When it comes to automating processes in the petrochemical industry, the stereotype is often triggered that production is complex, which means that everything that can be reached is automated there thanks to process control systems. Actually not quite like that.
The petrochemical industry is indeed quite well automated, but this is true for the main technological process, where automation and minimization of the human factor are critical. All related processes are not automated due to the high cost of APCS solutions and are carried out manually. Therefore, the situation when once every couple of hours an employee manually checks whether a particular pipe is properly heated, whether the necessary knife switch is turned on and the valve is pushed in, whether the bearing vibration level is normal, this is normal.
Most non-critical processes are not automated, but this can be done using Internet of Things technologies, and not process control systems.
Unfortunately, there is a problem here - the gap in communications between customers from the petrochemical industry and the hardware developers themselves, who do not have customers in the oil and gas industry and, accordingly, do not receive information about the requirements for equipment in terms of use in aggressive, explosive zones, in harsh climatic conditions, etc.
In this post, we will talk about this problem and how to solve it.
IoT in petrochemistry
To check some of the parameters, we use walking tours for the purpose of visual and tactile inspection of non-critical parts of the installation. One of the common problems is related to the supply of steam. Steam is the heat transfer medium for many petrochemical processes, and it is supplied from the heating plant to the end node through long pipes. At the same time, it should be taken into account that our plants and installations are located in rather difficult climatic conditions, winters in Russia are severe, and sometimes some pipes begin to freeze.
Therefore, according to the regulations, certain personnel must make rounds once an hour and measure the temperature of the pipes. On the scale of the whole plant - this is a large number of people who are almost exclusively engaged in this - they walk and touch the pipes.
Firstly, it is inconvenient: temperatures are low, and you need to walk far. Secondly, in this way it is impossible to collect and, moreover, use process data. Thirdly, it is costly: all these people have to do more useful work. Finally, the human factor: how accurately is the temperature measured, how regularly does this happen?
And this is just one of the reasons why plant and plant managers are quite concerned about minimizing the impact of human error on processes.
This is the first useful use case for IoT in manufacturing.
The second is vibration control. The equipment has electric motors, and it is necessary to perform vibration control. It is still carried out in the same way, manually - once a day people walk and measure the vibration level with special devices to make sure that everything is in order. This is again a waste of time and human resources, again the influence of the human factor on the correctness and frequency of such bypasses, but the main disadvantage is that you cannot work with such data, because there is actually no data for processing and it is impossible to switch to maintenance of dynamic equipment as it is.
And this is now one of the main trends in the industry - the transition from routine maintenance to maintenance on condition, with the proper organization of which, an active and detailed record of equipment operating time and full control of its current condition are kept. For example, when it’s time to check the pumps, you check their parameters and see that pump A has managed to accumulate the required number of engine hours for maintenance during this time, but pump B has not yet, which means that it can not be serviced yet, it’s too early.
In general, like changing the oil in a car every 15 kilometers. Someone can roll it out in six months, someone will need a year, and someone will take even more time, depending on how actively a particular car is used.
The same goes for pumps. Plus, there is a second variable that affects the need for maintenance - the history of vibration indicators. Let's say the vibration history was in order, the pump has not yet worked by the hour, which means that we are not servicing it yet. And if the vibration history is not normal, then such a pump must be serviced even if the hours are not running. And vice versa - with an excellent history of vibrations, we serve if hours have been accumulated.
If all this is taken into account and maintenance is carried out in this way, then it is possible to reduce the cost of maintaining dynamic equipment by 20 or even 30 percent. Taking into account the scale of production, these are very significant figures, without loss of quality and without a decrease in the level of safety. And this is a ready-made case for using IIoT in an enterprise.
And there are also many counters from which information is now taken manually (“I went - I looked - I wrote it down”). It is also more efficient to serve online, to see in real time what is being used and how. Such an approach will greatly help in solving the issue of using energy resources: knowing the exact consumption figures, you can supply more steam to pipe A in the morning, and more steam to pipe B in the evening, for example. After all, now thermal power plants are being built with a large margin in order to accurately provide all nodes with heat. And you can build not with a margin, but reasonably, distributing resources more optimally.
This is the fashionable data driven decision, when decisions are made on the basis of full-fledged work with the data that has been collected. Clouds and analytics are especially popular today, at Open Innovations this year they talked a lot about bigdata and clouds. Everyone is ready to work with big data, process it, store it, but first the data needs to be collected. Less is said about it. There are very few hardware startups right now.
The third IoT case is staff tracking, perimeter navigation, and so on. We use this to track employee movements and monitor restricted areas. For example, some work is being carried out in the zone, during which there should not be strangers in it - and it is possible to visually control this in real time. Or the lineman went to check the pump, and has been with him for a long time and does not move - maybe the person has become unwell, needs help.
About standards
Another problem is that there are no integrators ready to make solutions for industrial IoT. Because in this area there are still no established standards.
For example, how things are at our house: there is a wifi router, you can buy something else for a smart home - a kettle, an outlet, an IP camera or light bulbs - connect all this to an existing wifi, and everything will work. It will definitely work, because wifi is the standard under which everything is sharpened.
But in the field of solutions for enterprises, there are no standards of this level of prevalence. The fact is that the component base itself became affordable relatively recently, which allowed the hardware on such a base to compete with the human resource.
If you visually compare, then the numbers will be approximately the same scale.
One SCADA sensor for industrial use costs about $2000.
One LoRaWAN sensor - 3-4 thousand rubles.
10 years ago there were only process control systems in general, without alternatives, LoRaWAN appeared about 5 years ago.
But we cannot just take and use LoRaWAN sensors throughout our enterprises.
Choice of technologies
With home wifi, everything is clear, with office equipment everything is about the same.
There are no popular and commonly used standards in terms of IoT in the industry. There are, of course, a bunch of different industrial standards that firms develop for themselves.
Take, for example, wireless HART, which the guys from Emerson made - also 2,4 GHz, almost the same wifi. The coverage area from point to point is 50-70 meters. Considering that the area of our installations exceeds the size of several football fields, it becomes sad. And one base station in this case can confidently serve up to 100 devices. And now we are equipping a new installation, there are already more than 400 sensors at the initial stages.
And then there is NB-IoT (NarrowBand Internet of Things), provided by mobile operators. And again, not for use in production - firstly, it is tritely expensive (the operator charges for traffic), and secondly, too much dependence on telecom operators is being formed. If you need to put such sensors in rooms like a bunker, where the connection does not catch, and you need to put additional equipment there, you will have to contact the operator, for a fee and with unpredictable deadlines for fulfilling the order for covering the object with a network.
It is impossible to use pure wifi on the sites. Even home channels are jammed with both 2,4 GHz and 5 GHz, and we have a production site with a huge amount of sensors and equipment, and not just a couple of computers and mobile phones per apartment.
Of course, there are proprietary standards of sane quality. But this does not work when we are building a network with many disparate devices, we need a single standard, and not something closed, which will again make us dependent on one supplier or another.
Therefore, the LoRaWAN alliance seems to be a very good way out, the technology is actively developing and, in my opinion, has every chance to grow to a full-fledged standard. After expanding the RU868 frequency range, we have more channels than in Europe, which means that you can not be shy about the network capacity at all, which makes LoRaWAN an excellent protocol for periodically collecting parameters, say, once every 10 minutes or once an hour.
Ideally, we need to receive data from a number of sensors once every 10 minutes in order to maintain a normal surveillance picture, collect data and generally monitor the condition of the equipment. And in the case of crawlers, this frequency is equal to an hour at best.
What else is missing?
Lack of dialogue
There is a lack of dialogue between hardware developers and customers from the petrochemical or oil and gas industries. And it turns out that IT people make excellent hardware from the point of view of IT, which cannot be massively used in petrochemical production.
For example, a piece of iron on LoRaWAN for measuring the temperature of pipes: hung it on a pipe, attached it with a clamp, hung a radio module, closed the control point - and that's it.
The equipment in terms of IT is absolutely suitable, but there are problems for the industry.
3400 mAh battery. Of course, it is not the simplest, here it is thionyl chloride, which makes it possible for it to work at -50 and not lose capacity. If we send information from such a sensor every 10 minutes, it will drain the battery in six months. There is nothing wrong with a piece solution - I unwound the sensor, inserted a new battery for 300 rubles every six months.
And if it is tens of thousands of sensors on a huge site? This will take a huge amount of time. By removing the man-hours spent on rounds, we get the same time for system maintenance.
A rather obvious solution to the problem is to install a battery not for 300 rubles, but for 1000, but for 19 mAh, it will have to be changed every 000 years. This is fine. Yes, this will slightly increase the cost of the sensor itself. But the industry can afford it, and the industry really needs it.
No one is a kasdevite, so no one knows about the needs of the industry.
And about the main
And most importantly, what they stumble on precisely because of the banal lack of dialogue. Petrochemistry is a production, and production is quite dangerous, where the scenario of a local gas leak and the formation of an explosive cloud is possible. Therefore, all equipment, without exception, must be explosion-proof. And have the appropriate explosion protection certificates in accordance with the Russian standard TR CU 012/2011.
The developers just don't know about it. And explosion protection is not a parameter that can simply be added to an almost finished device, like a couple of additional LEDs. It is necessary to redo everything from the board itself and the circuit to the insulation of the wires.
What to do
It's simple - communicate. We are ready for a direct dialogue, my name is Vasily Ezhov, the owner of the IoT product at SIBUR, you can write to me here in a personal or email - [email protected]. We have ready-made technical requirements, we will tell you everything and show what and why we need equipment and what needs to be taken into account.
Right now, we are already building a number of projects on LoRaWAN in the green zone (where explosion protection is not a mandatory parameter for us), we are looking at how it is in general, and whether LoRaWAN is suitable for solving problems on such a scale. We really liked it on small test networks, now we are building a network with a high density of sensors, where about 400 sensors are planned on one installation. In terms of quantity for LoRaWAN, this is not much, but in terms of network density, it’s already too much. So let's check.
At a number of high-tech exhibitions from me, hardware manufacturers generally heard about explosion protection and its necessity for the first time.
So this is, first of all, a communication problem that we want to solve. We are very much in favor of cusdev, it is useful and beneficial to all parties, the customer gets the right piece of hardware for his needs, and the developer does not waste time creating something unnecessary or completely reworking the existing pieces of iron from scratch.
If you are already doing something similar and are ready to expand into the oil and gas and petrochemical sectors - just write to us.
Source: habr.com