Determining, setting up and optimizing a maintenance plan can be a very laborious and expensive process. Is this plan maintained annually within your organization? Is asset data collected and actively used in determining the maintenance strategy? An optimal maintenance plan delivers financial and qualitative results for your organization! Supply Value has a white paper predictive maintenance written that discusses the different types of maintenance, a maturity level of your maintenance plan and how it can be implemented. But what is an optimal maintenance plan? In this insight, we take you through a concrete example, where we can determine and optimize your asset maintenance strategy by using historical figures and simulation technology.
As an organization you naturally want to get the most out of your assets. Asset utilization is the indicator with which it is possible to gain insight into how your organization is performing, also compared to the competition. This is quantified as the ratio between the realized output and the maximum theoretical output, i.e. normal working time. There can be innumerable reasons for not meeting normal working hours. These can almost all be included in the Overall Equipment Effectiveness (OEE). The OEE consists of three categories, which in turn are subdivided into the six KPIs that determine asset utilization. These KPIs are also known as the 'Six Big Losses':
- Asset failures (unplanned)
- Changeover times, adjustments and maintenance (planned)
- Waiting and small interruptions
- Low production rate
- Starting up the asset
- Defects and Rework
If an asset is in a poor state of maintenance, this is more likely to lead to production defects, loss of speed or even asset failures. This has a negative impact on the OEE, and thus the asset utilization. Assets in use do not escape wear and tear and their condition deteriorates over time. We call this asset degradation. But how can you manage this in such a way that the costs of your maintenance remain acceptable? Below is a description of how the degradation can be mapped by means of a simulation and the optimal maintenance strategy can subsequently be determined. A fictitious case shows the effects of a number of asset characteristics.
Although a simulation is a simplified representation of reality, it can still provide good advice for the maintenance strategy. The goal is to find out at what level of asset degradation ideally preventive maintenance should be planned, i.e. the optimal limit value M. Historical asset data is extrapolated over a significantly larger time horizon, making analysis of the data more reliable. Before the simulation can be performed, there are a number of preconditions that must be met.
In order to determine an optimal maintenance strategy, it is necessary that at least the following data is available over a representative period of at least three months. More available data leads to a more accurate simulation and fewer assumptions. In any case, it concerns:
- Condition of the asset;
- Timestamps of the times when both corrective and preventive maintenance took place;
- Cost of maintenance, both corrective and preventive.
The simulation will be completed in a number of steps.
Based on historical data, common asset characteristics can be identified such as: asset failures, the effect of preventive maintenance, variation in degradation levels and the ratio between the costs of corrective and preventive maintenance. All these characteristics have an impact on the optimal limit value M and the average costs per hour for the life of the asset. In reality, many more characteristics can be of influence, the chosen variables are therefore illustrative. The characteristics are discussed below.
1. Asset failures
From the historical data, it is possible to find out what the asset's degradation is and how it has changed over time. It also charts how often the asset suffered a failure (failure level), what the time interval is and how stable this level is. Disruptions can occur around the same degradation level (stable) or at random times (unstable). The latter can be seen in figure 2.
2. Effect of preventive maintenance
This analysis also provides insight into the effectiveness of maintenance. Maintenance can be 'perfect', in which the condition of the asset is brought back to new condition, or 'imperfect', in which it is only partially restored. An example of the degradation progression is shown in figure 2. The maintenance is perfect, as the degradation level is always reduced to 0 with each maintenance.
3. Relegation Level Variation
Asset degradation does not always take place evenly, a certain degree of variation always applies. Figure 3 shows that the degradation shoots up in big leaps with some regularity. Think of a train that drives over a stone or a machine with oil leakage, the condition of the asset deteriorates enormously in a short time.
4. Cost ratio between preventive and corrective maintenance
Finally, the realized maintenance costs can be mapped out. We naturally want to see this minimized, while asset utilization does not suffer as a result. To properly make this trade-off, it is important to understand how the two types of maintenance (corrective vs. preventive) are related to the 'Six Big Losses', and thus the OEE. Green means that the type of maintenance makes a positive contribution to the relevant 'Loss', orange has virtually no effect, while red has a negative impact and thus increases the 'Loss'.
Prepare and extrapolate data
However, to be able to calculate the optimal limit value M, much more data is needed to draw well-founded conclusions. This data is generated by simulation. To perform this, the data is loaded into a tool and simulated over a significantly longer period of time. In the examples discussed in this insight, the historical data spans over 4,000 hours and is simulated over 150,000 hours. This example includes degradation, degradation level, corrective maintenance, preventive maintenance, and maintenance costs. The simulation should be tailor-made for each unique case, because each characteristic has a different effect on the limit value M.
When simulating the data, the current limit value M for preventive maintenance is taken as the starting point. If this is not available, an assumption will be made. Based on the data, the average cost per unit of time (GKPT) can be calculated for each possible limit value M. In order to subsequently optimize the maintenance strategy, the lowest GKPT is identified. Various scenarios can be compared: the current strategy, the optimal strategy or if only corrective maintenance would take place. Figure 4 shows the results of the optimized strategy from the example: GKPT of €4.86 at a threshold M of 50.
The simulation does not give a conclusive result, but it does provide excellent advice about the limit value M and therefore the optimal moment to perform preventive maintenance. Naturally, the determination of the maintenance strategy also depends on other insights, asset characteristics and company-specific KPIs. Think of the situation in which delivery reliability is one of the most important KPIs. It is not inconceivable to lower the limit value M so that asset disruptions (and corrective maintenance) can be prevented. In addition, as can be seen in Figure 4, the surrounding limits have a relatively similar GKPT.
This insight paints a picture of the power of simulation in determining your maintenance plan. In this example it concerns a simplified representation of reality. However, the simulation can be adapted and used on a wide range of complex maintenance issues. Feel free to contact our consultants to see what they can do for your maintenance strategy!