Modelling and analysis of the energy intensity in polyacrylonitrile
(PAN) precursor and carbon
fibre manufacturing
Antal Der
a,e,*, Nikolas Dilger
b,e, Alexander Kaluza
a,e, Claudia Creighton
c, Sami Kara
d,
Russell Varley
c, Christoph Herrmann
a,b,e, Sebastian Thiede
a,eaChair of Sustainable Manufacturing and Life Cycle Engineering, Institute of Machine Tools and Production Technology (IWF), Technische Universit€at
Braunschweig, Langer Kamp 19b, 38106, Braunschweig, Germany
bFraunhofer Institute for Surface Engineering and Thin Films (IST), Bienroder Weg 54e, 38108, Braunschweig, Germany cCarbon Nexus, Institute for Frontier Materials, Deakin University, Geelong, Victoria, 3216, Australia
dSustainable Manufacturing& Life Cycle Engineering Research Group, School of Mechanical & Manufacturing Engineering, The University of New South
Wales, Sydney, NSW, 2052, Australia
eOpen Hybrid LabFactory e.V., Hermann-Münch-Straße 2, 38440, Wolfsburg, Germany
a r t i c l e i n f o
Article history:
Received 28 February 2020 Received in revised form 5 March 2021 Accepted 9 April 2021 Available online 13 April 2021 Handling editor: Zhifu Mi Keywords:
Energy intensity Carbonfibre Manufacturing Life cycle evaluation PAN
a b s t r a c t
Carbonfibre manufacturing is characterized by a high energy demand due to long processing times and energy intensive thermal processes. Since energyflow related information about the carbon fibre pro-duction is rare, lacks in detail and ranges widely, the goal of this paper is to increase the transparency of energyflows in the carbon fibre value chain. To this end, the extended energy value stream methodology was modified and applied on a research scale polyacrylonitrile precursor line and a pilot scale carbon fibre line. The paper takes in a holistic factory understanding when quantifying energy demands and it provides highly-detailed energy related data across the carbonfibre manufacturing process chain. Energy demands are broken down into energy carriers, process steps and peripheral levels of the carbonfibre production system. The results of the paper lay afirst valuable step towards a comprehensive life cycle inventory of carbonfibre manufacturing. Furthermore, the high transparency of energy data provides a starting point for future research to increase the energy efficiency of carbon fibre manufacturing. With respect to the high share of the technical building services on the total energy demand, the paper em-phasizes the need for an integrated assessment and improvement of the manufacturing process and technical building services.
© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction 1.1. Motivation and goals
Carbon composite materials are of major importance in today’s manufacturing of lightweight structures. Due to their excellent mechanical properties and strength-to-weight ratio, carbon fibre reinforced plastics (CFRP) have become highly desired engineering materials for structural applications (Herrmann et al., 2018). Since 2010, the demand for carbonfibre reinforced plastics has more than doubled (Witten et al., 2018). The largest carbon fibre utilizing
sectors are aviation& defence, automotive, wind energy and sports leisure, with the aviation industry being the most maturefield of CFRP application and a share of 36% of global CFRP demand. The future demand is expected to increase as major aircraft manufac-turers tend to use large proportions of CFRP in new aircraft models, such as Boeing B787 and Airbus A350XWB that are currently in production ramp-up (Witten et al., 2018). Carbonfibre is defined as afibre containing at least 92 wt % carbon. It is a unique material to the extent that the material properties can be tailored within a vast range, allowing so for optimally engineering its properties for its applications (Huang, 2009; Newcomb, 2016). Carbon fibres are classified according to their mechanical properties, most commonly by modulus and tensile strength, with distinct re-quirements from each industrial sector (Das et al., 2016). To achieve differentfibre properties, process parameters and processing times need to be carefully adjusted, which result in a diverse energy
* Corresponding author. Chair of Sustainable Manufacturing and Life Cycle En-gineering, Insitute of Machine Tools and Production Technology (IWF), Technische Universit€at Braunschweig, Langer Kamp 19b, 38106, Braunschweig, Germany.
E-mail address:a.der@tu-braunschweig.de(A. Der).
Contents lists available atScienceDirect
Journal of Cleaner Production
j o u rn a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j c l e p r ohttps://doi.org/10.1016/j.jclepro.2021.127105
intensity per kg carbonfibre (Huang, 2009).
Carbonfibre manufacturing includes energy-intensive thermal processes as well as long cycle times, which result in high embodied energies and material cost. The cumulative energy de-mand of carbon fibre from state-of-the-art manufacturing tech-nology exceeds the one of engineering steel by a factor betweenfive to ten per kilogram of material (Duflou et al., 2012).Fig. 1compares in this regard qualitatively a carbon composite lightweight struc-ture with a steel reference strucstruc-ture in an automotive application (conventional vehicle with an internal combustion engine) over the whole life cycle. While the lightweight structure outperforms the heavier steel structure in the use stage by a reduced fuel demand per driven kilometre, it starts with a higher environmental impact at the beginning of the use stage. The difference can be traced back to the embodied energy of the carbonfibre and matrix material as well as increased manufacturing efforts. Since an electric vehicle’s environmental impact is largely affected by the environmental impact of the electricity mix in the use stage, the evaluation of the life cycle performance of lightweight structures in electric vehicles is accompanied by an increased complexity (Herrmann et al., 2018). This paper focuses, however, on the manufacturing stage of carbon fibers. In order to avoid problem shifting across life cycle stages by displacing carbon emissions from the use stage to upstream life cycle stages (raw materials and manufacturing), it is paramount to reduce the environmental impact and correspondingly the energy demand of carbonfibre manufacturing (Der et al., 2018).
In a previous study, a methodological framework has been introduced for evaluating the eco-efficiency of carbon fibre pro-duction. The framework outlines the need for a holistic perspective in the energy assessment at different levels of the carbon fibre process chain including the processes, production equipment, technical building services and operation strategies (Der et al.,
2018). Against this background, an increased energy transparency in carbon fibre production is desired. The paper is structured as following. After reviewing available life cycle inventory data and energy related information, a methodology is presented for the energyflow modelling of carbon fibre manufacturing. This work will generate fresh insight into where energy is consumed in the manufacturing of carbonfibres and what the energy intensity of carbon fibres is composed of. For this purpose, the polyacrylonitrile-based (PAN-based) carbon fibre production including the precursor fabrication and its successive conversion to carbonfibre have been modelled and analysed. Following a quali-tative assessment of the energy flows and the production pro-cesses, energy consumers have been prioritized. Data acquisition was conducted depending on the prioritisation based on own measurements, expert interviews and nominal values. The extended energy value stream mapping approach based on (Bogdanski et al., 2013) has been applied to assess the energy
demands of the production process and the technical building services as well. Following the data acquisition, the state-based energy demands have been aggregated and the results are pre-sented and discussed.
1.2. Carbonfibre production
In order to understand the distinct processes in the production of carbonfibre and their relation to the carbon fibre’s energy in-tensity, this section briefly explains the whole production process of PAN-based carbonfibres.Fig. 2zooms into the two main sections of the process chain that distinguishes between the precursor and carbonfibre line. Besides the PAN-based precursor, which is the most common precursor material and is characterized later on in detail, celluloses-, pitch- and lignin-based precursors can be used as precursor material (MInus and Kumar, 2005). The second pro-duction section converts the precursor material in a series of thermal processing steps and surface treatment to a carbon-rich, high-strengthfibre. While converting the precursor to a carbon-rich fibre, different chemical elements evolve out of the fibre, leading to a weight loss in thefibre. This is characterised by the carbon yield that is a physically limited factor. Typical carbon yields from PAN-precursors are between 50% and 60% (Park, 2015). 1.2.1. Precursor line
1.2.1.1. Polymerisation and dope preparation. Commonly, but not exclusively, PAN is polymerized in a solution or suspension by free radicals from acrylonitrile. Solution polymerisation of PAN can be performed by mixing acrylonitrile (AN) with other comonomers in presence of non-polar solvents and catalysts to initiate the reaction. The heat that is generated during this process will be absorbed by the solvent and the resultant PAN polymer used to prepare polymer dope for wet spinning. Generally, the dope is prepared by dissolv-ing 10e30% of PAN polymer in the solvent such as dimethyl sulf-oxide (DMSO), Dimethylformamide (DMF), dimethyl acetamide (DMAC) at temperatures around 60C. (Huang, 2009;Park, 2015). 1.2.1.2. Spinning. Depending on the dope that is used to produce PAN-precursor, different types of spinning are applied. The preva-lent process is wet spinning, although dry jet wet spinning can accomplish increased mechanical properties by being able to spin high polymer concentration dope (Huang, 2009). In the coagulation process of wet spinning, the dope solution is transferred at 25e120C in the coagulation bath, containing a solvent and a
non-solvent, through a spinneret with up to 500.000 holes (Gupta et al., 1991). The coagulation environment in terms of temperature and spinning bath configuration affects the coagulation rate and thereby thefibre quality (Chen et al., 2007;Morgan, 2005).
Fig. 1. Qualitative comparison of a lightweight structure with a reference structure over its life cycle, based on (Der et al., 2018).
1.2.1.3. Stretching and finish. After extrusion in the coagulation bath, thefilaments are further stretched to enhance the orientation of molecular chains along thefibre axis. Stretching procedures are carried out by stretching in the spinning bath or in-air and in-bath combined, followed by a stretching process in pressured steam (Morgan, 2005). Prior to collection, thefibre needs to undergo an application of a finish, which serves for lubrication and as an antistatic agent. Afterwards, thefibre is being dried and relaxed and can be spun on a spool in different tow sizes (Morgan, 2005). 1.2.2. Carbonfibre line
1.2.2.1. Stabilization/oxidation. This first heat treatment of the carbonization process is carried out to stabilize the PAN precursor fibre. The oxidation process takes place in an air atmosphere at temperatures between 200 and 300C. As the PAN precursorfibre has inferior heat conductivity characteristics and exothermic re-actions occur, the heating step must be taken out slowly to prevent heat building up in the tow, which would damage thefibre surface and lead to defects. Usually a stepwise temperature increase is applied, where heat is evenly distributed in the oven and the fila-ments (Morgan, 2005). To reduce relaxation during stabilization, tension must be employed on the fibre (Edie, 1998). Distinct chemical reactions take place throughout the stabilization, including cyclization, dehydrogenation and oxidation (Rahaman et al., 2007). The oxidation process is the most time-consuming step in the manufacturing process of CF, hence it is subject to constant review (Morgan, 2005).
1.2.2.2. Carbonization. After the stabilization step, the PAN pre-cursorfibre can withstand a heat treatment at higher temperatures. In the carbonization process, a thermal pyrolysis is performed in an inert gas atmosphere (Morgan, 2005). The carbonization step is carried out at temperatures between 500 and 1600 C. Being treated with high temperatures, most non-carbon elements in the precursor are volatilized in form of different gases as methane, hydrogen, carbon monoxide, carbon dioxide and ammonia (Edie, 1998). Commonly, the process is divided in two or three different parts, using different ovens for specific temperature zones. Starting with a low temperature [LT] furnace, thefibre is heated up to 950C
in an electrically heated slot furnace with different zones, that remove tar from thefibre. Continuing with a high temperature [HT] furnace, the temperature is exceeded to enhance the fibres modulus (Morgan, 2005). Increasing the temperature past 1600C in an ultra high temperature [UHT] furnace results in a reduction of tensile strength, but leads to increasing modulus (Huang, 2009). By removing the non-carbon elements from the fibre, only about 50e55 %wt. of the original precursor will remain, containing more than 98 %wt. of carbon (Huang, 2009).
1.2.2.3. Surface treatment. Surface treatment is applied on carbon fibres to additionally enhance the mechanical properties by altering the surface (Chand, 2000). Most commonly, some form of oxidation in liquid or gaseous form is applied with the most common being
electrochemical oxidation. The electrically conductive CF functions as an anode, while passing through an acidic or salt solution elec-trolyte bath. Post to the surface treatment, the surplus of elecelec-trolyte is washed off in a rinsing bath with warm water (Park, 2015). For handling, lubrication andfibre protection, a sizing layer is applied to thefibre surface. Sizing refers to the polymer solution such as epoxy formulations and resins that coats thefibres and is required as a part of the weaving process. Before and after the sizing, the fibre is dried which either happens with a contact dryer in form of a hot drive roller, or a non-contact-dryer using hot air. Finally, the tows are collected by online-winders or packaged in boxes for larger tow sizes (Dai et al., 2011;Morgan, 2005).
1.3. Derivation of the research gap
Fig. 3 presents a framework for the cradle-to-gate environ-mental assessment of PAN-based carbon fibre manufacturing following the ISO standardized methodology of Life Cycle Assess-ment (LCA). All input and outputflows constitute the life cycle inventory (LCI), which is linked with characterisation factors to environmental impact categories in the life cycle impact assess-ment. For the life cycle evaluation of carbonfibre production and for the interpretation of these results, it is important to keep in mind system boundaries.Fig. 3differentiates between three system boundaries that can also be used for comparing life cycle inventory data from literature. The cradle-to-gate perspective involves both process sections, the precursor production and the carbonfibre production. Energy and materialflows, emissions as well as raw materials and intermediate products are all part of this system boundary. In contrast to this, the gate-to-gate perspective includes only energy and materialflows that are linked with the production process. The life cycle inventory of raw materials is excluded by the gate-to-gate perspective. This paper takes in a gate-to-gate perspective on the carbonfibre process chain. It analyses highly-detailed energy flows quantitatively and material flows more aggregated on a qualitative basis.
In contrast to steel and aluminium, where LCI is also provided by associations such as the World Steel Association and the Interna-tional Aluminium Institute, life cycle inventories with respect to carbonfibres provided by related associations are not existing. LCI datasets on carbonfibre production in commercial LCI-databases (e.g. by GaBi) are rare and if existing they only provide an aggre-gated value without further possibilities to investigate the contri-bution of single processes to the total energy intensity. Scientific literature and published data about the energy intensity and life cycle inventory of the carbonfibre manufacturing process chain is limited. Publications with primary data were preferred in the following literature review. Cited values based on previous work were only included if the citation was based on unpublished work or the original publication was no longer accessible.Table 1and
Table 2compile reported data on the energy intensity of PAN-based precursor manufacturing and carbonfibre manufacturing, respec-tively. Energy related data for the carbon fibre manufacturing
Fig. 2. The carbonfibre process chain (compiled fromHuang, 2009;Morgan, 2005;Park, 2015).
process can be found within a few published life cycle assessments, life cycle inventory data or other analysations and assessments.
Reports on the energy intensity of PAN precursor production are scarce (Table 1). Das describes in a cradle-to-gate analysis, that the high energy demand of the production is caused by acrylonitrile as a raw material. The reported primary energy demand amounts to 245 MJ/kg PANfibre (Das, 2011). Another study presents the gate-to-gate energy intensities of the process steps polymerisation and spinning. The values lie in a range between 312 and 394 MJ/kg PAN fibre (Liddell et al, 2016,2017).
Regarding the energy intensity of carbonfibre manufacturing in
Table 2, some reports differentiate between the process steps (stabilization, carbonization and surface treatment), but leave out specific information about the examined system, while others only outline a combined number for the stabilization and the carbon-ization process. Likewise, the distinction of the total energy in-tensity on different types of energy is often missing. Liddell et al. distinguish the data into current typical, state-of-the-art and practical minimum, showing the energy reduction possibilities for
the process steps from precursor to carbonfibre. However, only current typical values are based on measurements (Liddell et al, 2016,2017). The data by Arnold et al. is based on a 1500 tonnes per year production capacity of 12 k tow size carbonfibre. They differentiate between electricity demand (125.4 MJ/kg CF) and heat demand (129.62 MJ/kg CF) (Arnold et al., 2018). Suzuki et al. pro-vide energy intensities based on different years, but give no further information than the overall energy intensity for the whole process chain (Suzuki and Takahashi, 2005). Das describes a system, where a 50 k tow size commercial grade carbonfibre with a yield of 48% is produced (Das, 2011).
The reported energy demand per kilogram PAN and carbonfibre ranges widely. System boundaries of the presented energy in-tensities, if detailed, are inconsistent among the presented publi-cations. The specification of fibre properties is often missing. Moreover, the level of detail for energy breakdown on the processes differs. While some publications break down the energy data on major processes, others simply display an accumulated value for the parts or the whole the process chain from precursor tofinal
Fig. 3. Framework for the cradle-to-gate environmental assessment of carbonfibre manufacturing following the Life Cycle Assessment (LCA) methodology.
Table 1
Reported energy intensities for PAN-based precursor manufacturing.
Das (2011) Liddell et al. (2016) Liddell et al. (2017)
Polymerisation [MJ/kg PAN] N/A 199 118
Spinning [MJ/kg PAN] N/A 195 194
Total [MJ/kg PAN] 245 394 312
System boundary cradle-to-gate, primary energy gate-to-gate gate-to-gate
Table 2
Reported energy intensities for carbonfibre manufacturing.
De Vegt and Haije (1997)
Suzuki and Takahashi (2005)
Griffing and Overcash (2009)
Das (2011) Liddell et al. (2016)
Liddell et al. (2017)
Arnold et al. (2018)
Stabilization [MJ/kg CF] N/A N/A N/A N/A 316 195 N/A
Carbonization [MJ/kg CF] N/A N/A 4.47 N/A N/A
Surface Treatment [MJ/ kg CF]
N/A N/A 0.05 N/A 25 24 N/A
Total [MJ/kg CF] 7.56 286e478 4.52 704 341 219 255.02 Fibre properties N/A N/A high strength N/A N/A N/A N/A System boundary gate-to-gate N/A gate-to-gate primary
energy
gate-to-gate gate-to-gate gate-to-gate Journal of Cleaner Production 303 (2021) 127105
carbonfibre. The energy related data is often based on estimations and rough calculations and does not specify information about the examined system. Details about system boundaries, used parame-ters, annual capacity, assessment/calculation methods and char-acteristics about the produced fibre grade are rarely given. Additionally, the reported values vary extremely by a factor of up to 100. Therefore, the data does not allow any comparison and is of limited use for the environmental evaluation of carbonfibre usage in lightweight structures as well as for a decision support towards potential improvement measures for advanced eco-efficiency in carbonfibre production. Although a few studies report on energy intensities, little is known in detail about the dynamic energy and resourceflows in carbon fibre production. A solid understanding of energy hotspots in the carbonfibre process chain is however the starting point for evaluating energy efficiency improvement mea-sures as well as future process development directions (e.g. changes in processing times, process operations and processing equipment). Furthermore, assessing the environmental impact in future production scale-up and energy scenarios will be only possible with a high energy transparency along the process chain. 2. Methodology for assessing the energy intensity of carbon fibre production
2.1. Methods& tools for assessing energy intensity in production systems
In order to describe a factory system pursuant to its inherent complexity, it can be broken down into different levels spanning from unit processes, over process chains to whole factories ( Reich-Weiser et al., 2010;Wiendahl et al., 2015). A holistic view on factory systems outlines the interdependencies between factory elements, such as production machines, technical building services, factory building and local climate and allows of a comprehensive assess-ment of dynamic energy and mediaflows (Hesselbach et al., 2008).
Fig. 4 adapts the concept of holistic factory understanding from
(Herrmann and Thiede, 2009; Hesselbach et al., 2008) in the context of carbon fibre manufacturing. The figure illustrates inflowing and outflowing energy and material flows on different levels of the manufacturing system (process, process chain and factory). Moreover, external and internal influencing factors (e.g. external: local climate determining heating and cooling demand, internal: waste heat and air emissions from production machines that expose the production environment to additional thermal loads) are highlighted between the factory objects that dynamically affect the energy and media demand of the production machines and technical building services (TBS).
Methods and tools for assessing energy demands in production facilities are facilitators for increasing the energy transparency in manufacturing companies. Assessing energy demands is thus afirst step towards improvements, regarding both costs and environ-mental impacts. In order to support the collection of LCI data of manufacturing unit processes, Kellens et al. provide a methodology that includes a time, power, consumables and emissions study (Kellens et al., 2012). The methodology distinguishes between different machine states (e.g. standby and processing mode) and allocates state-based energy, consumables and emissions data. The methodology suggests the measurement or estimation of related data. The topic of energy efficiency has been extensively researched in the previous years that led to a number of methods and tools (Menghi et al., 2019). In the following, different energy assessment perspectives as well as methods and tools are discussed.
Thiede et al. established a framework for the continuous improvement of the energy and resource efficiency in manufacturing environments. The framework consists of several methodological steps that support the identification of main en-ergy carriers, prioritisation of enen-ergy consumers and deriving the type of data acquisition. The framework suggests developing an energy model that is later used for the evaluation of improvement measures (Thiede et al., 2013). Energy Value-Stream Mapping (EVSM) may be applied for this purpose. EVSM can be used to assess the energy demand of manufacturing processes from a
Fig. 4. Holistic factory understanding of carbonfibre manufacturing (adapted fromHerrmann and Thiede, 2009;Hesselbach et al., 2008).
process chain perspective (Erlach, 2013). The approach is originally based on value stream mapping that aims at eliminating waste along the value chain from the lean management philosophy (Rother and Shook, 2003). Over the years, EVSM has been advanced in multiple approaches, either introducing new features to the baseline concept or extending already enhanced adaptations with new features. Hence, there exist various EVSM adaptations with a focus on sustainability assessment, e.g.(Edtmayr et al., 2016;
Faulkner and Badurdeen, 2014), integrated economic and envi-ronmental assessment, e.g. (Alvandi et al., 2016), and advanced energy modelling, e.g. (Bogdanski et al., 2013;Müller et al., 2013). Further approaches focused on IT-supported data acquisition and visualization for better decision support, e.g. (Li et al., 2017;
Sch€onemann et al., 2016), as well and integrated analysis of energy and materialflows, e.g. (Papetti et al., 2019;Schmidt et al., 2007).
Since energy value stream mapping takes a static view on the energy demand of production equipment, but machine tools have a dynamic electric load curve, EVSM has been enhanced in the concept of extended energy value stream method (e-EVSM) for a more comprehensive tool for energy assessment (Bogdanski et al., 2013). The e-EVSM approach addresses the dynamic energy de-mand profile of production machines by distinguishing between distinct machine states of ramp-up, idle and processing. Accord-ingly, e-EVSM allows of a more accurate modelling of the energy demand in process chains by matching state-based load levels with corresponding state-specific times. The e-EVSM approach also in-tegrates the energy demands induced by the TBS, as this is directly involved in maintaining the production conditions (illumination, temperature, humidity and media supply).
The TBS related energy demand can either be allocated to spe-cific production equipment or be considered as an energy overhead for the whole process chain (Bogdanski et al., 2013). One allocation procedure classifies auxiliary processes in different peripheral or-ders according to their proximity to the value adding process (Schenk et al., 2013). Posselt et al. included the peripheral orders in the extended EVSM approach to distinguish between direct and indirect energy demands (Posselt et al., 2014). The share of time and energy being used to actually create value for a finished product is considered as direct energy demand and value adding time. The indirect energy demand of auxiliary processes is divided into peripheral orders and can be either value or non-value adding. 2.2. Adaptation of an energy assessment methodology for carbon fibre production systems
In the following, the approach for assessing the energy intensity of precursor and carbon fibre production in a gate-to-gate perspective is introduced. To this end, the previously introduced approaches were adapted to the characteristics of a carbon fibre production system.Fig. 5illustrates with this regard the multi-step approach that leads to a high transparency over the energy de-mands in precursor and carbon fibre production line. Each step from A-E is explained in the following.
2.2.1. A: Consumer and key process identification
The first step aims at selecting the processes, machines and production equipment as well as energy carriers (electricity, com-pressed air, natural gas). All relevant energy consumers need to be identified and allocated to process steps along value chain. One process step can include several devices and production equip-ment. Multiple walks along the production line, engineering drawings, circuit diagrams and expert interviews can be employed to build up a detailed overview of the production facility. In order to handle complexity and focus the energy assessment, prioritisation of energy consumers and process steps may be necessary. Besides
energy related data, process times and material related data are taken into account during prioritisation. Efforts for data acquisition at this step should be kept low. Therefore, nominal power values and process specifications are sufficient.
2.2.2. B: Allocation to peripheral levels
This step aims at assigning the devices and production equip-ment to the different levels according to the model of production periphery by (Schenk et al., 2013) that is applied in the extended EVSM approach by (Posselt et al., 2014). The peripheral levels are determined based on the functional adjacency to the value adding process.Fig. 6illustrates the allocation logic according to the pe-ripheral levels. The value adding process is in the core of this sys-tem understanding. The first peripheral order is represented by linked equipment and can be defined as devices that are particu-larly assigned to a single production machine. Material trans-portation and energy distribution equipment that links two or more devices, such as transportation belts can be classified as the second peripheral order. The third periphery level (e.g. centralized equipment such as air compressors, steam generators or in-cinerators) has a lower proximity to the value adding process. Lastly, other technical building services, such as air-conditioning, heating, ventilation and lighting form the fourth peripheral order (Posselt et al., 2014).
2.2.3. C: Qualitative input/outputflows
An input/outputflow diagram can visualise the process chain and the connections between the consumers as well as energy and materialflows. Mapping the processes of the precursor line and carbonfibre line allows a detailed understanding of the process chains. While still considering process steps as black box models, this step aims at creating a complete list of energy and material inputs for each process step.
2.2.4. D: Data acquisition
Data acquisition aims at compiling energy and media demands for the production machinery at different operating states from different data sources. The goal is to systematically gather data on relevant energyflows and meta data for each process in accordance with steps A and C. The means of data acquisition (estimation, spot sampling, temporary sampling and continuous monitoring) is agreed on for each process and energy carrier (Posselt, 2016). Distinct machine-state-based energy demands (e.g. for standby, waiting and processing) are preferable for key energy consumers. For this purpose, data collection at the shop floor with interval spans from a couple of hours to days may be required. Spot sam-pling at the shopfloor and extracting nominal values from machine labels are feasible for production equipment that is less relevant from an energy demand perspective. Shopfloor measurements can be omitted at facilities, where an energy management system is installed. In these facilities, the data acquisition is limited to the analysis of existing energy data logs.Fig. 7displays on an exem-plary load curve of a heating element how the electrical load curve is used to derive state-based power demands. The load curve is analysed for every energy consumer. Two distinct states (ramp-up and processing) are identified and an average power demand is assigned in both states. This builds the basis for setting up the energy value stream map in the next step.
2.2.5. E: Evaluation and visualization
Setting up the extended energy value stream enables the energy assessment from different perspectives. The allocation of the pro-duction equipment to peripheral levels allows for distinguishing between the process-related and non-process-related energy de-mands. State-based energy demands broken down on the single
energy consumers in the production environment further enhance the energy transparency across the process chain.Fig. 8illustrates the methodology for calculating the energy intensities based on an extended energy value stream map. While energy measurements and energy assessments in a production context typically use kWh to display the results, the total energy intensity EITotal will be
calculated as MJ per kilogram PAN and MJ per kilogram carbon fibre. This conforms to the already published energy intensities and enables comparability (Tables 1 and 2). The total energy intensity EITotalis the sum of the process step-related energy intensities EIi.
The process step-related energy intensity EIiof the process step i
includes the state-based energy intensity of all allocated energy consumers across the peripheral levels 1e4 inside the given process step i.Fig. 8illustrates the calculation logic of the energy intensity only on thefirst peripheral level in detail. The calculation for the other peripheral levels is analogous to thefirst level. Every energy consumer is listed according to its allocation to peripheral levels
with its corresponding state-based power demands. The power levels from the energy data acquisition distinguish between ramp-up and processing loads. Taking thefirst peripheral level as an example, the energy intensity EI1st is the sum of the energy
de-mand during ramp-up and processing. Hereby, the energy dede-mand is the product of power demands and the respective time intervals (ramp-up time and cycle time). The energy demand of consumers that are not directly assigned to a single process step, e.g. the abatement system or lighting, is imputed to the process steps based on a process step-specific factor. This factor depends on the rate, to which the corresponding process step is accountable for the total energy demand. An example of this is the abatement system, where the factor is determined based on the exhaust gasflow rate from a given process step. In the case of lighting, the factor is calculated based on the production area of the corresponding process step.
Fig. 5. Methodology for assessing the energy intensity of carbonfibre production.
Fig. 6. Peripheral allocation of TBS equipment, based on (Posselt et al., 2014;Schenk et al., 2013).
3. Application of the energy assessment methodology in a case study for PAN-precursor and carbonfibre production
The previously presented methodology has been applied to determine the energy intensity of carbon fibre production at a research scale precursor line and a pilot scale carbonfibre line. The precursor line is a single tow wet spinning lab line, with a tow size of 0.5 kfilaments. Since the precursor line operates as a small-scale research line, it is operated for trial campaigns and not in a constant shift system. One campaign cycle usually runs until the prepared dope is consumed and is performed in approximately 6e7 h or less, depending on the dope mass. Being a continuous process, idle phases are non-existent during the regular operation state. Mea-surements took place in production during continuous operation, after the process conditions had stabilized. Measurements in the precursor line were based on basic clamp metering.
The carbon fibre line is a multiple tow line, designed to
carbonize PAN-based precursor to carbonfibre. It can operate with up to 30 tows with a maximum tow size of 24 kfilaments and 100 kg per bobbin on the creels. Its annual capacity is 110 tonnes of carbonfibre, when operated continuously. In the carbon fibre line, there are two distinct ramp-up scenarios depending on the pro-duction setting. A white warm-up is applied, when the precursor fibre is initially fed into the oxidation. As there is untreated pre-cursorfibre in every zone of the oxidation ovens, heat needs to be applied slowly to prevent quickly rising heat and ignition. When each zone of the oxidation ovens already has heat treatedfibre inside and the process is shut down, starting up again can be done faster, because the heat can be applied more quickly. This ramp-up scenario is called black warm-up. As oxidation and carbonization are continuous processes as well, there are no idle phases in normal operation. However, cleaning and maintenance cycles are per-formed regularly and during that time, the line enters an energy saving mode. In this mode, the line is stopped and most of the energy consumers are either taken off the power supply or run with reduced power. Measurements have been undertaken for black warm-up as well as during continuous operation at stable process conditions. Next to basic clamp metering of less relevant energy consumers, the energy data was acquired for oxidation and carbonization furnaces as well as the abatements system and drives in the carbonfibre line from continuous metering equipment.
After identification of the relevant energy consumers, the pro-duction equipment has been allocated to peripheral levels of the production system.Fig. 9 and Fig. 10present in this regard the allocation on the example of the research-scale precursor line and pilot-scale carbonfibre line. In addition to the peripheral allocation, thefigures match the energy consumers with energy and resource flows in a qualitative manner. Energy flows include electricity and compressed air, steam in the precursor line and natural gas in the carbonfibre line. Input material flows in the precursor line include the polymer powder, solvent, demineralised water, helium and the finish agent. Input material flows in the carbon fibre line consist of nitrogen, water, fresh air, electrolyte and emulsion. The only notable output materialflow depicted by the figures is exhaust air
Fig. 7. Identification of state-based power demands on an exemplary load curve of a heating element.
Fig. 8. Schematic energy value stream map for the calculation of the energy intensity in the precursor and carbonfibre line.
that needs to be treated in an abatement system before releasing it to the outside atmosphere.
3.1. Precursor line
Fig. 11 and Fig. 12 present the as-is results of the energy
assessment of the research scale PAN production. Since the line is a research scale wet spinning line with one single tow of 0.5 k, the only energy input to the process chain is electricity. Due to a limited data availability at the precursor line, power demands were solitary measured during processing. The presented results therefore only include the energy demand from the operating state and neglect
Fig. 9. Peripheral levels and allocated production equipment for PAN-based precursor manufacturing.
Fig. 10. Peripheral levels and allocated production equipment for carbonfibre manufacturing.
ramp-up energy demands. During the measuring campaign, the line was operating with a line speed of 56 m/min and a yield of 72%. Consequently, the throughput only comes to about 120 g PAN/hour. Examining Figs. 11 and 12, the energy demands may appear excessively high, which is presumably due to the small process scale of a single tow with 0.5 k. In a multiple-tow production sce-nario with larger tow size the energy intensity is expected to decrease rapidly.
The most energy demanding processes in the precursor line are the coagulation, the steam stretching and steam relaxing processes with about 50% of the total energy demand per kg PANfibre. All of these processes involve heating devices with high energy demands. Steam stretching with 834.4 MJ/kg PAN is the most energy inten-sive process step, followed by steam relaxing with 799 MJ/kg PAN. Coagulation requires 581.9 MJ/kg PANfibre. Altogether, the energy intensity of the investigated precursor line amounts to 4436.3 MJ per kg PANfibre for the investigated single tow precursor line.
Fig. 12breaks down the energy intensity of the PAN production into process steps and peripheral levels. As the chart illustrates, 67% of total energy demand in coagulation occurs directly at the pro-cessing equipment and thefirst peripheral order. This demand can be traced back to the heating devices that provides the coagulation bath temperature of 60C. On the contrary, the energy demand in steam stretching and steam relaxing is mainly driven in the third peripheral order by a central steam generator. The relevance of the energy demand from the second peripheral level is highest in process steps that transport thefibre to the next step and do not require thermal treatment, such as washing, drawing and spin finish application. Moreover,Fig. 12highlights the relevance of the peripheral production equipment on the energy intensity of PAN fibres. As the chart illustrates for the total of all production steps, the direct process energy demand from thefirst peripheral level only makes up a third of the total energy intensity, while the third peripheral level (in this case primarily steam generation) accounts for 53% of the total energy demand.
3.2. Carbonfibre line
Fig. 13 andFig. 14 summarise the as-is results of the energy assessment of the pilot scale carbonfibre production. The results have been calculated for continuous production of 24 k tow size carbonfibre with a ramp-up of 10 h and maintenance time of 12 h every eleven days. The carbon yield from PAN to carbonfibre was 44% for an annual production scenario. During the measuring campaign, the line was operating with a line speed of 120 m/h. The energy intensities include the natural gas demand in the abatement system and the electricity demand from the involved production equipment. The temperature range of the oxidation was between 220C and 250C. Carbonization temperature was 700C in the LT oven and 1300C in the HT oven. The exhaust gases from oxidation and carbonization were burned at 850C in the abatement system. With the given process parameters, 5.0 kg of carbonfibres were produced within an hour.
The oxidation and the carbonization furnaces are by far the highest energy consumers with about 97% of the total energy de-mand per kg carbonfibre. Considering only electricity, the oxida-tion furnace has the highest demand of 166.2 MJ/kg carbonfibre, followed by the HT furnace with 146.3 MJ/kg carbon fibre and
Fig. 11. Energy intensity of the PAN production separated into process steps.
Fig. 12. Breakdown of the energy intensity of the PAN precursor production line into process steps and peripheral levels.
subsequently the LT furnace with 111.7 MJ/kg carbonfibre. How-ever, when considering energy demands from both processes (electricity) and abatement system (natural gas), the LT furnace is responsible for 48%, the oxidation for 34% and the HT furnace for 16% of total energy demand. The differences in the natural gas demand result from the specific share of exhaust gases that needs to be treated in the abatement system. The described distribution of natural gas demand is also reflected on the energy breakdown on the peripheral levels and process steps that is displayed inFig. 14. While 76% of the energy demand in the HT furnace is caused by the heating elements that is part of machine andfirst peripheral level, the main energy demand in the LT furnace (81%) and oxidation (59%) occur in the abatement system (third peripheral level). The remaining energy demand in these process steps is linked to the direct production equipment in thefirst peripheral order, such as recirculation fans and heating elements. Altogether, the energy intensity of the investigated carbonfibre line amounts to 1150.5 MJ/ kg carbonfibre. The high energy demand of the abatement system that uses natural gas is reflected in the breakdown of the total
energy intensity on natural gas (60%, 692.6 MJ/kg CF) and electricity (40%, 457.8 MJ/kg CF).
The notable influence of the abatement system on the energy demand is also reflected by the last bar inFig. 14that illustrates the distribution of the total energy intensity of carbonfibre production in the carbonfibre line on peripheral levels. From the chart, it can be seen that 62.5% of the total energy demand is dedicated to the third peripheral level (due to the abatement system that treats the process gases coming from oxidation, LT and HT furnaces). The second biggest share with 36.5% is the process related energy de-mand from thefirst peripheral order.
The allocation of the energy consumers to peripheral levels of the production system allowed the differentiation between value adding and non-value adding energy demands. This is displayed by
Fig. 15for the carbonfibre line, divided into electricity and natural gas. Looking at the aggregated distribution for the total process chain, the non-value adding energy demand outweighs the value adding shares. This is especially true for natural gas, which is used for burning noxious exhaust gases from oxidation as well as LT and HT carbonization furnaces.
4. Interpretation of results
This paper concludes with the discussion of the energy assess-ment results and its comparison with the literature values that were presented in section1.2. The calculated gate-to-gate energy intensity of precursor production of 4436.3 MJ/kg PAN outweighs by far the previously reported gate-to-gate values of 394 and 312 MJ/kg PAN, respectively. However, it is not known, whether this difference comes from different system boundaries or from different line scales. Due to missing information regarding the system boundaries (e.g. included process steps and production peripheries) and the line scale in the literature values, it is not possible to resolve the exact reason for the difference. The pre-sented energy intensity of precursor production in this paper is based on a research scale line and includes process related energy demand from the direct production periphery as well as technical building services.
The gate-to-gate energy intensity of 1150.5 MJ/kg carbonfibre in the carbonfibre line is also much higher than gate-to-gate litera-ture values. The minimum and maximum gate-to-gate literalitera-ture values were found to be 7.56 and 478 MJ/kg carbon fibre,
Fig. 13. Energy intensity of the carbonfibre production separated into process steps.
Fig. 14. Breakdown of the energy intensity of the carbonfibre production into process steps and peripheral levels.
respectively. Similar to the precursor line, the differences cannot be resolved due to missing information in the literature regarding system boundaries and line scales. The most detailed literature values break down the energy intensity into a joint value for oxidation and carbonization (between 92% and 89% of total energy intensity) and surface treatment (Liddell et al, 2016, 2017). The presented energy data in this paper that is based on own mea-surements signalise an even higher relevance (97% of total energy demand) of the oxidation and carbonization process steps. A more detailed relative comparison on process step level is however not feasible due to missing details in literature. Regarding the carbon fibre line, energy data was acquired on a pilot scale carbon fibre line with an annual production capacity of 110 tonnes. The reported energy intensity in this paper includes process and non-process related energy demands from direct production equipment and from technical building services as well. The energy demand of the abatement system was allocated proportionally to the measured exhaust gas volumeflows from the process steps. The results un-derline the high relevance of the energy demand from the technical building services that amount to 62.5% of the total energy intensity in the investigated carbonfibre line. The energy hotspots in the carbonfibre line are oxidation and carbonization. Both of them are dominated by non-value adding energy shares from the abatement system, shown as“natural gas” inFig. 13. Comparing the energy demands of electricity and natural gas in the carbonfibre line with the most recent dataset in the literature reveals a close relation with the acquired data in this paper. The ratio of electricity to natural gas, that is used to treat waste gases, was 49:51 in literature (Arnold et al., 2018), while own measurements came to a ratio of 40:60. The higher portion of natural gas in this study can be explained by the type of abatement system used, which does not recover energy as commercial abatement systems do.
In order to reduce the total energy demand in the carbonfibre production including the precursor and the carbonfibre line, the most relevant energy consumers, known as energy hotspots, need to be known. Although Figs. 11 and 13 present the energy in-tensities for each process step regarding the electricity and natural gas demand, a breakdown of energy consuming production
equipment, which directly supports the identification of energy improvement strategies, is however missing. As the investigated production lines are at different scales, the top energy consumers are accounted for separately for both lines. This avoids an incon-sistent comparison of the energy demands from both lines.Fig. 16
andFig. 17illustrate with this regard the energy hotspots for the precursor line and carbon fibre line from an energy consumer perspective. Thefigures break down the top energy consumers by energy services instead of process steps. The precursor production is dominated by the energy demands from heating elements, fans and electric drives. These three energy services constitute round 97% of the total energy demand. Respectively, energy improvement strategies should focus on reducing the thermal energy demand in the precursor production. From an energy point of view, reducing process times or process temperatures can provide a huge potential for reducing the total energy demand. Regarding fans and drives, more efficient electric motors, avoiding overdimensioning and load adapted control of motors could be possible improvement measures.
In the carbon fibre line, from an energy point of view the abatement system is the largest individual energy consumer and the only natural gas consumer across the process chain. This is followed by electric heaters and fans that together constitute 98% of the total energy demand. The abatement system is responsible for burning noxious gases from the process steps of oxidation and carbonization, which themselves require elevated process tem-peratures. Waste heat recovery inside the LT and HT furnaces and in the abatement system have therefore substantial potential for reducing the energy intensity of carbonfibre production. Analo-gous to the precursor line, reducing process time and temperature would have a vast potential to decrease the energy intensity.
In the context of the cradle-to-gate energy intensity of carbon fibre production, the material yield of the carbon fibre line has an undisputable effect. At the current best practice of 55% yield in carbonfibre manufacture (Park, 2015), almost the double amount of precursor is required to produce one unit of carbonfibre. Lower yields result in an even higher precursor material demand and corresponding energy intensity of the precursor for the production of one unit of carbonfibre. Additionally, the carbon yield seems to have a large effect on the energy intensity of carbonfibre. A higher conversion yield from PAN to carbonfibre increases the output rate at constant temperatures and line speed. When assuming that the power levels remain constant in the production equipment, the specific energy demand per kilogram of carbon fibre decreases. This effect is exemplified by a simple comparison. Increasing the current yield of 44%e55% results in reducing the specific energy demand from 1150.5 MJ/kg CF to 960 MJ/kg CF. However, this energy reduction potential must be interpreted with caution because po-tential process parameter adjustments and resulting changes in the energetic behaviour of the production equipment are neglected.
While this paper makes a significant contribution towards en-ergy transparency in carbonfibre manufacturing, following short-comings and limitations need to be mentioned:
⁃ The examined precursor line is a research line operating at a scale with small output, while the carbonfibre line is a pilot line at larger scale that can produce about 110 t per year in an ideal production scenario. Due to the scale differences between the investigated production lines, their energy intensities cannot be compared consistently and do not represent industrial scales. ⁃ Due to its small scale of the precursor line, providing the energy
demands for individual process steps as well as value adding and non-value adding energy demands is not informative. Therefore, in contrast to the carbonfibre line, the results are not presented.
Fig. 15. Breakdown of the energy intensity of the carbonfibre production into process steps and value and non-value adding energy demands.
⁃ Measurements along the precursor line were based on basic clamp metering. Therefore, the measurements only represent a short timeframe and divergences from long-time values are possible. In order to avoid inaccuracies, measuring the ramp-up phase in the precursor line has been omitted. Long-interval measuring campaigns or continuous energy monitoring would enhance the accuracy but would require a tremendous amount of additional time, since every device needs to be measured separately.
⁃ Ramp-up values were only considered in the carbon fibre line for the heating elements. To fully reflect a ramp-up phase, the remaining energy consumers also have to be metered. ⁃ The applied methodology allows only a static view in carbon
fibre production. Dynamic interdependencies between process parameters, process conditions and product properties as well as the influence of different operation strategies and circum-stances, such as yearly output, production programme, number offibre variants, their individual processing times and param-eters, are neglected.
⁃ While energy flows were quantified in a high resolution, ma-terialflows were only depicted in a qualitative manner. Together with highly detailed energy flows, quantifying material flows with a similarly high detail level would close the gap regarding the missing information on life cycle inventories of carbon fibres.
⁃ More data sets should be acquired to validate current data and more information regarding the influence of relevant process parameters, such as line speed and different temperatures
should be collected. By that, data accuracy would increase and energy intensities could be provided for carbon fibres with different material properties and scale-up scenarios.
⁃ The evaluation examined the processes and the devices still in a black box model, neglecting the specific procedures, physical actions and chemical reactions inside the processes. In order to analyse the effect of specific process parameters and process conditions on the energy intensity, detailed models are needed that represent the underlying interrelationships of the physical and chemical processes as well as production system dynamics between the process steps and technical building services. Such a glass-box model could also be applied to assess scale-up sce-narios, which is essential to provide reliable LCI-data about carbonfibre.
5. Conclusion and outlook
Carbonfibre production involves energy intensive process steps involving extensive heat treatment and relatively low processing speeds that result in long cycle times. Since energyflow related information about the carbon fibre production is rare and not comprehensive, this paper contributes to increasing the trans-parency of energyflows in the carbon fibre process chain on a gate-to-gate perspective. After reviewing available life cycle inventory data and energy related information, a methodology has been presented for the energy flow modelling of carbon fibre manufacturing. The methodology has then been applied in a case
Fig. 16. Energy hotspots in the precursor line for the individual energy services.
Fig. 17. Energy hotspots in the carbonfibre line split into energy services.
study, exposing the energyflows within the production PAN based precursor and carbon fibre. The examined precursor line is a research line scale with small output, while the carbonfibre line is pilot line scale that can produce about 110 t per year in an ideal production scenario. Adapting the approach of the extended energy value stream mapping on carbon fibre manufacturing, relevant process steps and energy consumers are identified and charac-terised regarding their energy and resource demands. In order to embed the results into a holistic factory understanding, energy consumers are allocated to different peripheral levels of a pro-duction system and technical building services are also part of the analysis. The evaluation is enhanced by measuring energy demands during regular operation and ramp-up phases. Moreover, as energy consumers were allocated to the peripheral levels of the carbon fibre production system, a differentiation between value adding and non-value adding energy demands were made.
The results of the paper can be seenfirst as a valuable step to-wards a comprehensive LCI dataset of carbonfibre manufacturing. Second, the high transparency of energy data provides a starting point for future research to increase the energy efficiency of carbon fibre manufacturing. This is especially valuable information, as research activities in the context of carbon fibre manufacturing mainly focus on the process itself, while often neglecting technical building services. In summary, the results indicate that the sole focus on process-related energy demands does not represent the true energy demand of carbon fibre production. Therefore, the scope of carbonfibre related energy data and life cycle inventories should be extended to energy demands from the technical building services. To the best knowledge of the authors, this is thefirst paper to present highly-detailed energy flow data of carbon fibre pro-duction that compiles with a systematic holistic factory under-standing. Energy assessments should include natural gas besides electricity and pursue a holistic perspective including the material yield as well. In comparison to previous work on the energy assessment in carbonfibre manufacturing, this paper opens up the black-box perspective, of how energy intensities of carbon fibre production have been reported previously. This paper extends the scope of previous work horizontally by a wider system boundary including not only direct production equipment but also technical building services into the energy assessment. Previous work is also extended vertically by a more detailed analysis that included each individual energy consumer into the data acquisition. To summa-rise, having the limitations in mind (Section4), this paper provides novel and highly-detailed empirical energy related data about the carbonfibre process chain.
In order to overcome above mentioned limitations, future research work could focus on enhancing the model-based under-standing between process parameters and the energy intensity of carbonfibre manufacturing. An energy-oriented dynamic simula-tion model of carbonfibre manufacturing could have the potential to address the above mentioned dynamics and interdependencies. The model would be used to assess different production scenarios, future process changes and improvement measures. Moreover, an enhanced model-based understanding is expected to ease scaling up energyflows and life cycle inventory data on industrial scale production.
CRediT authorship contribution statement
Antal Der: Conceptualization, Methodology, Formal analysis, Investigation, Visualization, Writing e original draft. Nikolas Dilger: Conceptualization, Methodology, Formal analysis, Investi-gation, Visualization, Writinge original draft. Alexander Kaluza: Conceptualization, Visualization, Writing e review & editing. Claudia Creighton: Conceptualization, Validation, Resources,
Supervision, Writinge review & editing. Sami Kara: Conceptuali-zation, Supervision, Writinge review & editing. Russell Varley: Conceptualization, Validation, Resources, Supervision, Writing e review & editing. Christoph Herrmann: Conceptualization, Su-pervision, Funding acquisition, Writing e review & editing. Sebastian Thiede: Conceptualization, Supervision, Funding acqui-sition, Writinge review & editing.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
The authors gratefully thank the German Federal Ministry of Education and Research (BMBF) for funding this research work in the project Open Hybrid LabFactory - Asia Pacific. The authors would like to express their gratitude towards Carbon Nexus at Deakin University for enabling and supporting this research work. References
Alvandi, S., Li, W., Sch€onemann, M., Kara, S., Herrmann, C., 2016. Economic and environmental value stream map (E2VSM) simulation for multi-product manufacturing systems. Int. J. Sustain. Eng. 9, 354e362. https://doi.org/ 10.1080/19397038.2016.1161095.
Arnold, U., De Palmenaer, A., Brück, T., Kuse, K., 2018. Energy-efficient carbon fiber production with concentrated solar power: process design and techno-economic analysis. Ind. Eng. Chem. Res. 57, 7934e7945.
Bogdanski, G., Sch€onemann, M., Thiede, S., Andrew, S., Herrmann, C., 2013. An extended energy value stream approach applied on the electronics industry. In: Emmanouilidis, C., Taisch, M., Kiritsis, D. (Eds.), Advances in Production Man-agement Systems. Competitive Manufacturing for Innovative Products and Services. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 65e72.
Chand, S., 2000. Review Carbonfibers for composites. J. Mater. Sci. 35, 1303e1313.
https://doi.org/10.1023/A:1004780301489.
Chen, J., Wang, C., Ge, H., Bai, Y., Wang, Y., 2007. Effect of coagulation temperature on the properties of poly(acrylonitrile- itaconic acid)fibers in wet spinning. J. Polym. Res. 14, 223e228.https://doi.org/10.1007/s10965-007-9101-2. Dai, Z., Shi, F., Zhang, B., Li, M., Zhang, Z., 2011. Effect of sizing on carbonfiber
surface properties andfibers/epoxy interfacial adhesion. Appl. Surf. Sci. 257, 6980e6985.https://doi.org/10.1016/j.apsusc.2011.03.047.
Das, S., 2011. Life cycle assessment of carbonfiber-reinforced polymer composites. Int. J. Life Cycle Assess. 16, 268e282. https://doi.org/10.1007/s11367-011-0264-z.
Das, S., Warren, J., West, D., Schexnayder, S.M., 2016. Global Carbon Fiber Composites Supply Chain Competitiveness Analysis. Report No. ORNL/SR2016/100 -NREL/TP-6A50-66071.
De Vegt, O.M., Haije, W.G., 1997. Comparative Environmental Life Cycle Assessment of Composite Materials. Citeseer.
Der, A., Kaluza, A., Kurle, D., Herrmann, C., Kara, S., Varley, R., 2018. Life cycle en-gineering of carbonfibres for lightweight structures. Procedia CIRP 69, 43e48.
https://doi.org/10.1016/j.procir.2017.11.007.
Duflou, J.R., Deng, Y., Van Acker, K., Dewulf, W., 2012. Do fiber-reinforced polymer composites provide environmentally benign alternatives? A life-cycle-assessment-based study. MRS Bull. 37, 374e382. https://doi.org/10.1557/ mrs.2012.33.
Edie, D.D., 1998. The effect of processing on the structure and properties of carbon fibers. Carbon N. Y. 36, 345e362. https://doi.org/10.1016/S0008-6223(97) 00185-1.
Edtmayr, T., Sunk, A., Sihn, W., 2016. An approach to integrate parameters and in-dicators of sustainability management into value stream mapping. Procedia Cirp 41, 289e294.https://doi.org/10.1016/j.procir.2015.08.037.
Erlach, K., 2013. Energy value stream: increasing energy efficiency in production. In: Future Trends in Production Engineering. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 343e349.https://doi.org/10.1007/978-3-642-24491-9_34. Faulkner, W., Badurdeen, F., 2014. Sustainable value stream mapping (Sus-VSM):
methodology to visualize and assess manufacturing sustainability performance. J. Clean. Prod. 85, 8e18.https://doi.org/10.1016/j.jclepro.2014.05.042. Griffing, E., Overcash, M., 2009. Carbon fiber HS from PAN [UIDCarbFibHS]. Contents
of factory gate to factory gate life cycle inventory summary [WWW Document]. Chem. Life Cycle Database. https://businessdocbox.com/Green_Solutions/ 67164362-Carbon-fiber-hs-from-pan-uidcarbfibhs.html.
Gupta, A.K., Paliwal, D.K., Bajaj, P., 1991. Acrylic precursors for carbonfibers. Polym. Rev. 31, 1e89.https://doi.org/10.1080/15321799108021557.
Herrmann, C., Dewulf, W., Hauschild, M., Kaluza, A., Kara, S., Skerlos, S., 2018. Life cycle engineering of lightweight structures. CIRP Ann. 67, 651e672.https:// doi.org/10.1016/j.cirp.2018.05.008.
Herrmann, C., Thiede, S., 2009. Process chain simulation to foster energy efficiency in manufacturing. CIRP J. Manuf. Sci. Technol. 1, 221e229. https://doi.org/ 10.1016/j.cirpj.2009.06.005.
Hesselbach, J., Herrmann, C., Detzer, R., Martin, L., Thiede, S., Lüdemann, B., 2008. Energy Efficiency through optimized coordination of production and technical building services. In: 15th Conf. Life Cycle Eng (Sydney Aust).
Huang, X., 2009. Fabrication and properties of carbonfibers. Materials (Basel) 2, 2369e2403.https://doi.org/10.3390/ma2042369.
Kellens, K., Dewulf, W., Overcash, M., Hauschild, M.Z., Duflou, J.R., 2012. Method-ology for systematic analysis and improvement of manufacturing unit process life-cycle inventory (UPLCI)-CO2PE! initiative (cooperative effort on process emissions in manufacturing). Part 1: methodology description. Int. J. Life Cycle Assess. 17, 69e78.https://doi.org/10.1007/s11367-011-0340-4.
Li, W., Thiede, S., Kara, S., Herrmann, C., 2017. A generic sankey tool for evaluating energy value stream in manufacturing systems. Procedia CIRP 61, 475e480.
https://doi.org/10.1016/j.procir.2016.11.174.
Liddell, H., Dollinger, C., Fisher, A., Brueske, S., Carpenter, A., Cresko, J., 2017. Bandwidth Study on Energy Use and Potential Energy Saving Opportunities in U.S. Glass Fiber Reinforced Polymer Manufacturing.
Liddell, H.P.H., Brueske, S.B., Carpenter, A.C., Cresko, J.W., 2016. Manufacturing en-ergy intensity and opportunity analysis forfiber-reinforced polymer composites and other lightweight materials. Proc. Am. Soc. Compos. - 31st Tech. Conf. ASC 2016.
Menghi, R., Papetti, A., Germani, M., Marconi, M., 2019. Energy efficiency of manufacturing systems: a review of energy assessment methods and tools. J. Clean. Prod. 240, 118276https://doi.org/10.1016/j.jclepro.2019.118276.
MInus, M., Kumar, S., 2005. The processing, properties, and structure of carbon fi-bers. JOM (J. Occup. Med.) 57, 52e58.
Morgan, P., 2005. Carbon Fibers and Their Composites. CRC press.
Müller, E., Stock, T., Schillig, R., 2013. Dual Energy Signatures Enable Energy Value-Stream Mapping, pp. 1603e1611.https://doi.org/10.1007/978-3-319-00557-7_ 129.
Newcomb, B.A., 2016. Processing, structure, and properties of carbonfibers. Com-pos. Part A Appl. Sci. Manuf. 91, 262e282. https://doi.org/10.1016/ j.compositesa.2016.10.018.
Papetti, A., Menghi, R., Di Domizio, G., Germani, M., Marconi, M., 2019. Resources value mapping: a method to assess the resource efficiency of manufacturing
systems. Appl. Energy 249, 326e342. https://doi.org/10.1016/ j.apenergy.2019.04.158.
Park, S.-J., 2015. Carbon Fibers, Springer Series in Materials Science. Springer Netherlands, Dordrecht.https://doi.org/10.1007/978-94-017-9478-7. Posselt, G., 2016. Towards Energy Transparent Factories, Sustainable Production, Life
Cycle Engineering and Management. Springer International Publishing, Cham.
https://doi.org/10.1007/978-3-319-20869-5.
Posselt, G., Fischer, J., Heinemann, T., Thiede, S., Alvandi, S., Weinert, N., Kara, S., Herrmann, C., 2014. Extending energy value stream models by the TBS dimension - applied on a multi product process chain in the railway industry. Procedia CIRP 15, 80e85.https://doi.org/10.1016/j.procir.2014.06.067. Rahaman, M.S.A., Ismail, A.F., Mustafa, A., 2007. A review of heat treatment on
polyacrylonitrilefiber. Polym. Degrad. Stabil. 92, 1421e1432. https://doi.org/ 10.1016/j.polymdegradstab.2007.03.023.
Reich-Weiser, C., Vijayaraghavan, A., Dornfeld, D.A., Reich-Weiser, C., Vijayaraghavan, A., Dornfeld, D.A., 2010. Appropriate use of green manufacturing frameworks. In: 17th CIRP LCE Conf, pp. 196e201.
Rother, M., Shook, J., 2003. Learning to See: Value Stream Mapping to Add Value and Eliminate Muda. Lean Enterprise Institute.
Schenk, M., Wirth, S., Müller, E., 2013. Fabrikplanung und Fabrikbetrieb: Methoden für die wandlungsf€ahige, vernetzte und ressourceneffiziente Fabrik. Springer-Verlag.
Schmidt, M., Raible, C., Keil, R., Gr€aber, M., 2007. Energy and material stream mapping. Recover. Mater. Energy Resour. Effic. Davos 7, 3e5.
Sch€onemann, M., Kurle, D., Herrmann, C., Thiede, S., 2016. Multi-product EVSM simulation. Procedia CIRP 41, 334e339. https://doi.org/10.1016/ j.procir.2015.10.012.
Suzuki, T., Takahashi, J., 2005. Prediction OF energy intensity OF carbonfiber reinforced plastics for mass-produced passenger cars. In: The Ninth Japan In-ternational SAMPE Symposium Nov.29e Dec.2, pp. 14e19, 2005.
Thiede, S., Posselt, G., Herrmann, C., 2013. SME appropriate concept for continu-ously improving the energy and resource efficiency in manufacturing com-panies. CIRP J. Manuf. Sci. Technol. 6, 204e211. https://doi.org/10.1016/ j.cirpj.2013.02.006.
Wiendahl, H.-P., Reichardt, J., Nyhuis, P., 2015. Handbook Factory Planning and Design. Springer Berlin Heidelberg, Berlin, Heidelberg.https://doi.org/10.1007/ 978-3-662-46391-8.
Witten, E., Mathes, V., Sauer, M., Kühnel, M., 2018. Composites Market Report 2018 -Market Developments, Trends, Outlooks and Challenges 59.
