The subject of digital disruption informed by IoT (Internet of Things) and Industry 4 is omnipresent. These technologies advance productivity gains which many owner/operators of production plants in the sugar industry acknowledge and are investing in to support the future success of their businesses. When developing an appropriate strategy, one aspect, in particular, should be taken into account: automation. It forms the foundation for every digitalization project and is central to its success. The paper navigates through a two-stage process in which the author describes automation in the sense of an unshakable foundation for digitalization projects.
Digital disruption technologies advanced by the Internet of Things (IoT), cyber-physical systems and Industry 4.0 are increasingly being applied in the manufacturing sector to support automation. It is only a matter of time before there is a rapid uptake of the advances in sugar factories. Before examining the prospects ahead from implementing factory-wide digitalization, it is worth briefly reviewing the technological advances in the past that have informed manufacturing operations.
The three so-called industrial revolutions that have taken place to date are largely the subject of consensus among experts. Beginning with the development of the “Spinning Jenny”, the first industrial spinning machine, by the Englishman James Hargreaves in 1764, the first industrial revolution was characterized by the mechanization of manual work by machines using water and steam power. A good 100 years later, around 1870, conveyor belts were used on a large scale in the slaughterhouses of Cincinnati to transport the slaughtered pigs from one worker to the next. The assembly line is regarded as the first key technology of the second industrial revolution. Their era is marked by mass production and electrification based on the division of labor. Another 100 years later, the third stage: automation begins with the first programmable logic controller (plc) in 1969. Electronic and IT systems are crucial for increasing efficiency and fundamentally changing working methods and processes. And today, 50 years later? Some speak of the fourth industrial revolution, some of the next phase of the third industrial revolution. Only in future generations of history books will a uniform terminology have prevailed. What we have to note in any case today, are two aspects:
- The basis of digitization is the microelectronics of the third industrial revolution.
- If we compare the 50 years that have passed since the invention of the plc with the lifecycles of sugar factories, it is clear that it is practically impossible that all sugar plants around the globe have already completely implemented the third stage.
So, if the digitalization is considered the 4th revolution of industrialization, then most plants still have to master the 3rd revolution.
Automation in the sugar industry – draft of an ideal (but existing) image
In fully automated plants, there are no manual valves, motors, drives, or any other actuator out in the field, that is not operated centrally. There are no push buttons in the plant which an operator has to activate. You do not need people taking samples to the laboratory and you do not need to verify process values by checking personally. The plant can be fully operated from one central control room, just like a pilot can fly the plane from his seat (see figure 1). A fully centrally controlled and automatically operated sugar factory can be run by 3-6 operators only.
Figure 1. Fully automated factory operations managed by a few operators
A case in point is beet sugar plants in Germany, in particular, those run by Nordzucker AG. Here, the factory operations are effectively monitored and managed by 4 to 5 people in the control room. The fully trained operators are trusted to manage the plant, whether it is a weekday, nights or over weekends, with nobody from management around to seek for guidance!
In such plants, it is normal, that managers, operators or engineers access the KPI (key performance indicators) of the plant, or certain maintenance or laboratory data on the fly and with a tip of the finger from their smartphone or iPad. These includes performance of a centrifugal machine, for example, optimizing the washing times, and crystal quality data coming from the lab.
Such implementations require some means of data acquisition and data handling which allows the user to access only the requisite information. Preferably in a way which is already preprocessed to give some hints on possible improvements. A good example for such application and the needed preprocessing of data is the Siemens Sugar MIS. A sugar factory which delivers approx. 12.000 tags into such a system in a time range between 1 second and once a day (for some laboratory analysis) needs a sophisticated data compression.
By contrast: A glimpse in today’s reality
Anecdotal evidence suggests that currently, around 80% of all sugar factories in the world are far from even working on a centralized control system! What does reality look like in those 80% of the factories? If one goes through a sugar factory in southern Louisiana, in El Salvador, Indonesia, China or Kenya he or she will find some control areas where operators have a screen or maybe just a panel to see what may be going on (see figure 2).
Figure 2. Control panel is a cane sugar plant in Indonesia, a very typical installation for so many cane sugar plants globally.
You will not see too many active people, because the plant is running more or less on its own, with enough leeway just by enough mechanical flexibility. Most measurements, if existing in the first place have not seen a calibration ever since initial installation. The basic production data is shown on some white boards as written figures calculated from the day before end the operators are the real kings of the plant. They decide the performance they keep the plant running on whatever condition is not relevant for them (see figure 3).
Figure 3. Production “Dash-board” as seen in a cane sugar factory
Studies show, that approximately 65 to 80% of all installed PID (proportional–integral–derivative) control loops in this world are not tuned to best performance or even working in a satisfactory mode! Considering the amount of energy, being lost due to the lack of optimization in existing control loops the call for digitalization is understandable, however, this does not solve the root cause of the problem. Normal plant operators must be praised for their ability to overcome the failing automation systems. While they really know how to interpret a new noise, a smell or a leak, whether it is critical or just normal, they will react to it with a professional attitude and that may sometime means, just doing nothing and wait for the major failure which will shut down the plant.
In essence, the point here is, and it will be shown below in more detail later, that automation in itself is of no value as long as there is no additional application which will make it successful. And the measurement of success in automation is NOT the existence of a control room! The only true measurement for a successful automation is the reduction of human interface to the process. Each operation on the control system must be questioned and checked whether needed or not. Experience shows, that successful plants have come down to below 200 operations per day from raw material delivery to sugar drying. With such minimised operation a plant can be run like an airplane, where you can achieve a stable production and optimized energy usage while maintaining highest quality. On average, the energy efficiency of these plants is around 15 to 25% higher then the one with more than 200 operations per day!
A fully automated operation of the plant will show results in several other fields as well:
- Better utilization of installed production capacity. As the automation system supports rapid response, process operations can be maintained at upper limits and thereby boast increased efficiency compared with manual operations. For example, with manual operation a tank will have to be maintained at 60% level of the installed nominal capacity, but with automation, it would be safe to run it at 80 or 90% without a problem.
- Fast reaction to process problems, leads to reduction of losses and a higher yield.
- And, as recent comparisons have shown, potential of energy optimization of around 10-25% is possible as well as increased production effectiveness and stabilization at a higher level.
The path to digitalization in two levels
Most of the European sugar producers, in particular Nordzucker have invested in automation. Siemens has been involved in many of these projects. It has to be stress that the path to automation has been incremental to make sure it is built on a strong foundation. Experience gained by Siemens shows, that there are exactly two levels or stages necessary to create optimal conditions for a working digitalization strategy.
Level 1: Eliminate operator intervention!
Level 2: Create a basis for correct decisions!
The two levels embrace the following successive steps:
- Data acquisition
- Data verification
- Alarm management
- Optimization of control loops
- Integrated data management
- Predictive analytics
The basics for all such ideas lies in the fast, reliable and accurate acquisition of data. Traditionally, sugar producers consider this part of the digitalization story as a given status. Not much thought is put into this part. However, it is the key to any forthcoming conclusion.
Starting in the very beginning of the sugar production is the feedstock (sugar beet or sugarcane) and its quality impacted by weather and crop management.
Then, the logistics of the material transport, storage time on roads or at the factory provide the next set of decision parameters. But the real task is, again, in the pure production process. Literally thousands of datasets are being gathered every second. Hundreds more are provided from laboratory analysis, and again hundreds are additionally calculated throughout the day. These data are a treasure as such and in some cases, they are handled just as a sunken treasure: Gathered, stored and invariably forgotten.
In daily life most of the plant data are not considered valuable, due to the simple fact that they are not considered reliable. The validation of data is in most plants not considered important. One can see this in only a short view to the central operator screen. If you see even one Alarm blinking it is clear that the operator does not take it serious. And most of the cases this is because the alarm setting is wrong in the first place and has never been corrected. Other reasons may be that the transmitter itself need calibration which is not carried out. How to solve that? It is almost too easy, but it costs time and money. And the same management which wants to push digitalization does not spend the money on the daily optimization works in the plant.
Data verification on the other hand is a more complex task but possible in most plants without high investment. What is needed is the application of common sense to existing data. For example, a pump which is operating within its optimized working point has a defined mass flow at a certain speed. In case that the measured flow does not match the given speed of the pump, either the pump is failing or the flow meter needs calibration. Such deviation must be detected and the maintenance team needs to be informed. No general alarm needs to be sounded, it is enough that the maintenance crew will get an automatic message from the system and can act accordingly. In parallel this message must be recorded in the electronic shift book of the installed Management Information System. Once the deviation is analyzed and either solved or considered OK, the corresponding action must be recorded again, and logged away properly.
Similar to any other measurement. Each device will react to any kind of action in the flow of the process and thus it is possible to verify all signals. In applying this policy throughout the plant, it may be found that some instruments are not needed anymore and others are missing in order to analyze. These needs to be corrected. And then the integration of electronic shift book and laboratory reporting in line with the overall DCS (distributed control system) is a must for plants on the way to digitalization.
Alarm management is the next step which needs to be executed. Modern process control systems, such as Simatic PCS 7 allow highly sophisticated methods to group alarm messages, and to give an alarm hierarchy or alarm avoidance procedures. All such measures will result in a reduced work load for the operators while maintaining overall alertness. Combined with the possibility to open alarm handling SOP (Standard Operation Procedures) in case of alarms, the operators have the means to respond immediately.
Analysis of devices in the field such as pumps, motors, gear boxes, valves etc. under the aspect of condition monitoring are the last of measures and packages to be integrated into the DCS systems. This will allow an early detection of any deviation in performance and will give appropriate information to the maintenance department to act accordingly.
Optimization of control loops
Once the alarms are eliminated from the measurements, the control loops and interlocks need a detailed review. Loops or sequences which are running in manual that are not working need to be elicited. If the operator has to adjust setpoints or values, the loops may be working but are not optimized. In most cases these loops were probably running adequately during commissioning but afterwards nobody ever tuned them to the current process conditions. It might be necessary to store a fixed set of parameters for different plant conditions or to use auto correction of parameters based on rules of operation.
Such implementation again consumes time, know-how and money. And here it is not easy to decide who can really help. There are literally hundreds of “consultants” offering their service on such optimization. However, one needs good knowledge of advanced process control strategies to really provide added value to the plant.
It is extremely helpful to implement engineering standards for the structure of the automation system in order to help operators and plant engineers to work on the same side of the problem. Once fixed engineering procedures are implemented and accepted, optimization is standard.
A practical example
Studies assume that 75% of all installed control loops in process plants globally are NOT operating in the optimum performance range. That means, that even if you do have some control loops active in any kind of control system they are either in manual mode or not tuned to the best performance (see figure 4).
Figure 4. Semi tuned control loop in an automated factory
What you see here is a control loop or level control in a cane sugar evaporation unit. The level setpoint is at 31% and the process value, quite stable at 33% But the valve action is modulating between 0 and 65%!
As long as the rest of the evaporation is running stable (and no batch pan will pull vapor) the control action will keep the level constant. But as soon as you have such external effects on the control, you will immediately see the level fluctuates and thus the brix changing because the actuator (the valve) is already acting to the limit its performance under normal conditions! The control loop is not optimized in the sense of the process section.
In visiting ‘so called’ automated sugar factories you will find screens, full of alarms, manually operated pans or process sections and operators being very busy to just keep the plant up and running. Based on process know how, and standardized engineering guidelines it is possible to reduce the operator interference into the process down to one operation per hour. That already would allow the plant to almost run like an airplane. Once that is achieved, the way to digitalization is paved.
Integrated data management
The steps described so far lead to level 1. Once performed, you will have eliminated most of the operator’s interventions. Now it’s time to accomplish level 2. We have defined this level as achieving a foundation for correct decisions. Earlier, it was noted how to integrate the data from process, laboratory or energy production. The main data collection system within a modern sugar production plant is the central Control System. In most plants this is a so called DCS.
The name is misleading, but this is due to historic reasons, when in comparison to huge really central mainframe computers it became possible to use smaller control systems, which could be installed in a decentralized location in a factory but are still able to communicate all information to the central control room. These Systems are organized in an object-oriented way. This means each and every object in a factory which is providing data has a unique number, the so-called tag-number. Such a tag, can be a pump, a valve, a tank or just a transmitter or a single switch.
Traditionally all the tags are operated within the processor unit of the DCS. That means within the DCS all information to each tag is available. That also includes all information to Alarm settings, operator interfaces etc. Every additional system which now needs information from the plant can access the DCS and will be provided with the respective information.
The most relevant system for further data handling is the Management Information Systems MIS. Such a system uses the same object-oriented approach, which has proven to be so effective over the last 30 years. As each tag belongs to one process section an additional level of data management is introduced in the respective process section.
Each process section is defined by certain input parameters and respective output parameters. And as all process sections are closely linked to each other, either through product flow, or energy flow, or both, it is clearly visible, that disturbance in one section will lead to problems in the next one. These key parameters may differ from plant to plant, but in the end, they are the values which need to be shown in any kind of dashboard and which will indicate what the plant is doing at any given moment in time.
For further optimization it is a good idea to look into the development of the airplane industry. Not too long ago it was common that the crew of a commercial airliner consisted of 3 highly specialized pilots. One was acting as navigation and on-board engineer. S/He got eliminated as soon as it was possible to combine the available data into computers and into advanced process control procedures. The same is possible in sugar plants. Such ideas were developed in the 1990s. Based on advanced control strategies such as Fuzzy Control and Neuro-Fuzzy Algorithms it became possible to set some process section into a fully automated operation. In the beet sugar industry, this was, for example, drying cossette in high temperature dryers. Or the control of a milling line from bagasse humidity, imbibition water and raw juice brix with the target of optimized extraction.
However, at that time the overall availability of data was not given in the same way as today. Today’s control systems have in some cases the built-in functionality of model predictive control loops. Such controllers allow for a mathematical model of the loop, permanently checking the incoming parameters with the output parameters and thus executing a self-learning controller which will handle such complex controls, even if they have a long inherent dead time.
Added to that the possibility to analyze process, laboratory and any other value within the MIS Systems brings the possibility of a statistical process control package. Such package will combine historic data on a totally different method. These data will be analyzed as a pattern, and similar patterns from the past will produce a message that a certain event is likely to happen. What is known from the weather forecast, becoming more accurate every year is also possible in process control. Here with the advantage that the amount of data is limited to a fixed set of parameters in the surrounding of the plant. All these data are available in the above described MIS packages and “only” need the analyzing software attached to it.
Digitalization – still missing?
Digitalization only needs to be applied and it needs fundamental engineering understanding to make proper use of it. For instance, an app on a mobile device will only show what has been provided by the plant in the first place. Use this – but wisely. Examples in many plants around in the cane sugar industry show how “digitalization” – and not even automation is not carried out properly: You will find automation installations, where the plant managers and process people will proudly show you that the systems are working, fully operational for more than 15 years, and where even advanced monitoring systems are being installed. Showing the plant in 3-dimensional reality. Only the data shown on the screens are all not valid, because the instrumentation is either switched off, not calibrated (even worse) or not operational. So, the whole investment into the 3-dimensional digital twin of the factory is a complete waste of money without any benefit for the plant operations or the operators, who have decided to go back to the old panels!
Conclusion: it’s all about the foundation
To create optimal conditions for a working digitalization strategy, the two-step approach described is imperative. Once mastered, your foundation is in place and you are ready to build your digitalization building. What this will eventually lead to operationally at a sugar plant is:
- optimized processes and thus more output with less energy consumption
- less operator intervention and thus more efficiency and higher quality at the same time
- integrated data management and thus better decisions on a verified basis
Siemens is one of the market leaders in automation. The company’s automation specialists accompany customers not only to choose the best automation foundation but also on their way in the digital future, providing reliable products, complementing services and tailor-made solutions.