As plant engineers focus on the effects of startup procedures and load cycles as a way of improving power plant efficiency, ABB discusses how its BoilerMax control software package can optimize steam generators’ startup strategy.

By F. Ruediger, ABB AG & B. Weidmann, E.ON-Kraftwerke GmbH, Germany

In the search for improved power plant efficiency, engineers are focusing on the effects of startup procedures and load cycles. ABB has responded to these by developing its BoilerMax control software package that automatically calculates the optimal startup strategy for steam generators, an approach that can achieve savings of between 10-20 per cent of the costs for fuel and auxiliary power per startup. This control concept employs the principle of feed-forward control as a way of ‘acting rather than reacting’ to the process. BoilerMax uses a nonlinear model of the process, where the most important dynamic behaviour of the boiler components relevant to cold, warm and hot startups are simulated, while also taking into account basic conditions such as the maximum permissible loads of critical thick-walled components or the respective minimum flow rates.

BoilerMax installations

The first practical results with Boilermax were achieved in a pilot application in STEAG Saar Energie’s Weiher III power plant. Then, over the past two years, the system has been installed in several E.ON power plants. In the 622 MW Staudinger 4 gas-fired unit and the 900 MW Heyden 4 coal-fired unit, BoilerMax has been integrated with a Procontrol P unit control system. In the 420 MW Ingolstadt 4 oil-fired unit and the 450 MW Zolling 5 coal-fired unit, BoilerMax has been integrated into a new ABB 800xA control system installed in the course of a turbine retrofit project.

The 662 MW Staudinger 4 gas-fired power station
Click here to enlarge image

These projects were mostly executed in two stages. During the first stage, a physical unit model was created using recorded historical data and the optimization potential was identified. Following analysis of this potential, the startup optimization was installed in a second stage. However, at the Staudinger power plant, the implementation was completed in a single stage. This was because this is the best approach in the case of major instrumentation and control (I&C) retrofit projects, new plant or if the startup optimization is intended to provide an improvement in startup operation independent of the expected cost savings.

Operating principle

BoilerMax’s primary purpose is to minimize startup costs, while taking into account the given process-specific basic conditions. Fuel costs and thermal stress in critical thick-walled components are taken into account when computing the optimal set-points for the fuel supply and the HP bypass station.

During a startup procedure, the actual values, which are cyclically scanned for temperature, pressure and steam mass flow rates, are used to adjust a physical unit model. Based on this nonlinear model, BoilerMax applies its predictive optimization routines to the rest of the startup procedure. The resulting startup curves, computed on-line, are then integrated into the existing unit control concept and serve as correction set-points.

The prediction horizon is 60 to 90 minutes, covering the entire length of a boiler startup, up to the point where the turbine is rolled on steam. This way the most cost-efficient overall operating mode can be computed. The predicted data are updated every one to two minutes, which enables an adequate response to disturbance conditions. The required high computing power means that BoilerMax is run on a PC configured as an application server.

The startup costs to be minimized are fuel, auxiliary power and auxiliary steam costs incurred during a boiler startup, from ‘Fire On’ to ‘Generator On-Line’ or ‘HP Bypass Closed’.

Independent of the cost savings realized, the model-based, multi-variable controller also enables a predictive integration of thermal stress data into the closed control loop. The level of flexibility, such as covering different downtimes, is improved as the physical model is continually adjusted to match the current state of the plant. Moreover, the startup can be adapted to changing basic conditions, such as different fuel costs or maximum permissible loads, by modifying the target function and optimization constraints, respectively.

Reducing Fuel Consumption

Predictive startup optimization often makes it possible to run a startup using less fuel, while maintaining the usual startup time and stress loading on critical thick-walled components. A comparison of two startup procedures in Ingolstadt 4 is shown in Figure 1. The diagrams clearly show that it was possible to run a similar boiler startup, while at the same time realizing a 20 per cent reduction in fuel consumption. Such fuel savings are possible because

Figure 1: Comparison of two startups with BoilerMax (bold lines) and without BoilerMax (thin lines) in the Ingolstadt oil-fired power plant, unit 4. The upper diagram shows the fuel quantity F_F and the HP bypass position Y_HPB. The diagram in the middle shows the live steam flow F_LS and the generator output P_Gen. The bottom diagram shows th e temperature differentials DT_SH4H and DT_SH5H occurring in HP headers of the two last superheater levels.
Click here to enlarge image

the steam flow used for starting up can be decreased by the simultaneous coordinated reduction of the opening of the HP bypass station. In addition, the startup time is slightly reduced. With a higher level of automation, achieved by the optimized startup procedures, startup procedures generally become more consistent.

Figure 2 shows the startup costs as a function of the duration of the preceding standstill. In the case of frequent brief standstills, where numerous startups are run under similar conditions, the spread of startup costs is obviously reduced. The optimized startup costs achieved with BoilerMax are at the lower end of the cost range that is characteristic of operations without BoilerMax. On average, the startup optimization resulted in a 19 per cent reduction in the startup costs.

Figure 2: Startup costs as a function of the duration of the standstill in the Ingolstadt oil-fired power plant, unit 4, with and without BoilerMax.
Click here to enlarge image

In the case of brief standstills, startup costs are significant because a high live-steam temperature has to be built up, since the temperatures in the turbine are still high. On the other hand, startup costs also rise in the case of long standstills because the boiler cools down to lower temperatures.

On account of the reduced fuel consumption, steam production and, in some cases, a slower pressure build-up, the operator might get the impression that the entire startup procedure is somehow dragging on. However, it is important to consider that, during boiler startup, the target set-points needed for steam flow and pressure must be available only when the turbine is about to be rolled on steam, and the predictive optimization concept makes full use of this fact.

At Staudinger 4, startup times with and without startup optimization were also compared over a period of one month. The startup times with BoilerMax were not found to be longer.

Reduced startup times

Startup times can generally be reduced if it is possible to intensify the heat-up process. Higher heat-up stress is also acceptable. Applying predictive startup optimization in order to reduce the startup time is advisable if the permissible heat-up stress margin is not fully exploited, or if the load is distributed unevenly during startup.

Figure 3 shows a comparison of two startup procedures in the Zolling coal-fired power plant. When starting up without BoilerMax, the margins for heat-up stress were not used before the 48th minute. The maximum temperature differential in the HP outlet headers was approximately -20 K, whereas the admissible limit was in the region of -30 K. Only during subsequent loading of the turbine was maximum heat-up stress reached around the 60th minute.

Figure 3: Comparison of two startup procedures with BoilerMax (bold lines) and without BoilerMax (thin lines) in the Zolling coal-fired power plant. The upper diagram shows the fuel quantity F_F and the HP bypass position Y_HPB. The middle diagram shows the live steam flow F_LS and the generator output P_Gen. The bottom diagram shows the temperature differentials DT_SH5H [K] as well as the associated limit values DT_SH5H_min [K] occurring in the HP outlet headers.
Click here to enlarge image

The temperature differential limit is a function of the pressure, which is generally specified by the boiler supplier and is used by BoilerMax. As an option, the limits can be recalculated during the physical modeling of the boiler and agreed in consultation with the power plant owner.

As shown in Figure 3 predictive startup optimization made it possible to better utilize the margin as early as the 35th minute. This was achieved by increasing the rate of fuel supply in the beginning and at the same time opening the HP bypass station to a greater extent. This reduced the startup time by 33 per cent and the amount of heavy fuel oil needed for starting up by six per cent. Since a shorter startup time is accompanied by a lower demand for auxiliary power (light oil and electrical auxiliary power), the total startup costs were diminished by approximately 11 per cent.

Moreover, there is a high savings potential available in coal-fired power plants if the operation provides for early shifting from burning startup fuel to coal firing. In this case as well, it is important to be able to start up using a high overall quantity of fuel. With predictive startup optimization, the amount of fuel is not necessarily increased monotonically, but can also be reduced after an initial excess supply.

Operating and monitoring

The operating screen used at Zolling 5 can be seen in Figure 4. The left and the upper part show the process parameters that are relevant during startups. The lower right area is used for the actual BoilerMax application. The set-point settings for fuel and HP bypass control, as computed by BoilerMax, are shown along with the actual values.

Figure 4: BoilerMax operating screen used in the Zolling power plant
Click here to enlarge image

The process values shown primarily cover the live-steam parameters and temperature differentials in thick-walled components. In order to avoid cluttering of the display, the temperature differential readings are shown in graphical form (bar charts). Alphanumerical representation is limited to the maximum values for each superheater level and the associated limit values.

Visualizing the temperature differential limits is especially important because BoilerMax uses these values in a closed control loop to define the fuel and HP bypass control actions. It is therefore important to present information on the current heat-up stress and the margins currently available so that the operator will be able to understand the set-point settings computed by BoilerMax.

With a new 800xA process control system, the predictive data, which are computed on-line during each startup, are available directly at the operator’s workplace. Predicted startup data can be viewed in a regular operating trend that represents the values to be expected in the future.

Integration with unit control system

From a software point of view, BoilerMax has been implemented as system extension for the ABB’s Extended Automation System 800xA. This assures a high degree of transparency and flexibility regarding its capability of being integrated into operational I&C equipment. Figure 5 highlights two possible scenarios.

Figure 5: Integration of the BoilerMax PC
Click here to enlarge image

In the first case, the BoilerMax PC can be implemented independently and connected directly to a control cabinet. Viewing and operating are done from regular operating and monitoring stations.

In plants where the operating and monitoring stations are implemented as component parts of System 800xA, the BoilerMax PC can be integrated as an application server. This provides a particular advantage – all parameter settings and calculation results, including the predicted process values, can be made visible and be integrated into the display without additional effort. Furthermore, this facilitates staff familiarization with the BoilerMax solution and the PC can be included in the regular maintenance routines for the System 800xA.

At Ingolstadt 4 and Zolling 5, the BoilerMax PC has been integrated with the 800xA operating system installed during a turbine retrofit project. At Staudinger 4 and Heyden 4, the BoilerMax PC is linked to the Procontrol P process control system via a serial interface.

The unit model used for startup optimization is adjusted on-line by incorporating 100 to 200 measured values. In general, these signals are connected to the process control system as analog signals. As an alternative, a ProfiBus connection has been established between the newly installed turbine controller and the existing unit control system in the Ingolstadt power plant. The main advantage of this digital bus coupling is a higher flexibility because individual signals can be integrated additionally, which requires little extra work. Longer signal transfer times might be viewed as a possible disadvantage. However, in the Ingolstadt power plant no problems arose in this regard.

The optimization results are fed back into the control system using approximately ten signals. They are integrated into the existing control concept in the form of set-point corrections for fuel and HP bypass control.

Depending on the given circumstances, this integration may affect different system levels. At Zolling 5, the optimized set-point for startup fuel supply is only visualized and then applied manually by the operating staff. At Staudinger 4, BoilerMax automatically performs the control of both fuel and HP bypass. At present, however, BoilerMax needs to be activated before each startup. At Ingolstadt 4, BoilerMax is activated automatically. The more automated the integration of BoilerMax is, the higher are the savings potentials because a sustained improvement of cost-efficiency will ultimately be achieved only by repeated use of the optimization function. A higher degree of automation though poses higher demands on the robustness of the startup optimization, e.g. in view of automatic detection and handling of disturbances.