I came to IT from CHP. At the thermal power plant of the Novolipetsk Metallurgical Plant (NLMK) he worked first as a cauldron operator, then as an engineer of the production and technical department (PTO). During the period of work in vocational training, we actively solved tasks related to the timeliness of stopping turbogenerators for cleaning, and at the thermal power plant we have already begun to develop the first digital advisors, so far in the boiler room. I immediately came up with the idea of ​​making a digital assistant that would assess the condition of the turbine capacitor and recommend when it is profitable to stop the turbogenerator for cleaning. At Tsifra, we have a team of people who are looking for ways to do something more efficiently, and if someone expresses a promising idea, developers immediately appear next to him. My case of creating a digital product was unusual in that I played the role of both an expert and a developer at the same time. The second turned out to be more difficult, but my fellow digitizers actively helped me get my hands full in this matter. As a result, first I expressed an idea, then I became an expert, then I calculated a mat model, and then, unexpectedly for myself, I found myself in the role of a developer in a digital team of energy engineers.

I will talk about the main technological processes at the CHP. We burn natural, blast-furnace and coke gas on boiler units, obtaining heat. We heat water with this heat and turn it into energy steam of medium (32 ata) and high (100 ata) pressure. Then this steam causes various units to rotate: compressors, turbofans, turbogenerators (TG). In this article, we are most interested in TG. These units produce electricity, and the steam, which has given off almost all of its potential and thermal energy, is sent to a condenser – a heat exchanger that serves to convert steam into water. The steam must be converted into water so that it can be more easily compressed and sent back to the boilers, starting the steam power cycle anew.

Our CHP is outside

Our CHP from the inside

Capacitor

The TG condenser is approximately 5,000 thin metal tubes in a box the size of a car, inside which cold water from the circulation circuit flows, and steam cools between them, turning into condensate.

Condenser tube bundle

Let’s repeat, steam passes between the tubes, which, cooling, condenses, turning into water, called the main condensate. Inside the tubes, on the contrary, the circulating water is heated. And who will cool this water, you ask? For this purpose, the thermal power plant has its own circulation circuit, in which the water heated after the condensers is cooled in cooling towers with the help of air. The main task of the condenser is to turn all the steam fed into it into condensate. The transformation of steam into condensate is accompanied by its compression, that is, a decrease in volume.

Recall that a condenser is a closed vessel that does not communicate with air. So, when the volume of steam decreases in this vessel, rarefaction is created, i.e. pressure less than atmospheric, it is called vacuum. The greater the difference between the condenser pressure and atmospheric pressure, the greater the vacuum.

And how is it profitable for the turbine to work so that the vacuum is greater or less? Answer: for a more efficient operation of the turbine, it is necessary to have a greater vacuum. And how to get a bigger vacuum? Answer: it is necessary to ensure a higher rate of conversion of steam to condensate. Accordingly, the question is how to do it. Answer: it is necessary either to supply less steam so that it has time to condense, or to ensure optimal cooling conditions. And what about cooling? This is the maximum flow of cooling water and a clean condenser. And this depends on the efficiency of the turbine and the amount of electricity it can produce.

That is, it is beneficial to keep the condenser clean. But it is not easy to do this, especially in the summer, because cleaning requires stopping or unloading the TG, which is always a loss in electricity generation.

At the same time, as the steam moves inside the turbine, it is selected for various needs (production and heating). It is selected for production with even greater potential, when it has passed only a third of the turbine. But for heating, steam is taken almost before the condenser. In this case, we can get almost the same amount of steam and electricity without loading the capacitor. However, there is an inconsistency here. Such a maneuver is relevant for summer, when it is difficult to cool circulating water in cooling towers. And in the summer, as you know, no one needs heating. So it turns out that a universal way to simultaneously increase the production of EE on the turbine and increase its efficiency is to maintain the cleanliness of the condenser tubes.

The capacitor is contaminated by various things: most often it is sand, silt, shells. These objects enter the cycle through the windows of the cooling towers and with the feed water.

The condenser should be cleaned approximately once every four months. There are two cleaning options: with turbine stop and without. The stop option is used during planned repairs or severe contamination. After stopping, mechanical cleaning is carried out. A more non-standard option for cleaning the condenser is cleaning it “on the go”. To do this, we stop half of it, disassemble and wash each tube. Then we collect everything back, add water to the cleaned part, do the same with the other half. This method makes it possible to lose production less.

Cleaning

The graphic describes the process of obtaining profit from the operation of the turbogenerator. After cleaning, the capacitor linearly deteriorates its performance, accordingly, the cost of generation increases.

At some point, we take it out for cleaning, and then the profit is negative (we do not generate and pay the contractor for cleaning). Next, you need to determine the largest area under the graph and solve a simple optimization problem.

This will give us a theoretical idea when to clean it. That is, it makes no sense to clean it too early, we will simply sit on the generation for the time of cleaning (up to zero, if we are talking about a complete stop). Cleaning it too late is also expensive, because the cost of generation will be too high. It is necessary to clean exactly on time.

Previously, it was cleaned either according to the schedule of planned and preventive repairs, which were tied to the repair of adjacent equipment, the number of hours of operation of the equipment and seasonality, or for visible deterioration of indicators. But still, the most important purpose of the digital indicator is to choose the optimal time for cleaning the capacitor with an accuracy of up to a day. From afar, this task may seem very simple, but in practice it turned out to be the opposite.

Math model

The first thing that comes to mind is to focus on the consumption of circulating water on the halves of the condenser. The smaller it is nominal, the more pollution and clogging of the pipe system. The biggest problem here was the banal lack of flow-measuring devices on circulation water pipes. The construction of supply and discharge pipelines does not have the necessary lengths of straight sections for installation of these same washers.

How can you get out of such a situation? We coped with this task by calculating the flow of circulating water through the heat balance of the condenser. But even here there are pitfalls. For turbogenerators with heating and production selection, it is necessary to calculate the amount of steam going into the condenser through the material balance of the turbogenerator, which also involves a number of flow measurement points of various flows. This inevitably affects the accuracy of readings when calculating the flow of circulating water.

We decided that let the accuracy be as it turns out, just add other indicators of capacitor pollution, which will be calculated based on the readings of various combinations of sensors. For example, calculations of the heat transfer coefficient, temperature pressure and heating in the condenser, as well as the vacuum value, were added. Then they added to the list of indicators such a macro-parameter as the specific heat consumption for the production of electrical energy of the turbogenerator (how much steam is needed to produce a kilowatt-hour).

Of course, a reference point was selected for each indicator. If for circulation water consumption it was nominal, then, for example, for the heat transfer coefficient it was the theoretical coefficient, and for the specific heat consumption for production – the value of this indicator after cleaning. As a result, we use a formula in which we tie the steam consumption to the cost price, differentiate the dependence, equate it to zero, get the extremum, clean in it.

Other indicators serve to confirm the verdict of the main formula. For example, we see a decrease in the heat transfer coefficient – confirmation of contamination of the heat exchange surface, we see a decrease in the flow of cooling water – confirmation of clogging of the condenser tube system by circulating water.

Now briefly about why they could not decide earlier. Various integrators came to us, there was even an offer from one institute. However, when dealing with them, it did not leave such a feeling that they had never seen a capacitor live. The first few proposals were of the level of “let’s take a theoretical model of physical processes and compare with it” or “let’s put a fact on the mode maps and try to understand the difference.”

All these methods were suitable, but differed in the complexity of creating and maintaining models with their low accuracy in the conditions of our set of measuring devices: most likely, we would put them on the table, because even if there was learning (or numerical approximation as data accumulated) ), “finishing” would take several years.

And I was engaged in the optimization of boilers in my bachelor’s and master’s degrees, and I became an engineer as a CHP engineer for the efficient use of energy resources. I considered the first boiler model in Excel with macros, but it was missing. The second was already done in Mathcad, and then it was necessary to make some clarifications. There I already started to learn Java (now I understand that it was a bit of a wrong path), but I completed the task. Here I managed to calculate everything necessary – and now already in the IT team I am modeling using Python.

So I found myself in two roles here: first, as a manufacturer who does not trust office IT specialists, because they have never seen a live unit, and then as an IT technician, but already understands the limitations of the system imposed by the real world. But what about those who “understand” – we can say that those who most experienced the complexities of production conditions on their project.