Conclusion and Recommendations
List of Figures
Figure I – Cognitive Mapping
Figure II – Ishikawa Diagram
Figure III – Why Analysis Diagram
Figure IV – Problem Space
List of Tables
Table I – Simulated Plant Scenarios
Table II – Internal and External Stakeholders
Table III – Ideas/solution generation
Table IV – Stakeholders Interest – Influence Mapping
Table V – Communication and Action Plan
Table VI – Morphological Analysis Box
Figure V – Cross Consistency Assessment
Table VII – Attribute Weights
Table VIII – Solution Generation
Table IX – Comparative Analysis
Table X – Plus/Minus/Interesting Evaluation of Preferred Solution #
Table XI – Plus/Minus/Interesting Evaluation of Preferred Solution #
Table XII – Plus/Minus/Interesting Evaluation of Preferred Solution #
Table XIII – Castle Technique Evaluation
Table XIV – Force Field Analysis of Preferred Solution #
Table XV – Force Field Analysis of Preferred Solution #
Table XVI – Force Field Analysis of Preferred Solution #
Table XVII – Potential Problem Analysis of Preferred Solution
The objective of this report is to present the issues related to successive equipment/machine failure within a Biorefinery industry in Angola.
This report presents several complex problem-solving techniques that could be fully applied to the situation that unable the PC&M (Planning and Control Maintenance) department to interpret the information from large datasets from the machineries. The PC&M (Planning, Control and Maintenance) department is a subcontractor of the maintenance head division of Biofuel Inc. The analysis began by adopting brainstorming models such as Causal Mind Map followed by Cause-and-Effect Fishbone Diagram. As the understanding of the problem evolved a Why Analysis was used to identify the boundaries of the problem and the solution space. Once the boundaries have been identified, the recognition of stakeholders enabled the interaction amongst several stakeholders. The interaction was possible through mapping the stakeholders for a comprehensive joint problem solving. As the consensus was reached on the problem space, techniques such Morphological Analysis, Plus/Minus/Interesting, Castle and Force Field Analysis have been used for solution search and solution evaluation.
The combination of models and techniques allowed the team involved in the problem solving to identify three preferred potential solutions related to equipment/machine failure. Through further analysis using the Plus/Minus/Interesting and Force Field Analysis it was possible to narrow it down to a single chosen preferred solution. The preferred solution was chosen by carrying out a comparative analysis and the weighted score indicated the suitability of the solution for the given scenario of the complex problem. Knowing that solution implementation is not readily accepted automatically (Proctor, 2011) a PPA (Potential Problem Analysis) was presented in order to alleviate the risk of new problems occurring during the implementation of the chosen solution process (Proctor, 2011). The implementation plan took into consideration the budget, the time and possible project cost escalation.
Recommendations presented included stringent requirements for getting budget approval in order to expedite contractual agreements IoT (Internet of Things) providers, AI (Artificial Intelligence) companies and Outsource Expertise. It was assumed that if all of the requirements are met then the successive machine breakdown rate would achieve rates considerably low.
According to Czinkia and Hentschelb (2015) a problem is considered to exist whenever there is a gap between an initial and a desired situation or when a living creature has a goal but does not know how this goal is to be reached (Fischer et al., 2012). Problems are often categorized as simple, complex and complicated. Simple problems are referred for cases when there is a clear initial situation and a known way to achieve the desired solution (Czinkia and Hentschelb 2015), complex problems are referred for cases composed of many parts that interconnect in intricate ways and the domain has many features that are interdependent of the system/environment (Fischer et al., 2012), and complicated problems are referred for cases when there is a quantitative aspect that one could measure, count or weigh, i.e. if quantity becomes extreme then the problem will get complicated. McLeod and Childs (2013) used the Cynefin framework (Snowden, 2010) to elaborate further the problem categories and approach it into domains. Simple problems referred as simple domain are characterized by a clear cause and effect, there is often a right answer and a standard operating procedure. Complex problems referred as complex domain are characterized by unpredictability and flux, the cause and effect are only understood in retrospect; experimentation is required to find answers. Complicated problems referred as complicated domain are also characterized by a clear cause and effect, but, with multiple right answers, oftentimes, expertise is required to choose the appropriate answer (i.e., good, rather than best practice).
Within the biorefinery industry, a large amount of data from equipment and machineries is logged 24/7 to monitor the health state condition of the running equipment/machinery. The data logging process is carried out by many sensors such as accelerometers, thermocouples and operated computer vision reporting system. I am part of the PC&M (Planning, Control and Maintenance) department a subcontractor of the maintenance head division of Biofuel Inc. The department is responsible to operate and manage the amount of scattered data logged 24/7. Data analysis is often required to extract useful information that leads to important strategic decisions to be made about planning maintenance schedules to enhance business benefits such as improved performance, increased reliability and reduced cost (MathWorks, 2017). As the feedstock process demand increases, large datasets of information are produced and stored in multiple locations with multiple formats, and their analysis becomes initially a complicated problem due to the high dimensionality of responses from the equipment/machinery throughput and the number of observations (Riani, 2012). To capture the real scenario, it was necessary to draw a cause – effect diagram (see appendix I) rooted by the principles of causation (Aurisicchio et al., 2016) in the higher number of observations in the data. Using the data log history in retrospect (Snowden, 2010) it was evident a direct linear correlation between high process demand and poor interpretation in the data analysis. Thus, according to Gano (2007) this correlation is understood as a static cause that exists over time prior to an action bringing them together to cause an effect. The dilemma consists on the large volume of datasets from several signal sources that yields atypical observations on the data masking one another. Traditional methods of static data analysis reveal deficiencies to capture the whole monitoring scenario. To address the issue of deficiencies the PC&M team decided to introduce within the system dynamic data analysis as a pilot project through experimentation to find answers (Snowden, 2010). Both analyses run in parallel and have introduced many interdependent variables that brings complexity on their interpretation. As a result, disruptions on feedstock supply occurred more frequently, thus affecting the overall efficiency of the biorefinery and the revenues.
According to Hujainah et al. (2018) the stakeholder is any person or organisational group with interest in or ability to affect the system or its environment. The stakeholders play a key role in producing an accurate list of requirements that must be implemented. Several definitions available in the literature (Eden and Ackermann (1998) and Johnson and Scholes (2002)) presents different views and opinions about stakeholders as people or small groups with the power to respond to, negotiate with, and change the strategic future of the organization. With respect to my organisation I have adopted the stakeholder identification and mapping protocol to frame key potential stakeholders, that have and/or can influence or be affected by the problem (Kanchana et al., 2018). To identify stakeholders, I managed to adopt techniques such as organising participation and creating ideas for strategic interventions including problem formulation and solution search (Bryson, 2004). I have followed a basic procedure of a brainstorm session with my colleagues to draw out a list of potential stakeholders to mapping their potential interests, assess their influence and importance with respect to the company. Two meetings were held, I co-chaired one of the sessions along with the PC&M coordinator. Other participants of the meetings were two senior data analysts, a senior materials engineer and planners. The sessions were conducted with active collaboration from the team as their inputs allowed me to rank the stakeholders according to their importance within the organization and/or department considering the stakeholder’s power, legitimacy and attention-getting capacity (Mitchell et al. 1997). A list table of interest vs influence is shown on the Table IV.
In the follow up process of ranking the stakeholders, I had to deal with some minor disagreements within the brainstorming team and I faced further difficulties to develop a matrix diagram of interest versus influence diagram (Bryson, 2004, 30). To overcome this issue and to obtain coalition and support, I have engaged with my colleagues through an active group discussion to illustrate the implications of the resulting stakeholder placements (Bryson, 2004). According to Bryson (2004, 31) importance versus influence grids (see appendix II) typically help determine which players’ interests and power bases must be considered in order to address the problem or issue at hand. Upon reflection on the issues of successive disruptions and machineries breakdown the team concluded that a problem really exists. Several alternatives have been proposed through brainstorm activities identifying everything that could contribute to the solution of the problem (Timmermans, 2016).
Efficient maintenance plan model from data logs is crucial to avoid low productivity, downtime and poor machinery performance conditions. The PC&M team is unable to interpret the information from large datasets. The demand of keeping the machineries running with minimal disruption must be achieved.
Models help us understand a complex problem and its potential solutions through abstraction (Selic, 2003). Mental models and cognitive mapping are powerful tools to identify cause and effect relationship among various strategic elements (Ackermann et al., 2005); it is used for modeling dynamic systems as it describes expert knowledge of complex systems with high dimensions and a variety of factors (Soares et al., 2018). Several models such as Cause – Effect Ishikawa diagram and Cognitive Map are widely used to reach a consensus to identify the boundaries of the problem. Figure I shows a cognitive mapping developed by a team composed of planners, engineers, data analysts and the maintenance manager during meeting sessions where every participant was encouraged to brainstorm for questions on the problem rather than looking for answers. This approach was in line with Gergersen (2018) where the author points out that brainstorming for questions rather than answers makes easier to push past cognitive biases and venture into uncharted territory. From the cognitive map (Figure I), I identified the following categories related to personnel, data management and maintenance. In order to explore further the problem categories, I took the initiative to promote a brainstorming session in a more restricted environment with few participants to avoid altering the character of the group so that individuals may get airtime and avoid to forget what they had in mind and also to voice over shy individuals who may have a lot to offer but hesitate to contribute in the context of a large group (Buggie, 2003).
I have followed the Osborn’s (1991) key rules. The attendees on the session, were the data analysts, technicians and mechanics. Essentially, the environment was aimed to allow the participants to think up as many ideas as possible without fear of criticism. My role was as a leader responsible to maintain a rapid flow of ideas without evaluating or judging any ideas during the process. This was in line (Osborn’s, 1991) to avoid alienate the members. From the session, plenty of inputs on several causes of the problem was evaluated and the results allowed to produce a cause-and-effect Ishikawa diagram (see Figure II) (Chemweno et al, 2016).
The brainstorm for the Ishikawa diagram proved to be very lengthy to derive the causal analysis and initially required some expert personnel (Guerin, 2015 and Medina-Oliva et al., 2012), to overcome this issue I suggested to place all of the most important causes closest to the main axis and the least probable causes placed further away from the main axis. My suggestion was in line with Konstanciak (2012) to allow the team to focus on the most relevant aspects of the problem.
Abbildung in dieser Leseprobe nicht enthalten
There was some differing views and opinions on the application of mental models amongst the team. The differing discussions pointed out that mental models often emphasise a succession of discrete events rather than underlying patterns of behavior, i.e. ignores dynamic elements including feedback loops, time delays and accumulations (Axlerod, 1976 and Forrester, 1971). Despite this shortcoming, I managed to explain them that mental models with visualizations helps more quickly decide upon an appropriate course of action (Maynard and Gilson, 2014). To further elaborate possible links of machine failure I then proposed to conduct a “Why” analysis (see Figure III) on the Ishikawa diagram to evaluate qualitatively more links that have been hidden in the whole process.
Abbildung in dieser Leseprobe nicht enthalten
In evaluating the Ishikawa diagram (Hassan and Jalaludin, 2016) and “Why” analysis (Hassan and Jalaludin, 2016) I faced some challenges as a result on the lack of familiarity amongst some colleagues about the Ishikawa Diagram and the Why Analysis. It was necessary to delve into research literature to explain the suitable application of these diagrams. In practical aspects the combination of models was beneficial such as to find associations amongst all categories. It was evident that links of machine/equipment failure was associated with data log acquisition management. Table I highlights several scenarios in the data exploration approach that was derived from the operational perspective of the machines/equipment.
Abbildung in dieser Leseprobe nicht enthalten
In reflection, the combination of models using cognitive map, Ishikawa diagram and 5 Why, it was experienced that applying the Cause and Effect diagram technique proved to be difficult, and oftentimes is not an optimal strategy to adopt as it is too time consuming and requires expert personnel (Guerin, 2015 and Weber et al., 2012). The “Why” analysis was used to overcome the shortcoming on the Ishikawa diagram as it supported to find out the main possibilities that contributed towards a machine/equipment failure (Hassan and Jalaludin, 2016). For instance, the “Why” analysis gave us insights that vibration and high temperatures are caused by factors that are maintenance-related and data-related. The data relation is affected by personnel which lacks experience in interpreting and processing the information from the logs. Thus, some major features of the problem boundary are within the personnel and data related categories and, in my reflection, the solution space needs to be investigated within these boundaries.
- ISBN (eBook)
- Institution / Hochschule
- University of Lincoln