There is a large CHP. It works as usual: it burns gas, produces heat for heating homes and electricity for a common network. The first task is heating. The second is to sell all generated electricity in the wholesale market. Sometimes snow appears even in frost with a clear sky, but this is a side effect of cooling towers.
The average CHPP consists of a couple of dozen turbines and boilers. If the necessary volumes of electricity and heat generation are known exactly, then the task is to minimize fuel costs. In this case, the calculation is reduced to choosing the composition and percentage of loading of turbines and boilers to achieve the highest possible efficiency of the equipment. The efficiency of turbines and boilers depends heavily on the type of equipment, the time without repair, the operating mode and much more. There is another problem, when with known prices for electricity and heat volumes, you need to decide how much electricity to generate and sell in order to get the maximum profit from working in the wholesale market. Then the optimization factor – the profit and efficiency of the equipment – is much less important. The result may be a mode where the equipment is absolutely inefficient, but the entire generated electricity can be sold with maximum margin.
In theory, all this has long been understood and sounds beautiful. The problem is how to do this in practice. We started a simulation of the operation of each piece of equipment and the entire plant as a whole. We came to the CHPP and began to collect the parameters of all nodes, measuring their real characteristics and evaluating the work in different modes. Based on them, we created accurate models to simulate the operation of each piece of equipment and used them for optimization calculations. Looking ahead, I’ll say that we won about 4% of real efficiency simply due to mathematics.
Happened. But before describing our decisions, I will talk about how the CHP works in terms of decision-making logic.
The main elements of the power plant are boilers and turbines. The turbines are driven into rotation by high-pressure steam, rotating, in turn, electric generators, which generate electricity. The remaining energy of the steam goes to heating and hot water. Boilers are places where steam is created. It takes a lot of time (hours) to heat up the boiler and accelerate the steam turbine, and this is a direct loss of fuel. The same goes for load changes. You need to plan such things in advance.
The CHP equipment has a technical minimum, which includes a minimum, but at the same time, a stable mode of operation, in which it is possible to provide a sufficient amount of heat to homes and industrial consumers. Usually the required amount of heat directly depends on the weather (air temperature).
Each unit has an efficiency curve and a point of maximum value for operating efficiency: with such a load, such a boiler and such a turbine give the cheapest electricity. Cheap – in the sense of with a minimum specific fuel consumption.
Most of the CHP plants in Russia are with parallel connections, when all boilers operate on one steam collector and all turbines are also powered by one collector. This adds flexibility when loading equipment, but greatly complicates the calculations. It also happens that the equipment of the station is divided into parts that operate on different collectors with different vapor pressures. And if you add the costs of domestic needs – the operation of pumps, fans, cooling towers and, to be honest, saunas right behind the fence of the thermal power plant – then the damn leg will break.
The characteristics of all equipment are non-linear. Each unit has a curve with zones where the efficiency is higher and lower. It depends on the load: at 70%, there will be one efficiency, at 30% – another.
Equipment differs in characteristics. There are new and old turbines and boilers, there are units of different designs. By choosing the right equipment and loading it optimally at the points of maximum efficiency, you can reduce fuel consumption, which leads to cost savings or greater margins.
How does a thermal power plant know how much energy is needed?
Planning is carried out for three days ahead: in three days the planned composition of the equipment becomes known. These are the turbines and boilers that will be included. Relatively speaking, we know that today five boilers and ten turbines will work. We cannot turn on other equipment or turn off the planned, but we can change the load for each boiler from minimum to maximum, and collect and reduce power in turbines. The step from maximum to minimum is from 15 to 30 minutes, depending on the unit of equipment. Here the task is simple: choose the optimal modes and keep them in line with operational adjustments.
Where did this equipment come from? He was determined by the results of trading in the wholesale market. There is a market for power and electricity. On the capacity market, manufacturers submit an application: “There is such and such equipment, these are the minimum and maximum capacities, taking into account the planned output for repair. We can issue 150 MW at such a price, 200 MW at that price, and 300 MW at that price. ” These are long term applications. On the other hand, large consumers also submit applications: "We need so much energy." Specific prices are determined at the intersection of what energy producers can give and what consumers want to take. These capacities are determined for every hour of the day.
Typically, CHPs carry approximately the same load throughout the season: in winter, the priority product is heat, and in summer, electricity. Strong deviations are most often associated with some kind of accident at the station itself or at adjacent power plants in the same price zone of the wholesale market. But there are always fluctuations, and these fluctuations strongly affect the economic efficiency of the plant. The required power can be taken by two boilers with a load of 50% or three with a load of 75% and watch, which is more efficient.
Margin depends on market prices and the cost of generating electricity. In the market, prices may be so that it is profitable to burn fuel, but it is good to sell electricity. Or maybe so that at a particular hour you need to go to a technical minimum and reduce losses. You also need to remember about the reserves and cost of fuel: the same natural gas is usually limited, and over-limit gas is noticeably more expensive, not to mention fuel oil. All this requires accurate mathematical models in order to understand which applications to submit and how to respond to changing circumstances.
How it was done before our arrival
Practically on paper, according to not very accurate characteristics of the equipment, which have a large scatter from the actual ones. Immediately after testing the equipment at best, they will be plus or minus 2% of the fact, and after a year – plus or minus 7-8%. Tests are carried out every five years, often less often.
The next point is that all calculations are carried out in standard fuel. In the USSR, a scheme was adopted when it was considered a certain conditional fuel for comparing different stations on fuel oil, coal, gas, atomic generation, and so on. It was necessary to understand the efficiency in the parrots of each generator, and the equivalent fuel is the same parrot. It is determined by the calorific value of fuel: one ton of standard fuel is approximately equal to one ton of coal. There are conversion tables for different types of fuel. For example, for brown coal, indicators are almost two times worse. But calorie content is not related to rubles. It's like gasoline and diesel: it’s not a fact that if a diesel costs 35 rubles, and the 92nd costs 32 rubles, then the caloric value of a diesel will be more efficient.
The third factor is the complexity of the calculations. Conditionally, based on the employee’s experience, two or three options are calculated, and more often the best mode is selected from the history of previous periods for similar loads and weather conditions. Naturally, employees believe that they choose the most optimal modes, and believe that not one model will ever surpass them.
We come. To solve the problem, we are preparing a digital double – an imitation model of the station. This is when, using special approaches, we simulate all technological processes for each piece of equipment, reduce steam and water balances and obtain an accurate model of the operation of the thermal power plant.
To create the model we use:
- Design and passport characteristics of the equipment.
- Characteristics according to the results of recent equipment tests: every five years, the equipment is tested and refined at the station.
- Data in the archives of industrial control systems and accounting systems for all available technological indicators, costs and generation of heat and electricity. In particular, data from heat and electricity metering systems, as well as telemechanics systems.
- Data from tape and pie paper charts. Yes, such analogue methods for recording equipment operation parameters are still used at Russian power plants, and we are digitizing them.
- Paper magazines at stations where the main parameters of the modes are constantly recorded, including those that are not recorded by ACS TP sensors. The crawler walks once every four hours, rewrites the testimony and writes everything in the journal.
That is, we have restored data sets on what mode worked, how much fuel was supplied, what were the temperature and flow rate of the steam, and how much heat and electric energy was received at the output. From thousands of such sets, it was necessary to collect the characteristics of each node. Fortunately, we have been able to play this Data Mining for a long time.
Describing such complex objects using mathematical models is extremely difficult. And even more difficult is to prove to the chief engineer that our model correctly calculates the operating modes of the station. Therefore, we went along the path of using specialized engineering systems that allow us to compose and debug the model of thermal power plants based on the design and technological characteristics of the equipment. We chose Termoflow software of the American company TermoFlex. Now there are Russian counterparts, but at that time it was this package that was the best in its class.
For each unit, its design and basic technological characteristics are selected. The system allows you to describe everything in great detail both at the logical and physical level, up to indicating the degree of deposits in the tubes of the heat exchangers.
As a result, the model of the thermal circuit of the station is described visually in terms of energy technologists. Technologists are not versed in programming, mathematics, and modeling, but they can choose the unit construct, the inputs and outputs of the units, and specify the parameters on them. Further, the system itself selects the most suitable parameters, and the technologist refines them so as to obtain maximum accuracy for the entire range of operating modes. We set a goal for ourselves – to ensure the accuracy of the model 2% for the main technological parameters and achieved this.
It turned out to be not so simple: the initial data were not very accurate, so for the first couple of months we went to the thermal power station and manually wrote off the current indicators from the pressure gauges and tuned the model to the actual modes. First made models of turbines and boilers. Each turbine and boiler were calibrated. To test the model, a working group was created and representatives of the TPP were included in it.
Then they assembled all the equipment into a general circuit and tuned the model of the thermal power station as a whole. I had to work, as there were a lot of conflicting data in the archives. For example, we found modes with a total efficiency of 105%.
When you assemble a complete circuit, the system always considers a balanced mode: material, electrical and thermal balances are compiled. Next, we evaluate how everything in the assembly corresponds to the actual parameters of the mode according to the indicators from the devices.
As a result, we got an accurate model of the technological processes of the CHP plant, based on the actual characteristics of the equipment and historical data. This allowed us to predict more accurately than on the basis of test characteristics only. The result was a simulator of the real processes of the station, a digital double of the TPP.
This simulator made it possible to conduct analysis according to the “what if …” scenarios according to specified indicators. Also, this model was used to solve the problem of optimizing the operation of a real station.
It turned out to implement four optimization calculations:
- The shift supervisor knows the heat release schedule, the system operator’s commands are known, the electricity supply schedule is known: what equipment should take the load to get the maximum margin.
- Choosing the composition of equipment according to market price forecasts: for a given date, taking into account the load schedule and the outdoor temperature forecast, we determine the optimal composition of the equipment.
- Filing applications on the market a day in advance: when there is a composition of equipment and there is a more accurate price forecast. We count and submit an application.
- The balancing market is already within the current day, when the electrical and thermal schedules are fixed, but several times a day every four hours, tenders are launched on the balancing market, and you can submit an application: “I ask you to load me up to 5 MW.” It is necessary to find the share of additional loading or unloading, when this gives the maximum margin.
For the correct tests, we needed to compare the standard loading modes of the plant equipment with our design recommendations under the same conditions: equipment composition, load schedules and weather. Over the course of a couple of months, we selected four to six hours intervals of the day with a stable schedule. We came to the station (often at night), waited for the station to enter the mode, and only then considered it in a simulation model. If the shift manager was happy with everything, then the operating personnel was sent to twist the valves and change the equipment modes.
Compared indicators before and after in fact. At peak, day and night, on weekends and weekdays. In each mode, we got fuel savings (in this task, the margin depends on fuel consumption). Then they switched completely to new modes. I must say that at the station they quickly believed in the effectiveness of our recommendations, and towards the end of the tests we increasingly noticed that the equipment was operating in the previously calculated modes.
Object: CHP with cross-links, 600 MW of electric power, 2,400 Gcal – thermal.
Team: CROC – seven people (expert technologists, analysts, engineers), CHP – five people (business experts, key users, specialists).
Implementation period: 16 months.
- Automated business processes of conducting regimes and work in the wholesale market.
- We conducted field tests confirming the economic effect.
- Saved 1.2% of fuel due to the redistribution of loads during maintenance.
- Saved 1% of fuel due to short-term equipment composition planning.
- We optimized the calculation of the stages of applications for RSV by the criterion of maximizing marginal profit.
The final effect is about 4%.
The estimated project payback period (ROI) is 1–1.5 years.
Of course, in order to implement and test all this, I had to change many processes and work closely with both the management of the CHP and the generating company as a whole. But the result was definitely worth it. It was possible to create a digital double of the station, to develop optimization planning procedures and to obtain a real economic effect.