Identifying modular candidates in engineer-to-order companies

Abstract

The purpose of this paper is to address a gap of missing modularization methods for engineer-to-order (ETO) companies. The research project was initiated by clarifying the challenges facing ETO companies, based on these challenges synthesis of existing methods was done to conceptualize a method. This article presents the modular candidate identification (MCI) method aimed at identifying modular candidates in ETO companies. The method analyzes five dimensions, namely, market segments, customer requirements, product architectures, cost and lead time to find modular candidates. The method was applied in a Danish ETO company and shown to be successful in identifying two modular candidates. Both were recognized by management and redesigned in modular product development projects.

Keywords

product architecture modularity modularization product development ETO engineer-to-order concurrent engineering

Introduction

In today’s competitive market, customers demand more customized products at a lower cost and with faster lead time (Hvam, 2008). One way to achieve this is to develop modular product architectures that reduce internal complexity while maintaining external variety to satisfy market needs (Meyer and Lehnerd, 1997; Otto et al., 2016; Simpson, 2006). Equally, the decoupled interfaces of modular product architectures facilitate concurrent engineering and thereby bring about significant time savings through the parallel development of modules (Prasad, 1999).

Based on our engagement with several engineer-to-order (ETO) companies as part of a larger research project focused on strengthening the manufacturing ecosystem in the Danish industrial sector, we found that companies wanted to begin developing modular product architectures but first needed detailed knowledge to make informed decisions about which products and product families to include in modular product architecture initiatives.

The literature has an abundance of methods and successful case studies on modular product architectures, platform-based product development and modular product family design (Johnson and Bröms, 1995; Otto et al., 2016; Pirmoradi et al., 2014). In terms of ETO-specific literature, however, there are only limited examples of methods and successful case studies (Gepp et al., 2016). A reason for this is the lack of information available to decide which products to include in future modular initiatives (Bertram et al., 2020). Unaddressed problems such as complex product structures, open solution space of future variants and product portfolios consisting of prior customized products are described as reasons for the research gap in ETO-specific methods and cases (ElMaraghy et al., 2013; Fisk, 2004; Willner et al., 2016).

In this article, we aim to adress this research gap by presenting the modular candidate identification (MCI) method to help identify modular candidates using data from previous product development projects to help companies make informed decisions about which products to include for future modularization projects.

This paper is structured with the following sections. ‘Research approach’ explains the research approach adopted in the study. ‘State of the art’ discusses the relevant literature and the relevance of dimensions. The ‘MCI method’ section presents the method for identifying modular candidates. ‘Application of MCI method’ explains the findings from the application, ‘Discussion’ considers the results and ‘Conclusion’ sums up the main contributions.

Research approach

The study was executed in three phases: literature review, development of method and application of method. For each of the phases, the process is detailed below.

Literature review

The relevant literature was reviewed to clarify the study’s position in relation to earlier contributions (Blessing and Chakrabarti, 2009). Additionally, the literature was used to identify existing methods that relevant for development of the method.

This review is based on 15 literature reviews covering the topic of modular product architecture development: Bonvoisin et al. (2016), Campagnolo and Camuffo (2010), ElMaraghy et al. (2013), Fixson (2007), Gauss et al. (2021), Gepp et al. (2016), Gershenson et al. (2004), Greve and Krause (2018), Jiao et al. (2007), Krause et al. (2014), Otto et al. (2016), Piran et al. (2016), Pirmoradi et al. (2014), Salvador (2007), Simpson (2006), (2014). With more than 1200 individual references in total, these reviews give a comprehensive overview of the published research in the field.

First, a general view of key concepts and challenges of modular product architecture development in ETO companies was used to assess and understand the generalizability of problems faced by Danish ETO companies and formulate a set of criteria for solving the problem. Secondly, the existing methods and frameworks of modular product architecture development were investigated in relation to their suitability for finding modular candidates based on the ETO criteria.

Development of the MCI method

The theoretical development of the MCI method was based on the criteria faced by ETO companies, leading to the need for a new approach based on combining modified methods found in the literature review. It was decided that a mapping of market segments and customer requrements was essential to understand the variation and commonality required by the market and the mapping of product architectures equally important. Cost and lead time mapping were selected to ensure that the candidates had the potential to influence both cost and lead time, as these improvements are the main focus of this research project.

Each of the dimension was broken down further, and appropriate methods were found in the reviewed literature, combined with a backwards search of keywords. The methods chosen and presented are based on their visual nature for comparison across categories and projects and their suitability for comparing data across different formats and levels of quality expected to be found in the companies.

Application of the MCI method

To evaluate the method, a case study was conducted (McCutcheon and Meredith, 1993) in an ETO company with 12 previously different product deliveries. The criteria for the evaluation were based on the framework’s ability to meet a set of criteria based on ETO characteristics from the literary research and on a method that would work in practice to provide a clear idea of modular candidates for managers to help plan future modular initiatives.

Modular product architectures in ETO

This section defines key concepts and presents and discusses the key considerations for finding modular candidates in a ETO company.

We define a product architecture by a commonality of underlying technologies and solution principles across a range of products. A modular product architecture is one or more interfaces that are designed to limit the effect of customers’ requirements to single modules, thereby decreasing the internal variety of products. We identify a modular candidate by a potential for reducing cost and lead time by module-based development, an example could be that standardization of interfaces enabling carry over modules, or isolate a functionality change to single module.

ETO companies are defined by the customer order decoupling point being in the design phase of product development (Wikner and Rudberg, 2005). This enables ETO companies to gain a strategic advantage in securing sales and building relationships (Adamson et al., 2012; Koponen et al., 2019) In practice, ETO companies do not know the exact specification of the next product variant until an order is received from the customer. This leaves product managers and architects planning standardization tasks as modularization challenged, because they face an open solution space of potential product variants, as opposed to a fixed set of variants, and do not know which of these variants should be prioritized in product architectures. (Haug et al., 2009). This is a key problem for this research project to solve as it is as yet unaddressed in the literature (ElMaraghy et al., 2013).

Additionally, the literature search provided various characteristics of ETO companies that are important for finding modular candidates.

Having or acquiring the knowledge to make change to product architectures is a problem in ETO companies (Bertram et al., 2020). Based on this, the authors believe that a method for finding modular candidates should develop the knowledge needed to do so.

Products range from simple customized items such as radiators to complex ones with deep product structures and many levels of assembly, such as powerplants or ships (Gepp et al., 2016; Willner et al., 2016). Based on this, the authors believe that methods for describing product architectures should accommodate different levels of abstraction and that modular candidates are not only a redesign of complete product architecture but are more likely to be improvements to parts of existing product architecture also seen in (Farrell and Simpson, 2008; Haug et al., 2009; Johnsen and Helene, 2017). Cost, lead time and range of potential customizations are key differentiating parameters in competition (Wikner and Rudberg, 2005). To estimate if a modular candidate is good, it is essential that its cost and lead time potential are described, in addition to its architecture.

Products are in many cases developed in serial, and previously customized products are further customized to new customer specifications (Willner et al., 2016). This approach leads to an increase in solution space and complexity for every product (Fisk, 2004). This led the authors to conclude that the potential for modular architecture in ETO could be found by discovering commonality across product deliveries that have not been exploited because of there being a focus on developing single products in series.

To summarize, the criteria for a method aimed at making informed decisions on modular candidates must be adapted to the open solution space of potential variants, bring forward knowledge, model product architectures on different levels of abstraction, include lead time and cost dimensions and focus on finding potential commonality for future deliveries.

State of the art

This section describes the literature relevant for assesing and designing modular product architectures. The contributions are divided into five categories based on their various modelling approaches and assessed based on the criteria from the previous chapter.

Assessment methods

In relation to finding the right candidates, a few assessment methods were of interest. Otto and Hölttä-Otto (2007) suggested a multicriteria approach for assessing product platforms by using 19 different metrics to evaluate platform concepts from market potential to aftersales service.

Fixson (2005) introduced a multidimensional product architecture assessment tool, and using the architecture mapping and evaluation method, Mortensen et al. (2016) outlined a means to estimate the savings of shared architectures using cost and CAPEX avoidance as the main criterion.

Function-based methods

Functional methods use functions as a means to establish the modularity of product architecture, including function–means trees (Andreasen, 1980) and schematic clustering (Ulrich and Eppinger, 2012). In addition, several tools have been suggested to support modular development in connection with the quality function deployment (QFD) method. The module indicator matrix (MIM) operationalizes 12 module drivers (Erixon et al., 1996) as part of the modular function deployment (MFD) framework (Erixon, 1998).

Component-based methods

Component-based methods include visual methods as an interface diagram by Bruun et al. (2014). Earlier contributions were made by Harlou (2006), who presented his product family masterplan to develop architectures connecting market, product and production domains. Krause et al. (2014) described the module interface graph (MIG), a visual representation of spatial arrangement, module boundaries, components and interactions. To include manufacturing architectures, Løkkegaard et al. (2018) introduced business-critical design rules to link markets and product architectures with manufacturing. A variety of commonality indices are available to compare commonality across product families (Thevenot and Simpson, 2006).

Matrix-based methods

Design structure matrix (DSM) is commonly used to support product design by modelling product and interfaces in a matrix. This can be done either heuristically or by clustering the components using advanced clustering algorithms (Borjesson and Hölttä-Otto, 2014; Eppinger and Browning, 2012). As extensions of the DSM, a variety of commonality indexes are available to quantify and compare commonality across product families (Thevenot and Simpson, 2006).

Mathematical methods

Mathematical methods are gaining in prominence (Pirmoradi et al., 2014), and several mathematical principles have been tested to model and optimize product architectures. These include conjoint analysis (Kazemzadeh et al., 2009), fuzzy logic clustering (Agard and Barajas, 2012; Chung et al., 2008) and discrete modelling (Ferguson et al., 2011). Mathematical methods have proven successful in optimizing product architectures, including in several manufacturing and market dimensions (De Weck, 2006).

Conclusion

From reviewing the state of the art, we did not find any methods that explicitly helped ETO companies to find modular candidates, since the methods were mostly focused on the design or redesign of a single product architecture, platform or product family. Whether a functional, component-based, matrix or mathematical modelling methods was chosen, all methods are either reliant on an existing set of product variants to be redesigned, or on specific and exact inputs on market segments and customers’ needs. The quantitative assessment involves determine the difference between an as-is situation with a to-be as seen in (Gonzalez-Zugasti et al., 2001; Moon and Simpson, 2014; Mortensen et al., 2016) This is not possible in ETO as the implications of future variants are unknown. Although no methods helped explicitly in finding modular candidates, the state of the art provides a selection of methods that are useful for a new method to do so. The authors believe that an analysis of previous product deliveries is a good option for uncovering modular potentials, as it has also proven successful in the field of configuration (Johnsen and Helene, 2017; Kristjansdottir et al., 2017). There exists a common understanding across the literature that market segments and customer requirement are important inputs for designing the right product architectures (ElMaraghy et al., 2013; Jiao et al., 2007; Otto et al., 2016; Pirmoradi et al., 2014). It is of course essential to model product architectures (Bruun et al., 2014; Gauss et al., 2021; Mortensen et al., 2016; Otto et al., 2016; Pirmoradi et al., 2014). Additionally, as mentioned earlier, cost and lead time are broadly recognized as success parameters in ETO and other types of company (Meyer and Lehnerd, 1997; Willner et al., 2016).

As to summarize the newness lies in the modification and connection of existing methods that allows for identifying patterns suggesting a modular candidate in ETO. Utilizing the advantage of knowing the requirements of the individual customers and the possibility to understand if savings in lead time and cost are present to start a modular project. Additionally, the product development phase is included in the cost and lead time analysis.

MCI method

The framework is designed to find modular candidates by analysing and comparing previous product deliveries. It is important that the product deliveries are comparable and contemporary in terms of delivering the desired features, and that ease of use and customer satisfaction is considered in order to avoid design modules and architectures on inferior solutions.

The authors suggest executing the work in three steps. First, each project is mapped using the same data and granularity of decomposition. Secondly, the individual dimensions are analyzed across products and the key findings mapped. Thirdly, the key findings are analyzed across dimensions to find the best candidates across market segments, customer requirements, product architectures, lead time and cost. Figure 1 shows an example of the approach used to find an ideal modular candidate M, because commonality across segments and requirements are not standardized in the product architecture, the problem is contributing to longer lead time and higher cost.

Market segment analysis

Market segmentation is important for a company to be more effective in fulfilling customer needs and requirements, and different customer characteristics can be used (Smith, 1956). To get a visual overview, a market segment grid using the power tower method (Meyer and Lehnerd, 1997) is proposed. The characteristics for finding segments are:

• Demographics (Smith, 1956).

• Performance (Bonev, 2013).

• Product’s application area (Otto et al., 2016).

• Revenue and contribution margin for market tiers such as economy, medium and premium price (Meyer and Lehnerd, 1997).

Customer requirement analysis

Customer requirements are used to understand the required variation of products (Meyer and Lehnerd, 1997; Otto et al., 2016; Ulrich and Eppinger, 2012). The data for finding customer requirements can be found from specifications or requirements from the customers. The requirements to indicate modular candidates are:

• Varying customer requirements (Erixon et al., 1996).

• Common customer requirements (Erixon et al., 1996).

Product architecture analysis

Mapping individual product structures to find the underlying product architectures are key to understanding if the latter are efficiently designed to handle variations from different customers. The interface diagram is proposed based on the recommendation of Bruun et al. (2014) for mapping the individual structures and enabling the underlying product architectures to be found. These are:

• A high number of product architectures seen by numbers close to the number of product structures, as this limits standardization (Mortensen, 2016).

• Non-standard customizations to product architectures (Johnsen and Hvam, 2018).

Lead time analysis

Lead time is a competitive factor for most ETO companies (Tersine and Hummingbird, 1995; Willner et al., 2016). The processes for customer-specific products can be mapped in a way done by Hvam (2008) or Prasad (2016). The candidates are processes on the critical path with dependencies leading to longer lead times, as a decoupling of these processes would allow for concurrent engineering to decrease lead time (Prasad, 1999).

Cost analysis

Cost is mentioned across most product architecture literature as a driver for modularization (Meyer and Lehnerd, 1997; Simpson, 2006). Ideally for mapping, the cost units reflect the structure and tasks related to the different product structures. A variety of methods for structuring cost in ETO were presented in Hooshmand et al. (2016).

Modular candidates’ analysis

The main advantage of applying modular product architectures is the ability to deliver products that fulfil a variety of customer requirements while reducing their internal variety, thereby lowering cost and lead time (Simpson, 2006; Ulrich and Eppinger, 2012). Cross-dimensional analysis is a key step in this approach; even though the individual dimensions are useful for finding potential candidates, the best candidates are found in the cross-dimensional analysis an overview of candidates can be seen in Figure 2.

Candidates are:

• Reoccurring common requirements that are not designed as a carryover module in the product architecture (Erixon et al., 1996) while carrying high cost and prolonging lead time.

• Reoccurring variants that are not designed as interchangeable modules in the product architecture (Erixon et al., 1996) while carrying high cost and prolonging lead time.

• Many non-similar or non-comparable customer requirements causing non-standard customizations in complex integrated products. These can have a high negative impact on product profitability (Johnsen and Hvam, 2018) and project execution (Willner et al., 2016).

The MCI method was developed to identify modular candidates in the open solution space facing ETO companies and the additional characteristics identified in the literature search.

The MCI methods differentiates itself by finding candidates on previous customized products and the product development processes, by comparing the dimension individually the methods allow to focus on the main signs of candidates and adds the lead time and cos dimensions across the project including the product development phases, this is not possible with existing methods. Additionally, the visual modelling allows for cross discipline verification and cooperation.

Case application of MCI method

The MCI method was tested at an industrial company operating an ETO strategy. The company in this case is a large enterprise with more than 1000 employees and an annual turnover of €200 m. It has the capability to develop, customize and manufacture within and across the domains of mechanics, electronics and software. To increase competitiveness, the company wanted to cut costs and lead time by applying modular product architectures, but it lacked the knowledge to decide which products to include. This was caused as inability to predict future requirements, and that the customization of previously customized products left direct comparison between projects difficult

A single researcher in charge of the project had full access to documentation and experts across all areas of the company to assist and validate findings. The selection of projects included the 12 most recent product deliveries within one of the business areas of the company, with the oldest dating back 5 years. The 12 projects included four types of products three mechatronic from the same family off products with significant recurring customization in the software domain, the last type is mainly mechanical with significant cost associated with customizations of the mechanical parts, the mechanical product was both sold separately and together with the other products as add on one might call it, and do to its different nature requirements etc. was done separately to the associated mechatronic product. The product development activities varied from around 5.000 h to around 50.000 h if both a mechatronics and mechanics was included.

Several technology upgrades to products and data systems prior to this time had made earlier comparisons irrelevant. Approximately 30 experts from the company provided knowledge and feedback. The work was carried out over a 20-week period. The results were summarized on a single poster measuring 1190 mm × 2500 mm. The learnings and the results from case application are detailed below, and the key findings are shown in Figure 3. Due to confidentiality the cases are anonymized and simplified.

Market segment analysis

Data concerning customer characteristics were extracted from the sales management system and project folders. The market segment grid method was useful for manually mapping segments based on the application area and the vertically by price of the products. It showed three segments based on the application area one for each type of mechatronic product as seen in Figure 3. The segments were used to structure various dimensions of requirements and product architectures.

Customer requirement analysis

The customer requirements were extracted from the product requirements and specifications included in contracts signed with the customer. The requirements were grouped at product level and on the associated product architectures separately, mechanical, electronics or software 1 and 2. As the detail and wording of the requirements varied heavily between customers, this structure was found to be essential in ensuring suitable comparisons across products. To compare across projects, the requirements were grouped based on their environmental requirements which that reflected the application area used for segmentation. The work with restructuring requirements and coding these to allow comparison was time intensive and required verification by the involved systems engineers. The analysis suggested several indications of candidates. As an example, the software (see modular candidate 1) showed variations on one function type and a high number of non-comparable requirements. In the mechanics-specific requirements, a similar demand to varying functions was identified interestingly the same variation to functions was seen in application area 1 and 3 as well as potential common environmental requirements (see modular candidate 2)

Product architecture analysis

The product architectures were mapped using different systems and architectural diagrams from each of software, electrical and mechanical domains as (Bruun et al., 2014). The drawings and diagrams were then converted to the same decomposition level and visual appearance for comparison. The mapping showed three different product architecture principles for the mechatronic products, and one mechanical architecture common across application area 1 and 3.

When investigating the individual architectures, several indications of candidates were present.

In Software architecture 2, very little reuse was present between products, while complexity made comparison between products difficult. The mechanical architectures connected to application area 1 and 3 changes directly influenced individual elements of the architecture but cascading effects to the connected elements was seen in most architectures indicating that carry overs of more modules could be facilitated by standardized interfaces.

Lead time analysis

A lead time analysis was done by combining data from project plans, milestone reviews and registered hours in the ERP system. The project plans and milestone reviews were documented visually to give an overview of the processes of the critical paths in the projects; the processes required some work to ensuring same coding and allow for comparison. An interesting finding was that in application area one software architecture 2 was often delayed compared to the original project plan, and changes were needed after customer delivery in several projects

Additionally In Application area 1 and 3 the mechanical architectures were a main driver, and the Software architecture 2 caused problems causing rework and iterations prolonging lead time. These findings stood out when comparing lead times across but in individual projects displayed those unplanned changes had dispositional effects across the different domains.

Cost analysis

A cost analysis was done by accessing data directly from the ERP system. The product related data had already been organized according to product structures but the cost to the product development phase required coding as was the case in the lead time analysis. Cost analysis showed similar findings as the lead time analysis, as most costs were uptrained by development cost. The costliest elements to design and produce were products based on software architecture 2 in application area 1 and products based on the mechanical architectures. As with the lead time analysis, these findings stood out when comparing lead times across but in individual projects displayed those unplanned changes had dispositional effects across the different domains.

Modular candidate analysis

The modular candidate’s analysis was done by comparing main findings across the dimensions. It showed that on the highest product architecture decomposition level product architectures where each designed to serve specific segments. In each of the three application areas had a specific set of environmental requrements. However, when analyzing the architectures in the individual application areas two candidates were found as standing out.

Modular candidate 1 (seen in Figure 3) is the software architecture 2 in application area 1; the requirements analysis showed that the candidate was affected both by varying requirements for the same functionality and varying non-comparable requirements. A high degree of changes leading to very different redesigned architectures for every project suggests that the software architecture was not adapted to the variation required by customers, concluded by firstly the isolation of the predicted varying requirements and secondly because the complexity of the architectures led to high amounts of iterations and cascading effects in some projects. In terms of lead time potential, Software architecture 2 was the driver and responsible for delays in several product deliveries. The cost was, on average, a significant element as well and therefor a suitable candidate.

Modular candidate 2 (seen in Figure 3) was the mechanical architecture. The requirements section showed that for both application areas 1 and 3, varying requrements for the same functionality were found and the environmental requirements were almost the same. The mechanical architectures layout was close to stable across products, but the interfaces changed in size and form. This drove high cost and lead time in product design and production, and it was therefore seen as a candidate because a more explicitely defined modular architecture was believe to decrease the effect of variation.

Conclusion of framework application

The application of the MCI method in the case company helped the case company isolated two architectures showing potentials as modular candidates. Initially, the individual projects were deemed uncomparable as the products was developed to new customers from previusly customized products leading to high variation in the product development phase and in the documentation following. But by consistent coding and analysis of the dimension seperately main points of variation was found and compared allowing for comparison in otherwise projects otherwise deemed uncomparable by the engineers.

The result of finding two modular candidates was presented and approved by the relevant company managers and the lack of direct quantitative results was made unnecesary by the visual backwards tracking of findings to projects and thereby suefficient at displaying a potentials without directly quantifying potential savings.

The software architecture was redesigned in a modular product architecture development project and was to be used for all future products for product architecture 1. The mechanical architecture is currently being redesigned to ensure standardized use across future products. Additionally, the analysis provided a better shared understanding of the company’s situation and its products across different dicplines and underlined the potential of modular architectures in the company.

Discussion

This research paper was initiated based on an industrial problem, the problem was clarified and supported by the findings of a literature review (Bertram et al., 2020; Gepp et al., 2016; Haug et al., 2013; Willner et al., 2016). Though the industrial problem and the criteria for the framework cannot claim to represent all ETO companies, as definitions are broad and companies and industries differ (Gepp et al., 2016; Willner et al., 2016). However, from Mortensen et al. (2018), it is seen that other company types share the same problems. Based on a survey of 12 small-to medium-sized Danish companies that have run modularization projects, seven of them said they had ‘no real expectations of how modularization was going to be applied’ and eight of the 12 companies ‘had to change their scope during the project’. This suggests that the problem this paper addresses is not only relevant to ETO companies.

The MCI method differentiates it self by adressing a gap in ETO-specific methods for modularization (Gepp et al., 2016). The method is aimed at the first stages of a ETO companies modular journey by answering the question ‘where to start our modularization initiatives ?’

Existing methods are not suefficint as these in general has two different starting points one type of methods are aimed a redesignign a portfolio of existing variants as the other at developing a new family of products from a given set of segments and customer needs (Gauss et al., 2021). The results on the right architecture is made by comparing alternatives as for exampel the as-is to the to-be situation and from that advantages can be quantified and measuered (Gonzalez-Zugasti et al., 2001; Moon and Simpson, 2014; Mortensen et al., 2016). None of these starting points are present acording to the litterary review. The MCI method cannot provide a quantification but through backwards tracking of conclusions to projects allows for an understanding of the modular candidates and their potentiel effect on cost and lead time. Additionally, the method differentiates itself by including a view of archtiectures effect on product development process. By the ability to focus on recurring variation and commonality in the individual dimensions and comparing these instead of trying to understand every requirement and the effect on the architecture thereby better suited for complex products and developments

The MCI method was tested in a single case company, which limited the ability of this study to make general conclusions beyon the case company (McCutcheon and Meredith, 1993). The 12 product developement projects that were analyzed all gave comparable assessments and inputs towards finding modular candidates. This suggests that the steps used are applicable outside the case company, although additional cases are needed for more certainty. It should be further stated that the MCI method could be misleading if, for example, technology or market changes make the previous project unrepresentative for the future. Therefore, relevant experts should be consulted before final decisions are made on the products to include. The individual categories were analysed using modified methods from the literature. These methods were chosen due to their ability to present the findings visually to support decision-making from different experts in the domain. The data from the company needed to be sorted and rearranged to be comparable across projects; this was possible in the case company, although the data quality might not be sufficient in other companies for this approach to work. With an increased focus on digitalization (Legner et al., 2017), data availability should only increase in years to come. The importance of the individual dimensions is well documented in the literature, and these were found to be important for finding the right candidates in the case company.

Conclusion

The primary contribution of this paper is the MCI method, which was developed to fill a gap in knowledge of the open solution space problem faced by ETO companies starting a modular initiative. The solution space problem has been considered both in academia and industry. The MCI method is a sequence of three steeps and includes five dimensions. The dimensions were chosen based on existing literature for their importance in identifying modular candidates; they are market segmentation, customer requirements, product architectures, cost and lead time.

The method was tested successfully in a single case study, although it has not yet been proven more widely. Future research could test the method in other ETO companies, but also how the modular candidates are governed in future projects to avoid falling back to old habits.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Carsten Keinicke Fjord Christensen

Author biographies

Carsten Keinicke Fjord Christensen is a PhD candidate focusing on the development of modular product architectures and platforms especially in the ETO business.

Niels Henrik Mortensen is professor of product development and do research on methods for developing modular product families.

References

Adamson

Dixon

Toman

(2012) The end of solution sales. Harvard Business Review (90): 7–8.

Agard

Barajas

(2012) The use of fuzzy logic in product family development: literature review and opportunities. Journal of Intelligent Manufacturing 23(5): 1445–1462. DOI: 10.1007/s10845-010-0465-5.

Andreasen

(1980) Machine design methods based on a systematic approach-contribution to a design theory. Lund University.

Bertram

Mueller

Løkkegaard

, et al. (2020) Why engineer-to-order portfolio rationalization stalls: challenges in standardization, modularization, platform design and mass customization. Proceedings of the Design Society: DESIGN Conference 1: 2265–2274. DOI: 10.1017/dsd.2020.45.

Blessing

LTM

Chakrabarti

(2009) DRM, a Design Research Methodology. London: Springer London. DOI: 10.1007/978-1-84882-587-1.

Bonev

(2013) Performance measures for mass customization strategies in an ETO environment. In: Proceedings of the 20th Euroma Conference. Available at: https://findit.dtu.dk/en/catalog/2389429520 (accessed 18 December 2018).

Bonvoisin

Halstenberg

Buchert

, et al. (2016) A systematic literature review on modular product design. Journal of Engineering Design 27(7): 488–514. DOI: 10.1080/09544828.2016.1166482.

Borjesson

Hölttä-Otto

(2014) A module generation algorithm for product architecture based on component interactions and strategic drivers. Research in Engineering Design 25(1): 31–51. DOI: 10.1007/s00163-013-0164-2.

Bruun

HPL

Mortensen

Harlou

(2014) Interface diagram: design tool for supporting the development of modularity in complex product systems. Concurrent Engineering Research and Applications 22(1): 62–76. DOI: 10.1177/1063293X13516329.

10.

Campagnolo

Camuffo

(2010) The concept of modularity in management studies: a literature review. International Journal of Management Reviews 12(3): 259–283. DOI: 10.1111/j.1468-2370.2009.00260.x.

11.

Chung

Lee

Kang

, et al. (2008) A DEA window analysis on the product family mix selection for a semiconductor fabricator. Expert Systems with Applications 35(1–2): 379–388. DOI: 10.1016/j.eswa.2007.07.011.

12.

de Weck

(2006) Determining product platform extent. In: Product Platform and Product Family Design. New York, NY: Springer US, 241–301. DOI: 10.1007/0-387-29197-0_12.

13.

ElMaraghy

Schuh

Elmaraghy

, et al. (2013) Product variety management. CIRP Annals - Manufacturing Technology 62(2): 629–652. DOI: 10.1016/j.cirp.2013.05.007.

14.

Eppinger

Browning

(2012) Design Structure Matrix Methods and Applications. MIT Press. Available at: https://findit.dtu.dk/en/catalog/2442794148 (accessed 3 September 2019).

15.

Erixon

(1998) Modular Function Deployment. KTH.

16.

Erixon

von Yxkull

Arnström

(1996) Modularity – the Basis for Product and Factory Reengineering. CIRP Annals 45(1): 1–6. DOI: 10.1016/S0007-8506(07)63005-4.

17.

Farrell

Simpson

(2008) A method to improve platform leveraging in a market segmentation grid for an existing product line. Journal of Mechanical Design 130(3): 031403. DOI: 10.1115/1.2829889.

18.

Ferguson

Olewnik

Cormier

(2011) Exploring marketing to engineering information mapping in mass customization: a presentation of ideas, challenges and resulting questions. In: Volume 5: 37th Design Automation Conference, Parts A and B, 1 January 2011, ASMEDC, 609–623. DOI: 10.1115/DETC2011-48742.

19.

Fisk

(2004) Engineering complexity. Interdisciplinary Science Reviews 29(2): 151–162. DOI: 10.1179/030801804225012617.

20.

Fixson

(2005) Product architecture assessment: a tool to link product, process, and supply chain design decisions. Journal of Operations Management 23(3–4): 345–369. DOI: 10.1016/j.jom.2004.08.006.

21.

Fixson

(2007) Modularity and commonality research: past developments and future opportunities. Concurrent Engineering 15(2): 85–111. DOI: 10.1177/1063293X07078935.

22.

Gauss

Lacerda

Cauchick Miguel

(2021) Module-based product family design: systematic literature review and meta-synthesis. Journal of Intelligent Manufacturing 32: 265–312. Springer US. DOI: 10.1007/s10845-020-01572-3.

23.

Gepp

Foehr

Vollmar

(2016) Standardization, modularization and platform approaches in the engineer-to-order business Review and outlook. In: Annual Ieee Systems Conference (syscon). Available at: https://findit.dtu.dk/en/catalog/2384016297 (accessed 28 August 2019).

24.

Gershenson

Prasad

Zhang

(2004) Product modularity: measures and design methods. Journal of Engineering Design 15(1): 33–51. DOI: 10.1080/0954482032000101731.

25.

Gonzalez-Zugasti

Otto

Baker

(2001) Assessing value in platformed product family design. Research in Engineering Design 13(1): 30–41. DOI: 10.1007/s001630100001.

26.

Greve

Krause

(2018) An assessment of methods to support the design of future robust modular product architectures. International Design Conference - Design 2018: 2018. DOI: 10.21278/idc.2018.0249.

27.

Harlou

(2006) Developing Product Families Based on Architectures: Contribution to a Theory of Product Families (PhD Thesis). Orbit. Department of Mechanical Engineering, Technical University of Denmark. Available at: https://findit.dtu.dk/en/catalog/2304902160 (accessed 28 November 2018).

28.

Haug

Ladeby

Edwards

(2009) From engineer-to-order to mass customization. Management Research News 32(7): 633–644. DOI: 10.1108/01409170910965233.

29.

Haug

Hvam

Mortensen

(2013) Reducing variety in product solution spaces of engineer-to-order companies: the case of Novenco A/S. International Journal of Product Development 18(6): 531. DOI: 10.1504/IJPD.2013.058556.

30.

Hooshmand

Köhler

Korff-Krumm

(2016) Cost estimation in engineer-to-order manufacturing. Open Engineering 6(1): 13. DOI: 10.1515/eng-2016-0002.

31.

Hvam

(2008) Product Customization. Product Customization. Berlin, Heidelberg: Springer Berlin Heidelberg. DOI: 10.1007/978-3-540-71449-1.

32.

Jiao

Simpson

Siddique

(2007) Product family design and platform-based product development: a state-of-the-art review. Journal of Intelligent Manufacturing 18(1): 5–29. DOI: 10.1007/s10845-007-0003-2.

33.

Johnsen

Helene

(2017) Improving product configurability. ETO COMPANIES 3: 221–230.

34.

Johnsen

Hvam

(2019) Understanding the impact of non-standard customisations in an engineer-to-order context : A case study. International Journal of Production Research. 57: 6780–6794. DOI: 10.1080/00207543.2018.1471239.

35.

Johnson

Bröms

(1995) The spirits in the walls - A pattern for high performance at Scania. Target 11(3): 9–17. DOI: 10.1055/s-0037-1603729.

36.

Kazemzadeh

Behzadian

Aghdasi

, et al. (2009) Integration of marketing research techniques into house of quality and product family design. The International Journal of Advanced Manufacturing Technology 41(9–10): 1019–1033. DOI: 10.1007/s00170-008-1533-2.

37.

Koponen

Julkunen

Asai

(2018) Sales communication competence in international B2B solution selling. Industrial Marketing Management (82): 238–252. doi: 10.1016/j.indmarman.2019.01.009.

38.

Krause

Beckmann

Eilmus

, et al. (2014) Integrated development of modular product families: a methods toolkit. In: Simpson

Jiao

Siddique

, et al. (eds) Advances in Product Family and Product Platform Design: Methods & Applications. New York, NY: Springer New York, 245–269. DOI: 10.1007/978-1-4614-7937-6_10.

39.

Kristjansdottir

Shafiee

Hvam

(2017) How to identify possible applications of product configuration systems in engineer-to-order companies. International Journal of Industrial Engineering and Management 8(3): 157–165.

40.

Legner

Eymann

Hess

, et al. (2017) Digitalization: opportunity and challenge for the business and information systems engineering community. Business and Information Systems Engineering 59(4): 301–308. DOI: 10.1007/s12599-017-0484-2.

41.

Løkkegaard

Mortensen

Hvam

(2018) Using business critical design rules to frame new architecture introduction in multi-architecture portfolios. International Journal of Production Research 56: 7313–7329. DOI: 10.1080/00207543.2018.1450531.

42.

McCutcheon

Meredith

(1993) Conducting case study research in operations management. Journal of Operations Management 11(3): 239–256. DOI: 10.1016/0272-6963(93)90002-7.

43.

Meyer

Lehnerd

(1997) The Power of Product Platforms: Building Value and Cost Leadership. Free Press. Available at: https://findit.dtu.dk/en/catalog/2304900488 (accessed 20 November 2018).

44.

Moon

Simpson

(2014) Platform valuation for product family design. In: Simpson

Jiao

Siddique

, et al. (eds) Advances in Product Family and Product Platform Design: Methods and Applications. New York: Springer, 179–198. DOI: 10.1007/978-1-4614-7937-6_7.

45.

Mortensen

Hansen

Løkkegaard

, et al. (2016) Assessing the cost saving potential of shared product architectures. Concurrent Engineering Research and Applications 24(2): 1–11. DOI: 10.1177/1063293X15624133.

46.

Mortensen

Bertram

Lundgaard

(2018) Achieving long-term modularization benefits: a small- and medium-sized enterprise study. Concurrent Engineering 27(1): 14–27. DOI: 10.1177/1063293X18803145.

47.

Otto

Hölttä-Otto

(2007) A multi-criteria assessment tool for screening preliminary product platform concepts. Journal of Intelligent Manufacturing 18(1): 59–75. DOI: 10.1007/s10845-007-0004-1.

48.

Otto

Hölttä-Otto

Simpson

, et al. (2016) Global views on modular design research: linking alternative methods to support modular product family concept development. Journal of Mechanical Design 138(7): 071101. DOI: 10.1115/1.4033654.

49.

Piran

FAS

Lacerda

Antunes

JAV

, et al. (2016) Modularization strategy: analysis of published articles on production and operations management (1999 to 2013). International Journal of Advanced Manufacturing Technology 86: 507–519. DOI: 10.1007/s00170-015-8221-9.

50.

Pirmoradi

Wang

Simpson

(2014) A review of recent literature in product family design and platform-based product development. In: Advances in Product Family and Product Platform Design New York, NY: Springer New York, 1–46. DOI: 10.1007/978-1-4614-7937-6_1.

51.

Prasad

(1999) Enabling principles of concurrency and simultaneity in concurrent engineering. AI EDAM 13(03): 185–204. DOI: 10.1017/S0890060499133055.

52.

Prasad

(2016) On mapping tasks during product development. Concurrent Engineering 24(2): 105–112. DOI: 10.1177/1063293X15625098.

53.

Salvador

(2007) Toward a product system modularity construct: literature review and reconceptualization. IEEE Transactions on Engineering Management 54(2): 219–240. DOI: 10.1109/TEM.2007.893996.

54.

Simpson

(2006) Product platform and product family design. In: Simpson

Siddique

Jiao

(eds) Product Platform and Product Family Design: Methods and Applications. New York, NY: Springer US. DOI: 10.1007/0-387-29197-0.

55.

Simpson

(2014) Advances in product family and product platform design. In: Simpson

Jiao

Siddique

, et al. (eds) Advances in Product Family and Product Platf. Design: Methods and Applications. New York, NY: Springer New York. DOI: 10.1007/978-1-4614-7937-6.

56.

Smith

(1956) Product differentiation and market segmentation as alternative marketing strategies. Journal of Marketing 21(1): 3–8. DOI: 10.1177/002224295602100102.

57.

Tersine

Hummingbird

(1995) Lead-time reduction: the search for competitive advantage. International Journal of Operations and Production management 15(2): 8–18. doi: 10.1108/01443579510080382.

58.

Thevenot

Simpson

(2006) Commonality indices for product family design: a detailed comparison. Journal of Engineering Design 17(2): 99–119. DOI: 10.1080/09544820500275693.

59.

Ulrich

Eppinger

(2012) Product Design and Development. 5th ed. McGraw-Hill. DOI: 10.1016/0956-5663(92)90013-D.

60.

Wikner

Rudberg

(2005) Integrating production and engineering perspectives on the customer order decoupling point. International Journal of Operations & Production Management 25(7): 623–641. DOI: 10.1108/01443570510605072.

61.

Willner

Powell

Gerschberger

, et al. (2016) Exploring the archetypes of engineer-to-order: an empirical analysis. International Journal of Operations & Production Management 36(3): 242–264. DOI: 10.1108/IJOPM-07-2014-0339.