisation involves modelling a given set of alternative processes and optimising said model based on a given criteria (Bynum et al. 2021). The optimal solution generated from the mathematical optimisation model represents the optimal process configuration (Andiappan 2017). As reported in the literature, several mathematical models were developed for green H2 production via renewable energy sources, such as solar, wind, geothermal, tidal, hydro, and biomass. It is worth noting that these renewable energy sources were used to fulfil the energy requirements for the production of green H2. For example, Shaofu et al. (2021) proposed an optimised green H2 process that recycled waste energy. The study focused its analysis on energy and exergy performance of the green H2 process. Next, Fragiacomo and Genovese (2019) developed a mathematical model to analyse the energy performance of green H2 production from polymer electrolyte membrane (PEM) and alkaline electrolysis cell (AEC). The energy performance is evaluated in terms of energy efficiency and system performance. Fragiacomo and Genovese (2019) also applied the mathematical model to a case study in the year 2020. Besides, Gu et al. (2022), Zhao et al. (2022) and Frick et al. (2022) also perform technoeconomic feasibility analysis of green H2 production. However, these works solely focused on the energy consumption of a green H2 process. Economic aspects of these processes were not considered. It is important to consider the economic aspects of green H2 production to ensure economic feasibility.

Various works have been developed to optimise green H2 production based on economic aspects. For example, Wissocq et al. (2021) proposed an optimisation model for green H2 production that consisted of wind energy, PEM and compressed gas storage. In the model (Wissocq et al. 2021), the capital expenses and operating costs was minimised. In addition, Tebibel (2021) and Grüger et al. (2019) optimised green H2 system consisting of wind energy and AEC. Both models (Grüger et al. 2019; Tebibel 2021) maximised wind energy utilisation and minimised H2 production cost. On the other hand, Han et al. (2019) studied the cost and efficiency of solid oxide electrolysis cells (SOECs) in various gas mixtures. Besides, several previous works also considered the economic aspect of green H2 production, such as Kazi et al. (2021), Fan et al. (2021), Bonacina et al. (2021), Jalili et al. (2021), Park (2022), Galitskaya (2022), Cloete et al. (2022), Jang (2022) and Lal and You (2023). Nevertheless, both the important criteria mentioned above, economics and energy consumption, should be evaluated simultaneously in an optimisation model. With that, the optimisation model will determine the most economically feasible green H2 process with least energy.

Cao et al. (2021a) proposed an optimisation model that compromises the economics and energy consumption of green H2 production. The proposed model (Cao et al. 2021a) considered biomass and solar energies for PEM. Then, Alirahmi et al. (2022) and (Cao et al. 2021b) proposed a green H2 process that consisted of geothermal energy and H2 produced from PEM. In addition, Khouya (2020) developed a mathematical model to assess the performance of AEC H2 production supplied from wind and solar energies. Similarly, the model’s performance (Khouya 2020) was evaluated based on efficiency, energy and economic aspects. Besides, several works also considered the similar criteria in the optimisation model, such as Alirahmi et al. (2021), Li et al. (2022), Zghaibeh (2022) and Cao (2022).

Aside from producing green H2, the storage method should also be considered. Zhang et al. (2020) proposed an optimisation model for electrical storage of H2 in batteries and H2 storage in compressed gas. Both energy storage methods were compared quantitatively in term of costs and advantages. Next, a mathematical model was developed by Zheng et al. (2020) to optimise the input power of cryocooler for liquid H2 storage. On the other hand, Bai et al. (2021) focused on optimising H2 storage in metal hydrides.

Based on the articles reviewed above, several key areas are identified as gaps. These gaps are summarised as follows:

  • There are limited mathematical optimisation models that consider efficiency, energy consumption, economics, and storage of H2 simultaneously in green H2 production.

  • Previous studies have pre-selected specific green H2 production routes and did not conduct screening of all possible alternative pathways. As an example, only one type of renewable energy is supplied to one H2 production technology.

  • There are limited studies on optimising turquoise H2 production.

Therefore, this paper presents a mathematical optimisation model to optimise and screen various green and turquoise H2 production routes based on several aspects simultaneously. These aspects include economics, energy consumption, efficiency and storage of both green and turqoise H2. A case study is solved to illustrate the proposed approach. The following sections are distributed as follows. The next section describes methodology used to develop the optimisation model in this work. Following this, the developed model is applied in several case studies. The results are then analysed and discussed in the Discussion section. This is followed by recommendations and lastly the conclusions.

Methodology

Number of systematic approaches have been developed to synthesise process and systems in the literature. These include heuristic approaches, insight-based approaches, graphical approaches and mathematical optimisation approaches (Andiappan 2017). In heuristic approaches, decision-makers utilised rules derived from their own knowledge and experience to generate process designs. Then, modifications will be conducted to improve the performance of the base case. Nevertheless, these approaches are highly dependent on the experience of the designer. Besides, complex design decisions might be overlooked; thus, an optimal configuration for a process or system is not guaranteed. Insight-based approaches obtain targets for a system from thermodynamic and other physical science principles. However, such approaches are often used as tools to determine energy recovery opportunities and utility requirements that a process or system should deliver, instead of screening alternative technologies (Andiappan 2017). On the other hand, insight-based approaches are simple methods to solve optimisation problems involving one or two variables but may not feasible for much larger-scale optimisation problems (Bhatti 2002).

Hence, a mathematical optimisation/programming model will be developed to overcome the previous limitations. Mathematical optimisation is used as it considers possible alternatives to synthesise any system or process. This is a more comprehensive approach as compared to heuristic approaches. Mathematical optimisation determines the topology of energy systems. This presents a better advantage than insight-based approaches. Besides, mathematical optimisation approaches allow several equality and inequality constraints, unlike graphical methods limited to two variables (Andiappan 2017).

The methodology of the current research is shown in Appendix Figure A1. In this work, the research can be divided into six stages. In Stage 1, all technologies and possible H2 pathways is compiled into a single diagram, known as a superstructure. Figure 1 shows how a general superstructure looks like. It can be seen that the superstructure is a diagram presenting all possible interconnections for pathways that convert inputs into outputs. Next, the working principle of the technologies is evaluated. The operational data and technical information for the identified technologies are collected through literature review and state-of-art analysis. Examples of operational data and technical information collected are as presented in Table 1. The collected information for all technologies in this work are shown in Appendix Tables A1 to A15.

Fig. 1
figure 1

Generic Superstructure

Full size image
Table 1 Operational Data and Technical Information Required
Full size table

In Stage 2, mathematical equations are formulated based on the understanding of the working principles of the processes in the superstructure. The formulated equations represent the behaviour of technologies involved in overall H2 production. A general methodology is used in this work to formulate these equations based on Fig. 1. As shown in Fig. 1, r \(\in\) R represents the feed, being sent into technology h \(\in\) H to produce product p \(\in\) P. Based on different criteria, the optimisation objectives can be revised to meet the requirement. The objective functions explored in this work are discussed in Case Study section. Following the superstructure, the feed flowrate, \({\text{F}}_{r}^{\text{available}}\), can be sent into each technology h to be converted into different products p as shown in Eq. 1.

$${\text{F}}_{r}^{\text{available}}\ge \sum_{h=1}^{H}{F}_{rh}\forall r$$
(1)

Meanwhile, product p flowrate from specific technology h, \({F}_{hp}\), can be calculated utilising the conversion, \(\varphi_{rhp}\), from feed flowrate, \({F}_{rh}\). \({F}_{hp}\) is determined for calculating capital expenditure (CAPEX) and operating expenditure (OPEX) in latter parts. This is because some of the technologies depends on product flowrate instead of feed flowrate on cost calculations. Thus, \({F}_{hp}\) is represented by Eq. (2):

$$F_{hp}=\sum_{r=1}^RF_{rh}\varphi_{rhp}\forall h\forall p$$
(2)

Next, total flowrate of product, \({F}_{p}\), from all technology can be calculated similar to \({F}_{hp}\) above. \({F}_{p}\) is represented by Eq. (3):

$$F_p=\sum_{h=1}^H\sum_{r=1}^RF_{rh}\varphi_{rhp}\forall p$$
(3)

Following this, Eq. (4) is added to the model to describe how each technology h can operate within a certain operating range, given by its minimum (Fhmin) and maximum capacity (Fhmax). Fhmin functions as a minimum capacity required for feasible operation of a given technology (e.g., gas turbine needing a minimum a volume of gas to begin operation). This value can be obtained from information on minimum throughput, typically provided by vendors of a given technology. Meanwhile, Fhmax is specified based on the maximum allowable throughput of the technology. This also can be obtained from vendors. In addition, a binary variable, \({b}_{h}\), is included in Eq. (4) as shown below. If the value of this binary is one (1), it means that technology h is selected and vice versa.

$${\text{F}}_{n}^{\text{min}}{b}_{h}\le \sum_{r=1}^{R}{F}_{rh}\le {\text{F}}_{h}^{\text{max}}{b}_{h}\forall h$$
(4)

After flowrates of feed (\({F}_{rh}\)) and product (\({F}_{hp}\)) are determined, CAPEX, OPEX and energy consumption can be worked out easily. This is achieved by multiplying required flowrate with respective factor.

The equation that determines the energy consumption of technology h, (\(Ele{c}_{h}\)) is shown in Eq. (5). Here, the energy factor or specific energy consumption, Eh, is multiplied with either the input or output flowrate (i.e., \({F}_{rh}\) or \({F}_{hp}\)) of technology h.

$$\begin{array}{c}Ele{c}_{h}=\sum_{r=1}^{R}{F}_{rh}{\text{E}}_{h}^{\text{feed}}\forall h\\ or\\ Ele{c}_{h}=\sum_{p=1}^{P}{F}_{hp}{\text{E}}_{h}^{\text{product}}\forall h\end{array}$$
(5)

The same applies to the equations for OPEX and CAPEX of technologies considered. The OPEX factor, Oh, and the CAPEX factor, Ch, are multiplied with flowrate required to determine OPEX and CAPEX, respectively. Thus, OPEX and CAPEX of technologies, \(OPE{X}_{h}\) and \(CAPE{X}_{h}\), are shown in Eq. (6) and Eq. (7). It is worth noting that in some cases, the energy consumption in technology h can also be used instead of flowrate to determine the operating cost in Eq. (6).

$$\begin{array}{c}OPE{X}_{h}=\sum_{r=1}^{R}{F}_{rh}{\text{O}}_{h}^{\text{feed}}\forall h\\ or\\ OPE{X}_{h}=\sum_{p=1}^{P}{F}_{hp}{\text{O}}_{h}^{\text{product}}\forall h\end{array}$$
(6)
$$\begin{array}{c}CAPE{X}_{h}=\sum_{r=1}^{R}{F}_{rh}{\text{C}}_{h}^{\text{feed}}\forall h\\ or\\ CAPE{X}_{h}=\sum_{p=1}^{P}{F}_{hp}{\text{C}}_{h}^{\text{product}}\forall h\end{array}$$
(7)

Lastly, the OPEX and CAPEX of technologies can be summed in Eq. (8) to determine the total annualised cost (TAC). In Eq. (8), AFh is the annualising factor for the CAPEX considered.

$$TAC=\sum_{h=1}^{H}{\text{AF}}_{h}\times CAPE{X}_{h}+\sum_{h=1}^{H}OPE{X}_{h}$$
(8)

Nevertheless, if a process involves more stages than shown in the general superstructure (i.e., Fig. 1), the equations and steps can be repeated for the following stages accordingly. Following this, Stage 3 is where the equations above serve as basis to develop the mathematical optimisation model in a commercial optimisation software. The commercial optimisation software used is LINGO version 19 with global solver.

The mathematical optimisation model is solved using objective functions such as minimising cost, maximising H2 production. The optimal solutions generated in Stage 3 are then determined and examined in Stage 4. To validate the optimum results in Stage 3, the results need to be compared with the literature in Stage 5. Besides, benchmarking with available analysis, case studies, research works and literature can be performed to countercheck the results. If the result is acceptable, the final optimised network is accepted in Stage 6. In the event where the results are not sensible to be implemented in the industry, Stage 1 to Stage 4 will be revisited to revise and re-examine the data. Finally, the optimal H2 production process based on current research and developed model will be used as a basis for future analysis.

In the next section, several case studies are analysed to determine optimal H2 network based on respective objection function. The optimal H2 network will be recommended by model after inserting all the generic equations above into mathematical optimisation software.

Case Study

The electricty and transportation fuel demands in Malaysia are typically supplied from fossil fuels. As reported in Ritchie and Roser (2021), fossil fuels make up 93.61% of the total fuels used in Malaysia in 2021. In the same year, Malaysia formulated the 12th Malaysia Plan, which stated goals to achieve carbon-neutrality by 2050. The first step for the Malaysian government to achieve this goal is to stop building new coal-fired power plants (KPMG 2021). Therefore, alternative renewable and clean energy must be introduced to fulfil future energy demands in Malaysia. In this respect, H2 is seen as a potential alternative energy resource that is expected to play an important role in converting Malaysia into a carbon–neutral nation (Salim 2021).

Thus, the case studies presented in this section are aimed to design an optimal H2 production network from clean resources available in Malaysia. One of the widely available and accessible clean resources is the biomass wastes from oil palm industry. Oil palm industry is the primary producer of agriculture-based biomass in Malaysia. This leads to significant wastes disposal problem from massive production of oil palm biomass, such as empty fruit bunches (EFB), palm kernel shell (PKS), palm-pressed fiber (PPF), palm oil mill effluent (POME), etc. (Jafri et al. 2021). In general, 22% of EFB and 67% of POME produced from sterilisation and stripping of fresh fruit bunches (FFB) (Abdullah and Sulaiman 2013). In this case study, only EFB and POME are considered eventhough other wastes, such as PKS and PPF, are generated from oil palm industry. The reason for this is that EFB and POME are available in large amounts as compared to other biomass/wastes. For instance, 25 tonnes/hr of FFB is able to produce 6.2 tonnes/hr of EFB and 15 tonnes/hr of POME. However, only 2 tonnes/hr of PKS and 3.1 tonnes/hr of PPF are produced at the same time. Thus, high availability of POME and EFB ensures consistent supply of biomass to electricity generation and H2 production sections. In addition, based on the current industrial practise, PPF and PKS are mainly used as feedstock for the combined heat and power system in mills. On top of that, EFB has relatively higher annual energy potential per area of land, compared to PPF and PKS, which is 75 GJ-LHV/hectare/year compared to 49 and 17 GJ-LHV/hectare/year, respectively. Due to the higher energy potential, the OPEX in electricity generation can be reduced for a fixed energy demand (Kamahara et al. 2018).

As raw materials, EFB and POME have great potential to be converted into value-added products, including clean H2 (Jafri et al. 2021), biochar and biogas. Hence, this case study considers EFB and POME as the biomass source for electricity generation and H2 production. This is shown in the superstructure below (Fig. 2). Aside from EFB and POME, H2O and CH4 are also consideredas raw materials in H2 production section. H2O is supplied to electrolyser for production of green H2, while CH4 will undergo methane pyrolysis to produce turquoise H2. Interestingly, CH4 has potential to be converted into cleaner set of products, such as H2 and char, as shown in Fig. 2 below.

Fig. 2
figure 2

Case Study Superstructure

Full size image

Based on the literature review, all the related technology which able to convert EFB and POME into H2 and related supporting technologies are summarised in Table 2.

Table 2 Technologies Involved
Full size table

Subsequently, several case studies are conducted to determine the optimal H2 network based on specific objective functions. The objection functions considered in this research are shown in the Table 3, along with the purposes of each objective. Each objective function accounts for different decision-makers. This study addressed these objective functions in order to present their influence on the optimal configuration. Doing this will allow decision-makers to compare the optimal configurations from each objective function and determine the common technologies while drawing out the differences. This provides decision-makers an understanding on technologies that are critical in each objective function and provide flexibility for opting for other decisions later.

Table 3 Objective Functions
Full size table

For instance (as shown in Fig. 3), a given technology could be a common choice for each objective function. If the decision-makers are faced with a mandate to reduce energy consumption in the future after implementing a design, having knowledge of the technologies selected in the another case i.e., low energy requirement case will be important. From here, decision-makers will be able to know how many of the technologies selected in the case of minimum cost overlaps with the case of minimum energy requirement. With this, it will be clear to know which technologies that may need to be replaced or retrofitted in the future to bridge the remaining gaps and avoid a total system retrofit.

Fig. 3
figure 3

Principle for Investigating Several Objective Functions

Full size image

Each case study stated above will be split into 2 different scenarios. Firstly, Scenario 1 offers a case where the end users are co-located to this system and have pipeline connections available to transport the H2 produced. This means that the H2 produced is assumed to be equal to H2 demand, which is 5,700 kg/hr (Scenario 1). Following this, Scenario 2 offers a case where the end users are not in close vicinity of this system to have pipeline connections. It determines how much more H2 must be produced in order to be stored prior send to end user. Hence, the value of H2 produced is allowed to be varied based on how much H2 needs to be stored to meet the H2 demand. On top of that, percentage of costs increased from Scenario 1 to Scenario 2 can be compared. Therefore, the model developed is flexible with regard to the extent of H2 stage. Table 4 summarised all the assumptions have been made in this research:

Table 4 Case Study Assumptions
Full size table

Following this, the supply of EFB and POME is based on production rate and area of plantation from typical palm oil mill (Kim 2011). Based on the oil palm planted area, flowrate of EFB and POME can be calculated. The general wastes on EFB and POME are found to be 4.42 and 13.34 tonnes/hectare/year. This leads to 158,000 kg/hr and 478,000 kg/hr available EFB and POME to be supplied by a typical palm oil mill. Regarding H2 demand, 5,700 kg/hr will be used in the model. This is based on target of Sarawak Economic Development Corporation (SEDC) by year 2025. Initially, SEDC aimed to produce H2 at a rate of minimum 10,000 tonnes/year. Then, H2 production will be scaled up to 50,000 tonnes/year or 80,000 tonnes/year depends on the availability of power to SEDC (Kong 2021). Thus, model developed will be based on scaled up production of 50,000 tonnes/year.

Case Study 1: Minimising Total Annualised Cost (TAC)

As shown in Eq. 8, Total annualised cost (TAC) considered CAPEX and OPEX of technologies, and raw material cost required. Therefore, minimising Eq. 8 subjected to the Eqs. 1, 2, 3, 4, 5, 6, and 7, the optimised network is shown in Fig. 4 with the minimum TAC, $2.71 × 109. This resulting configuration has a $4.53 × 104 / kg H2 and $4.77 × 104 / kg H2 for Scenario 1 and 2, respectively.

Fig. 4
figure 4

Optimal H2 Network for Case Study 1

Full size image

As shown above, crystalline silicon photovoltaics cell (CS), dark fermentation coupled with microbial electrolysis cell (DFMEC) and compressed gas cylinder (compressed H2 gas) are being selected. These represent the electricity generation, H2 production and H2 storage technologies respectively. Interestingly, the technologies chosen for both scenarios are the same.

Regarding electricity generation section, CS is chosen among solar and biomass energies. Firstly, the efficiency of CS is the highest (22%) among solar energy (Benda and Černá 2020), along with the lowest CAPEX (1,130 $/kW) required (Tjengdrawira et al. 2016). Although OPEX required by thin-film solar cell (TF) is lower than CS, 0.038 $/kW compared to 0.042 $/kW (Bellini 2021), however, the efficiency of TF is comparatively lower than CS (efficiency of 16% compared to 22% from CS as mentioned) (Benda and Černá 2020). When the efficiency is lower, more solar energy is required to produce the electricity needed from H2 production and storage sections. The higher the solar energy requirement, the larger the area of TF solar panels, and the higher the CAPEX of TF calculated from the total surface area of the panel. As such, higher CAPEX and OPEX are required in TF to produce the same amount of electricity compared to CS. This shows that higher total TAC is required from technology with lower efficiency. Therefore, the efficiency of technology predominates the CAPEX and OPEX required in the electricity generation section.

This also applied to biomass energy, but the efficiency of biomass energy is not as straightforward as solar energy. Regarding biomass energy, the final producible electricity is highly dependent on available biomass and conversion of biomass into intermediate gas, followed by conversion of intermediate gas into electricity. As an example, electricity produced from pyrolysis and gasification of EFB depends on the composition of carbon monoxide (CO) within syngas (SG) produced from the decomposition of EFB. When 1 kg of EFB is supplied, only 0.35 kg and 0.8 5 kg of SG can be produced from pyrolysis and gasification (Bogale 2017). Following this, extremely low CO of 47.9 wt% (Zajec 2009) and 27.7 wt% (Mohammed et al. 2011) within SG of pyrolysis and gasification are found. Although 3.556 MW and 3.048 MW of electricity can be produced from 1 kg of CO in pyrolysis and gasification, however relatively high EFB flowrate is required to produce the same amount of electricity from solar energy. Similarly, anaerobic digestion of POME also suffered from low conversion of POME into biogas (BG), where only 0.0314 kg of BG can be produced from 1 kg of POME.

Contrastingly, combusting 1 kg of EFB is able to produce 4.555 kg of high-pressure steam (HPS). Nevertheless, electricity generated from combustion is limited to HPS production capacity from the steam turbine. The steam turbine can only be functioned with at least 5,000 kg of HPS per hour (Hong 2022). When the demand of H2 is low, the electricity required from H2 production and storage technologies is low. This leads to lower HPS is required to produce the essential electricity for both sections. Thus, the combustion of EFB will not be chosen as the minimum HPS flowrate requirement from the steam turbine cannot be met.

In terms of H2 production and storage sections, the justification for selected technologies is more direct compared to electricity generation. Starting from the H2 production section, DFMEC dominates the selection out of all green and turquoise H2. DFMEC required low CAPEX and OPEX, at the same time having a high conversion of POME into H2.

In terms of traditional water electrolysis technologies, polymer electrolyte membrane (PEM), alkaline water electrolysis (AWE) and solid oxide electrolysis cell (SOEC), endure from the low conversion of water and high CAPEX and OPEX requirements. Surprisingly, recent water electrolysis that is still under research stage, anion exchange membrane (AEM), presents an attractive conversion and OPEX in contrast with DFMEC. However, maximum production capacity of H2 from single AEM unit is limited at 0.899 kg per hour (SinoHy Energy 2012). Thus, higher CAPEX and OPEX are required to implement more AEM units in achieving the demand of H2. This is followed by huge increase of TAC required to reach the H2 demand.

On the other hand, turquoise H2 produced from methane pyrolysis shows high conversion of CH4 and low OPEX. Nevertheless, CAPEX required in such technology is exceptionally high compared to green H2. This causes methane pyrolysis is not being selected in the optimal H2 production network.

Regarding H2 storage, compressed H2 gas is chosen out of all physical and chemical-based H2 storage technologies. Compressed H2 gas is the most common technique to store H2 efficiently (95%) (Norouzi 2021). The main reason for compressed H2 gas to be widely used could be due to comparatively low CAPEX and OPEX requirements at $ 23.4 /(kg/hr) (Patonia and Poudineh 2020) and $ 0.4 /kg (Dias et al. 2020).

Case Study 2: Minimising Capital Expenditure (CAPEX)

In this case study, Eq. 7 is set as optimisation objective. Minimising Eq. 7 subjected to Eqs. 1, 2, 3, 4, 5, and 6, the optimal H2 network with minimum CAPEX is showed in Fig. 5. This resulting configuration has a $9.75 × 104 / kg H2 for Scenario 1) and $1.03 × 105 / kg H2 for Scenario 2. Note that biomass energies of anaerobic digestion and gasification are chosen in the electricity generation section. Both technologies generate electricity by sending intermediate gas, BG and CO respectively, into the gas engine. However, CAPEX factor of anaerobic digestion and gasification is not at the basis. Based on research article, the CAPEX factor for anaerobic digestion depends on the flowrate of POME ($ 4,292.4 /(kg POME/hr) (Wilson et al. 2016), while gasification depends on electricity produced ($ 562.744 /kW) (Susanto et al. 2018). Therefore, assuming 100 kW of electricity produced, anaerobic digestion only required $ 999 of CAPEX, compared to gasification at $ 56,300. Thus, anaerobic digestion will first be recommended by model based on current objective function. Although CAPEX of anaerobic digestion is lower than gasification, capacity of gas engine for BG is limited to 23 MW/h (Haga 2011). Thus, model is forced to select an alternative that also required a low CAPEX to fulfil the remaining energy requirement.

Fig. 5
figure 5

Optimal H2 Network for Case Study 2

Full size image

In short, biomass energy required lower CAPEX than solar energy. This is because set-up of biomass technology is way simpler yet ease in control compared to solar energy. As an example, a single gasifier can produce the same amount of electricity from a few hectares land area of solar panel. However, the larger the covered land area, the higher the cost to implement electricity generation system. This turns out that the initial investment in CAPEX of solar energy will be incredibly high. DFMEC with low CAPEX ($ 105.29 /(kg POME/hr)) (Aiken et al. 2019) and high conversion is recommended by the model in H2 production section. In addition, SG from the gasification of EFB consists of CO and H2. After passing through a separator, CO will be sent for electricity generation. At the same time, H2 from gasification did contribute a minute amount to the total H2 produced. Therefore, H2 produced in this network compromises all biomass technologies. Similar to Case Study 1, compressed H2 gas with the lowest CAPEX ($ 23.40 /hr) and highest efficiency is being chosen in H2 storage section.

Case Study 3: Minimising Operating Expenditure (OPEX)

To minimise OPEX, Eq. 6 is set as optimisation objective. Minimising Eq. 6 subjected to Eqs. 1 – 5 and 7, the optimal H2 network is presented in Fig. 6. This resulting configuration has a $5.68 × 104 / kg H2 for Scenario 1 and $6.02 × 104 / kg H2 for Scenario 2. In contrast to Case Study 2, TF and CS solar energies are chosen in the electricity generation section. Although there is no restriction on the surface area of solar panels, a maximum area of 1 × 107 m2 is assumed in the model. This explains why both TF and CS will be chosen together even when OPEX of TF is lower than CS, $ 0.038 /kW compared to $ 0.042 /kW. After maximum solar panel area of TF is reached, model selects an alternative that also required a low OPEX to fulfil the remaining energy requirement. In summary, Case Study 3 shows that solar energy required lower OPEX than biomass energy. This is due to OPEX of biomass energy also considered raw material cost, EFB and POME. Contrastingly, solar energy will be widely available and free of charge. Thus, raw material cost is considered an impactful additional cost in the annual production and storage of H2. Thus, this leads to relatively high OPEX requirements from biomass energy than solar energy.

Fig. 6
figure 6

Optimal H2 Network for Case Study 3

Full size image

In terms of H2 production section, methane pyrolysis with lower OPEX than DFMEC ($ 2.19 /kg H2) (Zore et al. 2021) will be prioritised. Nevertheless, current H2 production capacity of methane pyrolysis is limited to 400 kg/hr (Mesrobian 2021). Therefore, the remaining H2 demand will be met by DFMEC. As mentioned, DFMEC acquired both low OPEX ($ 2.58 /kg H2) (Kadier et al. 2016)and high conversion. Therefore, H2 produced in this network compromises both turquoise and green H2.

As the objective function is minimising OPEX, ammonia (NH3) with the lowest OPEX ($ 0.18 /kW) (Brown 2020) and high efficiency (93%) (Norouzi 2021) is being chosen in H2 storage section.

Case Study 4: Minimising Energy Requirement

In this case study, Eq. 5 is minimised subjected to Eqs. 1, 2, 3, 4, 6 and 7. Figure 7 presents the optimal H2 network of Case Study 4. In this case study, methane pyrolysis with the lowest energy requirement (5.23 kW/kg H2) (Machhammer et al. 2016) will be selectively chosen. However as mentioned in Case Study 3, the remaining H2 demand will be fulfilled by DFMEC due to restriction on production capacity in methane pyrolysis. DFMEC is being chosen due to relatively low energy consumption (21.6 kW/kg H2) (Mehmeti et al. 2018) as compared to other green technologies (range from 36.14 kW/kg H2 to 60.09 kW/kg H2). This resulting configuration has a $9.58 × 104 / kg H2 in Scenario 1 and $1.08 × 105 / kg H2 in Scenario 2.

Fig. 7
figure 7

Optimal H2 Network for Case Study 4

Full size image

Regarding the H2 storage section, liquid organic hydride (LOH) with the lowest energy consumption (0.7 kW/kg H2) (Andersson and Grönkvist 2019) is selected among all storage methods. Meanwhile, all electricity required from both sections will be fully supplied from green source of EFB. EFB will then undergo gasification to produce the relative electricity demand. Hence, the entire H2 network is green by using green sources and technologies to produce electricity and H2. After an optimal H2 network based on the respective function is obtained, a detailed discussion regarding the comparison between scenarios and case studies is explained next.

Discussion

In this section, a comparison between scenarios and case studies is shown.

Effect of H2 Demand

The effect of H2 demand on results of Scenarios 1 and 2 from all case studies were evaluated. This is presented in terms of percentage increment and summarised in Fig. 8.

Fig. 8
figure 8

Comparison between Scenarios

Full size image

In terms of cost (Case Studies 1 to 3), the percentage difference between both scenarios ranges from 5.26% to 7.83%. This is clear that amount of H2 produced will be higher in Scenario 2 compared to Scenario 1. In Scenario 2, 5,700 kg/hr of H2 will be stored in order to meet the demand. Thus, additional amount of H2 must be produced from chosen technology in each case study. Refer to Figs. 3 and 4 (Case Studies 1 and 2), additional 300 kg/hr of H2 will be produced if compressed gas cylinder is chosen as the H2 storage technology. Besides, additional 430 kg/hr of H2 must be produced in order to store H2 in terms of ammonia to meet the demand, as shown in Fig. 6 (Case Study 3). Lastly, based on Fig. 7 (Case Study 4), additional 730 kg/hr H2 is produced in order to store sufficient H2 to meet the demand when LOH is chosen as H2 storage technology. This overcomes the compression loss or low reaction efficiency to convert H2 from production technology to storage condition. Nevertheless, higher cost is required for all 3 sections, i.e., electricity generation, H2 production and storage, to compensate the increased H2 production and storage rates.

Next, Case Study 4 shows a difference of 13.43% in energy requirement. This proves that the more the H2 produced and stored, the higher the energy consumed from both sections. Besides, results from each case study and scenario showed that H2 production technology has higher energy requirement compared to H2 storage. This is because energy factor of H2 production technology is comparatively higher than H2 storage. More energy will be required in production technology to produce H2 at demand rate.

Comparison between Case Studies

A comparison between case studies was conducted based on results in Scenario 1. As the optimal H2 network for both scenarios were the same, the percentage difference between case studies were similar too. Thus, only Scenario 1 was compared and discussed below. Besides, Case Study 1 served as the base case in all conducted comparisons.

Firstly, the costs comparison between case studies are summarised in Fig. 9. As shown, raw material was compared individually to investigate the effect of flowrates on respective costs requirement. The raw material cost in Case Study 1 only considered POME, which was sent into DFMEC. This route used raw materials (i.e., solar energy) that did not have cost.

Fig. 9
figure 9

Costs Comparison between Case Studies

Full size image

In Case Study 2, a 0.99% increase in raw material cost as POME was sent into the anaerobic digester while EFB was sent to gasification. The limited electricity production capacity in the gas engine after anaerobic digestion, allowed only 53.544 kg/hr of POME to be utilised. Then, the remaining electricity requirement was fulfilled by gasification of EFB. Even when large amount of EFB is required to undergo gasification, the percentage of increment is low. This is due to the low unit cost of EFB compared to POME (i.e., $ 0.01202 /kg compared to $ 0.74 /kg). Hence, this proves that the OPEX in biomass energy is fully dependant on remaining operating system within equipment involved.

Interestingly, CAPEX of Case Study 2 was reduced by 57% once biomass energy is used as the electricity source. This is concluded after comparing this to Case Study 1, which utilised solar energy instead. However, this is accompanied by 148% increase in OPEX of the optimal network. As mentioned above, the equipment for biomass energy is relatively simpler with lesser production constraints that leads to low CAPEX required. Nonetheless, the effect of high OPEX requirement from biomass energy cannot be neglected.

OPEX in Case Study 3 can only be reduced by 8.8%, along with 558% increase in CAPEX required. Although methane pyrolysis and NH3 required relatively low OPEX, but the CAPEX of these 2 technologies computed large values. Other than CAPEX and OPEX, cost of methane is the highest among all raw materials considered. This led to a 18% increase in raw material cost, even after reduction in requirement of POME into DFMEC.

The TAC in Case Study 4 increased by 112%, even when energy requirement was the lowest. The main reason of such increment in raw material cost is due to supply of methane for H2 production. As shown in Appendix Table A5, methane is the most expensive raw material considered in this work. On top of that, flow of EFB into gasification also leads to an additional cost for raw material. Besides, contrasting effect on CAPEX and OPEX is expected in Case Study 4 when solar energy is substituted by biomass energy. Nonetheless, effect of OPEX is much greater than CAPEX, as shown in Case Study 2 and 3. Therefore, utilisation of biomass energy required 137% higher OPEX, compared to solar energy, to generate the same amount of electricity.

Table 5 Water Consumption Rate
Full size table

The comparison of energy requirement between case studies are summarised in Fig. 9. As indicated in Fig. 9, energy required in H2 production of Case Study 2 is lesser than Case Study 1. As EFB gasification is chosen as one of the electricity generation technology, small amounts of H2 was supplied from gasification in Case Study 2. This reduces small amount of energy consumption in H2 production technology to meet the same demand. Following this, compressed H2 gas is chosen as the final H2 storage method in both case studies. This led to same energy requirement in H2 storage section where technology efficiency and H2 production demand are the same. Hence, 0.01% reduction in energy requirement comes from H2 production section only.

When OPEX or energy requirement is minimised, methane pyrolysis was chosen to produce H2 up to the production capacity limit. This is because methane pyrolysis has the lowest OPEX and energy requirements. In addition, in H2 storage method of Case Study 3, NH3 with the lowest OPEX requirement is chosen instead of compressed H2 gas. At the same time, the energy required from NH3 is lower than compressed H2 gas, which is 2 kW/kg H2 (Andersson and Grönkvist 2019) compared to 6 kW/kg H2 (Sheffield et al. 2014). Therefore, total energy requirement from Case Study 3 had decreased by 19% compared to Case Study 1. For the H2 storage of Case Study 4, LOH with the lowest energy requirement was chosen. This contributes to 23% reduction in overall energy required from H2 production and storage sections. Based on Fig. 10, energy consumption in the H2 production section ranged from 70 to 90% of overall energy consumption. This proves that H2 production technologies is more energy intensive compared to H2 storage methods, as mentioned in section above.

Fig. 10
figure 10

Energy Comsumption Comparison between Case Studies

Full size image

Next, water consumption rates from each case study are summarised in Table 5. Technologies that required water consumption in this study were EFB combustion, EFB gasification and water electrolysis in green H2. In that respect, Case Study 1 and 3 do not involve any water consumption as neither of the aforementioned technology were selected. However, the water consumption in Case Study 2 and 4 solely comes from the gasification of EFB to produce the electricity required in the following sections.

Interestingly, electricity generation in Case Study 4 purely depends on gasification, while Case Study 2 only requires gasification when the production limit of anaerobic digestion is reached. However, Table 5 shows that Case Study 4 consumed lesser water than Case Study 2. As water consumption in gasification depends on the mass flowrate of EFB, while EFB flowrate depends on the electricity required. Thus, when energy requirement is minimised in Case Study 4, flowrate of EFB is lowered, water consumption in gasification will be reduced accordingly. A summary of technology recommendation based on each objective function is shown in the Fig. 11.

Fig. 11
figure 11

Recommended Technologies

Full size image

In the electricity generation section, the common solar and biomass energy technologies that the model recommended is CS and EFB gasification, respectively. Following this, DFMEC will always be predominated in all objective functions. Finally, compressed H2 gas was a more commonly chosen H2 storage method among the case studies. The main justifications for recommended technologies are summarised in the next section.

Recommendations

Figures 12 and 13 summarise the advantages and limitations of each recommended type of technology and the suitability of technology based on each case study. As shown below, solar energy requires lower OPEX and presents higher electricity production efficiency. This leads to relatively lower overall cost. Thus, solar energy was recommended when OPEX or TAC is minimised. Although biomass energy required lower CAPEX, however, such technology is limited by production capacity and high overall cost. Hence, biomass energy was only recommended when CAPEX is required to be minimised.

Fig. 12
figure 12

Recommendation for Electricity Generation Section

Full size image
Fig. 13
figure 13

Recommendation for H2 Production Section

Full size image

For the H2 production technology, electrolysis was not recommended due to high cost and high energy requirements. Besides, electrolysis has low efficiency and production capacity. This leads to a high number of electrolysis units required to produce specific H2 demand. Thus, the overall cost will be increased simultaneously when more electrolysis units are implemented.

On the other hand, H2 produce via methane pyrolysis showed advantages in acquiring the lowest OPEX, lowest energy consumption and highest yield of H2. Nevertheless, limited production capacity and high CAPEX requirements have restricted such technology from being widely implemented. Hence, methane pyrolysis can only be recommended when energy requirements or OPEX are to be minimised.

Last but not least, DFMEC will be predominated in all objective functions based on the following advantages. In contrast to electrolysis, DFMEC required low cost and consumed lesser energy to produce the same amount of H2. In addition, DFMEC has high efficiency to convert biomass (i.e., POME) to H2 within capacity. Therefore, DFMEC is a common technology selected for the H2 optimal networks in all case studies.

Alternatively, the H2 storage section is more straightforward than the electricity generation and H2 production sections. Therefore, when minimising TAC or CAPEX, compressed H2 gas with the highest electricity production efficiency and required low CAPEX and OPEX will be recommended. On the other hand, NH3 and LOH required the lowest OPEX and energy consumption, respectively. Thus, NH3 or LOH will only be recommended when it minimises OPEX or energy consumption.

Conclusion

In conclusion, a mathematical optimisation model to optimise the technoeconomic performance of green and clean H2 networks was developed. Several case studies have been conducted to recommend an optimal H2 network based on specific objective functions. The objective functions included minimising TAC, CAPEX and OPEX and minimising energy consumption. Furthermore, each case study is separated into two scenarios, when H2 demand equals to H2 produced (Scenario 1) or H2 stored (Scenario 2). Following this, the effect of H2 demand was discussed. Besides, the outcome from respective case studies were compared with base case, minimising TAC. In Scenario 2, the overall cost and energy consumption was higher than Scenario 1. More H2 will be produced to compensate H2 loss during compression or reaction stages in the storage section. Thus, the more H2 is being produced and stored, the higher the energy consumption from technology and the overall cost to meet the demand. Next, the optimal networks from the case studies have showed some common technologies that will always be recommended. First of all, DFMEC, a type of green H2 technology, was chosen in all case studies. DFMEC required low CAPEX and OPEX and provided high yield of H2 from biomass of POME. In addition, DFMEC required significantly lower energy compared to other technologies. Interestingly, DFMEC is not limited to any production capacity, unlike water electrolysis or turquoise H2. Hence, DFMEC was selected to meet the entire H2 demand or to produce the remaning H2 when other technologies have achieved maximum production limit. Besides DFMEC, turquoise H2 was recommended due to acquiring the lowest OPEX and energy requirement. Surprisingly, turquoise H2 was recommended by model prior to DFMEC. However, H2 production from methane pyrolysis is limited within a certain range. Therefore, the remaining H2 demand was fulfilled by DFMEC, as mentioned above. In electricity generation section, both solar and biomass energies were recommended in different case studies. To summarise, solar energy required lower OPEX while biomass energy required lower CAPEX. This is because biotechnology only involves simple implementation of equipment. However, the operation of biotechnology is more intensive compared to solar energy. On top of that, biomass energy is limited to electricity production capacity, especially in gas engine and steam turbine. Thus, solar energy with higher electricity production efficiency recommended when the overall cost is required to be minimised. Subsequently, compressed H2 gas with the highest efficiency was always recommended by the proposed model. Not only that, compressed H2 gas also required low CAPEX and OPEX and low energy requirement. These make compressed H2 gas the most common H2 storage technology in order to store the maximum amount of H2 with the lowest cost. However, alternatives were considered prior to compressed H2 gas when objective function is specified to OPEX or energy requirement. Although compressed H2 gas required low OPEX and energy, LOH is the least energy intensive storage method, while NH3 required the lowest OPEX to store H2. In addition, the efficiency of storage technologies does not deviate significantly. Hence, LOH and NH3 was recommended in most cases. However, current research does not calculate exact reduction in CO2 emission after utilising green sources and technologies for H2 production. Furthermore, model developed in current research does not consider profit from selling stored H2. Therefore, future work can be directed to consider how green the H2 network is. As such, accounting for exact reduction in CO2 emission from green technologies. Besides, the profit from selling green or turquoise H2 can also be considered in the future work.