Supply Chain Cost Analysis for Interior Lighting Systems Based on Polymer Optical Fibres Compared to Optical Injection Moulding

Abstract

Car interior design should evoke emotions, offer comfort, convey safety and at the same time project the brand identity of the car manufacturer. Lighting is used to address these functions. Modules required for automotive interior lighting often feature injection-moulded (IM) light guides, whereas woven fabrics with polymer optical fibres (POFs) offer certain technological advantages and show first-series applications in cars. In the future, car interior illumination will become even more important in the wake of megatrends such as autonomous driving. Since the increase in deployment of these technologies facilitates a need for an economical comparison, this paper aims to deliver a cost-driven approach to fulfil the aforementioned objective. Therefore, the cost structures of the supply chains for an IM-based and a POF-based illumination module are analysed. The employed research methodologies include an activity-based costing approach for which the data is collected via document analysis and guideline-based expert interviews. To account for data uncertainty, Monte Carlo simulations are conducted. POF-based lighting modules have lower initial costs due to continuous fibre production and weaving processes, but are associated with higher unit costs. This is caused by the discontinuous assembly of the rolled woven fabric which allows postponement strategies. The development costs of the mould generate high initial costs for IM light guides, which makes them beneficial only for high quantities of produced light guides. For the selected scenario, the POF-based module’s self-costs are 11.05 EUR/unit whereas the IM module’s self-costs are 14,19 EUR/unit. While the cost structures are relatively independent from the selected scenario, the actual self-costs are highly dependent on boundary conditions such as production volume.

1. Introduction

The automotive industry is globally one of the most important industries. In 2024, the global automotive manufacturing market size was estimated at USD 2.4 trillion, and this is predicted to double by 2034 [1]. In recent years, emerging technologies like alternative drive technologies have changed the landscape of automotive manufacturing substantially which has implications for interior design [2].

Against this background, interior illumination plays an ever-increasing role in the customer’s perception of the offered comfort of a vehicle [3]. This has a profound impact on the decision making of customers in regards to their buying behaviour, as well as on brand identity [4]. Furthermore, several functions are fulfilled by lighting: illumination of the car’s interior to facilitate orientation inside the vehicle, dynamic variation in the interior lighting depending on different scenarios occurring, improvement of the sense of security due to a better all-around visibility, and lastly a constant background illumination which keeps the driver awake as well as the passengers relaxed [5]. The scope of interior lighting in vehicles has changed from scenario-dependent ambient lighting to dynamically adapting atmospheric lighting [6].

Injection-moulded optics as shown in Figure 1a have been extensively utilised to provide a cost-effective and technologically viable solution to the customers’ needs of interior illumination. These injection mould parts have proven their worth in practical use as they can be produced in large quantities in a single-stage process with integrated coupling optics and micro-structuring [7]. However, woven fabrics with polymer optical fibres (POFs) as shown in Figure 1b hold distinct technical advantages. Due to their flexible nature, they offer increased drapability which is especially desirable in restricted spaces inside the car. These draping properties facilitate more flexible assembly and gluing processes. Also, the inherent thinness of the fabric reduces the required installation space and simplifies the possibility of an optional lighting package [8]. Furthermore, dynamic light effects are enabled using POF. Nonetheless, the prevailing use of injection-moulded parts is based on its very low production costs and the availability of manufacturing capacities all around the world. An example of a POF-based lighting from an Audi A3 is given in Figure 2.

The scenario to be reviewed in this paper is based on the exemplary and partially fictitious production of a small batch application of POF and injection moulding, respectively, in the European automotive sector for the sake of comparability. Two doors of a medium-sized car are to be fitted with illumination systems in which the fabric used for the interior cladding is woven from POF. The utilisation of both the POF fabric and the injection-moulded interior lighting are treated as special features as to make the costs for both solutions stand out more in the cost analysis. The choice of product example is relevant for certain manufacturing parameters, but does not prevent the later analytical generalisability of the derived cost structures for comparing the value chains of both technologies. Under the impression of this use case the research question to be answered is the following: “How do the cost structures in the supply chains for injection-moulded and fibre-based interior lighting modules differ?”

The approach selected to address the research question commences with the establishment of an exploratory comparative case study based on the framework conditions outlined in the preceding paragraph. To this end, a document analysis was conducted, and guideline-based expert interviews were undertaken to provide supplementary data and insights. A Monte Carlo simulation approach was then implemented to process the gained data as well as to account for potential uncertainties. The results obtained from these methodologies are subsequently utilised to perform a process cost analysis employing the activity-based costing (ABC) approach, alongside a scenario analysis. These steps facilitated a robust comparison between the two examined supply chains. Finally, all findings are compiled and managerial and practical implications are derived.

2. Theory

2.1. Case Study

To find satisfactory answers to the formulated research question, the overarching methodology being employed is that of a case study. A case study is an empirical tool for the handling of explorative research questions and utilises questionnaires, archive analysis, documents and interviews as sources of data [9]. In that context, explorative is defined as the first-time collection of data on a specific subject [10]. If several case studies are compiled into one, the resulting case study is categorised as comparative [11]. Comparative case studies are widely used in the area of supply chain and accounting [12,13,14,15,16]. In the wake of these characteristics and the circumstances of the case study in question, an exploratory and comparative case study was deemed to be most fitting. Especially since fibre-based illumination models are not properly fully established in the market, quantitative data was sparsely available and facilitated the need for a qualitive analysis using documents and expert interviews.

2.2. Light Guides

The main functionality of automotive interior lighting is illumination, with transporting the brand identity and catering to the overall design being supplementary functions [17]. In that context, light guides are transparent components which can transmit optical power over certain distances, often via total internal reflection [18]. Light guides can be produced inexpensively through injection moulding [7] or via continuous extrusion or melt-spinning, e.g., in the form of polymer optical fibres [19]. In either case, optical polymers are widely used for this purpose as they are rather cost-effective, lightweight and mechanically robust, as well as easily processable [20]. Additionally, these materials provide indispensable transparency in the transmissive spectrum and an amorphous inner structure to enable light guidance [21,22]. Light guiding systems rely on the same combination of components in both cases: a light source (nowadays almost exclusively LEDs), coupling, light guiding element and the light distributing decoupling [23]. While polymer optical fibres offer distinct advantages as mentioned before, the current oligopoly of 3 to 5 major POF manufacturers, all located in Asia, cause an obstructive price rigidity [19,24].

2.3. Activity-Based Costing (ABC)

To assess the cost structure of injection moulding, as well as melt-spinning of POF, the activity-based cost theory (ABC) has been applied along the supply chain of said illumination components. ABC is a kind of cost calculation which is based on a process-oriented approach and aims to allocate overhead costs according to their origin [25]. The main advantages of the utilisation of ABC are the more realistic assignment of general costs to the supply chain [26] and the possibility for the easy detection of non-value activities as well as responsibilities [27]. For ABC approaches, activities represent the most detailed level into which a process is broken down [28]. The activities are often categorised hierarchically according to their resource consumption. This includes unit-level activities (direct labour time, materials, machine hours, etc.), batch-level activities (machine set-up, logistics, order processing, etc.), product-sustaining activities (product and process development, design changes, etc.) and facility-sustaining activities (site management, building maintenance, heating, etc.) [29].

In general, ABC can be summarised in five steps which are as follows:

• Determination of processes and activities;
• Selection of cost drivers for each activity;
• Assignment of general costs to processes via resource drivers;
• Cost rate determination;
• Process cost determination [30,31].

The result of these steps is a comparatively precise account of either the production costs or the self-costs per production unit, depending on if the costs for research, development, marketing and administration have been accounted for in the calculation [32]. In Figure 3, steps 3 to 5 are shown for two exemplary resources and one exemplary activity.

In step 4, the cost rate CR_A of activity A is calculated according to Equation (1) as the product of the resource driver R_A,i of resource i multiplied by the respective cost rate RC_i of resource i. In the case of a calculation for a product, the cost drivers of all activities required for this product are determined. The total process overheads for the product α are calculated according to Equation (2) by adding up all products from cost driver C_A and cost rate CR_A of the respective activity A.

$${\mathrm{CR}}_{\mathrm{A}}=\sum _{\mathrm{i}\in \mathrm{A}}{\mathrm{R}}_{\mathrm{A},\mathrm{i}}·{\mathrm{Ri}}_{\mathrm{i}}$$

(1)

$${\mathrm{Total} \mathrm{overhead} \cos \mathrm{t}}_{\mathsf{\alpha }}=\sum _{\mathrm{A}\in \mathsf{\alpha }}{\mathrm{C}}_{\mathrm{A}}·{\mathrm{CR}}_{\mathrm{A}}$$

(2)

Since ABC was applied to a generic supply chain in the automotive industry as described in the first section, a short definition of this system shall be given as well. A supply chain (SC) can be defined as a combination of at least three enterprises which are participating in a flow of products, services, finances or information to cater to a consumer [33]. The approach to holistically plan and control these cross-company material, information and value streams for the purpose of increasing productivity, performance and profitability is called supply chain management (SCM) [34]. In the wake of mass customisation in the automotive industry, SCM in particular is required to deliver an appropriate response to these new challenges [35]. Since suppliers account for up to 75% to 80% of the value creation [36], and given the steep incline in the complexity of supply chains, their overall costs also need to be taken into consideration [37].

2.4. Supply Chain (SC) Uncertainty and Monte Carlo Simulations (MCS)

In the past, it has been observed that the costs of SCs were not allocated correctly to the processes involved; this leads to the costs not being activity-based and the level of aggregation being too abstract. To combat this issue, it was postulated to utilise ABC for SCs for the sake of enormous cost savings even though it was originally developed for in-enterprise usage [38]. Aside from cost savings, the transparency of the profitability of the SC, more precise cost information and in most cases higher profits are the main advantages to look at in supply chain costing [39]. The fundamental driver for the employment of ABC in SCs is the standardisation of cost structures and controlling data [40].

In order to process these complex and uncertainty afflicted datasets, Monte Carlo simulation (MCS) is the tool of choice in the presented scenario. This approach aims to illustrate reality using random samples generated from distributions with probabilistic traits [41]. The data derived from ABC can be outdated or rough estimates performed by workers which makes process costs subject to uncertainty [42]. As it is the case with any uncertain set of data, a probability distribution must be assigned. In the case of automotive interior lighting, two different distributions have been selected:

• Equal distribution: the same probability of occurrence is assigned to any value between a maximum value and a minimum value.
• Triangular distribution: the probability density function is a triangle defined by a minimum, a maximum and a most probable value [43].

Other distributions suffer from difficult estimations of parameters or being too uncommon to be known by interview partners, which is why they are not considered in this paper. MCS runs through a huge number of simulations, each time assigning a new value to an uncertain parameter according to the chosen distribution. This yields the cost per process step and in return the distribution of general costs [44].

2.5. Literature Review and Gap Analysis

Injection-moulded parts have been analysed economically with different focuses and via different methods. Karania et al. highlighted the high initial investment for injection moulding as a critical factor in the cost structure while comparing it with other processes such as fused deposition modelling [45]. Liebig and Mehler compared the cost structure for injection-moulded parts for production plants in different parts of the world [46]. The direct costs of different polymers for injection-moulded optical structures, as well as different injection-moulding techniques like compression or micro-injection moulding, have been compared by Burgdorf and Chi et al. regarding their impact on production costs [47,48]. Kazmer et al. focused on sustainability costing aspects like energy consumption and carbon footprint [49]. Multiple cost estimation methods for injection moulding have been studied, such as empirical costing [50], complexity-based parametric methods [51,52] and data-driven AI-based methods [53,54,55], as well as ABC approaches, which are summarised in

Table 1. ABC in the field of plastic injection-moulding.

Reference	Contributions	Limitations
Ruiz de Arbulo et al. [56]	Comparison of time-driven ABC (TD-ABC) and Value Stream Costing (VSC) for an automotive injection-moulding company in Spain TDABC gives accurate product costs, VSC an average value	Only time as cost driver No details on activity level published Case study refers to a boot locking knob, no relation to lighting
Park and Simpson [57]	Development of an ABC system to map costs of individual parts within a product family Cost modularisation is relevant for product family design decisions	Focus on cost-effective product families Case study refers to an injection-moulded switch assembly, no relation to lighting
Canciglieri and Young [58]	Information mapping for product development, exemplary for an injection-moulding company Mapping depends on communicating domains like mouldability, cavity insert design and cavity manufacturing	Activity-based analysis without focus on costing Focus on product development for a plastic box as the case study product
Lou et al. [59]	ABC implementation in a Taiwanese injection-moulding company Traditional costing underestimates injection-moulding costs much more than costs for insertion or assembling	Case study refers to injection-moulded phones, no relation to lighting Activities in a low level of detail (only 7 activities) No details (interviews, etc.) published
Alaoui et al. [60]	Activity-based E-liability allocation for an injection-moulded food bowl	Activity-based analysis without focus on costing, but on greenhouse gas emission accounting for Life Cycle Assessment
Tekin and Gözlü [61]	Calculation of costs from manufacturing losses via ABC in a Turkish injection-moulding company Speed loss and electrical breakdown cause the highest costs	Case study refers to white goods, no relation to lighting Focus on losses such as downtime, malfunctions and breakdowns instead of single product costs

.

For the textile supply chain, different cost estimation methods have been described as well: analytical costing, analogical costing and parametric costing (e.g., technical cost models) [62,63,64,65]. In addition to these, ABC analyses have been carried out and published for weaving, see

Table 2. ABC analysis in the field of weaving.

Reference	Contributions	Limitations
Klimaitienė and Kundzelevičius [66]	Time-driven ABC (TD-ABC) for a weaving service company Recording activity durations only with reluctance on the part of staff Only 69% of potential working time is used for profitable activities	Weaving service company (various materials and designs are woven on request) Outside of automotive context Fibre production not considered
Al-Eas and Al-Ghabban [67]	TD-ABC for weaving and knitting for PET fabrics in a cost leadership environment in Iraq Different results for ABC and TD-ABC	No focus on weaving Low-cost fabrics (in contrast to POF fabrics for lighting) Activities in a low level of detail
Rendall et al. [65]	Analysis of cost management practices for multiple textile companies Mostly low understanding of actual overhead cost distribution	Results outdated (from 1998) No details regarding ABC data published Multiple textile fields included (no focus on weaving)
Todingbua [68]	Cost analysis for an Indonesian weaving company Craftsmen only count material costs (no labour or overhead costs) Craftsmen value preserving traditional weaving techniques more than profitability	Traditional weaving by hand (no automated loom) No details regarding single activities published
Tsai et al. [69]	Linear optimisation of profit via the product mix with boundary conditions regarding environmental impact Textile supply chain from PET yarn twisting to finished woven fabric High potential for cost savings and economic returns	Fibre production not considered Focus on model optimisation Case study relates to a weaving company with a very high degree of Industry 4.0 adoption (IoT, machine vision, AI, AR, ERP systems, etc.) No details regarding ABC data published

. Since POF-based lighting is relatively new and less widespread, no analyses of cost structures have been published to date. Cost and market information on POF in general can be found in [19].

There are more ABC case studies for injection moulding than for weaving, see Table 1 and Table 2. Most of them are either not related to an automotive setting, lack a suitable level of detail and published data or are outdated. Fibre-production via melt-spinning has not been part of any ABC study.

To the best knowledge of the authors, there is no comparative cost analysis of comparable injection-moulded and fibre- or fabric-based products in general. Due to the absence of any cost analysis for lighting applications with POF fabrics or injection-moulded light guides, there is also no comparison for lighting applications in specific. This article aims to bridge this research gap to provide a comprehensive understanding of the cost structures in these supply chains.

3. Methodology

3.1. Calculation of Self-Costs

Self-costs are the total expenses incurred by a company in the production of goods or services, including both direct and indirect costs. Here, they are calculated as in Equation (3), with transport costs and customs only being applied for the POF spinning process due to the assumptions made. It can be used as a long-term lower price limit for the product [32].

$$\begin{array}{c}\mathrm{direct} \mathrm{costs} \mathrm{per} \mathrm{unit}\\ + \mathrm{variable} \mathrm{indirect} \mathrm{costs} \mathrm{per} \mathrm{unit}\\ + \mathrm{fixed} \mathrm{indirect} \mathrm{costs} \mathrm{per} \mathrm{unit}\\ + \mathrm{transport} \mathrm{costs} \mathrm{and} \mathrm{customs}\\ = \mathrm{self-costs} \mathrm{per} \mathrm{unit}\end{array}$$

(3)

3.2. Data Collection from Documents and Expert Interviews

Data collection is one of the challenges for this case study since supply-chain internal prices and cost structures are not publicly available. As many values as possible were determined on the basis of documents such as publicly available prices or technical data sheets.

A matrix of needed data (derived from the process chains in Section 4.1) is used as a guideline for the document analysis and interviews. Data sheets, price scales, comparison portals, depreciation tables and customs duties can be found on the providers’ websites. For further information that could not be clearly determined through the expert interviews, quotations were requested from companies. Further information was determined by means of a standard internet search. However, this information was only used when given by multiple sources or estimated to be plausible during the expert interviews. Technical parameters were taken from scientific publications.

However, the majority of the data had to be determined during expert interviews. Interviewees count as experts if they have clear, retrievable knowledge in the limited field of expertise. Criteria can be privileged access to information or responsibility for planning, implementing or testing a solution to a problem [70]. The aim is to interview at least one expert for each of the main processes based on the process chains. In order not to be dependent on one opinion, it is preferable to have multiple experts for the main processes. A total of five expert interviews were conducted as video conferences.

In addition to opening the interview and clarifying legal conditions (e.g., recording and publication of the interview), the expertise of the interviewees was requested at the beginning of the interview for verification purposes. This is followed by a formal introduction by the interviewer regarding the central research question, the case under consideration (door lighting module as an example of automotive interior lighting) and the methodology (ABC). In particular, the definition and function of resource and cost drivers are defined, as these are of great importance in the further course of the interview and the experts’ prior knowledge cannot be assumed.

In the second part of the interview, the actual process costs are determined. Depending on the expertise of the interviewees, questions are only asked [9,10] in their respective fields. This initially includes the direct costs, e.g., purchase prices for the polymer. In addition, fixed indirect costs with their respective processes are also asked, e.g., an estimate as a proportion of annual turnover with regard to resources not considered in the process chain, such as marketing in general or pre-development costs for injection moulds, as the process costs are only considered from SOP onwards. Next, the parameters required to calculate the production resource drivers are requested, e.g., for weaving the weft density or for injection moulding the number of cavities in the mould. The uncertainty represented by the distribution of parameters relates primarily to process variables, as certain process settings can be selected differently or fluctuate within process limits. In the case of the non-process parameter-dependent resource drivers, which are queried next, the uncertainty represented by the distribution relates in particular to process-inherent uncertainties that arise, for example, from different employees. When asking about resource drivers, the cost driver of the activity is always named, as the value of the resource driver to be estimated relates to the cost driver.

The third part of the interview asks about future developments and scenarios. To this end, an assessment of the functional differences between the two fibre optic systems is first obtained with an open question. This question does not primarily serve to gain knowledge, but opens up the discussion to open-scenario thinking. Possible answers from the experts could, for example, relate to the product development and manufacturing process, the areas of application or functionalities, as well as durability or disposal and recycling costs. In the next step, an assessment is requested as to which processes or process sections have the greatest need and/or the greatest potential for reducing the self-costs in the future. The expert is then asked to name further scenarios by asking about further changes. The procedure is repeated until no further scenarios with high relevance or probability of occurrence can be named. At the end of the interview, the interviewer is asked once again openly whether the expert has anything to add that has not yet been answered to any of the questions.

The complete interview guide for the expert interviews and the transcription of the interviews in German and the translation into English can be found in the Supplementary Materials S1.

For the second part of the interview, the content analysis of the interviews is limited to the collection and comparison of the largely quantitative data in the matrix proposed by Yin and Eisenhardt [9,10]. For the more qualitative data from the first and, in particular, third part of the interview, the responses of all interviewees are inductively categorised and structured using text-related headings, following Mayring [71].

3.3. Monte Carlo Simulations

Based on the data from the document analysis and in the second part of the expert interviews, the required resources with resource drivers and cost rates with cost drivers with a probability distribution can be determined or calculated for each process step of the supply chains. These are used in the Monte Carlo simulation to determine the distribution of the total self-costs of a component or a specific fibre length.

The number of simulation runs can be determined in an evolutionary-empirical consideration of the mean square pure error of median (MSPE_MED) and the mean square pure error of standard deviation MSPE_STDEV as a function of the number of simulation runs N, see Figure 4 [72].

The diagrams indicate that the course of the deviations in the simulation results changes only slightly with a further increase in the number of simulation runs from 30,000 and barely improves from approx. 500,000. For the “injection moulding” process chain, this already occurs with fewer simulation runs, which indicates fewer process steps and/or a lower fluctuation in their cost distributions. Due to the low simulation costs with a duration of less than 20 s, a conservatively estimated safety factor of 2 is used so that all simulations are carried out with 10⁶ runs.

4. Results

First the supply chains are derived and general parameters for the ABC are plausibilised. Then, the process costs, broken down into fibre spinning, weaving and injection moulding, are analysed and compared below. In the second part, two possible future scenarios are developed and their influence on the cost structures is analysed.

4.1. Derivation and Delimitation of the Supply Chains Under Consideration

The product development process in the automotive industry is complex and can involve more than 1000 interdisciplinary participants. After the design decision, the actual series development with all components takes two to three years until the start of production (SOP). Specifications are drawn up for the product development of all components of the car based on the approval criteria of the sales markets and the design specifications. Product development as a process is very individual for each product, each supplier and each car manufacturer and is not characterised by repetitive activities [73]. A process-orientated cost analysis is therefore only suitable to a limited extent.

Therefore, the process chain starting from SOP is considered below. No development and price negotiation processes are taken into account. It is assumed that all quotas have been negotiated and are only called up. All main and sub-processes for the fibre-based and injection-moulding-based supply chain are shown in Figure 5.

The main processes are subdivided into a maximum of four sub-processes in order to reduce the effort required through an appropriate level of detail in the process breakdown [74]. All single activities are shown in three event-driven process chains (EPCs) in the Supplementary Materials S2.

The process chains only include those main processes in which the supply chains differ. Polymethyl methacrylate (PMMA) is used as the light guidance material in both cases. The PMMA grade differs slightly for continuous and discontinuous extrusion processes. However, the polymer itself and the manufacturing process are the same in each case, meaning that polymer production and other upstream processes are not considered. The processes for further processing of the lighting module at the direct supplier or OEM may also differ. The main “module assembly” processes of the two supply chains are similar. Nonetheless, there may be differences with regard to component handling, functional testing due to the number of LEDs and additional mixed optics to be used for POF-based lighting, which is why this main process is taken into account.

Cost and resource drivers are determined for each activity from the established process chains in accordance with the procedure in Section 2.3. In some cases, costs and resource drivers are determined directly by means of document analysis or expert interviews; in other cases, they depend on process parameters. The determination of process parameters instead of the direct assessment of resource drivers is favoured, as the assessment of technical process parameters by the experts is considered more reliable. For example, the winding speed during spinning is a production parameter to be set, but the time required to produce a bobbin is not and therefore harder to estimate by experts. In addition, the reference of resource drivers to standardised process parameters results in consistent values, which is not the case when resource drivers are assessed separately and are mutually dependent. All process parameters, information on resources and activities associated with cost drivers and resource drivers can be found in the Supplementary Materials S3.

There are various manufacturing processes that are generally suitable for POF production [75]. The established spinning process for POF differs from the conventional melt-spinning process in that unpolymerised monomers remain in the spinning mass to reduce the viscosity during extrusion [76]. Higher productivity through melt-spinning, among other things, is considered in more detail in Section 5.1. In the fibre-based supply chain, it is also assumed that there is a company boundary between fibre production and weaving. This is plausible, as none of the five major POF manufacturers includes a weaving department. For both supply chains, it is assumed that module assembly takes place in-house.

4.2. Overall Parameters of ABC

The parameters listed in

Table 3. General parameters for the ABC.

Parameter	Value
Illuminated visible area	5 cm × 20 cm
Total number of units	60,000 units
Number of units for the order under consideration	4000 units
Annual production for the year under consideration	8000 units
Overhead factor for non-wage labour costs	1.7
	Spinning	Weaving	Injection moulding
Additional length due to coupling	–	10 cm	5 cm
Number of employees	50	100	100
Turnover	EUR 160 million	EUR 220 million	EUR 140 million
Production lines	4	50	30
Shift operation	4 shifts	3 shifts	3 shifts
Production days	351 d/a	256 d/a	256 d/a

are used for ABC. These are suggested or made plausible by the experts during the interviews. As POF lighting modules are just beginning to be established on the market, the comparison of injection moulding and POF is based on a small series in automotive manufacturing. A total quantity of 60,000 units is defined as a small-series scenario. This results from the registration of 30,000 cars of a model in Germany, such as the Peugeot 108, Citroën DS3 or Toyota Yaris Verso, or approximately four times the number in Europe [77,78]. The estimation of the European rather than the global market is explained by the typical separation of the American, European and Asian markets, for which suppliers usually differ. Two doors are fitted to each car. In the small car segment, an equipment rate of 25% is assumed for additional interior lighting equipment that increases the purchase price.

All personnel costs (including in production) are charged as overheads, as the costs are incurred due to continued wage payments even if the respective product is not produced. This is common practice in machine-intensive production [79]. Therefore, the direct costs only consist of the costs for materials and purchased parts. The personnel costs are based on standard wages in the respective industry (for a fictitious production site in Aachen, Germany). A distinction is made between salaries for technical and commercial employees. In each case, a middle level of the salary group and a middle level of seniority are assumed, with the exception of simple production employees for module assembly. For these, a lower salary group and short length of service are assumed. In addition to the direct labour costs in the form of gross wages, there are also non-wage labour costs (social security contributions such as employer contributions to pension, unemployment, long-term care and health insurance, voluntary or collectively agreed social benefits such as Christmas bonuses, childcare costs or company pension schemes, contributions to the employers’ liability insurance association and continued remuneration for sick days and other absences). For the latter costs in particular, only empirical values can be used. A factor of 1.7 is assumed as the most likely value for ancillary personnel costs, with a tolerance of 10% for the minimum and maximum values [80,81]. As the POFs currently used in the automotive industry are generally manufactured in Japan, the wages from the textile industry wage table are corrected with a factor of 0.587, which is derived from an international comparison of industrial labour costs from 2016 [82].

Large weaving companies, such as automotive suppliers, usually have a large number of machines, with individual weaving machines each having a product focus (e.g., rapier weaving machines for POF fabrics) [79]. The situation is similar for injection-moulding plants. According to the experts’ recommendation, 50 weaving machines and 15 injection-moulding machines are assumed in each case. Turnover was also estimated. A four-shift operation is assumed for fibre production, as it is generally not profitable to shut down the plant for continuous extrusion. Apart from one week of company holidays and one week of maintenance and servicing, this results in 351 operating days per year. Three-shift production is assumed for both the weaving mill and the injection-moulding plant. According to the interviewees, weekend shifts are not profitable, depending on the order situation. Assuming one week of company holidays, this results in 256 operating days per year for both.

4.3. Process Cost and Cost Driver Analysis for the Fibre-Based Supply Chain

Figure 6 shows the distribution of the self-costs for POF including shipping and customs costs to Germany. The mean value is 5.91 EUR/km with a standard deviation of 1.35 EUR/km. At 5.73 EUR/km, the median is lower than the mean. The distribution is not symmetrical, but slightly skewed to the right with an (empirical) skewness of 0.82. This is due to the asymmetrical triangular distributions of some parameters for POF spinning, e.g., the costs of the cladding polymer in the direct costs and the number of winding positions in the process parameters, which in turn skews the production speed and thus the influential cost driver of production “number of spools per month” to the right. The (empirical) kurtosis curvature of the distribution is 4.04 > 3, which means that the distribution is more steeply skewed than a normal distribution. Accordingly, the variance in the distribution results in more from rare, extreme events.

Compared to the net sales price for POF with a diameter of 0.25 mm of approx. 10.97 EUR/km (see individual costs of the weaving mill and Kröplin et al. [19]), the self-costs for the POF spinner are much lower. The mean value of the distribution results in a profit margin of 86% which was anticipated by one of the experts to be similarly high and also fits in with the price rigidity of the oligopoly on the POF market.

The second company in the fibre-based supply chain is the weaving mill. The distribution of self-costs for their product is shown in Figure 7. The mean and median of the cost price distribution is 11.05 EUR/unit, which can be regarded as the long-term lower price limit. The distribution is almost symmetrical with a skewness of 0.002, which matches the equality of the mean and median. This is due to the largely symmetrically estimated triangular distributions of the cost parameters for weaving and the upstream and downstream processes. The standard deviation is 0.34 EUR/unit. The kurtosis is 2.36 < 3, which corresponds to a flattened distribution compared to the normal distribution.

Due to the two-step production of spinning and weaving, one more company is involved in the fibre-based supply chain compared to injection moulding. This increases the SC coordination effort. The profit margin of the weaver is not considered here, but is estimated to be significantly smaller compared to the POF spinner and at most as large as that of OEMs in the automotive industry (approx. 7%) [83]. The different profit margins show the great market power of POF spinners on the one hand and the great cost-cutting potential on the other. Appropriate coordination of the supply chain, e.g., via contracts, volume discounts or revenue sharing, could potentially increase the SC’s overall profit [84]. A reduction in the margin of the POF spinner could lower the price for the weaver, who could then offer the fibre-based lighting module at a lower price, making it more profitable than injection moulding. As a result, the sales volume could be increased and the SC profit would be maximised. The coordination measure would need to ensure that the POF spinner would still benefit from a lower profit margin through higher sales.

4.4. Process Cost and Cost Driver Analysis for the Injection-Moulded Chain

Figure 8 shows the histogram of the relative frequencies for the self-costs of the lighting module with injection-moulded light guidance. The mean and median cost price is 14.19 EUR/unit. Similar to weaving, the skewness is very low at −0.0002 and the distribution is therefore almost symmetrical.

A kurtosis of 1.82 < 3 indicates an even more flattened distribution compared to weaving. In fact, the distribution shows a constant plateau between approx. 10.50 EUR/unit and 15.20 EUR/unit and resembles a uniform distribution in this area. This is also reflected in the standard deviation of 1.45 EUR/unit, which is more than three-times higher than for the fibre-based lighting module. One reason for the described plateau is the pre-development and manufacturing costs estimated as a uniform distribution, which are one of the main factors for injection moulding (~42% of the self-costs per unit). The pre-development includes an iterative design of the mould, which according to Altan et al. can account for up to 45% of the costs of an injection-moulded part [85]. This also fits with the findings of Cassettari et al. that uniform distributions lead to greater variations in the Monte Carlo results compared to triangular distributions [43]. The SCM also plays a role for the injection-moulding company, as suppliers also exist there, but are not considered here. The influence is estimated to be smaller compared to the spinning and weaving companies.

4.5. Comparison of the Cost Structures of the Two Competing Lighting Systems

The results of Section 4.3 and Section 4.4 show that the POF-based lighting module for the present case is on average 11.05 EUR/unit, which is 3.14 EUR/unit cheaper than the injection-moulded lighting module at 14.19 EUR/unit. This shows that the use of fibre-based lighting modules can be advantageous in the case of small series in the automotive industry, which is particularly cost-driven. Based on these findings, an answer to the research question, namely how the cost structures in the supply chains for injection-moulded and fibre-based interior lighting modules differ, can be given. To provide a comprehensive illustration of this matter, the self-costs of both light modules are broken down by type of cost, hierarchy level and process step in the following.

4.5.1. Breakdown by Direct and Indirect Costs

The direct costs for spinning account for 31% of the total self-costs. At 48%, the variable indirect process costs account for the majority. The direct costs for weaving account for 60% of the total costs. They are primarily made up of the cost of the LED module at 36.6%, the cover material at 18.0% and the POF at 4.2% of the total unit costs. Direct costs for injection moulding average only 37%, with the largest item being the LED module at 21.1%. The percentages are shown in Figure 9.

The share of variable and fixed indirect process costs is 27% and 14% for the weaver, while for injection moulding it is 9% and 54%. The high proportion of fixed costs (see “depreciation” in Figure 9) for injection moulding is due to the high initial costs for the mould. The small proportion of variable costs can also be attributed to the smaller proportion of labour costs, as more robot systems can be used. The higher proportion of labour costs for weaving is due to a lower degree of automation in production and assembly. It is roughly in line with the 20% share of labour costs calculated by Klimaitienė and Kundzelevičius for a weaving mill [66].

For weaving and injection moulding, the shares of “energy” and “other” are negligible. The energy costs of the spinning process account for 17.9% of the total fibre costs, highlighting the significant relevance of energy expenses in the spinning process. The category “other” is considerably higher for the spinning process at 25%, as it includes additional shipping costs (17.8%) and customs duties (2.9%).

4.5.2. Breakdown by Hierarchy Level

Similarly, the comparatively high proportion of batch costs for the spinning process can be justified as shown in Figure 10. This figure presents the self-costs per kilometre or unit according to the classification by Cooper and Kaplan at unit, batch, product and location levels [29].

Due to the high direct costs in all three process chains, the costs at unit level each represent a substantial share. The unit costs for weaving are notably higher due to the finishing process of the fabric as roll goods (see Figure 11). In injection moulding, the costs at product level are the highest, primarily because of the mould costs.

4.5.3. Breakdown by Process Step

Figure 11 shows Pareto diagrams for the main processes of the three companies, arranged according to the magnitude of variable indirect costs. For spinning, the actual fibre extrusion emerges as the most cost-intensive main process. This is consistent with the findings of Jiran et al., who analysed the production of hollow fibres using an ABC analysis [86]. At 1.99 EUR/km, this accounts for 34% of the self-costs of the POF at 5.91 EUR/km.

For injection moulding, the influence of the variable indirect process costs at 1.33 EUR/unit is relatively small. The two largest cost-causing main processes are module assembly (0.65 EUR/unit) and production (0.47 EUR/unit). For weaving, the variable process costs amount to 2.94 EUR/unit, with the actual weaving process contributing a negligible 0.02 EUR/unit. This is related to the small batch size considered and the high number of components that can be cut from the width of the fabric. In weaving, the costliest processes are production finishing (1.79 EUR/unit) and module assembly (0.90 EUR/unit). The module assembly is more expensive than in injection moulding because an additional mixing optic must be used before the POF bundle, which can be directly integrated into the injection-moulded part. In production finishing for weaving, this particularly includes the finishing of woven roll goods, encompassing cutting and surface treatment. In contrast, the result of the injection-moulding process is a ready-finished component with a micro-structured surface, which means that production finishing plays a minimal role there.

The production of standardised roll goods in weaving is a significant difference compared to injection-moulded light guides, as fibre-based lighting modules can be customised to the respective product during the finishing step. From a supply chain management perspective, this allows for component aggregation as an aggregation strategy. This corresponds to a postponement of product differentiation or variant formation and offers several advantages [87]:

• Higher forecast accuracy through aggregated planning for all activities prior to product differentiation, thereby reducing inventory levels and consequently holding and stockout costs;
• A risk-diversifying gain in information through disaggregated forecasts that are closer to the point of sale (later customer order decoupling point);
• Economies of scale for intermediate products (in this case: fibres and fabrics), whose demand is aggregated;
• A reduction in development costs, as the same parts or components are used.

In contrast, the costs associated with late variant creation represent a disadvantage. This is exemplified by the main process “production finishing” in weaving, which accounts for 61% of the variable costs and 14% of the total self-costs. This is the major reason why injection moulding can be advantageous when producing a high quantity of identical components. The high initial costs due to mould development are offset by the large production volume.

The usable lifespan of an injection mould in the automotive sector can be estimated to range between 100,000 and 250,000 cycles, significantly exceeding the number of cycles required for the total number of parts, which is 60,000 and is used for allocating the tool-related unit costs. This allows for a reduction in self-costs per unit for injection moulding at higher quantities. The marginal unit quantity at which injection moulding becomes more profitable than fibre-based lighting can be determined to be approximately 125,000 units. However, with larger quantities, other parameters also change (see Section 5.2). Furthermore, according to one of the experts, the initial costs in injection moulding scale disproportionately with component size due to more complex requirements. For tendentially larger illuminated areas, POFs are therefore advantageous or enable an economically viable production of light guide-based lighting solutions. This hinders the generalisability of the marginal unit quantity.

In addition to the generally lower development costs associated with postponement, the development cycles for fibre-based light guides are shorter than those for injection moulding according to the experts. This reduces costs for both the manufacturer and the customer. This is particularly advantageous in the automotive industry, as it allows for a shorter timeline for series development. It enables later design decisions, savings in the costly series development phase and quicker integration of employees who can be deployed in other projects [73].

By breaking down the self-costs by type of cost, hierarchy level and process step, a multi-layered answer to the research question was formulated. It has been shown that both technologies possess a unique cost structure. The indirect costs for the IM SC are much higher due to the depreciation of the mould. Broken down by hierarchy level, for the weaving SC the majority of self-costs appear on the unit level due to finishing whereas for the IM SC the majority of self-costs are on the product level due to the initial cost of the mould. To further explore these results, future scenarios are evaluated in the next chapter.

5. Discussion of Future Scenarios

Scenarios are a way of analysing and evaluating real future variants. Accordingly, a scenario analysis involves analysing and evaluating possible scenarios, meaning consistent possible future events or states. Assumptions about the future can be developed and important influencing variables identified [88]. Using open questions in guideline interviews with experts to develop scenarios is a well-established approach [89,90,91].

In the third part of the expert interviews, cost reduction potentials and other scenarios for both supply chains for light guide-based lighting were asked for. All scenarios given by the interviewees are categorised and evaluated in Figure 12. A discussion on their relevance is given in Appendix A. Two of the scenarios are calculated and discussed in detail.

5.1. Scenario of a Low-Cost POF

The scenario “reduction in the POF price” can be justified by various factors. Labour costs in Germany are on average 1.7-times higher than in Japan, which shows additional potential for reduction in self-costs [82]. This results in an increase in the price of fibre of approximately 7.2%. At the same time, however, customs duties on the total value are eliminated (2.9%) and shipping costs are drastically reduced, resulting in a cost reduction of approximately 17.9%. The calculation is based on an 18-tonne truck that can load 17 pallets, each of which can hold 120 spools with 12 km of fibre each. At a daily rate of 640 EUR, this corresponds to shipping costs of 0.0261 EUR/km. Further cost optimisation results from the production process itself. Melt-spinning as the most widely used process for fibre production offers great potential for price reduction for POF with acceptable quality losses for lighting applications. This is due to the following:

• Higher production speed for monofilament melt-spinning compared to a continuous extrusion process with in situ polymerisation [75] (70–150 m/min instead of 50–70 m/min);
• Greater competition on the monofilament melt-spinning market with more resilience for the automotive supply chain with multiple suppliers;
• Economies of scale in the polymer supplier’s supply chain.

This increasing competition among PMMA manufacturers is also leading to a reduction in price. Due to a lack of information, this is conservatively estimated at 2%, which reduces the fibre price by less than 0.1%. The self-costs for an injection-moulded part also decrease by the same amount. Furthermore, the profit margin is set at 10% due to potentially greater competition (minimum estimate by expert 4). Compared to OEMs, which have profit margins of around 7%, this estimate is not unrealistically low [83]. Self-costs are not affected by a lower profit margin. However, a comparison with the previous profit margin of 86% shows the enormous potential for price reductions. The inclusion of electricity costs for a change in production location from Japan to Germany is prevented by ambiguous data, as electricity in Japan can be 13% cheaper or 25% more expensive depending on the purchase quantity [92]. At 17.9%, the share of energy costs in the spinning process is comparatively high, which again makes the choice of production location relevant. At the same time, the choice of production location also depends in particular on transport costs to customers, which can make the production of POF in countries with comparatively high industrial electricity prices such as Japan and Germany plausible. The changes considered in comparison to Section 4.4 are summarised in

Table 4. Change in cost parameters in the “reduction in POF price” scenario and effects in relation to the original self-cost of 5.9096 EUR/km.

Change	Consequence in Cost Accounting	Effect on Average Self-Costs for POF
Production location Germany instead of Japan	Labour costs on average 1.7-times higher	6.3357 EUR/km	+7.2%
	No customs costs	5.7442 EUR/km	–2.9%
	Shipping costs decrease from 0.8768 EUR/km to 0.0261 EUR/km	4.8536 EUR/km	–17.9%
Melt-spinning with polymer pellets instead of continuous extrusion with in situ polymerisation	Increase in production speed to 70–150 m/min	5.3184 EUR/km	–10.0%
More competition and efficiency in PMMA production	PMMA price falls by 2% (conservative estimate)	5.9073 EUR/km	−0.04%
More competition in POF production	Reduction in profit margin to 10%	No effect on self-costs

.

A reduction in POF self-costs from 5.91 EUR/km to 4.43 EUR/km corresponds to a total reduction of 25% despite higher wage costs in Germany. In this scenario, the price for weaving would fall from 10.96 EUR/km to 4.88 EUR/km, a reduction of 55.5%. Further price reduction due to economies of scale is examined in more detail in Section 5.2. However, its potential self-cost reduction can be estimated by its increase in production volume by a factor of 600 ≈ 521 = 2⁹. The basic statement of the experience curve according to Henderson and the Boston Consulting Group is that for every doubling of the cumulative production volume, the inflation-adjusted unit costs related to value added potentially decrease by a constant percentage, e.g., 20% to 30. [93]. This corresponds to an expected reduction to a maximum of 0.8⁹ ≈ 13.5% of the original self-costs. According to Hall, the effect of the experience curve cannot be empirically limited to 20% to 30%, so deviations from the calculated reduction are to be expected [94]. In addition, a degressive experience curve is mentioned in expert interview 4, which is why a total reduction of only 25% to 50% may be more plausible.

The “low-cost POF” scenario illustrates that a significant reduction in the purchase price of polymer optical fibres is possible for the weaving mill as the producer of the fibre-based lighting module. The causes of the cost reduction can be technological, logistical and competitive in nature, with the reduction in profit margins due to the breakdown of an oligopolistic price rigidity offering the greatest potential.

5.2. Scenario of a Large-Scale Production

Again, European statistics for car model registrations are used to assume a typical annual registration figure of 225,000 registrations per year, as was roughly the case for the models like VW Tiguan, Renault Captur, Dacia Duster and Škoda Octavia [95]. Based on four lighting modules per vehicle, an eight-year model life and an equipment rate of 50%, this results in a total number of parts of 3.6 million and a monthly order of 37,500 units if distributed evenly. The higher equipment rate is due to the higher-class vehicles in the mid-range and SUV segments in this scenario. One of the experts predicts that, in general, more interior lighting will be integrated as standard in the future. For the injection mould, four cavities are assumed for large-scale production, i.e., four components are produced per injection cycle. The cost of the mould is assumed to be twice as high as in the simpler case. The parameters that have changed compared to small-scale production are summarised in

Table 5. Change in cost parameters in the scenarios “large-scale production” compared to the small-scale production.

	Large-Scale Production	Small-Scale Production
Lighting modules per vehicle	4 units/vehicle	2 units/vehicle
Equipment rate	50%	25%
Total number of parts	3,600,000 units	60,000 units
Part number for single order	37,500 units/order	4000 units/order
Annual production	450,000 units/a	8000 units/a
Number of cavities in the injection mould	4	1

. The hypothetically reduced POF price from Section 5.1 is not used here, but the currently available price.

Here, the self-costs of the fibre-based lighting module fall from an average of 11.05 EUR/unit to 9.34 EUR/unit (reduction of 15.5%). The majority of the reduction is achieved in the variable indirect process costs and the costs at batch and product level. The unit-related direct costs and fixed indirect process costs decrease only slightly. At 9.28 EUR/unit, the unit costs are almost exclusively at the unit level. Energy and personnel costs hardly change, which in the latter case is due to the unchanged main processes of production follow-up (i.e., packaging) and module assembly.

For the injection-moulded lighting module, the average self-costs fall from the original 14.19 EUR/unit to 8.19 EUR/unit (reduction of 42.3%). The fixed indirect costs increase due to the doubled mould price and the triple wear-related reworking of the mould, which accounts for 10% to 50% of the mould price. Nevertheless, the enormous reduction can be attributed in particular to the distribution of the high initial costs of the mould over the 60-fold number of parts. Specifically, the unit-related fixed process costs decrease from 7.64 EUR/unit to 2.25 EUR/unit. The costs at the product level also decrease accordingly and account for only 27.5% of the self-costs instead of 52.9%. Here too, the unit costs and thus in particular the module assembly cause the costs. Similarly, depreciation costs fall from an average of 7.72 EUR/unit to 0.31 EUR/unit for the same reason. As with the fibre-based lighting module, the unit costs fall slightly. The variable process costs fall from 1.33 EUR/unit to 0.79 EUR/unit, which is significantly more than for the fibre-based lighting module. This can be attributed in particular to the fourfold increase in parts production per injection-moulding cycle. This is also reflected in the fact that the process costs of the main “production” process fall from an average of 0.47 EUR/unit to 0.12 EUR/unit. If the single cavity from the small series was retained, the variable process costs would rise again by 0.40 EUR/unit and the total self-costs would rise accordingly to 8.60 EUR/unit. Again, energy and personnel costs change only slightly.

The scenario “large-scale production” shows that the unit costs for injection moulding can fall much more sharply than for weaving due to the high initial costs. Nevertheless, it remains true that fibre-based lighting can also become relevant in larger series due to technological differences. In the expert interviews, the following technological advantages for POF-based lighting modules were mentioned:

• Flexibility: textile drapability allows for other applications, reduces space restrictions;
• Installation situation: makes mounting concepts more flexible and simplifies bonding;
• Dynamics: enables brightness and colour dynamics by controlling individual fibre bundles;
• Less restrictions: also enables very large light guides (injection moulds become disproportionately expensive and the light guides become too thick);
• Component thickness: reduces installation space requirements.

The following advantages are mentioned for injection-moulded light guides:

• Higher brightness: tends to reduce the required number of LEDs;
• Single-material purity: increases recyclability, reduces disposal costs.

6. Conclusions

Using the product example of a fibre-based and an injection-moulded interior lighting module for ambient lighting in automobiles, the differences in the cost structures of their supply chains is investigated in a process-oriented manner. Activity-based costing allows for a high degree of cost transparency for supplier- and customer-specific product and pricing policies in supply chains typical of the automotive industry, where the depth of value added for OEMs is decreasing and overhead costs are therefore increasing.

Injection moulding is characterised by high initial costs resulting from the fixed costs for development and manufacturing of the injection mould, combined with comparatively low variable costs (see Figure 13a). The fibre-based supply chain—consisting here of POF spinning and weaving—is characterised by low initial costs and medium unit costs, which scale relatively little with production volume (see Figure 13b).

Therefore, from a purely cost-based view, fibre-based lighting modules are advantageous for production runs with lower quantities compared to injection moulding as the established commercial practice for light guides. In the case study examined, the break-even point of produced units up to which the textile production is advantageous in terms of costs is found to be 125,000 units. The disproportionate scaling of initial costs in injection moulding with the size of the light guide (see Figure 13c) component forbids us from generalising this break-even point. For the textile supply chain, technology (i.e., spinning process), transport (i.e., location of production) and supply chain management measures (i.e., postponement) can substantially reduce the fibre purchase price and thus the unit costs for lighting module.

Derived from these insights, the following managerial and practical implications can be formulated.

-

Cost transparency for investment decision: companies at the end of the automotive supply chain can consider the costing differences when deciding for a lighting technology route.

-

Strategic sourcing decisions for increased flexibility in production: maximum exploitation of advantages of the respective technology from a supply chain point of view.

-

Future scenario planning in the form of innovation opportunities: identification of potential future technological advancements to further reduce costs or improve functionalities.

All shown costs are based on estimates and, despite being verified by multiple experts, may differ from reality. They also may change in the future. Certain general overhead costs that are not caused by the considered resources (e.g., expenses for the management board and artificial overhead costs) are not directly considered, but estimated using a potentially inaccurate surcharge factor. This might distort the absolute values of the long-term lower price limit. In addition, the same boundary conditions were chosen for both supply chains (not differences in public levies, corporate structures, etc.). As Eisenhardt and Graebner describe, even a case study with only a few cases considered allows for generalisable conclusions to a certain extent [96]. Fundamental differences such as continuous production in spinning and weaving with subsequent finishing (later decoupling point) and production of ready-to-deliver components (earlier decoupling point) in injection moulding apply across company boundaries in a technology-specific manner. Thus, the inductive conclusions drawn from the case study for the qualitative comparison of cost structures are also valid beyond the case of door panel lighting considered.