In my previous article Maintenance Strategies Part 1 I introduced two theory’s of management — efficiency and effectiveness, and showed how maintenance strategies align themselves with these concepts depending on where the company is positioned in the boom and bust cycle.
In the boom times, effectiveness is the mot du jour, .i.e. to achieve the goal regardless of the cost; and in the bust times, efficiency takes the fore so to achieve the goal with fewer resources. However, the law of efficiency or work says that one cannot be 100% efficient and 100% effective at the same time. So what is the magic efficiency and effectiveness ratio?
Those that have studied management will know about contingency theory, which says (intuitively) that a function (task) takes on of various contingencies (dependencies) in the form external variables (Wiio and Goldhaber 1993). The effectiveness of a given task is contingent upon the demands imposed by the internal and external environments. The environment could be oil/gas share price (money), internal/external resources, limited employee competency, time, equipment and so on. The challenge at hand is performing the task under the constraints that are imposed by the environment. Moreover what is the correct balance between efficiency and effectiveness in order to produce the optimal outcome for a given situation?
My previous article used Rio Tinto as an example of how vehicle autonomy increased maintenance efficiency by prolonging vehicle life. Prior to the introduction of autonomy, preventative maintenance accounted for a large part of Rio Tinto’s operational budget. Rio Tinto realise that they were achieving effectiveness by selling more tonnes but not doing it efficiently, so they spent $400M and more to increase productivity, safety, mine efficiency as outlined in their Mine of the Future brochure.
In order for Rio Tinto to get an understanding of what was required to find the optimal efficiency/effectiveness ratio, Rio had to identify and evaluate all the components that contributed to the maintenance task and rank the components under a set criteria. In essence, what is required is to analyse the causal relations between the components that contribute to effectiveness and efficiency of a task.
Lesson # 1: Control
Organisation control is a major part of management theory, bureaucratic, normative and so on. Control theory says managers cannot control everything, so what should they control? The way managers answer this question has critical implications on the organisation. If managers control for just one thing, such as costs (efficiency), then other dimensions, such as employees, customer service and quality (effectiveness) suffer. The same applies for maintenance strategies, not all variables that contribute to the maintenance function can be controlled to achieve an optimal result.
In management a balance score card is used to define what is important and make comparisons. The same principle applies when formulating a maintenance strategy. In Rio Tinto’s case they defined all the variables that contribute to the maintenance function and then found their relationships.
There are a few methodologies that are able to help determine and quantify these relationships; DEMATEL and Pairwise are two, or my favourite approach is to apply Machine Learn techniques. These methods solve complicated and intertwined problems, and finding an optimal maintenance strategy is a complicated and intertwined problem when the efficiency and effectiveness paradigm is thrown into the equation.
For example, given five maintenance criteria such as work scheduling, work execution, inventory and personnel management, and quality control there are 4+3+2+1=10 relationships to consider, and for 10 criteria there are 45 relationships to consider.
In order for Rio Tinto to come to a determination on how automation brought an optimal outcome to their bottom-line, they had to perform a causal relationship study between all interconnecting components.
Lesson # 2: Productivity and continual improvement perspectives
In a managerial environment, the manager asks ‘can we continue to improve and create value?’
In most cases the answer is yes … to a point. Continuous Improvement (CI) is a principle that forms part of Total Quality Management (TQM), which says that in order to increase effectiveness and efficiency is to continually improve the process. In my last article, I spoke about how efficiency is defined as the amount of useful output (product/work) produced per the amount cost of resources or effort consumed. Well, productivity is a measure of how many inputs it takes to produce or create an output. The fewer inputs it takes to create an output, the higher the productivity. So applying this management concept we can say that in order to maintain a machine in the most productive way is to perform the minimum work possible that maintains the desired quality. This notion is contrary to the idea that time-based maintenance is crucial to maintain equipment effectiveness.
Inputs for a maintenance function
To truly understand the relationships between multitudes of variables, all the inputs that are associated with the maintenance function is required, and some of which are listed below. This list in a large organisation is huge; therefore the relationships are many and complex. Once all the tasks and responsibilities are documented, the relationships are determined and represented in a matrix and rank as a measure of importance.
Since it can become rather complex from here, the preferred method, from my point of view is to learn the relationships via k-means or similar statistical methods. The benefit of Machine Learning techniques is that it is dynamic, i.e. should the variables in the maintenance function change, then learning changes to suit. The Machine Learning algorithm then translates the importance and relationships into an easy to understand visualisation.
The effectiveness and efficiency ratio
As management theory says everything is contingent, here is no one size fits all. However, what I can offer is an formular. Since efficiency refers to an input-output ratio or comparison, and effectiveness refers to an absolute level of either input acquisition or outcome attainment (Pennings & Goodman 1977), then Efficiency is: Output/Input and Effectiveness is actual output – desired output.
To measure effectiveness mathematical values representing effort-input and production-output need to be created. Upon creating these values, create a matrix displaying the relationships between quantified values (via k-means, DEMATEL and Pairwise etc).
The end result will be on a form of a logit or s-curve. The cure shows that in all cases there is an ultimate efficiency ratio that can be approached but unfortunately can never be achieved. Pursuing the cause of better efficiency and effectiveness is usually worthwhile but up to a point, however it brings small returns as increasing complication, complexity and cost is required to achieve small gains. This is called the law of diminishing returns.
Rio Tinto’s situation exhibits the law of diminishing returns clearly. Automation drove efficiency and effectiveness to its maximum, but the cost to continually improve (CI) eventually becomes prohibitive.
To conclude, there is no one efficiency and effectiveness ratio that applies to all situations, it is contingent upon time and place. However, Machine Learning technology makes it easier to mitigate contingences by adapting and applying new dependency’s as they are found.
To know more about how Machine Learning can drive efficiencies in your business, contact me.