The main outcome of this thesis was the analysis of how different maintenance strat-egy related decisions affect the profitability of equipment owners and maintenance service providers. Finding ways to maximize profitability was the main concern here as this had not been the case in most of the previous studies in maintenance area, alt-hough it should have been. ABM was chosen to be the tool by which it was possible to show through simulations how profitability can be increased in maintenance con-text. The whole study has been quite theoretical and the AB model is a conceptual model, mainly due to the fact that no real world data was available for to be used as an input and reference to the model. However, the results of the simulations pointed out fundamental relationships between certain decisions and profitability, and know-ing these is valuable for researchers and practitioners.
Constructing and simulating with the AB model followed a standard modeling pro-cess presented in the introduction of this thesis. The theoretical review on the mainte-nance management literature was the foundation for the model. As being a conceptual
model, structural realism was achieved by relying on the structural realism presented in the literature. As being a rather experimental and explorative thesis – no particular problem was stated to be solved in the beginning – the actual tests by which the final results were obtained were fixed in the end of making the model. The model came out so that it would have been possible to analyze even more maintenance strategy related phenomena than what were chosen to be analyzed in this thesis.
The basic model structure was presented in the figure 7 (in the chapter 3.3) and can be summarized as follows: the four agents that are operating in the environment are customers, the machines of the customers, the maintenance company, and the work-ers of the maintenance company. The maintenance company is responsible for main-taining the machines of the customers, as these degrade over time and maintenance is required at certain point. At that point the maintenance worker goes to the location of the machine and maintains it. From this task, the maintenance company receives a payment from the customer, by which it covers the costs of having employed mainte-nance workers, and possibly makes profit. The machines of the customers produce revenue while running and after subtracting the maintenance costs from the revenue the profits of the customers are left. Different parameters and variables affect the abil-ity of the maintenance company to schedule the maintenance tasks as required.
In simplest terms, the answer to the main research question is that there are possible ways to optimize maintenance strategy in terms of maximizing profitability. The final results from the simulations answered to the question of how price and pricing logic (time- or value-based) affect the profitability of both the customers and maintenance companies in different maintenance points. And in addition, the simulation results answered how the error in machine condition measurement affects profitability in dif-ferent maintenance points while having a specific pricing logic. The specific results and their practical implications can be summarized as follows in the table 16.
Table 16. The results of the modeling and their practical implications.
THE RESULTS OF THE MODELING PRACTICAL IMPLICATIONS
1
Regardless of the pricing logic, it is possible to find an optimal static maintenance point, in which the average cumulative profits are maximized compared to the other possible static maintenance points.
Optimizing the maintenance point should be done in practice, as this maximizes profitability. In other words, if maintenance point is not optimized, prof-its are not likely to be maximized.
2
When the pricing logic of the maintenance services is time-based, the optimal maintenance point is different for customers and maintenance companies, regardless of the price.
In this setting, it is almost as if the best possible maintenance point for the customers is the worst for the maintenance companies, and vice versa. This is due to that the rationale in optimizing the maintenance point is different between the parties: the customers want to minimize the time needed for maintenance, while maintenance companies want to maximize it.
It is argued that time-based pricing of maintenance services leads to a zero-sum game that will result in unsatisfactory deal for the other party (customers or maintenance companies) in the long run.
3
When the pricing logic of the maintenance services is value-based, the optimal maintenance point for the maintenance companies approaches the optimal maintenance point for the customers, as the price for maintenance services is increased. However, if the price for maintenance services is low, the profits of maintenance companies are actually near to the lowest in the optimal maintenance point for the customers.
The above is due to that maximizing the profits of the customers (by this it is meant that the profits are max-imized with that given price, while increasing the price of course reduces the average profitability of the cus-tomers) tends to maximize the profits of the mainte-nance companies. However, customers must be willing to give a substantial share of their value to the mainte-nance companies. If the share is not substantial enough, there is no rationale for the maintenance com-panies to maximize customers’ profitability.
It is argued that value-based pricing of maintenance services can lead to a win-win game that could last in the long run, and should be favored compared to time-based pricing.
4
The error in machine condition measurement is a criti-cal parameter to making the optimal decisions on maintenance points. In general, the narrower the error range is, the higher the average profitability is for both the customers and maintenance companies. In addi-tion, the errors in condition measurement affect the optimal maintenance point (see more accurate explana-tions on this in the chapters 4.3 and 4.4). If we do not know the error range, or it is too wide, we cannot op-timize the maintenance point.
There is real systemic value in having more accu-rate condition measurement systems. The company (whether it was a maintaining or manufacturing company or both) could gain competitive advantage from having more accurate condition measurement systems than the competitors.
The results from this study are in line with the previous studies in maintenance man-agement. However, this study focused on finding ways to increase profitability rather than to improve operational efficiency in maintenance context. In addition, this study added the ABM perspective to the studies and elaborated more on the dynamics
with-in the mawith-intenance system. In terms of practical value, the results from this study support the recent developments in maintenance management. To the author’s under-standing, value-based pricing of maintenance services is more and more popular, as companies have noticed how this improves their performance. In addition, condition measurement systems are developing all the time, as companies want to act proactive-ly and know exactproactive-ly when maintenance is needed. There are huge business opportuni-ties for companies in improving condition measurement systems, and that is why most innovations in maintenance emerge in this area.
It is argued, that ABM is very eligible for studying maintenance related phenomena and could reveal new insights that are not to be found with other methods. For exam-ple, building the same kind of a model as in this study with system dynamics (which was the initial plan) would have been near to impossible, if we wanted that the num-ber of agents was adjustable in the model. In other words, system dynamics is a poor choice for the modeling tool when it is needed to change the structure of the system on the fly. In turn, with ABM, changing the system structure on the fly is one key fea-ture that supports using the methodology. In reality, not only the variables but also the structures change, so it is necessary that the modeling tool can cope with structur-al changes. However, not many tools are able to cope with structurstructur-al changes. It is quite bold to say this, but many studies are biased due to that the modeling tool that had been used was not eligible for the task.
To be a bit more objective, ABM is probably not the method without any disad-vantages. Although ABM has been praised this far for a few good reasons, the worst thing in it is probably the amount of work one must conduct in order to come up with a sophisticated model. Programming requires so much attention in many ways that quick results are unlikely. In turn, trying to get quick results most likely leads to too simple models or unacceptable errors. The time it took to make the model in this study was exaggerated a bit, due to that the author had to learn about everything (of maintenance strategy and ABM) from scratch. But still, if the methodology had been
familiar in the beginning, it would have taken a lot of time to come up with the model anyway. However, what one sacrificed in modeling time, gained in quality.