Internet of Things adoption for reconfiguring decision-making processes in asset management

1.1 Motivation

The Internet of Things (IoT) can be used to collect more and more data which can be used by decision-makers to acquire the necessary insights in a timely fashion. IoT and data analytics will transform complete supply chain processes (Kumar et al., 2016) and has the potential to revolutionize management (Fosso Wamba et al., 2015). But developing and managing this data to an acceptable level whereby the right information can be provided to the right people at the right time is a complex undertaking.

Modern economies are supported by large infrastructures of transport systems, water and waste disposal networks, and energy and telecommunications networks. As such, the proper management and maintenance of the assets making up the infrastructure is vital to prosperity. IoT has the potential to improve the management of these assets by providing insight into the utilization of the infrastructure and the quality of the assets for maintenance and replacement strategies. Asset management as a type of business process is highly dependent on large amounts of data from which relevant information can be created and is used for decision-making during the life-cycle of assets. Infrastructure assets are stationary systems (or networks) that serve society where the system as a whole is intended to be maintained indefinitely to a specified level of service by the continuing replacement and refurbishment of its components (Herder et al., 2008). Asset management is generally understood to be the set of activities of a business objective associated with: identifying what assets are needed; identifying funding requirements; acquiring assets; providing logistic and maintenance support systems for assets; and disposing or renewing assets so as to effectively and efficiently meet the desired objective (Hastings, 2010; Woodhouse, 1997).

1.2 Research problem, objective and approach

The term, IoT refers to a network of physical objects that are able to communicate their data over the internet (Hounsell et al., 2009; Ramos et al., 2008). IoT enables asset managers to access remote sensor data and to monitor and control the physical world from a distance, allowing many physical objects to act in unison, through means of ambient intelligence (Ramos et al., 2008). IoT can benefit asset management organizations by providing enough quality data to generate the information required to help asset managers make the right decisions at the right time (Brous and Janssen, 2015b). For example, IoT can be used to collect data to determine the position and length of traffic jams, and to redirect traffic or offer alternative multi-modal forms of transport by using location sensors and analyzing traffic flow. But IoT adoption can also impact the asset management organization in unexpected ways. Automating processes often necessarily leads to changes to organizational structures and cultures as tasks previously performed by people become automated, whilst other tasks and responsibilities which previously did not exist become apparent (Brous and Janssen, 2015a).

Technology and organization influence each other in many ways, and analytical efforts to treat these as distinct conceptual units are increasingly being called into question (Boos et al., 2013). As such, it is important to understand how IoT adoption affects decision-making in asset management business processes in order to be able to achieve expected benefits and mitigate known and unknown risk. This research fills that gap by analyzing the changes in the business processes of maintaining civil infrastructure assets brought about by IoT adoption. The research objective is to understand how IoT adoption affects decision-making processes in asset management. As there is limited knowledge in this field and deep insight into the field is needed, case study research is used. Case study is a qualitative research method particularly suited to researching contemporary phenomena that cannot be separated from the environment they are embedded in (compared to laboratory experiments, for example) and that have not been scientifically studied to a large extent so far (Benbasat et al., 1987). The case studies comprise two organizations that have similar goals to reduce confounding effects in the study. Two case studies of IoT adoption within the asset management domain are presented in this paper. The cases have been made anonymous in the interests of privacy. The first case, a water measurement network, is managed by a central government organization. We refer to this case as case “X” in this paper. The second case study, a pump management decision support system, is managed by a water board. We refer to this case as case “Y” in this paper.

This paper read as follows: Section 2 presents the literature background; Section 3 describes the research methods in detail; and Section 4 describes the case studies background and compares the cases. The case studies are discussed in Section 5 showing how business processes need to be changed driven by IoT. Finally, conclusions are drawn in Section 6.

2.1 Decision-making for performance management

Decision-making processes with regards to maintaining current service levels are generally focused on preventing malfunction or asset failure or to quickly assess damage to infrastructure after an event such that maintenance procedures could be directed to the areas that need immediate attention (Aono et al., 2016). As such, performance measurement is required for corrective, preventive and predictive maintenance as well as other supportive activities (Parida et al., 2015). According to Phares et al. (2004), routine inspections are the traditional method of gathering information regarding the performance of the infrastructure. Routine inspections are regularly scheduled inspections to determine the physical and functional condition of infrastructure assets. Generally, a series of condition ratings that describe the general condition of the asset are considered for performance measurement. According to Phares et al. (2004), these condition ratings, consider both the severity of deterioration and the extent to which it is distributed throughout each of the components. The more often these inspections take place, the higher the granularity. Often, traditional inspections are performed subjectively. In other words, inspectors visually inspect the asset and makes an expert judgment based on what they see and their past experience.

2.2 Decision-making for managing how infrastructure services are perceived

According to Parida et al. (2015), maintenance is often integrated into a production process involving human factors. As such, although difficult to measure (Simões et al., 2011), perception management of infrastructure services is an important aspect to the asset management decision-making process as the effectiveness of the different facets of the performance system is often dependent on the competency, training, and motivation of people interacting with the infrastructure and the asset management systems (Ljungberg, 1998). For example, when planning maintenance on a busy section of highway it is important to take into account how the maintenance will affect the public and how the public will perceive the need for the maintenance. Decisions such as when to perform the maintenance (at night or during the day), or how best to inform the public, for example to provide alternative routes, are important aspects of infrastructure decision-making which require large amounts of data. For example, IoT data can provide insight into such aspects as the busiest times of day or night, or popular destinations (for determining alternative routes). Therefore, the relationship between the organization and the customer may be critical due to the influence that service quality has on user satisfaction level and the fact that people are involved in decisions related to maintenance and execution of tasks (Ardalan et al., 1992).

2.3 Decision-making for improving infrastructure services

Traditional thinking about the strategic decision-making process rests on the belief that actors enter a decision-making process with a known set of objectives, gather information about these objectives and select an optimal alternative to achieve these objectives (Eisenhardt and Zbaracki, 1992). However, Eisenhardt and Zbaracki (1992) also suggest that whilst people may be rational, power tends to win battles of choice. For example, according to Smith (2014), competing interests mean that managers increasingly embed poorly managed tensions into their organization’s strategy which require ongoing responses rather than one-time resolutions. Despite this, Marquez and Gupta (2006) believe that having a strategic perspective to asset management is a key factor for success. For example, detections of damage or failure of critical public infrastructure may have significant societal and economic impacts (Tien et al., 2016).

2.4 Effects of IoT adoption on asset management decision-making processes

Infrastructure systems consist of many different types of assets that could have long life cycles. For example a bridge lasts for at least 30 years, however, regular maintenance might be needed. Civil infrastructure assets need to be maintained to ensure their optimal value over their entire (long) life cycles (Hassanain et al., 2003). As early as 2001 there were already many software tools for asset management (Hassanain et al., 2003; Vanier, 2001), and since then many data formats, data sources and pools of unstructured data have become available over the years. At a high level, Hassanain et al. (2003) suggest that asset management tooling should at minimum provide the following functionality:

• identification of assets;

• identification of performance requirements;

• assessment of asset performance;

• plan maintenance;

• manage maintenance operations;

• life-cycle costing analysis;

• life-cycle analysis and long-term service-life prediction; and

• central repository for asset information.

It is expected that IoT adoption will change performance measurement of infrastructure services, like applying statistical learning (Archetti et al., 2015). Second, IoT adoption is expected to change the perception of infrastructure services, like perceiving sudden changes in temperature by which a fire could be detected (Hentschel et al., 2016), or the deterioration of the quality of assets over time (Brous et al., 2017). Finally, IoT adoption is expected to change business processes, for example through self-organizing resource planning (Zhang et al., 2015). In the next sections, we discuss these effects of IoT on asset management.

2.5 IoT adoption expected to change performance management of infrastructure services

IoT services are knowledge intensive and require collection of appropriate data contents, data analysis and reporting (Backman and Helaakoski, 2016). As such, statistical learning and network science is expected to play a critical role in converting data resources into actionable knowledge (Archetti et al., 2015). Due to increasing pressure on budgets and personnel as well as increased utilization of civil infrastructure, public asset management organizations increasingly need to intelligently manage their infrastructure with fewer resources (Rathore et al., 2016). By managing and analyzing various IoT data, it should be possible to create new services to achieve an efficient and sustainable civil infrastructure (Backman and Helaakoski, 2016; Hashi et al., 2015). IoT may bring an improved understanding of complex processes which is expected to help improve the efficiency of transport management and infrastructure services, and help with effective reporting (Kothari et al., 2015). IoT infrastructure could potentially be used to reduce costs in terms of time and money (Aono et al., 2016), as traditional methods of inspecting infrastructure, such as highway structures and bridges, for damage are often reactive in nature and require significant amounts of time and use of costly equipment. By specifying events (Hashi et al., 2015; Tao et al., 2014), it should be possible to obtain a set of data before and after an event to be used for analysis and evaluations, taking the effect of the event into consideration.

2.6 IoT adoption expected to change perception of infrastructure services

Rathore et al. (2016) believe that smart management of traffic systems with the provision of real-time information to the citizen based on the current traffic situation should enhance decision-making. Jonoski et al. (2013) believe that an important potential of IoT adoption may lay in developing applications for the broadest user base of individual citizens, not only as recipients of information, but also as suppliers of data. As such, Jonoski et al. (2010) suggest that a possible approach toward meeting this challenge may be the development and deployment of information systems that integrate the operations of data collection, data and model integration, and information dissemination such that organizations can work together with the users of their information for mutual benefits. In this way, IoT is expected to be able to provide users with information on costs, time, environmental impact and perceived quality of services (Archetti et al., 2015).

2.7 IoT adoption expected to change business processes of infrastructure services

In order to keep civil infrastructure such as bridges safe and functioning, regular inspections to determine the condition of the asset are a necessity (Ahlborn et al., 2010; Neisse et al., 2016). For example, traditional inspections of bridges are usually visual assessments by trained personnel where all the asset’s component conditions are observed once every three to six years, and are summarized into one report (Phares et al., 2004). After the inspection is done, asset managers must decide what maintenance interventions are needed based on these inspection reports. However, inspection reports of bridges can be biased by subjective judgments of the experts or by lack of information (Kallen and van Noortwijk, 2005). IoT data may make it possible to remotely observe the condition of objects and thereby enhance the available information on the current condition of public infrastructure (Ahlborn et al., 2010) and their environments.

2.8 Summary of effects of IoT adoption on the asset management decision-making process

The expectation is that IoT will be used for key decision-making in operational activities. It is expected that IoT will be used in a variety of ways related both to the real-time measurement and analysis of data as to trend analysis of historical data over time (Brous and Janssen, 2015b). Expected benefits of IoT adoption for business processes include:

• self-organizing resource planning (Zhang et al., 2015);

• creation of new services to achieve a sustainable civil infrastructure (Backman and Helaakoski, 2016);

• improving the efficiency of infrastructure services (Kothari et al., 2015) and thus reducing costs in terms of time and money (Aono et al., 2016);

• automation of processes (Hentschel et al., 2016);

• timelier provision of information (Rathore et al., 2016) allowing for more accurate inspections and analysis (Ahlborn et al., 2010);

• greater frequency of inspections (Neisse et al., 2016); and

• reduce or remove the need for physical, on-sight inspections (Ahlborn et al., 2010).

But adopting IoT also has consequences for asset management business processes. Expected consequences for asset management business processes include:

• requires understanding of the conditions and factors for effective and sustainable adoption of new data sources (Brous and Janssen, 2015b);

• requires statistical learning and network science to convert data resources into actionable knowledge (Archetti et al., 2015);

• requires the development and deployment of information systems that integrate the operations of data collection, data and model integration and information dissemination (Jonoski et al., 2010); and

• requires defining events in terms of sensing (e.g. sound, light, etc.) for localized events. For example, when is a loud noise an accident or simply a car back-firing? (Hentschel et al., 2016).

3.1 Case selection

Whilst single cases are recommended where the case represents a critical test of existing theory, or where the case is a unique event, or where the case serves a revelatory purpose, a limited number of case studies is considered to be more successful with regards to theory formulation and testing (Yin, 2003). Using more than one case study provides us with the opportunity to build the theory irrespective of an organization, which improves the argument for generalization. The evidence from multiple cases is often considered more compelling and the research more robust (Herriott and Firestone, 1983). We followed a similar approach to that employed in studies conducted by Pagell and Wu (2009) and Wilhelm et al. (2016) and selected two different organizations by applying a number of criteria. Two organizations tasked with maintaining infrastructure were selected. Case X operates at the national level, whilst case Y operates at the regional level. This enables us to compare differences between the levels to determine possibilities for generalization. Any use of multiple-case designs should follow replication logic to guarantee external validity. For this reason, we defined the following criteria which were used to select the different cases:

1. The case must be a complex functioning unit.

2. The primary processes supported by the case must be focused on the management of civil infrastructure.

3. Asset management processes must be supported by the asset management data infrastructure, evidenced by the existence and monitoring of key performance indicators.

4. People operating within the case must be willing to cooperate in the research and must be willing to provide access to the information required for the research.

5. The case environment should be “data-rich.” This means that the organization should produce, manage and maintain at least 5 large data sets as well as a more than twenty small to medium data sets.

6. The asset management data infrastructure must include at least one use case of IoT adoption.

The research studies the impact of IoT adoption on business processes in asset management data infrastructures. The cases of IoT adoption to be investigated were selected based on the criticality of their use and their importance to the organization. To avoid confounding effects, the study was limited to the examination of cases of IoT adoption in asset management according to specific criteria as specified above.

3.2 Data collection

It is not only generalization that presents challenges when adopting the case study methodology; the reliability aspect should also be taken into consideration (Yin, 2003). Reliability refers to the demonstration that the operation of a study, such as the data collection procedures, can be repeated with the same results (Yin, 2003). Yin (2003) recommends employing a well thought out research protocol to ensure reliability. According to Yin (2003), a case study protocol is a formal document which describes the set of procedures involved in the collection of data for a case study. The protocol used in this research follows the advice from Yin (2003) and included the problem statement, a delineation of the unit of analysis, the steps (including the altering of the steps) to be taken, the procedures for contacting key informants and making field work arrangements, reminders for implementing and enforcing the rules for protecting (the privacy of) human subjects, a detailed line of questions, and a preliminary outline for the final case study report. The protocol used in the case studies includes a variety of data collection instruments including documentation, individual interviews and group discussions as suggested by Choi et al. (2016). The use of multiple research instruments is encouraged to ensure construct validity through triangulation, taking different angles toward the studied object (Runeson and Höst, 2008), which provide a stronger substantiation of our propositions (Eisenhardt, 1989). At the start of the research, in June 2015, group discussions were held with staff directly involved in the exploratory use cases or who were tasked with managing and maintain the systems. Between October 2015 and June 2017, individual interviews were held with personnel in the organizations under study. Internal documentation was selected which dealt with issues faced by the adopting projects. All interviews were documented in writing. The documents were then analyzed and transferred into an integrated case document (one for each case). The first versions of this document were then sent to the interview participants for feedback and clarification of open points. Once all the additional information feedback had been incorporated, the final version was reviewed and discussed with the main contacts at the organizations under study. Table I gives an overview of the sources used in the case studies.

4.1 Case X

The first case, the automatic measurement of water data, case X, is managed by a central government organization. Case X is a facility that is responsible for the acquisition, storage and distribution of data for water resources. Case X provides a complete technical infrastructure for the gathering and distribution of water data and delivers the data to various stakeholders within and outside the organization such as, hydro-meteorological centers, municipal port companies, flood early warning services and other private parties.

Case X was created from the merging of three previously existing monitoring networks and also includes data from third parties, including water data from foreign countries and other public organizations such as the shipping and transport industry, logistics, harbor management, meteorology, regional and local water management, and international water management. Case X has approximately 640 data collection points using a nationwide system of sensors. The data are then processed and stored in the data center and is made available to a variety of systems and users. Case X collects data regarding water levels, wind speed, wave heights, water temperature, astronomical tides, water currents, salt content, etc. These data are aggregated and calculated within models to accurately predict water levels and water quality. Based on these models, decisions are made to close storm surge barriers, close swimming areas, send out messages to shipping, etc. As such, we can classify case X services as being collaborative aware services (Gigli and Koo, 2011), as case X services are used to make decisions, and based on those decisions, to perform an action. Case X services have “terminal-to-terminal” communication, as well as “terminal-to-person” communication.

4.2 Case Y

Case Y is a decision support system for water management. Traditional water level measurement is performed using a level scale in fresh waterways such as ducts and locks. This is placed during construction of the asset and indicates the depth related to the soil (usually) a plurality of centimeters. Case Y automates this process with IoT measurements. The main pumping stations regulate the water levels in the region. Case Y manages approximately>3,500 km of polder ditches,>130 automated polder pumping stations,>20 automated inlets, >100 automated weirs,>100 remote level loggers,>80 smaller pumping stations,>200 smaller pumping stations,>3,000 fixed weirs and>2,000 fixed inlets. Case Y involves the regulation of the water level in streams, lakes, ditches, moats and canals. This is vital for industrial development, agricultural businesses, environmental management and recreation. The height at which the water level of an area is set depends on the use and function of that area. For example, although water levels in wildlife areas often fluctuates, farmers tend to prefer a relatively low water level to prevent their land from becoming too wet.

In the process screen of case Y, IoT measurements are displayed from telemetry, supplemented by estimations from the system itself. These include inland water levels, meteorological information and volumetric flow rate. The system reads precipitation from rainwater measuring stations every 15 min and water levels on the reservoirs which is measured at the polder mills. In addition, the system receives weather forecasts every 15 min via FTP. These are three files with 1 h, 3 h and 24 h forecasts of precipitation, wind and evaporation. The relevant level manager indicates which target level should be used and whether a precipitation protocol is active. The system in case Y then calculates the desired deployment of each reservoir mill for the next 24 h and makes a “request” for the use of pumping stations for the required time. As such, we can classify the services in case Y as being collaborative aware services (Gigli and Koo, 2011), as the services are used to make decisions, and based on those decisions, to perform an action. The services have “terminal-to-terminal” communication, as well as “terminal-to-person” communication.

4.3 Comparison of the case studies

Both cases are involved in water management in the public sector, and the number of measuring stations is similar. Significant differences are in the levels and how the systems are governed. Case X is a national system, which means that the sensors are spread over a wide geographical region and the asset management processes can affect large numbers of people. Case Y, on the other hand, is regional and has a mix of publicly and privately owned sensors. As such, the risk of poor system maintenance may be higher, but the geographical region is much smaller and there are far fewer people affected, significantly reducing the impact of risks involved. Table II demonstrates how IoT adoption affects asset management business processes and specifically how these effects are manifested in the cases.

As seen in Table II, all effects as expected in the literature were found in both cases. As such, we may infer that IoT adoption impacts the asset management process in a variety of ways, which will be discussed in the next section.