Get best of the week

Hardware testing in SIBUR

Expectations do not always correspond to reality. Unfortunately. And if another such inconsistency in life can become a reason for a sad grin, creating a new memechik or hastily rolling out a backup, then in the industrial sphere everything is a little different.

If some piece of iron that you made and began to use at the facility does not meet expectations, the consequences may be different. For example, a smartphone, which (according to the manufacturer) is declared to work at -40, turns off after a couple of minutes on these same -40.

Therefore, iron, that its own, that is vendor's, must be thoroughly and thoughtfully tested. It is most convenient to do this in a special laboratory.



My name is Justina, I am an engineer in Industry 4.0. Together with the owner of the IIoT product, Vasily Yezhov, we make up the SIBUR hardware development team. We already wrote that at the end of 2017, the digitalization process began in SIBUR, which continues to this day. We, like many other companies on the market, realized that now few products do without the integrated use of hardware and software - for a good result, it is necessary that both of these parts work efficiently. Here, as with smartphones and laptops, if the axis is in harmony with the hardware, you will be happy. If not, well, you get the point.

In our case, the situation is very complicated by climatic conditions: almost all devices at SIBUR's production facilities are located in open facilities and must withstand truly arctic temperatures. Therefore, all the equipment that is used in our digitalization projects must undergo fairly stringent tests before being in the field.

We have created our own laboratory in which we test both our own developments and vendor products. Why did you decide to do your own and not rent? Because paying contract laboratories or renting premises from research centers is too expensive, since most tests are quite lengthy. This is not a story in which you promptly throw a pack of modules for a test, they are driven away in half an hour, confirm the TTX declared by the manufacturer and give back to you. For example, we can carry out climatic tests for several days in a row, sometimes even several weeks, to make sure that the equipment can withstand operation in severe weather conditions.

It turned out to be more profitable to make your own laboratory at the PolyLab research center in Skolkovo, where SIBUR develops and tests polymer products. There was already a laboratory room and heat chambers that are suitable for testing iron. We just bought the missing test equipment (including the new heat chambers, those that were, did not fit us in the climate) and got to work.



What and how we test


First, SIBUR's own developments: explosion-proof temperature and vibration sensors, bluetooth beacons. In general, explosion-proof sensors are our hobby, they are guaranteed to withstand low temperatures up to -56 degrees Celsius. Secondly, devices from vendors. It is smartphones that pass the longest climatic tests, because we need to make sure that staff can walk with them in different conditions and they will not stop working at the most crucial moment. And thirdly, all other sensors from vendors, both with and without explosion protection.



In our laboratory, we test only consumer properties: we carry out climate tests, tests for strength and vibration, and other special tests. Previously, we also went to the contractors to test the vibration test bench, but recently we set up our own in Tobolsk, closer to the factories. When we develop the device from scratch (either ourselves or with the help of contractors), after passing all the internal tests, it is sent to a third-party laboratory for an explosion protection test - we ourselves are not entitled to carry out such tests without an appropriate certificate.

But testing in general is, in general, already the final stage. First, we go to production to our internal customer, see where the equipment will be used, in what conditions, what requirements it must meet. Then we decide on the supplier: whether we will do it ourselves, write a statement of work and give it to the contractors, or order the finished product from the vendors.

Further, the processes are slightly different, but regardless of the chosen supplier, we shift all the requirements that were determined at the first stage to the test program and test method, by which we test the finished iron. For each type of device, we have a separate program, on which the duration and parameters of the test usually depend.

Order from vendors


Each program and test procedure that we draw up for equipment from vendors is sent to them for approval. The vendor must confirm that it allows us to run the provided sample according to our methodology. If the vendor agrees, then the tests begin. If the vendor declares, say, the frost resistance of the smartphone, but is opposed to being tested in the cold, this is an occasion to draw a rather unambiguous conclusion.

Sometimes we purchase some semi-domestic sensors, which are usually made for housing and communal services or, for example, agriculture. The longest testing cycle was with smartphones. After all, with ready-made sensors like: if we want to test them, then on the market, as a rule, there are only 2-3 models of a specific sensor. But smartphones have to check for a long time. For each new iteration of the tests (with the load application at -40, with the target at -40, at -20, at +40), each of the tested smartphones must be recharged up to 100% again. Often, someone else comes running and brings several more devices after testing the main pack, and we have to test them.



We run them with different software. This includes at least the target software with which the phone will be used (for smartphones - an MTOiR proprietary application, mobile maintenance and repair), plus special load-bearing applications that test various nodes of the smartphone. With all this kit, we send the device to the heat chamber and watch how such a load passes at low temperatures.

One iteration takes an average of 6 days. Then we prepare a test report, send it to the customer. It takes another day, and then the total total test time is 7 working days. If we take into account the preparation time for the tests (installation of software, writing the program and test methods, coordination with the customer), the result is about three weeks. With sensors, the process is faster, because there is no need to wait for a complete discharge, as is the case with smartphones - the sensor battery is on average designed for 5 years of operation. If you test this parameter as well, the tests may take a little longer.

In our practice, there are cases when the vendor claims that smartphones or sensors withstand temperatures up to -40, and at -20 they turned off. Some models with the same -40 work less than an hour. Although we do not need it to work at -40 during the entire shift, as a person will not be at such a temperature on the street for such a long time, but at least 1 hour must be maintained to complete the necessary tasks.

Similar inconsistencies occur with both smartphones and home sensors. The reason is most often general: the developers declare the characteristics of the entire device by the components that were used in their manufacture: chips, diodes, capacitors and the rest. They look at the datasheets, see that the components withstand temperatures up to -40, and write that the sensor as a whole will work at these temperatures. But they forget that at low temperatures the characteristics of the constituent elements change significantly and problems with the connectivity of these elements to each other may begin.

We even encountered this problem in our own developments: the way the components work at 0 degrees or light frost is very different from the work of the same components at -40 and below. This point must be taken into account when developing circuitry, and when writing firmware, and when debugging a device.

Own development and contractors


It does not matter whether we make the device inside SIBUR or transfer the task to contractors, the development process is approximately the same: customer development, writing of technical specifications, selection of a development team (through tenders or internal selection). And then, together with the developers, we work on the product. On our side, we define the concept of the device: its characteristics, components, materials, appearance. And the developer is already making circuitry, writing the firmware, and participating with us in acceptance tests. Who cares, about the need for high-quality custdev, Vasily wrote here .

Most often, acceptance tests are carried out in our laboratory. Sometimes, if we don’t have any specific equipment, we still pay and go to third-party laboratories. But slowly everything is moving towards completely closing the cycle on itself: there are more and more orders for development, so the need for more specialized equipment is increasing, and we are buying it.

When the circuitry has passed our internal tests and we realized that it complies with the points of the program and test procedure, we send it for certification for explosion protection. To do this, submit a complete set of design documentation. If you do the opposite - first send tested for explosion protection, and then conduct internal tests - there is a chance that the circuitry will not pass our test and it will need to be changed, and then certified again.

After the certification stage, we again organize acceptance tests, but this time we also invite a functional customer from production. And if the test is successful, the equipment goes to production.

We did not immediately come to such a streamlined work scheme. When they made their first explosion-proof sensor, a development team and a delegation of metrologists from the factory were invited to the first internal test in the then-rented laboratory.

Of course, the first pancake turned out to be lumpy: all the observers who we carefully convened witnessed a little embarrassment - the piece of iron did not pass the test. But this story helped us improve the processes, because who wants to once again stand and blush in front of such a delegation.

Future plans


There are more and more orders for hardware development. This is due to the same digitalization in production - everything is moving towards predictive diagnostics, process optimization based on data. The same data driven approach that everyone is talking about. Many people need specific types of sensors, which simply do not exist on the market - they have to invent it themselves.

There was a need for vibration sensors. Even not so much internal in SIBUR as external, many companies, including metallurgical ones, are contacting us. There are also many requests for spectral gas analyzers. Now we are evaluating how efficient it is from an economic point of view, and if the development and production costs are justified, we will organize an additional laboratory specifically for testing gas analyzers. Under these projects, we will probably expand not only the laboratory, but also the staff ( here is a vacancy for the frontend developer IoT ).

And a little about people. We do not divide engineers into those who write TK and those who then conduct testing. Quite often, companies have two (at best) different people, one collects information, the other writes TK, the third one tests. We have the whole process of development from beginning to end oversees one person. He communicates with future users, finds out their needs, based on them draws up a technical task, and then tests it. This process integrity allows you to get a connected and high-quality result.

More articles:


All Articles