1. Introduction
Organisations in the Engineer-to-Order (ETO) domain typically manufacture products that are designed to unique customer specifications. Unlike the case of standardised or Commercial Off-The-Shelf (COTS) products, the customer is involved throughout the entire design and manufacturing process (Reference CutlerCutler, 2006). The products of ETO programmes are often described as complex, produced in low volumes, and developed in close collaboration with the customer (Reference Mueller, Bertram and MortensenMueller et al., 2020). They are likely to be complicated in their operational behaviour, but usually require complex design and manufacturing processes. They typically require a high level of customer and product support due to the unpredictable situations experienced throughout their life cycle (Reference McKendry, Whitfield and DuffyMcKendry et al., 2015). Companies that operate in the ETO domain have a Customer Order Decoupling Point (CODP) at the design stage. The CODP is the point in the development cycle where the customer starts influencing the product. The CODP also signifies the point where the customer specifications become frozen in most cases Olhager Reference Olhager(2010).
ETO products may usually be described as a system-of-systems and consist of bespoke or customised supplier systems, created to perform a specific function (Reference Hicks and McGovernHicks & McGovern, 2009). These products are developed through a process called New Product Development (NPD). The products (the developed system) and the business (the development system) exist together to create the system-of-systems. In this context, the developed system includes the physical products to be delivered, the manufacturing information required to create these products, as well as the information required to operate and maintain the products (Reference RingRing, 1994). The development system describes the engineering organisation, manufacturing supply chain, all development processes, and any other system elements required to generate the product system.
Ashby identified and defined the Law of Requisite Variety and explained the implications of this law for the control of complex systems (Reference AshbyAshby, 1956). In mathematical terms, variety is a measure of the number of possible states a system can have. The variety in a “controller” system must be equal to, or higher than the variety in the “situation” system in order to establish control. In terms of Operational Research, the Law of Requisite Variety implies that the desired outcome represents the ability to control the interaction between two systems.
In the context of NPD, the term variety is commonly associated with the number of product configurations or the level of customisation a company offers. However, Ashby’s concept of variety should be regarded as referring to the number of means of managing the potential states of the developed system. The identification and categorisation of these states will be required for assessing if the developing system possesses the requisite variety to address the complexity in the developed (product) system. In line with Ashby’s Law, the complexity of the system performing NPD must match or exceed the complexity of the environment in which it operates to effectively manage and control outcomes.
INCOSE (2015) defines a system as “a combination of interacting elements organised to achieve one or more stated purposes”. A system may be considered as either simple, complicated or complex. Simple and complicated systems can be understood using traditional Systems Engineering approaches. In complex systems, the elements have states that are interconnected in ways that cannot be fully understood, leading to uncertainty between cause and effect. Akers et al. Reference Akers, Keating, Gheorghe and Sousa-Poza(2015) stated that complexity in systems is often revealed as a collection of irreducible emergent characteristics that are not present in the constituent parts. Complex systems thus cannot be understood through simple reductionism.
Exploring the impact of complexity on NPD in the ETO context was undertaken in two steps. Section 2 describes the output of a descriptive literature review to identify the relevant dimensions of complexity in the context of NPD within an ETO domain. In Section 3 the authors’ industrial experiences were used to determine if and how these dimensions affected NPD in a real-world context. Section 4 discusses the findings from this case study and proposes an agenda for the research. Section 5 concludes the paper.
2. Literature review
A descriptive study was undertaken to analyse the current state of research in the literature and was guided by the following research questions:
-
RQ1: Which dimensions of complexity impacts the performance of NPD in the ETO context?
-
RQ2: Is there sufficient variety in the development system to address the complexity in the developed system?
The literature review was based on a scoping survey conducted using the Engineering Village search engine. Combinations of search strings were used to identify potential references to establish broad categories of complexity and was then followed by the descriptive study of dimensions of complexity associated with each category. These search terms used are recorded in Table 1.
Table 1. Search strings used to determine relevant dimensions of complexity
After reviewing the results of the searches in Table 1, ten sources were retained for the scoping survey and are summarised in Table 2. The categories identified were organisational complexity; process complexity; and product complexity. These categories and their associated dimensions are illustrated in Tables 3, 4 and 5, and includes an assessment of the extent to which the dimension was considered by the associated literature. In these tables sources marked with “F” indicates that a dimension was fully considered, while those marked with “P” indicates the dimension was partially considered.
Table 2. References that discussed dimensions of complexity impacting NPD in the ETO context
Table 3. Categories and dimensions of organisational complexity
Table 4. Categories and dimensions of process complexity
Table 5. Categories and dimensions of product complexity
Based on the outcome of the scoping survey and synthesis of the results, a descriptive study was undertaken to explore the organisational complexity, process complexity and product complexity of NPD in the ETO context.
2.1. Organisational complexity
Typical manufacturing organisations may be described in terms of differentiation (the type of elements in the system) and connectivity (how the elements are organised). Vertical differentiation describes the hierarchy of the organisation and horizontal differentiation describes how the elements in the hierarchy are split into smaller organisational units to achieve objectives. The connectivity describes how the vertical and horizontal elements interact with each other (Reference BaccariniBaccarini, 1996). Organising a business requires strategic decisions about the vertical and horizontal differentiation of elements, as well as the connectivity of these elements. Vertical differentiation refers to the hierarchical structure of an organisation and pertains to the levels of authority and decision-making (Reference Nowotny, Hirsch and NitzlNowotny et al., 2022). Higher levels of vertical differentiation are associated with more layers of management and centralised decision making. Vertical integration describes the organisation of successive stages of production or distribution within a single firm (Reference Gianfreda, Marciano and RamelloGianfreda, 2020). It pertains to the position of the organisation within its supply chain. Organisations that are vertically integrated control multiple stages of their supply chains, rather than relying on customer-supplier relationships for those stages.
Horizontal differentiation / integration refers to the organisation of teams or departments and how they provide specialised expertise at a certain level within an organisation’s structure (Reference RobbinsRobbins, 1989). Higher levels of horizontal integration is associated with cross-sectional teams combining skills and knowledge across different departments (Reference Nowotny, Hirsch and NitzlNowotny et al., 2022). Moorman et al. Reference Moorman, Deshpandé and Zaltman(1993) identified spatial differentiation as a dimension of organisational complexity. This dimension addresses the complexity that arises due to organisations being split over different sites.
Some of the dimensions pertaining to organisational complexity identified in Table 3 describe the vertical (variety of interactions; suppliers and subcontractors involved) and horizontal (complex organisational structure / temporary teams) differentiation. According to Bataglin et al. Reference Bataglin, Viana and Formoso(2022) and Bataglin et al. Reference Bataglin, Formoso and Viana(2024), operating in a multi-project environment further increases the sophistication required by Production Planning and Control (PPC) systems, as well as the communication between business units and teams. Mello et al. Reference Mello, Strandhagen and Alfnes(2015) added that integration between Engineering and Production across multiple companies (engaged in multiple projects) was the most critical factor in the coordination of the supply chain. Finally, Birkie and Trucco Reference Birkie and Trucco(2016) stated that inconsistent organisational goals increased what they termed internal complexity. An example of this was when lean practices were abandoned during peak production times to increase production rates. In addition to these dimensions that describe the formal structures, Sosa et al. (2008) added that informal structures should also be considered. Informal structures describe how communication flows inside teams and between teams.
2.2. Process complexity
NPD is a multi-stage-gate process, see (Reference Marquis and DeebMarquis & Deeb, 2018), and when conducted in the ETO context, it often includes multiple decision points and conditional workflow elements. Zhou et al. Reference Moorman, Deshpandé and Zaltman(2023) conducted a systematic literature review to determine the common metrics used to measure business process complexity. The authors reported that process complexity was primarily measured in four dimensions: activity complexity, control-flow complexity, data-flow complexity, and resource complexity. These factors largely described the dimension called production process interdependences / overlapping in Table 4. Alfnes et al. Reference Alfnes, Gosling, Naim and Dreyer(2021) considered the impact of the lead time in engineering and manufacturing processes and found that these lead times were a major source of uncertainty in Innovate-To-Order (ITO) and Redesign-To-Order (RTO) approaches. High numbers of engineering iterations and multiple hand-overs complicated estimating lead times in engineering, while the necessity for flexibility increased lead times in production.
Design change is an important engineering process, but was often associated with a negative impact on the cost and schedule of product development (Reference Fei, Gao, Owodunni and TangFei et al., 2011). However, Balogun and Jenkins Reference Balogun and Jenkins(2003) found that design changes could also be seen as a process of knowledge generation.
2.3. Product complexity
The common dimensions associated with product complexity addressed in the literature were summarised in Table 5. Barbalho et al. Reference Barbalho, Carvalho, Tavares, Llanos and Leite(2022) investigated the relationship between product complexity, team seniority, and project performance. Their interpretation of product complexity largely agreed with the dimensions listed in Table 5. The amount of technical uncertainty and completeness of the specifications played a role in defining the product complexity. Puddicombe Reference Puddicombe(2012) considered technical complexity and novelty as two dimensions that defined the environment of a product system. Products designed by ETO organisations typically had high levels of technical complexity and novelty, indicating that the environment was “uncertain”. Further complexity was added due to incomplete specifications during the development process. Kim and Wilemon Reference Kim and Wilemon(2009) found that ambiguity around product requirements was seen as a significant contributing factor in causing complexity in NPD.
2.4. Literature review summary
The organisational complexity describes the structural layout of the business, as well as its position within the supply chain. The vertical dimension describes the route of information and material flow through the business, which includes the involvement of an external supply chain. The horizontal dimension describes how the business organises their corporate resources to process the information and material flows. The level of complexity in the organisation is described by the items in Table 3. Three of these dimensions address the structural design of the organisation. The remaining two dimensions arise due to the interactions of humans in the organisational structures. It is due to this human involvement in the organisation that ETO-context businesses are complex, which implies that these businesses may exhibit unpredictable organisational behaviours.
NPD is arguably the most significant process in the ETO context. It often contains multiple decision points and conditional workflows, which complicates this process. The dimensions describing processes in the ETO NPD context, as shown in Table 4, describe complicated behaviours, rather than complex ones. Thanks to improvements in PPC, production processes interdependencies have come under control of well-trained and skilled project and production managers. Similarly, lead times in both engineering and manufacturing are now largely predictable. However, even though engineering changes cannot be anticipated a-priori, they can be controlled through appropriate engineering office procedures. As such, product complexity in the ETO NPD context is typically complicated, rather than complex.
Most of the dimensions of product complexity identified in Table 5 are quantifiable in nature, for example the number of functional requirements, or the number of components will be exactly known at some point in the design process. However, the dimensions of “product diversity / novelty”, and “technical uncertainty” are qualitative. Throughout the NPD process, the design is matured, and the quantitative dimensions are determined, while the technical uncertainty is reduced. The product complexity thus evolves through the different development phases. During the early development phases the product is a complex system, since the high amount of technical uncertainty makes it impossible to accurately predict the final behaviour. In addition to this, the product behaves in an unpredictable manner to design decisions during the development phase leading to multiple design iterations and adding to the product complexity. However, by the time the product is ready to enter service, the behaviour must be predictable in order to demonstrate that it meets the customer’s requirements. Assuming the requirements are accurate and complete, and that the product has been manufactured in accordance with the design, then the product has become a complicated (or perhaps a simple) system.
It should be noted that the literature review deliberately focused on NPD in the ETO context, as the influence of complexity in this case is expected to differ from other industrial contexts.
3. Case study consideration of complexity categories
Héroux-Devtek Inc. (HDI) is a medium-sized multi-national company providing landing gear systems and other specialised equipment as a Tier-1 supplier to the aerospace industry. The company has around 1,900 employees and is head-quartered in Quebec, with seven facilities in Canada, four in the United States, two in the United Kingdom and two in Spain. The company is the 3rd largest supplier of landing gear systems with a long-standing established history - including manufacturing the landing gear for the Apollo Lunar Excursion Module. HDI services a wide range of customers, with roughly 60% of their work being conducted on behalf of the defence industry. Half of the projects undertaken are of the MTO variety and half follow the ETO strategy. A recurring request by potential customers was for faster completion of development projects. HDI wanted to understand which factors contributed to increasing development times and started researching this topic. The dimensions of complexity identified in Tables 3, 4 and 5 were used by the authors to describe the realities experienced at HDI and applies to the global organisation. This case study was conceptually based on the work presented by Alfnes et al. Reference Alfnes, Gosling, Naim and Dreyer(2021) on their investigation into the impact of uncertainty and complexity in Norwegian ship building businesses.
3.1. Organisational complexity
HDI had a low level of vertical differentiation and exhibited a flat organisational structure with typically only four or five levels in the hierarchy for most teams. HDI operated in a multi-project environment and inconsistent goals increased internal complexity. Arbitration was performed at Director and VP level. ETO development projects were led by a Project Manager (PM) who was responsible for cost and schedule, and a Project Engineer (PE) who bore responsibility for the technical aspects. Technical decision-making was decentralised, while financial decision-making was centralised. HDI utilised a large and sophisticated supply chain and the vertical integration of the company fell between the delineations of a Type I and a Type II ETO (Reference Hicks, McGovern and EarlHicks et al., 2001). Critical, complex, and high-value manufacturing tasks were kept inhouse (Type I), but non-critical manufacturing tasks were delegated into their global supply chain (Type II).
The horizontal differentiation was less clearly defined, and aspects of both matrix organisation and a divisional structure were present. Project teams were temporary structures made up from functional teams, while manufacturing teams were more permanent and resembled divisional structures. The engineering organisation had functionally differentiated teams, although these teams were not structurally segregated from each other in an organisational sense. As the organisation grew through a mixture of acquisitions and organic growth, it had a large spatial differentiation.
3.2. Process complexity
HDI had an extensive set of procedures describing the various business processes undertaken by the organisation. The NPD process was described at a high level and relied on the skill and intervention of PMs and PEs to drive the process. NPD of new landing gear products invariably had numerous conditional workflow gates and multiple decision points. Typically, the clean-sheet development of a new landing gear system could require more than 100,000 engineering hours and thousands of individual activities. The development timescales typically lasted between five and eight years from winning the bid to delivery of the first products. During this time, most activities had a very high level of data dependency on other activities in the development process. The planning resolution for processes could span anything from tens of months (fatigue tests), down to hours (drawing of minor components).
Long lead items (for example forgings) were usually ordered at the conclusion of the Preliminary Design Review (PDR) and manufacturing drawings were completed as part of the entry criteria to the Critical Design Review (CDR). Design changes frequently competed with other development activities for engineering resource during this phase. It was not uncommon that the manufacture of first article components were already on the second or third major design modification.
3.3. Product complexity
HDI specialises in Landing Gear Systems (LGS) as seen in Figure 1. These typically consisted of the Main Landing Gear (MLG), Nose Landing Gear (NLG), and the hydraulic systems such as actuators and uplocks. The MLG and NLG interfaced with aircraft connection points powered by hydraulic actuators. Equipment such as uplocks, door actuators and steering actuators formed part of the hydraulic circuit. Hydraulic manifolds and selector valves controlled the sequence of events to prevent structures from contacting each other. Electronically powered sensors, such as micro-switches or proximity switches, provided state variables that informed to the system of the position of gears and doors.
Figure 1. Landing gear system produced by Héroux-Devtek (Copyright: Héroux Devtek Inc)
HDI’s products typically had a high part count and number of interfaces. Although the LGS had a relatively low number of functions, the development involved several engineering disciplines and required technical innovation or novelty to be successful. Providing hardware to the customer required a sophisticated supply chain and large numbers of complicated manufacturing operations.
New Product Development (NPD) at Héroux-Devtek
NPD was typically led by PMs and PEs utilising the skills of the engineering teams located in St. Hubert (Canada), Runcorn (UK) and Madrid (Spain). These teams provided the required skills in architecture and design, structural integrity, dynamics, performance, hydraulics, electronics, materials and processes, and testing to facilitate the development of a new LGS. Manufacturing plans were prepared by specialist Manufacturing Engineers (MEs) located in Longueuil (Canada), Runcorn, Nottingham (UK) and Madrid. A diverse and sophisticated global supply chain provided components that were assembled to parts that were manufactured by HDI’s in-house manufacturing facilities.
HDI often assisted customers with finalising the specifications based on their expert insights into the products. Repeat customers could request modification of existing products to meet updated requirements for new aircraft, which could be classified as RTO.
NPD followed a structured gateway process and usually consisted of the following phases: Bid, Concept Development Phase (CDP), Joint Definition Phase (JDP), Preliminary Design, Critical Design, Industrialisation, First Article Manufacturing, Qualification Testing, Safety-of-Flight (SOF), Full-Flight Clearance (FFC), Physical Configuration Audit (PCA), Project Closure and handover to serialised production. As was often the case with complex development projects, the stage boundaries were not clearly defined, and some aspects of the product development could be in different stages at any given time during the development phase.
3.4. Product LG1 - a case study
Product LG1 was a development project recently undertaken on behalf of an export military customer. The product consisted of the NLG, MLG, Actuators, Uplocks, LG Control Valves and all associated hydraulic and electronic dressings. Due to an offset requirement imposed by the contracting government, some development work was performed by a partner company in the destination country. The development team was spatially distributed. HDI’s partner company was both a supplier (providing hydraulic equipment hardware) and the customer (buying the complete LGS). This company served as the Tier-1 supplier for the Prime Contractor with the flow of requirements following a complicated path: the Air Force / Government set the overall requirements through their defence procurement agency; the Prime Contractor translated these into technical specifications to the Tier-1 provider; and HDI provided a design that met these specifications. There were large amounts of technical uncertainty due to several key specifications marked as “TBD” or “TBC” as late as the PDR. HDI and the Tier-1 supplier were tasked with demonstrating that the designs were future-proof against these unknown specifications during the early stage-gate reviews. This was done by approaching the development through the principles of Platform Design, using previous experience and knowledge of their product capabilities. The depth of organisational complexity in the engineering teams allowed HDI to continue with the NPD process despite the high level of uncertainty. The customer specification was subject to iteration and co-evolution during the NPD. Iteration occurred mainly due to outdated information found in the specification, while co-evolution was mainly due to maturation of sub-system requirements. Design changes were common during the development process. Product LG1 underwent 24 Engineering Change Requests (ECR) between the CDR and the assembly of the first article. Another 29 ECRs were incorporated before the product was fully industrialised. These changes were the result of experiences during the early manufacturing stages, or from issues discovered during the qualification testing campaign to improve the design, or to successfully meet the customer specifications. The NPD of Product LG1 provided an example of a complicated organisation, deploying a complex NPD process to develop a product that evolved from complex to complicated during the NPD process. However, despite this level of complexity, Product LG1 achieved SOF approximately two years before the original contract date and was hailed by HDI’s customer as a highly successful development project.
To investigate the reasons for this success, the development system was evaluated in terms of the requisite variety in complexity between the development system and the product system. Since there is no pre-existing framework to perform this analysis on NPD in the ETO context, it is proposed to compare the complexity present in the product and match that with complexity in the organisation or process and make a qualitative assessment if this complexity helped or hindered the development process. This is shown in Table 6 and were obtained through informal interviews with HDI personnel.
Table 6. Product LG1 - complexity dimensions addressing product complexity
4. Discussion
Dimensions of the organisational complexity and process complexity were found to both help and hinder during the development of LG1. Organisational complexity had a more helpful effect, while process complexity more frequently hindered the development programme. Only one dimension of product complexity (number of product components or parts) was “helped” by both the organisation and process complexity, while two (number of connections or interfaces; number of feedback loops) were hindered by both categories. The other five dimensions of product complexity were impacted in ways that were more difficult to interpret requiring further investigation. The original RQs could now be reviewed:
-
RQ1: Three categories of complexity were identified: organisational and process complexity described the developing system and product complexity described the developed system. Each of these categories had several interacting dimensions describing the overall complexity found in the system-of-systems formed by the combination of the developing and developed systems.
-
RQ2: This question could not be fully answered. It was unclear how to determine what “sufficient variety” would be for NPD in the ETO context. The identified categories and dimensions of complexity appeared to be sufficient to describe the system-of-systems, but their interaction, or how these interactions affected NPD, were not straight-forward to determine. It was proposed that further research would be required to answer this question satisfactorily.
Problems involving the interaction of complex systems were often described as wicked problems. These problems could be investigated using Soft Systems Methodology (SSM) or Systems Thinking. It was proposed to investigate the impact of complexity in the NPD of ETO products using a SSM approach.
5. Conclusion
This paper presented a summary of the impact that complexity had on NPD in the context of businesses that provide ETO products. The study addressed two primary research questions: which categories and dimensions of complexity had an impact on these activities, and did the requisite variety exist in the developing system (represented by the organisation and processes) to control the level of complexity present in the developed system (the product).
A literature review identified the categories of organisational, process and product complexity. Using the INCOSE definitions of complexity in systems as reference, the level of complexity in the organisation and processes were found to be ‘complicated’. The complexity in the product were found to evolve throughout the development lifecycle, starting off as ‘complex’ and becoming ‘complicated’ or ‘simple’ by the time the product was transferred to the customer. This was a necessary condition required to demonstrate that the product meets the customer’s specifications and requirements.
The results of a case study that evaluated the development of an aircraft LGS in an industrial setting were included. The NPD process was evaluated based on the previously identified categories and dimensions of complexity existing between the developing system and the developed system. The evaluation highlighted that while the categories could be identified using the state of the art available in literature, the interaction between the different categories was less clearly defined and would require further research to answer completely.
This paper added to the body of literature concerning NPD in the ETO context and evaluated the impact of complexity in the system-of-systems made up by the combination of the development system and developed system based on an industrial case study. The originality of the paper was in considering if the source of complexity helped or hindered the execution of the development process.
Open access
