Buffering Configuration: An Essential Decision for The Biogas Supply Chain

(1)

Buffering Configuration: An Essential Decision for

The Biogas Supply Chain

Master Thesis

MSc. Supply Chain Management

Words Count: 11.477 Date of Submission: 29 June 2018

Supervisors:

Prof. Dr. J. C. (Hans) Wortmann

Dr. Martin J. Land

by

Adrian Gilrandy

(2)

List of Figures

Figure 1 - Biogas Supply Chain Network ... 11

Figure 2 - Illustration of gas storage facility ... 15

Figure 3 - Hourly plot of natural gas demand in a one-year period ... 16

Figure 4 - Comparison of natural gas consumption between peak and low season in the rural household ... 17

Figure 5 - Methodology for Quantitative Modeling ... 18

Figure 6 - Conceptual Model of Biogas Buffering System ... 19

Figure 7 - Biogas supply pattern ... 21

Figure 8 - Expected gas demand in each hour ... 21

Figure 9 - Biogas supply and demand profile ... 22

Figure 10 - Simulation algorithm ... 26

Figure 11 - Different upgrading capacity consequences to storage capacity ... 28

Figure 12 - Trade-offs in choosing upgrading capacity ... 28

Figure 13 - Profile of minimum and maximum inventory within different upgrading capacity level .. 29

Figure 14 - Situation when the buffer plays its two different roles ... 30

Figure 15 - Situation when the buffer exists only to cover a deficit supply ... 30

Figure 16 - Cost comparison by choosing different dimension of pipeline 2 (P2) ... 31

Figure 17 - Cost comparison by choosing different dimension of pipeline 1 (P1) ... 31

Figure 18 - Cost structure comparison of different upgrading capacity setting ... 32

Figure 19 - Cost structure comparison of different pipeline dimension chosen ... 33

(5)

List of Tables

Table 1 - Inlet and outlet pipeline pressure ... 23

Table 2 - Energy use to transport biogas in pipeline 1 and 2 ... 23

Table 3 - Pipeline cost ... 24

Table 4 - The best alternative of buffering configuration ... 32

Table 5 - Statistics of the buffer used in the storage facility ... 33

Table 6 - Buffering configuration without storage facility ... 34

Table 7 - Sensitivity Analysis – Different Upgrading Capacity Without Storage Facility ... 35

Table 8 - Sensitivity Analysis – Different P1 Diameter Without Storage Facility ... 35

Table 9 - Sensitivity analysis - digester capacity ... 36

Table 10 - Sensitivity analysis - pipeline length ... 37

(6)

Acknowledgment

This last year is one of the most precious moments in my life. Having an opportunity to live in the Netherlands while studying at the University of Groningen reminds me of how hard life is. Being independent and punctual is the main lesson learned I got from this incredible journey. Finally, the journey will be ended as soon as I submit this document on Nestor. I would like to give my highest gratitude to both my thesis supervisors, Prof. Hans Wortmann and Dr. Martin Land, who guide me very patiently through all the learning process. I admire their dedication as academics to help their students who always struggle with a combination of academic and life problems. I also want to thank Mr. Jan Eise Fokkema for his sincere help by giving feedback and fruitful discussions to improve the quality of my thesis. Another person who is also really helpful in the making this thesis is Mr. Gerard Martinus from Gasterra. I appreciate him for his time to give me a short course of biogas system that allows me to understand this topic in a concise period.

For my wife Zahra Aldila, I do not know how to express my gratefulness that I have you. Thank you for your patience to accompany me accomplishing one of my life goal checklists. I know that it is not easy for you at first, living in a country with a completely different culture while taking care of the children as a full-time housewife, but you nailed it. I hope that this priceless experience will bring us into a better relationship for the next chapter of our life. And for my sons Alendra and Arsakha, you might not be aware of this living experience since you are still too young at the moment. However, I am just so thankful that you are here. Your presence is my all-time mood booster.

(7)

Abstract

Purpose – The study aims to design a buffering configuration to overcome the imbalance of supply and demand in the biogas supply chain, mainly due to the hourly demand fluctuation. Buffers accumulate when the biogas supply exceeds the demand and will be used in the deficit period. The study proposes the most economical configuration by making use of the gas storage facility and the pipeline linepack to store the buffer. The pipeline linepack is a unique characteristic that a pipeline can not only transport the gas from one point to another but also act as a storage facility.

Design/methodology/approach – The study investigates a biogas supply chain that consists of a centralized digester, local biogas markets, and an upgrading facility. All stakeholders are connected via biogas grids. Simulations are conducted to study the system dynamics how the biogas flows from a centralized digester to the local consumption given that the demand is stochastic and the surplus biogas is upgraded into an upgrading facility. Different buffering configurations are compared regarding the total cost which includes the pipeline, storage, production, and energy cost.

Findings – Results from simulation reveal that there is an optimal value of the upgrading capacity level and the pipeline dimension to ensure the total cost is minimum. The optimal upgrading capacity balances the trade-off between the production and storage cost while the optimal pipeline diameter minimizes the combination of pipeline, storage, and energy cost.

Research limitations/implications – Due to the computation complexity in modeling a dynamic linepack capacity, the study assumes that the linepack capacity is static. It is estimated by the average linepack based on the average calculated differential pressure.

Practical implications – Using a different approach compared to prior studies, this study modeled the biogas supply chain as a supply-driven chain. The biogas produced flows from the digester, and the local biogas consumption has priority. Such an approach applies to a situation where the biogas producers are located in the area without any connections to the natural gas grid.

(8)

1. Introduction

Current use of biogas is low compared to its theoretical potential (Visser, 2017) due to many challenges. One of these challenges is balancing supply and demand (Börjesson & Ahlgren, 2012; Vlap et al., 2015). As one of the energy sources derived from biological feedstock (Murray et al., 2017), biogas faces a typical problem of renewable energy alternatives. The renewable energy usually is produced at places and times when or where it is not always needed (Dinalog, 2018). In order to deal with a situation that there is an imbalance of the supply and demand, buffers are needed (Hopp, 2011). Buffer stock is a type of inventory that is built up as a reserve against fluctuations both from supply and demand (Hopp, 2011; Sutopo et al., 2013).

According to its supply-driven nature (Fokkema et al., 2018), biogas product flow is not triggered by either its demand schedule or forecast, but by its production rate and supply capacity (Hull, 2005). Hull also argues that the goal of such a system is to continue the supply source running at a full operating rate and channel it to the multiple markets available. Because the volatility of the market remains unpredictable in the biogas market, buffers are accumulated when supply exceeds demand in a particular period and will be used in the opposite situation (Hengeveld et al., 2014).

Biogas supply chains cope with the problem how much biogas is needed as a buffer, where to locate it, and in what form that the gas is stored, either as raw biogas or as upgraded gas. Such a problem will be dependent on the network configuration. A biogas distribution configuration is an arrangement of the transportation mode chosen and the natural gas grid dependence. The mode of transportation specifies how biogas is transported from producers to consumers, while the grid dependence is related to whether the biogas is injected into a natural gas grid eventually. Because biogas cannot be injected directly into the grid until the quality is upgraded to the natural gas standard (Berglund & Börjesson, 2006; Börjesson & Ahlgren, 2012; Hengeveld et al., 2014), upgrading facilities are required (Brijder et al., 2014; Petersson et al., 2009). According to Hengeveld et al. (2016, 2014), a centralized digester, the facility that produces biogas, combined with a centralized upgrading facility is the most economical configuration when the biogas is transported through the pipeline.

(9)

(Hengeveld et al., 2016). Arvesen et al. (2012) argue that linepack is an utilized and under-communicated short-term energy storage option, whereas it is cost-efficient and environmentally sound since it uses the existing gas pipeline infrastructure.

Taking the pipeline as a storage medium might have a significant impact on the economic aspect and can reduce the investment cost of building a storage facility. Therefore, this study will address the following research question:

How should a buffering configuration, making use of the linepack in particular, be designed to overcome the hourly imbalance of supply and demand for a centralized biogas digester?

Besides including linepack consequences, this study uses a different approach compared to prior studies mentioned as it does not assume that all biogas produced is upgraded into green gas. The use of biogas for local consumptions is prioritized, and the surplus will be enhanced in the upgrading facility to be injected into the natural gas grid. Such an approach will make that this study is beneficial to give insights to practice, particularly in a situation when the biogas is produced in a region that has no connection to the natural gas grid. Furthermore, from the theoretical perspective, this study supplements the prior research in answering how to encounter the imbalance of biogas supply and demand due to not only seasonal but also hourly fluctuations.

(10)

2. Theoretical Background

Understanding the supply chain of a biogas network is essential to a buffering decision. This section will discuss important aspects that lead to a buffering problem in the biogas system. It starts with comprehending what biogas is and how the production works. After that, mapping the biogas supply chain network is required to analyze the possible alternatives of the distribution configuration. Furthermore, the pipeline linepack and storage facility will be discussed as the options to design the buffering configuration to cope with fluctuations of biogas supply and demand.

2.1. Biogas Production

Biogas is one of the renewable energy sources that is increasingly used in many countries (World Biogas Association, 2017). The main constituents of this gas are methane (CH4) and carbon dioxide (CO2). However, it also contains considerable quantities of undesirable content: hydrogen sulfide (H2S), ammonia (NH3) and siloxanes (Abatzoglou & Sherbrooke, 2009). Many sources could be used to produce biogas, such as animal manure, organic household waste, food residues, and dry biomass. Such substrates are processed using either the (co)digesting or the gasification method (www.greengasgrids.eu, 2018). Among those two options, (co)digestion, i.e. production of biogas through anaerobic digestion plants (AD), is the more common technology applied (Hengeveld et al., 2014; Lindkvist et al., 2017; Murray et al., 2017). Raw biogas is produced by the AD and ready to distribute to the next step of processing or directly consumed for generating power by using the AD itself.

2.2. Biogas Supply Chain Network

(11)

Figure 1 - Biogas Supply Chain Network (adopted from Berglund & Börjesson, 2006; Börjesson & Ahlgren, 2012;

Hengeveld et al., 2014)

There are several stakeholders in the biogas supply chain network. Households, farmers, factories, and municipalities are examples of the initial stakeholder that can provide the raw material of biogas (Lindkvist et al., 2017; Murray et al., 2017). Furthermore, the chain is continued to a different ramification which is affected by the purposes of biogas utilization.

2.2.1. Biogas Producers

A supply chain network of biogas starts from the place where it is produced. Murray et al. (2017) classify that there are four types of biogas producers: (1) Landfill waste, (2) manure from farmers, (3) waste-water treatment plant in the municipality, and (4) forest and agricultural residues and energy crops. As we can see from where it is originated, biogas producers are spread in multiple locations, and the supply scales differ from one to another.

2.2.2. Upgrading facilities

Upgrading of biogas is necessary to meet requirements which are demanded not only by the application of the gas (burners) but also by the gas grid which transports the gas (Bekkering et al., 2010b). The upgrading process is required due to several reasons, such as to meet the requirement of gas appliance, to increase the gas heating value, and to standardize matter (Persson et al., 2006).

(12)

2.3. Biogas distribution configurations

There are multiple options of biogas distribution configuration that exists. As already mentioned in the previous chapter, the source of biogas or substrate is dispersed in terms of location. When the substrates from multiple locations are gathered into a single digester to process, it is called a centralized configuration. Conversely, a decentralized configuration deals with a situation in which every substrate owners produce biogas by their own digesters. Bekkering et al. (2010a) explain the trade-offs between choosing these two configuration options. The centralized configuration has a relative scale advantage for the capital costs and energy use of the digester. This central location could also be coupled with the upgrading and gas grid injection facility if the biogas will be upgraded to a natural gas standard eventually. On the other hand, the decentralized configuration has an upside of costs and energy use for the transport and logistics of the substrate and digestate as the by-product of digesting process that is needed to return.

Hengeveld et al. (2014) conducted a study to investigate the impact regarding the energy use and costs between those different configurations. Assuming that all biogas produced is upgraded into green gas, a model is developed to calculate the cost per m3_{of green gas produced for each configuration.} The result shows that decentralized digesters do not give any energy advantage per m3_{green gas} produced. The production cost of centralized configuration is lower than that of decentralized configuration, in the range of 5 €ct to 13 €ct per m3_{. To conclude, merging smaller digesters into a} smaller number of larger ones for the same biomass source area manages the cost of biogas transport more efficiently (Hengeveld et al., 2016, 2014). However, Hengeveld et al. (2014) also argue that a digester with a capacity larger than 1.200 Nm3_{/h not be efficient in managing the transportation of} substrates.

Another factor that determines a biogas distribution configuration is the mode of transport used.

There are two ways of transporting biogas: (1) using cylinders carried by trucks and (2) using pipelines (Börjesson & Ahlgren, 2012). Hjort & Tamm (2012) identified several factors that influence when to choose either using pipelines, truck, or a combination of both. They argued that pipeline transport has been shown as the most efficient alternative for bulk transport over a greater distance. For local distribution, when the distance is less than 30 kilometers, the low-pressure pipeline is strongly recommended. Otherwise, the truck is the most economical option. In addition, Hengeveld et al. (2016) also argue that creating a pipeline infrastructure is vital when the biogas could be distributed for local consumption. By not implementing such system, there will be an economic loss for society. A more detail description of pipeline transport will be discussed in the next chapter.

2.4. Pipeline transportation

(13)

Ma et al., 2018; Midthun, 2007). Pipeline linepack is defined as the total quantity of gas in the pipeline that can be used without any safety problem at any given time (Lapuerta, 2003; Ma et al., 2018). Lapuerta also adds that there is a minimum safe level of linepack that is needed to be maintained in order to keep the system works. The amount of gas in the linepack (𝐿𝑝) is determined by the equation (1) that is explained in Ma et al. (2018):

𝐿𝑝 =𝑍𝑉 𝑝𝑛 2 3 (𝑝1+ 𝑝2− 𝑝1𝑝2 𝑝1+ 𝑝2 ) 𝐸𝑞. (1)

Assume that the volume of a pipe is 𝑉, the linepack is affected mainly by the 𝑝1 and 𝑝2 that represent

the inlet and outlet pressure, respectively (𝑝1 > 𝑝2). Pressure at normal temperature is denoted by 𝑝𝑛,

where 𝑍 represents the compressibility factor which is associated with the dimension of the pipeline and the given pressure.

Equation (1) shows that the size of the linepack could be flexible by setting a pressure difference between the inlet and outlet of a pipeline. Such a condition is the reason why pipeline transport could also act as a storage facility. As the fundamental principle of storage is, linepack flexibility operates like a buffer that is filled first, then it will be emptied later (Keyaerts et al., 2010). Therefore, linepack flexibility would be beneficial to accommodate the imbalance between supply and demand as a short-term gas storage (Arvesen et al., 2012). Technically, according to Keyaerts et al. (2010), linepack flexibility is available in any pipeline transport with a fluctuated demand profile. This particular property of pipeline transport provides an economic value that such mode of transport would avoid extra investment of storage facility to overcome the variation in the gas supply and demand (Keyaerts et al., 2010).

Another essential aspect of pipeline transport is that a combination of pipelines dimensions and the differential pressure of the nodal input and output limits the flow of gas in the pipeline. Neacsu et al. (2013) explain the condition using equation (2) for the gas flow rate (𝑄𝑛):

𝑄𝑛 = 𝜋 4 𝑇𝑛 𝑝𝑛 √(𝑝₁2_{− 𝑝} 22) 𝑅 𝑍𝑇 𝑑5 𝑙𝜆 𝐸𝑞. (2) From equation (2), the pipe diameter (𝑑), the length of the pipeline (𝑙), and the absolute gas temperature (𝑇) influence the gas flow rate. The other parameters needed to calculate 𝑄𝑛 are 𝑇𝑛 (standard

(14)

Lastly, another important aspect to look at is the cost structure for pipeline transport. There are two types of cost embedded in this mode of transport. They are the investment cost which are also associated with its operational cost and the cost that is related to energy used to transport the gas through the pipeline. According to Hengeveld et al. (2014), the investment cost will be dependent on the pipeline dimension and the installation difficulty related to a geographical condition. A wider diameter with a longer distance costs a higher investment. Furthermore, a difficult pipeline installation, such as when it passes through roads, will also end up with a considerable amount of investment cost. Another type of cost incurred to transport gas using pipeline is the energy cost (Damen, 2007; Hengeveld et al., 2016; Keyaerts, 2012). The literature explains that the energy cost is also influenced by pipeline dimension, particularly the pipeline diameter. A higher energy cost is needed to transport gas using a smaller diameter pipeline because the smaller pipeline requires higher pressure. Therefore, the selection decision of pipeline dimension aims to find the balance between the pipeline cost and energy cost. 2.5. Gas storage facility

Even though a pipeline can be used to store the gas, the capacity might not be sufficient, so a gas storage facility still needs to be built. There are multiple types of storage facilities, such as abandoned oil and gas reservoirs, aquifers, salt caverns, or liquified storages. Those storages are different with respect to capacity, operational cost, and the capability to inject and extract the gas (Midthun, 2007). Similar to what has been explained previously about the linepack, storages are essential due to several reasons. They can be used to avoid bottlenecks in the system in high demand periods or to utilize market possibilities. Storages also allow gas producers to produce in time periods where demand is low and to utilize the available transportation capacity. Furthermore, they can be used as seasonal storages to smooth out seasonal effects. However, this option is very costly and would be the last alternative to consider (Bekkering et al., 2013).

Midthun (2007) describes some critical terms needed to understand how the mechanism of a gas storage works. They will be illustrated in Figure 2 and explained as follows:

• Storage capacity

The maximum volume of gas in the storages facility. The physical properties of the storage limit the storages capacity.

• The volume of gas in the storage

The total volume of gas in given storage at a given time. • Cushion gas

(15)

• Working gas

The gas volume available during normal operation of the storage. This volume corresponds to the total amount of gas in the storage deducted the cushion gas.

Figure 2 - Illustration of gas storage facility – from Midthun (2007)

Up to this point, this theoretical background section already covers the biogas supply chain that focuses on how the biogas distribution is designed in a particular configuration using pipeline transportation. Then, multiple options of storage alternative are extensively discussed as the solution approach to answering the research question of this study. In the last two chapters of this section, the characteristic of biogas supply and demand will be addressed to get a full picture of the situation to design a proper buffering configuration.

2.6. Biogas supply characteristics

Biogas supply is usually modeled by an assumption that the production rate over time is constant (Bekkering et al., 2015). However, this is not always the case because biogas production is dependent on the source availability of the biogas itself. When the biogas is produced by farmers, for instance, the source availability of biogas might be affected by the seasonal difference. It means that seasonal fluctuation may occur, but such information is still limited in the existing biogas literature. In many studies, the biogas supply chain is usually modeled as a demand-driven chain, but in fact, biogas is more appropriate to be categorized as a supply-driven product (Fokkema et al., 2018).

(16)

• Alternative markets should be developed so that if demand disappoints in one, another is

available. At the same time, any bottlenecks which might interrupt access to those markets should be removed.

• Inventory can be positioned en route to the markets if doing so enhances the security of

supply or contributes to profit maximization.

• Prices should be adjusted as necessary to stimulate or discourage sales. 2.7. Biogas demand characteristics

As opposed to the biogas supply modeling, literature confirms that the biogas demand fluctuates over the period (Bekkering et al., 2015; Hengeveld et al., 2014). There are two approaches to model how the biogas demand looks like: (1) based on the end-product chosen or (2) finding the substitute commodity. From the first perspective, biogas would be transformed ultimately for multiple different purposes, such as the production of electricity, heat, and vehicle fuel (Bekkering et al., 2010a). Each end-products has its own characteristic regarding the demand pattern. Electricity and heat consumption have a seasonal fluctuation because changes in temperature and humidity, for instance, influence the demand for space heating and cooling (Gadd & Werner, 2013; www.eia.gov, 2013). In contrast, the demand for vehicle fuel is relatively stable over the period regardless the seasonal changes that occur (Hakimelahi et al., 2016). By using the first approach, biogas demand could be modeled by choosing which derived product is the target of biogas production.

Figure 3 - Hourly plot of natural gas demand in a one-year period – from Bekkering et al. (2013)

(17)

seasonal fluctuation, Bekkering et al. use Fourier analysis that results in a sine function that is presented as a black line in the graph. Furthermore, the graph also clearly shows that in the peak season, the hourly demand has higher deviation than the demand in the low season (see figure 3).

A study from Trojanowska & Lipczyńska (2014) gives additional insights into the hourly fluctuation of the natural gas demand. They investigate the consumption pattern at rural household and analyze the pattern of peak and low season separately. The demand in each hour is represented as the percentage of maximum demand during the week. They reveal that there is a different pattern between workday and weekend consumption. In figure 4, it can be concluded that the natural gas demand for the rural household has a periodic hourly, daily, weekly and seasonal fluctuation.

Figure 4 - Comparison of natural gas consumption between peak and low season in the rural household – from Trojanowska

& Lipczyńska (2014) 2.8. Implications

According to literature, the biogas demand fluctuations create an imbalance of supply and demand over the period. Many studies investigated such problem and proposed multiple alternative solutions particularly for seasonal fluctuations. Bekkering et al. (2015) argue that there are three options to overcome the biogas seasonal fluctuation: (1) making use of flexible production digesters, (2) activating non-continuous biogas production digesters in the peak season period, and (3) building a seasonal gas storage. Results from the study show that the flexible digester is the most economical solution while the seasonal gas storage is very costly. However, current literature still cannot adequately answer how to deal with the biogas supply and demand imbalance problem because, despite the seasonal fluctuation, the hourly demand variation also occurs.

(18)

3. Methodology

According to Bertrand & Fransoo (2002), this study follows a methodology from Mitroff et al. (1974) which has been proven as an appropriate and widely used methodology for quantitative modeling research (see figure 5). It starts with a problem description based on reality and the problem is conceptualized into a conceptual model that defines all the parameters and variables needed. Then, modeling step is conducted to determine the input parameters and to develop the cost function represented by a mathematical formulation called a scientific model. In the modeling step, the biogas demand pattern is modeled according to literature while the biogas supply pattern follows the real-life data. To solve the problem, a simulation model is constructed using Microsoft Excel software and a solving algorithm is designed. Biogas supply and demand fluctuations are simulated on an hourly basis in a one-year period. One simulation represents one scenario of one possible buffering configuration. The simulation gives the value of variables explained in section 3.1. The simulation also calculates the total cost to implement a particular buffering configuration. All possible configurations are compared to find the configuration which has the least total cost. Insights from the simulation are discussed in section 4 complement theories to improve the decision-making process in the biogas supply chain.

Figure 5 - Methodology for Quantitative Modeling, adopted from Mitroff et al. (1974) and Bertrand & Fransoo (2002)

3.1. Conceptualization 3.1.1. Problem Description

A problem investigated in the study is related to a case based on literature, and a real-life problem in a biogas supply chain. The scope of the case starts with the biogas produced by the digester as the input and ends with delivering the biogas to the customer. The biogas produced is needed to satisfy local consumption which is fluctuating at an hourly basis. The surplus of biogas will be transferred to the upgrading facility to be injected into the natural gas grid eventually. Such an approach to model a

Conceptual Model

Result Interpretation Problem

(19)

biogas flow is called a one-way system (Martinus, 2018). Biogas is transported through a biogas grid that connects a centralized digester with the local consumption and the upgrading facility.

The fluctuation of biogas supply and demand require the existence of a biogas buffer. The buffer can be located either in a particular storage facility or in the biogas pipeline itself. Because the upgrading capacity influences the buffer size, a buffering configuration can be translated into three decisions: (1) the upgrading capacity level, (2) the dimension of the pipeline used, and (3) the biogas storage capacity needed. However, due to the stochasticity factor of supply and demand, those decision variables can only be calculated after simulating the biogas flow in each hourly period. A conceptual model is shown in figure 6 to illustrate the system dynamics.

3.1.2. Conceptual Model

Figure 6 - Conceptual Model of Biogas Buffering System

Figure 6 gives an interpretation of the biogas decision flow in each period. The conceptual model is a framework that grounds the simulation model. The biogas produced by the digester and local consumption in each period t, denoted by 𝑆𝐵𝐺(𝑡) and 𝐷𝐵𝐺(𝑡) respectively, are the inputs of the model.

Biogas flow decisions must be made each hour period. The computation procedure to calculate all of the variables in the model will be discussed in section 3.3. The set of parameters and variables will be explained as follows:

Set

S Set of scenarios investigated in the study (combination of picking different pipeline dimension for pipeline 1 and 2

I Set of pipelines (1=pipeline 1; 2=pipeline 2)

J Set of capacity alternatives for one upgrading facility (j = 1..n)

T Time period in hour unit (t = 1..8000; assuming that 1 year = 8.000 hours) Parameter

(20)

DBG(t) Local consumption demand for biogas at hour t, for all t ∈ T [Nm3_] Dia_Pi Diameter of the pipeline i, for all i ∈ I [mm]

Length_Pi Length of the pipeline i, for all i ∈ I [m]

Cap LP_Pi Capacity of linepack pipeline i, for all i ∈ I [Nm3_] Variable

Outflow_dig to Pi(t) Biogas outflow from digester to pipeline i at hour t, for all i ∈ I and for all t ∈ T [Nm3_]

Outflow_dig to Z(t) Biogas outflow from digester to storage facility Z at hour t, for all t ∈ T [Nm3_] Outflow_Z to Pi(t) Biogas outflow [Nm3_]

Outflow_Pi(t) Biogas outflow from pipeline i at hour t, for all i ∈ I and for all t ∈ T [Nm3_] Inv_Z(t) Inventory level at storage facility Z at hour t, for all t ∈ T [Nm3_]

Inv_Pi(t) Inventory level in pipeline i at hour t, for all i ∈ I and for all t ∈ T [Nm3_] Cap_UPj Capacity upgrading facility j, for all j ∈ J [Nm3_/h]

3.1.3. Defining scenarios

There are 25 different scenarios to be simulated which represent 25 buffering configurations. The number of scenarios is generated by every combination of 5 different diameter dimension type of pipeline 1 and pipeline 2: 110 mm, 160 mm, 200 mm, 250 mm, and 315 mm. These diameter options are commonly used for local distribution networks operated by a low-pressure setting. The length of pipeline 1 and pipeline 2 is set to 5.000 m and 500 m, respectively for all scenarios. Different alternatives of the pipeline length will be tested in a sensitivity analysis that is explained in section 3.3.3. A detailed description of each scenario is shown in Appendix 1. The simulation model will solve each scenario, and the cost parameter measures the performance of each configuration.

3.2. Modeling

Based on the scenario developed, the decision problem is modeled using mathematical expressions. To begin with, modeling the biogas supply and demand is needed.

3.2.1. Modeling the supply and demand pattern

3.2.1.1. Supply pattern

(21)

capacity because the real biogas production rate of the observed digester is a very small-scale digester with a capacity of 100 Nm3_{/h. This type of digester is not appropriate to be a model of a centralized} digester in the study. Therefore, the percentage drop based on the primary data is used to model the biogas supply pattern when the digester size is scaled up to certain criteria explained in the section 3.2.1.3.

Figure 7 - Biogas supply pattern

3.2.1.2. Demand pattern

The demand pattern is modeled based on the study of Bekkering et al. (2013, 2015). Given that the purpose of biogas is to substitute the natural gas, data of one of the real cases presented by Bekkering et al. (2013) is used in this paper. The data specifies a natural gas demand of a particular region in the Netherland which consumes 6.713.454 Nm3_{of gas per year, which implies an average} consumption of 766 Nm3_{/h. The maximum and minimum gas demand per hour equals to 1.417 Nm}3_/h and 41 Nm3_{/h, respectively. To model the consumption profiles, equation (3) is used and the resulting} expected demand in each hour 𝑡 (𝐸(𝐷𝑡)) is presented in the figure 8.

Figure 8 - Expected gas demand in each hour

𝐸(𝐷𝑡) = 𝑐1+ 𝑐2 . sin (

2𝜋𝑡

8000) 𝐸𝑞. (3)

𝑐1 is the average hourly gas demand (Nm3/h) while 𝑐2 represents the amplitude. The constant 8000

represents number of hours in a one-year period. To calculate 𝑐2, it is needed to estimate the SSF

(22)

the value of 𝑐2 is presented in equation (4). For the case chosen in this study, SSF = 1.2 is used which

avoids a focus on seasonal fluctuations and fits the case characteristics.

𝑐2= 𝑐1.

𝑆𝑆𝐹 − 1

𝑆𝑆𝐹 + 1 𝐸𝑞. (4)

As the gas demand fluctuated over the hours, the gas demand in each hour is modeled by equation (5) while 𝜀𝑡 denotes the error in hour 𝑡. 𝜀𝑡 is generated by a random number that is assumed normally

distributed with the mean = 0 and standard deviation = 0.25 which fits the case. The statistics parameter chosen models the demand pattern as close as possible to the real case that accounts the maximum demand gas = 1.417 Nm3_{/h and reaches the lowest point at 41 Nm}3_{/h. Furthermore, how the demand} fluctuates over the period based on those properties is presented in Figure 9.

𝐷𝑡 = 𝐸(𝐷𝑡) + 𝜀𝑡. 𝐸(𝐷𝑡) 𝐸𝑞. (5)

3.2.1.3. Supply vs demand profile

In this study, the biogas hourly supply profile is determined using two criteria: (1) the digester capacity > average biogas demand per hour and (2) the digester capacity < 1.200 Nm3_{/h. According to} Hengeveld et al. (2014), the size of 1.200 Nm3_{/h is considered as the largest digester capacity for a} single centralized digester to ensure the efficiency of transporting manure and biomass from decentralized locations of farmers, for instance. Therefore, the capacity of 1.000 Nm3_{is chosen as a} base level since it is higher than the average hourly gas consumption (766 Nm3_{/h) and lower than 1.200} Nm3_{/h. Different values of the digester capacity will be tested in a sensitivity analysis that is explained} further in section 3.3.3. The biogas supply from the digester and local consumption in each hour are plotted in figure 9 as the primary input of the study.

Figure 9 - Biogas supply and demand profile

3.2.2. Calculating parameters and variables

(23)

before being able to do the simulation in the study. However, both the linepack capacity and energy use calculation are dependent on the pressure difference between inlet and outlet of the pipe as the fundamental principle to transport gas through the pipeline (Chisholm, 2003). So, determining pressure for each different pipeline dimension is required.

3.2.2.1. Pressure calculation

According to Borraz-Sánchez & Haugland (2013), Hengeveld et al. (2016), Keyaerts (2012), and Neacsu et al. (2013), the pressure calculation is based initially on equation (2) explained in the theoretical background. The detailed calculation and procedure can be seen in appendix 2. Based on the calculation, the pressure needed for each scenario is presented in table 1 where the length of P1=5.000 m and P2=500 m.

Table 1 - Inlet and outlet pipeline pressure

3.2.2.2. Energy use and linepack calculation

Energy is needed to create the differential pressure between the inlet and outlet node of the pipe to transport gas from one point to another through a pipeline. The energy use calculation formula is based on a paper from Damen (2007), Hengeveld et al. (2016), and Keyaerts (2012). Furthermore, to calculate the linepack flexibility, equation (1) is used (Ma et al., 2018). Table 2 shows the calculation result of energy use and the linepack capacity for each different pipeline dimension chosen where the length of P1=5.000 m and P2=500 m.

Table 2 - Energy use to transport biogas in pipeline 1 and 2

3.2.2.3. Set of alternatives for the upgrading capacity

The capacity of the upgrading facility cannot be less than the average surplus biogas over the one-year period. According to the case investigated in the study, average biogas supply is 947 Nm3_/h and the average local consumption is 764 Nm3_{/h. The average surplus would be 183 Nm}3_{/h. Therefore,} the minimum capacity of the upgrading facility is approximately 200 Nm3_{/h. Every additional capacity} of 50 Nm3_{/h is considered as an option to choose the optimal upgrading capacity. Then, the maximum}

(24)

upgrading capacity studied is 1.000 Nm3_{/h because it makes no sense to set the upgrading capacity that} exceeds the digester capacity.

3.2.3. Developing the cost function

For calculating the cost, the assumption used is as follows: depreciation is 12 years for each investment cost and operational cost is 5% of total investment cost. Then a year period consists of 8.000 hours. The total cost of a buffering configuration chosen consists of energy cost, pipeline cost, production cost, and storage cost calculated as follows:

• Energy cost

Energy cost is calculated based on the energy used [kWh] that creates the differential pressure in the pipeline. The energy used is then multiplied by the electricity price that is assumed €0.14/kWh (Bekkering et al., 2010b; Hengeveld et al., 2016).

• Pipeline cost

Transportation cost is calculated based on cost parameter presented in table 3. Based on the same assumption used by Hengeveld et al. (2014), the pipeline installation is divided into two complexity classes: easy (60%) and difficult (40%).

Table 3 - Pipeline cost

• Production cost/upgrading cost

Following Bekkering et al. (2010), the production/upgrading cost [€/hour] will be calculated based on equation (6) where 𝑄 represents the upgrading capacity. The estimation of production cost is modeled as a function of 𝑄 that is derived from real investment costs of upgrading facility in several cases observed.

𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡 = (81.532 𝑥 𝑄

0.4551

12 𝑥 8000 ) 𝑥1.05 𝐸𝑞. (6) • Storage cost

According to Bekkering et al. (2013), storage cost [€/hour] can be calculated using equation (7) where 𝑄 denotes the capacity of storage facility. The storage cost is also estimated based on multiple real-life cases. It is clearly shown from equation (7) that a different scale of storage

(25)

facility influences the incremental cost for additional storage capacity. If 𝑄 > 10.000 Nm3_, the storage facility requires a very high pressure to operate and it needs an extra compressor.

𝑆𝑡𝑜𝑟𝑎𝑔𝑒 𝑐𝑜𝑠𝑡 = { ( 222 𝑥 𝑄 12 𝑥 8000) 𝑥1.05 ; 𝑄 ≤ 10.000 (2.240.000 + 18,85 𝑥 𝑄 12 𝑥 8000 ) 𝑥1.05 ; 10.000 < 𝑄 ≤ 300.000 𝐸𝑞. (7) 3.3. Model Solving

We use simulation to model the hourly flow of gas in the system and compare multiple scenarios to determine which buffer configuration leads to the lowest storage, transportation and production cost. 3.3.1. Inventory policy

The simulation will compute the gas flow decision for each hour. The decision is affected by multiple conditions. A set of procedures called an inventory policy must be designed to control the gas flow from one point to another in each hour. The policy organizes the biogas flow that the local consumption has priority.

Therefore, the policy creates a parameter named minimum inventory level which is determined by a set of procedure explained in section 3.3.2. A buffer in the biogas supply chain plays two different roles to overcome the imbalance between supply and demand. The first role is to hold the gas when the gas produced is greater than the gas consumed. The second role is to deliver the gas to the customer when the production rate is less than the consumption rate. The minimum inventory level is a certain threshold for the actual inventory level that must be maintained in order to perform the second role mentioned. In the context of the study, the minimum inventory level aims to guarantee that the local consumption is satisfied 100% at any point in time.

(26)

3.3.2. Simulation steps

The simulation is designed using Microsoft Excel software. Figure 10 presents the designed algorithm used in the simulation. The algorithm is divided into three steps. It starts with initializing all possible scenario and calculating the set of alternatives for the upgrading capacity (cap_UPj). Then the minimum inventory level is initially set as 0. Step 2 in the algorithm calculates the solution for each designed scenario. For every hour in a one-year period, all the biogas decision flow is calculated using the inventory policy algorithm (see appendix 1). Then, an iteration started to find the minimum inventory level (min_inv_Z) until 100% service level is guaranteed. Different alternatives to upgrading capacity level are also compared regarding the total cost. An optimal upgrading capacity level is selected when it gives the least total cost in a scenario. In the last step, the best buffering configuration is determined by the scenario which results in the lowest total cost amongst the set of designed scenarios.

Figure 10 - Simulation algorithm

Furthermore, according to Robinson (2003), a simulation which deals with a random number generation must perform multiple replications to ensure the model performance with sufficient accuracy. At least three to five replications must be performed based on the rule of thumb approach (Robinson, 2003). Three replications are conducted in the study as the minimum requirement, and the

Algorithm - Simulation Procedure

Step 1 Defining scenario and upgrading capacity

Initialize: scenarios ∀ s ∈ S; cap_UPj ∀ j ∈ J

Set min_inv_Z = 0

Step 2 Determining resulting flows, storage needs, and total cost Forall s ∈ S do

Forall j ∈ J do

Forall t ∈ T do

Forall i ∈ I

Compute: outflow_dig to Pi(t); outflow_dig to Z(t); inflow_Z(t); outflow_Z to Pi(t);

inflow Pi(t); outflow Pi(t); inv_Z(t); inv_Pi(t) → using algorithm inventory policy

End-do Repeat

min_inv_Z = min_inv_Z +1

Until ∑t∈T outflow_P1(t) = ∑t∈T DBG(t) → to guarantee 100% service level

Compute Total_costs

Set min_inv_Z = 0

End-do

Total_costs = Min[Total_costj ∀ j ∈ J] → to select the optimal upgrading capacity

End-do

Step 3 Finding the best solution

Select the best scenario for which Total_costs = Min[Total_costs ∀ s ∈ S]

(27)

replication procedure will continue if results from each replication differ regarding the solution of decision variables.

3.3.3. Sensitivity Analysis

A sensitivity analysis is executed in order to test how sensitive the solution is to the changes in the parameters. It is expected that the designed model in the study is suitable not only for a specific case but also for general cases in the biogas supply chain. The parameters varied are as follows:

• Storage facility size

The approach used in designing the inventory policy assumes that the service level must be set to 100%. The existence of storage facility will increase the service level since the linepack capacity is limited to hold all the accumulated buffers. However, if 100% service level is not mandatory to achieve, then a storage facility might not be necessary to build. The sensitivity analysis will examine how the buffering configuration would be without building the storage facility regarding the total cost and service level.

• Digester capacity

This parameter is the primary factor in deciding the set of alternatives for the upgrading capacity. In the initial case investigated in the study, the capacity of the centralized digester is set at 1.000 Nm3_{/h as the base level. This parameter will determine how much biogas is left} over after fulfilling the local consumption in a particular period. Changing the digester capacity will directly affect the system dynamics. The sensitivity analysis investigates how the optimal buffering configuration would react to a smaller or larger digester capacity. Firstly, the digester capacity is set to 900 Nm3_{/h which is less than the base level but still higher than the average} local consumption. Furthermore, the simulation sets a capacity of 1.200 Nm3_{/h which is} suggested to be the largest size of a centralized digester.

• Demand fluctuation

According to the case investigated in the study, the biogas demand fluctuates +/- 25% in each hour. The sensitivity analysis wants to study how significant the changes in hourly fluctuation influence the optimal buffering configuration. Two sets of randomly generated biogas demand are constructed under an assumption that the fluctuation has a mean = 0 and the standard deviation of 15% and 35%, respectively. Making a sensitivity analysis in the demand fluctuation is needed to set the boundary of demand fluctuations to ensure that the optimal buffering configuration is not adjusted.

• Distance/length of the pipeline

(28)

4. Result and Discussion

4.1. The trade-off between storage size and upgrading capacity

The storage facility size and the upgrading capacity is the output of the simulation. Choosing different capacities of the upgrading facility will affect the required storage facility. Figure 11 shows that initially increasing the upgrading capacity level will decrease the size of a storage facility substantially for every scenario investigated. However, the decreasing rate is slowing down at higher capacity when it reaches a point that any extra capacity does not reduce the size of the storage at all. Results presented in figure 11 also imply that a larger pipeline dimension contributes to decreasing the required storage capacity.

Figure 11 - Different upgrading capacity consequences to storage capacity

The relationship between the storage size and the upgrading capacity is associated with the utilization of the upgrading facility itself (see figure 12). A significant amount of biogas is built up and stuck for a long time period in the storage when the upgrading facility utilization is set nearly 100%. The trade-off between storage needed and upgrading capacity is then presented in figure 12 regarding the total cost. Figure 12 shows the result of simulating scenario 1 (both diameters of pipeline 1 and pipeline 2 are 110 mm).

(29)

As can be seen from the graph, the total cost first decreases strongly when the utilization drops from around 90% (upgrading capacity = 200 Nm3_{/h) to 60% (upgrading capacity = 300 Nm}3_{/h). The} cost continues to decrease until hitting its lowest point when the upgrading capacity is 400 Nm3_{/h with} 46% utilization. After that level, the cost steadily increase due to the higher investment needed for a larger upgrading facility while it has no significant benefit in reducing the storage cost. The result of the other scenarios also follows the same pattern and conclude that upgrading 400 Nm3_/h_{gives the least} cost for each configuration.

4.2. The role of the storage facility

According to the inventory policy designed, when there is a supply surplus in an hour, the surplus gas fulfills the linepack until it is fully capacitated. Then, the next priority is to satisfy the minimum inventory level of the storage facility. If those conditions are performed, then the remaining surplus gas will flow into the upgrading facility. However, the remaining surplus will be built up in the storage facility when the upgrading facility already operates at its maximum capacity. In this case, the storage facility plays its first role as a transit point for biogas waiting to be upgraded. The maximum inventory level within a one year period reflects the required storage size must be obtained to avoid the surplus biogas is wasted.

In the opposite situation, when there is a supply deficit in an hour, all the buffers both in the storage facility and the linepack will be used to meet the biogas demand. Therefore, the minimum inventory level parameter will guarantee that all the biogas demand must be 100% satisfied which is the second role of the storage facility. Figure 13 illustrates how the value of minimum and maximum inventory level changes according to a different option of the upgrading capacity. If there is a gap between the maximum and minimum inventory level, then the buffer plays both roles (see figure 13). Otherwise, the storage capacity is minimal because it is needed only to cover the deficit production period.

(30)

The storage facility plays both of its different roles when the upgrading facility has a capacity of up to 850 Nm3_{/h. Figure 13 indicates that a capacity higher than 850 Nm}3_{/h will lead to a condition} that there is no more buffer waiting to be processed in the upgrading facility.

Figure 14 illustrates how the storage facility deals with its multiple roles when the upgrading capacity is set to 400 Nm3_{/h. The red line in the graph represents the minimum inventory level needed} to maintain the service level for local consumption. When the actual inventory level (the blue line) falls below the red line, it means that the stored gas is used due to insufficient supply in the particular hour to cover the demand. Conversely, when the blue line is plotted above the red line, it exemplifies the situation when the biogas supply already satisfies the demand at that point in time.

Figure 14 - Situation when the buffer plays its two different roles

Furthermore, figure 15 shows the actual inventory level when the upgrading capacity is 1.000 Nm3_{/h. The actual inventory level never hits above the minimum inventory parameter. As the capacity} is extremely high, compared to the excess biogas supply over time, there will be no chance that the buffer can accumulate waiting to be upgraded.

(31)

4.3. Opt for the best solution

The simulation result of all 25 scenarios studied concludes that the optimal upgrading capacity is 400 Nm3_{/h. After finding the best option for upgrading facility, all configurations of different pipeline} dimension combination are compared regarding the total cost. According to the simulation result presented in figure 16, choosing a pipeline with a diameter of 200 mm is the best option to flow the gas from digester to the upgrading facility (pipeline 2) when the pipeline 1 diameter is 110 mm and the upgrading capacity is 400 Nm3_{/h. Section 4.4 will explain the occurrence of this optimal value for both} pipeline dimension and upgrading capacity that minimizes the total cost in more detail.

Figure 16 - Cost comparison by choosing different dimension of pipeline 2 (P2)

Then, determined that a 200-mm pipe is installed for the pipeline 2, figure 17 also confirms that the same pipeline dimension gives the lowest total cost compared to the other choice. Therefore, based on the simulation result, the best solution for designing a buffering configuration given a situation described in the problem description section is presented in table 4.

(32)

Table 4 - The best alternative of buffering configuration

4.4. The cost structure of different buffering configurations

The simulation result discloses that the best configuration costs € 30.91 per hour. Figure 18 and 19 exhibit the cost components divided into four different types of cost: production cost, storage cost, energy cost and pipeline cost. Figure 18 presents how a different upgrading capacity affects the cost structure while figure 19 analyzes the consequence when changing the option of pipeline dimensions.

Figure 18 - Cost structure comparison of different upgrading capacity setting

There is no difference regarding the energy and pipeline cost by altering the upgrading capacity since using the same pipeline dimension. However, storage cost decreases when the upgrading capacity increases. Given the set of example for capacity option varies from 300 – 500 Nm3_{/h (see figure 18),} the storage cost drops significantly changing the capacity from 300 to 400 Nm3_{/h due to considerable} storage size reduced. In contrast, the cost of upgrading the gas goes up as the need for higher investment to build larger upgrading facility. The total cost escalates when choosing the upgrading capacity greater than 400 Nm3_{/h since the increasing production cost outweighs the decrease of storage cost.}

Diameter P1 200 mm Diameter P2 200 mm Upgrading Capacity 400 Nm3/h Min INV_Z 1,023 Nm3 Max INV_Z 1,732 Nm3 Z Storage Capacity 1,819 Nm3 Utilization Upgrading Facility 46%

Pipeline Cost € 9.12 per hour

Energy Cost € 3.75 per hour

Storage Cost € 4.42 per hour

Production Cost € 13.63 per hour

Total Cost € 30.91 per hour

(33)

Figure 19 - Cost structure comparison of different pipeline dimension chosen

An analysis of cost dynamics is relevant to decide which different pipeline dimension is the most efficient option, since the storage, energy, and pipeline cost fluctuate. Both storage cost and energy cost are reduced when using a larger pipeline. The declining storage cost is because the larger pipeline has more flexibility in the linepack, so it will shrink the size of storage facility accordingly which directly decreases the storage cost. However, using the larger pipeline means a higher investment needed. Choosing a 200-mm pipeline is known as the best solution to the problem because of a significant drop in energy cost by changing pipeline diameter from 110 mm to 200 mm. Subsequently, a slight cost increase is expected because the pipeline cost proliferation offsets the storage and energy cost reduction.

4.5. The needs of a storage facility for implementation

To compare the cost implications between alternative buffering configuration possible, the required storage size is assumed to equal the maximum inventory parameter observed in the simulation increased by a percentage of cushion gas needed. Also, the minimum inventory level parameter is set to ensure 100% service level for local consumption. This approach is used only for calculation simplification. In the real case, it might not be the best option to calculate the storage size in such a way. Demand for the local consumption enormously fluctuates over the hours, so uncertainty is inevitable. To deal with this, attaining 100% service level might be too costly. Determining a minimum service level to achieve is a way more sensible in this case.

Table 5 - Statistics of the buffer used in the storage facility

Storage Cap. 1,819 Nm3 UP cap. 400 Nm3/h Max Inventory 1,732 Nm3 Ø P1 200 mm Min Inventory 1,023 Nm3 Ø P2 200 mm

Mean Stdev

Investment Cost Storage Facility = € 403,724 Storage for surplus

(Nm3)

Buffer for deficit (Nm3) 121.82

125.30 162.41

(34)

Based on the simulation result of the best configuration, the upgrading facility is set to 400 Nm3_/h and pipe diameter for both connections is 200 mm. According to the best-case scenario, the statistics of the buffer used in the storage facility are collected (see table 5). Then, mean (μ) of “storage-for-surplus” is the average accumulated surplus above the minimum inventory level parameter while the mean of “buffer-for-deficit” is the average of the buffer used in a shortage demand period. Results in table 5 show that the storage size is an overkill design. For only an average of 192.01 Nm3_buffer consumption, it is needed to guarantee a minimum inventory level of 1.023 Nm3_{/h. Consistently, the} maximum inventory parameter is designed to be 709 Nm3_{more than the minimum inventory parameter} while the average accumulated surplus to store is only 121.82 Nm3_.

However, the service level is not only dependent on the storage facility because the buffer is also located in the pipeline linepack. Therefore, a simulation is conducted to calculate what is the minimum service level obtained when the storage facility is excluded. Table 6 shows that without having a storage gas facility, the linepack capacity of a 200 mm pipeline already maintains a very high service level and fill rate. In this case, not building the storage facility would be an option since it will reduce the investment cost substantially.

Table 6 - Buffering configuration without storage facility

Compared to the optimal solution while maintaining 100% service level, a solution without a storage facility contributes to lower the utilization of the upgrading facility because of the less amount biogas upgraded. Some amount of biogas which is equivalent to a volume of 84.921 Nm3 _{biogas is} wasted due to the capacity limit of upgrading capacity during the one year period. Excluding the storage facility will change the cost structure of a buffering configuration that will be studied in section 4.6.1 in more detail.

Diameter P1 200 mm 200 mm

Diameter P2 200 mm 200 mm

Upgrading Capacity 400 Nm3/h 400 Nm3/h

Z Storage Capacity 1,819 Nm3 0 Nm3

Utilization Upgrading Facility 46% 43%

Biogas produced 7,580,524 Nm3 7,580,524 Nm3 Local consumption demand 6,113,731 Nm3 6,113,731 Nm3 Local consumption satisfied 6,113,731 Nm3 6,108,559 Nm3 Biogas wasted Nm- 3 84,921 Nm3 α-Service level for local consumption 100% 99.24%

Fill rate local consumption 100% 99.91%

Percentage of biogas wasted 0% 1.12%

(35)

4.6. Sensitivity analysis

Sensitivity analysis is done to analyze how the buffering decision reacts to the change of parameters. Insights from the sensitivity analysis are very important to generalize the biogas buffering problems investigated in the study. Crucial parameters that are suspected to have a significant influence to determine decision variables are evaluated.

4.6.1. Storage facility size

Section 4.5 reveals that building a storage facility might not be a necessity if the service level could be adjusted. In this section, we investigate how a decision of not building a storage facility would affect the solution of the optimal buffering configuration.

Table 7 - Sensitivity Analysis – Different Upgrading Capacity Without Storage Facility

Table 7 compares the simulation result when the upgrading capacity is altered. When the storage facility is not considered, a decision of setting the upgrading capacity level only influence the amount of wasted biogas and the production cost. Increasing the upgrading capacity will decrease the amount of wasted biogas while it increases the investment of building a larger upgrading capacity. Regarding the total cost, choosing the upgrading capacity as small as possible will lead to the least cost.

Table 8 - Sensitivity Analysis – Different P1 Diameter Without Storage Facility

Results presented in table 8 show that the linepack capacity influences the service level for the local consumption. A larger pipeline which can carry more buffer in the pipeline increases the service

Diameter P1 200 mm 200 mm 200 mm Diameter P2 200 mm 200 mm 200 mm Upgrading Capacity 200 Nm3/h 300 Nm3/h 400 Nm3/h

Z Storage Capacity 0 Nm3 0 Nm3 0 Nm3

Utilization Upgrading Facility 87% 58% 43%

Biogas wasted 478,717 Nm3 217,910 Nm3 84,921 Nm3

α-Service level for local consumption 99.24% 99.24% 99.24%

Fill rate local consumption 99.91% 99.91% 99.91%

Percentage of biogas wasted 6.32% 2.87% 1.12%

Pipeline Cost € 9.12 per hour € 9.12 per hour € 9.12 per hour

Energy Cost € 3.74 per hour € 3.74 per hour € 3.74 per hour

Storage Cost € 0.00 per hour € 0.00 per hour € 0.00 per hour

Production Cost € 9.94per hour € 11.96per hour € 13.63per hour

Total Cost € 22.80 per hour € 24.81 per hour € 26.49 per hour

Comparison Solutions for Different Upgrading Capacity

(Excluding The Storage Facility)

Diameter P1 160mm 200mm 250mm

Diameter P2 200 mm 200 mm 200 mm Upgrading Capacity 400 Nm3/h 400 Nm3/h 400 Nm3/h

Z Storage Capacity 0 Nm3 0 Nm3 0 Nm3

Biogas wasted 85,226 Nm3 84,921 Nm3 84,846 Nm3

α-Service level for local consumption 97.53% 99.24% 99.88%

Fill rate local consumption 99.72% 99.91% 99.98%

Percentage of biogas wasted 1.12% 1.12% 1.12%

Pipeline Cost € 7.75per hour € 9.12per hour € 11.87per hour

Energy Cost € 6.81per hour € 3.74per hour € 1.72per hour

Production Cost € 13.63 per hour € 13.63 per hour € 13.63 per hour

Comparison Solutions for Different P1 Diameter

(36)

level. As opposed to the change of upgrading capacity, choosing a larger pipeline does not necessarily increase the total cost because there is still a trade-off between pipeline cost and energy cost. It is concluded that even the storage cost is no longer considered, P1 diameter of 200 mm results in the most economical option. Increasing the pipeline diameter from 200 mm to 250 mm increases the total cost because a reduction of energy cost cannot compensate the increase of pipeline cost

4.6.2. Digester capacity

From a simulation we made, different level of digester capacity contributes to change the optimal buffering configuration (see Table 9). When the supply capacity is scaled down to 900 Nm3_{/h, the} storage size facility needed is inflated because the excess capacity is limited, so the deficit gas situation occurs more frequently. Such a condition enforces that more buffers are needed to overcome the deficit condition (local consumption > biogas supply in a certain period) that is reflected by a very high of minimum inventory level (Min_INV_Z). As opposed to that, increasing the digester capacity will reduce the storage size facility significantly because practically the biogas supply in almost every period is greater than the local consumption. Such a condition implies that there is not so much buffer needed to satisfy all the demand.

Table 9 - Sensitivity analysis - digester capacity

Changing the digester capacity also leads to a different optimal upgrading capacity level. It is so apparent that a larger digester will produce more surplus biogas that is needed to be upgraded eventually. Table 9 shows that upgrading capacity level of 300 Nm3_{/h is sufficient when the biogas} supply rate is 900 Nm3_{/h, while an upgrading capacity of 600 Nm}3_{/h is needed to balance the digester} at 1.200 Nm3_{/h capacity level. Moreover, each different digester size also has an implication to the} optimal utilization of the upgrading capacity. Low utilization of upgrading facility for less biogas supply is because most of the biogas produced is concentrated to be stored as a buffer for the local consumption. Meanwhile, as the surplus biogas increases substantially due to a larger digester, the utilization of the upgrading facility will increase accordingly.

Lastly, the digester capacity influences the change of the optimal pipeline dimension. The optimal P1 diameter shifts from 200 mm to 315 mm when the digester scale is decreased from 1.000 Nm3_{/h to 900 Nm}3_{/h. This phenomenon could be explained by a reason that a larger pipeline diameter}

Digester Capacity 900 Nm3/h 1,000 Nm3/h 1,200 Nm3/h Diameter P1 315 mm 200 mm 200 mm Diameter P2 200 mm 200 mm 200 mm Upgrading Capacity 300 Nm3/h 400 Nm3/h 600 Nm3/h Min INV_Z 5,600 Nm3 1,023 Nm3 50 Nm3 Max INV_Z 6,268 Nm3 _1,732 _Nm3 ₇₅₄ _Nm3 Z Storage Capacity 6,581 Nm3 1,819 Nm3 792 Nm3

Pipeline Cost € 13.99 per hour € 9.12 per hour € 9.12 per hour

Energy Cost € 0.64 per hour € 3.75 per hour € 3.89 per hour

Production Cost € 11.96 per hour € 13.63 per hour € 16.39 per hour

Buffering Configuration: An Essential Decision for The Biogas Supply Chain