PLANT

FIELD

The present disclosure relates to the domain of scheduling in power plants. More particularly, the present disclosure relates to scheduling load of a power plant.

BACKGROUND

Optimization is a technique of controlling a process, so as to optimize a specified set of parameters without violating constraints of the process. Conventionally, the optimization process in a power plant is carried out to increase efficiency, lower possible emissions, reduce cost, and maximize system availability for power generation. There are several systems that may be optimized independently in the power plant for better performance, for example, upgrading a specific component of equipment in the power plant can result in less fuel consumption. Also, the overall operation of the power plant may be optimized, by optimizing one or more factors that contribute to overall efficiency of the power plant. Typically, it is desired to optimize load scheduling in the power plant to minimize operational cost. Various conventional techniques exist for optimizing load scheduling. For example, load scheduling may be optimized based on a load demand i.e. the power plants are scheduled in such a manner that the load demand is met. As another example, load scheduling may also be optimized to meet a predetermined maintenance schedule.

As one can easily see, the operation of load scheduling has cost implications and the cost associated with load scheduling is referred as "cost of load scheduling". The cost of load scheduling may be determined from the capital cost of the equipments, fuel cost, cost of chemicals, cost of spare equipments and parts, and maintenance cost. Apart from the capital cost and the fuel cost, the maintenance cost is considered to be a significant expenditure for a power plant and a shift in the maintenance schedule may have significant change in the cost of load scheduling. The maintenance schedule of equipment may be based on regular intervals, elapsed time, or runtime meter readings. Therefore, it is often desired to adapt to any unforeseen shift or preplanned shift in the maintenance schedule to minimize the cost. Moreover, overall operation cost of the power plant also changes due to the shift in the maintenance schedule. Maintenance schedule is based on downtime resulting from scheduled maintenance of power plant components and unplanned or forced shutdowns because of sudden failure and repair activity. It is desirable to have planned and scheduled maintenance and avoid the unplanned maintenance. Therefore, maintenance activities are scheduled periodically and as frequent as possible either as recommended by the manufacturer or based on the operators past experience.

Delaying the planned maintenance schedule may increase the unplanned maintenance and the associated cost. Advancement of maintenance schedule may influence unnecessary maintenance activities and the maintenance costs. It is to be noted that there are multiple scheduling tools for scheduling production as well as maintenance but, this is often not based on the actual operating conditions and state of the component or operation under consideration.

In general, the maintenance actions required by the power plant components are notified by corresponding maintenance triggers in the form of an electronic representation which are the inputs for such scheduling tools. As per the maintenance triggers these tools will find the schedule for maintenance actions along with the production scheduling for the period of time. In such scheduling approaches, the optimization techniques used are only based on cost consideration and does not include actual operating conditions and state of the components.

With advent of advanced control system and with increased computational power available with such control system, more features are being included for optimization. In a control system, optimization may be carried out with an optimization module or a component that is already integrated with the control system or may be carried out separately based on the information available from the plant. However, it is common to find the former means i.e. having the optimization module already embedded in the control system. In mo^t cases, the optimization module utilizes a statistical or physics based model approach (first principle model) for evaluation of optimal settings. Other approaches such as that based on neural network or syntactic may also be practiced. In case of load scheduling operation, the optimized output values are the various set points to the controllers controlling the plant. The provided set points are such that the plant in overall sense functions to meet the requirements (example load demand, operation cost, efficiency, safety and regulatory requirements, maintenance requirements etc).

As mentioned earlier, in most cases, optimization is based on statistical or first principle model based approach. In such approaches, essentially there is at least one mathematical expression that relates a property of the plant as a function of measured or estimated parameters of the plant. Some examples of property of the plant are generator power output, boiler steam generation, fuel utilization, maintenance schedule, age or life expectancy of a particular unit in the plant etc. The mathematical models commonly used are related to performance of individual units in the plant or for overall coordinated functioning of the plant. In most cases, performance includes cost functions or these may be derived by suitable formulation of optimization problem. On specific aspect of load scheduling and influence of maintenance activity, one would recognize that it is common to find a predefined schedule prescribed for maintenance, though in practice maintenance activity may be an unforeseen activity carried out as a result of failure of one or several components in a power plant. As the cost of a power plant being unable for service is very high, the design of power plant is made having sufficient redundancy and margins to withstand any unusual loads or scenarios. In addition, there is adequate general knowledge or history present with the power plants about maintenance or service activities for the plant that one skilled in the art would recognize what kind of load or scenario is likely to cause failure of what component and the associated cost and downtime as a consequence of the maintenance activity. This knowledge can be efficiently utilized for scheduling the load for the power plant and include schedule for maintenance activity considering the state of the plant.

In light of the foregoing discussion, there is a need for an efficient technique for scheduling the load for a power plant and developing optimization module present in the control system to also take care of maintenance scheduling. SUMMARY OF THE INVENTION

Accordingly the present invention provides a method for optimizing load scheduling for a power plant having one or more generation units. The method comprises the steps of i) analyzing the operating state of one or more components of generation units having one or more risk indices associated with the one or more components of generation units, ii) updating the objective function that reflects the state of one or more components of generation units, iii) solving the objective function to optimize schedule of the one or more generation units and operating state of one or more components of generation units and iv) operating the one or more generation units at optimized schedule and operating state.

According to one aspect of the method, the step of analyzing includes capability assessment and operational cost assessment in the prediction horizon. Optimizing load scheduling mentioned herein includes production scheduling, maintenance scheduling or load for one or more generation units or a combination thereof. The method also includes optimizing risk indices for one or more components of the one or more generation units. Optimizing the risk indices is done by suitably changing the manipulated variables.

According to further aspect, the method includes postponing or advancing maintenance trigger for maintenance of the one or more components of the one or more generation units, which is based on the state of the component or the load demand. The objective function referred in the invention includes atleast one term for process control of the one or more components of the one or more generation units and atleast one term associated with maintenance of the one or more components of the one or more generation units. The step of updating includes updating the objective function with cost associated with postponing or advancing the maintenance of the one or more components of one or more generation units.

Accordingly the present invention also provides a control system for scheduling load of a power plant having one or more generation units. The control system comprises an optimizer having a single objective function for optimizing load scheduling which includes maintenance scheduling and for optimally controlling processes of the one or more generation units. The optimizer utilizes a plant model component and a failure model component for load scheduling optimization.

According to further aspect of the system, the optimizer comprises a scheduler analyzer for analyzing the operating state of one or more components of generation units having one or more risk indices associated with the one or more components of generation units. The optimizer is capable of scheduling maintenance based on risk indices associated with the one or more components of one or more generation units or demand forecast or improvements in the operating state affected by new manipulated variables or prescheduled maintenance effected by maintenance trigger or combinations thereof.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a block diagram of a system for scheduling load of a power plant, in accordance with which various exemplary embodiments can be implemented;

FIG. 2 is a block diagram of an optimizer for scheduling load of a power plant, in accordance with one embodiment;

FIG. 3 is a block diagram of a simplified generic fossil fired power plant (FFPP), in accordance with one embodiment; FIG. 4 shows an exemplary demand forecast profile.

FIG. 5 illustrates a method for scheduling load of a power plant in accordance with one embodiment. DETAILED DESCRIPTION

It shall be observed that method steps and system components have been represented by conventional symbols in the figures, showing only specific details that are relevant for an understanding of the present disclosure. Further, details that may be readily apparent to person ordinarily skilled in the art may not have been disclosed.

Exemplary embodiments of the present disclosure provide a method and system for scheduling load of a power plant. Normally power plants are scheduled to produce power for a period of time/prediction horizon varying from days to weeks, termed as short term load scheduling. Production schedule of power plant is been done on the basis of power/steam demand, availability of power plant components and the net revenue from production. As said, the optimizer uses the different cost factors include penalty for not meeting the demand, revenue of power sales, fuel consumption, emission reduction, components depreciation, startup and shutdown of components to load the power plant optimally. In addition to these cost factors the optimizer also use the cost associated with the maintenance schedule of each component assessed with consideration to the state of each component in terms of risk index, also referred as risk indices and EOH (Equivalent Operating Hour) compensation referenced with respect to its nominal functioning and life expectancy, to find the optimal production, maintenance schedules and the operating conditions and achieve maximized revenue and efficiency.

In accordance with a first aspect, a method for scheduling load of a power plant by an optimizer includes receiving one or more inputs, the one or more inputs are associated with a plurality of constituents of the power plant. The method includes calculating a risk index of at least one of the constituents of the power plant responsive to the one or more inputs. The method includes determining a load based on the risk index, the load is associated with an output power of the power plant. The method includes operating the power plant based on the load.

FIG. 1 is a block diagram of a system for scheduling load of a power plant, in accordance with which various embodiments can be implemented. The system includes an optimizer 105, a forecast module 125, a user input module 130, a plant database 135, a plant controller 140 and a power plant 145.

The optimizer 105 includes a model component 110, a failure model component 120 and an EOH compensation model component 115. The optimizer 105 receives one or more inputs from a forecast module 125, a user input module 130, a plant controller 140 and from a plant database 135.

The system includes a forecast module 125 to provide load forecast for the power plant 145 over a period of time. The demand for the load keeps fluctuating and hence there is a need for forecasting the load demand. The forecast module 125 may use user input data to provide forecast information or have dedicated forecast models based on statistical models or other techniques. One or more generators can be selected for operation based on the forecasted load demand. Furthermore, by switching off the generators based on the load demand, the operation cost of the power plant 145 can be minimized. The load demand forecasted is further sent as an input to the optimizer 105 for scheduling the load of the power plant 145. In addition to the above, the forecast module 125 is also used to provide a predetermined maintenance schedule for one or more constituents of the power plant 145. The predetermined maintenance schedule, type of maintenance and the period/periodicity of the maintenance schedule of one or more constituents of the power plant 145 are based on operator's experience or component manufacturer recommendations. The predetermined maintenance schedule over a period of time is further sent as an input to the optimizer 105 for scheduling the maintenance of one or more constituents of the power plant 145.

The system includes the user input module 130 for receiving a plurality of user inputs to the optimizer 105. The plurality of user inputs includes but is not limited to a cost of fuel, an emission penalty, an equipment life cost, and a spare unit operating cost. The user inputs are further sent as inputs to the optimizer 105.

The power plant 145 includes a plurality of units. Information related to the plurality of units of the power plant 145 and their operating conditions is stored in the plant database 135. An operating history, current status, manufacturing details and maintenance scheduling of the plurality of units of the power plant 145 is also stored in the plant database 135. The information related to the plurality of units of the power plant 145 is further sent as an input to the optimizer 105 through plant controller 140, for scheduling the load of the power plant 145. The system includes the power plant 145. The power plant 145 receives the load and the maintenance schedule determined through plant controller 140 and the power plant 145 is operated based on the load determined by the optimizer 105.

FIG. 2 is a block diagram of an optimizer 105 for scheduling load of a power plant 145, in accordance with one embodiment. The optimizer 105 includes a plant model component 110, schedule analyzer 113 and optimization solver module 118. Schedule analyzer 113 mentioned herein comprises EOH compensation model 115 and failure model component 120. The schedule analyzer 113 through various plant parameters (eg. measured variables, plant database) analyzes the factors as required for solving the objective function providing load schedule (production / maintenance schedule) and the load values (set points) for one or more components of the one or more generation units. Collectively, the schedule analyzer is said to analyze the operating state (capability assessment of a particular generating unit to effectively perform its function with regard to both process and cost effectiveness). Here, the capability assessment includes risk assessment, demand assessment and based on risk assessment and demand assessment, also assess the need for maintenance including suggestion of a schedule for maintenance. Further, the cost aspect herein above referred relates to operational cost that includes maintenance cost.

A plurality of manipulated values of load scheduling problem handled by the one or more units from can be fed as inputs to a plant model component 110 and the failure model component 120. Both plant model component 110 and a failure model component 120 also receive one or more inputs from the power plant 145 through plant controller 140, a plant database 135, a forecast module 125 and a user input module 130. The optimizer component 105 has optimization solver module 118 and that is used to find an optimal load schedule for the power plant 145, by the minimization of an objective/estimation function of the power plant 145 responsive to receiving the one or more inputs. The objective/estimation function to be minimized by the optimization module 118 of optimizer 105 includes a penalty cost for not meeting the demands, an operating cost due to fuel consumption, a start-up cost, a shutdown cost, an ageing cost, an emission cost and a maintenance cost. The optimization solver module 118 uses the well known max-min optimization technique for finding optimal load and maintenance schedules for the power plant 145. During the iterative process of minimizing the objective/estimation function the optimization module 118 of optimizer component 105 uses a model component 110 and a failure model component 120 of one or more constituents of the power plant 145. This optimization process continues till the optimization module finds the optimal load schedule for which the value of objective function is minimized.

The failure model component 120 receives the manipulated values and the inputs from the plant database 135 (the database has both present values and history information related to plant parameters as required for computations by the failure model component). The failure model component 120 then calculates a risk index of at least one of the constituents of the power plant 145 based on the manipulated values and one or more inputs received from the plant database 135. The risk index values for one more constituents of the power plant are then passed from the failure model component 120 to the EOH compensation model 115. The EOH compensation model 115 has cost factors associated with each risk index values of one or more constituents of the power plant 145. The optimizer 105 determines the load and the maintenance schedules based on the cost factors associated with calculated risk index values and operates the power plant 145 (loads the power plant) to meet the demand in the best possible manner. FIG. 3 is a block diagram representation of a simplified generic fossil fired power plant (FFPP) 145 that is controlled by a control system that includes an optimizer 105 to calculate the optimal solution for operating the power plant. The FFPP consists of three FFPP units 150, 155, 160 running in parallel. Each FFPP unit has three main equipments namely, a boiler (B) 165, a steam turbine (ST) 170 that is mechanically coupled with an electrical generator (G) 175. Under operation, steam loads, generally referred to as manipulated variables u _t , u ₂ and u ₃ are applied to the respective boiler to generate the output in the form of steam, expressed as yn, y ₂ i, y that is given to the steam turbine combined with electrical generator for electrical power generation. The electrical power output from the generator is expressed as yn, y ₂ 2, y ₃ 2- The control system 140 is used to monitor and control the different operating parameters of the power plant 145 to ensure the power plant is operated at the optimum conditions. For optimal running of the power plant, as explained earlier, one of the critical aspects is the optimal load scheduling between the different FFPP units and the calculation for the optimized solution is done at the optimizer 105.

In the exemplary embodiment, the objective of a load scheduling optimization problem is to meet the power demand by scheduling the load among the three FFPP units, subject to different constraints such as the minimization of the fuel cost, start up cost, running cost, emission cost and life time cost. The optimizer 105 receives inputs from the power plant, and applies optimization techniques for the optimal load scheduling. Based on the optimal solution, the control system 140 sends commands to different actuators in the power plant 145 to control the process parameters.

The objective function used for optimization is as follows:

minimize (J) = C _dem + C _fud + C _{slart shui} + C _emission + C _t y _e + C _maintenance — C _revenue

C _dem is the penalty function for not meeting the electric demands over the period of time termed as prediction horizon.

where, k _{dem E} i _ec (t) is the suitable weight coefficient and D _{dem Elec} (t) , for t = T,...,T+ M- dt is the forecast of the load demand within the prediction horizon M, ya is the electrical power generated by all the 'n' units. With reference to Fig. 3, n = 3. Cfi _t ei is the cost for fuel consumption represented in model for FFPP by the outputs yu, yu, JV31 and thus the total cost for fuel consumption is given by,

Cfi _iel =

^emission is the costs involved in reducing the pollutant emission (NO _x , SO , CO _x ) produced by the power plant and is given by,

T+M-dt n

^emission ^ ^' ^ ^i emission f j ₂ ( ) where k _iemjssion represents the positive weight coefficients and f(y _i2 (t)) represent the non-liner functional relationship between the load and the emission production.

C _s t _{art s} i _ml is the cost function for starting/shutting of one or more constituents of the power plant and is given by

Γ+Μ-2Λ n

C _s tart,s,„U = Σ Σ^ _/ ^ ^ί^^ + ^ - Μ,,ί ,θ} where k _{i slartlshut} represent the positive weight coefficients, u _H are the integer states (On/Off) of the units.

C _nfe describes the asset depreciation due to loading effect and is defined as,

where, the depreciation cost of each component is calculated as,

Load

LT compjoad

^Load base J ioUU where, Load and Load _base are the load, base load on each component of the power plant respectively. costEOH.comp is the cost per EOH for a specific component of the power plant and dt is the sampling time.

Cmaintenance , the cost of maintenance for one or more components of the power plants is defined as follows: C maintenance - c fixed +c Risklndex + c maintenance shift

where, Cf _aed is the sum of fixed maintenance costs for different components of the power plant. Cmaintenance s ift is the cost of depreciation of the component of power plant due to the shift in predetermined maintenance schedule. It is defined as follows:

C maintenance shift =

Load

where, LT c'ompjoad * At * cost _£CW∞mp and At is the shift in maintenance schedule from the predetermined schedule.

CRisk index is the compensation cost corresponding to risk index value provided by the failure model.

(-■revenue i ^{s tne term} f° ^r revenues obtained by the sales of electrical energy.

T+M -dt n

^C revenue = Σ t=T Σ Pi.Etec ⁽ #2 ⁽ 0

=l

where, Pi, EI∞ (0 is the cost coefficient for the electrical energy for sale.

The objective of the power plant is to maximize revenue and minimize maintenance and penalty costs. This directly depends on the time for active production (production schedule) and inversely to the maintenance time (maintenance schedule) when the production is stopped or not to its full capacity.

In the proposed formulation, operation of the plant activities including maintenance is based on actual operating conditions and is also capable of influencing the operating conditions for production schedule and maintenance activities. The maintenance cost of one or more constituents of the power plant is thus determined based on the operating conditions using a failure model component and an EOH compensation model component.

The failure model component is either based on first principle models (ageing models) or probability models based on statistical distributions relating operating condition parameters ( based on history/experimental data, the excepted life under defined operating conditions. Example- For electrical ageing the electric stress value and the time of operation). The failure model component includes the measures for severity, occurrence and detection of failures for different operating conditions using the FMEA technique. The measures are in terms of scores (eg score between 0-10), derived from the state of the plant also categorized or coded in terms of scores derived from manipulated variables or data from the plant database. Severity measure is an estimate of how severe the production schedule will be affected by a failure. In one embodiment, severity is defined to depend on multiple factors and each of the factors may be summated and scaled by appropriate weight functions associated with each of the factors. Some examples of the factors are provided for obtaining a measure for Severity. These are

a) the time of operation with respect to maintenance i.e. Severity is assumed to be high if the component within a unit for which the severity information is being coded is already due for maintenance. The severity measure is medium if it is approaching the pre-defined or allotted schedule for maintenance and is low if it is fresh from maintenance. (High, medium and low may have a corresponding numerical score associated with it). This factor is automatically coded from the history information associated with the component obtained from the plant database;

b) impact on downtime due to failure, this factor may be coded again based on the critically associated with the component either based on the service history available in the plant database or based on the judgment of skilled persons associated with the component. If there is no impact i.e. the functioning of the unit will not get impacted in any manner by failure of the component for any reasons including that there is standby component that improves reliability, the score may be taken as low and depending on the impact associated with the unit becoming unavailable for production, the score is judged to be high or medium. The impact is pre-defined function associated with each component;

c) cost for replacement associated with the component of the plant may also be coded based on relative cost for replacement of various serviceable/replaceable components in the power plant;

d) complexity of failure and repair for various component in the plant units may also be coded based on the expertise required to attend to the maintenance activity or/and based on the intricacies involved in the maintenance activities. In another embodiment, the factors for Severity may be derived from categorization of region of operation for various components in the power plant. Here, the calculation of the risk index value is based on the priorities from the region of operation, their initial conditions and the depreciation rate associated with the MVs for each components of power plant. Three different operating regions for one or more constituents of a power plant in association with the manipulated variables are defined. The regions of operation for one or more constituents of the power plant include a recoverable region, a replaceable region and a catastrophe region. Recoverable region is defined as a region where by suitably adjusting the operating conditions of units, the condition of the unit may be maintained in a manner that does not incur significant loss or any abrupt failure resulting in loss of service of the particular unit. This region is said to be with a low risk index value. Replaceable region defines a medium risk and denotes a condition wherein the unit or a major component in the unit is nearing its recommended maintenance period as prescribed by the manufacturer or as recommended based on the history information (experience) for its nominal functioning and on failure is likely to create a loss, though manageable from cost perspective, by small disruption in service or by failures resulting in a replacement activity carried out in small time or manageable by activation of the spare unit. Catastrophe region defines a high risk value where it is not any more likely to have any benefit by delay in maintenance activity i.e. any failure incurred either has huge impact in production schedule or in maintenance cost. The operating region information are designated values by plant engineers as suitable by the design or specification of the component and the service or other history information available in the plant database. The risk levels are coded into the severity measure in the failure model component. It is re-iterated that the various factors contributing to severity rating of the component may be summated and scaled appropriately depending on the role/importance of the component and various components in turn summated and scaled to represent the power plant unit.

Occurrence data is associated with factors such as the probability of failure of a particular component in the power plant unit. The probability of failure is further related to the operating conditions/state of the power plant components (eg probability of failure for every major component in the plant units based on the accumulated stress levels and the time of accumulation of these stresses). The probability value is also coded as low, medium and high for various components, summated and scaled to represent for a power unit. Depending on the component that is at risk (medium or high probability value), the manipulated variables may be so adjusted by the optimizer to minimize the risk of having the component fail. Another factor that may be used is the frequency of occurrence of fault. If the component is prone to failures, the factor may be coded as "high" indicating high risk. Detection is used to reflect how easy it is to detect the growth of defect leading to a failure of the component in operation in the power plant unit. On having a good detection mechanism in place to monitor defect in the control system, the value may be coded as "low" indicating low risk associated with the component. The value may be high for defects that appear suddenly or have multiple modes or are not monitored (directly or indirectly). It is to be recognized that the defects or failure modes that are monitored or is capable to be monitored has low risk as these defects based on the knowledge of failure mode and fault tree are controllable by adjusting the operating conditions (controlled through manipulated variables). For the components, where the defects are non-observable or there are means to infer or estimate, the risk may be considered as "medium".

The failure model component calculates the risk index corresponding to the operating state (derived from plant database, manipulated variables (MVs) or measured variables) for each power plant units/components. The risk index value of one or more constituents of the power plant is taken into account for scheduling maintenance and have scores (value of risk index) categorized further as follows:

Lower risk index: No need for maintenance schedule. The maintenance schedule may be allowed to be postponed if found justified based on cost or if the component vulnerable to failure is of the nature that can easily be replaced without affecting or with minimum affect to the plant. That is that the risk index does not affect the operation of the plant.

Medium risk index: Maintenance schedule may be delayed and may be accommodated when the power plant component is idle or not scheduled for production schedule if such a schedule is anticipated in near future (within the prediction horizon) or if the risk index value is expected to become lower in the prediction horizon. However, alteration in operating conditions recommended to reduce the risk level or at the least maintain risk within manageable limits. Eg. the plant may be operated in reduced stress conditions (not to overload or operated below full capacity). The alteration in operating conditions may also be made to reduce the risk value from medium to low i.e. have the plant/component of the plant under risk recover.

Higher risk index: Maintenance schedule is compulsory. Immediately or shortly schedule for maintenance. Here, the component under risk is likely to fail at short notice and will severely affect operation of the plant or influence downtime significantly. From this system, maintenance is scheduled based on the condition of the system and accounting the overall benefit (cost function) considering the load demand forecast over the prediction horizon, i.e., the optimization of the system decides when to go for maintenance. This is with reference to both risk index (indicative of the actual condition of the system) and operation/mairitenance/penalty costs. If high risk is found, a new set of manipulated variables is provided as optimizer tries to reduce the risk index, the associated EOH compensation and other costs (aging, maintenance) going high.

The calculated risk index values for one more constituents of the power plant 145 by the failure model component 120 is passed to the EOH compensation model 115. The EOH compensation model 115 provide the corresponding EOH compensation cost factors associated with the each category of risk index values (high/medium/low) of one or more constituents of the power plant 145. The EOH compensation model is based on lookup tables (eg. Table 1) for each component of the power plant which defines the EOH compensation values and the cost per EOH corresponding to each category of risk index value.

As an example, the use of EOH compensation model is illustrated for a boiler component in a power plant. Consider a following table to represent risk index values and the associated EOH compensation. The cost factor associated with the risk index is calculated as,

C Risklndex = ^E H Compensation * cost _£0¾com/ ,

Table 1: Coding for EOH compensation

For example, if a boiler of a power plant operating with a high risk value of 0.8 and the Costaw=100$ ( ^sa y) _> ^{men tne cost} factor associated is calculated as, C _RiskIndex =50* 100=5000$. This cost factor will be added into the objective function J. In the process of minimizing the objective function (J), the optimizer component tries to avoid such cost factors associated with the high risk index values, by scheduling maintenance activities for such components in the power plant.

In finding the optimal load scheduling of power plant, the optimizer always try to operate the units in less severe conditions/recoverable region. If a unit falls under the replaceable region, the optimizer try to reduce the manipulated variable values in order to reduce the risk value from medium to low. Similarly, in any case, if any unit falls under the catastrophe region, the optimizer will try to schedule the maintenance for the corresponding unit if in near future it is not foreseen to reduce risk level. The schedule analyzer uses the plant model to help in simulating the plant condition in the prediction horizon to find if the risk is expected to come down within the prediction horizon.

Further the initial starting points i.e. the service information related to accumulated stress and time of application of these stresses, for the defined operating regions are derived from the power plant operation history.

Depreciation/ Ageing of the equipment of a power plant is closely associated with the operating conditions of the power plant. As said, the manipulated variables is related with the probability of failure value (coded suitably to obtain Occurrence score in the Failure model) and also in calculation of the depreciate rate. For e.g., the rate of depreciation increases if the equipment is heavily loaded. With reference to the severity rating, the value of depreciation rate also varies. A delay in predetermined maintenance schedule, may lead to increased depreciation rate and the associated cost factors. But at the same time, if there is any useful tradeoff by setting the manipulated variables in such a manner so to accomplish delayed maintenance by managing depreciation rate from the time of such setting till the period after the predetermined maintenance schedule, the optimizer is likely to operate the plant then with such a setting for the manipulated variable. In such cases, the maintenance schedules can be delayed and the unit can be scheduled for production with loads as recommended by the optimizer for overall benefit. In one embodiment, the maintenance activity is scheduled as defined in the lookup table (Table 2) based on the risk levels and the load forecast. The column load forecast has relative codified information where the load requirement based on the forecast is categorized as "Low" when the load requirements are easily met by other units. The load forecast value is "Medium" when the load requirements are met by other units under low loss, by maintaining at nominal levels or by slight increasing in the output of unit beyond nominal levels. The load forecast value is "High" when the load requirements cannot be met without participation of the unit in question for maintenance (the penalty values are high). As an initial condition, the predetermined maintenance schedule for one or more constituents of the power plant (say, after every period of T _m , maintenance of a particular component of the power plant has to be scheduled) is obtained from the forecast module and used by the optimizer as initial conditions for scheduling maintenance activities. At any time t, the time for next maintenance is (T _m -t).

The system will optimize load schedule in the prediction horizon based on capability assessment, operational cost assessment and shifts in maintenance schedule. The risk levels and the EOH compensations are considered. If at the time (T _m -t), the risk index value is acceptable for production scheduling, the unit is scheduled for production. In case, the risk level is such that it suggests maintenance to be scheduled, the system tests for improvement by suitably changing the manipulated variables to create a condition where the risk level is expected to improve. The system may also adopt simulation as a means for test for improvement using plant model component. If at the simulation test or at the test in time (At) with new manipulated variables, the forecasted or actual risk index value is still high (no significant reduction in risk index value), the system will prompt/force the component/unit to go into maintenance within the preconfigured time (e.g., ramp down + ΔΤ, where ΔΤ is the reaction time to have the other units adjusted or have the required consents from operators).

If during the improvement test, the forecasted or actual risk value is getting reduced (significant improvement in the system) from running the plant with the new manipulated variables (MVs) , then the maintenance may not be carried out at (T _m -t) and the new maintenance schedule of T _m +mAt is recorded in the forecast module. It is to be recognized that the simulation activity is optional and the optimizer in one embodiment may be described without the simulation test to determine for improvement. However, at the minimum, there may be sufficient analysis to indicate there is cost benefit by altering the maintenance schedule and these analyses may be done by the schedule analyzer. Further, the schedule analyzer may also be used to indicate an optimal shift instead of postponement by mAt for maintenance schedule only on the cost basis using well known procedures such as that provided in the US Patent 6,999,829 including having an objective function based on cost to optimally determine the shift required in the maintenance schedule.

It is also to be noted that with postponement of maintenance schedule, the objective function needs to consider a higher value of depreciation cost till the time the maintenance is carried out. This factor is added as additional term in the objective function as a part that accounts maintenance cost.

The method of optimization is explained briefly with an example below with reference to Fig. 3. Let us assume the units 1, 2 and 3 are identical. Consider that the maximum load carrying capacity of units 1, 2 and 3 are 60 MW, 60 MW and 50 MW respectively. A typical demand forecast profile is assumed to demonstrate the optimal schedule of both production and maintenance activities over the prediction horizon as shown in Fig. 4.

All the three units (Unit 1, 2, 3) are optimally scheduled to produce 40 MW each, to meet the total demand of 120 MW till time t _\ . After time ti, the demand profile changes to 160 MW as inferred from 410 of Fig.4. Since the maximum load carrying capacity of Unit 3 is only 50 MW, as inferred from 440, it is scheduled by the optimizer to produce 50 MW. Remaining 110 MW is shared among units 1 and 2.

From 420 it is seen that, between time ti to t2, there exists a predetermined maintenance trigger for Unit 1 , denoted as T _m . The schedule analyzer and the optimization solver module estimates and takes into account various factors associated with load scheduling / maintenance scheduling of Unit 1. In this example, at the point around T _m the risk index for Unit 1 is medium. The medium risk value for risk index and high demand suggests for an immediate need for reduction of risk and at the same time meet the demand. In this example, for Unit 1, the optimizer is shown to time up with a value of 50 MW as the load that is optimal cost wise and may reduce the risk. New set point to reduce risk is sent for Unit 1 corresponding to 50 MW. To meet the demand and to have Unit 1 function at 50 MW, Unit 2 is scheduled to ramp up from 55 MW to 60 MW as inferred from 430. The optimizer checks for improvement in At within the prediction horizon and it finds that the risk on Unit 1 indeed reduces to low with At in the prediction horizon. Hence, the maintenance activities of Unit 1 are postponed to T _m +At. The optimizer records the suitable time for maintenance schedule for Unit 1 as T _m +At, where the At is the time within the prediction horizon and shift in maintenance schedule from T _m .

The demand forecast goes down just after the schedule T _m +At and the demand is low enough to be met by Unit 2 and Unit 3. As the demand is low and even though the risk index of Unit 1 is low, Unit 1 is taken for maintenance at T _m +At. At time t2, Units 2 and 3 are scheduled to take care of the total demand of 50 MW.

Table 2: Maintenance activity scheduling based on Risk Index and Demand Forecast

T _m is the schedule for maintenance; "m" is a predetermined number; and At is time within the prediction horizon.

Fig. 5 illustrates a method for optimizing load scheduling for a power plant in accordance with an embodiment of the present invention. The power plant herein referred has one or more generation units having one or more components thereof.

Optimizing load scheduling here includes production scheduling, maintenance scheduling or load for the generation units and the like. It also includes optimizing the risk indices for the components of the generation units, which can be done by changing the manipulated variables as described herein above. Optimizing load scheduling can also include postponing or advancing the maintenance trigger for maintenance of the components of the generation units which is based on the load demand or the state of the component.

Step 505, refers to analyzing the operating state of the components of the generation unit. The generation units have one or more risk indices associated with one or more components of generation units. The step of analyzing includes capability assessment, operation cost assessment and shifts in maintenance schedule therein within the prediction horizon. Step 510, involves updating the objective function that reflects the state of one or more components of the generation units. The objective function herein mentioned includes atleast one term for process control of the components and atleast one term associated with maintenance of the components. Also, the step of updating the objective function includes updating with respect to the cost associated with postponing or advancing the maintenance of the components of the generation units or with the cost associated with the lifecycle of the components taking into account the depreciation or degradation of the components therein.

In step 515, the objective function is solved, in order to optimize the schedule of the generation units operating state of the components of the generation units, as depicted in step 520.

Step 525 refers to the step of operating the power plant. This has reference to operating the generation units at optimized schedule and operating state thereof. It should also be noted that the control system offers means to allow an operator / user to override load scheduling (eg. maintenance schedule) or operating state by inputs through a suitable user interface directly by specifying a particular schedule or operating state for one or more components of the generation units in a power plant or indirectly by manipulating the certain parameters / variables related with schedule analyzer or forecast module.

While only certain features of the invention have been illustrated and described herein, many modifications and changes will occur to those skilled in the art. It is, therefore, to be understood that the appended claims are intended to cover all such modifications and changes as fall within the true spirit of the invention.