The best-demonstrated cycle time for optimising the OEE power factor

While the availability and quality factors represent parameters that are relatively easy to measure for the purpose of OEE optimization, the performance factor poses a challenge for many companies. However, the concept of best-demonstrated cycle time combined with granular output measurement solves this problem for companies in a simple way.

The special feature of the OEE power factor is that it must be measured against an expected value. The question is therefore what output do I expect per unit of time of continuous production — i.e. without downtimes and defective products — and what output does the plant actually generate. Putting these two values in relation describes the power factor of the OEE.

Nameplate capacity

A simple way to define the expected output is the “capacity nameplate”, i.e. the statement from the manufacturer of the system how quickly it manufactures any product. For certain scenarios, this information is actually of great importance — for example when products are manufactured on a plant for which it was not originally designed. If there are large discrepancies in this setup between the possible deployment of the system and the actual output of the produced product, the loss of performance identified in this way is sufficient for a first major optimization step. However, this approach does not allow actual product-dependent differences in cycle times to be included in the optimization.

Challenge of product-dependent cycle times

In operational practice, it is often necessary to identify performance losses as a function of the product. In this scenario, a product-specific default speed is required to calculate the power factor. In the case of manufacturing large quantities, this value is determined in departments called “work preparation” or “industrial engineering” either from experience or through time studies. There can be many reasons why values generated in this way do not correspond to shop floor reality:

1. The default values were estimates and have never been accurate.
2. In the course of production planning, default values are misused by using them to correct the planning for production expectations that have not been met in reality.
3. The default values were measured a long time ago, but the system or the environmental conditions have since been changed so that they are no longer correct — although these deviations can be a reality both upwards and downwards.
4. After determining the default values through continuous improvement measures, employees have found a more clever way of operating the plant, so that the original default values are exceeded.

The above list only addresses the issue of the actual incorrect default values. In addition, it is common in operational practice that employees do not comply with the default values unconsciously or consciously by choosing operating parameters — but this challenge is not intended to be addressed in this article.

Best-demonstrated cycle time

How do you now obtain reliable default values for the planning system on a large scale, i.e. for many products? This is where the best-demonstrated cycle time comes into play as an intelligent analytics function from oee.ai.

Based on a granular data model that endlessly stores plant status and system performance in minute increments, an algorithm can identify the best-demonstrated cycle time for each material number produced. This is done via a rolling window that slides across the time series and identifies the historically best period of time. This is visible in the following animation. The graph represents the quantity produced over time:

Animation: Identification of the best-demonstrated cycle time with the OEE.AI algorithm

The width of the window can be defined in order to identify a stable state with certainty. Times of faster cycle sequences may be possible in the short term, but they do not represent a stable state of the plant. Depending on the production process, windows of 15 to 30 minutes have proven useful.

When using the algorithm, only windows in which there has been no loss of availability are taken into account. If the algorithm does not find a window of constant production of, for example, 15 minutes, this indicates a very unstable production process, in which the priority of optimization should actually not be on the performance factor but on process stability.

The results of the algorithm are made available either via API or csv. In this way, they can be fed back into the planning system.

With this technology, the best, actually produced cycle times for each individual product from the past can simply be identified at the push of a button and used in the planning system — and then confirmed again in real time by oee.ai or deviations can be identified.

Do you have any questions about this procedure? Then feel free to contact us at info@oee.ai.