What should I implement first? APS or MES?
Controlling production on the shop floor is a process that has historically been important in industrial management. Productivity, repeatability, Taylor... It's all in the blood of industrial management. MES (Manufacturing Execution Systems) are here to improve the management of production activities, integrating management information from ERP software directly with shop-floor resources. Although it officially emerged in the 1990s, it was in the last decade that MES software became more widely used in Brazilian industry, with the possibility of managing production execution and the flow of materials quickly and assertively. A great solution for knowing what has happened and what is happening in your factory.
Production scheduling, one of the outputs of an MES solution, is a fundamental part of factory programming, because we need to know what has been done so that we don't produce again, taking up production capacity that would be available to make another product. An error like this can cause the schedule to start from a different premise than the real one, generating a series of invalid decisions. In other words, having an up-to-date scenario when planning or programming is important.
In addition to this timing aspect, accuracy can also be critical if it doesn't reach an ideal minimum level. Both for notes and for stock positions and movements, as an incorrectly recorded volume can trigger an order point for replenishment at an inopportune moment.
However, this premise of an up-to-date running scenario often raises doubts about the frequency, detail and accuracy needed from a scheduling perspective. This type of information requirement can give the impression that we need real-time reporting in order to be able to schedule, which is not true. At this point, we need to look at our current process and what we want for our future to see whether or not we are adequate.
If your company schedules once a day, at 7:30 a.m., this is the time every day when you need to have the running scenario up to date. If there is a shift change just before this time and the production leadership is organized to have all the notes in the system by the time the change takes place, you'll be able to schedule smoothly. If you have the same process in place for an afternoon shift change, you can already reschedule at that time too.
From the point of view of proactive scheduling, which has a standardized frequency and an initial but robust degree of maturity, the lack of a real-time note tends not to generate significant complications in its decisions.
Let's look at the example below, in which we have Scenario 1 of an initial schedule, at the beginning of production, and then two scenarios of a next scheduling cycle (Scenario 2 with MES notes and Scenario 3 via ERP). We have the premise that the MES will allow for more frequent production updates, as well as reducing the risk of human error due to manual entries. Let's analyze the situations in these two cases.
[caption id="attachment_3695" align="aligncenter" width="725"]
![What should I implement first? APS or MES?](https://cdn.prod.website-files.com/64f21c7ba1681bab1f84b99c/65403044b8aeca7a92af9d3a_large-Exemplo-Apontamentos-ERP-Vs-MES-02-1.png)
Execution scenarios and notes[/caption]
For the case with the longest lead time, which is operation A in R1, in scenario 2, the information is more accurate and we can see that there is already one more lead in this scenario than in scenario 3, which still has a larger balance to produce. However, this has not altered the sequence of this resource.
In the case of incorrect booking, in operation G in R3, it can be seen that in scenario 3 (via ERP) more was booked than was due, generating a lower balance to be produced than in scenario 2. This means that the completion forecast is incorrect with a gap of a few hours. Again, in this specific example, this has not altered the scheduling logic and sequence.
In short, the sequence and logic are the same. If there is a setup gain in doing A-B-C, in this sequence, it will be done even if it is done manually and there are some deviations in times and quantities. Of course, if the error or delay in booking is very high, the sequence will change. Here are some of the points that increase the risk of this happening:
- Very short cycle and process times, which generate many scheduling events, greatly increasing the sequence options and their dynamism (for example, an operation in a bottleneck sector has average process times of 15-20 minutes, which will imply a very large volume of orders scheduled in one shift);
- Very high volume of operations per order, with similar cycle times. If the factory has many macro-operations per order (for example, 10 or more), and they are balanced, any small variation has a domino effect that can be significant. In cases of better-defined bottlenecks, these variations tend not to impact the sequence itself as much, because the system as a whole depends on a few operations more specifically.
- Processes with many resource options to execute each operation. With more resource options, there are more possibilities for rescheduling as the execution deviates from what was planned. If a product in a given operation only has, for example, 2 or 3 machine alternatives, this risk is greatly reduced.
- High occurrence of corrective maintenance and unscheduled downtime, which can significantly reduce the adherence of the schedule compared to what is actually delivered by production.
In other words, proactive scheduling, which has a day and time to take place, can usually be carried out without the presence of an MES system, as long as we have a manual record of the work carried out at the same intervals as the scheduling (usually daily or by shift).
Another important point to evaluate when implementing an MES system, and one that consequently affects the frequency of entries, is the decision as to which data collection points will be used. It is often impossible to monitor all processes. To do this, it's worth understanding which processes are most critical in terms of capacity and reliability, these are probably the ones that should have greater control and deserve to be invested in. The others are often pointed out using backflush logic, which considers that if a subsequent operation has already been pointed out, then the previous one that doesn't have its own collection point has also been carried out in practice. The ideal minimum is to position these collection points in such a way that a production order doesn't go more than one shift without receiving a note at some stage.
However, even with a stable proactive schedule, we want to be able to reschedule at a different time when an unforeseen demand or supply situation arises, which we can call reactive. This type of scheduling is useful for resolving unforeseen situations, seeing all the impacts that a change in a Production Order can have on a factory as a whole. However, it should not be frequent, especially if we are making a change to the short-term frozen horizon, which should be changed as little as possible, because constantly making substantial changes to the plans we bring to the factory makes scheduling "nervous". We use this term because if the scheduler keeps changing production priorities every hour with the shop floor leaders, the reliability of that plan will be lost and confusion and potential errors will increase greatly. In other words, it's the kind of alternative to putting a "use sparingly" label on.
A programming process cannot be set up with the main focus on reactivity to very short-term unforeseen events. Normally, those who need to trigger this type of (re)action too much have some major problem behind them that generates this unpredictability, whether it's a faulty maintenance plan that causes a lot of machine breakdowns, or a lack of coordination between Sales and PPCP, causing constant changes in priorities. The introduction of APS generally helps to stabilize these processes that involve more areas, which is precisely why it starts at a lower frequency to get all parties used to the new model.
Even with these precautions, as you gain maturity, an on-demand decision at a critical moment, made assertively and quickly, is very valuable. But now, for reactive scheduling, do I need to have an MES beforehand? Many industries already have a standard procedure for taking notes (via ERP even) at the end of each unit of movement (pallet, reel, box, etc.) or at the end of each operation in a Production Order. If the cycle times of the processes we want to monitor are not too long, to the point where the complete processing of an order operation exceeds 2/3 hours, it is very likely that on-demand rescheduling is already possible. You won't be assertive 100% of the time, but you will be in the vast majority (remembering the example of the 3 scenarios we showed you).
An MES solution, in addition to its intrinsic advantages of providing greater control, is designed to give this whole process more speed, accuracy and control, allowing us to rely on assertive rescheduling with increasing frequency (if this increase makes sense and is productive). At the same time, this does not exclude our traditional process of a daily and/or weekly schedule, in which you set up your plan in a more proactive and structured way. In this way, MES improves reactive scheduling, but is not essential for proactive scheduling. Below is a graph showing the evolution of scheduling with maturity and the introduction of an MES system.
![](https://cdn.prod.website-files.com/64f21c7ba1681bab1f84b99c/6571d3332e96d5e78876dfd5_Design%20sem%20nome%20(19).png)
Where to start?
But then, even knowing that we don't need one to implement the other and vice versa, where do I start? I get this question a lot.
Although I'm a little suspicious because I work with APS directly and MES only indirectly, the answer is easy to get by understanding the company's priorities. Within an OEE analysis, if the productivity focus is on performance, MES is the one that will be able to show where you are falling short and why. However, if your main problem is availability, if there is a lot of waiting time, high setups and unbalances, APS will be the solution that addresses these problems. If you have very critical maintenance problems, MES can be a great way to work with predictive maintenance. If you have a lot of variation in demand and need to simulate scenarios, APS will do the trick. If you're experiencing productivity problems, don't know where your plant's problems are, have no idea how big they are and are looking for a complete diagnosis, start with MES. But if you already know where these problems usually are and you want to know what to do in order to make a decision, start with APS.
Factories are usually run by engineers. And engineers have a control freak, they want that diagnosis down to the comma. My tip is to pay attention to this perfectionism, because a regular biopsy with a concrete prognosis is worth more than a perfect autopsy.