A systems engineering–based approach for framing reliability, availability, and maintainability: A case study for subsea design

Framing reliability, availability and maintainability (RAM) aspects are critical for an engineering design, as RAM is concerned with the sustained capability of a system throughout its useful life. RAM analysts are responsible to consider both functional and dysfunctional behavior of a given system beyond the perspective of system designer. However, the system concept baseline developed by RAM toolset is often a partial view, which is either too abstract when preparing RAM analysis or too overloaded when integrating RAM analysis with design process. Such practice may not give systemic insights of the design concept, considering specific subsea design challenges such as limited accessibility and requirement for automate control. For this reason, it is of great importance to ensure an effective and sufficient communication between the domain of design and domain of RAM. Integrating with a well‐known engineering discipline, such as systems engineering (SE), may help analysts to create the collaborative design environment necessary to control the design risks for a system with high complexity. This article proposes a new framework that links SE with RAM engineering by connecting relevant concepts and models used. A novel subsea design concept is offered as a case study to demonstrate the key changes in subsea design activities for addressing RAM with the proposed framework.


INTRODUCTION
Reliability, availability, and maintainability (RAM) is concerned with the sustained capability of a system throughout its useful life. RAM plays an essential role in the engineering design process of subsea systems to create competitive advantages, such as reducing capital investment (CAPEX) and operational costs (OPEX), controlling the risk of redesign, and mitigating potential future production disturbances. 1 RAM of technical systems are receiving center stage attention in many sectors, such as automotive, 2 aviation, 3 nuclear, 4 oil and gas (O&G), 5 and railway. 6 RAM analysis based on feedback from existing legacy systems imposes constraints on systems requirements, architecture, and design. 7(p97) However, managing RAM is often viewed as a separate activity in many subsea engineering practices, and the relationship to other established engineering frameworks, such as systems engineering (SE), are often not developed. For example, in discussions that have taken place inside the research center of SUBPRO 8 with manufacturers of subsea systems, we see that they have established both RAM and SE processes, although the tasks may not be coordinated and there holistically address the generally high complexity associated with technical systems.
The authors investigate and suggest a new framework to integrate RAM engineering with SE. The International Council on Systems Engineering 12 defines SE as "an interdisciplinary approach and means to enable the realization of successful systems." RAM engineering shares some similarities with SE. For instance, they both employ models developed to give an abstract view about system behaviors and physical configurations, albeit for different analysis needs. This article provides a view on how to make specific couplings between SE and RAM engineering in terms of concepts and models used. RAM engineering is often considered as a specialty subset of SE, 7 and even then it seems that the specific interfaces between SE and RAM engineering are given limited attention. The authors select some literature from the SE community and discuss the interrelationship with typical RAM analysis methods and steps. A new framework is proposed on basis of this evaluation, to mirror SE for extending the current practice of framing RAM aspects in design.
A review of the literature uncovered references that discuss the potential integration and proposes some tools to support exchanges between RAM and SE. Jigar et al 13 presented ways to extend the existing availability allocation process to the relevant stakeholders involved by applying a SE approach. The work indicates that the availability allocation problem can be redesigned within SE principle, so that the analysis is conducted in an iterative and systematic manner. Garro and Tundis 14 showed the possible extension of reliability analysis of a system to that of the System of Systems (SoS) concept, to solve the main issues arising in system reliability analysis considering particular properties of SoS. Leveson 15  This article uses a subsea O&G production system to explain the foundation of the framework and demonstrate its applicability. Due to lower oil prices and changing field conditions, the Norwegian-based O&G industry is increasing the installation of subsea equipment to accommodate pressure assistance, O&G separation, and water treatment. 16 The marinization of topside technology (eg, fixed or floating facility) offers several benefits, such as increasing recovery from the field and saving costs associated with manning and maintaining the platforms. Hereafter, such innovations for improving current production solutions are referred as new subsea design. As of today, manufacturers and system integrators of subsea systems use internally developed procedures for framing RAM in the design, following standards such as ISO 20815 5 that link production assurance with reliability management in a wider context, and more detailed recommended practices such as DNV-RP-A203 17 and API-RP-17N. 18 However, the current practices are not optimized for recognizing new and specific design challenges or new operating environments.
For instance, failure mode, effects and criticality analysis (FMECA) is often used as "one size fits all" method for failure analysis, regardless of whether systems are installed subsea or topside. In the proposed framework, we will discuss how outdated practices can benefit by using SE methods as a foundation.
Subsea Production and Processing (SUBPRO) is an initiative funded by the Norwegian Research Council to address current and future challenges in subsea systems that require multidisciplinary collaboration. The project combines researchers and industry partners to address the gaps in knowledge and accelerate the level of innovation in O&G field development and operation. 8 The rest of article is organized as follows. Section 2 explains some of main characteristics of a typical design processes within SE and RAM, including highlighted similarities and differences. The new framework, referred to as RAM-SE, is introduced and explained in Section 3 and followed by a presentation in Section 4 about how these two discipline get advantages from such integration. A new subsea design concept is presented in Section 5 to demonstrate the application of the proposal.
The case study has been selected on the basis of systems relevant for the research based innovation center for SUBPRO. A summary with concluding remarks and suggestions for future research is given in Section 6.

RAM ENGINEERING AND SE
The following subsections give a brief introduction to the practice of RAM engineering and SE, including general considerations and practical challenges with respect to new subsea design. The discussions and reflections are based on literature review, investigation of the current industry practices, and feedback received from participants in the research project SUBPRO. 8

RAM engineering
RAM engineering aims at using engineering knowledge and techniques to control the risk of failures and reduce engineering uncertainties. 19 The main activities of RAM engineering covers (a) artificial experiments to test out the properties of a given system or parts, and (b) analysis and modeling techniques to reveal the cause-effect relationships between failure and specific conditions. 20 Activities, such as life time testing, carried out later, are of little relevance for this article and thus will not be further discussed. However, the current process may not be optimal for complex system design. Highly complex systems are characterized by highly coupled parts and nonlinear interactions. 23 Unfortunately, alone many RAM methods in Figure 1 are not well suited for identifying and studying the effects of these interactions. Using them in this way introduces design risks that stem from insufficient considerations of engineering aspects, and will be latent on the first day of operation. The traditional RAM models follow reductionism (or analytical reduction), which fosters a bottom up approach by assuming that parts are operated independently and are not subject to feedback loop and interactions. 15,23 Such "system concept" developed by RAM analysts is not efficient for a complex system, as the hierarchy structure does not explicitly express any dependencies. Taking subsea as an example, high-level complexity is introduced by modular and compact design, software implementation (programmed functionalities), digitalization for communication technologies, interconnected hardware devices, and use of new technologies under more demanding (eg, autonomous) operating environment. These issues require efforts to systematically manage complexity, otherwise the framing of RAM aspects could be incorrect.
In addition, the heterogeneity of the multidisciplinary context in the design phase also restrains the use of current processes. System designers (who are responsible to organize system models considering various engineering disciplines at stakes) may have conflicting interests with RAM analysts, reflected by inconsistency of their models and focus of their elaborations. New subsea design is a concurrent and collaborative process, where different engineering teams are involved including RAM analysts. The RAM issues for new subsea design must be considered as early as possible to support decision making about redundancy, modularization, strategies for interventions, and the like.
However, the effect of RAM considerations is not easily observed by other engineering teams, as confirmed by O&G industry partners who indicate that RAM analysis is not fully and actively used to support new subsea design. This said, many of the abovementioned methods do not have a well-defined interface with other analyses carried out in parallel phases of the design. A similar problem is also identified by Barnard 24 who points out that the overemphasis on probabilistic modeling frequently leads to misinterpretation of RAM analysis, which can lead to bad design or waste of engineering efforts.
For instance, a successful FMECA depends on a clear understanding of system concepts. 25 However, in practice one may start FMECA without establishing the holistic vision, due to the limited project time or independence of RAM analysis in the design process. The approach itself is unable to deal with critical combinations of failures modes, which means the failure or deviation is only analyzed individually within local perspective. 17 In the case of novel or unproven design, such as a new subsea design, many failures are systemic rather than the result of individual parts degradation, in particular for systems where software and communication technologies are used to implement a majority of the functionality. Systemic failures include "one of a kind" errors caused by improper operation procedure, software errors and flawed controls, and whose effects are complete or partial loss of functionality. Such failures may not be sufficiently identified through FMECA, which relies on a well-defined understanding of how the system can fail and the effects of failure. Therefore, the effect of failure at a system level is studied only partially. On the other hand, FMECA may take on a too large scope covering many trivial cases, which limits its support for decision making in design process. 26 It is therefore not ideal for engineers with different backgrounds to capture the useful concepts in their own models and analysis. Need to master complexity of design concept in a systematic and organized way before any specialty analysis.
The interactions between components/functions are not sufficiently considered in evaluating RAM performance. Example: The failure effect is only identified and evaluated on the selected hierarchical decomposition. The maintenance activities are evaluated in similar fashion.
The loss of RAM performance is beyond a chain of events. Need to organize the interactions between components/functions of system so the effect of failure is well understood.
The results of RAM analysis could be misinterpreted or misunderstood. Example: Probabilistic methods dominate in most practice. Human errors, software reliability, and systematic failures are not sufficiently covered in such analysis.
Need to communicate the result of RAM analysis in other ways than probabilistic based indicators so that systematic failures can be correctly communicated.
(Model-based) RAM activities are often "disconnected" from design process or have little interface with other engineering disciplines. Example: Heterogeneity in knowledge base Need to integrate RAM engineering with other engineering disciplines involved in design process by connecting the produced models and used concepts.
these requirements has been identified within the SE framework. SE includes methods to support design team coordination, ensuring that the system concept is communicated correctly and that the correct system concept is communicated. SE also includes analyses that can improve the basis on which the RAM analysis is carried out.

SE in subsea design
The core of SE is to apply system thinking to solve complex problems, where problems are viewed holistically instead of individually. 27 SE provides an iterative and systematic approach for problem solving, although the definition of SE varies across the literature. 28,29 The SE concept can apply to many industries to systematically analyze the given complexity, given two assumptions. 15 The first assumption is that the engineering effort for improvement on an individual component may not lead to an overall optimization. Returning to the subsea case, some subsea equipment cannot be replaced without pulling a whole module. This means that the effect of failure is not isolated to one component and one system function alone, but may include many others as well.

APPLYING SE TO INTEGRATE RAM IN SUBSEA DESIGN
This section will elaborate on SE activities with an outlook on RAM integration.

Requirement analysis
The SE engineering process starts with identifying the requirements of stakeholders. 7 A complex system often involves multiple disciplines and is verified by multiple analyses rooted in different domains. The stakeholders can be classified based on their contributions as "primary," "secondary," and "tertiary." 34 Both RAM analyst and system F I G U R E 2 A conceptual map of RAM and SE models designers who maintain a unified vision of the system concept are the primary stakeholders in new subsea design.
The glue that integrates the different contributing teams is the system level requirements that allow useful design concepts to be generated. 15 The study of operational concepts provides a preliminary overview to describe system missions, operating environment, and the internal/external interfaces. Much of the effort of a system designer is devoted to the functional requirements that define the behavior of system for fulfilling the needs, whereas RAM engineers aim to specify required RAM performance under different operating conditions. RAM requirements would be meaningless unless use profiles, environmental conditions, and operating conditions are specified. 36 The distinction between functional requirements and RAM requirements are important for eliminating inconsistencies between contributing engineering teams. Fulfilling the functional requirement does not implies the satisfaction of RAM requirement. The introduction or update of RAM requirements needs to update functional requirements and vice versa, but there are many constraints, for example, schedule and budget, on the simultaneous updates. In the context of subsea design, such conflicts can end up being more problematic, as most equipment and their interconnection cannot be modified after installation subsea. Therefore, it is more important to identify a best RAM performance considering the constraints of the operation and environment, rather than the theoretically optimal RAM performance. For example, the duplication of critical components (ie, redundancy) may add more flexibility in long-run subsea operation, but this decision implies costly installation and intervention due to the hiring of a larger vessel (ie, larger CAPEX).
The design should proceed with respect to these constraints and requirements to analyze functions and physical structure. Subsection 3.2 presents system architecture analysis as one of the most important SE activities and identify the role of RAM within.

System architecture and analysis
As stated above, RAM engineers are accustomed to focus on the hierarchical function structure, since failure can generally be described as the termination or loss of functions and each function could be analyzed independently. Such practice is suitable for a system with simple interactions, decoupled functions, and straightforward part-function relationships, but not complex systems. Complex systems are better served by the SE suite of tools to systematically develop a vision of behaviors, interfaces, elements, and control structure for a new subsea system. SysML are activity diagram and state diagram, respectively. As a specialized form of flowchart, the activity diagram uses "tokens" to illustrate the concurrency of flow of control and data. This semantic aligns the structure of activity diagrams with that of Petri nets accepted in RAM community, although the activity diagram is more concise than standard Petri nets, especially when it comes to modeling the reactivity of workflow. 37  Solely relying on functional architecture to analyze RAM performance of complex systems could be superfical and incomplete, as it only assists in identifying potential failure and repair events but not the associated cause and consequence. Therefore, the physical architecture of a design concept should be developed.

Architecture (physical) analysis
The physical (architecture) analysis defines the components that realize the identified functions. Depending on the role RAM analysts have in the design phase, a technical system is generally considered from a functional instead of architecture point of view. However, it shall not be the case for new subsea design. Even if the well-rounded functional analysis is completed, we may not be able to evaluate the potential failure modes due to the incomplete view of given system concept.
The most commonly used approach to study physical aspects of system is the physical decomposition, which is often used as the "checklist" for the dysfunctional analysis, such as physical FMECA. However, such breakdown structure does not help in the context of complex system as many parts are interrelated and ought not be analyzed individually.
Often times, studying physical aspects in RAM community is a brainstorming process that requires participations from multiple disciplines, Additional attention should be paid to system structure, that is, the modularity in subsea design environment. Modularity deserves attention even in the early phase of subsea design, and can be illustrated as shown in Figure 3. Some subsea functions are realized by components located within different modules, but the replacement takes place at a module level.
Design structure matric (DSM) is rather a straightforward modeling technique to handle the modularity replacement problem. 41 The component-based DSM is often adopted in SE even though it is not available in SysML and here recommended for new subsea design.
DSM is efficient in organizing the interactions between components and visualizing the shared patterns, and it can help designers to identify the relatively independent modules, and support some tasks such as RAM allocation.

RAM-SE FRAMEWORK
This section proposes a new step-wise framework for supporting RAM engineering in new subsea design. The proposed framework, shown in Figure 4, has been named RAM-SE. The RAM-SE framework revisits the current process of framing RAM aspects as given in Figure 1, and proposes several steps integrating both the SE and RAM community.

1.
Step 1: Operational analysis. The operational analysis introduced here takes place alongside requirement analysis introduced in Subsection 3.1. It covers the identification of interactions, environment, and boundaries of the system for an overall view but offers only an abstract conceptual view of the design. The main objective is to systematically formulate RAM and functional requirements of a system, based on the needs of identified stakeholders.

2.
Step 2: Design analysis. Hereafter, we use the term design analysis to cover both functional and architectural analysis introduced in Sub-

3.
Step 3: RAM analysis. As opposed to the static system structure formulated in design analysis, RAM analysis focuses on the "dynamic" changes within the system structure. Table 2

4.
Step 4: Joint concept analysis. This step is beyond the scope of

CASE STUDY
This section introduces an existing design concept-fiscal metering system. Adaptations must be made considering subsea specific issues.

System description
The fiscal metering is one vital part in O&G sector to precisely measure petroleum product exported from delivery to the eventual recipient, a schematic is given in Figure 5. The accuracy and validity of flow measurement are very important for contractual obligation between custody transfer parties (eg, consumer and supplier). Statoil 44 has proposed a design concept for subsea fiscal oil export system using ultrasonic flow meter (USM). The main advantage is that USM has no moving parts so the maintenance requirement is rather low. Figure 5 presents

FMECA
• Uses a basis for detailed RAM analysis and maintenance optimization and planning.
• Document the effect of failure on system.
• Systematically identify all operational modes and functions attached to each potential failure modes.
• Carry out an extended/revised type of FMECA that is able to involve dynamic aspects of key scenarios, see also the discussion in Ref. 52.
HAZOP • Review all system sections for abnormal operational situations for all modes of operations.
• Identify hazards and hazardous situations that must be encountered for or removed from design concept.
• Be less resource and time consuming.
• Instead of brainstorming, focuses on the solid system architecture to evaluate the possible hazardous situations.
Maintainability analysis • Establish maintenance strategies before put into the operation. 53 • Incorporate operational and maintenance mode in the design analysis.
• Develop the subsea system-specific or module-specific maintenance strategies.
CCF assessment • Encounter common mode errors that lead to the loss of independence.
• Systematically indicate the possible dependencies among functions and system architecture, such as proximity, overlaps in functionality, and dependencies on resources (eg, data, information, and power supply).
Zonal analysis • Encounter the malfunction that could result in serious effects on the adjacent components.
• Benefit from building a consistence system architecture that incorporates physical properties.

RAM allocation
• Decide the necessary improvement on component level to achieve the minimum required RAM performance in an optimal way.
• Benefit from building a consistence system architecture that considering modularity or other architecture aspects that may influence the efficiency of component improvement, for example, DSM.
Failure rate estimation • Provide failure rates and other input parameters for reliability modeling and calculation.
• Integrate a comprehensive set of influential factors on identified failures brought up by design analysis.
• Involve subsea designers as the experts via joint concept analysis for judging upon some particular issues, such as the excess of working loads, variations in internal or external pressures.

Reliability modeling and calculation
• Prepare a set of suitable models to be used for reliability and availability analysis.
• Identify relevant failure scenarios and evaluate model capacity in light of these.
• Identify the characteristics of architectures (eg, modularization, obsolescence, and degradation) and scenarios/events (eg, delay on repair, imperfect testing or harmful testing, failures of activation of backup) needed to be considered in suitable modeling approaches.
F I G U R E 5 Subsea fiscal oil export metering system 44 F I G U R E 6 Context model for design concept

Operational analysis
As shown in Figure 4, operational analysis frames the scope and paves the ground for both design analysis and RAM analysis by abstractly characterizing the life cycle, interactions, and externals of the system in question. Figure 6 presents a simplified context model for describing the surrounding elements (ie, blocks with gray) of USMs (ie, the block with black) and associated operational description and interface, in order to share this core concept agreed by various stakeholders. Based on Figure 6, it is assumed that each functional channel that fulfills the operational needs requires the signal interfaces between USM and SEU. There are two alternatives for configuration: configuration 1 is that all three USMs are connected to two SEUs, and configuration 2 is that one USM is connected to SEU and other two are To compare various maintenance strategies for USM assembly, the three possible maintenance strategies are as follows given the considerations from system designer: • Strategy I: The activities related to maintenance starts immediately when two USM functions are affected, the metering station is shut down during maintenance.

RAM analysis
RAM analysis starts with dysfunctional analysis as indicated in Figure 4. Here, FMECA is selected as hazard identification methods, and the part of the FMECA are presented in Table 3. The failure rate for each failure mode is shown in the last column of Table 3 With the information in Table 3  • The sensor lines are continuously checked, thus the delay for detecting failures on jumper and SEU can be ignored.
• All components are considered as good as new after maintenance.
The activities of maintenance are considered as perfect, thus no adverse effects are induced.
• Ideally, the subsea operator does not expect any retrieval during the operation until the metering system cannot perform the function as intended. Assuming that restoration duration = 8 hours and mobilization time = 1440 hours (ie, 2 months), and the intervention will be carried out after 20 years of installation (ie, 175 200 hours).
There are many suitable approaches for the following quantitative analysis, for example, Petri nets. Figure 8 presents partial Petri nets for case 1 (ie, configuration 1 following strategy I), where state-transitions in Figure 7 are mapping into Figure 8 by the predicates and assertions in the Petri nets. Predicate (represented by "?") is a formula to validate the transitions, and assertion (often represented by "!") is a formula to update the variables after the associated transition is fired. 48 The instruction for constructing Petri nets model can be found in articles of Signoret et al 48  The computation for RAM modeling is completed by the software GRaphical Interface for reliability Forecasting. 50 The simulation run is set to be 100 000 to get the result with confidence. The downtime and retrieval frequency of cases 1-6 is reported in Table 4 and measurement uncertainty of cases 1-6 is illustrated in Figure 9. From Figure 9 and Table 4, one may notice the following points: • The downtime reported in Table 4 not only considers the retrieval frequency of USM assembly but also the downtime to replace TA B L E 4 Downtime and retrieval frequency for cases 1-6 • Applying strategy II (cases 2 and 5) needs less maintenance than applying strategy I (cases 1 4) by paying the price of allowing an increase in measurement uncertainty.
• Applying strategy III (cases 3 and 6) results in the increment of measurement uncertainty in the last 5 years of lifetime (ie, the turning points in Figure 9) as the system is allowed to operate with single USM. The downtime due to maintenance is significantly reduced compared to strategies I and II for configuration 1 (cases 1 and 2), however, not for configuration 2 (cases 4 and 5).
• Configuration 2 (cases 4-6) has more maintenance needs than configuration 1 (cases 1-3), and the maintenance need does not vary too much given the different maintenance strategies. As result, the measurement uncertainty is decreased.
• The peak value of measurement uncertainty for configuration 2 (cases 4-6) comes earlier than configuration 1 (cases 1-3). The reason is that configuration 2 loses flexibility as the SEU is not fully redundant for each USM.

Joint concept analysis and communication
The objective of joint concept analysis is to present some common themes that cannot be solved or considered by any individual engineering discipline. Table 5 presents some major considerations derived from the selected analysis in RAM-SE framework. These considerations may either require designers to reevaluate the system concept, or RAM analysts to reconstruct the RAM model to achieve more realistic design implications. For example, the maintainability analysis shows that it is necessary to consider the separation between measurement instruments and sampling systems. Therefore, DSM is required for design analysis for mastering the interaction between these two modules and subsequent RAM analysis. Another example could be CCF assessment. The series connection of duty USM, master USM, and spare USM can introduce the common mode errors due to the same design, installation, and function. In this case study, common failure mode for USMs is mainly the deposits, for example, wax. The designer indicated that the implemented measure is to heat the flow, thus prevent wax formation. 44   • PT installed in the close location may cause the turbulences that influence USM performance.
• Develops strategy and associated equipment to reduce the effect of noise if cost and space allows, for example, noise trap or bends in piping.
• Keep the necessary distance between PT and USM, for example, at least three diameters of downstream. 54 CCF assessment • The series connection of USM offers better quality monitoring capacities but common mode errors of USM are introduced, which can influence the performance of USM and calibration process.
• Develops strategy for eliminating the potential factors on CCF, for example, improve manufacturing process and upgrade on-site calibration process by taking CCF into account, see also the guideline in IEC61508. 55 If not, CCF must be incorporated in relevant RAM modeling.
Maintainability analysis 53 • The sampling system has higher maintenance needs than metering module. This framework serves as a baseline for further refinement in order to direct future effort to improve the process of framing RAM in subsea design. The process described by the RAM-SE framework is highly simplified and idealized. First, RAM-SE framework only restrictively discusses interlinks between these two disciplines in light of models with high acceptance and commonality in each community, for example, SysML. This said, the design analysis and RAM analysis are conducted in sequence thus some overlaps may be latent as system theory or system thinking is indirectly placed in conducting RAM analysis. Additional research could develop RAM methods directly using system theory. One such pioneer work has been completed by Leveson 15 who use system theory to create a new accident model used for safety analysis. However, similar work has not been found in RAM domain yet. Moreover, the application is here only demonstrated within subsea design. One remaining work of this article can be to expand the analysis to consider other sectors to enrich the content of the proposed framework and hopefully bring ideas for transfer of knowledge from this article to other domains of interest. Our suggestion for improving this framework is to further test the proposal against an industry-size case.

ACKNOWLEDGMENTS
This work was carried out as a part of Subsea Production and Before starting on her PhD, she worked for several years in industry, including as instrumentation engineer (onshore) and automation and electrical supervisor (offshore) in Phillips Petroleum, automation leader at the factory of Nidar, and senior researcher at SIN-TEF, department for applied cybernetics. Her main research focus and interest concern functional safety and reliability of safetyrelated electrical/electronic/programmable electronic (E/E/PE) systems. She is also a member of IEC 61511 committee who maintains the standard on functional safety for process industry sector. As a co-director and responsible for reliability subject in the 8-year research centre on subsea production and processing (SUBPRO, see http://www.ntnu.edu/subpro), she has extended her research to cover reliability, safety, and condition-based maintenance of systems with particular demanding environmental and operational conditions, such as subsea systems. Lundteigen has a long-lasting and extensive contact network in Norwegian industry work, due to involvement in SIN-TEF projects over several years, many of the m through the PDS forum (http://www.sintef.edu/pds). She has more than 50 publications peer-reviewed papers, including around 20 papers in international journals. She has also been contributing to a high number of studies for industry companies.