Harvesting challenges for green biorefineries
A scenario analysis on harvesting patterns in Green Biorefinery systemsChristiaan Hurulean
Master’s Thesis
Harvesting challenges for green biorefineries
A scenario analysis on harvesting patterns in Green Biorefinery systemsChristiaan Hurulean
Master’s Thesis
Technology and Operation Management University of Groningen
Faculty of Economics and Business Supervisor: Dr. M.J. Land Co-Assessor: Dr. Ir. J.C. Wortmann
Acknowledgement
Abstract
Purpose: The purpose of this paper is to investigate the influence of a Green Biorefinery conversion process on the process of harvesting process. These types of systems are characterized by the rapid expiry rate of grass and the high variable growth rate. Green Biorefinery systems can play an important role in avoiding the energy-for-food discussion that is commonly used to argue against solely utilising land for energy purposes.
Methodology: A simulation study was performed on a green biorefinery system which uses grass to feed the conversion process. An extensive number of scenarios were created by combining multiple variables that influence the system on multiple dimensions. Beside modelling different grass availabilities by means of growth curves and area sizes, different process capacities and policies were modelled for both the harvesting process and the conversion process. This resulted in modelled harvesting patterns for all the scenarios. The analysis of this patterns provided insight in how the harvesting process is influenced by the system configurations.
Findings: This study shows that the high variable grass growth and the conversion process constraints have major impact on the harvesting process. It was found that a conversion process negatively influences the harvesting utilisation if a conventional continuous harvesting policy is applied. The results showed that this can be encountered by adapting an alternative harvesting strategy. Furthermore, the results show that economies of scale can be gained in terms of operational hours required for harvesting if the size of the system increases.
Implications: The findings indicate that a GBR system is able to partially cope with decrease of harvesting utilisation by adopting a bulk harvesting strategy. It was found that harvesting grass in large quantities once every few hours instead of continuous increases the efficiency of harvesting without negatively influencing the conversion process utilisation and lost production. Alternatively, low harvesting utilisation rates can be partially prevented by performing both harvesting and transportation tasks with the same equipment.
TABLE OF CONTENT
1 INTRODUCTION ... 8
2 THEORETICAL BACKGROUND ... 9
2.1 BIOREFINERIES ... 9
2.2 HARVESTING OF BIOMASS ... 10
2.3 RELATIONSHIP BETWEEN HARVESTING PATTERN AND GBR ... 11
2.4 MANAGING HARVESTING AND GBR OPERATIONS ... 12
2.5 REVIEW OF PREVIOUS RESEARCH ... 12
3 METHODOLOGY ... 13
3.1 THE MODELLED SYSTEM ... 13
3.2 SIMULATION MODEL ... 15
3.3 MODEL INPUTS ... 16
3.3.1 Grass availability class ... 16
3.3.2 Conversion dimensions ... 17
3.4 MODEL OUTPUTS ... 18
3.5 WORKING OF THE MODEL ... 19
3.6 ASSUMPTIONS AND PARAMETERS ... 21
3.6.1 Assumptions ... 21
3.6.2 Parameters ... 22
3.7 EXPERIMENTAL DESIGN... 23
4 RESULTS ... 24
4.1 DIFFERENT HARVEST CAPACITIES... 25
4.2 CONVERSION PROCESS CAPACITIES ... 27
4.3 GRASS GROWTH CURVES... 30
4.4 DIFFERENT HARVESTING POLICY ... 32
4.5 AREA SIZES ... 33
4.6 NO PRODUCTION LOSSES ... 37
5 DISCUSSION ... 38
5.1 LIMITATIONS ... 38
5.2 IMPLICATIONS OF MAIN FINDINGS ... 39
6 CONCLUSION ... 41
REFERENCES ... 43
APPENDIX A:PSEUDO CODES... 46
Appendix A1: Determine number of hectares to harvest, hourly policy ... 46
Appendix A2: determine which hectare to harvest... 48
Appendix A3: Determine number of hectares to harvest, every 3 hours policy ... 49
APPENDIX B:SCENARIOS ... 50
8
1 Introduction
In the past decade, the total number of biogas plants in Europe has almost tripled. Approximately 70% of this growth was realized in the agricultural sector (European Biogas Association, 2017). One of the arguments that is used to argue against agricultural biogas systems is the energy-for-food argument (Ignaciuk, Vöhringer, Ruijs, & van Ierland, 2006; Johansson & Azar, 2007). One might argue that land that is used for the purpose of energy cannot be used to provide food. A green bio refinery (GBR) can be created to avoid this problem. A GBR is a system which processes fresh biomass to create a range of products, which include food, chemicals, materials and feed beside energy (de Jong & Jungmeier, 2015). The conversion process of such a refinery result in new logistic challenges, which are the subject of this study.
The logistic challenges can partially be linked to the operational harvesting decisions. Creating a GBR by means of implementing a conversion process will impact the entire system, but especially the harvesting pattern. These harvesting challenges are results of multiple constraints that are added to the system by implementing a pre-treatment method. One aspect which is likely to significantly influence the system, is the limited amount of time in which the biomass must be processed. For the conversion process to be beneficial, fresh biomass must be processed within the first four hours due the rapid decrease of nutritive value after this period (Ir. M. Bouwer, 2019). As a result of this constraint, both the optimal moment of harvesting and the amount to harvest are impacted by the capacity of the conversion process.
9 inefficient logistics. This study aims to contribute to existing literature by focussing on the harvesting pattern of GBR system and how this processed is influenced by the added constraints. By computing this research, empirical evidence can be provided to support operational decision making to design the harvesting process to be more beneficial. Which in turn, will increase the likelihood for broader adoption, and a contribution to the energy transition.
Many studies address operational aspects and theories in a broad range of settings. However, no excising literature address the harvesting challenges of GBR systems, and especially how to coop with the unique type of variability. This study aims to contribute to this literature gap by generating insight into the relationship between the harvesting pattern of grass and the capacity of the conversion process of a GBR. From an operational expenditure point of view, it is desirable to harvest as much grass as possible at once, while on the other side the capital expenditures will significantly increase when the pre-treatment has to increase capacity to cope with these large amounts of fresh biomass.
In the upcoming section of this paper a theoretical background of this research topic will be provided by linking the process of harvesting to green biorefineries. Additional, prior research will be reviewed, and the gap which is addressed by this research will be described. In the subsequent section, an overview of the research design is provided, and specifically the way the modelling approach is used to reach the objectives of this research is described. Section four of this research will provide the findings of the simulation model used regarding the relation between the harvesting pattern and the conversion process capacity. The findings section is followed by the discussion and conclusion sections where the contributions of this research for both academic literature and operational managers will be discussed.
2 Theoretical background
This section will first provide background information about biorefineries and the harvesting process. This is followed by an explanation of how these two aspects are interconnected in a GBR system. This section will end with an overview of current literature in the field of logistical challenges of biorefineries, and how this study aims to contribute to this literature.
2.1 Biorefineries
10 marketable products (Corona et al., 2018), which makes it possible to use biomasses to its full potential. IEA Bioenergy Task 42 (2011) developed a classification system of biorefineries. In this system, biorefineries are classified based on four main features, which enables the refineries to be identified, classified and described based on their core differences. These main features are platforms, feedstock, product and conversion process (de Jong & Jungmeier, 2015).
According to de Jong & Jungmeier (2015), the platform characteristic of a biorefinery classifies it based on the intermediates or final products that enables different biorefinery systems and their processes to be connected. The feedstock characteristic classifies biorefineries based on the type of input used. Whole crop biorefineries process grains such as wheat and spelt (Cherubini et al., 2009). A second type of biorefineries are so called lignocellulosic biorefineries. These refineries use biomasses that have low moisture levels by nature (Ree et al., 2014). Thirdly, there are forest-based biorefineries which use biomasses like woodchips (Cherubini et al., 2009). The last class to which a biorefinery can be assigned to, based on its feedstock used, is that of a green biorefinery. This so called green biorefinery uses natural wet biomass such as grasses, and is subject of this study (B. Kamm, Kamm, Gruber, & Kromus, 2006; Kamm, 2012).
The next characteristic, i.e. product, classifies the biorefinery based on the produced products, the major groups are energy (e.g. biofuels, bioethanol), and other products (e.g. chemicals, nutrients and materials). Generally, biorefineries fractionate the biomass into a solid and a liquid part. The solid part can be used in multiple ways which include; (1) feed for livestock, (2) biogas production and (3) solid fuel and (4) as structural material if entirely dried (Corona et al., 2018). The liquid fraction consists, beside moisture, out of multiple valuable components such as proteins, vitamins and chemical (Corona et al., 2018). According to Corona et al. (2018), most often the targeted components of the liquid fraction are the proteins, due the high concentration thereof.
The last classification feature of biorefineries is that of the type of conversion process used. This process converses the raw biomass into a range of products. This can be done by means of biochemical, thermochemical, chemical or mechanical processes (de Jong & Jungmeier, 2015).
2.2 Harvesting of biomass
11 First of all, the growth of grass and therefore the availability of the used biomass is not consistent and depends upon multiple factors such as soil texture, precipitation and temperature (Graß, Heuser, Stülpnagel, Piepho, & Wachendorf, 2013). Although climate conditions are highly uncertain by nature, there are some trends. In the Netherlands, temperatures drop and precipitation increases during the winter period, and there is less rain fall and increased temperatures in the summer (IPCC, 2007). The multiple conditions that are of influence on the growth rate of the biomass result in a yearly growth curve which changes overtime. It is found that grass grows rapidly during the spring season with a peak of just over 90 kg dry matter per hectare per day and decreases until almost 0 at the end of November which holds on until March (Remmelink, Philipsen, Stienezen-, Tjoonk, & Kuiper, 2015).
Beside the unpredictable nature of the crop growth rate, the amount of crop harvested is also influenced by the amount of time used for harvesting. Generally, crops are harvested between May and October (Graß et al., 2013), which means that the demand of 12 months must be supplied within just 6 months.
Generally, it is not possible to harvest grass during relatively wet periods due the drying period required and the chances of rotting. However, this is not the case if the grass is used for a biorefinery system, since the conversion process requires a high level of moisture to extract as many of the valuable components as possible (Ir. M. Bouwer, 2019).
The system under investigation utilises grass that is seeded for crop rotation. In the agricultural sector crop rotation policies are adopted to increase the soil quality in the form of physical, chemical and biological properties (Ma, 2016) and decrease the possibility for crop deceases and pests (Karlen, Varvel, Bullock, & Cruse, 1994). Crop rotation is the practice of growing different types of crops on a specific field during different periods (Ouda, Zohry, & Noreldin, 2018). The time between different crops depends on the type of crop. Crop rotation is a general acknowledged effective strategy within the agricultural sector, which aims at maintaining the soil quality and increase of yield by growing different crop families in succeeding years (Ma, 2016).
2.3 Relationship between harvesting pattern and GBR
12 If the latter option is chosen, the utilization rate of the conversion process will be significantly small since it will only be utilized five times a year. Consequently, installing a conversion process requires to change the arrival pattern of the biomass. Besides requiring an evenly distributed supply, it is logistically challenging to cope with the variable grass growth during the harvesting period and the probability of dry periods, in which grass cannot be harvested. The harvest period is a time period in which a specific biomass can be harvested. As mentioned in the previous subsection, grass must have a high moisture content to be efficiently processed by the conversion process. This requirement influences the moments at which grass can be harvested. This is not only dependent on the weather but also on the time of harvesting. All types of biomass tend to have a high moisture content during the morning.
2.4 Managing harvesting and GBR operations
Multiple combinations between existing operational management theories and settings have been researched in the past. And many of these researchers implemented variability, such as variable demand (Chao, 1992; Vargas, 2009; Wagner & Whitin, 1958) or variable processing times (Hou, 2007; Palanivel & Uthayakumar, 2017) and even researched queueing theories for inventories and process with perishable goods (Syntetos, 2014). However, none of the existing literature addresses the unique type of variability which occurs in GBR systems.
On the other hand, there is existing literature which addresses the logistical challenges for GBR systems (Igathinathane et al., 2014; Rudi et al., 2017). However, this literature is limited to economic, environmental and social challenges of the transportation. No current research addresses the challenges of matching the supply quantities and arrivals with the conversion process properties.
2.5 Review of previous research
13 also implemented the environmental impact of the logistics into the model. Both studies (Sokhansanj et al., 2006; Tatsiopoulos & Tolis, 2003) designed an extensive numerical model to analyse the costs regarding the transportation of biomass. Prior quantitative studies that studied the biomass logistics focussed solely on the transportation logistic costs regarding biomass transportation without taking the optimal amount of biomass in relation to the GBR capacity into account (Chiueh, Lee, Syu, & Lo, 2012). Considering only the transportation costs of biomass lacks operational managerial insights to balance the transportation of biomass and the installed conversion process capacity.
In conclusion, the capacity of a GBR’s conversion process has a major impact on the operational aspect of harvesting biomass. The goal of this study is to analyse the relation between the conversion process capacity and the harvesting pattern. Logistic challenges arise due the requirement to process harvested biomass within the first four hours in combination with the variability in availability of grass. In this study, we investigate the relation between the harvesting pattern and the capacity of the conversion process. This study contributes to the energy-for-food discussion by providing operational insights in balancing the harvesting process with the conversion process to use biomasses to its full potential.
3 Methodology
This study analyses how the harvesting pattern is influenced by the capacity of a conversion process. To analyse the relation under different circumstances and gain a deeper understanding of the relation, multiple realistic scenarios will be analysed. This section will provide an overview of the methodology used during this study. First an overview of the system under investigation will be provided. This will be followed by a description of the research approach used. The two sections that follow provide insight in the required inputs and generated outputs respectively. This is followed by a section which describes the workings of the model used. The sixth subsection will provide an overview of the different assumptions made and the parameters used during this study. This section will end with an overview of the different experimental setups analysed in this study.
3.1 The modelled system
14 fixed central position. Therefore, no transportation times of this asset had to be integrated in the model.
Different challenges arise due the implantation of a conversion process in a conventional agricultural system. A major challenge that arises it the influence of this conversion process on the harvesting process due to the constraints that are added to the system, of which the limited amount of time to process the harvested fresh biomass is expected to have the most impact.
This study is focussed on the harvesting process and the way this process is influenced by an implementation of a GBR’s conversion process. Additionally, this study looks into the buffer behaviour in front of the dry digester under different settings. The dotted orange line in figure 1 shows where this part of the system is located within GBR system.
Figure 1: System overview
15 insight in the quantities to harvest during each hour, the outcomes can be projected on the different strategies described above. This was done to create a more general applicable model. Operational insights and impacts of the results on the two different harvesting architectures described in the previous paragraph will be discussed in section 5.
3.2 Simulation model
The aim of this study is to gain insight in the way the conversion process affects the conventional harvesting pattern for the collection of grass as input for a green biorefinery system. This will be computed by analysing when and how much grass has to be harvested for multiple conversion process capacities. Multiple scenarios will be designed by altering scenario dimensions. Additionally, to the different system parameters, this research will distinguish two different key classes, which are: (1) Grass availability class (2) Process capacities class. Both of these classes consist out of multiple dimensions which will be explained in more detail during this section. Next to the alterations made based on these input dimensions, two types of harvesting policies were tested, of which the exact workings will be described in section 3.5. The first policy is designed to analyse the impact of a conversion process on a continuous harvesting policy. The second policy tries to fill the buffer of the conversion process every few hours to increase the utilisation rate of the harvesting equipment during one hour of harvesting, we will refer to this policy as the “bulk” policy.
16 To compute the analysis and process the extensive amount of data, an Excel based quantitative model is used. The model (figure 2) is used to analyse the behaviour of the harvesting pattern in different scenarios based on alterations of the different classes and policies. A detailed description of the different aspects and corresponding values of figure 2 will be provided during the upcoming sections.
The model will provide insight in when certain number of hectares must be harvested to use the full potential of the biomass. Each simulation simulates the process on an hourly basis, for the duration of a year. No warm up period was needed because it is assumed that the buffer in front of the conversion process is empty at the start of each harvesting period. Besides, the available grass on each hectare does not differ largely because the available biomass is seeded a few months before. By creating different scenarios in the simulation study, it is possible to analyse the behaviour of the relation over different settings. A more detailed overview of the workings of the model and the specific scenarios used will be discussed in sections 3.5 and 3.7 respectively. During this study multiple assumptions will be made regarding the simulation model. These assumptions are explained in section 3.6.
3.3 Model inputs
As mentioned in the previous subsection, the scenarios that are used during this simulation are alterations based on two scenario classes. These two factors will be referred to as grass availability class and process capacities class. This section will provide an understanding of the different dimensions that are part of these classes.
3.3.1 Grass availability class
In this study two scenario dimensions influence the amount of grass that is located on the fields at certain moment in time. These elements are: (1) area size and (2) grass growth rate.
Area size
17 Grass growth rate
To determine the amount of grass available on a certain moment in time the amount of land which is available for grass production is required. Beside the surface of grass covered land, the growth rate of grass is required. The growth rate of grass varies over the year due to weather conditions. During the spring season, the weather conditions are optimal which results in a rapid growth of grass, but this growth rate weakens overtime.
Figure 3: grass growth curves
Figure 3 provides a visual overview of the grass growth rates used during this study. It can be seen that the different years show different patterns. The growth rates of 2014 and 2015 show relatively small differences as are those of 2016 and 2017. These patterns are based on measurements made by the Wageningen University, which were conducted in close range to the location of the university. The available data provides a growth rate expressed in kilograms of dry matter per hectare added per hour. If this is combined with the area size, the total amount of grass available for harvesting could be calculated. Since the green biorefinery under investigation consumes fresh biomass which consists out both the dry matter and moist, the data was recalculated to express ‘wet matter’ instead of solely dry matter.
3.3.2 Conversion dimensions
Conversion process capacity
18 processed within the first four hours after harvesting, the processing rate of the conversion process determines the maximum amount of grass (tons) that can be processed within a time period. This will therefore influence the quantity of grass that can be harvested within that same time period.
Harvesting process capacity
The hourly capacity of the harvesting equipment limits the maximum number of hectares to harvest. Based on the basic understanding a queueing theory, a relatively small harvesting capacity in relation to the conversion process capacity will likely result in a low utilisation rate of the conversion process capacity. On the other hand, if the harvesting process is significantly larger than the conversion process capacity, this will either result in production losses, or a low utilisation rate of the harvesting equipment. If this is the case, the most optimal number of hectares to harvest is likely to be significantly lower than the installed harvesting capacity. I.e. the harvesting can be done within just a part of an hour. By altering this dimension, the impact of this variable could be analysed to gain a deeper understanding of the relation between these two processes within the studied system.
3.4 Model outputs
This subsection provides a description of the results that are generated by the model. The goal of the study is to determine the effect of the conversion process capacity on the optimal harvesting pattern. The data rough data generated by the model represent hourly quantities and correspond with the following model aspects:
• Harvesting schedule • Buffer behaviour • Lost production
19 Since all the grass must be processed within the first four hours after production. It can already be concluded that under no circumstances the buffer size should exceed the process capacity of four hours. This factor is also modelled in the simulation, and the model holds track of when this period is exceeded by a specific amount of grass.
Beside the insights in which combination between harvesting and conversion process capacity are able the most amount of biomass, and how the harvesting pattern is behaving over time, additional operational insights will be provided in this report to provide support for operational decision making. The following additional outputs will be discussed in the findings section of this paper:
• Number of breaks. i.e. hours in which no harvesting activities take place, that lay within two hours in which grass is harvested.
• Conversion process utilisation • Harvesting utilisation
• Operational harvesting hours
3.5 Working of the model
This section provides a description of how the developed model works, i.e. how it converts the model inputs into the previous described outputs. A more detailed description of the model, in the form of pseudo codes can be found in appendix A.
For every hour during the year, the simulation model must make a harvesting decision. This is done based on the multiple aspects: (1) time, (2) day, (3) number of hectares that can be harvested, (4) number of hectares that must be harvested, (5) amount of fresh biomass in conversion process buffer.
20 continuous scenarios, the model tries to optimize the amount of grass harvest on an hourly basis. As an example, if there is 1ton grass left in the buffer, and the installed conversion process capacity is 8 tons per hour. The system wants to harvest as close to 7 tons of grass as possible. In the case that the three hectares with the largest amount of grass available have a grass availability of 3 tons each. The system will harvest 2 hectares instead of 3, due the absolute difference with 7 tons required to fully utilise the conversion process.
The bulk harvesting policy analysed during this study, is that of harvesting once every few hours depending on the expiry rate of the grass. This policy creates a system in which larger quantities are harvested at once, but where the number of hours that are used for harvesting is decreased. The number of hectares to harvest is based on the amount of grass the conversion process is able to process in 3 hours and the buffer of t-1. If we use the same example used for the explanation of the continuous policy, the model harvest as close to 23 tons as possible (3 times 8 tons per hour minus the remaining buffer of 1 ton). A more detailed description of the way the model calculates the number of hectares to harvest during a specific hour for both policies in the form of a pseudo code can be found in appendix A.
In all modelled systems and policies, the harvesting capacity limits the maximal number of hectares to harvest in the system. However, it is possible that the optimal number of hectares to harvest is a small fraction of the installed harvesting capacity. This results in an inefficient harvesting pattern since this requires the system to harvest only during a part of an hour. E.g. 20 minutes if the harvesting hour utilisation is 33%. This factor will be referred to as harvesting hour utilisation throughout this paper. Figure 4 provides visualizes example of this behaviour, in which the vertical lines represent the end of one hour and the beginning of the next hour. The blocks in between the lines represent harvesting actions. As can be seen, only parts of an hour are used. In practice, this inefficient pattern could be partially encountered by using the harvesting equipment for the transportation of grass to the conversion process facility, instead of assigning additional assets to this task.
21 After the required number of hectares is calculated by the model, it reduces the available grass on those fields to the minimal remaining quantity after harvesting. The model provides an hourly overview of the buffer size and the amount of grass and hectares harvested. Based on these model outputs several additional outputs and performance measurements, motioned previously, were calculated.
Within the simulation model, process properties must be taken into account. These properties influence the decision-making process in the quantitative model:
• All grass must be harvested at the end of the harvesting period.
• Grass that cannot be processed by the conversion process within the first four hours after harvesting will not be processed and is transferred to a storage facility (ir. M. Brouwer, 2019). This will result in production loss.
• Harvesting always starts at the hectare with the largest amount of grass availability.
3.6 Assumptions and parameters
This subsection discusses the different assumptions and parameters used during this study. 3.6.1 Assumptions
The following assumptions were made during this study: The harvesting period ends at 31 October
• Both the harvesting and the conversion process state are stable, and no failures or variability is modelled for their hourly processing capacity.
• Harvesting is always done in integer number of hectares. • Moisture content of grass is fixed.
• It is assumed that no harvesting actions take place on Sundays due religious reasons • It is assumed that the grass density versus grass length ratio is equal for all hectares • Grass is always mowed to the minimal length after harvesting
22 3.6.2 Parameters
Every quantitative study includes a set of parameters. This subsection provides an overview (table 1) and description of the different parameters used during this study. The description of each parameter includes, but is not limited to, how the parameter relates to the system, and how the specific value is chosen.
Table 1: parameters
Grass length and moisture content
Beside the actual availability of grass, operational aspects should be taken into account. As one can imagine, there is a certain range in which grass should be harvested. i.e. there is a minimal amount of grass that must remain on the land, and if grass exceeds a certain length it will become too hard to harvest even with professional equipment. The University of Wageningen, provides guidelines regarding the range in which grass should be harvested to reduce the impact on the land and grass quality. This range is provided in amounts of kilograms dry matter available per hectare. Since the process which is under investigation consumes ‘wet matter’ instead of dry matter, a basic formula is used to calculate the available wet matter based on the provided figures.
Available harvesting time
As a consequence of the growth rate of grass and weather conditions, the agricultural sector only harvest grass during a specific period. Generally, this period is between April and October. From an operational perspective, this means that all the available grass must be collected in this period, and that it therefore cannot be evenly distributed over the entire year. This period restricts the maximal utilisation of the conversion process to 50% on a yearly basis. During this study, the day on which the harvesting period ends was fixed, although in reality this might differ a bit depending on the weather conditions.
Parameters Value
Moisture content 60%
Minimal dry matter 3066 kg
Maximal dry matter 6018 kg
Dry matter after harvesting 1754 kg
Last day of harvesting season 31 October
Start time 06:00
End time 13:00
Days not to harvest Sunday
Conversion process time available per day 10 hours
23 A second aspect that influences the available operational time is that of moisture content of the grass. In a GBR system, a high moisture content is desired to extract the maximum number of valuable components. The moisture content of fresh grass is at its highest before 1 pm and therefore should be harvested before this moment in time. Both this element and the assumption that the earliest moment to harvest is 6 am results in a situation in which grass must be harvested between 6 am and 1 pm.
Another element that must be modelled to provide a realistic overview, is to take the available time within the harvesting season into account. One of the aspects that is taken into account which influences the available time is the fact that many businesses in the agricultural sector do not operate on Sunday’s due religious beliefs. This aspect is implemented in the model to create a more realistic picture of the real-life situation.
Conversion availability
Beside the hourly processing rate of the conversion process, the availability of the conversion process influences the maximum amount of grass that can be processed daily. As one might imagine, it is not operational possible to operate the conversion process at a 24/7 basis for local GBR systems. During this simulation it was assumed that the conversion process could operate 10 hours each day if deemed necessary.
3.7 Experimental design
In this analysis, multiple sets of scenarios will be compared, each of which consists out of multiple scenarios. Each set is a unique combination between the grass growth curve of specific years, and the number of available hectares. Within these sets multiple scenarios are designed based on alterations within the two scenario classes described in section 3.2 and 3.3.
Table 2 provides an example of a scenario set. It can be seen that the different scenarios within the first simulations are modelled with the same area size (200), grass growth curve (2014) and policy (Continuous). The differences between the scenarios are based on the available harvesting capacity, and conversion process capacity.
Table 2: overview of simulation 1
Growth rate from year: 2014
Number of ha 200
Policy Continuous
Harvest capacity
In ha per hour 4 8 12 14
Conversion capacity 6 1 2 3 4
in tons per hour 8 5 6 7 8
12 9 10 11 12
16 13 14 15 16
24 By comparing and analysing the different scenarios in this simulation it is possible to analyse the relation between the harvest and conversion capacity. After comparing the different scenarios in the first set, scenarios will be compared with 23 other scenario sets. These scenario sets are simulated with different combinations between grass growth curves, area sizes and harvesting policies. By doing this the impact of specific area sizes and grass growth patterns can be analysed. Table 3 provides an overview of the 24 different sets of scenarios. A complete overview of the different scenario sets and corresponding scenarios can be found in appendix B.
Table 3: Simulation overview
4 Results
In this section the influence of the different variables on the harvesting pattern will be discussed. First the influence of different harvesting capacities will be discussed for a given conversion capacity, which is followed by a discussion regarding the differences in system behaviour for different conversion process capacities. The base case that is used during the comparison between different scenarios is scenario 6 this scenario is selected because the pilot project on which this study is based operates a conversion process with a capacity of 8 tons per hour and wishes to utilize 200 hectares of grass land for the GBR system. Additionally, a harvesting capacity of 8 hectares a week is a realistic variable, that an average agricultural organisation is able to reach. From a more operational perspective, 8 hectares an hour can be realised with harvesting equipment with a working width of 8 meters, and a driving speed of 10 km/h. Although this approach is chosen to write a compact report, additional scenarios will be cited if they provide critical insight. For instance, if all scenarios with 8 tons per hour conversion process capacity, equal grass growth curve and area size show no impact of different harvesting capacities, but this variable has an impact on the behaviour for scenarios with larger installed conversion process capacities. The performance measurements of all computed scenarios can be found in appendix C.
After the comparison within one simulation set, the differences between sets will be discussed. That is, the influences of increased area sizes, different grass growth curves and harvesting policy.
200 ha 400 ha 800 ha 200 ha 400 ha 800 ha
2014 Set 1 Set 5 Set 9 Set 13 Set 17 Set 21
2015 Set 2 Set 6 Set 10 Set 14 Set 18 Set 22
2016 Set 3 Set 7 Set 11 Set 15 Set 19 Set 23
2017 Set 4 Set 8 Set 12 Set 16 Set 20 Set 24
25
4.1 Different harvest capacities
This subsection discusses the influence of different harvesting capacities on the system’s behaviour. Different key performance indicators for the different simulations compared in this section are summarized in table 4.
Table 4: Performance measurements for different harvesting capacities
The left half of table 4 provides an overview of the different variable settings for the different scenarios. It can be seen that in this section the experimental variable of hourly harvesting capacity was isolated to analyse the impact of alterations of this variable on the system. The four different scenarios used to describe the impact of different harvesting capacities have a conversion process capacity of 8 tons per hour, grass growth curve of 2014 and an area size of 200 hectares. These scenarios are used for the description, because they represent a realistic system, as described previously.
Figure 5 provides an overview of how often a specific integer number of hectares was harvested in each scenario. The different scenarios are displayed on the x-axis, in which the different columns indicate the number of times a specific number of hectares is harvested. The figure shows that this distribution is equal for all scenarios, and that the optimal hourly harvesting quantity is always within the range of the smallest harvesting capacity (4 ha/hour). Figure 5 also provide a visual explanation of the decreasing harvesting hour utilisation.
Figure 5: overview of how often a specific number of hectares is harvested over the entire year
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Compulsory to harvest (times) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 5 4 8 2014 Continuous 200 100,0% 0 885 45,6% 900 35,8% 6 8 8 2014 Continuous 200 100,0% 0 885 22,8% 900 35,8% 7 12 8 2014 Continuous 200 100,0% 0 885 15,2% 900 35,8% 8 14 8 2014 Continuous 200 100,0% 0 885 13,0% 900 35,8%
Scenario summary Performance measurements
0 100 200 300 400 500 600 4 8 12 14 Nu m be r o f t im es harvestng capacity
Times a number of hectares is harvested
26 We can conclude that a conversion process of 8 tons per hour, results in a situation in which it is only required to harvest between 15 and 30 minutes each hour, with the smallest harvesting capacity installed. In practice, it is likely that it is most efficient to transport the grass to the conversion process with the same asset used for harvesting. i.e. 30 minutes used for harvesting and x amount of time used for transportation.
Based on these scenarios we would conclude that the system is not limited by any of the modelled harvesting capacities. However, if we increase the conversion process capacity of the system and look at the influence of different harvesting capacities in systems with a conversion process capacity of 16 tons per hour minor differences can be found (table 5).
Table 5: Performance measurement overview of different harvesting capacities with a 16-ton conversion process capacity
Although all the scenarios are able to process 100% of the harvested grass, a harvesting capacity of 4 hectares per hour results in more operational hours. However, the operational hours are more utilised in comparison to the other scenarios. Figure 6 shows the times a number of hectares is harvested in a specific scenario. It can be seen that scenario 13 usually operates at its maximal capacity.
Figure 6: Number of times a number of hectares is harvested with conversion process capacity 16
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Compulsory to harvest (times) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 13 4 16 2014 Continuous 200 100,0% 0 435 99,3% 437 18,0% 14 8 16 2014 Continuous 200 100,0% 0 398 54,9% 450 18,1% 15 12 16 2014 Continuous 200 100,0% 0 398 36,6% 540 18,1% 16 14 16 2014 Continuous 200 100,0% 0 398 31,4% 540 18,1%
Scenario summary Performance measurements
0 50 100 150 200 250 300 350 400 450 13 14 15 16 N um be r o f t im es Scenario
Times a number of hectares is harvested
27 Based on table 4 and 5, we are able to conclude that the harvesting capacity does not have a major influence on the system, and that the system will only be bound by this capacity if the capacity becomes relatively small in comparison to the conversion process capacity.
4.2 Conversion process capacities
In this subsection the influence of the conversion process capacity on the system will be described and analysed. At first this is done for a single year (2014) with an area size of 200 hectares. This is done to isolate the specific conversion process capacity variables, and therefore its specific impact. This subsection compares the results of scenarios 2, 6, 10 and 14 respectively. Table 6 provides an overview of several performance measurements of the different scenarios.
Table 6: performance measurements for different conversion process capacities
First of all, we can point out that percentage of grass processed does not differ between the different scenarios shown in table 6. However, there are major differences noticeable in in the other performance measurements. The first of these performance measurements we address is that of the number of breaks during harvesting. Since the ‘demand’ of the conversion process is in tons of grass, but harvesting is done in integer number of hectares and the grass yield of hectares is highly variable, the harvested amount always differs from the exact optimum and switches over time as a result of the variable grass growth. Although the system is allowed to harvest based on availability, time and day, it is possible that the optimal decision for that hour is to harvest nothing. These situations occur when the buffer in front of the conversion process contains a large amount of grass relatively to the conversion process capacity. If this phenomenon occurs in between two hours were grass is harvested, a so called ‘break’ occurs. As one might imagine, it is probably more desirable to operate continuously during a day instead of scattered from an operational point of view. Figure 7 visualizes the number of breaks occurring over the whole period for the different scenarios. It shows that the decrease in this so called ‘breaks’ is nonlinear, and the number of breaks decreases for the first increases in conversion process capacities but increases between the capacities 12 and 16 tons per hour.
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Breaks Compulsory to harvest (times) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 2 8 6 2014 Continuous 200 100,0% 20 0 1059 15,8% 1173 47,7% 6 8 8 2014 Continuous 200 100,0% 16 0 885 22,8% 900 35,8% 10 8 12 2014 Continuous 200 100,0% 9 0 524 40,5% 578 23,8% 14 8 16 2014 Continuous 200 100,0% 10 0 398 54,9% 450 18,1%
28
Figure 7: Number of breaks over the whole period
As expected, it can be seen that an increase in conversion process capacity leads to a decrease in utilisation of the same process. The utilisation rate of the conversion process is based on the ratio between the amount it would be able to process during the period and what it actually processed. This calculation took into account that the process would only be operational for 10 hours each day, so that processing at full load for 10 hours a day during the entire harvesting season would result in a utilisation rate of 100%. The utilisation rate for each of the 4 scenarios is displayed in figure 8. The figure shows that the utilisation rates differ between 48% and 18% for scenario 2 and 14 respectively. In hourly perspective these percentages represent 973 (scenario 2) and 370 (scenario 14) operational hours during the year.
Figure 8: Conversion process utilisation
It was found that the operational hours regarding harvesting decreases when the conversion process capacity increases. This is logical, as an increased conversion process capacity results in a situation in which the optimal number of hectares to harvest on a specific moment increases as well which results in less operational hours. More interesting is the insight that this effect is not linear. Figure 9 shows that the first increase in capacity only results in a relatively small decrease.
0 5 10 15 20 25 6 8 12 16 N um be r o f t im es
Conversion process capacity
Breaks 0% 10% 20% 30% 40% 50% 60% 6 8 12 16 U til is at io n ra te
Conversion process capacity
29 However, a second increase (from 8 to 12 tons per hour) rapidly decreases the total amount of hours spend on harvesting.
Figure 9: Harvesting hours spend for scenarios with
different conversion process capacities Figure 10: The percentage of time that is utilised on average during an hour of harvesting
If we look at the harvesting behaviour at a more detailed level, we can see that the beside a decrease in operational hours, the hours are more optimized for increased conversion process capacities. Figure 10 displays the average part of an hour consumed during an hour at which the model decided to harvest. E.g. if 2 hectares are harvested in a 4 hectare per hour harvest capacity scenario, the consumed time is 0,5 hour. It can be seen that this increase in ‘harvest utilisation’ is not linear to the increase of conversion process capacity. And that the utilisation increases with a higher rate in comparison to the conversion process capacity.
Altering the conversion process capacity shows a difference in harvest behaviour. Figure 11 visualizes the harvesting pattern, were each ‘dot’ represents a harvesting hour of harvesting. On the y-axis the number of hectares is expressed, so that the ‘dots’ indicates the number of hectares harvested on specific moments in time.
Figure 11: Harvesting pattern for different conversion process capacities scenarios
0 200 400 600 800 1000 1200 6 8 12 16 H ou rs
Conversion process capacity Hours spend harvest
0% 10% 20% 30% 40% 50% 60% 6 8 12 16 U til is at io n
Conversion process capacity Harvesting hour utilisation
30 The figure shows that the optimal harvesting patterns differ for the different scenarios. And that in the case of scenario 2 and 6, with conversion process capacities of 6 and 8 respectively, the optimal quantity is always 1 or 2 hectares. There is only one noticeable spike for scenario 2 were the system is not capable of harvesting and process all grass within the model constraints. Were the optimal number of hectares increase when the capacity increases further. However, we see that none of the systems operate their harvesting capacity to their full potential. This indicates that from an operational perspective, a harvesting capacity relatively small to that of the conversion process is sufficient for operating the system on an hourly optimized basis.
4.3 Grass growth curves
This subsection provides insights in the way the system is influenced by different grass growth curves of different years. For this section scenario 6, 22, 38 and 54 were compared. All these scenarios have an area size of 200 hectares, 8 hectares per hour harvest capacity and 8 tons per hour conversion process capacity. By comparing these scenarios, the impact of different grass growth curves is isolated. Table 7 provides an overview of different performances measurements of the 4 scenarios mentioned previously.
Table 7: Performance measurements for different grass growth curves
Based on the overview in table 7 we can conclude that there are slight differences noticeable between the different growth curves that occurred during different years. Although there are some similarities noticeable between different grass growth curves, overall, they differ quite a lot, as explained previously. We can see that although the growth curves differ, the system was able to process 100% of the harvested grass for all scenarios. However, we see differences in utilisation rates for both the ‘conversion utilisation’ and the ‘harvesting hour utilisation’.
Figure 13 and 14 provide visualized overviews of the differences in utilisation rates for the 4 analysed scenarios. In absolute values, of the average grass processed by the conversion process for different grass growth pattern varies between 2,59 and 3,09 tons per hour for 2015 and 2017 respectively. In the case of the average harvesting quantity for the different scenarios, the number of hectares harvest varies between 1,73 and 1,94 for 2015 and 2017 respectively. These figures show
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Breaks Compulsory to harvest (times) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 6 8 8 2014 Continuous 200 100,0% 16 0 885 22,8% 900 35,8% 22 8 8 2015 Continuous 200 100,0% 27 0 818 21,7% 837 42,3% 38 8 8 2016 Continuous 200 100,0% 21 0 895 24,2% 899 36,8% 54 8 8 2017 Continuous 200 100,0% 17 0 939 24,3% 941 38,6%
31 that the grass growth curves have limited influence on the average optimal number of hectares to harvest, but have more impact on the utilisation and buffer of the conversion process.
Figure 12: Conversion process utilisation Figure 13: Harvesting hour utilisation
Figure 15 shows the total number of times a specific number of hectares is harvested over the entire period. Although the ratios differ between the scenarios, it can be seen that all the scenarios only harvest 1 or 2 hectares each time.
If we look with more detail at the differences between amounts of hectares harvested, a noticeable difference could be identified. Despite that the total harvested amount of grass and the total number of harvesting actions differing slightly between scenario 6 and 38, the ratio between harvesting either 1 or 2 hectares differs enormously. For scenario 6, approximately 55% of the times only 1 hectare is harvested, were this percentage is just 8% for scenario 38.
Figure 14: Times specific numbers of hectares are harvested for different grass growth curves
29% 30% 31% 32% 33% 34% 35% 36% 37% 38% 39% 40% 6 22 38 54 U til is at io n Scenario Conversion process utilisation
0% 5% 10% 15% 20% 25% 30% 35% 40% 6 22 38 54 U til is at io n Scenario Harvesting hour utilisation
0 200 400 600 800 1000 2014 2015 2016 2017 N um be r o f t im es
Grass growth year
Times a number of hectare is harvested
32
The differences in growth curves impacts the amount of grass available, it indicates that the steepness of the growth curves impacts the optimal number of hectares to harvest. This is quite logical, a rapid grass growth in scenario 6 and 22 creates a system were each of the hectares has a larger amount of grass available after period x in comparison to the other grass growth curves. Consequently, the optimal quantity of grass to supply the conversion process (Appendix A1) can more often be fulfilled with 1 hectare instead of 2. However, a smaller more continuous growth rate as that of scenario 38 creates a system were the grass has grown less between two harvest actions and 2 hectares are preferred more often. Since the system always prefers the hectare with the most grass available, the growth curve, and time, between harvest actions for a specific hectare determines the amount of available grass on that hectare at the moment that it is the hectare with the most grass.
4.4 Different harvesting policy
The results discussed in the previous subsections show that the conversion process capacity result in low utilization of the harvesting equipment. i.e. the number of hectares required to harvest to supply the hourly conversion process capacity are a fraction of the installed harvesting capacity. It is believed that a different harvesting policy is able to increase this utilization rate without majorly influencing the conversion process utilization and lost production. This subsection will provide insights in the influence of a different harvesting policy. This, so-called, “bulk” policy tries to fully utilize the harvesting equipment by harvesting the total amount of grass that can be processed by the conversion process within the time the fresh biomass expires. In this way, harvesting actions are executed every 4 hours, to supply the conversion process with the required amount of grass for 4 hours.
Table 8: Performance measurements different harvesting policies
Although the percentage of grass processed reduces if the bulk policy is applied, the operational hours spend on harvesting can be reduced with 66%. The fact that the percentage of grass processed differ between the different policies can be explained by analysing the workings of the policies. This will be explained by an example in which the conversion process capacity is 8 tons per hour, and the average grass available to harvest on a single hectare is 3,5 tons. For the first policy,
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Breaks Compulsory to harvest (times) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 6 8 8 2014 Continuous 200 100,0% 16 0 885 22,8% 900 35,8% 198 8 8 2014 Bulk 200 99,4% 0 0 301 60,2% 822 35,4%
33
the model will harvest 2 hectares (7 tons) since the grass yield of 2 hectares has a smaller absolute difference with the conversion process capacity than that of 3 hectares. Since the process rate is constant, all harvested grass can be processed in this situation. For the second policy, the system will try to harvest 3 times the conversion process capacity (24 tons). In this case the system will harvest 7 hectares (24,5 tons), which results in a production loss of 0,5 ton.
We can conclude although a bulk policy results in a higher lost production, the policy has a significant positive effect on the operational hours spend on both harvesting practices as the conversion process.
4.5 Area sizes
This subsection will provide insight in the impact of utilising different area sizes. The project which inspired this research aims to operate the number of 200 hectares used during the previous subsections. However, it is interesting to identify the scalability of a certain system and the impact of different area sizes in combination with the different conversion process capacities on the harvesting pattern and behaviour. This section focusses on scenario 6, 70 and 134 (table 9). All these scenarios are based on the 2014 growth curve, an 8 hectares harvest capacity and a conversion process capacity of 8 tons per hour.
Table 9: Performance measurements for scenarios with different area sizes
The simulation outputs resulted in the insight that the percentage of processed grass is not linearly related to the area size. It is logical that a decrease in processed grass was found, since the system tries to process a larger amount of grass with the same equipment. If we take a close look at the dataset it can be concluded that this is a result of moments were the harvest capacity is exceeded because the amount of grass available exceeds the upper bound. For scenario 6 this never happened. However, for the other two scenarios it occurred 1 and 2 times respectively. Which resulted in 103 and 892 hectares being harvested compulsory for scenario 70 and 134 respectively. Since the harvested grass must be processed within the first 4 hours, these situations resulted in enormous production losses.
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Breaks Compulsory to harvest (times) Compulsory to harvest (tons) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 6 8 8 2014 Continuous 200 100,0% 16 0 0 885 22,8% 900 35,8% 70 8 8 2014 Continuous 400 90,9% 5 1 1106 1486 14,4% 1355 62,8% 134 8 8 2014 Continuous 800 57,5% 57 3 9535 1478 20,4% 1598 74,1%
34
If we take a look at the number of hectares the system chooses to harvest for the different scenarios, neglecting the different exceeding values explained previously, differences can be identified (figure 16). It can be seen that the ratio between 1 hectare and 2 hectares seems relatively equal, but that the ratio for scenario 134 differs largely. Interesting is to see that the small difference in ratio between scenario 6 and 70 moves slightly the other way than expected. We expected that an increase in area size would increase the number of 1 hectare being harvested in relation to the number of 2 hectares being harvested. However, in this specific case we see a shift to the other side. With the ratio increasing from 0,82 to 0,92.
Figure 15: Number of times specific numbers of hectares are harvested in the system
The unexpected phenomenon can be explained the number of hectares exceeding the maximal level of grass available. Figure 17 provides a simplified visualisation of this phenomenon. In the first figure, a “normal” constant harvesting pattern is visualized. In this pattern none of the hectares exceed the maximal grass length. The yellow blocks represent hectares on which 2 hectares must be harvested to supply the optimal amount of grass, and green blocks are those that can supply this optimal amount by them self. The X’s represent the harvesting pattern. The white blocks represent hectares on which not enough grass is available to harvest. Notice that all blocks turn white after the hectare is harvested (represented with an X).
The second figure shows what will happen if a number of hectares must be harvested due the violation of the maximal grass length. It can be seen that due this phenomenon the pattern is “skipped”, and that the system requires to harvest two hectares each time instead of 1. The results
0 200 400 600 800 1000 1200 1400 1600 200 400 800 Ti m es Area size
Total harvesting actions
35
show that it is not guaranteed that compulsory harvesting actions lead to this phenomenon, but that it is an indication.
Figure 16: Visualisation of harvesting pattern
If we take a look at the utilisation of the conversion process, we can see an increase which, in its core, is a result of processing a higher amount of grass with the same equipment. However, some noticeable insights can be gained. Although the grass yield is approximately doubled between scenario 6 and 70, the utilisation of the conversion process only increased with 31%.
The next aspect that gained our interest is that of the number of breaks. The results show that an increase in area size increases the number of breaks in between harvesting actions. This can logically be explained since both the harvesting and conversion process capacity are equal for all scenarios. Because the system always harvests the hectare with the largest grass availability a pattern is created. The time between a specific hectare being harvest and becoming the hectare with the largest grass availability again takes a larger amount of time if the area size increases. This time between harvesting one specific hectare results in a larger grass yield per harvesting action. Due to the large amount of grass, the system harvests more grass per hour than actually being required to satisfy the hourly conversion process capacity. Grass that is not processed is placed in a buffer, and this buffer size influences the harvesting quantity for the upcoming hour (Appendix A). This behaviour can be examined in the average buffer size over the year (figure 19).
36
harvesting quantity for a specific time (t), the model decides more often to have a ‘break’ to reduce the buffer size.
Figure 17: Average amount of grass in buffer during the entire period
All the scenarios in the previous analyses had equal harvesting and conversion process capacities. Next, we computed an analyse of scenarios with different area sizes that had the same capacity to area size ratio (table 10). i.e. if the area size was doubled in size, both the harvesting and conversion process capacities would be doubled as well.
Table 10: Performance measurements of scenarios equal area size / capacity ratios
The results show that the behaviour of these three scenarios differ. Although it seems that it is not possible to create equal efficiency by increasing both the capacities with the same ratio as the increase in area size based on scenario 385, it seems that this is approximately true for an increase from area size 400 to 800.
Although scenario 385 has the highest percentage of processed grass, we found an enormous increase in breaks in this scenario. This is a result of the average grass availability of a harvested hectare. The average grass yield of one hour in this scenario was larger than the conversion process capacity, which automatically resulted in a constantly increasing buffer size. To decrease the buffer, breaks occurred more often.
0 1 2 3 4 5 6 7 8 9 10 6 70 134 Am ou nt in to ns Scenarios Average quantity in buffer
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Breaks Compulsory to harvest (times) Compulsory to harvest (tons) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 385 4 4 2014 Continuous 200 95,0% 280 1 311 1031 25,7% 1538 68,6% 70 8 8 2014 Continuous 400 90,9% 5 1 1106 1486 14,4% 1355 62,8% 386 16 16 2014 Continuous 800 91,7% 1 1 2051 1495 13,5% 1562 64,0%
37
4.6 No production losses
In the previous sections it was mentioned that almost all scenarios based on 200 hectares were able to process almost 100% of the grass yield. However, subsection 4.5 showed that this is not necessarily the case for the scenarios with a larger area size. In this subsection we ran additional scenarios to provide insight in the lowest capacity settings that are able to process at least 99% of the grass yield for different area sizes. This analysis was computed for both policies addressed during this study.
Table 11: performance measurements for different 99% continuous scenarios
The table above provides an overview of the performance measurements of the 99% scenarios that apply the continuous policy. Logically, to process all available grass in the different scenarios we see a shift in conversion process capacity. An increase in area size results in an increase in required capacity.
If we take a look at the differences between the conversion capacities of the three different scenarios, it can be seen that the required capacity does not perfectly linear. This is a result of the variable grass availability, and the fact that doubling the conversion process capacity does not automatically result in a doubling in the required hectares at a specific moment. The origin of this behaviour can be compared with the behaviour under different policies, which was explained in the previous section.
This analysis shows several interesting operational insights. Although there is no fundamental evidence that economies of scale can be gained for the installed conversion process capacities required. Table 11 shows that they can be gained on operational hours, and utilisation rates.
If we look at the scenarios that are able to process at least 99% of the harvested grass while using the bulk policy (table 12), we see that economies of scale can also be gained for these settings. It was found that the system performed more efficient from an operational perspective if the area size increases. i.e. relative less operational hours are used to process at least 99% of the grass yield.
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Breaks Compulsory to harvest (times) Compulsory to harvest (tons) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 387 4 5 2014 Continuous 200 100,0% 145 1 11 1103 27,0% 1386 58% 389 4 9 2014 Continuous 400 99,6% 10 18 463 1456 30,0% 1592 65% 388 8 20 2014 Continuous 800 99,0% 14 1 376 1300 40,0% 1306 55%
38
Table 12: performance measurements for different 99% bulk scenarios
Additionally, the second increase in area size results in a relatively smaller increase in conversion process capacity. Increasing the area size from 400 to 800 requires a capacity increase of 75%. However, the harvesting capacity has to increase with 138% to process supply the conversion process.
If we compare both the continuous and bulk policy scenarios analysed in this section, it was found that for a bulk policy the system requires larger installed capacities. As shown in section 4.5 applying a bulk strategy has a large impact on both the number of operational harvesting hours, and the utilisation rate of these hours.
5 Discussion
The aim of this study was to determine the impact of a GBR’s conversion process on the harvesting pattern, and how a certain system operates under different circumstances. The upcoming section will first discuss several remarks and potential shortcomings of this study and its model. After that the implications of the findings will be discussed.
5.1 Limitations
This subsection will highlight several limitations regarding the modelling approach. First of all, there are some limitations regarding the parameters used during this study. Although we are aware of the fact that the moisture content of grass varies over the year and even over the day, we used a fixed value to calculate the available grass in terms of grass weight. Beside this parameter, we picked a fixed value for minimal weight available before the system is allowed to harvest. The best length before harvesting are subject to different variables that were not modelled during this system. They include but are not limited to weather conditions, and ‘age’ of the grass. i.e. the minimal length before harvesting is longer for new sown grass than for ‘older’ grass. However, for this study we used averages regarding the moisture content, and guidelines regarding grass length provided by the Wageningen University. Therefore, we can assume that the chosen parameters are realistic, and that the usage of fixed parameters did not influence the usefulness of the results.
Scenario Harvest Capacity (ha/hour) Conversion Capacity
(ton/hour) Year Policy Area size
Percentage processed Breaks Compulsory to harvest (times) Compulsory to harvest (tons) Hours spend harvest Harvesting hour utilisation Hours spend conversion Conversion process utilisation 198 8 8 2014 Bulk 200 99,4% 0 0 0 301 60% 822 40% 270 8 16 2014 Bulk 400 99,6% 0 1 52 352 97% 864 42% 390 19 28 2014 Bulk 800 100,0% 0 0 0 332 80% 982 48%
